This is a reproduction of a library book that was digitized

by Google as part of an ongoing effort to preserve the

information in books and make it universally accessible.

http://books.google.com

SYNOPSIS

ELEMENTAKY

PUKE

RESULTS

MATHEMATICS

CONTAINING

PROPOSITIONS.lFORMUI^E, AND METHODS OF ANALYSIS,

WITH
ABRIDGED DEMONSTRATIONS.
supplemented by an index to the papers on purb mathematics which are to
be found in the principal journals and transactions of lfarnfd societies,
both English and Foreign, of thb present century.

iiy
G.

S.

CARR,

M.A.

LONDON :
FRANCIS HODGSON, 89 FARRINGDON STREET, E.G.
CAMBRIDGE : MACMILLAN & BOWES.
1886.
(All rights reserved )

4 A <

Libi-../

LONDON :
PRINTED BY C. F. HODGSON AND SON,
GOUGH SQUASE, FLEET STREET.
76

(o

Q<

or ttj-

PREFACE TO PART I.

The work, of which the part now issued is a first instal

ment, has been compiled from notes made at various periods
of the last fourteen years, and chiefly during the engagements
of teaching. Many of the abbreviated methods and mnemonic
rules are in the form in which I originally wrote them for my
pupils.
The general object of the compilation is, as the title
indicates, to present within a moderate compass the funda
mental theorems, formula?, and processes in the chief branches
of pure and applied mathematics *
The work is intended, in the first place, to follow and
supplement the use of the ordinary text-books, and it is
arranged with the view of assisting the student in the task of
revision of book-work. To this end I have, in many cases,
merely indicated the salient points of a demonstration, or
merely referred to the theorems by which the proposition is
proved. I am convinced that it is more beneficial to the
student to recall demonstrations with such aids, than to read
and re-read them. Let them be read once, but recalled often.
The difference in the effect upon the mind between reading a
mathematical demonstration, and originating one wholly or

IV

PBEFACE.

partly, is very great. It may be compared to the difference

between the pleasure experienced, and interest aroused, when
in the one case a traveller is passively conducted through the
roads of a novel and unexplored country, and in the other
case he discovers the roads for himself with the assistance of
a map.
In the second place, I venture to hope that the work,
when completed, may prove useful to advanced students as
an aide-memoire and book of reference. The boundary of
mathematical science forms, year by year, an ever widening
circle, and the advantage of having at hand some condensed
statement of results becomes more and more evident.
To the original investigator occupied with abstruse re
searches in some one of the many branches of mathematics, a
work which gathers together synoptically the leading propo
sitions in all, may not therefore prove unacceptable. Abler
hands than mine undoubtedly, might have undertaken the task
of making such a digest ; but abler hands might also, perhaps,
be more usefully employed,and with this reflection I have the
less hesitation in commencing the work myself. The design
which I have indicated is somewhat comprehensive, and in
relation to it the present essay may be regarded as tentative.
The degree of success which it may meet with, and the
suggestions or criticisms which it may call forth, will doubt1-x i have their effect on the subsequent portions of the work.
With respect to the abridgment of the demonstrations, I
may remark, that while some diffuseness of explanation is not
only allowable but very desirable in an initiatory treatise,
conciseness is one of the chief requirements in a work intended

PBEFACE.

for the purposes of revision and reference only. In order,

however, not to sacrifice clearness to conciseness, much more
labour has been expended upon this part of the subject-matter
of the book than will at first sight be at all evident. The only
palpable result being a compression of the text, the result is
so far a negative one. The amount of compression attained
is illustrated in the last section of the present part, in which
more than the number of propositions usually given in
treatises on Geometrical Conies are contained, together with
the figures and demonstrations, in the space of twenty-four
pages.
The foregoing remarks have a general application to the
work as a whole. With the view, however, of making the
earlier sections more acceptable to beginners, it will be found
that, in those sections, important principles have sometimes
been more fully elucidated and more illustrated by examples,
than the plan of the work would admit of in subsequent
divisions.
A feature to which attention may be directed is the uni
form system of reference adopted throughout all the sections.
With the object of facilitating such reference, the articles have
been numbered progressively from the commencement in
large Clarendon figures ; the breaks which will occasionally
be found in these numbers having been purposely made, in order
to leave room for the insertion of additional matter, if it should
be required in a future edition, without disturbing the original
numbers and references. With the same object, demonstrations
and examples have been made subordinate to enunciations and
formulas, the former being printed in small, the latter in bold

VI

PREFACE.

type.

By these aids, the interdependence of propositions is

more readily shown, and it becomes easy to trace the connexion

between theorems in different branches of mathematics, with
out the loss of time which would be incurred in turning to
separate treatises on the subjects. The advantage thus gained
will, however, become more apparent as the work proceeds.
The Algebra section was printed some years ago, and does
not quite correspond with the succeeding ones in some of
the particulars named above.

Under the pressure of other

occupations, this section moreover was not properly revised

before going to press.

On that account the table of errata

will be found to apply almost exclusively to errors in that

section ; but I trust that the list is exhaustive.

Great pains

have been taken to secure the accuracy of the rest of the

volume.

Any intimation of errors will be gladly received.

I have now to acknowledge some of the sources from which

the present part has been compiled.

In the Algebra, Theory

of Equations, and Trigonometry sections, I am largely in

debted to Todhunter's well-known treatises, the accuracy and
completeness of which it would be superfluous in me to dwell
upon.
In the section entitled Elementary Geometry, I have added
to simpler propositions a selection of theorems from Townsend's Modern Geometry and Salmon's Conic Sections.
In Geometrical Conies, the line of demonstration followed
agrees, in the main, with that adopted in Drew's treatise on the
subject.

I am inclined to think that the method of that

author cannot be much improved.

It is true that some im

portant properties of the ellipse, which arc arrived at in

PREFACE.

VU

Drew's Conic Sections through certain intermediate proposi

tions, can be deduced at once from the circle by the method of
orthogonal projection. But the intermediate propositions can
not on that account be dispensed with, for they are of value in
themselves. Moreover, the method of projection applied to
the hyperbola is not so successful ; because a property which
has first to be proved true in the case of the equilateral
hyperbola, might as will be proved at once for the general case.
I have introduced the method of projection but sparingly,
always giving preference to a demonstration which admits of
being applied in the same identical form to the ellipse and to
the hyperbola. The remarkable analogy subsisting between
the two curves is thus kept prominently before the reader.
The account of the C. G. S. system of units given in the
preliminary section, has been compiled from a valuable con
tribution on the subject by Professor Everett, of Belfast,
published by the Physical Society of London.* This abstract,
and the tables of physical constants, might perhaps have found
a more appropriate place in an after part of the work. I have,
however, introduced them at the commencement, from a sense
of the great importance of the reform in the selection of units
of measurement which is embodied in the C. G. S. system,
and from a belief that the student cannot be too early
familiarized with the same.
The Factor Table which follows is, to its limited extent, a
reprint of Burckhardt's " Tables des diviseurs," published in

* " Illustrations of the Centimetre-Gramme-Second System of Units.

London : Taylor and Francis. 1875.

Vlll

PREFACE.

1814-17, which give the least divisors of all numbers from 1

to 3,036,000. In a certain sense, it may be said that this is
the only sort of purely mathematical table which is absolutely
indispensable, because the information which it gives cannot
be supplied by any process of direct calculation. The loga
rithm of a number, for instance, may be computed by a
formula. Not so its prime factors. These can only bo
arrived at through the tentative process of successive divisions
by the prime numbers, an operation of a most deterrent kind
when the subject of it is a high integer.
A table similar to and in continuation of Burckhardt's has
recently been constructed for the fourth million by J. W. L.
Glaisher, F.R.S., who I believe is also now engaged in com
pleting the fifth and sixth millions. The factors for the seventh,
eighth, and ninth millions were calculated previously by Dase
and Rosenberg, and published in 1862-G5, and the tenth
million is said to exist in manuscript. The history of the
formation of these tables is both instructive and interesting.*
As, however, such tables are necessarily expensive to pur
chase, and not very accessible in any other way to the majority
of persons, it seemed to me that a small portion of them
would form a useful accompaniment to the present volume.
I have, accordingly, introduced the first eleven pages of Burckh
ardt's tables, which give the least factors of the first 100,000
integers nearly. Each double page of the table here printed is

* See " Factor Table for the Fourth Million." By James Glaisher, F.R.S.
London : Taylor and Francis. 1880. Also Camb. Phil. Soc. Proc, Vol. III.,
Pt. IV., and Nature, No. 542, p. 462.

PREFACE.

IX

an exact reproduction, in all but the type, of a Bingle quarto

page of Burckhardt's great work.
It may be noticed here that Prof. Lebesque constructed
a table to about this extent, on tho plan of omitting the
multiples of seven, and thus reducing the size of the table
by about one-sixth.* But a small calculation is required in
using the table which counterbalances the advantage so gained.
The values of the Gamma-Function, pages 30 and 31, have
been taken from Legendre's table in his "Exercices de Calcul
Integral," Tome I. The table belongs to Part II. of this
Volume, but it is placed here for the convenience of having
all the numerical tables of Volume I. in the same section.
In addition to the authors already named, the following
treatises have been consultedAlgebras, by Wood, Bourdon,
and Lefebure deFourcy; Snowball's Trigonometry ; Salmon's
Higher Algebra ; the Geometrical Exercises in Potts's Euclid ;
and Geometrical Conies by Taylor, Jackson, and Benshaw.
Articles 260, 431, 569, and very nearly all the examples,
are original. The latter have been framed with great care, in
order that they might illustrate tho propositions as completely
as possible.
G. S. C.
Hadley, Middlesex ;
May 23, 18S0.

"Tables diverses pour la d6composition dca nombres en leurs fucteurs

premiers." Par V. A. Lebesque. Paris. 1864.

EBB ATA.
Art. 13,
,, 66. Line 1,
>. 66. " 6,
90, it *,
99, n 1,
107, n 1.
108, j> 2,
131, ft 1, 2,
. ..
n 5,
133, M 3,6,7,
.
ft 8.
> ,
. .
>.
ft 10,
138, tt 4,

fir
tt

>
tt
ft
tt
ft
ft
ti
'
W
n

-w
3
j;
numerators 1, 1, 1
denominator r 1
taken
(196)
5
(-!)*
6s
4
204, 459
459
10.9.8
1.2.3
(J + 1)*
w^ , in numerator
(163)
(* + y + s)
0)
*2.= 1
(.r5-4.r + 8) on loft side
(234)
(29)
(267)
LLi
/> + 2
(P-l)
z=l
-l
ff(r,-l)
H(r+1, n-l)

read
II
IV
II
II
II
II
If
II

II
II
II
ft

1
x\
1, a, a*
n-l.
taken m at a time.
(360).
6.
(-1).
3*.
84.
102, 306.
9.
7.8.9.10.
(? + l)(4J Notation of (96).

140,
II
182, n .
II <-r
191, ft 4,
II (164).

220, ,
it
11 2 (* + ? + *).
221, n 4,
II square of (1).
tt
,."237, n
11 *" - - 1.
ft
238, n 6,
II 0r*-4a; + 8)3.
239, n 11,
248, tt 4,

II (28).
267, n 4,
II (266).
274, n 8,
II 2|_U.
II
,, 276, tt 13,

II p+L
>
II
tt
ft 14,
,. 283, n 8,
>
II a = 4.
288, tt 7,
tt
II n + 1.
,. 289, n 4,
>' H(n, r-1).
M
290, tt 2,
11
11 tf(, r).
325, P 17,

II P.
1)
n i^/jij, last line but ono II
n m
II l"2-J *
n 333, ft s,
3528
361, ft 7,

11 10284.
481, tt 6,
M -3
II n-l.
applying Descartes' rule
Deli.
614, ii 4,
M o3
m 517, ti 3,
II r\
Transpose F and /.
644, *y 1,
661, tt 1,
H *i
11 B.
tt
tt ,
n a
II O K.
704,
) (11, 12)
II (9, 10, 1).
729,
II (940)
II (960).
Article 112 should bo as follows:
1
= 1 +2y/3
= 1 + 2y3 + y2 = (l+2,/3 + t^ni-4y3)
1 + 2^3-^/2 = (1 +2v^3)--2 ^ 11 + 4^/3
73

TABLE OF CONTENTS.

PART I.
SECTION I.MATHEMATICAL TABLES.
Introduction. Tiie C. G. S. System of Units
Notation and Definitions of Units...
Physical Constants and Formula) ...
Table I.English Measures and Equivalents in C. G. S. Units
II.Pressure of Aqueous Vapour at different temperatures
III.Wave lengths and Wave frequency for the principal'
lines of the Spectrum ...
TV.The Principal Metals Their Densities ; Coeffi
cients of Elasticity, Rigidity, and Tenacity ; Expan
sion by Heat ; Specific Heat ; Conductivity ; Rate
of conduction of Sound ; Electro- magnetic Specifio
Resistance
V.The Planets Their Dimensions, Masses, Densities,
and Elements of Orbits
VI.Powers and Logarithms of T and e
VII.Square and Cube Roots of the Integers 1 to 30
VIII. Common and Hyperbolic Logarithms of the
Prime numbers from 1 to 109...
IX.Faotor Table
Explanation of the Table
The Least Factors of all numbers from 1 to 90000...
X.Values of the Gamma-Function

SECTION II.ALGEBRA.
Factors
...
Newton's Rule for expanding a Binomial . . .
Multiplication and Division
Indices
Highest Common Factor

PfMTO

1
2
4
4

7
8
30

Ndi'of
Article

1
12
28
29
30

xn

CONTENTS.
No. ol
Article

Lowest Common Multiple

Evolution
Square Root and Cube Root
Useful Transformations
Quadratic Equations
Theory of Quadratic Expressions
Equations in one Unknown Quantity. Examples
Maxima and Minima by a Quadratic Equation
Simultaneous Equations and Examples ...
Ratio and Proportion
The 7* Theorem
Duplicate and Triplicate Ratios ...
Compound Ratios
Variation
Arithmetical Progrkssion ...
Geometrical Progression
Harmonical Progression
Permutations and Combinations
Surds
Simplification of i/a+ </b and va+ /&...
Simplification of Va+ \/b ...
Binomial Theorem ...
Multinomial Theorem
Logarithms ...
...
...
...
Exponential Theorem
Continued Fractions and Convergents ...
General Theory of same
To convert a Series into a Continued Fraction
A Continued Fraction with Recurring Quotients
Indeterminate Equations ...
To reduce a Quadratic Surd to a Continued Fraction
To form high Convergents rapidly
General Theory
Equations
Special cases in the Solution of Simultaneous Equations
Method by Indeterminate Multipliers
Miscellaneous Equations and Solutions ...
On Symmetrical Expressions
Imaginary Expressions
Method of Indeterminate Coefficients ...
Method of Proof by Induction
Partial Fractions. Four Cases ...
Convergency and Divergency of Series ...
General Theorem of <j> (x) ...

33
35
38
45
50
54
58
59
68
70
72
74
76
79
83
H7
94
108
121
124
125
137
142
149
1G0
167
182
186
188
195
197
199
211
213
214
219
223
232
233
235
239
243

CONTENTS.

Expansion of a Fraction
Recurring Series ...
...
...
...
...
...
The General Term
Case of Quadratic Factor with Imaginary Roots...
Lagrange's Rule
...
...
...
...
...
Summation of Series by the Method of Differences
Interpolation of a term
...
...
...
...
Direct Factorial Series ...
...
...
...
...
Inverse Factorial Series
Summation by Partial Fractions ...
...
...
...
Composite Factorial Series
Miscellaneous Series
Sums of the Powers of the Natural Numbers
...
Sum of a+(a+d)r+(a + 2d)t*+&o
Sum of nT-n (n-l)r+&c

Xlll

...
...
...
...
...
...

...

Polygonal Numbers
Figurate Numbers
Hypergeometrical Series
Proof that e" is incommensurable .. .
...
...
Interest
...
...
...
...
...
...
...
Annuities
...
...
...
...
...
...
...
Probabilities...
...
...
...
...
...
...
Inequalities
Arithmetic Mean > Geometric Mean
...
...
Arithmetic Mean of mth powers > to* power of A. M.
Scales of Notation
Theorem concerning Sum or Difference of Digits
Theory of Ncmrers
Highest Power of a Prime p contained in | m ...
Fermat's Theorem
~
Wilson's Theorem
Divisors of a Number
...
...
...
...
Sr divisible by 2n + l
SECTION III. THEORY OF EQUATIONS.
Factors of an Equation
To compute /(a) numerically
...
...
...
Discrimination of Roots
...
...
...
...
Descartes' Rule of Signs ...
...
...
...
...
The derived Functions of/ () ...
...
...
...
To remove an assigned term
...
...
...
To transform an equation ...
...
...
...
Equal Roots of an Equation
Practical Rule

No. of
Artic'o
248
251
257
258
263
264
267
268
270
272
274
276
279
285
287
289
291

...
...
...
...
...
...
...
...

...

...
...
...
...
...
...

295
296
302
309
330
332
334
342
347
349
365
369
371
374
380

400
403
409
416
424
428
430
432
445

XIV

CONTENTS.
No. of
Article.

Limits of ttie Roots

...
Newton's Method
...
Rolle'e Theorem
...
Newton's Mettiod of Divisors
Reciprocal Equations
Binomial Equations
Solution of "1 = 0 by
Cubic Equations
Cardan's Method
...
Trigonometrical Method
Biquadratic Equations
Descartes' Solution ...
Ferrari's Solution ...
Euler's Solution
Commensurable Roots

...
...
...
...

...
...
...
...

...
...
...
...

...
...
...
...

...
...
...
...

De Moivro's Theorem...

...

...
...

...
...

...
...

...
...

...
...

...
...

...
...

...
...

...
...

...
...

Incommensurable Roots
Sturm's Theorem
...
...
...
...
...
...
Fourier's Theorem ...
...
...
...
...
...
Lagrange's Method of Approximation
...
...
...
Newton's Method of Approximation
...
...
...
Fourier's Limitation to the same ...
...
...
...
Newton's Rule for the Limits of the Roots
...
...
Sylvester's Theorem
Horner's Method
...
...
...
...
...
...
Symmetrical Functions of the Roots of an Equation
Sums of the Powers of the Roots ...
...
...
...
Symmetrical Functions not Powers of the Roots...
...
The Equation whose Roots are the Squares of the
Differences of the Roots of a given Equation
...
Sum of the mlh Powers of the Roots of a Quadratic
Equation
...
...
...
...
...
...
Approximation to the Root of an Equation through the
Sums of the Powers of the Roots
...
...
...
Expansion of an Implicit Function of a
Determinants
Definitions ...
...
...
...
...
...
...
General Theory
...
...
...
...
...
...
To raise the Order of a Determinant
...
...
...
Analysis of a Determinant ...
...
...
...
...
Synthesis of a Determinant
...
...
...
...
Product of two Determinants of the nth Order ...
...
Symmetrical Determinants ..
...
...
Reciprocal Determinants ...
...
...
Partial and Complementary Determinants

...
...
...

...
...
...

448
452
454
459
4GG
472
480
483
484
489
492
496
499
502
506
518
525
527
528
530
532
533
534
538
541
545
548
551
554
556
564
568
569
570
574
575
576

CONTEXTS.

XV
No. of
Articli;.

Theorem of a Partial Reciprocal Determinant ...

577
Product of Differences of n Quantities
578
Product of Squares of Differences of same
579
Rational Algebraic Fraction expressed as a Determinant 581
Elimination
Solution of Linear Equations
582
Orthogonal Transformation
584
Theorem of the ?! 2lh Power of a Determinant
585
Bezout's Method of Elimination ...
586
Sylvester's Dialytic Method
587
Method by Symmetrical Functions
588
Elimination by Highest Common Factor
593

SECTION TV.PLANE TRIGONOMETRY.

Angular Measurement
Trigonometrical Ratios
Formulee involving one Angle
Formulae involving two Angles and Multiple Angles
Formulae involving three Angles ...
Ratios of 45, 60, 15, 18, &c
Properties of the Triangle
The s Formulae for ain^A, &c.
The Triangle and Circle ...
...
Solution of Triangles
Right-angled Triangles
Scalene Triangles.Three cases ...
Examples on the same
...
.;.
Quadrilateral in a Circle ...
Bisector of the Side of a Triangle...
Bisector of the Angle of a Triangle
Perpendicular on the Base of a Triangle ...
Regular Polygon and Circle
Subsidiary Angles
Limits of Ratios
De Moivre's Theorem
Expansion of cos nO, &c. in powers of sin 0 and cos 0
Expansion of sinO and cos0 in powers of 0
Expansion of cos" 0 and sin" 0 in cosines or sines
multiples of 0 ...
Expansion of cos n6 and sinnfl in powers of sin 0
Expansion of cos nO and sin nd in powers of cos 0
Expansion of cos8 in descending powers of cosO
Sin o + c sin (a +/3) + Ac., and similar series

600
606
613
627
674
690
700
704
709
718
720
859
733
738
742
744
746
749
753
756
758
764

of
772
775
779
780
783

Xvi

CONTENTS.
No. of
Article.

Gregory's Series for 0 in powers of tan d...

...
...
Formulae for tho calculation of * ...
...
..
...
Proof that jt is incommensurable ...
...
...
...
Sinas = n sin(a; + a). Series forse
Sum of sines or cosines of Angles in A. P.
...
...
Expansion of the sine and cosine in Factors
...
...
Sinn^ and cosnf expanded in Factors ...
...
...
Sin0 and cos0 in Factors involving 0 ...
...
...
e* 2cos0 + e~* expanded in Factors
...
...
...
De Moivre's Property of the Circle
...
...
...
Cotes's Properties ...
...
...
...
...
...
Additional Formula
Properties of a Right-angled Triangle ...
...
...
Properties of any Triangle...
...
...
...
...
Area of a Triangle ...
...
...
...
...
...
Relations between a Trianglo and tho Inscribed,
Escribed, and Circumscribed Circles
...
...
Other Relations between the Sides and Angles of a Triangle
Examples of the Solution of Triangles ...
...
...

791
792
795
790
800
807
808
815
817
819
821
823
832
835
838
841
850
859

SECTION V.SPHERICAL TRIGONOMETRY.

Introductory Theorems
Definitions ...
...
...
...
Polar Triangle
Right-angled Triangles
Napier's Rules
...
...
...
Oblique-angled Triangles.
Formula? for cosa and cos.4
...
The S Formula) for biji^A, sin|a, &c.
sin A
sin B
sin 0
:
=
= ;
...
...
sin a
sin ft
sine
cos b cos 0 = cot a sin 6 cot A sin G
Napier'B Formuhe ...
...
A.
Gauss's Formula? ...
...
...
Spherical Triangle and Circle
Inscribed and Escribed Circles
...
Circumscribed Circles
...
...
Spherical Areas
Spherical Excess
...
...
...
Area of Spherical Polygon ...
...
Cagnoli's Theorem ...
...
...
Lhuillier's Theorem
...
...

...

...

...

870
871

...

...

...

881

...
...

...
...

...
...

882
884
on*

...

...

...

BlM

...
...
...

...

...

...

...

895
&*$
897

...
...

...
...

...
...

898
900

...
...
...
...

...
...
...
...

...
...
...
...

902
903
904
905

CONTENTS.

XV11
No. at
Article.

Polyhedrons
The five Regular Solids
The Angle between Adjacent Faces
...
...
Radii of Inscribed and Circumscribed Spheres ...

...
...

906
907
909
910

SECTION VI.ELEMENTARY GEOMETRY.

MISCELLANEOUS PROPOSITIONS

Reflection of a point at a single surface ...

...
...
do.
do.
at successive surfaces
...
...
Relations between the sides of a triangle, the segments
of the base, and the line drawn from the vertex ...
Equilateral triangle ABC; PA*+PB1 + PC
Sum of squares of sides of a quadrilateral
...
...
Locus of a point whose distances from given lines or
points are in a given ratio ...
...
...
...
To divide a triangle in a given ratio
...
...
...
Sides of triangle in given ratio. Locus of vertex
...
Harmonic division of base
...
...
...
...
Triangle with Inscribed and Circumscribed circles
...
The Problems of the Tangencies
Tangents and chord of contact. /3y = us
To find any sub-multiple of a line
...
...
...
Triangle and three concurrent lines ; Three cases
...
Inscribed and Escribed circles ; s, s a, &o.
...
...
Nine-Point Circle
Construction or Triangles
...
...
...
...
...
Locus of a point from which the tangents to two circles
have a constant ratio .. .
...
...
...
...
Collinear and Concurrent Systems
Triangle of constant species circumscribed or inscribed
to a triangle ...
...
...
...
...
...
Radical Axis
Of two Circles
Of threo Circles
Inversion
Inversion of a point
...
...
...
...
...
do.
circle
do.
right line
Pole and Polab
Coaxal Circles
...
Centres and Axes of Similitude
Homologous and Anti-homologous points
...
...
do.
do.
chords
C

920
921
922
923
924
926
930
932
933
935
937
948
950
951
953
954
960
963
967
977
984
997
1000
1009
1012
1016
1021
1037
1038

xviii

CONTENTS.

No. of
Article.
Constant product of anti-similitude
..,
...
... 1043
Circle of similitude
...
...
...
...
... 1045
Axes of similitude of three circles
...
...
... 1046
Gergonne's Theorem
...
...
...
...
... 1049
Anharmonic Ratio and Pencil ...
...
...
...
... 1052
Homographic Systems of Points
1058
Involution
1066
Projection ...
...
...
...
...
...
...
... 1075
On Perspective Drawing ...
...
...
...
... 1083
Orthogonal Projection
...
...
...
...
... 1087
Projections of the Sphere .. .
...
...
...
... 1090
Additional Theorems
Squares of distances of P from equidistant points on a
circle
1094
Squares of perpendiculars on radii, &c. ...
...
... 1095
Polygon with inscribed and circumscribed circles. Sum
of perpendiculars on sides, &c...
... 1099

*
SECTION VII.GEOMETRICAL CONICS.
Sections of the Cone
Defining property of Conic PS = ePM ...
...
Fundamental Equation ...
...
...
...
Projection from Circle and Rectangular Hyperbola
Joint Properties of the Ellipse and Hyperbola
Definitions ...
...
...
...
...
...
CS : CA:CX
PSPS,= AA'
CS* = AC T BC
SZ bisects L QSB
If PZ be a tangent PSZ is a right angle
Tangent makes equal angles with focal distances
Tangents of focal chord meet in directrix
...
CN.CT =AC
OS :PS=e
KG : NC = BC : AC
Auxiliary Circle ...
...
...
...
...
PN : QN = BC : AO
PA3 : AN. NA' = BC : AC
Cn.Ct = BC
SY.S'T = BC
PE = AG
To draw two tangents
...
...
...
...
Tangents subtend equal angles at the focus
...

...
...
...
...

1151
1156
1158

11G0
1162
1163
... 1164
1166
1167
... 1168
... 1169
1170
1171
1172
... 1173
1174
1 1 76
1177
1178
1179
... 1180
... 1181

CONTENTS.

XIX
No. of
Article.

Asymptotic Properties of tiie Hyperbola

RN'-PN* = BG'
PR.Pr = BG'
GE = AG
FT) is parallel to the Asymptote ...
...
...
...
QR= qr
Th = PI and QV=qV
QR.Qr= PL1 = RV'-QV1
4PH.PK=CS'
Joint Properties of Ellipse and Hyperbola resumed. Con
jugate Diameters
QV : PV. FF = CD1 : CP1
PF.CD=AC.BG
PF.PG = BG' and PF.Pff = AC*
PO.Pa = CI?
Diameter bisects parallel chords ...
...
...
...
Supplemental chords
...
...
...
...
...
Diameters are mutually conjugate
...
...
...
CV.GT=GPt
GN=dB, CR=pN
CN* VR* = AC\ DR* PN* = BC*
CP1 GTS = AC PC
PS. PS = CD*
OQ.Oq:OQ'.Oq' = CB,:CD'>
SB: QL = e
Director Circle
Properties of Parabola deduced from the Ellipse
...
The Parabola
Denning property PS = PM
Latus Rectum = 4AS
If PZ be a tangent, PSZ is a right angle
Tangent bisects I SPM and SZM
ST=SP = SO
Tangents of a focal chord intersect at right angles in
directrix ...
...
...
...
...
...

AN = AT
NG = 2AS
PN*=AS.AN
SA :SY:SP
To draw two tangents
...
...
...
...

SQ: SO: SQ'

l OSQ = OSQ' and QOQ' = -J QSQ'
Diameters
The diameter bisects parallel chords
...
...
...

1183
1184
1186
1187
1188
1189
1191
1192

1193
1194
1195
1197
1198
1201
1202
1203
1205
1207
1211
1213
1214
1216
1217
1219
1220
1222
1223
1224
1225
1226
1227
1228
1229
1231
1232
1233
1234
1235
1238

XX

CONTENTS.

QV* =4J>S.rV
OQ.Oq : UQ' : Oq = PS : PS
Parabola two-thirds of circumscribing parallelogram ...
Methods of Drawing a Conic
To find the axes and centre
...
...
...
...
To construct a conic from the conjugate diameters
...
Circle of Curvature
'.
Chord of curvature = QV* + PV ult
bemi-chords of curvature, ,

...

In Parabola, Focal chord of curvature = 4SP ...

do.
Radius of curvature = 2SP -f- ST
Common chords of a circle and conies are equally
clined to the axis
...
...
...
...
To find the centre of curvature ...
...
...
Miscellaneous Theorems

No. of
Article.
1239
1242
1244
1245
1252
1253
1254
1258

...

125iJ

...
...
in
...
...

1260
1261
1263
1265
1267

INDEX TO PROPOSITIONS OF EUCLID

REFERRED TO IN THIS WORK.

Tho references to Euclid are made in Roman and Arabic numerals ; e.g. (VI. 19).

BOOK
I.
I.
I.
I.
I.
I.
I.
I.
I.
I.

I.
I.
I.
I.

I.

4.Triangles are equal and similar if two sides and tho included
angle of each are equal each to each.
5.The angles at the base of an isosceles triangle are equal.
6. The converse of 5.
8.Triangles are equal and similar if the three sides of each are
equal each to each.
16. The exterior angle of a triangle is greater than the interior
and opposite.
20.Two sides of a triangle are greater than the third.
26.Triangles are equal and similar if two angles and one corres
ponding side of each aro equal each to each.
27. Two straight lines are parallel if they make equal alternate
angles with a third line.
29.The converse of 27.
32. The exterior angle of a triangle is equal to the two interior
and opposite ; and the three angles of a triangle are equal
to two right angles.
Cou. 1.The interior angles of a polygon of n sides
= ( 2) T.
Cor. 2.The exterior angles = 2*-.
35 to 38. Parallelograms or triangles upon the same or equal
bases and between the same parallels are equal.
43. The complements of the parallelograms about the diameter
of a parallelogram are equal.
47. The square on the hypotenuse of a right-angled triangle is
equal to the squares on the other sides.
48. The converse of 47.

XX11

INDEX TO PROPOSITIONS OF EUCLID

BOOK II.
II. 4.If a, b are the two parts of a right line, (a + 6)' = a, + 2ab + b1.

II.

II.

II.

II.

If a right lino be bisected, and also divided, internally or

externally, into two nncqnal segments, then
5 and 0. The rectangle of the unequal segments is equal to tho
difference of tho squares on half" the line, and on the lino
between the points of section; or (a + b) (a b) = a*b*.
9 and 10. The squares on the same unequal segments are together
double the squares on the other parts ; or
(a+&)+ (-&)* = 2a, + 26,.
11. To divide a right line into two parts so that tho rectangle
of the whole line and one part may be equal to tho square on
the other part.
12 and 13.The square on the base of a trianglo is equal to the
sum of the squares on tho two sides plus or minus (as the
vertical angle is obtuse or acute), twice the rectangle under
either of those sides, and the projection of the other upon it ;
or a' = Z' + cs-2ccos4 (702).

BOOK III.
III.
III.
III.
III.
III.
III.
III.
III.

III.

3.If a diameter of a circle bisects a chord, it is perpendicular to

it : and conversely.
20.The angle at the centre of a circle is twice the angle at tho
circumference on the same arc.
21.Angles in the same segment of a circle are equal.
22. The opposite angles of a quadrilateral inscribed in a circle
are together equal to two right angles.
31. The angle in a semicircle is a right angle.
32. The angle between a tangent and a chord from the point of
contact is equal to the angle in the altemato segment.
33 and 34. To describe or to cut off a segment of a circle which
shall contain a given angle.
35 and 30.The rectangle of tho segments of any chord of a
circle drawn through an internal or external point is equal
to tho square of the semi-chord perpendicular to the
diameter through the internal point, or to the square of tho
tangent from the external point.
37. The converse of 30. If the rectangle be equal to the square,
the line which meets the circle touches it.

REFERRED TO IN TIIIS WORK.

XX1U

BOOK IV.
TV. 2.To inscribe a triangle of given form in a circle.
IV. 3.To describe the same about a circle.
IV. 4.To inscribe a circle in a triangle.
IV. 5.To describe a circle about a triangle.
IV. 10.To construct two-fifths of a right angle.
IV. 11.To construct a regular pentagon.

BOOK VI.
VI. 1.Triangles and parallelograms of the same altitude are
proportional to their bases.
VI. 2.A right line parallel to the side of a triangle cuts the other
sides proportionally ; and conversely.
VI. 3 and A.The bisector of the interior or exterior vertical angle of
a triangle divides the base into segments proportional to
the sides.
VI. 4.Equiangular triangles have their sides proportional homologously.
VI. 5. The converse of 4.
VI. 6.Two triangles are equiangular if they have two angles equal,
and the sides about them proportional.
VI. 7. Two triangles are equiangular if they have two angles equal
and tho sides about two other angles proportional, provided
that the third angles are both greater than, both less than,
or both equal to a right angle.
VI. 8.A right-angled triangle is divided by tho perpendicular from
the right angle upon tho hypotenuse into triangles similar
to itself.
VI. 14 and 15.Equal parallelograms, or triangles which have two
angles equal, have the sides about those angles reciprocally
proportional ; and conversely, if tho sides are in this pro
portion, the figures are equal.
VI. 19. Similar triangles are in the duplicate ratio of their homo
logous sides.
VI. 20. Likewise similar polygons.
VI. 23. Equiangular parallelograms are in the ratio compounded of
the ratios of their sides.
VI. B. The rectangle of the sides of a triangle is equal to the square
of the bisector of the vertical angle pZ?w the rectangle of
the segments of the base.

XXIV

INDEX TO PROPOSITIONS OP EUCLID.

VI. C.The rectangle of the sides of a triangle is equal to the rect

angle under the perpendicular from the vertex on the
base and the diameter of the circumscribing circle.
VI. D.Ptolemy's Theorem. The rectangle of the diagonals of a
quadrilateral inscribed in a circle is equal to both the
rectangles under the opposite sides.

BOOK
XI.
XI.

XI.
XI.
XI.
XI.

XI.

4.A right line perpendicular to two others at their point of

intersection is perpendicular to their plane.
5.The converse of 4. If the first line is also perpendicular to a
fourth at the same point, that fourth line and the other
two are in the same plane.
6.Right lines perpendicular to the same plane are parallel.
8.If one of two parallel lines is perpendicular to a plane, the
other is also.
20.Any two of three plane angles containing a solid angle are
greater than the third.
21.The plane angles of any solid angle are together less than
four right angles.

TABLE OF CONTENTS.

PABT II.
SECTION VIII.DIFFERENTIAL CALCULUS.

No. of
Article.

Introduction
1400
Successive differentiation
...
...
...
...
1405
Infinitesimals. Differentials ...
...
...
...
1407
Differentiation.
Methods
1411-21
Successive Differentiation
Leibnitz's theorem
...
...
...
...
...
1460
Derivatives of the nth order (see Index)
...
... 1461-71
Partial Differentiation
1480
Theort of Operations
1483
Distributive, Commutative, and Index laws...
...
1488
Expansion of Explicit Functions
Taylor's and Maclaurin's theorems ...
...
...1500,1507
Symbolic forms of the same ...
...
...
...
1520-3
f(x + h, y + k), 4c.
...
1512-4
Methods of expansion by indeterminate coefficients.
Four rules
1527-34
Method by Maclaurin's theorem
...
1524
Arbogast's method of expanding ip (z)
...
...
1536
Bernoulli's numbers ...
...
...
...
...
1539
Expansions of f (x + h) 0 (a>). Stirling and Boole
1546-7
Expansions of Implicit Functions
Lagrange's, Laplace's, and Burmann's theorems, 1552, 1556-63
1
1555
Cayley's series for

1572
1580
1600
1604

Abel's series for <j>(x + a)

Indeterminate Forms
Jacobians
...
Modulus of transformation

XXVI

CONTENTS.
No. of
Article.

QlANTlCS

Euler's theorem
Eliminant, Discriminant, Invariant, Covariant, Hessian
Theorems concerning discriminants ...
Notation A = 6c/*, &c.
Invariants
Cogredients and Emanents
Implicit Functions
One independent variable
Two independent variables
n independent variables
Change of the Independent Variable
Linear transformation ...
Orthogonal transformation
Contragredient and Contravariant
Notation zx =p, &c, q,r, s, t ...
Maxima and Minima
One independent variable
Two independent variables
Three or more in dependent variables ...
Discriminating cubic ...
Method of undetermined multipliers ...
Continuous maxima and minima

1620
1621
1626-30
1635-45
1642
1648-52
1653-5
1700
1725
1737
1760
1794
1799
1813
1815
1830
1841
1852
1849
1862
1866

SECTION IX.INTEGRAL CALCULUS.

Introduction
1900
Multiple Integrals
...
...
...
...
...
1905
Methods op Integration
By Substitution, Parts, Division, Rationalization,
Partial fractions, Infinite series ...
...
... 1908-19
Standard Integrals
...
1921
Various Indefinite Integrals
Circular functions
...
...
...
...
...
1954
Exponential and logarithmic functions
...
...
1998
Algebraic functions
...
...
...
...
...
2007
Integration by rationalization...
...
...
...
2110
Integrals reducible to Elliptic integrals
...
... 2121-47
Elliptic integrals approximated to ...
...
...
2127
Successive Integration
...
2148
Hyperbolic Functions cosh*, sinhas, tanha
2180
Inverse relations
...
...
...
...
...
2210
Geometrical meaning of tanh S
...
...
...
2213
Logarithm of an imaginary quantity .. .
...
...
2214

CONTENTS.

XXV11
No. of
Article.

Definite Integrals
Summation of series ...
...
...
...
...
2230
Theorems respecting Limits of Integration
2233
Methods of evaluating Definite Integrals (Eight rules) ...
2245
Differentiation under the sign of Integration ...
...
2253
Integration by this method ...
...
...
...
2258
Change in the order of integration ...
...
...
2261
Approximate Integration by Bernoulli's Series ...
...
2262
The Integrals B(l, m) and T(n)
2280
logTfl+n) in a converging series ...
...
...
2294
Numerical calculation of T (a;)
...
...
...
2317
Integration of Algebraic Forms
2341
Integration of Logarithmic and Exponential Forms
...
2391
Integration of Circular Forms
...
...
2451
Integration of Circular Logarithmic and Exponential Forms
2571
Miscellaneous Theorems
Frnllani's, Poisson's, Abel's, Kummer's, and Cauchy's
formulas ...
2700-13
Finite Variation of a Parameter ...
2714
Fourier's formula
...
...
...
...
...
2726
The Function i//(
2743
Summation of series by the function ^() ...
...
2757
4* (x) as a definite integral independent of ^(1)
...
2766
Numerical Calculation of log T(x)
2771
Change of the Variables in a Definite Multiple Integral
2774
Multiple Integrals
Expansions of Functions in Converging Series
Derivatives of the nth order ...
...
...
...
2852
Miscellaneous expansions
...
...
...
...
2911
Legendre's function Xn
...
...
...
...
2936
Expansion of Functions in Trigonometrical Series
...
2955
Approximate Integration
2991
Methods by Simpson, Cotes, and Gauss
...
...
2992-7

SECTION X.CALCULUS OF VARIATIONS.

Functions of one Independent Variable
Particular cases. . .
Other exceptional cases
Functions of two Dependent Variables
Relative maxima and minima...
Geometrical applications

3028
3033
3045
3051
3069
3070

XXV111

CONTENTS.
No. of
Article.

Functions of two Independent Variables

Geometrical applications
...
...
...
...
Appendix
On the general object of the Calculus of Variations...
Successive variation ...
...
...
...
...
Immediate integrability
...
...
...
...

3075
3078
3084
3087
3090

SECTION XLDIFFERENTIAL EQUATIONS.

Generation of Differential Equations
Definitions and Rules
Singular Solutions
First Order Linear Equations
Integrating factor for Mdx + Ndy = 0
...
...
Riccati's Equation
...
...
...
...
...
First Order Non-linear Equations
Solution by factors
...
...
...
...
...
Solntion by differentiation
...
...
...
...
Higher Order Linear Equations
Linear Equations with Constant Coefficients
...
Higher Order Non-linear Equations
Depression of Order by Unity...
...
...
...
Exact Differential Equations
Miscellaneous Methods
Approximate solution of Differential Equations by
Taylor's theorem
Singular Solutions of Higher Order Equations
Equations with more than two Variables
Simultaneous Equations with one Independent Variable...
Partial Differential Equations
Linear first order P. D. Equations ...
...
...
Non-linear first order P. D. Equations
...
...
Non-linear first order P. D. Equations with more
than two independent variables .. .
...
...
Second Order P. D. Equations
Law of Reciprocity
Symbolic Methods
...
Solution of Linear Differential Equations by Series
...
Solution by Definite Integrals
P. D. Equations with more than two Independent Variables
Differential Resolvents of Algebraic Equations
...

3150
3158
3168
3184
3192
3214
3221
3222
3236
3237
3238
3251
3262
3270
3276
3289
3301
3320
3340
3380
3381
3399
3409
3420
3446
3470
3604
3617
3629
3631

CONTENTS.

XXIX
No. of
Article.

SECTION XII. CALCULUS OF FINITE

DIFFERENCES.
Formula foe First and nth Differences
Expansion by factorials
Generating functions ...
The operations E, A, and dx ...
Herschel's theorem
A theorem conjugate to Maclaurin's ...
Interpolation ...
Lagrange's interpolation formula
Mechanical Quadrature
Cotes's and Gauss's formula ...
Laplace's formula
Summation of Series
Approximate Summation

3706
3730
3732
3735
3757
3759
3762
3768
3772
3777
3778
3781
3820

SECTION XIII.PLANE COORDINATE GEOM[ETRY.

Systems of Coordinates
Cartesian, Polar, Trilinear, Areal, Tangential, and
Intercept Coordinates

4001-28

ANALYTICAL CONICS IN CARTESIAN COORDINATES.

Lengths and Areas
Transformation of Coordinates
The Right Line.
Equations of two or more right lines...
General Methods
Poles and Polars
The Circle
Co-axal circles ...
The Parabola
The Ellipse and Hyperbola...
Right line and ellipse ...
Polar equations of the conic ...
Conjugate diameters ...
Determination of various angles
The Hyperbola referred to its Asymptotes
The rectangular hyperbola

4032
4048
4052
4110
4114
4124
4136
4161
4200
4250
4310
4336
4346
4375
4387
4392

XXX

CONTENTS.
No. of
Article.

The General Equation

The ellipse and hyperbola
...
...
...
Invariants of the conic
...
...
...
The parabola
Method without transformation of the axes ...
Rules for the analysis of the general equation
Bight line and conic with the general equation
Intercept equation of a conic ...
...
...
Similar Conics
Circle of Curvature
Contact of Conics
...
...
...
...
Confocal Conics

...
...
...
...
...
...

...

4400
4402
441 7
4430
4445
4464
4487
4498
4522
4527
4550

ANALYTICAL CONICS IN TRILINEAR COORDINATES.

The Right Line
...
...
...
...
...
...
Equations of particular lines and coordinate ratios
of particular points in the trigon
...
...
Anharmonic Ratio
...
...
...
...
...
...
The complete quadrilateral
...
...
...
...
The General Equation of a Conic
Director- Circle ...
...
...
...
...
...
Particular Conics
Conic circumscribing the trigon
...
...
...
Inscribed conic of the trigon ...
...
...
...
Inscribed circle of the trigon ...
...
...
...
General equation of the circle...
...
...
...
Nine- point circle
...
...
...
...
...
Triplicate-ratio circle ...
...
...
...
...
Seven-point circle
...
...
...
...
...
Conic and Self-conjugate Triangle
On lines passing through imaginary points ...
...
Carnot's, Pascal's, and Brianchon's Theorems
The Conic referred to two Tangents and the Chord of
Contact
Related conics ...
...
...
...
...
Anharmonic Pencils of Conics
Construction of Conics
...
Newton's method of generating a conic
...
...
Maclaurin's method of generating a conic ...
...
The Method of Reciprocal Polars
Tangential Coordinates
Abridged notation
...
...
...
...
...

4601
4628
4648
4652
4656
4693
4697
4724
4739
4747
4751
4754
47546
4754c
4755
4761
4778-83

4803
4809
4822
4829
4830
4844
4870
4907

CONTENTS.

XXXI
No.ot
Article.

On the Intersection of two Conics

Geometrical^meaning of */( 1)
...
The Method of Projection
Invariants and Covariants
To find the foci of the general conic ...

...

...

...

...

4916
4921
4936
5008

...

...

5100
5134
5160
5167
5172

...
...

...
...

...
...
...
...
...
...

...
...
...
...
...
...

...
...
...

...
...
...

...

...

THEORY OF PLANE CURVES.

Tangent and Normal
Radics of Curvature and Evolute
Inverse Problem and Intrinsic Equation
Asymptotes
Asymptotic curves
...
...
...
Singularities of Curves
Concavity and Convexity
...
...
Points of inflexion, multiple points, &c.
Contact of Curves
Envelopes
Integrals of Curves and Areas
Inverse Curves
Pedal Curves
Roulettes
Area, length, and radins of curvature...
The envelope of a carried curve
...
Instantaneous centre ...
...
...
Holditch's theorem
...
...
...
Trajectories
...
...
...
...
Curves of pursuit
...
...
...
Caustics
Quetelet's theorem
Transcendental and other Curves
The cycloid
The companion to the cycloid...
...
Prolate and curtate cycloids ...
...
Epi trochoids and hypotrochoids
...
Epicycloids and hypocycloids...
...
The Catenary
TheTractrix
The Syntractrix
The Logarithmic Curve
...
...
The Equiangular Spiral
...
...
The Spiral of Archimedes
The Hyperbolic or Reciprocal Spiral .. .

...
...
...

...
...

5174
5175-87
5188
5192
5196
5212
5220
5229
5230-5
5239
5243
5244
5246
5247
5248
5249
5250
5258
5260
5262
5266
5273
5279
5282
5284
5288
5296
5302

XXX11

CONTENTS.
No. of
Article.

The Involute of the Circle

TheCissoid
The Cassinian or Oval of Cassini
...
Tli o Lenin iscate
...
...
...
The Conchoid
TheLimacon
The Versiera (or Witch of Agnesi)
The Quadratrix
The Cartesian Oval
The semi- cubical parabola
...
...
The folinm of Descartes
...
...
Linkages and Linkwork
Kempe's five-bar linkage. Eight cases
Reversor, Multiplicator, and Translator
Peancellier's linkage ...
...
...
The six-bar invertor ...
...
...
The eight-bar double invertor
...
The Quadrnplane or Versor Invertor

...
...

...
...
...
...
...
...
...

5306
5309
...
5313
...
5317
...
5320
5327
5335
5338
5341
...
5359
...
5360
5400
...5401-5417
...
5407
...
5410
...
5419
5420
...
5422

The Pentograph or Proportionator ...

...
...
The Isoklinostat or Angle-divider ...
...
...
A linkage for drawing an Ellipse
...'
...'" ...
A linkage for drawing a Limacon, and also a bicircular quartic
...
...
...
...
...
A linkage for solving a cnbic equation
...
...
On three-bar motion in a plane '
...
...
...
The Mechanical Integrator ...
...
...
...
The Planimeter

5423
5425
5426
5427
5429
5430
5450
5452

SECTION XIV.SOLID COORDINATE GEOMETRY.

Systems op Coordinates
The Riqht Line
The Plane
Transformation of Coordinates
...
The Sphere
...
...
The Radical Plane
...
...
Poles of similitude
...
...
Cylindrical and Conical Sdrfaces
Circular Sections
...
...
Ellipsoid, Hypbrboloid, and Paraboloid

...

...

...

...

5501
5507
5545
5574
5582
5585
...
5587
5590
...
5596
5599-5621

CONTENTS.

XXX111
No. of
Article.

Central Quadeic Surface

Tangent and diametral planes...
...
...
...
5626
Eccentric values of the coordinates ...
...
...
5638
Confocal Quadrics
5656
Reciprocal and Enveloping Cones ...
...
...
5664
The General Equation of a Quadric
5673
Reciprocal Polars
5704
Theory of Tortuous Curves
5721
The Helix
5756
General Theory of Surfaces
General equation of a surface .. .
...
...
...
5780
Tangent line and cone at a singular point ...
...
5783
The Indicatrix Conic
5795
Enler's and Mennier's theorems
...
...
5806-9
Curvature of a surface...
...
...
...
...
5826
Osculating plane of a line of curvature
...
...
5835
Geodesics
5837-48
Invariants
5856
Integrals for Volumes and Surfaces
...
5871
Guldin's rules
5879
Centre of Mass
5884
Moments and Products of Inertia ...
5903
Momental ellipsoids ...
...
...
...
... 5925-40
Al omental ellipse
...
...
...
...
...
5953
Integrals for moments of inertia
...
...
...
5978
Perimeters, Areas, Volumes, Centres of Mass, and Moments
of Inertia of various Figures
Rectangular lamina and Right Solid...
...
...
6015
The Circle
6019
The Right Cone
6043
Frustum of Cylinder ...
...
...
...
...
6048
The Sphere
6050
The Parabola
6067
The Ellipse
6083
Fagnani's, Griffith's, and Lambert's theorems
...6088-6114
The Hyperbola
6115
The Paraboloid
6126
The Ellipsoid
6142
Prolate and Oblate Spheroids
6152-65

PREFACE TO PART II.

Apologies for the non-completion of this volume at an

earlier period are due to friends and enquirers. The labour
involved in its production, and the pressure of other duties,
must form the author's excuse.
In the compilation of Sections VIII. to XIV., the
following works have been made use of :
Treatises on the Differential and Integral Calculus, by Bertrand,
Hymer, Todhunter, Williamson, and Gregory's Examples
on thesamesubjects; Salmon's Lessons on Higher Algebra.
Treatises on the Calculus of Variations, by Jellett and Tod
hunter ; Boole's Differential Equations and Supplement ;
Carmichtiel's Calculus of Operations ; Boole's Calculus of
Finite Differences, edited by Moulton.
Salmon's Conic Sections ; Ferrers's Trilinear Coordinates ;
Kempe on Linkages (Proc. of Boy. Soc, Vol. 23) j Frost
and Wolstenholme's Solid Geometry ; Salmon's Geometry
of Three Dimensions.
Wolstenholme's Problems.

The Index which concludes the work, and which, it is

hoped, will supply a felt want, deals with 890 volumes of
32 serial publications : of these publications, thirteen belong
to Great Britain, one to New South Wales, two to America,
four to France, five to Germany, three to Italy, two to
Russia, and two to Sweden.
As the volumes only date from the year 1800, the

xxxvi

PREFACE.

important contributions of Euler to the " Transactions of

the

St. Petersburg Academy," in

excluded.

the last century, are

It was, however, unnecessary to include them,

because a very complete classified index to Buler's papers,

as well as to those of David Bernoulli, Fuss, and others in
the same Transactions, already exists.
The titles of this Index, and of the works of Euler
therein referred to, are here appended, for the convenience
of those who may wish to refer to the volumes.
Tableau general des publications de l'Academie Imperiale de
St. Petersbourg depuis sa fondation. 1872. [B.M.C.:* R.R.
2050, e.]
I. Commentarii AcademiaB Scientiarum Iraperialis Petropolitanae.
1726-1746; 14 vols. [B.M.C: 431,/.]
II. Novi Commentarii A. S. I. P.
1747-75, 1750-77 ; 21 vols.
[B.M.C: 431,/. 15-17, .1-16, J.1,2.]
III. Acta A. S. I. P. 1778-86; 12 vols. [B. M. C : 431, A.3-8; or
T.G. 8, a. 11.]
IV. Nova Acta A. S. I. P. 1787-1806; 15 vols. [B.M.C: 431,
/(.9-15, i. 1-8; or T.G. 8, a. 23.]
V. Leonhardi Euler Opera minora collecta, vel Commentationes Arithmeticae collectae ; 2 vols. 1849. [B. M. C : 8534, ee.]
VI. Opera posthuma uiathematica et physica ; 2 vols. 1862. [B.M.C:
8534,/.]
VII. Opuscula analytica ; 1783-5; 2 vols. [B.M.C: b0,i. 15.]
Analysis infinitorum. [B. M. C : 529,6.11.]
G. S. C.
Endsleigh Gardens,
London, N.W., 1886.

* British Museum Catalogue.

i
EllEATA CONTINUED FROM PAGE x.
(Correction* xehich are important are marked with an asterisk.)
Page
*

Art.
11
II
If
ri
pi

u
VI

M
* 11
M
If
II
I
II
II
If
II
II
>*
II
II
)
If
M
II
II
II
M
*
1
II
II

II

1,
8,
123,
259,
276,
291,
292,
322,
361,
469,
470,
489,
565,
S93,
604,
713,
897,
922 ii.,
949,
1076,
1158,
1178,
1241,
1413,
1491,
1849,
1903,
1925,
1954-6,
2030-2,
2036,
2140,
2136,
2354,
2392,
2466,
3237,
3761,
4678,
4680,
4692,
4903,
5154,
5330,

Line 7,
read weight.
for volume
11 gramme-million
n gramme-six.
ii 10,
ii 5,
M 1-4071499
ii 4971499.
ii 6,
II 6679358
M 1-1447299.
II 2*/15.
ii 2.
II 2^5
ii 2,
II a3 + 02
II (a3+ 0s)".
ii 6,
II 3s + -l
II
ii 1,
II S
II Tii 3&4, II a
II a.
ii 9,
II 45 and 13
11 35 and 10.
ii 7,
II 3528
II 8684.
ii 3&9, II -6
II -16.
ii 1,
II *m
II x .
ii 9,
II -12.
3
a
number
of
rows
two columns.
14,
II
If
it
ii 11&12 t II Ban&B
>l if, and ik.
II one-sixtieth
II one-ninetieth.
ii 2,
,ii 2,
II ii.
II III.
ii 6,
II cosic
II sin Jc.
1) *2 + 2<~
II 2*3 + A
ii last, II D
II C.
dele " The projections . . . are parallel."
ii 2,
ii 1217.
ii last, II 1201
ii PS1.
ii 4,
II PS
ii 1.
ii conjugate.
*J parallel
ii + dttdv.
ii 8,
19 dudv
ii =
II ii I,
ii II +
i 1,
footnote,
ii /(*)
supply dx.
supply x.
erroneous, because I in (1427) is necessarily an
ii a.
II ax
ii 1,
applies to the whole denominator.
ii 1,
ii 2293.
ii last, II 2294
M +
II
II * -1.
II X*
*
II 2 "
II L2
II (-!)
II (-l)x
M 1,
supply w,.
dele 2 in the second term.
supply the factor 4 on the left.
dele 2 in the second term.
II 6,
supply the factor 4 on the left.
II
If 6156.
II 4,
II 3166
II b.
II 2,
II m
and refer to Fig. 129, Art. 5332 on the cardioid is wanting.

MATHEMATICAL TABLES.
INTRODUCTION.
The Centimetre-Gramme-Second system of units.
Notation.The decimal measures of length are the kilo
metre, hectometre, decametre, metre, decimetre, centimetre,
millimetre. The same prefixes are used with the litre and
gramme for measures of capacity and volume: >*>4i>f*~
Also, 107 metres is denominated a metre-seven; 10~7 metres,
a seventh-metre ; 1015 grammes, a gramme-fifteen ; and so on.
A gramme-HaillieH is also called a mega gramme ; and a
millionth-gramme, a microgramme ; and similarly with other
measures.
Definitions.The C. Gr. S. system of units refers all phy
sical measurements to the Centimetre (cm.), the Gramme (gm.),
and the Second (sec.) as the units of length, mass, and time.
The quadrant of a meridian is approximately a metreseven. More exactly, one metre = 3-2808694 feet = 39-3704-32
inches.
The Gramme is the Unit of mass, and the weight of a
gramme is the Unit oficeight, being approximately the weight
of a cubic centimetre of water; more exactly, 1 gm. =
15-432349 grs.
The Litre is a cubic decimetre : but one cubic centimetre
is the 0. G. S. Unit of volume.
1 litre = -035317 cubic feet = -2200967 gallons.
The Dyne (dn.) is the Unit offorce, and is the force which,
in one second generates in a gramme of matter a velocity of
one centimetre per second.
The Erg is the Unit of work and energy, and is the work
done by a dyne in the distance of one centimetre.
The absolute Unit of atmospheric pressure is one megadyne
per square centimetre = 74-964 cm., or 29-514 in. of mercurial
column at 0 at London, where g = 981-17 dynes.
Elasticity of Volume = k, is the pressure per unit area
upon a body divided by the cubic dilatation.

e.*sclJr

MATHEMATICAL TABLES.

Rigidity = n, is the shearing stress divided by the angle

of the shear.
Young's Modulus = M, is the longitudinal stress divided
by the elongation produced, = 9nk -=- (3/c+to).
Tenacity is the tensile strength of the substance in dynes
per square centimetre.
The Gramme-degree is the Unit of heat, and is the amount
of heat required to raise by 1 C. the temperature of 1 gramme
of water at or near 0.
Thermal capacity of a body is the increment of heat
divided by the increment of temperature. When the incre
ments are small, this is the thermal capacity at the given
temperature.
Specific heat is the thermal capacity of unit mass of the
body at the given temperature.
The Electrostatic unit is the quantity of electricity which
repels an equal quantity at the distance of 1 centimetre with
the force of 1 dyne.
The Electromagnetic unit of quantity = 3X 1010 electro
static units approximately.
The Unit of potential is the potential of unit quantity at
unit distance.
The Ohm is the common electromagnetic unit of resistance,
and is approximately = ] 09 C. G. S. units.
The Volt is the unit of electromotive force, and is = 108
C. G. S. units of potential.
The Weber is the unit of current, being the current due to
an electromotive force of 1 Volt, with a resistance of 1 Ohm.
It is = -nj C. G. S. unit.
Resistance of aWire Specific resistanceX Length--- Section.
Physical constants and Formula?.
In the latitude of London, g = 32-19084 feet per second.
= 981"17 centimetres per second.
In latitudo X, at a height h above the sea level,
g = (980-6056 2-5028 cos 2\ '000003 h) centimetres per second.
Seconds pendulum = (09-3562 -2536 cos 2\ -0000003 h) centimetres.
THE EARTH.- Semi-polar axis, 20,854895 feet* = (5-35411 x 108centims.
Mean semi-equatorial diameter,
20,926202 * = 6 37824 x 108
Quadrant of meridian,
39 377786 x 107 inches* = 1-000196 x 107metres.
Volume, lp08279 cubic centimetre-nines.
Mass (with a density 5|) = Six gramme- twenty- sevens nearly.
* These dimensions are taken frjin Clarke's " Geodesy," 1880.

MATHEMATICAL TABLES.

Velocity in orbit = 2933000 centims per sec. Obliquity, 23 27 16".*

Angular velocity of rotation = 1 -H 13713.
Precession, 50"-20.* Progression of Apse, H"-25. Eccentricity, e = '01679.
Centrifugal force of rotation at the equator, 3'3912 dynes per gramme.
Force of attraction upon moon, "2701.
Force of sun's attraction, '5839.
Ratio of g to centrifugal force of rotation, g '. rJ' = 289.
Sun's horizontal parallax, 8"7 to 9'.*
Aberration, 20"11 to 20"' 79.*
Semi-diameter at earth's mean distance, 16' 1""82.*
Approximate mean distance, 92,000000 miles, or 1'48 centimetre-thirteens.t
Tropical year,
305 242216 days, or 31,556927 seconds.
Sidereal year,
365'256374
,,
31,558150

Anomalistic year, 365"259544 days.

Sidereal day, 86164 seconds.
THE MOON.Mass = Earth's mass x 011304 = 6 98 10* grammes.
Horizontal parallax. From 53' 56" to 61' 24".*
Sidereal revolution, 27d. 7h. 43m. 1 l'46s. Lunar month, 29d. 12h. 44m. 2'87s.
Greatest distance from the earth, 251700 miles, or 4'05 centimetre- tens,
Least

225600

363
Inclination of Orbit, 5 9'. Annual regression of Nodes, 19 20'.
Rule.(The Year +l)-i- 19. The remainder is the Gulden Number.
(The Golden Number 1) x ll-j-30. The remainder is tlm Epact.
GRAVITATION. Attraction between masses )
mm
,
m, m at a distance I
j
p x 1*543 x 107
The mass which at unit distance (1 cm.) attracts an equal mass with unit
force (1 dn.) is = v/(l'543x 107) gm. = 39*28 gm.
WATERDensity at 0O, unity ; at 4, 1 000013 (Kupffer).
Volume elasticity at 15, 2'22 x 1010.
Compression for 1 megadyne per sq. cm., 4'51 X 10"' (Amaury and
Descamps).
The heat required to raise the temperature of a mass of water from 0 to
< is proportional to t + '00002/' + 0000003^ (Regnault).
GASES.Expansion for 1 C, -003665 = l-=-273.
Specific heat at constant pressure
1-408
Specific heat at constant volume
Density of dry air at 0 with Bar. at 76 cm. = '0012932 gm. per cb. cm.
(Regnault).
At unit pres. (a megadyne) Density = '0012759.
Density at press, p p x 1'2759 x 10"'.
Density of saturated steam at 1, with p taken ") __
'7936093;?
from Table II., is approximately
)
(1 + VUolMt) 10s'
SOUND.Velocity = \/ (elasticity of medium -4- density).
Velocity in dry air at t = 33240 y/(l + '003664) centimetres per second.
Velocity in water at 0 = 143000

LIGHT.Velocity in a medium of absolute refrangibility n

- 3 004 x 10lu-4- fi (Cornu).
If P be the pressure in dynes per sq. cm., and t the temperature,
fi-1 = 2903 X 10" 13i'-Kl + '003660 (Biot & Arago).
Those data are from the "Nautical Almanack" for 188:!.
f Transit of Venus, 1ST I, " Aslrom. S c. Notices,' Vols. 37, 88.

MATHEMATICAL TABLES.

Table I.
Various Measures and their Equivalents in C. G. S. units.
Dimensions.
1 inch
= 2-5400 cm.
1 foot
= 304797 ,,
1 mile
= 160933 ,,
1 nautical do. = 185230
1 sq. inch = 6'4516 sq. cm.
1 sq. foot = 929 01 ,,
1 sq. yard = 83GM3
1 sq.mile = 2"59 x 10,0
1 cb. inch = 16 387 cb. cm.
1 cb. foot = 28316

1 cb. yard = 7G4535

1 gallon = 4541
,,
= 277 274 cb. in. or the vo
lume of 10 lbs. of water
at 62 Fab., Bar. 30 in.
1 grain
1 ounce
1 pound
1 ton
1 kilogramme
1' pound Avoir.
1 pound Troy

Mass.
= "06479895 gm.
= 28-3495
= 453-5926
= 1,016047
= 2-20402125 lbs.
= 7000 grains
= 5760

Pressure.
1 gm.persq.cm.= 981 dynes per Bq.cm.
1 lb. per sq. foot = 479

1 lb. per sq. in. = 68971

76 centimetres3)
,014,000
of mercury
at0C. 5
lbs, per sq. in. _ 70.3Q7 _
1
gms. per sq. cm.
-014223
Force of Gravity.
upon 1 gramme = 981
dynes
1 grain = 63-56777

1 oz.
= 2-781 1x10*
1 lb.
= 4-4+97 x 10s
1 cwt.
= 4-9837 x 107
1 ton
= 9-9674 x 10*
1
1
1
1
1
1

Work (g = 981).
gramme-centimetre = 981
ergs.
kilogram-metre
= 981 X 10"
foot-grain
= 1-937 x 10*
foot-pound
= T356 x 107
foot-ton
= 3-04 x 1010
'horse power' p. sec. = 7'4G x 10

Heat.
Velocity.
1 gramme-degree C. = 42 x 10* ergs.
= 191 x 10s
1 mile per hour 44704 cm. per sec. 1 pound-degree
1
pound-degree
Fah.
= 10G x 10*
1 kilometre = 27777

Table II.
Pressure of Aqueous Vapour in
dynes per square centim.
Temp.
20
-15
-10
0
0
5
10
15
- 20
25
30

Pressure.
1236
18G6
27K0
4150
6133
8710
12220
1G930
23190
31400
42050

Temp.
40
50
60
80
100
120
140
1G0
180
200

Pressure.
73200
122600
198500
472900
1014000
19S8000
362G000
6210000
10060000
15600000

Table III.
Values for the principal Lines of
the Spectrum in air at 160 C.
with Bar. 76 cm.
Wave-length
in ccntiras.
A
7-604 x 10-'
6-867
B
G-5G201
C
D (mean) 5-89212
5-26913
E
4-86072
F
4-30725
a
3-96801
3-93300
H,

No.ofvibrations
per second.
3-950 x 10"
4373

4577

5-097
,,
5-700

6179
0-973
7-569

7636

MATHEMATICAL TABLES.

abo a 2 .
=5<e SO

00 rH O O rH io
OO
i-i
o
i
oo
io
to
1
1
oo
1
co iO
o 1
01 CJ (O O r( H
CS CO
CS i-l
rH
-a

Density.

cm.ec.
Conduction of insper
of
Rate Sound

iriHNOOJtfCOiON
OOlONClHH(N
OCMOrHOOOOO

x
1
II
Cs -rjf
CO
to
i>- 1 CO
ci w
a n (O
.o CIO
co o (N
cm 1 i{3
N h r-i cm cb co ib ib
*
Relative Conduc

354936 0-118 0-883 1000 0132 338-034 101-064 14-789 24-648

DD

tivity.

-HO
-rjl CO |1
00
O ,1 O
00 CO
1^ CO
Cs N
iO r-. . OID
COO
r* rH CS 00 CO 1 CO 'coco 1
o jori 0335
0557 0949
Co
S3
g|

St
1098
0927 1770
0330
xx
CO O'^oWHio^N'fH
rH
HCOO
uisd lO oo
i oDC5No>-'OC<io)aco
c 12 odtJi
O >1 lOl-HHHHHOKNO
OO OOOOOOOOOO
a o OO OOOOOOOOOO
c f
000033 000037
000180
000015
.22 3= 2be
ta"5 S,
Tenacity.

of
Elasticity volume

0 0 86 0 0 61 0 0 54 0 0 54 0 0 40 0 0 63 0 0 8
noO
rH
~ S
E S S
1 I Ico 1 oo 1 co co r^
1
<?*
cs T1 T1
CM Ai cb
CO do
iONCO
>o

000027 0 0 45

S
S S S
E
rH
1 1 53 1 1 00
-# 1 CO
-f co -h 1 1 >o
i-i
ip
tp os "O #
CO
*
O
-H O -H .H
O

Diameter

888000 3000 7700 7926 4500 92000 75000 36000 35000

Miles.
in

BO00rH-<7jlCN-X>r^
CM CM (M CM r-l
HfjOiOH'iMOCl
ci io co ira cm |1 |1

of
Time Rotation.

CO CO -"Jl OS o
.a. O
O
to CN CM CM CM rH
Inclination of
Orbit Etocliptic.

CO rH
CO

, 11
O
Sidereal Revolution Days.
in
1
1

Sun.
Least distance from

Rigidity

1I
1 1 1 1 1 x
*
||*
NtDCJOO
O
CO tO H
-rfl
cb ib ob
oi

dGirseta necset
Earmteha'ns
Sun.
from

059x10"

075 1-349 1963 2139

1-234 1

0-603

US O
-<Jl 00
CM O
CO OS
iO
O CM
CO 11 rH
00
OS
to
CO
iO rH CM
t CO rH rH CM rH
OSiirOOiOOr-ICM
tOOiOOOOOCMCMCM
Ost^OMOSiOCMOOl^
r^-rjiibtbcMostbcb
OOCMtOOOCOiOOOCM
cm co to co to r-i
-* rH
O CO
Oo
tO
OOCMOCMCMtOtO
iO -* CM tO 00 * rH O
NXCOHH-*OSlff)
OHCOXiOOCCN
WNcseeasoNN
OOrH-JflCsdcOS
rH CM

d1.ista=nce

,
1

it
co
co i uc> r-. cs
cm
tDttiONifNJSN^iOSOT)!
010CO-*X-*CNIOOOCNr-IOS
^aCOrHOCOCOWNNIsNOl

cscooooooooocmcc
CO cm r>- f CM r1 00
COOOtOiOCOCOCOOO
CD OI H
KN
-<J(
o '-O
O >C
Tfl O
O CI
CM
66hih*66o
iH CM CO
JupiterSaturn Uranus Neptune

Mercury
Venus
Mars

Iron,
cast
Platinum

Mercury
Gold

Lead

Silver Copper

Sun

Iron,
wrought

Brass,
drawn

Tin,
cast
Steel

flint
Zinc,
cast Glass,

MATHEMATICAL TABLES.

Table VI.Functions of ir and e.

= 31415926
a-5 = 9-8696044
x* = 31 0062761
= 17724539
log10x = i-4971499
log.* =

r"1 =
3183099
x 1=
1013212
- =
-0322515
200-f- w = 63-6619772
180 -5- x = 57-2957795
= 206264"8

Table VII.

e
e'
e->
e'
lor10e
log, 10

=
=
=
=
=
=

271828183
7 3890561 1
0-3678794
01353353
0-43429448
230258509

Table VIII,

No. Square root.

Cube root.

If.

iog10 y.

log.A".

1-4142136
1-7320508
20000000
22360680
2-4494897
26457513
2-8284271
3-0000000
31622777
3-3166248
3 4641016
3-6055513
37416574
38729833
4-0000000
41231056
4-2426407
4-3588989
44721360
4-5825757
4-6904158
4-7958315
4-8989795
50000000
5-0990195
5-1961524
52915026
5-385164S
54772256

1-2599210
1-4422496
1-5874011
1-7099759
1-8171206
1-9129312
2-0000000
2-0800837
21544347
2-2239801
2-2894286
2-3513347
24101422
2-4662121
25198421
25712816
2-6207414
26684016
2-7144177
2-7589243
2-8020393
2-8438670
28844991
2-9240177
29624960
3-0000000
3-0365889
3-0723168
3 1072325

2
3
5
i
11
13
17
19
23
29
31
37
41
43
47
53
59
61
67
71
73
79
83
89
97
101
103
107
109

3010300
4771213
6989700
8450980
1-0413927
1 1 139434
1-2304489
1-2787536
1-3617278
1-4623980
1-4913617
1-5682017
1-6127839
1-6334685
1-6720979
1-7242759
1-7708520
1-7853298
1-8260748
1-8512583
1-8633229
1-8976271
1-9190781
1-9493900
1-9867717
2-0043214
20128372
2-0293838
20374265

69314718
1-09861229
1-60943791
1-94591015
2-39789527
2-56494936
2-83321334
294443898
313549422
336729583
343398720
361091791
371357207
3-76120012
3-85014760
397029191
407753744
411087386
4-20469202
4-26267988
4-29045944
4-36944785
4-41884061
4-48863637
4-57471098
4-61512052
463472899
4-67282883
4-69134788

2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30

Note.The authorities for Table IV. are as follows :Columns 2, 3, and

4 (Mercury excepted), Everett's experiments (Phil. Trans., 1867) ; g is here
taken = 9814. The densities in these cases are those of the specimens
employed. Cols. 5 and 7, Rankine. Col. 6, Watt's Diet, of Chemistry.
Col. 8, Dulong and Petit. Col. 10, Wertheim. Col. 1 1, Matthiessen.
Table V. is abridged from Loomis's Astronomy.
The values in Table III. are Angstrom's.

BURCKHARDT'S

FACTOR TABLES.

Fob all nttmbees fbom 1 to 99000.

Explanation.The tables give the least divisor of every

number from 1 up to 99000 : but numbers divisible by 2, 3,
or 5 are not printed. All the digits of the number whose
divisor is sought, excepting the units and tens, will be found
in one of the three rows of larger figures. The two remaining
digits will be found in the left-hand column. The least divisor
will then be found in the column of the first named digits, and
in the row of the units and tens.
If the number be prime, a cipher is printed in the place of
its least divisor.
The numbers in the first left-hand column are not conse
cutive. Those are omitted which have 2, 3, or 5 for a divisor.
Since 22.3.52 = 300, it follows that this column of numbers
will re-appear in the same order after each multiple of 300 is
reached.
Mode op using the Tables.If the number whose prime
factors are required is divisible by 2 or 5, the fact is evident
upon inspection, and the division must be effected. The
quotient then becomes the number whose factors are required.
If this number, being within the range of the tables, is yet
not given, it is divisible by 3. Dividing by 3, we refer to the
tables again for the new quotient and its least factor, and so on.
Examples. Required the prime factors of 310155.
Dividing by 5, the quotient is 62031. This number is within the range
of the tables. But it is not found printed. Therefore 3 is a divisor of it.
Dividing by 3, the quotient is 20677. The table gives 23 for the least factor
of 20677. Dividing by 23, the quotient is 899.
The table gives 29 for the least factor of 899. Dividing by 29, the quo
tient is 31, a prime number. Therefore 310155 = 3.5.23.29 .31.
Again, required the divisors of 92881. The table gives 293 for the least
divisor. Dividing by it, the quotient is 317. Referring to the tables for 317,
a cipher is found in the place of the least divisor, and this signifies that 317
is a prime number.
Therefore 92881 = 293 x 317, the product of two primes.
It may be remarked that, to have resolved 92881 into these factors with
out the aid of the tables by the method of Art. 360, would have involved
fifty-nine fruitless trial divisions by prime numbers.

Is o

rs.
00

O c:
en
cm

t>
s*

i>.

O O Ft

I-- n

m o i> cn
C4

-f ~

is.

tO

~ o o en rs.
<-
* en

r* l> W N d
(, -
C4

HIOH

00

O ts

o F-i r- ts.
F-. ^f O

is. o Ci ts

o o o

cn en cn O eo
IO wm ct
*

o o cn o o

o r* o

o r*. o fi
F- T

cn ts
00

o rs co

o>

Ft r. *i
tCM

en cn en rs
M
CI r~ -.

r- o

CO

r- is. o c .
to
r-

CO O ts o o

CO

Ft Ft Is.
t -T

o r- r* en o

O r- o cn eo
rs tt

Ft O O
CO

o Ft ts cn o

OM t
C Ft

00
CO

Ft

Ct

CO
o ~

O I s. o

m o ph r- o
C

rt * rs

-- r* ct

on

O O O h

M ONftON
CM

CO

00
cn rs n ci o

CO

O ts

m o on r-

d o ^ r- h
(H
M
en

r* o r*

is. rs.

Ct CO r f

CO is. o

OJ O

o o **

en o r
r>
*-

CT. Ci is.

-H O is. o
en

i-> en cn is
Ft
H

-i .. o o rHO

at f- r- o
C
FN

o O o c- en en

O O ts- IS. O

~ o -.

US

to

i-t

is. n en - r*

r*

O r-i

at at o ts. en o

o o o at

o ts- en o o
Ct

p-i ct

en c*

ts.
ts o O -H O

en

to

en

rs o rs en
CO
CO

CO
CO

o
1>

CO

o t oao
a

r*. rs cn o
eo
>-

-- co

r* o r- o

rs

Oi r*. r-

a. t* o .

to
o en

W+ O Is.

fj*

o o cn I*.

en

rt

r- en is.

ts. o> w n t

Ct

i-<

US

t- O ts
US

rH ts ts.

ro M ^i Ft Ft

o is
Ft en *

en o

ts rs o

*r t ,

CO
rs o

O Is m

en cn r*.

CO

r* en o
c*

en ro o

o O ts- o tQ

C*

Ft

CO

CO

o en

OOIO

rs en r - o

r-

O Is. ,

tt t** t*i at n

US < Ft

Is. |s. O
CO

Ft o o ts cn

en o

CO

Is. Is,
CO

rs cn eo

kO

r*. Ft r- en
sr

r- m o

o eo rs

en o r* o i>.

rs en o

en o r- - t>

-t o rs -t
eo
<*

o o o

en * - f-

I-- C Ct

to
o fro o o

o o cn rs

o rs e
CO

o o

o m is-

O ts en t-% fh

en t- o en
to m
"

~ r* i-h rtO
^

r o is- en en

o ts en co
ct -

Ci r- a>

o en

O is.

* ct

eo en r* o en
US ^i
-

O i> m

t- en o o r-

o en
(0
r*
ct
o t>- en t>. at

-r

CO

CM
US

cn

Ct C4

Ft C

s*

r- o en o o

O ts

o o

cm

r-. r- r--

en cu o en
ftr.

O l^

-f

y. cn cn
Ft Ft o eo Ft

rs en r*
ct
to

O m>

CO

o cn o

cn is
ct en

Ft C- CO

O IS H O Is
Ft
TT

o rs
to

r- o>

en is o

en Ft m ts o

s cn F-.

ts. ~ o is. o

en

Ft

CO

Ft Ft

T*t

rs o r o

en

o o o
n

r* o o c?>

Is. Ci o o

^ o r* > en

r- i-- i e?>

r- en r* o en i>

en ^

^r Ct

-- c*

-i

CO
CO o

en O

o c- en o

o en en t- o

CO
CO
is ~
CO

O O rs.

en o r* r*

o ts.

pi h. O O n

en r
Ct
n

o o en r- o

o t-H o

J> o O rs. r
en

r- o en

o o
CO
CO

<-<

i- o -t

o r O rs

o en

- h O Ms

O r- i o

Ft Ft Ct ^

o o co -t m
o cn rt

o ro rs
sf

o o r- en cn

o r- M cn o

o rs. o

f- o r- rs

CO

Ft O tS
CO
-f

r- t-. en ts

o o en o

cn ts

CO Cs o

O is o o o

o O ts m
Ct

o o o r-. cn

O O O i
Ft eo

rs co r-

cs is cn o o

o o en o o

rs. r-

o cn is o n

r* rs o

F- o

en f- o ts.

O is o

Oi o

o o o o o

r o o r-

i-h co rs. cn co

OHr-ftCO
rt CM CM CO CO

feSg

00

en

CM

rH

Ft sf

t>> cn en o Ft
Ct Ft
Ft

CO

CO

s*
CM

CO
O rt

ts,

CO

CO

US d

CM

r-> cn oao
Cl

r* ph

o r- o <~>

en ts.
ct

f- o en en r*

o t>- en en t-

nF.*

CM Ft

CM
--< r O)
04

en -. o

o en o *> en

r* o t o ^

O O O) t- o

ts. o en
P-*
-J.

ts. i s. ts. m

o o

o o r- o -t t*
eo

CM
CM

CM

CO

Ci

ft

CO

r- at o to r*

*i

t C*

en
r-t

en -

O is

r- rs

t0 " O CA

ct

Ct

t- o o

W o o o
ct

f. a o th i-i c*

en i-- en

en p*

o o

O Is

o en o en o

h<-< o On
eo

r-. o en o
t
CI

o o eo - o

o r- o en r*

CM
ts O t>

en o t o

en r is. o

CO

en

I"- O ~

CO
rH

O
o o en o o

o r* r* o r*

CO

CO

-.
CO

O
rH
Ft

C4

~
-i

Ct

O
rs o o o o

r* r* o o

*- r* o o o

en t-.

en o o o r>. o

CO

o
o
o

o
Q O

o o o

o o o o o

is.

o o o o o

tS. O IS.

O
Ci co eft *** t
O O *-< r-t H

-> CM CJ CO CO)

t-t CO J> Ci CO
^* ^ ^i* ^}* irt

OS -- t- r-i CO

its cc co r- 1*

Js* C3> CO CJ r-( Is,

t> l> CC 00 CI sTS

OO OO -t

1 9000
O Ji r

O O ts. Oft en

PI N o N fCO

Ft

Ft

C.

--

t-

1 -

Oi o i-

PS rs p;

C7S

.*

N C. Is
Ft Ft CO F-

r* O PS o

Oi N O

PS Ft O

N PS O N N
t 00
PS

O ts.

O Si N
US

ts. O O is.
Ft

00
r. t- ir.
x - *

r o ps

ps

f- is n

o PS o
PS ~ rt

n o o o ps

N- ?S CI

O O 03
n

oj ps r* os
ps
n>

CO
00

ps n o . oj
CM

CC

Ft

n a ts.
* n -

Oi PS
00 o

Oi t- C PI

i>.

CO
00

"-

US

MSPS O p)-
O PS - r

-<

ps
H

-h
(!)

O O i-t ts.
^-

-t

Oi r* is.
Ft

> PS r- Oi
r c*
ci

Ft
l-t

r**
Ft

O O Oi -> ~
O i-< C*

Pi O
to ci

ts. t-- Oi r

<-< pj
*

M r- t- n

rs. oi O

fs. ts. -t O
FPS

N O O Oi

O O M " O

t* r *i

is. r* p; isps
C* Ft

ps Ft is p:
< Ft
C

co
co

i>. r* oh

F- n r* o

Oi O O PS

N s .-: f-.
N CO *C

t- ps r *-
d i-i

C h M
~ O
-+

O Oi
PS
C-

r o o

O Oi N Ft Oi

m
co

r*

t" Pi PJ t>.
.-i
c* ^-

Oi ts. :i
Is.
Ft

Ft o o o

Ft

Oi PS "-<
-^ C* CO PS

o r* i- o

PS O* -i I*.

Ci Is is. i pJ
US O *
*i"

O -t PI p; N

PS r

i- PS r-S O

o o

is O PS Is.
us

PS
F-

r- ps o o

Is, O O P) Ft
O Ft

PS

* Ft

OJ

N O rs

O O Oi
CI

O i t>. r

Ft i> o PJ
r*
ps -t

O -t ts o
(OflH

P! O N

O 1- t- PS

MS

O OlO ON
CI

Ft
Ft

O Ft Oi PS Is

N O O

I- C !"

O O. PS PS

i o

I> O Ft
-r

-> P) PS

r* o

F-r

PS Ft ps o
Cf.*

r- t^ p*. Oi c* PS
PS ^ US
Ft

O PS 1^.

o n i o

PS O Is

NO K NO
Ft
us

O Ft PJ IS IS

o o t^
OS -

PS o o o

O PS f- o

O N PS is

-*

O is 1- Oi
Ft PS
us

O O PS

t PS

ts. fs. O

o o - o n
*
C*

I-. O Ci Is O
PS
ci

00
c r> n

r> en r* r
rps

o o o n ps

t^

p)
CO

r*

Ft

"*

Ft

S-

JF)

-r

Ft
Ft

t*
0 C l.

n n ps ps n>
- ps
r* *r

ps

oi

f- 00 ~

ps

f-

r-,

*-l

^1

Ft

1>
l* o

ps n n ~- o

PS

es a t* - i- is Ft
-j. tj. n

n. 3? Cft
- CI

O O n N PS

f- o n p:

oo -

M00 O"

O ps p: ps n

r- r
to
co
w

Is p) O

- O N

lO

e *

PS
Ft

tS
S

Ft
tS.

Ft

C*

Ft

Ft

Ft

oi

CO

oo o

i-

n ps " n
** ts

- n

c. oi n

n o o

-~*

r> o C. C O i"
tO

o a r. - o
c fh

n n o n

1"

PS
Ft

~ n O PS

I^ Ci

c* **

co

O O r PS r*
CO
to

f- to

N O n oi o

o
coo

O ps ts O O

- O M MP) *>
f-i
PS

O C: O

n c r- c Ci

PS
Ft

Ft
^i

p)

ps
US

oo o

rs rs O
n
f-

n n
to
t

f)
CI f.

op:

C*
Ft

PS
Ft

i o

is
Ft

Ci
CI

1*

r-

^H

IS, tS. IS
(O

3!
Ft
Ft

tS. O

00

fT)

Ft

tS -t

IS Ft
i
1

PS

Ft
I-t

US

Ft

N N
<0

Ft
*

Ft
F-

eo
C r- o

M l>
CI

n o n n o

Ft

CO
n n
* CI

o o o o
PI

-t

Ft

-* P3 t
PS

PS N Oi
Ft
CI

N ts Oi Ft |s o

ts. f. PS

F- O O P.

Ft

Ft

Ft

Ft

CI

N Ft

PS

CO
t* c o

n n en ps c

* *

pi r ps

en

Oi

F- Ft CM

CM
Oi ps o o r-

Oi O l>> o o

PS t>- ^- t
^ PS

Oi Ft ts. O

O is.

ts O

co

n o O PS
m
-r

c o
c*

co

r* r ~

PS PJ t>. o o
c>

PS

N N rS
F- CI

O O N

rs r. - o

r ps

is. o o Oi O

PS PS ts. .
Ft CI

Oi O Oi Ft
CI
Ft PS

O O N O

o o o

ps n ts o

WON

O Oi

v.

t-, c o

I* ps o

- O O

n
CI

t* Oi O

p: Oi r-
~
Ci -H

O - NO Is

PS

O ts. o c-

f- Oi N
* CI

PS O Ft N

OS n PS

O OS o n n
CI
PS

o oi n i h c

n o mc l-

Ft
Ft

PS

N PS ts

o n

f-i Oi P)

PS o o o t^
C*

OO-Oft

O N O - N

ts. Ft o PS
PS
Ft

Oi O N O Ft

O N- O C O t5*
F*

hOOOIO

o o o i-s.

Oi ts. PS o
C*
CI

PS o

ts Oi Oi
F- CI

o r- o

O r- ps PS ts.
c

O Oi ts.

PS o o o

N Ft Oi O

r* r* eo

o o o

OIOO-O

is O o

O N N PS

O C i-t fH N

co r d cc Oi
CI 04 N CO CO

i-l ts. i-H CO l>t}< s# iq xo ia

OS CO OS r-t t>
i: c C 1- is

F-t CO N. OS CO OS
00 CO CO 00 o> OS

^
N o o

o r> ec o o o

<- o r* ps o r*
00

o
~

3 N

cs t*. r* o

o o o

n o o cs o

o o
CO

O
OS N>
-f u= *o

--eoNOieoo
CX C t>- t*

oO co os era 3s jj

 Ol C r^.

to

to
;0

eo eo

to o

s
eo t>

1-

to Ol CO Co
Ft

Ol ro O O O ro

to o> ro
00

FH

F.

CO
O

Ft
t>.

Ft
Ft

JO
Ft

O to

to oi r-

to, o eo
O
to
CO

eo oi o
o

to O O

CO I> o CO
Ft CI
00

to to to Ol
Ft CO

eo co

l> to to o
CO

to CO to
to
- <- Ft

Ft
CO

O to CO

Ol Ft o o Oi
CO -V

CO O O to

to to to Oi

FH

tO

^ to to oa

CO
C

CO
Ft

Ft

to O O O to

io

to

00

CO O fh
F- ,t

O O CO

Co

CM

to O to

-*

C to CO

to

FH

C0

Ft

-1

tO

CO

Ft

o eo o

FH

CO
Ol

toFt

to

ro

tO -H O CO
to
FH

Ol CO O
to

Ol 1^ to O
O CO

Ft tO
to

O eo O

to Ol
CO

Ol CO to M to

O CO *

to o o

Co r* oi

o o

to O O to

CO

Ft

CO

00

to e<

t- O CO

Ft
Ft

oi to eo
iO
tO

to
*

to Ol eo
CO
CO

m
O to

en

eo to to

rCO

Ft

o fh eo o o

to Co .

CO

co

CO

Ft

F*

ro oi to Ft
tO Ft
Ft

Ft

Ft

Oi Ol to
F-

to Oi

O CO to to
FH
tO

CO to Ft to o
co eo Ft

to CO Ft
*r to

to m CO CO Ol

Ft Ol o o

O to.

CO
CO

CO Ol Ft Ol o ro
Ft

O to eo to

ro

IO

CM

fh

Ft

LO

CD

CO

O O Co CO to
to

CO

fh
i-t

Ft
to

to to to

j>

IS

Ft

CO
CM

r O to

CO

tO

FH

00

CO to CO O CO
CKOO
Ft

o o

CO to ' >-
Ft
* O

Ft

FH

Ol

Ft O Oi CO
tO
CM -

O Co O

Ft

eo to o

tO

Co o oi
Ft

Ft

Ft tO

to to CO Ft

CO o o

CO O to O Co

OOOi

Ft

O Ft O Oi t>
CO

tO

CI

O CO to Ol o

CO

Ft

Ft

CM

eo

iO
to to
to
*

o o

o c o

to CO

-*

fh O

O CO O to O

O C31

Ol

t> o o

t>. t>
00
CO

eo

JO
CO

e o

CM
CO

en

I>

to

Oi eo eo Oi
00

tO. Ft

O tO

Ol O

5
=
CT> ^

to o ro to

cm

m n n
to Ft (0

CO CO o ~

CO to o
CD

CJl to O
to

O to

CO

Ol Ft
O tO

Ft to
Ft

C> Ol to O
Ft

Ft

Ft

co,

O Ft CO to CO O

CO

CO

tO Ft

-*
Ft

CO

tO Ft

f- O to O co r~

tO Ol Ft
Ft C* *

Ft

Ft

Ft

tO FH

CO F-t

O to o

Ol Ft to CO

eo oi o
CM CM

O O to

Ft O Oi o
eo
Ft

to CO CO
O to.

tO Ft

O CO CO

eo co to.
F- C*

Ft

CO

FH

Ct

fh to eo fh
FH

to O O to

IO

to O Oi Ft CO

J<

C*

O Co oi
Ft CM

Ft

CO CO FH
Ft tO Ft

co
O

Ft

tO,

Ft

O to to to O
eo Ft o

co

Ft

FH

to fh eo

CO

to

O - t>

O to eo O Ol

Ft

Ft

Ol CO CO

CO

FH
Ft

CO JO tO
CO Ol

Co CO to O

Oi Co o O to
Ft
tO

O F.

O O O to

to Oi to eo o
V C*
CO

O CO o

CO Co o
Ft
Oi tO

to O

to o o

O Oi Ol o o

Ft
-

O CO Ft
Ft CO

to eo

tO tO tO Ft
Ft (O
Ft

to o
o to

co o o o

t> o eo

CO

eo
r- o o

to
to

to to to
to m

O to Ft CO

CO CO to O

r- O to

o eo

to

Ol O -t
Ft
CO Ft

tO Ft

CM
to
CD
CM

Ft

CO O Ol

CO Ft

Ft

CO

o
eo

eo Ft

Ft
Ft

Ol tO
CN

C4

o eo

to Oi

Co

to ft

0)

CO
CM

CO

to CO -t
to -

to to
Ft *

CO Ol o
co go

to CO I~

tO Ft

CM
CO

r-

CO

CO

O to O to

c* to

to Ft o o
CO

eo to eo to ro
r-. ~r -r

CO t>

Co CI
to tO

ro

1^

Ol to to
IO
CO

Ft

to O

to O to CO

t> CO O to
to

CI !- O

eo
CO

>

to

Ol

Ft

Ft

CJl

O to o o o
i-t
CNI

CM

Ft O tO
00
r-l
to to Co Ol
eo fh c*

co eo o o o o
C*

MS

Ft

fh O CO Ol
#

Ft Ft

CO
00

FH
Ft

to

rH

FH

f* to
CO

eo F- Ft oi
to

o o

CA to o o
CO to

to to O Ol o
eo
Ft

o o

CO o

to CO Co Ol
O tO

o
CO

Ft

eo to to Ft o
C* Ft
Ft

CO to CO H to
Ft

Ft

to o o

1-1

00

|o Ol ~
O

oi

fh

O to ro
CO

CO -t o 1^
tO Ft

o
o t> eo CO
d

to

I-

tc

(0 CO CO to ^*
CO Ft to
to

O Ol

ci o i> t
o to

CO CO CO
tO Ft CN

to O to

O ro ro
c-t

to o O O Oi

-t t> tO
to
CO

CO to 0

O O CO

CO

Ft
Ft

ro
Ol

to Ft Ol
tO Ft

Co

c*

CO
f

Ol to O
Ft

Ft
*

r^

Ft
Ft

tO fo IO CO
Ft tO
Oi

p:

c*
CO o Co Ol to

o o eo F*

eo

Ft
CO

Ft

CO

to

CO

Co

eo o

eo eo o
w
o

to eo o o t>

O Ft CO o

Co ro

eo
o

Ft

to

O o.
Ft

FH

CO to O t>

Ft

CM

O to o

C*

eo ro to oi
-f CO
to

O O

Ft

oi Ft ro to Ft
CO CO fh
Ft

O O FH to O
to CO

to o o

Co CO to.
O Oi

o o o

O to

oi o co eo

FH
fi

O (0 Ft

tO

Ft

cn
o

to

Ol to

to <*

O O CO O to
Ft

tO Ft

to to

CO Ft Ft
Wf.*

en

OS

o o o

o
o>

co
o

-8

o>
CO

-t to to to CO
tO tO
-t Ft

0>

ro to

cn

I-

CA
CN

Co

to

O eo f00

Ft

CO Co CO
-J. T

Ol CO F-

<n

FH

eo to

tO

CO
CT

rt

O)

IN rH CO l>
^H r-t rH

os pa eft *-< ty

-. (M >4 CO CO

fH CO to Ol CO
^ ^ ^# ^ trt

oi Ft i> r-t eo
\q ep co t> to

to. OS CO Oi i-t to
t> to 00 00 01 Oi

rH CO Co, Ol CO
-.

Ol rH to. fh CO
r-l CM CM CO

J>>01 CO
W tP

9000-18000
c in.

o o r* tN. ok

!* r- o

o o oi r.
eo
ct Ft

CO

~ _ r^. ro t* to so
c* eo

i Oi in O

r
- n

t o o> ~t o
in.

o on ono
fh
oi CO

coo

on m

- in. o

C r. o

in o eo

o o r*

o t> in. to ai to
CO ~- CN

eo O O tv o r-

O O ro, o
p*
mm
CO

o r* f-

eo

CO

Ok
CT

Ok o.
r to

O t* t^ Ft

CO

CT

eo ** Oi r* o
Ft

s
s

Off)
CI
X

t* Ok O i- O
BO
o

eo O O eo t-

o o o

o r* tN. eo -

CT

ci

o o eo
o

oo t* ^^ n r* 0k
-
o

r o t'. ci
CO
~

m f* in.

co
ct

0
to co

t- r o
FlhH

o ~ **.

an

c. .

O O O O r

F*

Ft
F-

CO
CT

Ft

~+

I"- CO
Ft

Ft
Ft

Ok o O O Ft

r*" co co co

Oi Ft r* co

CT

Ft

O t- O r^t
CO

t* r eo to
^ r> c*

r co Ci Oi

o o r* Ft to

ok
Ft

rN m eo In. in.
Ok Ft

r^ O

I- r

f- o o e-

Ci f- eo t>> O
o o ~r

CO O O Cn.
O
H

eo o r*. eo O

in. Oi r*
CO CO CT O

Ft i>. oi co r*

Ft
Ft

r*

r* CO Ft yi Ft
Ft Ft CO O

Ci r* Ok
CT
IO

O !>. CO CO

Ft

Oi o co r-
Ft

o t>. to r
o *

O Oi eo r1*
CT CO

o ~ rN Ft oi
-t
CO Ft

r* r* o Ft
cn to
CO

co r>- o t
Ft
ok

o ok Ft r* Ft
t>.
Ft

o o

tN tN
Ft

r* (n n r n

C03iO

r* o o

r* co t^

In. CO to
ts
HH

CT

* *o
O

Ok O

Ft

CO

r-

0k

Ft

Ft

Ft

t*.

Ft

Ft

<*

Ft

CT

CT

CN

O f>. Ft

Ft

**

Ft

Ft

Ft

Ft
CO

Ft

tO

r o o o o

o r* Ft eo

Ft

Ft

FtO

Ft

t*. r*
Ft

O Ft f. CO o
CO

i^
Ft

r co r-*

r* t* ct r- o

*t

r* Ok

r in.

en

t- o> O t CO

(O

o r -t oi o
Ok Ft CT

eo t- r o o f-

i- o o r* o

i> O

o r* r-

O O I^ O CO

oo a r*on
c CTi
-r

Ft

r* o co r*

00

O 0k 91
O CI

Ft

Ft oi r o O

CO

Ft
Ft

&
**

r* Oi co

CO

t> Ft O Ok o

o eo

r* o Ci o

t> to

0)

** o

o o a. o

Ok

co r

Ft

r- Ft
_

r. e)
Ft Ft

-t

CT

CT

F-*

tN
Ok

Ft

CO Ok in. in.

Ft

!)

eo eo o r* -t

CO Ci O O tN, in,

00

tN. Ft

Ft

CO

Ft o o r- r^
Ft
o

hMO Oh
CO Ft

ok F- co

r* i> Ft o r-

O O O ro In. to

Ft O Ft O
r*
Ft

o r* o Ft Ft
CO Ft

r- Ci o r* o

r* r* eo

O tN tN. O CO

CO Ft o o

Ft Ok CO t^ Ok
CO CT -t
00

O I^ Ft o

eo O O Ok O

ro in. tN co o
Ft Oi
O

Ft Oi co r-

r o o r Ok

Ft

Ft

tN. Ft

CT

* CT

Ft

00
r- f-. n

eo

c*

O Ci ro r*.

r- oi o

Ft

00
o
eo -

in. eo r t* feo
r

eo to co o r-
~

o o r *i o

Ft

ct

O O CO o

r* o o f. Oi iFt IN. tO

is, Oi l" Ft Oi
F.HICN

Ft CO tN o o

O O r>0 oi

Ci O-

eo
CT

o -t tN. to

-t
t. Ft

Ok tt.
Ft

to

Ci

In

CO
- - o

a o o

r r* eo ro eo

oi o r*
M
f- (O

r* r* r* o o

r* o co o ci

- >

CO

*-

r* O eo t- O

Ft

t- Ok

CT

CT

00
CI

F-

O Oi O CO

* CO

r* o <- eo
Ft

CT

tN.

Ft

CT

go

CO

**

03
o c o

ct *

t. i* ~

<-

eo pi r- o>

Ci O -* r

Oi Ft
Ft Ft

r*. o o co o

o r o o r*
CO

co o o r-

o o to r r

O Ok o

O co r- O

r- r-

C3
CQ
m o r* y.
Of*
r*.

cn

Oi
CM

-r

Ft

to

CT

Ft to Ft

t* CO

1H
o o e*

in. o

- - n r o
eo * CT r*

r* r- o r*
CO
<o

ONAAOO

Oi r* co < o

Ft

in. o

to o o
M

r- e -> r o
r.

In. O In. CO o*
* -t o

CT

CT

Oi o* r* r
r
ct
-

t to r* O

Ci l* Oi O CO
tO
CT
O

r* Oi t rj o
Ft (O *

OlNhtO
CO
Ft

oi eo O O O
CO 4*

co r- o o r*.

F- C o

Ft

Ft

t* Ft
CO Ft

00

eo
CO

-t
CO

Ok Ci in, o

O f- t^ K CO
O
CT O

eo Ft r
r*
^ Ft

- o t o
to
^

Ft

tN

t^ O r* -

o c o o

CO Ft eo O Ft
ct Ft Ft
m

r>. r* -t to

f-, r* eo o

Ft
O

l>

Ft

n t ho o
m co

O C Ok in
Ft eo

tN Ft

eo co r* o

o eo o Ft r*

o o r*

in. co r*
Ft to

CT

Ft
-

Ft

to Oi
Ft Ft

IN.

CO
-f

fO
>n

CD

- o o r* c* co

<*

Ft

-t

CO

t -

Ft

-o

Ft

Ft

O
Ci o
tN

bo eo r* o f- i-

Ft *

Ok

t>. ro Ci f* O

<- eo

t* Oi
CT

fFt

e>.
Ft

Ff

F-.

tN

IN.

o
nao
X

o t*

Si O r-

CT

-t eo tN.

Ft

to to o o ci f
-t

CT

I-

Ft

Ft

CO tO O Oi

co r- o f- o
CO
co

Oi r* co o O

r- o Ft ok

Ft ?o o r>.
Ft f.
Ft

Ft
Ft

O CO o

CO
CT

00

Ok co - o r*

CN. o t

o t* o o r*

O t- I-

ct

o o o r o o

Ft o r^ Ok
Ft
(OCT

NNOOnO

Ft

-t

CO

00

Ft

Ft

Oi
a

05
r* c

* o n

r- o eo o r*
CO iO
'

o o r> a r- o
CT

Ft

00 X 3S 35 35 II

tO

Ft

n -ft f^ i^ h
O C rH i-t 01

t>.

O t>> Ft Ok o

r Ok to to
*fltN

tN. tN tN.
CO

eo r OS co O
CI tM CI CO CO

rH r>. F-t CO t>.

^* ** it* IO o

if? C CC t> tt

rw co r>- r5 co 2
X X X X Ci 05

Ft
Ft

 r o r

r- o o

NMQh
*i
cn eo

r*
FN

<7> O CO
C*
fn

f- eo O r^
eo *n

r- o CO eo

CO

M F- o> O

I> CO
FN

Oi FN
FN *

FN

I>

to

(O

r^. fn
tO

N-

Od

r* fn r^ o

***

r* o

co

00
CD

t^
*

tN.
FN>

O r- eo (^

fn o fn o cn o
fn

eo o Oi r o
#-
co

CN

to

FN

J>

Njl

FN

CD

CO
CO O O O-

O r- i> Oi
o
tO

-H

I-

i>. r* t^>
CO

O t>. O O fn r-j.

F-

FN

00
1Q

fn CO CO
CO O 00

o i eo eo r

t-* CO I- -CI Cl CO

r-

co r* o

CN

FN

eo i> fn r* O

eo o eo

C3 FN O FN

t* a o> h

r^- t^ t*
to
eo
O

tO

C*

FN

FN

FN

r- fn
ITS fn

co

Oi 1-* CO
tO
F-

O I> FN o

r*
CO

CO

Oi o Cd

r- Ci

FN

O o O r>

r- oi eo o
fn r

CO o

CO ^- I> Oi

r fn oi o o
c eo

r- Ji

fn CO O
*H
CI

CO

r-

eo fn t^ rfn
nnn

i-

^>

CO

1- O J) h
CO
CI

O O I'* r- fh
OS fn

r- o n n o

fn Oi O O eo

O o i- r- t>.
to

CN

FN

CO

CO o **
o fn
>o

co o r-
CN

t*

t*. O FN
FN
CO

CO

O O CO -n

5
CO CO O:
CO n
CN

FN

FN tO

r* o f- fn

FN

Oi C- t- Oi
FN
tO 00

o r* r* fn

fn i c: r

O r* r* O

FN

o r n

CA
C*

O 0> O r*
CO
to

r . ** cq
co m co CO

Ols - ON
to fn

h an o n

Oi FN

FN

i-
-N CO

FN

t-

CO

C*

t-> o fn o r*

CO O fn o
Ct

FN

OOlHhO

o fn o r*
tO FN
FN

CO c wi i> r;
r
fn

a r- o

o a> r- r-

r-

O CO O Oi

Ci

eo r* - o o
FN

00
C4

FN

fn o r

FN

C*

CO

FN

FN

Cft FN

C>

CO
FN

Ch Oi CD t o
fn to r*

CI o

n ono

eo fn eo

I-- o c

FN

FN

tt)

eo t* r r- eo
i
fn
eo m

ci eo

t-N CO
FN If)

CO
CN

CO
CO

CO
FN

eo eo
to fn r*

FN

Oi

CO

CO

^t

r-t

t- c o

FN
FN

FN Oi FN r^

oi eo i>
CN *

o co

O CO Oi

o fn eo o

Oi

r*

co r* ci o
00
CO

to
eo

r eo r* r^
FN

Oi O eo Oi O
CO
CN O

J> o

o fn eo

CO
rH
CO

-<*

r^ o fn rCO *

HIO

CO

CO

co
O Oi fn CO
CO <* fn

OlO NAO h
FN

o
1>

ear

-I r ** o> *n

O Ol ts -

CO

i- co r- r-
tn fn
fn eo

to

us
09
5

I> CO O Fr
fn

CD

CO

FN

C*

1.*

FN

t-

o
FN
CO

CO)
FN

f FN

C3
CO

ci

eo

O r* o fn fn
eo
eo fn

eo fn eo t^

o o co eo o

fn r eo

FN

Ct

FN

C4

r* fn eo
FN
CO

Oi :o r* t CO
T >o
Oi

r* ~ o ~ r

r:

r- r*

ci

fn oi r^ l>-

eo fn o r* o
c* ~

-n r*

US

l- O Oi

t>- t* o

t^. r* fn f. oa Cs
Otfi

FN

lO

CD

0>

CO

Oi l- CO
04
tO

fn eo eo r* co
FN FN co eo CN

t^ Ci
CO

r oi co o

f^ Oi
Oi FN

fn
FN

I--

-n r r*

Nro -n
eo
eo

O O IN.

o r^ oi r

o Oi o
CO
CN

eo fn f^

CO

t* r*. fn t>
CO

CO
CN

CO

Ci r^ Oi t* co
* tO CO CO

I-* O. Oi
tO
Ft

eo r eo

i o eo c^

fn Oi

O CO CO o

M" FN

rN C)

CM

CO
CO

en

FN

(0

CO O 0> r*

CO t o

o o
CO

FN

oa
r*. i Oi

eo

F-

t-%

CT

fn
FN

r>- rCt

O ~n ro fn
CO FN FN

c r> a o

co
1

tO

FN

C3

rH
n (n m r-t*
fn

t cr>

r^ eo O at
CM
to

FN

eo f* eo Ci eo
tO
C* FN

r* r^ fn
CO co

eo fn co r fn

o O O CO o

fn

t^ CO

*r ~

T-l

co

r* #- ^ o
o

r- co os
CO

FN

FN

C*

FN

I>

l>

FN

FN

FN

-r

rH

iN
oisc ;

O r*. o o

FN

CO fn to r r

FN

FN

O o Oi eo

r*. t*- o> r t>.

OCT

CO

FN

r-l
i-J

J> Oi t~
C*
CO

Q r* A O Oi
FN

FN

r*

o o o o

O K CD

CO

Oi fn
1^. FN

f. O

r- t^
FN

eo o o r t-d
CO

i> co fn rFN

CO

t- f- eo

FN !> t^- FN o
-t
cn - *o

o o o r*. F- co
eo o

o t* o a

r-

co os i> eo
* r*
co

o an o

i>. oi o t>
CO
s

O O fn eo

fn fn n
FN

o o> o r fn

o ooin
O

Ci

FN

r* o o o
o
o
fn O o o

-*

FN

r o o

r*. eo fn os
ci a ci

Oi 1 CO O
FN

-N

t* O

FN

|>

eo

CO Ci

r- co
CO

r* a o w

FN

O F- O O

r- eo *>
tO

fn
FN

FN

FN tO

FN

:o oi
tl>

i-

FN tO

Oi ^n o r
CO

FN

r* co r* eo o

OI FN
FN CO

r^ r*
c*

eo i Oi r Oi
Ci
fn IO Ct

r* ^ fn

r^ CO r
Oi Tj

Ci r- eo
CO
~*

f- eo co
to to C*

o F- eo o
eo co

r+ co eo t>
c o

Oi t>1" CO

Oi eo to
tO FN (O

o r fn fOk tO
FN

fn r

r* Oi fn

CO CO O I>

eo o

eo eo oj r* oi
C ^ FN
U)

r*

F-l CO I*n Cft CO

OS r-l I-* rH CO
r- O) (M CO CO

!> OS CO
co co -*

O.

FN

FN

Nf

!>.

fn

CD
r*. eo fn o -n co
CO FN FN
CN

CO
C73

I> O n M> H
-f

F-

eo co

r* fn

FN

N0

t- O Ci -n O

CI

1^-

CO CV O l> FN

o i-* o eo fn oi

CI

FN
FN

eo fn en

CD

00
CO
00

FN

fn r*

-00

00

co r- c o
<-* ci n

n fn eo fn
* f.

t^. co cj r* co
O Ct fn c*

t~-
-T

FN
FN

FN
I-

Oi o r- fn o r^>
CO

o o o o

CO

CO
00

r* co o co .
t*
C ^

co r. oi
FN

FN t> Oi

tO

t>. O h N O
*-
eo
to
CO

r f- r*

CO

CO C

rtco
<- * rH CO |>
OOHHH

FN

FN

O O f*

OOhNF.

0JC0 0)HS
rn 5 cm eo oo

F- eo c a.
fn

FN

i-t

^ ^ ^fl to

CO 1> C7i CO

eji -4 1^ r-i co

lo cr ex) t> t-

fr-- OS C?0 Cl ^^ t>

t>l> OC CO CS 05

rtco

o o

18000 27000
O p*.

1 i- (^ o

o * o n

1^ IM O
CO

e . m 0 r*

p-

p*

CO "0

co eo r*.

0B

.- o
X

at o

p*

o
C4

eo P- O p*
io

O O CO

1-

co r>. c

P- c CO

CO p- CO

eo P -

I-

o o-

CO

t- o
l-~

r- p.
r- (O

I-

r- co
CN

eo i-

t~

o 01

r- co
Ul

C^ PrH
eo

CD
O C P*

p. o

o i eo
o ~ -h

r- r- rn o o
^ to

co t^ eo pi
M
CN
r-

' eo r

r r*. O
c r-i

< p eo C p
to cn
CO

r r- o t*.

p>. eo
~-

n
CN
*J*

r- eo r ci

r* r-

Ci f
r- r.

co

r* O co
r- c

- r r

O r- ro p-

r*
CO
CO

CO

CN

rH

CO

3i
CN X

p p*
rH
rH
-a

CO

r r p*

CD

rr:o

D
p- co

i-

CO

. __.

c *

r- ro o r*

a o p-

l- CO .
to

p- r-

o*m o

r-

~.

r^

CJ

09

i~ o

CO O

r- co

r*

r-- co

CN

r- . p.

o r- r*
CO

eo O
CO

CO

-J.

p
CO

l^.

r- o

i r-

i CO

CO

O CO

-H
-

CO
iO

p*

OOO)

1 -

r r- ci
o
c*

co
O

CJ

cn

O)
to c

r^- CO CO O
-T

i-

8
C!

r- . co

r* n

eo r- o

- o o

CO

r-

i-

S) P^ r-.
X

C O

r*

wt

a r*

CD

o r- o

CO

CO
CO
R 0k m

- c r.
~
eo

r eo ~

-*

r- i- tt o CO

r r- o

t t>. ~- o
CO
-

>

Oi '

CO P

r- O p

co

C C-* o

9
a

I-

o
eo

CO CO o

-"o-

n e

rr r-

I -

H(T)0001
I-

CO r- <-.

1^

i-

0)

rt -

CO
QO

PrH

O P-

CO
rH

CO

Ol O P*. CO
X
rH

1^

CO CO P o
CN -.

O.
fO -f

< p p

p- co co p* r
^ CN CO
<-

p- p

--

P-

o c

CN

P- i*
rH

=. c o

CO

rH

rT

PO

-0
r-

rH

CO
-r

Ph

P-
rH

rr

r- O

r^ O P*
CO

riNMsoiO
CO
r- ^<

l~

^.

O P l^

"
CO

JO r*.

o CO CO P*
M
r- CN

FN

p*

r^. p- CO P
CO
CN

O CO CO CO
rH

r- P^ o

O P P* P-.
CO
CO

rH

P rH

-h

eo

CO

r- o r~

rH

P* rl Ol CO

c o o
m

p- C

r-

-t

t-

CO

c<

eo
p n

T-

CO

r* ^i o ^->

ci

rH
^-

rrH

Ol
r*

CM

cn

ro r* Oi --
cn
^- r-

r
34 (O

co o r r-

o o r*

o eo o

r* o

eo

o ~

p- c

p- r- o . 01

X * r-t r-i

O c
cn

p- ~ o

CN CO

r* co r~

ro r-

-^

CO

CO P00

r- t-
CO

(C

t-

CO

CI

CO
CI

p-

r- O

a ro p-

p*
10

Ol

CO

CO

00

r-

eo o

CO

1~

r*

p*

CO

-r

o
CO

P* CO o

31
CN

p-
X

co r* r- p,
_
tO -H

31

O CO

PCO

rH
rH

Ol P-
CO CO
rH

eo o <

CO

eo p p- p*

CN

r-

O ~. ro

p. o

P- Tj<

CO

o
o

cn

eo to

m*

on t^
CO

.- CO

lO

rH

-T

Ci |>

I-

3-,

i-

CN

P-

rH

P- O

P- CO P*

t-

CO

rH

CO

rn
r-

Ol
rH

CO

P-

CO

o
C r p-

r.

r- eo

CO CO r- P*.
c* -h
r-

r- co t CO

O O

CO t

eo
o

r eo rc
p. o

t- r~ o

r-- eo
eo

N r- N O O

O c-

eo c r*

r* eo O

o r* co
m

C 1*

eo si r* p-
r- ci
'.
00

a o

-s

O >. CO
eo o

rH

I-

CI

co O I-
CO eo

r CO CO
CN
-h

CO 1"^

[-

CO

P*

1^ o
CD CN

r^

p*

CO

p-

C P* p-

eo

C)

CO P- o

tO

i~-

-f

r- O c: o

I^

CO
r-

r*

rH

rH

P O

r-.

CrH

-,

t-

P* rH
CO

P, tO
r- P-

T-.

I-

CN

CO

ri * MS o ?;
rr.

1^

I -

a.
a

CO o

CO
r-

CO
-f

i CO Q

1- r~.

i^

pCO

-M
eo

P-

CO

1- c
rH

p- O
P*

r*

rH

rH r O r- Cl
^<
hCI

Ol CO en rH t-

HWNOMC1

.o cr cr

oo ao oo oo ei ci

p-

CO

r-

00 -*

r -

i'

eo r- C
00 O

r. r*
CO

on

Ol CO cs
71 <ri co co

00
00

CO

i-

l-H
H

o r-

CO

P>

at

rH

CO

C. PCN

00
o r* r

r- r* r*

.- *

co r* o
-h
eo

3 M

oo
i- .o

a u e <r i^ i^

qo ac ci 3i s: ^i

TO 0)

f-* f'- i-C

rH .-I 0<l

HNHCON
"H- "* VO W5 lO

o is eo *~
I*.

rs

o rs rs rs

as

n a> o o n
cm eo
m

CO t
O CO

O ^ Is, o CO
^t *
CO

rs rs m co rs

o o -< ~ o

r- co o

OOCiCi ^
to
c r*

eo rs rs co rs

o t- - r*

oi rs ^h rs co

rs

o eo o
rH

r* --" oi
m
m *

onhoo

CO

H -

O O Ol
IO
tfl

rs co -> rs

O CO
rs

co O rs

o o

o rs o

o o rs m

CO M

Ci ci n

m to

03

ts rs rs eo
m eo C

is -. CO
tO CO M

rs r- ts ts
M to

O ts CO CO m
-to-*

ts eo
CO

rs o co
to

n jihooi
m co eo
to

o m rs.

O O ts i o

o rs co

o eo m
cs to

m rs rs

CO ts O

us
eo o rs m is

rs

Ol

00

n n

Ol CN tO M

o o o rs

O Ol Is O

OJ
*

o
co

O o r-

Cl Is CO O ts

r <- o m eo
*
m m

rs ts o o
O

o eo oi co
m
m ct

CO O* O Ol

-<
M C> O O O
C* rs

r* eo

(^

r eo cti o

r> to

o eo

cn

5
co

r is co o
CO
^1

CM Cl CI

CO *- m

Ol O fs CO
ON
tO

- m rs -i oi o
to m m <r m

m eo eo rs

-- o cn
o

to

s*
o r oi o

O CO

co o r- o

m rs o o ts

CO

r r*> co
<-t
r*

rs o ro Oi cs
ci *f n o

MM)

r* O O

CO CO !0
Cs ts: d

oi rs co o is
ci
rs

oi co r* o

CO rs co o o

CO

ts rs eo

CO m

rs o oi oi rs
c* ao

m e mi

is O
CT
M

co Oi rs
C4 m
tO

Ol ts CO
CN
*Q)

Ol ts O CO O
Is,

ts ts rs

eo m mj
m eo mj

eo ts -
tO
m

rs eo rs
eo e* m

rs

-f-

CO
CO
CO

M Mf

tn < rs
; m
CO

ts o rs

m oi o rs co
MM

C*

CO
eo rs o
to

CO
rH

CO

CO o
rs
m
rs o m eo

Cs O O Ol ro

- rs o
eo m m

co rs

is ts ci r-
m Ol tC m

rs oi m is

o t CO

Ol Ol CO o
CI <-t tO

eo rs o

eo o rs -.
m
eo

m rs co rs.

O CO Ch
MJ

rs rs eo oi
CM

is. m is oi co
-* _
CO <-l

o ~ ^- r. o
o ^

o m o rs ts

e rs,

oi ts ts ts m
to eo
eo m

cn oi rs
o

m eo
m
rs o

rs eo o ts ts

<f

Is Ol o
M C*

co eo m co
m cm
co rs

O rs o

o rs tn

oi rs -h . o
e
~

o is. oi rs
c M

rs

o o rs o

CO Ol CO

rs oi m
o ts.

r* r* n

i- -^ ^. o rs

(o

(s, O *-
<-t eo
eo

rs co o rs oi

r- t~- O
to

ci o

ts ts CO Is

~ r* co

eo i>. co o r
CO
rs
+

o co r- rs oi

co O rs o r*

o -* rs o

O r- 0>
** CI -f

CO

- n is a>
* cn
m

CO - oi CO
M eo
MM*

r* is rs

m is, r* r

N n Ol O O
(OhO

r> c*

mm

pi o o> r*

eo

to m

I CO

M tO

is o is rs
A

O
rH

m eo

-*

ts ts

to

o eo rs eo
m

wm

c^

Ol

eo m c*

ts co rs rs
,r

Ol Ol O O CO
M C*
M

rs

m o rs o
o

o rs is ov

o o

is. o -* co ts

O r- ts CO o
M
C*

O CO

0 M C< M

rs n o ts
to

O Ol -*
to

O Ol
M
CO

M M CO
tO
m
Cl

eo rs

o rs rs o oi
tO
ci

m o o ts o

oi eo eo

is O ts m

o O O m

is eo o

rs m o eo
m
e

o is eo m ci
e m ci

CO t- ts.
* M

<- rs rs oi

o is o rs

co <- cr

M tO

O
H

ts r- CO

tO M *

nS;

M tO

oi is, o eo

CO
rH
CO
1H

o
00

-.tO

T-l

o
CO

CO
CM

r* - oi ts

rs rs o

o>
o

ts is, o CO

o o c r*
r* eo ^

r rs co
M

U3
rH
CM
rH

Ch r ^- r
c*

M IO

tO

o to o i>

S!
S

CO CO o

s
s

co o r co m
i-i

rs o rs co
<o M
o n w r fh
M <f

m is o oi rs
m
tO

eo rs o
00
**

CO O

^o rs rs

c* o
i^ ~ o

is Is Ol

M CO

CO oi

o> r* o

Ol fs Ol CO

co M rs *

^H

C. ts rs o o
m
U3

COo
rs co o co
m Ol m
CO CD

Ng

rs ^ is rs eo
-*

CO.

-J-

o eo oi eo m
Hi

o>
rs o O O O

r o> r* o
. C ^H

r* co
>o
o

eo - O

co o ts rs o ts

mm

tO

o r-i cs ~

i> O co

^t ^N

rs o r* ts ts ~*
sO

tO M

00

M * O

CO
cm

a - r n o
~ -f o

o rs o a rs
to
ao

co o r*

M CO ST o

r -< r t*

o i- o r-

o co ts o o

is o o
M
o M

CO
<- t CO Ol

eo to -* -

ts cj co rs ts
O C* M

O O O Ci

eo o m
rs
co

m rs o eo co o

CO
CO

CO
O O Cf) o

M M

m m o

co

rs c> o eo
m
o

r- c- r o oi

c - - a
CM
CO - a

c- ~. o
mcd
m

O t>. P5 1- is
- n ci
^

OiOOHH

CO

eo

rs oi m rs
m co to

CO CO rs

3^
-ej (O tO
^

NON-M

o ~ o

o co rs ro o rs

OJ 05 CI ^H t^
m M 94 90 CO

^ ^tp ^" *Jt i(0

g
HNHWI>
o

mi

mi

<- r* n

rs oi co rs o

CO Ol Ol o

CO

<-< CO rs

M CO M

-*

Cl < ts i-H CQ
lOOOtsN

CO
CO

CO

CO

^t

CO o

co co m rs

IO

M O M

ts CS CO Cl mi ts
ts, rs> x 00 Cl O

ts rs o o ts

CO CO
CM M

O O Cl
-r
m

rs ts o eo eo
* eo
m v

m ts O

<-i CO rs Cl CO

Co mi rs. pi co rs, Cl CO
mi cm m ro co

- ts Ol

m to

CN m CM

eo Ol rs .

rs

OO O

27000 36000
0 ps

i> o n c a o

rs is ft o

a
- is ps

to is. o a is

- o o
o

ts 01 po is O
eo i^
.

O r- O ft is
C* T

CO

z ts n

O o ps m ts
ts m* *

ft r ft
ui
r* PS

fs rs o O

ft o r o n ts

o a ts

ts O ts

ts ft rs

ft

a is r
ps ro

r> ts to
PS
to -*

a ts
l> M

O PS ~ ts CO PS

ps ft Pi rs t> ps
r* oo t>-
^h

OS

fs ft O O 2 r-

PS f* o PJ
w* m
ir
i^

n - o

Q ft ts o i a

r>. ft o r^ r

O ts ft PS
Pi ~
CO

IS PS Is
d
*n *

ts PS s

<- O ts ts O

Cs ft O ft

rs ps ~* ps ts
c o

ft PS tS M
CJ -f
00

PS ft CO o

i ts fs ft ts ft
CO CI -Jj
O*

PS Is

O Is ft
CO

ft PS PS ts
tr. c

ft PS ts ts

r^ M P5 IS
4> r- *fl

is m is ft r

PS m ts
O* iO PS

is O ^

O ts ts fs

O PS ft

^- m

to o

O i PS ft

<o *

PS ts o r* o

ts ft PS o

O tS -H PS
~ CO

ts PS ft -^ ts
T

ft p; is o
M m* M

ft PS Is
e ts

t- i- a

Is o PS

ts (VS IS ft ft
to CN
O *J

O ts

O O PS ts o

ts ts 1

t* O ts -H
CO

ft p: ps

PS -i is
T
OS

ts O ~ ft

is . ts O ft
ph CO
o

O l w - o
CM

ts m ^

ts ts CO PS

Is O ft

O ts

PS

O -H O ft ft
^
O PS

ft is ts P3

PS O O ts

O O O PS

OS

iQ _

CO
U0

5
5
5!
00
CO

c- r ft ft

~ PS

^i PS

ft is ts O

c*

CN

.-. c*

^h

ts fs ft ~
-H tS P)

CO
o o

e ps rs ~ is

-.

O r- ft ft
m so

to ^ f#J O ts

ft ft ts ts r

ps ps o ps is
C* CO

ft PS O

is CO
* CO CO

PS ts O ts

PS

ts ts rs ts

ts PS O ts
^

ft IS fs ^1

tS ts ft

O ^ ts PS
.
ts

r* -i O ps

CO ft I> ft ft
ps
e is

O ts -o ^

- PS ts
PS
o

ft ps rs
ts Is.

O PS o

ts O .- tots

O PS ts

O PS PS

ts ps O O PS
CN
**

to CO

-* PH

I- CI

CO
i- r
O
~

ts ft ps o -* rs
"lO w
r*

* r r* r*
PS

PS

<> IQ

CO

-- *en t-

o o> o

f> O a

a ts rt oi n

m
f-

C-i o

- O rt - ! b

i-< PS

CO
ft is O

ft ps t n o
CO

ri

CO

-*

t0

<N

p-

ft

ci a
- s

~ ps ts o n

c - o
ps

ts ~ ft o
_ ~* ts

PS ft o

^ * <N

ts Is ft ts O

r: rs

i Ci ts CO is

r<- o o is

ts l-> ft PS
PS
CO ~

J> - PS ts ft
h in ii ji

PS PS ts to
CI
ts CO

(O PS o <
tO ts
P-

ft PS ts PS

ft o r O ts o
CO
-r

c* r* is

ft PS PS ft o
H rt lO W

O ts ts PS

O ts

ts

PS O ts

OOhON

PS ft O i

ft O O ft

ts O ts

ft ts ts O rt
CO tO -h
w*

ft PS t>. >

c*

<-

(0 -

ci a
- O PS
(0
M

is rs

os o-otor*

*H PS

i-t

ON M) ft

PS O O > ft
CO ^
CO d

ps O O r-

p*

r* fs . ro
^-

O O r ft p)
o

=-. c n

ps ts a

O h N bOl
H I'.tfi
CM

ps
r rt

ts rs o
- CJ

p* *

lO

00

>

P4 *s

to

00
O
2 1- r--

ts r. ps o

~ PS PS

CI

-H IS O ft
PS
-<

rs -. (>. o
PS

O PS Is C7i O

PS tS PS PS ts
!,Is f4
PS

O is O is

O r Olf>
tO ^H

O ts PS ft
f-.
ts

O PS PS

r^ _ ^ rs o
O PS
**

is ts o o

o p: ts ts
P
ts PS

ft O 1 IS ft

O ft PS ts r-l
- C* ^ PS

ts ts ft P-

ts

r^ ts o
0*

n0
Ng

to

c*

in

co ps

ts Ol O ts ft

-- PS O is
~4 PI

PS ft PS ts
(M >* -t

ts O O ft
tO

ts ts r> ts
to
CS

is O ft ft

ts O ft
PS

PS ts ft

PS P-. ft

* C C*

i O O ts

o
PS Is O

O O ts
ts

n h ms

PS - -t

O O i ts o

PS fs t> ft ts ts

O ft O PS ts

PS ft ft

^- ts o O ft

rs ps O ts

Is is O O O -H
PS

O Is (S
O ft

PS ts o o

Is CO Is t
qm ro ~*

ft Is

ft is ~- P-
>m
to PS

is O ~ ~
in
m to

O ! PS CS ft

is ts Is Is
O
rf PS

ft Is -. ft
C*
<-. rH

-H tS ^ PS
ts
ri O

O O ts PS

O ts ps rs

-^ ts ft

CO

PS ft ts H

_< c* * to

ts O ft Is PS
ts tO
CM

ts ts PS

tS O O r*

PS ft Is
r-l
U] H

ts PS ft ft
~r t -

t- tS PH ^

ts ts PS

ts ft PS PS
r- ts p-

O -h ts O ts
-t
ft

rs o O
PS

O O PS ft is

to

TS ^) H Is H

co ts as co os
CM CM CM CO CO

HNHPi N
^TplOUSUS

O CO OS r-( ts
Hi CO sC l>t>

rH co rs as ro os

OOHHN

CO

o>
- a ts

a ps ps

O Is o

o n is pi ft
cm .-.
ts -.

r* r* i ps
~
CO

o -- w a ro
CO

ps r* ps ft r ts
t*
m i- O f

o o n on h
O

co
o>
o
o>

ps ps r o

w*

* PS FN

e ~ *

WS>

00
o a is

ft n ft t^ n

mom
cn

m
r

ts

00
ts. o o

a o

ts to rs rs
~

*o -*
r- o rs o

ooMnoo
1-t
00
PS t o t* o
<-H PS

a r O

fs o ps

c> - is.
*r iO O

r* o rs rs
C X

r- ft ^ r*
PS

is n r - ft o
o c*
c ei

m -+ O ft
co
r*
o

-< CO />> OS CO OS

Hr*HP)toi
co / - r. *. :j

00

^ c*

oo oo co oo cs a

O co
< Oi

r-

r.
ci

CO

o r c
H

**

o *+

oo

01

co

in

**

co

O)

i^ r- r~
o en

a;

I~

f* CO

3.
X

CO

l>

r.

rj

t-

to

- C! ft
y# <n *n

c*

t^- CO
r-. tC

co

Oi ~ CO
-H -H

o O r*

r*

l-

r O

O !>

CO

i^

CM

CO

CJ. CO

I"

i-

r* r*

r* o

co

t^

"

O O ~ CO

t- t>. i f.
-r *->

CO

r* o o

i~

55
CO

CO CO Cl
^- Tf CN

In r- M h. O
CO tC ^< Ol

co r*
to

r c

co

-r

01

31 r- O O

CO -^ O t^ O
lO 00

pi t^ .

CO
*-

CO

- .
c

i^

cm
co
o

co
c*

i^

Ol t^ o
CO

I- O t>- o

CO

10

CO Cf)
00 to

i^

to

t^

CO

1-*.

a
o

(O CI

CO

>

CO O
I-

O fO

CO

oi co ^- r>- o
CO
CO

CO

~-

t Ol Ol
CO

r^ .

O r^. r- co
-* ^- CO CN

OflN

O 1- i

t O CO ~

Cl

ft O O O t

co oi r-

O " CO t^ CO
!-, _ ci

tvp) r o rt o

CO

CM

en r

co

Oi

i-

i- r d t^.

o ^~

CO
Cl

p- CO

CO

n r* o

o c

r* co
CM

i- O
CO
CM

co

CO

ci

CO

CD
CM

r co r- r>.
CN c*

r~ o

CO

^ co

-H l-

00

Cl

CO CO

C4

--

o c*

a r

-=

CO t> r^

rf

rt ^- r-

I- Ol

r*

^t e*

9a

CO

CO

r*

a
c

c^

CO

Cl

O r^

T-4
CM

CO

Oi r~
CN
p-

CM
Cl

[-

co

in

O CTi

l^ CO Oi CO
00

O tCO

<#

t-- .- CO
*+ tN

t- t>. o

l>. 1^

i-

. r-l tTi Ifl -^

CO

t-t
Oi r
pi -J-

co
cn -

i-

co
00

I-

CO

r- eo ~h

ci

C7l
CM

i^

CO r~
t

a, o co

l> O

1-

i-

CO

o r-

CO

r> i>. o co O

o r-- ^*
co
**

CO o

ai r* o

r- O O O

r- pi o

t> O Ol i CO
~- --. <

hnO-O

CO O r-

-h ci r o r>.

r* oi
to
*# t-

i^a*

r- [s n r.
c. -f r-

P5 i>, i CO
CN (O
CO tO

CO

fO Ol o o

i- o l
CO

O CO
CN

Ol O Cl
Cl
o
-o

o r-

Cl o -.
r*

CO o o

1-1

r-l

a m

1-

to

00

i^ ^ ^H
CO

O I I"*
Ok

CO
CI

t-

i>

Cl

co rc*

**

r- CO
CO o

.- KJ

o
t^

t-

or*

O CO CO

r*

CO

o
31
CO I-

9K i-.
PI

o r*

c-

"g

to

Oi CO
Of -H

CD

I-

I-- O
CO

CO

o
l-^ f-

CI

O CO
O

CO

o
co

CO O

C) CO

i-*

CO

.
X

*+

O t-

CO
Cl

^o
CO

i-

co
CI

Cl

co Ol
*f pi

i-

co o

CD

*i CO

Oi

i*

1*

r-

CO

~.

a
o

CO

c^
0

t>

t> 1- o
CO
to

O Ol Ol t* l>
in co
*r

CO
Cl

CO

00 T

o t-
(O

CJ Ol
H ^-

r>

Oi o

t^ ~H I-. |>.
ift Cl CO

o - o

l-

ci r> i^ oi
Ol H

ro r>. ci co .
^o o

oi r o
00 ~

o co r> r

r- r* co

O O l

CO

i^

1^.

cn
CO

00

CM
i-l

t- [- ~%o CO CO

CO eo
CO CN O

i^

c.

CO

1^

I~

CO

c*

en
o

r-

1^

!> CO

(33

i>

i"*.

j.

r>.

=-

3
CO

1^

OO
i CO
oi t-

<*

CO

t-^

co r*
CO

ro

ro

I-

Cl

to
CO

p 10

00

r* co co
o

t t

co

10

CO

CO

3
00

CO

CO

CO

00
t-

e m

a.

o o

CO

00

CO
00

O o
C>

co

t-

cn

r^

rt

i^

CO

t>. C

^-

t-

CO
CI

co

i-~

t^ C IC

o w

01 CO o

r-

O CO o

i^

B* Q

_-. CO
O CO

Cl
3

O O CO

i^

o o

l"

a r*

CO

c-

CO

CO

i-~

.-. o

CI

1 -

CO

CO
O

CO
-T

CO

ci r- -h
tO Cl
(D CO

r* co r* co t
nntb

Ol -i Oi
tO ~-

Ol Ol

-' CN

r
^ a

to

r- o

i co r>. r-

O t^

~ CO

CO

CO CO t-

O O ^ Ci Ol

!> O t^

t^ l> t

CO Ol CO Ol

r^ o *-*

Ci c r co co
00 to
c*

Ol l- CO In.

xk n

co oi o r*

Ol t- O CO
*
CN

-^ co r- oi co
o o o o -

0lHI>r-CO
r-t Cq CM JO CO

00
1~

to

Id

CO

CD

0l

r>.
CO
CD

co
to

r-

to

I-

(JO

i CO

t-*

z.

o
CO

-
CO
CD

I"

CO

CI

r- r* r- r*
~ -r

o>

CO

co CO

CO

1^-

r*

p-

CO

t o

cog

O O H t-H pH

cO r* i- O
CO

CO
CN

t*

CO

CO
CO CO * t* CO t* 31 CO
r- w ri co n 5H tP ^ -^ kO

I-

CO

O
O

CO ;>

(Jl _. t^. _ co
kQ CT1 CC t^ t^-

t^ 3i CO 31 <
1?- t^ X X 31

I2i

CO CO -*.

36000 45000
is

CO Is 0) O

co
ro rs

O is o

- eo

O Ci

ts is ts

eo

*o

p- ts
P"
V
CN

O) O r* ci t> f.
O
-.
o

CD

I" O

is co
C ^
i-i CO

CO

r> o r> o co
CN
CM

O Is O CO

o rs co rs ~h

eo

tO

c^

eo is

is
CO

-5

n co ts. co

ts

JO I- Oi pp-
rs

o o

i-

ts

CO o
Ci
*

CO
M3 P-. ^Ji

PH
CO

IS

ts O

2S

CN

CNts eji is.Ci eo

ts O
PH tO

Is

*S O . ts

is ts CO

o o o o
o

eo co is .
CN ts
o Ci

ts 1- O ts
_ p-. (O

Q CO p* l^.
CO
"

CO is JO
1^.
[<)-

O CO o
c*

Ci

n ts

c: o> eo

tS ts O
CN
~
is

ts eo

^ o

ts ts CO O
C

C, Is i- JO
-r pCM

CO *

<- CO ts.

is JO ts
is

is Ci CO i~

is

Is. , o
t-

"

!> M l- Cft
^
r- CO 30

ts 9) ts o

n r- oi

CO

eo Ci

at
CO

1 -

o o

is

CO c
CM

ts PCM

ts eo

rs ph

IS

ts

IS C IS
PH CN

C"

P-"

CO
r- (>.
o

rO I* M t^.
CO p- eO

m co f C o

ro
o CN

CO
e o r

rs o

- o c

n -.
p. (O

ts o co o> *o

M l cm

'-

t^ C^

<c

ts, o o p-

r*

O -* rs

CO o o r-
c*

c.

O Is is <-

CO r: ts
CN ts

C is ts Is
p- o

ts ~

00

OO-COO)
eo

is n is

is

eo co ~

?l

i -

r-l
CO

I"*

eo rs. o

^ C N O C 1-

c*

00
C3

CM

tO *f IO

O CO p-

t |s

CO

o o
to

en

eo rs
** v to

i~

C CO

at
3)

is Z.

C is is

eo

C4
coo

C rs CO O

l> o n o o to
<-

05

eo o ts> is
X r.

eo t o r*>
-r

CD

Ci > r
o
to

eo CM

Tf PH

ro is c
q*

-r

c is o rs
to

ts CO Oi ts

o o

is ts i- eo
^- ^

y\ is
CM

ts
0 O ts

ts ts

O CO CO
ts

r-

I,

# PH

O Ci
CI

r: i-

ts Is

ts

o
<0

ts ts

is

CO

O ts

c* ~ ~+

CI

tc

Is

ts

1-1

o t

ro m ro eo
ts o
^

r* r*. oi r
- t^

r-ro ts - tors oooi

i- co r* o

o
r-4

CM

oo n

CO
T

eo ci co ts

CO -

p- ts CO CO
x in
CO

rs ts to co

CO

is

ts PH O
CO p- >

ts eo

i^ ~j is r:

Is
tO -N

is C, CO
CO p*

ts O
* CM

s O O

O O 1- p-

CO

-4 Oi ts CO

^H

I - eo eo

s.

to

eo ts O

Is

is

ts O

c-

C.

Is Is O

eo
01 -

i-

o o
to

Ci -H
Ifi
PH

O p-

O
rs c o

O r* r t^ O)
00 eo
* *

CO ~ t* ~

ci *

ts H rs < eo
-z, -. ~

r-
<#
^
o

C r- e)
CI CD

r- ts O O

CO

r- co r-

CO O p-

Oi Is
to cm
^

to r- r co r* o

ro eo eo rs

c?> r o*

~ rs o

ro is
<4

A -*

C t t> is
M
^
M

*s
nS

to

z.

CO

Ci

r^

ts IS

O eo

is rs

CM

o o
ts

Is ts
CO

CO

^- o
to

eo

is I-
p- eo

eo rs

ts ^ eo

1^ 1^ CO

I- Is

ts ts CO

1C3

o
a

CO

CO O ts o

ts rs

1-

ts |H
Tj-

is

o o

ts O

ts

CO

J^

r.

CO

CO

~ CN
1^

ci r^ o o r^ c
an
Oi

CM

o eo ts

C*

O t- Ci

O ts

p- ic

CO

CO
tftO

ts, o is co

< r>. t>- o co -^

rm ts,

n a t*

CO
00

ro rs, ts. o
00
o

o c " a t- o

r* o eo en to
* to

Cl CO

c n c
d

CO ts Oi p- <-"
CO 3 T

r- co o o ~ o

c c ts <M

CI CIO

is. o

<- *-

o
CO
00

CO tO

o
00

o ~ ts

t O

eo .-* ts
o . CO

-oaro n
>->
c*
e*

rs so
Ci
go

C C C ICO

r Q O

-^ r* ts.

00
CO

3 rs.
-* ifi iO

O ? Ol t> -" -

p-l 50 t* CI CO <3S
co ,0 c co i> t>

Ci
CM

Oi ts
C4

O O ts

ts eo

is -H tS JO
^ P, ,-, >.

Ci CO
O
pi

ts eo is
-t
to

ts pi CI ts CO

ts

Ci rs

C i - : )

CO

ts Pi

CN CO

is I- JO

C; ts rs

eo ts

CO ts Ci
-.
m CI

O ts

ro rs

ts

^- pt C*

a is
CO
p->

Is
c-

CO ts ts

O ts rs
CO ts

eo -- ts eo co

ts.

ph p4
CO -

eo ts eo o p-i

ts O CO ts O
CO

Ci

ts

Cl

CO
C

is

p-

PH P-. Oi <I3 Is

O O ts

W IS p- Ci

^h CN O

to

p- eo

[* O O l> In n
j

*a

r*
ng

p- ts Ci ts

H r^ ^ co r* o>
00 GO C71 OS Oi Ji

CO OB i-h l> i-(

COrsOCCOl
CJ CM CI CO CO

co
t- m

i^

*
t-

C 1- -

c*

O O i-H i-t W

is

i-h rs. p-t CO rs.

IS,

eo o

Is Ci O
n Is

ts O

ts <0
d

CO t>

rj

eo
p-l

ts eo

to

1*

Is

ta
ts eo

cji co c*j pH rs
Ifl C tt ts ts

JO

i-h co rs ci co tj

co eo is Q
CD *
m

C M (ft Is

o o ~ t>
co

C ro ts c

CO

^- rt - Cft i>
C fl [ o
CI

i^. i-* In o

o n <n o

r- ir. i- IOl CN

Ol C CO
P4
M
CN

CO Ol Ol Ol
A
i* 'J. -r t

o eo eo o

- r o n i>

IN Ol
-
rN

In CO CO O

CN

l> f-

C. 1^ l~

^- O in eo

Ol O CO
wm

pm

,-.

^>

CM
CO

CO
t- C

00
CM

co is eo
m

CN

Cl CO Is
CM (O
CO

l- o

r- Oi Is
~
m

Ol O Ol Cl
ts
e* m

r- Ci cn

r*o

o n

CM
Cl

tN o - l>
CO
i-i CO

1^ l-~ o
Ct
>M

ts ts

CO

co Oi

Ft

CN ~
CN

CO
CM

~ cn
m
is

tf. ~ _I"*

O) eo eo is
cn
m cm

Oi is o tN i

CO

> IN CO

CO M

1-4

to

CO
CN

In CO IN Ol

cm

to ri t> o

r- l- "!

In Oi eo
-.
M

is is ro
CN

tN CO Ol
tN US

oi eo o oi in
O M
Ol M

<- eo

ro tN CO In
CO
' c

Ol IS tN

Ol IN,
CO
CO

in r*
Ol .o

O m is ~ ro
-1
f

ts
s? M

is Is.

CO

CO

CM
CM

CO
CO

n o n n
.-4 ^

OJ
CO

-> Cn O O In
tO CM

CO IN tN O
CN
i0
CN
-.

o oi cn r m
3 CN
r-
CN

ci r- ro
M
to

r-. O tN
CO
_
_

1^ O In In
CO ^

CO ^- CO In CO
CO Ol
p- mt O

CO
CM

CO m ^- ^ "tN" M~0~ O ts"

CN T C0
i-t
o

ro o co oi

CO o
M

tO

O CO O. In
OlMH

i-

* in

Ol Is IS.
Cl
M

ci in o n
^f
eo
ii

O O tN M
tC

M M tS

CO Ol Ol In O
i- c*

O O CO In

i* cn
eo o

tN r-l ^-. O
~

O Ol

O IN,

In CO In Ol
CO ~
In

ts Is Oi O
~
CN

o M
^ Ol
o>

eo O CM i-i

Cl O I- r. O
CO

CN

O tN ro O

Is Oi O CO ts

I-

i> eo

CM
tN.
rH

y-4

OS oi tN
- CN CO m

r* o t> m **
m

**

ro in r
to
~
r* * ** bs
to

n o o m
O !> - Ol c

Ol -*

CO
CN
-<
CN

CO IN CO IN Ol IN
-
CN V i-i CO

CO n In o
Ol CN

tN ro eo Oi
*
HlS

1-t

K t> O - O

no

CO

oi o m

o
in O m O m
CN
hMM

.-.

-i r eo

Ol -H tN

o * m

In Cl Ol IN
CN

1-H

Oi

tQ

cn

Ol Is ro
CO
CN

c o o o

Oi
o CO

is eo a

!-. ~ e ^ o

CO

h o in a> o

CO CO
n CO

o a In " r: r

eo o

CO tN tN CO
IN
*
Q

o h n b o h
CO

CO

t- _ CO Ol
HIO
CM **

eo o ~
Ol

r- MN
r-J.

Ol [N H Ol O)

O ts CO

IN M

^J-

r-

CN

p-I

*C5

ci ts i co

en o
CN

r* o o to -

O Ol In Ol Ol
C4
J-"

<-

p: c c> i->
CO
m
o

K r- 1* O
4> CN -h

IN C CI Ol IN

O co ro o

O m

CO

Ol In

ro c c-o i - w

Oi ~ o o

CO

Ol

CN -i

CO

tN tN CO
*
MM

T. 1- C C
* CO

Ol in rM
CN

Is ~ C

O Cl o

!-. I- T,
M M tN

%jf tN

-i

CO

CO

89
912

Q O O *
CN

i CN

CO

Is

CO
GO

Oi O C. -i
!>,
tO *

K I- M 1- Is
M
CO

o r- o ~ o

Is CO CO ^
^Is TO

C Is ~ ts

1" 1>- O -'

CN
-h

CO "~" CO CO

in r* < o co

o
o

CO

CO ft tN

oi eo >
CO CO CN

eo CO in is

ro ts

CO i to

M CO

Ol In
CM Ol

Is ^- Ol Oi Is

* ts M

ci ro
O Oi

CO CO ro

is tN
M
CN

rs t*. m
eo
co

eo in o in
reo

1^ CO ts

"

I O is O

Oi

In In.
CO

co co ro oi
CO In *

ts r*
Ci

IN f -O

tN. IN
tO tN

CO CO
M tS M

Ol fN tN
CN .- CN

CO In ro
CM
M

Cl

o. C C Cl

CN

tN _ Is f>.

Cl

In Ol r: Ol M
O I"- K

IN . CO
M
Oi M

Cl ts i-i

ro tN co in

CO Is Ci
CO M M

ro Ci

PO In Is
CO
M

m CO o o
l;
CO

ts Cl

tN eo co Oi in
CN V In Ci

o o ~

O t~

Ml CO in Ci CO

Oi , ts, n co

n. Jj co

oooo

mi CM CM CO CO

CO CO N-f

GO
1- O 1

CO
CO

co m is
CN ' V

^H Is
~ ~ M ~
CM

O O M t^ O
CN ->

m O CO -*

i'. t- CS Ol
CN
CN

CO

**

in t% - o n
CO CO ~

Is, o _ t^

CO
CO CO
- i- m

- O In Ol S l-

eo

co

CO
a

IN Is (S o

Is ro Ol
r- M tO

i- m r--

is eo e eo

^- r- m i-

CO I- l CO ^
*0
CO CN -

C I*
tN

tN tN Ol
CO
CO CM

In Ol CO CO
Ol IN O ^

tN -i CO In

eo ~ oi oi

CO
1>
3

eo

to

tN.
Ol In o ts

r^ O

"

Cl

IN CO t

CN

C4

tN.
is co

in c i-

CO
CO
tN o eo
Ol CO

m o i> o n
-i
. is

Oi Ol i-t o
CO PI t- r-

r- o o oi

r- I^ Ol CO

00

O ~ O In O

o in co eo co
_ CN

c
co

In CO In tN
rr ro

CO

In

co

CO
CM
CO

is r* ** i co
^m eo

ts co
"
CO

CO - CO P* CO
CO
CN

p. CO c c
to
r-.

i^. r* tN

n *

to

CO

co

CO

tso. cmoi en i mi-

* M H o

> ts m

I- fO P5 t~*

CO

n o m

r^ a. m w

to

tN o

O CO t "

PS

CO O tN tN
CN
~

tN oi O

tN H tN O

ct *

O Ol CO

o
co

CN

-<

K3
3i c i" eo o
CO

CO

r*. i>
CO O

In I- Ol In

eo co Ci
CM m
mm
"3

-C QON

r-.
:-: r ci
csi

ro ^ r^ oi
t>. CO

In O CO fr>

r-^ CO t- 5i CO
TJ1 Tj ^ -^ L5

CI - l H
IC to CO In t^

O o - In eo
to
-. to ^"

*iq

O O ~ r- m.

Qi CO Qi i^ t**
^- cm rt cc

In. OS CO Cl I In
l> tNOC <X> Ci Ci

*3

45000 54000
ONO

or*

tO - O rt

o O CO

O O
O rt

ts CO CO Oi
rt ts
tO ts

O ts Oi rt is

-OO OO)

r*. o

IS rt CO O

CO JO ts i CO
CN O
Oi rt

O ts ts CO
CO o

O O) H -

r- r* rt

rt ts O Oi
*P
CO
C*
CI

is CO Ci
rt CO rt

rt i^ i> is rs
rt Oi
OtO

o r o r^.

r CO Oi r- rrt CO

Oi O ts ts
Ci
rt

ts ts O CO
O. ^
ts

ts Oi O Oi

-OrsOrt
rt
to
CI

CO is O
rt
rt ts

ts Ci O CO Oi

CO

Is ts ^ CO ts
-f
rt rt CO

Ci o> is o
C ts
CM ^

-- oi ts co Ci

- - rt O imrt CO ts
tf

SO O CO ts
If
^

O rt Oi ts
CI
rt

c* rt

CO Is CO rt
CN CO
rt
rt

CO
O O rs

r- 31

n O

o a* is rt ~

* co

*- O r-

co is

CO Is 1- rt
*->
CO CI
CI

c* ^

o r- rt

CO
CO

~ oi (* rs
rt rt )

c*

-m ~

CO
CO

<

CO

N O" O
CO

oioio rt o CO **
c*
>o

O o ts so

!> -t toH to-H

-. -i ^

r- o C- O

.*i

1^ ' -. -o

-*

02
so O O

CT. ~> O O rs Ci

CI

s rt]

rH

ts CO CO Ci CO
. o to

is ro
is

is o so r*
Ci ro
ci

co is ts

O co rs

O CO ro O
wa <-i

o co co r*

Ft t~- O CO

is to

r* r*

rs o

CO

to

00

o r- co o
o so ~

h" r* r* r*
*
CO rt

co t- r*
-.

oi r*.

is Oi Oi O so
* CO
rt

ts IS

Ci -: rt o o
CI rt rt rt

t* r t^ o

ts Oi rt rt
rt
CO rt

is ts CO

ts oi ?o
is CI

ts CO ts O
to

CO ts rt ts

Oi ts CO

Oi

to

r-

to r*>
CM

TJ1

CD

IH

C CI
CI

rt O t-
CO rt

m rs,

rt

to rr

rt

rt ^r

i-l

c - o
as

co

r- r- o r- ro
-< uo
C*

Oi t"" O CO

*0 ^H

ts rt t
CI
CO

rt CO O is O

rt rt O Oi
CO rt
rt

is fs Is O O jrj

r- co t^ co

CO O t rt CI
C
C rt Is

Oi O ts Oi
CD
CO is

ts Ol CO
CM CN
rt

cv is _~: ro
io
rs to rt

rt ts CO Is

rt to

rt

CI

o o so

r o* -^ ro rs
to 1 to o co

O O 3

Ci Ci

o * t* O r*

r* Cl CO CT)

c rt cn

rt

CO

CO

CI
CI

r- Oi rt

CO

i> o r-

to

-. -.

CO CO* so

CTi - O -

-^

* CO

rt is

SO

O is

oo

*r

o
ai e
a m
is ts

rs r oi t^ ts co
*> rt C*
d

so co r
to
m

rs ts
to

n O r- -

r> o

rs ~- rs
rs,
co
CO rt

eo

r r*.
A

i-i

CO Is

co rt is

rt rt CO
tO rt
rt

to

v* t* r- co

*s
^1
C3

o F- r>. c. t>O
CN CO

CI O SO o
rt
CI #
CM

CO rt

IS is I- rt IS
rt
U0 ts

Is |S C.
tO
rt rt

r* o so
c*
r-

i^. o o ro

ts i- o is Oi
Oi
rt

ts

t i, j*;
rt -J.

C O I', o

o co so r-
rt r*

Is ts CO CO o
tO
CI rt

rt rt

^- (^ t^
CO

n -- o n t>
IN ^

CD

<o

I-I

to

en
Ci o o
c

O O ts r-

L-

O O 1^ O I> o
t m

oo

CI
i-* r* o
CO

is r^ t>tn rt **

CO O Is

O rt o i*
rt
CO

ts J. CO Is
rt
rt

i* o a n h
r- CO

SO O O

o rt i rt a
Tf C* tO

is O 1- O rt
-t
CO

rt rt

Cl t>. I" CO
i
CO

t>a i> o> o

Oi ts co
rt
ts rt

is C0 o so
rt
CN

so o: SO fs

SO O ro

is ts

o ci ts
Oi
ts

Oi ts
Oi

SO Is |s rt ts
rt *# CO rt

O Is CI

rt Oi ts
rt C* rt

O ts co rt ts

O SO

00
C so rs

rt *

1 - O i - c. c

CO Oi is O CO

c*

00
5. Oi O
K is

r is rs

a r. a

so o so

i-i

tO

rt)

CD * d ^

. co co rs

so r-

C* MO

r- is A

tw o r-

O Ml t^ ft

rs h m o
to O
^"

Boaeob

to
r* o

<- r- is CO is Ci
tO
tO O

HfthOOM

ao n t> -
e*
ci
-

co r* r-

is rs

c ;o

CI
CO

en

ci

r*.

r* t- v*

IS, rt

tO O "*

.-

is

r * rt

i G*i

Is. ts o. o
CO
rt rt

is CO CO

co o is t- "

CI

l O SO rt
CO

rt ff CO CO
rt d US
r^

oi rt rt rs is
to SO CO
*

O ts

Oi co Oi

CO Is t> CO
C Oi
rt

co r- rt oi

CO Is ts

rt ts

Oi rt Oi
rt
CI

ts co ro rt

n t a ffi o
rt c* to

rt

is I-
Oi CO

r is o a o
mo

is r: co oi
o> to
ci

O Is

C. is

c < ts

CO O. rt o
(O rt tO

is CO

0 rt is

rt is CO

rt i* Oi O rs
rt
O

I- r- CO t

CO rt CO

is o o
CO *

Ci ts O Ci is
C*
C;

Ci Oi O I- o
!s
CI

rt r* rt o
^
CO
*-

Is is CO Is

ts |s Oi

t- rt O rs

Oi

C5 fs CJJ CO OS
CI C4 C CO CO

HNHCOts
* -# o \n lo

os co :r> r-1 ts
\a (o x ts is

rt* co rs os co ^i
00 00 00 CG OS OJ

co
o

CI SO

CO

rt rt |s

<~>

rt rt is fs O

co

a o o

c o ~

I> Q 00 I.

co rs co
c
d
-

CO r ro
H CH H

rt

to

r*. co !- r*

r* r^ o o
-

I- r- t>. i

rt ^

CO

CO rt co

to

CO
> C

r o r rs
to

CO

NM31H h n
ri
C* ^

00

rs r-
*
eo

# CO

on o M3 rt

CO -0 is
P4 <0

o i rs o> o
- CO * OI

SO CO t^.

CO J"s 05 CO Oi
CC tt) C C l- t"*

rH 1^. * |- Cl

WfflHb"H

3SS

X; v - i ~ r^ JJ

OOHHN

ro ji r is

rt r* e*
r- so cr> co
rt to CI o

c*

uo rt

rt to

en

t-

o>

t^ o

n c

i-

-2

-2

CM

|>

OJ

en

3i
CM

r-. i

Ci

Is
OJ

t^.

c.
rs

r*
n

Is
en

1-

O en
-r

t-

r-

en
Cl

o D

Cl

Cl

'"

f~
to

I-

Is en
Oi (H

CS

en

t^

i-

l~

cm

2
[^

CNt
l-l

is

Cl

i^

i-

i-

OJ

at

i'

CM

a a

rn

CM
Ci

r^ o

l^

Cl

c
C;

o o

to

m en

n
CM

n O

Cl

I-.

O
-r

*T

ci
en

r*.

ci
cm

m
m

i-

r* i>

en

!>.

a
Ci

r-

P)
Cl

m
00

en

t^

eo

en

Ci
f-

eo

CM
r-.

*+

to

en
C*

r^.

I>

en o

[^

Cl

or*

f
tN

en

i-

en
o

'^

r>-

en

i-

en
CO

ts

o o r-

r* o
to -*

en

CO

o n

en

i^

BQ

as

CM
CM

CO

(~

1^

en

eo

T-l

i -

eo c

^i

c o

r*

c o

CD
O

CM

en

i-

r*

r^

i -

PS

r*. im

r>.

Ci

en
to w

O r- m

.--.

o
n
M*

-* -

1*

ci

en
c

n o

CJ

en r*

r^-

|s

BO

''

Is

PI

t^ -H

en

e* ci

o
n n
CM

i-

en S P9

cm

ai

a a r^

i^

r^ CO
O)

-. o

PI

Cl

I--

C:

Oi r

*". en

t^.

Ci

zz

i~

o en

en

Cs

en o

-r

l>

a e

CJ

Cl

ci
ts ifl

r*.
to

r* *-.

-r

i- Is

t^

c.

^- en
** 10

CJ

en P9 n
a

en

r-

is O

en r

(}
f

D> t+>
io eo

CM

PS

PS

[^

l>

to

1". o
eo

en o

t>-

'

t^.

wg

C)

1^

i-

c<

CI

o en

tH
CD

n
a

i-

i*

OJ

r^

o a

O en

<->

ts

r^.

en o

-t

en
C

1-

I-

i-

i^

r^ en

PS

40

09

**

[^
CI

CI

at
10

i-

Cl

1^

OJ

Ol

00

t ^*

i-

en

at

CM

00
Bt

i-

Tf"

00
CM

t^

1-H

o r m
n

OS
O

oo

I -

CM

oS

1^

i
s

C!
n c.

m
o

en

<M

U5

CM

<N

00

00

i-

CN

-o
- ^ .;. -

r* o
CM

en

i- t>.

- _

- -

ci a

ts

en o

en

^
r"

o r-

en
P3

i^

-
~ en

CO
00

-r

c.

i^
is

c.

eo

en ci

m en

CM

i^

co
o

en

i-

en
c

***
o

-r

Cl
Cl

"

is

l-

Ol

U3
at

i-

Cl

l^

o o

t^

en o

Cl l>

t^ Ito

1Cl

i~

Cl

r en

r-

i--

t- ~

o o

o r*.

J.

m
BE

eh

o en

Ci

en

CM

t^

r-

CM

r-
en

i -

ts
-

i r* eo
Cl
-

PJ i^
Is

,
Cl

Ci
Cl

i*

(^

Cl

ts

en

i-

CO

r* t-

n
ts

c.

en

c.

P)

O
00

(- eo

9)
Cl

t^

to

n c

CO

r- ~
Cl ^
CM

c*

t>*

en

en

CD

eo
C4

00

00

en

Is

eo

00

--

en

f.

Cl
is

Cl

ts.

i>.

Ci

l>-

to

PS

PS

i-~

I-.

en

i-

o z,

en oi o r* c.

Cl

t*

i-

n a Q

t^ o

f* Cl

en ~fh

is

t-

c.

i~

ts

r- is

r-

eo

Cl
CM

t>. o

I<

r-

cj ts en

i-

PS

eo

is. O

Cl

.- o

en
CM

r.

ts

CJ

!>.

** o

00

co

c.

1~

i -

r-

.-

en

en

co
o

CJ

I-l
CO

00

a
m
c*

en

T
CO

*J

i^

r* o

en c*

m
M

o o r--

CO

OJ
CD

OD

CM

a o o

Cl

!*

o pi t*

I-

e n

- .-n c
a
o

i^

i-

-*

rs m

en en
CO

Is

[-.

'

I-

c.

r-

to

i-

en

i^

r>en

1^

c:

O
m o

Ci

:-.

en

Cl

c.
lO

IS

en

l-~

r Is

en

en

PS
-f

i-

cn

"

>o

CO

cz

Cl

CM

to

5
1^.

en
-r

en

I -

CD

31

o
o

Pi

1^

i^

C4

o c

en
o m

O I'-

t>.

(A

BQ

o
Cl

-1-

o
at

1-

to

i^

1en

I-

to
i-

i> en

--c"

CO

Cl

a 01

I -

en
X

CO l'-

1-

Cl

en

1~

P!

i-t

1^

'.-.

r* O

r-

en c.

01

i^

CO

<- CO t> 51 CO
** * ^ T1" tO

Ci ' I> ^
W > N

r- rs co
t^ !> ac

1^

I1-

rGO O

L~

r. en
en f

en

en

-r

u>3

I-to

<
3

ci r*

r- en
-* CM

i
^

1-* o SO
CM

CM

Cl

CO

C7i

ts

o ts a

eo

MP

>

CN

r- Oi co

co r* c?3 co

t-t oj <m ro co

CO CO Tp

54000-63000
i^o a
eo

o r* r* o or*
eo
ci

r*

i -*
-i to O
(O

r* co r

p* CO o

CO

r* r* o
eo -

CO

or* o

r* eo
eo eo

31

eo Ci o r r*

r*

CM
Q -.

eo O Oi CO r*
r*
. _
_

m*

r* -*
- r* eo

ro eci ^^ e?o r*
c* > c* r*

o P* eo Oi
i.O

or*

CO
CM

t~

00

(O OllS

co 0i
Cl Cl
Cl

r* co
eo

Ci

o o

pf r*

r*

Cl

eo

CO
C4

r* r* Cl o
ci eo

co eo
c-i to
CN

r* co r*
10 CO

o eo r*
eo
CN

r* r* r* r*

OlOO CA
IO
o

eo t* . r* o

r* o

oi co r*

co n o

o r* r* r*
o

Ci r*
Cl

O r*

CO
*

r*

P* CO
p*

-t-

CO

-^

Cl

cr.

n ci

o t* co r*

r* C

r* r* o ro o r*
ca
co
n

t*

- C.

eo o en r* eo
*
o

r* Oi r* CO

o r* r*

d Oi r*
n h tf
Cl

co r*
ci

O CO

eo p*

CO

P-

- O O

r* r* -o r* *o
c* r* r* o
c

r* ro

r* co so r* o
O Cl

Oi o

CO

o r* o o

CO
CN

r* O
eo

CO

ci

n oi o
r*

O O * r* O O

r* Oi
CO CO

r*
<o

r* :o ci

t* eo CM

Oi r*.
Cl

I*, o r* o
ci o

c o
to

eo co r* r* r*
o
e

I*. FH
CO
-*<
CI

o o

r ro
m r*

C r* eo

O r*

o> tn

i> r* r*
eo

CM

SON

O i* o

O Oi O
CO
CO

M O CO ci
CO
O CO
IN

r* r*

o r*
r*
# to

o r* ro r*
to o

Oi O co r* o

-2

r*. p*
at

Cl ^H

o o

>-> n d

a* p- c .*

*s=

r*

CO

Oi

mm r*

Oi O r*
Cl

" o

O Oi o r*

s
s

CO

Cl ~ o
ro CP ~

eo

Oi

o o

Ci o o r* co
o

CO 0i
Oi

Cl CO Ci Oi
c* 04 U*
Cl

CO

eo
r*

r* -

O P*

p*

r*

to Cl

r* ci

CO

p* r* CO Oi
*i

eo r* r*
ci r* eo

CM

or* o r*

CO

CO
Cl

OI

OOP*

r* r*
to

O P*
eo

P~

CO
Cl
CM

oi r* o
eo

CO

o o

O -O

t*
CO

eo r* r* o o

CO

o Oi

IO
oi r* o

r* -. ci
eo

Ci

P* co eo r*

Cl

r*

r* o r* o

CO
Cl

eo O

r*

CO

r* o eo eo
to

-*

r* r* o o CO
^* eo
r*

r* o Oi o

o r* r* o to

r^

r* o m

t* o o
o

to r* O

r* o r*
to

O CO

CO

Oi o O r* n
c*

r^

r* r*
CO
r*

o o o r*

co r* Oi r* o
to o

to
eo

Oi
CO

C*

* t*

CM

r* O O !*

p* Ci
to

r* O O oi p*
O

CO -*

ro oi
ci

r* eo eo o

ci

r* oi Oi O
r

o co eo
ci

r* r* o

to

00
Ol I* o CO -
C4 fH

00

r* r*

CO

OS

CI

o r* o

r* eo

g Z

^ :o c* Ci
o ~

CO

O O O r*
eo

_ t~. _ _ o Ci

^ to "*

Ci
* Ci 4
CI

r* eo o

o o o

Oi r*
CI

CN

r*
Cl

Ci
CD

co r*

r*

CO oi P*
C* SI

CM

eo r* r*
to *
eo

CO o

f0 r* o

r* o r* o

o o

oi r* o o r*

CO
CO CO

o o o

t* r*

eo O

CO

r-

00

r- eo O
r*

r* r* co Ci
eo
d

e*

r* r* o eo

r*

m o i p* o o

CO

os

o r* r*

tO tO

to

iO

CO r*

in

eo p*
to

C* K r*
eo 01

CO P- CO
Cl
to

Oi r* Cl

Cl

a.

r*

O O r*

CO o r* p*
Oi

CO

eo

r* o

r* r* o O

eo t*
eo

r*

CO

CO O Ci p*

OI

- co r* o eo
n ci

eo Oi P* M
Cl CO

Oi ro I* O

CO

t*

t*

O Oi

to
Cl

CO

o eo

Ci o r* Ci

r*

CO
CM

r~

oi eo
iO Cl ->

r* wi r*
ci c*
c

c;

O P-

O O o Oi
to

-* o

r* o

CO

r* eo r* r*
to

r* o o o

c*

I* P* CO

o r* eo r*

r* o

p*

CO

r* o>

O r*

CO -i

r*

CO

IO

CI

r- - moo n
ro

Cl

ro ~ co
Ci

Cl

no r*
Ci
CD

C*

co eo r* ci
<o -*
ci

r* o t- o
r*

CO

CN

CO

O o o

CO

p.

co

eo

co
CO

e* r* r*
c*
us

Oi o
CI

eo

CO

r*

O P*

CO

p* r*

CO
OK*-

-AN

i^ r- eo r*
Cl

O* d

r- ro r* o Oi

o
co

CI

O O O Oi o CO

CO

o r*

Ci CO

r*

r* *-
eo r- O o

r* ^
O Cl
CO

r* r*

Cl

i^
to
*n

Ci CO <f- * N

r* o

r*

o o r o

Cl Oi
Oi
Cl

eo r*

r~ fo Oi r*
Cl o

CO

CO Oi O Oi

r oi o

CO

r~
Ci

eo r* eo r* o oi
ci

r*

O r* r* eo -*
^- eo
0*

o o o

r* eo r* o

CO ^- Ol
f- n oo

t- r* -
eo

nor*

r* eo O
C*
~

Oi
CI
<-

c* t^.
m eo

t^- r* o ~* r*
o
^ o

r* to eft o o o

- O O o> i* r.

o r* t- O eo
-* CD

r*

eo r* M

J4

1-

CO
CN

Cl

CN

Cl

Ci

p*
eo eo

CO

Ci
Cl

Ci

*
CO

o eo

r*

Oi Oi r-. r*
CO Cl Oi
Cl

r* O
eo

r* P* CO
r*
eo ^ ~

o eo

Cl

O r* eo

eo
r*

^ P*

~-

^>4 CO t* Cl 00 Ci

a
00

r* o

CI CI

<*

>-< t>- -^ CO t> o>

00 0D Oi Ci 3> 05

85

to

- CO o
A Cl
o

CO

co eo
r* .

O Oi

P* CO O
CO

CN

Oi t* r*

r*
CO

eo o
c*

r* t* Ci
33

O P*
Oi

co i r*

:o eo
oi r*

-* r* o r* r*

Oi O
Cl

r* o

r*

Ci O

en eo os *-**>
U3COCO !>*

00

sss

CJ

O t)^ ^H CM

r* o r*
to
CO

co r Oi CO Oi
n N n co eo

r*

Cl

i-i t*. im co r*.

eo p*
eo
Cl

r* O

Ol

o o

r* m (^

IO

CO
r*

P0
*

IP* CO
~

PM

Ol CO
r. ci

r
CO

r* O

Ol CO

r o r* ci

S^

Ol Ol o r-

CO
*

co

r*

CO -

r-

00

"

t- Ol Ol
O Cl

o.

one*

-S

1-1
-i O O

Oi
Ol

C-*
CI

CO
c*

CO

CO
d

1-1

o o
1-1
1-1

CO

CO

Ol CI

t^ O co

ro

ro

oi r* r r* r<o
n
^

00

Ol

Ci

co

CO

o>

ci

a i> o -<

CO -

CO

CO

Oi

o o

CO

Ol

Cl

CI

Ol

e-

o r* t* ro o

-
CO

Ol

t-- Ol
00

IO

o
o o

o o
O

O)

Cl

o
to

<-"

CO
ci

to
c*

o
o

r>

CO
C

-!

Ol

r*

0)00 CO
^H Ol
to

t^.
PI -T

tr*

t'-

r*

C^

t-

CO CO
IO

--

1-

co t^>
r* ci
ci
CO

r co r.

CO

o
m n o

r*
co

t*

CO

o o r ro
o to

CO

o o
PS

CO
O

-H
00

ro O o r

O to

Cl

CO
~

CO
CM

CO

r-.
-.

I"-

CO
X

r t> t^
o
CI

Ol

r*.

CO

i>. CO
CI
CN

r* co ^ o
.- c*

Ol
Ol

o *
CO
_
_ o
,,

Ol r t r

oi t^ r* o

W)

t-

oS

CI

OJ h

O (0

r>. o r oi

* n c

CI

^
-*

CO
Cl

Ol

CO

o
to

t>- o

t>. O. f>.

Ol

t*01 r O
Cl to
CI -

CO

to n

I Oi

ro

Ol

CO - o =

r
CO

Ol
00

o
CA

Cl

t>.
^

CO
rf

r*

Cl

h on r*OI
-a

CO

CO

CO

to

o n o
o to

CO

1-

OOr>

CO
Cl

on c-

1^

^ o

CO

Ol

Ot

-* e

Cl

r0

c* ot

co o o o
o to
ro

CO

O Cl

ro

CO
to

t^

o>

Oi r* O
CI

r*
Ol

w)

1-

Ol
QU

00
CO

-0

noeo

r*

00

r* co
c

00

r* to
*

o o
CO

r-

CO

- oi

co (0
(O

00

i-

i>.

t>- r* pi O
O
X'

o - o r*
CO

CO

r*

CI

OlOiHH
t>. CI * <-

t* -H

Ol t

00

00
o o

ro
o

lO
CM

o o

Ok O

00
r* o
X

i- o

CSJ

# ci

r* o

a.
c*

t^

o o o

f. co
"- CI

Ol O CO

E-. r*
~

Ol-

CO

I>

- Oi r co

r-

to

Ol

~
CO

Ol Cl

to

pi r* o oi
O

ro

CO
CI

O co O r* -

co
O

c* n-

Ol

O- O

Ol

CO

I>

,-r

-t

ro

!>

CO

I*.

ro

CO

oo

CI

co o o

00
00

i*
CO

QM1 O
C* X O)

CO

~ t>~ Oi

CO

CI

O Ol o

1^

CI

Ol ~
- to

I-* CO
m Ci

1^-

co

CO

r*

t^

to

MOO-

CO
X

Ol

CO

r co
O

CO

-*

CO

-*

.-t

*n

f*

CO
Cl

o O O

Ol

if

i> o r-

CO
Cl

CO

i~

or*
Oi

ro

l>

ro o O
O

Ol
CM

CO

Cl

CO
o co
CO
CO

CM Cl
O CI

o co

O o
cs

to

Ol
CI

oi
~

ro

CO

CO

1-

wm

<- O CO

I'* toCO to<-

o>
r*

I-

CO

co ~ r* r o

CO

o r* ro i-
-*
Ol -* to

CO

1 CO

I-*

CO

co
P)
Cl

CO

PS

CO
o

to A

to

CO

r* o

O -

o o

to
C4

ci
-7

co

r- o i>.

CO

O O

r>.

Cl
(X)

Ol

r*

CO

to

r*

OOP)

eo
o

1-1

CO
mu

CO

C*

n oi r

>- r O O
CO

r* o o o

Oi 1^.

c* >n

CO
C

t^

CO

Oi

CI

to

<-

CO

Ol i*

>4 SO M

^-

oi co r o

o>
00

r^ i*

CO

Ol o
t-

CO

Cl

Cl

r CO

n
-

01

t>-

o o

C* Ol CO

CO
**

r*
Ol to

I- CO

Ol o o
*+ PS

<x>

* *

en

Ol
CM -*

o r* co o
ci

o> o

Oi

in
o o

Ol

ro

o o

1-1

CO
U5

<o o r*

PI
CI

r* O

r>. o oi

r* pn

CM

us

00

Ol

c*

ro

to

CO

r- ci o o

c*

Ol " ^^
t* CO <-

^ r+ o
^-t

pa

d
C*

r*

*t
CO

o o ** t>

ps
CO
a o o

CO
i

CO
o

CM

n oinix
CO

CI

i^>

I-

co o> r* o

c* * -*

CA

**>

r
Tf

iOl

!>

l^

CO

CO

t CO

CO

I-

tO

CO

Cl

CO
CM

CO

CO

CO

l>

o o
o ^

CO

O O Ol
CO w

1^

!*

or*

Ol

ro

CO
CO

CO CO

O o

CO

co

haoc*
~*
^ o r- o

CO

CO
-

Ol CO
-i

to

ci

CI

o o

CI

CO
~r

CO
-.

CO
C*

CO

c* r* t* r* r*
CO

to

CO

C- Oi
CO to

CA
CO

CO

co
co

o a o ~

to r* O

1- I*

CO

r>r>no o
to
^
CO

CO
CO

Ol
PI

o co
o

01
Ci

PS
CI

n r* o o

Ol o

r* f> O

01

-<

CO

r* o ! r*
**

1^

CI

Ol

o o
to

-f
CI

Ol

p-> CI o

f
Ol

ci

^ I>. r-l CO l>

it co t>a

O -iNH CO

>: '^ C L- !

Ol ^0
i-i to

co

r* C3i co c - t>
fc- t* X 00 7> Ji

oS

Ol
CO

CO
Cl

ei
HCONOCO

oooo

CJ W 0> r-i CO
i-l CM CO CO

63000 72000
c eo
. c

con
to

c c to ~

Cl

to

Oi

Ol

.'.

o c> r* 1- c

i-

to o

03

eo o o ~

to

- eo

"

Oi
~

to

c*

* to
to 1

to r
C IT

y.

CD

o a

o
to

CO

CO

to IO

Cj
lo

to
CO

to

CO

o o

eo ro
t a

to to O

^-

o t> o

to
CO

Cl
CM

Oi
to Pi

to

CT

CO
Oi

Cl

C M o

0 o

r. a

o o r w* oo eo
re -*
c

r-i o eo O to o
NJlM

C. O 1- Oi to

. CM

o o eo f

ao eo
t^

o eo o o

to

c.

t Oi
Ct

lo

to

o
n

eo i o.
01

H 0 I*
0

r-.
CO

to O

O O

lo

r^ o
eo

to

o eo
o

t-

-1

ro

Oi

Cl

* eo

to.

lo

Ci
1 -

to ro

ro

O ro
i

to

1-

to

ro o Oi ro
to
CD
BQ

l*

Oi

.
T

r-.
~

to

Cl

ro

to
Cl

Ci

C*

C" cc to

1-4

lo

Cl

o
c*

C4

rt O r-. tto

ro

to

lo

i-

o-

lo

O lo

CO

o
r* o to

i* a o

c o

O 1* o c>
* to.

O O O I- o

r* o o to o to

ro o to eo GO M

"

to CO

to. o> o
ci Oi

i to CO
O to
Cl
M
C*

o o o o

eo o

M.00

to

O ro
Ci

t-

c, o
n

i~

Cl

CO o

lo

to o

C4

c a

ci

ro. o

n to ro o

-r

to o
o 3

ro

lo

ro to.

CO

Cl

in

t*

O O O CO

to

?0

lo

to

Oi

ro

eo io

C4

S3

i*

1-

to -*

t-

lo

eo

t~

l-

i -

l-

m
r-

lo

~>

to

CO
to

Cl

to
1-4

CO o
Cl

ro

CO

CO

Ol c. eo to to
ct
CN r- Oi .-.

*0

"^

to

n
Cl

to

ro

Oi
<*

to eo Ci
"
to

o
co S r* to.

l^

to to o

to

* i

c:

o o o

<o go

O to. O

at
00
to, o i- r

I-

eo to, o

CO

C n ft r> o
, ro <c

ro (o ro to,
** eo <o

c to o o

O to o to. ro
Oi iO

to

*r

Oi O Ci Co
CD
OS

eo

BQ

eo ro

to to.

IO

to

eo eo

<o

CO

to

ifl

a n
c.

to

i-

"

ro

GO

<-i

to

"

a ro

C"

ro

-t

to

Ci

O b-

[~

CO
to.

^t

Cl

w*

I CO

-H

Oi a o to

o
C> CO o
*

to

lo

a to eo

CO
1-

Oi

re

00
to

-> o

.-?

ro

* CN

r- t--
~ to o
co

O r- -o
a

ceo

c^

ro

K 3* O Ok 3
CO

CI

eo Oi to o

O l* lo C

o a eo r- ro
o
to. c*
*

0 Ci O

o to eo

o -^ a o

i-

a i>

to

:i .

CO

1-

CO

Oi

to.

t^

C r- c

i-

O r- Ci O
CM

to

EN

to Oi

to O

o c

f -

O eo

l*

CO

c o

c
00
CD

fc"

to

to

d
^ o c Pi to.

Oi
IO

-f
ci

i- c

at

CO

ti
o

CO

o >. oi o

ct

00

ro

io

o o

t>

CO
IO

to. O O

>- i
^- IO

C;

to

CO

eo O to

ro

CO

Ci

Ci

c,

Oi to i

to

CJ

to

re

--

to

to

Cl

to.

lo

i-^

to

'3

Ci

I-

to

4f

lo

O to

ro
c*

to

Cl
CO

PH

O (o o to

CD
ft (0 to>
C GO

Ci

to

t*.

to. CO
CO IO

to
CO

CD
l o

- o o

(01* to Oia Oi -
p"1 o eo to ro o
E
to.
cn

to

~r

CO

en

O r~ -* i^ O

ro ?o
to C

to

Oj

ro

to

to.
O

Ci

CJ

I-

^ e*

2"

"

(a-

i-

o
o

ro
t-

o
Ci

Oi to eo
to

o ro
O

CO

c*

Ci

G
lo

to ro

a to
~
CO

to

to

CM
to.

to

ro Oi

CO

c o

ro c*

l* *

O to O

CO

l^

ro

Oi

CO

F-l

IO,

ro

Cl

-no

to O O eo Oi

t* ci

Ife

n o i* n
o

'- O ro,
o

* to to, c " ^ - -

r; r- a

C O O r- O

t^ ^-

* e*

lo

-i

to

eo
o

o ~ n

ro

o o

CO

to

Ci
to

St ro O
X
(0

to O

Ci ro

ro

i*

Ci

to
to

CT

c c ro

l^

ci

c.

i-

ro

to

ro ro to
T

O ro

to

c;
ro

C-

" to, O

Oi

O
CO

Ci
CO

to .

tn eo

o to ro o to
p- CO

Cl

a> o *

c. ro o

a so to

* r
r o

*-

'-

Cl 1^ ^ o

W3

AO C ft hft
c*
ro ro

c to co o
.
T
C*

ro

Ci

t* o
m

f* O O I- o *0

io

cn

ro

t^

to
(O Oi ro

CO

ro
c*

_,
G

re

"

lo

rto

rr

Ci o o

to

O eo i

Ci

C-

Cl

Tf

re

CO

~
-4

to. eo

a to

-h

c* a

i-

to CO

O O CiO

-* ..-;

CO o

to, Oi lo
to

to

Oi O
ro

lo

lo

ro to eo
O

to

ro

Ci

CO
-r

t^

ro

CO

O)
Cl

lo

00

to

Ci
Cl

y.

to

to

to CO
c*

CO

<

Ci
ii-

to CO

ro

00 -

eo

Cl

c o

r-i

Oi

O rt t* o

1-4

Co Oi

to

-H
CO

eo

ft
Jl

- C id

or

'

vSoSkS

S5
00 CC OS 51 OS 3>

to

O CHH

ci

CT)

co to, Ci oi
OI (M ) CC CO

to.
-i

11 to |1 CO to.

<} rf iO iD iO

CTi CO "> i-H t>

m c to to r>

f-t CO to. Ci eo> or*

00 00 CO 00 O C5

in
o

ro
9$

<*

In

t^

c,

c.

a a

"

ro

< r* O

in in

CO

CO

e c

tc

o n i>O

CO
C4

o n

t r r i"

.-o

1^

r* r' o
tc

h-

CO 1-* CO O
CN

i>

IN

ro

CI

P)

ro

a r r r
w

O r^

3 o t* a
o

o o e

t^ o o

Cs t

o ^-

In

eo O

In CO

CN

In

ro

CN w*

O r.

r t>-

r^. CO

In s l>.

e a r*

q
CO

ro
c

o at

i-

z.

[N

co

O CD
GO

a
IN

c.
cn

r to

CO
CD

o; O

l"

<-

eo

IN

IS

In

I'

r>

c
c

1^ o

i-

i- e n

O r*

l-

t+

c-

1-

"

Cl

Oh

i- c

t- CO

CD

CO
r-

a
go

a
c.

c.

ro
CI

c*j r* t^.

I^

r-

i -
CJl 0

i^

nrso
CO

l-

i^

CD
1^.

rr

!> CO O)
CN CN
O X

O r-
IO i
o

CJ

Ol Cl
CN ~

Ci
X

I-

c r*

CO In tO

ro CO

<*

O M

n CO

n
i-

O CO i
C-l

IN

In

IN

-.8

CO

1-

tN

Ol

Oi in
cm >n

tN tN

p-

In

Oi

ro
CN

MhOO

c<

CO

o o o o

1 -

"

1 "

CO

c.

Cl

i^

Ci

ro

o
en
tN

CO

r-

In

tN
00
In
In
Cl

In

CO
CA

IN tN

Ol

In.

Cl
Ol ^

In
r-

eo

IN

In

CO

tN

ro

In

O O

In

IN

CO
CM

i-.

ro

c, o
CM
Cl

CO

IN

IN
tN

=
o

CO

h- o

ro
Cl
CN

CO In Ol Ol
~
CO

eo

Cl

C. :-.

CN

In
p-

IN

CO
c.

Cl

CO

IN

Ol IN

x r. -

Ci

1-

00
IN CO

In

00

or*

c.

In

XI

Ol
IN

In

Ol

CO

In

f- CO
In

Ol
<N

IN O

In Ol O
" IN

o o

ro
CO

O M<! - O

In

co

tN

a
in r- O

Cl

ICO

*-.
CO

Ol
CN

in eo

CN

C O GJ

CO

CO

t*

H-

lA O

o ^

CO

"

ro

in

CO
p-l

CO
-*

Ci i

In In

tN

o eo co

In

eo a>
Ol ~

O tN
c-

CO IN.
tN

tN In

In

CO

In

CM

CO

CD

n c

in

r co r-

-* =

M !>. O
O

o o

e *

o c c

CO
Jn

CO
M

C-

CO

co

CO

c
CM
CO

Cl

eo

I-

Cl

O tN

In
CO

ci

00
CO

t- -! r*
to ~

ro

00

tN 1

CO

ao

g -

|N

ec

i- r

in

o o

a c

cm

00

heo

IN

r.
D

tN

Ol

CO

en
C ro
C

CO
^

*8
i-.

CM
at

Ol ^

00
O

r-

^-

31

i-*

r- o O

ci
CO

O CO CO c>

-- /.

CO

n ci

t^ to
I-

en
iO

tt l> O
-CO

I In

c o tn

CO

IN

CO

CO
Cl

r-.

i-

c n a

CO

c r:

y.

110

"

CM -" CO -"

CN o

IN

In

, -

I,

ID

-r

>

CN C

tN Oi tN
i- CN

IN

2,
CO

CO O

In

o o

In

tN

ro ro
IO CM

O CO CO

ro Cl
IQ Ol -.

O tN

tN

o
co

o
c.

eo
co

i
U3

~
CO

I -

IN

cn

a
o

i-

o o

1- c-.
to

c. ec a,

O eo tN

In

tN M

l-:

c.

>*
-*
*

tN

In

o o

a
in

o o

O O CO

1-

i-

=
i^

r- tf to

l- ri

CO *

1-

a.

r C) co t*

eo

CO o
Ol

In

IN
CO

CO
1

i -*
IN
CN

CM

In

co

w n

ro in
c-i oi
ci

in
00
CO

o i^ r-t o r-

04

i~

c.

Jf

r i"* r**

O O t- m

CO

*"

O r-

CO -

i~

CO
Cl

i-n

CO 1^ o

r- c
X

In

C.

co r !-
co
eo

lm*

i^

C.
CO
CM

IN

-H

O CO o

tN
CO

In

CD O

CO

ro eo
T

c.
tN

CO
Cl

o o

Ol

tN Ol
Ol

O Ol

Ol

CO
m

Ci
CN

r>

tN

IN

Ol

CM
CJ

CM

ro
CO

tN In

CO

to

tN

IN

co

CO

to
CM

co

~ |>.

IN In

CN

CO
In

Oi
eo

in in

i'

in

co

I".

M
X

In

CO

r: O O
CO

CO 1^

a a oi

r- c

r* o

O CO

1^

O t CO
CO T

C. Ci

eo t-. o

r- CO

IN Cl

CO

In

In

IN

In

IN

Cl
CO

CO
CD

:-

IN

iro
tc ^

tN

IN O

z. t> i- O I-

i-4 In, r-l CO L

OOhhh

as co o ph 1^

i - CO
0 ci

a
ro i- CD
00 ~ CN

IN,

Cl

-0
Cl

CO O IN
-
CO

|N

In

CO

In
tN

pi M *J CO CO

~
to

IN

c.

In

Ol
Ol

o o

IN
iO

01

CO

tN Ol

IN

pi

In
Ifl
Cl

CN

IN

CO

eo

CO

tN

CO

CN

IN . I
CO -

In

rN

04

at

CO

~* ^ Ol

i-i CO t> 3i CO
nfi ^i ^1 ^i iO

c CO CO
O

o
co

r*

In IN i-r

CN

tN

co

CM

Z.
F*

I-

--o

"

CO

T-i

c o

co

CO

a a

co

CD

80

T
Cl

CN

CO

ca
co

In

00

CN

ro

CO

rN CO

c*

IN
-i

(C

CN

CO

<-
ifl

to

Ci

Cl

In

C35 i t>. pH CO
Ifl CO CD l> t>.

Oh"

1-

CNOJ

tN as co
^InOC

as ! tN

O t- O tN.
CO

;* tN

tN

CO

t-H CO In. OS CO

Ci r-HN,H

oooo

CN CO. CO CO

CO >

NO
CO CO

72000 81000
: ro i^

o ps r- eo

c o
o

co t* O co r OS

ct Ok

rot*

co o r o o
Q0
~>

cs r* co
ct

-r

cj

"

r* a o

l> o O OS
c*

I-

tO

p:
i-i

PI

OS r,=

r*

OS

OS < o

r- o

i^

PS

PS

r*

V CO
T
C*

1^

a t^
O
-i o

o
eo r* eo
(0
CO

eo r> eo

t-

OS o
CO

o o

r* co

CO

O OS r t>.

PC

o
O I*
00

o
PI

I-

>.

z,

Z-

I -

~.

I -

Ofth
PS

o
- r* r- - ro

t* co o o o

O r

" tO

e*
r^ os r t>

r- o o o

r*

OS
t* OS

rt r* Ci

Ci

H
to

J
-*

ft
to

ts O

c.

c co n
ir

tO

*m

i> *

C-t

B*

I-

CO

's05

.-0

PS
CM

OS

c*
r* r

to

r OS t

ps o r t r>- r*
r
eo o

os r c.
CO
CO

p:
CO

O O

i-

OS
r

ro

tci
Ot

i^
ps

I*. O ~ t* o
m
R ft

Cl

r* co to ps
CD

c*

p: i

t-

CO PS *> OS
ct r>
OS

i a i^
o C*

o c*

t^

p;

o O OS
-^
c*

t^.

O O r
ws

OS lCO

OS

cs
c*

o o

o
n

I*-

r OS o
CO c*

r * V.

t>.

l^

r^

eo i

PS OS

t*

.
CM

t* t* CO

t- CO

Or; 1 PS
to c*
c

t^

Ci o c o CO
CO
C*

l> os ~

i^ r
o

0) t- o o r

(O

co o r* co

PS
US

O CI o

00

t* y.

CM

Ci -

O p)

tn o
to

l-

CO

P*.
to

00
r- - o

t* o o r* o o

c h a

l-- eo r"

*>

coo

O l"
tO Ct

-*

r- r-i o o r-

CO
00

1 o

_ t, _

i- o

o o t>

C. I* !O
CO

n n

eo

-^

.
Ci

t- o

l-~

03

r*

co

co r* r OS

**S5

> o r o o
.-
ct

r*

SI OS

a
Cl

co r*- ps O r
OS w*
<

o --

CO

CO

~*

i^

OS

OS

I-

>. o

m
C4

to

ro r*.

PS

CM

CM

PS

r*
ro

PS
1-

o >.

ro
M

1
~ os ps c

co r r>

o o r^

y.

ro

C t

eo ~ eo

t en co ^
^-i o r

1 o. O O O

r* r* i

Ol

OS CO

to c*

r- O
PS

OS CO
~

CO f
OS

CM

P0

O <-

t*

CO
rr

t^

l^ u;

O r~

r- t* O CO

i-

r*

t^ -r

CO
CM

Oi

co r O r*
PS

c*
c<

to
CJ

o o

OS OS

I-

J)

r* i o -
to

r- o o

Ci

r-

1-*

CO -" ^ to

r r*.
c*

*o o

CO

C*

-*

co CO l r- r o
CM CM
Ct

-ro

r* os r* ~

~ o

ot

*"*

p-

ct
os r*. r eo O

ro t*

o n o o r co

01
CI

O r* r--

3.

r* r-

c*

00
U3

us

r i co r^

a to

PS

r*- os
X

CO

r*

p-.

t>- o

PS

^H

Q1^

r*

-*

PS

OS OS

t*

OS
CM

r* ^ co co
<# to -> Ci

PS

1- c\
-Of*

CM

O t>. PS ws
O to

O t- r^

1 o

CO

~ c*

CO

CO

(M

iO

OS PS

r.

i-

ro

OS

I-

i-

FH

PS

o.

o o

o o

t-.

CM

o o

eo o

O r-

r- eo *^

t^ CO

r*

t^.

CSS

co
r-.

CM

r* o

OS O OS r>.
to
01

OS

>

eo

<

V,
BO

O OS r t^
co to
-
Ct

OS r* t*

ct

a i e

O p*. eo

noob

r* o r*
ct

C r; o

r- O

. c eo ro i>-

!!>

C i-

CM

C K m

i-

CO

^ OS

ps
O r- t* c* OS
r*
Mg
CO
Ct

r ro
V 4< CH IO

r* o

O ) o n
CO o

t^

OS r*

o c*

en

eo
^H O OS
c< r*
(O

h ft o CO

CO

i*

r* os

n j o

cs

PN
CO

ii
O

r*

r* o t* o

pfl t*

OS

r*- t>- o

o o

r-

"
i

CO t- t>. OS ro
OS
CO o

1--

i^.

o os eo o

CO o

co r- . co

r>.

PS
c*

PS

OS
in

t^ CO

to **

CO

CM

o o

r co
o

CO

i CO

o*

m o

~ OS r os
~ PS
ct

CO

CO

CO

i-

PS

O) t^
CO

^- c*
r* os eo
ct

. o
<M
CO

r* t-

r* O r^

CO ~ O

OS

^-. o

CO
- a
^
o

o o r* o Ok o
CO
ct

r o Ji )
t* c

O O OS r*- O
eo go

r r* os
O
M to

P4

r co

o o

l-

I-

O OS

PS
CM

p^

eo ps r*
to
0

PS

PS ct
to
ct

PS

CO

t*

to

t*

CM

CO
CO
-

r M r
p m

OS
CM

PS

o *

".

CO

r*
OS

-) c

OS CO OS t*

PS

CO
eo eo
c

eo

9 *>

to

r*
(O

ps co r*
r* - oo

H
00
CM

O CO O CO -<
eo
Ct
CI

O CO r-I"- t*

r- Pi Ok
Ct 3

O ** O 3*
CO

i r- r* ~

r. t*
O iO

-h co ** os co 9s

i-tt^-H CO J^
00 CO CI 35 A

i-

OS t to

PO

CO

ro

t* CO

C4

a to

w rp*

^f

PS

1^

Cl

<-* cs

t*

r*

CO

tfS

c ^

CO

I- o
o
CM

o o r* o>
c*

OOHHN

O CM 94 CO CO

to
CM
CO

o to

t--

i-

c*

co r* oi co cj>

o o

CO

OS r
eo

t- o O OS

HI>HCOr
Tj< Tf IO U3 iO

Ol CO 3S i-4 l>W3 CO CD t>-

I-

- -*

OS ~ OS OS
C4
ct

OS

rH CO t> OS CO o?
00 CO CD 00 OS OS

oik r n

ts
Ci

O 00 Is
CM
to oi ts o
ci

en
O o c ts

CO CO
CM WS tO

~.

n t s - o

O c* r*

CO ts t^

O CO I-

cm eo
cm

O ts ts fs

eo

c*

OOr^

o o o

rs n

i- rs -.

rs

o ts co
cm

00
CO

to to

en

C;

n is o O) o

00

eo eo ci

i to

ts 'o co ro ts
tc .-i CM

^ ts O St
ts
M M O
CM

Ci

i-

CO CO
*
- CI

Is O ~

ts CO
03
O

01
CI

Nt* O t>
CO
ts ^

Oi ts CO

CO o 01
OMh

<-

Is is oi

is r-.

CO
00 -*

70

tCO

O O Is. ts.

;:

pt

o o-n h

ts CO CO

Is.

(O f

- (O

1^

I- O |s-

rs

is

t-

01

o
ts
ts

o o o

o rs> eo

is oi n M t*

OmH
CM

n
rs o is
to

r* 9> t<*

.-..-. rs
eo
w -

CO
ts

t^ CO

1-

cm
i -

co o to
ts ->
o

ci
IX

O ro O

ro co oo
ci

~ cm
p: ffl o -

o r o> ro

CI

Is f.
O

r-t

Is

CO

C*

O is ^ ts O

CO

ts ts -.

Is O

ro

1 -

CO Is

Ci O Is ts

i-

r^

TO

O Ol Oi

- i- o
tfS CO

is

c -^
CM

'

IS

CO

c-. ts o

Is.

O Ol ts r+ -H
Oi
-J> ^r

Ci ir>

CO
ti;
(O

ts CO CO O ts
00

Oi co

ts. O CO CO
d
tN
^C*

ro Ol

Is ts Ci

OMs. O
CO ^

^- ts CO

O ifl

CO Ci O r-

co

Is

CO

CO

CM
Is

ts CO

-t

00
CO

C.

is

O is Is

a - .-,

Is

r.
4

CO o o

is

Is -. O

oi co rs

m:

ro

O ts
ts
tO

CI

ts o o

s~

to

fs

tS

i-

Is

|S rt

^>Is

is O
CO

1
1

- O is

TO

CO Ci O is
CM

O ts 1^. CO Is
Ol ~
US

01

CI

~t

CM

O is 3: Ol Oi
CM CO c*
(.^ CM

r.
CI

eo

Ifl
?1

<o

Is

CM
|s -; co ts
CM CO rj>
CM CN

Is

ts

in

ls

ro

CO O

*j*

eo is o
n n -

CM

xm ri-s. f O-^

Ol 31 ts
ci
CI

Is Ol

i-

CM

O CO O CO
.-.
is

Is

M)

c*

7^

is

eo
ao

Oi O O O

ts, O O CO
Oi

- * t

CO Oi ts CO Oi

ts rs

CO

"

ts <o o
CO 7-.
c>

ts

Ol Is

so

01 ts oi is
Oi ~

CO I-* O CO
CO

ts rs ^ o co
to
*
00

CO

CO |s.

n
o

CO

CO

I"*
i.~.

Oi co r*.
d Is.
W4 *4

ts

3.
tS tO

r* ts
eo

3 co eo i>
<- CM

C:

-i

Ol is

O IsO Cl-

Is

01

Oi O
CM

ts CO ts
Oi
CM

<* "

Is

ts is CO ~

eo

O - ts eo o

Is

Cl

Is O

CO
CI

O is o

rs

Is

o ~ o

Is - rt - O

c.

eo ts

I-

01

Is

^r oo t r rs eo ts eo
<-*
eo c*

|s

!_T

Oi ts eo

IS

TO
J-

70

1 CI

rs ts o O

Is

to O CO ts

70

Is

-1

CO

Is ts

"

iO

CO CO ts- o
J - *

Ts O
co ci
CM

ts rs Oi

D Is. O O is

h Ol Ol Is Oi

Ol CO

co eo <- r*

Oi

r- Oi ts
to CN
d

O CO

ts CO Ol -h
rs- X CD IO

CO o Ci

is

-- r- Ol ~

is

OOlOO

r* o r-

co

c* *-

-cH

o r- r co

CO

CO

is Is O CI o
cm
ID

eo is t^.
Oi

i.

c
C CO

to

!- d to

eo

to ^

O Oi

i -

Oi o

co
ft
r-

i-

TO

O ts ts

CI

CM

ts o t>

ID
CM

'_T

Is

ts

CM

CO
Ol

CO ts

rt> CO ts
CO -f

i-

si*

c*

<1

h - Ol

o - oi o>

i^. Oi

w r. o

at r- o

ts o eo

Ol is CO
Oi
CN

Is

eo

H CJllSrt |S

I^

. - r< - --

CO i-

1 ts CO Oi
f CO
CI

t^

! ro ~

or-

r* o

co ci if* r*

CO
1-H

~ o
CO

ts . to
-h M CO

01 is. CO

to Oi ci

t oi ci o
cm ~
cm

CO
O QO

co co -r ^

-; l"
00
CO
C;
CI

ffl

(S

"CO

Is ^- CO Oi ts

1st*

TO CO
CM Ol * CM
CM r-

O Ol Oi

eo

ts CO O

Oi Is
ts tO

01

Is o ts*Ol

CM

eo

^.

Is.

CM

eo
o m

00
CM

^- o is, o r
CO

so eo

S*

CM

is

CO

ts

O Is - K CO
HOB
CI

CO O ts o
if
CM

(s O Ol ts. o
to
~

~ o

is Oi o o -

is ts fO

CO O ts

^- r^
to

CO O Is - ,

m
i-

CO

Ol

Is

ts to eo

is

tS tO

i -

CO ts

ci

i.

eo

Ol

Is

Oi ts is ts
I-*

Oi

CO

|s n
HCI

is |s
CO
IO
Ci

CO o

IS

o o

c *

O ts Is ts

CO

O
CM

i.-i

CO
CM

CM

eo ts

-*

.- ts 3. cr
~4
tO
CM

?v

o"

9)

is

O eo

CO

Is
CO

o ts eo

is

ts

s*

h o n"
CO

CM

a
o

'

CM

eo
CO

CM

CO

Ci
i-

Cj
CM

rs o _ co

o
us

US

CO

eo

co

PQ

CM
CM

*i

ci

CO
00

s|<
ts

CO

CO
ci

CO

-c,
is

ts

CO

S3

00

CO

00

I"

(3

00

CM

CM

n
at

Ol Is M Ol
M
M

ao

ts

CO

~ cm

00

(M
00

--

05

w Ol - o

ts a ts

Ol ts

e
-4 ts l~ CO t-.
O O i1 rt H

O O is

05 COOHI>
. (M 0M CO CO

HCONCICO

CO

,-H

, 7

CO Is O ts

CJ> -< ts ,-H -TO

W3 C CD I> *-

Ol
00
CO

TO

ts. a eo as >-' ts
ts. i> x CO Ci rs

fl

- eo ts Oi
~* -.
CM

H CO Is D CO

oooo-

CO
CI

1 ts

"

90

ts CO
* -O

ts as CO,
co ro s*

81000-90000
o e ft
p-> tf)

ft t* en

to en en r* r*
C* f*

r* o o

*+

o o co .-> r*

ft r*
CM o

~ r** o n-

O ft r* ft

en

en <~> en CO
o -< ft ct
C4

t-

o o r* o

r* o ft c r*
eo

CO

o r*

ft o r* o

r* r* r* e
CO

O r* O O t*

o en o r* ft

ft
CO
r*

eo r* r-

r* r* ft en

en r* o ft
ft
en

or* o o
t*.

CO O O

r* r* r*

ft

Or* ft

ft CO
C* -!
X

ft o r- ft
c
* en

"r* ft O en

r* r* ft o o o

CO

c*

o or*
*

o en r* o o en

ct

r*

ft ft o eo r*

CN

**

ft

eo r* o
C*
to

*-oo
-

o o o r* eo
r*

O r* CO
d eo i-i
-

r* to ft
*

r* r o r o
c- '

en r* r* o ft
eo w> to

m*k

tf

"

eo ro

i* ro r ft o

<m ft t*

O CO

r- r*

cm

ft eo i1*

o
o

CN

-* ft

to
CN

CN

r*. ft O
r*

r*. cn O ft r*
<- X
X

o o en

r- ft

r* en
n *r ft >A

en

cn

co

r* ft t* r*
s to
eo <-

* r* - o ft-

t^

.*

ft

CO

ft

Cl

ft

r* r* o

r*. r* ft ft r* eo o r*

o ft r* ft r*

Or*

k - r * ft O
T G to CO r*

o to r* r*

ft

r* ft co
~

en t* t* co r*
ce
to eo
cn
<-

t* to
O

co ft r* Io
CN

CN

ft r* en ft
X

ft

en o r*> o
c* en
en

ft so ft r* ft

o
o r- to
en *n

r* o r*
en r*

en r- ft o r*
*o

CO I-* O
-f.

o en o ft en r*
r*

O O r*

10 1^.

O ft
CN

o o

o t> r* r*
CO

f0 o o o

en o

r* ft en

CO
CN

r* co o o

M t*> t* o
r*
ft

r- o o

o r*

r* t*. O ci ro
Cl
X
CN

ft ft

CN -

co r* r* o
wm en

OtO M CN

Cl

r*

c* -* o
c o

r-^ co r*
c
r* o oi

00

CO

o en r*

en O O O

s
r* CO CO

so to

r* r* en ft r*.

CO C* *

CO

r*
- a

r* o r*

*r* r* o COr* r* *eo

r* t*-

c o ft c c r*

o n r* r* eo eo

ci

10

CD
CD

co e*

CN

C)

r* r* ft
en

ft

r-

co en
ot -> r*

I*, en

ft

ft ft en o

r* t* ro o
X

o r* r*
en

on nm
r* en <*

O ft o o o
r*

r* r* r
CO

r* Oft

r* r~ o o en

r* r*

ft

ex, en r* m r*
en

ft o r*

ft
10
en

ft r* en

ft ft o ft r*
CN
r*

ft ft r*
ft ro ft
cn

ft CO

or*

ro o o r-

r* r* r*

CO o
- 1*

eo i* ro
o ft

=
CN

r*

1^

s
O
CO

* CM

ro
03
<o

ft

CN
CN

CO

O CO
c

ft r* o

i-t o

O ft o

co r* o

CO

IO
CO

CN
CN

>o
o i*. r*

r- r* ~ ft
^ art a -r f
e
c*

o ro r*

r* C
tn

r* ft ;- eo rc* o

ft t* ft

c ft
r*

er o o r r* n

ir ft r*

rt N O

fi

to ft eo
M c

r* r* ro o
r* *

3- o ft ft

C i*.

r* ;- r* r*

to

i o o n h n

*o

en

a O ft
x

c o

3
9

r*

en

o c- o
en

CN
CN

o r* o

co r- O
c*

ft

ft

r*

t*

- en

CN

c w r* ro ft to r* r* en o o
o

o
r*
*
CN

CO

r* r* en m o

o to o o en

r* ft i* ro en
CN
d r*

r*
X CO

r^

r- ft r*

ro ft en en

r*

ft

c en

O CN X
CN

O o -

en en ft
I-

CO

r* i*
I-*

ft I*.

r* x

o
en

CN

C* CO I^
CN <i

^- r* r*
-

CN

CO
CO
CN

en -o

to ft
>-
en r*

t^ o o
en

r*

ft o r*
co
to

eo t- ro
- lC <o

r- i- ft

CO
ft co

o ~ r*

ft r* ft
~
r*
^

r- e*

ft ro O o e>
ro o

r* n r- o
n
w

ft r*

eo eo esi eo
e
: cm

r- ft r*
n #

t*

{*

r co eft r* ft

_ -f CM

ft r- ec
x
*- ^

r-
cm eo

m t r- t*

i- r* eso

ft M M ^^ c; ft
r* e <-

ro

o o

r* <~ r* o
Cl

r*
CO
CO

to
g
1* o
& ft - X

t* p: en r*
ft O CN CO
o r* ft en
r* *

CO
C

O r*. !
Cl

CN

O r* ft o

r* co r*
en
to
c*

ft

r* o en sn en

r* O

r* ft

IO

CN

o.

C ft C
CO
X

OCrt
CM

CN

r* r*

r* c* to O
ft
X

o r* en cn
en cn

ft o

r* O

~ ft en c
eo en

t*

eo i*>

to

r* o

X' CN ?

r~ r* r* r* en

r* o jo

o en en

O ft

r* r* o r*
c
en

c c* o c en

eo r*
CN

r>- r* ft

en o i*
c*

en
CN

ft O ft

-t

r-

r* o tn r* r* o
in

l-

en r* o r*
to

en O
r*

c o o r* ft
c*

en ft t*
ct ~ x en

i~

en r*

O en ft o o

r* en
m ~i n

ft -

en r r* r*
c*
X CO

O O ft en o o
CN

en r*. OS
Ol Ol M X

f-< r* i TO r-*

CSS 50 7) ^ t>
.0 !T CC t> 1*.

-t co r*. 7i to as

ft

l" CO O

r*

ft m
: -

o r*> o eo

r* i* ft r*
x <o m

3L * O
. .

ft O r*. r* ft
c
* -O
o

r* r- ft i' :*3 i

ft r* e

r* to o t*

O Ol i~ o o

1 -

CT.

t~

CO

-r
CN

<oS

Q
i-i r*, n I* Clrt 31
roo c c c c t* r-

CN

CN

CD
CO

OC 00 05 3 3>

32

rH ^- ?1

f CN

O
CO

rf- *? tn ia o>

or*

oi n o
00

o o

Oi O O

n n

00

n o> o
-. to
d
eo O n CO

55
n to

r-t

Oi

N O

OnOOO

Oi o r* o* r-

n o r-. eo

- N n O OI
ci

CO

-*

n
to

r- ^ ^ ^

to

eo r Oi t" oi

r-*

eo

CC

CM

r-

n r-

eo n n <

eo to o
O CM

r eo o *-

O r F-t oi eo

o> eo -h ^

Oi N
Oi -1

CO

CN

n eo

eo eo
H to ct

r-i

o o

O co eo N
CN h

p^ r* r* Oi
*
d
r-

?0 O rto
r

N OIhNO
O CO

r* eo ro -* o

eo - eo

CN

-i

m O n. o> o
r^
CO

N Ol-t N
<ONnh

r- r* m ~+
rn

CN

OI

o
eo

Cl
O N CO
n
Cl

CO -- |S
Oi eo -1

1^. !0

OI N Oi
i
N

O -. O H

N N O
to

O
pi

o o

00

N
CO

CO
CN

r- r eo
eo *-t
r-t

Oi O CO

n n Oi
CO
<-.

~ CO
r-t eo to

r>. r*

N O n Oi
-. w* C
o

n t* h o t*

CTt CO r*.
CO
-t

co Oi oi eo

O n ~ o

o r* o o rn

Oi CO Oi N o
C)

O rs

o r* eo

NO

CO

Oi

iO N

2S

N N eo ^

H M

2= =to

CO

N i

r-*
CO

Oi
1

Oi

CO CO
ci
o

<-< 01

N Oi N

Oi NN O
~t
O
ci

CO ^ eo 00
Cl

CO

r eo n

oino n r*
x to
CI

eo Oi o
w
Cl

N O

CN

eo

o o n o o

eo -H is
oo ~ eo

CN

NO

to d n -#

t* - oi r

o eo o

00
CN

co

00

N O eo to O
Oi
ci
CO
Cl

o o o

Oi

Oi r i- Oi
CO
CI

n to r-

to

o n c
to

eo r Oi r-
M
CI tO

oi

CO CI
Oi r-i

O 10 N
- Ok

CN

co
CO
CM

ts N I" cn N
Mnei

NNfflH

oj eo n o -

eo

o n n o
CO

CO

CJk

CI

n o o

CN

CM

o o r

-h t* r*
Ci
f*

Oi

00

n h OC4N
ei rm d
d

CO

o
co

00
00

00

o o eo n

CO
CO

CO CO N
NN
eo
- CI

ION
d !*

r-l
oo
00

NObno

h n b M O
tO - CO
CI

Ci

CO

O eo i eo
-

-t

- N

ts

c*

Cl

-I eo co
- CN
Cl

N Oi N N *N >- Oi
tO
Cl
CN

NO.
CM N

Ci is o

" Ci is
ii NIC *

NCl-J

eo 0 eo
Cl iO
00

eo o n

o N Oi O

O N o o

- to
eo
ci

N N

1-1
CO

o a
00

CO -
Cl

o eo m o n
m* n

CO

Ci

CO

o o o n
d rH
d
O

Oi CO
to to

-NO)
P*
Cl

MO

00

eo

O N

^> -H o < ^H

N Oi
r*

r- o o eo
-.
Cl

C4

Oi
eo
eo -x
Cl

o>

o - ** o

o o o o

eo

O N N
O

i eo N Oi
C-t
tO

Ci CO
(0

Ci
*

CO

Oi ts o
CO CO

ts eo to n

r-

is O
tO
Cl

CI

-^

Oi n
to
-*

r*. to Oi r-

N r- Oi o
--> CI

O r*. r i>.

t* r o
CO
eo

-- N
rt

CO

r- r o

-> N

eo Oi
CM W

N CO

eo o

N CO

^-

TO

CO

O O N
CO

CO

Cl

Ok

-*

Oi

N
-

N Is |s

* IO r-

n eo oi
N

eo
eo o eo

r-< r*

N rl Oi
i-i o
o
C*
w*

o o o n o

o n

CO

* Oi

N ~

|S

eo oi c* t^ to
r cn
# to
CI

eo eo
<-

-. *

l- M eo O
CD CN

NOin ON
00 i-

n n -* ~* n
to
-t m

eo Oi ^ o

o o> to h n
ci - to

O N O

~ i- r

p- t^ (O t> o

to eo

to O r- to

OS

r* eo eo r* Oi

O O Oi Oi
CO

-4

^<

O r->

eo

oi i- m
e*
ci

J.

3;

tO

Is

O N Oi N

to O Oi Oi
CN

to eo to
O
JO
-x
CN

CO N
CN t
Cl

N N O

CO N CO

Cl

O Is

Is

N CO
r-C
C*

ts
Ct
T<

CO

I*

OS

* to

CN

eo
O

N
CN

o
eo
n

CM

r-

r-

tO

CO

cm

CO

eo n
CO
eo

o> eo co

CN

CI

oiNt*n
eo to t>- ~+
C*
CN

CO
CO

Q "

ci - eo r*

Oi N
CN

N-O
CN
-.

eo
CO

r- eo

o o eo n

eo

r* o

t-i
eo
O

is

i N O TO

<-

Oi

N ~t |s
r- tO

is Is
rH

CO

CN

Oi CO

Ci O

eo o n

ts r-i ts Oi

~r

Oi
00

N.

CO
CM

CO N N
CN
>-"

eo

N N

CN

1-1
C2

o o n -
O
eo

o r- eo o>
CO Oi

r* o . p) o

N eo N- -h
*0 Cl eo to

N O <-* N

N
CO

.=

CM

o o o
00

n :o N
eo ~

tO
CN

r* Oi o
ct eo

Oi r- eo
ci
to so

r- r* o ro o
Oi
rn

r* eo

- oi o m

~ eo eo r*
eo * co

co oi r* o

O N O N -x

Oi Oi r-> N to eo

tO

Oi

Is

cn

CN

J
GO

N N

o cnocn

co

r*

N O N
w*
CN

O (N

TO

ts

'J-

i-i
P) N to n
-eoi

o o o . o

k o n oo
ci
~

to -x o

CO
r-l

Oi eo

eo n o; r
c*
r*

^^ Oi CO OS
-^ Oi

CM ~

(O N Oi

O eo r* eo
.-.
eo

o t- o eo Oi
to
CN

a r- ~ r*. -

CN

o n
pi

co * n

ri oi eo M

- - M O
(D - tO

o> to
o c

r Oi eo n

o o n o

O r 3
o>
t-

tN NO

onoeo

eo -n N O CO N

Qi CO 05 - t*
a -?i co eo

^ ^ ^ tjt uS

d
CN

o
N r-> CJi
CD

-. o
CO

eo r* CO
CO

Oi !> O

Ptrtrt

r-l >

CI

r-t

O N N Oi
CO

o o

fH Jfc- 1-4 10 J>

O O ft <-< r-t

CN

tO

fH CO l> OI CO

Cj -< r* ih eo
kQ CO CON- 1>

n. co co as f- 1>
l> t 00 00 0) 5i

Cl

N O eo

O N N Oi
Cl CN
CN

Oi JO
IO CO

CO

ON
to

CO N N
COO
*
CN

r-l

ll
O

CO
Cl

N Oi r-t N
r- CO

O O Oi

eo

o Oi
Cl

N O to CO
Cl
OI N
CN

eo ts o
W
O

Heor-^eo
o o OO

fJJ r-l N ^H CO
r-t CM Cl TO CO

i> C3S CO
CO CO) -*

f-

Ci

Cl
N

o
r-l

Ci Oi
Cl Cl 00

to o>

CO

eo

90000 99000.
c -

r>- i O O O

* -t

co o o o

<

-t o*.

CO O CO t".
CO
o
<

r* co Oi

c,
cs

o 1^ o
eo

r- O

r- o o. r:
~ o

ro I-* t* ro

O Oi
i
r

-# co Oi S r>. Oi
cs
cs
Cl

r r r>. o o

Cl l-r m

r- o tCO ro
co

r
00

or* ro

r>. ~ ci - eo

-no
OlOO

t- so
*

Ci

O r*

r-* eo t* ro
o ~
eo eo o
to
-*

co

i^

- t _ c
c c* o ^

Ci

<*

CO
00

o r r* o o
M

CI p-i
Cl

n o> r t*; -*
M

C*

pN

CO

Oi CO -

o r* i^ Oi
o cs

n o o>

CO
GO

CO o t
c

CO r>. co r*
50 CO ID ~

o o
t** CO

<

P5 -H C
e
P<

r* r*. eo
te

~ r-

t^. r* o m

r- r*.

" a n r> o ro

O r Ci " r- Oi

i r* m

Ci

Oi

CA I> fo CO

t*

eo

p-

Cl

Ci t^. ro

o <- r ci

ro

O t> ff> O -

Ol
Cl

HF l?+ ~

O Oi r* O t-

eo o co r r*
<-i
ex eo

p- ai o r

i- i- i- eo r*
O
30 ro

r* -.

O O r Oi

eo co

Cl

i% ro t^- ro
^- <o r*

-. 1- l~ T-.

1-

^ O CO t^ Oi
CO CO

Oi t>. O O
C*

CO CCS CO cc

CO

Oi 1lO

Oi
cs

O r-l C* CO co
C*

t^ o

CO -

t>- ro

r* r- o co

oe w

p-

CO

O CO

to

CD
o o

o
c*

O t CO Ci A O

cs

r co oi r*

00
CD

C. ci t>-
CS O

p- *

co

r* p- n

O ro ^ r- eo ro
t - * M tO
c* r*

CO

cs
Oi Ci y.

1^

^ i -
-

GO

O O

to

cs

c
r*- co ~ o
^ C<

o r*

Ci M O f-

c
r* co o
'- c*

r* r- o o oi ro

""

c
ro o r

i Ci Oi r*
fcO O Cs O

O O 1^

ro O co r

C* Oi

r^ ci
CO
cs
CS

t- ai

I~

CO O; r-

V*. Cl

r ^- co
O
-t Cs

r>.
cs

-r

ro i- eo ~ ro eo
CD
-. W

Ci

1-

CO
H r O

or*

r* r- i o O t*
n -i eo
c*
e co r- cs
~ c*
co

i>. r-

en
to

~ O to t^. Oi t*

CO

o
r- eo

o r*- r o o

r* - fo r^
~
n io

ro
CO

O Ci

Oi CO !>

Oi

r- i-* t r- iCO
Ol ~ CO

cs

i - c

:r i

3 i h G ft

c
cs

o r.
(S

CO r ro
~ c*
r

1^-

a n n o r o
F-

-#

O t^ CO

r Oi r
(O
i-t c*

I> l-~ Cl
1*+

CO CO

c*

r^

r- O i
CO

to

o to
o

CO

CO

[^ O M
to

p-

00 M * -

r*. -*

CO

* CD ^-

t Ci ro

t-. ro t-* O
Oi Oi

ro i~- r*.

-J*

ro o

r* o ci o
CO

o r^ ci
cs ~

Ci =

I- ?*.

r- o ci r co

p- r- ro CO

r^ r-

eo r o r>-

to <-

* CO

c
O

rto

t-- r* fr* ~* Cl
C* OS

Ci Ol t-

~- 1- a

e
* co r

r-

CO CO

i- r p- ro O

eo

^
^

VI

ftO -

^ o r* *- r
o
r~
Oi
o o t-

M <o ~

r*. r* Oi

CO
c-i
Oi
CO

Oi o

r^ rB1

GO
CO

c*

r o - pi r

ro i^- i

r- oi t- o
p-

C r- t>.
CO

CO

o r- oi

CO

1^ t* i-

CO

t-

> to

to

eo oi r*- -h t*
o oi eo fh

C r:
tc cs
cs

o ro
-*

- !> CO ~ CO

p- CO

- CO r~ CO
CO co r-

CO CO f

i^

Oi Cl

1CO

o r* Ci oi

ro Oi Oi t"
*
i Ol

O O t- o

CO ~ Ci

3)

r- >- CO O CO

O r* o

O CO o

i^ c,

ro

r oi

Oi o

p- t* o

c ci r*-

i*

O [> Ol fl tf
CS

i^

co ir- '

*-. Oi

i--

Cl

r- ro o r r*

CO

CO
r* > ci

Ci O

ci r- o to
o
w
c*

n a oi r*

O r

r- ~ o ro
o

<- r p- ro

-r

C9
CO

Oi 1M

en

03

on o oin
CD
CI

c*

c* eo r- co
Ci

CO

o o

O I*

o* t^ t^ p- r*
io
con
Ok

-r

r* r^- r^ co
~+ CO
c

^- eo ro
cs

iO
CN

Ci

t* r- r* eo
- c*

r^ r* o

". 1~

eo ro i-

eo

C eo ro r~

CO 1Ci
CN

(^ Cl Ci

eo ro
cs

CO

t^ o

C:

n r* t O ?!

.-: iCO O

c* r.

Cl

o -*

r* --

00

^ * ci

04
.

1 Ci t-

r* r^ c- r- r-

r eo cs
C*

n n h

oi

a i" n o
CD

r- ro r

ro n
O
P

C Ci r-

ro r--

-^

f- CO
~
CO

t ci ci o a

1^ CO
CS

*. T-*

* r* os

oi r o

r* Oi r: o
^- co r~

CO c*
Ol

O 1^ c

o o n n

~ to

r ?. i-

ro o co t* o

r-
Oi

o ~ ~ r*

t^ r* ro

eo oi

O t>-

O ro CO r>.

CS 00 CO

r* i ro ro
,- -j.

Ol t- r^ t>. O
ro
tO -*
W
p-

~+ o r* f' eo

CO O Ol o

z eor~- r^ t* ~

CO

r-

.- co

GO
O

-*

a i^
tO C*

r* ro o o ro

C t- o
to

OS o r r
Ci
CO

-O o

o r* co
fplrtMl

CO

r* Oi O

lO

ro

eo

O ro
00

r* o co o
c -

eo

OOHHN

rt r* C3i co Oi
NNMCOCO

t-H r r-1 00 t
tj tji *a to ta

k O O) O

<>
co r^ eo
3i
tB
'
- t^

* t~ i co r^ C5

o r- *^

~ *

to

X s: Ti t. Si n

Ci co "; ~* t
lO cc cc t> t-

i-t co r rji co Ci
00 00 00 X O Ci

30
n

Log T {n).
0

97497 95001 92512 90030 87555 85087 82627 80173

1.00
1.01 9.9975287 72855 70430 68011 65600 63196 60799 58408 56025
51279 48916 46561 44212 41870 39535 37207 34886 32572
1.02
27964 25671 23384 21104 18831 16564 14305 12052 09806
1.03
05334 03108 00889 98677 96471 94273 92080 89895 87716
1.04
1.05 9.9883379 81220 79068 76922 74783 72651 70525 68406 66294
62089 59996 57910 55830 53757 51690 49630 47577 45530
1.06
41469 39428 37407 35392 33384 31382 29387 27398 25415
1.07
21469 19506 17549 15599 13655 11717 09785 07860 05941
1.08
1.09
02123 00223 98329 96442 94561 92686 90818 88956 87100

9
77727
53648
30265
07567
85544
64188
43489
23449
04029
85250

1.10 9.9783407 81570 79738 77914 76095 74283 72476 70676 68882 67095
1.11
65313 63538 61768 60005 58248 56497 54753 53014 51281 49555
1.12
47834 46120 44411 42709 41013 39323 37638 35960 34288 32622
30962 29308 27659 26017 24381 22751 21126 19508 17896 16289
1.13
14689 13094 11505 09922 08345 06774 05209 03650 02096 00549
1.14
1.15 9.9699007 97471 95941 94417 92898 91386 89879 88378 86883 85393
83910 82432 80960 79493 78033 76578 75129 73686 72248 70816
1.16
69390 67969 66554 65145 63742 62344 60952 59566 58185 56810
1.17
55440 54076 52718 51366 50019 48677 47341 46011 44687 43368
1.18
42054 40746 39444 38147 36856 35570 34290 33016 31747 30483
1.19
29225 27973 26725 25484 24248 23017
1.20
1.21
16946 15748 14556 13369 12188 11011
1.22
05212 04068 02930 01796 00669 99546
1.23 9.9594015 92925 91840 90760 89685 88616
1.24
83350 j 82313 81280 80253 79232 78215
73211 72226 71246 70271 69301 68337
1.25
1.26
63592 62658 61730 60806 59888 58975
1.27
54487 53604 52727 51855 50988 50126
1.28
45891 45059 44232 43410 42593 41782
1.29
37798 37016 36239 35467 34700 33938

21792
09841
98430
87553
77204
67377
58067
49268
40975
33181

20573
08675
97318
86494
76198
66423
57165
48416
40173
32439

19358
07515
96212
85441
75197
65474
56267
47570
39376
31682

18150
06361
95111
84393
74201
64530
55374
46728
38585
30940

1.30
30203 29470 28743 28021 27303
1.31
23100 22417 21739 21065 20396
16485 15850 15220 14595 13975
1.32
1.33
10353 09766 09184 08606 08034
1.34
04698 04158 03624 03094 02568
1.35 9.9499515 99023 98535 98052 97573
94800 94355 93913 93477 93044
1.36
90549 90149 89754 89363 88977
1.37
1.38
86756 86402 86052 85707 85366
83417 83108 82803 82503 82208
1.39

26590
19732
13359
U7466
O2048
9710o
92617
88595
85030
81916

25883
19073
12748
06903
01532
96630
92194
88218
84698
81630

25180
18419
12142
06344
01021
96166
M776
87846
84371
81348

24482
17770
11540
05791
00514
95706
91362
87478
84049
81070

23789
17125
10944
05242
00012
95251
90953
87115
83731
80797

79250
77027
75243
73894
72976
72484
72416
72766
73531
74708

79008
76829
75089
73783
72908
72459
72432
72824
73630
74848

78770
76636
74939
73676
72844
72437
72452
72886
73734
74992

78537
76446
74793
73574
72784
72419
72477
72952
73841
75141

78308
76261
74652
73476
72728
72406
72506
73022
73953
75293

1.40
1.41
1.42
1.43
1.44
L.45
1.46
1.47
1.48
1.49

80528
78084
76081
74515
73382
72677
72397
72539
73097
74068

80263
77864
75905
74382
73292
72630
72393
72576
73175
74188

80003
77648
75733
74254
73207
72587
72392
72617
73258
74312

79748
77437
75565
74130
73125
72549
72396
72662
73345
74440

79497
77230
75402
74010
73049
72514
72404
72712
73436
74572

Note.This table ie taken from Vol. II. of Logundrc's work, and not
from Vol. I., as statd in the T'rp'hre: the nnmhi rs g'v>;n in Vol. X. being
inarcurate in the seventh decimal place. In Vol If. the values are given to
twelve places of decimals. The figure here printed in the 8. venth place is

Leg T (n).
n

31
5

1.50 9.9475449 75610 75774 75943 76116 76292 76473 76658 76847 77040
1.51
77237 77438 77642 77851 78064 78281 78502 78727 78956 79189
79426 79667 79912 80161 80414 80671 80932 81196 81465 81738
1.52
82015 82295 82580 82868 83161 83457 83758 84062 84370 84682
1.53
84998 85318 85642 85970 86302 86638 86977 87321 87668 88019
1.54
88374 88733 89096 89463 89834 90208 90587 90969 91355 91745
1 55
92139 92537 92938 93344 93753 94166 94583 95004 95429 95857
1.56
96289 96725 97165 97609 98056 98508 98963 99422 99885 00351
1.57
158 9.9500822 01296 01774 02255 02741 03230 03723 04220 04720 05225
1.59
05733 06245 06760 07280 07803 08330 08860 09395 09933 10475
1.60
1.61
1.62
1.63
1.64
1.65
1.66
1.67
1.68
1.69

11020
16680
22710
29107
35867
42989
50468
58303
66491
75028

11569
17267
23333
29767
36563
43721
51236
5910G
67329
75901

12122
17857
23960
30430
37263
44456
52007
59913
68170
76777

12679
18451
24591
31097
37966
45195
52782
60723
69015
77657

13240
19048
25225
31767
38673
45938
53560
61536
69864
78540

13804
19650
25863
32442
39383
46684
54342
62353
70716
79427

14372
20254
26504
33120
40097
47434
55127
63174
71571
80317

14943
20862
27149
33801
40815
48187
55916
63998
72430
81211

15519
21475
27798
34486
41536
48944
56708
64826
73293
82108

16098
22091
28451
35175
42260
49704
57504
65656
74159
83008

83912 84820 85731 86645 87563 88484 89409 90337

1.70
93141 94083 95028 95977 96929 97884 98843 99805
1.71
1.72 9.9602712 03688 04667 05650 06636 07625 08618 09614
1.73
12622 13632 14645 15661 16681 17704 18730 19760
1.74
22869 23912 24959 26009 27062 28118 29178 30241
33451 34527 35607 36690 37776 38866 39959 41055
1.75
44364 45473 46586 47702 48821 49944 51070 52200
1.76
55606 56749 57894 59043 60195 61350 62509 63671
1.77
67176 68351 69529 70710 71895 73082 74274 75468
1.78
79070 80277 81488 82701 83918 85138 86361 87588
1.79

91268
00771
10613
20793
31308
42155
53331
64836
76665
88818

92203
01740
11616
21830
32377
43258
54467
66004
77866
90051

91287 92526 93768 95014 96263 97515 98770 O0029 01291

1.80
1.81 9.9703823 05095 06369 07646 08927 10211 11498 12788 14082
16678 17981 19287 20596 21908 23224 24542 25864 27189
1.82
29848 31182 32520 33860 35204 36551 37900 39254 40610
1.83
43331 44697 46065 47437 48812 50190 51571 52955 54342
1 84
57126 58522 59922 61325 62730 64140 65551 66966 68384
1.85
71230 72657 74087 75521 76957 78397 79839 81285 82734
1.86
85640 87098 88559 90023 91490 92960 94433 95910 97389
1.87
1.88 9.9800356 01844 03335 04830 06327 07827 09331 10837 12346
15374 16893 18414 19939 21466 22996 24530 26066 27606
1.89

02555
15378
28517
41969
55733
69805
84186
98871
13859
29148

30693 32242 33793 35348 36905 38465

1.90
46311 47890 49471 51055 52642 54232
1.91
62226 63834 65445 67058 68675 70294
1.92
78436 80073 81713 83356 85002 86651
1.93
94938 96605 98274 99946 01621 03299
1.94
1.95 9.9911732 13427 15125 16826 18530 20237
28815 30539 32266 33995 35728 37464
1.96
46185 47937 49693 51451 53213 54977
1.97
63840 65621 67405 69192 70982 72774
1.98
1.99
81779 83588 85401 87216 89034 90854

44736
60622
76802
93275
T0039
27093
44435
62062
79972
98165

40028
55825
71917
88302
04980
21947
39202
56744
74570
92678

41595
57421
73542
89957
06663
23659
40943
58513
76368
94504

43164
59020
75170
91614
08350
25375
42688
60286
78169
96333

the ono nearest to the true value whether in excess or defect. This table, and
the table of Least Factors, have each been subjected to two complete and in
dependent revisions before finally printing off.

1
:

ALGEBRA.

FACTORS.
1

a2-b2 = (a-b) (a+b).

a*-b3 = (a-b) (a2+ab+b2).

a3+63 = (a+b) (a2-ab+b2).

And generally,
4 anbn = (ab) (a"-l+an-2b+ ... + bn-1) always.
5

a"bn = (a+b) (a"-1 a"-2b+ ...

if n be even.

a* + bn = (a+b) (an-1a"-ib+ ...+b"-1) if n be odd.

(r+a) (tr+6) = ,r2+(a+&) tr+a6.

8
9

(#+a)
U+c) = o?+(a+b+c)x2
(+6)2 = 2+2a6+62.
+(bc+ca+ab)x+abc.

10
11

(a-6)2 = a2-2a&+62.
(a+b)3 = a3+3a26+362+&3 = as+63+3a& (a+b).

12

(a-5)8 = a3-3a26+3a&2-63 = 3-63-3a6 (a-b).

Generally,
(a6)7=a77a66+21a56235a463+35a36421a266+76667.
Newton's Rule for forming the coefficients : Multiply any
coefficient by the index of the leading quantity, and divide by
the number of terms to that place to obtain the coefficient of the
term next following . Thus 21X5-;- 3 gives 35, the following
coefficient in the example given above.
See also (125).
To square a polynomial : Add to the square of each term
twice the product of that term and every term that follows it.
Thus,
(a+b+c+d)*
- at+2a(b+c+d)+b2+2b(c+d)+c2+2cd+d2.
F

ALGEBRA.

34

13

a4-f a2b2+b* = (a2+ab+b2) (al-ab+b2).

14

a*+b* = (a*+ab V2+b2) (a*-ab </2+b2).

16
17

(a+b+c)2 = a2+b2+c2+2bc+2ca+2ab.
(+ft+c)s = a8+63+c3+3(62c4-6ci+c2a+cas
+a26+a62)+6a6c.

Observe that in an algebraical equation the sign of any

letter may be changed throughout, and thus a new formula
obtained, it being borne in mind that an even power of a
negative quantity is positive. For example, by changing the
sign of c in (16), we obtain
(a + b-c? = a* + b2 + c2-2bc-2ca+2ab.

18

a2+b2-r+2ab = (a+b)2-c2 = (a+b+c) (a+b-c)

by (l).

19

a2-b2-c2+2bc = a2-(b-c)2 = (a+b-c) (a-b+c).

20

a3+63+c8-3afec = (a+b+c) (a2+b2+c2-bc-ca-ab).

21

bc2+b2c+ca2+c2a+ab2+a2b+a!i+b*+ci

22

= (a+b+c) (a'+V+c2).
bc2+b2c+ca2+cia+ab2+a2b+'Mbc

23

= (a+b+c) (bc+ca+ab).
bc2+bic+ca2+cia+ab2+a2b+2abc=(b+c)(c+a)(a+b)

24

6c2 + blc + ca2 + c2a + ab2 + a2b - 2abc - as - - 1?

= (b+ca) (c+a-b) (a+bc).

25

b^-tfc + ca'-c'a + aV-a'b = (b-c) (c-a) (a-b).

26

2b2c2+2c2a2+2a2b2-a*-b*-c*
= (a+b+c) (b+c-a) (c+a-b) (a+b-c).
Js+2a?y+2xy2+y* = (oc+y) (rf+xy+tf).

27

Generally for the division of (x + y)n (xn + yn) by x2 +xy + y2

see (545).

MULTIPLIOATION AND DIVISION.

._^

J
MULTIPLICATION AND DIVISION,
BY" THE
J METHOD OF DETACHED COEFFICIENTS.

28

Ex. 1:

35

(at-3atbt + 2abt+b*)x(ai-2ab,-2bs).
1+0-3+2+1
1+0-2-2
1+0-3+2+1
-2-0+6-4-2
-2-0+6-4-2
1+0-5+0+7+2-6-2
Result a7-baibt + 7a"b, + 2a,bs- 6a6"-267.

Ex. 2:

(x7-5zi + 7x*+2xi-6z-2) + (xt-3z' + 2z + l).

1+0-3+2+1)

1+0-5+0+7+2-6-2(1+0-2-2
-1-0+3-2-1
0-2-2+6+2-6
+2+0-6+4+2
-2+0+6-4-2
+2+0-6+4+2

Result zs 2x ct

Synthetic Division.
Ex. 3 : Employing the last example, the work stands thus,
-0
+3
-2
-1

1+0-5+0+7+2-6-2
0+0+0+0
+3+0-6-6
-2+0+4+4
-1+0+2+2
1+0-2-5

Result x" -2x 2.

[See also (248).

Note that, in all operations with detached coefficients, the result must be
written out in successive powers of the quantity which stood in its successive
powers in the origiual expression.

30

ALGEBRA.

INDICES.
29

Multiplication:

aiX a* = ai + i = a*, or /ae;

a" Xa"= a" "= a"", or

Division :

a'-i-o* = a*~* = a*, or

-Ii_ 1
>-n
aan , or
a" -+a = a"

Involution :
Evolution :

(a*)* ="* = a*, or

l/a* = a* x 1 = a*
or %*.
a- = L
a"

a = l.

HIGHEST COMMON FACTOR.

30 Rule.To find the highest common factor of two ex
pressions : Divide the one which is of the highest dimension by
the other, rejecting first any factor of either expression which
is not also a factor of the other. Operate in the same manner
upon the remainder and the divisor, and continue the process
until there is no remainder. The last divisor ivill be the
highest common factor required.
31

Example.To find the H. C. P. of

3*,-10r'+15.K + 8 and
2xt-6x'+ ia> + l3z+6.
1- 2- 6+ 4 + 13+ 6
3 + 0-10+ 0 + 15+ 8
3
-3 + 6 + 18-12-39-18
3- 6-18 + 12+39 + 18
2)6+ 8-12-24-10
-3- 4+ 6 + 12+ 5
3+ 4- 6-12- 5
2 ) -10-12 + 24 + 44 + 18
-3- 9- 9- 3
- 5- 6 + 12 + 22+ 9
- 5-15-15- 5
3
+ 5 + 15 + 15+ 5
-15-18+36 + 66 + 27
+ 15 + 20-30-60-25
2 )2+ 6+ 6+_2
1+ 3+ 3+ 1

Result H. C. F. = a* + 3x> + 3z + l.

EVOLUTION.

37

32 Otherwise. To form the H. C. F. of two or more

algebraical expressions : Separate the expressions into their
simplest factors. The H. (J. F. ivill be the product of the
factors common to all the expressions, taken in the loivest
powers that occur.
LOWEST COMMON MULTIPLE.

33 The L. C. M. of two quantities is equal to their product

divided by the H. C. F.
34 Otherwise. To form the L. C. M. of two or more
algebraical expressions : Separate them into their simplest
factors. The L. C. M. xoill be the product of all the factors
that occur, taken in the highest powers that occur.
Examine.The H. C. F. of a\b-x)^d and as(6-z)Ve is aJ(6-a:)V;
and the L. C. M. is a'(b - a;)'Vde.

EVOLUTION.
To extract the Square Root of
. 3a Va

3-/a , 41a ,-,

Arranging according to powers of a, and reducing to one denominator, the

.
,becomes
expression
35

16a5- 24a' +
41a -24a* + 16 .

Detaching the coefficients, the work is as follows :

16-24 + 41-24 + 16(4-3+4
16
8-3
-3
8-6+4

_
,
Result

-24 + 41
24- 9
32-24 + 16
-32 + 24-16
4a 3a*
, , 1,
~. + 4 = a-|, va+

ALGEBRA.

38

To extract the Cube Root of

37

S^-SoVv/y + 66^-63^ v/y +33*y-9^yy+y.

The terms here contain the successive powers of x and \/y ; therefore,
detaching the coefficients, the work will be as follows:
I.
II.
III.
63 7
12
8 -36 + 66- 63 + 33 - 9 + 1 (2 - 3+1
-6 J
-18+ 9)
-8
6-9 + 1

12-18+ 9 1
+ 9)

-36 + 66-63 + 33-9 + 1

+36-54 + 27

12-36 + 27
6-9+1

12-36 + 33-9 + 1
-12+36-33 + 9-1

12-36 + 33-9 + 1
Result

2x'3x>/y + y.

Explanation.The cube root of 8 is 2, the first term of the result.

Place 3 x 2 = 6 in the first column I., 3 x 21 = 12 in column II., and 2= 8
in III., changing its sign for subtraction.
36-7-12 = 3, the second term of the result.
Put -3 in I.; (6-3) x (-3) gives -18+9 for II.
(12-18 + 9) x 3 (changing sign) gives 36-54 + 27 for III. Then add.
Put twice (3), the term last found, in I., and the square of it in II.
Add the two last rows in I., and the three last in II.
12-T-12 gives 1, the third term of the result.
Put 1 in col. I., (6-9 + 1) x 1 gives 6-9 + 1 for col. II.
(1236 + 339 + 1) x 1 gives the same for III. Change the signs, and
add, and the work is finished.

The foregoing process is but a slight variation of Horner's

rule for solving an equation of any degree. See (533).

Transformations frequently required.

38

a
c then
T
-7,
If
= 4
o
a

a-\-b = !c4-d
L_
^=^5
ab
ccd
a

39

If

*+3, = n

^ ^ = i(+<0.

and ocy = b)

= \ {ab).

40

(*+yY+ (.r-y)2 = 2 (J+y2).

41

(x+yY-(x-yy =

[68.

EQUATIONS.

42

(*+yY = (xyy+4sxy.

43

{*-yf = (x+yy-4w

44

Examples.

39

.[38.
2^/o,-6, 3vV-b*
9(0'-^) = 4(6,-d1),
a; = v/5c,+4?.
To simplify a compound fraction, as
1
-ab + b* + at + ab + V
~1
T
'
a'ab+b' ct + ab + V
multiply the numerator and denominator by the L. C. M. of all the smaller
denominators.
Result
(a?+abb<+(a>-ab + b>) a' + 6'
(o, + o6 + 6,)-(o,-a6+6,)
ab '

QUADRATIC EQUATIONS.

45

If

a*+6H-c = 0,

46

If

a.ri+26.r+c = 0;

i
an even number,

47

<r =

* = -6v^-4c.
that is, if the coefficient of a; be

6 + \/62 ac

.
a

Method of solution without the formula.

Ex.:

Divide by 2,

23?-7x+3 = 0.
a?-

-| = 0.

40

ALGEBRA.

Complete the square,

ar -z + ^J = = .

Take square root,

x = ~,
x=_ =3

or -.

48 Rule for "completing the square" of an expression like

a?\x : Add the square of half the coefficient of x.

49

The solution of the foregoing equation, employing formula (45), is

1
x -6v/6'-4ac = 7v/49-24 = 75_
o or .
2a
4
4
2

THEORY OF QUADRATIC EXPRESSIONS.

If a, /3 be the roots of the equation ax^+bx+c = 0, then

50

aa*+bx+c = a (xa)(xB).

51

Sum of roots

a+/8 =
a

52

Product of roots

aB = .
a

Condition for the existence of equal roots

53

b2 4ac must vanish.

54 The solution of equations in oue unknown quantity may

sometimes be simplified by changing the quantity sought.

*,(!

&+^; + j^_=
6V + 5J-1
3x+l

a).

6('3.c + l) _
6x> + bz-l

= 5^

<*>

EQUATIONS.

thus

41

= 14.
y
y'-liy + 6 = 0.
y having been determined from this quadratic, x is afterwards found from (2).

55

y -\

Ex.2:

gf+L+x+L = 4,
x
x

Put x + = y, and solve the quadratic in y.

56

Ex.3:

*? + x+-y2x' + x + 2 = ^+l.
2xi+x + Z^2xt+x + 2 = 2,

Put

2x* + x + 2 + 3</2x,+x + 2 = 4.
^2x* + x + 2 = y, and solve the quadratic
y* + 3y = 4.

57

Ex.4:
as* 4- z3 - 16
+ 3
" 3'

A quadratic in
-

58

To find Maxima and Min ima values by mean s of a

Quadratic Equation.

Ex.Given
y = 3 c! + 6x + 7,
to find what value of x will make y a maximum or minimum.
Solve the quadratic equation
2x, + 6x+7-y = 0.
Thus

x = -3A,-12

[46

In order that x may be a real quantity, we must have 3y not less than 12 ;
therefore 4 is a minimum value of y, and the value of x which makes y a
minimum is 1.
G

42

ALGEBRA.

SIMULTANEOUS EQUATIONS.

General solution with two unknown quantities.

Given
59

aix+b1y = cl\

r_ cL68 c26^

_ c,a, ^a.

General solution with three unknown quantities.

60

Given

alx-\-bly-\-clz = d^
a.iX-^-h.zy+CiZ = d2 > ,

O! (6,c3 63c2)+a2 (63^ 6if3)+3 (biC3 62c,)'

and symmetrical forms for ?/ and 2.

Methods of solving simultaneous equations between two

unknown quantities x and y.

61 I. By substitution.Find one unknown in terms of the

other from one of the hvo equations, and substitute this value
in the remaining equation. Then solve the resulting equation.
Ex.:
From (2), y = 4.

62

x + 5y = 23
(1)]
7y = 28
(2) ) "
Substitute in (1) ; thus
x + 20 = 23, .r=3.

IL By the method of Multipliers.

Ex.:

8as
+ 5y = 365
2x-Sy=

(1)
)
(:>)}

Eliminate x by multiplying eq. (1) by 2, and (2) by 3; thus

6.-r + 102/ = 72,
6x 9y = 15,
1% = 57, by subtraction,
'J = 3;
x = 7, by substitution in eq. (2}.

EQUATIONS.

63

III- By changing the quantities sought.

Ex. 1:

x-y- 2
(1)7
x'-y'+x + y = 30
(2) ) '
Let
x + y = u, x y = v.
Substitute these values in (1) and (2),
v = 2)
uv + u = 30) '
.-. 2tt + = 30,
u = 10 ;
.'.
s + y=10,
x-y = 2.
From which
x = 6 and y = 4.

64

Ex.2:

2^+10?^= 9
(1)
a; y
+y
*+7y' = 64
(2) J
Substitute z for 'L^l in (1) ;
xy
,.
2,+ L0 = 9;
z
2z,-9z+10 = 0.

From which

z = A or 2,
x+y
xy

From which
Substitute in (2) ; thus
or

65

5
2

7 y.
a; = 3y or o
y = 2 and a: = 6,
2/ = 7^ and x = 2 \/7.
v7

Ex.3:

3a; + 5y = zy
2x + 7y = 3xy
Divide each quantity by xy ;
1+A-1
y
x

(1) ]
(2) )
(3)

-+-=3
(4)
y
x
Multiply (3) by 2,- and (4) by 3, and by subtraction y is eliminated.

43

ALOEBRA.

44

66

IV. By substituting y = tx, when the equations are

homogeneous in the terms which contain x and y.
Ex. 1:

52x* + 7xy = by'

(1) ~)
5x-3y = 17
(2) ) '
From (1),
52z* + 7tx> = 5<V
(3) ~)
and, from (2),
bxZlx = 17
(4) ) '
(3) gives
52 + 7< = 5<*,
a quadratic equation from which t must be found, and its value substituted
in (4).

x is thus determined ; and then y from y = tx.

67

Ex.2:

2x*+xy + 3yi

3y2x
From (1), by putting y = tx,
a?(2 + f+3<J)
from (2),
x(3t-2)
squaring,
3? (9^-12f + 4)
9i'-12f + 4
a quadratic equation for
t being found from this, equat:on

16
4

(1)7
(2)5"

16
(3) 7 .
4
(4) 5 '
16;
2 + f+3<*,
) will determine x ; and finally y = tx.

RATIO AND PROPORTION.

68

If a : b :: c : d; then ad = be, and

a-\-b _ c+d
b
d '

on
69

a b _ cd
~b
d~;

!L ;
0
d
a+b _c+d
a=b~c~=d'

ic
c

p
ti.

a-|-c+e+&c.
" T=
= 7=&C'; tbenT = 64-rf+/4-&cGeneral theorem.

70

If 4r = 4- = A = &c. = /c say, then

b
d
J
k _ \ pan+qc'>+ren+&c.} j

where p, 9, r, &c. are any quantities whatever.

in (71).

Proved as

RATIO AND PROPORTION.

45

71 Rule.To verify any equation between such proportional

quantities : Substitute for a, c, e, Sfc, their equivalents hb, kd,
kf, Sfc. respectively, in the given equation.
Ex.If a : b :: c : d, to show that
1/0 ft
y/a s/b
/cd
Vc^/d
Pat kb for a, and kd for c ; thus
v/o-/>_ Vkb-b _ Sb v/fe^l _ y/6 .
/c^d Vkd-d VdJk^l ^d'
also
a,

-v/gy/b _ Skby/b _ Vd(Sk-l)

Vb (-y/fc-1) _ v/&
s/cs/d
</kd-y/d
</d'

Identical results being obtained, the proposed equation must be true.

72

Hal b I el.di e &c, forming a continued proportion,

then

a '. c '.: a2 '. b2, the duplicate ratio of a I b,

a l d'.l a3 : bs, the triplicate ratio of a '. b, and so on.

Also

v'a : v'ft is the subduplicate ratio of a : b,

a1 : 6*

73

is the sesquiplicate ratio of a : b.

The fraction

is made to approach nearer to unity in

value, by adding the same quantity to the numerator and

denominator. Thus
a+x
, ,
a .
-r-1 is nearer to 1 than -7- is.
b+x
b
74 Dev.The ratio compounded of the ratios a : b andc : d
is the ratio ac : bd.
75 If a : b :: c : d, and a : b' :: c' : d' ; then, by compound
ing ratios, aal : bb' :: cc' : dd'.
VARIATION.
76

If a <xc and b<xc, then (a+6) a c and Vab oc c.

77
.1
.j and,
a a
ft .
.If a a 6?
. >, then
ac a bd
and coca)
c
a
78

If a a 6, we may assume a = nib, where ?;& is some constant;

46

ALGEBRA.

ARITHMETICAL PROGRESSION.

Generalform of a series in A. P.
79

a, a + d, a+2d, a+3d,

a + (n l)d.

a = first term,
d common difference,
/ = last of n terms,
* = sum of n terms ; then
80

l=a+{n-l)d.

81

# = (a+l)|.

82

*={2a+(n-l)d}.

Proof.By writing (79) in reversed order, and adding both series

together.

GEOMETRICAL PROGRESSION.

General form of a series in G. P.

83

a, ar, ar", ar3,

arn'\

a = first term,
r = common ratio,
I last of n terms,
* = sum of n terms ; then
84

l = ar*1.

oc
85

s= a

r* 1
r1

or

1-r"
a
.
1 r

If r be less than 1, and n be infinite,

86

* = r > since r* = 0.
1 r
Proof. (85) is obtained by multiplying (83) by r, and subtracting one
peries from the other.

47

PERMUTATIONS AND COMBINATIONS.

HARMONICAL PROGRESSION.

87

i b, c, d, &c. are in Harm. Prog, when the reciprocals

, ~, , -4-, &c. are in Arith. Prog.,

abed
88 Or when a : b :: a b : b c is the relation subsisting
between any three consecutive terms.
89

nth term of the series = ,

"6
-.
(n l)o (n 2)

90

Approximate sum of n terms of

the

[87, 80.
Harm.

Prog.

,, ^-r, &c, when d is small compared with a,

_ (a+rf)"-an

PROor.By taking instead the G.P. ?, + ; a ls, H

a - . + ... .
'
a + d (a+d)1 (a + d)'

91

Arithmetic mean between a and 6 = a~j"A

92

Geometric

do.

= v ab.

93

Harmonic

do.

= ^r.
a+b

The three means are in continued proportion.

PERMUTATIONS AND COMBINATIONS.

94
time

The number of permutations of n things taken all at a

= n (n 1) ( 2) ... 3 . 2 . 1 = n ! or n(">.

Proof dy Induction.Assume
Now take n + 1 things. After each
arranged in n ! ways, making in all
n + 1 things; therefore, &c. See
Induction.

tho formula to be true for n things.

of these the remaining n things maybe
nXn! [that is (n + 1) !] permutations of
aUo (233) for the mode of proof by

ALGEBRA.

48

95 The number of permutations of n things taken r at a

time is denoted by P (n, r).
P (n, r) = n (n 1) (n-2) ... (n-r+1) =n(r).
Proof.By (94) ; for (n r) things are left out of each permutation ;
therefore P (n, r) = n ! -r- (n r) ! .
Observe that r = the number of factors.

96 The number of combinations of n things taken r at a

time is denoted by C (n, r).
U l*' r;

1.2.3 ...r

= TT

n!
= C (n, n r).
r ! (n r)
For every combination of r things admits of r ! permuta
tions; therefore C (n, r) = P (n, ?) h- r !

97 0 (n> r) is greatest when r = fyi or % (n+\), according

as n is even or odd.

98 The number of homogeneous products of r dimensions

of n things is denoted by H(n, r).
H(n rx
1 ' }

n (n+1) (n+2) ... Q+r-1) _ (n+r- 1)"

1.2 ...r
=
~H
*

When r is > n, this reduces to

qq
9a

(r+1) (r+2) ... (n + r-l)

(V-l)!
n

Proof.H(n, r) is equal to the number of terms in the product of the

expansions by the Bin. Th. of the n expressions (I ax)'1, (1 bx)'1,
(1-cz)-', Ac.
Put a = I = c = Ac. = 1. The number will be the coefficient of xr in
(1-*)-". (128, 129.)

PERMUTATIONS AND COMBINATIONS.

49

100 The number of permutations of n things taken all to

gether, when a of them are alike, b of them alike, c alike, &c.
n!
a ! b ! c! ... &c.
For, if the a things were all different, they would form a\
permutations where there is now but one. So of b, c, &c.
101 The number of combinations of n things r at a time,
in which any p of them will always be found, is
= C (n p, rp).
For, if the p things be set on one side, we have to add to them
rp things taken- from the remaining np things in every
possible way.
102

Theorem:

C(n-1, r-l)+C(n-l, r) = C(n,r).

Peoof by Induction ; or as follows : Put one out of n

letters aside; there are 0(n 1, r) combinations of the re
maining n 1 letters r at a time. To complete the total
C(n, r), we must place with the excluded letter all the com
binations of the remaining n 1 letters r 1 at a time.
103 If there be one set of P things, another of Q things,
another of R things, and so on ; the number of combinations
formed by taking one out of each set is = PQR ... &c, the
product of the numbers in the several sets.
For one of the P things will form Q combinations with
the Q things. A second of the P things will form Q more
combinations ; and so on. In all, PQ combinations of two
things. Similarly there will be PQR combinations of three
things; and so on. This principle is very important.
104 On the same principle, if p, q, r, &c. things be taken
out of each set respectively, the number of combinations will
be the product of the numbers of the separate combinations ;
that is,

=C(Pp).C (Qq) . C (Rr) . . . &c.

ii

50

ALGEBRA.

105 The number of combinations of n things taken to at a

time, when p of the n things are alike, q of them alike, r of
them alike, &c, will be the sum of all the combinations of
each possible form of to dimensions, and this is equal to the
coefficient of xm in the expansion of
(1 +x + x2+ ... +a?) (1 +x+x?+ ... +x")(l+x+a?+ ... +*r)--- .
106 The total number of possible combinations under the
same circumstances, when the n things are taken in all ways,
1, 2, 3 ... n at a time,

= (H-l)(?+l)(r+l)...-l.
107 The number of permutations when they are taken to
at a time in all possible ways will be equal to the product of
to ! and the coefficient of xm in the expansion of
f, . . * . a5 .
, xf] L , , x% , Xs .
, a;l
.&c.

SURDS.

108 To reduce \/2808. Decompose the number into its

prime factors by (360) ; thus,
^28118 = V-A* . 3" . 13 = 6 V13,
Vu>h b">di = av 6V c = a5 b> c' b' c' = a V c1 Vbc*.

109

To bring 5%/3 to an entire surd.

5J/3 = V'5'.3= Vl875,
xl yl z' = zn y* zn = v/ar0 y zl:

110

To rationalise fractions having surds in their

denominators.
J_ _ y7

_1

/49

_ |/49

SURDS.

111

y-^r0 = 3JS^ = 3(9+^

since

(9- ,/80) (9 + ^80) = 81 - 80, by (1).

112

1+2./3-./2

51

_ i+2y3+y2 _ i+2y3+y/2
(l + 2v/3)J-2
11 + 4^3
(1 + 2^/3 + y2)(ll-4y/3)
73

113

/3-,/2

3-2"

Pnt 3* = x, 21 = y, and take 6 the L. C. M. of the denominators 2 and 3, then

1 _a*+*y+aY + gY+yy* + jf by(4).
.. ,
therefore

1
8*+8*2*+8l2*+32*+3l2*+2*

-=
;
i
3l-2*
3-2*
= 3y9 + 3y72 + 6 + 2e/648 + 4y3+4v/2.

114-

Here the result will be the same as in the last example

if the signs of the even terms be changed.

[See 5.

115 A surd cannot be partly rational ; that is, y/a cannot

be equal to ^/b + c.
Proved by squaring.
116

The product of two unlike squares is irrational ;

v/7x y/3 = n/2], an irrational quantity.

117 The sum or difference of two unlike surds cannot

produce a single surd; that is, ^a-\-y/b cannot be equal
to \/c.
By squaring.
118

If a + ^/m = b + v/n ; then a = b and m = n.

Theorems (115) to (118) are proved indirectly.

119

If

Va+ Vb = \/x+ v'y,

then
\/a V6 = Vx s/y.
By squaring and by (118).

ALGEBRA.

52

120

To express in two terms v/7 + 2^6.

Let
then

x+y = 7

and

/7+2/6 = /*+ /y ;
by squaring and by (118),

x-y = y/V-(2^/6y = /49-24 = 5, by (119) ;

.. * = 6 and y = 1.
Result

/6 + 1.

General formula for the same

121

VaVb = >/\{a+Va*-b) V}(a-vV-ft).

Observe that no simplification is effected unless a2 b is a

perfect square.

122

To simplify

Assume
Let

Va+ /6.
Va+ */b = x-\- /y.
c Va* b.

Then x must be found by trial from the cubic equation

4r*3cx = a,
and

y afc.
No simplification is effected unless aa b is a perfect cube.

Ex.1:

S/7 + 5/2 = x+ /y.

c = /49-50 = -1.
4c' + 3* = 7;
Result

Ex. 2:
Cubing,

= 1, y = 2.
1 -t- /2.

/gyS-ll^ = /*+Vy, two different surds.

9/3-11/2 = x /*+ 3js /y + 3y /as + y/y ;
9/3 = (z-r3y)/a!] .
11/2 = (3x + y)/y 3 '
.-. j = 3 and y = 2.

(118)

53

123 To simplify
v/(12+4./3 + 4v/5 + 2./lo).
Assnme
./(12 + 4./3 + 4,/5+2-/15) = y/x+ </y+
Square, and equate corresponding surds.
Result ^3+ y4+ v/5.

124 To express \/A+B in the form of two surds, where A

and B are one or both quadratic surds and n is odd. Take q
such that q (AiBi) may be a perfect 7ith power, say pn, by
(361).

Take s and t the nearest integers to \Zq(A+By

and X/q (A B)*, then

VA+B=-- {V*+t+2p s/s+t-2p).

Example :
Here

To reduce ^89^3 + 109^/2.

4 = 89^/3, B = 109^2,
.4' 5* = 1 ; .-. p = 1 and q = 1.
Vg (A + B)* = 9 +/ )
/ being a proper fraction ;

!/q (A- By = 1-/1 '

Result

.-. f = 9, t = 1.

| ( v/9 + 1 + 2 v^ + 1-2) = ^3 + v/2.

BINOMIAL THEOREM.

125

(+&)" =

a-+a-6+ W("71} a W+ w
^!
126 General or (r + l)th term,

~2> -'&+ &c.

o!

n(n 1) (n 2) ... (n r+1) _r , ,

r!
127

or

.a-'b*
(n r) I rl
if n be a positive integer.
If b be negative, the signs of the even terms will be changed.

54

ALGEBRA.

If n be negative the expansion reduces to

128
(a+brn =
i
a -n" Aa -i-u
"
129

...in
t
" *6 n(n+l)(n+2)
!^!i-a ...,.,
" sos+&c.

General term,
(_1)rn(4-l)(n4-2),..(n+r-l)q_.,r6r

Eider's proof.Let the expansion of (l+.r)n, as in (125),

be called f(n).
equation

Then it may be proved by Induction that the

f(m) */() = /(+*)
(1)

is true when m and n are integers, and therefore universally

true ; because the form of an algebraical product is not altered
by changing the letters involved into fractional or negative
quantities. Hence
f(m+n+p + &c.) =f(m)Xf(n)Xf(p), &c.
Put m = n =p = &c. to h terms, each equal -r-% and the
theorem is proved for a fractional index.
Again, put n for m in (1) ; thus, whatever n may be,
/(-)X/(n)=/(0) = l,
which proves the theorem for a negative index.
130

For the greatest terra in the expansion of (a + b)n, take

,
, part. of- (n
+ l)b or J(n-l)b
r = the integral
1j
',
a+b
ao
according as n is positive or negative.
But if b be greater than a, and n negative or fractional,
the terms increase without limit.
Examples.
Required the 40th term of ^1 J .
Here r = 39 ; o = 1 ; b=-~; n = 42.
By (127), the term will be

BINOMIAL THEOREM.

Required the 31st term of (a a:)~*.

Here r = 30; b= x; n = 4.
By (129), the term is
31. 32.33 M,(
(. n,4.5.6...30.
A)
1.2.3...30

131

55

*J..._31.32.33
~ 1.2.3 r .

fg

Required the greatest term in the expansion of ^ when x = -J-f.

-j= (1 +*)"*.
(1+*/

Here n = 6, o = 1, 6 = x in the formula

(n-l)6_oxH_ OQi .
1 trr * '

therefore r = 23, by (130), and the greatest term

_ ,
^
132

1V,5.6.7...27/14\"_
' 1.2. 3. ..23X17/

24.25.26.27 /14\
1.2.3.4 U7/ *

Find the first negative term in the expansion of (2a + 36)J,t.

We must take r the first integer which makes nr + 1 negative; there

fore r> + l= Jst+l = 6fi therefore r =7. The term will be
V-V-Y-H'iH) (2a)"*(36)7 by (126)
= _ 17.14.11.8 5.2.1 I7
7!
(2tt)*'
133

Required the ooeffioient of

in the expansion of

.
\2 3x/

.(,*+,){,+.(*)+(*)+

+(*)"+(*)"+.)%*.

}.

the three terms last written being those w hich produce z** after multiplying
by the factor (l+3a5 + faj*) j for we have

giving for the coefficient of x" in the result

?(}r+"(i)"*(fr-(f)"
The coefficient of a:" will in like manner be 9

56

ALGEBRA.

134

To write the coefficient of a?*1 in the expansion of (a1 -A

The general term is

(2n + l)!
,(. 1
(2n + l-r)!r!
V

(2n + l)!
,..+,
(2n+l-r) ! r!

Equate 4 4r+2 to 3m + l, thus

+l
r = 4n 3m
5
Substitute this value ofr in the general term; the required coefficient becomes
(2n+l)!
[i(4n + 3m + 3)] ! [i(4-3m + l)] !*
.1

i ^1 -I- 1

The value of r shows that there is no term in a*"'1'1 unless

integer.

is an

135 An approximate value of (l+a:)", when x is small, is

l-\-nx, by (125), neglecting x* and higher powers of x.
136 Ex.An approximation to \/y99 by Bin. Th. (125) is obtained from
the first two or three terms of the expansion of
(1000-1)* = 10-^ . lOOO"1 = 10-To = 9fS nearly.

MULTINOMIAL THEOREM.
The general term in the expansion of (a + bx + cxi-\-&c.)n is

137
where

n(n-l) (n-2) ... (p+l)rW<iijtWW

9! r! *! ...
jf) + j + r++&c. = n,

and the number of terms p, q, r, &c. corresponds to the

number of terms in the given multinomial.
p is integral, fractional, or negative, according as n is one
or the other.
If n be an integer, (137) may be written
138

, "! , ,ap6^rf/,...^+8r+s'.
pi ql rl si
[Deduced from the Bin. Theor.

MULTINOMIAL THEOREM.

57

Ex. 1.To write the coefficient of as6c5 in the expansion of (a + b + c + df'.

Here put n= 10, x= I, p = 3, q 1, )- = 5, s = 0 in (138).
Resul t

10!
= 7.8.9.10.
3! 5!

Ex. 2.To obtain the coefficient of a? in the expansion of

(\-2x + 3x'-4a>y.
Here, comparing with (137), we have a=l, b = 2, c= 3, d = -4,
p+q + r+ s = 4,
q + 2r + 3s = 8,

1
0
0
0

0
2
1
0

1
0
2
4

2
2
1
0

The numbers 1, 0, 1, 2 are particular values of p, q, r, s respectively,

which satisfy the two equations given above.
0, 2, 0, 2 are another set of values which also satisfy those equations ;
and the four rows of numbers constitute all the solutions. In forming these
rows always try the highest possible numbers on the right first.
Now substitute each set of values of p, q, r, s in formula (138) succes
sively, as under:
~ l1 (-2) 31 (-4)' = 576
4!
-^l" (-2)' 3 (-4)' = 384

2! 2!
|j 1 (-2)' 3s (-4)' = 864
^[l0(-2)34(-4)0= 81
4!
_
Result

1905

Ex. 3. Required the coefficient of .r* in (1 + 2x 4ss-2a:s)_4.

Here o= 1, o = 2, c = 4, d = 2,' = \ ; and the two equations arc
p+q+ r+ s = i,
q + 2r + 3s=
!

r,

-?
_7

f8

1
0
2
4

4,

0 ' 1

2
1
0

0
0
0
i

ALGEBRA.

58

Employing formula (137), the remainder of the work stands as follows :

(-i)(-f)l-*2l(-*)-(-2)l = -3
^\-\) (-|)l-*2-(-4)(-2)=

n(-i)(-f)(-i)1'*2'(-4),(-2),=

15

^ ("I) (" I) ("I) (-l)l-*2.(_4).<-2)-=

Result

f_
22 1

139 The number of terms in the expansion of the multi

nomial (a + b + c + to n terms)r is the same as the number of
homogeneous products of n things of r dimensions.
and (98).

See (97)

The greatest coefficient in the expansion of (a + b + c +

to m terms)", n being an integer, is
140

(g!r(g+i)w' where qm + k = n'

Proof.By making the denominator in (138) as small as possible. The

notation is explained in (96).

LOGARITHMS.
142

loga AT = x signifies that a" = N, or

Dkf. The logarithm of a number is the power to which the

base must be raised to produce that number.
143
144

loga = l,

log 1 = 0.

log MN = log M+ log iV.

log^ = log Jf log A7,
log (M)H = n log M.
log \/M = 1. log M.

[142

EXPONENTIAL THEOREM.

145

59

ioR = |2&.
log, b

That is The logarithm of a number to any base is equal to

the logarithm of the number divided by the logarithm of the
base, the two last named logarithms being taken to any the
same base at pleasure.
Proof.Let \ogca = x and \ogrb = y; then a = cx, b = c". Eliminate c.
c = ax = b";

r. a=b",

that is, log,,a =

y'

146

log4 a = ,r
loga b

147

log10tf =

148

Q. e. d.

Put c = a in (145).

by(145).

L- = -43429448 . . .
log, 10

is called the modulus of the common system of logarithms ;

that is, the factor which will convert logarithms of numbers
calculated to the base e into the corresponding logarithms to
the base 10. See (154).

EXPONENTIAL THEOREM.

149
where

a* = 1 + ex + ^f! + ^ + &c,
c = (a-\)-\(a-\y+\ (a-l)s-&c.

Proof. ax = [l + (o 1)}*. Expand this by Binomial Theorem, and

collect the coefficients of * ; thus c is obtained. Assume cs, cs, &c, as the
coefficients of the succeeding powers of x, and with this assumption write
out the expansions of ax, a", and a**". Form the product of the first two
series, which product must be equivalent to the third. Therefore equate
the coefficient of a: in this product with that in the expansion of axt". In
the identity so obtained, equate the coefficients of the successive powers of
y to determine c c &c.

ALGEBRA.

60

Let e be that value of a which makes c = 1 , then

150

e-=l + .r+|^+^ + &c.

151

e = l + l + l + l+&c.
= 2718281828...

[See (295).

Proof.By making x = 1 in (150).

152

By making x = 1 in (149) and ,r = c in (150), we obtain

a = ec ; that is, c = log a.

154

Therefore by (149)

log, = (o-l)-i(o-l)*+i(a-l)-&c.

155

log (1 +.*) =

>
+ .r1
3

156

*s
a?4 -&c.
log (1 -.*) = -.r- - - ^- - ^
2
4

157

logl=2),+ +|L+*c.j
+ &C.

[154

Put ?- - for ,e in (157) ; thus,

?n + 1
158

i*- - J^5 +i(^t)*+ieii) +M

Put
159

for x in (157); thus,

!og(n+l)-logn
= 2S-J
1
+
U_ + Ac.iS
? 2/i + l +
T3(2w+l)sT5(2+l)'^

CONTINUED FRACTIONS.

61

CONTINUED FRACTIONS AND CONVBRGENTS.

To find convergents to 3' 141 59 = 314159
100000'
in the rule for H. C. F.
160

100000
99113
887
854
33
29
4
4

314159
300000
14159
887
5289
4435
854
66
194
165
29
28

15

Proceed as

The continued fraction is

3+1
7+1
15 + &c.
or, as it is more conve
niently written,

25
3+*
* &c.
7+ 15 +

The convergents are formed as follows :

3
7
15
1
25
1
7
4
3. 22 333 355 9208 9563 76149 314159
1' 7' 106' 113' 2931' 3044' 24239' 100000'
161 Rule.Write the quotients in a row, and the first two
convergents at sight (in the example 3 and 3 + y). Multiply
the numerator of any convergent by the next quotient, and
add the previous numerator. The result is the numerator of
the next convergent. Proceed in the same way to determine
the denominator. The last convergent should be the original
fraction in its lowest terms.

162

Formula for forming the convergents.

If &=it Pzi) P- are aI1y consecutive convergents, and

q-s q-i 2
(*n-2> o-i flM the corresponding quotients; then
P = "nPn-l+Pn-2, tf = ?-! + ?.-!

62

ALGEBRA.

The ntt> convergent is therefore

The true value of the continued fraction will be expressed

by
163
f = aP-iJrP-*
in which a'H is the complete quotient or value of the continued
fraction commencing with a.
164

Pq-ipn-xqn = 1 alternately, by (162).

The convergents are alternately greater and less than the

original fraction, and are always in their lowest terms.
165 The difference between Fn and the true value of the
continued fraction is
<

and >
7</+i
9 (<7+7+i)
and this difference therefore diminishes as n increases.
Proof.By taking the difference, Pn - a'P+l+Pa.
q
a.qn,i+qH

(163)

Also F is nearer the true value than any other fraction

with a less denominator.
166 FnF+1 is greater or less than J72 according as Fn is
greater or less than Fn+l.
General Theory of Continued Fractions.
167

First class of continued

fraction.

F=Pl - I?*

*8

i + Off a3 + &c.

Second class of continued

fraction .

y_h

bs-

Oj a.2 <zs &c.

au bu &c. are taken as positive quantities.

CONTINUED FRACTIONS.

63

, - , &c. are termed components of the continued fraction. If the components be infinite in number, the continued
fraction is said to be infinite.
Let the successive convergents be denoted by
&_ = k;
j,
a,
168

Pi = k^k.;
g2
ax + a,

P*=h_lh_h. andsoon.
y3
!+ a2 + a8

The law of formation of the convergents is

For F,

For V,

[Proved by Induction.
The relation between the successive differences of the
convergents is, by (168),
Jgg

ffn+l

Pn _ -j- b + iqn-i (pn

Pn-\ \

Take the sign for F, and the + for V.

170

171

Pnq,-i-p.-iq. = (-l)'_lMA 6,.

(168)

The odd convergents for F,

&c, continually

decrease, and the even convergents,

, &c, continually

increase.
(167)
Every odd convergent is greater, and every even con
vergent is less, than all following convergents.
(169)
172 Def.-If the difference between consecutive convei*gents diminishes without limit, the infinite continued fraction
is said to be definite. If the same difference tends to a fixed
value greater than zero, the infinite continued fraction is in
definite ; the odd convergent?; tending to one value, and the
even convergents to another.

64

ALGEBRA.

173 F is definite if the ratio of every quotient to the next

component is greater than a fixed quantity.
Proof. Apply (169) successively.

174 F is incommensurable when the components are all

proper fractions and infinite in number.
Proof. Indirectly, and by (168).

175 If a> be never less than ft-p-l, the convergents of V are

all positive proper fractions, increasing in magnitude, p and
qn also increasing with n.
By (167) and (108).
176 If, in this case, V be infinite, it is also definite, being
= 1, if a always =b + l while b is less than 1, (175); and
being less than 1, if a is ever greater than 6 + 1.
By (ISO).
177 V is incommensurable when it is less than 1 , and the
components are all proper fractions and infinite in number.
180 If in the continued fraction V (167), we have an == b -f 1
always; then, by (168),
Pn = &i + bt b2 -+- bi bt b3 + . . . to n terms, and qn = pn + 1 .
181 If, in the continued fraction F, a and bn are constant
and equal, say, to a and b respectively ; then pn and qn are
respectively equal to the coefficients of ain_1 in the expansions
ofc

b
---rj
1ax bir

andj

a+
r bx 1 ax b:r

Proof.pn and qM are the nth terms of two recurring series.

and (251).

182

See (168)

To convert a Series into a Continued Fraction.

The series

1 + + + ... +l

is equal to a continued fraction V (167), witli n + 1 com

ponents; the first, second, and -|-lth components being
1

ir.r

Mn-i^'
[Proved by Induction.

CONTINUED FRACTIONS.

183

65

The series

vvy

wxv2

wlvt...vn

is equal to a continued fraction V (167), with n + 1 compo

nents, the first, second, and ?? + lth components being
[Proved by Induction.

184 The sign of x may be changed in either of the state

ments in (182) or (183).
185 Also, if any of these series are convergent and infinite,
the continued fractions become infinite.

186

To find the value of a continued fraction with

recurring quotients.

Let the continued fraction be

x =

where y = ^" + 1
i+. .. + +#
*
o+i+ ... +n+.y
so that there are m recurring quotients. Form the nth con
vergent for x, and the mth for y. Then, by substituting the
complete quotients an + y for an, and an+m -\-y for a+m in (168),
two equations are obtained of the forms
v_AyB

Ey + F

from which, by eliminating y, a quadratic equation for de

termining x is obtained.

187

If

^-r
. ^-r
,+ ... ++

be a continued fraction, and

El
fx

Ejl
?
K

66

ALGEliHA.

the corresponding first n convcrgonts ; then

developed
?

by (168), produces the continued fraction

b.-i
bs
b
,-2+ ... + ao + a,

a* +

the quotients being the same but in reversed order.

INDETERMINATE

188

Given

EQUATIONS.

a.r-\-bt/ = c

free from fractions, and a, j3 integral values of x and y wnich

satisfy the equation, the complete integral solution is given by
X = a bt
y = P+at
where t is any integer.
Examplk.Given

f>x + % = 112.

Then x = 20, y = 4 are values ;

x = 20- St \
y = 4 + 5* \ '
The values of x and y may be exhibited as under:
t = -2 -1
0
1
2 3 4 5
6
7
i= 26 23 20 17 14 11
8 5
2 -1
y = _<3 -1
4 9 14 19 24 29 34 39
For solutions in positive integers / must lie between "s" - 6* and J- ;
that is, t must be 0, 1, 2, -i, 4, 5, or 6, giving 7 positive integral solutions.
189

If the equation be
.r hi/ = c

the solutions are given by

r = a + bt
y = 0+at.

INDETERMINATE EQUATIONS.

67

Example :
4x Sy = 19.
Here x = 10, y = 7 satisfy the equation ;
a; =10 + 3/}
furnish all the solutions.
y = 7 + 4/)
The simultaneous values of /, x, and y will be as follows :
t = -5
-4-3-2-1
0
1
2
3
as = -5
-2
1
4
7 10 13 16 19
y =-13 -9 -5-1
3 7 11 15 19
The nnmber of positive integral solutions is infinite, and the least positive
integral values of x and y are given by the limiting value of /, viz.,
t>

and t> ;
3
4
that is, t must be 1, 0, 1, 2, 3, or greater.

190 If two values, a and (3, cannot readily be found by

inspection, as, for example, in the equation
17x+lBy = 14900,
divide by the least coefficient, and equate the remaining frac
tions to t, an integer ; thus
y+*+3=1146+-?4a; 2 = 13*.
Repeat tho process ; thus

Put
.-.

and

.-.
M=
t=
s _
=

< + 2 = 4w.
1,
2,
13/ + 2 __
^=7
= a;.

y+x + t = 1146, by (1),

y = 1146- 7 - 2 = 1137 = fi.
The general solution will be
x =
7 13*,
y - 1137 + 17*,
Or, changing the sign of / for convenience,
x=
7 + 13/,
y = 1137-17/.

(1);

ALGEBRA.

68

Here the number of solutions in positive integers is equal to the number of

integers IV*
lying between

7 andA 1137
^
^ ;

or

7 and 66| ; that is, 67.

1o

191 Otherwise. Two values of x and y may be found in

the following manner :
17
Find the nearest converging fraction to .
4
This is .

[By (160).

By (164) we have

17x3-13x4 = -l.
Multiply by 14900, and change the signs ;
17 (-44700) + 13 (59600) = 14900 ;
which shews that we may take
J

S a
1/3 =

44700
59600

and the general solution may be written

x = -44700 + 13<,
y = 59600-17*.
This method has the disadvantage of producing high values of a and /?.
192 The values of x and y, in positive integers, which
satisfy the equation ax+by = c, form two Arithmetic Pro
gressions, of which b and a are respectively the common
differences. See examples (188) and (189).
193

Abbreviation of the method in (169).

Example :
11* I8y = 63.
Put x 9z, and divide by 9 ; then proceed as before.

194

To obtain integral solutions of a<v-\-by-\-cz = d.

Write the equation thus

a-x + by = dcz.
Put successive integers for z, and solve for x, y in each case.

REDUCTION OF A QtJABEATIO SURD.

TO

REDUCE

QUADRATIC

SURD

69

TO

CONTINUED FRACTION.
195

Example :
4
v/29 =

5 + v/29--5 =

v/29 + 5 __
4

-3 _
2 + v/29" . 4

2 +

v/29 + 3
5

-2 __
1 + v/295

1 +

v/29 + 2
5

-3
1 + v/295

1 +

/29 + 3 __
4

-5
2 + v/294

2 +

5 +
v/29 + 5'
5
v/29+3'
5
v/29+2'
4
v/29 + 3'
1
v/29 + 5'
4

v/29 + 5 = 10 + v/29--5 = 10 +
v/29 + 5*
The quotients 5, 2, 1, 1, 2, 10 are the greatest integers
contained in the quantities in the first column. The quotients
now recur, and the surd v/29 is equivalent to the continued
fraction
1111
1
1_ 1_ 1_ 1
5+ 2+ 1+ 1+ 2+ 10+ 2+ 1+ 1+ 2 + &c.
The convergents to v/29, formed as in (160), will be
5
I*

11
2'

16
"8"'

27
5 '

70
13'

727
135'

1524
283'

2251
418'

3775 9801
701' 1820"

196 Note that the last quotient 10 is the greatest and twice
the first, that the second is the first of the recurring ones, and
that the recurring quotients, excluding the last, consist of
pairs of equal terms, quotients equi-distant from the first and
last being equal. These properties are universal. (See 204
-210).

To form high convergents rapidly.

197

Suppose m the number of recurring quotients, or any

ALGEBRA.

70

multiple of that number, and let the mth convergent to \/Q be

represented by Fm ; then the 2w,h convergent is given by the
formula

F2m = I [f +

] by (203) and (210).

198 For example, in approximating to v/29 above, there are five recurring
quotients. Take m = 2 x 5 = 10 ; therefore, by
Fn=\{FM+fJ,
Fi0 =
m,
f
Therefore

tne 10th convergent.

p = }(9801
,0G
1820 )} = 192119201
Fx
_+29x
~_

the 20th convergent to \/29 ; and the labour of calculating the intervening
convergent* is saved.

GENERAL

199

THEORY.

The process of (174) may be exhibited as follows :

SQ+<-i - a ,

^Q + Cn _

200

Then

l
l
Q = fl,+ .;+

r.+1

l
,-)- AC.

The quotients au Oj, a3, &c. are the integral parts of the frac
tions on the left.

REDUCTION OF A QUADRATIC SURD.

201

71

The equations connecting the remaining quantities are

Cl = 0

r, = l
Q-c\

_ Q-cl
rs =

cs = a.2r.2c2

_ Q Cn
I

-1

^*

The ?ith convergent to \/Q will be

202

= a"P"1+P"-2.

[By Induction.

The true value of \/Q is what this becomes when we

substitute for an the complete quotient ^^~'--Cbj of which <z
is only the integral part.
203

This gives

v/g = (/Q+Cn)Pn-i+rPn->

By the relations (199) to (203) the following theorems are

demonstrated :
204

All the quantities a, r, and c are positive integers.

205

The greatest c is c2, and c.2 = ax.

206

No a or r can be greater than 2av

207

If ru = 1 , then c = a,.

208

For all values of n greater than 1, a c is <r.

209 The number of quotients cannot be greater than 2a*.

The last quotient is 2av and after that the terms repeat.
The first complete quotient that is repeated is ^-^^~ Cs!, and
<ht r cs commence each cycle of repeated terms.

72

ALGEBRA.

210

Let am, rm, rm be the last terras of the first cycle ; then
i'm-i,
are respectively equal to a2, r2, c2; am_t, rm_t,
cm-i are equal to a3, rs, cs, and so on.
[By (187).

EQUATIONS.

Special Cases in the Solution of Simultaneous Equations.

211

First, with two unknown quantities.

ap+bjy =
(ux + bti/ = ca)

x _ cfiic^x
ajtta^bi

_ Cid, caa,
b^b^

If the denominators vanish, we have

=
a,
o2

and a; = oo, y = oo ;

unless at the same time the numerators vanish, for then

at

b2

Cg '

0 '

0 '

and the equations are not independent, one being produced by

multiplying the other by some constant.
212 Next, with three unknown quantities. See (60) for
the equations.
If dlt d2, d3 all vanish, divide each equation by z, and we

T and il , two
have three equations for finding the two ratios

z
z
only of which equations are necessary, any one being deducible from the other two if the three be consistent.
213

To solve simultaneous equations by Indeterminate

Multipliers.

Ex.Take the equations

x + 2y + 3z + 4w = 27,
:ix + by +

1z +

v = 48,

bx + 8y + 102 2w = 05,
7x + 6y+ 5z + iw = 53.

MISCELLANEOUS EQUATIONS.

73

Multiply the first by A, the second by 5, the third by 0,

leaving one equation unmultiplied ; and then add the results.
Thus

(4 + 35 + 50+7) as+(24 + 55+80+6)y

+ (34 + 75+100+5) z+(44+5-20+4)w;
= 274+485+650+53.

To determine either of the unknowns, for instance x,

equate the coefficients of the other three separately to zero,
and from the three equations find A, B, 0. Then

= 274 +485 + 650 + 53

4+35+50+7
'

MISCELLANEOUS EQUATIONS AND SOLUTIONS.

214
*61 = 0.
Divide by Xs, and throw into factors, by (2) or (3).

See also

(480).
215

^-7^-6 = 0.

x = 1 is a root, by inspection; therefore x + 1 is a factor.

Divide by * + l, and solve the resulting quadratic.
216

a?+Wx = 455.
aJ*+16V = 455* = 65 X 7x,
a> + Gb* + ( 6|-J= 49*8+65X 7x+ ^ J,
s , 65
w , 65
x* + T = 7z+,
a? = 7x\

.".

x = 7.

Rule. Divide the absolute term (here 455) into two

factors, if possible, such that one of them, minus the square
of the other, equals the coefficient of x. See (483) for general
solution of a cubic equation.
L

74

ALGEBRA.

217

x'-y* = 14560,

Put

x z-\-v

and

x-y = 8.
y = z v.

Eliminate v, and obtain a cubic in z, which solve as in (216).

218

.r5y = 3093,

x-y = 3.

Divide the first equation by the second, and subtract from

the result the fourth power of xy. Eliminate (af+y1), and
obtain a quadratic in xy.

219

On forming Symmetrical Expressions.

Take, for example, the equation

(y c){z b) = a\
To form the remaining equations symmetrical with this, write
the corresponding letters in vertical columns, observing the
circular order in which a is followed by b, b by c, and c by a.
So with x, y, and z. Thus the equations become
(yc) (zb) = a*,
(za) (xc) = b1,
(x-b)(y-a) = c*.
To solve these equations, substitute
x = b + c + x',

y = c + a + y',

z = a + b + z';

and, multiplying out, and eliminating y and z, we obtain

x = hc (b + c)-a(bt+ct)t
be co ab
and therefore, by symmetry, the values of y and z, by the
rule just given.
220

.-.

yi + z*+yz = a>

(1),

zi + x*+zx = bi

(2),

x*+y> + xy=c>

(3);

3 (yz + --.r+ryY = 2b1ct+2cia*+2a,b*-a*b*-c*

(4).

IMAGINARY EXPRESSIONS.

Now add (1), (2), and (3), and we obtain

2(x+y+z)*-3 (yz+zx+xy) = a* + b* + c2

(5).

From (4) and (5), (x+y + z) is obtained, and then (1), (2),
and (3) are readily solved.
221

o*-yz = at

(I),

y*-zx = b*

(2),

z*-xy=<?

(3).

Multiply (2) by (3), and subtract the square of (1).

Result

x (Sxyz a?8 i/3 r*) = 62cs a*,

ftV-a1

c*a?-b*

aW-c*

K)'

Obtain A* by proportion as a fraction with numerator

= a?yz = a*.
222

x = cy + bz

(1),

y = az + coc

(2),

z = bx+ ay

(3).

Eliminate a between (2) and (3), and substitute the value of

x from equation (1).
Result

T^Ti
-i a2a16* 11 caa = 1

IMAGINARY EXPRESSIONS.

223

The following are conventions :

That
a*) is equivalent to ay/( \); that a-\/{ 1)
vanishes when a vanishes; that the symbol a^/{ 1) is sub
ject to the ordinary rules of Algebra. \/( 1) is denoted
by *.

70

ALGEBRA.

224

If a + i"/3 = y -f- i"8 ; then n = -y aud /3 = 8.

225 a + </3and a 1'/3 are conjugate expressions ; their pro

duct = <r+/3J.
226 The sum and product of two conjugate expressions are
both real, but their difference is imaginary.
227

The modulus is +

228

If the modulus vanishes, a and f3 must vanish.

229 If two imaginary expressions are equal, their moduli

are equal, by (224).
230 The modulus of the product of two imaginary expres
sions is equal to the product of their moduli.
231 Also the modulus of the quotient is equal to the
quotient of their moduli.

METHOD OF INDETERMINATE COEFFICIENTS.

232 If A+Bx + Cr+...= A'+B'x+Cx!+... be an equa

tion which holds for all values of .r, the coefficients A, B, &c.
not iuvolving r, then A = ./', B= ft, C = C, &c. ; that is,
the coefficients of like powers of x must be equal. Proved by
putting t = 0, and dividing by x alternately. See (234) for
an example.

233

METBOD OF PROOF BY INDUCTION.

Ex.To prove that

l+S+*+...+rf = ( + !) (2+D.
6
Assume

l+2+3>+ ... +' =

PARTIAL FRACTIONS.

...

77

l + 2'+3'+-+n'-Kn + l)' = w(n"fl)/2,' + 1)+(-fl)'

_ n(n + l) (2nl) + 6 (n + 1)' _ (n + 1) {ra (2n + l) +6 (n + 1)}
6
6
_ (n+1) ( + 2) (2n + 3) _ n'(n'+l) (2n' + l)
6
~
6

where n is written for n + 1 ;

...
1 +2' + 3'+. ..+'* = *'(.*' + 11&^+J1.
6
It is thus proved that if the formula be true for n it is also true for n + 1. .
Bat tbe formula is true when n = 2 or 3, as may be shewn by actual
trial ; therefore it is true when n = 4 ; therefore also when n = 5, and so on ;
therefore universally true.
234 Ex.The same theorem proved by the method of In
determinate coefficients.
Assume
l+2, + 3,+ ...+'
= A + Bn
+ <V
+Dn*
+&c.;
.-. ! + & + $'+. ..+nt + (n + l)1 = A + B(n + l) + 0(n + iy + D(n+])s+&c.;
therefore, by subtraction,
' + 2n + l = + C(2n+l) + D(3n' + 3+l),
writing no terms in this equation which contain higher powers of n than the
highest which occurs on the left-hand side, for the coefficients of such terms
may be shewn to be separately equal to zero.
Now equate the coefficients of like powers of n ; thus
.-. D

]_
3 5

2C+3D = 2,

.-.

*,

B+C+D = 1,

.-.

1
6 '

3D = 1,

and .4 = 0;

therefore the sum of the series is equal to

n_ r?_ n* _ n(n + l) (2n+l)
6
2
3
"6

PARTIAL FRACTIONS.
In the resolution of a fraction into partial fractions four
cases present themselves, which are illustrated in the follow
ing examples.

ALGEBRA.

78

235 First. "When there are no repeated factors in the de

nominator of the given fraction.
Ex.To resolve

SSnme

-;
^rz
^ into partial fractions.
(xl)(x2)(x3)

(*-l)(*-~2) (*-3) ~*=I+i^2+*-3;

Sx-2 = A(x-2) (a-3) +B (*-3) (x-l) + C (<r-l) (*-2).

Since J, B, and (7 do not contain x, and this equation is true for all values
of x, put x = 1 ; then
3-2 = 4(1-2) (1-3), from which A = i
Similarly, if * be put = 2, we have
6-2 = B (2-3) (2-1);
and, putting a; = 3,
9-2 = 0(3-1) (3-2);

Hence

236

3x-2
1

=
(*-l) (as-2) (x-3)
2(3-1)

.-.

B = -4;

.\

0 = ^.

4 (7
.
t-2 2(i-3)

Secondly.When there is a repeated factor.

7^ \0a?+6v
Ex.Resolve into partial fractions (* l)s(ai+2)
. .
.
Assume

7x'-10x*
A
^ ; B; H^ G 7 H1 D
+ 6x =
+
(* 1)V + 2)
(*-l)
(*-l)
*-l " + 2

These forms are necessary and sufficient. Multiplying up, we have

7x*- 10x*+6x = A (* + 2) +B (x- 1) (x+2) + G (x- 1)' (at + 2) +D (*- 1)'
(1).
Makea: = l;
.\ 7-10 + 6=4(1 + 2);
.-. A = 1.
Substitute this value of A in (1) ; thus
7x'-10x1 + 5x-2 = B (x-l)(x + 2) + G (x-lY(x + 2)+I) (x-iy.
Divide by x 1 ; thus
7x1-3x+2 = B(x + 2) + C(x-l)(x + 2)+D(x-iy
(2).
Make x = 1 again, 7-3 + 2 = 5(1 + 2);
.-. 5 = 2.
Substitute this value of B in (2), and we have
7z*-5z-2 = G (x-1) (*+2)+D (a-l).
Divide by x-1,
7x+2 = C(*+2) + I> (z-1)
Put * = 1 a third time, 7 + 2 =C (1+2); /. G = 3.

(3).

PARTIAL FRACTIONS.

79

Lastly, make x = 2 in (3),

-14+2 = 2) (-2-1);
Result

.-. D =4.

- + ? + -A_ +
(z-l/^Or-l)' *-l 3 + 2

237 Thirdly.When there is a quadratic factor of imaginaryroots not repeated.

Ex. Resolve -z., . A

rr into partial fractions.

Here we mast assume

Ax + B + Cx+D
(x'+l)(x' + x + l)
*+l
fl^+as + l '
Xs +1 and ai' + as + l have no real factors, and are therefore retained as
denominators. The requisite form of the numerators is seen by adding
together two simple fractions, such as 1 .
x+b x+d
Multiplying up, we have the equation
1 = (Ax+B) (x'+x+D + iCx+D) (x* + l)
(1).
Let
ss+l = 0;
.-. a? = -1.
Substitute this value of a? in (1) repeatedly ; thus
1 = (Ax + B) x = Ax'+Bx = -A + Bx ;
or
BxA 1 = 0.
Equate coefficients to zero; .". B = 0,
A = -1.
Again, let
x'+x+l = 0;
.-. xt=-x-l.
Substitute this value of a,-* repeatedly in (1) ; thus
1 = (Cx + D) (-*) = -Cx*-Bx = Ox+0-Dx;
or
(C-D)x + C-l =0.
Equate coefficients to zero ; thus 0 = 1,
D=l.
1
_ x+l
x
Hence
(x,+l)(x1 + x + \) x'+x + l rf + l'

238 Fourthly.When there is a repeated quadratic factor

of imaginary roots.
Ex. Resolve 7r^t?f

. o\* into partial fractions.

ALGEBRA.

80

Assume
40* -103
_ Ax + B
(x + iy (-4t! + 8)5 (a!'_4J; + 8)s

Cx+D
Ex+F
(a;!-4a! + 8), +a;-4r + 8

4O5-103 = {(Jj!+B)+(Cx+D)(a!,-4a! + 8) + (Ex + FKa!s-4j! + 8)3} (* + l)J

+ {G + fl-(a; + l)} (x>-4x + 8)s

(1).

In the first place, to determine A and B, equate x* 4a: + 8 to zero ; thus

3*= 43-8.
Substitute this value of x% repeatedly in (1) , as in the previous example,
until the first power of x alone remains. The resulting equation is
40x-103 = (17A + 6B)x-4SA-7B.
Equating coefficients, we obtain two equations
17A +
= 103}'
40 |
484
+ 6B
75 =

,
,.,4
fromwhlch
B=
= 2l.

Next, to determine G and D, substitute these values of A and B in (1) ;

the equation will then be divisible by x1 4x + 8. Divide, and the resulting
equation is
0 = 2x + 13+ {Cx+D+(Ex + F)(x*-4x+8)} (x + 1)1
+ {G + H(x + 1)} (C'_4J! + 8)1

(2).

Equate x2 4# + 8 again to zero, and proceed exactly as before, when

finding A and B.
Next, to determine E and F, substitute the values of G and D, last found
in equation (2) ; divide, and proceed as before.
Lastly, G and H are determined by equating x + 1 to zero successively,
as in Example 2.

CONVERGENCY AND DIVERGENCY OF SERIES.

239 Let o1 + ai + a34-&c. be a series, and a an+l any two

consecutive terms. The following tests of convergency may
be applied. The series will converge, if, after any fixed term
(i.) The terms decrease and are alternately positive and
negative.
(ii.) Or if
greater than unity.

is always greater than

some quantity

SERIES.

81

(iii.) Or if is never less than the corresponding ratio

in a known converging series.
(iv.) Or if ( ^2 n\ is always greater than some quantity greater than unity.

[By 244 and iii.

(v.) Or if lH^bt ti 1 1 log n

is

always

greater than

some quantity greater than unity.

240 The conditions of divergency are obviously the converse
of rules (i.) to (v.).

241

The

series a1 + a2x + asa? + &c. converges, if ^2i is

always less than some quantity p, and x less than .

[By 239 (ii.)
242

To make the sum of the last series less than an assigned

quantity p, make x less than

~^,> ^ being the greatest co

efficient.

General Theorem.
243 If <f> (x) be positive for all positive integral values of x,
and continually diminish as x increases, and if m be any posi
tive integer, then the two series
*(l)+*(2)+*(8)+*(4) +
<f> {l)+nuf> (m)+mV (m2)+m3< (ms)+
are either both convergent or divergent.
244

Application of this theorem.

1

To ascertain whether the

1

is divergent or convergent when p is greater than unity.

M

82

ALOEBEA.

Taking m = 2, the second series in (243) becomes

1,2,4,8,,
1+V + V + &+&C
'
a geometrical progression which converges ; therefore the
given series converges.
245

The series of which is the general term is

n (log n)p
convergent if p be greater than unity, and divergent if p be
not greater than unity.
[By (243), (244).
246

The series of which the general term is

1
n\ () X2 ()
\r (n) { Xr+1 ()}"*

where X (?t) signifies logn, X2() signifies log {log (to)}, and
so on, is convergent if p be greater than unity, and divergent
if p be not greater than unity.
[By Induction, and by (243).
247

The series al + a2 + &c. is convergent if

nan log (n) log2 (n)
logr () {logr+1 (n) }"

is always finite for a value of p greater than unity ; log2 (n)

here signifying log (log n), and so on.
[See Todhunter'8 Algebra, or Boole's Finite Differences.

EXPANSION OF A FRACTION.

248

A fractional expression such as -

' - mav
1 6,c + ll2 6ar
be expanded in ascending powers of x in three different ways.
First, by dividing the numerator by the denominator in
the ordinary way, or by Synthetic Division, as shewn in (28).
Secondly, by the method of Indeterminate Coefficients
(232).
Thirdly, by Partial Fractions and the Binomial Theorem.

SERIES.

To expand by the method of Indeterminate Coefficients,

proceed as follows :
Assume , p ~}f 0 , = A + Bx + Cx2 + Dx* + Ex* + Ac.
1 oj: + lla ox*
4xl0x2 = A + Bx + Cx2 + Bx* + Ex* + Fz*+...
-6Ax- 6Bx>- 6Cx*- 6Bx*- SEx*-...
+ UAz! + nBxs + nCxi + lWxr'+...
- 6Ax*- 6Bxl- 6Cx*-...
Equate coefficients of like powers of x, thus
A =
0,
B- 6A =
4,
G- 65 + 11^1 = -10,
D-60+UB- 6A =
o,
E-6D+UG- 6B =
o,
F-6E + UB- 60 =
o,

. B = 4;
: G = 14;
: D = 40;
: E = 110;
\ F = 304;

The formation of the same coefficients by synthetic division is now

exhibited, in order that the connexion between the two processes may be
clearly seen.
The division of 4ai 10a:2 by 1 Gx+ Ua:'J 6.c5 is as follows:
+ 6
-11
+ 6

0+4-10
24+84 + 240 + 660
-44-154 - 440-1210
+ 24 + 84 + 240 + 660
0 + 4 + 14 + 40 + 110 + 304+
A B C B E
F

If we stop at the term 110x4, then the undivided remainder will be

;i04j^ 97Ou." + 6G0j;7, and the complete result will be
4, + 14z' + 40.1;' + 110,'+ 30-970^+66O^
I bx+ llx* bx"
249 Here the concluding fraction may be regarded as the
sum to infinity after four terms of the series, just as the
original expression is considered to be the sum to infinity of
the whole series.
250 If the general term be required, the method of ex
pansion by partial fractions must be adopted. See (257),
where the general term of the foregoing series is obtained.

ALGEBRA.

84

RECURRING SERIES.

a0-\-al.v-\-a2a:i-\-a^-^-&c. is a recurring series if the co

efficients are connected by the relation
251

= Pi - 1 + Pi 0 - 2 + + Pm - m-

The Scale of Relation is

252
1 pixpttf pmxm-

The sum of n terms of the series is equal to

253

[The first m terms

pxx (first m 1 terms + the last term)
p2<x? (first m 2 terms + the last 2 terms)
psj? (first m 3 terms + the last 3 terms)
Pm-i^ 1 (first term + the last m 1 terms)
pmxm (the last m terms)]
[1p^rp^ .../>J?"]-

254 If the series converges, and the sum to infinity is re

quired, omit all " the last terms " from the formula.
255 Example. Required the Scale of Relation, the general
term, and the apparent sum to infinity, of the series
4e + 14./ + 40.e5 + 110.b1 + 304c5-854c6 + ... .
Observe that six arbitrary terms given are sufficient to determine a Scale
of Relation of the form 1 px qx' rx*, involving three constants p, q, r,
for, by (251), we can write three equations to determine these constants ;
namely,
110= 40p + 14g+ 4r"|
The solntion gives
304= U0p+ 403 + 14r k
p = 6, q = -U, r = 6.
854 = 304p+l 10g+40r J
Hence the Scale of Relation is 1 6x + 11^ 6x\
The sum of the series without limit will be found from (254), by putting
P\ = 6, Pi = 11, Pi 6, to = 3.
The first three terms
= 4r + 14ar+40j^
6 xthe first two terms = 24*1 84a*
+ 11a x the first term
=
+ 44as'
4-c-lOar

RECURRING SERIES.

a-

85

4s- IPs'

l-6a; + ila!2-6a!';
the meaning of which is that, if this fraction be expanded in ascending
powers of x, the first six terms will be those given in the question.
256 To obtain more terms of the series, we may use the Scale of Relation ;
thus the 7th term will be
(6 X854-11 x 304 + 6 x 110) a? = 2440a;7.

257 To find the general term, S must be decomposed into

partial fractions ; thus, by the method of (235),
43-10^
__ 1
l-6o; + lla!J-6J l-3a;

2
12s

3
1-a:'

By the Binomial Theorem (128),

i\-=
1 das ,
%~ =
1 2x

1+& + 8V +

+3V,

2 + 2'a!+2V+

+2"+1a;n,

-r-5- --Z-Zx-Ztflx
Hence the general term involving x" is
(3" + 2"*l-3)a;\

-3x*.

And by this formula we caii write the " last terms " required in (253), and
so obtain the sum of any finite number of terms of the given series. Also,
by the same formula we can calculate the successive terms at the beginning
of the series. In the present case this mode will be more expeditious than
that of employing the Scale of Relation.

258 Ifj in decomposing S into partial fractions for the sake

of obtaining the general term, a quadratic factor with ima
ginary roots should occur as a denominator, the same method
must be pursued, for the imaginary quantities will disappear
in the final result. In this case, however, it is more con
venient to employ a general formula. Suppose the fraction
which gives rise to the imaginary roots to be
L + Mx _
L + Mx
a + bx+a?
(p x)(q x)'
p and q being the imaginary roots of a + bx+x2 = 0.
Suppose

p = a + ifi,
q = ai(5, where i=y/ 1.

86

ALGEBRA.

If, now, the above fraction be resolved into two partial

fractions in the ordinary way, and if these fractions be ex
panded separately by the Binomial Theorem, and that part of
the general term furnished by these two expansions written
out, still retaining p and q, and if the imaginary values of p
and q be then substituted, it will be found that the factor will
disappear, and that the result may be enunciated as follows.
259

The coefficient of a:*"1 in the expansion of

L+Mx
(o8+/82),1-2o.r+^

ViS*

will be
L

+ PK

{naniP- C (n, 3) a-3F+ C (n, 5) os/85- . .. }

M
'TP)

+ C(n-l,5)a"-6j85-...}

260 With the aid of the known expansion of sin nO in

Trigonometry, this formula for the nth term may be reduced to

in which

6 = tan-1,
a

d> = tan"1 r . \\r .

L + Ma

If n be not greater than 100, sin (n0 <p) may be obtained

from the tables correct to about six places of decimals, and
accordingly the nth term of the expansion may be found with
corresponding accuracy. As an example, the 100th term in
the expansion of -1 -\-x 2 is readily found by this method
. ,

41824 <

t0 be -1Q41- S^-

To determine whether a given Series is recurring or not.

261 If certain first terms only of the series be given, a scale
of relation may be found which shall produce a recurring

BECUBBING 8EBIES.

87

series whose first terms are those given. The method is

exemplified in (255). The number of unknown coefficients
p, q, r, &c. to be assumed for the scale of relation must be
equal to half the number of the given terms of the series, if
that number be even. If the number of given terms be odd,
it may be made even by prefixing zero for the first term of
the series.
262 Since this method may, however, produce zero values
for one or more of the last coefficients in the scale of relation,
it may be advisable in practice to determine a scale from the
first two terms of the series, and if that scale does not produce
the following terms, we may try a scale determined from the
first four terms, and so on until the true scale is arrived at.
If an indefinite number of terms of the series be given,
we may find whether it is recurring or not by a rule of
Lagrange's.
263

Let the series be

S = A + Bx + Gx2 + Dxs + &c.

Divide unity by S as far as two terms of the quotient, which

will be of the form p + qx, and write the remainder in the form
S'x*, S' being another indefinite series of the same form as S.
Next, divide S by S' as far as two terms of the quotient,
and write the remainder in the form S"a?.
Again, divide S' by S", and proceed as before, and repeat
this process until there is no remainder after one of the
divisions. The series will then be proved to be a recurring
series, and the order of the series, that is, the degree of the
scale of relation, will be the same as the number of divisions
which have been effected in the process.
Example. To determine whether the series 1, 3, 6, 10, 15, 21, 28, 36,
45, . . . is recurring or not.
Introducing x, we may write
8 = l+3a: + 6a;a + 10a!!,-r-15a!4+21a;6+283; + 36x7+45a!8... .
o

Then we shall have = 1 3x+ ... with a remainder

Therefore

Sx1 + 8x> + 15as4 + 24a5 + 35a6 + &c.

S'= 3 + 8x + 15z'+24a* + 35zi+&c.,
8
8'

1 .as
3
9

ALGEBRA.

88

with a remainder

i- (^+3x' + 6x, + l(V + &c. ...)

y
Therefore we may take S" = 1 + az + 6z, + 10x' + &c.
8" = 3x without any remainder.
Lastly "3>,a
Consequently the series is a recurring series of the third order. It is, in
fact, the expansion of

1 3x+3*a*'

SUMMATION OF SERIES BY THE METHOD OF

DIFFERENCES.

264 Rule.Form successive series of differences tintil a series

of equal differences is obtained. Let a, b, c, d, &c. be the first
terms of the several series ; then the nth term of the given
series is
265 +(.-i)^("-y<;->+(-')(-^-)rf+
The sum of n terms
noo
266

i n(n 1) . . n(n
= na + \ 2
b+

2)
.
"c + &c-

Proved by Induction.
Example:

a...l+ 5 + 15 + 35 + 70 + 126+ ...

b ... 4 + 10 + 20 + 35 + 56+.. .
c ... 6 + 10 + 15 + 21 + ...
d...4+ 5+ 6+...
e ... 1+ 1 + ...
The 100th term of the first series
-l+QQ 4

99-98 fi 99.98.97 , 99.98.97.96

1.2 + 1.2.3
+ 1.2.3.4 "

The sum of 100 terms

- 100+ 1QQ-"-i a. 100-99-98fi 100.99.98.97 ,
1.2
+ 1.2.3
+
1.2.3.4

100.99.98. 97.96
1.2:"3.4.5
*

v
FACTORIAL 8ERIES.\^t,. .

89

267 To interpolate a term between two terms of a series by

the method of differences.
Ex. Given log 71, log 72, log 73, log 74, it is required to find log 72'54.
Form the series of differences from the given logarithms, as in (266),
log 71
a. .. 1-8512583
b. .. -0060742
c ... - 0000838
d ... -0000022

log 72
log 73
1 8573325
1-8633229
-0059904
-0059088
- -0000816
considered to vanish.

log 74
1 8692317

Log 72'54 mast be regarded as an interpolated term, the number of its

place being 2 54.
Therefore put 2'bi for n in formula (265).
Result
log 72-54 = 1 8605777.

DIRECT

268

Ex.:

FACTORIAL SERIES.

5. 7. 9 + 7. 9. 11+9. 11. 13 + 11. 13. 15 + ...

d = common difference of factors,
m = number of factors in each term,
n number of terms,
a = first factor of first term d.

nth term = (a + nd) (a+M + ltf)

(a+n+m 1 d).

269 To find the sum of n terms.

Rule.From the last term with the next highest factor take
thefirst term with the next lowest factor, and divide by (??i+ 1) d.
Proof.By Induction.
Thus the sum of 4 terms of the above series will be, putting d = 2, m = 3,
= 4,a = 3,

11.13.15.17-3.5.7.9
8=
(3 + 1)2

Proved either by Induction, or by the method of Indeterminate Coefficients.

ALGEBRA.

90

INVERSE FACTORIAL SERIES.

270

Ex.: g 7 g + 7 9 n +y n .i3 + n. 13.15 + -

Defining d, m, n, a as before, the

i

nth term =

(a+nrf) (a+n+ld) ... (o + +m-ld)

271 To find the sum of n terms. Rule. From the first
term wanting its last factor take the last term wanting its first
factor, and divide by (to 1) d.
Thus the sum of 4 terms of the above series will be, putting d = 2, m = 3,
1
1
.
Q
5.7
13.15
n = 4, a = o,
(3-1)2
Proof.By Induction, or by decomposing the terms, as in the following
example.
272 Ex.: To sum the same series by decomposing the terms into partial
fractions. Take the general term in the simple form
2
(>-2)r(r + 2)'
Resolve this into the three fractions

Substitute 7, 9, 11, &<:. successively for r, and the given series has for
its equivalent the three series
8 I
5
7
J. (
2 _ 2
8 l~* 7
9
+ 8LfI
T

911
2 _ 2
11
13

1+
1 + _L
9
11
13 +

13

2 + 3)
2 _ 2 7
2n + 3 2 + 5)

,1.1,1?
2n+3 2n + 5^2 + 7)'

and the sum of n terms is seen, by inspection, to be

ifi
i
L +_i_]=l( i
i
)
8 ( 5
7
2 + 5 2n + 7)
4(5.7 (2n + 5) (2ra + 7) 5 '
a result obtained at once by the rule in (271), taking * for the first
5.7*9
term, and -^rr

for the nth or last term.

(2u + 3) (2n + 5) (2n + 7)

FACTORIAL SERIES.

91

273 Analogous series may be reduced to the types in (268)

and (270), or else the terms may be decomposed in the manner
shewn in (272).
Ex.:

_J_+_4_ + _Z_ + 10
1.2.3 2.3.4 3.4.5 4.5.6
has for its general term
3n 2 + 2)_ ~ n1 +, ^+1
5 ~ ^2
4 .bj
n (n+l)(
and we may proceed as in (272) to find the sum of n terms.
The method of (272) includes the method known as " Summation by
Subtraction," but it has the advantage of being more general and easier of
application to complex series.

COMPOSITE

274

FACTORIAL SERIES.

If the two series

be multiplied together, and the coefficient of x* in the product

be equated to the coefficient of x* in the expansion of (1 a;)"8,
we obtain as the result the sum of the composite series
5.6.7.8x1.2+4.5.6.7x2.3+3.4.5.6x3.4
+ 2. 3. 4. 5X4. 5 + 1. 2. 3. 4X5. 6 = ^j^p.

275

Generally, if the given series be

PA + P.O.+ -+P.-1Q.-1

where
and

Qr = r (r+1) (r+2) ... (r+q-l),

Pr = (nr) (n r + 1) ... (n r+p 1) ;

the sum of n 1 terms will be

7>! <7!
(n+p+q1) !
(jH-9+1)!'
(-2)!

(1),

92

ALGEBRA.

MISCELLANEOUS SERIES.

276

Sum of the powers of the terms of an Arithmetical

Progression.
1+2 +3 +...+ =^+n

=Sl

l+2^+...+u2="(/t + 1),(2" + 1)

= 8,

l+23+33+... + s= ^n(w+1)
= S3
l+2'+3* | ...+* = ( + l)(2+l)(32+3n-l)_ ^
30
[By the method of Indeterminate Coefficients (234).
A general formula for the sum of the rth powers of
1.2.3 ... n, obtained in the same way is
Sr = ^T*r+,+K+-4,,'-,+ ... Ar.in,
where Au At, &c, are determined by putting p = 1,2, 3, &c.
successively in the equation
1
2(p + Y)\
(j>+2)!

x
Ao
,
A,
r(p)rr(r-l)(p-l)r" r(r-l)...(r-p+l)

277
"+(+
( +2</)"*+ . . . + (a+nrf)"
= (n + 1) "+^HiH,"" 'rf+SX'(m) 2i a ad*
+ S3C (m, 3) a--srfs+&c.
Proof.By Binomial Theorem and (276).
278

Summation of a scries partly Arithmetical and

partly Geometrical.
Example.To find the sum of the series 1 +3* + 5*2+to n
terms.

MISCELLANEOUS SERIES.

93

Let

s = l + 3a: + 5a;! + 7*1,+ ... + (2-l)iB"-I)

sx =
z+3J + 5x*+ ... + (2-3) a;"-1 + (2n-l) x*,
.". by subtraction,
s(l-x) = l + 2x+2a?+2a>+. .. + 2z"-1-(2n-l) x"

279

= l+2z-~
1 x
_ ^(2n-l)z"
1-a

(2n-l)x,
2(l-g"-'
(I-*)'

A general formula for the sum of n terms of

a+(a+rf) r+(a+2d) r'+(a+3rf) **+&c.

is

g a~ ("+n 1 d)rn , drjl-r"-1)

1-r

(l-r)8 '
Obtained as in (278).

Rule.Multiply by the ratio and subtract the resulting

series.

280

281

1
i OS

1 X

TT^-Tz = l+2tf+3*f+4*+...
+ cl?"-1+ (n+1)^*-^n+1

282

(n-l)cJ-+(n-2)*2 + (n-3)lr3+... + 2*"-2+^-1

= (.-1) =g*gg.

283

1 +n+

By(253).

+ n(n-y,(n~2) +&c = 2-,

1 _ +
!

_ n(W-l) (n-2)+&c = Q
o!
By making -a-=.h in (125).

.eirrat^

ALGEBRA.

94

284

The series

i _ -3 , (n-4) (n-5) _ (n-5) (n-6) (n-7) j_

2 +
a!
4!
l(

1)r.1(-r-l)(n-r-2)...(-2r+l)

consists of or -- terms, and the sum is given by

S =

3
if n be of the form 6m + 3,
n

8 =

0 if be of the form 6m + l,

S = if n be of the form 6m,

n
8 =

if ft be of the form 6m + 2.
n

Proof.By (545), putting p = x + y, 9 = xy, and applying (546).

285

The series

nr-n (-l)r+ n(n~1) (n-2)r

- nin
3j

takes the values

0,

ft!,

according as r is

<ft, =ra, or = n-f-1.

2)/
vr . ,
'- (n 3)r+&c...

^ft(?i + l)!

Proof.By expanding (e1 1)", in two ways: first, by the Exponential

Theorem and Mnltinomial ; secondly, by the Bin. Th., and each term of
the expansion by the Exponential. Equate the coefficients of xr in the two
results.
Other results are obtained by putting r = n + 2, ft + 3, &c.
The series (285), when divided by r!, is, in fact, equal to
the coefficient of xr in the expansion of

POLYGONAL NUMBERS.

95

286 By exactly the same process we may deduce from the

function {exe'x}n the result that the series
nr-n(n 2)r+n("~1) (n-4)r-&c.

takes the values 0 or 2". n ! , according as r is < n or = n ;

this series, divided by r ! , being equal to the coefficient of xr
in the expansion of
+ ST +!

POLYGONAL NUMBERS.

287

The ?ith term of the rth order of polygonal numbers is

equal to the sum of n terms of an Arith. Prog, whose first

term is unity and common difference r 2 ; that is
= | (2+(n-l)(r-2)} =
288

(-l) (r-2).

The sum of n terms

= n(n+l) , n(n-l)(n + l)(r-2)
2
6
By resolving into two series.

Order.

th torrn.

1
2
3
4
5
6

1
n
i(n+l)
n'
ii(3n-l)
(2n-l)n

l
l
l
l
l
l

l
2
3
4
5
6

1
1
3
4
6 10
9 16
12 22
15 28

1
5
15
25
35
45

1
1
6 7
21 28
36 49
51 70
66 91

n + _0^)(r_2) i, r, 3 + 3(r- 2), 4 + 6 (r-2), 5 + 10 (r-2),

6 + 15 (r-2), &c.

96

ALGEBRA.

In practiceto form, for instance, the 6th order of poly

gonal numberswrite the first three terms by the formula,
and form the rest by the method of differences.
Ex.:

6
5

[r-2 = 4]

15
9

28
13

45
17

21
4

FIGURATE

289

66 91 120 ...
25 29 ...
4 4 ...

NUMBERS.

The 11th term of any order is the sum of n terms of the

preceding order.
The ?ith term of the ;,th order is
n(n+l)
(n+r-2) = H{n>r_^
(r 1) !

290

The sum of n terms is

n(n+l),

Order.

[Bj98.

(n+r-1) = H(f|> r)

n,h term.

Figurate Numbers.

1, 1,

1,

1,

1,

1, 2,

3,

4,

5,

11

1, 3,

6, 10,

15,

21

n (n + l)
1.2

1, 4, 10, 20,

35,

56

n(n + l)(n + 2)
1.2.3

1, 5, 15, 35,

70, 126

1, 6, 21, 56, 126, 252

A. ( + l) (n + 2) (?i + 3)
1.2.3.4
n(n + l)(n + 2)(n + S)(n+4)
1.2.3.4.5

97

HYPERGEOMETEWAL SERIES.

HYPEEGEOMETRICAL SERIES.

291

i-+^.+*!fpJ&*p*
l.y
1.2.y(y+l)
+

g(g+l) (a+2)j8(j8+l)QB+2) - , 1.2.8.y(y+l) (y+2)

* +&C'

is convergent if a; is < 1 ,
and divergent if x is > 1 ;

(239 ii.)

and if x = 1, the series is

convergent if y a /3 is positive,
divergent if y a (3 is negative,

(239 iv.)

and divergent if y a /3 is zero.

(239 v.)

Let the hypergeometrical series (291) be denoted by

F(a, /3, y) ; then, the series being convergent, it is shewn by
induction that
292

FM+^r+1) = j_

concluding with
1-A-,
1 &C ...

1-A-2r-l
1 kirztr

where 1elt h2, k3, &c, with z2r, are given by the formulae
J.

_ (q + r-l)(y + r-l-i8).f
(y+2r-2) (y+2r-l)

,
2r
w

_ (ff+r) (y+r-a) x
~(y+2r-l)(y+2r)
_ F(a+r, fl+r+l, y+2r+l)
F(a+r,+r,y+2r)

The continued fraction may be concluded at any point

with kirzir. When r is infinite, ?.ir = 1 and the continued
fraction is infinite.
o

ALGEBRA.

98

293 Let
/(y) = 1 + I^ + 1.2.y(r+l) + 1.2.3.y(yr+l)(7+2)+&C-

the result of substituting - for x in (291), and making

j3 = a = oc, .

Then, by last, or independently by induction,

/(y+l)_l Jh Jh
p.
f{y)
1+ 1+ 1 + ... + 1 + &C.
WiihPm-(y+m-l){7+tn)'

294

In this result put y = % and

for x, and we obtain by

Exp. Th. (150),

= \ U. -l
the r*h component being . * ...
e*+e"
1+ a+ 5 + &c.
v
6 2r-l
Or the continued fraction may be formed by ordinary division
of one series by the other.

295

e" is incommensurable, m and n being integers.

m
the last and (174), by putting x =
n

From

INTEREST.

If r be the Iuterest on 1 for 1 year,

n

the number of years,

P
A

the Principal,
the amount in n years.

Then

296

At Simple Interest

A = P (1+wr).

297

At Compound Interest A = P(l + r)\

By (84).

INTEREST AND ANNUITIES.

298

99

Butt it
if the payments of")
of
.
Interest be made qi A = P(l+ ) .
times a year
)
^
{1 '

If A be an amount due in n years' time, and P the present

worth of A. Then
299

At Simple Interest

P=

300

At Compound Interest

P =
(1 + r)" .

301

- .
1+nr

By (296).
ByJ (297).
y '

Discount = A -P.

ANNUITIES.
302

303

The amount of an Annu-")

,
..
ity of 1 in n years, [ = n+ nvn
) r.
at Simple Interest . . . )
*
Present value of same

= w+|(-l) r

By (82).
By (299)

l+r
304

Amount at Compound")
(l+r)B 1
Interest
) _ "(i+r)__i
Present worth of same

305

""*' i+r)~n>
(1+r) 1

Amount when the pay-"\

/', , r \n*
ments of Interest/
\
q) ~
are made q times per C ~~ ~,
^~yi
annum
J
(H
) ""*

Present value of same

(LL

0+t)"-1

% (85).
By (300)

* ' lb^'

100

306

ALOEDHA.

Amount when the pay

ments of the Annuity
are made m times per
annum

(l+r)'-l
m {(l+r) 1}
l-(l+r)-

Present value of same

=
m {(l-fr)--l}

307

Amount when the In- '

terest is paid q times
and the Annuity m
times per annum ...

Present value of same

-K'+t)5-1)"

PROBABILITIES.

309 If all the ways in which an event can happen be m

in number, all being equally likely to occur, and if in n of
these m ways the event would happen under certain restrictive
conditions ; then the probability of the restricted event hap
pening is equal to 11
Thus, if the letters of the alphabet be chosen at random,
any letter being equally likely to be taken, the probability of
a vowel being selected is equal to
The number of un
restricted cases here is 26, and the number of restricted
ones 5.
310 If> however, all the m events are not equally probable,
they may be divided into groups of equally probable cases.
The probability of the restricted event happening in each
group separately must be calculated, and the sum of these
probabilities will be the total probability of the restricted
event happening at all.

PROBABILITIES.

101

Example.There are three bags A, B, and 0.

A contains 2 white and 3 black balls.
S

A bag is taken at random and a ball drawn from it. Required the pro
bability of tho ball being white.
Here the probability of the bag A being chosen = J-, and the snbseqnent
probability of a white ball being drawn = f.
Therefore the probability of a white ball being drawn from A
3
5
15'
Similarly the probability of a white ball being drawn from B
" 3.
7 ~ 7
And the probability of a white ball being drawn from G

=X v A=A
3
9
27'
Therefore the total probability of a white ball being drawn
_ i2_
1
4 _ 401
15
7
27
945'

If a be the number of ways in which an event can happen,

and b the number of ways in which it can fail, then the
311 Probability of the event happening = a ,.
a-\-b
312

Probability of the event failing

a+b"
Thus

Certainty = 1.

If p, p' be the respective probabilities of two independent

events, then
313
314
315

316

Probability of both happening

= pp'.

>>

of not both happening = 1pp'.

of one happening and one failing

= p+p'2pp'.
of both failing = (!-;>)(! -//).

102

If the probability of an event happening in one trial be p,

and the probability of its failing q, then
317

Probability of the event happening r times in n trials

= C(n, r)prqnr.

318

Probability of the event failing r times in n trials

= C (n, r) pn'Tqr.
[By induction.

319 Probability of the event happening at least r times in

trials = the sum of the first n
1 terms in the expansion
of (p + q)n.
320 Probability of the event failing at least r times in n
trials = the sum of the last nr + 1 terms in the same ex
pansion.
321 The number of trials in which the probability of the
same event happening amounts to p'
_ log (1-/0
log(l-p)"
From the equation (1 p)x = 1 p'.

322 Definition."When a sum of money is to be received if

a certain event happens, that sum multiplied into the proba
bility of the event is termed the expectation.
Examplk.If three coins be taken at random from a basr
containing one sovereign, four half-crowns, and five shillings,
the expectation will be the sum of the expectations founded
upon each way of drawing three coins. But this is also equal
to the average value of three coins out of the ten ; that is,
-njths of 35 shillings, or 10s. 6d.

323 The probability that, after r chance selections of the

numbers 0, 1, 2, 3 ... n, the sum of the numbers drawn will
be , is equal to the coefficient of .e* in the expansion of
(/ +

+ .T* + + O r + ( H + 1) r.

PROBABILITIES.

103

324 The probability of the existence of a certain cause of

an observed event out of several known causes, one of which
must have produced the event, is proportional to the a priori
probability of the cause existing multiplied by the probability
of the event happening from it if it does exist.
Thus, if the a priori probabilities of the causes be Pu P2
... &c, and the corresponding probabilities of the event hap
pening from those causes Qlt Q? &c, then the probability
of the rth cause having produced the event is
PrQr

325 It P'i, P'i &c. be the a priori probabilities of a second

event happening from the same causes respectively, then,
after the first event has happened, the probability of the
second happening is

=?

'.

POP'
For this is the sum of such probabilities as ~jpn{ which is
the probability of the rth cause existing multiplied by the
probability of the second event happening from it.
Ex. 1. Suppose there are
4 vases containing each 5 white and 6 black balls,
2 vases containing each 3 white and 5 black balls,
and 1 vase containing
2 white and 1 black ball.
A white ball has been drawn, and the probability that it came from the group
of 2 vases is required.
Here

P, = -|-,

P, = -|,

Ps = |

Q> = TV

' = !'

*=

Therefore, by (324), the probability required is

4.5
7.11

2L3
7.K
2.3
7.8

1.2
7.3

_ 99
427'

Ex. 2. After the white ball has been drawn and replaced, a ball is
drawn again ; required the probability of the ball being black.

104

ALGEBRA.

Here

P( = A,

P; = |,

P; = |.

The probability, by (325), will be

4.5.6
2.3.5 1.2.1
7.11.11 7.8.8 7.3.3_. 58639
112728
4.5
2.31.2
7.11 + 7.8 7.3
If the probability of the second ball being white is required, Q,QiQa
must be employed instead of PiPJPj.

326 The probability of one event at least happening out of

a number of events whose respective probabilities are a, b, c,
&c, is

Pl Pi+PsPi+&c.

where

Pi is the probability of 1 event happening,

Pi
i)
>
2

and so on.

For, by (316), the probability is

l-(l-a) (1-6) (1-c) ... = 2a-2a& + 2aic-, ...

327 The probability of the occurrence of r assigned events
and no more out of events is

where Qr is the probability of the r assigned events ; Qr+l the

probability of r+1 events including the r assigned events.
For if a, b, c ... be the probabilities of the r events, and
a', b', c ... the probabilities of the excluded events, the re
quired probability will be
abc...(l-a')(l-b')(l-c')...
= abc ... (1 - 2a' + 2a'&'-2a'&V +...)
328

The probability of any r events happening and no more

is

2Qr 2Qr+i+2Qr+2-&c.
Note.If a=b=c= &c, then 2Qr = G (n, r) Qr) &c.

INEQUALITIES.

105

INEQUALITIES.

330

.2

," lies between the greatest and least of

Oi + 02+...+0

the fractions , -, ... -j-2-, the denominators being all of

the same sign.
Proof.Let k be the greatest of the fractions, and any other; then
ar<kbr. Substitute in this way for each a.
fraction.

331

^ > Vab.

332
or,

Similarly if h be the least

a1+a2+...+an > v
n

--

Arithmetic mean > Geometric mean.

Proof. Substitute both for the greatest and least factors their Arith
metic mean. The product is thus increased in value. Repeat the process
indefinitely. The limiting value of the G. M. is tho A. M. of the quantities.

333

^ > (!+)".

excepting when m is a positive proper fraction.

Proof:
where x =

334

<- + &-= (^)"{(l+)" + (l-ai)"},

r.
a+b

Employ Bin. Th.

>\

)'

excepting when m is a positive proper fraction.

p

ALGEBRA.

106

Otherwise. Tlie Arithmetic mean of the 111th powers is

greater than the 111th power of the Arithmetic mean, excepting
when m is a positive proper fraction.
Proof. Similar to (332).
left side, employing (333).

336
then

Substitute for the greatest and least on the

If * and m are positive, and x and mx less than unity ;

(l+#)-"> 1 mx.

(125,240)

337 If ^, m, and n are positive, and n greater than m ; then,

by taking x small enough, we can make
1+nx > (l+.r)*.
For x may be diminished until 1 +nx is > (1 wi*)_1, and this
is > (l+x)m, by last.

338

If x be positive,

If x be positive and > 1 ,

log

< x.

(150)

log (1+*) > .r

(155, 240)
m

If x be positive and < 1,

339

log

> x.

(156)

When n becomes infinite in the two expressions

1.3.5 -(2n-l)
2.4.6... 2n

,
'

3.5.7.-(2 + l)
2.4.6... 2n '

the first vanishes, the second becomes infinite, and their

product lies between and 1.
Shewn by adding 1 to each factor (see 73), and multi
plying the result by the original fraction.

340

If m be > n, and n > a,

SCALES OF NOTATION.

341

107

If a, b be positive quantities,
uab" is > ( -^-)

Similarly

a" bb (f > ^n+b+cy+b+c

These and similar theorems may be proved by taking loga

rithms of each side, and employing the Expon. Th. (158), &c.

SCALES OF NOTATION.

342

If N be a whole number of n+ I digits, and r the radix

of the scale,

N = p r" +p-i r" " 1 +P -2 r"~2 + +lh >' +Po >

where pn, p-i> Pa are the digits.

343

Similarly a radix-fraction will be expressed by

where plt p2, &c. are the digits.

Examples:

3426 in the scale of 7 = 3.7s + 4 . 7J + 2.7 +6 ;

1045 in the same scale = i + 2i +
+ f;n
<7a
Ha
H*

344 Ex. To transform 34268 from the scale of 5 to the

scale of 11.
Rule. Divide successively by the new radix.
11 : 34268
llj 1343-t
11 1 40-3
19

Result 193, in which t stands for 10.

ALGEBRA.

108

345 Ex. To transform ^Ot'l from the scale of 12 to that

of 7, e standing for 11, and t for 10.
Rule. Multiply successively by the new radix.
tQel
7
5*657
7
61931
7
1-04977
0 2971

Result

5610 ...

346 Ex.In what scale does 2t7 represent the number 475
in the scale of ten ?
Solve the equation
Result

2iJ + 10r + 7 = 475.

[178

r = 13.

347 The sum of the digits of any number divided by r 1

leaves the same remainder as the number itself divided by
r 1 ; r being the radix of the scale.
(401)
348 The difference between the sums of the digits in the
even and odd places divided by r-f-1 leaves the same re
mainder as the number itself when divided by r-f-1.

THEORY OF NUMBERS.

349

If o> is prime to b, ~ is in its lowest terms.

Proof.Let =
a fraction in lower terms,
o
o,
Divide a by ax, remainder as quotient ql ,
b by 6 remainder 62 quotient qx ;
and so on, as in finding the H. C. F. of a and a and of b and 6, (see 30).
Let an and b be the highest common factors thus determined.

THEORY OF NUMBERS.

Then, because
and so on ; thus

=
O
t,

.-. "-==2A=.,
0
0 JjO,
0,

- = -1- = 1 = &c
o
o,
o,

109

(70)

= 4s->

Therefore a and 6 are equimultiples of aH and bn ; that is, o is not prime to 6

if any fraction exists in lower terms.

350

K a is prime to b, and -^r = -r- 5 then a' and fc' are

bo
equimultiples of a and 6.
Proof.Let ~ reduced to its lowest terms be
Then ?- = , and,
b
9
2"
since p is now prime to q, and a prime to b, it follows, by 349, that ?~ is
neither greater nor less than ~ ; that is, it is equal to it.
b
351

Therefore, &c.

If ab is divisible by c, aud a is not ; then b must be.

Proof.Let

~=q;
.-. = -9-.
c
CO
But a is prime to c; therefore, by last, 6 is a multiple of c.
352
to c.

If a and b be each of them prime to c, ab is prime

[By (351).

353 If abed... is divisible by a prime, one at least of the

factors a, b, c, &c. must also be divisible by it.
Or, if p be prime to all but one of the factors, that factor
is divisible by p.
(351)
354 Therefore, if a" is divisible by p, p cannot be prime to
a, ; and if p be a prime it must divide a.

355 If a is prime to b, any power of a is prime to any

power of b.
Also, if a, b, c, &c. are prime to each other, the product
of any of their powers is prime to any other product of their
powers.

ALGEBRA.

110

356 No expression with integral coefficients,

A + Bx + Cx2+
can represent primes only.

such

as

Proof.For it is divisible by x if A = 0 ; and if not, it is divisible by A,

when x = A,
357

The number of primes is infinite.

Proof.Suppose if possible p to be the greatest prime. Then the pro

duct of all primes up to p, plus unity, is either a prime, in which case it
would be a greater prime than p, or it must be divisible by a prime ; but
no prime up to p divides it, because there is a remainder I in each case.
Therefore, if divisible at all, it must be by a prime greater than p. In
either case, then, a prime greater than p exists.
358 If a be prime to b, and the quantities a, 2a, 3a,
(b 1) a be divided by b, the remainders will be different.
Proof. Assume manb = ma rib, m and ?j being less than 6,
o
nn
Then by (350).
b
to m
359 A number can be resolved into prime factors in one
way only.
[By (353).
360

To resolve 5040 into its prime factors.

Rule. Divide by the prime numbers successively.

2x5! 5040
2 1 504
21 252
2 1 126
7 1 63
319
3

Thus 5040 = 2*. 3'. 5. 7.

361 Required the least multiplier of 4704 which will make

the product a perfect fourth power.
By (196),
Then

4704 = 2s. 3. T.
25.3,.7tx2,.3'.7s = 28.34.7'= 844,

the indices 8, 4, 4 being the least multiples of 4 which are not less than
5, 1, 2 respectively.
Thus 2\3'.72 = 3584 is the multiplier required.

Ill

THEORY OF NUMBERS.

362 All numbers are of one of the forms 2m or 2m + 1

2n or 2m 1
3noral
4n or 4m 1 or 4m +2
4m or 4m1 or 4m 2
5m or 5m 1 or 5m 2
and so on.
363

All square numbers are of the form 5m or 51.

Proof.By squaring the forms on, 5n 1, 5n 2, which comprehend

all numbers whatever.
364 All cube numbers are of the form 7" or 71And similarly for other powers.
365 The highest power of a prime p, which is contained in
the product m ! , is the sum of the integral parts of
m

tn

p*

p3

For there are

- factors in m ! which p will divide ; ; which

P
P
it will divide a second time ; and so on.
The successive
divisions are equivalent to dividing by
m m
wi f m t
pp.pp1... &c. = pp P' .
Example. The highest power of 3 which will divide 29!. Here the
factors 3, 6, 9, 12, 15, 18, 21, 24, 27 can be divided by 3. Their number is
09
= 9 (the integral part).
The factors 9, 18, 27 can be divided a second time.
29 = 3 (the integral part).
One factor, 27, is divisible a third time.

Their number is

29
- 1- = 1 (integral part).

9 + 3 + 1 =13; that is, 3" is the highest power of 3 which will divide 29!.
366 The product of any r consecutive integers is divisible
by r ! .
Proof:

i Lr.'jS*

ja necessarily an integer, by (96).

112

ALGEBBA.

367 If be a prime, every coefficient in the expansion of

(a + b)n, except the first and last, is divisible by n.
By last.
368 If be a prime, the coefficient of every term in the ex
pansion of (a + b + c ...)", except a", b", &c, is divisible by n.
Pkoof.By (367).

369

Put 0 for (6 + c+ ...).

Fermat's Theorem. If p be a prime, and iV prime to

p ; then
Proof:

N13-1 1 is divisible by p.
N" = (1 + 1 + ...) = N+Mp.

By (368).

370 If J? be any number, and if 1, a, b, c, ... (p 1) be all

the numbers less than, and prime to p ; and if n be their
number, and x any one of them ; then of 1 is divisible by p.
Proof.If x, ax, bx ... (pl)x be divided by p, the remainders will be
all different and prime to p [as in (358)] ; therefore the remainders will be
1, o, 6, c ... (p 1) ; therefore the product
x'abc ... (p 1) = abc ... (pl)+Mp.

371 Witeon's Theorem. If p be a prime, and only then,

1 + (j3 1) ! is divisible by p.
Put p 1 for r and n in (285), and apply Fermat's Theorem
to each term.
372 If p be a prime = 2n + l, then (!/+( 1)" is divisible
by p.
Proof. By multiplying together eqni-distant factors of (j> 1) ! in
Wilson's Theorem, and putting "ln + 1 for p.
373

Let N = ap V cr . . . in prime factors ; the number of in

tegers, including 1 , which are less than n and prime to it, is

Proof. The number of integers prime to N contained in a' is a'

Similarly in Z/7, if, &c.

Take the product of these.

THEORY OF NUMBERS.

113

Also the number of integers less than and prime to

(NX If X &c.) is the product of the corresponding numbers
for N, M, &c. separately.
374 The number of divisors of N, including 1 and N itself,
is' = (p + 1) (? + l) (r+l) .... For it is equal to the number
of terms in the product
(l+a+ ... +a?) (1 + 6 + ...+&*) (l + c + ... +<f) ... &c.
375 The number of ways of resolving N into two factors is
half the number of its divisors (374). If the number be a
square the two equal factors must, in this case, be reckoned
as two divisors.
376 If the factors of each pair are to be prime to each other,
put p, q, r, &c. each equal to one.
377

The sum of the divisors of N is

a?+1-l b"+1-l cr+1-l
a-l

'

6-1

'

c-1

Proof.By the product in (374), and by (85).

378

K P be a prime, then the p 1th power of any number

is of the form mp or rap + 1.

By Fermat's Theorem (369).
Ex.The 12tt power of any number is of the form 13m or 13j + 1.
379

To find all the divisors of a number ; for instance, of 504.

I.

n.

504
252
126
63
21
7

2
2
2
3
3
7

1
2
4
8
3
6 1.2
9
18 36
7
14 28
84 168 63

24
72
56
126

21
252

42
504

Explanation. ReBolve 504 into its primo factors, placing them in

column 11.
Q

ALGEBRA.

114

The divisors of 504 are now formed from the numbers iu colnmn II., and
placed to the right of that column in the following manner :
Place the divisor 1 to the right of column II., and follow this rule
Multiply in order all the divisors which are written, down by the next number
in column II., which has not already been used as a multiplier : plate the first
new divisor so obtained and all the following products in order to the right of
column II.
380

Sr the sum of the rth powers of the first n natural

numbers is divisible by 2n + l.
Proof :

as {x*- 1') (a;1 - 2') ... (a? - if)

constitutes 2n + l factors divisible by 2n+l, by (3G6). Multiply out, re

jecting x, which is to be less than 2n + l. Thus, using (372),
ar-S^-' + Stz"-*- ... S..ix' + (-l)*([ny = M (2 + l).
Put 1, 2, 3 ... (n 1) in succession for x, and the solution of the (n 1)
equations is of the form

8, = M (2?i + l).

THEORY

OF

EQUATIONS.

FACTOKS OF AN EQUATION.

Generalform of a rational integral equation ofthe nth degree.

400

p0x"+p1xH-1+pix*-i+ ... +p-ix+pn = 0.

The left side -will be designated f(x) in the following

summary.

401 If /(*) be divided by as a, the remainder will be f(a).

By assuming f(x) = P (as a) + B.
402

If a be a root of the equation / (x) = 0, then /(a) = 0.

403 To compute /(a) numerically; divide f(x) by as a,

and the remainder will bef{a).
[401
404

Example.To find the value of 4a"-3a;,,+12;B4-ie, + 10 when a = 2.

4-3 + 12 +0 -1 +0 +10
8 + 10 + 44 + 88 + 174 + 348
4+5 + 22 + 44+87 + 174 + 358

Thus /(2) = 358.

If a, b, c ... h be the roots of the equation /(as) = 0;

then, by (401) and (402),
405

fix) = p0 (xa) (xb) (xc) ... (xk).

By multiplying out tho last equation, and equating coefficients with

equation (400), considering pa = 1, the following results are obtained:

116

406

THEORY OF EQUATIONS.

P\=
_
*
_ __
^8 ~"

the
( the
\
( the
\

sum of all the roots of / (x) .

sum of the products of the roots taken
two at a time.
sum of the products of the roots taken
three at a time.

/ly _ Cthe sum of the products of the roots taken

'
) Pr X
rata time.
( l)"p =

product of all the roots.

407 The number of roots of f(x) is equal to the degree of

the equation.
408

Imaginary roots must occur in pairs of the form

a+py/^l,

0-/3v^l.

The quadratic factor corresponding to these roots will

then have real coefficients ; for it will be
tf-Zax+a'+p*.

[405, 226

409 If f(x) be of an odd degree, it has at least one real root

of the opposite sign to pn.
Thus x!\ = 0 has at least one positive root.

410 If /(*) be of an even degree, and pn negative, there is

at least one positive and one negative root.
Thus xl 1 has +1 and 1 for roots.

411 If several terms at the beginning of the equation are of

one sign, and all the rest of another, there is one, and only
one, positive root.
Thns xi+2a>*+3x*+x* 5x 4 = 0 has only one positive root.

412

If all the terms are positive there is no positive root.

413 If all the terms of an even order are of one sign, and
all the rest are of another sign, there is no negative root.
4J4

Thus a4 se*-|- as* x + 1 = 0 has no negative root,

DISCRIMINATION OF BOOTS.

117

415 If all the indices are even, and all the terms of the same
sign, there is no real root ; and if all the indices are odd, and
all the terms of the same sign, there is no real root but zero.
Thus x*+x*+l = 0 has no real root, and as5 + as* + as = 0 has no real root
but zero. In this last equation there is no absolute term, because such a
term would involve the zero power of x, which is even, and by hypothesis is
wanting.

DESCARTES' RULE OF SIGNS.

416 In the following theorems every two adjacent terms in

f(x), which have the same signs, count as one " continuation
of sign"; and every two adjacent terms, with different signs,
count as one change of sign.
417 f(x)> multiplied by (x a), has an odd number of
changes of sign thereby introduced, and one at least.
418 fix) cannot have more positive roots than changes of
sign, or more negative roots than continuations of sign.
419 When all the roots of f(x) are real, the number of
positive roots is equal to the number of changes of sign in
f(x) ; and the number of negative roots is equal to the number
of changes of sign in f(x).
420

Thus, it being known that the roots of the equation

as4-10as+35.is>-50as + 24 = 0
are all real ; the number of positive roots will be equal to the number of
changes of sign, which is four. Also /( as) = as* + lOas* + 35as* + 50as + 24 = 0,
and since there is no change of sign, there is consequently, by the rule, no
negative root.

421 If the degree oif(x) exceeds the number of changes of

sign in f(x) and /( #) together, by ft, there are at least /*
imaginary roots.
422 If between two terms in f(x) of the same sign, there
be an odd number of consecutive terms wanting, then there
must be at least one more than that number of imaginary
roots ; and if the missing terms lie between terms of different

THEORY OF EQUATIONS.

118

sign, there is at least one less than the same number of

imaginary roots.
Thus, in the cubic equation x*+4x7 = 0, there must be two imaginary
roots.
And in the equation * 1 = 0 there are, for certain, four imaginary roots.
423 If an even number of consecutive terms be -wanting in
fix), there is at least the same number of imaginary roots.
Thus the equation as'+l = 0 has four terms absent ; and therefore four
imaginary roots at least.

THE DERIVED FUNCTIONS OF f(x).

Rule for forming the derived functions.

424 Multiply each term by the index of x, and reduce the
index by one ; that is, differentiate the function with respect
to x.
Example.Take
/ (*) =
ars+ x*+ xs x'x-l
f(x)= bx*+ 4xs + 3x'-2x-l
f(x) = 20z*+12a:*+6a>-2
f(x)= 60^ + 24.6+6
f{x) = 120a +24
f (*) = 120
f (x)> f 00 > &c- are called the first, second, &c. derived functions of/ (as).

425 To form the equation whose roots differ from those of

f(x) by a quantity a.
Put x = y + a in f{x), and expand each term by the Binomial
Theorem, arranging the results in vertical columns in the fol
lowing manner :
/(a+y) =

(a+yY+(a+yy+(a+yy-(a+yy-(a+V)-l
as + a* + a a? a 1
+ ( 5a* + 4aJ + 3a1 2a - l)y
+ (10a* + 6a + 3a - l)y*
+ (10a" + 4a
+ l)y*
+ ( 5a + l)y*
+ '/

TBANSFORHATION OF AN EQUATION.

119

Comparing this result with that seen in (424), it is seen that

426

/(+) =Aa)+f<a)y

so that the coefficient generally of if in the transformed

equation is J , v

427 To form the equation most expeditiously when a has a

numerical value, divide f(x) continuously by xa, and the
successive remainders will furnish the coefficients.
Example.To expand /(y + 2) when, as in (425),
Divide repeatedly by z 2, as follows :
1 + 1 + 1- 1- 1-1
+ 2 + 6 + 14 + 26 +50
1 + 3 + 7 + 13 + 25 +49 =/(2)
2
+ 2 +10 + 34 + 94
1 + 5 +17 + 47 + 119 = /'(2)
2
+ 2 +14 + 62
1 + 7+31 +109 =
That these remainders
+ g+lgj
are the required coefficients
1 + 91+49 =
is seen by inspecting the
2
Li
form of the equation (426) ;
for if that equation be di
i
vided by xa = y repeat
edly, these remainders are
obviously produced when
a = 2.
Thus the equation, whose roots are each less by 2 than the roots of the
proposed equation, is ^5 + llj(4+4V+10V+119f +49 = 0.
428 To make any assigned term vanish in the transformed
equation, a must be so determined that the coefficient of that
term shall vanish.
Example.In order that there may be no term involving y* in equation
(426), we must have /*(a) = 0.
Find /4(o) as in (424);
thus
120a + 24 = 0;
.-. o = -|.
The equation in (424) must now be divided repeatedly by z+\ after the
manner of (427), and the resulting equation will be minus its second term.

120

429

THEORY OF EQUATIONS.

Note, that to remove the second term of the equation

f^x) = 0, the requisite value of a is =

; that is, the

coefficient of the second term, with the sign changed, divided by

the coefficient of the first term, and by the number expressing
the degree of the equation.
430 To transform f(x) into an equation in y so that y <p (z),
a given function of x, put x = <f>~1(y), the inverse function of y.
Example.To obtain an equation whose roots are respectively three times
the roots of the eqnation xl6x + l=0. Here y = 2x; therefore x = ~,
and the equation becomes

^- + 1 = 0, or y* 54y + 27 = 0.

431 To transform f(x) = 0 into an equation in which the

coefficient of the first term shall be unity, and the other
coefficients the least possible integers.
Example.Take the equation
288z,+240a:,-176;r- 21 = 0.
Divide by the coefficient of the first term, and reduce the fractions ; the
5
11 m -
7 = 0.
equation becomes

6
lo
yo
Substitute
for ar, and multiply by k% ; we get
y

6 r

18 y

96

Next resolve the denominators into their prime factors,

. . 5t . IIP
7W _Q
*
2.3 ^
2.3' V
2s. 3
The smallest value must now be assigned to Jc, which will suffice to make
each coefficient an integer. This is easily seen by inspection to be 2*. 3 = 12,
and the resulting equation is y*+10y* 88y 126 = 0,
the roots of which are connected with the roots of the original eqnation by
the relation
y = 12x.

EQUAL ROOTS OF AN EQUATION.

By expanding f(x+z) in powers of a by (405), and also

by (426), and equating the coefficients of a in the two ex

EQUAL ROOTS.

121

pansions, it is proved that

432

f{*) = fisL + J^ + 1S*L + &c,

(a; a)
b)
{vc)

from which result it appears that, if the roots a, b, c, &c. are

all unequal, f(x) and f (as) can have no common measure in
volving x. If, however, there are r roots each equal to a,
s roots equal to b, t roots equal to c, &c, so that
then
433

/-^ + tfj +

and the greatest common measure oif(x) and fix) will be

444
ix-ay^ix-by-'ix-c)'-1...
When a; = a, fix), fix), ,..fr~l{x) all vanish.

Similarly

when x = b, &c.

Practical method offinding the equal roots.

445

Let /(*) = X, Xj Xj Xj Xj . . . X~ where

Xi = product of all the factors like (as a),
Xj =

(j; a)',
Xj =

(a; -a)5.
Find the greatest common measure of f(x) and /' (x) = I\ () say,
F.C*) and
= *!(),
F,(as) and .P; (*) = *", (as),

Lastly, the greatest common measure of Fm.\(m) and F'm.,(x) = Fm(x) = 1.

Next perform the divisions
f{z) + F.OO =
say,

And, finally,

^ (a;) -h

(as) = X,,

(a) +
=

= Xm_
= X,,..
E

[T. 82.

122

THEORY OF EQUATIONS.

The solution of the equations X, = 0, X, = 0, &c. will furnish all the

roots of/ (x) ; those which occur twice being found from X, ; those which
occur three times each, from X, ; and so on.

446 K /(*) has all its coefficients commensurable, X^X^X^

&c. have likewise their coefficients commensurable.
Hence, if only one root be repeated r times, that root must
be commensurable.
447 In all the following theorems, unless otherwise stated,
/(*) is understood to have unity for the coefficient of its first
term.

LIMITS OF THE BOOTS.

448 If the greatest negative coefficients mf(x) and /( )
be p and q respectively, then p-\-l and (q + 1) are limits of
the roots.
449 If n_r and xn~' are the highest negative terms in f(x)
and /(a1) respectively, (1+ \Zv) and (l + v'j) are limits
of the roots.
450

If k be a superior limit to the positive roots of /( J ,

then -j- will be an inferior limit to the positive roots of f{x).

451 If each negative coefficient be divided by the sum of all
the preceding positive coefficients, the greatest of the fractions
so formed + unity will be a superior limit to the positive
roots.
452

Newton's method.Put x= h +y in f(x) ; then, by (426),

f(h+y)=f(h)+yf(h) + f(h) + ... +j*>(*) =0.

Take h so that f(h), f(h), f (h) .../" (h) are all positive;
then h is a superior limit to the positive roots.
453 According as /(a) and/ (b) have the same or different
signs, the number of roots intermediate between a and b is
even or odd.

INTEGRAL BOOTS.

123

454 Rolle's Theorem.One real root of the equation f (x)

lies between every two adjacent real roots off(x).
455 Cob. 1.f(x) cannot have more than one root greater
than the greatest root in f (x) ; or more than one less than
the least root in /"(*).
456 Cos. 2.If f{x) has m real roots, fr(x) has at least
mr real roots.
457 Cob. 3.If f(x) has n imaginary roots, f(x) has also
P at least.
458 Cob. 4.If a, /3, y ... k be the roots of f(x) ; then the
number of changes of sign in the series of terms
/(). /(). /()> /(?)... /(-)
is equal to the number of roots oif(x).

NEWTON'S METHOD OF DIVISORS.

459

To discover the integral roots of an equation.

Example.To ascertain if 5 be a root of

x'-te+mxt-nSx + YOh = 0.
If 5 be a root it will divide 105.
next coefficient. Result, 155.

Add the quotient to the

If 5 be a root it will divide 155.

the next coefficient ; and so on.

Add the quotient to

If the number tried be a root, the divisions will be effectiblo

to the end, and the last quotient will be 1, or po, if^o be
not unity.

5)105
21
-176
5) -155
-31
86
5^55
11
-6
5) -5
-1

460 In employing this method, limits of the roots may first

be found, and divisors chosen between those limits.
461 Also, to lessen the number of trial divisors, take any
integer m ; then any divisor a of the last term can be rejected
if a m does not divide /(m).
In practice take m = +l and 1.
To find whether any of the roots determined as above are
repeated, divide f(x) by the factors corresponding to them,
and then apply the method of divisors to the resulting equation.

124

THEORY OF EQUATIONS.

Example.Take the equation

z+23*-17zl-26x*+83x,+ 72x-144, = 0.
Putting z=l, we find /(l) = 24. The divisors of 144 are
1, 2, 3, 4, 6, 8, 9, 12, 16, 24, &c.
The values of a m (since m = 1) are therefore
0, 1, 2, 3, 5, 7, 8, 11, 15, 23, &c.
Of these last numbers only 1, 2, 3, and 8 will divide 24. Hence 2, 3, 4, and
9 are the only divisors of 144 which it is of use to try. The only integral
roots of the equation will be found to be 2 and 3.
462 If /(<) and F(X) have common roots, they are con
tained in the greatest common measure of /(a?) and F(X).
463 If /(*) has for its roots a, <j>(a), b, <f>(b) amongst others ;
then the equations f(.r) = 0 and/j: <f>(a;) j = 0 have the common
roots a and b.
464 But, if all the roots occur in pairs in this way, these
equations coincide.
For example, suppose that each pair of roots, a and b, satisfies the equation
a+b = 2r. We may then assume a b 2z. Therefore / (z + r) = 0. This
equation involves only even powers of z, and may be solved for z1.
465 Otherwise: Let abz=z\ then /(x) is divisible by (z a) b)
= 2rx + z. Perform the division until a remainder is obtained of the
form Px + Q, where P and Q only involve z.
The equations P= 0, Q = 0 determine z, by (462) ; and a and b are found
from a + b = 2r, ab z.

RECIPROCAL EQUATIONS.

466

A reciprocal equation has its roots in pairs of the form

a, ; also the relation between the coefficients is

a
Vr Pn-r, OT else fT = pn.r.
467 A reciprocal equation of an even degree, with its last
term positive, may be made to depend upon the solution of an
equation of half the same degree.

BINOMIAL EQUATIONS.

468

Example:

125

4a>l-24a?+b7&-7&i? + b7a?-24x + 4> = 0

is a reciprocal equation of an even degree, with its last term

positive.
Any reciprocal equation which is not of this form may be
reduced to it by dividing by x-\-l if the last term be positive ;
and, if the last term be negative, by dividing by x1 or x'2 l,
so as to bring the equation to an even degree. Then proceed
in the following manner :
469 First bring together equidistant terms, and divide the
equation by x* ; thus
4 if + 7?) - 24(^ + ?) + 67 ( + v) " 73 = "
By putting x + = y, and by making repeated use of the
1
/
1 V
relation a? -\- = \ x -\- J 2, the equation is reduced to
a cubic in y, the degree being one-half that of the original
equation.

Put p for x + , and pm for x, + .

x
xM
470 The relation between the successive factors of the form
pm may be expressed by the equation
Pm = PPm-lp,n-2471

The equation for^?m, in terms of p, is

p = p-mp-*+ m(-$)

...

+ ( 1)' m(m~ r~ D (m 2r+l) pm-Sr^^

By (545), putting q = 1.

BINOMIAL EQUATIONS.
472

If a be a root of xn~i =0, then am is likewise a root

where m is any positive or negative integer.

473

If a be a root of x" + 1 = 0, then a2"*+1 is likewise a root.

126

THEORY OF EQUATIONS.

474 If m and n be prime to each other, * 1 and as" 1

have no common root but unity.
Take pmqn = 1 for an indirect proof.
475 H n be a prime number, and if a be a root of x* 1 = 0,
the other roots are a, a2, a? ... a*.
These are all roots, by (472). Prove, by (474), that no two can be equal.
476 If n be not a prime number, other roots besides these
may exist. The successive powers, however, of some root
will furnish all the rest.
477 If
1 = 0 has the index n mpq ; m, p, q being
prime factors ; then the roots are the terms of the product
(l+a + 2 + ... 4--1)(l+0 + 0' + ... + /3"-1)
X (1+7 + 7*+ - +7*_1)
where a is a root of as" 1,
0

f-l,
7

x9 I,
but neither n, j3, nor 7 = 1.
Proof as in (475).
478

If n = m\ and
a be a root of xml = 0,
/3

xm-a = 0,

af-0 = O;
then the roots of * 1 = 0 will be the terms of the product
(l+a + a2+ ... +,i) (l+/3 + /32+...+/3-1)
X(l + 7 + 7a+ - +7"'1)479
+ 1 = 0 may be treated as a reciprocal equation, and
depressed in degree after the manner of (468).
480

The complete solution of the equation

xn -1 = 0

is obtained by De Moivre's Theorem.

The n different roots are given by the formula

(757)

2r7T ,
/r . 2rir
v 1 sin
n
n
in which r must have the successive values 0, 1, 2, 3, &c,
concluding with ~, if n be even ; and with n~ , if n be odd.
x = cos

CUBIC EQUATIONS.

481

127

Similarly the n roots of the equation

xn + 1 = 0

are given by the formula

= cos (+l)'^ZIgin (gr+Dw
n
n
r taking the successive values 0, 1, 2, 3, &c, up to w~ , if
g
n be even ; and up to

482

^
, if be odd.

The number of different values of the product

Am &*

is equal to the least common multiple of m and n, when m and

n are integers.

CUBIC EQUATIONS.

483

To solve the general cubic equation

a?+pa?-\-qx+r = 0.

Remove the term px2 by the method of (429).

formed equation be

Let the trans

a?-\-qoc-\-r = 0.

484 Cardan's method.The complete theoretical solution

of this equation by Cardan's method is as follows :
Put

x = y+z

(i.)

tf+f+(3yZ+q) (y+z)+r = 0.
Put

Syz + q = 0;

/.

V = ^-

Substitute this value of y, and solve the resulting quadratic

in if. The roots are equal to if and z3 respectively ; and we
have, by (i.),

128

THEORY OF EQUATIONS.

The cubic must have one real root at least, by (409).

Let in be one of the three values of \
t
of the three valnes of | ^ \J ~&
486

t" + /~~" + ^;1^> and n

"i
V 4
*7 )
27 } ^'

Let 1, a, a' be the three cube roots of unity, so that

=_ A +-1

and a3 = - 1 - 1 v^8.

[472

487

Then, since v^w' =

the roots of the cubic will be
m + n, am + a1n, a'm + an.
Now, if in the expansion of

by the Binomial Theorem, we put

/* = the sum of the odd terms, and
v = the sum of the even terms ;
then we shall have m = ft + v,
and n = fi v
or else
m fi + v V 1, and n = fi v s/ 1 ;
according as /~T" + j?= is real or imaginary.
By substituting these expressions for ?ij and n in (487), it appears that
488

(i-) If -;- + ^ be positive, the roots of the cubic will be

2ji, fi + v</ 3, ft -iV-3.
(ii.) If + X- be negative, the roots will be
2ft,

-ft + v</3,

-fi-y^/3.

(iii.) If ^ + ^ = 0, the roots are

2m,

m,

m ;

since m is now equal to ft.

489

The Trigonometrical method.The equation

a? + qx + r = 0
may be solved in the following manner, by Trigonometry,
when -|- ^ is negative.
Assume x = cos a.

Divide the equation by n9; thus

cos3 a + -2- cos o +

= 0.
n"
n"
But

cos3 a - A cos a - 521^ = 0.

By (657)

BIQUADRATIC EQUATIONS.

129

Equate coefficients in the two equations ; the result is

a must now be found with the aid of the Trigonometrical

tables.
490

The roots of the cubic will be

n cos a, n cos (fw+a), n cos (fir a).

491

Observe that, according as + ^ is positive or nega-

tive, Cardan's method or the Trigonometrical will be practi

cable. In the former case, there will be one real and two
imaginary roots; in the latter case, three real roots.

BIQUADRATIC EQUATIONS.

492

Descartes' Solution.To solve the equation

x* + qa? + rx + * = 0

(i.)

the term in a? having been removed by the method of (429).

Assume

(o?+ex+f) (a?ex+g) = 0

(ii.)

Multiply out, and equate coefficients with (i.) ; and the fol
lowing equations for determining /, g, and e are obtained
g+f= q+e\
493

g-f= j,

gf=s

e+2qei+(qils) e*-r* = 0

(iii.)
(iv.)

494 The cubic m e1 is reducible by Cardan's method, when the biquadratic

has two real and two imaginary roots. For proof, take a ifi and a y as
the roots of (i.), since their sum must be zero. Form the sum of each pair
for the values of e [see (ii.)], and apply the rules in (488) to the cubic in e*.
If (he biqiuuhatic has all its roots real, or all imaginary, the cubic will have
all its roots real. Take a i(3 and a iy for four imaginary roots of (i.),
and form the values of e as before.
495 If a* F> y* be the roots of the cubic in e', the roots of the biquadratic
mil be _I(a + /; + y), J(a+/3-y), (/3 + y-a), l(y + a-/3).
S

130

THEORY OF EQUATIONS.

For proof, take w, x, y, z for the roots of the biquadratic ; then, by (ii.), the
gum of each pair mast give a value of e. Hence, we have only to solve the
symmetrical equations
y + z = a,
w + x = a,
z +x = /3,
w + y l3,
x+y = y,
vi + z = y.
496

Ferrari's solution.To the left member of the equation

rf+paf+qaf+nv+s = 0,

fc8
add the quantity ax* + bx + , and assume the result

497 Expanding and equating coefficients, the

cubic equation for determining m is obtained

following

8m*4*qm*-f- (2prSs) m-\-4qsp2sr 0.

Then x is given by the two quadratics
2ax+b

498 The cubic in tn is reducible by Cardan's method when the biquadratic

has two real and two imaginary roots. Assume a, /3, y, S for the roots of the
biquadratic ; then a/3 and yl are the respective products of roots of the two
quadratics above. From this find m in terms of afiyl.
499
have

Euler's solution. Remove the term in x3; then we

of+qtf+rx+s = 0.

500 Assume x = y + z + u, and it may be shewn that y*, a8,

and u* are the roots of the equation
+ 2r+
501

16

64

U-

The six values of y, z, and u, thence obtained, are

restricted by the relation

yzu =

8
Thus x = y+z+u will take four different values.

COMMENSURABLE ROOTS.

131

COMMENSURABLE ROOTS.

502

To find the commensurable roots of an equation.

First transform it by putting * = -|- into an equation of

the form

xn+p1xn~1+p2xn~i+ ... +/> = 0,

having jfl0 = 1, and the remaining coefficients integers.

(431)

503 This equation cannot have a rational fractional root,

and the integral roots may be found by Newton's method of
Divisors (459).
These roots, divided each by k, will furnish the commen
surable roots of the original equation.
504

Example.To find the commensurable roots of the equation

81a;5-207a!4-9a!, + 89a!s + 2a!-8 = 0.

Dividing by 81, and proceeding as in (431), we find the requisite substitu

tion to be

x = 3L.
y

The transformed equation is

y6- 23y'~ 9y> + 80ly' + l62y- 5832 = 0.
The roots all lie between 24 and 34, Joy (451).
The method of divisors gives the integral roots
6, 4, and 3.
Therefore, dividing each by 9, we find the commensurable roots of the original
equation to be
J, and .
505 To obtain the remaining roots ; diminish the transformed equation by
the roots 6, 4, and 3, in the following manner (see 427) :
1 -23- 9 + 801 + 162-5832
6
6-102-666 + 810 + 5832
1 -17-111 + 135 + 972
4
- 4+ 84 + 108-972
1 -21- 27 + 243
3
3- 54-243
1 -18- 81
The depressed equation is therefore
y* - I8y - 81 = 0.
The roots of which are 9(l+\/2) and 9(1 ^2); and, consequently, the
incommensurable roots of the proposed equation are 1+ </2 and 1 v2.

132

THEORY OF EQUATIONS.

INCOMMENSURABLE ROOTS.
506 Sturm's Theorem.If /(*), freed from equal roots, be
divided by fix), and the last divisor by the last remainder,
changing the sign of each remainder before dividing by it,
until a remainder independent of x is obtained, or else a re
mainder which cannot change its sign; then f(x), /'('')> ant^
the successive remainders constitute Sturm's functions, and
are denoted by /(a?), fx(x), /,(*), &c

fm{x).

The operation may be exhibited as follows :

/(*) =
-/*(<*)>
/ (*) = Htft () ~/a (*).
M*) = q*f*{x) -/(*),

507 Note.Any constant factor of a remainder may .be

rejected, and the quotient may be set down for the corres
ponding function.
508 An inspection of the foregoing equations shews
(1) That /.(*) cannot be zero; for, if it were, fix) and
fx (x) would have a common factor, and therefore fix) would
have equal roots, by (432).
(2) Two consecutive functions, after the first, cannot
vanish together ; for this would make/ (a?) zero.
(3) When any function, after the first, vanishes, the two
adjacent ones have contrary signs.
509 If, as x increases, fix) passes through the value zero,
Sturm's functions lose one change of sign.
For, before / (z) takes the value zero, /(x) and/^ar) have contrary signs,
and afterwards they have the same sign ; as may be shewn by making h
small, and changing its sign in the expansion off(x + h), by (426).
510 If any other of Sturm's functions vanishes, there is
neither loss nor gain in the number of changes of sign.
This will appear on inspecting the equations.
511 Result. The number of roots offix) between a and b is
equal to the difference in the number of changes of sign in
Sturm's functions, when x = a and when x = b.

INCOMMENSURABLE ROOTS.

133

512 Cor.The total number of roots of / (x) will be found

by taking a = + oo and b = oo ; the sign of each function
will then be the same as that of its first term.
When the number of functions exceeds the degree of f(x)
by unity, the two following theorems hold :
513 If the first terms in all the functions, after the first, are
positive; all the roots off{x) are real.
514 If the first terms are not all positive; then, for every
change of sign, there will be a pair of imaginary roots.
For the proof put x = + oo and oo , and examine the number of
changes of sign in each case, applying Descartes' rule.
(416).
515 If <f> () has no factor in common with / (x), and if <p (x)
and/"(#) take the same sign when /(a;) = 0; then the rest of
Sturm's functions may be found from f(x) and <p (x), instead
oif(x). For the reasoning in (509) and (510) will apply to
the new functions.
516 If Sturm's functions be formed without first removing
equal roots from f(x), the theorem will still give the number
of distinct roots, without repetitions, between assigned limits.
For if/ (x) and /,
be divided by their highest common factor (see 444),
and if the quotients be used instead of/ (as) and /, (as) to form Sturm's func
tions ; then, by (515), the theorem will apply to the new set of functions,
which will differ only from those formed from f(x) and f1 (x) by the absence
of the same factor in every term of the series.
517 Example. To find the position of the roots of the equation
x*-4^+x, + Gx + 2 = 0.
Sturm's functions, formed according to / (as) = x* 4a* + x'+ Gx+ 2
the rule given above, are here calculated.
2x'-6x'+ x+ 3
/.() =
The first terms of the functions are all
bx'-lOx- 7
positive ; therefore there is no imaginary /.(*) =
x- 1
root.
12
/*(*) =
The changes of
sign in the func
1
2
3
X = -2 -1 0
4
tions, as x passes
through integral
/(*) = + + + + + + +
values, are exhi
/l(*) = + + + +
bited in the adjoin

/.(*) = + + + +
ing table. There
= + + + +
are two changes of
sign lost while x
=
+ + + + + + +
/*(*)
passes from 1 to
0, and two more
No. of changes ") 4
2
2
2
4
0
0
lost while x passes
of sign
)
frutn 2 to 3. Thero

134

THEORY OF EQUATIONS.

are therefore two roots lying between 0 and 1 ; and two roots also between
2 and 3.
These roots are all incommensurable, by (503).

518 Fourier's Theorem.Fourier's functions are the fol

lowing quantities /(*), f{x), f'{x)
/(*).
519 Properties of Fourier's functions. As x increases,
Fourier's functions lose one change of sign for each root of
the equation f(x) = 0, through which x passes, and r changes
of sign for r repeated roots.
520 If any of the other functions vanish, an even number
of changes of sign is lost.
521 Results. The number of real roots of f(x) between a
and /3 cannot be more than the difference between the number
of changes of sign in Fourier's functions when x = a, and the
number of changes when x = /3.
522 When that difference is odd, the number of intermediate
roots is odd, and therefore one at least.
523 When the same difference is even, the number of inter
mediate roots is either even or zero.
524 Descartes' rule of signs follows from the above for the
signs of Fourier's functions, when x = 0 are the signs of the
terms in/ (x) ; and when # = oo, Fourier's functions are all
positive.
525

Lagrange's method of approximating to the incom

mensurable roots of an equation.

Let a be the greatest integer less than an incommen

surable root of /(#). Diminish the roots of f{x) by a. Take
the reciprocal of the resulting equation. Let b be the greatest
integer less than a positive root of this equation. Diminish
the roots of this equation by b, and proceed as before.
526 Let a, b, c, &c. be the quantities thus determined ; then,
an approximation to the incommensurable root of f(x) will be
the continued fraction

x= a+7

b+ c +

INCOMMENSURABLE ROOTS.

135

527 Newtori's method ofapproximation.If cx be a quantity

a little less than one of the roots of the equation f(x) = 0, so
that
+ h) = 0; then cx is a first approximation to the
value of the root. Also because
f(Cl+h) =f(c1)+hf'(cl) + ^f"(c1) + &c

(426),

and h is but small, a second approximation to the root will be

In the same way a third approximation may be obtained from

c2, and so on.
528 Fourier's limitation of Newtoris method.To ensure
that cu c2, cs, &c. shall successively increase up to the value
ct + h without passing beyond it, it is necessary for all values
of x between cx and c^h.
(i.) Thatf(x) andf'(x) should have contrary signs.
(ii.) That f{x) andf"(x) should have the same sign.

Fio. 1.

Fio. 2.

A proof may be obtained from the figure. Draw the curve

y =f(x). Let OX be a root of the equation, and ON = c^i
draw the successive ordinates and tangents NP, PQ, QB, &c.
Then OQ = c2, OS = cs, and so on.
Fig. (2) represents c2 > OX, and the subsequent approxi
mations decreasing towards the root.
530 Newtoris Rule for Limits of the Roots.Let the co
efficients of f(x) be respectively divided by the Binomial
coefficients, and let a0, a1} a2 ... an be the quotients, so that
/() = Oosf+nOiX*-1 + n.^l~l) aixn-i-'r ,.,+nan_lx+an.

136

THEORY OF EQUATIONS.

Let Au A2, A3 . . . An be formed by the law Ar = ar nr_1ar+1.

Write the first series of quantities over the second, in the fol
lowing manner :
a0,
a-j,
at,
(7S
(in-i>
rt>
A0, Au Ai} As
An_lt
An.
Whenever two adjacent terms in the first series have the
same sign, and the two corresponding terms below them in
the second series also the same sign; let this be called a
double permanence. When two adjacent terms above have
different signs, and the two below the same sign, let this be
known as a variation-permanence.
531 Rule. The number of double permanences in the asso
ciated series is a superior limit to the number of negative roots
off(x).
Hie number of variation-permanences is a superior limit to
the number of positive roots.
The number of imaginary roots cannot be less than the
number of variations of sign in the second seriesr

532 Sylvester's Theorem. Let f(x + \) be expanded by

(426) in powers of x, and let the two series be formed as in
Newton's Rule (530).
Let P (A) denote the number of double permanences.
Then P (A) - P (jt) is either equal to the number of roots
of /(), or surpasses that number by an even integer.
Note.The first series may be multiplied by \ n_, and will
then stand thus,
faW,
L2/"-2W, 18/(X)...1/(X).
The second series may be reduced to
^iWr
where

eB_2(A)...e?(A),

Gr (A) = {/- (A)}2 - =*1 /-1 (A) /- (A).

it '"' /

533 Horner's Method.To find the numerical values of the

roots of an equation. Take, for example, the equation
x* -4a +x*+6x + 2 = 0,
and find limits of the roots by Sturm's Method or otherwise.

INCOMMENSURABLE BOOTS.

137

It has been shewn in (517) that this equation has two

incommensurable roots between 2 and 3. The process of
calculating the least of these roots is here exhibited.
-4
o
^2
2
0
2
2
2
D, 40
4
44
4
48
4
52
4
Z), 560
1
561
1
562
1
563
1
D, 5640
4
5644
4
5648
4
5652
4
r>, 5,656

o,

ot

ct

ct
ot

+1
-4
-3
0
-3
4
100
170
276
192
408
208
07600
501
69101
502
08723
503
61*28000
22570
0951170
22592
0973708
22008
0990370
11
09974
11
09985
11
09990
7

B,

Bt

B<

Bt

Bt

B,

+6
-6
0
-6
-0000
1104
-4896
1872
-3024000
68101
-2955839
08723
-2887110000
27804704
-2859311296
27895072
-2831410224
139948
283001074
139970
-282801704
700
- 28285470
700
-28284770
21
-2828450
21
-2828435

A,
A,
A3
A,
A,
A,
A,

+ 2(2-414213
0
20000
-19584
4100000
-2955839
12041610000
-11437245184
604304816
-506003348
38361408
-28285470
10075998
- 8485308
1590030

282843)1590630 (562372
1414215
28284) 176415
169706
2828) 6709
5057
282) 1052
848
28) 204
197
2) 7
5
2
Root = 2-414213502372.

Method.Diminish the roots by 2 in the manner of (427).

The resulting coefficients are indicated by Av 2? C D,.
By Newton's rale (527), jrjrr i that is, ^ is an approximation to
the remaining part of the root. This gives "3 for the next figure ; '4 will bo
found to be the correct one. The highest figure must bo taken which will
not change tho sign of A.
Diminish the roots by -4. This is accomplished most easily by affixing
ciphers to Av Bv C Dlt in the manner shewn, and then employing 4 instead
of -4.
Having obtained At, and observing that its sign is +, retrace the steps,
T

138

THEORY OF EQUATIONS.

trying 5 instead of 4. This gives At with a minus sign, thereby proving the
existence of a root between 24 and 2'5. The new coefficients are At, Bit C2, Dr
-z * gives 1 for the next figure of the root.
Affix ciphers as before, and diminish the roots by 1, distinguishing the
new coefficients as Ait Bt, C\, Dr
Note that at every stage of the work A and B must preserve their signs
unchanged. If a change of sign takes place it shews that too large a figure
has been tried.
To abridge the calculation proceed thus :After a certain number of
figures of the root have been obtained (in this example four), instead of
adding ciphers cut ofl' one digit from Bv two from C\, and three from D4.
This amounts to the same thing as adding the ciphers, and then dividing
each number by 10000.
Continue the work with the numbers so reduced, and cut off digits in like
manner at each stage until the D and C columns have disappeared.
A7 and B7 now alone remain, and six additional figures of the root are
determined correctly by the division of A7 by B7.
To find the other root which lies between 2 and 3, we proceed as follows :
After diminishing the roots by 2, try G for the next figure. This gives A,
negative; 7 does the same, but 8 makes A2 positive. That is to say, f(2'7)
is negative, and/ (28) positive. Therefore a root exists between 2*7 and
2'8, and its value may be approximated to, in the manner shewn.
Throughout this last calculation A will preserve the negative sign.
Observe also that the trial number for the next figure of the root given at
each stage of the process by the formula , {, will in this case be always
too great, as in the former case it was always too small.

SYMMETRICAL FUNCTIONS OF THE ROOTS OF

AN EQUATION.

Notation.Let a, b, c ... be the roots of the equation

f{x) =. 0.
Let sM denote am -f- b'" + ..., the sum of the mth powers of
the roots.
Let s,tP denote a'"bp + bmap + amep+ ... through all the
permutations of the roots, two at a time.
Similarly let sm>p denote ambpcq-\-anbpdq -\- ... , taking all
the permutations of the roots three at a time ; and so on.

SYMMETRICAL FUNCTIONS- OF THE ROOTS.

534

139

SUMS OF THE POWERS OF THE BOOTS.

where m is less than n, the degree of / (x).

Obtained by expanding by division each term in the value off(x) given
at (432), arranging the whole in powers of a, and equating coefficients iu the
result and in the value off (x), found by differentiation as iu (424).
535

K m be greater than n, the formula will be

Sm~\~Pl Sm-l~\~Pt *-2"t" ~\~PnSm-n = 0.

Obtained by multiplying /(as) = 0 by as"", substituting for a: the roots

o, h, c, &c. in succession, and adding the results.
By these formula; 3, ss, &c. may be calculated successively.
536 To find the sura of the negative powers of the roots, put
m equal to n 1, n 2, n 3, &c. successively in (535), in
order to obtain s_u s_2, s_s, &c.

537 To calculate sr independently.

Rule : sr = r X coefficient of x~r in the expansion of
log tS^l in descending powers of x.
Proved by taking f (x) = (x a) (x- b) (as c) ... , dividing by x", and
expanding the logarithm of the right side of the equation by (150).

538

SYMMETRICAL FUNCTIONS WHICH ARE

NOT POWERS OF THE ROOTS.

These are expressed in terms of the sums of powers of

the roots as under, and thence, by (534), iu terms of the roots
explicitly,
^m.p
&m&p
*+p >
539

$m,p,q ~

8m+p8q

Sm+){Sp

SJ)+liSm-\-2sm + p + q .

The last equation may be proved by multiplying sm p by

e?; and expansions of other symmetrical functions may be
obtained in a similar way.

540 If <p (x) be a rational integral function of x, then the

symmetrical function of the roots of / (x), denoted by

140

THEORY OF EQUATIONS.

<j> (a) +<p (b) + f (c) + &c, is equal to the coefficient of .e"-1 in
the remainder obtained by dividing <p (x) f (x) by /(as).
Proved by multiplying the equation (432) by

\ and by theorem (401).

541 To find the equation whose roots are the squares of

the differences of the roots of a given equation.
Let F (x) be the given equation , and 8r the sum of the r01
powers of its roots. Let f(x) and sr have the same meaning
with regard to the required equation.
The coefficients of the required equation can be calculated
from those of the given one as follows :
The coefficients of each equation may be connected with the
sums of the powers of its roots by (534) ; and the sums of the
powers of the roots of the two equations are connected by the
formula
542

2sr = nS.ir-2rS1S,r_1+ 2|,(^r~1) S2S2r_2- ... +nS4r.

Rule.2sr is equal to the formal expansion of (SSyr by

the Binomial Theorem, with the first and last terms each mul
tiplied by n, and the indices all changed to suffixes. As the
equi-distant terms are equal we can divide by 2, and take half
the scries.
Demonstbation.Let a, b, c ... be the roots of F(x).
Let
<j> (x) = (x -a)-'+(x -&)*+

(i.)

Expand each term on the right by the Bin. Theor., and add, substituting
Su Sj, &o. In the result change x into a, b, c ... successively, and add the n
equations to obtain the formula, observing that, by (i.),
f (a) +(&) + ... = 2sr.
If n be the degree of F (x), then \n (n 1) is the degree
Gt f{x).
'
By (96).
543

The last term of the equation / (,c) = 0 is equal to

nF(a)F((3)F(y)...
where a, /3, -y, ... are the roots of F'{x). Proved by shewing
that
F'(a)F'(b) ... = n'F(a) F(p) ...
544 If F(x) has negative or imaginary roots, f(x) must
have imaginary roots.

SYMMETRICAL FUNCTIONS OF THE ROOTS.

545

141

The sum of the mth powers of the roots of the quad

ratic equation

a? px-\-q = 0.

(_1)rm(m-r-l) ... (m-2r+l) - r+&0>

Ir
By (537) expanding the logarithm by (156).

546 The sum of the mth powers of the roots of xH 1 = 0

is n if m be a multiple of n, and zero if it be not.
By (537) ; expanding the logarithm by (1 5G).
547

If

<f>(x) = a0-\-a1x+atxt + &c

(i.),

then the sum of the selected terms

amxm + am+n xm+n + am+2n xm+* + &c.
mil be *= {an-m<f>(auc)+Ptt-m<t>(fa)+yn-m<l>(yx)+&c.}
where a, /3, y, &c. are the th roots of unity.
For proof, multiply (i.) by a"", and change * into ax; so with /3, y, &c,
and add the resulting equations.

548 To approximate to the root of an equation by means of

the sums of the powers of the roots.
By taking m large enough, the fraction -ii will approximate to the value of the numerically greatest root, unless
there be a modulus of imaginary roots greater than any real
root, in which case the fraction has no limiting value.
2

549

Similarly the fraction

'" '"+2
Sm-lSvi+l

^ approximates, as m
*m

increases, to the greatest j^roduct of any pair of roots, real or

imaginary ; excepting in the case in which the product of the
pair of imaginary roots, though less than the product of the
two real roots, is greater than the square of the least of them,
for then the fraction has no limiting1 value.

THEORY OF EQUATIONS.

142

550

Similarly tlte fraction

fs^fg+j approximates,

as i/i increases, to the sum of the two numerically greatest

roots, or to the sum of the two imaginary roots with the
greatest modulus.

EXPANSION OF AN IMPLICIT FUNCTION OF *.

Let

y(Ar"+)+/(5llr6+)+...+/(Str'+) = 0

(1)

be an equation arranged in descending powers of y, the co

efficients being functions of x, the highest powers only of x
in each coefficient being written.
It is required to obtain y in a series of descending powers
of x.
First form the fractions
ab
a ^3'

ac
a y

ft d
a 8

as
aa

^\

Let a = t be the greatest of these algebraically, or

an
if several are equal and greater than the rest, let it be the
last of such. Then, with the letters corresponding to these
equal and greatest fractions, form the equation
Av?+

(3).

Each value of u in this equation corresponds to a value of y,

commencing with ux1.
Next select the greatest of the fractions
klr
KA

km
K fl

ks
K O"

hn
= t' be the last of the greatest ones.
KV
the corresponding equation Ku"-\- ... +iW = 0
Let

Form
(5).

Then each value of u in this equation gives a corresponding

value of y, commencing with ux1'.

EXPANSION OF AN IMPLICIT FUNCTION.

143

Proceed in tliis way until the last fraction of the series (2)
is reached.
To obtain the second term in the expansion of y, put
y =

in (1)

(6),

employing the different values of u, and again of t' and u, t"

and v, &c. in succession ; and in each case this substitution
will produce an equation in y and x similar to the original
equation in y.
Eepeat the foregoing process with the new equation in y,
observing the following additional rule :
When all the values of t, t', t", Sfc. have been obtained, the
negative ones only must be employed informing the equations
inn.
(7).
552 To obtain y in a series of ascending powers of x.
Arrange equation (1) so that a, |3, y, &c. may be in as
cending order of magnitude, and a, b, c, &c. the lotvest powers
of x in the respective coefficients.
Select t, the greatest of the fractions in (2), and proceed
exactly as before, with the one exception of substituting the
word positive for negative in (7).
553

Example.Take the equation

(x3 + zt) + (3x'-5zi)y + (-4<e + 7x1l + z',)yi-if = 0.

It is required to expand y in ascending powers of x.

The fractions (2) are

q~2' |E| ' or 1 * and *'

The first two being equal and greatest, we have 2 = 1,

The fractions (4) reduce to \^ = = (.
Equation (3) is
1 + 3n -W = 0,
which gives
u = l and \, with f = l.
Equation (5) is
4ws it5 = 0,
and from this
u = 0 and 4*, with f = \.
We have now to substitute for y, according to (G), either
2(1 + 2/0, f(-i+y,),
or xl(-4i+y1).
Put y = x (1 + ?/,), the first of these values, in the original equation, and
arrange in ascending powers of y, thus
-4z+ (-5*'+) y, +
y\ - lOxY, - 5x'y\ - x*y[ = 0.
The lowest power only of x in each coefficient is hero written.

144

THEORY OF EQUATIONS.

The fractions (2) now become

4-3
4-3 _ 4-5
4-5 _ 4-5
0-1'
0-2'
0-3'
0-4'
0-5'
or
1, |, |, i,
From these t 1, and equation (3) becomes
4 5u = 0; .'. w = - \
Hence one of the values of yl is, as in (G), yx = x ( + 2/i)Therefore
y = x {1 + x (-f+y,)} = x-$x* + ...
Thus the first two terms of one of the expansions have been obtained.

DETERMINANTS.
554

Definitions.The determinant

is equivalent

to axb2 2 6I5 and is called a determinant of the second order.

A determinant of the third order is
ax a2 a, | = Oj (b2c3 b3ct) + a^b^ 6^) + s(V -^O6i 62 *s
<?1 C<z t*3
Another notation is 2 a^Cg, or simply (a^Cg).
The letters are named constituents, and the terms are
called elements.
The determinant is composed of all the
elements obtained by permutations of the suffixes 1, 2, 3.
The coefficients of the constituents are determinants of
the next lower order, and are termed minors of the original
determinant. Thus, the first determinant above is the minor
of c3 in the second determinant. It is denoted by C3. So the
minor of at is denoted by Au and so on.
555

A determinant of the nth order may be written in either

of the forms below

b, b2

6.

a.2l an ..
or

^1 l2 lr ' In

, a

In the latter, or double suffix notation, the first suffix indicates

the row, and the second the column. The former notation
will be adopted in these pages.

DETERMINANTS.

145

A Composite determinant is one in which the

ax a2 a3
number of columns exceeds the number of rows,
bv b2 bs
and it is written as in the annexed example.
Its value is the sum of all the determinants obtained by taking
a number of rows in every possible way.
A Simple determinant has single terms for its constituents.
A Compound determinant has more than one term in some
or all of its constituents. See (570) for an example.
For the definitions of Symmetrical, Reciprocal, Partial,
and Complementary determinants; see (574), (575), and (576).
General Theory.
556 The number of constituents is n*.
The number of elements in the complete determinant is \n_.
557 The first or leading element is a^c^ ... ln. Any
element may be derived from the first by permutation of the
suffixes.
The sign of an element is + or according as it has
been obtained from the diagonal element by an even or odd
number of permutations of the suffixes.
Hence the following rule for determining tho sign of an
element.
Rttle. Take the suffixes in order, and put them bach to
their places in the first element. Let m be the whole number
ofplaces passed over ; then ( 1) will give the sign required.
Ex. To find the sign of the element a46jC6e7,e, of
(a,6,c,d4e6).
< h cb
Move the suffix 1, three places
1 4 3

2, three places
12 4

3, one place
12 3
In all, seven places; therefore ( l)7 = 1 gives the sign

the determinant
<*i ei
5 2
3 5
4 5
required.

558 If two suffixes in any element be transposed, the sign

of the element is changed.
Half of the elements are plus, and half are minus.
559 The elements are not altered by changing the rows into
columns.
If two rows or columns are transposed, the sign of the
u

146

THEORY OF EQUATIONS.

determinant is changed. Because each element changes its

sign.
If two rows or columns are identical, the determinant
vanishes.
560 If all the constituents but one in a row or column
vanish, the determinant becomes the product of that con
stituent and a determinant of the next lower order.
561 A cyclical interchange is effected by to 1 successive
transpositions of adjacent rows or columns, until the top row
has been brought to the bottom, or the left column to the
right side. Hence
A cyclical interchange changes the sign of a determinant
of an even order only.
The rth row may be brought to the top by r 1 cyclical
interchanges.
562 If each constituent in a row or column be multiplied
by the same factor, the determinant becomes multiplied by it.
If each constituent of a row or column is the sum of m
terms, the compound determinant becomes the sum of m,
simple determinants of the same order.
Also, if every constituent of the determinant consists of
m terms, the compound determinant is resolvable into the sum
of 7H2 simple determinants.
563 To express the minor of the rth row and ktb column as
a determinant of the to 1th order.
Put all the constituents in the rth row and kth column equal
to 0, and then make r 1 cyclical interchanges in the rows
and h 1 in the columns, and multiply by ( 1)("-*><-*).
r. . _ f_iyr-l**-l)*n-l)^

564 To express a determinant as a deter

minant of a higher order.
Continue the diagonal with constituents
of " ones," and fill up with zeros on one side,
and with any quantities whatever (a, /3, y, &c.)
on the other.

10 0 0 0
a 1 0 0 0
/8 e a h g

ih bf
n s f c

DETERMINANTS.

U7

565 The sum of the products of each constituent of a

column by the corresponding minor in another given column
is zero. And the same is true if we read ' row' instead of
'column.' Thus, referring to the determinant in (555),
Taking the pth and qtb columns,
Taking the a and c rows,
apAq+bpBq+ ...+lpLq = 0.

o1C1+2C4+ ...+C. = 0.

For in each case we have a determinant with two columns identical.

566 In any row or column the sum of the products of each
constituent by its minor is the determinant itself. That is,
Taking the pth column,
Or taking the c row,
apAp+bpBp+...+lpLp = A.
567

...+cnCn = A.

The last equation may be expressed by "2cpGp = A.

Also, if (apcq) express the determinant

then

2(apcq) will represent the sum of all the determinants of the

second order which can be formed by taking any two columns
out of the a and c rows. The minor of (ap, cq) may be written
(Ap, Gq), and signifies the determinant obtained by suppress
ing the two rows and two columns of ap and cq.
Thus
A = 2 (ap, cq) (Ap, Cq). And a similar notation when three or
more rows and columns are selected.
r

568

Analysis of a determinant.

Rule. To resolve into its elements a determinant of the

n** order. Express it as the sum of n determinants of the
(n 1)'* order by (560), and repeat the process with each oj
the new determinants.

<h
6,
"i
</,

Example :
+ a, bt bg bt - at bt 6, b,
i <h " = *i 6, b, bt "i h h
b, bt bt
rt et ct
c, c, c.
C, Ct Cj
<*< Cj C,
C| <* <"
d2 d, dt
d, d, d,
d, dt d,
dt ds d,
</, d\ dt
Again,

bt b, 6,
+ 6, cs ci 1 + &a
d,d,
c, c, c,
d,dj
^d,
<h d, d,
and so on. In the first series the determinants have alternately plus and
minus signs, by the rule for cyclical interchanges (5G1), the order being even.

148

THEORY OF EQUATIONS.

569

Synthesis of a determinant.

The process is facilitated by making use of two evident

rules. Those constituents which belong to the row and
column of a given constituent a, will be designated " o's con
stituents." Also, two pairs of constituents such as ap, cq and
aq, cp, forming the corners of a rectangle, will be said to be
" conjugate" to each other.
Rule I.No constituent will be found in the same term
with one of its own constituents.
Rule II.The conjugates of any two constituents a and b
toill be common to a's and b's constituents.
Ex.To write the following terms in the form of a determinant :
a bed + bfijl +fh* + 1 edf+ cghp + 1 ahr + elpr
fhprablractflfhg bdf^efhl cedp.
The determinant will be of the fourth order ; and since every term must
contain four constituents, the constituent 1 is supplied to make up the
number in some of the terms. Select any term, as abed, for the leading
diagonal.
Now apply Rule I.,
a is not found with e,/,/, g,p, 0...(1). c is not found with/,/, l,r, 1,0... (3).
b is not found with e, h, 7i,j,l,0 ..(2). rfis not found with g, h,h,l,r,0...(4).
Each constituent has 2 (n 1), that is, 6 constituents belonging to it, since
n =?4. Assuming, therefore, that the above letters are the constituents of
a, b, c, and d, and that there are no more, we supply a sixth zero constituent
in each case.
Now apply Rule II.The constituents common
to a and b are e, p ; to a and c-/, /;
to 6 and c1, 0 ;
to a and dy, 0 ;
to 6 and dh, It, 0 ; to c and d4, r, 0.
The determinant may now be formed. The diagonal
a '.; J 9
being abed ; place e, p, the conjugates of a and b, either as
V b 1 h
in the diagram or transposed.
f 0 c r
0 k I d
Then / and/, the conjugates of a and c, may be written.
1 and 0, the conjugates of b and c, must be placed as indicated, because
1 is one of p'& constituents, since it is not found in any term witli p, and.
must therefore be in the second row.
Similarly the places of g and 0, and of I and r, are assigned.
In the case of b and d we have h, h, 0 from which to choose the two
conjugates, but we see that 0 is not one of them because that would assign
two zero constituents to b, whereas b has but one, which is already placed.
By similar reasoning the ambiguity in selecting the conjugates I, r is
removed.
The foregoing method is rigid in the caso of a complete determinant

149

DETERMINANTS.

having different constituents. It becomes uncertain when the zero con

stituents increase in number, and when several constituents are identical.
Bnt even then, in the majority of cases, it will soon afford a clue to tho
required arrangement.

570

PRODUCT OF TWO DETERMINANTS OF

THE nth ORDER.
(P)
! a2 ... an
b1 b* ... bn

h h

(Q)

A A- A
X.X,.

The values of Au Bt ... Lx in

tuo first column of 8 are an
nexed. For the second column
write b's in the place of a's.
For the third column write c's,
and so on.

OS)
A1 At. ..An
By Bt. Bn

u L*.

Ln

Ai = ala1+aia2+...+anan

B1 = aipi + aiPi+...+ anfi

Ly = a1X1 + a2X2+...+aXn

For proof substitute the values of A 2? &c. in the determinant 8, and

theu resolve S into tho sum of a number of determinants by (562), and note
the determinants which vanish through having identical columns.

Rule. To form the determinant 8, which is the product of

two determinants P and Q. First connect by plus signs the
constituents in the roivs of both the determinants P and Q.
Nuw place the first row of P upon each row of Q in turn,
and let each two constituents as they touch become products.
This is the first column of S.
Perform the same operation upon Q vrith the second row of
P to obtain the second column of 8 ; and again with the third
row of P to obtain the third column of 8, and so on.
571 If the number of columns, both in P and Q, be n, and
the number of rows r, and if n be > r, then the determinant
S, found in the same way from P and Q, is equal to the sum
of the 0 (n, r) products of pnirs of determinants obtained by
taking any r columns out of P, and the corresponding r
columns out of Q.
But if n be <r the determinant 8 vanishes.
For in that case, in every one of the component determinants, there will
be two columns identical.

150

THEORY OF EQUATIONS.

572 The product of the determinants P and Q may be

formed in four ways by changing the rows into columns in
either or both P and Q.
573 Let the following system of n equations in a^ ... xn
be transformed by substituting the accompanying values of
the variables,
al*\+agxi+ ...+anxn = 0,
arxs= a^-f a2+... + a

Mi+M*+ ...+&#. = 0,

xt = A+ A&+ ... + A,fn

llxl+lixi+...+lna!1t = 0,

a?= X1+Xi!,+ ...+X,.

The eliminant of the resulting equations in t 2 . . . 5 is

the determinant S in (570), and is therefore equal to the
product of the determinants P and Q. The determinant Q is
then termed the modulus of transformation.
574 A Symmetrical determinant is symmetrical about the
leading diagonal. If the P's form the rth row, and the K's
the kth row ; then Bk = Kr throughout a symmetrical deter
minant.
The square of a determinant is a symmetrical determinant.
575 A Reciprocal determinant has for its constituents the
first minors of the original determinant, and is equal to its
n 1th power; that is,
A1 ... An

Jj

576

Jj

a^ ...
a
I . .. an

hI, .

IL

n_1

Proof.Multiply both sides

of the equation by the original
determinant (oi)b). Tbe con
stituents on the left side all
vanish excepting the diagonal
of As.

Partial and Complementary determinants.

If r rows and the same number of columns be selected

from a determinant, and if the rows be brought to the top,
and the columns to the left side, without changing their order,
then the elements common to the selected rows and columns
form a Partial determinant of the order r, and the elements
not found in any of those rows and columns form the Com
plementary determinant, its order being nr.

DETEBMINANTS.

151

Ex. Let the selected rows from the determinant (a^c,^^) be the
second, third, and fifth ; and the selected columns be the third, fourth, and
fifth. The original and the transformed determinants will be
and
\
h j &i 6.
6.
8 c c5 I cl
C] c% c$
Cg
e 6i e i e,
tij d, d, dt di
i i
e, e, e, e4 ej
<7, i4 4
The partial determinant of the third order is (bicte6'), and its comple
mentary of the second order is (a, a",).
The complete altered determinant is plus or minus, according as the
permutations of the rows and columns are of the same or of different class.
In the example they are of the same class, for there have been four trans
positions of rows, and six of columns. Thus ( l)10 = + 1 gives the sign
of the altered determinant.
577 Theobbm.A partial reciprocal determinant of the rth
order is equal to the product of the r 1th power of the
original determinant, and the complementary of its corres
ponding partial determinant.
Take the last determinant for an example. Here =5, r=3 ; and by the
theorem,
Bt Bt Bs = A' a, o,
where B, 0, E are the
respective minors.
Ot G, Os
d, d,
Et Et B,
Proof.Raise the Partial Reciprocal to the original order five without
altering its value, by (564) ; and multiply it by A, with the rows and
columns changed to correspond as in Ex. (576) ; thus, by (570), we have
B,
B, B,
bt 6. &, 6, = A 0 0 6, 6, = A8 a, n,
0 A 0 c, c,
C,
a. o>
o*
"i
0 0 A e, ea
Et
. e* e6
0 0 0 11,0,
0 0 0 i 6
a,
<h
<h
0
0 0 d\ t?,
0 0 0 0 i
*t
<k
578 The product of the differences between every pair of n
quantities a^,
... an,
(i aa)(i

*) ... (i a)

X (as aj)(a2 a4) ... (a2 a)

X(asai)...(asa) } =

1
8 a.
a!

a!

an-\ an-l a~l a-l

1
2
3 '
*
Pboof.The determinant vanishes when any two of the quantities are
X (n-l~ )

THEOEY OF EQUATIONS.

152

equal. Therefore it is divisible by each of the factors on the left ; therefore

by their product. And the quotient is seen to be unity, for both sides of the
equation are of the same degree ; viz., \n(n 1).
1
579 The product of the squares of the
=
differences of the same n quantities
}
Proof.Square the determinant in (578), and
write *T for the sum of the rth powers of the roots.

8n-\ *

Stn-i

580 With the same meaning for slf s2..., the same deter
minant taken of an order r, less than n, is equal to the sum
of the products of the squares of the differences of r of the n
quantities taken in every possible way ; that is, in C (n, r)
ways.
Ex.:

= (a, - a,)' + K - a,)' + &c. = 2 (a, - a,)',

2 (<*.-.)' C^-**)' (<*-)'

$0 ^1 ,<?3
*! C| Ss

The next determinant in order

= 2 (a, -a,)1 (ai-a,)' (oi-<0' (aj-a,)' (a, a4)s (as-o4)'.
And so on until the equation (579) is reached.
Proved by substituting the values of *, ... Ac., and resolving the deter
minant into its partial determinants by (571).

581

The quotient of
a()xm+a1<vm-1+ ...+arxm-r+ ...
60 of1 + &x
1 + . . . + br xn ~ r + . . .

is given by the formula

q<)X*-* + q1jpm~n~1 + ... + qrxn
+ ....
where
9r

60

a.

K
bt

60
h

0
60

1
a.,

-r+i

br 6r_! 6r_2 ... 6xar

Proved by Induction.

ELIMINATION.

153

ELIMINATION.

582

Solution of n linear equations in n variables.

The equations and the values of the variables are arranged

below :
a1<r1+a2a?2+ ...+dna?n = ,

xtA = A1^l-\-B1it+...+L1(n

blxl+b2a-2+...+bnx = f

x2A = Ati1+Bait+ ...+L2,

lx a.\+l2 cr2+... + / xn =

^BA =

where A is the determinant annexed, and Au Blt

&c. are its first minors.

la, ... a
h ... I

To find the value of one of the unknowns xr.

Rule.Multiply the equations respectively by the minors
of the rth column, and add the results. xr will be equal to the
fraction whose numerator is the determinant A, with its r**
column replaced by
2 ... , and whose denominator is A
itself.
583 If 15 2
and A all vanish, then xx, x2 ... xn are in
the ratios of the minors of any row of the determinant A.
For example, in the ratios

Cx : C2 : Cs : ... : C%.

The eliminant of the given equations is now A = 0.

584

Orthogonal Transformation.

If the two sets of variables in the n equations (582) be

connected by the relation
* + *; + ... + * = ! + + ... + :
a),
then the changing from one set of variables to the other, by
substituting the values of the 's in terms of the x's in any
function of the former, or vice versa, is called orthogonal
transformation.
When equation (1) is satisfied, two results follow.
I. The determinant A = l.
x

154

THEORY OF EQUATIONS.

II. Each of the constituents of A is equal to the corre

sponding minor, or else to minus that minor according as A is
positive or negative.
Proof.Substitute the values of
, ... , in terms of z
equation (1), and equate coefficients of the squares and products
variables. We get the n' equations
o1 + 6l + - 1
a,a, + 6,6, + = 0
a,a, + 6,6, +
i 1 i
o,a, + 6,6, + = 0
OjO, + 6,5, +
a\ + b\ + = 1
a,o, + 6,6, + = 0
a,a, + 6,6, + = 0
a* + 6* +
a,a + 6,6,, + = 0 J
Also A =

0,6, ... Z,
0,6, ... 2,
0,6, ... Z,
a,6H ... l

o^a, + 6,6 + = 0 J

z, ... xn in
of the new
=0'
=0
= 1

asan+ bsbn + = 0 .

Form the square of the determinant A by the

rule (570), and these equations show that the
product is a determinant in which the only con
stituents that do not vanish constitute a diagonal
of ' ones.' Therefore
A' = 1 and A = 1.

Again, solving the first set of equations

for o, (writing oj as a,o &c.), the second
set for a,, the third for a,, and so on, we
have, by (582), the results annexed ; which
proves the second proposition.

fa,A = 4, +Afi+A,0 + = At
a,A = Afi+A, +A,0+=At
a,A = Afi+ Afi+A, +=At
&c.
&c.

585 Theorem.The n 2th power of a determinant of the

nth order multiplied by any constituent is equal to the corre
sponding minor of the reciprocal determinant.
Proof.Let p be the reciprocal determinant of A, and ft the minor of Br
in p. Write the transformed equations (582) for the x'a in terms of the 's,
and solve them for ,. Then equate the coefficient of xr in the result with ita
coefficient in the original valne of
Thuspi,= A (/3,<e1+...+/3rav+...), and {, = 6>,+ ... + 6,*,+ ... ;
.-. Aft. = P6, = A-'6r by (575) j
.-. ft = A-1 br
586

To eliminate * from the two equations

orm + bxn-1 + cxm-2 + ... = 0

(1),

rfar+b'a*-1+c'a*-,+... = 0

(2).

If it is desired that the equation should be homogeneous

in x and y ; put instead of x, and clear of fractions.
following methods will still be applicable.

The

I. Bezout's Method.Suppose m > n.

Rule.Bring the equations to the same degree by multi
plying (2) by x'*. Then multiply (1) by a', and (2) by a, and
subtract.
Again, multiply (1) by a'x-\-b', and (2) by (ax + b), and
subtract.
Again, multiply (1) by a'af + b'x + c', and (2) by (ax*+bx+c),
and subtract, and so on until n equations have been obtained.
Each will be of the degree m 1.
Write under these the m n equations obtained by multiflying (2) successively by x. The eliminant of the m equations
is the result required.
Ex.Let the equations be ( axi+bx*+cz* + dzt + ez+f = 0,
iax' + b'z*+cz +d?

=0.

The five equations obtained by the method, and their eliminant, by (583),
are, writing capital letters for the functions of a, b, c, d, e, f,
A^+B^+C^ + D^c + E, = 0
A^+B^ + C^+D^+Es = 0
a'z,+b'xt+c'zt + d'z
=0

and

ax*+b'x* + c'x+d' - 0 J

A1
A,
A,
a
0

5,
B,
B,
b'
a

0,
C,
Gt
c
V

D,
B,
Dt
d'
c

?, = 0.
E,
Et
0
d'

Should the equations be of the same degree, the eliminant will be a sym
metrical determinant.

587

II. Sylvester's Dialytic Method.

Rule.Multiply equation (1) successively by x, n \ times;

and equation (2) m 1 times ; and eliminate x from the m + n
resulting equations.
Ex.To eliminate z from

a3t + bxt+cx+d = 0 I
px+qx+r = 0 I

The m + n equations and their eliminant are

px*+qx+r
px* + qx1 + rz
=
pz' + qif + rx1
=
ax*+bx,+cx + d =
az'+bz' + caf+dx
=

0'
0
0
0
0

and

0
0
p
0
a

0
p
q
a
b

p
q
r
b
c

q
r
0
c
d

r = 0.
0
0
d
0

156

588

THEORY OF EQUATIONS.

HI- Method of elimination by Symmetrical Functions.

Divide the two equations in (586) respectively by the

coefficients of their first terms, thus reducing them to the
forms
f{x) =aT + p1afH~1+ ...+pm = 0,

4>(x) = xn + qlx"-1 + ... +qn = 0.

Rule.Let a, b, c ... represent the roots of f(x). Form
the equation <j> (a) $ (6) <j> (c) ... = 0. This will contain sym
metrical functions only of the roots a, b, c ... .
Express these functions in terms of plt p2 by (538), Sfc,
and the equation becomes the eliminant.
Reason of the rule.The eliminant is the condition for a common root of
the two equations. That root must make one of the factors <p (a), f (6) ...
vanish, and therefore it makes their product vanish.

589 The eliminant expressed in terms of the roots a,b,c ...

oif(x), and the roots a, /3, y ... of f (x), will be
(a-a) (a-p) (a-y) ... (&-) (6-/3) (b-y) ... &c,
being the product of all possible differences between a root of
one equation and a root of another.
590 The eliminant is a homogeneous function of the co
efficients of either equation, being of the nth degree in the
coefficients of /(), and of the mth degree in the coefficients
of <p (x).

591 The sum of the suffixes of p and q in each term of the

eliminant = mn. Also, if p, q contain z ; if p2, q2 contain z2 ;
if p%, q% contain z3, and so on, the eliminant will contain zmn.
Proved by the fact that pr is a homogeneous function of r dimensions of
the roots o, b, c ..., by (406).

592 If the two equations involve a; and y, the elimination

may be conducted with respect to x ; and y will be contained
in the coefficients pu pt . . . , qlt qs . . . .
593

Elimination by the Method of Highest Common Factor.

Let two algebraical equations in x and y be represented by

A = 0 and B = 0.

ELIMINATION.

157

It is required to eliminate x.
Arrange A and B according to descending powers of x,
and, having rejected any factor which is a function of ?/ only,
proceed to find the Highest Common Factor of A and B.
The process may be exhibited as follows :
txA = qtB +r1Rl

<h> ca cs c4 are the multipliers re

quired at each stage in order to avoid
fractional quotients ; and these must
be constants or functions of y only.
?x> ? 1i> 1* are the successive quo
tients.

ctB = q2B1 -+- r2B2

ctB2

rxBu r2B2, r3Bs, r4 are the successive remainders ; ru r2, r, r4

being functions of y only.
The process terminates as soon as a remainder is obtained
which is a function of y only ; r4 is here supposed to be such
a remainder.
Now, the simplest factors having been taken for cl} c2, c8, c4,
we see that
1 is the H. 0. F. of

and

It

11

and r2

tt

and r3

^4

11

11

C-^~ and r4
d2as

C1C2

fj = 0 and B 0
rf = 0 and R-l = 0

0)1
(2)

The values of * and y,

which
satisfy
simulta
neously
the
equations
A=0 and B=0, are those
obtained by the four pairs
of simultaneous equations
following :
The final equation in y,
which gives all admissible
values, is

tllWj _ o
d2 ds dt
If it should happen that
the remainder r4 is zero,
= 0 and Bs = 0
(4)
a4
the simultaneous equa
tions (1), (2), (3), and (4) reduce to
^ = 0 and B2 = 0
<*s

r1=0and^- = 0;

(3)

^ = 0 and = 0 ;

^ = 0and|2 = 0.
a3
itj

158

THEORY OF EQUATIONS.

594 To find infinite values of x or y which satisfy the given

equations.
Put x = . Clear of fractions, and make z = 0.
z
If the two resulting equations in y have any common
roots, such roots, together with x oo, satisfy simultaneously
the equations proposed.
Similarly we may put y = .

PLANE TRIGONOMETRY.
ANGULAR MEASUREMENT.
600 The unit of Circular measure is a Radian, and is the
angle at the centre of a circle which subtends an arc equal to
the radius. Hence
601
602

Circular measure of an angle = .

radius
Circular measure of two right angles = 3"14159 ...~w.

603 The unit of Centesimal measure is a Grade, and is the

one-hundredth part of a right angle.
604 The unit of Sexagesimal measure is a Degree, and is
the one-ninetieth part of a right angle.
To change degrees into grades, or circular measure, or
vice verm, employ one of the three equations included in

fin*
W0

D - G - 2R
90 " 100 ~ r '

where D, 6, and G are respectively the numbers of degrees,

grades, and radians in the angle considered.

TRIGONOMETRICAL RATIOS.
606 Let OA be fixed, and let the
revolving line OP describe a circle
round 0. Draw PN always perpen
dicular to AA'. Then, in all posi
tions of OP,
PN
prp = the sine of the angle AOP,
ON
pr-p = the cosine of the angle A OP,
PN
= the tangent of the angle AOP.
ON

PLANE TRIGONOMETRY.

160

607

If P be above the line AA', sin AOP is positive.

If P be below the line AA', sin

608

If P lies to the right of

If P lies to the ie//

is negative.

cos AOP is positive.

of BP/, cos AOP is negative.

609 Note, that by the angle AOP is meant the angle through which
OP has revolved from 0.1, it< initial position ; and this angle of revolution
mav have any magnitude. If the revolution takes place in the oppos'te
direction, the angle described is reckoned negative.
610

The secant of an angle is the reciprocal of its cosine,

or

cos A sec A = 1.

611

The cosecant of an angle is the reciprocal of its sine,

or

sin A cosec.4 = 1.

612

The cotangent of an angle is the reciprocal of its tangent,

or

tan A cot A = 1.

Relations between the trigonometrical functions of the

same angle.
613

siuM+cosM = 1.

614

sec2 A = l+tan2.4.

615

cosecM = l+cotM.

[I. 47

[606

616

If tan A
it
617

sin A =
b
b
cos A

[606
Va' + lr

TRIGONOMETRICAL RATIOS.

619

The Complement of A is =

620

The Supplement of A is = 180 -A.

621

sin(90-yl) = cos^,

161

90 A.

tan(90-^) = cot4,
sec (90 A) = cosec A.
622

sin (180- A) =

sin A,

cos (180 A) = -cos 4,

tan (180- J) = -tan A.
In the figure
Z QOK = 180^.

[607, 608

623

sin( A) = sin 4.

624

cos( A) =

cos A.
By Fig., and (607), (608).
The secant, cosecant, and cotangent of 180 A, and of
A, will follow the same rule as their reciprocals, the cosine,
sine, and tangent.
[610-612

625 To reduce any ratio of an angle greater than 90 to the

ratio of an angle less than 90.
Rule.Determine the sign of the ratio by the rules (607),
and then substitute for the given angle the acute angle formed
by its two bounding lines, produced if necessary.
Ex.To find all the ratios of 660.
Measuring 300 (= 660-360) round
the circle from A to P, we find the
acute angle AOP to be 60, and P lies
below JA1, and to the right of BP/.
Therefore
sin 660 = - sin 60 = cos 660=
cos 60=
|,
and from the sine and cosine all the remaining ratios may be
found by (610-616).
Inverse Notation. The angle whose sine is as isdenoted
by sin-1 x.
Y

162

PLANE TRIGONOMETRY.

626 All the angles which have a given sine, cosine, or tan
gent, are given by the formulas
sin"1 .r =

nw+(

(1),

cos1^= 2nv$

(2),

tan'.i =

(3).

nir+0

In these formulae 6 is any angle which has x for its sine, cosine, or
tangent respectively, and n is any integer.
Cosec-1!, sec"1 a;, cot-1 a; have similar general values, by (610-612).
These formulae are verified by taking A, in Fig. 622, for 0, and making
n an odd or even integer successively.

FORMULAE INVOLVING TWO ANGLES, AND

MULTIPLE ANGLES.
627

sin (-4 + 10 = sin A cos//+cos4 sin B,

628

sin (A B) = sin A cos J? cos .4 sin B,

629

cos (A-\-B) = cos A cos// sin A sin B,

630

cos (.4 if) = cos A cosfi+sin A sin B.

Proof.By (700) and (701), we have

sin C = sin A cos B + cos A sin B,
and
sin 0= sin (A+B), by (622).
To obtain sin {A B) change the sign of B in (027), and employ (623),
(624),
coa(A + B) = sin {(W-A)-B], by (621).
Expand by (628), and use (621), (623), (624). For coeiA-B) change the
sign of B in (629).
631

tHn(A+B) = t;il\A+iB.
1 tan A tan B

ann
632

,
/ A
u\
tan A tan B
tan {A - B) =

-.
1 + tan 4 tan//

633

cot (A+B) =

634

cot (A-B) =

cot A cot Bl
cot A + cot B
cot A cot fi+1
cot B cot A
Obtained from (627-630).

MULTIPLE ANGLES.

168

635

sin 2A = 2 sin A cos A.

[627. Puts =4

636

cos 2A = cos2 4 - sin2 A ,

637

=2 cos2 4-1,

638
639
640

=1 2 sin2 A.
2 cos2 A = 1 + cos 2A .
2 sin2 A = 1 - cos 2A.

[629,613
[637
[638

641

sin A = J^ZA.

[640

642

cos^ = V/IpI.

[639

643 i^4--\!T~cosA -1~cosA -_ sin^__

2

vl + cos-4

sin 4

l + cos.4'
[641, G42, 613

646

648

cos^ =

1-tan2^_.
1 + tan2^
cos A

2tan^sin A =

[643, 613

l + tan242
1

l + tan4 tan 42
649

sin (45+ ^ ) = cos (45- A^yjLtlsin A

[641

650

sin A
eos(45+|-) = sin (45-|) = yj*^

[642

651 tan(45+^) = Vf"r8in^-1+sin^- C8^

sin A
652

tan24= 2tanf,.
1 tan2 A

653

cot 2A = 52*!4=i.
2 cot A

cos A

1 sin 4'
[6ii. F*tB=A

PLANE TRIGONOMETRY.

164

t^+A) = \^A
654
655

tan (45-^) - ! 7 *a" A. .

1 -f- tan A

[631, 632

656

sin 3.4 = 3 sin A 4 sinM.

657

cos 3.4 = 4 cos^l 3 cos A.

658

,
n *
3 tan A tan8 A
iAn3A~
i-3tan'i
By putting B = 2A in (627), (629), and (631).

659

sin (A + B) sin (A-B) = sin* A - sin2

660

cos (^+.8) cos(^-B) = cosM -sin2

= cos2 If cos2 .4.

= cos2!? sin2 .4.

From (627), <fcc.
661

sin A + cos
= y/l + sin .4.
~
~
[Proved by squaring.

662

sin

cos
2

663
664
when

= V^l sin .4.

2

^1
1
sin = {v/l + sin.4 \/l sin^l).
A
2
A
1
COs^- = -i {v/l + sin^ + A/l-sin^j,
A
lies between 45 and +45.
2

665 In the accompanying diagram the

signs exhibited in each quadrant are the
signs to be prefixed to the two surds in

the value of sin ^ according to the quad-

j _ + \/ +

rant in which ^ lies.

V/l_ _\y/

A
_
For cos change the second sign.
2

\.

.
_i_^X
/ \

\,

Proof. By examining the changes of sign in (661) and (662) by (607).

MULTIPLE ANGLES.

165

666

2 sin A cos B = sin (4 +

+ sin (AB).

667

2 cos A sin If = sin (A + B)- sin (4 - B).

668

2 cos 4 cosjB = cos(4+jB) + cos(^ B).

669

2 sin A sin 5 = cos (4 5) - cos (A + B).

[627-630

670

sin ^1 + sin B = 2 sin

cos

671

sin A sin B = 2 cos

sin

672

cos .4 + cos B = 2 cos^-t-^ cos

673

cos jB cos ^4 = 2 sin

sin

Obtained by changing .4 into "t^, and 5 into

, in (666-669).

It is advantageous to commit the foregoing formulae to memory, in words,

thus
2 sin cos = sin sum + sin difference,
2 cos sin = sin sum sin difference,
2 cos cos = cos sum + cos difference,
2 sin sin = cos difference cos sum.
sin first + sin second = 2 sin half sum cos half difference,
sin first sin second = 2 cos half sum sin half difference,
cos first + cos second = 2 cos half sum cos half difference,
cos second cos first = 2 sin half sum sin half difference.
674

sin (A+B+C)
= sin A cos B cos C + sin B cos C cos A
+ sin C cos A cos B sin A sin B sin C.

675

cos (A+B+C)
= cos A cos B cos C cos A sin B sin C
cos B sin C sin A cos C sin A sin JB.

676

tan (A+B+C)
_ tan 4 + tan iB + tan C tan ^ tan JB tan C
1 tan B tan 17 tan C tan 4 tan A tan JB '

Proof.Put B + C for J? in (627), (629), and (631).

166

PLANE TRIGONOMETRY.

If A+B+C = 180,
677

678

ABC
sin A + sin B + sin C = 4 cos cos cos.
A
St
M
ABC
sin A + sin 5 sin C = 4 sin sin cos.
2
2
2
ABC
cos^l -f cos2? + cosC = 4 sin sin sin + 1.
M
A
J*
cos.4 -4- cos.B cosC = 4 cos4 cos^sin'^ 1.
2
2
2

679

tan A + tan B + tan C = tan 4 tan B tan C.

680

cot ~ + cot ^ + cot ^ = cot ^ cot ^ cot ^.

JZ
&
&
A

681

sin 2A + sin 2B+ sin 2C = 4 sin 4 sin 5 sin C.

682

cos2^+cos2J3 + cos2C = -4cos^ costfcosC 1.

General formulae,
applying (666-673).
If A+B + C = 7r,
.0n 4 nA . nB
683 4 sin sin

including the foregoing, obtained by

and n be any integer,
. nC
sin

. nir
= sin (!^- nA ) + sin ^ n j + sin^ n&J sin -.
2
nctA
684

.
nA
nB
nC
4 cos cos cos-^nir

= cos^^ nA^j +co8^^ nB\ + cos^j^ nCj + cos

If 4+5 + 0 = 0,
685

4 sin ^ sin ^ sin

= sin nA sin nB sin(7.
AAA

686

4 cos ?^ cos ^ cos ^ =

*U
A

cos?i.^+cos7iZ?+cosn(7+l.

Rule.7/', in formulae (683) to (686), hoo factors on the left

he changed by writing sin for cos, or cos for sin, then, on the
right side, change the signs of those terms which do not contain
the angles of the altered factors.

COMMON ANGLES.

167

Thus, from (683), we obtain

.aH
. nA
nB
nC
687 4 sin cos cos
7T
sin^ nAj + sin

n2?^-f- &va{^~-nG^ + si n

A Formula for the construction of Tables of sines, co

sines, &c.
688
sin (n + l)a sinna = sin Ma sin (n l)a A; sin na,
where a = 10", and k = 2 (1-cos a) =-0000000023504.
689 Formula for verifying the tables
sin^i + sin (72+vl)-8in (72- A) = sin (36 + .4)-sin (36-A),
cos ^l+cos (72+.4) +cos (72+.4) = cos (36+^) +cos (36-.4),
sin (60 + 4) -sin (60- 4) = sin 4.

RATIOS OF CERTAIN ANGLES.

690

sin 45 = 008 45 = -^,

691

sin 60 = ^3,

692

X5- = ^i,

tan45 = l.

cos 60 = ^

tan 60 = y/'d.

-15" = ^,

tanl5 = 2-y3;
cot 15 = 2 + ^/3:
693

sinl8 = ^i,

cos 18 ^J^5,

tan 18 = v/5~l^5
694

sin 54 = ^1,

cos 54 =^-^5,

tan54 = V*F*695 By taking the complements of these angles, the same

table gives the ratios of 30, 75, 72, and 36.

PLANE TRIGONOMETRY.

168

696

Proofssin 15 is obtained from sin (45-30), expanded by (628).

697

sin 18 from the equation sin 2x = cos 3a;, where x = 18.

698

sin 54 from sin 3a; = 3 sin a; 4 sin' a;, where z = 18.

699 And the ratios of various angles may be obtained by taking the sum,
difference, or some multiple of the angles in the table, and making use of
known formulas. Thus
12 = 30-18, 7i =
&c, &c.

PROPERTIES

OF

700

c = a cos B+b cos A.

701

a b
=
sin A
sin B
sin t
2 = 6s+c*-26c cos A.

702

THE

TRIANGLE.

Proof.By Euc. II. 12 and 13, o' = tf+c'^c.AD.

cos A =

703

26c

'

If 8 = ^^, and A denote the area ABG,

nr\A

l(sb)(sc)

/*(.)
[641, 642, 703, 9, 10, 1.

705

706

707

sin A = ^- \/* (* a) (sb) (sc).

be .
A = sin A = Vs (su) (sb) {sc),

[635, 704

[707, 706

= \ v/2V+2ca2+2-&2-a*-6W.
708

169

PROPERTIES OF TRIANGLES.

The Triangle and Circle.

Let
r = radius of inscribed circle,
r= radius of escribed circle
touching the side a,
B = radius of circumscribing
circle.
709

r = -.
s
rn
[From ~.
Fig., A. = ra +, rb . re
. B . C
a sin sin
2
2

710

r =
cos
2
[By a = root + rcot^.
A
s
B
(J
a cos cos

711

712

[By A = ^,+^-^1.
a
a
a
B +r tan C .
[From a = r tan
a
j2

r =
COS
2*

713

It =

a
2 sin A

abc
4A '

[By (III. 20) ami (706)

715

[702
Distance between the centres of inscribed and circum
scribed circles
716
= y/R*2Rr.
[936
Radius of circle touching b, c and the inscribed circle
717

r'= r tan2 \ (B+ C).

[By sin 4 = !
2
r+r

170

PLANE TRIGONOMETRY.

SOLUTION OF TRIANGLES.
Right-angled triangles are solved
by the formulae

c*=a*+bi;

718

fa = c sin A,
\ b = ccos A,
\a = b tan A ,
&c.

719

720

Scalene Triangles.
Case I.The equation

a
b
[701
sin A
sin B
will determine anv one of the four
quantities A, B, a, b when the re
maining three are known.
721
The Ambiguous Case
When, in Case I., two
sides and an acute angle
opposite to one of them
are given, we have, from
the figure,
e sin A
sinC =
a
Then G and 180'- C are the values of G and G', by (622).
Also
because
722

and

/> = c cos A v a* c2 sin* A,

=AD + DC.

When an angle B is to be determined from the equation

sin B = -sin A,
a
is a small fraction ;. the circular measure of B may be approximated

a
to by putting sin (B+C) for sin .4, and using theorem (796).

SOLUTION OF TRIANGLES.

171

723 Case II.When two sides b, c and the included angle

A are known, the third side a is given by the formula
o2 = 62+c2 26c cos A,
when logarithms are not used.

[702

Otherwise, employ the following formula with logarithms,

c%a
724

a BC
bc ,A
tan = cot-.

Obtained from b + c = sin-|-sinOv

-=:
:7,(701),
' and then applying
vr J B
(670) and (671).
Q
- having been found from the above equation, and
being equal to 90 , we have

B and 0 having been determined, a can be found by Case I.

726 If the logarithms of b and c are known, the trouble of taking oat
log (6 c) and log(/> + c) may be avoided by employing the subsidiary angle
6 = tan"1, and the formula
c
727

tanH-B-C)=tan(fl-^)cot^,

[655

Or else the subsidiary angle 0 = cos"1 , and the formula

728

tan i (B- 0) = tan3 1 cot 4-

[643

If a be required without calculating the angles B and C, we may use the

formula
/, . > - A
[From the figure in 960, by
f,an
2
drawing a perpendicular
79
a = cosi (B G)
from B to EG produced.
730 If a be required in terms of b, c, and A alone, and in a form adapted
to logarithmic computation, employ the subsidiary angle
/.
-l / 4fce
\
6s=Bm
((b^yco*, A*)>

and the formula

i(b+cy
a = (b + u) cos 0.

[702,637

172

PLANE TRIGONOMETRY.

Case III.When the three sides are known, the angles

may be found without employing logarithms, from the formula
[703

cos A

731

2bc
732

If logarithms are to be used, take the formulas for

sin^, cos^, or tan^; (704) and (705).

A
LI
U
QUADRILATERAL INSCRIBED IN A CIRCLE,
a'+b'-^-d*
2(ab+cd) '
From AC* = a* + 6' 2ab coaB = c' + i1+
2cd cos B, by (702), and B + D = 180.
733

cos B

734

sin B _
==

735
and
736

2Q
ab+cd

[613, 733

Q = y/Js -)(*- b) (s-c) (.-</)

= area of A BCD,
s = % (a + b + c + d).
Area = %ab sin B+\cd sin B ; substitute sin B from last.
A C> = (c+/>d)(d+bc) m
(o+c)

733

Itadius of circumscribed circle

737

\/{ab+cd) {ac+bd) {ad+bc).

If ^4Z) bisect the side of the triangle ABC in D,

738

t^BDA=J^?.

739
740

cot 2L4D = 2 cot ^1 + cot B.

JD1= i(l, + c,+ 26ccos/l) = K&' + ^-ia').

If AD bisect the angle A of a triangle ABC,

742
743

tan BDA = cot

2

= ^ tan 46 c
2

AD = - cos .
b+c
2

[713, 734, 736

SUBSIDIARY ANGLES:

173

If AD be perpendicular to BG,
_ he sin A _ 6' sin g + c' sin B
a
b+c
745

BD~CD = ^ = a
a
tan 2? + tan 0

REGULAR POLYGON AND CIRCLE.

Radius of circumscribing circle = B.
Radius of inscribed circle
= r.
Side of polygon
= a.
Number of sides
= n.
746

R=
2

cosec ,
n

cot .
n

Area of Polygon
748

= in a2 cot = \n R2 sin = nr2 tan

4
n
*
n
n

USE OF SUBSIDIARY ANGLES.

749

To adapt a + b to logarithmic computation.

Take

0 = tan"1^ ; then a + fc = aeeo'0.

750

For * take ^ = tan_1("^); tbns

751

tt_b = Qy/2coB(g+45)
COS0
To adapt a cos G + b sin (7 to logarithmic computation.

Take 0 = tan-1 ^- ; then

o
a cob (7 6 sin 0 = y(a3 + &') sin (0 0).
[By 617
For similar instances of the use of a subsidiary angle, see (726) to (730).
752 To solve a quadratic equation by employing a subsidiary
angle.
If j?2px + q = 0 be the equation,

174

PLANE TRIGONOMETRY.

Case I.If g be <p*, put

= sin1 fl ; then
P
x = 2p cos1 ~, and 2y sinJ |-.

[639, 640

Case II.If g be >p\ put 9 = sec' fl ; then

P
X = p (1 tan fl), imaginary roots.
Case III.If g be negative, put
x=

-Q

sin

cofc 0-

[614

= tan1 fl ; then
and

8
y/g tan .

[644, 645

LIMITS OF RATIOS.
_ tan fl
,

when 0 vanishes.
For ultimately ^ = = 1. [601,606 0

754

n sin - = fl when n is infinite.

755

^co8"^") = ^ when n is infinite.

By putting 1 for fl in last.

l-sin! J 2, and expand the logarithm by (156).

DE MOIVRE'S THEOKEM.
756

(cos a+ i sin a) cos /3+ i'sin ft) ... &c.

= cos(a+j8+7+ ...) + isin(a++7+ ...),

where i = V 1 .
[Proved by Induction.
757
(cos 0+ 8in fl)" = cos nfl+i sin nfl.
Proof.By Induction, or by putting a, /I, &c. each = in (756).
Expansion of cosnB, fyc, in powers sin 6 and cost).
758

cos nfl = cos" fl-C(n, 2) cos""* fl sin2 fl

+ C (, 4) cos-4 0 sin4 6 &c.

759 sin nfl = n cos'^fl sin fl-C(n, 3) cos""3 fl sin,fl+&c.
Proof. Expand (757) by Bin. Th., and equate real and imaginary
parts.

TRIGONOMETRICAL SERIES.

175

7fi0 tannfl n tan 0~C (n 3) tap30+&c1 DU

~ 1 - C {n, 2) tan2 0+ C (n, 4) tan4 0 - &c. "
. In series (758, 759), stop at, and exclude, all terms with indices greater
than n. Note, n is here an integer.

Let sr = sum of the G{n, r) products of tana, tan/3, tany,

&c. to n terms.
761

sin (a+#+y+&c.) = coso cos/3 ... {sv *s+*s &c).

762

cos(a+/?+y+&c.) = coso cos/J ... (1 *2+*4 &c).

Pboof.By equating real and imaginary parts in (756).

763

tan (a+j8+y+&c.) = *i -*.+* ~7 +&<?.

Expansions of the sine and cosine in powers of the angle.

764

sin0 = 0-.||+|j--&c.

cos0 = l-|l+|^+&c.

g
Proof., Put for 6 in (757) and n oo , employing (754) aud (755).
n

766
768
770

e* = cos0+isin0.

e"*9 = cos 8i sin 0. By (150)

e+e"' = 2cos0.

e'9 e", = 2isin0.

ea e~*
*lanp-e+e-,9-

1+itnnd _ciie
1i tan0

Expansion of cosn 8 and sin" 6 in cosines or sines of

multiples of 0.
772
2*1cosB0 = cos n9+n cos (n 2) 0
+ C(n, 2) cos (- 4)0+ C(n, 3) cos (n- 6)0+ &c.
773 When n is even,
2""1 (- 1)1B sin" 0 = cos n0-n cos (n 2) 0
+ C (n, 2) cos (n-4) 0 C(n, 3) cos (n-6)0 + &c,
774 And when is odd,
2'-1( ifr sinB0 = sin w0 n sin (2) 0
+ C (n, 2) sin (n-4) 0 C (n, 3) sin (n6) 0 + &c.

170

PLAXE TRIGOXOilETRY.

Observe that in these series the coefficients are those of the Binomial
Theorem, with this exception : If n be even, the last term must be divided by 2.
The series are obtained by expanding (e" e")" by the Binomial Theorem,
collecting the equidistant terms in pairs, and employing (768) and (769).
Expansion of cos n9 and sin n9 in powers of sin 0.
775

When n is even,
a
i
2 i a i "2 (2 -2) 4 a
cos no = 1 sin1 a ^
p
sin* 0

6!
776

When n is odd,

cos nfi = cos 0{ 1 - 2^1 sin* 6 + g=l^gzg sin* 6

777

_(n'-l)('-ff)(.'-y)gi||,#+Ae^
o!
;
When n is even,
sin n0 = n cos 0 ^ sin 0 11 ^ ^ sin3 5+
(We.4t) sin^_ (w..g) (n.,4Q (w.-ff)

^,

5!
/ !
3
778 When . is odd,
a = n sin
0
a (2
1) sm'fl
i/i +
, 1
n(tr l)(n'
. 5 0
sin n0
v
,'
rr 3") sin
o!
o!
n(n2-l)' (2-.T)
. 7, &c.
.

y a,
i-5(V-54) sin7
0+
Peoof.By (758), we may assnme, when n is an even integer,
cosnfl = 1+A, ein16+Ai8inie + ... +Ausin'r0+....
Put 6+x for 0, and in cos n6 cos nx sin n# sin na substitute for cos nx and
sin ru; their values in powers of nx from (764). Each term on the right is of
the type A2r (sin 0 cos a: + cos 0 sin.r)2r. Make similar substitutions for cos^
and sinz in powers of x. Collect the two coefficients of a? in each term by
the multinomial theorem (137) and equate them all to the coefficient of .< 'on the left. In this equation write cos*0 for 1 sin'0 everywhere, and then
equate the coefficients of sinIr0 to obtain the relation between the successive
quantities Alr and Alrtl for the series (775).
To obtain the series (777) equate the coefficients of x instead of those of*1.
When n is an odd integer, begin by assuming, by (759),
sin nd = Al sin 6 + At sin' 6+ &c.

TRIGONOMETRICAL SERIES.

177

779 The expansions of cos nQ and sin nO in powers of cos 0

are obtained by changing 0 into
in (775) to (778).
780

Expansion of cos nO in descending powers of cos 0.

2 cos n6 = (2 cos 6)n- n (2 cos fl)-s+OT ("-3) (2 cos 0)-4A1

+ (_1)rn(n-r-l)(n-r-2)...(n-2r+l)(2cosg)B_2r+
up to the last positive power of 2 cos 0.
Proof.By expanding each term of the identity
log(l-r)+log(l- ~) = log jl-(* + i-*U
by (156), equating coefficients of z", and substituting from (768).
783 sina+c sin (a+ft)+r sin (a+2ft) + &c. to n terms
_ sin a c sin (a ft) c" sin (a+/?) + cn+1 sin (a+i 1)8}
1^2cCos^3+?
If c be < 1 and n infinite, this becomes
mni

_ sing c sin (a ft)

l-2ccosft+c2 '

785 cos o+c cos (a+^+c2 cos (a+2ft) + &c. to n terms

= a similar result, changing sin into cos in the numerator.
786

Similarly when c is < 1 and n infinite.

787 Method of summation. Substitute for tlie sines or

cosines their exponential values (768). Sum the two resulting
geometrical series, and substitute the sines or cosines again for
the exponential values by (766).
788

c sin (a+ft) + ^sin(a4-2ft) + ^sin(a+3ft) + &c. to

infinity

789

= ecco*p sin (a+c sin ft) sin a.

c cos (o+ft) + |^ cos (a+2ft) +

infinity

cos (a+3ft) + &c. to

= eecoM cos (a-j- c sin ft) cos a.

Obtained by the rnle in (787).

2 A

PLANE TRIGONOMETRY.

178

790 If, in the series (783) to (789), 0 be changed into

the alternate terms will thereby be; changed.

the signs of

Expansion of 0 in powers of tanO (Gregory's series).

791

6 = tan 0 -

+ t^l -&c.

The series converges if tan 8 be not > 1.

Proof.By expanding the logarithm of the value of e* in (771) by (158).
Formulae lor the calculation of the value of n by Gregory's
series.
792
794

= tan-1I + tau1I = 4tan-1I-tan-1-L

4
2
3
5
239

t791

=4 tan1 1 - tan1 JL + tan1 JL.

Proof.By employing the formula for tan (AB), (631).

To prove that it is incommensurable .
795 Convert the value of tan 8 in terms of 0 from (764) and (765) into
Q foi
J3
a continued Iraction, thus tan 0 =
or this result may
1 o 5 7 &c.
may be obtained by putting iO for y in (294), and by (770). Hence
.
e_ _ P e &
tan H
3 5 7- Ac.
Put Z for 8, and nssnme that ir, and therefore , is commensnrable. Let
2
4
- = , m and m being integers. Then we shall have 1 = nl- -"^
4
en on in <sc.
The continued fraction is incommensurable, by (177). But unity cannot be
equal to an incommensurable quantity. Therefore f is not commensurable.
796

^ sin* = m sin (as + a),

797

" tan* = ntany,

where m

x = n sin a+ sin 2a + sin 3a + <fec.

2
3
i'
x- = y m sin 2i/+ - siu4i/ ' sin6i/ + &c.,

.
1+H
Proof.By tubs''tinting the eip 'iiential values of the sine or tangent (769)
and (770), find then eliminating x.
n
798 Coefficient of a;* in the expansion of eax cos bx =
cos n$,
where a = r cos 8 and b = r sin 0.
For proof, substitute for cos hx from (768); expand by (150); put
a = rcosfl, b = r sin 6 in the coefficient of x", and employ (757).

TRIGONOMETRICAL SERIES.

799

When e is < 1,

179

/\ _ e i = 1 + 26 cos 0 + 2lJ cos 20 + 21* cos 36 + ...,

1 e cos 6

where 6 =
1+ VI -eJ
26 and 2 cos 0 = x-\ 1 , expand the fraction in two
For proof, put e = -tz
1+tr
x
feries of powers of x by the method of (257), and substitute from (768).
800

sin a+ sin (a+j8) + sin (a+ 2/8) + . . . + sin { a + (n - 1)

sin (a-f

801

sin^

sin -42
cosa+cos(a+)+cos(a+2) + ...+cos(a+(n-l)i8}
cos^aH ) sin^
sin A
2

802 If the terms in these series have the signs + and

alternately, change j3 into P + jt in the results.
Proof.Multiply the series by 2 sin

and apply (669) and (666).

803

2ir
If 0 = in (800) and (801), each series vanishes.

804

Generally, If j3 = , and if r be an integer not a

multiple of n, the sum of the r,th powers of the sines or cosines

in (800) or (801) is zero if r be odd; and if r be even it is
= F(r' f); by (772) t0 (774)-

General Theorem. Denoting the snm of the series

805
c + v + c/+...+c/byii'((i!);
then c cos a + c, cos (a + j8) + . . . + e cos (0 + /3) = e'*JXe") +e-'*.F(-*) } ,
and
806 esina + c1sin(a+/3) + ... +cnsin(a f n/3) = ~.{e''F (eis)-e-uF(e-i3)}.
Proved bv substituting for the sines and cosines their exponential values
(76C), Ac.

180

PLANE TRIGONOMETRY.

807

Expansion of the sine and cosine in factors.

2.rBy cos n6 + y-'

= p-^cosfl+^j p-2*yco8(0+^)+3fj ...

to n factors, adding to the angle successively.
Proof.By solving thequadraticon the left, we get x=y(cosn6 + is\nnd)" .
The n values of x are found by (757) and (626), and thence the factors. For
the factors of z*y" see (480).
808

sin n<f> = 2-1 sin <f> sin

sin (j> + ^ . . .

as far as n factors of sines.

Proof.By putting x = y \ and 6 = 2$ in the last.
809 If n be even,
sin n<f> = 2"'1 ain<j> cos^sin2-^- sin*^ ^sin' sin2<^ &c.
810 If n be odd, omit cos^ and make up n factors, reckoning
two factors for each pair of terms in brackets.
Proof.From (808), by collecting equidistant factors in pairs, and
applying (659).
811

cos n<f> = 2""1 sin^ + ^-j sin (^<f> +

Proof.Put </> +
812

... to n factors.

for </> in (808).

Also, if n be odd,

cos n<f> = 2"'1 cos ^ ^sin2^- sin2 ^ ^sin2 ^ sin2 ^V

813 If n be even, omit cos ^.
Proof.As in (809).
at a
814

e-i1 sm

sin
2w sin
.
ilir ...sinv
. (n l)ir
n = a"
n
n
n
n
Proof.Divide (8<9) by sin*, and make <j> vanish ; then apply (754).

816

{i-^j(i-(ar}j,_rj

Proof.Put ? = - in (809) and (812) ; divide by (81-4) and make n

infinite.
w

ADDITIONAL FORMULAE.

817

181

e*-2cosfl + e-*

2 y = 1 ,
Z and
0 for 6 in (807),
Proved by substituting a- = 1 + ,
2n
2
n
,6 by patting
making n infinite, and reducing one series of factors to 4 sin'
z=0.
2
De Moivre's Property of the
Circle. Take P any point, and
POD = 0 any angle,
BOG = COD = &c. = ;
n
0P = x; 0B = r.
819
2xnr* cos n$ + r2n
= P.B2 PC3 PD2 ...ton factors.
By (807) and (702), since PB1 = a?-'2xr cob 0 + t*, &c.
820

If x = r,

2r sm^ = PB.PC.PD ... Ac.

821

Cotes''s properties.If 6 =

2tt
n
^r _ PB.PC.PD ... &c.
x+rn = Pa.Pb.Pc ... &c.

822

ADDITIONAL FORMULA?.
823

cot -4+tan A = 2 cosec 24 = sec A cosec 4.

824

cosec 2A + cot 2A = cot A .

sec 4 = 1 + tan 4 tan ^- .

826

^
.id
-4-4
cos -4 = cos* sin*.
2
2

ggry

i.tan 4+sec
A I
AtL I -4
A/I = tan (^45+

828

tan A+ tan S

-s = tan 4 tan P..

cot 4+ cot B

829

sec2 ,4 cosec2 4 = sec2 A+ cosec2 A.

PLANE TRIGONOMETRY.

182

830

If A+B+C=l,
tan B tan C+tan C tan A +tan .4 tan B = 1.

831

If 4+B+C = ir,
cotU cotC+cotC cot^+cot^4 cotB = 1.

832

Bin1|-+8in-1-i = |..

tan'A + tan1!:^.

In a right-angled triangle ABC, G being the right angle,

833

cos2B = f4=:K.

tan 25=

834

tan|^=v/^).

ft+r = i(a+6).

In any triangle,
835

sm^(A-B) = ^cosiC.

83g

sin ^1 6 _ 2 6'
sin A-\-B
c2

c
tan -J- .4 -f tan ^i?
tan ^4 tan ^B

837

a (2+62+r) = 6c cos 4 + ca cos B+ab cos C.

838

Area of triangle ABC = \bc sin .4

_ ! a sin B sin C _ , 2 ,*\ sin -4 sin .B
-2""^nT3
tK
b) sin {A-By

839

840

= i (+6+c)2 tan ^ tan ^B tan C.

c
a 6'

2"6c cos \A cos B cos \C.

u+b-\-c

With the notation of (709),

841
842
843

** = i (a+6+c) tan \A tan

2Br =

abc
"r,

tan \C.

A = y/rr.nr,.

cos J -f 6 cos B+c cos C = 4/1 sin J sin B sin C.

ADDITIONAL FORMULA!.

183

844 iJ+r = \ (a cot A+b cot B+c cot C) = sum of per

pendiculars on the sides from centre of circumscribing circle.
This may also be shown by applying Enc. VI. D. to the circle described
on R as diameter and the quadrilateral so formed.
845

ra rb rc = abc cos \A cos \B cos \C.

846

r = y/M+Jir.rJ + y/ir.n)^

847
849

1=1
r
ra + 1+-!.
rb
rc

tanM=V
'
v rbre

If 0 be the centre of inscribed circle,

OA = ,^c, cos \A.
a+6+c
3

850

a (6 cos C-c cos 5) = 6s -c2.

851

6 cos B+c cos C= ccos^-C).

852

cos .4 +6 cos i?+c cosC = 2a sin 5 sinC.

853

cos^+cosB+cosC = 1 +

854

2a sin B sin C
a+6+c

K s = \(a + b + c),
1 cos2 a cos2 6 cossc+2 cos a cos b cose
= 4 sin* sin (# a) sin (* 6) sin (sc).

855

l+cos2a+cos2&+cos2c+2cosa cos 6 cose

= 4 cos* cos( a) cos (sb) cos (sc).

act*
856

857

,
a
b
c
4 cos cos cos
u
m
m
= cos*+cos (a a)+cos (* 6)+cos (sc).
4 sin

sin -|- sin ^

= sin*+sin (* a) + sin (* i) + sin (sc).

858

^ = 6(l+^+| + ...) = 8(l+|+^ + ...).

Pboof.Equate coefficients of 0" in the expansion of

(815) or of cos 6 by (705) and (81C).

by (764) and
0

184

PLANE TRIGONOMETRY.

859

Examples of the Solutions of Triangles.

Ex. 1 : Case II. (724).Two sides of a triangle 6, c, being 900 and 700
feet, and the included angle 47 25', to find the remaining angles.
tan
= h-=? cot 4 = I cot 23 42' 30" j
2
b+c
2
8
therefore

log tan (BC) = log cot log 8 ;

a
therefore
L tan \ (B-0) = L cot 23 42' 30"-3 log 2,
10 being added to each side of the equation.
.-.
L cot 23 42' 30" = 10-3573942* f;. (B-C) = 15 53' 19 55"*
3 log 2 = -9030900 | and 1 (B + C) = 66 17' 30"
.-.

itani(B-C) = 9 4543042 ( .-.

B = 82 10' 49 55"
And, by subtraction, C = 50 24' 10 45".

Ex. 2: Case III. (732).Given the sides a, b, c = 7, 8, 9 respectively,

to find the angles.
. A
l(s-b)(s-e)
/4.3 = Vl0;
fY
tan2=V
(,-) =Vl2T5
therefore

Lt&n~=10 + l (log 21) = 9 650515;

therefore
\A = 24 5' 41-43".*
\B is found in a similar manner, and C = 180 A B.
Ex. 3.In a right-angled triangle, given the hypotenuse c = 6953 and
a side 6 = 3, to find the remaining angles.
Here cos A = 3 . But, since A is nearly a right angle, it cannot be
6953
determined accurately from log cos A. Therefore take
. A = V~~2
/l-cos4 =V6953;
/3475
8lnT
therefore

Lsin^ = 10 + \ (log 3475 -log 6953) = 9 8493913;

therefore

~ = 44 59' 1552" *

therefore

A = 89 58' 31 04" and B = 0 1' 28 96".

* See Chambers's Mathematical Tables for a concise explanation of the

method of obtaining these figures.

SPHERICAL TRIGONOMETRYT

INTRODUCTORY THEOREMS.
870 Definitions.Planes through the centre of a sphere
intersect the surface in great circles ; other planes intersect
it in small circles. Unless otherwise stated, all arcs are
measured on great circles.
The poles of a great circle are the extremities of the
diameter perpendicular to its plane.
The sides a, b, c of a spherical triangle are the arcs of
great circles BO, GA, J 2? on a sphere of radius unity; and
the angles A, B, C are the angles between the tangents to the
sides at the vertices, or the angles between the planes of the
great circles. The centre of the sphere will be denoted by 0.
The polar triangle of a spherical triangle ABC has for its
angular points A', B', C, the poles of the sides BC, GA, AB
of the primitive triangle in the directions of A, B, C respec
tively (since each great circle has two poles). The sides of
A'B'C' are denoted by a, b', c'.
871 The sides and angles of the
polar triangle are respectively the
supplements of the angles and
sides of the primitive triangle ;
that is,

a'+A = b'+B = e'+C = n,

a+A'=b+B'=c+C'=ir.
Let BC produced cut the sides A'lV,
CA' in G, H. B is the pole of A'C,
therefore BH = 2

Similarly OG = ~,

therefore, by addition, a -)- GIIjt and GH=A\ because A' is the pole of BC.
The polar diagram of a spherical polygon is formed in the same way, and
the same relations subsist between the sides and angles of the two figures.
2 B

186

SPHERICAL TRIGONOMETRY.

Rule. Hence, any equation between the sides and angles

of a spherical triangle produces a supplementary equation by
changing a into w A and A into w a, fyc.
872 The centre of the inscribed circle, radius r, is also the
centre of the circumscribed circle, radius R', of the polar
triangle, and r + R'= ^t.
Pkoof.In the last figure, let 0 be the centre of the inscribed circle of
ABC; then 01), the perpendicular on BO, passes through A', the polo of BC.
Also, OD = r; therefore OA'=U r. Similarly 0B'= OC=\k r; there
fore 0 is the centre of the circumscribed circle of A'B'C, and r + R'= ^t.

873 The sine of the arc joining a point on the circumference

of a small circle with the pole of a parallel great circle, is equal
to the ratio of the circumferences or corresponding arcs of the
two circles.
For it is equal to the radius of the small circle divided by the radius of
the sphere ; that is, by the radius of the great circle.

874

Two sides of a triangle are greater than the third.

[By XI. 20.

875 The sides of a triangle are together less than the cir
cumference of a great circle.
[By XI. 21.
876 The angles of a triangle are together greater than two
right angles.
For ir A + rr B + tt C is <2w, by (875) and the polar triangle.

877

If two sides of a^triangle are equal, the opposite angles

are equal.

878

[By the geometrical proof in (89-1).

If two angles of a triangle are equal, the opposite sides

are equal.

[By the polar triangle and (877).

879 The greater angle of a triangle has the greater side

opposite to it.
Proof. If B be > A, draw the arc BD meeting AC in B, and make
L ABD=A, therefore BB = AB; but BD + I)C>BC, therefore AOBC.

880

The greater side of a triangle has the greater angle

opposite to it.

[By the polar triangle and (879).

OBLIQUE-ANGLED TRIANGLES.

187

RIGHT-ANGLED TRIANGLES.
881 Napier's Rules. In the triangle ABC let G be a right
angle, then a, {\ir B), ( c), (\ir A), and b, are called
the five circular parts. Taking any part for middle part,
Napier's rules are
I. sine of middle part = product of tangents of adjacent parts.
II. sine of middle part = product of cosines of opposite parts.
In applying the rules we can take A, B, c instead of their
complements, and change sine into cos, or vice versa, for those
parts at once. Thus, taking b for the middle part,
sin b = tan a cot A = sin B sin c.
Ten equations in all are given by the
rules.
Proof.From any point P in OA, draw
PR perpendicular to 00, and RQ to OB;
therefore PRQ is a right angle ; therefore OB
is perpendicular to PR and QR, and therefore
to FQ. Then prove any formula by proportion
from the triangles of the tetrahedron OPQR,
which are all right-angled. Otherwise, prove
by the formulas lor oblique-angled triangles.

OBLIQUE-ANGLED TRIANGLES.

882

cos a = cos b cos c+sin b sin c cos A.

Proof.Draw tangents at A
to the sides c, b to meet OB, 00
in D and E. Express DW by
(702) applied to each of tho
trinngles DAE and DOE, and
subtract.
If AB and AG are both > -,
produce them to meet in A', the
pole of A, and employ the tri
angle A'BC.
If AB alone be > - , proZ
dace BA to meet BG.

The supplementary formula, by (871), is

883
cos A = cos/f cos C+sin B siuCcoso.

188

SPHERICAL TRIGONOMETRY.

884

sin A = >iu (*7fe) siM (*-c) .

2
sin 6 sin c

885

cos4 = J a*"* sin (j-aT

2
v
sin 6 sin e

886

tan4 = v/Si" .(V~6) 'I" ('TC) where s = \{a+b+c).

2
*
sin* sin (* a)

pROOr.sin' = 1(1 cos A). Substitute for cos A from (872), and
a
throw the numerator of the whole expression into factors by (673). Similarly
for cos---.
The supplementary formuke are obtained in a similar way,
or by the rule in (871). They are
887

a
cosT

888

. ft
s.n-

/cos (3 B) cos (S-Cj

*
sin B s*in C

sin B sin C
tan-|

889

cos (S B) cos (SC)

where S = 1(4+5+ tT).
890

Let
= ^

<r = \/ sin * sin (s //) sin (sb) sin (s c)

+ - cos a cos 6 cos c cos-' a cos'- b cos2 c.

Then the supplementary form, by (871), is

891

S = y/ cos 6' cos (SA) cos {S B) cos (if C)

= ^Vl 2 cos ^4 cosii cosC cos* .4 cos2 if cos-'C.

892

sui .1 = t: .
sin b sin c

sin a =
sin B sin C

[By sin A = 2 sin ^ cos 4 and (884, 885), Ac

893 The following rules will produce the ten formulae
(884 to 892)
I. Write sin he/ore each factor in the s values of sin-,

OBLIQUE-ANGLED TRIANGLES.

189

A
A
cos -, tan , sin A, and A, in Plane Trigonometry (704707), to obtain the corresponding formula? in Spherical Trigo
nometry.
II. To obtain the supplementary forms of the five results,
transpose large and small letters everywhere, and transpose
sin and cos everywhere but in the denominators, and write
minus before cos S.
ofti

sin A
sin a

sin B
sin b

sin C
sin c

Proof.By (882). Otherwise, in the figure of 882, draw PN perpendi

cular to BOC, and NR, NS to OB, 00. Prove PRO and PSO right angles
by 1. 47, and therefore PN = OP sin c sin B = OP sin 6 sin 0.

895
cos 6 cos C = cot sin b cot ^4 sin C.
To remember this formula, take any four consecutive angles
and sides (as a, G, b, J), and, calling the first and fourth the
extremes, and the second and third the middle parts, employ
the following rule :
Rule. Product of cosines of middle parts = cot extreme
aide X sin middle side cot extreme angle X sin middle angle.
Pkoof. In the formula for cos a (882) substitute a similar value for
cos c, and for sin o put sin 0 - .
Mil .1

896

NAPIER'S FORMTJL/E.

<2> *("+*> =StfetK-

Rule. in Me vaZ?<e o/ tan^(A B) change sin <o cos to

o&Wn tan^(A + B). To o&fain (3) ai (4)/rom (1) and (2),
transpose sides and angles, and change cot to tan.
PfiOOF.In the values of cos^l and cos B, by (883), put m sin a and
msino for sin A and sin B, and add the two equations. Then put
m = ""^sinff and transform by (670-672).
sin a sin 6

190

SPHERICAL TRIGONOMETRY.

897

GAUSS'S FORMULAE.
(1)

sm\(A + B) _ cosj(q 6)
COS }jC
cos^c

(2)

sili } (A B) _ sin (a b)
cos
sin \c

(3)

qos\{A + B)
sin
cos
sin 1 61

cosj(a+6)
cos
_ 8in|(a+6)
sin j;c

From any of these formulae the others may be obtained

by the following rule :
Rule. Change the sign of the letter B (large or small) on
one side of the equation, and write sin for cos and cos for sin
on the other side.
Proof.Take sin a (.4 + 2?) = sin \A cos J-B + cos |.A sin ^B,
substitute the s values by (884, 885), and reduce.

SPHERICAL TRIANGLE AND CIRCLE.

898 Let r be the radius of the in

scribed circle of ABC ; ra the radius of
the escribed circle touching the side a,
and R, Ba the radii of the circumscribed
circles ; then
(1)
(3)

tan r = tan \A sin (s a) =

sm s
2 sin a
sin^ sini# sin^C
sin A

(4)
2 cos \A cos \B cos \C
22
cosS+cos (> .4)+cos (SB)+&c.
Proof. The first value is found from the
right-angled triangle OAF, in which AF = sa.
The other values by (884-892).

SPHERICAL TRIANGLE AND CIRCLE.

899

(1)

191

tan ra = tan \A sin s =

sin (* a)

(3)

= - f*n a sin IA cos IB cos iC

sin .4

(4)

2
2 cos^l sin^ZJ sin^C
22
cos S cos (S J)-f cos (SB)+ cob {SC)'

Proof. From the right-angled triangle O'AT, in which AF'= s.

Notk.The first two values of tan ra may be obtained from those of
tanr hy interchanging s and s a.

900

(1) tan R =

tan la
cos {SA)

cos S

(3)

sin \a
sin A cos \b cos
_ 2 sin \a sin ^6 sin \c

(4)
__ sin^+sin (s a)4-sin
2<r

6)-f-&c.

Proof.The first value from the right-angled

triangle OBD, in which t OBD = 8 A. The other
values by the formula? (887-892).
901

(1)
'

t*nRa= tanj
cos 6

(3)

sin \a
sin 4 sin j 6 sin \c

(4)

2 sin ^ cos \b cos

(T
sin s sin

(5)

coB(g-il)

u) -f sin (, b) + sin (a c)
2<T

Proof.From the right-angled triangle O'BD, in which / O'BD = irS.

192

SPHERICAL TRIGONOMETRY.

SPHERICAL AREAS.

902

area of ABC = (A + B+C-n) r- = Er3,

where

E = A-\-B+Ctt, the spherical excess.

Proof.By adding the three lanes

ABDC,

BCEA,

GAFB,

"\
AL~~

and observing that ABF = CDE,

we get

903

( + + ) 2^=2^ + 2ABO.
\ it
it
it I

I >^

}P
J

\
^~~~-/D
\
^/

AREA OF SPHERICAL POLYGON,

n being the number of sides,

Area = {interior Angles ( 2) w} r2
= {2w Exterior Angles} r*
= {2ir sides of Polar Diagram} r2.
The last value holds for a curvilinear area in the limit.
Pboof.By joining the vertices with an interior point, and adding the
areas of the spherical triangles so formed.
904

Cagnoli's Theorem.
siniE = vlgjf sin (*~ ") sin(* 6) sin (*<)}

2 cos ha cos lb cos ^c

Proof. Expand sin [(A + B)--\ (*- C)] by (628), and transform by
Gauss's equations (897 i., iii.) and (669, 890).

905

Llhuillier's Theorem.

tan \E = ^/(tan \s tan ^ (s a) tan ^ (sb) tan^ (* c) }.

Proof. Multiply numerator and denominator of the left side by
2 cos ; (A + BC + v) and reduce by (667, 668), then eliminate J- {A+B)
by Gauss's formulae (8!)7 i., iii.) Transform by (672, 673), and substitute
from (886).

POLYHEDRONS.

193

POLYHEDRONS.
Let the number of faces, solid angles, and edges, of any
polyhedron be F, 8, and E ; then
F+S = E+2.

906

Proof. Project the polyhedron upon an internal sphere. Let m =

number of sides, and s = sum of angles of one of the spherical polygons so
formed. Then its area = { (ro 2) r} r3, by (903). S am this for all the
polygons, and equate to 4-iri*.
THE FIVE REGULAR SOLIDS.
Let m be the number of sides in each face, n the number
of plane angles in each solid angle ; therefore
907

mF = nS = 2E.

From these equations and (906), find F, S, and E in terms of m and

, thus,
1 _ m M , 1_ 1 \
JL_ (_L+ 1 _ 1\
F
2 \m
n
2/' S
2 \m n 2/' E
m n 2"
In order that F, S, and E may be positive, we must have - - + ->
m
n
2'
a relation which admits of five solutions in whole numbers, corresponding to
the five regular solids. The values of m, , F, S, and E for the five regular
solids are exhibited in the following table :

908

Or,

Tetrahedron

Hexahedron

12

Octahedron

12

Dodecahedron

12

20

30

Icosahedron

20

12

30

The sum of all the plane angles of any polyhedron

Four right angles for every vertex less eight right angles,
2 c

194

SPHERICAL TRIGONOMETRY.

909 If I be the angle between two adjacent faces of a

regular polyhedron,
sin

= cos
n

sin .
m

Proof. Let PQ = a be the edge, and 8

the centre of a face, T the middle point of
PQ, 0 the centre of the inscribed and circum
scribed spheres, ABC the projection of PST
upon a concenti ic sphere. In this spherical
triangle,
A =
and B = = PST.
C=
2'
n
m
Also
STO = \I.
Now, by (881, ii.),
cos A = sin B cos BC ;
that is,

cos = sin sin -J- J.

n
to
Q. e. d.

If r, i2 be the radii of the inscribed and circumscribed

spheres of a regular polyhedron,
910

r = 4- tan
2

cot ,
m

R = 4 tan J tan .
2
n

Proof. In the above figure, OS = r, OP = B, Pr=;

OS = PT cot tan \I.

Also OP = FT cosec ,40, and by (881, i.),

sin AC = tan BC cot /I = cot i/ cot ; therefore, Ac.

and

ELEMENTARY

GEOMETRY.

MISCELLANEOUS PROPOSITIONS.
To find the point in a given line QY, the sum of whose
distances from two fixed points 8, S' is a minimum.
Draw SYB at right angles to QY,
making YB = YS. Join BS', cutting
QY in P. Then P will be the required
point.
Proof.For, if D be any other point
on the line, SD = DB and SP = PB.
But BD + DS' is >BS'; therefore, &c.
B is called the reflection of the point S,
and SPS' is the path of a ray of light
reflected at the line QY.
If S, S' and QY are not in the same plane, make SY, YB equal perpen
diculars as before, but the last in the plane of S' and QY.
Similarly, the point Q in the given line, the difference of whose distances
from the fixed points S and B' is a maximum, is found by a like construction.
The minimum sum of distances from 8, 8' is given by
(SP + S'Pf= SS'*+4<SY.S'Y'.
And the maximum difference from 8 and B' is given by
(SQ-R'QY= (SRy-4SY.R'Y.
Proved by VI. D., since SBB'S' can be inscribed in a circle.
921 Hence, to find the
shortest distance from P
to Q en route of the lines
AB, BG, CD; in other
words, the path of the ray
reflected at the successive
surfaces AB, BC, GD.
Find Pu the reflection of P at
the first surface ; then Pj, the
reflection of P, at the second sur
face ; next Ps, the reflection of P,
at the third surface ; and so on if

ELEMENTARY GEOMETRY.

196

there be more surfaces. Lastly, join Q with P the last reflection, cutting
CD in a. Join aP,, cutting BO in 6. Join 6P cntting AB in c. Join cP.
PcboQ is the path required.
The same construction will give the path when the surfaces are not, as
in the case considered, all perpendicular to the same plane.
922 If the straight line d from the vertex of a triangle
divide the base into segments p, q, and if h be the distance
from the point of section to the foot of the perpendicular from
the vertex on the base, then
&+C = p,+q2+2di+2h (p-q).
[II. 12, 13.
The following cases are important :
(i.) When p = q,

ft'+c2 = 2q3+2(P ;

i.e., the sum of the squares of two sides of a

triangle is equal to twice the square of half
the base, together with twice the square of the
bisecting line drawn from the vertex.
(ii.) When p = 2q,

26i+c2 = 6^+3^.

(II. 12 or 13)

(iii.) When the triangle is isosceles,

b1 = c2 = pq+(P.

923 If 0 be the centre of an equilateral triangle ABC and

P any point in space. Then
PAt+PB1+PC3 = Z(POi+OA%).
Proof. PB' + PC = 2PD, + 2PD1.
Also

PA* + 2PL* =60L* + 3PO',

and

BO = 20D ;

(922, i.)
(922, ii.)

therefore, &c.
Cob.Hence, if P be any point on the
surface of a sphere, centre 0, the sum of the squares of
its distances from A, B, G is constant. And if r, the radius
of the sphere, be equal to OA, the sum of the same squares is
equal to 6r*.

MISCELLANEOUS PROPOSITIONS.

197

924 The sum of the squares of

the sides of a quadrilateral is equal
to the sum of the squares of the
diagonals plus four times the square
of the line joining the middle points
of the diagonals.
(922, i.)
925 Cor.The sum of the squares
of the sides of a parallelogram is
equal to the sum of the squares of the diagonals.
926 In a given line AG, to find a point X whose distance
from a point P shall have a given
ratio to its distance in a given
direction from a line AB.
Through P draw BPO parallel to the
given direction. Produce AP, and make
CE in the given ratio to GB. Draw PX
parallel to EG, and XY to GB. There are
two solutions when GE cuts AP in two
points.
[Proof.By (VI. 2).

927 To find a point X in AG,

whose distance XY from AB parallel
to BG shall have a given ratio to its
distance XZ from BG parallel to AD.
Draw AE parallel to BG, and having to
AD the given ratio. Join BE cutting AG in
X, the point required. [Proved by (VI. 2).

928 To find a point X on any

line, straight or curved, whose
distances XY, XZ, in given direc
tions from two given lines AP, AB,
shall be in a given ratio.
Take P any point in the first line.
Draw PB parallel to the direction of XY,
and BG parallel to that of XZ, making
PB have to BO the given ratio. Join PO,
cutting AB in D. Draw BE parallel to
OB. Then AE produced cuts the line in
X, the point required, and is the locus of
such points.
[Proof.By (VI. 2).

198

ELEMENTARY GEOMETRY.

929 To draw a line XY through a given point P so that

the segments XP, PY, intercepted by a given circle, shall be
in a given ratio.
Divide the radius of the circle in that ratio,
and, with the parts for sides, construct a triangle
PBG upon PC as base. Produce OB to cut the
circle in X. Draw XPY and GY.
Then
PB+BG = radius ;
therefore
PD = DX;
But
CY=CX;
thereforePD is parallel to OT (1.5, 28); therefore
Ac., by (VI. 2).

930 From a given point P in the

side of a triangle, to draw a line PX
which shall divide the area of the tri
angle in a given ratio.
Divide BG in B in the given ratio, and
draw AX parallel to PB. PX will be the line
required.
ABB : ADG = the given ratio (VI. 1), and
APB = XPB (I. 37) ; therefore, Ac.

931 To divide the triangle ABG in a given ratio by a line

XY drawn parallel to any given line AE.
Make BB to BO in the given ratio. Then
make BY & mean proportional to BE and BB,
and draw YX parallel to EA.
Proof.AB divides ABO in the given ratio
(VI. 1). Now
ABE : XBY :: BE : BB,
or
therefore

(VI. 19)

:: ABE : ABB;
XBY = ABB.

'

932 If the interior and exterior vertical angles at P of the

triangle APB be bisected by straight lines which cut the base
in G and D, then the circle circumscribing GPD gives the
locus of the vertices of all triangles on the base AB whose
sides AP, BP are in a constant ratio.

MISCELLANEOUS PROPOSITIONS.

199

Proof.
The / GPD = i (APB+BPE)
= a right angle ;
therefore P lies on the circumference of
the circle, diameter CD (III. 31). Also
AP : PB :: AC : CB :: AD : BB
(VI. 3, and A.), a fixed ratio.
933 AD is divided harmonically in B and C ; i.e., AD : DB :: AO : OB ;
or, the whole line is to one extreme part as the other extreme part is to the middle
part. If we put a, b, c for the lengths AD, BD, CD, the proportion is
expressed algebraically by a : b :: a c : cb, which is equivalent to
i + T = ~a
b
e
AP:BP = OA:00=OC: OB
934 Also
and
AP3 : BI=OA: OB,
(VI. 19)
AP'-AC1 : CP2 : BI-BC\
(VI. 3, & B.)
935 If Q be the centre of the inscribed circle of the triangle
ABC, and if AQ produced meet the circumscribed circle,
radius B, in F; and if FOG be a diameter, and AD perpendi
cular to BG ; then
(i.) FC =FQ= FB = 2/i sin^-.
(ii.) z FAD=FA 0= (B- C),
and Z CAG = i(B+C).
Pboo? of (i.)
tFQC= QOA + QAC.
But QAC = QAB = BOF; (III. 21)
.-. FQC = FGQ;
.: FO = FQ.
Similarly FB = FQ.
Also / GOF is a right angle, and
FGO =FAG = \A; (III. 21)
.-.

F0 = 2B sin 4--

936 If B, r be the radii of the circumscribed and inscribed

circles of the triangle ABC (see last figure), and 0, Q the
centres ; then
OQa = R2 2Rr.
Proof. Draw QH perpendicular to AO ; then QH = r. By the isosceles
triangle AOF, OOJ = B2-AQ.QF (922, iii.), and QF = FG (935, i.), and
by similar triangles QFC, AQH, AQ : QH :: GF : FG ;
therefore
AQ.FG = GF.QH= 2Rr.

200

ELEMENTARY GEOMETRY.

The problems known as the Tangencies.

937 Given in position any three of the following nine data
viz., three points, three straight lines, and three circles,it is
required to describe a circle passing through the given points
and touching the given lines or circles. The following five
principal cases occur.
938 ! Given two points A, B, and the straight line CD.
Analysis.Let ABX be the required
circle, touching GD in X. Therefore
GX,= GA.GB.

(III. 36)

Hence the point X can be found, and the

centre of the circle defined by the inter
section of the perpendicular to GD through
X and the perpendicular bisector of AB.
There are two solutions.
Otherwise, by (926), making the ratio
one of equality, and DO the given line.
Cor.The point X thus determined is the point in CD at
which the distance AB subtends the greatest angle. In the
solution of (941) Q is a similar point in the circumference GD.
(III. 21, A L 16)
939 H. Given one point A and two straight lines DC, DE.
In the last figure draw AOG perpendicular to DO, the bisector of the
angle D, and make OB = OA, and this case is solved by Case I.
940 HI. Given the point P, the straight line DE, and the
circle ACF.
Analysis. Let PEF be the required
circle touching the given line in E and the
circle in F.
Through H, the centre of the given
circle, draw AHCD perpendicular to DE.
Let K be the centre of the other circle.
Join HK, passing through F, the point of
contact. Join AF, EF, and AP, cutting
the required circle in X. Then
DHF = LKF;
(1.27)
therefore HFA = KFE (the halves of equal
angles) ; therefore AF, FE are in the same
straight line. Then, because AX.AP = AF.AE,
(III. 36)
and A F. AE *= AG. AD by similar triangles, therefore AX can be found.
A circle must then be described through P and X to touch the given line,

THE PROBLEMS OF THE TANGENCIES.

201

by Case I. There are two solutions with exterior contact, as appears from
Case I. These are indicated in the diagram. There are two more in which
the circle A G lies within the described circle. The construction is quite
analogous, C taking the place of A.

941 IV. Given two points A, B and

the circle CD.
Draw any circle through A, B, cutting the
required circle in C, D. Draw AB and DC,
and let them meet in P. Draw I'Q to touch
the given circle. Then, because
PC . PD = PA . PB = PQ>, (in. 36)
and the required circle is to pass through
A, B ; therefore a circle drawn through A, B, Q
must touch PQ, and therefore the circle CD,
in Q (HI. 37), and it can be described by Case
I. There are two solutions corresponding to
the two tangents from P to the circle CD.

942 V. Given one point P, and two circles, centres A and B.

Analysis.Let PFO be the required circle touching the given ones in F

and 0. Join the centres QA, QB. Join FO, and produce it to cut the
circles in E and H, and the line of centres in 0. Then, by the isosceles
triangles, the four angles at E, F, G, H are all equal ; therefore AE, BG are
parallel, and so are AF, BH; therefore AO : BO :: AF : BE, and 0 is a
centre of similitude for the two circles. Again, Z HBK = 2B7LK, and
FAM=2FNM (III. 20) ; therefore FNM=HLK= RGK (III. 21) ; therefore the triangles OFN, OKG are similar; therefore OF. OG OK. ON;
therefore, if OP cut the required circle in X, OX.OP= OK. ON. Thus
the point X can be found, and the problem is reduced to Case IV.
Two circles can be drawn through P and X to touch the given circles.
One is the circle PFX. The centre of the other is at the point where EA
and HB meet if produced, and this circle touches the given ones in E and H.
943 An analogous construction, employing the internal centre of simili
tude ff, determines the circle which passes through P, and touches one given
circle externally and the other internally. See (1047-9).
The centres of similitude are the two points which divide the distance
between the centres in the ratio of the radii. See (1037).

2d

202

ELEMENTABT GEOMETRY.

944 Cor.The tangents from 0 to all circles which touch

the given circles, either both externally or both internally,
are equal.
For the square of the tangent is always equal to OK. ON or OL . OH.

945 The solutions for the cases of three given straight lines
or three given points are to be found in Euc. IV., Props. 4, 5.
946 In the remaining cases of the tangencies, straight lines
and circles alone are given. By drawing a circle concentric
with the required one through the centre of the least given
circle, the problem can always be made to depend upon one
of the preceding cases ; the centre of the least circle becoming
one of the given points.
947 Definition.A centre of similarity of two plane curves
is a point such that, any straight line being drawn through it
to cut the curves, the segments of the line intercepted between
the point and the curves are in a constant ratio.
948 If AB, AG touch a circle at B and
C, then any straight line AEDF, cutting
the circle, is divided harmonically by
the circumference and the chord of con
tact BG.
Proof from AE . AF = AT?.
AT? = BD.DG+AD*,
and
BD.DC=ED. DF.

(ILL 36)
(923)
(III. 35)

949 If > (3 > 7> hi the same figure, be the

perpendiculars to the sides of ABC from
any point E on the circumference of the
circle, then
fly = a2.
Proof.Draw the diameter BH=d ; then EI? = fid, because BEH is a
right angle. Similarly EG1 = yd. But EB .EG ad (VI. D.), therefore Ac.

950 If FE be drawn parallel to the base

BG of a triangle, and if EB, FG intersect
in 0, then
AE : AG :: EO : OB :: FO : OG.
By VI. 2. Since each ratio = FE : BG.
Coa.If AC = n . AE, then

BE= (n+l)OE.

MISCELLANEOUS PROPOSITIONS.

203

951 The three lines drawn from the angles of a triangle to

the middle points of the opposite sides, intersect in the same
point, and divide each other in the ratio of two to one.
For, by the last theorem, any one of these lines is divided by each of the
others in the ratio of two to one, measuring from the same extremity, and
mast therefore be intersected by them in the same point.
This point will be referred to as the centroid of the triangle.
952 The perpendiculars from the angles upon the opposite
sides of a triangle intersect in the same point.
Draw BE, CF perpendicular to the sides, and let
them intersect in 0. Let AO meet BC in D. Circles
will circumscribe AEOF and BFEC, by (III. 31) ;
therefore
Z FAO = FEO = FOB ;
(III. 21)
therefore Z BDA = BFC = a right angle ;
i.e., AO is perpendicular to BO, and therefore the
perpendicular from A on BO passes through 0.
0 is called the orthocentre of the triangle ABO.
Cob.The perpendiculars on the Bides bisect the angles
of the triangle DEF, and the point 0 is therefore the centre
of the inscribed circle of that triangle.
Proof.From (III. 21), and the circles circumscribing OEAF and OEGD.
953 If the inscribed circle of a triangle ABC touches the
sides a, b, c in the points D, E, F; and if the escribed circle
to the side a touches a and b, c produced in D' , E', F' ; and if
s = \(a + b + c) ;
then
BF = BD'= CD = s-c,
and AE' = AF = s;
and similarly with respect to
the other segments.
Proof.The two tangents from
any vertex to either circlebeingequal,
it foliows that CD+c = half the
perimeter of ABO, which is made np
of three pairs of equal segments ;
therefore
CB = s c.
Also
AV+AF = AC+ CD'+AB+BD
therefore AE' = AF' = i.

204

ELEMENTARY GEOMETRY.

The Nine-Point Circle.

954 The Nine-point circle is the circle described through
D, E, F, the feet of the perpendiculars on the sides of the
triangle ABC. It also passes through the middle points of
the sides of ABC and the middle points of OA, OB, OC ; in
all, through nine points.
Proof.Let the circle
cut the sides of ABO
again in O, H, K; and
OA, OB, 00 in L, M, N.
/.EMF=EDF (III. 21)
= 20DF (952, Cor.);
therefore, since OB is the
diameter of the circle cir
cumscribing OFBB (III.
31), M is the centre of
that circle (HI. 20), and
therefore bisects OB.
Similarly 00 and OA
are bisected at N and L.
Again, I MQB = MED (III. 22) = OCD, (III. 21), by the circle circum
scribing OEGD. Therefore MO is parallel to 00, and therefore bisects BO.
Similarly H and K bisect OA and AB.

955 The centre of the

nine-point circle is the
middle point of OQ, the
line joining the orthocentre and the centre
of the circumscribing
circle of the triangle
ABC.
For the centre of the N. P.
circle is the intersection of
the perpendicular bisectors
of the chords DG, EH, FK,
and these perpendiculars
bisect OQ in the same point
N, by (VI. 2).

956 The centroid of

the triangle ABC also lies on the line OQ and divides it in
B so that OB = 2BQ.
Pkoof.The triangles QBG, OAB are similar, and AB = 2HG ; there
fore AO = 2GQ ; therefore OB = 2BQ ; and AB = 2BG ; therefore B is the
centroid, and it divides OQ as stated (951).

CONSTRUCTION OF TRIANGLES.

205

957 Hence the line joining the centres of the circumscribed

and nine-point circles is divided harmonically in the ratio of
2 : 1 by the centroid and the orthocentre of the triangle.
These two points are therefore centres of similitude of the
circumscribed and nine-point circles; and any line drawn
through either of the points is divided by the circumferences
in the ratio of 2 : 1. See (1037.)
958 The lines BE, EF, FD intersect the sides of ABC in
the radical axis of the two circles.
For, if EF meets BC in P, then by the circle circumscribing BCEF,
PE.FF= PC.FB; therefore (III. 36) the tangents from P to the circles
are equal (985).

959 The nine-point circle touches the inscribed and escribed

circles of the triangle.
Proof.Let 0 be the orthocentre, and I, Q
the centres of the inscribed and circumscribed
circles. Produce AI to bisect the arc BC in T.
Bisect AO in L, and join GL, cutting AT in S.
The N. P. circle passes through G, D, and
L (954), and I) is a right angle. Therefore
GL is a diameter, and is therefore = R = QA
(957). Therefore GL and QA are parallel.
But QA = QT, therefore
GS = GT = CT sin 4 = 2E sin' A.

(935, i.)

Also
ST=2GScosO
(6 being the angle GST = GTS).
N being the centre of the N. P. circle, its
radius = NG = \R ; and r being the radius of
the inscribed circle, it is required to shew that
NI = NG-r.
Now
Nl* = SN' + SP-2SN. SI cos 0.
Substitute
8N=^R-G8;

(702)

SI = TI-8T = 2R sin 4 -2GS cos 0 ;

and GS = 2B sin' \A, to prove the proposition.
If / be the centre of the escribed circle touching BC, and r its radius, it
is shewn in a similar way that NJ = NQ+ra.

To construct a triangle from certain data.

960 When amongst the data we have the sum or difference
of the two sides AB, AG ; or the sum of the segments of the
base made by AG, the bisector of the exterior vertical angle ;
or the difference of the segments made by AF, the bisector of

206

ELEMENTARY GEOMETRY.

the interior vertical angle; the following construction will

lead to the solution.
Make AE = AD = AG. Draw
DK parallel to A F, and suppose
EK drawn parallel to AG to meet
the base produced in K; and
complete the figure. Then BE
is the sum, and DD is the differ
ence of the sides.
BK is the sum of the exterior
segments of the base, and BH is
the difference of the interior seg
ments. /. BBH = BEG = \A,
AABO = EAG = i(B + C),
/ BCB = \BFB = i (G-B).

961 When the base and the vertical angle .are given ; the
locus of the vertex is the circle ABC in figure (935) ; and the
locus of the centre of the inscribed circle is the circle, centre
F and radius FB. When the ratio of the sides is given,
see (932).
962 To construct a triangle when its form and the distances
of its vertices from a point A' are given.
Analysis.Let ABG be the required triangle. On
A'B make the triangle A'BG' similar to ABC, so tbat
AB : A'B :: CB : CB. The angles ABA', GBG' will
also be equal; therefore AB : BC :: AA' : CO, which
gives GC , since the ratio AB : BC is known. Hence
the point G is found by constructing the triangle
A'CC. Thus BC is determined, and thence the tri
angle ABC from the known angles.

963 To find the locus

of a point P, the tan
gent from which to a
given circle, centre A,
has a constant ratio to
its distancefrom a given
point B.
Let AK be the radius of
the circle, and p : q the
given ratio. On AB take
AG, a third proportional to
AB and AK, and make

ABiDB^f : q\
With centre D, and a radius

TANGENTS TO TWO CIRCLES.

equal to a mean proportional between DB and DC, describe a circle.

be the required locus.

207

It will

Proof.Suppose P to be a point on the required locus. Join P with

A, B, C, and D.
Describe a circle about PBC cutting AP in F, and another about ABF
cutting PB in G, and join AO and BF. Then
PIC = AP'-AK1 = AlA-BA . AG (by constr.) = AP'-PA . AF (III. 36)
= AP.PF (II. 2) =GP.PB (III. 36).
Therefore, by hypothesis,
y : q' = GP.PB : PB1 = GP : PB = AD : DB (by constr.) ;
therefore L DPG = PGA (VI. 2) = PFB (III. 22) = PCB (III. 21).
Therefore the triangles DPB, DCP are similar ; therefore DP is a mean pro
portional to DB and DC. Henco the construction.
964 Coe. If p = q the locus becomes the perpendicular
bisector of BG, as is otherwise showu in (1003).

965 To find the locus of a point P, the tangents from which

to two given circles shall have a given ratio. (See also 1036.)
Let A, B be the centres, a, b the radii
(o>i), and p : q the given ratio. Take c, so
that c : b = p : q, and describe a circle with
centre A and radius AN = Va1 c*. Find the
locus of P by the last proposition, so that
the tangent from P to this circle may have
the given ratio to PB. It will be the re
quired locus.
Proof. By hypothesis and construction,
2l_ _ PK*
ca
PIP + c* _ ylP'-ft' + c' _ ^P'-^Jv"'
</
PT1
ts
fr'+is
PPJ
BP1 '
Cor.Hence the point can be found on any curve from
which the tangents to two circles shall have a given ratio.

966 To find the locus of the point from which the tangents
to two given circles are equal.
Since, in (965), we have now p = q, and therefore c = b, the construction
simplifies to the following :
Take AN= v/(a,-6'), and in AB take AB : AN : AG. The perpen
dicular bisector of BG is the required locus. But, if the circles intersect,
then their common chord is at once the line required. See Radical Axis
(985).

208

ELEMENTARY GEOMETRY.

Collinear and Concurrent systems ofpoints and lines.

967 Definitions.Points lying in the same straight line are
collinear. Straight lines passing through the same point are
concurrent, and the point is called the focus of the pencil of
lines.
Theorem.If the sides of the triangle ABC, or the sides
produced, be cut by any straight line in the points a, b, o
respectively, the line is called a transversal, and the segments
of the sides are connected by the equation
968
(Ab : bC) (Ca : aB) {Be : cA) = 1.
Conversely, if this relation holds, the points a, b, c will be
collinear.
Proof. Through any vertex A draw AD
parallel to the opposite side BO, to meet the
transversal in D, then
Ab :bO = AD:Ca and Be : cA = aB : AD
(VI. 4), which proves the theorem.
Note.In the formula the segments of the
sides are estimated positive, independently of
direction, the sequence of the letters being pre
served the better to assist the memory. A point may be supposed to travel
from A over the segments Ab, bC, &c. continuously, until it reaches A again.
969 By the aid of (701) the above relation may be put in
the form
(sin ABb : sin bBC) (sin CAa : sin aAB) (sin BCc : sin cCA) = l
970 If 0 be any focus in the plane of the triangle ABC, and
if AO, BO, CO meet the sides in a, b, c; then, as before,
(Ab : bC) (Ca : aB) (Be : cA) = 1.
Conversely, if this relation holds, the lines Aa, Bb, Gc will
be concurrent.
Proof.By the trans
versal Bb to the triangle
AaC, we have (968)
(Ab : bO) (OB : Ba)
x (aO : OA) = 1.
And, by the transversal
Cc to the triangle AaB,
(Be : cA) (AO : Oa)
x (aC : CB) = 1.
Multiply these equations
together.

COLLINEAR AND CONCURRENT SYSTEMS.

209

971 If be, ca, ab, in the last figure, be produced to meet the
sides of ABC in P, Q, It, then each of the nine lines in the
figure will be divided harmonically, and the points P, Q, B
will be collinear.
Proof.(i.) Take bP a transversal to ABC; therefore, by (968),
(CP : PB) (Be : cA) (Ab :bC) = l;
therefore, by (970),
CP : PB = Ca : aB.
(ii.) Take CP a transversal to Abe, therefore
(AB : Be) (cP : Pb) (bC : CA) = 1.
But, by (970), taking 0 for focus to Abe,
(AB : Be) (cp : pb) (bC : CA) = 1 j
therefore
cP : Pb = cp : pb.
(iii.) Take PC a transversal to AOc, and b a focus to AOc; therefore, by
(908 & 970),
(Aa : aO) (OC : Co) (cB : BA) = 1,
and
(Ap : pO) (OC : Ce) (cB : BA) = 1 ;
therefore
Aa : aO = Ap : pO.
Thus all the lines are divided harmonically.
(iv.) In the equation of (970) put Ab : bC = AQ : QC the harmonic
ratio, and similarly for each ratio, and the result proves that P, Q, R are
collinear, by (968).
Cor.If in the same figure qr, rp, pq be joined, the three
lines will pass through P, Q, B respectively.
Peoof.Take 0 as a focus to the triangle abe, and employ (970) and the
harmonic division of be to show that the transversal rq cuts be in P.

972 If a transversal intersects the sides AB, BG, CD, &c.

of any polygon in the points a, b, c, &c. in order, then
(Aa : aB) (Bb : bC) (Cc : cD) (Dd : dE) ... &c. = 1.
Proof.Divide the polygon into triangles b}' lines drawn from one of the
angles, and, applying (968) to each triangle, combine the results.

973 Let any transversal cut the sides of a triangle and

their three intersectors AO, BO, CO (see figure of 970) in the
points A', B1', (J, a, b', c, respectively ; then, as before,
(Ab' : b'C) (Ca : a'B) (Be' : c'A') = 1.
Proof.Each side forms a triangle with its intersector and the trans
versal. Take the four remaining lines in succession for transversals to each
triangle, applying (968) symmetrically, and combine the twelve equations.
2 E

210

ELEMENTARY GEOMETRY.

974 If the lines joining corresponding vertices of two tri

angles ABC, abc are concurrent, the points of intersection
of the pairs of corresponding
sides are collinear, and con
versely.
Proof. Let the concurrent lines
Aa, Bb, Cc meet in 0.
Take be,
ca, ah transversals respectively to
the triangles OBG, OCA, OAB, ap
plying (908), and the product of the
three equations shows that P, R, Q
lie on a transversal to A BC.
-j-

975 Hence it follows that, if the lines joining each pair of

corresponding vertices of any two rectilineal figures are con
current, the pairs of corresponding sides intersect in points
which are collinear.
The figures in this case are said to be in perspective, or in
homology, with each other. The point of concurrence and
the line of collinearity are called respectively the centre and
axis of perspective or homology. See (1083).
976 Theorem:When three perpendiculars to the sides of a
triangle ABC, intersecting them in the points a, b, c respec
tively, are concurrent, the following relation is satisfied ; and
conversely, if the relation be satisfied, the perpendiculars are
concurrent.

Abt-bC2+Cai-aB2+Bci-cAi = 0.
Proof. If the perpendiculars meet in 0, then Ab* bCm = AOiOC1,
Ac. (I. 47).
Examples.By the application of this theorem, the concurrence of the
three perpendiculars is readily established in the following case3:
(1) When the perpendiculars bisect the sides of the triangle
(2) When they pass through the vertices. (By employing I. 47.)
(3) The three radii of the escribed circles of a triangle at the points of
contact between the vertices are concurrent. So also are the radius of the
inscribed circle at the point of contact with one side, and the radii of the two
escribed circles of the remaining sides at the points of contact beyond the
included angle.
In these cases employ the valnes of the segments ffiven in (953).
(4) The perpendiculars equidistant from the vertices with three con
current perpendiculars are also concurrent.
(5) When the three perpendiculars from the vertices of one triangle upon
the sides of the other are concurrent, then the perpendiculars from the
vertices of the second triangle upon the sides of the first are also concurrent.
Proof. If A, B, 0 and A', B", C are corresponding vertices of the tri
angles, join AB', AC, BC, BA', CA', CB", and apply the theorem in conjunc
tion with (I. 47).

TRIANGLES CIRCUMSCRIBING A TRIANGLE.

211

Triangles of constant species circumscribed to a triangle.

977 Let ABG be any triangle, and 0 any point ; and let
circles circumscribe AOB, BOG, COA. The circumferences
will be the loci of the vertices of a triangle of constant form
whose sides pass through the
points A, B, 0.
Proof.Draw any line hAc from circle
to circle, and produce bC, cB to meet in a.
The angles AOB, COA are supplements
of the angles c and b (III. 22) ; therefore
BOG is the supplement of a (I. 32) ; there
fore a lies on the circle OBC. Also, the
angles at 0 being constant, the angles a, b, o
are constant.

978 The triangle abc is a maximum when its sides are per
pendicular to OA, OB, OC.
Proof.The triangle is greatest when its sides are greatest. But the
sides vary as Oa, Ob, Oc, which are greatest when they are diameters of tha
circles; therefore &c, by (ILL 31).

979 To construct a triangle of given species and of given

limited magnitude which shall have its sides passing through
three given points A, B, G.
Determine 0 by describing circles on the sides of ABG to contain angles
eqnal to the supplements of the angles of the specified triangle. Construct
the figure abcO independently from the known sides of abc, and the now
known angles ObC = OAC, OaG OBG, &c. Thus the lengths Oa, Ob, Oc
are found, and therefore the points a, b, c, on the circles, can be determined.
The demonstrations of the following propositions will now be obvious.

Triangles of constant species inscribed to a triangle.

980 Let abc, in the last figure, be a fixed triangle, and 0
any point. Take any point A on be, and let the circles cir
cumscribing OAc, OAb cut the other sides in B, G. Then
ABG will be a triangle of constant form, and its angles will
have the values A . Oba + Oca, &c. (III. 21.)
981 The triangle ABG will evidently be a minimum when
OA, OB, OC are drawn perpendicular to the sides of abc.
982 To construct a triangle of given form and of given
liinited magnitude having its vertices upon three fixed lines
be, ca, ab.

212

ELEMENTARY GEOMETRY.

Construct the figure ABGO independently from the known sides of ABC
and the angles at 0, which are equal to the supplements of the given angles
o, b, c. Thus the angles OAC, &c are found, and therefore the angles ObC,
&c, equal to them (III. 21), are known. From these last angles the point (J
can be determined, and the lengths OA, OB, 00 being known from the inde
pendent figure, the points A, B, C can be found.
Observe that, wherever the point 0 may be taken, the angles AOB, BOO
COA are in all cases either the supplements of, or equal to, the angles c, a, b
respectively; while the angles aOb, bOc, cOa are in all cases equal to Cdtzc,

Aa, Bb.
983 Note.In general problems, like the foregoing, which,
admit of different cases, it is advisable to choose for reference
a standard figure which has all its elements of the same affec
tion or sign. In adapting the figure to other cases, all that
is necessary is to follow the same construction, letter for
letter, observing the convention respecting positive and
negative, which applies both to the lengths of lines and to
the magnitudes of angles, as explained in (607609).
Radical Axis.
984 Definition.The radical axis of two circles is that
perpendicular to the line of centres which divides the dis
tance between the centres into segments, the difference of
whose squares is equal to the difference of the squares of the
radii.
Thus, A, B being the centres, a, b the radii, and IP the
the radical axis,
APBP = ai bi.

985 It follows that, if the circles intersect, the radical axis

is their common chord ; and that, if they do not intersect, the
radical axis cuts the line of centres in a point the tangents
from which to the circles are equal (I. 47).
To draw the axis in this case, see (966).

RADICAL AXTS.

213

Otherwise : let the two circles cut the line of centres in C, D and C, D'
respectively. Describe any circle through C and D, and another through
C and I/, intersecting the former in E and F. Their common chord EF
will cut the central axis in the required point I.
Proof. IC .ID = IE . IF = IC. ID" (III. 36) ; therefore the tangents
from I to the circles are equal.

986 Theorem.The difference of the squares of tangents

from any point P to two circles is equal to twice the rectangle
under the distance between their centres and the distance of
the point from their radical axis, or
PK2-PTi = 2AB.PN.

Proof.

PK'-PT' = (AF'-BP1) - (a'-f) = (Atf-Btf) - (AP-BP), '

by (I. 47) & (984). Bisect AB in G, and substitute for each difference of
squares, by (II. 12).

987

Cok. 1.If P be on the circle whose centre is B, then

PK2 = 2AB . PN.

988 Coe. 2.If two chords be drawn through P to cut the

circles in X, X, Y, F; then, by (III. 36),
PX.PX'-PY.PY' =2AB.PN.
989 If a variable circle intersect two given circles at con
stant angles a and /3, it will intersect their radical axis at a
constant angle ; and its radius will bear a constant ratio to
the distance of its centre from the radical axis. Or
PN:PX= a coso-6 cos j8 : AB.

ELEMENTARY GEOMETRY.

214

Proof.In the same figure, if P be the centre of the variable circle, and
if PX=PY be its radius; then, by (988),
PX (XX'- YY') = 2AB . PN.
But
XX" = 2a cos n and YY' = 26 cos ft ;
therefore
PN : PX = a cos a b cos/3 : AB,
which is a constant ratio if the angles u, ft are constant.
990 Also PX : PX the cosine of the angle at which the
circle of radius PX cuts the radical axis. This angle is
therefore constant.
991 Cob.A circle which touches two fixed circles has its
radius in a constant ratio to the distance of its centre from
their radical axis.
This follows from the proposition by making a = ft = 0 or 2jt.
If P be on the radical axis ; then (see Figs. 1 and 2 of 984)
992

(i-) The tangents from P to the two circles are equal,

or
993

PK = PT.
(ii-)

(986)

The rectangles under the segments of chords

through P are equal, or PX . PX' = PY.PY'.

(988)

994 (iii.) Therefore the four points X, X', Y, Y" are concyclic (III. 36); and, conversely, if they are concyclic, the
chords XX', YY' intersect in the radical axis.
995 Definition.Points which he on the circumference of
a circle are termed concyclic.
996 (iv.) If P be the centre, and if PX = PY be the radius
of a circle intersecting the two circles in the figure at angles
a and (5; then, by (993), XX' =-YY', or a cos, a = b cos j3 ;
that is, The cosines of the angles of intersection are inversely
as the radii of the fixed circles.
997 The radical axes of three circles (Fig. 1046), taken two
and two together, intersect at a point called their radical centre.
Proof.Let .4, B, Cb - t ic centres, a, b, c the radii, and X, Y, Z the points
in which the radical axes cut BG, CA, AB. Write the equation of the defini
tion (984) for each pair of circles. Add the results, and apply (970).
998 A circle whose centre is the radical centre of three
other circles intersects them in angles whose cosines are
inversely as their radii (996).

INVERSION.

215

Hence, if this fourth circle cuts one of the others or

thogonally, it cuts them all orthogonally.
999 The circle which intersects at angles a, /3, y three fixed
circles, whose centres are A, B, G and radii a, b, c, has its
centre at distances from the radical axes of the fixed circles
proportional to
b cos /3 c cos 7 C COS y a COS a a cos a b cos |3
CA
~'
BG
'
And therefore the locus of its centre will be a straight line
passing through the radical centre and inclined to the three
radical axes at angles whose sines are proportional to these
fractions.
Peoof.The result is obtained immediately by writing out equation (989)
for each pair of fixed circles.

The Method of Inversion.

1000 Definitions. Any two points P, P*, situated on a
diameter of a fixed circle
whose centre is 0 and radius
4, so that OP.OP=, are
called inverse points with re
spect to the circle, and either
point is said to be the inverse
of the other. The circle and
its centre are called the circle
and centre of inversion, and
h the constant of inversion.
1001 If every point of a plane figure be inverted with
respect to a circle, or every point of a figure in space with
respect to a sphere, the resulting figure is called the inverse
or image of the original one.
Since OP :k: OF, therefore
1002
OP : OP' = OP2 : k3 = ft* : OP'\
1003 Let D, V , in the same figure, be a pair of inverse
points on the diameter OD . In the perpendicular bisector of
DH, take any point Q as the centre of a circle passing through
D, D', cutting the circle of inversion in B, and any straight
line through 0 in the points P, P. Then, by (III. 36),
OP . OP' = OD . 01/ = 0BZ (1 000) . Hence

216

ELEMENTARY GEOMETRY.

1004 (i-) P, P are inverse points; and, conversely, any

two pairs of inverse points lie on a circle.
1005 (") The circle cuts orthogonally the circle of inver
sion (III. 37) ; and, conversely, every circle cutting another
orthogonally intersects each of its diameters in a pair of
inverse points.
1006 (iii.) The line IQ is the locus of a point the tangent
from which to a given circle is equal to its distance from a
given point D.
1007 Def.The line IQ is called the axis of reflexion for
the two inverse points D, J7, because there is another circle
of inversion, the reflexion of the former, to the right of IQ,
having also D, D' for inverse points.
1008 The straight lines drawn from any point P, within
or without a circle (Figs. 1 and 2), to the extremities of any
chord AB passing through the inverse point Q, make equal
angles with the diameter through PQ. Also, the four points
0, A, B, P are concyclic, and QA . QB = QO . QP.

(0

Proof. In either figure OP : OA : OQ and Or : OB : OQ (1000),

therefore, by similar triangles, L OPA = OAB and OPB = OBA in figure
(1) and the supplement of it in figure (2). But OAB = OBA (I. 5), there
fore, &c.
Also, because l OPA = OBA, the four points 0. A, B, P lie on a circle in
each case (III. 21), and therefore QA . QB = QO . QP (III. 35, 36).

1009 The inverse of a circle is a circle, and the centre of

inversion is the centre of similitude of the two figures. See
also (1037).
Proof.In the figure of (1043), let O be the point where the common
tangent BT of the two circles, centres A and B, cuts the central axis, and let
any other line through 0 cut the circles in P, Q P\ Q'. Then, in the demon
stration of (942), it is shown that 0P0Q'= OQ.OP =k3, a constant
quantity. Therefore either circle is the iuverse of the other, h being the
radius of tho circle of inversion.

INVERSION.

217

1010 To make the inversions of two given circles equal

circles.
Rule. TaJce the centre of inversion so that the squares of
the tangents from it to the given circles may be proportional to
their radii (965).
Proof. (Fig. 1043) AT: BR=OT: OR=OT : Jc\ since OT : h : OR.
Therefore OT' : AT = Ar1 : BR, therefore BR remains constant if OT1 oc AT.

1011 Hence three circles may be inverted into equal circles,

for the required centre of inversion is the intersection of two
circles that can be drawn by (965).
1012 The inverse of a straight line is a circle passing
through the centre of inversion.
Proof.Draw OQ perpendicular to the
line, and take P any other point on it. Let
Q, P be the inverse points. Then OP . OP=
OQ . OQ ; therefore, by similar triangles,
L OPQ = OQP, a right angle ; and OQ' is
constant, therefore the locus of P is the
circle whose diameter is OQ".
1013 Example.The inversion of a poly
gon produces a figure bounded by circular
arcs which intersect in angles equal to the
corresponding angles of the polygon, the
complete circles intersecting in the centre
of inversion.

1014 If the extremities of a straight line PQ in the last

figure are the inversions of the extremities of PQ, then

PQ:FQ=V{OP.OQ) : S(OF.OQ).
Proof.By similar triangles, PQ : PQ = OP : OQ and PQ : PQ =
OQ : OP. Compound these ratios.

1015 From the above it follows that any homogeneous

equation between the lengths of lines joining pairs of points
in space, such as PQ .RS.TU= PR. QT . SU, the same
points appearing on both sides of the equation, will be
true for the figure obtained by joining the corresponding
pairs of inverse points.
For the ratio of each side of the equation to the corresponding side of the
equation for the inverted points will bo the same, namely,
^{OP.OQ.OR ...) : V{OP .OQ.OIf ...).

2p

218

ELEMENTARY GEOMETBT.

Pole and Polar.

1016 Definition.The polar of any point P with respect
to a circle is the perpendicular to the diameter OP (Fig. 1012)
drawn through the inverse point P.
1017 It follows that the polar of a point exterior to the
circle is the chord of contact of the tangents fron the point;
that is, the line joining their points of contact.
1018 Also, PQ' is the polar of P with respect to the circle,
centre 0, and PQ is the polar of Q'. In other words, 0iy
point P lying on the polar of a point Q', has its own polar
always passing through Q'.
1019 The lino joining any two points P, p is the polar of
Q\ the point of intersection of their polars.
Proof.The point Q' lies on both the lines PQ', p'Q'. and therefore has
its polar passing through the pole of each line, by the last theorem.

1020 The polars of any two points P,p, and the line joining
the points form a self-reciprocal triangle with respect to
the circle, the three vertices being the poles of the opposite
sides. The centre 0 of the circle is evidently the ortbocentre
of the triangle (952). The circle and its centre are called the
polar circle and polar centre of the triangle.
If the radii of the polar and circumscribed circles of a
triangle ABC be r and R, then
r* = 4/i2 cos A cos B cos C.
Proof.In Fig. (952), 0 is the centre of the polar circle, and the circles
described round ABO, BOG, COA, AOB are all equal ; because the angle
BOO is the supplement of A ; Ac. Therofore 21t . OD = OB. 00 (VI. C)
and r8 = OA . OB = OA .OB. OC + 2E. Also, OA = 2 R cos.-l by a diameter
through B, and (III. 21).

Coaxal Circles.
1021 Definition. A system of circles having a common
line of centres called the central axis, and a common radical
axis, is termed a coaxal system.
1022

If 0 be the variable centre of one of the circles, and

COAXAL CIRCLES.

219

OK its radius, the whole system is included in the equation

OF-OK2= 82,
where 8 is a constant length.

1023 In the first species (Fig. 1),

OP- OK2 = 82,
and ? is the length of the tangent from I to any circle of the
system (985). Let a circle, centre I and radius 8, cut the
central axis in D, D'. When 0 is atD or If, the circle whose
radius is OK vanishes. When 0 is at an infinite distance,
the circle developes into the radical axis itself and into a fine
at infinity.
The points D, D' are called the limiting points.
1024 In the second species (Fig. 2),
OR*-OP = S2,
and S is half the chord BB' common to all the circles of
the system. These circles vary between the circle with
centre I and radius 8, and the circle with its centre at infinity
as described above. The points B, B' are the common points
of all circles of this system. The two systems are therefore
distinguished as the limiting points species and the common
points species of coaxal circles.

220

ELEMENTARY GEOMETRY.

1025 There is a conjugate system of circles having R, R1 for

limiting points, and D, D' for common points, and the circles
of one species intersect all the circles of the conjugate system
of the other species orthogonally (1005).
Thus, in figures (1) and (2), Q is the centre of a circle
of the opposite species intersecting the other circles or
thogonally.
1026 In the first species of coaxal circles, the limiting
points D, D are inverse points for every circle of the system,
the radical axis being the axis of reflexion for the system.
Proof. (Fig. 1)
OP-l* = OB?,
therefore
OD . OB1' = OK\
therefore D, If are inverse points (1000).

(II. 13)

1027 Also, the points in which any circle of the system

cuts the central axis are inverse points for the circle whose
centre is / and radius 8.
[Proof. Similar to the last.
1028 Problem.Given two circles of a coaxal system, to
describe a circle of the same system(i.) to pass through a
given point; or (ii.) to touch a given circle ; or (iii.) to cut a
given circle orthogonally.
1029 I. If the system be of the common points species, then, since the
required circle always passes through two known points, the first and second
cases fall under the Tangencies. See (941).
1030 To solve the third case, describe a circle through the given common
points, and through the inverse of either of them with respect to the given
circle, which will then be cut orthogonally, by (1005).
1031 II. If the system be of the limiting points species, the problem is
solved in each case by the aid of a circle of the conjugate system. Such a
circle always passes through the known limiting points, and may be called a
conjugate circle of the limiting points system. Thus,
1032 To solve case (i.) Draw a conjugate circle through the given point,
and the tangent to it at that point will be the radius of the required circle.
1033 To solve case (ii.) Draw a conjugate circle through the inverse
of either limiting point with respect to the given circle, which will thus be
cut orthogonally, and the tangent to the cutting circle at either point of
intersection will be the radius of the required circle.
1034 To solve case (iii.) Draw a conjugate circle to touch the given one,
and the common tangent of the two will be the radius of the required circle.
1035 Thus, according as we wish to make a circle of the system touch, or
cut orthogonally, the given circle, wo must draw a conjugate circle to cut
ortlwgonally, or touch it.

CENTRES AND AXES OF SIMILITUDE.

221

1036 If three circles be coaxal, the squares of the tangents

drawn to any two of them from a point on the third are in
the ratio of the distances of the centre of the third circle from
the centres of the other two.
Peoof.Let A, B, C be the centres of the circles ; PK, PT the tangents
from a point P on the circle, centre G, to the other two ; PN the perpen
dicular on the radical axis. By (986),
PK1 = 2AO . PN and PT1 = 2BG . PN,
therefore
PK1 : PT' = AC : BC.

Centres and axes of similitude.

1037 Definitions. Let OO be the centres of similitude
(Def. 947) of the two circles in the figure below, and let any
hne through 0 cut the circles in P, Q, F, Q. Then the
constant ratio OP : OF = OQ : OQ' is called the ratio of
similitude of the two figures ; and the constant product
OP . OQ' = OQ . OF is called the product of anti-similitude.
See (942), (1009), and (1043).
The corresponding points P, F or Q, Q on the same
straight line through 0 are termed homologous, and P, Q' or
Q, P are termed anti-homologous.
1038 Let any other line Opqp'q be drawn through 0.
Then, if any two points P, p on the one figure be joined, and
if P,p', homologous to P, p on the other figure, be also joined,
the lines so formed are termed homologous. But if the points
which are joined on the second figure are anti-homologous to
those on the first, the two fines are termed anti-homologous.
Thus, Pq, Qp' are anti-homologous lines.
1039 The circle whose centre is O, and whose radius is
equal to the square root of the product of anti-similitude, is
called the circle of anti-similitude.

1040 The four pairs of homologous chords Pp and Fp',

Qq and QJq, Pq and Fq, Qp and Qp' of the two circles in the
figure are parallel. And in all similar and similarly situated
figures homologous fines are parallel.
Peoof.By (VI. 2) and the definition (947).

222

ELEMENTARY GEOMETRY.

1041 The four pairs of anti-homologous chords, Pp and

Qq, Qq and Pp, Pq and Q'p, Qp and Pq, of the two circles
meet on their radical axis.
Proof.
OP.OQf= Op. Oq =
where h is the constant of inversion ; therefore P, p, Q', q are concyclic ;
therefore Pp and Q'q' meet on the radical axis. Similarly for any other pair
of anti-homologous chords.
1042 Cor. From this and the preceding proposition it
follows that the tangents at homologous points are parallel ;
and that the tangents at anti-homologous points meet on the
radical axis. For these tangents are the limiting positions of
homologous or anti-homologous chords.
(1160)
1043 Let C, D be the inverse points of 0 with respect to
two circles, centres A and B ; then the constant product of
anti-similitude
OP . OQ or OQ . OP = OA.OD or OB . OC.

CENTRES AND AXES OF SIMILITUDE.

Proof.By similar right-angled triangles,

OA: OT: OC and OB: OR: OD;
therefore
OA . OD = OB . OC
and also
OA . OC = OT = OP . OQ,
and
OB.OD = OR1 = OP . 0$ ;
therefore
OA . OB . OC . OD = OP . OQ . OP" . OQ,
therefore kc, by (1).

223

0),
(III. 36)

1044 The foregoing definitions and properties (1037 to

1043), which have respect to the external centre of similitude
0, hold good for the internal centre of similitude 0', with the
usual convention of positive and negative for distances
measured from 0' upon lines passing through it.
1045 Two circles will subtend equal angles at any point on
the circumference of the circle whose diameter is 00', where
0, 0' are the centres of similitude (Fig. 1043). This circle is
also coaxal with the given circles, and has been called the.
circle of similitude.
Proof.Let A, B be the centres, a, b the radii, and K any point on the
circle, diameter 00'. Then, by (932),
KA : KB = AO : BO = AO' : BO' = a : b,
by the definition (943) ;
therefore
a : KA = b : KB ;
that is, the sines of the halves of the angles in qnestion are eqnal, which
proves the first part. Also, because the tangents from K are in the constant
ratio of the radii a, b, this circle is coaxal with the given ones, by (1036, 934).
1046 The six centres of similitude P, p, Q, q, B, r of three
circles lie three and three on four straight lines PQB, Pqr,
Qpr, Bpq, called axes
of similit ude.
Proof. Taking any
three of the sets of points
named, say P, q, r, they are
Bhewn at once to ,be colliuear by the transversal
theorem (968) applied to
the triangle ABC.
For the segments of its
rides made by the points
P, q, r are in the ratios of
the radii of the circles.

224

ELEMENTARY GEOMETRY.

1047 From the investigation in (942), it appears that one

circle touches two others in a pair of anti-homologous points,
and that the following rule obtains :
Kur/E. The right line joining the points of contact passes
through the external or internal centre of similitude of the two
circles according as the contacts are of the same or of different
hinds.
1048 Definition. Contact of curves is either internal or external ac
cording as the curvatures at the point of contact are in the same or opposite
directions.

1049 Gergonne's method of describing the circles which, touch

three given circles.
Take Par, one of the four axes of similitude, and find its poles a, /3, y
with respect to the given circles, centres A, B, 0 (1016). From 0, the
radical centre, draw lines through a, /3, y, cutting the circles in a, a', b, b',
c, c. Then a, b, c and a', b', c will be the points of contact of two of the
required circles.
Pkoof.Analysis.Let the circles
E, F touch the circles A, B, C in
a, b, c, a', b', c. Let be, b'c meet in
P ; ca, cV in q ; and ab, a'V in r.
Regarding E and F as touched by
A, B, G in turn, Rule (1047) shews
that are', bb', cc meet in O, the centre
of similitude of E and F ; and (1041)
shews that P, q, and r lie on the
radical axis of E and F.
Regarding B and G, or G and A,
or A and B, as touched by E and F
in turn, Rule (1047) shews that P, q, r
are the centres of similitude of B and
C, G and A, A and B respectively;
and (1041) shews that O is on the radical axis of each pair, and is therefore
the radical centre of A, B, and 0.
Again, because the tangents to E and F, at the anti-homologous points
a, a, meet on Pqr, the radical axis of E and F (1042) ; therefore the point
of meeting is the pole of aa' with respect to the circle A (1017). Therefore
ad passes through the pole of the line Pqr (1018). Similarly, bb' and co
pass through the poles of the same line Pqr with respect to B and C. Hence
the construction.

1050 In the given configuration of the circles A, B, G, the

demonstration shews that each of the three internal axes of
similitude Pqr, Qrp, Bpq (Fig. 1046) is a radical axis and
common chord of two of the eight osculating circles which,
can be drawn. The external axis of similitude PQR is the

ANHARMONIC RATIO.

225

radical axis of the two remaining circles which touch A, B,

and G either all externally or all internally.
1051 The radical centre 0 of the three given circles is also
the common internal centre of similitude of the four pairs of
osculating circles. Therefore the central axis of each pair
passes through 0, and is perpendicular to the radical axis.
Thus, in the figure, EF passes through 0, and is perpen
dicular to Pqr.
Anharmonic Ratio.
1052 Definition.Let a pencil of
four lines through a point 0 be cut
by a transversal in the points A, B,
C, D. The anharmonic ratio of the
pencil is any one of the three frac
tions
AB .CD
AD.BC

or

AB. CD
AC.BD

,..
AC.BD'

1053 The relation between these three different ratios is

obtained from the equation
AB . CD+AD .BC = AC. BD.
Denoting the terms on the left side by p and q, the three anharmonic
ratios may be expressed by

p : q, p:p+q, q:p+qThe ratios are therefore mutually dependent. Hence, if the identity
merely of the anharmonic ratio in any two systems is to be established, it is
immaterial which of the three ratios is selected.
1054 In future, when the ratio of an anharmonic pencil {0, ABCD} is
mentioned, the form AB . CD : AD . BG will be the one intended, whatever
the actual order of the points A, B, C, D may be. For, it should be observed
that, by making the line OD revolve about 0, the ratio takes in turn each of
the forms given above. This ratio is shortly expressed by the notation
{0, ABCD}, or simply {ABCD}.
1055 If the transversal be drawn parallel to one of the lines, for instance
OD, the two factors containing D become infinite, and their ratio becomes
unity. They may therefore be omitted. The anharmonic ratio then reduces
to AB : BC. Thus, when D is at infinity, we may write
{0, ABC oo} = AB-.BC.

1056

The anharmonic ratio

ABCD
sin A OB sin COD
AD.BC
sin AOD sin BOC'
2g

226

ELEMENTARY GEOMETRY.

and its value is therefore the same for all transversals of the
pencil.
Proof.Draw OR parallol to the transversal, and let p be the perpendi
cular from A upon OR. Multiply each factor in the fraction by p. Then
substitute p.AB= OA.OB sin ,405, Ac. (707).
1057 The anharmonic ratio (1050) becomes harmonic when its value is
unity. See (933). The harmonic relation there defined may also be stated
thus : four points divide a liue harmonically when the product of the extreme
segments is equal to the product of the whole line and the middle segment.

Homographic Systems of Points.

1058 Definition. If x, a, b, c be the distances of one
variable point and three fixed points on a straight line from a
point O on the same ; and if x, a, b', c be the distances of
similar points on another line through 0 ; then the variable
points on the two lines will form two homographic systems
when they are connected by the anharmonic relation
1Q5Q
(at a) (bc) _ (x'a) (b'e)^
(<r c) (ab)
{x c) (a'b)'
Expanding, and writing A, B, G, D for the constant coeffi
cients, the equation becomes
1060
Axx + Bx+ Cx+D = 0.
From which

1061

~|S. and *=-44

1062 Theoeem.Any four arbitrary points x^, x2, xs, xt on

one of the lines will have four corresponding points x\, x.z, a%, x\
on the other determined by the last equation, and the two sets
ofpoints will have equal anharmonic ratios.
Proof. This may be shown by actual substitution of the value of each as
in terms of a;', by (1001), in the harmonic ratio {a;1;,.TJ.T<}.

1063 If the distances of four points on a right line from a

point 0 upon it, in order, are a, a', /3, /3 , where a, /3 ; a, /3' are
the respective roots of the two quadratic equations
aa-2+2^+6 = 0, aV+2/t'*+6' = 0;
the condition that the two pairs of points may be harmonically
conjugate is
1064
ab'+a'b = 2M\

INVOLUTION.

227

Proof.The harmonic relation, by (1057), is

(a-o-) 03-/3') = (-/3-)('-i3).
Multiply ont, and substitute for the sums and products of the roots of tho
quadratics above in terms of their coefficients by (51, o2).
1065 If
ut be the quadratic expressions in (1063) for two pairs of
points, and if u represent a third pair harmonically conjugate with w, and u
then the pair of points u will also be harmonically conjugate with every pair
given by the equation u, + Att, = 0, where \ is any constant. For the con
dition (1064) applied to the last equation will be identically satisfied.
Involution.
1066 Definitions.Pairs of inverse points PF, QQ', &c, on
the same right line, form a system in involution, and the rela
tion between them, by (1000), is
op .of = oq. oq = &c. = tf.
A
I

0
I

P Q
I I

b
I

q
I

P
I

The radius of the circle of inversion is k, and the centre

0 is called the centre of the system. Inverse points are also
termed conjugate points.
When two inverse points coincide, the point is called a
fonts.
1067 The equation OF = l shows that there are two foci
A, B at the distance h from the centre, and on opposite sides
of it, real or imaginary according as any two inverse points
lie on the same side or on opposite sides of the centre.
1068 If the two homographic systems of points in (1058)
be on the same line, they will constitute a system in involu
tion when B = C.
Pkoof.Equation (1060) may now be written
Ax3t' + H(x + *>')+B = 0,

a constant.

Therefore

is the distance of the origin 0 from the centre

A.
of inversion. Measuring from this centre, the equation becomes IX
representing a system in involution.
1069 Any four points whatever of a system in involution on
a right line have their anharmonic ratio equal to that of their
four conjugates.

228

ELEMENTARY GEOMETRY.

Proof.Let p, p'\ q, q; r, /; *, s be the distances of the pairs of inverse

points from the centre.
In the anharmonic ratio of any fonr of the points, for instance {pq'rs},
substitute p = i'-f-p', q = k*-i-q, &c, and the result is the anharmonic
ratio {p'qr's}.

1070 Any two inverse points P, P are in harmonic relation

with the foci A, B.
A

Proof. Let p, p be the distances of P, P from the centre 0 ; then,

pp = *, therefore

^ = ,
k
p

that is,

therefore | = ^2 .
p k
kp
# = #

<933>

1071 If a system of points in involution be given, as in

(1068), by the equation
Axx'+H(x+x)+B = 0
(1);
and a pair of conjugate points by the equation
ax* + 2hx + b = 0
(2) ;
the necessary relation between a, h, and b is
1072
Ab+Ba = 2Hh.
Proof.The roots of equation (2) must be simultaneous values of x, x in
(1) ; therefore substitute in (1)
x+z =
a

and xz = .
a

(51)

1073 Cor. A system in involution may be determined from two given

pairs of corresponding points.
Let the equations for these points be
aaj*+2Aa + 6 = 0 and oV + 2hx + V = 0.
Then there are two conditions (1072),
Ab + Ba = 2Hli and AV + Be! = 2Hh\
from which A, H, B can be found.
A geometrical solution is given in (985). G, D ; C, V are, in that con
struction, pairs of inverse points, and I is the centre of a system in involution
defined by a series of coaxal circles (1022). Each circle intersects the
central axis in a pair of inverse points with respect to the circle whose centre
is 0 and radius c.
1074: The relations which have been established for a system of eollinear
points may be transferred to a system of concurrent lines by the method of
(1056), in which the distance between two points corresponds to the sine of
the angle between two lines passing through those points.

METHODS OF PBOJECTIOK

229

The Method of Projection.

1075 Definitions.The projection of any point P in space
(Fig. of 1079) is the pointy in which a right line OP, drawn
from a fixed point 0 called the vertex, intersects a fixed plane
called the plane of projection.
If all the points of any figure, plane or solid, be thus pro
jected, the figure obtained is called the projection of the
original figure.
1076 Projective Properties.The projection of a right line is
a right line. The projections of parallel lines are parallel. The
projections of a curve, and of the tangent at any point of it, are
another curve and the tangent at the corresponding point.
1077 The anharmonic ratio of the segments of a right line
is not altered by projection ; for the line and its projection are
but two transversals of the same anharmonic pencil. (1056)
1078 Also, any relation between the segments of a line
similar to that in (1015), in which each letter occurs in every
term, is a projective property.
[Proof as in (1056).
1079 Theorem.Any quadrilateral PQRS may be projected
into a parallelogram.
Construction.
Produce PQ, SB to
meet in A, and PS,
QE to meet in B.
Then, with any
point O for vertex,
project the quadri
lateral npon anyplane pah parallel to
OAB. The projected
figure pqrs will be a
parallelogram.
Proof. The
planes OPQ, OBS in
tersect in OA, and
they intersect the
plane of projection
which is parallel to
OA in the lines pq,
ts. Therefore pq and
t are parallel to OA,
and therefore to each
other. Similarly, pi,
qr are parallel to OB.

230

ELEMENTARY GEOMETRY.

1080 Cor. 1.The opposite sides of the parallelogram pqrs meet in two
points at infinity, which are the projections of the points A, B; and AB
itself, which is the third diagonal of the complete quadrilateral PQRS, is
projected into a line at infinity.
1081 Hence, to project any figure so that a certain line in it may pass to
infinity Take the plane of projection parallel to the plane which contains the
given line and the vertex.
1082 Cob. 2.To make the projection of the quadrilateral a rectangle,
it is only necessary to make A OB a right angle.

On Perspective Drawing.
1083 Taking the parallelogram pqrs, in (1079), for the original figure,
the quadrilateral PQRS is its projection on the plane ABab. Suppose this
plane to be the plane of the paper. Let the planes OAB, pab, while
remaining parallel to each other, be turned respectively about the fixed
parallel lines AB, ab. In every position of the planes, the linos Op, Oq, Or,
Os will iutersect the dotted lines in the same points P, Q, R, S. When the
planes coincide with that of the paper, pqrs becomes a ground plan of the
parallelogram, and PQRS is the representation of it in perspective.
AB is then called the horizontal line, ab the picture line, and the plane of
both the picture plane.

1084 To find the projection of any point p in the ground

plan.
Rule.Draw pb to any point b in the picture line, and draw OB parallel
to pb, to meet the horizontal line in B Join Op, Bb, and they will intersect in,
P, the point required.
In practice, pb is drawn perpendicnlar to ab, and OB therefore perpendi
cular to AB. The point B is then called the point of sight, or centre of vision,
and 0 the station point.

1085 To find the projection of a point in the ground plan,

not in the original plane, but at a perpendicular distance c
above it.
Rule. Take a new picture line parallel to the former, and at a distance
above it = c coseca, where a is the angle between the original plane and the
plane of projection. For a plane through the given point, parallel to the
original plane, will intersect the plane of projection in the new picture line
so constructed.
Thus, every point of a figure in the ground plan is transferred to the
drawing.
1086 The whole theory of perspective drawing is virtually included in
the foregoing propositions. The original plane is commonly horizontal, and
the plane of projection vertical. In this case, cosec a = 1, and the height of
the picture line for any point is equal to the height of the point itself above
the original plane.
The distance BO, when B is the point of sight, may be measured along
AB, and bp along ab, in the opposite direction; for the line Bb will continue
to intersect Op in the point P.

METHODS OF PROJECTION.

231

Orthogonal Projection.
1087 Definition. In orthogonal projection the lines of
projection are parallel to each other, and perpendicular to the
plane of projection. The vertex in this case may be consi
dered to be at infinity.
1088 The projections of parallel lines are parallel, and the
projected segments are in a constant ratio to the original
segments.
1089 Areas are in a constant ratio to their projections.
For, lines parallel to the intersection of the original plane and the plane
of projection are nnaltered in length, and lines at right angles to the former
are altered in a constant ratio. This ratio is the ratio of the areas, and is
the cosine of the angle between the two planes.

Projections of the Sphere.

1090 In Stereographic projection, the vertex is on the sur
face of the sphere, and the diameter through the vertex is
perpendicular to the plane of projection which passes through
the other extremity of the diameter. The projection is there
fore the inversion of the surface of the sphere (1012), and the
diameter is the constant k.
1091 In Globular projection, the vertex is taken at a dis
tance from the sphere equal to the radius 4- \/2, and the
diameter through the vertex is perpendicular to the plane of
projection.
1092 In Gnomonic projection, which is used in the construc
tion of sun-dials, the vertex is at the centre of the sphere.
1093 Mercator's projection, which is employed in navigation,
and sometimes in maps of the world, is not a projection at
all as defined in (1075). Meridian circles of the sphere are
represented on a plane by parallel right lines at intervals
equal to the intervals on the equator. The parallels of lati
tude are represented by right lines perpendicular to the
meridians, and at increasing intervals, so as to preserve the
actual ratio between the increments of longitude and latitude
at every point.
With r for the radius of the sphere, the distance, on the chart, from the
equator of a point whose latitude is X, is = r log tan (45 + \K) .

232

ELEMENTARY GEOMETRY.

Additional Tlieoreins.
1094 The sum of the squares of the distances of any point
P from n equidistant points on a circle whose centre is 0 and
radius r
= n (ri-\-OPi).
Proof.Sam the values of PIT; PC, &c, given in (819), and apply (803).
This theorem is the generalization of (923).
1095 In the same figure, if P be on the circle, the sum of
the squares of the perpendiculars from P on the radii OB, OCt
&c, is equal to ^r3.
Proof.Describe a circle upon the radius through P as diameter, and
apply the foregoing theorem to this circle.
1096 Cor ].The sum of the squares of the intercepts on the radii be
tween the perpendiculars and the centre is also equal to ^nr3.
(L 47)
1097 Cor. 2.The sum of the squares of the perpendiculars from the
equidistant points on the circle to any right line passing through the centre
is also equal to ^nr3.
Because the perpendicnlars from two points on a circle to the diameters
drawn through the points are equal.
1098 Cor. 3.Th e sum of the squares of the intercepts on the same
right line between the centre of the circle and the perpendiculars is also
equal to %ni*.
(I. 47)
If the radii of the inscribed and circumscribed circles of a
regular polygon of n sides be r, B, and the centre 0 ; then,
1099 I. The sum of the perpendiculars from any point P upon the sides
is equal to nr.
1100 II. If js be the perpendicnlar from 0 npon any right line, the sum
of the perpendiculars from the vertices upon the same line is equal to np.
1101 III. The sum of the squares of the perpendiculars from P on the
bides is = n (?* + OP3).
1102 IV. The sum of the squares of the perpendiculars from the vertices
npon the right line is = n (p* + {lt*).
Proof.In theorem I., the values of the perpendicnlars are given by
r OP cos ^0-\
with successive integers for m. Add together the n
values, and apply (803).
Similarly, to prove II. ; take for the perpendiculars the values
r-Rcos(e+^-).
To prove III. and IV., take the same expressions for the perpendiculars j
square each value; add the results, and apply (803, 804).
For additional propositions in the subjects of this section,
see the section entitled Plane Coordinate Geometry.

GEOMETRICAL

CONICS.

THE SECTIONS OF THE CONE.

1150 Definitions.A Conic Section or Conic is the curve
AP in which any plane intersects the surface of a right cone.
A right cone is the solid generated by the revolution of
one straight line about another which it intersects in a fixed
point at a constant angle.
Let the axis of the cone, in Fig. (1) or Fig. (2), be in the plane of the
paper, and let the cutting plane PMXN be perpendicular to the paper. (Bead
either the accented or unaccented letters throughout.') Let a sphere be inscribed
in the cone, touching it in the circle EQF and touching the cutting plane in
the point S, and let the cutting plane and the plane of the circle EQF inter
sect in XM. The following theorem may be regarded as the defining property
of the curve of section.
1151 Theorem.The distance of any point P on the conic
from the point S, called the focus, is in a constant ratio to
PM, its distance from the line XM, called the directrix, or
PS : PM= PS' : PM' = AS : AX = e, the eccentricity.
[See next page for the Prw/.]
1152 Cor.The conic may be generated in a plane from
either focus S, S', and either directrix XM, X'M', by the law
just proved.
1153 The conic is an Ellipse, a Parabola, or an Hyperbola,
according as e is less than, equal to, or greater than unity.
That is, according as the cutting plane emerges on both sides
of the lower cone, or is parallel to a side of the cone, or in
tersects both the upper and lower cones.
1154 All sections made by parallel planes are similar; for
the inclination of the cutting plane determines the ratio
AE :AX.
,
1155 The limiting forms of the curve are respectivelya
circle when e vanishes, and two coincident right lines when e
becomes infinite.
2 H

234

GEOMETRICAL CONICS.

Proof of Theokem 1151.Join P, S and P, 0, cutting the circular section

in Q, and draw PM parallel to NX. Because all tangents from the tame
point, 0 or A, to either sphere are equal, therefore BE = l'Q - PS and
AE = AS. Now, by (VI. 2), EE : NX = AE : AX and NX = PM;
therefore PS : PM = AS : AX, a constant ratio denoted by e and called
the eccentricity of the conic.

Referring the letters either to the ellipse or the hyperbola in the

subjoined figure, let G be the middle point of AA' and N any other point on
it. Let PL), lilt' be the two circular sections of the cone whose planes pass
through G and Ar; BCB? and PN the intersections with the plane of the
conic. In the ellipse, BG is the common ordinate of the ellipse and circle;
but, in the hyperbola, LG is to be taken equal to the tangent from G to the
circle DH.
1156
is

The fundamental equation of the ellipse or hyperbola

PN* : AN. NA' = BC* : AC*.

Proof.PN> = NB . NB' and BC* = CD. CP/ (III. 35, 36). Also, by
similar triangles (VI. 2, 6), NB : CD = AN: AG and NB? : CB'=A'N : A'C.
Multiply the last equations together.

TEE ELLIPSE AND EYPEEBOLA.

235

1157 Cob. 1.PN has

equal values at two points
equi- distant from AA'.
Hence the curve is sym
metrical with respect to
M and BB.
These two lines are
called the major and mi
nor axes, otherwise the
transverse and conjugate
axes of the conic.
When the axes are
equal, or BC = AG, the
ellipse becomes a circle,
and the hyperbola be
comes rectangular or
equilateral.
1158 Any ellipse or hyperbola is the orthogonal projection
of a circle or rectangular hyperbola respectively.
Proof.Along tbe ordinate NP, measure NP' = AN . NA' ; therefore by
the theorem PN : P"N = BG : AG. Therefore a circle or rectangular hyper
bola, having AA' for one axis, and having its plane inclined to that of the
conic at an angle whose cosine = BG-i-AC, projects orthogonally into the
ellipse or hyperbola in question, by (1089). See Note to (1201).

1159 Hence any projective property (1076-78), which is

known to belong to the circle or rectangular hyperbola, will
also be universally true for the ellipse and hyperbola respec
tively.

THE ELLIPSE AND HYPERBOLA.

Joint properties of the Ellipse and Hyperbola.
1160 Definitions.The tangent to a curve at a point P
(Fig. 1166) is the right line PQ, drawn through an adjacent
point Q, in its ultimate position when Q is made to coincide
with P.
The normal is the perpendicular to the tangent through
the point of contact.

236

GEOMETRICAL C0NIC8.

In (Fig. 1171), referred to rectangular axes through the

centre G (see Coordinate Geometry) ; the length CN is called
the abscissa ; PN the ordinate ; PT the tangent ; PO the nor
mal ; NT the subtangent ; and NG the subnormal. 8, S" are
the foci; XM,XM' the directrices; PS, PS' the focal dis
tances, and a double ordinate through S the Latus Rectum.
The auxiliary circle (Fig. 1173) is described upon AA' as
diameter.
A diameter parallel to the tangent at the extremity of
another diameter is termed a conjugate diameter with respect
to the other.
The conjugate hyperbola has BG for its major, and AC for
its minor axis (1157).
1161 The following theorems (1162) to (1181) are deduced from the
property PS : PM = e obtained in (1151).
The propositions and demonstrations are nearly identical for the ellipse
and the hyperbola, any difference in the application being specified.

1162

CS : CA : CX, and the common ratio is e.

1163 In the ellipse the sum, and in the hyperbola the dif
ference, of the focal distances of P is equal to the major axis,
or
PSJPS= A A'.
Proof.With the same figures we have, in the ellipse, by (1151),
e = PS+pff, ^ PS+par ^ e m as+a's = a* tWore 4c
pm+pw
xx 'mmau'ue ax+a'x
xx:'
For the hyperbola take difference instead of sum.

THE ELLIPSE AND HYPERBOLA.

1164

237

CS2 = AW-BC* in the ellipse.

[For B8 = AC, by (1163).
CS2 = ACl+BC* in the hyperbola.
[By assuming BG. See (1176).

1165
BC*=SL.AC.
Pboof. (Figs, of 1162) SL: SX=OS: OA,
(1151, 1162)
.-. SL.AC=CS.SX=CS(CX~ GS) = GA1 GS> (1162) = B(? (1164).
1166 If a right line through P, Q, two points on the conic,
meets the directrix in Z, then SZ bisects the angle QSR.
\z'
\
P^r

s'<\'
S

S' J

XT/5

Proof.By similar triangles, ZP : ZQ MP : NQ = SP : SQ (1151),

therefore by (VI. A.)
1167 If PZ be a tangent at P, then PSZ and PSZ' are
right angles.
Pboof.Make Q coincide with P in the last theorem.
1168 The tangent makes equal angles with the focal dis
tances.
Pboof.In (1166), PS : PS' = PM : PM' (1151) = PZ : PZ'; therefore,
when PQ becomes the tangent at P, Z SPZ = 8PZ', by (1167) and (VI. 7).
1169 The tangents at the extremities of a focal chord inter
sect in the directrix.
Pboof.(Figs, of 1166). Join ZB ; then, if ZP is a tangent, ZR is also,
for (1167) proves RSZ to be a right angle.
1170

CN.CT=AC\

PB00F._(Fig8. 1171.) H = H (VI. 3, A.) = ||; (1151) = ||,

therefore
therefore

TS+T8 =
0**=*,
TS'-TS
NX'-NX9
2GS
2GN'
CN.CT=CS.CX = AC.

(1162)

238

1171

GEOMETRICAL CONICS.

If PG be the normal,
GS.PS- GS' : PS' = e.

Proof.By (11 G8) and (VI. 3, A.),

GS _Gg_ GS'+GS _ 2CS _
PS
PS'
Pb' + PS
2CA
But, for the hyperbola, change phis to minus.
1172 The subnormal and the abscissa are as the squares of
the axes, or
AG : NC = BC2 : AC*.
Proof. (Figs. 1171.) Exactly as in (1170), taking the normal instead
of the tangent, we obtain ^ =
cs
CN~ CG_ OA* CS1
CN
CA> '

.\ ^ = ^-= ^ (11G2),
CG
GS
OS1
BCf
NG
, (H64).
NO
AG

1173 The tangents at P and Q, the corresponding points on

the ellipse and auxiliary circle, meet the axis in the same point
T. But in the hyperbola, the ordinate TQ of the circle being
drawn, the tangent at Q cuts the axis in N.
i
^-r-\
E,

Proof.For the ellipse : Join TQ. Then CW. CT = CQ* (11 70) ; there
fore CQT is a right angle (VI. 8) ; therefore QT is a tangent.
For the hyperbola : Interchange N and T.

THE ELLIPSE
i
AND HYPERBOLA.

239

1174
PN : QN = BC : AC.
Proof.(Figs. 1173). NG . NT = PN1, and CN.NT = QN\ (VI. 8)
Therefore NG : NC = PN1 : C^1; therefore, by (1172).
This proposition is equivalent to (1158), and shows that an ellipse is the
orthogonal projection of a circle equal to the auxiliary circle.
1175 Cob.The area of the ellipse is to that of the auxiliary
circle as DC:.iC (1089).
1176
PN2 : AN. NA' = BC* : AC2.
Peoof.By (1174), since QN1 = AN .NA' (111.35,36). An independ
ent proof of this theorem is given in (1156). The construction for BC in
the hyperbola in (1164) is thus verified.
Cn . Ct = BC2.
1177
Pkoof.(Figs. 1173.)
Ct _ J'iV . Cn . Ct
FN'
CT NT
NT''
CN.CT CN.NT
AN-NA'^1-55^
Therefore, by (1170) and (1176), On . Ct : AC = BC1 : AO*.
1178 If SY, /STY* are the perpendiculars on the tangent,
then Y, Y' are points on the auxiliary circle, and
SY.SY' = BC2.

Peoof.Let PS meet SY in W. Then PS = PW (1168). Therefore

S~W=AA' (1163). Also, SY= YW, and 80 = OS'. Therefore CY= \S W
AC. Similarly CY' = AG. Therefore Y, Y' are on the circle.
Hence ZY' is a diameter (III. 31), and therefore SZ = S'Y', by similar
triangles ; therefore SY. SZ = SA . SA' (HI. 35, 36) = CS1 ~ OA1 (II. 5)
= BC (1164).
1179 Cob.If CE be drawn parallel to the tangent at P,
then
PE=CY=AC.
1180 Pboblem.To draw tangents from any point 0 to an
ellipse or hyperbola.
Constbcction.(Figs. 1181.) Describe two circles, one with centre 0 and
radius OS, and another with centre and radius = AA', intersecting in M,

240

GEOMETRICAL C0NIC8.

Jf. Join MS", M'&. These lines will intersect the curve in P, P, the
points of contact. For another method see (1204).
Proof.By (1163), PSPS = AA' = KM by construction. There
fore PS = PM, therefore Z OPS = OPM (1. 8), therefore OP is a tangent
by (1168).
Similarly P'S = P'M', and OP is a tangent.
1181
focus.

The tangents OP, OP subtend equal angles at either

Proof.The angles OSP, OSP" are respectively equal to OMP, OM'P",

by (I. 8), as above ; and these last angles arc equal, by the triangles OS'M,
OSM", and (I. 8). Similarly at the other focus.
Asymptotic Properties of the Hyperbola.
1182 Def. The asymptotes of the hyperbola are the
diagonals of the rectangle formed by tangents at the vertices
A, A', B, ir.
1183 If the ordinates EN, BM from any point B on an
asymptote cut the hyperbola and its conjugate in P, P", D, IJ ,

241

THE HYPERBOLA.

then either of the following pairs of equations will define both

the branches of each curve
RN2-PN2 = BC2 = PN2-RN2
RMi-DMi = AC2 = DM2-RM2

(1),
(2).

Proof.Firstly, to prove (1) : By proportion from the similar triangles

T>xrn
ntn we have
u
R^1 = HC*
PN*
BZW, 04(7,
_
by (1176), since

AN. NA' = ON*-AO*.

By (II. 6)

Therefore

vwr- = fki>
by the theorem (69) ;
AO
AC "
therefore
RN'-PN* = BC.
Also, by (1176), applied to the conjugate hyperbola, the axes being now
,
CN1
ON*
. 7 . ., ^ tnanles
. .
, ;
reversed,
PN*-BC
= AC
BC = RN1'
therefore
P'N'-BC = RN* or PN'-RN* = BC.
Secondly, to prove (2) : By proportion from the triangles RMO, OBC, we
RM1
AC
DM1
have
CM2
BC
CM'-BC"
by (1176), applied to the conjugate hyperbola, for in this case we should
have
BM . MI? = CM*- BO*.
Therefore E3f^f3/' = j~ ; therefore RM^-BM* = AC
Also, by (1176), since CM, D'M are equal to the coordinates of V,
CM1
BO1 = RM*'
CM1
VU*-AO*
^AC
by 8imilar triangl6S :
therefore

D'M*- AC = RM* or D'M,-RM1 = AC.

1184 Cor. 1. If the same ordinates RN, RM meet the

other asymptote in r and r', then
PR.Pr = BC2 and DR.Dr^AC2.
(II. 5)
1185 Cor. 2.As It recedes from G, PR and DR con
tinually diminish.
Hence the curves continually approach
the asymptote.
1186 If XE be the directrix, CE = AC.
Proof. CE : CO = CX : CA - CA : CS and CS = CO.
1187 PD is parallel to the asymptote.
Po
RN1
BC* _ RN'-PN* ,Qei.
PN*
Peoof- rm* =
= im'-DU* (1183) = mF m'
Therefore RN I PN = RM : DM; therefore, by (VI. 2).
2 I

(1164)

GEOMETRICAL CONICS.

242

1188 The segments of any right line between the curve and
the asymptote are equal, or QR = qr.

Proof.
and

QR : QTJ qR : qU'\
Qr : Qu = qr : qu )

Compound the ratios,

and employ (1184).

1189

Cor. 1.PL = PI and QV = qV.

1190

Cor. 2.CH = HL.

Because PD is parallel to IC.

(1187)

1191 QR.Qr = PL* = RV'-QV*^ Q'V*-RV.

Proof QR : QU = FL : PE ) Compound theratios. Therefore, by (11 84),
and
Qr : Qu = PI : Pe )
QR . Qr = PL .PI = PL' (1189).
1192
Proof.
and

4PH.PK=CS*.
FH : PE = GO : Oo] ;. PH. PK : PE .Pe = CO* : Oo*
PK : Pe = Co : Oo) = CS1 : ABC1'; therefore, by (1184).

Joint Properties of the Ellipse and Hyperbola resumed.

If POP' be a diameter, and QV an ordinate parallel to the
conjugate diameter CD (Figs. 1195 and 1188).
1193

QV2 : PV.VP = CD* : CP*.

This is the fundamental equation of the conic, equation (1176) being the
most important form of it.

THE ELLIPSE AND HYPERBOLA.

Otherwise :
In the ellipse,

243

Q V2 : CP2 - C V2 = CD2 : CP2.

In the hyperbola, QV2 : CV'-CP2 = CD2 : CP2 ;

and

Q'V2 : CV*+CP2 = CD2 : CP2.

Proof.(Ellipse. Fig. 1195.)By orthogonal projection from a circle.

If C, P, P", D, Q, V are the projections of c, p, p', d, q, v on the circle ;
ji'5 = po . vp' and c(? = cp*. The proportion is therefore true in the case of
the circle. Therefore &c, by (1088).
(Hyperbola. Fig. 1188)
CP1

C\T

CP1

'CVi^CPi

07s- CP* r CV^-rCP1'

'

1194 The parallelogram formed by tangents at the extremi

ties of conjugate diameters is of constant area, and therefore,
PF being perpendicular to CD (Figs. 1195),
PF.CD = AC.BC.
Proof.(Ellipse.') By orthogonal projection from the circle (1089).
(Hyperbola. Fig. 1188.) CL.Cl = 4PH.PK = CO.Co (1192) ; there
fore, by (VI. 15), ALCl = OCo = AC.BC.
If PF intersects the axes in G and ?',
1195

PF.PG = BC2 and PF .PG' = AC2.

Proof. PF.PG = PK.PN= On . Ct = J3C3 (1177).

PF.Pa.
1197

Cor. PG . PG' = CD2 = PT . Pt.

Similarly for
^
By (1194)

244

GEOMETRICAL CONICS.

1198 The diameter bisects all chords parallel to the tangent

at its extremity.
Proof. (Ellipse. Fig. 1195.) By projection from the circle (1088)
QV = VOJ.
(Hyperbola.) By (1189.)

1199 Coe. 1.The tangents at the extremities of any chord

meet on the diameter which bisects it.
Proof.The secants drawn through the extremities of two parallel chords
meet on the diameter which bisects them (VI. 4), and the tangents are the
limiting positions of the secants when the parallel chords coincide.

1200 Cor. 2.If the tangents from a point are equal, the
diameter through the point must be a principal axis.
(i. 8)
1201 Cor. 3.The chords joining any point Q on the curve
with the extremities of a diameter PP , are parallel to con
jugate diameters, and are called supplemental chords.
For the diameter bisecting PQ is parallel to P"Q (VI. 2).
diameter bisecting P'Q is parallel to PQ.

Similarly the

1202 Diameters are mutually conjugate ; If CD be parallel

to the tangent at P, CP will be parallel to the tangent at D.
Proof.(Ellipse. Fig. 1205.)By projection from the circle (1088).
Note. Observe that, if the ellipse in the figure with its ordinates and
tangents be turned about the axis Tt through the angle cos"1 (BO -j- AC), it
becomes the projection of the auxiliary circle with its corresponding ordinates
and tangents.
(Hyperbola. Fig. 1188.)By (11S7, 1189) the tangents at P, D meet the
asymptotes in the same point L. Therefore they are parallel to CD, GP (VI. 2.)

If QT be the tangent at Q, and QV the ordinate parallel

to the tangent at any other point P,
1203
CV.CT=CFi.
(l)

(2)

O,

Proof. CB bisects PQ (1199). Therefore PW is parallel to QB.

Therefore, by. (VI. 2), CV : CP = CW :CR=CP :CT.

THE ELLIPSE AND HYPERBOLA.

245

1204 Coe.Hence, to draw two tangents from a point T, we may find

CFfrora the above equation, and draw QVQ' parallel to the tangent at P to
determine the points of contact Q, Q'.
Let PN, DN be the ordinates at the extremities of con
jugate diameters, and PT the tangent at P. Let the ordinates
at N and R in the ellipse, but at T and G in the hyperbola,
meet the auxiliary circle in p and d ; then
CN = dR,

1205

CR = pN.
d,

//v
f 1
r a J

it

Proof.{Ellipse.) Op, Cd are parallel to the tangents at d and p (Note to

1202). Therefore pCd is a right angle. Therefore pNC, OBd are equal
right-angled triangles with CN=dR and GB=pN.
(Hyperbola.)
CN . CT= AO' (1170),
and
DB.CT= 2\CDT = 2CDP = AC .BO (1194) ;
CN_AC = pN
7
CN = PN=TN
DB
CB ,(8,mllar
. ., tnangles0. . , .
pN
DR~ BO
PN 1 ' h
Bat 55T=M (VI. 8) ; .-. CB = pN. Also Cp = Od; therefore the tripN IN
angles CpN, dCB are equal and similar; therefore CN=dB and dB is
parallel to pN.
1206

Coe.

DR:dR = BC : AC.

Proof. (Ellipse.) By (1174). (Hyperbola.) By the similar right-angled

triangles, we have
dB : pN = OB : TN= DB:PN;
therefore
dB : DB= pN : PN= AO : BO (1174).

In the same figures,

1207

(Ellipse.)

CN2+CR2 = AC2; DR2+PN2 = BC2.

1209

(Hyperbola.) CN2- CR2 = AC2; DR2-PN2 = BC2.

Proof.Firstly, from the right-angled triangle CNp in which pN CB

(1205).
Secondly, In the ellipse, by (1174), DR'+PN* : dR'+pN* = BC1 : AC,
and dRi+pNi=AC1, by (1205). For the hyperbola, take difference of squares.

246

1211
1212

GEOMETRICAL COXICS.

(Ellipse.)
CF+CD1^ AC*+BC\
(Hyperbola.) CP" -CI? = AC-BC.

Proof. (Figs. 1205.)

triangles CXP, CRD.

By (12051210) and (L 47), applied to the

The product of the focal distances is equal to the square

of the semi-conjugate diameter, or
1213
PS. PS = CD3.
Proof. (Ellipse. Fig. 1171.) 2PS . P& = (PS + PSy - PS*- P&*
= 446,,-2C.$,-2CTJ(922,i.) = 2 (AC1 + BC1-CPt) (1164) = 2^^1211).
(Hyperbola.)Similarly with 2PS .P&= PS'+PS'-(PS'-PSy= Ac.
1214 The products of the segments of intersecting chords
QOq, Q'Oq' are in the ratio of the squares of the diameters
parallel to them, or
OQ.Oq : OQ.Oq' = CI? : CD 3.
Pboof. (Ellipte.) By projection from the circle (1088) ; for the propor
tion is true for the circle, by (III. 35, 36).
(Hyperbola. Fig. 1188.) Let 0 be any point on Qq. Draw IOi parallel
to Ee, meeting the asymptotes in J and i ; then
OR . Or- 0Q.0q=QR. Qr (II. 5) = PL' (1191)
(1).

OR
PL
,
Or
PI
OR.
Or
PL*
CI?
,11C..
N0W Ol=PB,and Oi=Pe' WXn=EZPe=BC^(U8i)wu
r_ ^
OR.Or-PV = ^
CV i or, .by n.
CD*
Therefore
(1), Q/OQ.Oq
g.,^. = ^
Similarly for any other chord Q'Oq' drawn through 0.
Therefore
OQ.Oq: OQJ. Oq = CD* : CLP.
1215 Cor.The tangents from any point to the curve are
in the ratio of the diameters parallel to them.
For, when 0 is without the curve and the chords become tangents, each
product of segments becomes the sqnare of a tangent.
1216 If from any point Q on a tangent FT drawn to any
conic (Fig. 1220), two perpendiculars QR, QL be drawn to the
focal distance PS and the directrix XM respectively ; then
SR : QL = e.
Pboof.Since QR is parallel to ZS (1167), therefore, by (VI. 2),
SR : PS = QZ : PZ = QL : PM;
therefore
SR : QL = PS : PM = e.
Cor.By applying the theorem to each of the tangents from Q, a proof
of (1181) is obtained.

THE ELLIPSE AND HYPERBOLA.

247

1217 The Director Circle.The locus of the point of inter

section, T, of two tangents always at right angles is a circle
called the Director Circle.

Proof.Perpendiculars from S, & to the tangents meet them in points

T, Z, Y, Z', which lie on the auxiliary circle. Therefore, by (II. 5, 6) and
(111.35,36),
TCi~AC,= TZ.TZ'=SY.gY' = BO1.
(1178)
Therefore

TO* = AC BC,

a constant value.

Note.Theorems (1170), (1177), and (1203) may also be deduced

at once for the ellipse by orthogonal projection from the circle ; and, in all
such cases, the analogous property of the hyperbola may be obtained by a
similar projection from the rectangular hyperbola if the property has already
been demonstrated for the latter curve.

1218 If the points A, S (Fig. 1162) be fixed, while C is

moved to an infinite distance, the conic becomes a parabola.
Hence, any relation which has been established for parts of
the curve which remain finite, when AC thus becomes infinite,
icill be a property of the parabola.
1219 Theorems relating to the ellipse may generally be
adapted to the parabola by eliminating the quantities which
become infinite, employing the principle that finite differences
may be neglected in considering the ratios of infinite quantities.
Example. In (1193), when P" is at infinity, VP" becomes = 2CP; and
in (1213) PS" becomes = 2GP. Thus the equations become

921 = and PS=^.

PV
OP
20P
Therefore QV = 4PS . PV in the parabola.

248

GEOMETRICAL CONICS.

THE

PARABOLA.

If 8 be the focus, XM
the directrix, and P any point
on the curve, the defining pro
perty is
1220
and
1221

PS = PM
e = 1.

(1153)

Hence
AX = AS.

1222 The Latus Rectum = 4A S.

Proof.
SL = SX (1220) = 2AS.
1223 If PZ be a tangent at P, meeting the directrix in Z,
then PSZ is a right angle.
Proof.As in (1167) ; theorem (1166) applying equally to the parabola.
1224

The tangent at P bisects the angles SPM, SZM.

Proof.
is common to the triangles PSZ, PMZ; PS = PM and
Z PSZ = PMZ (1223).
1225

Cor.

ST=SP = SG.

(1.29,6)

1226 The tangents at the extremities of a focal chord PQ

intersect at right angles in the directrix.
Proof.(i.) They intersect in the directrix, as in (1169).
(ii.) They bisect the angles SZM, SZM' (1224), and therefore include a
right angle.
1227 The curve bisects the sub-tangent. AN = AT.
Proof. ST = 8P (1225) = PM = XN, and AX = AS.
1228 The sub-normal is half the latus rectum. NG = 2AS.
Proot.ST=SP=SG and TX = SN (1227). Subtract.

THE PARABOLA.

249

1229
PN^AAS.AN.
Proof.PIP =TN.NG (VI. 8) =AN. 2N0 (1227) = 4AS . AN (1228).
Otherwise, by (1176) and (1165) ; making AG infinite. See (1219).
1230 The tangents at A and P each bisect SM, the latter
bisecting it at right angles.
Proof.(i.) The tangent at A, by (1. 2), since AX = AS.
(ii.) PT bisects SM at right angles, by (I. 4), since PS = PM and
I SPY = MPY.
1231

COE.

SA : SY : SP.

[By similar triangles.

1232

To draw tangents from a point 0 to the parabola.

Construction. Describe a circle, centre

0 and radius OS, cutting the directrix in
M, M'. Draw MQ, M'Q' parallel to the axis,
meeting the parabola in Q, Q'. Then OQ,
OQf will be tangents.
Proof. OS, SQ=OM, MQ (1220) ;0
therefore, by (I. 8), I OQS = OQM ; there
fore OQ is a tangent (1224). Similarly
OQ' is a tangent.
Otherwise, by (1181). When 8" moves
to infinity, the circle MM' becomes the
directrix.
1233

Cos. 1.The triangles SQO, SOQ' are similar, and

Proof.
Similarly

SQ : SO : SQ'.
/. SQO = MQO = SMM' = SOQ'.
SQ'O = SOQ.

(III. 20)

1234 Coe. 2.The tangents at two points subtend equal

angles at the focus ; and they contain an angle equal to half
the exterior angle between the focal distances of the points.
Proof.
Also

Z OSQ = OSQ', by (Cor. 1).

/.QOQ' = SOQ + SQO = *-OSQ=%QSQ'.

1235 Def.Any line parallel to the axis of a parabola is

called a diameter.
1236 The chord of contact QQ' of tangents from any point
0 is bisected by the diameter through 0.
Proof.This proposition and the corollaries are included in (1198-1200),
by the principle in (1218). An independent proof is as follows.
2 K

GEOMETRICAL CONICS.

250

The construction Being as in (1232),

we have ZM=ZM'; therefore QV=V<$
(VI. 2).
1237 Coe. 1. The tangent
BR' at P is parallel to QQ' ; and
OP = PV.
Proof. Draw the diameter liW.
QW= WP; therefore QR = R0 (VI. 2).
Similarly Q'R' = RO.
1238 Coe. 2.Hence, the dia
meter through P bisects all chords parallel to the tangent at P.
If QVbe a semi-chord parallel to the tangent at P,
1239

QF = 4PS.PF.

This is the fundamental equation of

the parabola, equation (1229) being the
most important form of it.
Proof.Let QO meet the axis in T.
By similar triangles (1231),
/ SEP = SQR = STQ = FOR ;
and tSPR =OPR (1224). Therefore
l'R%= PS.PO = PS. PVand QV= 2PR.
Otherwise: See (1219), where the
equation is deduced from (1193) of the
ellipse.
1240 Coe. 1.If v be any other point, either within or
without the curve, on the chord QQ' , and pv the corresponding
diameter,

vQ . vQ' 4pS .pv.

(II. 5)

1241 Coe. 2.The focal chord parallel to the diameter

through P, and called the parameter of that diameter, is equal
to 4SP. For PV in this case is equal to PS.
1242 The products of the segments of intersecting chords,
Q,Oq, Q'Oq, are in the ratio of the parameters of the diameters
which bisect the chords ; or
OQ.Oq: OQ.Oq' = PS : P S.
Proof.By (1240), the ratio is equal to 4PS.pO : iPS.pO.
Otherwise : In the ellipse (1214), the ratio is = -^f^ = -rJt D-c (1213)
~ t*h' WDen &

a* infinity and the curve becomes a parabola (1219).

ON CONSTRUCTING THE CONIC.

251

1243 Cor.The squares of the tangents to a parabola from

any point are as the focal distances of the points of contact.
Peoof.As in (1215).

Otherwise, by (1233) and (VI. 19).

1244 The area of the parabola cut off by any chord QQ' is
two-thirds of the circumscribed parallelogram, or of the tri
angle formed by the chord and the tangents at Q, Q'.
Proof.Through Q, q, q', &c, adjacent points
on the curve, draw right lines parallel to the
diameter and tangent at P. Let the secant Qq
cut the diameter in 0. Then, when q coincides
with Q, so that Qq becomes a tangent, we have
OP = PV (1237). Therefore the parallelogram
Vq = 2Uq, by (I. 43), applied to the parallelogram
of which OQ is the diagonal. Similarly vq = 2uq',
Ac. Therefore the sum of all the evanescent par
allelograms on one side of PQ is equal to twice
the corresponding sum on the other side ; and
these sums are respectively equal to the areas
PQV, PQJJ (Newton, Sect. I., Lem. II.)

1245

Practical methods of constructing the Conic.

To draw the Ellipse.

Fix two pins at S, S" (Fig. 1162). Place over them a loop of thread
having a perimeter SPS = Si>' + AA\ A pencil point moved so as to keep
the thread stretched will describe the ellipse, by (1163).
1246
Otherwise. (Fig. 1173.) Draw PEK parallel to QC, cutting the
axes in JEL, K. PK = AC and PH = BC (1174). Hence, if a ruler PHK
moves so that the points H, K slide along the axes, P will describe the ellipse.

1247

To draw the Hyperbola.

Make the pin 8' (Fig. 1162) serve as a pivot for one end of a bar of any
convenient length. To the free end of the bar attach one end of a thread
whose length is less than that of the bar by AA' ; and fasten the other end of
the thread to the pin 8. A pencil point moved so as to keep the thread
stretched, and touching the bar, will describe the hyperbola, by (1163).
1248 Otherwise :Lay off any scale of equal parts along both asymptotes
(Fig. 1188), starting and numbering the divisions from 0, in both positive
and negative directions.
Join every pair of points L, I, the product of whose distances from C is
the same, and a series of tangents will be formed (1192) which will define
the hyperbola. See also (1289).

252

GEOMETRICAL C0N1CS.

1249

To draw the Parabola.

Proceed as in (1247), with this difference : let the end of the bar, before
attached to &, terminate in a " T-sqnare," and be made to slide along the
directrix (Fig. 1220), taking the string and bar of the same length.
1250 Otherwise: Make the same construction as in (1248), and join
every pair of points, the algebraic sum of whose distances from the zero
point of division is the same.
Proof.If the two equal tangents from any point T on the axis (Pig.
1239) be cut by a third tangent in the points II, r; then BQ may be proved
equal to rT, by (1233), proving the triangles SBQ, SrT equal in all respects.
1251
SQT.

Cob.The triangle SBr is always similar to the isosceles triangle

1252

To find the axes and centre of a given central conic.

(i.) Draw a right line through the centres of two parallel chordB. This
line is a diameter, by (1198) ; and two diameters so found will intersect in
the centre of the conic.
(ii.) Describe a circle having for its diameter any diameter PP1 of the
conic, and let the circle cut the curve in Q. Then PQ, PQ are parallel to
the axes, by (1201) and (III. 31).

1253

Given two conjugate diameters, CP, CD, in position

and magnitude : to construct the conic.

On CPtake PZ = CVi+CP; measuring

from C in the ellipse, and towards 0 in the
hyperbola (Fig. 1188). A circle described
through the points C, Z, and having its
centre 0 on the tangent at P, will cut the
tangent in the points where it is intersected
by the axes.
Proof. Analysis: Let AC, BC cut the
tangent at P in T, t. The circle whose
diameter is Tt will pass through C (III. 31),
and will make
CP . PZ = PT . Pt (III. 35, 36) = CD (1197).
Hence the construction.

Circle and Radius of Curvature.

1254 Definitions.The circle which has the same tangent
with a curve at P (Fig. 1259), and which passes through
another point Q on the curve, becomes the circle of curvature
when Q ultimately coincides with P ; and its radius becomes
the radius of curvature.

CIRCLE OF CURVATURE.

253

1255 Otherwise.The circle of curvature is the circle which

passes through three coincident points on the curve at P.
1256 Any chord PH of the circle of curvature is called a
chord of curvature at P.
1257 Through Q draw RQ' parallel to PH, meeting the
tangent at P in B, and the circle in Q' , and draw QV parallel
to PR. RQ is called a subtense of the arc PQ.
1258 Theorem.Any chord of curvature PH is equal to
the ultimate' value of the square of the arc PQ divided by the
subtense RQ parallel to the chord : and this is also equal to

Pboof.RQf = RP'RQ (HI. 36). And when Q moves np to P, RQ'

becomes PH; and RP, PQ, and QV become equal because coincident lines.
1259 In the ellipse or hyperbola, the semi-chords of cur
vature at P, measured along the diameter PC, the normal
PF, and the focal distance PS, are respectively equal to
CD2
CD2
CD\
CP'
PF '
AC '
the second being the radius of curvature at P.

limit when VP" becomes PF = 2CP.

(ii.) By the similar triangles PHU, PFC (III. 31), we have
PU.PF= CP.PH= 2CI?, by (i.)
(iii.) By the similar triangles PIU, PFE (1168), we have
PI.PE=PU.PF = 2CD8, by (ii.) ; and PE = AC (1179).

254

GEOMETRICAL CONICS.

1260 In the parabola, the chord of curvature at P (Fig.

1259) drawn parallel to the axis, and the one drawn through
the focus, are each equal to 4SP, the parameter of the dia
meter at P (1241).
Proof.By (1258). The chord parallel to the axis = QT*+PV= 4PS
(1239) ; and the two chords are equal because they make equal angles with,
the diameter of the circle of curvature.

1261 Cor.The radius of curvature of the parabola at P

(Fig. 1220) is equal to 2SP* + SY.
Proof.(Fig. 1259.) $PU - \PI sec IPU = 2SP sec P8Y (Fig. 1221).

1262 The products of the segments of intersecting chords

are as the squares of the tangents parallel to them (1214-15),
(1242-43).
1263 The common chords of a circle and conic (Fig. 1264)
are equally inclined to the axis ; and conversely, if two chords
of a conic are equally inclined to the axis, their extremities
are coneychc.
Proof.The products of the segments of the chords being equal (III.
35, 36), the tangents parallel to them are equal (1262). Therefore, by (1200).

1264 The common chord of any conic and of the circle of

curvature at a point P, has the
same inclination to the axis as
the tangent at P.
Proof.Draw any chord Qq parallel
to the tangent at P. The circle circum
scribing PQq always passes through the
same point p (1263), and does so, there
fore, when Qq moves up to P, and the
circle becomes the circle of curvature.

1265 Problem.To find the centre of curvature at any given

point of a conic.
First Method.(Fig. 1264.) Draw a chord from the point making the
same angle with the axis as the tangent. The perpendicular bisector of the
chord will meet the normal in the centre of curvature, by (1264) and (III. 3).
1266 Second Method. Draw the normal PG and a perpendicular to it
from G, meeting either of the focal distances in Q. Then a perpendicular to
the focal distance drawn from Q will meet the normal in 0, the centre of
curvature.

MISCELLANEOUS THEOREMS.

255

Proof.(Ellipse or Hyperbola.) By (1259),

the radius of curvature at P
= ? = If PG (1195) = pj* fg (1194)
= PG sec' S"PG = PO.
For
AC = PE (by 1179).
(Parabola.) By (1261). The radius of curvature
= 2SF'-i-SY = 2SP sec SPG = PO.
For 2SP = PO, because SP = SG (1125).

Miscellaneous Theorems.
1267
then

In the Parabola (Fig. 1239) let QD be drawn perpendicular to PV,

QZ), = 44S.PF.
(1231,1239)

1268 Let RPR be any third tangent meeting the tangents OQ, OQ' in
fi, R; the triangles SQO, SiJiJ', SOQ' are similar and similarly divided by
SR, SP, SR (1233-4).
1269 Cor.
OR. OR = RQ.RQ.
1270 Also, the triangle PQQ' = 20RR.
With, the same construction and for any conic,
1271
OQ : OQ' = RQ.RR : RQ' . PR.
1272 Also the angle
RSR = %QSQ'.

(1244)
(1215, 1243)
(1181)

1273

Hence, in the Parabola, the points 0, R, S, R are concyclic, by (1234).

1274

In any conic (Figs. 1171), SP : ST = AN : AT.

Peoof.- ||=|| = ^ (69,1163)

(1170) =^ (69).

1275 Cob.If the tangent PT meets the tangent at A in R, then SR

bisects the angle PST (VI. 3).
1276
1277

In Figs. (1178), SY', S'Y both bisect the uorroal PG.

The perpendicular from S to PG meets it in CY.

1278 If CD be the radius conjugate to CP, the perpendicular from D

upon C Y is equal to BC.
1279

SY and CP intersect in the directrix.

1280 If every ordinate PN of the conic (Figs. 1205) be turned round N,

in the plane of the figure, through the same angle PNR, the locus of P" is
also a conic, by (1193). The auxiliary circle then becomes an ellipse, of
which AC and BC produced are the equi-conjugate diameters.
If the entire figures be thus deformed, the points on the axis AA' remain
fixed while PA7, DR describe the same angle. Hence CD remains parallel to
PT. CP, CD are therefore still conjugate to each other.

256

GEOMETRICAL CONICS.

Hence, the relations in (1205-6) still subsist when CA, CB are any con
jugate radii. Thus universally,
1281

PN: CR = BR: GN or FN. CN = BR . CR.

1282 If the tangent at P meets any pair of conjugate diameters in T, T,

then PT.PT is constant and equal to CB3.
Proof.Let CA, CB (Figs. 1205) be the conjugate radii, the figures
being deformed through any angle. By similar triangles,
PT : CN = CB : CR } ' therefore PT.PT : PN.CN = CB* : BR. CR.
Therefore

PT. PT = CB*, by (1281).

1283 If the tangent at P meets any pair of parallel tangents in T, T,

then PT.PT = CB', where CB is conjugate to CP.
Proof.Let the parallel tangents touch in the points Q, Q'. Join PQ,
PQ', CT, CT. Then CT, CT are conjugate diameters (1199, 1201). There
fore PT.PT = CI? (1282).
1284

Cor. QT. Q'T = CI/1, where CB" is the radius parallel to QT.

1285 To draw two conjugate diameters of a conic to include a given

angle. Proceed as in (1252 ii.), making PP in this case the chord of the
segment of a circle containing the given angle (III. 33).
1286 The focal distance of a point P on any conic is equal to the length
QN intercepted on the ordinate through P between the axis and the tangent
at the extremity of the latns rectum.
Proof.(Fig. 1220). QN : NX = L8 : SX = e and SP: NX= e.
1287 In the hyperbola (Fig. 1183). CO : CA = e.
- (1162,1164).
If a right line PKK" be drawn parallel to the asymptote CR, cutting the
one directrix XE in K and the other in K ; then
1288
8P - PK = e . GN-AG; S'P = PE'=e. ON+AO.
Proof.From CR = * . CN (1287) and CE = AC (1186).
1289 Cor. Hence the hyperbola may be drawn mechanically by the
method of (1249) by merely fixing the cross-piece of the T-square at au
angle with the bar equal to BCO.
1290

Definition. Confocal conies are conies which have the same foci.

1291 The tangents drawn to any conic from a point T on a confocal conic
make equal angles with the tangent at T.
Proof.(Fig. 1217.) Let T be the point on the confocal conic.
SY: SZ= SZ' : ST (1178).
Therefore ST and ST make equal angles with the tangents TP, TQ ; and
they also make equal angles with the tangent to the confocal at T (1168),
therefore &c.
1292 In the construction of (1253), PZ is equal to half the chord of
ourvature at P drawn through the centre C (1259).

DIFFERENTIAL CALCULUS.

INTRODUCTION.
1400 Functions.A quantity which depends for its value
upon another quantity x is called a function of x. Thus, sin x,
log x, of, ai+ax+xi are all functions of x. The notation
V =f{x) expresses generally that y is a function oix. y = sin*
is a particular function.
1401 f(x) is called a continuous function between assigned
limits, when an indefinitely small change in the value of x
always produces an indefinitely small change in the value of
A transcendental function is one which is not purely
algebraical, such as the exponential, logarithmic, and circular
functions ax, log*, sin*, cos a?, &c.
If f(x) =/( #), the function is called an even function.
If /(*) = /(#), it is called an odd function.
Thus, z* and cos x are even functions, while x* and sin z are odd functions
of x ; the latter, but not the former, being altered in value by changing the
sign of x.

1402 Differential Coefficient or Derivative.Let y be any

function of x denoted by /(), such that any change in the
value of * causes a definite change in the value of y ; then *
is called the independent variable, and y the dependent variable.
Let an indefinitely small change in x, denoted by dx, produce
a corresponding small change dy in y; then the ratio -A in
the limit when both dy and dx are vanishing, is called the
differential coefficient, or derivative, of y with respect
to x.
2 L

T)TFFEIIENTTAL CALCULUS.

258

1403 Theorem.The ratio dy : dx is definite for each value

of x, and generally different for different values.
Proof. Let an abscissa ON
i/
(Defs. 1160) be measured from 0
equal to x, and a perpendicular or
dinate NP equal to y. Then, what
ever may be the form of the function
y =f(x), as x varies, the locus of
P will be some line PQL. Let
OM = x', MQ = y be values of x
and y near to the former values, ^trLet the straight line QP meet the
axis in T; and when Q coincides with. P, let the final direction of QP cut the
. axis in T'.
Then SP
^| or xx = ~.
QS
NT And, ultimately,
J when
,
pv and SP vanish,
they vanish in the ratio of PiV : NT.

Therefore ^ = 4^r, = tan PTN, a

dx NF
definite ratio at each point of the curve, but different at different points.
1404

Let NM, the increment of x, be denoted by h ; then,

when h vanishes,

^ _

f(.x) _ f (#)

a new function of.,r, called also the first derived function.

The process of finding its value is called differentiation.
1405

Successive differentiation. If ^ or f'{x) be differ-

entiated with respect to x, the result is the second differential

coefficient of f(x), or the second, derived function; and so on
to any number of differentiations. These successive functions
may be represented in any of the three following systems of
notation :
dy
iP>i
d3tf
d*y
d*y ,
dx
da*'
dP'
dP'
dP''
/w. rw. /"> rw>
rw;
y*.
Vix>
Vix,
y*x,
y**
The operations of differentiating a function of x once,
twice, or n times, arc also indicated by prefixing the symbols
.
dx' da?
or, more concisely,

<** or d
(d\'
dx*'
dx' \dx)'
dx, d.^, . . . d^.
* See note to (148").

(d\\dx) '

introduction:

259

1406 If after differentiating a function for x, x be made

zero in the result, the value may be indicated in any of the
following ways :

/'(0),

yMt

dM.

If any other constant a be substituted for x in yx, the

result may be indicated by. yx .
1407 Infinitesimals and Differentials. The evanescent
quantities dx, dy are cajlecl infinitesimals; and, with respect
to x and y, they are called differentials, dx2, d~y are the second
differentials of x and y ; dx3, d3y the third, and so on.
1408 The successive differentials of y are expressed in terms
of dx by the equations
dy=f{x)dx; <Py =f" (x) dx* ; &c, and d*y =f*(x) dxn.
Since f (x) is the coefficient of dx in the value of dy, it has
therefore been named the differential coefficient of y or /()*
For similar reasons f (x) is called the second, and /" (x) the
n"1 differential coefficient of f(x), &c.
1409 Two infinitesimals are of the same order when their
ratio is neither zero nor infinity.
If dx, dy are infinitesimals of the same order, dx2, dy1, and
dxdy will be infinitesimals of the second order with respect to
dx, dy; dx3, dx2dy, &c. will be of the third order, and so on.
dx, dx2, &c. are sometimes denoted by x, x, &c.
1410 Lemma.In estimating the ratio of two quantities, any
increment of either which is infinitely small in comparison
with the quantities may be neglected.
Hence the ratio of two infinitesimals of the same order is
not affected by adding to or subtracting from either of them
an infinitesimal of a higher order.
Example. ^ ^X = ~- dx = ~-, for dx is zero in comparison with
,
dx
dx
dx
the ratio
Thus, in Fig. (1403), putting FS=dx, QS= dy; we have
ultimately, by (12f>8), QR = Icdx1, where k is a constant. Therefore
PN RS
dii-kdx* dy
.....
. , .
kt' =
=
~~ = j ln "ie hmit, by the principle jnat enunciated ;
i
Jra
dx
dx
that is, QR vanishes in comparison ivith PS or QS even when those lines themtdves are infinitely small.
* The name is slightly misleading, as it seems to imply that /' (x) is in
some sense a coefficient of/ (as).

DIFFERENTIAL CALCULUS.

260

DIFFERENTIATION.
DIFFERENTIATION OF A SUM, PRODUCT, AND QUOTIENT.
Let u, v be functions of x, then
141 1
d (u-\-v)
du , dv
dx
dx
dx '
+ Am
1412

d (uv)
du ,
dv
3* = v-T-+u
da?
dx
dx

1413

*(JL\ =
dx \ v }
\

dx

dx J

Pboof.(i.) d(u+v) = (u+du + v+dv) (u+v) = du+dv.

(ii.)
d(uv) = (u + du) (v + dv)uv = vdu+udvdudv,
and, by (1410), dudv disappears in the ultimate ratio to dx.
.... ,
j/ti\
u + du
it
vdu udv
(in.)
a II =
;
=
,
\vI
v + dv
v
(y + dv) v
therefore Ac, by (1410) ; vdv vanishing in comparison with *.
Hence, if u be a constant = c,
1414
</ (cv) _ c dv
dx
dx
1

d / c\ __
dx\v /

c dv
v* dx'

DIFFERENTIATION OF A FUNCTION OF A FUNCTION.

If y be a function of z, and z a function of x,
1415

|L = ^L.^.

Proof.Since, in all cases, the change dx causes the change dz, and the
change dz causes the change dy ; therefore the change dx causes the change
dy in the limit.
Differentiating the above as a product, by (1412), the successive differ
ential coefficients of y can be formed. The first four are here subjoined for
the sake of reference. Observe that (i/r), = i/2,.
1416

y* = yzzx.

1417

y* = 3fc,*i+y2Sb.

1418

y^ = ys + 3ytt s,^ +

1419

y* =

+y* (3a$, + 4s,sta)

DIFFERENTIA TION.

261

DIFFERENTIATION OF A COMPOSITE FUNCTION.

If and v be explicit functions of x, so that u = $ (x) and
v = t(x),
j^2Q
dF(u, v)
<#F du , dF dv
dx
du dx
dv dx '
Here dF in the first term on the right is the change in
F(u,v) produced by du, the change in u; and dF in the.
second term is the change produced by dv, so that the total
change dF(u, v) may be written as in (1408)
dF, + dFt= ^du+ M-&o.
du
dv
DIFFERENTIATION OF TEE SIMPLE FUNCTIONS.
Since 4^- =f(x~^~ty
^g-Q ^ yanishes, we have the
dx
h
following rule for finding its value :
1421 Rule.Expand f(x+h) by some known theorem in
ascending powers of h ; subtract f (x) ; divide by h ; and in
the result put h equal to zero.
The differential coefficients which follow are obtained by
the rule and the theorems indicated.

1422

y = *r.

& = ruf-\

PEOOF.-Here /(* + *)-/() _ (x+h)'-af _ nx"-*h + C(n, 2)ay+

h
h
h
(125) = nx'-i+Oin, 2) "-*&+... = tia""1, when h vanishes.
1423

Cor.
*=n(n-l)...tn-r+l)x~

1424

y = logax;

PBOOF.-By(145),

h
Expand the logarithm by (155).
1425

Cob.-

^- = jt^.
1_ f iog<
x log, a (.
V

^ = (-1)"-M"-1.
c/o?"
tfMoga

Ul.
as / J
as

Put^-l^
ana r-n-l.

DIFFERENTIAL CALCULUS.

262

a"

1426

a'^-a' _ a'(ah-l)
h
h

Proof.
1427

Cob.

Expand a" by (149).

g. = a-(l0g.a)-

Function.

Method of Froof by Rule (1421)

and Limits (753).
Expand by (627, G29), and
put 1 cos h = 2 sin1 .

Derivative.

1428

Bin x.

COS*.

1429

cos x.

1430

tan*,

sin x.
seca x.

1431

cot x.

cosec2 x.

1432

sec x.

1433

cosec x.

cot*cosec*.

1434

sin'
cos

1436

tan-1 x \
cot

i
1+*4"

1438

sec-1 x
cosec

tan* sec*.

1
v{i-x*y

1
~ ^7 ^

Expand by (631), observing

(1410).
By cota! = , and(1415).
tana
By secas = , and(1415).
cos
Similarly.
If sm~lx = y, = siny,
therefore
= cos y =
'hi'
1
therefore dy__
dx

2y

Similarly for the rest.

Examples.
1440

(1422)

1441

(1422)

1442

{(a+s*)' (&+*)*} = 3(a + a!,),2*(6+!i!,),+2(6+a!,)3a!,(a+!!,) '

, = 6x [a+x*y (6 + xs) (b+vx+W).
(1412, '15, '22)

1AA.O
Xit*0

(e'-e~I\ - f+O (e*+e-')- (eW) je'-e-') _

4
Ve*+e-*/.
(ex+e-*)s
(e' + e-)1'
(1413, 1426)

1444

Oog tan a-)' = 2 log tan c secJ x = 41g*ana!, (1415, '24, '30).
tun c
sin

SUCCESSIVE DIFFERENTIATION.

263

Some differentiations are rendered easier by taking the logarithm of the

function. For example,
1445
., .
therefore
therefore

y = y/(t^y' tWoro loe v = * loe C1 -"O-l los C1 +

'

1 dy = _
1 ^--^
2a! - _
3 (__^
2 ;
A = y =M^!) =

-8,(2-,*)

1446

y = (sin as)* ; therefore log y x log sin as ;

1.x
therefore

y, = log sinar+ cos as :

(1415, '24, '28)
y
Bin as
therefore
jU = (sin a1)1 (log sin a: + x cot a;) .
Otherwise, by (1420), y, = x (sin a:)*'"1 cos a; + (sin as)* log sin as
(1426)
= (sin x)' (x cot x + log sin a;) .

SUCCESSIVE DIFFERENTIATION.
1460

Leibnitz's Theorem.If n be any integer,

= yx+y(-i)xx+C (n, 2) y(_S!)Xir+ ...
...+C(n,r)

Proof.By Induction (233). Differentiate the two consecutive terms

C ("> r)
+ 0 (n + y<-,-l)xZ(r+l)>
and four terms are obtained, the second and third of which are
C(", r) y(.-r).*(fi). + 0 (, r + 1) &_,),(,,).
= {C(n, r) + C {n, r+l)}y[H_r)miT.1)a = C (n + 1, r+1) y(irn-7n) *<r+i)
by (102).
This is the general term of the series with n increased by unity. Similarly,
by differentiating all the terms the whole series is reproduced with n in
creased by unity.
DIFFEEENTIAL COEFFICIENTS OF THE nth ORDER.
1461
1462

(sin ax)nx = a" sin (ax+^nir).

(cos ax)nx = a" cos (ax+lmr).

1463

(e") = aTe**.

1464

(e"y) = ** (<*+<**)" y,

By induction
and (1428)"

(1426)

where, in the expansion by the Binomial Theorem, dxy is to

be replaced by yrx.
(1400, '63)

DIFFERENTIAL CALCULUS.

264

1465

(e" cos

where

= r*ef" cos (kr + n<),

a = r cos ^ and b = r sin ^.

Proof.By Induction. Differentiating once more, we obtain

r"e" {a cos (6x + n<p) 6 sin (6x + n^)}
_ r'e{Cos^ cos (Jz + n?) sin^ sin (bx+n<p)}
= r^'e-'cos (6x + w + l^).
Thus n is increased by one.
1466
1467

Cr"-1 log x) = \n-l -r- .r.

(1460), (283)

(1423>

1468

(tan-1 *)=(- 1)""1 |n-l sin" 0 sin n0,

where

0 == cot-1 x.

Proof.By Induction. Differentiating again, we obtain (omitting the

coefficient)
(n sin""1 d cos 8 sin nO+n cos nO sin" 6) 0r
= n sin"' 1 0 (sin nO cos 0 + cos nO sin 6) ( sin* 0).
Since, by (1437), 6. = -(1 + z5)-1 = -sin'0.
Therefore (tan-1*)^!,. = (1)" [n_sin"+I0 sin ( + l) 6,
n being increased by one.

1469

(y^rp)

= (-l)*l_8in*+10 sin(n+l)0.

1470

(rr-i)

= (-1)* \n_sin*^ 0 cos (n+1) 0.

PROO,-By (1460),

(1436,1468)

(^p)^

Then by (1469).
1471
*(-d* (l-^2)"-1 =

Jacobi's Formula.
1.3... (2n-l) sin (n cos"1^) + n.

Proof.Let y = \a?; therefore

(2T*)~ = -(2n+l) (jr*),..

Also (jT*).. = (yy-*).

Expand each of these values by (1460) and eliminate (y*"*)(_ijw the deriva
tive of lowest order. Call the result equation (1). Now assume (1471) true
for the value n. Differentiate and substitute the result, and also (.1471) on
the right side of equation (1) to obtain a proof by Induction.

THEOBY OF OPERATIONS.

265

1472 Theorem.If y, z are functions of x, and n a positive

integer,
zy = (yz)n (y,)(-D*+C (n, 2)(#*8x)(B_2),... +(-l)"ysBX.
Proof.By Indaction. Differentiate for x, substituting for zxy,^ on the
right its value by the formula itself.

PARTIAL DIFFERENTIATION.
1480 If u =f(x> y) be a function of two independent vari
ables, any differentiation of u with respect to x requires that y
sltould be considered constant in that operation, and vice versa.
d2n
Thus, -j-t or ulx signifies that u is to be differentiated succesaar
sively twice with respect to x, y being considered constant.
dsn
1481 The notation
or u^^* signifies that u is to be
differentiated successively twice for x, y being considered
constant, and the result three times successively for y, x being
considered constant.
1482 The order of the differentiations does not -affect the
final result, or
= uyx.
Pboof.Let u=f(xy); then u.=f(x + h- V]~f^' ^ in limit. (1484)
= ^ -fj* h y + *)-/("? + jO -f(x + h.y)+f(x,y) Jn lim.t
**
dy
hk
Now, if u, had been first formed, and then utI, the same result would have
been obtained. The proof is easily extended. Let ux = v ;
then
%, = vn =
= Uytz ; and so on.

THEORY OF

OPERATIONS.

1483 Let the symbols

% prefixed to a quantity, denote
operations upon it of the same class, such as multiplication or
differentiation. Then the law of the operation is said to be
distributive, when
* (n+y) = * (*)+* (y);
* See note to (1487).
2 M

266

DIFFFBENTIAL CALCULUS.

that is, the operation may be performed upon an undivided

quantity, or it may be distributed by being performed upon
parts of the quantity separately with the same result.
1484: The law is said to be commutative when
that is, the order of operation may be changed, 4> operating
upon ** producing the same result as ^ operating upon <&.
1485 mx denotes the repetition of the operation $ m times,
and is equivalent to <!><i> ... x to m operations. This definition
involves the index law,
which merely asserts, that to perform the operation n times in
succession upon x, and afterwards m times in succession upon
the result, is equivalent to performing it m + n times in suc
cession upon x.
1486 The three laws of Distribution, Commutation, and the
law of Indices apply to the operation of multiplication, and
also to that of differentiation (1411, '12). Therefore any
algebraic transformation which proceeds at every step by one
or more of these laws only, has a valid result when for the
operation "of multiplication that of differentiation is sub
stituted.
1487 In making use of this principle, the symbol of dif
ferentiation employed is -=, or simply dx prefixed to the
quantity upon which the operation of differentiating with
respect to a: is to be performed. The repetition of the operad?
d?
tion is indicated by -y-j-,
-j-j,
2d* s, &c, prefixed to the
function. An abbreviated notation is dx, d^, d^, d^^, &c.
Since dxXdx = dx in the symbolic operation of multiplication,
it will be requisite, in transferring the operation to differentia
tion, to change all such indices to suffixes when the abbreviated
notation is being used.
Note.The notation y yu, w^, da3v, &c. is an innovation. It has, how
ever, the recommendations of definiteness, simplicity, and economy of time in
writing, and of space in printing. The expression 55 requires at least
dX il If

fourteen distinct types, while its equivalent it^, requires but seven.

For

THEOBY OF OPERATIONS.

267

such reasons I have introduced the shorter notation experimentally in these

pages.
AH such abbreviated forms of differential coefficients as y y 'y . . . or y y y... ,
though convenient in practice, are incomplete expressions, because the inde
pendent variable is not specified.
The operation duty, and the derived function m^j,, would be move
accurately represented by (<PX)\ and (ul)\, the iudex as usual indicating the
repetition of the operation. But the former notation is simpler, and it has
the advantage of separating more clearly the index of differentiation from the
index of involution.
In the symbols y% and y^, the figure 2 is an index in each case : in tho
first, it shows the degree of involution ; in the second, the order of differentia
tion. The index is omitted when the degree or the order is unity, since we
write y and yt.
The suffix takes precedence of the superfix. y\ means the square of y.
,(y*) would be written (?/s), in this notation.
As a concise nomenclature for all fundamental operations is of great
assistance in practice, the following is recommended : or ym may be read
"y for x," as an abbreviation of the phrase, " the differential coefficient of y
for, or with respect to, x." Similarly, 4^-, ~ , ^ u , or the shortened
d'xT dxdy dx'dy'
forms y^, un, u.u3f, may be read "y for two x," "ufor xy," "u for two x, three
y," and so forth.
The distinction in meaning between the two forms y and y is obvious.
The first (in which n is numerical and always an integer) indicates n succes
sive differentiations for x ; the second indicates two successive differentiations
for the variables x and z.
d?^- or yM, may be read, for shortness,
The symbols di or y^, and
dx0
dx0
"y for x zero," "y for two x zero'"; d^^t^ (xy) can be read "d for two x
three y of <p (xy)."
Although the notation v, is already employed in a totally different sense
in the Calculus of Finite Differences, my own experience is that the double
signification of the symbol does not lead to any confusion : and this for the
very reason that the two meanings are so entirely distinct. Whenever the
operation of differentiation is introduced along with the subject of Finite
Differences, the notation
must of course alone be employed.

Thus, in differentiation, y
1488

The distributive law dx (u+v) = dxu+dxv.

1489

The commutative law dx (dyu)

or

1490
that is,

(1411)

= dt (dxu)

djyU

= dyxu.

(1482)

dmmdnxU

=d{m+n)xU.

(1485)

The index law


DIFFERENTIAL CALCULUS.

268

1491 Example.
(dxd,Y (d, d) (d, d,) = d,dmd.dyd,d,+drd, = itm2dm+&^.
Here djdt ci/Z, or dri/ = dt by the commutative laio.
(1489)
djit = du by the index late.
(l-i'JO)
Also (dj, 2tZ + d3>) it = t^M 2(ZM + dj w, by the distributive law.
Therefore, finally, (rfx (J,)'u = tZ^w 2cZ.n,M + <Z;iv?*.
Similarly for more complex transformations.
1492 Thus d, may be treated as quantitative, and operated upon as snch
by the laws of Algebra ; d" being written <Z,, and factors such as dxdt, in
which the independent variables are different, being written dn, &c.

EXPANSION OF EXPLICIT FUNCTIONS.

TAYLOR'S THEOREM.EXPANSION OP f(x + h).
1500
/(*+*) =/<*)+*/>>+ ^r(*)+ +/(+).
where 0 is some quantity between zero and unity, and n is
any integer.
Proof. (i.) Assume f(x + h) A+Bh+Ch*+&o.
Differentiate both sides of this equation,first for x, and again for h,and
equate coefficients in the two results.
1501 (ii.) Cox's Proof.Lemma.If f(x) vanishes when x a, and also
when x = b, and if f(x) and f (x) are continuous functions between the
same limits ; then f (x) vanishes for some value of x between a and b.
For f (r) must change sign somewhere between the assigned limits (see
proof of 1403), and, being continuous, it must vanish in passing from plus to
minus.
1502 Now, the expression
/(a+)-y(a)-./(a)-f>(a)
-[gl5Sr{/<-+)-/W-V(.)-...-^/'(.)J

vanishes when x = 0 and when x=h. Therefore the differential coefficient

with respect to x vanishes for some value of x between 0 and h by the lemma.
Let Bh be this value. Differentiate, and apply the lemma to the resulting
expression, which vanishes when x = 0 and when x = ddh. Perform the
same process n+l times successively, writing 8h for OOh, &c, since 0 merely
stands for some quantity less than unity. The result shews that
(+)--j=r {/(a+*)-/w-vw-- -fjr-w {
vanishes when x = Oh.
is proved.

Substituting Oh and equating to zero, the theorem

EXPANSION OF EXPLICIT FUNCTIONS.

269

1503 The last term in (1500) is called the remainder after

n terms. It may be obtained in either of the subjoined forms,
the first being due to Lagrange,
*Lf>(X+0h)

1504

or

(l-^.-y(+flk).

Since the coefficient , diminishes at last without

limit as n increases (239, ii.), it follows that Taylor's series

is convergent if f" (x) remains finite for all values of n.

1505 If iQ any expansion of f(x + h) in powers of h some

index of h be negat ive, then f(x), f (x), f" (x), &c. all become
infinite.
1506 If the least fractional index of h lies between n and
n+1; then f**1 (x) and all the following differential coeffi
cients become infinite.
Pkoof.To obtain the value of /"(), differentiate the expansion n times
successively for h, and put h = 0 in the result.

MACLAURIN'S THEOREM.
Put x = 0 in (1500), and write x for h; then, with the
notation of (1406),
1507 /<*) =f(o)+*f(o)+-^r(o)+...+^rm,
where 0, as before, lies between 0 and 1.
Putting y =f(x), this may also be written
1508

y-y+xJFrT^d4 + 1.2.3 d4 +

1509 Note.If any function f(x) becomes infinite with a finite value of
x, then f (x), f'(x), &c. all become infinite. Thus, if f (x) = sec-1 (1+x),
f(x) is infinite when x = 0 (1438). Therefore /"(0), / (0), &c. are all
infinite, and /(*) cannot be expanded by this theorem.

Bernoulli's Series.Put h = x in (1500); thus,

1510 /(o) =/w-<w+^rw-nj/+^

DIFFERENTIAL CALCULUS.

270

1511

if </>

= 0 and (y) = *; tlien

&=

1.2.3 ^s*+&c-

Proof.Let y = frl (a) =/(*), and let y + *=/(a)+A);

therefore
r+fc = ^(i/ + fc)=0.
Therefore y + fc =/(0) =/(x) -*/'(*)+

/"()- Ac-, by (1510);

which proves the theorem.

EXPANSION OP f(x + k, y + h).
Let f(xy) = i*.

Then, with the notation of (1405),

1512
f(x+h,y+k) = u+(hux+kut)+^T2(hsult+2hkuxy+^ut0)

1513

The general term is given by . (hdx-\-kdy)*u,

1n

where, in the expansion by the Binomial Theorem, each index

of dx and dy is changed into a suffix ; and the coefficients
dx, dtx, &c. are joined to u as symbols of operation (1487) ;
thus ua is to be changed into u^.
Proof.First expand f(x + h, y + k) as a function of (x + h) by (1500);
thus, f(x + h, y + k) =f(x, y + ) + hf. (x, y + fc) + ^
(x, y + k) + Ac.
Nest, expand each term of this series as a function of (y+k).
writing u for f(xy),
f(x,y + k)=

u +

}>f.(z,y+k)=

hu, +

ku, + Xk*ih,
hku^, +

J- fu (*, y + k) = j-j h*thM + ^ Vku.^ +

~WuM>
Ir.

+^k'uy, +

Thus,
+.

+,4-''^"^+
liL

Wufcjy +

h*
11
^fu(*,y + k) = jj- '''"a. + r fc^-Mte, +
^ /*(* 2/ +*) =

^4"i+

The law by which tlie terms of the same dimension in h and k are formed,
is seen on inspection. They lie in successive diagonals ; and when cleared
of fractions the numerical coefficients are those of the Binomial Theorem.

EXPANSION OF EXPLICIT FUNCTIONS.

271

The theorem may be extended inductively to a function of

three or more variables. Thus, if u =f (x, y, z), we have
1514
f(x+h, y+k, z+l) = u+{hux+ku,+luz)
+i (hiu^+kSuH+Pu^+2klu+2lhii,x+2hkuxy)+... ,
the general term being obtained as before from the expression
J- {hdx+ kdt+ld,)nu.
L?L
1515 Cor.If u =f(xyz) be a function of several inde
pendent variables, the term (hux-\-kut + luz) proves, in con
junction with (1410), that the total change in the value of u,
caused by simultaneous small changes in x, y, z, is equal to
the sum of the increments of u due to the increments of x, y, z
taken separately and superposed in any order.
This is known as the principle of the superposition of small
quantities.
1516 To expand f(x, y) or f(x, y, z, ...) in powers of x, y,
&c, put x, y, z each equal to zero after differentiating in
(1512) or (1514), and write x, y, ... instead of h, h, &c.
1517 Observe that any term in these series may be made
the last by writing x + Qh for x, y + 9k for y, &C, as in (1500).
SYMBOLIC FORM OF TAYLOR'S THEOREM.
The expansion in (1500) is equivalent to the following
1520

f(*+h) =

Proof.By the Exponential Theorem (150), writing the indices of d. as

suffixes (1487),
eW'/(*) = (l+hd.+ih'd* +.-..)/(*) =f(x)+hf,(x)+lh%(.x)+...,bY(im).
Cor.
A /(*) =/(*+*)-/(*) = (eM'-l)/(.r),
therefore

A2/(,r) = {eM'-\ff{x),

and generally An/(.r) = {eM,l)nf{x),

the index signifying that the operation is performed n times
upon / (*).
1521

Similarly f(x+h,ij+k) = ehd-+'"i<>f(x,!/).

Peoof.
^'+ify(,y) = {l + hd,+kd, + l(hd.+kd,y + l(hd.+kd,y+...}f(x,y)
= f(.x+h,y + k), by (150) and (1512).

DIFFEllENTIAL CALCULUS.

272

1522

And, generally, with any number of variables,

Cor.As in (1520),
(*,*....) = (cM-+M'+-l)/(af,y,...).

1523 If u =f(x,y) = <p(r,6), where * = rcos0, y = rsin0;

and if x = r cos (0 + w), y' = r sin (0 + w) ; then f(x'y') is
expanded in powers of by the formula

Proof.By (1520), r being constant,

f (r, +) = e"'' f (r, 0) = e-'VOs, y).
Now x and y are functions of the single variable 0; therefore
u, u^x. + u,)/, = u, ( r sin 6) + , (r cost)) = xu^yu^
The operation will be transformed by the same law (1492) ; therefore
d, = xd,ydT; therefore
f(x',y') =e(xd'-*dJf{r,y) = l+w(xu,-yut) + Wi.*->h,-2*y^ + yW) + &c.

1524 Examples. The Binomial, Exponential, and Loga

rithmic series for (l+x)n, a", and log(l+,r), (125, 149, 155),
are obtained immediately by Maclaurin's Theorem (1507) ; as
also the series for sin a; and cos a; (764), and tan-1* (7U1).
The mode of proceeding, which is the same in all cases, is
shewn in the following example ; the test of convergency
(1504) being applied when practicable.
1525

tan* = * + i

+ 1 * + ^ a? +

to.

Obtained by Maclaurin's theorem, as follows :Let

/ (?) = tan x = y )
Therefore yt = z and z, = 2yz ;
j
_ seci x z ) y and z being used for shortness.
f (x)
f'(x)
f (x)
f (x)

= 2 sec' a: tan x = 2y:,

= 2(zy,+yzr) = 2(z' + 2yV))
= 2 (%z+ V+ Vz) = 8 (2y**+y*).
= 8(2zs + 8yV+3i/y + 2i/*z) = 8(2z*+Uy'xi + 2y'z),

/*(*) = 8(\2yz* + 22yz+...) = 272y*+...,

/"(*) = 272z'+...&c.,
the terms omitted involving positive powers of y, which vanish when x is
zero, and which therefore need not be computed if no term of the expansion
higher than that containing x7 is required.

EXPANSION OF EXPLICIT FUNCTIONS.

273

Hence, by making x = 0, and therefore y = 0 and 2=1, we obtain

/(0) = 0; /(0) = 1; /"(0)=0; /'"(0)=2; /"(0)=0; /*(0) = 16;
/"(<>) =0; /TU(0) = 273.
Thus the terms up to z7 may be written by substituting these values in
(1507).
In a similar manner, may be obtained
1526

sec * = 1 + 1 * + A a* + iL

+ &c

Methods of expansion by Indeterminate Coefficients.

1527 Rule L.issttme f (x) = A + Bx + Cx24-&c. Differ
entiate both sides of the equation. Then expand f'(x) by some
htmvn theorem, and equate coefficients in the two results to
determine A, B, C, fyc.
1588

Ex.

8in-., = , + .^ + ^| + y^ + &c.

Obtained by Rule I. Assume

sin-'z = A + Bx+Cx,+Dz'>+Ext+F,i+
Therefore, by (1434), (l-x1)'* = B+2Cx + 3Dx' + 4Exs+5Fxl+
Bat, by Bin. Th. (128), (1 -)"* =

14 '+ 144 *+

Equate coefficients ; therefore 5 = 1; 0 = 0; B = --z\ E = 0; F=

;
2.3
2.4.5
&c. By putting x = 0, we see that A=f(0) always. In this case
A = sin"1 0 = 0.
In a similar manner, by Rule I.,
1529

1530 Rule II. Assume the series, as before, with unknovn

coefficients. Differentiate successively until the function re
appears. Then equate coefficients in the two equivalent series.
1531 Ex.To expand sin x in powers of x.
Assume
sins = A + Bx+Cx'+Dx,i + Ext+Fx'k +
Differentiate twice,

cos* = B + 2Cx+ W.e2+ 4K.,'-t 5Fx'+

-sins = 2C+3.2Dz + 4.3^J + 5.4iV+

Put x= 0 in the first two equations ; therefore A 0, B=l.

Equate coefficients in the first and third series.
2 N

274

Thus

Therefore

DIFFERENTIAL CALCULUS.

-2C=A,

.:C-0;

-S.2D=B,

-4>.ZE = C,

.'. E = 0;

-5AF=D,

'D=-1^;
JF = -

-, &c.

sin a; = * . ^ - + = . * . , Ac., as in (764).

1.2.3 1.2.0.4.6

1532 Role III.Differentiate the equation y = f (x) twice

with respect to x, and combine the results so as to form an
equation iny, yx, and y2x. Next assume y = A-f-Bx + Cx4-f-&c.
Differentiate twice, and substitute the three values of y, yx, y^
so obtained in the former equation. Lastly, equate coefficients
in the result to determine in succession A, B, C, Ifc.
1533 Ex.To expand sin mO and cos mO in ascending powers
of sin 9 or cos 0.
These series are given in (775-779). They may be obtained by Rule III.
as follows :
Put
x = sin 0
and
y = ammd = sin (m sin-1 a;).
Therefore

y, = cos (m sin"1 x)

.
(1434)
(i.)
vl-is'
yu = sin (m sin 1 x) - to' 1 +cos (to sin x)
-.
1 a
(1 ar3)*
Therefore, eliminating cos (m sin_1x), (1 a;') y2,xy,+m*y = 0
(ii.)
Let
y = A+A& +Af?+
+ ^*"+
(iii.)
Differentiate twice, and put the values of y, y and
in equation (ii.) ;
thus 0 = r>i1(A + Alz+A,x'+Aix,+ ...+A,x" +
)
-x(Ai + 2Aix + 3Alz*+...+nAHx"-i+
)
+ (l-a>,){241+2.3^+
+ (n-l)^la!-J+n(n+l)A.i^-1 + (n + l)(+2) AK.1x"+...}.
Equating the collected coefficients of x" to zero, we get the relation
i"'=(fl)(n-ri!)i'

(1Y)

Now, when x = 0, y = 0; therefore ^4 = 0, by (iii.). And when x = 0,

yx = m, by (i.) ; and therefore A, = m, by differentiating (iii.). The relation
(iv.) furnishes the remaining coefficients by making n equal to 0, 1, 2, 3, Ac.
in succession.
Cos mO is obtained in a similar way.

1534 Rule IV.Form the equation in j, yx, and y^, as in

Rule III. Talce the nth derivative of this equation by applying
Leibnitz's formula (1460) to the terms, and an equation in

EXPANSION OF EXPLICIT FUNCTIONS.

275

y(n+ u Jfn+Dn an<L Jnx is obtained. Put x = 0 in this; and

employ the resulting formula to calculate in succession jSx0,
Juts S'c- 171 Maclaurin's expansion (1507).
1535

Ex.

C*'-1* = 1 + ax +
, as(a2+22)

+ a @1 a*

4 , a(a2+l)(a2+32)

. . .

Obtained by Rule IV.

Writing y for tbe function, the relation found is

(1 z!) y^xy.a^y = 0.
Differentiating n. times, by (1460), we get
(1ar) ylm.v. - (2+ 1) ajy{ti,. (ft'+n') yM = 0.
Therefore yt*,T,z<> = (<*+') y^, a formula which produces the coeffi
cients in Maclauriu's expansion in succession when y^ and y^ have been
calculated.
1536

ARBOGAST'S METHOD OF EXPANDING <f>(z),

where

z= a+

-f y^z? + t "?3

<r3+ &c

Let y = /> (z).

When a: = 0, y = <j>(a) ;
Maclaurin's theorem (1508),

(i.)

therefore, by

y = 4> () 4- xy* + Y^y^ + 17273 y4r0 +

Hence, in the values of

y^, &c, at (141 G), x has to be
put = 0.
Now, when x=0, z=a; therefore y,, y^, &c. become <j>'(a],
f'\a), &c. ; and z^, z.^, z^, &c. become alt at, a3, &c. Hence
y*o = 4> () on
#sxo= f (o) a2 + <f(a)a2,
y&ro = <" () o* + 3<" (a)

1537

+ <' (a) a3, &c.

Exam ple .To expand log (a + bx + c.b5 + du? + &c.) .

Here a, = b, a,= 2c, a, = 6(2, ^'(a)=. ?" () = ~ 7i> f"(a) = ^r6t

ft
ft
lncreforo

yM = , y^ =
H
, 2/3*1 = nf
j H
.
a
a.
ft
ft
ft
ft
Therefore, substituting in (ii.)i we obtain as far as four terms,

DIFFERENTIAL CALCULUS.

276

1538

Ex. 2.To expand (a + a^ + 0^+ ...+amx")r in powers of x.

Arbogast's method may be employed ; otherwise, we may proceed as

follows. Assume (a0+axx -(-o^c5+ ...")' = A0+Alx-\-A^e,+
Differentiate for x; divide the equation by the result; clear of fractions,
and equate coefficients of like powers of x.
BERNOULLI'S NUMBERS.
1539

ji, = 1 -

-B. | +B, |- to,

where B2, Bt, &c. are known as Bernoulli's numbers.

values, as far as BIS, are

Their

B*=io> 2?8=42' Bs = 30' Bw = 66'

_ fiM
J 12 ~ 2780'

t? _ 7
14 ~ 6 '

7? _3617 _ 43867
18 ~ 510 ' ^18~ "798"'

They are found in succession from the formula

1540

C (n, 2)

C (n, 3) J3M_34- ...

... + C(n,2)Bt-\n+\ = Qt

the odd numbers J?3, Bs, &c. being all zero.

Proof.Let y =

Then, by (1508),
T i.
2/ = y<>+y<*+y*<> ^ + 2/3.0 jj- +

^ + &c.

Here j/^ = ( \)'*lBlr. Now y0 = 1 and yrf = \, by (1587). Also

ye1 = ?/ + ar. Therefore, by (14(J0), differentiating n times,
<? {>/+
+ C (m, 2)?/(.2),+ ... + ny, + ?/} = yM.
Therefore
y(,,-i)o + C7(, 2)
... +nyM+y0 = 0.
Substitute J?-i, -Z>'-j, &c, and we get the formula required.
B3, Bt, B7. &c. will all b found to vanish. It may be proved, a priori,
that this will be the case : for
1 *S4-1
-ir + * = i I.
LO*A
e*-l
2
2 <r*-l
Therefore the series (1-539) wanting its second term is the expansion of the
expression on the right. But that expression is an even function of x (1401) ;
changing the sign of x does not alter its value. Therefore the series in ques
tion contains no odd powers of x after the first.
1542 The connexion between Bernoulli's numbers and the sums of the
powers of the natural numbers in (276) is seen by expanding (1 e1)"1 in
powers of e', and each term afterwards by the Exponential Theorem (150).

EXPANSION OF EXPLICIT FUNCTIONS.

277

1543
^=f-Bl(,-1)|+.,,-1)g-B,,-1)|+&0.
1544
^} = 2[^(2!-l)|--JB^-l)^+l?6(2''-l)^-&c.].

1545

4l = ?=i - i and 7TI = x" 1 7TV and by (1539)-

1+2^+-^ + ^; +

- i22*2n-B-

Peoof.In the expansion of ^- (1540) substitute 2t'0 for x, and it

becomes the expansion of 0 cotfl (770). Obtain a second expansion by differ
entiating the logarithm of equation (815, sin 0 in factors). Expand each
term of the result by the Binomial Theorem, and equate coefficients of like
powers of 0 in the two expansions.
STIRLING'S THEOREM.
1546
where

(*) = Af (x)+A,h {f (*+A)-f (*)}

+A,h* {*"(*+*)-*" (*)J+&o.f
= ( l)B#2n-r [2n_ and ^2B+1 = 0.

Proof. .4,, .4,, &c. are determined by expanding each function of

by (1500), and then equating coefficients of like powers of x. Thus
k-A*=0i

i-^-A^-'

[t-ol-[H'=0;

&c-

To obtain the general relation between the coefficients : put <p (x) = e",
since Alt Av &c. are independent of the form of </>. Equation (1546) then
produces

= 1 AJi AJt' AJf Ac. ;

c L
and, by (1539), we see that, for values of n greater than zero,
A2ntl = 0 and A*, = ( !)".%, -f-| 2w.
BOOLE'S THEOREM.
1547

{f
(*)}
+ 4tfc jf(*+A)+fW} +&c.

Peoof.Av Ar A &c. arc found by the same method as that employed

in Stirling's Theorem.

278

DIFFERENTIAL CALCULUS.

For the general relation between the coefficients, as before, make ^ (x) = e't
and equation (1547) then produces
-^=1 = AJi+AJt+AJP + Sn. ;
and, by comparing this with (1544), we see that
A1H = 0 and
= (-1)"-'^ j^.

EXPANSION OF IMPLICIT FUNCTIONS.

1550 Definition.An equation / (x, y) = 0 constitutes y
an implicit function of x. If y be obtained in terms of x by
solving the equation, y becomes an explicit function of x.
1551 Lemma. If y be a function of two independent
variables x and z,
dAF(y)y.}=d.{F{y)yx}.
Proof.By performing the differentiation s, we obtain
F'(y)y,y, + F(y)y and JP'(y) y.y.+ F{y) ym,
which are evidently equal, by (1482).
LAGRANGE'S THEOREM.
1552 Given y = z-\-xj>(y), the expansion of u=f(y) in
powers of x is
Ay) = /() +** ()/ to + + 1 a [ {* (*) }V] +
PitOOF.Expand u as a function of a;, by (1507) ; thus, with the notation
aj*
xn
of (1406),
m = M,+aJj0+ io+ ... + wo+&c.
Here 0 is evidently / (z).
Differentiating the equation y = z + x<j> (y) for as and z in turn, we have
y* = 1> (.y) + x (y) y- and y. = l
(2/) y.Therefore y, = ^ (y) y, ; and, since ux=f(y)y, and m, =f(y)ya
therefore also
mx = <f> (y) u,
(i.)
The following equation may now be proved by induction, equation (i.)
being its form when n = 1.
Assume that
=
[{-p (?)}" ]
(ii.)
Therefore
u,... = dl9_1)md. [{f (y) }" u,] (1482)
= io.-ii.'i. [{* (?)}",] (1551) = d. [{f (y) }"",], by (i.)

EXPANSION OF IMPLICIT FUNCTIONS.

279

Thus, n becomes
But equation (ii.) is true when n = 1 ; for then
it is equation (i.) ; therefore it is universally true.
Now, since in equations (i.) and (ii.) the differentiations on the right are
all effected with respect to z, x may be made zero before differentiating instead
of after. But, when as = 0, u, = /'(z) and ^(j/)=^(z), therefore equa
tions (i.) and (ii.) give
- = 9 CO/ 00 ;
= <*c-i). [{ CO}"/ CO]1553

Ex. 1.Given y'ay + b = 0 : to expand logy in powers of .

b -I-
t/* ; therefore, in Lagrange's formula,
Here y
x = ;
a
Therefore

z = ;
a

f(y)=\ogy;

$>(*/) =/;

and

y = z+xy\

it, = log z ;

.= i,.-i (*"

= z* = z' ;
z
= (3n_1) (3u-2) - (2'l+1) **

Therefore, substituting the values of a and z, (1552) becomes

,
, b . b' 1 ,
, (3-l) (3-2) ... (2 + l) 6s" 1 ,
l0gl/ = loga +
a*
a + ... +^
1! . 2 . . '. " ^-^n,
arn
a" +

1554

Ex. 2.Given the same equation : to expand y* in powers of .

/ (y) is now y", and, proceeding as in the last example, we find

, 6" f , , 6s 1 , n(n + 5) 64 1 , n(n + 7)(n + B) b' 1
y= !+-; I^
g-fH
r^r
a" I
ci" a
1.2
a* a*
1.2.
n(n + 9) (n+lOj (n + 11) fc" 1
+
r^nr-.
^^+&c-j
1.2.3.4
If n = 1,

y = 1 1+ +3 + 12- - + 55-, +&c

a \
a a
a* a"
a0 a
a" a

CATLEY'S SERIES FOR .

1555
^=^-~[^()]8.+...+^^[{^)r"i]+...,

where

J. = ^-r.
*(0)
Pkoof. Differentiate Lagrange's expansion (1552) for z, noting that
Replace x by
- Put / (y) = ; and therefore

/(O = r-r, since / is an arbitrary function.

?(2)

Then make y=0.

280

DIFFERENTIAL CALCULUS.

LAPLACE'S THEOREM.
1556

To expand / (y) in powers of x when

y = F{z+xtf> (y)}.

Rule. Proceed as in Lagrange's Theorem, merely sub

stituting F (z) for z in the formula.

1557 Ex. 3.To expand e" in powers of 3 when y = log (z+x siny).
Here
f(y) = e'; F(z) = logz; </>(</) = siuy;
In the value of
(1552), if (z) becomes f {F (z)} = sin logz;
/(z) becomes /{-F(z)} = e"*' = *; therefore /'(z) = 1.
Thus the expansion becomes
xn
e = z + 3 sinlogz + ... + j <*<-i), (sin logz)*.
[n_

1558 Ex. 4. Given sin y = x sin (i/ + a) : to expand y in powers of x.

Here
y = sin-1 (as sin */ + <z), with z = 0.
f(y) = V, F (z) = sin"1 z ; ^ (1/) = sin (f/ + a).
^> (z) in (1552) becomes <p{F(z)} = sin (sin-1 z + a),
/(z) becomes /{.F(z)} = *'(*) = sin^z ; therefore f(z) = (1-z1)-*.
Thus
y = x sin (sin-1 z + a) (1 z')-^
+ fzs<Z>{sin1 (sin-l*+o)(l-,)~*}+4x*<t {sin* (sin"1 z + a) (1 -**)""*; +with * pat =0 after differentiating. The result is, as in (796),
y = x sin a + \x* sin 2a + \x* sin 3a + Skc.

BTTRMANN'S THEOREM.
1559 To expand one function / (y) in powers of another
function \p (y) .
Rule.Put x = \p (y) in Lagrange's expansion, and there
fore <p (y) = (y z) -rrf, (y) ; therefore
1560 /(y) = /(*)+* {J^/W}

Here ?/ = z signifies that after differentiating z is to be sub

stituted for y.

EXPANSION OF IMPLICIT FUNCTIONS.

1561 Coe. 1. Since x = *(?/), V =

becomes, by writing x for $(y),

281

I therefore (1560)

/!r.W} =/+...+ ^{(jg)Vw}.+...

But since the variable ?/ is changed into z after differentiating,
it is immaterial what letter is written for y in the second
factor of the general term.

1562

Coe. 2.If f(y) be simply y, the equation becomes

1563 Cos. 3.If z = 0, so that y = X(j>(y), we obtain the

expansion of an inverse function,

1564 Ex. 5.The series (1528) for sin_1a! may be obtained by this
formula ; thus,
Let sin-1 x y, therefore x = sin y = \p (y), in (1563) ; therefore
sin-'* = * -JU
\sint//,y-o

1565

Ex. 6.-If y =

1.2 Vsin'y/irt

+TT^[-^-) +&c1.2.3 \ sin1 yl w

then, by Lagrange's

theorem (1552), since y -$- + ^rx. we

. fx\'. /a>\"*
w|w+2r-l
Pnt * = 2Jt, thus
1566
(i-v/(i-^)r=r+wpt,+
\
2
/
Change the sign of n, thus

+i.+2r-ir<f+
I r I w+r

1567

(1+ ^-^)" = 1
(-ir
^*
\
2
/
I r_ I nzr
7t + 1 or ft
"4~ 1 terms, according as n is even
This last series, continned to
or odd, is eqnal to the sum of the two series, as appears by the Binomial
theorem.
2 0

DIFFERENTIAL CALCULUS.

282

Also, by Lagrange's Theorem,

1568

,ogJ = kgf + (|),+ ... + |^(f)%

or, by putting x = 2 </t,

1569

1,1=^=^-,+ ... + ^+

1570 Ex. 7.Given xy = log y ; to expand y in powers of x.

The equation can be adapted as follows :
y = s**,
therefore xy = ze*".
Put ary = j/', therefore y' = ze"', from which, by putting z = 0 in (1552),
y may be expanded, and therefore y.
Ex. 8.To expand e* in powers of ye*.
Here = ye*, f (y) =
= e"**, if we take t = 0.
1571

1572

e" = l + aye^ + a (a -26)

Therefore

+ o (a- 36)'

ABEL'S THEOREM.
If <P (*) be a function (developable in powers of e*; then

^ (*+a) = * (*)+a*'(*+6) + a("~2&) f (*+25)+...

, a (a r6)r 1 ,r, ,
.
+ t 2
y <f>r(*+rb) +
ProofLet
f (y) = A0 + A^ + A^ + A^" +
(i.)
Put y = 0, 1. 2, 3, &c. in (1571), and multiply the results respectively by
Aq, A1e', A^e11, Ac. Then the theorem is proved by equation (i.).
1573

Coe.If </> (x) = xn, Abel's formula gives

+ C(n,r)a(a-rb)r-' {x+rb)r+ &c.

INDETERMINATE FORMS.
1580 Forms -JJ-,

*.

RULE._# lM Je a fraction which

iaies etJfter o/ these forms when x = a;

ii^ = S^l2l or
<a" (a)
^ (a)
^ (a)
^ j the first determinate fraction obtained by differentiating

INDETERMINATE FORMS.

283

the numerator and denominator simultaneously and substituting

a for x in the result.
1581 But at any stage of the process the fraction may be
reduced to its simplest form before the next differentiation.
See example (1589).
Pboof.(i.) By Taylor's theorem (1500), since <p (a) = 0 = \p (a),
<j> (a + h) _ q> (a) + h<p' (a + Oh) _ <j>' (a + Bh) _ $ (a)
\p(a + h) ~ 4,(a)+hi (a + Oh) ~ \j,< (a + Oh) ~ ^'(a)'
when h vanishes.
<M H <>-*<>.

^ = ^rr^y

which is of the first form, and therefore

_
f(o) {f (a) ridli-)
-{T(a)p
(14l4) "{*()} ^'(<0
f'()'

Therefore
^1
^%
(a) " f ()'

1582 Vanishing fractions in Algebra are of the indeter

minate form just considered, and may be evaluated by the
rule, or by rejecting the vanishing factor common to the
numerator and denominator.

x'as
0 = (x
3
Ex.When
x=a;
-5 5 = > a)/v(x' + ax + a')' = 3as- = a.

1583

Form 0 X oo.

Rule.If <p(x)X^ (x) takes this form

when x = a, put <f> (a) X ^ (a) = ^> (a) -f-

, ivhich is of the

/om .
1584 Forms 0, oo, 1. Rule.If {*(x)}*w fofces any 0/
<Aese forms when x = a, ./md </te Zimii 0/ Me logarithm of the
expression. For the logarithm = i(a) log^(a), which, in each
case, is of the form 0 X 00 .
1585 Form oc oo. Rule. If <f>(x) \(,(x) takes this form
when x=a, we have e*(a)"*(a) = 77-7 = ; and if the value of
this expression be found to be c, by (1580), the required value
will be log c.
1586

Otherwise : <(>(&) $ (a) = <j> (a) j 1

tAe/om 00 XO (1583).

C f wfticA i of

DIFFERENTIAL CALCULUS.

284

1587

Ex. 1.With x=0,

y=
e* 1

= ~=- (1580) = 1.
0
e*

Also, with x = 0,
_ e'lxe' _ 0 _ e'-e'-xe' _
x _ 0 _
2/'0- (^-l)1
0
2(<r,-l)e" 2("-l)
0~

1 _ _1
2e* ~ 2'

1588

Ex. 2.With a = 1 ;
log a; = 0 = cos1 ^>
(rx) (1580)
,-,eans = -.
1
cot (w) log* = -etan {-nx)
0
rx
w

1589

Ex. 3.With x = 0;

'
c
mar"
( )"-"
'
by (1581), differentiating n times and reducing the fraction to its simplest
form after each differentiation.
1590 Ex. 4.With x = 0 ;
y = (1 + az) = 1*.
By (1584), logy = lgV + *) = <> = a
as
0
1+

=a.

1591 Ex. 5.With z =

y = (T-a;)"" = 0.
By (1584), log y = sin x log (tt- z) = lo? ('-) =
cosec x
oo
(x x) cos as
/ 1 - ^, ,\
0
z sm a: cos ?
.-,
(1580) = -V = cos x (v x)
. . y = \. .
sin x = 0 ;
1592

If /(*) and x become infinite together, then

* =/(*) =/(*+!)-/().

Proof.-

= f
(1580) =/^ + 1)~/^) (1404),
(S
00
1
1
since, when a: = oo , h may be taken = 1.

1592

Indeterminate forms involving two variables.

Rule. First : If the values x=a, y=b make the frac

tion ^~^^ = "^"' the true value is =: ~, if }t and \f/9 both

vanish.
1593 Secondly : If $x : ix = }t :
the fraction is k.
Peoof.(i.) By (1703)

= k,

Ae true value of

= f'+fr^, and y being an arbitrary

function of ,that is, independent of z,the value of the fraction is inde

terminate unless and
both vanish.
(ii.) If we substitute <pt = k\p, and <(> = hp^ the fraction becomes = it.

JACOBIAN8.

285

JACOBIANS.
1600 Let u, v, w be n functions of n variables x, y,z (n=B).
The following determinant notation is adopted :
Xu X- X..
d (uvw)
~ d {xyz)
Wx U'y Wg

_
y yv y*
Zu Zv Su,

d(xyz)
d (uvw)

The first determinant is called the Jacobian of u, v, to with

respect to x, y, z, and is also denoted by J (umv), or simply
by/.
d (uvw) ^ d (xyz)
j
d (xyz)
d {uvw)
Proof.If the product of the two determinants be formed by the role in
(570), first changing the columns into rows in the second determinant (559),
the first column of the resulting determinant will be
u,x + u,y +sz = ^ ") and the whole
V. wu
1 0 0
u,x, + u,y, +u,zv = t, > , determinant
0 1 0 = 1.
. w.
u,x,+ uty, + u,z = uK)
will be
V. w
0 0 1

1601

Theobem.-

1602 ltu,v,w are n functions of n variables a, (3, y (=3),

and a, /3, -y, functions of x,y,z;
d (uvw)
d (afty) _ d (uvw)
d (ay)
d (xyz) ~ d (xyz) '
Proof.Form the product of the two determinants, changing columns
into rows in the second as in (1601). The first colnmn of the resulting
determinant will be
u, v, wz
u.at+UpPt+u^y, =
an<i *ne whole ded (uvw)
.o+mh''* + mt7 = u [ i
terminant will
v, w,
~~
d(xyz)'
uc v, w,
+ ft +w,y = u, )
be
since rows and columns may be transposed (559).
1603 Cob.If a, /3, -y are only given as implicit functions
of x, y, z, by the equations <j> = 0, \ = 0, ^ = 0, involving the
six variables ; then
ii*xL y d (aPy) = / i\ d (tvf)
d(a(3y) A d(xyz)
K
> d(xyz) '
Proof.By (1737), <t>.a,+<t>ffi, + <t>,y, = <t> where f, is the partial
derivative of <)>. Thus u v w in the determinant, are now replaced by
1>'> ~X*> ~ 4>x i 80 with y and z ; and by changing the sign of each element,
the factor (1)* is introduced (562).

286

DIFFEREXTTAL CALCULUS.

1604 If , v, w, n functions of n variables x, y, z (n = 3), be

transformed into functions of , n, t by the linear substitutions
x= Oif+<V?+s{')
y = b^+btTj+b^ i ;
z = c^+CzT) +<*3 J

t^en
or

d(uew) _
d(urw)
d (ftf)
d (xyz) '
J' = MJ,

where M is the determinant (a^Cg) called the modulus of

transformation (573).
Peoof. J =
.
to.

2f = <*i &i ci
a, 6, c,
a, 6, c,

J* = ut

w.
,
Form the product If/ by the rule in (570). The first element of the
resulting determinant is u^a, + utbl +
= i*,xf + u,y^+ulz^ = u^. Similarly
for each element. Then transpose rows and columns, and the determinant
J' is obtained.
1605 When the modulus is unity, the transformation is said
to be unimodular.
1606 If5 in (1600), <f> (uvw) = 0, where <p is some function ;
then J (uvw) = 0; and conversely.
Proof.Differentiate <p for as, y, and a separately, thus
1>., + <P.v, + <l>wv>, = 0;
similarly y and z ; and the eliminant of the three equations is J (uvw') = 0.
1607 If = 0, v = 0, w = 0 be a number of homogeneous
equations of dimensions m, n, p in the same number of vari
ables x, y, z; then J (uvw) vanishes, and if the dimensions
are equal
Jy, Je also vanish.
Proof.By(1624), xu,+yu,+zu, = mu~\
xv, +yvt + zv, = nv > ; .'. Jx = Almu + Blnv + Ctpw.
xw, + yw + zw, = pw J
By (582), Av Blt Cx being the minors of the first column of/. Therefore,
if u, v, w vanish, J also vanishes.
Again, differentiating the last equation, J+xJz = Almu, + Binv,+ C1picm.
Therefore, if m = n = p, J+xJ, = m (il,,+B,t>#+ C,wx) = mj.
Therefore J, vanishes when / does.
1608 If m = 0, v=0, w=0 are three homogeneous equations
of the second degree in x, y, z, their eliminant will be the
determinant of the sixth order formed by taking the eliminant
of the six equations u, v, w, Jx, Jt, /.

QUANTICS.

287

Proof.J is of tbe third degree, and therefore Jz, J, J, are of the second
degree, and they vanish because u, v, w vanish, by (1607). Hence u, v, w,
J Js, J, form six equations of the form (x, y, z)j = 0.

1609 H n variables x, y, z (n = 3) are connected with n

other variables , y, I, by as many equations u0, v=0, w=0 ;
tben

rfcr dy dz _ d (time) _^ d (uvte)

d drj d
d(r)Q ' d (xyz) '

Peoof.By (562) we have

. * M,
Ax dy dz
V,
=
di di\ d
w, wr W,

u,z(
v,x(
v,z{ =
w,x( vy, w,z{

n(
vi

vi
%

QUANTICS.
1620 Definition.A Quantic is a homogeneous function of
any number of variables : if of two, three variables, &c, it is
called a binary, ternary quantic, &c.
The following will
illustrate the notation in use. The binary quantic
oar* + 3bx*y + 3cxy* + df
is denoted by (a, b, c, d\x, y)s when the numerical coefficients
are those in the expansion of (x+y)s. When the numerical
coefficients are all unity, the same quantic is written
(a, b, c, d\x, y)s. When the coefficients are not mentioned,
the notation (x, y)z is employed.

ETJLER'S THEOREM OP QUANTICS.

If u =f(x,y) be a binary quantic of the ntb degree, then
1621
1622

xwx+yuy = nu.
. ^uix-\-2jeyuxv-\-yiuiy==n(n l)u.

1623

(xdx+ydv)ru = n (n 1) ... (n r+1) .

(1492)

Proof.In (1512) pat h = ax, h = ay; then, because the function is

homogeneous, the equation becomes
(l + a)*tt = t*+a (xu, + yuy) + |a' (a'ttj, + 2xyury + y*u*) +
Expand (1 + a)*, and equate coefficients of powers of o.

288

DIFFERENTIAL CALCULUS.

The theorem may be extended to any quantic, the quan

tities on the right remaining unaltered. Thus, in a ternary
quantic u of the nth degree,
1624

<M**4-y**+*w = nu ;

and generally

1625

(.vdx+ydy+zdzyu = n (n-1) ... (n-r+1) u.

Definitions.

1626 The Eliminant of n quantics in n variables is the

function of the coefficients obtained by putting all the quantics
equal to zero and eliminating the variables (583, 586).
1627 The Discriminant of a quantic is the eliminant of its
first derivatives with respect to each of the variables (Ex.
1631).
1628 An Invariant is a function
equation whose value is not altered
of the equation, excepting that the
the modulus of transformation (Ex.

of the coefficients of an
by linear transformation
function is multiplied by
1632).

1629 A Covariant is a quantic derived from another quantic,

and such that, when both are subjected to the same linear
transformation, the resulting quantics are connected by the
same process of derivation (Ex. 1634).
1630 A Hessian is the Jacobian of the first derivatives of a
function.
Thus, the Hessian of a ternary quantic u, whose first
derivatives are ux, ut, u, is
vxy "xz
,
Uyx Uiy Uyz
d(xyz)
^zx Wzy "i;
1631 Ex.Take the binary cubic u =
derivatives are
u, = 3aa? + 6bxy + 3cy' ,
u, = 3bx* + 6cxy + My*.
Therefore (1627) the discriminant of u
is the annexed determinant, by (587).

axt-r3bj?y + 3cxy't + dy%.

3a
0
36
0

66
3a
6c
36

3c
66
3d
6c

Ite first

0
3c
0
3d

1632 The determinant is also an invariant of u, by (1638) ; that is, if u

be transformed into v by putting x = n+/3ij and y = a'^-f-^ij; and, if a
corresponding determinant be formed with the coefficients of v, the new
determinant will be equal to the original one multiplied by (a/3' a'(3)'.

QUANTICS.

289

1633 Again,
= 6ax + 6by, w2> = 6cx + Qdy, u,s = 6bx + 6cy.
Therefore, by (1630), the Hessian of u is
i.2 J = (ax + by)(ex+dy) (bx + ey)'
= (acb*) x,+ (adbc) xy + (bd<?) y*.
1634 And this is also a covariant; for, if u be transformed into v, as
before, then the result of transforming the Hessian by the same equations
will be found to be equal to te2, vlf. See (1652).
1635 If a quantic, u =f(x, y,z ...), involving n variables
can be expressed as a function of the second degree in
Xu X2 ... XM_!, where the latter are linear functions of the
variables, the discriminant vanishes.
Proof.Let
= <f>X\ + \pX,X, + x X,X, + &c.,
where
X, = a-p + -j- cxz + &o.
The derivatives u ut, &c. must contain one of the factors X,X, ... X.! in
every term, and therefore must have, for common roots, the roots of the
simultaneous equations X, = 0, X, = 0, ... X_! = 0; 1 equations being
required to determine the ratios of the n variables. Therefore the dis
criminant of u, which (1627) is the eliminaut of the equations ux = 0, uy = 0,
etc., vanishes, by (588).
1636 Cob. 1.If a binary quantic contains a square factor,
the discriminant vanishes ; and conversely.
Thus, in Example (1631), if u has a factor of the form (Ax+Byf, the
determinant there written vanishes.
1637 Cor. 2.If any quadric is resolvable into two factors,
the discriminant vanishes.
An independent proof is as follows :
Let u = XY be the quadric, where
X= (ax + by + cz+...),
Y = (a'z + b'y + c'z + ...).
The derivatives u,, u^, u, are each of the form pX+qY, and therefore have
for common roots the roots of the simultaneous equations X = 0, Y = 0.
Therefore the eliminant of uz = 0, = 0, &c. vanishes (1627).
1638

The discriminant of a binary quantic is an invariant.

Proof.A square factor remains a square factor after linear transforma

tion. Hence, by (163(5), if the discriminant vanishes, the discriminant of
the transformed equation vanishes, ami must therefore contain the former
discriminant as a factor (see 1628). Thus the determinant in (1631) is an
invariant of the quantic u.
The discriminant of the ternary quadric
1639

ty+c+2fyz+2gssx+ 2hxy
2 F

290

DIFFERENTIAL CALCULUS.

is the eliminant of the equations

1640

\ux = ax+hy+gz = Q^
\u hx+by+fz = 0 j
= gx+fy + CZ = QJ

1641

that is, the

determinant

a h g
h b f
8 f c

= abc+2fgh-af2-bg* ch? = A.

1642 The following notation will frequently be employed.

The determinant will be denoted by A, and its minors by
A, B, G, F, G, H.
Their values are readily found by differ
entiating A with respect to a, b, c,f, g, h ; thus,
A = be -f = Aa, B = ca-g* = A4,
C = ab-Jr = Ac,
F = gh-af=\A G = hf-bg = \\, H=fg-ch = ^Ah.
1643

The reciprocal determinant is equal to A2 or

A H G
H B F

G F C
1644

a h g
h b f

By (575).

g f c

The discriminant of the quaternary quadric

= arf+bif+ez* + dw2+2fyz+2gzx+2hxy
+ 2pxw + 2qyw + 2rzw

is the eliminant of the equations

1645

i = ax+by+gz+pw = 0
= hx-\-by-\-fz +qic = 0
\uz = gx+fy +cz +rtp = 0
hlw px+qy+rz+dw = 0

that is,
the
deter
minant

ah g p
h bfq
gfcr
p q r d

The determinant will be denoted by A', and, by decom

posing it by (568), we have
1646
1647
1648

A'= d.A-Ap2-Bq2-C^-2Fqr-2Grp-2Hpq.
A' = i (pA'p+qA'q+rA'r) + d.A.
Theorem.If <j> (sty) be a quantic of an even degree,

is an invariant of the quantic.

QUANTICS.

291

Proof.Let the linear substitutions (1628) be

z = a + bn,
y = a', + b'n
(i.)
Solve for 4 and ij. Find
>/*, t\ and substitute in the two equations
d, = dfi + d^n,;
= d(^ + d\n,The result is
d, = {ad, + 6 (-cZf )} - Jlf ; -4 = {aV, + (-d>)} - M
(ii.),
where M = ai' a'6, the modulus of transformation. Equations (i.) and (ii.)
are parallel, and show that the operations dy and dx can be transformed in
the same way as the quantities x and y ; that is, if <j> (x, y) becomes \p (> t)>
then f (df, d,) becomes yp (dt, > d( ) -3- If", where i is the degree of the
quantic ^. But <j> (d,, dx) <f> (x, y) is a function of the coefficients only of
the quantic ^, since the order of differentiation of each term, is the same as the
degree of the term ; therefore the function is an invariant, by definition (1628).
1649 Example.Let <p (r,, y) = axi+bx,y + exti'+fy*. The quantic must
first be completed; thus, ^ (xy) = ax^ + bx'y + cx^y1 + exy'+fy', (o = 0) ; then
<r(dr d*) f (*./) = (adty bd^. + cd^ ed^+fd,,) (j>(x,y)
= a.24/-6.6e + c.4c-e.6i+/.24a = 4 (12a/-36e + cs).
Therefore 12a/ 36e+c' is an invariant of (j>, and = {12AF3BE + C)+Mt,
where A, B, C, E, F are the coefficients of any equation obtained froui <p by a
linear transformation.
But if the degree of the quantic be odd, these results vanish identically.
1650 Similarly, if <j> (x, y), \p (x, y) are two quantics of the
same degree, the functions
<f>{dy,dx) + (x,y) and + (d,, dx) <f> (x,y)
are both invariants.
1651 Ex.If <t> = ax*+2bxy + cy* and i/ = a'x* + 2b'xy +c'y* ; then
(ad-ty 2bd, + crffa) (aV + 2b'xy + c'y') = ac' + ca' 2bb\ an invariant.
1652

A Hessian is a covariant of the original qur.ntic.

Proof.Let a ternary quantic u be transformed by the linear substitu

tions in (1604) ; so that u = <p (x, y, z) = ^ (, n, ). The Ht-ssiaus of the
two functions are

and ^ ( VVV~ (1630). Now

d (xyz)
d (ij; )
"d(Mf,c')"_ ^ d (f , f ) _ ^ <j(",w.tQ _ ^ rfOv-vO.
d(fij;)
d(*yz)
d(tii()
d(xyz)
The second transformation is seen at once from the form of the determi
nant by merely transposing rows and columns; the first and third are by
theorem (1604). Therefore, by definition (1629), the Hessian of It is a
covariant.
1653 Cogredients.Variables are cogredient when they are
subjected to the same linear transformation ; thus, x, y are

292

DIFFERENTIAL CALCULUS.

cogredient with x , y when

x=a+h\
y = cg+drj )

and

*'=f
y' = cg'+drj' J

1654 Emanents.If in any quantic it = <p (x, y), we change

x into x+px', and y into y-\-py', where x , y are cogredient
with x, y ; then, by (1512),
<f> (x+pv, y+py)
= u+p (x'dx+y'd,) u+ip* (x'dx+y'd,yu+&c.,
and the coefficients of p, p2, ps, ... are called the first, second,
third, ... emanents of u.

1655 The emanents, of the typical form (xd^+y'd^)" u, are

all covariants of the quantic u.
Proof.-If, in <j> (x, y), we first make the substitutions which lead to the
emanent, and afterwards make the cogredient substitutions, we change
x into x + px', and this into al + brj + p (a + brf).
And if the order of these operations be reversed, we change
x into al + bri, and this into o (+pf) + b (n + pn)The two results are identical, and it follows that, if <j> (x, y) be transformed
by the same operations in reversed order, the coefficients of the powers of p
in the two expansions will be equal, since p is indeterminate. Therefore, bj
the definition ( L629), each emanent is a covariant.
1656 For definitions of contragredients and crmtravariants, see (1813-4).
For other theorems on invariants, see (1794), and the Article on Invari
ants in Section XII.

IMPLICIT

FUNCTIONS.

IMPLICIT FUNCTIONS OF ONE INDEPENDENT VARIABLE.

If y and z be functions of x, the successive application of
formula (1420) gives, for the first, second, and third deriva
tives of the function <p (y, z) with the notation of (1405),
1700
1701

<kG/*) = <Mx+"Mx.
<M#*0 55 ^.vi+2&..y,*,+&.s;+&jfe+*.ai.-

IMPLICIT FUNCTIONS.

1702

293

fc, (yz) = ^jd+3^,^*,+3^y,4+^

+3 {tf>iyyx+4>yizx) y^+3

By making z = a; in the last three formulae, and conse

quently zx= 1, 2^= 0, or else by differentiating independently,
we obtain
1703

1704

(*?/) = ^,i+2^yx+^fa+^ytI.

1705

<f>*x(*y) = ^$+3&,*3+arW+&.-

1706 In these formulae the notation <px is used where the differentiation
is partial, while <f>, (x, y) is used to denote the complete derivative of <f> (x, y)
with respect to x. Each successive partial derivative of the function <p(y, z)
(1700) is itself treated as a function of y and z, and differentiated as such by
formula (1420).
Thus, the differentiation of the product <pty, in (1700) produces
The function ^ involves y and z by implication. If it should not in fact
contain z, for instance, then the partial derivative
vanishes. On the other
hand, y y^, &c. are independent of z ; and z zu, &c. are independent of y.
DERIVED EQUATIONS.
1707 If # (%y) = 0, its successive derivatives are also zero,
and the expansions (1703-5) are then called the first, second,
and third derived equations of the primitive equation <p(xy) = 0.
In this case, those equations give, by eliminating yx,
17ao
A/U0

d!t _4>*.
dl ~
'

ty*,$*4> h*& ~
da*'
#

1710 Similarly, by eliminating yx and y^, equation (1705)

would give y^ in terms of the partial derivatives of <p (xy).
See the note following (1732).
1711

If

1712

and

+()-0

and

lg=0;

$N

g = McM.
da?
#
Proof.By (1708), ^. = 0. Therefore (1704) and (1705) give these
values of yu and yu.

DIFFERENTIAL CALCULUS.

291

1713 If tyx aT1d <t>y both vanish, yx in (1708) is indeterminate.

In this case it has two values given by the second derived
equation (1704), which becomes a quadratic in yx.
1714 If fix, $xy> and 02y also vanish, proceed to the third
derived equation (1705), which now becomes a cubic in yx,
giving three values, and so on.
1715 Generally, when all the partial derivatives of <j> (x, y) of
orders less than n vanish for certain values x = a, y = b, we
have, by (1512), <j>(a, b) being zero,
<(> (a + h, b + k) = j (hdx+kdM)* $ (xy)iafi + terms of
higher orders which may be neglected in the limit, (x, y are
here put = a,b after differentiation.) Now, with the notation
of 1406),

%.o+% = 0;

therefore

*
_**-=^;
h
<pVib
dx
the values of which are therefore given by the equation
1716
1717

{hdx + kdyY $ (#, y),0>4 = 0.

If yx becomes indeterminate through x and y vanish

ing, observe that

^- in this case, and that the value of

the latter fraction may often be more readily determined by

algebraic methods.
If * and y in the function <f> (x, y) are connected by the
equation \p (,c, y) = 0, y is thereby made an implicit function
of x, and we have
.
1718
1719

4>x{*,y) =
<M*,y)= { (<M,-<M,) V +

V,

-2(^,-^x^)^}h-^.
Proof. (i.) Differentiate both <j> and \p for x, by (1703), and eliminate y,.
(ii.) Differentiate also, by (1704), and eliminate y, and y.
If u, y, z are functions of x, then, as in (1700),
1720

<rV (uyz) = <l>uux+<f>vyx+<l>,zx.

IMPLICIT FUNCTIONS.

1721

295

(uyz) = fautil+^yl+^zl

1722 To obtain $x(zyz) and ^(xyz), make tt = a: in the

above equations.
Let U = <f>(x, y, z, ) be a function of four variables con
nected by three equations u = 0, v 0, iv = 0, so that one of
the variables, , may be considered independent.
1723

We have, by differentiating for ,

<k^i+<k,yf+<k%+& Ut = (K
uxxt+u yt+uzzt+u(
= 0 / .

dU_ d(<f>uvic) 1 .
d
d(^yzj) J '

vx x(+vv y(+v, z(+v

w*^f

wtere ^ _ (twre)_
1
d(xyz)'

1 794.

dx _ _d (urnv) J_

= 0 C'
= <K

dy _ _ d (uvw) 1

dz _ _ d (uvw) 1

Observe that Z7f stands for the complete and ^ for the
partial derivative of the function Z7.
Proof. (i.) U is founil by taking the eliminant of the four equations,
separating the determinant into two terms by means of the element <f>^ Up
and employing the notation in (1600).
(ii.) Zj, yp and
are found by solving the last three of the same equa
tions, by (582).
IMPLICIT FUNCTIONS OF TWO INDEPENDENT VARIABLES.
1725 If the equation $ (x, y, z) = 0 alone be given, y may
be considered an implicit function of * and z. Since x and z
are independent, we may make z constant and differentiate
for x; thus, for a variation in x only, the equations (1703-5)
are produced again with $ (x,y, z) in the place of ^> (x, y).
1726 If x be made constant, z must replace x in those equa
tions as the independent variable.
Again, by differentiating the equation $ (xyz) = 0 first for
x, making z constant, and the result for z, making x constant,

DIFFERENTIAL CALCULUS.

296

we obtain
1727
fcz+fasyt+h^+hiyxyz+fayzz = 0.
From this and the values of yx and yz, by (1708),
1728

yxz =

^" ^vz - r

^ ^x ^:

"_
1729 If x, y, z in the function $(x, y, z) be connected by the
relation \f> (x, y, z) = 0, y may be taken as a function of two
independent variables x and z. We may therefore make z
constant, and the values of $x (,r, y, z) and fa (x, y, z) are
identical with those in (1718, '19) if x, y, z be substituted for
x, y in each function.
1730 By changing x into z the same formulae give the
values of
(x, y, z) and fa (x, y, z).
1731 On the same hypothesis, if the value of <px (x, y, z), in
forming which z has been made constant, be now differen
tiated for z while x is made constant, each partial derivative
fa
&c. in (1718) must be differentiated as containing x, y,
and z, of which three variables x is now constant and y is a
function of z.
The result is
1732

*(*,y,*)= {(*-^-**,)-(^-*^)*.*S

In a particular instance it is generally easier to apply such rules for

differentiating directly to the example proposed, than to deduce the result in
a functional form for the purpose of substituting in it the values of the partial
derivatives.
1734 Example.Let <p (x, y, z) = lx + my +nz and \p (x, y, z) = x*+ y* +z%
= 1, x and z being the independent variables ; if, (x, y, z) aud p., (x, y, z) are
required. Differentiating <f>, considering z constant,
. /
x
T . dy
i
a>
dy
tL,
x
4>m (%t 2/j z) f+ -f- = I m ; since -r~ = , =
;
dx
y
dx
if,
y
,
+ x1
<t>u(*,y,
-). = ~OT y yzi/r =m y'
z-yr-;
a result which is otherwise obtained from formula (1719) by substituting the
values
<p. = I,
<f>f = m, f, =n;
0te = flf = f = 0 ;
i>. = 2x,
= 2y, *. = 2;
^ = ^, = ^ = 2.
Again, to find f (x, y, z), differentiate for z, considering x constant in
the function
d<t> = i
, m
x ; thus -r^<P<b = +m
, xy, = mr,
xz since

dv = J/,
_
z.
J
=
ax
y
dxdz
y
rf
dj
\pk
y

IMPLICIT FUNCTIONS,

297

1735 Let U = <f> (x, y, z, , j) be a function of five variables

connected by three equations u = 0, v = 0, w = 0 ; so that ttoo
of the variables , r; may be considered independent. Making
i) constant, the equations in (1723), and the values obtained
for U(, x(, yt, 2f, hold good in the present case for the varia
tions due to a variation in , observing that <(>, u, v, w now
stand for functions of jj as well as of .
1736 The corresponding values of Z7,,
by changing into ?j.

yv, z, are obtained

IMPLICIT FUNCTIONS OF n INDEPENDENT VARIABLES.

1737 The same method is applicable to the general case of
a function of n variables connected by r equations u=0, v=0,
w = 0 ... &c.
The equations constitute any nr of the variables we
please, independent : let these be ,),.... The remaining r
variables will be dependent: let these be x,y,z...; and let
the function be U $ (x, y,z ... , ij, I ...).
For a variation in only, there will be the derivative of the
function TJ, and r derived equations as under.
1738

MHft+#.% + - + k = Ue .
u9xt+ury(+u,zt + ... + u( = 0,
vx -I't+Vy y(+v, st + ... + vt = 0,
&c,

involving the r implicit functions xt, yf, zf, &c.

of the r equations, as in (1724), gives
1MQ
17d9

da? _

d {uvw ...) 1
d(iyz...)~J''

1740

,
T
d(uvw...)
where J = ^ftj.

The solution

dy _
d(uviv ...) 1
.
d%--d{x&...)j"'&C"
dU
d(Suvw) l
Also ^ = 3^^.

The last value being found exactly as in (1723).

1741 With replaced by j we have in like manner the
values of
?/,, z,, 7, ; and similarly with each of the inde
pendent variables in turn.
1742 If there be n variables and but one equation
(*> 1J, 2 . . .) = 0, there will be n 1 independent and one depen
2q

DIFFERENTIAL CALCULUS.

208

dent variable. Let y be dependent. Then for a variation in

x only of the remaining variables, the equations (1703-5)
apply to the present case, <f> standing for <f> (x, y, z ...). If x
be replaced by each of the remaining independents in turn,
there will be, in all, n1 sets of derived equations.

CHANGE OF THE INDEPENDENT VARIABLE.

If y be any function of x, and if the independent variable
pc be changed to t, and if t be afterwards put equal to y, the
following formulae of substitution are obtained, in which p = yx:
1760

4 = J!l = !.

^Differentiating these fractions, we get

1762

*fy - yx*-**y* *

1766

- m
y

= ppI+pW

no; Ex.If x = r cos 0 and i/ = r sin 0 ; then

1768

_ r9sinfl+r cosfl.
(Lv ~~ re cos 0 r sin 0'

tfy/ _
r-+2r; rrit
dx1 ~ (r cos 6 r sin 0)3 '

''^^ROOF.Wiiting 0 for t in (1760) and (17C2), we have to find x y

Xu, Di> ; thus,
x, = r, cos 0 r sin 0 ; ss = ru cos 0 2r, sin 0 r cos 0 ;
yt = r, sin 0 + r cos 0 ;
= r sin 0 + 2r, cos 0 r siu 0.
Substituting these values, the above results are obtained.
ot'J To change the variable from x to t in (a+bx)n y, where
"(a -'&#) = e', employ the formula
1770

(*-jT=2) ...

fift:V?hich, multiplication by dt by the index law signifies the

repetition of the operation dt (1402).

CHANGE OF THE INDEPENDENT VARIABLE.

Proof.

299

d,{(a+bx)"y,tlt} {n {a + bx)"-'lbijnt+{a+bx)'y{H^),} xt.

Now bx, = e' = a + bx.

Substitute this, and denote (a + bx)"yx by UJ

nU.+ yU^, or P.,, = b (</,-n) UH.

o
-Therefore U = b ('i, n 1) ?7_j and 7n_[ = i (a*, n 2) Z7.2, &c,
and finally
U, = b (dtl) Ux.
Bat
If, = (a+te) y, = bz,y'M = by,.
Therefore
U = V' (cZ,-T^T) (dt-n2) ... (<i,-l) y,.
1771 Cor. (a+x)nyx and a;"?/^ are transformed by the
therefore

d, ( U.)

same formula by putting 6 = 1.

1772 Let V i F(x, y), where x, y are connected with , j
by the equations t/ = 0, ?; = 0. It is required to change tho
independent variables x, y to , tj in the functions Vx and Vy.
1773 Rule.To find the value of VxDifferentiate V, u, v,
each with respect to x, considering , rj functions of the inde
pendent variables x, j; and form the e.liminant of the resulting
equations; thus,
1774

vjx+vvVx-va

-vx
Of;
o)

.'.

= 0.
i

Similarly, to find V.
1775

Ex.Let x = r cos 0 and y = r sin 6 ; then

r
r
Proof. u = r cos 9 x, c=rsin9 y,
V,
V.
-V,
and the determinant in (1774) takes the form cos 9 r sin 0 1 = 0.
annexed by writing r and 9 instead of H, and if.
sin 9
rcos9
0
A similar determinant gives Vr
To find Vu, substitute Frcos9 y,s,n- in the pl.ee of Fin the value of
r
V, ; and in differentiating for r and 0, consider Vr and V, as functions of
both r and 9. Similarly, to find V-tv and Vn. Tlius,
sin' 9 , , sin' 9
1777 Fll=FJrcos39 + (l7,-Fr,) 2 sin 0 cos 9 t
Kr
r K2, ,.
r
1*
cos2 9 , ,, cos! 9
1778 F2, = F* sin' 8-(Ir,- F) 2Ei2?Lfl
r
r*
By addition these equations give
1779

F+ F% = Fir + -J- ^ + 4 F.

300

DIFFERENTIAL CALCULUS.

1780 Given V = f(x, y, z) and E, v, I known functions of

x, y, z; Vt, V^, Vs are expressed in terms of Vx, V,, Vt by
the formulae
, dV _d(r,V) ..... r
dV _d(VVC) . T dv_d(m)
u'
m-di^z)-udrj
d (xyz)
d(j
d (xyz)
Proof.Differentiate V as a function of ( V(Z + Vv7)* + V{(* = V
f it ! with respect to independent variables 1 y _^ y _j_ y ^ _ y
x, y, z. The annexed equations are the result. )
' '
"
Solve these by (582) with the notation of (1600). (. Vt , + V n, + V^, = V,.
1781 Given V =f (x, y, z), where x, y, z are involved with
, n, in three equations = 0, v = 0, w = 0, it is required
to change the variables to , ?j, in Vx, Vy, and Vx.
Applying Rule (1773) to the case of three variables, we have
v* vs -vx
%

w,i?x+ fL+ ux = o

f , {
V( f>, v;
W( ? W{

= 0.

The determinant gives Fx in terms of V(, T\, Ff and the deri

vatives of u, v, w. Vy and Vz are found in an analogous manner.
1782

Similarly with n equations between 2n variables.

1783 Ex.Given
x = r sin 0 cos

y r sin 5 sin gS ;

z = r cos

The equations u, v, w become *

r sin d cos ij> x = 0 ; r sin 0 sin 0 y = 0 ; rcosfl a = 0.
Writing r, 0, <]> instead of , q, Z, the determinant becomes
v,
v.
rt
-v.\
sin 0 cos ^ r cos 6 cos r sin 6 sin ^ 1
= 0. *
sin 0 sin ^ rcosOsin^
r sin 6 cos ^
0
cosfl
rsinS
0
0
From which V, is obtained. Similarly, V, and Fs ; and, by an exactly similar
process, the converse forms for Vr, V and Vr The results are
* In writing out a determinant like the above, it will be found expeditious
in practice to have the columns written on separate slips of paper in order to
be able to transpose them readily. Thus, to find the coeflacieut of V bring
the second column to the left side, and, since this changes the sign of the
determinant, transpose any two other columns, so that the coefficient of V,
may be read off in the standard form as the minor of the first element of the
determinant.

CHANGE OF TEE INDEPENDENT VARIABLE.

1784

r.

1785

r.

1786

v.

1787

v.
v,

1788
1789

301

it- a
. i -rr cos 0 cos <b
Vr sin 0 cos 0 + V,

T7- sin <j>

r sin 0
rr
a0 sin
<pj. +i V,
ir cos
cos 0
T r sin
- 0 sin rA +.
F, r-^r.
r
r sin 0
T7Vr COS 0fl rr
V, sin 0 .
r
F, sin 0 cos ^ + F, sin 0 sin ^ + F, cos 0.
V, r cos 0 cos0 + Vy r cos0 sin 0 7.r sin 0.
F, r sin 0 sin <j>+Vyr sin 0 cos 0.

1790 To find V, directly; solve the equations it, v, w, in (1783), for r, 0,

and f ; tlte solution in this ease being practicable ; thus,
r=
+ *+*),
0 = tan"1 ^/(a:'+'/'),
0 = tan"1 K
z
as
Find r 0 0, from these, and substitute in V, = V,rx+ V,0,+ F40,. Simi
larly, V, and V,. Also Fr = F,a;r+ F,i/r+ V,zr. Similarly, V, and F,.
1791 To obtain V^, substitute the value of V, in the place of V, in the
value of V,, in (1785), and, in differentiating V V Vt, consider each of
these quantities a function of r, 0, and <j>.

To change the variables to r, 0, and

in 7^+ V2y+ V^,
the equations (1783) still subsisting. Result
1792

V^+V^+Vi,
= Vir+ 1 Vr+ i ( Fe cot 0+ Vie+ Vy, cosec2 0).

Proof.Put r sin 0 = p, so that x = p cos 0 and y = p sin 0,

therefore, by (1779), 7^+ F* = Vu + Vf+ \v^
(i.).
P
P
Also, since z r cos 0 and p = r sin 0, we have, by the same formula,
V+ F = F+ i
(ii.).
? F,+ 1
r F,
Add together (i.) and (ii.), and eliminate V, by (1776), which gives
Vt = Vrmn0+V.^.
r
If V be a function of n variables x, y, z ... connected by
the single relation,
1793

A,z+y2+s;2+... = r*

V+ P*+ Vu+&c. = V2r+^Vr.

(i.).

DIFFERENTIAL CALCUI,U8.

302

Proof. F. = Vrrm=s Y, , since r, = , by differentiating (i.),

r
r
therefore
Fu = Vj + rrr-=^ = FJr^- + 7r
- -J).
Similarly

F* = F* "fr + F' (7 ~

) ' &0-

Thus, by addition,

LINEAR TRANSFORMATION.
1794 H y =/(*> lh z)>
(1781) take the forms
y aJ+btf+cA V ,
z = 3f+M+c3^

and if the equations , o, tc in

then

j Av = B1.r+B.ii/-\-BAz,
A = C^+Cy+C,*,

by (582), A being the determinant (o^iC,), and ^, the minor

of alt &c.
1795 The operations dx, dt, d will now be transformed by
the first set of equations below ; and d(, d^, di by the second
set.
d = (Atdt+B^+Ctdi) vA

,/ = M,+M,+M, (

Proof.By d = d^l. + d^. + d^, and c?^ = d,X+dyy( + d,Z and the
values of l Xp &c, from tbe preceding equations.

1797 From (1705), Vt = a^ + a^ + a^.

again upon Vt, we have

Operating

Vtt= (a^ + nJg + a^,.) Vt = al(Vt)M+aa(Vt)IL+at(Vd

and by substituting the value of Y(, and similarly with Fi,,
V2i, we obtain the formula?,
1798
V = b\ V^+b\ Vu+bl Ft,+26i6sF,.+26,61 V+2bA V [ .
V*=c> F+cJ F+f Vu+2ctcs Vst+2ciCl F+2Clct F<>

CHANGE OF THE INDEPENDENT TABIABLE.

303

ORTHOGONAL TRANSFORMATION.
1799

If the transformation is orthogonal (584), we have

and since, by (582, 584), A = l, Al = a1, &c;

(1794) now become
1800

x-

y =
= 3
And equations (1795) become
1802

equations

dx = aydt+byd,+Cyds
i i- ci oA
id^c.di V,
dy = a.,d+b.zt

3-'
d( = axdx + a2dy 4- a,dz -\
rf, = bydx+b2dy+bzdz V.

dt = a3rff+63</,+tsu,fThe double relations between x, y, z and , ?, I, in the six

equations of (1800-1), and the similar relations in (1802-3)
between dxdydz and d(d,,d(, are indicated by a single diagram
in each case ; thus,
1804

a?
z

dt d d(

a. by Cy
a2 bt ca

dx

a. 6. cs

d.

Uy by Cy
2 ^2 Cj
^3 ^3 *"8

1806 Hence, when the transformation is orthogonal, the

quantities x, y, z are cogredient with dx, dyt dMi by the defini
tion (1653).
1807 Extending the definition in (1629), it follows that any
function u = <p (x, y, a), when orthogonally transformed, has,
for a covariant, the function $ (dx, dy, dz) u. That is, if by
the transformation,
u = <f> {x, y,z) = \]i (, rj, ),
then also

<f> (dx, dy, d,) u = / (r/f, dv, d{) u.

1808 But if a be a quantic, then, as shown in (1648),

<j> {dx, dv, dz) u is always a function of the coefficients only of u,
and the covariant is, in this case, an invariant.
1809 Ex.Let u or f (as, y, z) = 03' + byt+czi+2fyz + 2gzx + 2hxy, '
.: f((ldd,) u = au,I + bu.iv + ctiu + 2futx + 2gu+2kurt,
= 2 {a1 + V + c' + 2f + 2<73 + 2A}, and this is an invariant of u.

304

1810

DIFFERENTIAL CALCULUS.

When V=f (x, y, z) is orthogonally transformed,

Proof.By adding together equations (1798), and by the relations

a] + b] + c] 1, <fec, and
a, + 6, 6, + c,c, = 0, &c.,
established in (584).
1811 If two functions u, v be subjected to the same ortho
gonal transformation, so that
u = <f>(x,y,z) = *(i7,C) and v = + (x,y,z) = *(^,0 i
then
1812
nndlet

<j> (dx, dv, d.) v = 4> (dt, d d{) v.

Ex.Let u = ax, + by' +c** + 2fyz + 2gzx +2hxy = <t>
= aV + 6V + e C + 2/V + 2g'ii + 2Kb, = *,
v = x1+yt+tt = F + if+f^ * and

Then
(c? cZ,, <Z.) = avu + bv2, + cva + 2fv + 2gv:x + 2/!^ ,
and
* (dp
d{) v = a'v^ + b'v2n + c'v.^ + 2fvni+ 2g'v(( + 2h'v^.
But Vt, = 2, and r,, = 0, Ac. Hence the theorem gives a+b+c
= a' + b'+c; in other words, a + b+c is an invariant.
1813 Contragredient. "When the transformation is not
orthogonal, (1795) shows that da is not transformed by the
same, but by a reciprocal substitution, in which au blt cy are
replaced by the corresponding minors Alt Blt C^. In this case
dx, dy, d, are said to be contragredient to x, y, z.
1814 Contravariant.If, in (1629), the quantics are sub
jected to a reciprocal transformation instead of the same, we
obtain the definition of a contravariant.
1815 When z is a function of two independent variables x
and y, the following notation is often used :
te __
dz
dp _dtz _
da-1'
dy~q'
d.v~d? = r'
(Jp _ (fy _ dz _ g
dy ~ dx
dx dy
'

liq^ _ cfs __ f
dy ~~ dy* ~

Let <j> {x, y, z) = 0.

It is required to change the inde
pendent variables from x, y to z, y. The formula? of trans

MAXIMA AND MINIMA.

305

formation are
1816

= _1_ .

dz

1819

d2x
^4
=

dx __ __ q
p'
dy "
p!
'Zspqtp^rq*.
P*

d*x
t
dz1
P
d\v _ qrps
dydz
p3

'

Peoof.Formulae (1761, 1763) give xz and asto, because, since y remains

constant, <p may be considered a function of only two variables, x and z.
Formulae (1708-9) give xt and x,v in terms of partial derivatives of <p,
since z is now constant, and <f> may be taken as a function of the two variables
x and y.
But <f> (x, y, z) = 0 is equivalent to \j/ (x, y)z = 0 ; and the partial
derivatives of p with respect to x and y are the same as those of yp ; and
therefore the same as those of z when x and y are the independent variables.
Hence z may be written for <j> in the formulae.
dydz
\ pit
\ p I x dz
p*
p
p
The independent variable is here changed from z to x, without reference
to the equation y = 0 ; and this is allowable, because y is constant for the
time being in either case.

MAXIMA AND MINIMA.

Maxima and Minima values of a function of one
independent variable.
1830 Definition.A function <p (x) has a maximum value
when some value x = a makes $ (x) greater than it is made
by any value of x indefinitely near to a. Similarly for a
minimum value, reading less for greater.
1831

Illustration. If the

ordinate y in the figure be al

ways drawn = /()> it has
maximum values at A, 0, E,
and minimum at B and D
(1403).
Note. For the algebraic
determination of maxima and
minima values by a quadratic
equation, see (58).

1832 Rule I.A function $ (x) is a maximum or minimum

when <j>' (x) vanishes, and changes its sign as x increases from
plus to minus or from minus to plus respectively.
2 E

306

DIFFERENTIAL CALCULUS.

1833 Rule II.Otherwise <j> (x) is a maximum or minimum

when an odd number of consecutive derivatives of <j> (x) vanish,
and- the next is minus or plus respectively.
Proof. (i.) The tangent to the curve in the last figure becomes parallel
to the x axis at the points A, B, U, D, E as x increases ; therefore, by (1-103),
tanfl, which is equal to/"(.r), vanishes at those points, while its sign changes
in the manner described.
(ii.) Let /*(x) be the first derivative of fix) which does not vanish when
h" f*(a + t)h).
x=sa, n being even; therefore, by (1500), /(oA) = /(<*)+ t
The last term retains the sign of /" (a), when h is small enough, whether h
be positive or negative, since n is even. Therefore / (*) diminishes for any
small variation of x from the value a if f"(a) be negative, but increases if
/" (a) be positive. Hence the rule.
1834 Note. Before applying the rule for discovering a
maximum or minimum, we may evidently
(i.) reject any constant factor of the function
(ii.) raise it to any constant power, paying attention to sign ;
(iii.) take its reciprocal ; maximum becoming minimum, and
vice versa ;
(iv.) take the logarithm of a positive function.
1835 Ex. 1.Let
f(x) = x7-7x4-35x+l,
therefore
f' (x) = 7x'-28xs-35 = 7 (x'-5) (x*+l).
Also f"(n>) = 7 (tix5 l'ix2). Therefore x \/5 makes <f (x) vanish, and
9" (x) positive ; and therefore makes <j>(x) a minimum.
1836 Ex. 2.Let
<j>(x) = fx-3)"(x-2)". Here
f (a>) = 14(ji-3)is(x-2)u + ll(x-3),<(x-2)10= (x-3),'(x-2),0(2.rx-61),
and we know, by (444) or by (1460), that, when x = 3, the first thirteen
derivatives of <j> (x) vanish; and 13 is an odd number. Therefore 9 (x) is
either a maximum or minimum when x = 3.
To determine which, examine the change of sign in 9(x). Now (,r 3)"
changes from negative to positive as x increases from a value a little less
than 3 to a value a little greater, while the other factors of p (x) remain
positive. Therefore, by the rule, <f> (x) is a minimum when x = 3.
. Again, as x passes through the value 2, 9' (.'-') does not change sign, 10
being even. Therefore x = 2 gives no maximum or minimum value of 9 (x).
Lastly, as x passes through the value ~ the signs of the three factors
in <p'(x) change from ( ) ( + ) ( ) to ( ) ( + ) ( + ); thatis,9'(x) changes
from + to ; and, consequently, 9 (,x) is a maximum.
1837
of y.

Ex. 3.Let 9 O, y) = x*+2x'y-y' = 0.

To find limiting values

MAXIMA AND MINIMA.

307

Here y is given only as an implicit function of x. Differentiating, in

order to employ formulae (1708, 1711),
f, = 4x*+4scy, <l>u = 12x*+4y, <f>v = 2z'-3y' ;
y, = 0 makes <j>, = 0. Solving this equation with <p (x, y) = 0, we get
z = l, y = 1 when yr vanishes.
And then
= = ^- = 8, positive ; therefore, when x = 1,
y has 1 for a minimum value.
Similai-ly, by making ?/ the independent variable, it may be shewn that,
8 x has both the maximum
...
when y = ,
and minimum values 4- */6.
1838

A limiting value of

subject to tlie condition

is obtained from tbe equation

(,r, y),
\fi
y) = 0
fyx^a = fyytyx

(i.),
(u>)

Simultaneous values of x and y, found by solving equa

tions (i.) and (ii.)> correspond to a maximum or minimum
value of <j>.
Proof.By (1718), <f> being virtually a function of x only ; and, by (1832),
f.(*y) = Q1839 Ex.Let <f> (x, y) = xy and \f> (x, y) - 2x'-xy + y" = 0
(i.)
Equation (ii.) becomes y ('iy1 x) =x(6z* y).
Solving this with (i.), we find y* = 2.e* and a' (4a; 1/2).
Therefore x=\\/2, y =
are values corresponding to o, maximum
value of <p. That it is a maximum, and not a mini mum, is seen by inspecting
equation (i.)
1840 Most geometrical problems can be treated in this way, and the
alternative of maximum or minimum decided by the nature of the case.
Otherwise the sign of <j>u{xij) may be examined by formula (1719) for the
criterion, according to the rule.
Maxima and Minima values of a function of two
independent variables.
1841 Rcle I.A function <f> (x, y) is a maximum or minimum
when <(>x and <py both vanish and change their signs from plus to
minus or from minus to plus respectively, as x and y increase.
1842 RuJJB II. Otherwise, <px and <j>y must vanish ; <j>.,x<p2y<t>ly
must be positive, and <j>.u or <f>iy must be negative for a maximum
and positive for a minimum value of
Proof.By (1512), writing A, B, 0 for f.^, (pIy, <f>2l/, we have, for small,
changes h, k in the values of x and y,
t(x + h, y + k)-<j> (x, y) =
+ (Ah* + 2Bhk+ Ck*) + terms
which may be neglected, by (1410).

308

DIFFERENTIAL CALCULUS.

Hence, as in the proof of (1833), in order that changing the sign of h or

h shall not have the effect of changing the sign of the right side of the equa
tion, the first powers must disappear, therefore px and <t>y must vanish. The
next term may be written, by completing the square, in the form
ltd { ( ^X"'"^)
^* | ' an(^' * ensnre this quantity retaining its
sign for all values of the ratio h : k; AC IP must be positive. <f> will then
be a maximum or minimum according as A in the denominator is negative
or positive.
It is clear that A and B might have been transposed in the proof. Hence
B must have the same sign as A.
1843 A limiting value of
<j> (x, y, z),
subject to the condition
i/ (r, y, z) = 0
is obtained from the two equations
1844
1846

(ii.),

or, as they may be written,

(i.),

<f>A = 4>A

(ni.) ;

^ = ^ = ^

(iv.)

Simultaneous values of x, y, z, found by solving equations

(i., ii., iii.), correspond to a maximum or minimum value of <p.
Proof.By (1841), <(> being considered a function of two independent
variables * and z, and, by (1729, 1730),
^C*> V, *) = 0 gives (ii.), and 0:(ar, y, z) = 0 gives (iii.)
The criterion of maximum or minimum in (1842) may also be applied
without eliminating y by employing the values of <pu and <p1: in (1719, '30).
1847 Ex.Let
<p(x,y,z) = x* + tf[+z*
and xf, {x,y,z) = axt+by>+cz*+2fyz+2gzx+2hxyl = 0..(i.)
Equations (ii.) and (iii.) here become
x
y
z
1
v
:
= -~
r =
;
= ~T, , Say, (IV.)
ax + hy+gz
kc + by+Jz
<jx-\-fy + cz
B
Therefore, by proportion (70) and by (i.), xt+yi-\-zt = B~x =
From equations (iv.) we have
1848

1849

ax + hy+gz = Bx )
hx+by+fz = By i
gz+fy +cz = Bz )

a B
h
h
b-B
9

g
f

= 0,

c-B

or
(B-a)(B-b)(B-c) + 2fgh
-(B-ajf-iB-fygt-iB-cjh^O, or (see 1641)

il'-il8 (a+b+c)+R (bc+ca+ab-ft-gt-h^-A = 0,

MAXIMA AND MINIMA.

309

This cubic in B is the eliminant of the three equations in

x,y,z. It is called the discriminating cubic of the quadric (i.),
and its roots are the reciprocals of the maxima and minima
values of a^ + y2+zs.
1850 To show that the roots of the discriminating cubic
are all real.
Let Rv Ri be the roots of the quadratic equation
Il-b
f
(R-b)(R-c)-f = 0
(v.)
1851
f
R-c
K, > b and c, and b and e > i?,.
Make It = 22, in the cubic, and the result is negative, being minus a
Rqnare quantity, by (v.)- Make R = R2, and the result is positive. There
fore the cubic has real roots between each pair of the consecutive values + oo ,
Rl, 7?j. oo ; that is, three real roots. But since the roots are in order of
magnitude, the first must be a maximum value of R, the third a minimum,
and the intermediate root neither a maximum nor a minimum.
Maxima and Minima values of a function of three
or more variables.
1852 Let <f> {xyz) be a function of three variables.
Let
to*. <r>z> ty*> <t>zx, <b*y be denoted by a, b, c,f, g, h; and let
A, B, C, F, G, H be the corresponding minors of the deter
minant A, as in (1642).
1853 Rule I.^ (x, y, z) is a maximum or minimum when
<f>s, <pz all vanish and change their signs from plus to minus
or from minus to plus respectively, as x, y, and z increase.
Otherwise
1854 Rule II. The first derivatives of $ must vanish; A
and its coefficient in the reciprocal determinant of A must be
positive ; and <p will be a maximum or minimum according as
a t negative or positive. Or, in the place of A and a, read B
and b or C and c.
Proof.Pursuing the method of (1842), let , >/. be small changes in
the values of 2, y, z. By (1514),
9 0+, y+i, *+09 (*> y- z)
= % + t,?, + Cf. + i (oi1 + W + c? + 2fn( + 2gtt + 2htn) + 4c.
For constancy of sign on the right, <j>T, f <p, must vanish. The quadric
may then be re-arranged as under by first completing the square of the terms
in , and then collecting the terms in (, /, and completing the square. It
thus becomes JL J (a*+foj + <tf)*+ (C'~F^ +
?}.

310

DIFFERENTIAL CALCULUS.

Hence, for constancy of sign for all values of , j, it is necessary that G

and BC F* should be positive. This makes B also positive. Bv symmetry,
it is evident that A, B, C, BC-F*, CA-CP, AB-H'- will all 'be positive.
The sign of a in the first factor then determines, as in (1842), whether <p is a
maximum or a minimum.
1855 The condition may be put otherwise.
Since
BCF* = aA by (577), the condition that BC*F* must be
positive is equivalent to the condition that a and A must have
the same sign. Hence we have also the following rule :
1856 Rule III.tj>x,
$T must vanish ; the second of the four
determinants below must be positive, and the first and third
must have the same sign: that sign being negative for a max
imum and positive for a minimum value of <p (x, y, z).
1857

4>t* fat,
fax fan

fa, fat fa:

fax fay

<r>2x fay faz <ro

fax fay faz fa*

fax fay fa.

fax fay <f>iz faw

fax <f>wy faz fa*

1858 The theorem can be extended in a similar manner to

<f> (x,y,z,w, ...), a function of any number of variables. Form
the successive Hessians of <p (1630) for one, two, three, &c.
variables in order as shown above ; then
1859 Rule.In order that <p (x, y, z, w, ...) may be a max
imum or minimum, ix,$j,$i,fmt &c. must vanish; the Hessians of
an even order must be positive ; and those of an odd order must
have the same sign, that sign being negativefor a maximum and
positive for a minimum value of the function <p.
For a demonstration in full, see Williamson's Biff. Calc, 4th Edit., p. 433.
1860

Ex.Required a limiting value of the function

u = ax1 + bif + c;5 + 2/yz + 2gzx + 2hxy + 2px + 2qy + 2rz 4- d.
The condition in the rule produces equations (1), (2), (3). Equation {V)
results from Euler's theorem (1621), thus; introducing a fourth variable w,
as in (1640), we have
xu,+ y",+ zu, + wuw = 2u,
which reduces to (4) by means of (1), (2), (3), and the value of , putting
w = 1.
a h g p
1861 \u, = ax + hy + gz+p = 0. ..(1^
Ju, = hx + by +fz +q = 0 . ..(2)
h bf q
9f c r
= gx+fy +cz +r = 0. ..(3)
p q r du
px + qy +- rz + d = u. (4) J

MAXIMA AND MTNIMA.

311

The determinant is the eliminant of the four eqnations, by (583), and is

equivalent, by the method of (1724, Proof, i.), to A' Aw = 0, or u = A' -4- A
(Notation of 1646).
To determine whether this value of it is a maximum or minimum, either
of the conditions in (1854, '6) may be applied ; and since, in this example,
1*2, = 'la, u29 = 26, &c, the letters a, 6, c, /, g, h may be considered identical
with those in the rule.
1862 To determine a limiting value of <j> (x, y,z, ...), a func
tion of m variables connected by n equations 2^ = 0, u2 0, ...
= 0.
Rule. Assume n undetermined multipliers Xlt X4, ... X
with the following m equations:
*x + \ (ui)x + x2 fa), + . . . + xn (un)x = o,
<j>y+\ (Uj)y + X4 (u2\ + . . . + X (un)y = 0,
making in all m-f-n equations in m-f-n quantities, x,j,z, ...
and \u X2, ... Xn. The values of x,j,z,
found from these
equations, correspond to a maximum or minimum value of <p.
Proof.Differentiate <j> and ,, ,, . .. on the hypothesis that x,y,z, ...
are arbitrary functions of an independent variable t. Multiply the resulting
eqnations, excepting the first, by A A ... A in order, and add them to the value
of f,. The coefficients of x y z ... may now be equated to zero, since the
functions of t are arbitrary, producing the equations in the rule.
1863 Ex. 1.To find the limiting values of r1 = x, + y, + zi with the
conditions
Ax* + By* + Ci* = \ and Ix + my + nz = 0.
Here m = 3, n = 2 ; and, choosing X and /i for the multipliers, the equa
tions in the rule become
2x + 2A\x + /il = 0
(l)-j
Multiply (1), (2), (3) respectively by
2y + 2B\y + fim = 0
(2) > .
x, y, z, and add; thus fi disappears,
2z + 2C\z+ri7i = 0
(3))
and we obtain
x*+y*+ii*+(Ai*+By*+Cz*) \ = 0,
therefore \ = r*.
Substitute this in (1), (2), (3) ; solve for x, y, z, and substitute their values
in
lx+my+nz = 0.
1864

The result is

f - + -+
- = 0, a quadratic in r*.
Ar\ Br1 Crl
The roots are the maximum and minimum values of the square of the
radius vector of a central section of the quadric Ax* + By* + Cz* = 1 made by
the plane Ix+my + nz = 0.
1865 Ex. 2.To find the maximum value of u = (a; + l) (2/ + 1) (z + 1),
abject to the condition
N = a'6'c".

312

DIFFERENTIAL CALCULUS.

This is equivalent to finding a maximum valua of

log (z + 1) +log (y + 1) +log (* + 1),
subject to the condition log .AT = x loga+ y log b + z logc.
The equations in the rule become
-J-+Xloga = 0;
_L.+Xlog& = 0;
J_+\lOgC = 0.
x+1
y+l
z+1
By eliminating X, these are seen to be equivalent to equations (1846).
Multiplying up and adding the pqnations, we find X, and thence x + \, y + l,
e + 1 ; the values of which, substituted in u, give, for its maximum value,
u = {log (Nabc) }'-=- 3 log (is log b> log c\
Compare (374), where a, 6, c and x, y, z are integers.

1866

Continuous Maxima and Minima.

If <px and <pt, in (1842), have a common factor, so that

*, = P+(xty),
+,= W(X,y)
(i.),
where P and Q may also be functions of x and y ; then the
equation \p (x, y) = 0 determines a continuous series of values
of x and y. For all these values $ is constant, but, at the same
time, it may be a maximum or a minimum ivith respect to any
other continuous values of <f>, obtained by taking x and y so
that
{xy) shall not vanish.
1867 In this case, $tefiy<&i vanishes with \p, so that the
criterion in Rule II. is not applicable.
Pkoof.Differentiating equation (i.), we have
= Q.++ Q4>, >'
fv* = QA+ Q-h > '
If from these values we form <p^^!t <pIy X ^yx, \(/ will appear as a factor of
the expression.
1868

Ex.Take z as <j> (xy) in the equation

z2 = a'-b1 + 2b^(x'+,/)-xi-y*

(i),

zz,-=x[ - r l)
and
=y(
f
lV
\vV + 2/3)
I
" ^^(x' + y')
)
The common factor equated to zero gives x* + y1 = V, and therefore z = o. . . (ii.)
Here a is a continuous maximum value of z, and a a continuous minimum.
Equation (i.) represents, in Coordinate Geometry, the surface of an anchor
ring, the generating circle of radius a having its centre at a distance b from
the axis of revolution Z. Equations (ii.) give the loci of the highest and
lowest points of the snrface.
For the application of the Differential Calculus to the
Theory of Curves, eee the Sections on Coordinate Geometry.

INTEGRAL

CALCULUS.

INTRODUCTION.

1900 The operations of differentiation and integration are

the converse of each other. Let /(*) be the derivative of
$(x); then <j>{x) is called the integral oif(x) with respect to x.
These converse relations are expressed in the notations of the
Differential and Integral Calculus, by
*fctei=/(*) and by

1901 Theorem.Let the curve y=f(x) be drawn as in

(1403), and any ordinates LI, Mm, and let OL = a, OM=b;
then the area LMml = <j>(b)<l>(a).
Proof.Let ON be any value of x, and PN
the corresponding value of y, and let the area
ONPQ = A ; then A is some function of x. Also,
if NN'= dx, the elemental area NN'FP = dA Q
= ydx in the limit ; therefore dA
= y. Thus A
dx
is that function of x -whose derivative for each
value of x is y or / (x) ; therefore A = f (x) + G,
where C is any constant. Consequently the area
'
LMml = <{> (6) 9 (a), whatever C may be.
The demonstration assumes that there is only one function (f> (x) corres
ponding to a given derivative f(x). This may be formally proved.
If possible, let \p (x) have the same derivative as <p (x) ; then, with the
same coordinate axes, two curves may be drawn so that the areas defined as
above, like LMml, shall be <(>(x) and ^{x) respectively, each area vanishing
with x. If these curves do not coincide, then, lor a given value of x, they
have different ordinates, that is, f (as) and \J/' (x) are different, contrary to the
hypothesis. The curves must therefore coincide, that is, <j>(x) and ifi(x) are
identical.
2 s

314

INTEGRAL CALCULUS.

1902 Since <{>(b) f(a) is the sum of all the elemental areas
like NN'P'P included between LI and Mm, that is, the sum of
the elements ydx or f(x) dx taken for all values of x between
a and b, this result is written

r
1903 The expression on the left is termed a definite integral
because the limits a, b of the integration are assigned.*
When the limits are not assigned, the integral is called inde
finite.
1904

By taking the constant (7=0 in (1901), we have the

area

ONPQ = <f> (x) = (f(x) dx.

Note.In practice, the constant should always be added to

the result of an integration when no limits are assigned.
MULTIPLE INTEGRALS.
1905

Let f(x, y, z) be a function of three variables ; then

/V Pyt zi

the notation

f(x, y, ) dx dy dz

*/Xi *Jyi zi

is used to denote the following operations.

Integrate the function for z between the limits z=zx, z=zs,
considering the remaining variables x and y constant. Then,
whether the limits Zj, z2 are constants or functions of x and y,
the result will be a function of x and y only. Next, consider
ing x constant, integrate this function for y between the limits
yx and y2, which may either be constants or functions of x.
The result will now be a function of x only. Lastly, integrate
this function for x between the limits a^ and xt.'
Similarly for a function of any number of variables.
1906 The clearest view of the nature of a multiple integral
is afforded by the geometrical interpretation of a triple in
tegral.
* The integral may be read " Sum a to b, J (x) dx"; J signifying "mm."

315

INTRODUCTION.

Taking rectangular coordinate axes, let the surface

2 = <p (x, y) (XYom in the figure) be drawn, intersected by the
cylindrical surface y.\f,(x) (RMNnm), and by the plane
it = zl (LSml). The volume of the solid OLMNolmn bounded
by these surfaces and the coordinate planes will be
1 I
Jo Jo

<j> {x, y) doc dy = 1 1

'
'
Jo Jo

I
Jo

doc dy dz.

Peoof.Since the volume cut off by any plane parallel to OYZ, and at a
distance x from it, varies continuously with x, it must be some function of x.
Let Fbe this volume, and let dVbe the small change in its value due to a
change tlx in x. Then, in the limit, dV PQqp x dx, an element of the solid
shown by dotted lines in the figure. Therefore
d = PQqp =
V) dy, by (1902),
x being constant throughout the integration for y. The result will be a
function of x only. Making x then vary from 0 to x, we have, for the whole
f*i r f+W
)
fx, ( f+M r ((*.) -i
")
jj
<j> (a, y) dy j dx = j jj
[J
rfz J tty j <fo,
since ^ (z, y) = z. With the notation explained in (1905), the brackets are
not required, and the integrals are written as above.
1907 If the solid is bounded by two surfaces zl = 0, (x, y), z, = <pt (x, y),
two cylindrical surfaces ?/, = >h (*), 2/j = (x), and two planes x = xx, x = ar
the volume will then be arrived at by taking the difference of two similar
integrals at each integration, and will be expressed by the integral in (1905).
If any limit is a constant, the corresponding boundary of the solid
becomes a plane.

INTEGRAL CALCULUS.

316

METHODS

OF

INTEGRATION.

INTEGRATION BY SUBSTITUTION.
1908

The formula is

j <f> (x) dx = J <f> (#) ^ dz,

where z is equal to f(x), some function chosen so as to facili

tate the integration.
Rule I.Put x in terms of z in the given function, and
multiply the function also by x,; then integrate for z.
If the limits of the proposed integral are given by x = a,
x = b, these must be converted into limits of z by the equation
x = f-1(z).
The following rule presents another view of the method of
substitution, and is useful in practice.
1909
then

Rule II.If f(x) can be expressed in the form F(z)zx;

J <f> (x) dx = f F (z) zxdx = J F (z) dz.

jT-jtt
therefore

Here
Ex 3

Substitute z = x + ^(x* + at) ;

g- = 1 + , f , = *+ y/f* + ') =
dx
' ' ^/{x> + at)
^/{xt+at)

= a-4a>-*,
f l~ca!'

JF"(*) = -,
^

+ a'/

. = - (525-'+ 23-*).

= f x~* c
cx y(cV + x"' + a)

1
, gy/(tt-2c) + y/Q+ag' + cV)
vYa-2c)
l+cx*
V>-2c) g6
1+
_ixy/(2ea)
_
1
or
7 . cob
-.
./(2c -a) '
</(2e
1 + cx'
By (1927) or (1926). Here * = -' + .

METHODS OF INTEGRATION.

317

In Examples (2) and (3) t.he process is analytical, and leads to the dis
covery of the particular function z, with respect to which the integration is
effected. If z be known, Rule I. supplies the direct, though not always the
simplest, method of integrating the function.
INTEGRATION BY PARTS.
1910

By differentiating uv with respect to x, we obtain the

general formula

^uxvdx = uv ^uvxdx.

The value of the first integral is thus determined if that

of the second is known.
Rule. Separate the quantity to be integrated into two
factors. Integrate one factor, and differentiate the other with
respect to x. If the integral of the resulting quantity is
known, or more readily ascertained than that of the original
one, the method by " Parts" is applicable.
1911 Note. In subsequent examples, where integration by Parts is
directed, the factor which is to be integrated will be indicated. Thus, in
example (1951), "By Parts jea'dx" signifies that e" is to be integrated and
sin bx differentiated afterwards in applying the foregoing rule. The factor 1 is
more frequently integrated than any other, and this step will be denoted by J dx.

1912

INTEGRATION BY DIVISION.
A formula is

I (a+bxn)p dx=a f(a+bxn)p-x+b fj?" (c+^r")""1 dx :

V
*/
The expression to be integrated is thus divided into two terms,
the index p in each being diminished by unity, a step which
often facilitates integration.
Similarly,
J (a + bx" + cxm)p dx
= a (a+bx* + cxmy-x + bxn (a + bx" + cx",)"-i + cxm (a+bxn+cx"')'"1.
1913

Ex.To integrate

j" S(x' + a') dx.

By Parts J dx,

+ a') dx = x J(x* + a') - j"

By Division,

f v/ (* + **) dx = f

+ \

Therefore, by addition,

= JaV(, + a,) + Klg{*+ A*' + ')}, by (1909, Ex. 1).

INTEGRAL CALCULUS.

318

INTEGRATION BY RATIONALIZATION.
1914 In the following example, 6 is the least common de
nominator of the fractional indices. Hence, by substituting
z = a:*, and therefore xz = 62s, we have

= 6 j'^z6 + z4-zs+z-z + l-^) dx.

Each term of the result is directly integrable by (1922) and
(1923). For other examples see (2110).

INTEGRATION BY PARTIAL FRACTIONS.

1915 Rational fractions can always be integrated by first
resolving them into partial fractions.
The theory of such
resolutions will now be given.
1916 K $(%) and F(x) are rational algebraic functions of x,
4j> (x) being of lowest dimensions, and if F (x) contains the
factor (x a) once, so that

1917

* = (<*-)*(<*)

(i);

then $M = -A- +
and A = fei
F(x)
xa
$(x)
F (a)

(2).

Proof.Multiply equation (2) by (1), thus

<p (as) = A4> (x) + (x-a) X (*).
Therefore, putting x = a, </> (a) = A<i* (a). Also, by differentiating (1), and
putting x a afterwards, F' (a) = ^ (a). Therefore A = f (a) -r- F' (a).

1918

Again, if F(x) contains the factor (xa), n times, so

that
Assume

F (x) = (xa)* $ (x).

4 = 7 A\ + , A\ 1 + ... + -ik+XM
F Or)
(.r a)" T (x-a)n~l T T a?a T i/> (j?)

To determine

... ^4.

Multiply by (x a)n; jwtf

x=a and differentiate, alternately.

1919

If jF(a:) = 0 has a single pair of imaginary roots

STANDARD INTEGRALS.

319

a + i/3; then, applying (1917), let

-A-iB-

ted -A\iB-

and the partial fractions corresponding to these roots will be

iB
x a i/3

A+iB _ 2A (x-a)+2BP
I _ i ;o
/ .it i
x a-\-i/3
(xa)2-\-ft

For practical methods of resolving a fraction into partial fractions in the

different cases which occur, see (235-238).

INTEGRATION BY INFINITE SERIES.

When other methods are not applicable, an integral may
sometimes be evaluated by expanding the function in a con
verging series and integrating the separate terms.
E,

JC*.^.+.+ W+5^ + _|V5 + toi

(150)

STANDARD INTEGRALS.

1921 Some elementary integrals are obtained at once from

the known derivatives of simple functions. Thus the deri
vatives (1422-38) furnish corresponding integrals. The fol
lowing are in constant use :
1922

(V<r = ^l.
J
m+1

f^ = log*.
J x

1924

fa* da? = r-^L.

J
log a

f e* = e*.
J

1926

f
* = -cos-1-a-or-lsin-1^.
J xv (or or)
a
x
a
x

1927

$i7$?r^l0g+A-sY

[Subs.1

tSob"itate

320

INTEGRAL CALCULUS.

dx

= log{a- + v/(^s)}.

(1909, Ex. 1)

1928
1929

C
dx
gin 1 or
J Vid'-x1)
a
C
xdx

cos 1 .
a

(1434.)

1930
1931

By Parts, Division, and adding results (1913), we obtain

f v/(j72 a!) dx = x v^V a2) *a2 log {x + ^(x* a2) } .

/
^
1932

By Parts, Division, and difference of results,

I vV <0 = W(*0 T is log {*+ vV*)}

1933

f vV-*) <r = a2 sin"1 . + |* ^(a2-^2).

J
a
[As in (1931)

1934

f jff* g = ia2 sin"1 - -U y/^-a*).

J y/(a*-ar)
a
[As in (1932)

1935

f^_-l tan"1 or -i-cot"1

J jr-|-er
a
a
a
a

(1436)

1936

f *L. = JL log
J afa*
2a
x+a

[By Partial fractions

1937

f_**
J crar

lj.oge5.
2a
ax

do.

1938 J*sni &dx = cos .r.

J cos x dx = sin .r.

1940 ^tanxdx = log cos r.

j*cot ,r (fa* = log sin .r.

1942 \ seexdx = log tan ^+^.

Jcosecr<r = logtan ^ .

Method. (1940, '2), substitute cos as.

(1941, '3), substitute sin a;.

1944

j sin-1 x dx = x sin-1 x + y/(l x*).

1945

j cos~1xdx= xcos^x

a?).

CIRCULAR FUNCTIONS.

1946

J tan"1 x dx = x tan-1 x I log (1 4-^).

1947

f cot"1 x dx = x tan-1 x + \ log (1+^2).

1948

J sec-1 x dx = x sec-1 x log {x +

1949

J cosec_1.r dx =

321

1) } .

cosec_1cr + log {x-{-v/(xil)} .

Method.(1944) to (1949), integrate by Parts, dx.

1950

J log X dx = X log X X.

1951

f Jv
= *
Ja + 6coScr
\/(a

1 Q*9

or

[By Parts, Jifc

^ ( tan | f
")
2 V\a+6/)

lootanj.v
g v/(6+)-v/(6-)taui^

according as a is > or < b.

[Subs, tan |a;, and integrate by (1935 or '37)

VARIOUS INDEFINITE INTEGRALS.

GENERALIZED CIRCULAR FUNCTIONS.
1954

J*sin*rfci\

cos" dr.

j cosec"tfj\

Method. When n is integral, integrate the expansions in (772-4).

Otherwise by successive reduction, see (2(Jt!0). For j" eosec" cte, see (2058).
1957

1 twn xdx =
i
^-H
=- &c.
J
n 1
n6
no
Peoof.By Division; tan*.? = tan"~*x sec2 x tan""2 x, the first term of
which is integrable ; aud so on.

J (a+6 cos a1)"

(a* o*)B * J

Method.By substituting tan

sin a in the place of cos a, substitute
1959

f"***,.
J cos no?

= tan -|- w (~pD* Similarly with,

as.

fPi!^.
J sin nx
2 T

fiiH^
J cos nr

f^^.
J sm no?

INTEGRAL CALCULUS.

322

Method. By (809 & 812), when p and n are integers, the first two
functions can be resolved into partial fractions as under, p being < n in the
first and < n 1 in the second. The third and fourth integrals reduce to one
or other of the former by substituting \ir x.
1963
= - sr"" (-ir 1 8in <2r-^ecp
with e = jl.
cos n.e
n rml
cos as cos (2r 1)0
2n
1 964 25l* = _i_ 2'"--' (-1)- "ip,rg cosPr0, with e = A
Bin nx n mux
cos cos rO
h
The fractions in (1963) are integrated by (1952) ; those in (19(34) by (1990).
Formula? of Reduction.
1965

1966

{^^d.r=
J cosp.r

2rooB(-l).rrfr_rcoS(n-2).r(/r
J
cosp_ij?
J
cosp<r

COS ^ d> = -2 fh ("-I)* dlV + reoS(n-2).r ^

Jf sinp,

1967

fe^tf> =
J 8inpj?

2rcos(-l).r
J
Bin" x

1968

f~^=
J cosp x

2fgin(*-11>Jfto- fsin^-2)^^
J
cosp 1jd
J
cosp.r

Proof. Tn (1965).
Similarly in (1966-8).
1969
=

2 cos a cos ( 1) x = cos ns + cos (n 2) x, <fcc.

J* sinp ;r sin n.r dx

sinp .r cos jij ,
;

1970

cosp.rcosn,r ,
!+

1972
=

p
C . _,
,
ix
,
i l sm" 1x cos (n 1) .rrf.r.

j cosp x sin n.r do?

1971

Csm(n-2)xdv
j
smpx

p
C
__,
. ,
i\
j
f 1 cosp \r sin (n 1) .r a.r.
+" J

1 sinp .r cos n.r eir

sinp .r sin n.r

p-fn

f .
. ,
,
,
1 sinp \r sin (n lKr a.r.
V+n J

I cosp .r cos n.r dx

cosp.rsinn.r ,
p
C
-i
/
i\~,7

I cosp 1 x cos (n 1) .r dx.

p+n
j)+n J

CIRCULAR FUNCTIONS.

323

Proof.(1969). By Parts, J sin nx dx. In the new integral change

cos?ia:cosa: into cos(n \) x sin nx sin x. By successive reduction in
this way the iutegral may be found. Similarly in (1970-2).
Otherwise, expand sin's or cos's in multiple angles by (772-4), and
integrate the terms by the following formulae.
1973-1975
f sin px sin nx dx = Un(P~*)* _ * (p+*)*\
J
*
2 V
pn
p+n
)
and so -with similar forms, by (666-9).
1976

fS~5*f and f sin Pwdx are fonnd from

J cos no?
J coswir
f t* cos px sin _p.r ,
2 f zp+n_1 rfa
J
cosw*
J l + z2tt

when p and n are integers, by equating real and imaginary

parts after integrating the right side by (2023).
Proof.Put cosar + i sin x = z ; therefore izdx = dz. Multiplying nume
rator and denominator of the fraction below by cosnz+t sinra#, we get
cos px + i sin fx
g cos (p + n) x + i sin (p + n)x
^ zP7>
cos nx
1 + cos 2na> + 1 Bin 2na
1 + zu '
therefore

[ cy + <""P da> = _ U f
J
cos n.E
J 1+z

mwo
1978

fcoBpd?dj7
, fsin<r<v
.
, .
l r*
and 1 r*
are found m the same
J sin nx
J sin nx
way from
f cos^+?:sin^ dx = 2 f 2-^J-z.
Proof.As in (1970), by multiplying nnmerator and denominator of
cos
px + i sin *
px ,by cosj t sin na:.
c:
loon
f fos.rrfr
n f sin .r </,r
1980
J ^(cos iw-)
J v'(cosn.r)'
Putting y-COffiB + tsin.B we find ^^-{n.^ and therefore
y(cos iue)
ydx = 2~~*L' Hence, multiplying by i, we have
ft cos a; sin x ^
f dij
7(cos nx)
~~ J 2-!/"'
The real part and the coefficient of t in the expansion of the integral on
the right by (2021, '2), are the values required.

INTEGRAL CALCULUS.

324

1982

f
, * , = -^-rr tan' '(tan a /-U.
J a cos1 x + b sin* x
\
V u/

1983

g
= -5r-j {6 log (a cos x+ b sinx) + ax}. [Subs, tan x
J a + b tan x
a +0
f grr =
n tan- ^ffl + *\
[Substitute cot*
J a+bem x
/(ir+ab)
cot x va

1984

1985
1 nan
1986

I
J
f
J

[Subs, tan*

Sln y cos x dx = i tan"1 (a cos x) 2!L? [Substitute a cos*

ltd cos x
a
a
a;
ro
, sin x
q co&xdx
3
i = -7771l
r ten ->/-,sin xf: sin!-
[Subs,
l-i^cos'x
uV(l <r)
,/(l a*)
a5

1987

f 008X^(1 a* sin' x) dx = Jsinx\/(1 a'sin,x)+ ^ sin*'(a sin 2).

1988

[Substitute a sin a
I sinasv^Cl 'sin*x)c?x = jCosj!v^(1 a'sin**)
J
1
a1 log ja cosx + v^l <i! sin'x) jj. [Subs." a cos as

1989

j sin* (1 d'sin'x)' f?x = A cosx (1 a'sin**)*

+

1 OOn
r.

1QQ1

a*) J sinxv/(l a1 sin's) dx.

[Subs, a COS 0!

lg { (" cosec x + b cot x)6 tan ^z}

J sin m (<i + 6 cos x)
o2 >'J
1
sin x (a + 6 cos x)

ft 6 cosx
(s &1) sin x

lr sin x
(as 6'j (<i + 1 cos x)

f
tanr<?.r
_
1
CQa .^ns x \/(b ff)
J v/(a+6 tauax)
,/(6-a)
[Subs, cos x v/(6 a)

1992

f v/(a + 6sin'x) & = ^ ^ _v/fi_cosx

J
sin x
V( + o)

log {-/a cotx 4- >/(a cosec* x + b)}.

Method.By Division (1912), making the numerator rational, and in
tegrating the two fractions by substituting cotx and cos 2 respectively.
1993

[
<lr
- c ( f
dx
f
,]*
1
J a + 'lb eosx + ccus2x
m(.J2ecosx + & m J 2ccosx-t-6-r # ) '
where m = ^{6* 2c (a c)}. Then integrate by (1953).
1994

f
.
= f
.
.
J a cos a' + b sin tt' + c
J V (+') cos 0+c

Method.Substitute 0 = x a, where tan a = .

(1953)

EXPONENTIAL AND LOGARITHMIC FUNCTIONS.

325

^ggg

f F(sin x, cos x) dx
J a cos x-\-b siii.f+c'
F being an integral algebraic function of sin x and cos x.
Method.Substitute 0 = x a as in (1094), and the resulting integral
.takes
, the
,, form
.
fI J-!
/"(sinfl, cosfl)
(79 = fI 0'-r(cos ft) <70 +, fI sin
ft)
0 J/ (cos
J ^lcosft + i>'
J .4cosft + ii J A cos 0 + B'
since / contains only integral powers of the sine and cosine, and may there
fore be resolved into the two terms as indicated.
To find the tirst integral on the right, divide by the denominator and
iutegrate each term separately. To find the second integral, substitute the
denominator.

1996

f:(Oi-rA cos

F (cos x) dx
(a2+62 cos*) ... (a+6 cos a;)'

where F is an integral function of cos x.

Method.Resolve into partial fractions.
form J( A+Bcos a; (1951).
v
'

1997

Each integral will be of the

cos x + B sin x + C ^
f
J a cos x -f-6 sin tt+c

Method.Let 0 (as) = a cos x + 1 sin* + c ; .". <f>'(x) = a sin x + 6 cos x.

Assume A cos x-t B sin x + C = \<p (r) +fi<i>' (x) + v. Substitute the values of
f (.r) and <p' (a), and equate the coefficients to zero to determine A, fi, v. The
integral becomes

and the last integral is found by (1991).

EXPONENTIAL AND LOGARITHMIC FUNCTIONS.
1998

J e*F(x) dx can be found at once when F(;e) can be

expressed as the sum of two functions, one of which is the

derivative, of the other, for
JV{*(#)+f (.r)} dx = tT+{x).
1999

^eaxcosnbxdx and ^e* sin" bxdx are respectively =

acoshv+nbsmbx
n _j 6*
, +. n(n
l)b2 IC e7X cos"._.2 to
7 ,
; , ,,,
e cos"
-Vi3,

INTEGRAL CALCULUS.

326

and
a sin kr 6 cos 6.r e" sin""1 bx + n("
(e* sin-* fi.rrf.ra-+n262
er+n-6- J
Proof.In either case, integrate twice by Parts, ^e^dx.
Otherwise, these integrals may be found in terms of multiple angles by
expanding sin" a: and cos" x by (772-4), and integrating each term by (19512).
2000

Je*8in*# cosmtrrf.r is found by expressing sinncr and

cos" a; in terms of multiple angles.

Ex. : j e* sin5 a; cos' x dx. Put e" = z in (768),
(2t sin a;)5 (2 cos*)' = (z-z"1)5 (z+z"1)1
= (z-z"1)' (z2 - z"2)2 = (z7-z-7)-3 (z5-z-5) + (z'-z-') + 5 (z-z-1) ;
2* e* sin* a: cos* a; = e* (sin 7a; 3 sin ox + sin 3x + 5 sin a;) .
Then integrate by (1999).
2001

Theorem.Let P, Q be functions of x ; and let

$Pdx = Pu

SP&xdx = P2,

SP.QJx = P3, &c.

Then

PQndJC = P1Q-PiQ'-1+n (-l) P,Q-2- ... [n_Pn+l.

Proof.Integrate successively by Parts, j" P dx, &c.
2002

Theorem.Let P, Q, as before, be functions of x; and

g-

(n-i)v,g
.g
P
1
CP
~ (-l) (n-2) (n-3)"g,Q"-
+f^=T J
Proof.Integrate successively by Parts, [ (w 1) Qx ^

Examples.
2003

j"*--1 (log
=

*(-!)

... +(_1)-L Y

METUOD.-By (2001). P, = , P, = C, P,=

&c.

ALGEBRAIC FUNCTIONS.

327

2004
Ja?M+nxdvC = l of {\-\-nx \ogx-\-% {nx logcr)2+&c.} da;,
J
[By (149)
and each term of this result can be integrated by (2003).
2005

Cr-Xdx

'V U-l ~l~ (m-1) (n 2) + (n-1) (n-2) (n-3) +

pm-l
n 1 J loff
log a,1 '
Method.By (2002). Px = xm, P, = mxM\ P, = mV-1, &c.

The last method is not applicable when n = 1. In this

case, writing I for loga;,
oftftfl flrm~1</tr
,
, ,
, . m2/2 , msf ,
m*l*
, 9
2006 jTo^=logZ+m/+ra+r^+ox^+&cMethod :

"1
gill lOg X

log x = xlog x .

Expand
the numerator byJ (150),
v
v
y' and inte-

See also (21 61-6) for similar developments of the expo

nential forms of the same functions.
PARTICULAR ALGEBRAIC FUNCTIONS.
^007

J X(X-1) = ^Zl { (l-a;-1 ~ aF1 }

w C
1
+ n-2 i (l-x)"-2

Lt +
x"--)

n(n + \) ... 2 (n-1) ,

x
1.2... (n-1)
gl-x'
n being even.

2008

J (i+av* (!-*) = tot^ 008" v (rrf?)[Subs.

onm
2010

f
1 i sc/2 + v/(a;,-l)
J(l+!B')^-l)=72l0g
(!+')

(1918)

INTEGRAL CALCULUS.

328

P011
*S"-LA

dr
=
I
sin-\lV*4g=M), ae>bc
(c + ex1) S(a+bx*)
y/{acti-bc2)
V V aex' + ac I
i
lo/yra+^)+^f^-^)
{be? -act)
J{c + eJ)

Q012

= _g_1og ^<Hfc*-W.

2013

r
a+^w*
J (1-^)^(1+**)

ae<le

CSabs. -i-

[Subs. Xy/2
1-X*

^2 + /a+i()
w

[Subs. aV2
1 + X*

2014
J (1+^^(1-1-^)

2015

f ^gffl - ^ ^

2016

j<wfe = ^{^^^---^}[Substitute zx/(1+j) = a;v/2 in (2015-6)

2017
UA<

^
= i4 log
Jf (l-aj(2^-l)
B(2a-l)*+ai - 14 tan"' (2^-lj*
[Substitute z = 2 (2^1)*

2018

f
1 = tan-1
?
-.
J (l + x^lv/Cl+^-z8)*
{ ^(.l+x*;-*5}*
[Substitute 7[-7^--

2019
f
* , = log V^+Vl^- -L tan- 2%^.

2020

[Subs.

<?

f (1 + x) y(f+3a + 3j>) educes to (2010) by substituting ^

INTEGRATION OF ,rnl
7.
If I and are positive integers, and Z 1 < n;* then, w
being even,
* If l=n, the value of the integral is simply log (as"1)

a-'"1
INTEGRATION OF xn
+ l .

2021

329

J^ = 4,0^-1>+^'I<'^+I>

+ - 2 cos rlfilog(a*2x cosr/8+ 1) -- 1 sinr/yStan"1 Z^?

sin r/3
where 8 = , and 2 denotes that the sum of all the terms
n
obtained by making r = 2, 4, 6 ... n 2 successively, is to
be taken.
If % be odd,
2022

J* = JLlog(-l)

+i
1 ') -n 2 cosrZi81og(*>2-2^cosr)848V
n 2 sin r/^tan-1^!^^
M
sin*v8
with r = 2, 4, 6 ... n 1 successively.
If w. be even,
2023

= ~ 7 S cos rl& lo&

cos rfi+1)

,
H

2 < .
JOi
_,<r cosr/8
5 sin r/p tan 1 :
n
sin rp
with r = 1, 3, 5 ... n 1 successively.
If n be odd,

- !:ScosWp*log(r8-2<rcosrj8+l)+- SsinWp'tan-1^^!?^
n
n
sin rp
with r = ] , 3, 5 ... 2 successively.
Proof.(2021-4).
of (1917). We have

xi-\
Resolve | into partial fractions by the method

= L = 8ince a" = T 1. The different

F (a)
Jttt"-1

valnes of a are the roots of x" 1 = 0, and these are given by x = cos r/3
t sin r/3, with odd or even integral values of r. (See 480, 481 ; 2r and 2r + l
of those articles being in each case here represented by r.) The first two
terms on the right in (2021) arise from the factors x 1 ; the remaining
terms from quadratic factors of the type
(x cos r/3 t sin r/3) (x cos r/3 + 1 sin r/3) = (x cos r/3)1 + sin1 r/3.
These last terms are integrated bv (1923) and (1935). Similarly for the
- (2022-4).
2u

INTEGRAL CALCULUS.

830

2025

If, in formulae (2021-4), + 2 Qtt - r/3) sin rlB be

added to the last term for the constant of integration, the

integral vanishes with x, and the last term becomes
-i- 2 v
loi -i * sin rB
=P ism rlB tan 1 n ,
n
lx cos rp
reading - in (2021-2), and + in (2023-4).

2026

Jf4
a+bx* = I(4)'
a \ b I
J(^^
l+zH

where a::n = Lxn.

onot?
2027

Then integrate by (2023-4).

(,.*m-1+lrn-m-1 ,
4^ .
,
_,tr-cosr
1
5-r-i
dx = sin /nr/J tan 1 :
J
.t +1
n
sin rp

where B = n-^n, and r = 1, 3, 5, ... successively up to ft 1

or n 2, according as n is even or odd.
2028
I
J

r
x +1

dx = 2 cos mrB . log (.r2 2.r cos rB-\- 1 ),

n

with the same values of r; but when n is odd, supply the

additional term
( 1)"* 2 log (a; + 1 )-=- /.
Proof.Follow the method of (2024, Proof).
Similar forms are obtainable when the denominator is as" 1.

2029
C
xl~ldx
1
v
/
n i a -i <r co8<A
j x*-'2x cos nfl+l = n^g S CS (-04.tM
2 sin (n /) <.lg (.r* 2.r cos <f>+l),
where f = 0-\
and r = 0, 2, 4, ... 2 (ft 1) successively.
ft
But if the integral is to vanish with x, write tan-1 X S1P ^ ,
as in (2025).
1-zcos*
Proof.By the method of (2024).The factors of the denominator are
given in (807 1, putting y equal to unity.

,
-8
INTEGRATION OF J xm (a + bx*)q.

2030

331

f^ffi0" d* = T 2 ^ { cos (Hfi+W

Xlog(^-2^cosri8+l)-2sm(W)8+imir)tan-1^^y^) ,
sin rp )
with the values of j3 and r in (2021-4).
Pkoof.Differentiate the equations (2021-4) i times with respect to I,
bj (1427) and (1461-2). If m be negative, integrate m times with respect
to I, and the same formula is obtained by (2155-6).
In a similar manner, from (2027-8) and (2029), the general
terms may be found for the integrals
2032
f {^(-1)"^ ^ (log*)* dx and C xl-'(\ogx)mdx
J
J xin 2lr"cosM0+l'
INTEGRATION OF $ xm (ax-\-bxnf dx.
m-r- 1
Rule I. WJien
is a positive integer, integrate
n
i
by substituting z = (a + bx")q. Thus

2035

| xm (o+&r) rf^ = X J

dz.

Expand the binomial, and integrate the separate terms by (1922).

2036 But if the positive integer be 1, the integral is known
at sight, since m then becomes = n 1.
2037

Rule II. When HLt -j_ JL

L
n
q
substitute z = (ax"n + b)q. Thus

is a, negative integer,

Expand and integrate as before.

2038 But, if the negative integer be 1, the integral is
found immediately by writing it in the form
x

(ax-*+b)q dx = I x-n-\ax-n + b)* dx

= r3

v (*-+ 6)

INTEGRAL CALCULUS.

332

2039

Examples.
Tofind jx*(l + x*)*dx. Here m=i, n=i, p=2, q=3, ^i = 3,

a positive integer. Therefore, substituting

x = 6y' (y5 1), and the integral becomes

y = (l + x*)*,

x = (y* 1)',

the value of which can be found immediately by expanding and integrating

the separate terms.
2040

dm = ^ (a + bx')*.

For m"*"^ = 1 (2036) ; that is, m + 1 = n, and the factor xs is the derivative
n
of i.s4.
2041

j G +

dx, or jV1 (l+x*) dm.

Here m = -, n= J, p = 2,

g=3,

TO - +

= 2, a negative integer.

Therefore, substitute

y=
1)*, x = (,/ 1)-', xr = Qy1 (y* T)~*. Writing the integral in
the form below, and then substituting the values, we have
j x-1 (x-* + 1 ) x,dy = - 6 j" y* (y'-l) dy,
which can be integrated at once.
2042

- = J[.-'(a+^-'ifa.
Jf x (a + ij")
therefore, by (2038), the integral

Here^
n + -2.
g = -l;

= I x""-1 (ax- + 6)-' dx = - log (ax-"-f *)

J
rki
REDUCTION OF j j?m (a+kr")1' <r.
When neither of the conditions in (2035, 2037) are ful
filled, the integral may be reduced by any of the six following
rules, so as to alter the indices m and p, those indices having
any algebraic values.
2043 I. To change m and p into m-f n and p 1.
Integrate by Parts,
J*xmdx.
2044 H. To change m and p into m n and p + 1.
Integrate by Parts, Jxn_1 (a + bxn)pdx.
2045 HI. To change m into m + n.
Add 1 to p. Then integrate by Parts, Jxmdx; and also
by Division, and equate the results.

REDUCTION OF xm(a + bx')p dx.

333

2046 IV. To change m into m n.

Add 1 to p, and subtract n from m. Then integrate by
Parts, j"xmdx; and also by Division, and equate the results.
2047 V. To change p into p + 1.
Add 1 to p.
Then integrate by Division, and the new
integral by Parts, Jxn_1 (a + bxn)p.
2048 "VI. To change p into p 1.
Integrate by Division, and the new integral by Parts,
jx'-1 (a + bxn)p.
2049
I.
II.
III.
IV.
V.
VI.

Mnemonic Table foe the same Rules.

m+n,
m n,
m+n
in n

p 1
p+1

P+ 1
p-1

By Parts (m).
By Parts (p).
(p + 1), Parts (m) and Division.
(p + 1, m n), Parts (m) and Division.
(p+1), Division, and tlie new integral by Parts (p).
Division, and the new integral by Parts (p).

By applying the rules, Formulae of reduction are obtained.

Thus, any of the six values below may be substituted for the
integral
jV" (a + bxn)p dx.

2050-2055
m-\-l
TT

m+l J

jT~n+1(a+bx,,)p+l

m-n+l f m_,

. . ._+, ,

in. ^(^r1 - *(++p+i) (V-(a+ ^

a (m+l)
a(wi + l)
J

6(m+np+l)
v
^

6(m+np+l)J

rm+1(a+&rB)p+1 . tn+n+np+1 f , . Xp+1 ,

an
+ an (p+1) J ^(+^)P+1^

vi. ^'fa+y +
f Tm+np+1
>/i+n/>+lJ

^,

33-4

2056

INTEGRAL CALCULUS.

Examples.
To find J ^f^-*') dx. Apply Rule I. or Formula I. ; thus
jaj-'Ca'-as')*^ =-^-J(o-a!,)t+J jx-1(o,-a!,)-,Ji (1927).

2057

To find [ --dx. Apply Rule II. or Formula II.; thus

J (o1 x1)*
x< (a'-xT* ds = x* (a'-Jyt-S x> (a'-z*)-* dx (1934).

2058

coBec"6dd. Substituting sin d = x, the integral becomes

J sin O^dx = ^z(l-xt)-i dx.

Apply Rule III. ; thus, increasing p by 1 and integrating, first by Parts

x~mdx, and again by Division ;
f
( l-x') dx = "'"a-**)*
dx,
J
1 ?
1 m J
[x-m(l-x*)*dx = jx (l-x,)-*dx-jx, (1 -*')-* <ix.
Equating the results, we obtain
2059

\x--a-Jy*dx=^(i-x>)*+^\z>-(i-x>r*d*.

By repeating the process, the integral is made to depend finally upon

farI(l-x,)-*da
or
j (l-a1)-* das,
according as to is an odd or even integer (1927, '29).
2060 sin" ddO is found in a similar manner by Rule IV. The integral
to be evaluated is j x" (1 x')-* dx ; and the integral operated upon is
Jx""* (1 x1)' dx. Otherwise apply Formula IV. See also (1954).
2061

To find [ } ^r j. Apply Rule V. p = r, and increasing p by

J (x + a )
1, we have, first, by Division,
j (x' + a*)1" dx = | x* (x5 + a')

dx + a^ (x + a')

dx.

Integrating the new form by Parts, \ x (z, + a,)~r dx, we next obtain

Substituting this value in the previous equation, we have, finally,

2062
C
d.r
_

2r-3
(
dx
J (,i''-+<rT
2(r- 1) o- (.i-i+ol)r-1 "i" r (2r-2) J {.r'+ay-1'

REDUCTION OF j" sin" 0 cos" Odd.

335

This equation is given at once by Formula V. Thus r is

changed into r 1, and by repeating the process of reduction,
the original integral is ultimately made to depend upon (1935)
for its value if r be an integer.
Another formula for this integral is

2063

J{y+jBp = i.2.../r-i) W*\Vfi tan 73>

Proof.Write 0 for a5 in (1935), and differentiate the equation r 1

times for ft by the principle in (2255).
2064

To find J(o*+a^)*"(fe.

Apply Rule VL

By Division, we have

f (a*+*)*"<k = a' f (ft' + x5)*"-1 <te + f ** (a' + ar*)'""1 cfa.

The last integral, by Parts, becomes
f x> (a> +x%)i"-1dx = JL x (rf+ar')*" - f (<*' +**)*".
J
'
n
n J
Substituting this value in the previous equation, we obtain

2065

f (<t+aF)*dx = *!+ )" + ? f (a*+a*)to-*dar,

J
n+1
n+1 J
a result given at once by Formula VI.
If n be an odd integer, we arrive, finally, by successive
reduction in this manner, at J'(a2+,'c2)4(Z,c (1931).
2066 The integral jsinm0 cos" Odd is reducible by the fore
going Rules I. to VI., if, in applying them, n be always put
equal to 2; if p be changed into p2 instead of p + 1; and
if Division be always effected by separating the factor
cos!0 = 1 sin*0.
Pboof. J sin" 0 cos" 0 dB = j" xm (1 a1)**"1' dx, where 2 = sin 0. Thus
n = 2 always, and the index | (p 1) is increased by 1 by adding 2 to p.

Thus, Rule I. gives the formula of reduction

2067
wi + 1

m -f- 1 J

INTEGRAL CALCULUS.

336

But the integral can be found by substitution in the fol

lowing cases :
If r be a positive integer,
2068

J co&ir+lxzva*xdx =

z^'z'dz, where z = sinaj.

2069

\ sin2r+1.r cospa rfcT = \ (1 z^'afdz, where z = cosx.

If m-\-p = 2r,
2070

f sin"d7 cospJ7(ir = f (l+zi)r~1zmdz, where z = tanx.

FUNCTIONS OF a + bxCJC*.
The seven following integrals are found either by writing
2071

a + bx + cx* = {(2ea; + &)8 + 4ac-&2j +4c,

and substituting 2cx -\-b; or by writing

2072
a + bx-cx* = {4ac+bl-(2cx-b)1j -~ 4c,
and substituting 2cx b.

orm

f
dx
1
log 2cH-ft--x/(y-4gc)
J a+te+ca* ~ -/(6*-4ac)
8 2c^+6 + v/(64-4ac)
or

m
T^r- tan 1 v/(4ac-62)'
v/(4ac-6i)

according as 5* > or < 4ac (2071, 1935-6).

3074.

f
dv
1
log v/(y+4ac) + (2CJ-6)
J a+bx-cj* ~ #+4a<? K v/(&4+4ac)-(2ftr-6)'
(2072, 1937)

2075

2076

1_

-i

2cx b
(2071-2, 1928-9)

FUNCTIONS OF a+bz+cx*.

337

2077

jVCa+for+c*8) dx = Jc"4 J v/(y2+4ac-52) dy.

2078

jV(a+to-c.r2) d.r = c_* jV(4ac+&2-y2) dy,

where y = 2cx + b.

2079

The integrals, are given at (1931-3).

(+6.p+Ccra)p
J (y2+4c b1)"'
L+WF^-'H'i
[By (2071), the integral being reduced by (2062-3).

J (a+bx+cx*)*
I i {2cx+b) da?
(
bl\ C
dx
~ 2c J ( + + c^)" + Vm
2c/ J (a+bx+ ex*)"'
The value of the second integral is (<^4-foE-(-ca:2)1_,' + (1p),
unless p 1, when the value is log (a + bx + cx2) .
For the
third, see (2079).
Method.Decompose into two fractions, making the numerator of the
first 2cx + b ; that is, the derivative of a + bz + cx*.
2081

"f**i dx may be integrated as follows :

I. If fe2 > 4ac, put a and j3 for -

^)} an&t
Ac
by Partial Fractions, the integral is resolved into
_1
j fc2 dx - fJ*+2 dxl.
c(a ft) ( J x* a
J x*p
)

(1936)

H. If fc2 < 4ac, put = 2 and 2^(ac)-& = TO* and the

c
c
integral may be decomposed into
2082

1
\ ((<1-Pn)*+Idjc _ C{q^pn)x-qm rf >
Imnc C J xr+mx+n
J
or rnx-\-n
j

the value of which is found by (2080).

III. If 62 = 4oc,
2083

f tt$t = <r^7-!+ -77CTtan"W) <2062>

Jo+ir+cr
2a + for 1 x/(2a6)
\ V2o/
2 x

338

INTEGRAL CALCULUS.

2084

} a+bj?+cx* ~

h(2a+bj*)'

2085

Jf a+6ct-+ca4 =^(-1
b is/{2ab)
\ v 2a)
.
REDUCTION OF jV (a + bx" + c.v*n)p dx.

2a-\-hi~)
(
)

Note.In the following Formula of liedaction, for the sake of clearness,

xm (a + bxn + cx**)r is denoted by (m,p), and the integral merely by \ (m, p).
2086

J (m, p) = (m+1, p)

bnp j(m+n, p 1) 2ch;> J (m+2,p 1)

2087

ft"

(1).

i (m,p) = (m-n + 1,^+1)

_(m_n+l) J(m_,|, + l)_2cn(/>+l)J(m + ,iO-(2).

2088

2cn(p+l) f (m,p) = (ro-2n+l,p+l)

-(m-2n+l) J(m-2,p+l)-6(/>+l) j (m-n,p)...(3).

2089

(m+np+1) J (ro, p) = (m+1, />)

+ anp ^ (m, p l)cnp f (m+2n,/> 1)

2090

(+ 2np + l) j (m, p) = (m+l,p)

-\-2anp ^ (m, pl)-\-bnp |*(m+n,p 1)

2091

(4).

(5).

6(m+np+l) f (m,j>) = (m-n+l,p+l)

a(m m + 1) I (m n,p) c(m+2np+n + l) I (m+7j,/>)

^
"
(0).
2092

bn(p+l)((m,p)=-(m-n+l,p+l)

+ (m+2/>+n+l) f(m-n,/>+l)-2an(j>+l) ((m-n,p)

J
J
(7).

REDUCTION OF J as* {a + bx* + cx'-)p dx.

2093

339

cn{p+\) Um,p) = (m-2 + l)jP+l)

+an(p+l) I (m 2n,j)) (m+np n + \) l(*~2n,p+l)

-_
J
()

2094

on(p+l) ( (m,/>) =-(m+l,;>+l)

+(m+np+n+l)f(m,|>+l)+cn(p+l)J(m+2n,i)...(9).

2095

2an(p+l) f (to,j>) = -(m+l,j+l)

+(m+2np+2n+l) f(m,jp+l)-ftji(|+l) f(m+n,p)

J
^
(10).

2096

a(m+l) f (m,p) = (m+l,p+l)

-b{m+np+n+l)((m-\-n,p)c(m +2p+2+l)f(
"

2097

(11)-

c (m+2np+l) J (to, j>) = (m-2n+l,

6(m+p n+1) I (to n,p) a(m 2n+l) I (to 2/i,/))

J
J
(12).
Proof.By differentiation, we have
2098
J
= m j" (ml,p)+bnp ^ (m+n1, p - 1) + 2cnp J (wi 4-2n 1, |> 1).
Formulae (1), (2), and (3) are obtained from this equation by altering the
indices m and p, so that each integral on the right, in turn, becomes
Again, by division,
2099 jO,J>) = *$(m,p-l) + b$ (m+n,p-l)+c$(m+2n,p-l)...(A).
And, by changing i into m n, and /> into p+ 1,
2100 J (-! P+l) = a (m-n,p) + b(in,p) + c ^(m+n,p)
(B).
Formulas (4) to (12) may now be found as follows :
(4) , by eliminafciug J (m + n, p1) betweon (1) and (A)
(5) , by eliminating J (m + 2n, p 1) between (1) and (A)
(6) , by eliminating J (mn, p + l) between (2) and (B) ;
(7) , by eliminating J (m + n, p) between (2) and (B);

340

INTEGRAL CALCULUS.

(8) ,
(9) ,
(10) ,
(11) ,
(12) ,

from
from
from
from
from

(4), by changing m into

(4), by changing p into
(5), by changing p into
(6), by changing m into
(6), by changing m into

m 2n, and p into^ + 1 ;

p + 1;
p+1 ;
m + n;
mn.

If a and /3 are real roots of the quadratic

a + bxn + cx2n = 0, then, by Partial Fractions,

equation

and the integrals are obtained by (2021-2).

But, if the roots are imaginary,
, !+!
2108

= Kf)"ji=raary (2029)

AvliLi'e

cos 0 = -1
2v/(ac)
' r/'

2103

and

z = ^-^'*as.

is reduced to (2079-80) by (2097).

J(V
(a

&
(2075)
2104 J (*+*V(+4*+s*) ~" j'^(A+By+Ctf)
where y = (x+h)~l, A = c, B = b2ch, C = a bh + chr.

2105

2106

=
coS-'l+^ = _i=co8h-'l+^.
(+A)vV-l
yi-A1
ijs
1 __ sin"1^-^.
_:_-il + '<a'
[By (2181).
x+h
J-x+h) Sl-z* -/h'-l

Method.Substitute (x + h)'1, as in (2104). Observe the cases in which

h=L
yr~'dy
2107

J(*+A)V(+^+er!) ~

S^(A +By+Cy*Y

with the same values for A, B, C, and y as in (2104).

integral is reduced by (2097).
91 OR
4iU

f
(j+m) rf.r
J (.*+/*) V{a + b*-+c.t*y

The

INTEGRATION BY RATIONALIZATION.

341

Method. Substitute 0 by putting x = ft tan (0 + y), and determine the

constant y by equating to zero the coefficient of sin 20 in the denominator.
The resulting integral is of the form I (" ' '
dd.
J s/(P+Qcoa20)
v't-F+y cosiitfj
jrms, and integrate by substituting
substitui
Separate this into two terms,
sin 0 in the first
and cos 0 in the second.

<l>(ir) dx
'(x)y/{a+bx+ca?)'
where $ (x) and F(x) are rational algebraic functions of x,
the former being of the lowest dimensions.

2109

JTC

Method.Resolve ^,

' into partial fractions.

The resulting integrals

are either of the form (2107), or else they arise from a pair of imaginary roots
of F (x) = 0, and are of the type I
-^%-, .; X ,
rr. Substitute
xa in this, and the integral (2108) is obtained.

INTEGRATION BY RATIONALIZATION.
In the following articles, F denotes a rational algebraic
function. In each case, an integral involving an irrational
function of x is, by substitution, made to take the form
F(z)dz. This latter integral can always be found by the
method of Partial Fractions (1915).

8110 ^H^)f'(7$f-l/-Hfc
Substitute -~ = z\ where I is the least common de-

f+9*
nominator of the fractional indices ; thus,
X~b-gz"

dz~

(b-gz'y '

\f+gx) ~*

*''

the powers of z being now all integral.

2111 p-^^,(^),(^)U.^.
Reduce to the form of (2110) by substituting as*.

INTEGRAL CALCULUS.

342

2112

\ F {^(a+y/mx+n)} dx.

2113

j F{x, x/(bxcx*)} dx.

And therefore

V{bx ox*) =

2114

Subs. y/(a + y/mx + n).

Substitute x = -J^>
g =-

jV fa, ^/(a+iaH-ca-2)} rf*.

Writing Q for a-j-foe+ca^, -f7 may always be reduced to

the form ^t-^ft^M, in which A, B, G, D are constants or
rational functions of x.
Eationalizing this fraction, it
takes the form L+M\/Q.
Thus the integral becomes
I Ldx+ 1 M
the first of which two integrals is
J.J
,
r l\fQ
rational, while the second is equivalent to jjk
which
is of the form in (2075).
J ^

2115 Otherwise. (i.) When c is positive, the integral may be made

rational by substituting
a cz*
dx
2c (bzczi a)
,, , , , ,N
, la * , \

(ii.) When c is negative, let , /$ be the roots of the equation

a + bx ex* 0, which are necessarily real (a, b, and c being now all
positive), so that a + bx ex1 = c (% a) (/3 x). The integral is now made
rational by substituting
cz3 + l

dz

(cz'+l)3

'

c^ + l

In each case the result is of the form J F(z) dz.

2116

j\rwF {.rB, Va+bx'+cx**} dx,

when 2Lt_ is an integer, is reduced to the form (2114) by

substituting xn.

2117
f F [x, v/(a+fcr),

Substitute

= f$?,

INTEGRALS REDUCIBLE TO ELLIPTIC INTEGRALS.

and, therefore, x = a , fz. ,

flur 6

\/(a + bx) = z }a,

s/{gz-b)

/i
vA^-ftf)
V(j+gv- V{(JZ*_b)>

^ _ 2(bfa>/)z
~dz- {g?-hy '

The form f-Ffz, \Z(gz2b)} dz

comprehended in (2114).
2118

343

is obtained,

which is

\x*F{x*, /(a+fiV"), bx%y/(a+btx*)} dx,

when m~^~

is an integer, is reduced to the form F(z)dz

by substituting z = bxn + \/(a + b2x2n), and therefore

m+l ,

2119

(a

* F (xn) efa? is rationalized by substituting

m+l or m
+l
either (a + bxn)q or (ax~n + b)q1 according as ^
l--^
n
n
q
is integral, whether positive or negative.
a?"-1!? {.rm, x", (a+bxn)q} dx, when is either a
positive or negative integer, is rationalized by substituting
(a+bxn)q.

INTEGRALS REDUCIBLE TO ELLIPTIC INTEGRALS.

2121

j^ta v^o+kt'+^+d.r'+e.r4)} dx.

Writing X for the quartic, the rational function F may

always be brought to the form r^yL
and this again, by rationalizing the denominator, to the form

344

INTEGRAL CALCULUS.

M+N\/X, where P, Q, P' , Q', M, iVare all rational functions

of x. j" Mdx has already been considered (1915).
JNy/X dx = I : dx or \
By substituting a: = ^

jf, where B is rational.

, and determining j? and g^ so

that the odd powers of ?/ in the denominator may vanish, the

last integral is brought to the form

2122

JTi^wW

B being a rational function of y, may be expressed as the

sum of an odd and an even function; thus the integral is
equivalent to the two

The first integral can be found by substituting \/y.

The second, by substituting J^ for y1, can be made to
depend upon three integrals of the forms

2124 JVi-J?i-*v

dx

v/l-fcV dx
Vl-x*

(l+f)v'l-rM-^*
By substituting ^ = sin-1 a;, the above become
2125

(*$.,

(Vl-tfsin^d*.

<*P

J (1 +n sin2?) ^l-K1 sin2?'

These are the transcendental functions known as Elliptic
Integrals. They are denoted respectively by
2126

F(k,4>),

E(k,4>)t

n (,*,?).

INTEGRALS REDUCIBLE TO ELLIPTIC INTEGRALS.

APPROXIMATIONS TO F(k,

345

AND E (k, f) IN SERIES.

When k is less than unity, the values of F(k,<t>) and

E(k, <j>), from the origin <p = 0, in converging series, are
2127

Wti^t-v^+ftt^-rtt^'"

2128

E (k,<f>) = <f>+ *At-ggA<+*gg A.-...

I c_.iy+il'3-5 ... n 3 fen j
+( 1J
2.4.G...n
2-1

,
w n r*TP

o
C-'

[w being an even integer.

,1

sin 4<&

4 sin 2<& . .

sin6<&

6sin4< , 15sin2<A

in,

A _ sinnA
nsin(n 2)^
C (n, 2} sin (n 4) <f>
An
n
n2
n4
- C<n' 3)nS^n~6) * + ...
(, jn) f
Pkoof.In each case expand by tbe Binomial Theorem ; substitute from
(773) for the powers of sin^, and integrate the separate terms.
The values of F(k,<j>) and E{k,<p), between the limits
f = 0, <j> =
are therefore
2129

2130

But series which converge more rapidlv are

2131
F(^)=^(1+(i)V+(2L|)W(i^)V+4c.}
2 Y
( CKJ
v, . -

' .V !

346

INTEGRAL CALCULUS.

2132

where

n = i , ,),-rf
l+v/(l *-)

2133 f 77 , , y.f" 3^ r. ^lien ^O'O can be expressed in the form (x

)f[x~\

)' *s integrated by

substituting x-\ .
If b is negative, and F(x) of the form (x+ t)f(x

);

substitute x
x
0104.

F (<r) dr

Substitute

#+ =55.
a?
Hence
|L = (
=
^ ) dz,
y/*>
V2v/s-2 2^8+2'
and the integral takes the form

where P, Q are rational functions of z. Writing Z for the

cubic in z, we see that the integral depends upon
f
Pd*
and
f'Q(z2)dz

J V'Z(z'2)
J VZ(z2)
the radicals in which contain no higher power of z than the
fourth. The integrals therefore fall under (2121).
OI QK

f
F ('r) ffl'
J <y'(a+kv-+CJl+(hva+a^y

Expressing F(x) as the sum of an odd and an even func

tion, as in (2128), the integral is divided into two; and, bysubstituting ,r2, the first of these is reduced to the form in
(2121), and the second to the form in (2134) with a = 0.

INTEGRALS REDUCIBLE TO ELLIPTIC INTEGRALS.

2136

Jv

347

7 " 1 ,lr

Put x = y-\- a, a being a root of the equation a-\-bx-\-c^-\-dd?= 0;

and, in the resulting integral, substitute zy for the denominator.
The form finally obtained will be

which falls under (2134), P and Q being rational functions

of z.
91 Q7

F(x)d.v

Expressing F(x) as the sum of an odd and an even func

tion, as in (2123), two integrals are obtained. By putting
the denominator equal to z in the first, and equal to xz in the
second, each is reducible to an integral of the form
(P+Q \Za+/3z4) dz,
which falls under (2121).

2138

2139

J?^-- 71 *(;*}

iw>

S^fa-l'tir*}

Proof.Substitute cos "'a in (2138), and 2 tan-1 a in (2139).

2140

according as 6 is > or < 2a.

PkookSubstitute accordingly, x = 2u sin1 <f> or x = b sins0.

2141

,
Jf-,
v(a #)(# )(.r+c)

V( + c)

\v+c

Proof. Substitute as = (a fc) sin2^>, * being < a and >6.

INTEGRAL CALCULUS.

349

SUCCESSIVE INTEGRATION.
2148 In conformity with the notation of (1487), let the
operation of integrating a function v, once, twice, ... n times
for x, be denoted either by
I v, \ v, ... \ v,
or by
d_x, d
/x Jix
%,'nx
the notation d_x indicating an operation which is the inverse
of dx.
Similarly, since yx, yUs y^, &c. denote successive
derivatives of y, so y_x, y^, ij-sx, &c. may be taken to repre
sent the successive integrals of y with respect to x.
2149 Since a constant is added to the result of each in
tegration, every integral of the th order of a function of a
single variable x must be supplemented by the quantity

In 1

n 2

Jnx

where Oj, a2, as ... an are arbitrary constants.

Examples.
The six following integrals are obtained from (1922) and
(1923).
"When p is any positive quantity,
2150

L" = (p+i)o+2)...a+) +L-

When p is any positive quantity not an integer, or any

positive integer greater than n,
2151

illDl

,f n

When p is a positive integer not greater than n, the fol

lowing cases occur
2152

2153

Lh-tjEj^+l0-

LJ=^?<*1**-*+Lj<

SUCCESSIVE INTEGRATION.

2154

jf

(<|?1

349

)_l|+r

o.

For the integral within the brackets, see (2166).

The following formula is analogous to (1461-2)
2155

f Sin } ax = 8in \ (ax-inn) + f 0,

JnxCOS)
a" COS J V
2 J^2nx

SUCCESSIVE INTEGRATION OP A PRODUCT.

Leibnitz's Theorem (1460) and its analogue in the Integral
Calculus are briefly expressed by the two equations
2157

/>() = (dx+8xyuv,

D_nx(uv) = (dx+8x)-uv;

where D operates upon the product uv, d only upon u, and 8

only upon v. Expanding the binomials, we get
2159

D^uv) = unxv +(-!) xvx +w^"^tt(,-g).t?to +&c-

2160 D.^uv) =u.nicv n_(.+1)a,t>af+^jilZit_(.+!)xt>to &c.

Proof.The first equation is obtained in (1460). The second follows
from the first by the operative law (1483) ; or it may be proved by Induction,
independently, as follows
Writing it in the equivalent form
I (uv) = I uv n I
uvx+ n
[
uvu &o
(i.),
Jm
}nx
J(l)
1.55 J(b2)i
make n = 1 ; then

| (uv) = f uv~ |
| uvlx&c
(ii.),
a result which may be obtained directly by integrating the left member suc
cessively by Parts. Now integrate equation (i.) once more for x, integrating
each term on the right as a product by formula (ii.), and equation (i.) will
be reproduced with (w + 1) in the place of n.
2161

f eaxxm= eax(a+dx)-nx'n+\ 0.

Or, by expansion,

2162
[ w = fl }
*<*+*) fo"1? -'-&<,. ] + f 0.
Jm
a" L
a
1.2a*
) J*
If m be an integer, the series terminates with ( l)"1 n'""-i-am.

INTEGRAL CALCULUS.

350

Similarly, by changing the sign of m,

2163
f 1 1 f J_ + nm + n faD m Cw+i)
Jm x"
aulx'* axm'1
1.2 a'x'*1

&c ")
')

f 0
Jx '

Proof.Putting tt = e*, = a" in (2158), the formula becomes

[ e^as" = (d,+ 3,)
= (46*+ e^a.)""*" = e" (a +3,)-" as".
J MX
Here eax is written before
within the brackets, because S does not
operate upon ex. Observe, also, that the index n affects only the opera
tive symbols dx and Zx, but it therefore affects the results of those operations.
Thus, since dxr"x produces aeax, the operation dz is equivalent to aX, and is
retained within the brackets, while the subject e"x, being only now connected
as a factor with each term in the expansion of (a + x)~", may be placed on
the left.
2164

fe"*- dx = e- \arJ
a C
a

nice

f*"

e* ( 1

ILi!?^}} *--&c. I
or
)
m

, "('+!) it J

Proof.Make n = l in (21G2) and (21G3).

2166

f ^(logo.')" = f e{p+n)xXm.

[Sabs. log*.

Hence the integral of the logarithmic function may be

obtained from that of the equivalent exponential function
(2161).
For another method, see (2003-5).

HYPERBOLIC FUNCTIONS.
2180 Definitions.The hyperbolic cosine, sine, and tan
gent are written and defined as follows :
2181
cosh x = \ (er+e-x) = cos (ix).
(768)
2183

sinh x = | (exe~x) = i sin (ix).

2185

tanh.r=

e\r~e~\ =-ttan(i>).
(f-\-e~x

(770)

HYPERBOLIC FUNCTIONS.

351

By these equations the following relations are readily

obtained.
2187

cosh 0=1;

2191

sinh0 = 0;

cosh oo = sinh x = ao .

cosh* a? sinh* x 1.

2192

sinh (*+#) = sinh* coshy+cosh* sinh?/.

2193

cosh (*+?/) = cosh** cosh 7/+ sinh* sinhy.

2194

tauHv+y) = l^\^Y.
1+tanh* tanh y

2195

sinh 2*

2196

cosh 2* cosh2 *+ sinh8 x.

2 sinh * cosh *.

= 2 cosh2 07-1 = 1+2 sinh2*.

2197
2199

sinh 3* = 3 sinh *+4sinh3*.

2200

cosh 3*

4 cosh3* 3 cosh x.

2201

tanh 2* =

2 tanh x
1+taulr*'

2202

tanh 3* =

3 tanh x + tanh3 x
1+3 tanh-* "

2203

sinh * = ^^= ;

2205

,r 1
tanh4
2 = v/^
V cosh
(ins 1 *+
.
1

2208

cosh * = ]+\m"\* ;
1 tanh" fx

cosh|- = ^J^H1.
cosh x 1
sinh *

sinh *
cosh x + 1 '

sinh * = 12t.Mj;tr
1 tanh'' fx

2210

INVERSE RELATIONS.
Let u = cosh*, .-. * = cosh1w = log (tt+-s/ua 1).

2211

v = sinh *, .*. * = sinh-1 v = log (v-\-\/v2-\-l).

2212

to = tanh*, .-. x = tauh-1w=i\og(j^^.

INTEGRAL CALCULUS.

352

GEOMETRICAL INTERPRETATION OF tanh S.

2213 The tangent of the angle which a radius from the centre
of a rectangular hyperbola makes with the principal axis, is
equal to the hyperbolic tangent of the included area.
Proof.Let 0 be the angle, r the radius, and S the area, in the hyperbola
y' = l or r2=sec2fl; then
S=

sec20ci0 = |logtan(^ + 0).

(1942)

Therefore e23 = ] + i9 therefore tan 0 = e*~e"* = tanh 8.

1 tan0
e^ + e"

(2185)
v
1

VALUE OP THE LOGARITHM OP AN IMAGINARY QUANTITY.

2214

log (a+ib) = ilog (2+62)+itan-1-* .

a
l+i
f( =itan-1A
1-i

By (771).

DEFINITE INTEGRALS.
SUMMATION OP SERIES BY DEFINITE INTEGRALS.
2230

/ dx = [/()+/(a+cir)+...+/(a+n(r)] d*,

where n increases and dx diminishes indefinitely, so that

ndx b a in the limit.
2231

Ex. 1.To find the sum, when n is infinite, of the series

- + -ZT + -TO
n +l n+2

^ + ^ + _i^ +
a
a+dx a+2dx
2232

thus,
n+n

dx

+ *L = P^ = log2.
la
Ja x

Ex. 2.To find the sum, when n is infinite, of the series

ra'+l2

s + 22
ffa

?i2 + 32
rfs

ra2 + n2
d

da;
f1 (fa

,,,,

THEOREMS RESPECTING LIMITS OF INTEGRATION.

353

THEOREMS RESPECTING THE LIMITS OF

INTEGRATION.
f*o
|
Jo

2233
2234

Pa
(*) dx = \ <f> (ax) dx.
J

[Substitute a-x.

fV (*) dx = 2 fia<t> (<*) dx,

Jo
Jo
or zero, according as ^(*) = + <j>(a x) for all values of x
between 0 and a.
Ex.

I Biuxdx = 2 Psinziz.
Jo
J0

I cosa;da; = 0.
Jo

If
= 0( #), that is, if $(x) be an even function
(1401) for all values of x between 0 and a.
2236

Ex.

f <f> (x) dx = f< (<r) dx = i- f <j> (x) dx.

J a
Jo
JiJ-a
m
*

j r cos 3 da: = j 2 cosasdas = - J 2 ^cosxdx.

If <j,(x) = <j>( x), that is, if $(x) be an odd function

for all values of x between 0 and a.
2238
Ex.

f i> (<v) dx = f"<f> (x) dx

J-o
Jo
j wBiaxdx = ysinxdx
J~s
Jo

and

and

f <f> (x) dx = 0.
J-o

j2 sina;cZa5 = 0.
J-^"

Given a<c<6, and that = c makes <t>(x) infinite, the

value of f4$(x)dx may be investigated by putting p = 0,
Jo
after integrating, in the formula
2240
(b<i>(x)(Lv = (e~'l<l>(x)dx+(b <f>(x)dx.
Jo
%)a
J c+n
If the function <f> (a?) changes sign on becoming infinite, this
expression, when ft is an indefinitely small quantity, is called
the principal value of the integral.
2 z

354

INTEGRAL CALCULUS.

which is the principal value. If, however, ^ he made to vanish, the expres
sion takes the indeterminate form oo oo .

2241

Given a <c< b, the integral I

,1 will always be

J a \3S

C)

finite in value while n is less than unity.

Proof.Let /* in (2240) he taken so near to c in vnlue that \p (x) shall
remain finite and of the same sign for all values of x comprised between
c/i. Then the part of the integral in which the fraction becomes infinite,
I***1 ilt
and which is omitted in (2240), will be equal to
-, multiplied by
a constant whose value lies between the greatest and least values of 4> (x)
which occur between \p(cft) and \p(c-rfi). By integration it appears that
the last integral is finite in value when n is < 1.

2242

f/(*) dx = (b-a)f{a + 6(b-a)},

a. a

where 6 lies between 0 and 1 in value.

The equation expresses the fact that the area in (Fig. 1901), bounded by
the curve y =/(.?), the ordinates /(o), f{b), and the base b a is equal to
the rectangle under b a and some ordinate lying in value between the
greatest and least which occur in passing from /(a) to f(b).

If ip (x) does not change sign while x varies from x = a to

x= b,
2243 (/(*) * (*) dx =/{+ 0 (b-a)} fV (*) dx.
2244

If </>(#> ) is a symmetrical function of x and ,

rt(..i)*..r#(.,i)*t.
Jo

x) X

Jo

.*'/ X

Proof. Separate the integral into two parts by the formula

= I +I ,
. ...
1 ..,,,..
,
Jo
Jo Jl
and substitute in the last integral.
x

METHODS OF EVALUATING DEFINITE INTEGRALS.

2245 Rule I.Substitute a new variable, and adjust the
limits accordingly.
For examples, see numbers 2291, 2308, 2342, 2345, 2416, 2425, 2457,
2500, 2605, Ac.

METHODS OF EVALUATING DEFINITE INTEGRALS.

355

2246 Rule II.Integrate by Parts (1910), so as to intro

duce a known definite integral.
For examples, see numbers 2283, 2290, 2430, 2453, 2465, 2484-5,
2608-13, 2623, 2625, &c.

2247 Rule III.Differentiate or integrate with respect to

some quantity other than the variable concerned; if a known
integral is thus obtained, evaluate it, and then reverse the
operation of differentiation or integration before performed
with respect to the secondary variable.
For camples, see numbers 2346-7, 2364, 2391, 2417, 2421-4, 2426, 2428,
2497-8, 2502-4, 2571, 2575-6, 2591, 2604, 2614, 2617-8, 2632, &c.

2248 Rule IV.Substitute imaginary values for constants,

and thus transform the expression into one capable of inte
gration.
For examples, see numbers 2490, 2494, 2577, 2594, 2598, 2603, 2606,
2615, 2641-2.

2249 Rule V.Expand the function, if possible, in a finite

or converging series, and integrate the separate terms.
For examples, see numbers 2395-7, 2402-3, 2418-9, 2479, 2506, 2571,
2593, 2598, 2614, 2620, 2625, 2629, 2630-2, 2639.

2250 Rule VI.Decompose the integral into a number of

partial integrals, and change all these by some substitution
into integrals having the same limits.
By summing the
resulting series, a new integral is obtained which may be a
known one.
For examples, see numbers 2341, 2356-61, 2572, 2638.

2251 Rule VII. Separate the function to be Integra ed

into two factors, and replace one of them by its value in the
form of a definite integral taken between constant limits with
respect to some new variable. The double integral so obtained
may frequently be evaluated by changing the order of integra
tion as explained in (2261).
For examples, see numbers 2507, 2510, 2573, 2619.

2252 Rule VIII.Multiply a known definite integral which

is discontinuous between certain values of a constant which it

INTEGRAL CALCULUS.

356

contains, by some function of that constant, such that the

integral of the product with respect to the constant is known.
A new definite integral may thus be obtained.
For examples, see numbers 2518, 2522.
Particular artifices not included in the foregoing rules are employed in
2293, 2305, 2310, 2314-5, 2317, 2367-9, 2404-15, 2422, 2429, 2456," 2495,
2514, 2518, 2585, 2600, 2626, 2635, 2637.
Additional formula? for integration will be found at 2700, et seq.

DIFFERENTIATION UNDER THE SIGN OF

INTEGRATION.

f4
Let u =

f(x) dx, where a, b, and f(x) are independent

of each other ; then

2253

^=/(6)

^d

*i = -/().

Proof.Let
u = </> (6) <p (a).
Therefore ub = <(>' (ft) = /(&)
and
ua = p'(<i) = /(<*).
Let u = If4 / (x, c) dx.

Then, when a and b are inde-

pendent of c,
2255

c= f {/(*> c)} dx

Proof.
= I

and

unc = f [f(x, c)} ^dx.

= | j f(x, c + h) <fo j" f (*, c)

| -r- h

dx (since ft is constant relatively to x) = I

\

v
dc

dx.

But if a and & also are functions of c,

Proof.The complete derivative of u with respect to c will now be

ve+u,lbe+uaac. But ub=f (ft, c) and ua f (a, c), by (2253-4).

APPROXIMATE INTEGRATION.

357

INTEGRATION BY DIFFERENTIATING UNDER THE SIGN OF

INTEGRATION.

2258 Ex.1. [artrdx=[(f)nadx = d\<f*dx (2256)

= dna(ea*a-1) = eax(x + daya-\
by (1464), a and x being transposed.
2259

Ex- 2

x"eax sin lxdx = dn*\eax sin bxdx.

The last integral is given in (1999), pntting n = 1.

2260

Ex. 3-

INTEGRATION UNDER THE SIGN OF INTEGRATION.

When the limits are constant,

2261

/(*, y) dxdy =

f(x,y)dydx.

That is, the order of integration may be changed.

But an exception to this rule occurs when, at any stage
of the integration, an infinite value is produced. The double
integrals above will not then have the same value.

APPROXIMATE INTEGRATION.
BERNOULLI'S SERIES.

2262 /(#) dx = af(a) - *-f{a) + j^/'to - ...

Pboof.Integrate successively by Parts, Jd., \xdx, &c.

f(x) into/(js) in (1510).
*

Or change

2263

/(*)& = (6-)/(a) + fc|l!/'(a) + ^=^/ + &c.

358

INTEGRAL CALCULUS.

Proof. Put/(a) for y (a) in the expansion of the right side of equation
(1902), by Taylor's theorem (1500) ; viz.,
f/()*> = *(&)-t(a) = (&-a)*'(a)+&=ip-V(a)+Ao.

The following is a nearer approximation :

Let (b a) = nh, where n is an integer ; then
2264 /(*) dx = h{if(b)+f(a)+f(a+h)+...+f(b-h)}

-T^{/v(6)-/v()}+&c
Proof.Expand (e x l)-7-(e T 1) by ordinary division, and also
by (1539), and operate upon f(x) with each result; thus, after multiplying
by h, we obtain, by (1520),

h\f(x)+f(x + h)+f(x + 2h) + . ..+f(x+^-Lk)}

which expression, by changing a: into a and + n7i into 6, is equivalent to

fx+nA

f(x) dx.

2265
J. /WtflC

T2~ +

Proof. Assume x = ce*x.

1

1.2.3

+1.2.3.4

+<^C-

Then x is equal to the coefficient of

-).

Thus

2hc* , 8W , 4W ,
1.2
1.2.3 1.2.3.4,
Substitute dx for *, and therefore dxe * for c in this equation, and operate
with it upon \ <j> (x + h)dx, employing (1520). Finally, write f(x) for <{>'(%),
and a for x.
* = C+-^r + 7~^ +

A more general result, obtained in the same way, is

2266 Jo f(x)dx = nhf(a-h)+n(n+2)f(a-2h)
+ (+3)2|-^/"(a-3A)+&c.

THE INTEGRALS B (I, m) AND T(h).

359

THE INTEGRALS B(Z, m) AND T{n).

EULER'S FIRST INTEGRAL B (I, m).
The three principal forms are
2280

I.

B(l,m)= (\vl-1(lx)m-1dx= B (m,Z). [By (2233)

Jo

2281

II.

B(Z,m)=( -JL-L^d*. [By substituting-?- in I.

Jo
L~x

2282

HI.

B(Z,m) = \ -dx. [By substituting

in I.
Jo (l+O-'J
*

When Z and m are positive, and I is an integer,

2283

B(Z,m) = 1

"

If m be the integer, interchange I and m.

m are integers, the forms are convertible.

If both Z and

Proof.Integrate (2280) by parts, thus,

f B*-l(l-)"",da = f ^-'(l-a)-.
Jo
W Jo
Repeat this step successively.
EULER'S SECOND INTEGRAL JT(ft).
ft being a real and positive quantity,
2284

r () = Ce-'ar-^dx =

(log jY~*r.

The second form being obtained by substituting e'x in the first.

2286

r(i) = i,

r(2) = i.

2288

r(n + l) = r(n) =n(n-l) ... (n-r) r(n-r).

2290

r (n + 1) = |_n_, when ft is an integer.

Proof.By Parts, f e-'x^dx = 1 +

f e-'ajVas.
Jo
we*Jo
ft Jo
The fraction becomes zero at each limit, as appears by (1580), diflerenliofing
the numerator and denominator, each r times, and taking r>n.

INTEGRAL CALCULUS.

3G0

2291

^e-^-'dx = 11^ = (V"1 (logi)"~\ir.

Proof.Substitute fas in the first integral, and so reduce it to the form

(2284). In the second integral, substitute log 2, reduciner it to the former.
When n is an integer, (2291) may be obtained by differentiating n 1
times for k the equation J e'kxdx =
Jo
*
"When fi ia an indefinitely great integer,
2293

r (n) =

/
W(+1) ...
+
1
1
Proof. log = lim. /j (1 st") (1583). Give it this value in (2285),
x
1_
and then substitute y =
; thus, in the limit,
r() =M-,jl(l-ir^),-,d = r"ry,'-,(l-y)"-,rfy. Then, by (2283),
changing /1 finally into

+ 1 in the fraction.

logr(l+) IN A CONVERGING SERIES.

2294

Let n be <1, fi an indefinitely great integer, and

+

then

2295

logr(l+n).
= (logri-S1)+^>8-i^>s+iSX-iS6ns+&c.

2296

= ilog-^ + aog/x-S,) n-Sin*-\Ssn*-&c.

+ (i-sO + "'(i-s)+&o
o
2298

= *log^- -|logl^ +-422/843n

0 sin nir
1 n

- 06735305- 00738555- 0011927n7- 0002231n9-&c.

PHOOF.-By (2293), r(l + n) = (m+1^ ^
since

= 1, when u = 00 .

Whence

iogr (1 +) = hZ/.-i (l +u)-i (1+ 1 ) -z (1+ -J)

-1 (1+ -).

TEE INTEGRALS B(l,m) AND T(n).

361

Developing the logarithms by (155), the series (2295) ia obtained. The next
series is deduced from this by substituting
+ iStn* + yS6?t6 + <S8W + &c. = log 7iv log sin wr,
a result obtained from (815) by putting d = mr and expanding the logarithms
by (156).
The series (2297) is deduced from the preceding by adding the expression
0 = -ilog^+W+^- + ^-+<fec., from (157).

2305

B(i,m) = miM.
r(/+m)

Proof.Perform the integrations in the double integral

first for x, by formula (2291), and then for ?/, by (2281), and the result is
B (1, m) r (Z + ih). Again perform the integration, first for y, by (2291), and
the result is T(l) T(w), by (2284).
Note.The double integral may be written by the following rule :
Write xy for x in V (I), and multiply by the factors of r(m + l). We
thus obtain
j j e-x" (xij)'-1 x'e-xxmdxdy,
which is equivalent to the integral in question.
2306

B (/, m) B (Z+to, n) = B (to, n) B (m+n, I)

= B(n, I) B{n+l,m)

2307
2308

=rfflr(TO)r(w)t
r(/+TO+n)
^"^{a-xY^dx = alM B (I, to).
Jo
"

[By(2305)_
[Substitute K
a

If p and q are positive integers, p<.q, and if m =

^2
2309

f-^Liir^s *
Jo 1-f.r9
2grsin
sin toit

IT
of9
2(/tanTO7r
2310
X
Proof. (i.) In (2023) put l = 2p + 1, n=2q, and take the value of the
integral between the limits <x>. The first term becomes logl =0; the
second gives the series
] Bin + sin - + ...+8in v ?
' \ = 2(7
2o
2o
J
osir
q (.L
2q
2q "'
2q
qsinmir
by (800). The integral required is one-half of this result, by (2237).
(ii.) (2310) is deduced in a similar manner from (2021).
3 A

INTEGRAL CALCULUS.

362

2311

<Lv = -^,
Jo l+.r
8iUHi7r

f_ dxr=-^,
Jo 1 x
xnumv

where m has any value between 0 and 1.

Pkoof. By substituting x1" in (2309-10).

Also, since m =

, by

taking the integers /> and q large enouerh. tlie fraction may, in the limit, be
made equal to any quantity whatever lying between 0 and 1 in value.
2313

r(m) r(l-) =

. n ,
m being < 1.
SHDHTT
Proof.Put l+m = 1 in the two values of B (I, m) (2232) and (2305) ;

thus,

2314

r (m) r (1-m) =

Cos.

f- da = -r-^,
Jo 1
sm i/t.T

by (2311).

r(i) = v/.

The following is an independent proof:

r (J) = f e *z~* dx = 2 [ e-t'dy = 2 f e'^dz.
Jo
Jo
Jo
Now form the product of the last two integrals, and change the variables to
r, 0 by the equations
y = r sin 0 \ from
(1609) dydz _ <| (^IdrdO = rdrd0. Hence
a = rcosfl )
> j \
j
d (rO)
{r(i)}' = 4f[ e"(" <*')dyd* = 4:[ i ''e^rdrdd = n;
Jo Jo
Jo Jo
the limits for r and 0 being obtained from
r,= 2/' + 2,) tan 6 = 0-.
z
2315

r(I)r(A)r(A)...r(^) = ^.

Proof.Mnltiplv the left side by the same factors in reversed order, and
apply (2;J13) thus

. 7t . 2r
.
Bin siu . . . sin
n
n

2316

"-T-,

by (814).

, ^rWr(I+I)..,(I+!=!)=^.
r

TJIE INTEGRALS B(Z,m) AND r(n).

363

Proof. Call the expression on the left <j>(x). Change x to x+r, where
r is any integer, ami change each Gamma function by the formula
T(x+r) = as(rl r(as) (2288). The result after reduction is <t>(x). Hence
f(jr) = <p(x + r), however great r may be. Therefore <f>(x) is independent of
x. Bnt, when x = , <f>(x) takes the value in question by (2315). There
fore <j> (x) always has that value.
The formula may also be obtained by means of (2294).

NUMERICAL CALCULATION OP T(x).

2317 All values of F(x) may be found in terms of values
lying between r(0) and
When x is >1, formula (2289) reduces r(x) to the value
in which a; is < 1 ; and when x lies between 1 and \, formula
(2318) reduces the function to the value in which x lies
between 0 and \.
Values of T(x), when x lies between 0 and 1, can also be
made to depend upon values in which x lies between ^ and -j,
by the formulas,
, v
2318 TO = 21-%AjyMT,
i

I = ^
rf^Ar2-cos^ r(l=)

Proof.To obtain (2318), make n = 2 in (2316). To obtain (2319),

pat m = i(l+;e) in (2313), and change x into \e iu (2318), and then
eliminate T (^r")-

Methods of employing the formulae

2320 (i ) When x lies between f and 1, reduce T(x) to
r(l-*), by (2313).
2321 (ii.) When x lies between | and f , reduce by (2319),
the limits on the right of which will then be and ^.
2322 (iii.) When x lies between 0 and
reduce by (2318) ;
F(+*) will then involve the limits ^ and f, and will be
reducible by case (ii.)
If 2x is <, reduce T(2x) by (2318), writing 2x for x. If
this gives 4x <
reduce again by the same formula, writing
4*E for x, and so on.

INTEGRAL CALCULUS.

364

2323 The figure exhibits the

curve whose equation in rect
angular coordinates is y=T(x).
Let the unit abscissa be OA =
AB=\.
Then the ordinates
AD, BG are also each = 1 by
(2286-7).
The minimum value of T(x)
is approximatelyO'8556032 cor
responding to x = 1 "461 6321.
The values of log T (x) in
the table at page 30 correspond
to ordinates taken between
AD = Y{\) and BG = T(2).

INTEGRATION OF ALGEBRAIC FORMS.

2341
Proof.Add together (2281) and (2282). Separate the resulting in1 in the last part.
tegral into f + r
I , and substitute
Jo Jl
X
i ^(l-.r)"-1 J _
B (I, m)
)o
/ , \i+m dx = am(l+ay
m/i , [Substitute
2342
x+a '

2343

2344

J1*'-1 (1 *n)'B_1^ = -~~B{^' m)-

f1

dr

[Substitute x\

Lfif/ m) [Sabs.

a; 4-6

The integral is also equivalent to

sm2'-10cos2'M-10
, I 2
Sill"
^ <Z0, and similarly in other cases.
'Jo (a sin2 0+6 cos2 0)

[Substitute

2345
(a + ftci-)""1

INTEGRATION OF ALGEBRAIC FORMS.

2346

r (a+bx)'"+a
Jo

n{m)

365

1
ambn'

Proof.Differentiate (2345) m 1 times for a.

2347

f
Jo

Tli

(2255)

= C (m, n) ^ .
sin mir

m< 1.

Proof.Substitute x=. ay in (2311), and then differentiate n times for a.

no,Q
2348

r00*'""1^
1 -f-
Jo 1+*

IT
WMT
cosec ,
n
n

P".J?m_1C?J?
IT
.WITT
1
j- = cot ,
Jo 1 or
n
n

where m and n are any positive quantities, and m is < n.

i_
Proof.Change m into in (2311-2), and then substitute as".

"When n is positive and greater than unity,

2350

1
; = cosec-.
\
- = - cot-.
Jo l+<r
n
n
Jo 1 <v
n
n
Proof.Substitute xm in (2311-2) and change m into .
71

2352

f' ^
Jo \flx*

Proof.Substitute

* cosec*

r_^-_. = ^Cot.
Jo

x
in (2350), and
,T
in (2351).
yi+* v
"
Vl-z"

When m lies between 0 and 1,

2354

Joiq-p = --secT-,

Jo_ = -_tan_.

ProofMake n = 2 and write ot + 1 for m in (2348-9).

l+a?

Bin mir

Jo

1 x

tanm7r

where m lies between 0 and 1.

Proof.Separate (2311-2) each into two integrals by the formula
j = j + jy, and substitute x'1 in the last integral.
Otherwise, in (2601) substitute e~nI, and change a into ira Jjt.

INTEGRAL CALCULUS.

366

iutr
2~*
Proof.From (2354-5) by the method of (2356).

2360

1
Jo

- do- = - cosec .
1+iT*
n
n

2361

r^c""1** =-<*-
Jo
1a.'"

n
Proof.In the same way, from (2348-9).

2362

j
Jo

.,
1+or"

Jo

1o^"

2363

rfj = .5- sec -5-.

2
2/t
dx =
2;i

tani^.
2

Proof.In (2601) substitute e~ ln and put a =

In (2595) substi-

tute e and put a = .

n
/
j
^(n-l)
I , , " , = -Itt r. ^ o w being an integer.
Jo (o^+o4)"
21"" ' 2a2"-1
5

Phoof. By successive reduction by (20G2), or by differentiating

t^ a = , m 1 times with respect to a1.
(2255)

2364

2365

^ ldv
vcosecmn
Jo(l + fio')(l-o-)"~ (1+6)" '

2367

[Snbg.lte
L~""'

(23n)

f(-iS>=k'-

Proof.The value when a = 1 is logrc.

The difference, when the value

which, by substituting ,r" in the second integral, is seen to be zero.

I*' (a;) being any integral polynomial,
2368

^ Gr) jfa _ ^_ wnere a is equal to the constant

term in the product of jP() and the expansion of (1

LOGARITHMIC AND EXPONENTIAL FORMS.

367

Proof.By successive reduction by (2053), we know that

where <p (x) is some integral polynomial and .rl is a constant. Therefore the
integral in question = Air. To determine A write the last equation thus,
(-*)> J* (i-*)

Expand each binomial ; perform the integrations and equate the coefficients
of the two logarithmic terms in the result.
F{x) being an integral polynomial of a degree less than n,
2369

f'5Lto-^^(.)k,lzf|.
Ja
(xc)n
n 1 acn 1 (.
ac)

PBO0F.- J.(-c)"
f ZSSL dx = I w-1
1
f }Uc"-'
2 \s-c/ |J
J.

(2255)
v
'

But -F () = f (x, c) + F
(c) where / is of a dimension lower than n1
x c
xc
(421), and therefore dln.\)ef (*, c) = 0. Hence the integral on the right
dn-l1 f*
r Ffc) ,
dnl ( x./ m
<fc"

INTEGRATION OF LOGARITHMIC
EXPONENTIAL FORMS.

2391

jVta,.*-^

AND

|t'^ = lg(P+l).

Proof. These are cases of (2292). Otherwise; to obtain the first

integral differentiate, and to obtain tho second integrate, the equation
| x" dx
witn respect to p (2255 and 22G1).

2393

jV(Iog*)-*r = (-1)- ffiffi-

Proof.See (2292). Otherwise, when n is either a positive or negative

integer, the valne may be obtained, as in (2391), by performing the differ
entiation or integration there described, n times successively, and employing
formate (21GC), and (21G3) in the case of integration.

2394

f1
dx = log[ .
Jo log*
q-\-\

[By (2392).

INTEGBAL CALCULUS.

3G8

2395

pi2^rf, = -^^- = 4"^.

Jo
1 X
II
Jo (f 1

2397

f1 (H*)*'1 dw = Jo
l+o?

= - f igl dr.
Jo e'+l

2n

Proof.Expand by dividing by lfai, and integrate by (2393) ; thus

^'=-^(1-^-^*4
The first series is summed in (1545). The difference of the two series mul
tiplied by 2**",ia equal to the first; this gives the value of the second series.
2399

Proof.As in (2395-7), making n = 1.

The series (2399) may also be summed by equating the coefficients of 0J
in (704) and (815).

Proof.The integral is half the sum of those in (2399, 2400).

Jo

if

12

Jo

Proof.Expand the logarithms by (155) and (157) and integrate the

terms. The series in (2400-1) are reproduced.

2404

Let

f 1-2&U) dx = <f>(a),
Jo
<r

a being < 1.

Substitute 1 x = y ; therefore, writing I for log,

9^ ;
J.-ai-y
J.i-y J. i-y
The second integral by (2399), and the third by Parts, make the right side
= - 4- + I (1 - <0 - f" Z ^ ~?/) <ty- Therefore

Jo
y

LOGARITHMIC AND EXPONENTIAL FORMS.

2405

869

^(a)+^(l-a) = logo log (l-a)-K-

Ua=i,

l2^Zl^ = i(log2)^^.

Again,

p (*) = f
Jo

"Ml- J

J0

dx, .: (by 2253) 9' (a) - *

'

l-<(a-l)

J0

1-x 1-*

J,

...
*

1-z a

= *{U1 -)}'-(*) '

Put - for x ; then
2406

k-*)+4,(-*J) = t{kg(l+*)}f-

Also,

Hence

*(*)=-[*

+ ^ +-^+&o.)

* () + *(-*) = - i (* + + |- + Ac.) = f (*').

Eliminate <p (x) by (2406) ; thus

2407
Let

= i{log(l+.r)}a.
r = x1, and therefore a! =

.: by (2407),
or

11-.

+ v/5 = ft say,

|* (/?)-* O) = |-{*

J-*(l-/3)-*(/J) = |W3);

and by (2405)

[v /? = 1 -ft and 1 + /3 =1 ;

$(\-ft) + <p (ft) = 2 (Ift)1-

* (ft) = (Z/3)'- 1 and 9 (1 -/J) = QftY- 1 * that is,

2410 Let a be > 1, then <f> (a) contains imaginary elements, but its
value is determinate. We have
Jo

Ji

35

6
3 B

Jl

INTEGRAL CALCULUS.

370

the integration by 2399, and I (-1) = rri by (2214).

The last integral

Substitute = y in the last integral, and it becomes

a
Hence, when a is > 1,
2411

<f>(a)+<f> (1) = -+mloga+(\ogay.

If a = 2, this result becomes, by employing (2405),

2412

2413

f2 log(1-tr) d* = - ^ + log2.
Jo
a?
4
Let

+(*) = r^- log^dr.

Therefore

f () =^ * (f)

g=J) = ^

therefore

+ * (1=^) = f (z 1S i. +

2414

+(*)+Hrr) = *logir log r=J

**) dx, therefore

The constant vanishes, by (2403) and (2401), putting 0 = 1.

Let * = f+'
2415

therefore = >/2 1 i then, by (2414),

f-logiJ^ = -Hlog(y2-l)}'

2416
Proof.Substitute p = tan"1* ; then, by (2233),

2417 By differentiating or integrating the equations (2341)

to (2363) with respect to the index m, the integrals of func

LOGARITHMIC AND EXPONENTIAL FORMS.

371

tions involving logx are produced; thus, from (2356), by

integrating for m between the limits \ and m, we have
2418

| \fl\r~
Jo (1+^) logj?

= loS tan^2

(1943)

Otherwise, this result may be arrived at by forming the expansion of the

fraction in powers of x, and integrating the terms by (2392) ; the reduction
is then effected by 815-6.
In a similar manner, we obtain the more general formula
2419

2420

f1 (x""1 a?""1) dx _ 10 n+P m+2p n+3p

Jo (l-\-ocp)logjc
n m+p n+2p m+Hp'"

l(T+^)lo^^ = 10gtan2^-

Proof.Integrate (23C0) for m from 2n to m.

2421

Jo

(k^3

= (p+Dlog(/)+l)-(r+l)log(r+l) + (r-jp){l+log(9+l)}
Proof.Integrate (2394) for^> between the limits r and j).

J- (,-,),-+(g)f+(,-?^

= log{(i)+l)(l'+1H''-r)(g+l)(,+1)(r-3')(r+l)(r+1)(j'-*)}.
Proof.Write (2421) symmetrically for r, p ; p, q and q, r. Multiply
the three equations, respectively, by q, r, p, and add, reducing the result by
(2394).
2*23

f!2^V)<fc = ^,g(l+o4).

Proof. Differentiate for a, and resolve into two fractions.

integration for x, and integrate finally with respect to a.

Effect the

2424

Proof.In (2423) put a = 1, and substitute = y ; multiply up by b,

and integrate for b between limits 0 and , and in_the result substitute by.

372

INTEGRAL CALCULUS.

2425

2426

f <?"**' da> i
Jo
Z ' (;

|V*-V-db = L^^D k"^ y*.

Proof.Substitute ke'.
ii times for h.
CCD

Otherwise, differentiate the preceding equation

,j
e

,
2428

[Substitute

~c

rftr = loga-log6.

r(e'bx e
(cb)e-"\,
, . ,
bba<
Jo ( Fn = c-h+Xo^-

Proof.Making n = 1 in (2291), | e~axdx = . Integrate this for a

Jo
a
between limits b and a to obtain (2427) ; and integrate that equation for b
between limits 6 and c to obtain (2428).

Jo \

or

Proof.Make c = a in (2428).
Otherwise.Integrating the first term by Parts, the whole reduces to
L
x
Jo
J0
*
The indeterminate fraction is evaluated by (1580) and the integral by (2427).

2430

f
Jo (

a*2

2d-

3 dx
)

= | {a2+362-46+2&2(loga-log6)}.
Proof. By two successive integrations by Parts, z~3dx, &c,
(" e- '-e'" ,
e-ht-e-" , ae~" -be'" . 1 f Pe ^-Je j,
5
=
iTi
+
]
1- 7T
dx],
x*
2e*
2e
2 J,
x
Also

I dx = e
I dx.
Jf *
t
it x
Substitute these values, and make e r= 0. The vanishing fractions are found
by (1580), and the one resulting integral is that in (2427).
In a similar manner the value of the subjoined integral may be found.

INTEGRATION OF CIRCULAR FORMS.

2431
r^e-b*eax
Jo t

(ab)e-ax

(a 6)'e

a*

1.2.**

373

( b)se-ax)
1.2.3^ J

INTEGRATION OF CIRCULAR FORMS.

Notation.Let a signify the continued product of n
factors in arithmetical progression, the first of which is a, and
the common difference of which is b, so that
2451
a{bn) = a(a + b)(a + 2b) ... {a+(n-l)b}.
Similarly, let
2452
al:l = a(a-b)(a-2b) ... {a-{n-l)b}.
These may be read, respectively, "a to n factors, differ
ence b"; " a to n factors, difference minus b."
2453

I smnwdx = --I sin"-2crdr.

Jo
n Jo

Proof. By (2048) ; applying Rule VI., we have, by division,

j^sin'ajdj; = j" sin""1 a; das j"sin""2ai cos'asd*,
and by Parts,

f sin""' x cos* x dx = X C8 X + ^r I Bm'xdx.

J
n 1
n 1J

f*"sin"" *as cos's

. cfc = - 1 -If1" siunxdx.
Jo
n-ljo
The snbstitntion of this value in the first equation produces the formula.
Therefore

If n be an integer, with the notation of (2451),

f**
2'B>
M'
l(n) ir
2454
sin*+1tfdr = -fo and
sin2" a d.r = ^ |-.
JO
JO
2^
*
Proof.By repeated application of formula (24o3).

TFa/Zi's's Formula.If m be any positive integer, we have

2456

2m [^j > |- > (2m- 1) j^j .

374

INTEGRAL CALCULUS.

And since the ratio of these limits to each other constantly

approaches unity as m increases, the value of either of them
when m is infinite is \tt.
Ex.With m 4,
lies in magnitude between
2'. 4'. 6s. 8
,
2'.4Vr,!.7
and
1 .3*. 5'. 7*
1 .3'. 5*. 7s'
Proof.Pnt 2m = n, then
f**
f*"
n 1 fir
sin^'arfz,
I sin" xdx,
and
I sin*"1 xdx
Jo
Jo
* Jo
are in descending order of magnitude ; the first and second because sin a; is
< 1 ; the second and third by (2453) ; then substitute the factorial values by
(2454-5).

2457

2458

f*T
IT
1 tan4"1-1 6d6 = -rr.
Jo
2 sin mw

[Subs. x = tan*f in (2311).

[By (2234).

f am* xdx = 2 f sin\r<r.

Jo
Jo

2459

1 Bin"a? cos* xdx = 2 1 sinn^cosp^l,rfa,, or zero,

Jo
Jo
according as
is an even or odd integer.
[By (2234).

2461

| sin*,r cosp,rrflf = | sinptr cos".re?.r.

Jo
Jo

2462

f^sin"* cosp .rdr =

[By

[Subs.sin1 z(2280).

Let either of the integers n and p, in (2461), be odd, ai*l

the other either odd or even ; thus, let n be odd and
= 2m +- 1 , then
2463

( *%in*"+1* cos!' xdx =

_?\L+iy

(2i)

Proof.Transposing the indices by (2401), we have, by Parts (206/),

P'sin'a; cos""lxdx =
^'biu'-'x cos""1 xdx.
Jo
P+l)0
By repeating the reduction, the integral finally arrived at is
2464

I ' sin"
Jo

x cos x <lx =

?
.
P + Zm + 1

INTEGRATION OF CIRCULAR FORMS.

375

If both the indices are even, then

2465
] sin2M.rcos^*^=^i|r ..

(2451)
fin-

Proof. Reduce by Parts as before. The final integral is

8m,m*lpzdx,

the value of which is given at (2-155).

2466 Should either of the indices be a negative integer, the

value of the integral is infinite, as the foregoing reduction
shows, for the factor zero will then occur somewhere in the
denominator.
2467

\ smnx sinpxdx = \ costuv cosp.vd,v = 0,

(i

Jo

when n and p are unequal integers.

2469

I sinn.r cos p<r da? =

3 or zero,
Jo
n'p2
according as the difference of the integers n and p is odd or
even.
[By (1973-5).
2470

I sin2 nxdx = \ cos2 nxdx = jir,

Jo

Jo

when n is an integer.
Proof. Express in terms of cos 2nx, and then integrate.

2472

f \maxdx = ( *cosnxdx = f'(l x2)~dx.

Jo

Jo

Jo

The following four integrals (2473-9) all vanish for in

tegral values of n and p excepting in the cases here specified.
2474

j/in".* sin nxdoc = (-iycht=ty ~,

when p and n are both odd, and n is not greater than p.

2475
v

But if p be even, and n odd, the value is

; 2p"2 ("-/

rii-(p-2)'i~t tf-ip-Ay
t-rfCMpn

INTEGRAL CALCULUS.

376

2476

^sin"x cosnxdx = (-1)" c(p,_J?) JL,

when p and n are both even, and n is not greater than p.

2477
,
v

But if p be ocM and n even, the value is

tf? 1 f ;>
' I^V-n*
w(

2478

C(p, l)(p-2)
(p-2)4-n4

1)griiiI^zM].

['cos^cosn^tir = c(p,=^)i,

when p and n are either 6oA odd or

greater than p.

2479

C(p,2)(p-4)
(p-4)

eve, and n is not

| cosp a: sin nxdx, when p ~ n is odd, takes the value

(
1
C(p,l)
,
C(p,2)
2*-* ? *-p* + (j,_2)" "T" n2-(p-4)a +

)
S'

the last term within the brackets being

*i

wnenn 1*even,
^s ocW
or and

^(P'^fr)
;
when
andpn is
odd.
even

Phoop.(For 2474 to 2479.)Expand by (772-4), and apply (2467-2470)

to the separate terms.
Cokollahies.n being any integer,
2480

2482

j cos" .r cos no? rfj? =

cos" a: cos nxdx =

f sin2"^cos2n^dr = (

Pain*"*1.* sin(2+l) xdx = (-1)" * .

Jo
2""

2484

1 cosp.r cos nxdx =

Jo

p n

I cosp *x cosnxdx.
Jt

INTEGRATION OF CIRCULAR FORMS.

377

2485
CcospJcsmnxdx =

f cosp"2<r sinnarfan* Jo

r*
p' ir

Proof. (For either formula) By Parts, J cos x (fx; and the new integral
of highest dimensions in cos x, by Parts, J cos"'^ siaxdx.

2486 \ cos" 2cr cos nxdx = 0. 1 cos"

Jo
Jo
Proof.Make y = in (2484-5).

smnxcLv =

T.
n1

When & is a positive integer,

2488

y cos"-2*.? cosn<r<r = 0,

2489

f^cos"*2*^ cosn.rrf.r = (w

^TT-

Proof.The first, by putting p = n 2, ?i 4, ... n 2k successively in

(2484) and employing (2486). The second, by putting p = 11 + 2, + 4,
... n-\-2k successively aud employing (2481).

When k is not an integer,

cosn_2*,r cosn.rrf.r = 22 "B+1 sin kirli(n 2k+l, k).
2490 Jo
Proof.In (2706) take f(a)'= a-2*, and transform by (706). The co
efficient oft vanishes by (2230), aud the limits are changed by (2237).

2491

j^'cos"* sin dx = -L (2+ I + 1 + ... +

Proof. By successive reduction by (1970), making m = n, aud the

integral definite.
2492 When p and n are integers, one odd and the other even,
Clr

,
(1)1 (+>! f+, with w odd,
1 cos'.r cos nxax = -L
-. <
-u
Jo
(n pyf+L)
I > Wlth P evenProof.Reduce successively by (2484). The final integral, acoordiug as
p is odd or even, will be
fi
,
cosiwir (-1)*
f*
,
srinjwr (_!)*(-
I cosxeosnarrfx =
=;
or I cosna;rte =
=
.
Jo
1 n
1 "
jo
n
n
3 c

378

2493

INTEGRAL CALCULUS.

\ cosp.r cos nxdx = pi ^sln^x coap~*j,dx,

where n and p are any integers whatever such that p

is > 0.
Proof.When p n is odd, each integral vanishes, by (2478) and (2459).
When p-n is even, let it = 2fc; then, by (2488),
I cos" Ilz cos nick; =
= :1 ?
J0
2"-"'-1
1<*
2J"*') 2
= 0 + 2fe)'-"i' f*w8in.iB cos^x^, (by 2465).
li"
Jo
Bat n + 2k =p, and by (2234) the limit may be doubled. Hence the result.
2494

\ x cos"~*.r sin nxdoc = a , wrr.

Jo
2*(n 1)
Proof.Tn (2707), put k= I and /(as) = a;""2. Give e'"* its value from
(76G). The imaginary term in the result vanishes, and the limits are changed,
by (2237). Finally, write z instead of 0.
2495
l /"(cos a?) Bin*txda?= 1.3... (2n 1) I f(cosx) cos7i.rrf.r.
Jo
Jo
Proof. Let z = cosas. By (1471), we have
d(.-.,.(l-*,)"-*= (-ir'1.3 ...-(2n-l)R-^
71
Also, by integrating n tines by Parte,

(i.)

j' ^(z)(l-z)-fe = (-l)-J'

= -1.3...,(2n-i)Jl/(Oi.(5^)&. by(i.)
Then substitute z = cosa:.
Otherwise.Let /(z) =

J,z + yljZs + &c. = S^z",

.-./(*) =Sp(j-l) ... o,-h+i)^-,

'

2496

j" /(cosr) cosn^ffa = 2Jp j" cos's eoanxdx

f -..
^ Bin
. . a* = 5^.
Jo
cos" .r+o
2o

(1982)

INTEGRATION OF CIRCULAR FORMS.

379

2497

f*
cos2,rrfr
T~~>
5
r~n-" = i77- r Differentiate (249G) for a.
o (<rcos .r-fo- sin .r)2
4a b

2498

1 T-5
3 ,
. ;rr. = -t-jt. [Differentiate (249G) for I.
Jo (a- cos^r+b1 sm- J'Y
4ub*

Jo (a2cos'!*'+62siu2^)''!

4abi,aA')'
[Add togetlier ('J497-8)

okaa
*

f1*
_ 7T /3
2
3\
Jo (a2 cos3 .r+62 sin2 ,v)a ~ UiubW + a* / ~*~ b* /

OCQ1

f**
rff
TT (O
3
3
5\
Jo (a2 cos2 +6* sin2 .r)4 - 32a6 ' a,i+ b1 + a'fr* + 67'

(2500) and (2501) are obtained by repeating upon (2499) tbe operations
by which that integral was ittelf obtained from (2490).

2502

fF7^d-' = ^^-

Proof.Denote the integral by u.

Proof. Differentiate for a.

then integrate for a.
2504

Integrate for x by partial fractions, and

$ = * ,g [(, + )*(,+ )1}.

Proof.From (2503) we obtain

|" tan-'y , _ ^ Iog(l -fa)
Joa!(a! + KJ) '
2
a2
"
Integrate for a between limits

and oo , and in the result substitute bx.

b

INTEGRAL CALCULUS.

380

2505

f tan"'aT'-tnu"lfer civ = ^ log"

Jo

a'

Proof.Applying (2700),

2
(0) here vanishes.

6
Also, by Parts, we have

f*tan-'(fa)jJ= * fid*
J
x
2 J
a
a
since Zuj is infinite and therefore tan"1 (hx) in every element of the
a
integral. Hence the required value is
_*
t [" dx
x . a

2506

P
Jo l+cos"*'

= *
4

Proof.(i.) Substitute ir x, and tho integral is reproduced, and is

thus shown to be
=
I + cos'y
~ dy9 =
2 J0l

2 v(tan cos t tan

cosO)'4
= .

(ii.) Otherwise, expand by dividing by the denominator, and integrate

each term of the result by Parts. Employing (2478) we obtain the series
r(l-i+I-i+&o....) = .

[By (2945).

Proof.By the method of (2251), putting

V*

V* Jo

the integral becomes

[ [ e-^cosx dxdy= \- f \ e'** ooaxdydx

(2261)

= f fg (2610) =
. (2348)
'J, 1 + 2/
V <*
The second integral is obtained in a similar manner.

2509

1 cos y*dy =
= I siupdy.
JO
<- \ a-'O

Proof.Substitute y%, and (2507-S) arc produced.

INTEGRATION OF CIRCULAR FORMS.

881

When ft and^J are integers,

2510

f"
Jo of

d* = r-5-4
tt r F*p-le-" sin" x dz dx.
1.2 ... (p 1) Jo Jo

The integration for * in the double integral is given in

(2608-9), and the original integral is thus reduced to the
integral of a rational fraction.
Pkoof.By the method of (2251), putting
= ! Ce-"z-ldz.
>
r(P)J0

2511

2512

2513

[By (2291).
1 yK
>

dz
1^* = ^-,.

= 6rro+9) - T

[By (2510).

ggg

r^jfrS25H&, = iogZ,
Jo
A'
q

Proof.By (2700). Transforming the numerator by (673), and putting

\ (l> + 2) = a> 2 (P~ <Z) = ^> this becomes
5U

2515

rsina^sin^
Jo
*

lt
2

a+b
ab

rc08y,p"cog>M?to = ^(p-g).
/0
^

Proof.Integrate (2572) for r between the limits p and q.

If a and 6 are positive quantities,
rens
2516

sinOA' COS&<r ,
7T n
l
dx = -o or0
Jo
a?
2
according as a is > or < 6.
Proof.Change by (666), and employ (2572).
+
2518

C sin ax sin fop ,

ira
I
?
^ = T

or

irb
2'

according as a or b is the least of the two numbers.

INTEGRAL CALCULUS.

382

Proof.From (2515), exactly as in (2513).

Otherwise, as an illustration of the method in (2252), as follows.
noting the integral in (2516) by w, we have, (i.) when b is > o,

that is,

= P f "fr"* cns6* <B<b = r 5!^H!I*5d*.

2
Jo Jo
*
Jo
x

De

(2261)

(ii.) When b is < a,

If a is a positive quantity,
fttmrt
2520

f" sin8 <r cos a.r j

IT ,ci
\
a
I
3
dx = (2 a) or 0,
Jo
itr
4

according as a is or is not less than 2.

Proof. sin's cos ax = \ sin x {sin (1 +a) a+sin (1 a) *} ;
and the result then follows from (2518), the valne of the integral being in
the two cases -^- = 0 and (a 1) = -^-(2 a).

2522

~d* = l

or

-(!-)

according as a is > or < 2.

Proof.Denote the integral in (2520) by ; then, when a is > 2, the
present integral is equal to
^uda = f ^ (2-fl) da + | <U = "f
And, when a is < 2,

fi*da = (-^- (2 a)

= ^

INTEGRATION OF CIRCULAR LOGARITHMIC

AND EXPONENTIAL FORMS.

2571

r^irir = tan<
Jo
a.*
a

Proof.Differentiate for r, and integrate by (2584).

Otherwise. Expanl sinrx by (764), and integrate the terms by (2291).
Gregory's series (791) is the result.

CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS.

383

2572
Jo

Proof.(i.) By making a = 0 in (2571).

(ii.) Otherwise. By the method of (2250). First, observing that the in
tegral is independent of r, which may be proved by substituting rx, let r = 1.
Then

r~-^ = r *+r~*+r~*+**
Jo *
Jo *
Jt *
J*r *
Now, n being an integer, the general term is either
rsin x dx = faydy
substituting as = (2 1) jt-H/,

or

-1- c/j;
J^Z , by substituting x = (2n l)ir-y;
J (la-i)* *
Jo (5J 1) 2/
j "in a; ^ f gjn f \
1 1_ 1.
\ _|_ 1
&c. ^ Ju
"Jo *
Jo
Cxy sr + y 3jr-y 8jr+y 5*y
)
= | j'siny tan| dy (2913) = j'sin' %dy = |.

2573

1 Y-r dx = -^e r= \
dx.
Jo
*
Jo r-\-or

Pboof.(i.) By (2251), putting

= 2 f e-"**ydy

(2291),

the integral takes the form

2j J C08ra;e-ll,ic'''i/(ia!dy = 2 j j e^y e'** coarxdydx

(ii.) Otherwise.

= ^Trf e'^'^dy (2614) = ^'(2604).

Jo

By the method of (2252), putting u = j" sln 't3; ens 7)3 ^

it follows from (251G) that

f e-rfa = [60e-rfa+ f ^e~ada = ^-e~b.
Jo
Jo
Jo "
2
Therefore
= f f si" a* CS to e-&to = f^dx,
2
Jo Jo
*
Jo 1 + *

by (2583).

Proof.For (2575) differentiate, and for (2576) integrate equation

(2573) with respect to r.

INTEGRAL CALCULUS.

384

2577

fV".r-l8in (&r) dx =

Proof.By (2291),

F(w>

8in(tan-'l).

j^e'**x-ld* = M.

Put A; = a + i&, and a = r cos 6, 6 = r sin 5 ; thus

^"e-iTxn-idx _ (c08 K0_t-sm) Ei^),

'

by (757). Substitute on the left side for "*** from (767), and equate real
and imaginary parts. Otherwise, as in (2259).

2579

f^W^^'S}

Proof.Make a=0 in (2577).

sin
2581

'

N
=
2sin(^)
v ' cos\ 2 /

Jo

Proof.Put n = 1 m in (2579), and employ

rr(l-n) =
SIUNIT

2583

( e~" sin bxdx = -J^-nJo

o-+6*

8111 J/iT

\ c"" cos &rd.r = -5-^-5-,.

Jo
<r+&-

Proof.Make = 1 in (2577-8).
Otherwise.Directly from (1999), putting n = 1, and a for a.

2585

f^*^<F**-%$&(<f).

where n is a positive integer > 1.

Proof.In (2577), put tan"1 = fl, thus, writings for n,
a
^e-M^f,-1 sin bxdx =
cos" $ sin 0.
Jo
P
Multiply this equation by V'^db = a" tan"~'0 sec Odd.
aud integrate from 6=0 to oo , by (2579). Then the corresponding limits
in the integration for 0 will be 0 and ir.

CIRCULAR LOGARITHMTC AND EXPONENTIAL FORMS. 385

2587

P'sin*-" 6 8in (P0) dd = ~c-s(^) A-

Proof.Put n=zp-l in (2585).

2589

(V**-1 ^ j (x tan 0) dx = T(n) cos0 j n0.

Proof.In (2709), let </> (x) = cos (x tan 0) ;

.-. by
(7G5), 4
A = 1,
l,
by(7G5),
vanishing.
1

A = -^p

At = rYYl' &C'' A" A* &C'

Therefore

i- 'tan'0 + -jjtan'fl-, tan 6+ ... -

I e"1**-1 cos (a: tan 0) dx

p

The series on the left = \ (1 + t tan 9)-" + (1 t tan 0)-a, which by tho
values (770) and (768) reduces to cos a0 cos* B. Then change a into n.
Similarly, with sine in the place of cosine.
2591

r e""-e~" sinker = tan-1! -tan-1-?-.

Jo
x
bo

Proof.Integrate (2583) for a between a = a and a = (3.

/100
J e-** cosa<r sinB.r7,r,

2592

where n is any positive integer.

See (2717-20) for the values of this integral.

2593

m
1

sin mxdx =
Jo e"' e~"
2 en,+2cosa+e-",

a being <*.
Proof.The function expanded by division becomes
(f+e-"*) sin ma (e-'x + e-mi + es'rx + &c.)
Multiply in and integrate by (2583).

The result is

{(2-l)T-o},+m,+ {(2ra-l)ir + a}, + w!"

But this series is also produced by differentiating the logarithm of equation
(2953). Hence the result.
3 D

INTEGRAL CALCULUS.

386

2594

/* e"r
sin a
1 e^e-**

cos mxdx = e"+2cosa+e

'
m .
Jo

Peoof.Change m into iO in (2593), thus

r (*"* + e-ax) (e*J-e-Bf)
_
sin d
Jo
e** e - "
coa a + ous 0 '
Now change a into im and write a instead of 0.
OKQK

_ tanja

Jo eT'-e-T* *

f*sin mjrrij? _ 1 eim e~**

" Jo e"-e-" ~ 4 c^+e-*"'

Proof.Make m = 0 in (2594), and a 0 in (2593).

2597

\
-r'
.
Jo
e"e-" sn nuv djc =
2 ekm-e
u

Proof. Make a=v in (2593).

2598

f" v2n~xdv
22n 1
-^^ii^.
Jo eTX e~'x
*n

Proof.Expand sin mx on the left side of (2596) by (7G4). The right

Bide is = \i tan ($tm) by (770). Expand this by (2917), and equate the
coefficients of the same powers of to.
ntnn
2599

f eax-erax .
,
2 (em c) sin a
1 -i
r- sin m#a# = -7-rz
^k-

ocaa

eax+e-ax
Jo e^'+e-*"

2(em+e-n)cosa
e2m+2cos2a+e-:!m

Proof. To obtain (2599), put a + ^T and a

successively for a in
equation (2593), and take the difference of the results. (2600) is obtained
in the same way from (2594).

2601

! +

tfa = sec .

Proof.Make m = 0 in (2600).

2602

f" sin far)' rf.r = Tcos (cr)1 d.v =

Jo
Jo
iCy/Z

CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS. 387

Proof.By (2425)
Pnt. a = 1+t
^^-

[VB * dx = ^
Jo
2a
Substitute on the left from (766), and equate real and

imaginary parts.

2604

e-{"+$dx=^e->\

Proof.Denote the integral by u. Differentiate the equation for a, and

substitute in the resulting integral to prove that = 2m, and therex
da
fore = Ce'2*. When o = 0, we get
^e-^dx = C, .: G-\-Jic (2425).

2605

jV^V. = 0^--

Proof.Substitute

2606

and integrate by (2604).

fe-(--)--[(^+^)sin^]^
\Ar
2

taoo.* cos/2 gin tf+ 0'

*(2sin*+|).
sm\

Proof. In (2605) put le = cos0 + isin0 ; substitute from (766), and

equate real and imaginary parts.

2608

f
Jo

e-~sin"'*te = .
1.2.3 ...(2n+Jl
(oi+l)(ol+34) ... (a2+2n+l2)'

/W0
e

2609
Jo

-a* 2 J
sin2" xdoc =

1.2. 3... 2/1

a (a2+22) (a2+42) ... (a2+2n2)

2610
fc cos2-1 vdv a

a(2n + l)2
Jo
XUX - 2+(2rt+l)2+(2+2l7+l2)(2+2^T2)
+

a (27i +l) 2n(2n-\) (2,i-2)

(' + 2Hn')(a' + 2^T'j (a2 + 2n=3J)

+
"'

<tj 2n + l
(a' + S^ + V) ... (a' + l) '

IXTEORAL CALCULUS.

383

i =
e ax cosin x ax
2611

a
a 2n
-J
-(2n-l)
Jrfl,+(2) + (ai+2ns)(a2+2n-2s)

f
a 2n f2r - 1 )(2n-2K2n-3)
(a1 + 2) (a1 + 2 - ?) (a1 + 2^4J) + ... + (a1 + 2 ) . . . (a' + 2')

Pboof of (2G08- 11). Reduce successively by (1999). The integral

part after each reduction disappears between the limits in the cases (2608-9),
but not in the cases (2610-1). See also (2721).

2612
f* e co8l+,*dr =
1-2.3 ... (2n+l)
,e"+e~%)
J-W
(a2+l)(i+32)...(a4+2^+T!)^
2613
f * e-**cof?*xdx =
J_i,

1-2.3 ...2n

(e*-e~^).

a (a2+22)(a2+4*) ... (a*+2n)

Peoof.By successive reduction by (1999).

2614

fVaV cos 2kr<7.r =

Jo
2a

Proof.Denote the integral by u, then

- - [e-Jx 2a! sin 26* dx=-C~
cos 2&z <fo = - ^ ,
d6
Jo
Jo a
the second integration being effected by parts, Je""2*5 2as dx.
Therefore
log m = log C- ~ ; and 6 = 0 gives (7 = ^ (2425).
Otherwise.Expand the cosine by (765), and integrate the terms of the
product by (2426). Thus the general term is

v/t / b \2n
1
= (1)" 7t I ) r
, which gives the required resnlt by (150).
2a \ a / 1.2 ... n

2615

fVaV cosh 2ta> =

e^'.
Jo
2a

PKOOr. Cliange i into t'6 in (,2614).

(2181)

GIBGULAB LOGARITHMIC AND EXPONENTIAL FORMS. 889

2617
2618

Ce-"x sin 2bxdx = ^ e"4*.

Jo
J
JV**^1 sin (26*+

d*=&?1-tfyi (6e"6')-

Proof.To obtain (2617), put a = l in (2614), and difforentiato for 6.

To obtain (2618), differentiate, in all, n + 1 times for 6.
2619

f C0SX~e~aX dx = log a.
Jo

Peoof. By (2251) putting = [ e^<2y (2291), and changing the

order of integration, the integral becomes
(cos-e-"I)e-a:yiyd.E=

(e-*cos -<>*) dy dx

=r(^-^)^0(2584'229i)=ioga-

2620

JJlog (1 2a cos x+a2) dx = 0,

when a is equal to, or less than, unity ; but is equal to

2?rloga, when a is greater than unity.
Proof. (i.) o= 1. By (2635), since
log 2 (1 cos x) = log 4 + 2 log sin
(ii.) a < 1. By integrating (2922) from 0 to
(iii.) a > 1. As in (2926), integrating from 0 to ir.

2622

j^log (1 - n cos x) dx.

When n is less than unity, the values of this integral depend

on those of (2620). See (2933).
C
x sinxdx
ti
/i i \
i
/i . 1\
according as a is less or greater than unity.
Proof.Integrate
log (1 2a cos x + a1) dx by Parts, \dx, and apply
(2620).
Jo

2625

l coarx log(l 2a cos,r+a2) dx = , or

Jo
r
according as a is less or greater than unity.

5
r

390

INTEGRAL CALCULUS.

Proof.Substitute the value of the logarithm obtained in (2922). T

integral of every term of the resulting expansion, excepting the one in which
u = r, vanishes by (2467).
oco
^7

C* sin .v sin r.r dx _ Trar~1

J0 i_2acoSlr+a4-~2"'

ira (r+1)
"T~f

according as a is less or greater than unity.

Proof.Integrate (2625) by Parts, J cos rx dx.
C
cos rxdx
mf
J0 l-2acos*+a* = 1=*'

2629

,
. ,
^ < 1"

Proof.The fraction = cos rx (1 + 2a cos * + 2a1 cos 2a + 2a* cos Sx + . . . )

-J-(l-a'), by (2919), and the result follows as in (2G25).

3630

dx
7r
\-\-ae
1+x2'
1
2a
cos
cx+a*
~~
2(1
-a4)
l-ae~c"
Eih T=

Pkoof.Expand the second factor by (2919), and integrate the terms

by (2573).
ocoi
2631

dx = w\og(l-ae
,
J. log
(l2a coscx+a1)>
)

Proof. Expand the numerator by (2922), and integrate the terms by

(2573).

2632

Jo

xs'm cxdx
ir
-2a cos ar+a") ~~ 2 (ee-a)"

Pkoof.By differentiating (2631) for c.

Otherwise.Expand by (2921), and integrate the terms by (2574).
2633

p'log(l+ccos.r)^,= x ^__(co8-ic).j,
J0
coso?
(.4
)

Proof.Put a = 1 in (1951), and take the integral between the limits

0 and ^jt, then integrate for b between limits 0 and c ; the result is
f** W +**) <fe = 2 f
1
tan- JB db,
Jo
cos*
Jo-Zl-i1
Vl+t
and the integral on the right is found by substituting cos"1 b.

CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS. 391

flog
(1 4-c cos x) dx
, = ir sin
. ,1 c.
l
Jo
cos J?

2634

Proof.As in (2633), by taking 0 and jt for the limits of x.

2635

pog sin mim = \ log \ = j^g*^ dx.

Proof. I

sinxdx=\

cos x dx (2233).

Add these integrals and sub

stitute 2x, applying (2234) to the result.

2637

lg sin xdx = ~ log

Proof. J Z* log sin as cte = j^C^-*)' log sin as

by(2233). Equatetho

difference of these integrals to zero.

2638

\ x log sin* x dx = -nV log 2,

JO

n being an integer.

Proof.Method of (2250),
,-rZ sin' x dx = zZ sin1 x dx + zZ sin' a: cfa +- ... + I
zZsin'zf/z

Jo
J ir
J(n-llir
= [ xl sinJ x dx + \ (ir + y) I sin' y dy + ... + | {(n 1) 7T + J/J Zsin' y dy.
Jo
Jo
-o
Each integral reduces by (2635) and (2637) ; for example,
j (*+y)ltnn*ydy = 2 F (T+y)lemydy = 2* j Zsinyiy + 2

Zsini/tfy

= - 2*' log 2ir1 log 2 = - 3*' log 2.

The result is - {1 + 3 + 5 + . . . + (2n- 1) } ir' log 2 = - nV log 2.

2639

f
( * 1^2
+I-1Y
Joe-"
! dx = \
2\em
m)

Proof.Developo sin rtix by (764) ; integrato the terms by (239(5), and

sum the series by (1539).

2640

f
..'a
Jo

sin ira# ,
dx =
(f + l

tt
e

INTEGRAL CALCULUS.

392

Proof.Develope sin mx by (764) ; integrate the terms by (2398). The

resulting series is = 5 Y Z: cosec imr, by (2918), which is equivalent to the
above by (769).

2641

P (m log ^-C09 (" loS*) djc = jlogi^.

Jo
log^
l + /r

2642

f1 rin ( log ^)~si ( lQg *) dx = tan-1 m-tair1 .

Jo
log cT

Proof.Put p = tnt and q = m in (2394), and equate corresponding

parts. See (2214).

2643
psinMog)^ = tan.ln
Jo
logtv

rvers(nlog.F)d:c==
Jo
log J?

1
1+ n2'

Proof.Put m = 0 in (2641) and (2642).

MISCELLANEOUS THEOREMS.

FRULLANI'S FORMULA.
2700

f &gg)=i**) d* = *(o) iogi+ fi+M dx,

Jo
jc
a
J i #

A being = 00 , and the last term generally = 0.

Proof.In the integral [ ^ W (2j tiz substitute = aaj and * = 6x,
Jo
and equate the results thus,
*
or

'o

f'tC^-tWA-fj SL^E) & = j" t^) & = f (0) log A.

0
o
T
Then make 7j infinite. For applications see (2513) and (2505).

MISCELLANEOUS THEOREMS.

2701

393

RH-^(M dx = (a_6) r^M ^

Jo
or
Jo a?

++-(0)(iog'-+*)-(-)iffl+(t4^<fc.
a
with h = co ,
Proof.-

rfa | f"^!^ j = f * ^1 & - ^

Jo
Jo
(2257) = f M<k-V (0) la-*-?i
Jo
A
Jo 35
by making 6 = 1 in the proof of (2700). Integrate for a between limits
a and 6, thus

J" ti^l dx _ J* tiM dx

= (<*-&) f tfel^ + f (0) {ala-blb-a + b}- (^MB,

and the left is

= J" tl^^M dx- \[^dx.

a
POISSON'S FORMULAE.

Jo

1 2cco8jp+r

lt?JX

c being < 1 .
Proof.By Taylor's theorem (1500), and by (2919), the fraction is equal
to the product of the two expansions
2 (/ 00 +/(<*) cos x + ~~ f" (a) cos 2x +

f" (a) cos 3z + . . . J

and
{ 1 + 2c cos a; + 2c* cos 2x + 2c' cos 3x + . . . }
divided by (1 e'). By (2408) the integral of every term of the product
vanishes, except when it is of the form 2 f cos* nx, and this is = v, by
(2471). Hence the result.
J"
2703
r/(+0+/(a+0 (1_c cos<r) ^ = , {/(+,)+/()} .

2704
Jo

1 2c cos.i'+c2

Proof.As in (2702), adding unity to each side of (2919), and employing

(2921, 2467, 2470).
3 E

394

INTEGRAL CALCULUS.

ABEL'S FORMULA.
Given that F{?' + a) can be expanded in powers of e~m,
then
27Q5

r F(.v+iat) + F(*-iat) rf# = ,F(tr+o).

Jo
1+*
Proof.Assume F(x + a) = A + Ale"' + Ate-,a + Aie-'a + &c.,
.: F(x+iat) + F(xiat) = 2A + 2Alcosat + 2A,cos2at+&c.
Substitute and integrate by (1935) and (2573).
Ex.Let F(z) - , then f

KUMMER'S FORMULA.
2706

f " /(2tr cos

e*tf = sin far f(1
/(*) dz.
J-j
Jo
Proof.If /t = .re2*, then ? + A = 2* cos Be* by (766). Substitute these
values in the expansion of / (x + h) by (1500); multiply by e2"** and in
tegrate ; thus, after reducing by (7G9),
f** /(2*cos 8e) e**eZ0 = sinibr

- *M +

- Ac. 1

Again, putting h = xtp in (1500), multiplying by p*"1 cfy, and in

tegrating, we have I
(a x<f) df = the foregoing series within the
Jo
brackets. Equating the two values and changing ^ into 1 the formula
is obtained.
For an application see (2490).
2707

When k is an integer,

f** /(2.r cos

J -i

e**0d* = ffCQ,sA',r f '(1 -z)k-lf{xz) dz.

2si
Jo

Proof.Divide eqnation (2706) by sin k-r, and evaluate the indeterminate

fraction by (1580), differentiating with respect to k.
For applications see (2490), (2494).
2708

If X be a function of a* so chosen that

fV(*> *) ** =
Jo
and if the series

f*/(*> ) rfa>
. a

co.

A0f(r,0)+AJ(x,l)+Aif(z,2) + &c. ... =f (*)... (ii.)i

MISCELLANEOUS THEOREMS.

395

where <p is a known function, then

Cx<f>(x?)dx
... =

(iii.)
J^X/Or, 0)d.v

Pboof.Multiply (ii.) by X, and integrate from a to 6, employing (i.)

2709

If the sum of the series

J0+Alx+Atx2 + Asx:i + &c. ... =ff>(x)

be known, then
A0-\-A1a-\-Aia(a-\-l)-\-Aia(a-\-l)(a-\-2) + &c. ...
,ix0
e~xaf~l <j>(x) doc
..=o
e xaf~xdx
o
Proof.In (2708) let X = e-*a;-> and f(x,k) = x". Then since, by
Parts, we have [ ex**t-1dx = a (a + 1) ... (a + fc-1) f e-'af-'da,
Jo
Jo
it follows that Ok = a (a + 1) ... (a + k 1). Hence, conditions (i.) and (ii.)
being fulfilled, result (iii.) is established.
For an application see (2589).
Theokem.Let

f (x+iy) = P -\-iQ

(i.)

2710 Then Xf

rr
*/ a a f

<M

= -r

**

Peoof.Differentiating (i.) independently for x and y,

/ (x + iy) = P, + %Q if (x + iy)=P + i%,
.-. Px+ iQx = Qy-iP, .: Px = Q and Qx = -P,.
Hence by (2261) the equalities (ii.) and (iii.) are obtained.
Ex.Let f(x + ty) = e-<**<r>* = e'^e"' (cos 2xyisin 2xy).
Here P = e~"V cos 2xy, Q = e~^^ sin 2xy, therefore, by (iii.),
(V*' (e^cos 2/3as-ea'cos 2aa:) <fa = PV (e'^sin 2by-e-"' sin2ay) iy.
Ja
Ja
Put a = a = 0, 6 = 00; therefore
( - (e^ cos 2/S* 1) dx = 0, .-. e"* fe-^cos 2/3* da; = f e"^ dx.
'o
Jo
Jo

INTEGRAL CALCULUS.

396

CAUCHY'S FORMULA.
2712

Let j

^F^r?) dx = Ain, n being an integer, then

JM(if}*

r
Proof.In the integral

l
r" F (z*) c?z = 2Aln, substitute z = x >

and it becomes f (a - -1)*" (as + -i-) P [ (I_ 7)'! f = 2a,

Let the integral sought be denoted by C>n, then

This is proved by substituting in the first integral.

Therefore by addition

fK'+^M

Now, in the expansion of cos(2 + l) 0 (776), put 2cos0 = x-\

2isin^ = 2
SB
F j ^as j I

and

, where x = e'9 by (768-9), and multiply the equation by

and integrate from 2 = 0 to x = ca .

Then, by (i.) and

(ii.), the required result is obtained.

2713

Ex.Let F(x) =e-, then

a.erurj2._
Hi -ai> 1
1-8 ... v
(2
1) v, and
, =
-i/T
a,uu, -40
-a0
/

Therefore
_e-"v'* j

x2"e

n(n + 1) (-!)

dx
(-2)W , (-3)">

FINITE VARIATION OF A PARAMETER.

2714 Theorem (2255) may be extended to the case of a
finite change in the value of a quantity under the sign of
integration.

MISCELLANEOUS THEOREMS.

397

Let a be independent of a and b, and let A be the differ

ence caused by an increase of unity in the value of a, then
I Aj> (x, a) dx = A f ^ (x, a) dx.

2715

Ex. 1.

Ce-"dx=, .'. f Ae-<"dx= A , that is

Jo
"
Jo
a
1
e-"(e-* 1) dx =
a (a -hi)
J0
f
Also, by repeating the operation,
A"e-*<e = A", that is
u
2716

fV"(e-'-l)^ =
j^fe- v
Jo
a(a-f-l) ... (a+)

2717

Ex. 2.In (2583-4) put it for a and (2a -to) for b, then
A sin (2a-m) *<k = A ^j^^J,

(*).

e"*x Acos (2a to) a!cZj) = A -j

ft" + (,2a wi)"
c
In (ii.) let to = 2p, an even integer, then
AJp cos (2a - 2p) 3 = cos (2a + 2p) x 2p cos (2a + 2p 2) * + . . .
... +coB (2a 2p) x
= cos 2ax [cos 2pa: 2p cos (2p 2) a + 0 (2p, 2) cos (2p 4) a...
. . . + cos 2j)x\
sin 2az [sin 2px2p sin (2p 2)
...
... sin 2pa;],
The coefficient of cos 2ax, in which equidistant terms are equal, is
= ( l)p22j,sin!',>a! (773) ; while the coefficient of sin 2ax vanishes because
the equidistant terms destroy each other. Therefore
A2" cos (2a 2p) x ( 1)" 2!p cos 2ux sin2"*.
Hence (ii.) becomes
/loo
** cos 2ax sin2^rf^ = L_J)f A2*
2*
^+(2a-2p)2'
2718 J *2719 Again, in (i) let to = 2p + l, an odd integer, then
A**'1 sin (2a-2p-l) * = sin (2a + 2p + l) x-(2p + l) sin (2a + 2p-l) *
+ C(2p + 1, 2) sin (2a + 2p-3)
sin (2a-2p-l)#
= sin 2aas [cos (2p + l) x- (2p + l) cos (2p l)z+
cos (2p + l) x]
+ cos2ax [sin (2p+ 1) ai-(2p + 1) sin (2p-l)a;+ ... + sin (2p + l)sj.
The coefficient of sin 2ax vanishes as before, while that of cos 2aas is
= (-l)2**lBn**,8> (774).

INTEOEAL CALCULUS.

398

Therefore equation (i.) becomes

2720
V** cos 2cuv sin?p+\rtfcr = Lz^r2*+I
1<

2a-2p-l
Ar+(2a-2p-l)s

To compute the right member of equation (2718), we have

tfp
i. f
1
ft* + (2a -2^)'
Ljk, + (2a + 2p)'
_
2p
g(2p,2)
_
1
1
4, + (2u+2ij-2)' T fc* + (2a + 2i-4)* "
*"+(2a-2P),J'
Let a=0, then the equidistant terms are equal, and we obtain in this case
2721

*
(~l)"1.2...2p.2
ff+(2a-2p) - A: (tf+4)(**+16) ... {*+(2p)t}*

Thus formula (2609) is proved.

Similarly, by making a = 0 in (2720) after expansion, formula (2608) is
obtained.

Let p be any integer, and let q and a be arbitrary, but

q<2p in (2722), and <2p + l in (2723).
2722
rcoaftLggin** .
Jo

(-1)"
rA
dx
~ 2*r(g+l) J.
**+(2a-2/)fl*-

2723
fcos2<fcrsin*p+1.r .
(-1)"
rA^1(2a-2p-l)^rf
= 22"+1r(g+l) J,
3+(2a-2s)
'
where A has the signification in (2714).
Peoof.Employing the method of (2510), replace

q being integral or fractional ; therefore

f S2^ax **L? dx = f fees 2ax BhPzeSdM dx,
J
r(2+i)J.J.
by changing the order of integration. Substitute the value in (2718) for
the integral containing x, writing the factor z" under the operator A, since it
is independent of a.
Similarly, with 2p + l in the place of p, we substitute from (2720).

MISCELLANEOUS THEORbitg.

399

It may be shown that, whenever a > p, formula (2722)

reduces to
2724
o

rfa? =
*

-A(2a-2p)*.
22p+1r(g+l)sin^

For a complete iuvestigation, see Cauchy's " Memoire de l'Ecole Polytechniqne," tome xvii.
2725

and

Ex.-Let a = 2, p = 1, q = \,
rCOHl* z\n*xdx =
x
A, (2a_2)i)
Jo
8r(f)sin-|A'(2a-2)* = (2a+2)*-2 (2a)* + (2a-2)* = 6* - 2 . 4* + 2*.

FOURIER'S FORMULA.

2726
when a = co and A is not greater than \ir.
Pkoof.(i.) Let <t>(x) be a continuous, finite, positive quantity, de
creasing in value as x increases from zero to h.
rh .
r* r r
r
r*
r^Hx)to=\'+\'
(L)>
J0sina!rw
J0 J* +[>+..,+['
J?r
J(r-i)- +\Jr?
w'
a
a
a
a
being the greatest multiple of contained in h. The terms are altero
a
nately positive and negative, as appears from the sign of sinaa;. The fol
lowing investigation shows that the terms decrease in value. Take two
consecutive terms
(n-H)ir
(ti).
f sin
ax

,
v
,
f
' sin
, > dx.
,

d> (x) ax,

m <t> (x)
nw
sin x
J
sin a;
a
a
Substituting x - in the second integral, it becomes
in+lL.

/
-L

sin

(-7)
and since decreases as a increases, an element of this integral is less than
the corresponding element of the first integral.

400

INTEGRAL CALCULUS.

Now, by substituting ax = y, we have

'

a Bin
a

when a is infinite, because then <j>

Blr
= Q (0) and o sin lj- = y.

Hence the sum of n terms of (i.) may be replaced by <p(0) {"'^^dy,

Jo y
which, when n is infinite, takes the value ^>(0) fir by (2572) ; while the sum
of the remaining terms vanishes, because (the signs alternating) that sum is
less than the n-\-lih term, which itself vanishes when n is infinite.
(ii.) If <p(x), while always decreasing, becomes negative, let C be a con
stant such that C+<p (x) remains always positive while x varies from 0 to h.
The theorem is true for C+<j> (x), and also for a function constant and equal
to 0, and it is therefore true for the decreasing function <p whatever its sign.
If <j> (*) is a function always increasing in value, <p(x) is a decreasing
function. The theorem applies to the last function, and therefore also to <p (x).
2727 Cor.Hence the same integral taken between any
two limits lying between zero and ^tt, vanishes when a is
infinite.

2728

C?^<f,(x)<LT
Jo sin a;

= n{i<f>(0)+<t>(Tr)+<j>(2v) + ...+<l>(n-l)ir+<l>(nir)},
when a is an indefinitely great odd integer, and nir is the
greatest multiple of n less than h. But when a is an indefi
nitely great even integer, the second and alternate terms of
the series have the minus sign.
n

f* sin o.i! , , ,
f sin cue . , x , . f* sin ax . , x ,
,. ,
Jo SlaX
J0 SmX
Jr BmX
decompose the second integral into 2n others with the limits 0 to Jjt, ^jt to ir,
ir to |ir, ... (2n 1) gir to nir ; and in these integrals put successively x = y,
iry, ir + y, 2ir ;/, 2ir + y, ... niry. The new limits will be 0 to \tt, \ir to
0 alternately, with the even terms negative, so that, by changing the signs of
the even terms, the limits for each will be 0 to
Also, if a is an odd in
teger, B' ax is changed into ?HL*2 by each substitution, so that (i.) becomes
sin x
sin y
^^{*(y)+<P(*-y)+<l>(*+y)+-..+*(n*-y)}dy
0
, f sin ax ,
J sin*
But, when a is even, the substitution of
j for x makes sm>a^ minus
sin y

MISCELLANEOUS THEOREMS.

401

whenever r is odd. The limit of the first part of (iii.) is

-~-{tf> (O) + 2?(*-) + 20 (2*) +-...+ 2$. (w-l) w+f (*)}, by (2726).
In the last part of (iii.) put x = nr+y, and the integral becomes
L2f (tw+y) dy = -1 ?(*), if h-mr is >

by (2725).

If A mx lies between -ir and x, decompose the integral into two others ;
the one with limits 0 to tt will converge towards $ir(p (nir), while the other
with limits \ir to h nir becomes, by putting y = irz,

0[(m+1)t t]d* = 0,

the limit by (2727). Hence the last term of (iii.) is \nri> (wr).
tuting these values, (2728) is obtained.

Substi

2729

Ex.By (2614), [% cos 2Jz dse =

2a
Jo
Put 6 = 0, 1, 2 ... n successively, and add, after multiplying the first equa
tion by i, thus
e-<Ac> (i + cos 2z + cos 4x + . . . + cos 2z) dx
i

The left side

= [%-<*- sin (^ + *)* ^ by (801),

Jo
81n*
and, if n = oo , becomes
JL j| + e-^ + e-^+e-**I>_K,.}) by (2728) ;
/
i -9'+...}.
.-. x ji + ev + e-,J + e-',,+ ...} = ^{l
+ e _I""^-e - 2+e
Put ira = a and = /3 ; therefore
a
2730 V*{\+e-ai+e-iai+e-+...)
= ^{H^+e-ne-^+...},
with the condition a/3 = jr.

2731

Ct^^r)dx=m,

when a is an infinite integer.

Proof.The integral may be put in the form
I s'n ax * (x) dx, where * (x) = 2IjS a (j-Y
3 F

INTEGRAL CALCULUS.

402

therefore, by (2726), when h is > \ir, and by (2728), if h is >

the value
is 1** (0), since in (2728) * (a-), * (2*-), Ac. all vanish. But * (0) = f (0).
Hence the theorem is proved.
When a and /3 are both positive,
2732
/a
2733

^)dr = 0=f^ *(*)dr.

%/ a
a?
P
/ a

= +(<!).
*X*

PR00F.-(i.) f = f - j" = f *(<>)- f * (0), by (2729).

(ii.) j' =| + =f,(0)+f *(()),
by substituting x in the second integral.

2734
PnooF.

f f <j>(.r)cosuxdud.v = -^-<(0),
Jo Jo
2

when a = co.

nx = I costix du. Substitute this in (2731).

Ja

When a and /3 are positive, the limit when a is infinite of

2735

f I <f>(x) cos tu cos uxdudx,

Jo Ja

or of

f f ^(.r) sin/w sin uxdudx,

0 Ja

is ^tt^ (/), if / lies between a and /3, |;r< (/) if t = a, and zero
for any other value of t.
Pb >of.When a = oo we have, by (GC8), and integrating with respect to u,
J" j " f , *) cos * cos f & <* = i J" (*- * () db + * j"

0 f> () tf*

= |(p-,^f>(,+0A+jp*,fi(-_|)4i
(i.),
J-< z
J+ 2
by substituting u = a; < and z = x + i in the two integrals respectively.
When a is infinite, the limit of each integral is known.
When n and /} are positive and t lies between them in value, the
limit of (i.) is \w $ (t), by (2732-3)
(ii.)
When a and /3 are positive and t does not lie between them, the
value is zero, by (2732)
(iii.)

MISCELLANEOUS THEOREMS.

403

If a = t in (i.), the first integral becomes = \ir$ (<) by (2731), and

the second vanishes as before ; so that the value, in this case, is \ir <f> (t) . . . (iv.)
The same demonstration applies in the case of (2736), transforming by
(669) instead of (668).

Hence, by (ii.), if t be always positive,

2737

1 1 <f> (&) cos tu cos ux du dx =

Jo Jo
&
= \ 1 <f>(x) sin tu sin uxdudx.
Jo Jo

2739

Ex.Let

0 (as) = e-,
e~ax cos tu cos ux dudx
fix. =

e'at.
2

Therefore; by (2584),

f" oobJw du _ * g.
Jo a
which is equivalent to (2574), with t = l.
The expressions in (2737-8) being even functions of u, we have, supposing
t to be always positive,
(z)cosiitcosM2cZMrfa; = ir^(<) = |

| <p (x) sin <w sin ux du dx ...(i.)

Replacing f (z) by o>( a), and afterwards substituting x, these

equations become
f[

<;> (x) cos itt cos ux du dx = ir<p ( <)

= I

^> () sin w sin uxdudx

From (i ) and (ii.), by addition and subtraction, we get

2740

I I ^> (x) cos tu cos uxdu dx = ir[<j> (t) +<)> (t)],

J - . -co

2741

^ (a) sin <u sin was iwia: = it

(<)^ (<)]

Whence, by addition,
2742

cos w (tx) dudx = 2ir<f> (f),

the original formula of Fourier's.

(ii.)

INTEGRAL CALCULUS.

404

THE FUNCTION
The function dx log T{x) is denominated ^(x).
2743

+C) -kg*-

11
,

1
a,+2

1
^

when /li is an indefinitely great integer.

Proof.By differentiating the logarithm of (2293).

2744

Cor. *(l) = log/i-l

2'

31

'*

when fi = ao ,
= -0-577215,664901,532860,60 ... (ZfoZer).
All other values of \f>(x), when a; is a commensurable
quantity, may be made to depend upon the value of ^(1).
When x is less than 1,
2745
* (1 -x)
f = w cot irr.
Proof.Differentiate the logarithm of the equation
r(a)r(l->) = 7r-T-8inxa! (2313).

2746

tAGr)+tA(^+^)+v(^+4)+...+v(^+^)
= n^r(kr) logn.

Proof.Differentiate the logarithm of equation (2316).

2747 To compute the value of \L[ -} when - is a proper
st\qn /
q
fraction.
Find i/- ^Lj from the two equations
2748
2749

iA(l-ii)->/(>) = ircot-7r,
19
q

(2745)

+(l-JL^++(}) = 2 j^(l)-logg+cos^log^2vers^
+ COS

log (2 vers^)+cos ^ log(2 vers S)+&c. j .

THE FUNCTION \\>{x).

405

The last term within the brackets, when q is odd, is

C03 (7-1) f log (2 vers (2^-?) ;
and when q is even, the last term is + log 2 according as p
is even or odd.
Proof.Equation (2743) may be written
^(x) = --i+z2-4r+zl
r-+,ci7
x
z+1
s + 2
x + n fi + L
H being an indefinitely great integer.
Replace x successively by , , ... 21? 1 ; where q is any in
teger; thus
2 1
1
1

(f).-

2 +
^ 12

2_2 +n
3+
T *

2? + l + 's

33+l + i

22 +' 2^ S

32 + 2T 1
(i)

*(1)

=-1

+22-

+Zf-

+Z|-

Now, if

be any one of the angles , , ... ^2DjT we shall have

1 2 2
2
1 = cos q<p = cos 2q<f> = cos 3g0 = &e
(ii.),
cos <p = cos (g + l) 0 = cos (2g + l) 0 = cos(3j + l) 0 = &c
(iii.),
cos 0 + cos 20 + cos 30 + ... +cos (q 1) 0 + 1 = 0 by (803)
(iv.)
By means of the relations (ii.) and (iii.), equations (i.) may be written
cos 0 4* ( ) = q cos* + cos0Z2
\qI

2 cos (2 + 1) 0 + cos 0Zf ,

2+1

cos 20i ( ) = --2- Cos20 + cos20Z2

2-cOB(2 + 2)0 + cos20Zf-,

cos 3^ 4- (y) = -|- cos 30 + cos30Z2 -2j^cos(2+3)0 + cos30J|-,

cos (q 1)01^ f5-M =

\ q I

2-cos(g-l) 0 + cos(y 1)0Z2

g-1
2-2_coe (2g-l)0 + cos (2 l)0i| 1

* (1) = - 1
11 -.
12
Upon adding the equations, the coefficient of each logarithm vanishes, by (iv.)
The remaining terms on the right form a continuous series, and we have
cos0^ i^j) +cos 20^ (y ) + + 008

t>4>

q {cos 0 + \ cos 20 + i cos 30+in inf.}

= \q log (2 -2 cos 0) by (2928)

C1)
(v.)

INTEGRAL CALCULUS.

406

2x = u.
Let

Then, by giving to tp in equation (v.) its different values ,

2ii>, 3w ... {q 1) <, we obtain q 1 linear equations in the unknown quanti

ties i (~~)> ^
^ ^
To solve these equations for \p (^~)
being an integer less than q, multiply them respectively by
cospiii, cos 2pu> . . . cos (q l)pu,
and join to their sum equation (2746), after putting z = and n = q.
The coefficient of \f/ [j in the result, & being any integer less than q, is
cospu cos A-ai + cos 2/>a; cos 2A-w-f ... +cos (j 1) pu> cos (j 1) kin 4- 1.
By expanding each term by (6G8), we see by (iv.) that this coefficient
vanishes excepting for the values k = q p and k = p, in each of which
cases it becomes = \q. Hence, dividing by \q, we obtain
^(izJ^J+f ^ =s2+(l)-2i2+ooBjHZ(2-2ooe-)
+ cos 2pul (2-2 cos 2w) +
+ cos (q 1) pwl {2-2 cos(g-l) }.
The last term = cos^>wJ(2 2cos w) = the third term; the last but one
= cos 2pul (2 2 cos 2<^ = the second term, and so on, forming pairs of
equal terms. But, if q be even, there is the odd term
cos \qp<* log (2 2 cos qv) = 2 log 2,
according as p is even or odd.
Examples.By (2748-9) we obtain
2750 >KJ) = Kl)-31og2+!-,
2752
2754

*(i) = *(l)-81og2|,

*(i) = *(i)-*iog3+273. *(*) =*(i)-* log s-gjg.

^q) = ^(i) -2 log (2).
DEVELOPMENTS OF ^(a+.r).

When a; is any integer,

2755

(*M)-*()+i+^i+^1+...+5ji=I.

Proof.By (2289), putting = a + a; 1 and r = - 1,

T (a+x) = (a+*-l) (a+x-2) ... (a + 2)(o+l) a T(a).
Differentiate the logarithm of this equation with respect to a.
2756

(a+*) = *()+-- *
+ ;.r ^"ffi
a
2(a+l)
da (a+l)(a4-2)
.rQr-l)(>-2)(,y-3)
4a(a + l)(+2)(a+3)^
'

THE FUNCTION iff (or).

407

If a; be a positive integer, the number of terms in this

series is finite, and the value of \p(a+x) can be found from
that of if, (a).
Hence, by this or the preceding formula, in conjunction
with (2747), the value of
when N is any commensurable
quantity, may be found in terms of ^(1).
Proof.Let vj/ (a + x) = A + Bx + Cx (x - 1 ) + T>x (x 1 ) (x - 2) + &c.
Change x into x + 1 ; then,
A^(+x) = ty{a+x+l)-<l>(a + x) = dx {log T (a+x+l)-log T (a+x)}
= dx\og (a + x) (2288) = ~- ,
Ax = 1, Ax (x - 1) = 2x, Ax (x - 1) (x - 2) = 3x (x- 1), Ac.

Therefore

= +2ae + 3Dx(x-l) + 4#*(x-l)(x-2) + ,

a+x
A -J = 2(7+2. 3Dx + 3.4.Ex(x-I) + ,
o+x
A1 = 2.3D+2.3.4. Ex + ,
a+x
A* = 2.3.4 + .
a+x
Put x = 0 in each equation to determine the coefficients A, B, C, D, &c.;
thus

A = $ (a), B = -, 20= A-i- = 11

o' "
a
a+1
a

1
o(a + l)'

2.3I> = A' = A ~* =
,
a
a (a+1) a(a+l)(a + 2)
1
o
o o
2 3 4E = A* = A
=
.
, and so on.
^ o
0(rt4i)(0 + 2)
o(o+l)(o+2)(a + 3)'
SUMMATION OF SERIES BY THE FUNCTION t/r(tr).
2757 Formula 1.

.+^+^. + ... + ^-

Proof.Let S, denote the n terms of the series to be summed.

We have

S..,-S. = f -(|+n+l)=|[^(A+B + 2)-*(A+n+l)](2288)

S.tl-^(A++2)=S.-*(|+n+l).
Hence the difference is independent of n, and therefore
S.-A+(A+.+,).S._^(|+1).i_?+(i.+l).

INTEGRAL OALOALUS.

408

2758

Ex.

l+I + I+...+ 5-LI = l-^(|)+^(n+l).

2759 Formula II. -2b

" '
"
b+c^b+2c

6+(2n+l)c

Proof.The series is equivalent to

a .
a
a
( _o_ ,
a

g
"?
b
b + 2c b+4c
b + 2nc
Ib + c 6 + c + 2c" fc+c + 27icj'
and the result follows by Formula L
2760 Formula III.
1
1.1
6
b+c^b+2c

in inf.

~ b(b+e) + 2er\ 2c )
Proof.Make n = oo in Formula II.

V\ 2c ) 5

The last two terms become equal .

2761

Ex. 1.In (2760) let b = c=l, then

l-*+i-i + *c = i+i
(2)-* (4)} = log 2.
For * (2) = 1+* (1), by (2755) ; * (f) = 2+* (1) -2 log 2, by (2754-5).
2762

Ex- 2.In (2760) let 6 = 1, c = 2, then

1-1+ W+Ao. = l+i*

2763

Proof.
But
therefore

(!) = f

+(!+) = flf!z^ <**++(!)

^(l + a)=^(l)+a-^^ + i^l)^-&c. [by(2756)

1-a)* = a _

i + (-!) (pg) z, _

* (1 + a) = f 1~^1~!8^<fa+^ (1).

Substitute 1 a; in the integral.

2764

*(l+a)-+(l+6) = Tflzf6 dx.

[By (2763)

THE FUNCTION

409

Ex.Put 6 = a; then
+ (l+o)-*(l-o) =i+iKa)-iKl-) (2756) =I-ooto (2745).
a
a
2765

j o s-1- ax = a

Therefore

jr cot ira.

^ {x) AS A DEFINITE INTEGRAL INDEPENDENT OF \\> (1).

2766

+W = -|;(E1-+^)*..

Peoop. <J> (x)

w = losu
sr
x

51
x+

x+ft_
1 with u = a> (2743).
K
'

Bat

++ +
1
_ f1 a*-1-^-1 dg
X
X+l
" X+fl 1
Jo
1 z
by actual division and integration.
Also

Put * =

Jog^j'^-gJ) dz

(2367).

in the first integral, therefore

^ / = V <h = ii \ y
J
dy.
Jo
ijr
Jo
i2T
Replace y by z, and suppress the term common with the second integral
of (i.), and we get

\f> (x) = j | j - C

j dz.

Put z*1 = m, and this becomes

U(i-*)

But when (i = &> the product ^(1fl) has logi* for its limit (1584) ;
m
and
= 1. Hence the result.

2767

W-j-jr--^*

g-'z'-'dz ;
o
But, by (2427),

d.r (a) =

e-z*"1 log z cfe.

Jo
logz = f e~"~8""da,
Jo
a
3 a

410

INTEGRAL CALCULUS.

Jo Jo

_ f [-JY*"*"'<fc-p~"+*"'*'''k] *

which establishes the formula since

d,r(x) + r(x) = i.log r (*) = * (*).

2768

2769

3770

logr(*) =[(*_l)e--?^=]#

= fT^-*+llf**
Jo L 1
J z

W-j-[^_I^,]

Proof.Integrate (2767) for x between the limits 1 and x, observing;

that logr(l) = 0; thus
iogr(*)=
j(a-i)e -i 1
* ;
|
Jo 1
log(l + a)
J a
Subtract from this the equation obtained from it by making x = 2, and
multiplying the result by x 1. We thus obtain
^^[[(-.kh.)--'1^-;^"-]^
Substitute = log (1 + n), and (2768) is the result. To obtain (2769), sub
stitute z = (1 + a)"1. Lastly, (2770) is the result of differentiating (2768)
for x.

NUMERICAL CALCULATION OF logr(^).

2771
The second member of (2768) can be divided into two parts, one of
which appears under a finite form, and the other vanishes with x. If we put
p = H-i-e-f)j- andQ=ioi^'
then

log r (a>) = f"(P+ Qe~t*)

(i.)
Jo
If Q be developed in ascending powers of , the terms which contain
negative indices are
+ - = R iay.
4
2i

NUMERICAL CALCULATION OF logT^).

411

Pat F(x) =^(P+Re-t*)d,

=[[(<*-'>- T^H+(j+lMf
and

w(x)=T (Q-R) e

= f* (j-L.-

<*>
| ... (iii.)

Then, by (i.),
log T (as) = F (x) + r (a)
(iv.)
F (as) can now be calculated in a finite form, and ts- (x) will have zero for its
limit as x increases.
First, to show that F (^) and sr (|) can be exactly calculated,

and, by substituting
<-j;(nb-^-i)'-'f

<->

Again, putting x = 1 in (iii.), we have

-o)=i;(r^-l-i)-'f
and, by substituting A,
-w-fGrb.-JriHf
The difference of (vi.) and (vii.) gives
o = r f-i_ '0

- i=p ) 8-t i

(vin.),

since
1-e-f l-e~
Subtract (viii.) from (v.), thus
(I) - i

l-e-'

- 6-f) = -J - M C2429).

Also, by(iv.), F(J)+(|) = ir(J) = iU, .-. F(t)=ilog(2w)-i...(iz.)

.F(;e) may now be fouud by calculating F(.t) F() as follows:
By (ii.), F(x)-F(i) = r[(x-i)e-*+

.-. by(iv.)
2772

(-*-a-*)] |

Jo
^
2 Jo
= i-x+ (*-i)loga (2427-8),
.-. F(a;)=ilog(2jr) + (ai-$)]ogai-*, by(ix.)j
logr(*) = Jlog(2Jr) + (a-J)loge-a+<r(*J
I' () = e-*x*-*y/(2w)e-M

(x.) ;
(xi.)

INTEGRAL CALCULUS.

412

When x is very largo, e" W differs but little from unity. For r {x) diminishes
without limit as x increases, by the value (iii.)
Replacing r (z) in (x.) by its value (iii.)i and observing that
log T (x + 1 ) = log x + log T (as),
we get log r(z + l)=i log (2*0 + (x+ J) log xx
+[y*A-i)'-f

Now, by (1539),
\l-e-f
f
2/4
where 6 is < 1. Also

1.2

1.2.3.4

1...2n

1...2n+2'

So that equation (xii.) produces

2773

log r

= !2gW +

1.2.c

+ 1) log #-#

.
-jQBin+i
3.4ra_r-'~,~ (2n+l)(2n+2)a-"+1'

This series is divergent, the terms increasing indefinitely. The comple

mentary term, which increases with n and is very great when n is very great,
is, however, very small for considerable values of n. For instance, when
= 10, the values obtained for log V (11), by taking 3, 4, 5, or 6 terms of
the series, are respectively,
1G-090820096, 1610441,5343, 16-104412565, 16-104112563.

CHANGE OF THE VARIABLES IN A DEFINITE

MULTIPLE INTEGRAL.

2774 Let x, y, z be connected with , v, I by three equations

u = 0, v = 0, w = 0.
Then, when the limits of the integral containing the new
variables can be assigned independently, we have

JtiJmJi,

d (uvw)
d(xyz)

TRANSFORMATION OF A MULTIPLE INTEGRAL.

413

where 4> is what F becomes when the values of x, y, z, in terms

of $, v, I, obtained by solving the equations xi, v, w, are sub
stituted.
Peoof.

[ [ [ F (x, y, z) dxdy dz = f f f * ({, ,, 0 g ^ - H dr, d(.

To find Xt, consider r\ and constant, and differentiate the three equations
u, v, w for ,, as in (1723). To find y , consider and x constant, and differ
entiate for r/. To find z^, consider x and y constant, and differentiate for (.
We thus obtain
d (uvw) d (uvw) d (uvw)
d (uvw)
dxd1ch__ d (jyz) d~(^rj~ d (gg _ _ d ggj
di, dr) d

d {uvw) d (uvw) d (uviv)

d (xyz) d (j/zt) d, (zi,r})

d (uvw) '
d (xyz)

observing that two interchanges of columns in a determinant do not alter

its value or sign (559).
Similarly in the case of any number of independent variables.

When, however, the limits in the transformed integral

have to be discovered from the given equations, the process
is not so simple.
In the first place, we shall show how to change the order
of integration merely.
2775

Taking a double integral in its most general form, we shall have

rJ n (x)

r ,- ()

F(x,y)dxdy=S,\

F(x,y)dydx

(i.)

The right member will generally consist of more than one integral, and 2
denotes their sum. The limits of the integration for x may be, one or both,
constants, or, one or both, functions of y. "9 is the inverse of the function \f/,
and is obtained by solving the equation y = ^(x), so that x = ((/). Simi
larly with regard to ^> and <b.
An examination of the solid figure described in (1907), whose volume
this integral represents, will make the matter clearer. The integration, the
order of which has to be changed, extends over an area which is the projec
tion of the solid upon the plane of xy, and which is bounded by the two
straight lines x = a, x = 6, and the two curves y == \p (x), y = (x).
The summation of the elements FQqp extends from a to b, and includes
in the one integral on the left of equation (i.) the whole of the solid in
question.
But, on the right, the different integrals represent the summation of
elements likePQ^), but all parallel to OX, between planes y = a, y = ft, &c.
drawn through points where the limits of x change their character on account
of the boundaries y = \p (x), y = (p (x) not being straight lines parallel
to OX.

414

INTEGRAL OALCULUS.

2776 Example. Lot the figure represent the pro

jected area on the xy plane, bounded by the curves
y = \p (a;), y = <j> (z), and the straight lines z = a, z = 6.
Let y = <p (x) have a maximum value when z = c.
The values of y at this point will be <p(c), and at the
points where the straight lines meet the curves the
values will be <p(a), <t>(b), <p(a), \l>(b).

According to the drawing, the right member of

equation (i.) will now stand as follows, U being written

for F(x,y),
r+ (o) rb

pf, (o) rb

rt (J) rb

Udydx+\
Udydx+\
Udydx +
}*(b)}v()
U(a))a
J*()J*()
The four integrals represent the four areas into which the whole is divided
by the dotted lines drawn parallel to the X axis. In the last integral, 4>, (y)
and *s(y) are the two values of a: corresponding to one of y in that part of
the curve y = <j> (x) which is cut twice by any x coordinate.

2777 To change the order of integration in a triple integral,

from z, y, x to y, x, z, we shall have an equation of the form
[r, ff. (*) [$, {x,y)

re, Wi () f *.. <, *)

F(x,y,z)dzdydz = 2\

F (x,y,z) dzdxdy

Here the most general form for the integrals whose sum is indicated by 2
is that in which the- limits of y are functions of z and x, the limits of x func
tions of z, and the limits of z constant. Referring to the figure in (l'JOG),
the total value of the integral is equivalent to the following. Every element
dxdydz of the solid described in (1907) is multiplied by F (xyz), a function
of the coordinates of the element, and the sum of the products is taken.
This process is indicated by one triple integral on the left of equation (ii ) ;
the limits of the integration for z being two unrestricted curved surfaces
z = 0i (a;, y), z = <pi (x, y) ; the limits fory, two cylindrical surfaces y = (^(x),
y = ij (a:) ; and the limits for x, two planes x = xv x = xv
But, with the changed order of integration, several integrals may be
required. The most general form which any of them can take is that shown
on the right of equation (ii.)
Solving the equation z = ^, (x, y), let
yx = <t>, (z, as), y, = *, (z, x) be two resulting values of y ; then the integra
tion for y may be effected between these limits over all parts of the solid
where the surface z = <px (x, y) is cut twice by the same y coordinate.
The next integration is with respect to x, and is limited by the cylindrical
surface, whose generating lines, parallel to OY, touch the surface z = ^(z, y).
At the points of contact, a; will have a maximum or minimum value for each
value of z ; therefore dl/<f>1(x, y) = 0. Eliminating y between this equation
and that of the surface, we get x = , (z), x = , (z) for the limits of x.
Lastly, the result of the previous summations is integrated for z between
two parallel planes z = z,, e = z drawn so as to include all that portion of
the solid over which the limits for x and y, already determined, remain the
same.

TRANSFORMATION OF A MULTIPLE INTEGRAL.

415

The remaining integrations will take place between z = z, and similar

successive parallel planes ; and, according to the portion of the solid which
any two of these planes intercept, the limits of x for that integral will bo one
or other of the bounding surfaces, curved or plane, the limits of y, one or
other of the curved surfaces.
The general problem to change the variables in a multiple
integral, and determine the limits from the given equations,
may now be solved.
2778

First, in the case of a double integral,

fxt rci (x)
F(x,y)dxdy
(in).,
Xl J Ml (x)
to change from x, y to , j, having given the equations u = 0,
v = 0, involving the four variables.

To change y for n, eliminate between these equations ; thus y =/ (a>, ij)

and dy = (x, 17) dr). Substituting these values, we shall have
F(x, rj) dy = F{x,f(x, ,)}/, (*, n) dr, = F, (x, n) dn.
Also, if f)j corresponds to y the equations yl = fix (x) and yx = f(x, Jjj) will
give ij, = \pt (x). Similarly 17, = \p3 (x).
Hence the integral (iii.) may now be written
fx, [Mx)
fn f*n(i)
F, (x, r,) dxdr, = 2
Fx(x,r,)dr,dx
(iv.),
JxiJ^itx)
J H J ,(,,)
the form on the right being obtained by changing tlw order of integration, as
explained in (2775).
Next, to change x for I, eliminate y between the equations u = 0, v = 0 ;
thus, x = g (, if) and dx = <7f (, fj) d. Substituting as before, we shall have
F^x, v)dx = F3^,V)di.
Also, i, corresponding to as,, the equations xx =
(17) and x, = g ({ 17) pro
duce , = m, {if), and in the same way , = m, (ij). Hence, finally,
F(x,y)dxdy = 2\

F^,r,)dndi,

(v.)

In the last transformation from x to , the most general form of the

integrals which may be included under 2 has been chosen. When any of the
limits of x are constants, the process is simplified.

2779 Again, to change the variables from x, y, z to t,n,Z,

in the triple integral,
{xpMx)rxi(x,y)
F(x,y,z) dxdydz
(vi.),
X! J ih (x) J xi (x,)
having given the equations u = 0, v = 0, 10 = 0 between the
six variables x, y, z, , y, I.

416

INTEGRAL CALCULUS.

First, to change from z to (, eliminate 4 and 17 between the three

equations, and let the resulting equation be z =f(x,y, ). From this
dz = f{(z,y, i) d{; therefore
F 0, y, z)dz = F {x, y, f (x, y, () } f{(x, y, 0 d( = Fl (x, y, f) .
Also, if f, corresponds to the limit z\, the equations z, = Xi (^i y)
ei =/(< 2/. fi) give = ^ (x, y). Similarly f, =
y).
The integral (vi.) may therefore be written
f r+i (x) /+. (x, )
rf, r, (f) r*i t *)
i^O,,,, 0^*^ = 2
Fi(x,y,0^dxdy
JxiJi^iCx) J*i(x,)
Jfi J*,(f) J*,(S
(vii)
the last form being the resnlt of changing the order of integration, as ex
plained in (2777). We have now to change from y to >;; we therefore
eliminate z and 4 from the equations u, v, w, obtaining an equation of the
form y =/(, x,
and prooeed exactly as before. The result, as respects
the general form of integral in (vii.), will be
f?i
fMf.*)
F,(x,n,:)dCdxdn
(viii.)
JhJVM Jai(J,x)
The order of x and i) has now to be changed by (2775). Since K is a
constant with respect to integrations for x and 17, ,()>
w^ a'80 be
constants, while X^f, a), Xs(, x) will be functions of the single variable a;.
Suppose r; = X,(f, a;) gives as = A,(f, ij). Similarly, x = A,(, ij) maybe
the other limit.
At the point where a; = ,() and ij = X,(, a;), we shall obtain by
eliminating a, 6ay, j = yu1 ). Similarly, from x=1rl(i) and j = Xa(x)
snppose, we get i/ = n,() for the next limit; then a general form for the
transformed integral will be
(MO rA(c,)_
i
J; (ar, ?, 0 d{dr,dx
(ix.)
0m<0Ja,;*)
It now remains to change from the variable x to 4. Eliminating y and z
between the equations , , w, we have a result of the form *=/(4,ij, )
Substituting for a: and <fx as before, we arrive finally at the form
f& [MO fviKi)
,
F,({,,,0
(*)
It should be noticed that the limits x = Aj( n), x = A,(Z, 7), in (ix.), are
not necessarily different curves. They may, in some of the partial integrals,
be different portions of the same curve. This was exemplified in the last
integral of (2776).

MULTIPLE INTEGRALS.

The following theorems, (2825) to (2830), which are

given for three variables only, hold good for any number.
Let x, y, z be quantities which can take any positive values

MULTIPLE INTEGRALS.

417

subject to the condition that their sum is not greater than

unity; then
2825

rrr^^ = ^Tlt
JoJo Js
r(/+m+n+l)

Here *+2/+z = l is the hunting equation.

ProofIntegrate for z ; then for y by (2308) ; finally for x by (2280),
and change to the gamma fnnetion by (2305).

2826

flTf-,-f-<gd,d{ = ^

rfiWiSWi)
y kli

when ^^ +(^)+(~ ) = * *s ^e Smiting equation.

Peoop.Substitute x =

V = (-j)"' 2 = (^)'> and *PPlJ (2825).

2827 When the limiting equation is simply + + = A*,

the value of the last integral becomes
r(Qr(m)r() \
r(i+m++l)'
2828 The value of the same integral, taken between the
limits h and h-\-dh of the sum of the variables, is
r(Z) r(m) r(n) T(/+m+n)
Proof.Let u be the value in (2827) ; then, by Taylor's theorem, the
value required is
ddh = (l+m+n) V r(0r(m)r(n) dh}
dh
v
'
T(Z + w + n + l)
which reduces to the above, by (2288).

2829

j]J'"1y""1*""1/(a7+y+*)
rm r(m) r() p ( }
3 H

rfA

418

INTEGRAL CALCULUS.

if x+y + z = h and h varies from 0 to c. In other words, the

variables must take all positive values allowed by the condi
tion that their sum is not greater than c.
Proof.For each value of 7 tho integration with respect to x, y, z gives,
by (2828),
/(h) -' rffrfw,)Tfr dh,
the variations of a;, y, z not affecting h. This expression has then to be in
tegrated as a function of h from 0 to c.

2830

^'"^""^"'/[(-iT+^y+l-)! didl>di

f(h)h*'<*r ldh,
\p

r/

with the limiting equation ^^ -f- (^0 + ^^ = c.

Proof.From (2820) by substituting x = ^ J", &c.

2831 If
z be 7i variables, taking all positive values
subject to the restriction ar + if+z*+ ... > 1 ; then
dxdi/dz.&c.
7r4(n+1)
JJJ ^(l-^-^-^-Ac.)

2"r{i(+l)}'

But, if negative values of the variables are permitted, omit

the factor 2" in the denominator.
Proof.In (2830) pnt l = m = Sx.=l; a = /3 = &c.=l; jp = g = <fcc.=2;
f(h)
^ ; c=l; and the expression on the right becomes
{rq)tP a*-1 ,
2T(i)Jo(l_/0i
The integral is = B (*n, J) (2280) = g&ff ^ (\) (2305)'
Hence the result.
1 i t>l + Uf .
But if negative values of the variables are allowed under the same re
striction, .r" + y* + + ... 1, each element of the integral will occur 2* times
for once under the first hypothecs. Therefore the former result must be
multiplied by 2".

MULTIPLE INTEGRALS.

419

2832 If positive variables, x, y, z, &c, are limited by the

condition a!2-j-y2 + 22-|-&c.> 1, then
Iff
2-1r{i(n + l)} Jorv
where = a2 + b2 + c2 + &c.
Peoof.Change the variables to , n, by the
orthogonal transformation (1799), so that
a' + b'+c' + Ac. = **, and ax+by + cz + &c. = let.
The integral then takes the form
fCkOdtdidti.. with +
+ Ac. > 1.
fff
Now, integrate for 17, , Ac., considering con
stant, by adapting formula (282G). The limiting
equation is

aba
Jc k
Jc
a_ b' c
k Jc k
r'
a " b'I"

(v{tej)1+ ( 7ih^))'+ &c- io 71-1 topri.

Therefore put I = m = &c. = 1 ; p = q = Ac. = 2 ; o : /3 = Ac. = >/(l-P).
The result is

f f (*)
*r^""
iMi (-!) + !}
Jo
2"
which is equivalent to the value above.
2833 With the same limiting equation for n variables and
the same value of h,

JIf

Peoof.-"Making the same orthogonal transformation as in (2832), the

Considering constant, the integration for the remaining variables is effected

by (2830). Adapting the integral to that formula, we have
(*) r? ff
ah, dj...

for the limiting equation.

420

INTEGRAL CALCULUS.

Here I = to = &c. = 1 ; p = q &c. = 2 ; a = p = &c. = ^(1 ?) ;

f (/t) = .. ; c = 1 ; and the reductions are similar to those in (2832).
(1)
2834 If in (2832-3) negative values of the variables are
admitted (since the limiting equation is satisfied by such),
each element of the integral with respect to n, I, &c. will then
occur 2*_1 times, and therefore the result in each case must
be multiplied by 2*_1, and the limits of the integration for
will be 1 and 1 instead of 0 and 1.

EXPANSIONS OF FUNCTIONS IN CONVERGING

SERIES.
The expansion of a function by Maclaurin's theorem (1507)
can be at once effected if the nth derivative of the function is
known, or if merely the value of the same, when the inde
pendent variable vanishes, is known. Some nth derivatives of
different functions, in addition to those given at (1461-71),
are therefore here collected. When the general value would
be too complicated, the value for the origin zero alone
is given.
DERIVATIVES OF THE nth ORDER.
The following is a general formula for calculating the nth
derivative of a function of a function.
If y be a function of z, and z a function of x,

where r = 1, 2, 3, ... n successively, and a is put =z in each

term of the expanded binomial, after differentiation.
Am
A
A
Proof.Assume y= A,yx+ ^!/iz+
y* + + -g- yTo determine any coefficient Ar, form r equations from this by mating
y = z, z', z*, ... zr in succession : multiply these r equations respectively by
rz-\ -C (r, 2) z~\ G (r, 3) ", ... (-l)r*'z-r, and add the results. All the
coefficients excepting Ar disappear. This is shown by differentiating the
equation
0x)r = 1-rx + C (r, 2)x'-C(r, 3)x'+...xT

EXPANSIONS OF FUNCTIONS-

421

successively for x, and making x zero after each differentiation. Thus, finally,
As = ( xyijjgg? 0(r, 2)(z,)nx | C(r, 3)
^ 1)r>i (aQ j

with a put = z, after expanding and differentiating the binomial.

2853 Examples.The formula may be applied to verify
equations (1416-19).
Jacobi's formula (1471) may also be obtained by it.

2854
_

rf<+i) sin-1*

1.3... (2n-l)
2(1_^)(1-^)U

n
Z\ 1SC(n>2) yi
2n-l + 2-1.2n-3

1ff(!'3) *Z?+...Z\, where Z = \

2n-1.2^3.
2*i^5~
'
V
'
~ ~-l+f
\ 1.2 3.2
o
3
1
Proof.
(sin"1
= j (1-*)"* (l+a>)-*} (1434).
Expand the right member by (1460).
2855 tf^tan-1*.
This derivative is obtained in (1468).
The following is another method, which also includes the
result in (1469).

... by (1425)

(1).

Put x = cot 0, therefore * t = ^/(l + x') (cos 9 t sin 0), which values
substituted in (1) convert the equation, by (757), into
(tan"1 x) = (-1)"-1 |w-l sin" 0 sin n9.
2856

tf{e,,0C,acos(a?Sina)} = excoso cos (x sin a +na).

Proof.By Induction.

2857

LAGRANGE'S METHOD.
Lemma.The nth derivative of a function u=f(x)

will, by Taylor's theorem (1500), be equal to 1.2 ... n times

the coefficient of hn in the expansion of f(x-\- h) in powers of
h by any known method.

422

INTEGRAL CALCULUS.

Let u = a + bx + cx2, and therefore ux=b-\-2cx; then

drx(a + bx + cx2)n is equal to either of the following series,
with the notation of (2451-2).

2859 or, putting 2 = m and 4ac &z = <?a,

(r)

(2)

(2p)

-s

Proof.Changing x into x + h in w", it becomes (u+umh+ch*)*. Then,

by (2857), drxun will be = | r times the coefficient of hr in the expansion of
this trinomial. (2858) is the result, and it may be obtained by expanding
{(u + uxh) + ch'}" as a binomial, and collecting the coefficients of V from the
subsequent expansions. The value (2859) is found by taking

expanding, collecting coefficients of hr, and multiplying by 1.2... r, as

before.

2860
Ex.To find d{a,+t?)\ Applying formula (2859), we have
= '+', ux = 2x, q 2a, r = n. Therefore
n-i

In (2-l)"
(In 1)
+ M.'(n-l)'(n-2)(n-3)
aV-, + &c. 7 .
1.2.2... (2n-3)
T
j

,(2r)

2861

d^e" = e* 5a"(2lr)+... + ^-a-'-(2^)"-8'-+&c.L

with r = 1, 2, 3, &c. in succession.

Proof.By the method of (2857). Putting e0'***^ = e^ e2"1* e*', expand
the factors containing /t by (150), and from the product of the two series
collect the coefficients of hn.

2868 <.:-'=w>1(-'+d+...

EXPANSIONS OF FUNCTIONS.

423

Proof. d^ (cosa-H tain x1) = dxe^. Expand the right by (2861),

putting i*-r = e""""1', since, by (766), ev' = isin
= i. Also put
= cos J x>+ (n-r)* I + .g.n p+ (n-r) ir J >
and then equate real and imaginary parts.

2864

^-q^ (-l)n[e-*+{4-l-2',}e(-1)*
+ \0 (+l, 2)- 2" (n + l)+3|(''-a)*
+ jC(n + l, 3)-2"a( + l,2) + 3n( + l)-4''}e("-3>;t + &c.^j

Proof.Let m be the function. By differentiating u it is seen that

(e'+l)""tl = i,e" + i>.iel"-1+i,.ie"-,"+... + i1e',
the .A's being constants. To determine their values, expand = (e*+l)_1,
and also (e* + l)"*', by the Binomial theorem; thus
mm = (-1)" {e-*-2*e-tx+3ne-*x-4!ne-lx+&c.\,
(e'+l)"*^ e("*"I+(n+l)eM+C(+l,2)e<"-I>*+C(+l, 3) e("-J)* + &c.
From the product of the two expansions the coefficients An, An_lt &c. may
be selected.
2865

?xo tan"1* = ( 1) 2 | n 1 or zero,

according as n is odd or even.

Proof.By Rule IV. (1534). The first and last differential equations
(see Example 1535) are, in this case,
(l+3?)yu+2zyx = 0
(i.);
2/(n+2)*o+(+l)yo = 0
(ii.);
with j/jo = 1 and
= 0.
Otherwise.By (1468), putting as = 0.
2867

rf^osin-1* = 1.3s. 5*... (n-2)2 or zero,

according as n is odd or even.

Proof.By differentiating (1528).
Otherwise.As in (2865) where equations (i.) and (ii.) will become in this
case
(l-*a)2/2* = a:y*
()
y(n)xo = Vo ()

2869

rf,0 (sin"1 xf = 2.22.42.63 ... (n-2)2 or zero,

according as n is even or odd.

Proof.As in (2865) ; equations (i.) and (ii.) being identical with those
in (2867).

INTEGRAL CALCULUS.

424

2871

<**o cos (m sin-1 x)

= (-l)^m2(m2-23)(m2-42) ... [m2-(n-2)2],

or zero ; according as n is even or odd and > 0 if even.

2873

<xo sin (w sin-1 x)

= (-l)^m(m2-l)(m2-32) ... [m1 (n-2)*],

or zero ; according as is otfo? or eren and > 1 if odd.

2875
^k> cos (m cos-1.r)
= (-l)^m(m2-l)(TOs-32) ... [m*-(n-2)t] sin^,
2876

or

= (-l)^m2(m2-22)(m2-42) ... [m2-(n-2)!] cos^,

according as n is odd and > 1, or even and >0.
2877
rfruo sin (m cos-1 <r)
= (-l)^m2(m2-22)(m2-42) ... [m2-(n-2)2] sin i/mr,
2879

or

= ( lf^ nt (m* 1)(m*3) ... [m2-(n-2)!] cos>Mr,

according as . is even and > 0, or odd and > 1.
Observe that, in (2871-3), sin"1 0 = 0, and in (2875-9),
cos-1 0 = , are the only values admitted.
A
Proof.For (2871-9). As in (2865) ; equations (i.) and fii.) now be
coming in each case
(I-*1) yix-xyx+m'y = 0
(i.)
y(.o = (*-m') y^
(ii.)
Otherwise. By the method of (1533).
2880

Let y = x cot x, then

n 1
. n
,
,
r< /
\ ' nr
,
n cos ir = y0 sm ir + . . . +yrx0 C{n,r) sin
ir + . . .
+#(-d *o sin

with integral values of r, from 0 to n 1 inclusive.

EXPANSIONS OF FUNCTIONS.

425

2881 Thus, denoting yxa shortly by y, we find, by making

n= 1, 2, 3, &c. successively in the formula,
_
Vx-

2
_
3, y*-

8
_
32
128
15, 2/ - - 2T' y8_~l5"'

Proof.Take the nth derivative of the equation as cos a; = y sin as by

(14G0), reducing the coefficients by (1461-2), and putting as finally = 0.
2882 The derivatives of an odd order all vanish. This may be shown
independently, as follows :
Let y = f (as), then <j> (x) is an even function of as (1401) ; therefore
.: ftl (0) = -

2883

(0) ;

.-. fl (0) = 0.

dnx0{(l+a*f sin (m tan"1*)} = (-l)"r m'.*? or zero,

according as n is odd or even.

2885

^{(l+tfV cosCmtan"1*)} = ( 1)^~(.5 or zero,

according as n is even or odd.

Proof. As in (2865). Equations (i.) and (ii.), both for (2883) and
(2885), are now (l+as')i/te 2 (m 1) xyx + m (m-1) y = 0
(i.),
and
yin,nx> = (mn) (mn1) y^
(ii.)
Formula (ii.) gives the factors in succession, starting with y0 = 0, yx = m
in (2883) ; and with y0 = 1, yx = 0 in (2885).
2887

rfo{(l+**)

4 cos (m tan"1 a?)} = (-1)2 m(n> or zero,

according as n is even or otZcZ.

Proof.Change the sign of m in (2885).
Note.In formulfB (2883-7) zero is the only admitted value of tan"1 0.
2889

d-*>(?=i) = (-1)T_ljB- or zero'

according as n is
2891

or odd ; by (1539).

When p is a positive integer,

^(^"e^cosftj?) = nlpl(a2+b-) 2 cos Unp) tan1 f

or zero, according as n is > or < p.
Proof.Put j = e" cos bx and z = xv in (1460), employing (1465).
3 I

42 fl

INTEGRAL CALCULUS.

MISCELLANEOUS EXPANSIONS.
The following series are placed here for the sake of
reference, many of them being of use in evaluating definite
integrals by Eule V. (2249). Other series and methods of
expansion will be found in Articles (125-129), (149-159),
(248-295), (756-817), (1460), (1471-1472), (1500-1573).
For tests of convergency, see (239-247).
Numerous expansions may be obtained by differentiating
or integrating known series or their logarithms. These and
other methods are exemplified below.

2911

cot.r=l
x

+ 1
ir x
Tt-\-x

1
2tr x

+ _i
^2tt+x

L_ , _i
3it x^'Sn+x

&c.

Proof.By differentiating the logarithm of equation (815).

2912

7rcot = i- + -Lj+ 1
1
x
x
x+1^ x 2
,
1,1,1,0.
+ 7+2 + x=3 + 7+3+&C-

Proof. By changing x into *x in (2911).

2913

tan.r = TJ
\tr~x

_L_ + ^_
jpr+x %Ttx
1
+ -i
|ir+,r x

Proof.By changing x into

2914

cosec x = -i- -f- 1

x
itx

_J_ + &c.
fw+j?

x in (2911).

*
n-\-x

1,1
4
T 2ir+x T 3tt x

^
2irx
1
3it+x

1
U&C.
s
4irx^

Proof.By adding together equations (2911, 2913), and changing x

into \x.

EXPANSIONS OF FUNCTIONS IN SERIES.

2915

^--4
1
am vvi mm
sinmir
m ' 1 m

427

\-\-m

1
.
1
2m 1 2+m 1 3 m

-&c.
3+m

Peoop.By putting x = m* in (2914).

2916

ct* = A-^_^-^_&.
x
\2_
l
[6_

For proof see (1545).

be to (1541) not (1540).

The reference in that article (first edition) should

2917
tan, = *S^B++*<?-1)BS+ ^y-1? ^+&c.

2918

cosec, = 4-2?t> 1)/?2,

,
|_2_

Pboof.By (2916) and the relations

tan x = cot v 2 cot 2x,
cosec as = cot \x cot .

2919
=s
s = l+2a cos.r+2a2cos2,r+2a3cos3r+&c.
1 2acos,r+a2
2920

cos,
-, ,
1 2a cos,+

= r-+
r^5
1 a2 1
aa (cosx+a cos2cr+a2cos3,+a3cos4#+&c...)
2921
sin ,?
s= sin
1 2a cos,+'

sin2,+a2 sin3,+88in4,+&c-

Pboof.By (784-6) making a=j3 = x and c = a.

When a. is less than unity and either positive or negative,

INTEGRAL CALCULUS.

428

2922
1
a2
as
-3-log(l+2acoScr+az) = a cos.r g-co82#+ cos3ir &c.

2923

S3
tan-1
= a sin j?
sin 2d'+ -rr sin&r &c.
1+a cos a?
2
3

Peoof.Potting z = a (cos 3 + t sin as), we have

log (l + ) = log (1 + a cos z + ia sins)
= Hog (l + 2a cobs H-a')-H tan-'

asmx (2214),
1 + a cos as
z*- + &c.
4
imaginary parts.

z* +
z*
t
o
Substitute the valne of z and equate real and
2924 Otherwise.To obtain (2922),
log (1 + 2a cos 3 + a') = log (1 +ae**) +log (1 + cmT<x).
and also

log (1 + z) = s

Expanding by (154), the series is at once obtained by (768).

2925 OtJierwise.Integrate the equation in (786) with respect to a, after
changing a and (3 into x, and c into a.
2926 When a is greater than unity, put
log (l+2acos,r+aB) = loga2+log (l+2a_1 cos .r+a~2),
and the last term can be expanded in a converging series by
(2922).
2927
log 2 cos \x

cos<r ^cos2a'+3 cos3a? ^cos4r+&c.

2928
log2sin^<r = cos x |cos2cr ^cos3r jCOs4r &c.
2929

\x =

sin# |siu2cr+^sm3u,' sin4r+&c.

2930 \ {irx) =

sintr+i sin 2r+| sin3r+i sin4r+&c.

Proof.(2927-30) Make a= 1 in (2922-3).

2931
2932

\n = sin 0?+^ sin &r+y sin 5^+&c

*= 2v/2(l+|-l-|+i+-1V-&c).

Proof.Add together (2929-30), and put x=

When n is less than unity, and a = 1 + \/(l+'a),
2933

log(l+ncos.r) = log(l+2acoBJ?+a2) log (1+a2),

EXPANSIONS OF FUNCTIONS IN SERIES.

429

and is therefore equal to twice the series in (2922), minus

log(l-j-a2). But if a be greater than unity, expand, as in
(2926), by
2934

log(l+ncos.r)
= log (1+2C-1 cos *+a~2)+log a2-log (1+a2).

2935
(l+2a cos,r)* = A+A1 cosr+.42 cos2x-\-As cos3#+&c,
where
A = l + C(n, 2) 2a'+... + C(n, 2p) 0(2P, p)a+...,
Al = 2a\n+G(n, 3) 3a*+... + C(, 2p + l) C(2p + l,p) at>+...},
,
ana

a
Ar-i (nr+1) arAr
r+i
7;
ttx

(n+r+1) a

If to be a positive integer, the series terminates with the

71 + 1th term, and the values of A and Ax are also finite.
Proof.Differentiate the logarithm of the first equation ; multiply up
and equate coefficients of sin rz after transforming by (C66) ; thus Artl is
obtained.
To find A and Alt expand (1 + 2a cos x)n by the Binomial Theorem, and
the powers of cos a afterwards by (772).

LEGENDRE'S FUNCTION X.
2936
with

(l-2a.H-a2)-* = l+X1a+X2a2+...+Xa+
X" = !
2" \_n_ dxn (tf'-l)"
v
'

Peoof.Expand by the Binomial Theorem, and in the numerical part

of each coefficient of a" express 1.3.5 ... 2n 1 as | 2n -5- 2*
.

Consecutive functions are connected by the relation

2937

dxXn+1 = (2n+l) X+dxX_,

Proof.Differentiate the factor once under the sign of differentiation in

the values of -2Ctl and X_t given by the formula for Xn in (2936).
A differential equation for Xn is
2938

(1-^) t/StoXn-2o?^X.+n (n+1) X = 0.

430

2939

INTEGRAL CALCULUS.

When p is any positive integer,

lp+2l'+3p+... + (n-l)p
Bsn"s

]JLl\p+l
p+1

-&c.

2
2 p
p + \2 \p-l
\p-l

,4 |p-3^16 \p-5

concluding, according as n is even or odd, with

(-l)lP+1

or

1 e"*-l
Proof. e~
E 1
*
e*-l' Expand the left side by division, and
each term subsequently by (150). Again, expand the first factor of the right
side by (ISO), and the second by (1539), and equate the coefficients of in
the two results.
See (276) for the values of the series when p is 1, 2, 3, or 4. Bnt the
general formula there is incorrectly printed.
Let the series (2940-4) be denoted by Sin, S'in, s2n, s'in+1, as
under, n being any positive integer ; then
2i-i
2940 S^=l++ + +&c....:
\2n
2941

irnBia.

S^l-A + ^-A+&c....:
2n

2942

^sl + 4 + i + 4+&c....= ^^

Peoof.(i.) S2n is obtained in (1545).

(ii.) S2x-S-2n = 2 (A + JL +4o.) = 2^.

This give

(iiL) *. = i(<S+.)-

2943

, = l+^+-L+-L+&c. =-^^1"
y> JL

OQAA

'

=l__i

L_

wrf2 cot TTX

42M+1 i_2
J7

Peoof.By differentiating equation (2912) successively, and putting

X = \ in the result. To compute d cot xx, see ( 1525).

EXPANSIONS OF FUNCTIONS IN SEBIES.

431

2945 The following values have been calculated by formulas

(2940-4).
1T8
7T1
7T*
St =
s4=
*6'
90'
^-945'
9450'
IT1
s,
31
7T6
127^ .
7**
6
30240'
12'
720'
1209600 '
v
w4
17^
Si
*6
960'
*8
=
96'
8'
161280'
r
61tt7
*3 =
6
1536'
*7
T'
32'
184320'

<*0

!_cosa

\/

a>;L

(27r-a)5JL

(2^ + a)5J

x[1-(i^][1-(C$5p]
QQ4.7

cos^+cosa_ri
l + cosa
L

g' "1 fl
(-a)JL

1 l"l
(r + a)'JL

^ 1
(3T-a)'J

X[1-(^][1-(^7J-"&Ct,
A (a x)'-. Expand, the sines ,by
Peoof. cos x cos a = sin
.i (a+ .z)' sin
?-i
1 cos a
sin
(815). The two n+ 1th factors of the numerator divided by the corresponding
ones of the denominator reduce to
/.
2ax + x'\ /, laxx1 \
\
4nV-oV \ +4nV-a'/
= (1 ) (14^-Wi+
x U1-0-M
= (1
* ,Wi)
\
(2nir-a)V \
(2 + o)V
Similarly with (2947) employing (816).
2948 ..+to|..= (i+^-J

2949

(i+i-J

-$-..-(1-5) (1+j-J (i-^-J

V

Proof,

4ir a/ \

4jr + a/ \

ojt 0/

cos a; 4- tan sin a = 008

Expand the cosines by (816),
2
cos -fa
and reduce. Similarly with (2949), employing 815.

INTEGRAL CALCULUS.

432

2950

^-.[i+fcn[i+(n

2951

tif.-.

[i+(^)l[i+(g)'J[l+(|)]

Proof.Change B into ix in (815) and (816).

Often
2952

eT-2cosa + e"1
2(1 -cos a)

- [+ (f)"] [ + (^-J'J [1+

Onto
953

f1+ (fibA - *

e1 + 2cns8 + e"*
2(1+ cos a)

= M^S\
[terS] H^n
Proof.Change a into w; in (2946-7).

FORMULAE FOR THE EXPANSION OF FUNCTIONS

IN TRIGONOMETRICAL SERIES.
2955

When x has any value between I and I,

K) = g jj w

7*--"*(t ^^p^ * (*).

where n must have all positive integral values in succession

from 1 upwards
But, if x = l or Z, the left side becomes l<p(l)-\-l<j>( I).
Proof.By (2919) we have, when h is < 1,

, = l + 2/icos0 + 2/iJcos20+2/i*cos30+&c....
1 2Acos0+/ts
Put 0 = - (v

^ ; multiply each side by <p ()> and integrate for v from

l
I to I; then make h = 1. The left side becomes, by substituting = u x,
f1
()ffa
_ f'~" (l-ht)<t>(x + z)dz
I-1 l-2Acos!!l^^-)+A,

'-^(l-^' + ^sin5^

When ^ = 1 each element of the integral vanishes, excepting for values of v

which lie near to *. Therefore the only appreciable value of the integral
arises from such elements, and in these z will have values near to zero, both
positive and negative, since x has a fixed value between I and I. Let these
values of z range from /3 to a. Then between these small limits we shall
.
sin* jtz a-'z'
,
. , ,
., .
have
___= and <p (x+z) = <j> (x),

EXPANSIONS OF FUNCTIONS IN SERIES.

433

and the integral takes the form

when 7i is made equal to unity, which establishes the formula.

In the case, however, in which x=l, sins^ vanishes at both limits, that
is, when 2 = 0 and when Z 21. We have therefore to integrate for %
from ft to 0, and also from 21 to '21 + a, a and ft being any small
quantities. The first integration gives Up (I) as above, putting = 0. The
second integration, by substituting y = z + 2l, produces a similar form with
limits 0 to a, and with f (x 2l) in the place of <p (x) giving lip ( I) when
x = I. Thus the total value of the integral is l<p (I) + lip ( /). The result
is the same when x = I.
That the right side of equation (i.) forms a converging series appears by
integrating the general terms by Parts ; thus
f' , >
nir (vx) ,
I I . , s . Mir (v x) )
I (>) cos
civ = 'f M sin ^
'- \
J.,
4
t (.
I
) -i
4> \V) sm t
"i
J-i
1
which vanishes when n is infinite, provided <p' (o) is not infinite.
Hence the multiplication uf such terms by h" when it is infinite produces
no finite result when h is made = 1, although 1 is a factor of indeterminate
value.
2955 A function of the form <p(x) cos nx, with n infinitely
great, has been called " a fluctuating function," for the reason
that between any two finite limits of the variable x, the func
tion changes sign infinitely often, oscillating between the
values f[x) and ~^(x). The preceding demonstration shows
that the sum of all these values, as x varies continuously
between the assigned limits, is zero.
By similar reasoning, the two following equations are
obtained.
2956 If # has any value between 0 and I,
*{*) = ; fV 00 dv+ 1 2? fV () cos nw (v-'v) dv ... (2).
But if x = 0, write ^(0) on the left; and if x=l, write
If x has any value between 0 and I,
2957

0 = gj*() dv+ 1 srJV () cos ^iMf) <*,... (3).

3 k

TNTEGBAL CALCULUS.

434

But if x = 0, write ^(0) on the left; and if x = l, write

fr(Q.
2958

tf> &r) = j fV(r)

Io

y 2r cos^ fcos^fa) rfe

t
t Jo

This formula is true for any value of x between 0 and /,

both inclusive.
But if x be > 1, write <jt(x~2ml) instead of <f>(x) on the
left, where 2ml is that even multiple of I which is nearest to x
in value.
If the sign of x be changed on the right, the left side of
the equation remains unaltered in every case.
2959

<K*) = 42:siu^fsm^<Kr)rf0

(5).

This formula holds for anv value of x between 0 and I

exclusive of those values.
If x be > /, write +^(.r~2m/) instead of <j>(x) on the left,
or according as x is > or < 2ml, the even multiple of I
which is nearest to x in value.
But if * be 0 or I, or any multiple of I, the left side of this
equation vanishes.
If the sign of x be changed on the right, the left side is
numerically the same in every case, but of opposite sign.
Pboof.For (2058-9). To obtain (4) take the sum, and to obtain (5)
tnke the difference, of equations (2) and (8). To determine the values of
the series when a; is > /, put x. = '2nd /, so that x is < I.
Examples.
For all values of x, from 0 to n inclusive,
2960

a.= ^-l;cos^^+25f+&c.(.

Proof.In formula (4) put <f> (as) = j: and I = it, then

{* v cos nv av
, = rt>siniv ,h costm)!*
2=- or n<J,
, - I =
,
L
n- J0
n1
according as n is odd or even.
Similarly, by formula (5), equation (2929) is reproduced.
For all values of x, from \v to \tt inclusive.
2961

*=-}slnxw- + i

Pboof.Change x into \* x in (29G1).

&cj.

EXPANSIONS OF FUNCTIONS IN SEEIES.

9QR9
*y0'5

435

T eax~e~ax _ 8111 x
2 sin 2 v 3 ain 3.r

2" e^ ?-' ~ JT^+T ~

+ ir+S* ~
'

Proof. In formula (5) put <b (.c) = e"* e"ox and J = * ; then
["(e^-a-") sinWy = (-l)wt'W
Jo

" +M

2963 If <t> (%) be not a continuous function between x = 0

and = Z, let the function be <f>{x) from ,^ = 0 to =
and
*l>(x) from * =
to x = l; then, in formulae (4) and (5), we
shall have f(x) or
respectively on the left .side, according
to the situation of x between 0 and a, or between a and /.
But, if x a, we must wrrite i
+ 'AC")} for the left
member.
Proof.In ascertaining the value of tho integral in the demonstration
of (2955), we are only concerned with the form of the function cl we to tho
value of a: in question. Hence the result is not affected by tho discontinuity
unless x a. In this case the integration for z is from fi to 0 with <p (a;)
for the function, and from 0 to a with (x) for the function, producing
fc() + i*(a).
2964

Hence an expression

involving

series of sines of consecutive multiples of

in

an

infinito

may be found,

such that, when x lies between any of the assigned limits

(0 and a, a and
/; and c, ... k and /), the serie 5 sh ill be equal
respectively to the corresponding assigued functions
/.(*)> /,(*),
provided that the integrals

/.(*).../.(*),

can all be determined.

2965 The same is truj, reading cosine for sin-', throughout,
with the additional proviso [as appears from formula (4)] that
the integrals
( fx (x) da,
0
can also be determined.

\ ft (x) dx, ... \ fn(x)dx

Ja
Jk

2966 Ex. 1.To find in tho form of a series of cosines of multiples of

i a function of a; which shall be equal to the constants a, |j, or y, according
as z lies between 0 and a, u and b, or b and n.

INTEGRAL CALCULUS.

436

Formula (4) produces, putting Z = ir,

1 | f dx + P/3 dx+ i'ydx )
X Jo
va
lb
+ v"" - cos im J f a cos nx dx + f ft cos iix dx+ (" y cos iix dx\
*

tJi
JJ*
>
= - {o(a-/3) + t03-y) + Jry}
+ 2 j cosiii
it -' Jl

/5) sinna + (/3 y)siu n6 + y sin jiit}.

2967 Ex. 2.To find a function of x having the value c, when x lies
between 0 and a, and the value zero when x lies between a and J.
By formula (-1), we shall have
f
JiTry / x ,
f
nir ,
cZ . nira
COS - - A (t!) (tJ! = c I cos j- O* = sin ,
.0
Jo
since ^ (d) = c from 0 to a, and zero from a to I.
Iherefore <j> (x) = + -

< sin

cos + - sin cos

1 . 3a
8** . , 1
+. ynn-roo8-r+&cj.
When = a, the value is i 0 (ri)+0] = U, by the rale in (2963).
may be verified by putting a = 1 in (2'J23).

This

2968 Ex. 3.To find a function of a; which becomes equal to lex when
x lies between 0 and \l, and equal to k (l x) when x lies between |Z and I.
By formula (4),
f f () cos^- dv = [V kvcos^ d+ f * (l-*) cos
dv.
Jo
'Jo
'
Ji/
'
This reduces to lf. f 2 cos V cosnn- ^1 ) = 4H'
.. or U,
2
/
*V
77 II
according as n is, or is not, of the form 4m + 2. Also
f 0) dv = k \

v dv + Ic

APPROXIMATE
2991

Let

(l-v) dv = ;

INTEGRATION.

/ (,r)
be the integral, and let the curve
Ja
V =f(x) he drawn. By summing the areas of the trapezoids,
whose parallel sides are the n + 1 equidistant ordinates

AFl'llOXIMATE INTEGRA TION.

437

Vu> Vi> Dm we find, for a first approximation,

f/(.r) dx = t^ (y.+2y1+2^+...+2^.1+yB)

(i.)

SIMPSON'S METHOD.
2992 If V\ be the ordinate intermediate between y0 f(a)
and y2 = /(>), then, approximately,
| f{x) dx =

(y+fyi+yt)

(ii.)

Proof.Take n = 3 in formula (i.) ; write y} for ys, and suppose two

intermediate ordinat.es each equal to yv The area thus obtained is equal to
what it would be if the bounding curve were a parabola having for ordiuates
y<>> V\i ?/j parallel to its axis. Otherwise by Cotes's formula (2095).
2993 A closer approximation, in terms of 2n-\-\ equi
distant ordinates, is given by Simpson's formula,
)/&)

= i [y,+yto+4i (#1+//5+ ... +yta_i)

+2(y4+3f*+...+y*i-0]

(i")

Pkoof.Wo have
| / (z) da = JV CO * + ( "/ ()*>+... + \n / CO
n
M
Apply formula (ii.) to each integral and add the results, denoting by yr the
value of y corresponding to x = .
2994 When the limits are a and b, the integral can be
changed into another having the limits 0 and 1, by sub
stituting x = a + (b a) y.
COTES'S METHOD.
Let n equidistant ordinates, and
abscissas, be
Vo, y Hi-- - y-u V

the

corresponding

0, , ...
, 1.
fh fh
fh
A formula for approximation will then be

2995

and

f/(*) dv = Ayo+A1yl+... + Ary,.+ ... + Any

where

Ar = .(""^"^ f*
tte.
r I r Jo no.' r

(iv.),
(2452)

INTEGRAL CALCULUS.

438

Pkoof.The method consists in substituting for f (x) the integral

function
* (.) = (-1)- [("">" . ... + (-1/ yr ...
Lnx\n

(rix r) \ r \ r *

+ -!)" ;
U-y-l
(llX ll.) I H J
r taking all integral values from 0 to n inclusive. When x = r, we have
i^(r) = yr; so that vt (x) has n+1 values in common with / (x). The
approximate value of the integral is therefore I \p (x) dx, and may be written
hs in (iv.)
J'
By substituting 1 x, it appears that
J0 nx r
J0 >ix r)
and therefore
= -l_r. Consequently it is only necessary to calculate
half the number of coefficients in (iv.)
2996 The coefficients corresponding to the valnes of n from 1 to 10 are
as follows. Every number lias been carefully verified, and two misprints in
Bertrand corrected ; namely, 2089 for 2989 in line 8, and 89500 for 8'JoOO in
line 11.
n=l:
= 2:

A% ~ ~6'

_ 2
3'

n= 3 :

= A* = '

= A,=
*.
i
8
-A
16 '
" A> ~ 45

2
~ 15"

= 5: 4> -J - 19
~ 5 ~ 28a'

= 44 = -ii,
1
yo

25
= A = J4l'

~ 4 6 -ii
~ 840'

= A5 = A
ao'

= A,= 9 A - 34
20'

n = 4: A.
=

= 6:

n = 7: \

= 90 '

- 751
7~ 17280'

A - A.6 - 17280'
3"7

,
49
~ 5 ~ 040'
^ _ ,i _ 989
, *~A
_ A ~
_ 14175*
5248

2989
17280*

2944

.
454
^-~2885-

464

APPROXIMATE INTEGRATION.

a = Aa
a = 28-57
n = 9o : An
,
0
9 89600
. _ , _ 1209
3 ~ 6 ~ 5600'

.
1574.1
A.. = Af,=
, ,
1
8 89600
. _ . _ 2889
4 ~ 5 ~ 44800'

439

. = A. = 27
A.,
,, ,
2
7 224U

-in A0
i _
a ^j^gg,
16067 ^
j _ 4,
< _ lT96
26575
, _
- _ - jgj^,
1^175
-10:
- Aw-, ^a
- 4,
. _ . _ 5675
3
7~1*474'

. _ .
A*-A~

48-25
iiyjjH'

. _

17807
24948'

GAUSS'S METHOD.
2997 When / (a?) is an integral algebraic function of degree
2n, or lower, Gauss's formula of approximation is
(/(*) = ^^(*o) + ...+-4r/(*r) + ...+^(*-)

(V),

where xa ... xr ... xn are the n + 1 roots of the equation

+ (*)sf,.+.{^,(*-l)-+1}=0

(vi.),

and 4r = f' (Z^L Gr-^1)(.r-rr+1);.(*-*) ^

Jo (^r .r0) (*r J?r-i)(*r ^r+l) (^r~
_ . _ (yii.)
The formula is evidently applicable to a function of any
form which can be expanded in a converging algebraic series
not having a fractional index in the first 2n terms. The result
will be the approximate value of those terms.
Peoof.Let
^(x) = (x x0)(xxl) ... (xxn),
and let
/ (x) = Qxj, (x) + B
(viii.),
where f (x) is of the 2mth degree, Q of the n 1th, and B of the nth, since ^ (aj)
is of the n+ l"1 degree.
Then the method consists in choosing a function (x) of the n + l'h degree,
bo that j Q J< (x) dx shall vanish ; and a function B of the n,h degree, which
shall coincide with / (as) when x is any one of the n + 1 roots of

(z) = 0.

(i.) To ensnre that j Q \j/ (as) dx = 0. We have, by Parts, successively,

writing N for 4> (x), and with the notation of (2148),

= z| N-px'-l[ N+p(p-l)^ (x'~i^2xN)

=

&c.

&c.

= x" [ N-px"-1 f N+pip-^x*-^ N...\p f

(ix.)
Jx
J2
Jte
J(J>*1)
Now Qi// (x) is made up of terms like x" \p (x) with integral values of p from
0 to n 1 inclusive. Hence, if the value (vi.) be assumed for ^(x), wo

INTEGRAL CALCULUS.

440

see, by (ix.), that j" Q,^ (*) <&> wiH vanish at both limits, because the factors
x and x 1 will appear in every term.
(ii.) Let H be the function on the right of equation (v.) Then, when
x = xr, we see, by (vii.), that Ar = 1, and that the other coefficients all vanish.
Hence 11 becomes f (.r) whenever x is a root of \(/ (x) = 0.
The values of the constants corresponding to the first six values of n,
according to Bertrand, are as follows. The abscisses values, only, have been
recalculated by the author.
u = 0:

x, = -5,

A,= L

=1 :

x = -21132487,
x, = -78807513.

A = Al = 5,

log = 9-6989700 ;

n = 2:

x =-11270167, A0 = At = A.
x, = -5 ;
x, = -88720833, Al = t,

log = 9-4436975 ;

H = 3:

= 4:

= 5:

log = 9-6478175.

0 = -069-43184, An = ^s = 1739274, log = 9-2403681 ;

x, = -33000948, A, = A3 = 3260726, log = 9-5133143 ;
x, = -66999052 ;
xt = -93056916.
x0 =-04691008, An = . tj = 1184634, log = 9 0735834 ;
x, = -23076534, A, = At = 2393143, log = 9 3789087 ;
At = 2844444, log = 9 4539975 ;
x, = -5,
x, = -76923466;
xt = -95308992.
x =
x, =
, =
x, =
ar, =
x5 =

-03370524, Af, = As = 0856622,

-16939531, A = ^ = 1803808,
-38069041, At = As = 2339570,
-61930959 ;
-83060469 ;
-96623476.

log = 89327895 ;
log = 9 2561903 ;
log = 9 3091360 ;

As a criterion of the relative degrees of approximation obtained by the

foregoing methods, Bertrand gives the following values of
f !g(I+jQ
J.
1+*' dx = 48 logfa 2 = -2721982613.
Method of Trapezoids,
n = 10,
Simpson's method,
n = 10,
Cotes's

n= 5,
Gauss's

n= 4,
For other formulae of approximation,

-2712837.
"2722012.
-2722091.
-2721980.
see also p. 357.

CALCULUS

OF

VARIATIONS.

FUNCTIONS OF ONE INDEPENDENT VARIABLE.

3028 Let y =f(x), and let V be a known function of x, y,

and a certain number of the derivatives yx,
y^, &c. The
chief object of the Calculus of Variations is to find the form
of the function f(x) which will make
(i-)
a maximum or minimum. See (3084).
Denote yx, y^, y^, &c. by p, q, r, &c.
For a maximum or minimum value of TJ,BTJ must vanish.
To find BU, let By be the change in y caused by a change in
the form of the function y =f(x), and let Bp, Bq, &c. be the
consequent changes in p, q, &c.
Now,
p = yx.
Therefore the new value of p, when a change takes place in
the form of the function y, is

Similarly,

Bq = (Bp)x,
Br=(Bq)x, &c
(ii.)
Now BU= \XlBVdx (1483). Expand by Taylor's theorem,
rejecting the squares of By, Bp, Bq, &c, and we find

or, denoting Vy, Vp, Vt, ... by N, P, Q,

BU= r(NBy+PBp + QBq + ...) dx
3 L

(iii.)

442

CALCULUS OF VARIATIONS.

Integrate each term after the first by Parts, observing that by

(ii.) | Ip dx = Sy, &c, and repeat the process until the final
integrals involve hjdx.

Thus

^NZydx is unaltered,
PSpdx = Pfy-f P.Zydx,
f QBpdz = Q$p-QJ&y+^Qu$ydx,
Rlrdx = R$q-RJfy+RtJty f B^lydx,
ii
3029

Hence, collecting the coefficients of ly, $p, iq, &c,

8U= F1(N-Pm+Qu-R9m+...)*ydx

+8y, (P-Qx+Rj-...)1-By0(P-Qx+Rix-...\
+8pl(Q-Rx+S2x-...)1-Zp0(Q-Rx+Stx-...)0
+8q, (iJ_<Sj + Tto-...)l-89 (R-Sx + Ttx-...\+&c. (iv.)
The terms affected by the suffixes 1 and 0 must have *
made equal to sb, and x0 respectively after differentiation.
Observe that Px, Qx, &c. are here complete derivatives ;
y, p, q, r, &c, which they involve, being functions of x.
Equation (iv.) is written in the abbreviated form,
3030
W=(X>K8y&c+H1-H0
(v.)
The condition for the vanishing of SZ7, that is, for mini
mum value of 17, is
3031
3032

K=N-Px+Qix-R3x+&c. = 0
and

H^-H^-Q

(vi.),
(vii.)

Proof.For, if not, we mast have

J J0
that is, the integral of an arbitrary function (since y is arbitrary in form) can
be expressed in terms of the limits of y and its derivatives ; which is impos
sible. Therefore HiH0 = 0. Also K = 0 ; for, if the integral could vanish
without A' vanishing, theform of the function ly would be restricted, which is
inadmissible.

FUNCTIONS OF ONE INDEPENDENT VARIABLE.

443

The order of K is twice that of the highest derivative contained in V. Let

n be the order of K, then there will be 2n constants in the solution of equa
tion (vi.) and the same number of equations for determining them. For
there are 2n terms in equation (vii.) involving 3y &/0, dp,, &c. If any of
these qnantities are arbitrary, their coefficients must vanish in order that
equation (vii.) may hold ; and if any are not arbitrary, they will be fixed in
their values by given equations which, together with the equations fnrnished
by the coefficients which have to be equated to zero, will make up, in all, 2
equations.
PARTICULAR CASES.
3033 I.When V does not involve explicitly, a first
integral of the equation K = 0 can always be found. Thus,
if, for example,
a first integral will be
V=

Pp
+ Qp,-Q*p
+Rpt,Rxpx +Rixp
+ Spzx Sxp23,+S2xpxSSj.p+ c.

The order of this equation is less by one than that of (vi.)

Proof.We have
V,= Np+Pq+Qr+Bs.
Substitute the value of N from (vi.), and it will be found that each pair of
terms involving P, Q, It, &c. is an exact differential.
3034 ELWhen V does not involve y, a first integral can
be found at once, for then N = 0, and therefore K = 0, and
we have
and therefore

PKQiz-\-R3x&c. = 0 ;
PQ^+R^Sdc. = A.

3035

III.When V involves only y and p,

V=Pp+A,
by Case I.

3036

IY.When V involves only p and q,

V= Qq+Ap+B.

See also (3046).

Proof. K = P,+ Qu = 0, giving, by integration, P = Q,+A.

Also
Vm = Pq+Qr = Aq + Q^+Qr.
Integrating again, wo find V= Qq + Ap + B,
a reduction from the fourth to the Becond order of differential equations.

CALCULUS OF VARIATIONS.

444,

3037 Ex.To find the brachistochrone, or curve of quickest descent,

from a point 0 taken as origin to a point Xjj/U measuring the axis of y down
wards.*
Velocity at a depth y = -/2gy.
Therefore time of descent
Here

= [ '
jiff dx.
J * v 2yy

V = . /iE! =

E!__ +A, by Case IIL

By reduction, y (l+p*) = 2a, an arbitrary constant.

.A
That is, since p = tan 0, y = 2a cos1 0, the defining property of a cycloid
having its vertex downwards and a cusp at the origin
Hi -Si reduces to j {(pty\ (j%)J = 0.
If the extreme points are fixed, lyl and hj0 both vanish.
The valnes xu yv at the lower point, determine a.
Suppose xv but not y is fixed. Then 8y, is arbitrary ; therefore
its coefficient in (3) (PQx+Sx.)l must vanish; that is, (F,), = 0, or
J . P
|=0, therefore p, = 0, which means that the tangent at the
lower point is horizontal, and the curve is therefore a complete half cycloid.
3038 In the example of the brachistochrone, it is useful to notice that
(i.) If the extreme points are fixed, Sy fyt both vanish.
(ii.) If the tangents at the extreme points have fixed directions, !pv Spl
both vanish.
(iii.) If the curvature at each extremity is fixed in value, 2pv Sq Jp fgr,
all vanish.
(iv.) If the abscissae xa,
only have fixed valnes, fy, lyl are then
arbitrary, and their coefficients in Ht H0 must vanish.
3039

When the limits

S?7in(3029)

variable, add to the value of

V1dx1-V0dxu.

Peoof.The partial increment of U, due to changes in x\ and ar0, is

~d^+
dx, = Fl<b,-F,(fab.
ax^
tte0
3040

When

equations,

By (2253).

and yu x0 and y0 are connected by given

y1 = + (jbJ ,

EULE.Put
hi= {*'G*i) Pi} dart

and

y9 = x (<>)

8y0 = {x'('ro)~Po} d**,

The Calculus of Variations originated with this problem, proposed by John Bern ulli
in 1696.

FUNCTIONS OF ONE INDEPENDENT VARIABLE.

445

and afterwards equate to zero the coefficients of dx^ and dx0,

because the values of the latter are arbitrary.
Proof.y, + 2y, being a function of xv
therefore yl+i>1dxl =

(x^ dxu neglecting Spdxv

Ex.In the brachistochrone problem (3037), the result thus arrived at

signifies that the cycloid is at right angles to each of the given curves at its
extremities.

3041 If V involves the limits x0, xlf ?/0, yu P* Pit &c-> the
terms to be added to dU in (3029), on account of the varia
tion of any of these quantities, are
d*T {K+Vtipl+Vpiq1+...} dx

J Xq

In the last integral, hj0, iylt Sj>0, &c. may be placed outside
the symbol of integration since, they are not functions of x.
Hence, when Finvolves the limits x0, xu y0, yu p0, pu &c,
and those limits are variable, the complete expression for
SZ7is
3042 8*7 = (tl{N-Px+Q2x-R3x+&c.} Bydx
Jx
+ { Pi+ P( VXl+ V9lPl+
Jxa

...) dx} dxx

- { y>-S*\V*.+ VPo+ ^?o+ ...) dx} dx,

+ {(P-Q*+Rt*-...)i+^VSldx} 8y,
-{(P-Qx+Ru-...)o-rVVadx} Sy0
Jxt
%3Xx
-{(Q-R,+ Slx-...)0-i!*1Vr,dx}8pa + &c.
Hi

446

CALCULUS OF VABIATIONS.

3043 Also, if yi = \f> (ajj) and y0 = \ 0ro) be equations re

stricting the limits, put
tyi = W(xi)Pi} d*i

and

fyo = {%(?)&} &c*-

(3040)

The relation K = 0 is unaltered, and, by means of it, the

additional integrals which appear in the value of E^ H^
become definite functions of x.

3044 Ex.To find the curve of quickest descent of a particle from some
point on the curve ya= x(xa) * the carve y, = \ (,).
As in (3037), t =

* ; ("*' /
dr, F= a A1^, and contains y,
v/(2fi0 JxoV y-y.
Vy-y,
and ye. Equation (3042) now reduces to
SU = f"(N-P,) lydx + V1dxl-{Vt~\XlV^Podx} dm,

+ P1ty,-{P.-

(1)-

Now E"=0 gives N-P, = 0; therefore F = Pp-M (3035) ;

therefore

Clearing of fractions, and putting A =

^T!

JT +
, this becomes

(y-y.)(i+^) = 2a
AISO
Hence
therefore

P = V. = ,

(2).

r- = -7^

(3).

F = I; F* = -F, = -IT = -P. (by tf = 0),

f"V^ix = P.-P, = ^iEl

Substituting the values (2), (3), (4), in (1), the condition jET, H produ
(l+p\)dxl-(l+ptpl)dxt+pliyl-plSgt = 0.
Next, put for fy, and Bij0 the values in (3040) ; thus the equation becomes
{ 1 +jp, f (,) } dx, - { 1+p, x (*.) } dx, = 0
(5) ;
dz,,
being arbitrary, their coefficients must vanish ; therefore
Pit' Oi) = 1 and p, x' (0) = - 1
That is, the tangents of the given curves 4> and x at the points ^0 and :r,y,
are both perpendicular to the tangent of the brachistochrone at the point x,y,.
Equation (2) shews that the brachistochrone is a cycloid with a cusp at
the starting-point, since there y = y0, and therefore p = 00 .

FUNCTIONS OF ONE INDEPENDENT VARIABLE.

447

OTHER EXCEPTIONAL CASES.

(Continued from 303G.)

3045 V.Denoting y.y^y^ ... y by y,pltpt ...pn;

and
V VPx, Vp ... VPn by N, Pu Pt ... Pn;
let the first m of the quantities y,Pi,pi} &c. be wanting in the
function V; so that
V = f (x, pm, pm+l ... pn).
Then K= dmxPn-d{m+1)xPm+l + ... (-l)"-dL.P. = 0,
which equation, being integrated m times, becomes
*m

"'j.T^m+1-\-dixr +|

... (

lj

*(-i)ir-l

= c0 + c1a:+...+cll,_1asm-1

(i.),

a differential equation of the order 2 m.

3046

VI.Let x also be wanting in V, so that

then K = 0 is the same as before, and produces the same

differential equation (i.) From that equation take the value
of P, and substitute it in

Each pair of terms, such as Pm+2p+s digPm+ipm+l, is an

exact differential ; and we thus find
' C-\-Pm+lPm+l~r\Pm+tPm+i

dxr m+tPm+y T

+(P.l,-d,P.j..1+4.P.l'.-)- ... (-l)-"-1^-i)-P.JP-+i

+ J (c0 + Ci* + + Cm-iaj*-1) pm+1 dx.
The resulting equation will be of the order 2n m 1, or
m+1 degrees lower than the original equation.
3047 VII.If V. be a linear function of pH, that being the
highest derivative it contains, Pn will not then contain pn.
Therefore dnrPH will be, at most, of the order 2n 1. In
deed, in this case, the equation K = 0 cannot be of an order
higher than 2n 2.
(Jellett, p. 44.)

448

CALCULUS OF VARIATIONS.

3048 VIII.Let
volves ; then, if Pm
x and of derivatives
cannot generally be

pm be the lowest derivative which V in

= f(x), and if only the limiting values of
higher than the m"1 be given, the problem
solved.
(Jellett, p. 49.)

3049 IX.Let N = 0, and let the limiting values of x alone

be given ; then the equation K = 0 becomes
Px-Qix + RSx-&c. = 0,
or, by integration , P Qx + Bir &c. = c,
and the two conditions furnished by equating to zero the co
efficients of Sylt St/0, viz.,
(P-Q,-|-&c.)1 = 0,
(P-Q,+&o.), = 0,
are equivalent to the single equation c = 0, and therefore
2?! H0 = 0 supplies but 2?i 1 equations instead of 2n, and
the problem is indeterminate.

3050

Let Z7 =[*>&+ 7', where

V=F(x,y,p,q ...) and

V =/(*, s^, y0, yltpt,pu &c.)

The condition for a maximum or minimum value of U arising

from a variation in y, is, as before, K = 0; and the terms to
be added to S^H^ are
If the order of V be n, and the number of increments dx0, Sy0,
&c. be greater than n + 1, the number of independent incre
ments will exceed the number of arbitrary constants in K, and
no maximum or minimum can be found.
Generally, U does not in this case admit of a maximum
or minimum if either V or V contains either of the limiting
values of a derivative of an order = or > than that of the
highest derivative found in V.
(Jellett, p. 72.)

FUNCTIONS OF TWO DEPENDENT VARIABLES.

3051 Let V be a function of two dependent variables y, z,
and their derivatives with respect to x ; that is, let
V =f{x,y,p,q... z,p',q' ...)

(1),

FUNCTIONS OF TWO DEPENDENT VARIABLES.

449

where p, q, ... , as before, are the successive derivatives of y,

and p', q, ... those of z.
Then, if the forms of the functions y, z vary, the condition
for a maximum or minimum value of U or
Vdx is
W= (X\KSy+K'Sz) (Ix+Hl-Ho+H^-H^ 0 ... (2).

Here K', H' involve z, p', q',

precisely as K, H involve
y>P> ? the values of the latter being given in (3029).
3052 First, if y and <z are independent, equation (2) ne
cessitates the following conditions :
K = 0, K=0, Jlr1-Jlr,+jHrl'-J%'=0
(3).
The equations K = 0, K = 0 give y and z in terms of ,
and the constants which appear in the solution must be deter
mined by equating to zero the coefficients of the arbitrary
quantities

8y0, Bylf 8pa, 8^ ... Sz0, Szu o>0' tyi'

which are found in the equation

fl1-flo+flrl'-fl0'=0

(4).

3053 The number of equations so obtained is equal to the

number of constants to be determined.
Proof.Let
V = f (*, y, y yt. ... yx, ,**. *,)>
K is of order 2it in y, and .'. of form <f> (x, y,yx ... y2, z, z, ... *(-tn)X) ... (i.),
A" ia of order 2t in z, and .*. of form y (a;, y,yx ... f(m*u)m x> z* *ju>) (")
Differentiating (i.) 2i times, and (ii.) m + n times, Sm + n + 2 equations are
obtained, between which, if we eliminate z, zz ... z(3m^n)x, we get a resulting
equation in y, of order 2 (m + n), whose solution will therefore contain
2 (m + n) arbitrary constants. The equations for finding these are also
2 (m+n) in number, viz., 2it in T, H0 and 2w in Hi
d(J54 Note.The numberofequations fordetermining the constants is not
generally affected by any auxiliary equations introduced by restricting the
limits. For every such equation either removes a term from (4) by an
nulling some variation (fy, Sp, &c.), or it makes two terms into one ; in each
case diminishing by one the number of equations, and adding one equation,
namely itself.
3055

Secondly, let y and z be connected by some equation

3 M

450

CAL0ULU8 OF VARIATIONS.

4> (xyz) = 0. y and z are then found by solving simultane

ously the equations
<j> (x, y,z) = 0

and

K : <f> = K' : <f>z.

Proof <p (x, y, *) = 0, and therefore <p (x, y + 2y, z + iz) = 0, when the
forms of y and z vary. Therefore ^Sy + f^Sz = 0 (1514). Also KSy+K'&s = 0,
by ('2). Hence the proportion.

3056 Thirdly, let the equation connecting y and z be of the

more general form
<t>{*>y>p>q ... ,pyQ.' ...) = o
(5).
By differentiation, -we obtain

i,*y+tP$P+tM+ ... *>+tv+tv+ - = o...(6).

If (which rarely happens) this equation can be integrated so
as to furnish a value of Bz in terms of By, then Bp , Sg', &c.
may be obtained, by simple differentiation, in terms of By, 8p
Generally, we proceed as follows :
BV = My +PBp + QBq + ...+N'Bz+FBp' + Q'Bq' +
(7).
Multiply (6) by X, and add it to (7), thus
BV=(N+\<pl))By + (P+X^)Bp+...
...+(N'+\<pz)Bz+{P'+\fp)Bp' +

(8).

The expression for BJJ will therefore be the same as in (2), if

we replaced by N+\<f>y, P by P+\<ppi &c, thus
3057 S7= r*,[{(Ar+X^)-(P+X<fc));c+...} fy
J"
+{(2V'+Xk)-(P'+Xfc,),+ ...} SS]rf.r
+ {P+X^-(Q+X^)x+...}18y1
-{P+X^-(Q+X^)x+...j0S^0
+ {Q+\<f>q-(R+hf>r)x+...}1Sp1
-{Q+X^-(il+X^+...}0Si>0
&c.
&c.
+ similar terms in P, Q ...p, q ... &c. ...(0).
3058 To render BJJ independent of the variation Bz, we must

FUNCTIONS OF TWO DEPENDENT VARIABLES.

451

then equate to zero the coefficient of $z under the sign of

integration; thus

N'+\<j>z-(P'+HP)1,+(Q'+\4>q)2x~&c. = 0

(10),

the equation for determining A.

3059

Ex. (i.)Given V= F(x, y,p, q ... zf, where

z = vdx and v = F(x,y,p, q ...).
The equation <p is now 2J via = 0 or vzx = 0,
$p = V

f = * 1

.&

0? = "> &c->

pt = 0,
fV= li
0-= ^> tne re8' vanishing.
Substituting these values in (9), we obtain
iU= {"[{N+Xvi-iP+XvJ.+ iQ+XvJto- ...} $y + \N'+\x}&:]dx
+ \P+\v,-(Q+\vt),+...\llyl-{P+\v9-(Q+\vt)x+...),Syt
+ {Q+\vq-(R + \vr)x+...}1Spl-{Q + \vt-(R+\vr)x + ...}SFo+&c.
For the complete variation DJJ add Vldx1 V^dx0. To reduce the above so
as to remove $z, we must put JT+X,,= 0, and therefore X = J N'dx. Let
X = u be the solution, u being a function of x,y,p,q ... . Substituting this
expression for X, the value of IU becomes independent of fos.
Ex. (ii.)Similarly, if z in the last example be = J v (2148), <j> becomes
v Zf, = 0; and, to make N'+X^, vanish, we must put X = \'N*.
3061

Ex. (iii.)Let U= f Sl + y%+zxdx

Here N=0; N'=0;

P=

*V

(1).

; F =

P'

,; Q=0;

QJ = 0; and the equations K=0, K'=Q become

Fx = 0, K = 0, or

=a,

v/l+^+i."
Solving these equations, we get
yx = to ; 2X = n ; or y = mas + A ;

p'

=6.

l+^ + n'1
z = ?iz + B.

3062 -Ft'rstf, if * i/n z x y0, z be given, there are four equations to

determine m, n, A, and B.
This solves the problem, to find a line of minimum length on a given
carved surface between two fixed points on the surface.
3063

Secondly, if the limits 0,, x0 only are given, then the equations
(P), = 0, (P)=0, (!"), = 0, (!"),= 0,
are only equivalent to the two equations to = 0, n = 0, and -4 and 2? remain
undetermined.

CALCULUS OF VARIATIONS.

452

3064

Thirdly, let the limits be connected by the equations

yi> j) = . + (x y> ) = "We shall have
+ p, + ^ pi) dxl + 0,, y, + vfo z, = 0.
Substitute
= m,^, ^>Zi = n^,,, Pi m. Pi = * I thus
(l+mmT+MnJ da;, + m1fy, +
= 0.
Eliminate dxl by this equation from
.
FIb, + (P)Ity1 + (JOi*i = 0,
and equate to zero the coefficients of fy, and iz, ; then
= (P^l + ro^+nn,) ; n,^ =
+w1+nn1).
Replacing F ?! by their values, and solving these equations for m and n,
we find ?ii = -wi n = n,.
Similarly from the equation i/< ('0> Voi *o) = 0 we derive to =
n = nv
Eliminating xv y zlt z,, y0, z0 between these equations, and
yi = mx1+A; t = , + ?; y0=nix0 + ^l; 0=nx,+ B;
* (*u y *i) = 0 ; 4> (*, y0, z0) = 0 ;
four equations remain for determining to, n, A, and B.
3065 On determining the constants in the solution o/(3056).
Denoting p,q,r ... by j>i,^2,i?s ... , we have
V= F(x,y,ppt ...pn, z,p'u p'i ...p'm);
and for the limiting equation,
4> {#> i/> Pl> Pt> Pn > *> Pu Pi, P'm) = 0.
Vis of the order n in y and m in z.
<f> is of the order n in y and m' in z.
3066 Rule I.If m le > m', and n either > or < n', fAe
order of the final differential equation will he the greater of the
two quantities 2(m + n'), 2(m' + n); and there will be a
sufficient number of subordinate equations to determine the
arbitrary constants.
3067 R^e H.7/*m be < m', and n < n', the order of the
final equation will generally be 2(m'-f-n'); and its solution
may contain any number of constants not greater than the least
of the Uco quantities 2 (m' m), 2(n' n).
For the investigation, see Jellett, pp. 118127.
3068 If V does not involve x explicitly, a single integral of
order 2(m+n) 1 maybe found. The value of V is that
given in (3033), with corresponding terms derived from z.

FUNCTIONS OF TWO DEPENDENT VARIABLES.

453

Proof. dV= Ndy + P, dp, + ...+PndPn+N'dz + Pldp'1 + . . . + F, dp,.

Substitute for N and N' from the equations K= 0, .ff'= 0, as in (3033),
and integrate for V.

RELATIVE MAXIMA AND MINIMA.

3069

In this class of problems, a maximum or minimum

Cxi
value of an integral, Z7X =
V1dx, is required, subject to the
condition that another integral, JJ% =

Vtdxy involving the

0

same variables, has a constant value.

Rule.Find the maximum or minimum value of the func
tion Ui + a^2 1 that *s> toke V = V1 -f- &V9, and afterwards
determine the constant a by equating Ua to its given value.
For examples, see (3074), (3082).

GEOMETRICAL APPLICATIONS.
3070

Pboposition I.To find a curve s which will make

| F (x, y) ds a maximum or minimum, F being a given function

of the coordinates x, y.
The equation (5), in (3056), here becomes
where p = x p' = y * and y being the dependent variables, and s the in
dependent variable.
In (3057), we have now, writing u for F(x, y),
N = ux, N'=ur, f, = 2p, tr = 2p';
the rest zero. The equations of condition are therefore
I v,-d,(\x.) = 0 and u,d,(\y.) = 0
(1).
Multiplying by x y, respectively, adding and integrating, the result is
X = u,
the constant being zero.*
Substituting this value in equations (1), differentiating ux, and wj and
pntting , = uxx,+uy we get
= **u
= y
Bee Todhunter's " History," p. 406.

(2),
(3).

CALCULUS OF VARIATIONS.

454

Multiplying (2) by y , and (3) by as and subtracting, we obtain finally

(y.<*u z.Vu) = vxy.uvx or
3071
u = dud^_^,dx
(4)
p
dx ds
dy ds
P being the radius of curvature.
To integrate this equation, the form of u must be known,
and, by assigning different forms, various geometrical theorems
are obtained.
3072

Proposition II. To find the curve which will make

jF(x,y)ds + \f(x,y)dx

(1)

a maximum or minimum, the functions F and / being of given

form.
Let
F(z, y) = i and f(x,y} = v.
Equation (1) is equivalent to J(t+e,) ds.
In (3057) we now have V= u + vp ; and for <f>, p'+p'1 = 1, as in (3070).
Therefore
N = ux +pvx ;
P=Vr = v;
<t>,= 2p-,
N' = Uy+pVy ;
fp. = 2p' ; the rest zero.
Therefore, equating to zero the coefficients of x and By, the result is the two
equations
ux + pvx (v + \p), = 0,
ur+pv9(\p), = 0;
or d, (\x,) +v. = ux + x,vx,
d,m,)
=u + x,vv.
Multiplying by se y, respectively, adding, and integrating, we obtain, as in
{3070), X = u, and ultimately,
oflwo
1
1 (du dy
du dx
dv\
3U/d
p~
u\dx ds
dy ds^ dy)'
3074 Ex.To find a curve s of given length, such that the volume of the
solid of revolution which it generates about a given line may be a maximum.
Here /(y**, a')ds must be a maximum, by (3069), o' being the arbitraryconstant. The problem is a case of (3072),
u = a', ux = 0, J =0, v = y\ v = 2y.
1 = '
2vf.
Hence equation (3073) becomes
P
a
Giving p its value, Q
ppy
-1=

^ fwhere p =
and integrating, the result
\
dx)
w which ,=f y+y)^

FUNCTIONS OF TWO INDEPENDENT VARIABLES.

455

FUNCTIONS OF TWO INDEPENDENT VARIABLES.

3075

Let

V = f(x,y,z,p,q,r,s,t),

in which x, y are the independent variables, and p, q, r, s, t

Btand for zx, zy, z^, z^, zty respectively (1815), z being an in
determinate function of x and y.
Let

U\\ Vdocdy,
J xo vyo
and let the equation connecting x and y at the limits be
$ (#, y) = 0. The complete variation of U, arising solely from
an infinitesimal change in the form of the function z, is as
follows :
Let V Vp, &c. be denoted by Z, P, Q, B, S, T.
Let

<t> = {P-Bm-\Sy) Bz+^SBq+BBp,

i = (Q-Ty
Bz+iSBp + nq,
X=(Z-P-Qy + B2x+8xy+Tiy) Bz.
The variation in question is then

3076

W =(*,-

**

4>dy\
Peoof. 2 I ' PVcfocfy = |

xdxdy.

f^Fiajrfy

J *o J ifo

as appears by differentiating the values of f and \p.

by (2257), and

But

j*^ 3y =

Hence the result.

3077 The conditions for a maximum or minimum value of
17" are, by similar reasoning to that employed in (3032),
* = 0,

* = 0,

x = 0.

CALCULUS OF VARIATIONS.

456

GEOMETRICAL APPLICATIONS.
3078

Proposition I. To find the surface, S, which will

make jj^G1'. y> z) dS a maximum or minimum, .Fbeing a given

function of the coordinates x, y, z.
Here, putting F (x, y, z) = u,

[Jellett, p. 276.

V = u \/l +p* + <f ;

and F F Vt are all zero.

gg =
p (dt+P^.)+J1+ii)r-pr
dx
Ji+})* + g*\dz r dzl
(i+p + g>)S'
dQ_

Id*

du\

w(l^j=gg

The equation x = 0 or
P Q, = 0 gives
+ r-2pq,+ (l + ;>') <
1
/ du
(l+p' + g1)*
ni/l+p'+^V dx

du
H dy

du\ _ Q
dzl

If 22, B' be the principal radii of curvature, and I, m, n

the direction cosines of the normal, this equation may be
written
nsiMA
1 1 I d du ,
du ,
du\
n
It
11
u \ dx
dy
dz)
and according to the nature of the function u different
geometrical theorems may be deduced.

3080

Proposition II. To find the surface S which will make

|f F(x, y, z) dS+ [[/(as, y, z) dxdy

a maximum or minimum ; F and / being given functions of

the coordinates x, y, z.
Let F (x, y, z) = u and / (*, y, z) = v. Proceeding throughout as in
(3078), we have
V = w/l +p*+qt+v,
Z = /l+p* + qtuI+v
and the remaining equations the same as in that article if we add to the
resulting differential equation the term -*- on the left.

APPENDIX.

457

This equation may then be put in the form

nno-t
3081

1 d du ,
du , du
dv\
R1.1
+ R=-u{ld^
+ md^
+ n^-^>

where I, m, n are the direction cosines of the normal to the

surface.
3082 Ex.To find a surface of given area such that the volume contained
by it shall be a maximum.
By (3069), the integral [ [ (za</l+p' + q') dxdy
must, take a maximum or minimum value. The problem is a case of (3080).
We have
= a, v = z, = 0, , = 0, u, = 0, v, = 1 ;
and the differential equation of the surface (3081) reduces to
a

3083

or -|-4=4.

APPENDIX.
ON THE GENERAL OBJECT OF THE CALCULUS OP
VARIATIONS.

3084 Definitions.A function whose form is invariable is

called determinate, and one whose form is variable, indeter
minate.
Let du be the increment of a function u due to a change
in the magnitude of the independent variable, 8u that due to a
change in the form of the function, Du the total increment
from both causes ; then
Du = du+foi.
Thus, in (3042), the terms involving d\ and <e0 constitute du,
and the remaining terms du; the whole variation being Du.
lu is called the variation of the function u.
3085 A primitive indeterminate function, u, of any number
of variables is a function whose variation is of arbitrary but
constant form ; in other words, 0% = 0.
3 N

458

CALCULUS OF VABTATIONS.

3086 Let v F.u be a derived function,that is, a func

tion derived by some process from the function u ; F denoting
a relation between the forms, but not between the magnitudes,
of u and v.
The general object of the Calculus of Variations is to
determine the change in a derived function v, caused by a
change in the form of its primitive u.
The particular derived functions considered are those
whose symbols are d and j , denoting operations of differ
entiation and integration respectively.
SUCCESSIVE VARIATION.
3087 Let the variation of the variation, or second variation
of V due to a change in the form of the involved function,
y =f(x), be denoted by S (SF) or S2V; the third variation by
8s F, and so on.
By definition (3085), y being a primitive indeterminate
function, and By its variation, S2y = 0
(1).
3088 The second variation of any derivative of y is also
zero, i.e., 8*p, S2q, &c. all vanish.
Proof P(y) = S (lVnx) = {B (By)} = (^) = 0 by (1).
3089 If V=f(x, y, p, q, r, &c. ...), where y is a primitive
indeterminate function of as, then
8"V=(Sydy+8pdp+?><Jd+...)*V,
where, in the formal expansion by the multinomial theorem,
Sy, Sp, &c. follow the law of involution, but the indices of dr,
dpi &c. indicate repetition of the operation dt, dp, &c. upon V.
Proof.First,
SV= (Zyd,+$pdp+Sqdq+ ...) V.
In finding
each product, such as By drV, is differentiated again as a
function of y, p, q, &c. ; bnt, since the variations of By, Bp, &c. vanish by (2),
it is the same in effect as though By, Bp, &c. were not operated npon at all.
They accordingly rank as algebraic quantities merely, and therefore
PV= (Bydt + Bpd+Bqdt+...yV.
Similarly for a third differentiation ; and so on.
IMMEDIATE INTEGRABILITY OF THE FUNCTION V.
3090 Dep.When the function V (3028) is integrable
without assigning the value of y in terms of a;, and therefore

APPENDIX.

459

integrable whatever the form of the function y may be, it is

said to be immediately integrable, or integrable per se.
3091 The requisite condition for V to be immediately
integrable is that K = 0 shall be identically true.
Pboof.

Vdz must be expressiblo in the form

where ^ is independent of the form of y. Hence, a change in the form of y,

which leaves the values at the limits unaltered, will leave
fi
pi
3 Vdx = 0; that is,
KZy = 0.
J*,
J*.
But the last equation necessitates K = 0, since oy is arbitrary. And K=Q
must be identically true, otherwise it would determine y as a function of x.

DIFFERENTIAL

EQUATIONS.

GENERATION OF DIFFERENTIAL EQUATIONS.

3050 By differentiating ordinary algebraic equations, and
eliminating constants or functions, differential equations are
produced.
Some methods are illustrated in the following
examples.
3051 From an equation between two variables and n
arbitrary constants, to eliminate the constants.
Rule. Differentiate r times (r<n), and from the r + 1
equations any r constants may be eliminated, and thus C (n, r)
different equations of the rth order (3060) obtained, involving
drv dr-1v
-r-^,
tJlt &c. Only r + 1, however, of these equations will be
independent. By differentiating n times and eliminating the
constants, a single final differential equation of the nth order
free from constants may be obtained.
3052

Ex.To eliminate the constants a and b from the equation

y = ax* + bx

Differentiating, we find

-J^ 2ax + b

(i.)
(ii.)

Eliminating a and b in tnrn, we get

x *JL + bx = 2y, x^~ = aJ + y
(Hi., iv.)
dx
ax
Now, differentiating (iii.) and eliminating 6 produces the final equation of
the second order,
x>dl!L-2x^L+2y = Q
(v.)
dx
dx
The same equation is obtained by differentiating (iv.) and eliminating a.
3053 To eliminate the function <p from the equation z = f (v),
where v is a function of x and y. We have
zx = +'(v)vx,
z, = f
Therefore
zxi\ = zt vx.

GENERATION OF DIFFERENTIAL EQUATIONS.

461

3054 To eliminate f from u = f (v), where u and v are func

tions of x, y, z.
Consider x and y the independent variables, and differ
entiate for each separately, thus
ux+utzx = f'(v) (*.+.*.)

and, by division, 4>'(v) is eliminated.

3055

To eliminate flt f2> f* from the equation

F {*, y, z,
... *()} = o,
where au a2, ... a are known functions of x, y, z.
Rule.Differentiate for x and y as independent variables,
forming the derivatives of F of each order, up to the (2n l)th
in every possible way ; that is, F ; F*, F ; F^, Fry, F2y ; fyc.
There will be 2n2 unknown functions, consisting of fu f2, ... fn
and their derivatives, and 2n2 + n equations for eliminating
them.
3056 To eliminate
tf>i(S), ft{K), ... tf>() between the
equations
F{x,y,z,Z, ^(0,
...*.(*)} =0,
f{x,y,z,l, ^(5), ^(Q ...

= 0.

Role. Consider z and! functions of the independent

variables x, y, and /orm i&e derivatives of F and f up o Me
2n 1th order in the manner described in (3055). There will
be 4n2-f n functions, and 4n2 + 2n equations for eliminating
them.
3057

To eliminate f from the equation

F{x, y, z,w, f(a,p)} = 0,

where a, /3 are known functions of x, y, z, w.

Rule.Consider x, y, z the independent variables. Dif
ferentiate for each, and eliminate f, fa, f? between the four
equations.

462

DIFFERENTIAL EQUATIONS.

DEFINITIONS AND RULES.

3058 Ordinary differential equations involve the derivatives
of a single independent variable.
3059 Partial differential equations involve partial deriva
tives, and therefore two or more independent variables are
concerned.
3060 The order of a differential equation is the order of the
highest derivative which it contains.
3061 The degree of a differential equation is the power to
which the highest derivative is raised.
3062 -4 Linear differential equation is one in which the
derivatives are all involved in the first degree.
3063 The complete primitive of a differential equation is
that equation between the primitive variables from which the
differential equation may be obtained by the process of differ
entiation.
3064 The general solution is the name given to the complete
primitive when it has been obtained by solving the given
differential equation.
Thus, reverting to the example in (3051), equation (i.) is the complete
primitive of (v.) which is obtained from (i.) by differentiation and elimination.
The differential equation (v.) being given, the process is reversed.
Equations (iii.) and (iv.) are called the first integrals of (v.)> an(^ equation
(i.) the final integral or general solution.

3065 A. particular solution, or particular integral, of a

differential equation is obtained by giving particular values to
the arbitrary constants in the general solution.
For the definition of a singular solution, see (3068).
30S6 To find when two differential equations of the first
order have a common primitive.
Rule. Differentiate each equation, and eliminate its
arbitrary constant. The two results will agree if there is a
common primitive, which, in that case, will be found by elimi
nating yx between the given equations.
Ex.Apply the rale to equations (iii.) and (iv.) in (3052).

SINGULAR SOLUTIONS.

463

3067 To find when two solutions of a differential equation,

each involving an arbitrary constant, are equivalent.
Rule.Eliminate one of the variables. The other will also
disappear, and a relation between the arbitrary constants will
remain.
Otherwise, if V=C, v = c be the two solutions : Fand v
being functions of the variables, and G and c constants ; then
dV dv _ dV dv
dx dy
dy dx
is the required condition.
Proof. Fmust be a function of v.
and Vy= <j>vvr; then eliminate <j>.

Let V= <j> (v) ; therefore Vx = <pvv

Ex.tan-1 (x + y~) + tan-1 (as y) = a and as' + 2bx = y1 + 1 are both solu
tions of 2xyy, = sf+y'+l. Eliminating y, x disappears, and the resulting
equation is 6 tan a = 1.

SINGULAR SOLUTIONS.

3068 Definition. "A singular solution of a differential

equation is a relation between x and y which satisfies the
equation by means of the values which it gives to the differ
ential coefficients yx, y^, &c, but is not included in the com
plete primitive." See examples (3132-3).
3069 To find a singular solution from the complete primitive
* (*, V, c) = 0.
Rule I.From the complete primitive determine c as a
function ofx, by solving the equation yc = 0, or else by solving
Xc = 0, and substitute this value of c in the primitive. The
result is a singular solution, unless it can also be obtained by
giving to c a constant value in the primitive.
3070 If the singular solution involves y only, it results from
the equation yo = 0 only, and if it involves x only, it results
from xc = 0 only. If it involves both x and y, the two equa
tions xc = 0, yc = 0 give the rame result.

464

DIFFERENTIAL EQUATIONS.

3071 When the primitive equation <f> (xyc) = 0 is a rational

integral function, <f>0 = 0 may be used instead of x = 0 or
y = 0.
Proof.Let (ft (x, y, c) = 0 be expressed in the form
y = /(*,c)
(l).
Then, if c be constant,
yx=/
(2);
and, if c varies,
!/* =/x+/:C
(3).
When c is constant, the differential equation of which (1) is the primitive is
satisfied by the value of yx in (2). But it will also be satisfied by the same
value of yx when c is variable, provided that either / = 0 or fx = oo , and in
either case a solution is obtained which is not the result of giving to c a
constant value in the complete primitive; that is, it is a singular solution.
But /,. = 0 is equivalent to ye = 0, and fx = oo makes y, = oo , and therefore
x = constant.

GEOMETRICAL MEANING OF A SINGULAR SOLUTION.

3072 Since the process in Rule I. is identical with that

employed in finding the envelope of the series of curves
obtained by varying the parameter c in the equation
<f> (x, y, c) = 0 ; the singular solution so obtained is the equa
tion of the envelope itself.
An exception occurs when the envelope coincides with one
of the curves of the system.
3073

Ex.Let the complete primitive bo

y = cx+ vl <?,

therefore yc = x

; ye = 0 gives c= ,
vl c*
v-'l+sr'
Substituting this in the primitive gives y = v\+x%, a singular solution. I'
is the equation of the envelope of all the lines that are obtained by varying
the parameter c in the primitive ; for it is the equation of a circle, and the
primitive, by varying c, represents all lines which touch the circle. See also
(3132-3).
3074 "The determination of c as a function of x by the solution of the
equation ye = 0, is equivalent to determining what particnlar primitive has
contact with the envelop at that point of the latter which corresponds to a
given value of x.
"The elimination of c between a primitive y=/(*>c) and the derived
equation yr = 0, does not necessarily lead to a singular solution in the sense
above explained.
" For it is possible that the derived equation yt = 0 may neither, on the
one hand, enable us to determine c as a function of x, so leading to a singular
solution ; nor, on the other hand, as an absolute constant, so leading to a
particnlar primitive.
.

SINGULAR SOLUTIONS.

465

" Thus the particular primitive y = e" being given, the condition yc = 0
fives e" = 0, whence c is +oo if x be negative, and oo if x be positive.
t is a dependent constant. The resulting solution y = 0 does not then
represent an envelope of the curves of particular primitives, nor strictly one
of those curves. It represents a curve formed of branches from two of them.
It is most fitly characterised as a particular primitive marked by a singularity
in the mode of its derivation from the complete primitive."
. [Boole's "Differential Equations," Supplement, p. 13.
DETERMINATION OF A SINGULAR SOLUTION FROM THE
DIFFERENTIAL EQUATION.

3075 Rule II.Any relation is a singular solution which,

ivhile it satisfies the differential equation, either involves y and
makes py infinite, or involves x and makes I ] infinite.
3076 " One negative feature marks all the cases in which a solution
involving y satisfies the condition p = oo . It is, that the solution, while
expressed by a single equation, is not connected with the complete primitive
by a single and absolutely constant value of c.
" The relation which makes py infinite satisfies the differential equation
only because it satisfies the condition yc = 0, and this implies a connexion
between c and x, which is the ground of a real, though it may be unimportant,
singularity in the solution itself.
" In the first, or, as it might be termed, the envelope species of singular
solutions, c receives an infinite number of different values connected with the
value of x by a law. In the second, it receives a finite number of values also
connected with the values of * by a law. In the third species, it receives a
finite number of values, determinate, but not connected with the values of x."

Hence the general inclusive definition

3077 "-^ singular solution of a differential equation of the
first order is a solution the connexion of which with the com
plete primitive does not consist in giving to c a single constant
value absolutely independent of the value ofx."
[Boole's " Differential Equations," p. 163, and Supplement, p. 19.

RULES FOR DISCRIMINATING A SINGULAR SOLUTION OF

THE ENVELOPE SPECIES.

3078

Rule III. Wlien py or i \ is made infinite by

equating to zero a factor having a negative index, the solution

" may be considered to belong to the envelope species."
3079 "In other cases, the solution is deducible from the
3 o

466

DIFFERENTIAL EQUATIONS.

complete primitive by regarding c as a constant of multiple

value,its particular values being either, 1st, dependent in
some way on the value of x, or, 2ndly, independent of x, but
still such as to render the property a singular one."
[Boole's " Differential Equations," p. 164.
3080 Rule IV.A solution which, while it makes py infinite
and satisfies the differential equation of the first order, does not
satisfy all the higher differential equations obtained front it, is
a singular solution of the envelope species.
Ex. : yx = my " has the singular solution y = 0 when m is >1.
m-r
Now
yTX = in (m 1) ... (mr+l) y " ,
and, when r is > m, the value y = 0 makes yrx infinite. The solution is,
therefore, by the rule of the envelope species.
3081 Rule V." The proposed solution being represented by
u = 0, let the differential equation, transformed by making u
and x the variables, be ux+f(x, u) = 0. Determine the in
tegral f 4t as a function of x and u, in which U is either
Jo U
equal to f (x, u) or to f (x, u) deprived of any factor ichich
neither vanishes nor becomes infinite when u = 0. If that
integral tends to zero with u, the solution is singular" and of
the envelope species.
[Boole, Supplement, p. 30.
3082 Ex.To determine whether y = 0 is a singular solution or par
ticular integral of
yx =. y (log y)1.
Here u = y, and
dV
=
.
Joy (logy)'
logy
As this tends to zero with y, the solution is singular.
Verification.The complete primitive is y = ec~x, and no constant value
assigned to c will produce the result y = 0.
3083 Professor De Morgan has shown that any relatiou
involving both x and y, which satisfies the conditions pt oo ,
px = oo , will satisfy the differential equation when it does not
make y^, as derived from it, infinite ; that it may satisfy it
even if it makes y^ infinite ; and that, if it does not satisfy the
differential equation, the curve it represents is a locus of
points of infinite curvature, usually cusps, in the curves of
complete primitives.
[Boole, Supplement, p. 35.

FIRST ORDER LINEAR EQUATIONS.

467

FIRST ORDER LINEAR EQUATIONS.

3084

M+N-%L = 0,

or

Mdx+Ndy=0,

AT and N being either functions of * and y or constants.

SOLUTION BY SEPARATION OF THE VARIABLES.
3085 This method of solution, when practicable, is the
simplest, and is frequently involved in other methods.
Ex.

zy (1 + *') dy = (1 + y1) dx,

therefore

' = .f*
l+y%
x(l+x')'
and each member can be at once integrated.
HOMOGENEOUS EQUATIONS.
3086 Here If and N, in (3084), are homogeneous functions
of x and y, and the solution is affected as follows :
Rule.Put y = vx, and therefore dy = vdx+xdv, and
then separate the variables. For an example, see (3108).

3087

EXACT DIFFERENTIAL EQUATIONS.

Mdx-\-Ndy = 0 is an exact differential when

My = Nxi
and the solution is then obtained by the formula
$Mdje + ${N-dy{$Mdx)} dy = C.
Peoof.If V=0 be the primitive, we mast have VX = M, Vy = N;
therefore F = Mr = N^. Also V = | Mdx + <l> (y), <p (y) being a constant
with respect to *.
Therefore
N = V = d \ Mdz+f" (y),
therefore
<j> (y) = J {N d \Mdx] dy+O.
3088 Ex.
(x*-3x*y) dx + Of-x3) dy = 0.
Here Mr = Sx1 = N,. Therefore the solution is
0=

j" { f-J-d, (| -A) ] dy

468

DIFFEBENTIAL EQUATIONS.

3089 Observe that, if Mdx+Ndy can be separated into two

parts, so that one of them is an exact differential, the other
part must also be an exact differential in order that the whole
may be such.
3090 Also, if a function of x and y can be expressed as the
product of two factors, one of which is a function of the
integral of the other, the original function is an exact differ
ential.
3091

Ex.- 1 cos * dm- 4 cos * dy = cos^ V^-'dV = 0.

y
y
v
y
y
y
Here is the integral of the second factor. Hence the solution is
sin = G.
y

INTEGRATING FACTOR FOR Mdx + Ndy = 0.

When this equation is not an exact differential, a factor

which will make it such can be found in the following cases.
3092 I- When one only of the functions Mx + Ny or
Mx Ny vanishes identically, the reciprocal of the other is an
integrating factor.
3093 II.If, when Mx+Ny = 0 identically, the equation
is at the same time homogeneous, then x~(n+1) is also an in
tegrating factor.
3094

III.If neither Mx + Ny nor Mx Ny vanishes identi

cally, then, when the equation is homogeneous,

=^ is an

integrating factor ; and when the equation can be put in the

form 4>(xy)xdy + x(xy)ydx = 0, Mx_-^ is an integrating
factor.
Proof.I. and III.From the identity
Mdx + Ndy = J | (Mx + Ny) d log xy + (Mx - Ny) d logy | ,

FIRST ORDER LINEAR EQUATIONS.

469

assuming the integrating factor in each case, and deducing tho required
forms for M and N, employing (3090).
II.Putt>=-^-, M=xn<t>(v), N = xnip(v), and dy = xdv+vdx in
x
Mdx + Ndy and Mx + Ny.
3095 The general
Mdx+Ndy = 0 is

form

for an

integrating

factor

of

f
H e Nv*-Mv*
where v is some chosen function of x and y ; and the condition
for the existence of an integrating factor under that hypothesis
is that
jf
3096
tt must be a function of v.
NvxMvy
Peoof.The condition for an exact differential of Mfidx + Nfidy = 0 is
(Mfi)r = (Nfi)x (3087). Assume ft = <j> (v), and differentiate out; we thus
obtain

ft=L.
NvxMvy

The following are cases of importance.

3097 I-If an integrating factor is required which is a
function of x only, we put p = $ (x), that is, v = x ; and the
necessary condition becomes
must be a function of x only.

3098 n.If the integrating factor is to be a function of xy,

the condition becomes, by putting xy = v,
M N

3099

must be a function of xy only.

HI-If the integrating factor is to be a function of

, the condition is
x
A i-ty*. J^*) must be a function of $L.
Mx+Ny
J
J x
If Mx+Ny vanishes, (3092) must be resorted to.

DIFFERENTIAL EQUATIONS.

470

In this and similar cases, the expression found will be a

function of v =

if it takes the form F (v) when y is re33

placed by vx.
3100 IV". Theorem. The condition that the equation
Mdx+Ndy = 0 may have a homogeneous function of x and y
of the nth degree for an integrating factor, is

3101

The integrating factor will then be obtained from

_ JP(u)du
H = x*e

Proof.Put n=v = x*$ () in (3097), thus

1= M,-Nx
v
NvxMvy'
Perform the differentiations, and, by reduction, we got
V(u) _ J(N,-M,)+nN
i//(m)
Mx+Ny
The right member must be a function of u in order that ip () may be found
by integration.
3102 Ex.To ascertain whether an integrating factor, which is a homo
geneous function of x and y, exists for the equation
(y' + axy*)dy ayldx + (x+y)(xdyydx) = 0.
Here
M = (ay*+xy + y*),
N = (tf+axif + xy + x*).
Substituting in the formula of (3100), we find that, by choosing n = 3,
the fraction reduces to ax & t
, and, by putting y = ux, it becomes ,^tt,
^
y
a function of u.
fa
Su au
, =
a 30 ilog u,
I r
J vr
u
H = x 3e *

" = y-'e

the integrating factor required. It is homogeneous, and of the degree 3

in x and y, as is seen by expanding the second factor by (150).
3103 If by means of the integrating factor n the equation
fiM dx+ftNdy = 0 is found to have V=G for its complete
primitive, then the form for all other integrating factors will
be pf(V), where /is any arbitrary function.

FIRST ORDER LINEAR EQUATIONS.

471

Pkoop.The equation becomes

fiMf (V)dz+pNf(V) dy = 0.
Applying the test of integrability (3087), we have
W(F)},= W(F)}..
Differentiate out, remembering that
0.3Q, =
7, = /itf,
rx = ?M,
and the eqnality is established.
3104 General Rule.Ascertain by the determination of an
integrating factor that an equation is solvable, and then seek to
effect the solution in some more direct ivay.
SOME PARTICULAR EQUATION'S.
3105

(ax+by+c) dx+(a'x+b'y+c) dy = 0.

This equation may be solved in three ways.

I.Substitute

x = a,

y = j (3,

and determine a and /3 so that the constant terms in the new

equation in and i may vanish.
II.Or substitute ax + by + c = ,

a'x + b'y+c' = j?.

3106 But if a:a'=b: b', the methods I. and II. fail. The
equation may then be written as a function of ax + by.
Put z = ax -{-by, and substitute bdy = dzadx, and after
wards separate the variables x and z.
3107

HI-A third method consists in assuming

(Ar, + C)dt+(A'Z+C) dn = 0,

and equating coefficients with the original equation after sub

stituting

= x+m^y,

v x+viiy.

m1} ma are the roots of the quadratic

am2 + (b + a')m + b'= 0.
The solution then takes the form
i
| (amx a) (x-\- m, y) +
~e'\ ami~a'
;
i V{{am.ia')(x-\-iniy)-\-c)nic } a""-a'

DIFFERENTIAL EQUATIONS.

472

3108 Ex.
(3y-7x + 7)dx + (7y-3x + 3)Jy = 0.
Put x = 4 a, y = ij (3, thns
(3;-7) #+ (7.;-3) d*7 = 0
(i.),
with equations for a and /3, 7a 3/3 + 7 = 0; 3a 7/3 + 3 = 0;
therefore
a = 1, /3 = 0
(ii.)
(i.) being homogeneous, put i> = t>, and therefore dij = vd+$dv (3086) ;
/.

(7,-7){(K+(7o-3)?^ = 0,

or 4L+pp^dv = 0.

iir 7
The second member is integrated, as in (2080), with 6 = 0, and, after
reduction, we find
5 log(ij + ) + 2 log(ij ) = 0.
Putting = asI and ij = y, by (ii.) the complete solution is
(y + x-iy(y-z + iy= G.
3109 When P and Q are functions of only, the solution
of the equation
4L+Py = 0

is

y = CeS**

(i.)

by merely separating the variables.

3110 Secondly, the solution of
l!jL+Py=Q
is
y = c-$Pdx { C+jV^dr} .
This result is obtained by the method of variation of
parameters.
Rule.Assume equation (i.) to be the form of the solution,
considering the parameter C a function of x. Differentiate (i.)
on this hypothesis, and put the value o/yx so obtained in the
proposed equation to determine C.
Thus, differentiating (i.), we get yx = Cxe ^MxPy,
therefore

Q = Cxe~$

therefore

G = J QJ Pdx dx + C

Then substitute this expression for G in equation (i.).

Otherwise, writing the equation in the form (PyQ)dx+dy = 0, the
integrating factor J*** may be found by (3097).
3111
y,+Py=Qf
is reduced to the last case by dividing by yn and substituting
z = y1.

FIRST ORDER LINEAR EQUATIONS.

473

*3212
P^x+Ptdy+Q {xdy-ydx) = 0.
Plt Pa being homogeneous functions of x and y of the jjth
degree, and Q homogeneous and of the qth degree, is solved
by assuming

P, = ^(i),

P, = ^(f),

-*x({>

Put y = ox, and change the variables to x and v. The result

may be reduced to
dx
x+(v)
=
x()^-p+2
dv <j> (v) + v$ (v)
<j> (v) + vip (v) '
which is identical in form with (3211), and may be solved
accordingly.
3213

(4, + B,x + C,y) (x dy-ydx)- (A, + B%x + C,y) dy

-(A, + Bjs+Osy)dx = 0.
To solve this equation, put ,B = i + a,
y = ; + /3,
and determine a and /3 so that the coefficients may become homogeneous,
and the form of (3212) will be obtained.
RICCATI'S EQUATION.

3214

ux+bu2 = ctf

(A).

Substitute y = ux, and this equation is reduced to the

form of the following one, with n = m + 2 and a = 1. It is
solvable whenever m(2t + l) = At, t being 0 or a positive
integer.

3215

*y*-ay+W = "

(B).

I.This equation is solvable, when n = 2a, by substituting

y = vx", dividing by afa, and separating the variables. We
thus obtain

= xa~1dx.
c bv*
Integrating by (1937) or (1935), according as b and c in
equation (B) have the same or different signs, and eliminating
v by y = vx", we obtain the solution
2xaJ(bc)

3216

iWl^V

(1)'

* The preceding articles of this section are wrongly numbered. Each number and
reference to it, up to this point, should be increased by 100. The sheets were printed off
before the error was discovered.

3 p

474

DIFFERENTIAL EQUATIONS.

3217 or ^=^(_|)^anjC-*3^4=M]

(2).

3218 II. Equation (B) may also be solved whenever

n 2a'= t a positive integer
2n
Rule. Write z for y in equation (B), and nt+a for a
in the second term, and transpose b and c if t be odd.
Thus, we shall have
xzx (nt + a) z + bz- = cxn (when t is even)
(3),
xzx (nt + a) z + cz2 = bxH (when t is odd)
(4) .
Either of these equations can be solved as in case (I.), when
n = 2 (nt + a), that is, when -' t.

2 having been de-

All

termined by such a solution, the complete primitive of (B)

will be the continued fraction
a
x*
xn
xn
y =
b +" 2n+a ,

,(tl) n+a

,gv

c
b
"'
k
z "
where k stands for b or c according as t is odd or even.
3219 III. Equation (B) can also be solved whenever
- = t a positive integer. The method and result will be
the same as in Case II., if the sign of a be changed throughout
and the first fraction omitted from the value ofj. Thus
.?*
w"
a?
y ~~ na , 2na .
. (t l)n a , xn

Proof.Case II.In equation (B), substitute y = A H

, and equate
a
Vx
the absolute term to zero. This gives A = or 0.
Taking the first value, the transformed equation becomes
as-^ (n+a)y,+eyj = bx*.
Next, put y1 =

+ ', and so on. In this way the t transformed

c
y*
equation (3) or (4) is obtained with z written for the t^ substituted variable yt-

FIRST ORDER NON-LINFAR EQUATIONS.

475

Case III.Taking the second value, .4 = 0, the first transformed equa

tion differs from the above only in the sign of a ; and consequently the same
series of subsequent transformations arises, with a in the place of a. The
successive substitutions produce (5) and (6) in the respective cases for the
values of y.

3220

Ex.

,+ = -*.

(3214)

Putting = -*-,
ju=xy.-y
as
ax
x
and the equation is reduced to xyx y +1/ = ex* of the form (B). Here
si, 6 = 1, = !, and r*=2, Case HI.
By the rule in (3218),
changing the sign of a for Case III., equation (3) becomes
xzz-$z+z = ex*.
Solving as in Case I., we put a = vx$, &c. ; or, employing formula (1)
directly,
z = v/ca;*
1
i
v
Cee*x'1

and then, by (6), y =

1 . a*
3c + z

is the final solution.

FIRST ORDER NON-LINEAR EQUATIONS.

3221

Type

&)+*.(r+~+*+'.- m.
where the coefficients Pl5 P2, ... P may be functions of x
and y.
SOLUTION BY FACTORS.

3222

If (1) can be resolved into n equations,

Vxi>i = 0, yxpi = Q, ... y*p = 0

(2),

and if the complete primitives of these are

' i = Cu

I % = c^j ...

r = c

\y)>

then the complete primitive of the original equation will be

(Fl_c)(Fa_c) ... (F-e) = 0
(4).

DIFFERENTIAL EQUATIONS.

476

Proof.Taking n = 3, assume the last equation. Differentiate and

eliminate c. The result is
(Fs- vsy (7,- r,y (7,- 7l)WIdrl<iFi = o
(5).
By (2), dVx = /i, (y,
dx, Ac, where /i, is an integrating factor. Sub
stitute these values in (5), rejecting the factors which do not contain differ
ential coefficients, and the result is
(y.-jpi)(y*-j'0(y.-j') =
which is the differential equation (1).
3223 Ex.Given
yl+3y,+2 = 0.
The component equations are y, + 1 = 0 and yI + 2 = 0, giving for the
complete primitive
(y+x-e) (,, + 2x-c)=0.

3224

SOLUTION WITHOUT RESOLVING INTO FACTORS.

Class I.Type <j> (x, p) = 0.

When only is involved with p, and it is easier to solve

the equation for ,r than for p, proceed as follows.
Rule.Obtain x = f (p). Differentiate and eliminate dx
by means of dy = pdx. Integrate and eliminate p by means
of the original equation.
Similarly, when y = f(p), eliminate dy, &c.
3225

Ex.Given x = ay,+byl, i.e., x = ap + bp1

dx = a dp + 2bp dp,
therefore dy = pdx = apdp + 2bp*dp,

(1),

therefore

y=&+
+ C.

o
Eliminatingp between this equation and (l),the result is the complete primitive
(ax + 6by be)1 = (Gay 4c' ac) (a* + 4ibx).
3226

Class II. Type

(;>)+.#(/>) = x(p)Rule.Differentiate and eliminate y if necessary. Inte

grate and eliminate p by means of the original equation.
If the equation be first divided by i/< (p), the form is
simplified into
3227
y = *<Hp)+x(p)Differentiate, and an equation is obtained of the form
xp-\-Dx = Q, where P and Q are functions of p.
This may be solved by (3210), and p afterwards eliminated.

FIRST ORDER NON-LINEAR EQUATIONS.

477

3228 Otherwise, a differential equation may be formed

between y and p, instead of between x and p.
3229 Or, more generally, a differential equation may be
formed between x or y and t, any proposed function of p, after
whicb t must be eliminated to obtain the complete primitive.
3230 Clairaut's equation, which belongs to this class, is of
the form
y = px-\-f{p)
Rule.Differentiate, and two equations are obtained
(1)
px = 0, and.-. p = c;
(2)
x+f(j>) = 0.
Eliminate p from, the original equation by means of (1), and
again by means of (2). The first elimination gives y = cx-j-f (c),
the complete primitive. Hie second gives a singular solution.
Proof.For, if Rule I. (3169) be applied to the primitive y = cx+f (c),
we have x +/' (c) = 0 ; and to eliminate c between these equations is the
elimination directed above, c being merely written for p in the two equations.

3231

Ex.1.

y = px + x-/T+p~1.

This is of the form y = xfip), and therefore falls under (3227).

entiating, we obtain

x dp + dx \/l +p* +

f V

Differ

= o,

since dy=pdx; thus

in which the variables are separated.

Integrating by (1928), and eliminating p, we find for the complete
primitive as'+y* = Cx.

3232

Ex.2.

y=px+-/V-ay.

This is Clairaut's form (3230).

dx I

Differentiating, we have

v/(is-aV)i

The complete primitive is y = cx+ >/Qf aV) ;

and the elimination of p by the other equation gives for the singular solution
aV iV=oV, an hyperbola and the envelope of the lines obtained by
varying c in the complete primitive, which is the equation of a tangent.
3233 Ex. 3. To find a curve having the tangent intercepted between
the coordinate axes of constant length.

478

The differential equation which expresses this property is

p
y =p*+

r
Differentiating gives

(i)-

-^-[x+^ |=0

(2).

Eliminating p between (1) and (2) gives,

1st, the primitive
y = cx + aC
vl + c*
2nd, the singular solution x'+y* = *l

(3) ;
(4).

(3) is the equation of a straight line ; (4) is the envelope of the lines
obtained by varying the parameter c in equation (3).

3234

Class III.Homogeneous in x and y.

Type

p) = 0'

Rule.Ptit y = vx, and divide by xn.

Solve for p, and

eliminate p by differentiating y = vx ; or solve for v, and

eliminate v by putting v = *-\ and in either case separate the
variables.
3235 Ex.
y = pz + x\/l+p'.
Substitute y = vx, and therefore p = v + xvx. This gives v =p +
Eliminate p between the last two equations, and then separate the variables.
The result is

*L+^*l-0,
x
1 + ir
from which
x (u'+l) G or x* + y% = Cx.
The same equation is solved in (3131) in another way.

SOLUTION BY DIFFERENTIATION.
3236

To solve an equation of the form

F {< (<*> y> y*)> $ 0*. .*/> y*j) = 0-

Rule.Equate the functions <j> and \p respectively to arbi

trary constants a and b.
Differentiate each equation, and
eliminate the constants. If the results agree, there is a common

HIGHER ORDER LINEAR EQUATIONS.

479

primitive (8166), which may be found by eliminating yx between

the equations <(> = a, <p = b, and subsequently eliminating one
of the constants by means of the relation F (a, b) = 0.
Ex.
-n.+fW-iftid = o.
Here the two equations xyyx = a, f (y% y*y*x)
on applying the test, are found to have a common primitive.
eliminating y we obtain
f{y'-(x-ay} = b.
Also, by the given equation,
a + b = 0.
Hence the solution is
f{y'~ (s + i)*} = b.

Therefore,

HIGHER ORDER LINEAR EQUATIONS.

3237

Type g+P1!Q + ...+iW-!g-+P.y=0,

where Px ... P and Q are either functions of x or constants.

Lemma.If ylt y2, ... y be n different values of y in terms
of x, which satisfy (3237), when Q = 0, the solution in that
case will be

y C.y, + Cey2+ . . . + Cnyn.

Proof. Substitute y y ... yH in turn in the given equation. Multiply

the resulting equations by arbitrary constants, 0 C ... C respectively;
add, and equate coefficients of Pu P ... P with those in the original
equation.
LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS.
3238
3239

y*+aiy(n-i)*+...+a(-i)#x+a,,y = Q

(1).

Case L When Q = 0.

The roots of the auxiliary equation

w"+a1m*~1+ ...+a_1m+a = 0

(2)

being ni^, m?, . . . mn, the complete primitive of the differential

equation will be
3240
y = C1emi*+Ciemt*+... + Cnem*
(3).
If the auxiliary equation (2) has a pair of imaginary roots

480

DIFFERENTIAL EQUATIONS.

(aib), there will be in the value of y the corresponding

terms
3241
Aeax cos bx+ Be1" sin for
(4).
If any real root m' of equation (2) is repeated r times, the
corresponding part of the value of y will be
3242
(A,+Alx+At**+...+Ar_1af-1) **.
And if a pair of imaginary roots occurs r times, substitute
for A and B in (3241) similar polynomials of the r 1th degree
in x.
Peoof. (i.) Substituting y = Gemx in (1) as a particular solution, and
dividing by Cemx, the auxiliary equation is produced, the roots of which
furnish n particular solutions, y = Clem'x, y = C,em'x, <fec, and therefore, by
the preceding lemma, the general solution will be equation (2).
(ii.) The imaginary roots a ib give rise to the terms Ceax"bx + G,eax~'bx,
which, by the Exp. values (766), reduce to
(C+ C) eax cos bx + i (G- (T) e" sin 6*.
(iii.) If there are two equal roots m, = m,, put at first vu = h, + h. The
two terms C1e"'x+ Cielm>*h)x become em>x ((7, + C,e**). Expand ehx by (150),
and put G1 + C, = A, Cth = B in the limit when Zt=0, C, = oo, Cs = oo .
By repeating this process, in the case of r equal roots, we arrive at the form
(_A0+A,x + A^ + ...+Ar.1xr-i)e""x;
and similarly in the case of repeated pairs of imaginary roots.

3243 Case II. When Q in (3238) is a function ofx.

First method.By variation of parameters.
Putting Q = 0, as in Case L, let the complete primitive be
y = Aa + Bp + Cy + &c. to n terms
(6),
a, (3, y being functions of x of the form emx. The values of
the parameters A, B, C, ..., when Q has its proper value
assigned, are determined by the n equations
3244

Axa
A.*,
4r2*

+BX(3
+B.px
+ -&A*

+ to to terms = 0,
+

=0,
+

=0,

^*a(-l).i; + -Sx^(n-l)J.-+

= Q,

Ax, Bx, &c. being found from these equations, their integrals
must be substituted in (6) to form the complete primitive.

HIGHER ORDER LINEAR EQUATIONS.

481

Peoop.Differentiate (6) on the hypothesis that A, B, O^&c. are func

tions of x ; thus
yx = (Aaa+ B(ix+ ...) + (Aa+B,P+ ...).
Now, in addition to equation (1), n 1 relations may be assumed between
the n arbitrary parameters. Equate then the last term in brackets to zero,
and differentiate y, in all, n 1 times, equating to zero the second part of
each differentiation ; thus we obtain
yx
= Aax
+ B/3X
+&c. and Axa
+Bxfi
+&c. = 0,
i/u
= Aa^ +Bfiix
+&c. and Axax
+BXPX
+ <fcc. = 0,
J(.-i) = ^(-i)I + M-i)i + fc and Axain_Vx + Bx(ilH_1)x+&C. = 0.
The n quantities A B &c. are now determined by the n 1 equations on
the right and equation (1). For, differentiating the value of y(-i)X) we have
Vhx= {A"nx+BPxx+&c.} + {Axain_l)x+BxPin_i)x+&c.},

and if these values of yx, y^, ... yM be substituted in (1), it reduces to

4(-i)*+-S*A-i)+*c'=: Q>
for the other part vanishes by the hypothetical equation
since the values of yx, ... y(1,_i)OT and the first part of ynx, are the true values
in this equation.

3245 Case II.Second Method. Differentiate and eliminate

Q. The resulting equation can be solved as in Case I. Being
of a higher order, there will be additional constants which
may be eliminated by substituting the result in the given
equation.
3246 Ex.-Given
y.u-7yt + \2y = x
(1).
1st Method.Putting x = 0, the auxiliary equation is m' 7m + 12 = 0 ;
therefore m = 3 and 4. Hence the complete primitive ofyi,- 7y + 12y = 0
is
y = Atf+Be1'
(2).
The corrected values of A and B for the primitive of equation (1) are found
from
A,eu+ B.e*=0~)
.-. Az = -xe-3' and A= ^^le^+a.
3Axetx+4!Blce? = xy
d
B,= xe-u and 3= -^-^-"+6.
16

Substituting these values of A and B in (2), we find for its complete primitive
3247

y = oe^+6et*+^^-.
2nd Method.

yu-7y.+ 12y = x
3 Q

(1).

482

DIFFERENTIAL EQUATIONS.

Differentiating to eliminate the term on the right, we get

2/i,-7yte+12yto= 0.
The aux. equation is m* 7m* + VZm* = 0 ; therefore m = 4, 3, 0, 0.
Therefore
y = Aeu + Bf + Ox + D
(2) ;
yM = 4,Aet'+3Be''+C; ylz = 16Ae"+9Beu.
1 ; D = 7 ;
Substitute these values in (1) ; thus G =
\.6t
1 *1'J-'
x +
7 as before.
therefore, substituting in (2), y = Aeu+ Beu+

3248 When a particular integral of the linear equation

(3238) is known in the form y =f(x), the complete primitive
may be obtained by adding to y that value which it would
take if Q were zero.
Thus, in Ex. (3247), y = ^ + t^-t is a particular integral of (1) ; and
the complementary part Aeu+Be** is the value of?/ when the dexter is zero.

3249 The order of the linear equation (3238) may always

be depressed by unity if a particular integral of the same
equation, when Q = 0, be known.
Thus, if
yu+Piyu+Pty,+P>y = Q
(1),
and if y = z be a particular solution when Q = 0 ; let y = vz be the solution
of(l). Therefore, substituting in (1),
(^+P,*u+P,*-+i>1*)'+*c- = 0,
the unwritten terms containing v
and vb.
The coefficient of v vanishes, by hypothesis ; therefore, if we put = w,
we have an equation of the second order for determining u. u being found,
v = judx+O.
3250 The linear equation
(a+bxygn+A (a+te)-lylu_1)m+B {a+bx)T-*y{n_tim+...
...+%= Q,
where A, B, ... L are constants, and Q is a function of x, is
solved by substituting a + bx = el, changing the variable by
formula (1770),
t = log (a + bx).

and in

the

complete

primitive

putting

Otherwise, reduce to the form in (3446) by putting

a + bx = A, and solve as in that article.

HIGHER ORDER NON-LINEAR EQUATIONS.

483

HIGHER ORDER NON-LINEAR EQUATIONS.

3251

T9pe

SPECIAL FORMS.
3252

F(x, y, y(,+1)- yx) = 0.

When the dependent variable y is absent, and yrx is the

derivative of lowest order present, the equation may be de
pressed to the order n r by putting yrx = z. If the equation
in z can be solved, the complete primitive will then be
y= f z+i 0
J rx
J rx
3253

(2149).

F(y, yrx, ycr+D* ... y) = 0.

If # be absent instead of y, change the independent vari

able from x to y, and proceed as before.
Otherwise, change the independent variable to y, and
make p (= yx) the dependent variable.
For example, let the equation be of the form
3254
F(y,yy^ys,)=0
(1).
(i.) This may be changed into the form
F (y, x,
=0
by (1761, '63, and '66) ;
and the order may then be depressed to the 2nd by (3252). The solution
will thus give * in terms of y.
3255 (ii.) Otherwise, equation (1) may be changed at once into one of
the form
F (y, p, p ply) = 0, by (1764 and '67),
the order being here depressed from the 3rd to the 2nd. If the solntion of
this equation be p = <j> (y, c,, Cj), then, since dy =pdz, we get, for the com
plete primitive of (1),

3256

x= \

H-- + c,.

yx=F{x).

Integrate n times successively, thus

y = f F(x) + c,xn~l + Cix-2 +...+en.
J nx

DIFFERENTIAL EQUATIONS.

484

3257

y*=F(y).

Multiply by 2yx and integrate, thus

3258
y = ^k.-i),},
an equation between two successive derivatives.
Put y(n-\)X z> then zx = F(z), from which

*=\m+
If, after integrating, this equation can be solved for z so
that z = $ (x, c), we have y^-i)x = $ (x, c), which falls under
(3256).
3259 But if z cannot be expressed in terms of x, proceed as
follows :
y(-i). = *;

y{-) = \zdse = \4l$)z'

I* dz f dz
yi*->)a~ )F(z) )F(z)Zi
Finally,

y=

... fJ*L*;

the number of integrations and arbitrary constants introduced

being n1.
3260
y~
Put y(,_2)J. = z; then 2^ = ^(2), which is (3257), the
solution giving x in terms of z and two constants. If z can be
found from this in terms of x and the two constants, we get
*

or

yln-t)* = tfacuCt),

which may be solved by (3256).

3261 But if z cannot be expressed in terms of x, proceed as
in (3259).

3262

DEPRESSION OP ORDER BY UNITY.

When
F (4?, y, yx, y^, . . . ) = 0

is rendered homogeneous by considering

* y> yx, yix, y**, &c.

HIGHER ORDER NON-LINEAR EQUATIONS.

485

to be of the respective dimensions 1, 1, 0, 1, 2, &c. ; put

x = ee, y = ze, &c.
The transformed equation will contain the same power of e in
every term, and will reduce to the form
F(z, z zm ...) = 0,
the order of which is depressed by unity by putting zt = u.
3263 When the original equation is of the 2nd order, the
transformed equation in u and z may be obtained at once by
changing x, y, yx, y^ into 1, z,
u+mm,, respectively.
The solution is then completed, as in example (3264).
Pboof.We have
as = e* ; y = ze' ;
y =*.+; Urn = e-'
+
yto = e_!* (**.) ; and so on.
The dimensions of as, y, y &c. with respect to e, are 1, 1, 0, 1, 2, &c.
Therefore the same power of e* will occur in every term of the homogeneous
equation.
3264 Ex. :
= (y-xyxy.
Making the above substitutions for x, y, y and y2x, the equation becomes
*u +
Put z, = w ; thus
u'+u = ut = mm,,

= J-

therefore u'+l = u

fu

dz,

therefore taa~1u=az (1935), therefore z, = w = tan (oz),

therefore dd = cot (a z) dz, therefore 0 = log 6 sin (a z) (1941).
But
therefore

0 = log as and z =
bx = cosec

^-j,

or

foe = sec

^,

by altering the arbitrary constant.

3265

When

F(xty,yx,y^...)=zQ

is made homogeneous by considering x, y, yx, y^, &c. to be of

the respective dimensions 1, , n 1, n 2, &c. ; put
.r = e9,

y = se"",

and depress the order by putting z = u, as in (3262).

3266

When the original equation is of the 2nd order, the

DIFFERENTIAL EQUATIONS.

486

final equation between u and z may be obtained at once by

changing
v,y,yx>yz* into 1, z, u+nz, uuz+(2n 1) u+n (n 1) z,
respectively.
3267 Ex. :
V2xyx = xy
(1).
With the view of applying (3265), the assumed dimensions of each
member of this equation, being equated, give
n 2+ 1 = 1 + n,
therefore n = 4.
Thus as = e*; y = zeu ; yx = e3,(z.+4a); iJix = a(1.+7.+12).
Substituting in (1), e disappears ; and by putting z, = ,
= u the equa
tion is reduced to
(w'+4ttz) dtt+(7u,+40ttz + 48z,-) dz = 0,
which is linear and of the 1st order. This equation is also obtained at once
by the rule in (3266).
3268

When

F{x, y, yr, y*, ...) = 0

is homogeneous with respect to y, ya,

&c, put y = d*",
and remove e as before by division. The equation between u
and x will have its order less by unity than the order of F.
3269 Ex. :
yu+Py,+ Qy = o
P and Q being functions of x.

(l),

Here
y = J""- y = uy, yu =(,+') y.
Substituting, the equation becomes + + Pu+Q = 0, an equation of the
1st order. If the solution gives w = <p (x, c), then J <j> (x, c) dx = log y is
the complete primitive of (1).

EXACT DIFFERENTIAL EQUATIONS.

3270

Let

d U = $ (x, y, yx, y^, ... yj) dx = 0

be an exact differential equation of the n* order.

derivative involved will be of the 1st degree.

The highest

3271 Rule foe the Solution (Sarrus).Integrate the term

involving ynx with respect to y(n-i)x only, and call the result Uj.
Find dU1} considering both x and y as variables. dU dUj
ivill be an exact differential of the n 1th order.

MISCELLANEOUS METHODS.

487

Integrate this with respect to y(n_2)X only, calling the result

TJ2, and so on.
The first integral of the proposed equation will be
U = U1+U2+...+Un = C.
3272 Ex.: Let dU= {yt+(2xy-l)y'x+xylx+xiySx} dx = 0.
Here
TJX fy>x> dUx = (2a!yit+!Bsyto) da ;
dU-dU, = {y%+ (2^-1) y,-^} dx = 0;
.-. Ut = xy dUt = (yx+xylx)dx, dU-dUldU1=(y' + 2xyyx)dx;
therefore
TJs = xyi, and 17= as'ifa, ay^+a^' = 0
is the first integral.
3273 Denoting equation (3270) by dJJ - Vdx, the series of
steps in Rule (3171) involve and amount to the single condi
tion that the equation
with the notation in (3028), shall be identically true. This
then is the condition that V shall be an exact 1st differential.
3274 Similarly, the condition that V shall be an exact 2nd
differential is
P-2<k+3JR2x-4S3x+&c. = 0.
3275

The condition that V shall be an exact 3rd differential

is

Q_3#x+^S2,-^Ts,+&c. = 0,

and SO on.

[Euler, Comm. Petrop., Vol. viii.

MISCELLANEOUS METHODS.
3276
y*,+Pyx+Qifl = Q
where P and Q are functions of x only.
The solution is y = ^Pd" {2^Qe~^Pdxdx)-'dx.

0),

Proof. Pat JPix = z, and multiply (1) by z ; then, since zx = Pz,

Q*yl = 0.

Pnt zy, = u~\ .-. uu. = Qz'2,

.-. u = v/(2 \ Qz~2dx), &c.

3277
ytx+Qyl+R = 0
where Q and 72 are functions of y only.

(1),

DIFFERENTIAL EQUATIONS.

488

The solution is z = jJQ" (2^Be^Qd"dy)-idy.

Proof.Put J9* = z, and mnltiply (1) by z.
(zy*), = Rz,
zy* (zy,), = B**y
.: (ay.)' = 2jBztdy, &c.
3278

y*+Py*+Qyn* = 0,

where P, Q involve x only.

Put yx = z, and the form (3211) is arrived at.
3279
y^+Pyi+Qyl^O.
This reduces to the last case by changing the variable from x
toy by (1763).
3280

For a few cases in which the equation

yx+Py2+Qy+B = 0
can be integrated, see De Morgan's " Differential and Integral
Calculus," p. 690.
3281
yir ax+by.
Put ax + by = t (1762-3). Eesult tu = bt.
or (3257).
3282

Solve by (3239)

(l-xt)yix-xyx+qty = 0.

Put sin-1 a; = t, and obtain yu + q*y = 0.

Solution,
y = A cos (q sin-1 x) + B sin (q sin-1 x).
3283

(l+*>)yte+*y,gV = ^/(l+ax1) =

an(^ ^a^n yw?2y = 0 as above.

3284 Liouville's equation, y^+f^x) yx-\- F(y)yx=^ 0.

Suppress the last term. Obtain a first integral by (3209), and
vary the parameter. The complete primitive is
^F^dy=A]e-^)dxdx+B.

3285

Jacobi's theorem.If one of the first integrals of the

equation

yu =f(x, y)

is

yx = (ft (x, y, c)

(i., ii.),

APPROXIMATE SOLUTION.

489

the complete primitive will be

^<f>c(dy-(f>da?) = c.
PftoOF.Differentiating (ii.), we obtain <p, + $<pv = f
y), and differen
tiating this for c, <t>+<j>c<pv+(l><p,c = 0. But, by (3187), this is the condition
for ensuring that (f>cdy fc<l>dx = 0 shall be an exact differential ; therefore <j>c
is an integrating factor for equation (ii.), ym <j> (as, y, e) 0.
Equations involving the arc s, having given
3286

ds* = dtf + dif

3287

or

sx = Vl+fx.

* = ax+by.

Here \/l +yl = a + byx. Find yx from the quadratic equation.

3288

xu - a.

Change from s to x by (1763) ; .'. s^s^ = a, .'. s~2 = 2ax+ct

APPROXIMATE SOLUTION OF DIFFERENTIAL EQUATIONS

BY TAYLOR'S THEOREM.
3289 The following example will illustrate the method :
Given yix = xyx-{-y,
.-. yto = (xl+2) yx+xy.
Generally, let ynx = AHyx+By ;

y(n+1)x = AH+1yx+Bn+1y.

But, by differentiation,
yin+1)x= (Anx+A'n+B)yx+(AH+B'n)y,
.-. An+1= Ax+A'+Bn
But

and Bn+l= An+B'n.

At= x, jB2 = 1, .-. As=x>+2, Ba = x;

^, = ^ + 5*, i?, = a!a+3, &c.

Now, when x = a, let y = b and yx = p', then, by Taylor's

theorem (1500),
y = a+p (x-a) + (AiP+B2b) &^ + (Asp+B3b)k^f-

which converges when a; a is small.

3 R

+ &c,
[De Morgan, p. 692.

DIFFERENTIAL EQUATIONS.

490

SINGULAR SOLUTIONS OF HIGHER ORDER

EQUATIONS.

DERIVATION FROM THE COMPLETE PRIMITIVE.

3301

Let

y = <j> (*, y, yti y^ ... y(n-1)x)

(1)

be the differential equation, and let its complete primitive be

y=f(x,a,b,c ... s)

(2),

containing n arbitrary constants.

3302 Rule. To find the general singular solution of (1),
eliminate abc ... s between the equatiotis
y = f.

yi = fx>

and

ya = fx y(n- x = fcn-u x

f"a fAa

f(n-l)x

fh fhr
bx f|Abzx

tb(n-l)x
= 0

(3)

(4).

fAs fAsx xiz

f
f (n1) i
As
The result is a differential equation of the n 1th order, and
the integral of it , containing n 1 arbitrary constants, is the
singular solution.
Proof.Differentiate (2), considering the parameters a, b ... s variable,
thus y, = fI+fax+... +/,,. Therefore, as in (3171),
y, =/ if fa o.+ft, K+ .../, * = 0,
yu = fu if fax<**+fbxl>x+ /.**. = 0, as well;
and so on up to ?/ = /M. Eliminating a b ... s, between the n equations
on the right, the determinant equation (4) is produced with the rows and
columns interchanged.

3303

Ex. :

y-vy.+tfyu-fi,- (y.-*y*)' = 0

The complete primitive is

y = -5- + bx + a, + bi

From which
and the determinant equation is
i*' + 2a, m
x +2b, 1

yx = ax + b
=Q or

(*)
(2)(3)i

J+2lxz=2a
2

(4).

SINGULAR SOLUTIONS OF HIGHER ORDER EQUATIONS. 491

Eliminating a and 6 from (2), (3), and (4), we get the differential equation

the integral of which, and the singular solution of (1), is

(16y + 4aj'+ a;4) = x y/(l +
+ log {x +
+ z') } + 0.
[Boole, Sup., p. 49.
3304 Either of the two first integrals' (3064) of a second
order differential equation leads to the same singular solution
of that equation.
3305 The complete primitive of a singular first integral of
a differential equation of the second order is itself a singular
solution of that equation ; but a singular solution of a singular
first integral is not generally a solution of the original equa
tion.
Thus the singular first integral (5) of equation (1) in the
last example has the singular solution 1 6y + 4a? +aj* = 0, which
is not a solution of equation (1).

DERIVATION OF THE SINGULAR SOLUTION FROM THE

DIFFERENTIAL EQUATION.
3306 Rule.Assuming the same form (3173), a singular
solution of the first order of a differential equation of the nth
order xoill make

j ^""^
infinite ; a singular solution of the
d (V(n-l) x)

second order will make

and SO on.
3307

j (j^)
d(y(-i),)

j (7"^)
both infinite ;
d(y(n_2)I)
[Boole, Sup., p. 51.

Ex.Taking the differential equation (3303) again,

y-xy^^y^-y^-^xyu)1 = 0

(1),

and differentiating for yz and yu only,

{\x* + 1x (yMzyb)-2y} d(ylx)-{x+2(yz-xylx)}d(yx) = 0.
The condition ^S?"? = 00 requires
<* (y.)%xi+2x (yvyu)tyu = 0.
Substituting the value of ylx obtained from this in equation (1), and rejecting
the factor (sc'+l), the same singular integral as before is produced (3303,
equation 5).

DIFFERENTIAL EQUATIONS.

492

EQUATIONS WITH MORE THAN TWO VARIABLES.

3320

Pdx+Qdy+Rdz = 0

(1).

P, Q, B being here functions of x, y, z, the condition that this

equation may be an exact differential of a single complete
primitive is
3321

P (Q-RJ+Q (Rx-P,)+R (P,-Q.) = 0.

Proof.Let /* be an integrating factor of Pdx + Qdy + Rdz = 0. Then

fiPdx = ftQdy + fiRdz, and, by (3187), for an exact differential, we must
have (nQ)z =
Write this symmetrically for P, Q, and R, differentiate
out, and add the three equations after multiplying them respectively by P,
Q, and R.
To find the single complete primitive of equation (1).
3322 Rule.Consider one of the variables z constant, and
therefore dz = 0. Integrate, and add <j> (z) for the constant of
integration. Then differentiate for x, y, and z, and compare
with the given equation (1). If a primitive exists, ^(z) will be
determined in terms of z only by means of preceding equations.
The complete primitive so obtained is the equation of a
system of surfaces, all of the same species, varying in position
according to the value assigned to the arbitrary constant.
3323 Ex.: (x-3y-z)dx + (2y-3x)dy+(z-x)dz = 0
(1).
Condition (3321) is satisfied ; therefore, putting dz = 0, we have
(x3y-)dx + (2y-3x)dy = 0.
Applying (3187), Mr = 3 = N and integration gives
\x* 3xyzx+y'+f () = 0.
Differentiating now for x, y, and z,
(z3yz)dx + (2y - 3x) dy + {f' (z) - x } dz = 0.
Equating coefficients with (1), $'(z)=z, therefore ^(z) = J*'+(7.
Hence the single complete primitive is
xi+2y'+zt-6xy-2zx = C,
the equation of a system of surfaces obtained by varying the constant G.
3324 When the equation Pdx + Qdy+Rdz = 0 is homo
geneous, put x = uz, y = vz. The result, when the coefficient
of dz vanishes, is of the form
3325

Mdu+Ndv = 0,

EQUATIONS WITH MOBE THAN TWO VARIABLES.

solvable by (3184).

Otherwise it is of the form

3326

= Mdu + Ndv,
z

493

and the right will be an exact differential if a complete primi

tive exists.
3327 Ex.:
yzdx + zxdy + xydz = 0
Condition (3321) is satisfied. Putting
x = uz, y=vz, dx = udz + zdu, dy = vdz + zdv,
(1) becomes
and the solution is

(1).

+ ^ + -f* = 0,
log (zm'd*) = G or xxjz = 0.

When the equation

Pdx+Qdy+Rdz = 0
has no single primitive:

(1)

3328

(2)

Rule.Assume <j> (x, y, z) = 0

and differentiate ; thus

tf>xdx+4>ydy+^dz = 0

(3).

The form of <f> being given, eliminate z and dz from (1) by

(2) and (3). The result, being of the form
Mdx+Ndy = 0,
can be integrated, and the solution taken with (2) constitutes a
solution of equation (1), and represents a system of lines (by
varying the constant of integration) drawn on the surface
*0,y,z) = 0.
3329 Ex.:
(l + 2m)xdx + (l-<B)ydy + edz = 0.
The condition (3321) not being satisfied, assume x' + y'+z* = i3 as the
function <f>, therefore xdx + ydy + zdz = 0 ; and by eliminating z and dz,
2mdxydy = 0, the integration of which gives y'4mx = O, a cylindrical
surface intersecting the spherical surface in a system of curves (by varying
0), whose projections on the plane of xy are parabolas.
The condition that
3330
Xdx+ Ydy+Zdz+Tdt = 0,
where X, Y, Z, T are functions of x, y, z, t, may be an exact

DIFFERENTIAL EQUATIONS.

494

differential, may be shewn, in a manner similar to that of (3321) ,

to be expressed by any three of the equations
3331

Y(Zt -T.)+Z(T,-Yt) + T(Y,-Z,) = 0,

Z(Tr-Xt) + T(Xz-ZI)+X(Zt-T2)=0,
T (X,- Yx) +X(Yt -T,) + Y(Tr-Xt) = 0,
X{YS-ZV) + Y(Z,-X,)+Z(X,- Yt) = 0,

the fourth being always deducible from the other equations.

If this condition is fulfilled, the solution of equation (3330)
is analogous to (3322).
Integrate as if z and t were constant, and therefore dz and
(// zero, adding for the constant of integration <t> (z, t).
Differentiate next for all the variables, and determine $ by
comparison with the original equation.
3332 If a single primitive does not exist, the solution must
be expressed by simultaneous equations in a manner similar
to that of (3328).

SIMULTANEOUS EQUATIONS WITH ONE

INDEPENDENT VARIABLE.
GENERAL THEORY.
3340

Let the first of n equations between n + 1 variables be

Pdx+P1dy+Ptdz+ ...+Pndw = 0

(1),

where P, Px ... P may be functions of all the variables.

Let x be the independent variable. The solution depends
upon a single differential equation of the nth order between
two variables.
Solving the n equations for the ratios dx :dy:dz: &c, let
dx

"

Ql
dx' ~ Q '

di/

dz _

Al. Q
dx~ Q'

__ dw

^il
dx ~ Q '

Differentiate the first of these equations n 1 times, substi

tuting from the others the values of zx ... wx, and the result

SIMULTANEOUS EQUATIONS.

495

is n equations in yx, yu ... y^, and the primitive variables

x,y,z ... w.
EUminate all the variables but x and y, and let the differ
ential equation obtained be
F(x,y,yx ... yj) = 0.
Find the n first integrals of this, each of the form
F (e, y,yx ... //(_!) x) = G, and substitute in them the values of
y*> y*r> ynx> m terms of x, y, z ... w, found by solving the n
equations last mentioned.
Thus a system of n primitives
is obtained, each of the form F{x,y,z ... w) = G.
3341 The same in the case of three variables.
Here % = 2. Let the given equations be
P1 dx+ Q.dy+R.dz^O,
Pidx+Qidy+Ridz = 0.
inerelore QA_QA~ B.P.-PA ~
From these let
Therefore

yx = i> (a, y, z),

y^ = <j>x +

zx ^ (x, y, z).
yx + $zzx.

Substitute the value of zx, and eliminate z by means

Vx = $ (*> 2/> z)- An equation of the 2nd order in x, y, yx,
is the result.
Let the complete primitive of this
y = X (*> a> fyThen we also have <j> (x, y, z) = dx% (x, a,
These two equations form the complete solution.

of
y^
be
b).

FIRST ORDER LINEAR SIMULTANEOUS EQUATIONS WITH

CONSTANT COEFFICIENTS.
3343 In equations of this class, the coefficients of the
dependent variables are constants, but any function of the
independent variable may exist in a separate term.
Such equations may be solved by the method of (3340),
but more practically by indeterminate multipliers.

3344

Ex. (1):

^+7x-y = 0,

^+2x + 5y = 0.

Multiply the second equation by m and add. The result may be written
feJS)+<fc+0{.+ ^lfJ_,
(,).

DIFFERENTIAL EQUATIONS.

496

To make the whole expression an exact differential, put

= in.

This

2m + 7
1 + m / = -
m=
-,

gives

fo\
(2);

(1) now becomes

ii^+2Hl + (2m + 7)(x+my) = 0,

at
and the solution is x + my s ce-'2"*7" and x+m'y = ee-****7".
Solving these equations, and substituting the values (2),
iy = c-(,"')(-c'e-(8 i)t= e-6'|(c-c') coat-i(c + c) sin*J,
* = "" { (+<!) -<+ (^-"-t1') H
or y = e-"(Ccosi-C'sinO,

= |e-"{(C+C) cos< + (0-C) sin t}.

3345 Ex. 2 :
x, + 5.e + jr = b*, y, + 3y-x = e".
Multiply the second equation by in, and add to the first
^; + (0 m) -j a; +
^ = e' + me.
at
C
5m )
Put V7" ^m = m, thus determining two values of m, and put x + my = z ; thus
Z( + (5_,,) , - e'+me". This is of the form (3210).
Note.The equations of this example, written in the symmetrical form
of (3342), would be

3346

-j^5
= a ^ o = *
e' oxy e" + x3y

General solution by indeterminate multipliers.

T ,

dx

dy

dz

be given with
Pi = (hx+biy + CiZ+di,
Pt = aixJrhy->rciz-{-di,
P3 = (hx + b3y + csz + ds.
Assume a third variable t and indeterminate multipliers I, to, n such that
dt
Idx + m dy +ndz
ldx-\-mdy + ndz
(1).
t ~ Wl+mPt + nPl~ \(lz + my + nz + r)'
The last fraction is an exact differential, and, to determine X, I, m, n, r,
we have
a1l + a1m + ain XI,
6, l+btm+btn = Xm,
= 0.
it
X
Zi,
Cji + c,w + c,n = Xm,
ci
c,
c,X
(Z^ + rfjTO + djM = Xr,

SIMULTANEOUS EQUATIONS.

497

The determinant is the eliminant of the first three equations in I, m, n.

The roots of this cubic in X furnish three sets of values of I, m, n, r, which,
being substituted in the integral of (1), give rise to three equations involving
three arbitrary constants ; thus,
X
i
c,< = (\x + rn^y + Wjz + r,)\
ett = (Z^c + m,y + + r,)Aj,

Eliminating t, we find for the solution two equations involving two

arbitrary constants.
A similar solution may be obtained when there are more than three
variables.
3347

Toso.vep^p = ;p^L_i = ^? = &c....(l),

where

P = ax+by+ c,

Assume

p = at+bt)+c(,

Px = a^x + 1$ + cx,
= a^ + h-fl + c^,

&c.

Ac.,

and take

=
(2),
Pi
Pi
P
the solution of which is known by (3346). Substitute = x{, q = 7/f, and
these equations become
xd +dx _ ydZ + jdy _ d!_
Pi
Pt
~ P '
and therefore
=
= d!.
Piap Ptyp p
Dividing numerators and denominators by , the first equation in (1) is pro
duced, and therefore its solution is obtained by changing , 77 in the solution
of (2) into xZ and y(.
Certain simultaneous equations in which the coefficients
are not constants may be solved by the method of multipliers.
Thus,
3348 Ex. (1): xt+P(ax + by) = Q, yt+P (a'x + l'y) - B,
P, Q, B being functions of t. Multiply the second equation by m, add, and
determine m as in (3344). The solution is obtained from
x+my = e-^'^C+^e^^iQ+mB) dt\,
with two values of m.
3349

(3210)

Bx.(2): xt+^(x-y) = 1, yt+ (x + 5y) = t

are equations solvable in a similar manner, and the results are

[Boole, p. 307.
3 S

498

DIFFERENTIAL EQUATIONS.

REDUCTION OP ORDER IN SIMULTANEOUS EQUATIONS.

3350 Theorem.n simultaneous equations of any orders

between n dependent variables and 1 independent variable are
reducible to a system of equations of the first order by sub
stituting a new variable for every derivative except the
highest.
3351 The number of equations and dependent variables in
the transformed system will be equal to the sum of the indices
of order of the highest derivatives. This will, therefore, in
general be the number of constants introduced in integrating
those equations. If, after integrating, all the new variables
be eliminated, there will remain n equations in the original
variables and the above-named constants. These equations
form the complete solution.
In practice, such reduction is unnecessary. The following
are methods frequently adopted :
3352 Rule I.Differentiate until by elimination of a vari~
able and its derivatives an equation of a higher order in one
dependent variable only is obtained.
3353 Rule II.Employ indeterminate multipliers.
3354 Ex. (1) :
x = ax + by, yu = ax + b'y.
By Rnle I., differentiating twice for t and eliminating y and yu, wo obtain
xlt(a + b') x2t+ (ab'a'b) x = 0,
which may be solved by (3239).
Otherwise by Rule II., exactly as in (3344), we find
am' + (a b')mb = 0,
and for the exact differential
(as + my)u = (a + ma) (x + my),
the solution of which, by (3239), is
x+my = Gie^a'ma'), + Ote--/iatma')t
in duplicate with the two values of m.
3355 Ex. (2) : Xjt - 2ayt + bx = 0, yn + 2axt + by = 0.
Differentiate, and eliminate y, yt, yu i thus
xu + 2 (2a' + b) xlt + Vx = 0,
and solve by (3239). Otherwise assume
x = I cos at + 1) sin at,
y = ijcosoi , sin at,
and the given equations reduce to

{ = -(+&) 4,
which are solved in (3257).

,M = -(a'+6)n,
[Boole, p. 311.

PARTIAL DIFFERENTIAL EQUATIONS.

499

3356 Ex. (3).Let = 0, v = 0, w = 0 bo three equations in x, y, z, t,

involving derivatives of t up to xlt, ytt, z1t.
To obtain an equation between x and t. Differentiate each equation
6+ 7 = 13 times, producing 3 + 13 x 3 = 42 equations involving derivatives
of t up to xw, yw, zMt. Between these 42 equations eliminate y, yt, ... yltt,
z, tf, ... z10t, in all 41 quantities, and an equation of the 16th order in x and t
is the result.
[De Morgan.
3357 If a number of equations involve the quantities x, afe,
xu, &c, yt, ySf, y6t, &c, all in the first degree, these quantities
may be eliminated by assuming
x = L smpt,

y = M cos pt.

3358 If there be n linear homogeneous equations in n vari

ables se,y,z, ... and their derivatives of the 2nd order only,
the equations may be solved by putting
x = L srnpt,

y = Msinpt,

3359 Ex.:
Putting

xn=zax + by,
x L einpt,
(a+p>)L + bM=0~>
gL + (f+p>)M = 0S'
p and the ratios L : M are thus found.
Suppose
L=kb and
then
x= kb sin pt,
and k and t are arbitrary constants.

z = Nsinpt,

&c.

ytt = 9^+fyy = M ninpt,

. a+p\ b
_ fl
"
g, f+p>
'
M=k(pi + a),

y = k (p1 + a) Binpt,

PARTIAL DIFFERENTIAL EQUATIONS.

3380 An equation is termed a general primitive or a com

plete primitive of a partial differential equation, according as
the latter is obtained from it by eliminating arbitrary functions
or arbitrary constants, as illustrated in (3150-7).
LINEAR FIRST ORDER P. D. EQUATIONS.
3381 To form the P. D. equation from the primitive
u = <f> (r), where u and v are functions of x, y, z.

DIFFER ENTIAL EQUATIONS.

500

Kule.Differentiate for x and y in turn, and eliminate

f(v). See (3054).
Otherwise.Differentiate the equations u = a, v = b; thus
^dx+Uydy+Ujdz = 0,
vIdx+vydy+vIdz = 0.
TKerefore

f = |=f,

*~

*=

Then the P. D. equation trill be

Pzx + Qzy = R.
Proof.Since z is a fanction of a and y, zJLe+Zfdy = dz. But dx = kP,
dy = kQ, dz = kB, therefore kPzt+kQzr = kB.
3382

Ex.The general equation of a conical surface drawn through the

point (a, b, c) is

UlZ^. =(k (--\

xa
\xal
the form of <p being arbitrary.
Considering z as a function of two independent variables x and y, differ
entiate for x and y in turn, and eliminate <f> as in (3154). The result is the
partial differential equation
(xa) z, + (yb) zy+ec = 0.
3383

To obtain the complete primitive; that is, to solve

the P. D. equation,

Pzx+Qzv = B,

P, Q, B being either functions of as, y, z or constants.

Rule.Solve the equations
dx
dy _ dz
P ~ Q ~ R'
Let the two integrals obtained be
then

u = a,

v = b;

u = <f> (v)

will be the complete primitive.

Propositions (3381) and (3383) extended to any number
of variables.
3384 To form the partial differential equation from the
primitive

<f>(u,v, ... w) = 0

(1),

where u, v, ... w are n given functions of n independent vari

ables x, y, ... 2 and one dependent t.

PARTIAL DIFFERENTIAL EQUATIONS.

501

Rule.Differentiate for all the variables thus,

fu du+^vdv+ ... +<t>v&w = 0

(2).

Therefore, since <j> is arbitrary, du,dv ... dw must separately

vanish, giving rise to the n equations
du = uxdx + uvdy-f-... +utdt = 0,
dv = v1dx4-vydy+...+Ytdt = 0,
dw = wxdx+wydy + ... +wtdt = 0.
Solving these for the ratios, by (583), we get
dx

dy

dz

dt

/Qx

P, Q ... B, S being functions of the variables or else constants.

Noio, t being a function of all the rest,
t1dx+tydy+... +tzdz = dt

(4),

therefore, by (3) and (4), the partial differential equation

required is
3385

Ptx+ Qty+ ...-rRte = S.

3386 If u,v ... w be n functions of n variables, x,y ... t, the

condition of interdependence of the functions or existence of
some relation expressed by equation (1) is / (, 0 ... to) = 0
(see 1606) ; that is, the eliminant of equations (2) must vanish.

3387
tion

Conversely, to integrate the partial differential equa

Ptw+Qt+...+Rts = S

(1).

Rule.Solve the system of ordinary equations

dx
dy
(i
dz
dt
P-=Q = &C-==R=S

,0s
(2)'

and let the integrals obtained be u = a, v = b, ... w = k ;

then (p (u, v, ... w) = 0 will be the complete primitive.
If P, Q . . . R, S are linear functions of the variables, the
integrals of equations (2) can alivays be found by the method of
(3346).

DIFFERENTIAL EQUATIONS.

502

Note.Suppose, in equation (1), that any coefficients P, Q

vanish; then, by (2), dx = 0, dy = 0, and therefore the cor
responding integrals are x = a, y = b. The complete primi
tive thus becomes
f (, y, u, v ... w) = 0.
3389 When only one independent variable occurs in the
derivatives of the partial differential equation, the equation
may be integrated as though the others were constant, adding
functions of the remaining variables for the constants of
integration.
3390

(Ex- !) :

, y Integrating for x as though V were conM3

v' y* to*
stant, the complete primitive is
z = y sin"1 + f(y).
Some equations are reducible to the above class by a transformation.
Thus:
3391 Ex. (2):
%xv = x*+y\
Put *.=*,
therefore uv = xt+yi,
therefore n = z, = xiy+$y*+f(x),
therefore
z = \xyy + \xyl + J f (as) dx + \p (y) ,
or

* = i(t*y+*y*) + xQ*)+4'(3/)-

3392

Ex. (3):

Solving by (3283),

(a-a)*,+ (y-&)*, = c-.

^=^ = ^

The integrals are

iog(2/-6)-iog(*-o) = iogo >
log ( c) log (x a) = log C ) '

0T yjzt-c Z~C = C
xa
' a a
'

therefore

= o f-
is the complete primitive.
xa
\x a/
For the converse process in respect of the same equation, see (3382).

3393 Ex. (4).To find the surface which cuts orthogonally all the
spheres whose equations (varying a) are
x'+if + z'^ax = 0
(1).
Let <j> (x, y, z) = 0 be the surface. Then
(xa) f. + y^+zif), = 0
by the condition of normals at right angles. Substitute the value of a from
(1), and divide by f,\ thus,
(x*-yi-z')z,+2xyz, = 2zx.

PARTIAL DIFFERENTIAL EQUATIONS.

By (3383),
= gives

503

= *9- = A,
x y1 z1 2xy 2zx
c for one integral.

Substituting y = cz, we then have

dx
dz
2!3-(c,+ l),-2i!'
which, being a homogeneous equation in x and z, may be solved by putting
z = iws (3186). The resulting integral is tf
= G. Hence the comz
plete primitive is x
~*~z = <j> ^ ^-J and the equation of the surface sought.

3394

Ex. (5).To find an integrating factor of the equation

(i*y-2yl) dx+^xy'-Zx') dy-0

(1).

Assuming z for that factor, the condition (Mz) = (Nz)x (3087) pro
duces the P. D. equation
(xy>-2x*) zx+ (Sy-afy) *, = 9 (^-y) z
(2).
The system of ordinary equations (3283) is
dx
dy
dz
xy"-2xi ~ 2yix,y ~ 9 ('-y') z'
The first of these equations is identical with (1) (and such an agreement
always occurs).
Also

Its integral is

+ -K = c.
y
x
ydx+xdy
_
dz
xyt-2xty + 2xyi-xiy
9 (as'-t/V

which reduces to

+
+= 0;
x
y
z
and thus the second integral is xsy*z = e'.
Hence the complete primitive and integrating factor is

Any linear P. D. equation may be written as a homogeneous

equation with one additional variable ; thus, equation (3387)
may be written
3395

Pux + Quy +...+ But = Sut.

SIMULTANEOUS LINEAR FIRST ORDER P. D. EQUATIONS.

3396 Pbop. I. The solution of such equations may be made
to depend upon a system of ordinary 1st order differential

504

DIFFERENTIAL EQUATIONS.

equations having a number of variables exceeding by more than

one the number of equations.
Let there be n equations reduced to the homogeneous form
(3395) involving one dependent variable P and n+m inde
pendent. Select n of the latter, x,y ... z, and let the remaining
m be E,v-..Z- From the n equations find Px, Pt ... P, in
terms of P(, P^ ... Pf, and arrange the results as under :
P, + a1Pt + b1P1,...+k1P< = 0)
Py + a*Pt + biP,...+k2P( = ol

(1).

Multiply these equations by Xu X2, ... X respectively, and add ;

thus,
+

... + XP, + 2 (Xa) P, + 2 (X6) P, ... +2 (\k) P( = 0

(2).
From this, as in (3387), we have the auxiliary system
die _ d^
X^"- X2

dz_ _ rf$ _
_ dt
X ~2(Xa)~~2(X&) ~2(Xfc)

,<y\
Wl

and, by ehminating \, X2 ... X,

d^ a^dxa^dy ...andz = 0")
dri bidxbidy ... bndz = 0[

^\

dl ^dxktdy ...kndz = oJ
Then, if = a, f> = 6, &c. be the integrals of (4), they
will be values of P satisfying the equivalent system (1), and
the integral of that system will be F(u, v, ...) = 0.

3397 Pbop. II. To integrate a system of linear 1st order

P. D. equations.
Let
A = adx + bdy ... +kd2,
so that AP = 0 represents a homogeneous linear P. D. equa
tion of the 1st order.
Rule."Reduce the equations to the homogeneous form (1);
express the result symbolically by
AXP = 0,

A2P = 0,

... AP = 0,

PARTIAL DIFFERENTIAL EQUATIONS.

505

and examine whether the condition

is identically satisfied for every pair of equations of the system.

If it be so, the equations of the auxiliary system (Prop. I.) will
be reducible to the form of exact differential equations, and
their integrals being u = a, v = b, w = c, ..., the complete
value of P will be F (u, v, w, ...), the form of F being
arbitrary.
" If the condition be not identically satisfied, its application
mil give rise to one or more new partial differential equations.
Combine any one of these with the previous reduced system, and
again reduce in the same way.
" With the new reduced system proceed as before, and continue
this method of reduction and derivation until either a system
of P. D. equations arises, between every two of which the above
condition is identically satisfied, or, which is the only possible
alternative, the system Px = 0, Py = 0, ... appears. In the
former case, the system of ordinary equations corresponding to
the final system of P. I). equations, will admit of reduction to
the exact form, and the general value of P will emerge from
their integrals as above. In the latter case, the given system
can only be satisfied by supposing P a constant."
3398

"Ex.:

Pm+(t+xi,+xz)P,+(y + e-Sx)Pt=:0,
Ps + (xzl + y -xy) Pr + (zt - y) P, = 0.
Representing these in the form A^ = 0, A3P = 0, it will be found that
(A,A, A,A,)P = 0 becomes, after rejecting an algebraic factor, xPz + Pt = 0,
and the three equations prepared in the manner explained in the Rule will
be found to be
Px + (3x*+t)Pz = 0, P,+yPlc = 0, Pt + xPz = 0.
No other equations are derivable from these. We conclude that there is but
one final integral.
" To obtain it, eliminate Px, Pv, Pz from the above system combined with
Pxdx + Pvdy+Ptdz+Ptdt = 0,
and equate to zero the coefficient of Pt in the result. We find
dz {t + Sx1) dx-ydy-xdt = 0,
the integral of which is

at xt x' gj/' = c.

" An arbitrary function of the first member of this equation is the general
value of P."
[Boole, Sup., Ch. xxv.
For Jacobi's researches in the same subject,' see Crelles Journal, Vol. lx.
3 T

DIFFERENTIAL EQUATIONS.

506

NON-LINEAR FIRST ORDER PARTIAL DIFFERENTIAL

EQUATIONS.
3399

Type

F(x,y,z,zzy) = 0

(1).

Charpits's Solution. Writing p, q instead of zx and zy,

assume the equations
^L = dy=

dz

(2).

-gP
9-pqp
q*+pq.Find a value of p from these by integration, and the corres
ponding value of q from the given equation, and substitute in
the equation
dz pdx+qdy
(3),
and integrate by (3322) to obtain the final integral.
Proof.Since dz = pJ.c + qdy, we have, by the condition of integrability,
pf = qx. Express pt and qx on the hypothesis that 2 is a function of x, y ;
p a fnnction of x, y, z ; q a function of x, y,z,p; considering x constant when
findings,,, and y as constant when finding qz. Equating the values of p, and
q, so obtained, the result is the equation
Ap,+Bp,+Cp. = D,
in which A, B, C, D stand for qm 1, qpq?, q,+pqzHence, to solve this equation, we have, by (3387), the system of ordinary
equations (2).
3400 Note.More than one value, of p obtained from equations (2) may
give rise to more than one complete primitive.
The first two of equations (2) taken together involve equation (3).

DERIVATION OF THE GENERAL PRIMITIVE AND SINGULAR

SOLUTION FROM THE COMPLETE PRIMITIVE.
Rule. Let the complete primitive of a P. D. equation of
the 1st order be
z = f (x, y, a, b)
(1).
3401
between

The general primitive is obtained by eliminating a

z = f {x, y, a, tf>(a)}

and

f, = 0

(2),

the form, of <j> being specified at pleasure.

3402 The singular solution is obtained by eliminating a and
b between the complete -primitive and the equations
f. = 0,

fb = 0

(3).

PARTIAL DIFFERENTIAL EQUATIONS.

507

Proof.By varying a and b in (1),

P = /* +fa a. +fb
q=fs, +fa a +fobv
Therefore, reasoning as in (3171), we must have
/,+/&. = 0 and faa,+f>l, = 0
(3),
therefore either / = 0, fb = 0, leading to the singular solution ; or, elimi
nating fmfb,
<!&, = 0,
and therefore, by (3167), b = <p (a). Multiply equations (3) by dx, dy re
spectively, and add, thus fada+fbdb = 0. Substitute b = <p (a) in this and
in (1), and the equations (2) are the result.
SINGULAR SOLUTION DERIVED FROM THE DIFFERENTIAL
EQUATION.
3403 Rule.Eliminate p and q from the differential equa
tion by means of the equations
zp = 0,

zq = 0.

Proof.Let the D. E. be z = f(x, y,p, q), and the C. P. z = F(x, y, a, b).

Now p and q being implicit functions of a and b, we have, from the first
equation,
za = zpa + zq qa, zb = zrpb + zqqb.
Hence the conditions z = 0, zb = 0 in (3) involve, and are equivalent to,
sr, = 0, zq = 0.
3404 All possible solutions of a P. D. equation of the 1st
order are represented by the complete primitive, the general
primitive, and the singular solution.
[Boole, p. 343.
3405 To connect any given solution with the complete
primitive.
Let 2 = F(x, y, a, b) be the complete primitive, and
z = (x, y) some other solution.
Determine the values of a and b which satisfy the three
equations

F = f,

Fx = $x,

Ft = $r

If these values are constant, the solution is a particular

case of the complete primitive ; if they are variable so that
one is a function of the other, the solution is a particular case
of the general primitive ; if they are variable and unconnected,
the solution is a singular solution.
3406 Coe.Any two solutions springing from different
complete primitives are equivalent.

DIFFERENTIAL EQUATIONS.

508

3407

Ex.:

By (3299),

z=px + qy+pq

(1).

^- = dy = ^ = ^

(2),

andwehave

q = J+^'

Hence (2) becomes

(-&dx = dy = (P+y)'* = $ ;
xy+z
2pz + yzp'x
0
q = z aa:. Substituting in dz=pdx+qdy,
a+y
dz = fidx+ -a,rii/
a+y

.. dp = 0, = o :

A = _?* = Jj+yJ '

(3).

By (3322), making constant,

+
= 0,
zax a + y
therefore
log (zax) + log (a+y) = $ (z)
(4).
Differentiate for x,y,z, and equate with (3), thus p'(ar) = 0, therefore
f (z) = constant (say log 6); therefore, by (4), z ax + by + ab, the C. P.
of(l).
3408 To find a singular solution by (3402), we must eliminate a and 6
between za = 0, z6 = 0; that is, x + b = 0 and y + a = 0,
therefore
z = xyxy + xy = xy
is the singular solution.
To find the general primitive by (3401), eliminate a between the two
equations = ax + (y +o) <j> (a) and x + (y + a) <tf (a) + <p(a) = 0.
NON-LINEAR FIRST ORDER P. D. EQUATIONS WITH MORE
THAN TWO INDEPENDENT VARIABLES.
3409 Peop.To find the complete primitive of the differ
ential equation
F(a;1,x2 ... a?3,pi>Pi ./>) = 0
C1).
,
where

3410

dz
r>\ = -j,
oaJi

dz
lh = ~r~>
rfa'g

o
&c-

Rule.Form the linear P. D. equation in 4> denoted by

_ f/dF ,
dF\d<I>
dF /d* .
d*\">
n
* 1 (dTr +P' dz ) dp; - dpT (dTr +Pr di) I = '

7ie summation extending from r = l fo r = n.

auxiliary system (3387) n 1 integrals
4>x = a1,

4>2 = aa, ... *w_1 = af,.1

-Fro??i Ae

SECOND ORDER P. D. EQUATIONS.

509

may be obtained. From these equations, together toith (l),find

Pu Pa Pn in terms of x15 x2 ... xn, substitute the values in
dz = p^Xi+padxsj ... +pndxn,
and the integral of this last equation will furnish the solution
required in the form
f (x1} Xg . . . xn, z, ax, a? an) = 0.
[Boole, Diff. Eq., Ch. xiv., and Sup., Ch. xxvii.

SECOND ORDER P. D. EQUATIONS.

3420

Type

F (at, y, z, zx, z, z^, zxy, as*) = 0.

The derivatives zx, zy, z^, z^, z* are briefly denoted by

p, q, r, s, t respectively.
2 being a function of the two independent variables x and
y, the following values are of frequent use
3421

dz = pdx+qdy;

dp = rd#-\-sdy ;

dq = sdx+tdy.

If u be any function of x, y, and z, the complete deriva

tives of u are indicated by brackets, thus
3422

00 = ux+puz,

(ut) = tiy + que.

A linear 2nd order P. D. equation is of the type

3423

Rr+Ss+Tt=V

(1),

in which B, S, T, V are functions of x, y, z, p, q.

Proposition.Any P. D. equation of the 2nd order which
has a first integral of the form u =f{y), where and v involve
x, y, z, p, q, is of the form
3424

Rr+Ss+Tt+Uirt-s*) = V

(2),

where B, 8, T, U, V are functions of x, y, z, p, q, and

3425

U=upvg-uqvp

(3).

Proof.Differentiate u = f (v) for x and y separately, considering x, y, z,

p, q all involved in u and v, and eliminate f'{v). The result is equation (2),
with the values

510

3426

DIFFERENTIAL EQUATIONS.

= % (vy) (tt,) vp,

fiT= vq {ux) - (vx) uv

fiS = vp (ux) -(,) up-vq (ut) + (vt) uq,

nU = UpVq- uqvp, MF = (,)(,)-(,)(,),
with the notation (3422), p being an undetermined constant.
3427 Cob. The condition to be fulfilled in order that
equation (1) may have a first integral of the form n =/(?) is
upvquqvp = 0.
SOLUTION BY MONGE'S METHOD OF
3428

Rr+Ss+Tt=V.

Rule. Write the two equations

Rdtf-Sdxdy+Tdt* = 0

(1),

Rdpdi/-Vdxdi/+Tdq(Lv = 0

(2) .

Resolve (1) into its factors, producing the two equations

dy mxdx = 0

and

dy m2dx = 0.

From dy = niidx and equation (2) combined, if necessary,

with dz = pdx + qdy, find two 1st integrals u = a, v = b;
then u = f (v) will be one 1st integral of the given equation.
Similarly from dy = m.2dx find another 1st integral.
3429 The final 2nd integral may be found from one of the
1st integrals by Lagrange's method (3383).
3430 Otherwise, determine p and q in terms of x, y, z from
the two 1st integrals; substitute in da = pdx + qdy, and then
integrate by (3322) to obtain the final integral.
3431 If equation (1) is a perfect square, there will be only
one 1st integral, and Lagrange's method only is applicable.
Proof. By (3427) we may put uq = mup, vq = mvp ; and also
dz = p dx + q dy (3321) in the complete derivatives
(du) = %itdx + uydy+uMd*+u9dp+uqdq = 0, (dv) = &c. = 0;
.-. by (3422)
(,) dx + (,) dy + ut (dp+mdq) =01
(3)
(O dx+ (O dy + vp (dp+mdq) = 0 3
K
Solving these equations for the ratios dx : dy : dp+mdq, we obtain at once
dx
dy+mdx
mdi/
dp + m dq
~B~
8
T
V
L ;'
with the values of B, S, T, Fin (3426).

SECOND OBBEB P. D. EQUATIONS.

511

Equations (1) and (2) are the result of eliminating m from (4). These
two equations with dz = p tlx + qdy suffice to determine a first integral of
(3428) when it exists in the form u =/ ().
3432

Ex. (i.): q{l+q)r-(p+q+2pq)s+p(l+p)t = 0.

Solving the quadratic equation (1), we find

pdx + qdy = 0, or (1+p) das+ (1+gr) dy = 0
First,
dz = pdx + qdy = 0,
.'. z = A.
Monge's equation (2) is q (1 + q) dp dy +p (1 +p) dq dx = 0,
which, by pdx = qdy, gives j~|jr~ =
Hence a first integral is

; and, integrating,

(5).

= ^>

= ^^

Next, taking the second equation of (5) with

dz = pdx + qdy, dx+dy + dz = 0, .'. jj + j/ + z = C.
Also, by (5), equation (2) now reduces to qdp = pdq, and by integration,
p = qD ; therefore the other first integral is p = q\p(x+y + z).
For the final integral integrate^ q4> = 0; i.e., zx \pz = 0, by (3383) ;
.-. & =

^
-=^,
^(x + y + z)
0'

.-. * = .4, and rfx = ,

+
l-K*+2/ + *)

J 1 l//(i! + y+)
Hence the second integral is xf (x + y+z) = F (a).
3433

Ex. (ii.):

*ar-a% = 0.

(i.) Here, in (3428), E = 1, S = 0, T=-a\ V=0; therefore (1)

and (2) become
dtfa'dx1 = 0, dpdy a1dqdx = 0.
From (1) dy + adx = 0, giving y + ax = c, and converting (2) into
tip + a cZgr = 0, which gives p + aq c ; therefore a first integral is
p + aq = </>(y + ax)
(3).
Similarly, from (1), dyadx = 0 gives rise to another first integral
p-aq = \f/(y-ax)
(4).
Eliminating^ and q by means of (3) and (4) from dz = pdx + qdy, we find
dz = (2a)"1 {<p (y + ax){dy + adx)\j/(yax')(dy adx)},
therefore, by integrating, z = * (y + ax)
(i/ ax).
For the symbolic solution of the same equation, see (3o(56).
SOLUTION OF THE P. D. EQUATION.
3434

Rr+Ss+Tt+U(rt-s2) = V

(1).

Let TOu mt be the roots of the quadratic equation

3435

m2-Sm+RT+ UV=0

(2).

DIFFEREXTIAL EQUATIONS.

512

Let j = a, i\ = b, and 4 = a, vt = b' be respectively

the solutions of the two systems of ordinary differential
equations.
3436

TJdp =
dy-Tdx\
Udq = m^dxRdy > (3),
dz = p &e+ qdy)

Fdp = itijdy Tdx\

Udq = nt^dx Rdy > (4).
dz = p dx -f- qdy )

Then the first integrals of (1) will be

i =/("i)>

To obtain a second integral :

3437 1st-When m,,
are unequal, assign any particular
forms to fi and/*, then substitute the values of p and q, found
from these equations in terms of x and y, in dz = pcLv-\-qdy,
which integrate. Otherwise, assign the form of one only of
the functions fuft, involving an arbitrary constant C, solve
for p and g, and integrate dz = pdx-\-qdy, adding an arbitrary
' function of C for the constant of integration.
3438

2ndly.When mlt mt are equal, and therefore, by (2),

Si = 4(RT+UV)
(5).

Equations (3) and (4)

reduce to
3439

coincide, and,

TJdp = \Sdy-Tdx
TJdq = \S dx-Edy
dz =

p dx+ qdy

since

m = \S,
(6),
(7),
(8).

Here py = qx, and therefore the last equation is integrable

if the values of p and q, obtained by integrating (6) and (7),
be substituted in it. Let u = a, v = b be the integrals of (6)
and (7) ; and let

z = <j> (x, y, a, b, c)

(9)

be the integral obtained from (8).

The general integral is found by making the parameters
, b, c vary subject to two conditions b=f(a), c = F(a);
that is, by differentiating
z = + {x, y, a, f(a), F(a)}
for a, and eliminating a.
3440 The general integral therefore represents the envelope
of the surface whose equation is (9).

SECOND ORDER P. D. EQUATIONS.

513

Proof. (Boole, Sup., p. 147.) Assuming a 1st integral of the form

u=f(v), eliminate p and v from equations (3126) by multiplying (i.) by
(x) (ii.) by (,),, (iv.) by ()(), (v.) by
and adding. Again,
eliminate fi and v by multiplying (i.) by ()*, (ii.) by ()2, (iii.) by (mi)(mj/),
(v.) by (ux) p+ (), and adding. The two resulting equations are
R (,) u, + T (,) u, - 17 (,) (,) + F, 4 = 0
|
JB(I)'+S(!tl)(,) + r(!1jJ+F{K) ,+(,),} = o)

Multiply the 2nd of these by m, divide by V, and add to the 1st equation ;
the result is expressible in two factors either as (11) or (12),
{R{u,) + rrh(y,)+VuP\{mx (u,) +T (ut) + VuQ} = 0
(11),
{R (tO +, (u) + Vup} {rnt (,) + T () + Vut] = 0
(12),
mx, jo, being the roots of the quadratic (2). By equating to zero one factor
of (11) and one of (12), we have four systems of two linear 1st order P. D.
equations. Taking each system in turn with the equations
()+rn+SM = 0>
(,)+*,+ fit, = 0,
and eliminating (ux), (uy), , uq, we have the de
R m. V 0 I
terminant annexed for the case in which the 1st
nh T 0 V
= 0,
factor of (11) and the 2nd of (12) are equated to
1
0 r s
sero. In this case, and also when the 2nd factor
0
1st
of (11) and the 1st of (12) are chosen, trans
posing OTj, m, in the determinant, the eliminant is equivalent to
V{Rr+St+Tt + U(rt-s*)-V} = 0,
having regard to the values of mlmJ and m1 + ma from (2).
When the 1st factor of both (11) and (12) is taken, the 2nd order P. D.
equation produced by the elimination is
Vt-R(rt-t*) = 0,
and when the 2nd factor of each is taken, the elimination produces
Vr-T(rt-ss) =0.
Hence the hypothesis of a 1st integral of (1), of the form u =/(),
involves the satisfying one or other of the systems of two simultaneous equa
tions, (13) or (14), below :
fi (.)+h (,) + F,= 0")
*,(.)+ T{u,)+Vu,= 0)

(13x
''

R (,) + w, (u)+Vup = 0~)

m,(u.) + T (,) + F4 = 0j "'K

Now multiply the 2nd equation of (13) by X and add it to the 1st.
In the result, collect the coefficients of u ut, up,
ue. The Lagrangean
system of auxiliary equations (3387) will then be found to be
dx
dy
dj>
B+Xmt" w,+\r V ~ W~ Rp + mlq + \(Tq + 'mip) ~ 0~'
Eliminating X, equations (3) are produced.
the same way, equations (4) are produced.
3 TJ

Treating equations (14) in

DIFFERENTIAL EQUATIONS.

514

3441

POISSON'S EQUATION.
P = (rt-si)nQ,

where P is a function of p, q, r, s, t, and homogeneous in r, s,

t ; Q is a function of x, y, z, and derivatives of z, which does
not become infinite when rts* vanishes, and n is positive.
Rule. Assume q = <{> (p) and express s and t in terms of qp
and r; tf/i?fs, rt s* vanishes, and the left side becomes a func
tion o/p, q, and qp.
Solve for a 1st integral in terms o/p and q, and integrate
again for the final solution.
PROOF.
s = q, = qpp, = qp r ; t = q = qpp = qpr;
therefore r/ ss = 0. Also P is of the form (r, s, t)m = r" (1, qp, qp)m.
Hence the equation takes the form (1, qp, qP)m = 0.
LAPLACE'S REDUCTION OF THE EQUATION.
3442

Rr+Ss+Tt+Pp + Qq+Zz = U

(1),

where B, S, T, P, Q, Z, U are functions of x and y only.

Let two integrals of Monge's equation (3428)
Rdy2-Sdxdy + Td,vi = 0
*>e

f (, y) = a,

^ (x, y) = b.

Assume

I = ^ (x, y),

n = t(x,y).

3443 To change the variables in equation (1) to $ and n,

we have
P = 2, = 2t,+2,n,;
</ = zg = Zf$, + 2,lf ' '
r = zu = zu^+2zhKtv,+zt,vl+ztKtx+Zr,ntx ',

(1701)

S = 2x = %^H, + 22,x' + %,(^', + 51,nx)+2i^xv + 2,,'xir.

The transformed equation is of the form
z^+Lzt+Mzi+Nz = V
where L, M, N, V are functions of $ and >j.
may be written in the form

If

(2),
This equation

(dt+M)(a\+L) z+(N-LM-Lt) z = V

(3).

N-LM-Le = 0

(4),

we shall have
(dt + M) z = V

with

(<7, + 1) = z,

LAW OF RECIPROCITY.

515

and the solution by a double application of (3210) is obtained

from

By symmetry, equation (1) is also solvable, if

N-LM-M, = 0

(5).

But if neither of these conditions is found to hold, find z

in terms of z from (3). It will be of the form
z = Az't+Bz' + G,
where A, B, C contain and j/. Substitute this for z in
(d^+L) z = z, and the result is of the form
z^+L'zt+M%+N'z' = V.
The same conditions of integrability, if fulfilled for this equa
tion, will lead to a solution of (1), and, if not fulfilled, the
transformation may be repeated until one of the equations,
similar to (4) or (5), is satisfied.
3444

Cor.The solution of the equation

z^ + az( + bz^ + abz = V

is

z = e-a"^() + e-*^(,) + e-fl,'-6t|je<,''+if Vdr,dt

3445 For the solution of equation (2), when L, M, V contain also z, see
Prof. Tanner, Proa. Lond. Math. Soc, Vol. viii., p. 159.

LAW OF RECIPROCITY.
3446

[Boole, ch. xt.

Let a differential equation of the 1st order be

y, z, p, q) = 0

(1).

Let the result of interchanging a; and p, y and q, and of

changing z into px-\-qyz, be
(P> <7> px+qyz, x,y) = 0

(2) ;

then, if z = rp (x, y) be the solution of either (1) or (2), the

DIFFERENTIAL EQUATIONS.

516

solution of the other will be obtained by eliminating S and v

between the equations
= W ( i),

y W (6 v)>

3447 Ex.Let
*=p?
be the two reciprocal equations.
The integral of (2) is

(1),

* = 1 + w+ ( *?)
px + qy-z = xy

= apy+/(

(2),

.-. * (,) =^ + x/(

, q hare now to be eliminated between

* = *-jf (|) +/(-f ),

y = +r(f).

=&

(3)-

Each form assigned to/gives a particular integral of (1). If /("y) = a y + i,

the equations (3) become x 17 + 6, y = + a,
and the elimination produces z = (2: b)(ya).

r = 17,

3448 In an equation of the 2nd order, the reciprocal equa

tion is formed by making the changes in (3446), and, in
addition, changing
r into -
rts

* into

t into
rts-

'
rtr

then, if the 2nd integral of either equation be z = \fi(x, y), that

of the other will be found by the same rule.
3449 The above transformation makes any equation of the
form
<t>{p,<l) r+ifi(p,q)s+x(p,q)t = 0
dependent for solution upon one of the form
x
y) r-i Q> y) s+<t> (*' y) 1 = 3450

And, in the same way, an equation of the form

Rr+ Ss+Tt= F(rf-J*)

is dependent for solution upon one of the form
Rr+S+Tt= V.
See Be Morgan, Camb. Phil. Trans., Vol. VIII.

SYMBOLIC METHODS.
FUNDAMENTAL FORMULAE.
Q denoting a function of 6,
3470
(d,m)-lQ = er*$e Qdd.

SYMBOLIC METHODS.

517

Proof.The right is the value of y in the solution of d,y my = Q by

(3210). But this equation is expressed symbolically by (d,m) y = Q (seo
1492), therefore y = (tZsw)"1^.
Let x = e9, therefore d9 = xda and xdO = dx.
Hence
(3470) may be written
3471
(**. m)-1 Q = xm jcr-1 Qdx.
3472

Cob.

3473

or

{di-m)-^Q = Cemii
(^-m)-1 0 = CV.

Let
denote a rational integral function of m ; then,
since deem9 = mem9, d29 em* = mieme, &c, the operation de is
always replaced by the operation mX .
3474
3475

Hence, in all cases,

F(de)eme = e""F(m).
F

eM9 Q = emF (de+ to) Q.

Formula (2161) is a particular case of this theorem.

3476
emiF(de) Q = F (de-m) em Q.

Also, by (3474-6),
3477
F (m) = e-m9F(de)
3478

F (de+m) Q = e-^F (de) em Q.

3479

F (de) Q = e-mF(de-m) emi Q.

To the last six formulae correspond

3480
F (xdx) xm = xmF(m).
3481

F (xdx) xmQ = xmF (xdx+ to) Q.

3482

*mF (xdx) Q= F (xdx-m) xmQ.

3483

F (to) = x~mF (xdx) xm.

3484

F (xdx+m) Q = x~mF (xdx) xmQ.

3485

F (xdx) Q = x~mF (xdx-m) xmQ.

If U=a + bx + cxl + &c., then, by (3480),

3486

F{xdx) (7= F(0) a+F(l) bx+F{2) cx*+&c.

DIFFERENTIAL EQUATIONS.

518

3487

F~\xdt) U= F-^O) a+F-J(l) bx+F-*(2) cx*+&c.

3488

F{adx,ydy, zdz, ...) *>V ... = F(m, n,p ...) .r"yzp...

3489

(dt-2) ... (rf,-n+l) u,

or, more succinctly, writing D for de,

D(D-1) ... (D-n + 1) n
3490

or

D^u

(2452).

Otherwise Jifu^ = xda(jsdx 1) ... (<rdx n+1) u.

Proof.As in (1770). Otherwise, by Induction, differentiating again,

and remembering that xt = x.
Note.In the symbolic solution of differential equations,
we may either employ the operator xdx directly, or the
operator de after substituting ee for x. Formula? (3480-5) or
(3474-9) will be required accordingly.
3491
{+{p)f]*Q
= enrt<f>(D+nr) <f>(D+n=l.r) ... <f>(D+r) Q
= +(D) 4>{D-r) *(Z>-2r) ... <j> {D-(n-l) r\ e*Q.
Pboof.By repeated application of (3475) or (3476).
For ready reference, formula? (1520, '21) are reprinted
here.
3492
3493
Let

f(x+h) =
f(x+h, y+k) = eM^f{x, y).
a0 + a1x + aix2...+anxn =f(x),

then, denoting de by D,
3494 /( D) uv = uf{D) v+ u.f (D) v+ *Lf (D) +&c.,
where f (D) means that D is to be written for x after differ
entiating f(x).
Proof.Expand uv, I) .uv, D* .uv ... D'.uv by Leibnitz's theorem (1460);
multiply the equations respectively by a0, alt as ... a, and add the results.

3495

uf(D)v =/(!>). uv-f'{D) utv+-Lf'D.uilv-&e.

SYMBOLIC METHODS.

519

Proof.Expand uv uvlt, uv^ ... uvH, by theorem (1472), and proceed as

in the last.
3496

F (rffa) sm mx = F(- ros) mu mx.

v ' COS
v
' COS

A more general theorem is

3497
F(<n*)(uim+u_im) = F(-m*)(uhu+u_im),
where u and it have the meanings assigned below (3499),
and i = V 1 .
Theorem.If <p and ^ denote any algebraic functions of x
and y, it may be shown, by (3474) and (3475), that
3498

* (dx+y)

= $ (d,+x) *(y).

3499 Let u, or, more definitely, nn = (x, y, z, ...)*, represent

a homogeneous function of the th degree in severable vari
ables, and let
3500

ft = xdx+ydv+zd,+&c.

Then, by (3480),
3501
ftu = nu, iPu n2u,
3502

Hence

ifu nsu,

&c.

F'() = F(n)u.
REDUCTION OF Ffa) TO/(ir).

3503 Let u be any implicit function of the valuables, and

let 7r = 7TJ + 7T2, where
operates only upon x as contained in
u, and 7r2 only upon x as contained in iru, &c. after repetitions
of the operation tr. Then
3504
3506

ft\U = iru,

ir\u = (w 1) fl-M,

< = (w r+1) ... (ir 2)(ir 1) irw.

PEOOF.
7r,M = (7T ffj) ?t = 1TW,
since has here no subject to operate upon.
TTjM = (lT T,) 7TM = (ff 1) 1TM,
for, m being of the 1st degree, tr, and 1 are equivalent as operators.
next step, 7r, and 2 are equivalent, and so on.

In the

Cor.When u is a homogeneous function, we have, by

DIFFERENTIAL EQUATIONS.

520

(3501), trru = nru, therefore ir and n are equivalent operators

upon u.
3507

Hence (3500) may be written

= (n-r+1) ... (-2)(w-1)hm = n^u,

which is Euler's theorem of homogeneous functions (1625),

since in that theorem the operator is confined to u.
3508

As an illustration, let tm = {xd, + ydv~) u = *],

then wjw = (x'd.iz + 2x>j dxy + y'diy) i

** = (""i + 'j)
= ^m + t, jtm.
Here
wynt = (a'(7, + yd^) {xd^ + ydy) u,
the operation being confined to x and y in the second factor (3503), and there
fore producing (xd, + ydg)u merely.
Hence v'u = (x*du + 2xy dx + y,d2l/ + xdz + ydl/~) it, which proves (3505).
If JJ = u0 + u1 + h.2 + ... , a series of homogeneous functions
of dimensions 0, 1, 2,
then, by (3502),
3509
3510

F(n)U=F (0) u0+F (1) ,+F (2) 2+ . . . ,

F-i()U= F-i(0)uQ+F-i(l)u1+F-i(2)u2+...

3511
3512

Ex. 1 :

a*U= u0+ ati^ + a*u2 -f- . . . ,

a-*U= u0+(Tlu1 + a-iui+ ...

Ex. 2 : If it have the meaning in (3499),

3513
F (tt) e* = F (0) 1 + F (n) u + F (2n) ^ + ^(3n)

+ Ac.,

and similarly for the inverse operation F-1 (jt).

Proof.By (3502) applied to the expansion of the subject by (150).
3514

C{n,m)un _ 2 A:"dpxyHl<nizrdr:... ((
ml

where

p + q + r+ = w

pi ql rl ...
and

^ I = 1.2 ... jp.

Proof.Equate coefficients of a" in the expansion of

(1 + a)* = (1 +a)z"< (1 4 a)""' (1 + a)*"' ... [T,
reducing by (3490).
3515 The general symbolic solution of the equation
F(d9)u = Qi&
u = F1 (de) Q+F l (de) 0, by (1488-90).

SYMBOLIC METHODS.

3516

521

The solution of the equation (3238), viz.,

i#(-u *+ +(-i)y*+a),y = Q

(1),

where Q is a function of x, is most readily obtained by the

symbolic method. Thus mu m2, ... m, being the roots of the
auxiliary equation in (3239), and A,B,C ... iV'the numerators
of the partial fractions into which (mn + a1mn~1+ ... +aH)~1 can
be resolved, the complete primitive will be
3517
y={A(dx-miyi+B(dje-mi)-\.. + N(dx-?nn)-i}(Q+0),
where

(dmm)-1Q = emx$e-U!Qdx,

(3470)

and the whole operation upon zero produces, by (3472), for

the complementary term,
3518
C1em+C2e"*...+Cnemx.
Proof.Equation (1) may be written
(dMe + aldin.vx+aidln_Vx ... +a) y = Q,
or
(d. - m,) (d,m,) ... (d,-m) y = Q,
by (3515), y = { (^-^(d.-m,) ... (i.-m.) }-> (Q + 0),
which, by partial fractions, is converted into the formula above.
If r of the roots mu m2, ... are each
m, those roots give
rise in (3517) to a single term of the form
3519

(A+Bd.+Cdt,... +Rdrt)e\ eQ.

J rx

Proof.By (1918), the r roots equal to m will produce

\A'{dm-m.yr+jB'(d.-tny+x ... +B'(d,-m)-1\ Q>
or
(4 + 5^+0(4, ... +RdTX)(d,-m)-rQ.
3520

But, by (3470),

(^-w)-JQ = (i-)-le"* Je-'-'QdaJ

= e** f | e-,Me"'1 je"'" Qda: j cfo = e"*| e""* Qrfas,

3521
Here

Ex. (1) :

and so on.

^-^-5^-3 = Q.
m8 -m2 - 5m - 8 = (m - 3) (m

and

11
1
1
(m-3)(m+l)2
16(m.-3) 16(m + l) 4(w + l)2'
therefore
y = TV (4_3)-'Q-TV (d. + l)"1 Q-JR+l)"1 Q
iV* J " 31
- tVJe* ' Q<* - e" jjV Qcfc2,
3 x

(3520)

592

3522

DIFFERENTIAL EQUATIONS.

ex. (2):

ir+2w = g,

therefore
Here

n = {d.^ + a3) " 1 Q.

(m'+a1)-' = (2ia)-|{(-to)-,-(m+i)-1},

therefore

u = (2ia)-' {(d,-)"1 Q-^+ta)"1 Q}

= (2ta)-' {eiox je-" Q^-e"*" Je'" Qdx]

(3470)

= a"1 sincw;JcosaxQ(7j; a ~ 1 cos <; J sin ax QcLr,

by the exponential values (766-7).
3523

Cor. 1.The solution of w^-f^u = 0 is

u = A cos ax +13 sin tu;.

3524

Cor. 2.The solution of u^atu = 0 is

u = Ae" + Be'.

Change a into ia in the fifth line of (3522), and put Q = 0.

3525

When Q is a function whose derivatives of the ntb and

higher orders vanish, proceed as in the following example.

Ex. (3):
u + a*u = (l+z)*,
therefore
= (c^+a')-1 (l + z)'+ (<*.+ a*) "'O
= (a-,-o-dto+o-6dl,-&c.)(l+2a;+a>1) + (c,+a,)"10
= a"1 (l+x)12a~*+A coaax + B sin a*,
the last two terms by (3523).
Exceptional Case of the Inverse Process.
3526

Ex. (4) :

itji + a1" = cos rue,

.-. u= (ct+o^-'CcoBwas+O) = H^+a')"1 (e'" + "'** + )

= J(e," + e-")(-n,+a,)-,+^coso + sino by (3474) and (3523)
= cos nx (a'')"'+ Ac.
Now, if = a, the first term becomes infinite. In such cases proceed
as follows :
Put A = A'-(a*-*)-\ and find the value of C08n^-C08aa!,
a n
n = a. By (1580) it is = E. Thus the solution is
At
x sin ax
+ A' cos ax + B sin ax.
2a
The same result is obtained by making Q c cos ax in the solution of (3522).
For another example, see (3559).

SYMBOLIC METHODS.

523

3527 Ex. (5) :

y-9y.+20y = arV,
therefore y = {(4-4)(c7I-5)}-1.1!!e:u+{(^-4)(^-5)}-,0
= es'{(^-l)(rfi-2)}-1a;,!+4eto+J5es*.
(3475, 3517, 3472)
Now

(TO'_3m + 2)-1 = i(l-3-^=^)"1

Hence the solution becomes

3528 Ex. (6) :

(d,-a)*u = e"*,
therefore u = (i,-o)-"eM= e?*(d,ynl (347G) = C'f, + f o).
Vm! Jiix /
3529

Ex. (7) :

(2149)

(d.+afy = sin mx,

therefore
y = (rf, + a) " 2 sin mx + (d, + a) ' 2 0
= (dx-a,y (rf^-a5)-2 sin mx+e-" (t/,)"2 0 [by (3478) with Q = 0]
= (-^-a*)-' (<*,-a)' sin f*+e-a(4;e + .B) (by 3496)
= (m* + a5)-2 (in' sin wis 2am cos ma + a8 sin mx) + e'ax (Ax + B).

REDUCTION OF AN INTEGRAL OP THE nth ORDER.

3530

jlg = ^Ti { xn~^Qdx~ fa"1)

+ C (n, 2)

JW*

jQ.r'dz' ... Jq*"-1^,

where to 1 ! = 1 .2 ... n.
Peoof.By (3489) dn,Q = e-'(d,-n + l)(d,-n + 2) ... d,Q
therefore
d_nxQ= {(d,-n+l)(d,-n+2) ... d.}-l<f*Q
= -T-{(d.-n + l)-i-(n-l)(d,-n + 2)->
n_i+ C(n,2)(d,-n + 3)-1-&c.}e~Q

(1),

(3517)

= -1 {- ((|,)-'e,-(n-l) e("-2"W1e," + &c.} Q.

n 1 !
Then replace e' by x.
The equation
3531 aarymx+bxnynx+&c. = 4

C^+&c. = Q

DIFFERENTIAL EQUATIONS.

524

may, by (3489), be transformed into

{a(xdx)+b(xdxy:l + &o.} y = Q

or

F(xdr)y = Q.

The solution is then obtained from

3532

y =

(y Q + F~* (xdx) 0.

The value of the 1st part is given in (3487).

3533 If o, (3, y, &c. are the roots of F(m) = 0, the second
part gives rise to the arbitrary terms

3534 If a root a is repeated r times, the corresponding

terms are
or { C, (log *)-+ Ct (log xy-*+ ... + cr}.
Proof.The partial fractions into which F~l (xd,) 0 can be resolved, as
in (3517), are of the type C (x^-m)-1 0, m being a root of F(x) = 0. But
(zd, to)_10 = Cxm (3473), 0 being an arbitrary constant.
For a root to repeated r times, the typical fraction is C (zd* m)~', p
being less than r. Now
(xd.-m)' Czm Qogx)'-' = (d,-m)> Ce" t)""1 = e" (d,)' GO1"1 (3475) = 0,
therefore

(xdxm)- 0 = Cx" (log z)"-1.

The equation
3535
aym9+%M(,+ &c. =f(ee, sin 6, cos 0)
is reducible to the form of (3531) by x = e"; or, substituting
from (768), it may be written
F(d,)y = Z(Ame't),
and the solution will take the form
3536

y = a^f'W+'-'WO.

for the last term of which the forms in (3533-4) are to be

substituted with x changed to e6.
3537

Ex. (1 ) :

= ax" + bxn
xdx(xdx-l)(zdx 2)y = axm+bx",
.: y = {xdx(xdx-\)(zdx-2)}-x (axa + bxK) + {xdx(zdx-l)(_zdx-2)}-10

~m(m-l) (to-2) (n-l)(-2)

'
by (3180) and (3533). A result evident by direct integration.

SYMBOLIC METHOTHiK

3538

525

Ex. (2):
x,ylx+3xyx + y = (1-z)"2.
By (3490)
{xdx(xdx-l) + 3xdx + l} y = (zdx + l)*y = l + 2z+3zi + &c.,

.-. y= (xdx+l)-2(l+2x + 3x>+) = (0+l)-'+2(l+l)-, + 3(2+l)-V+

(3480)

= 1+ -I +4 +&o.+ Ai + J* =_I iog (l-x) + &c.

16
xx,
x

3539 Ex. (3) : yjx+(4z-l) yx+(W-2x+2)y = 0.

Let t = d, + 2x. Then the equation may be written
T(*-l)y = 0, .-. y = {*(*] )}"'0 = (t - 1)"1 O-ir"1 0.
Let (tt-1)-10 = m, /. (a l)w = 0, or x+(2as l)u= 0, .-. = Ae*.
3540

The solution of a P. D. equation of the type

air^z+bir^z+Scc. = u1-\-ui-\-&c,

where uu u2, &c. are homogeneous functions of the 1st, 2nd

degrees, &c. in x, y, and v1 = xdx + ydt (3503), is analogous to
(3531), and is obtained from that solution by substituting
!, i*2, &c. for Bx, Cx2, &c. ; and, for such terms as Cx", an
arbitrary homogeneous function of x and y of the same degree.
3541
where

Solution of
F (it) u = Q,
F(w) = nn+A1n*-1+Aiirn-i+An,

and

Q = w0+m1+M2+&c.,

a series of homogeneous functions of x, y, z, ... of the respec

tive dimensions 0, 1, 2, &c.
Here
u = F~l (tt) Q+F~l (n) 0.
3542 The value of the 1st term is given in (3510). For the
general value of the last term (see Proof of 3533), let
F (to) = 0 have r roots = to ; then
3543

C(ir-m)-*0= C{u(\ogxy-1+v(\ogvy-i...+w},

where u,v, ... w are arbitrary functions of the variables all of

the degree to.
3544

Cor.

(-m)-10=(,y, ...)",

that is, a single homogeneous function of the variables of the

degree to (1620).

DIFFERENTIAL EQUATIONS.

526

3545 Ex. : ar^ + 2xyzrll + y,zJf a (xzx + yz) + az = um + tt,

um, un being homogeneous functions of the mth and /Ith degrees. The equation
may be written

(ir cm^ + a) z = um + un;

or, by (3505) ,
( - a) (x 1) z = um + un,
therefore z= { (v- a) (* - 1)
( + ) + {( a) (-l)}-l0
=

''Is
I
M"
-L 7T -L TT
(m-a)(m- 1) (n-o)(n-l)
a

The first two terms by formula (3502) ; the last two terms are arbitrary
functions of the degrees o and 1 respectively, and result from formula (3543)
by taking p 1 and m = a and 1.
3546 To reduce a P. D. equation, when possible, to the
symbolic form
(n*+Ain-i+AiTl-*... + An)u=Q
(1),
where

n = MdJ. + Ndl/ + &c.,

and Q, M, N, &c. are any functions of the independent

variables.
Consider the case of two independent variables,
(Mdx+Ndyyu = M2iiix + 2MNHXU+N2uiy
+ (HMX+NM) ux + (MNx+NNy) u, ... (2).
Here the form of n is obtainable from the right by con
sidering the terms involving the highest derivatives only, for
these terms are algebraically equivalent to (Mda+Nd9)*.
The reduction being effected, and the equation being
brought to the form of (1) ; then, if the auxiliary equation
3547

m" + A1mn-1 + A2m-'l...+An = 0

(3)

have its roots a, b, ... all unequal, the solution of (1) will be
of the form
3548
u = (U-a)-1Q + (U-b)-iQ + &c
(4).
The terms on the right involve the solution of a series of linear
first order P. D. equations, the first of which is
3549

Mu. + Nu, + ...-au = Q,

and the rest involve b, c, &c.

If equal or imaginary roots occur in the auxiliary equation,
we may proceed as in the following example.

SYMBOLIC METHODS.

527

3550 Ex.:
(1 + *y zlx - 4*y ( 1 + *) x, + 4eY* + 2a! (1+ x2) zx + 2y
- 1) z, + a*z = 0.
Here II = (1 -f x2) dx 2x>jdll, and the equation becomes (IIa + a*)z = 0.
Let the variables x, y be now changed to , ij, so that II = d(. Therefore,
since H ({) = 1,
n () = (1 +.r3) lx-2xyl = 1.
Therefore, by (3383) ,

=
= d*,
1+ar
2xy
from which, by separating the variables and integrating, we obtain
x>y + y = A
and, by (1436),
= tan-'a; + 5
Also, since EL (?) = i/j = 0, (l + x1) iM2xynv = 0.
Therefore

(1),
(2).

= _^L_ = *?,

the solution of which is equation (1). Thus

=tan-1 and ri = x2y + y.
The transformed equation is now (iZ^ + a5) z = 0,
and the solution, by (3523), is
z = <p (;) cos ai + >p (>;) sin a4,
arbitrary functions of the variable, which is not explicitly involved, being
substituted for the constants (3389). Therefore finally,
z = <f> (x'y + y) cos (a tan "'!) + > (x'y + y) sin (atan"'a;).
MISCELLANEOUS EXAMPLES.
3551

Uix+V*+Uu = 0.

Put djg + du. = a*. Thus w^+a'tt = 0, the solution of which, by (3523),
, is
u = <j> (y, z) cos ax + \p (y, z) sin ax,
arbitrary functions of y and z being put for the constants A and B. Expand
the sine and cosine by (764-5) ; replace a! by its operative equivalent, and,
in the expansion of sina.T, put ai/< (ij, z) \ (y, z) ; thus
u = f(y, )--fj (<*,+ W <p (y, *) + Jj

<k) t (y, )-&c.

(y.

x(s

[See (3626) for another solution

3552
Here

= vyx.

u = (dx+d+d.)-1 (ei/z + 0)
= {<J_.-L,, (d,+4)+d-*(d, + ^5-...}(*y* + 0).
Operating upon xyz, we get
m = ix>yz -fr* (z + y)+ r\x\

DIFFERENTIAL EQUATIONS.

528

the rest vanishing. For symmetry, take ^rd of the sum of three such
expressions ; thus
u = i{ A (** + yl + - i (A + V + y'z + yz* + z*x + za>*) + \xyz (x + y + z) } .
Operating upon zero, we have, in the first place, d.z0 = <j> (yz)
instead of a constant, therefore d.^Q =
(yz), &c.
The result is
{l-x(d,+d,) + &(d,+d,y-...}t(yz) = e-Wf (yi) = *(</-*, z-x)
(3493) the complementary term.
3553

Otherwise, putting dv + dz = D, we have, by (3478),

(d.+S))-lxyi = e-^d.^xyz = e~^d.x{x (y+x) (z+x)}, (3493)
= e-'*>{lz,yz+{xt(i,+z) + $xt}
= lxi(y-z)(z-x) + $z'(y + z-2x)+ixt,

which agrees with the former solution.

3554

aux+61(+CMir = xyz.

Substitute x al, y = brj, z = c(, and the equation becomes

which is solved in (3552).
The same methods furnish the solution of
3555
aux +bu+ cux = xnynzp.

3556

xZx+yZy = 2xy\/a2z\

Put
.'. irz = a cos v . in,
3557
Put

.'. 2 = a sin (ry + c).

axux-\-byuy-\-czuxnu = 0.
x = ir, y = vb, = r;
.-.

3558

z ss a sin v,
.'. wv = 2xy,

m^ + i^-nu = 0,

.-. by (3544) u = (x, y , z7)".

The solution of any P. D. equation of the type

F(xdx, ydy, zdz, ...) = 2Axmy"zp ...

is, by (3488) and (3557),

u=x
A*mfsP
F(m, n,p, ...)

!
F(xdx,ydt, zdz, ...)

o.

SYMBOLIC METHODS.

529

3559 Ex. :
xux+yuy au = Q
where Qm = (x, y)n (1620).
Here u = (m a)~lQm+ JJa. When a m, this solution becomes inde
terminate. In that case, as in (3526), assume
ua= va.-. u = Qa~Qa + ra.
ma
in a
Diflferentiate for a, by (1580), putting Qa first in the form

thns

u = i Qm (log x + log y) + Vm.

Similarly, the solution of

3560
is

xux+yMv+sua mu = Qm
w = iQm (log x + log y + log 2) + Vm.

3561

^Wx+yug+zu, c.

The solution, by (3560), is

u = ^c(loga;-|-log2/ + log)-r-70.
3562

*u ftw^+a1** = 0

or

ad,)1* =0.

* = (d,-od,)",0 = (4-ad,)"1 ^ (y) e"^

(3472)
by putting a<i for w and ^ (y) for (7. The second operation produces, by
(3476),
z = eaxd{x<p(y) + 4'(y)} =x<t>(y+ax) + +(y + ax).
(3492)

3563

tfzteifziy+xzxyzy = 0.

This reduces to
(xd. + yd^) (xdx yd) z as 0.
Here = xdi+ydy, and = 0 in (3544),
therefore

z = (, y)+ ("S ~ ) >

the second term being obtained by substituting y~l = y, and so converting

the second factor into (xd.+y'dy). The above may also be written

JF and / being integral algebraic functions.

3564

to-%+26^+2aX = 0.

Putting y = aij, this equation is equivalent to

(d.-d, + 26) (i, + d,)* = 05
3 Y

DIFFERENTIAL EQUATIONS.

530

putting x = log a' and i) logtj', this gives, by (3544),

z = (e*, e_,)-*"*+(e*, e") = e-Ia**f (e"n) + i (e*-*)
= e-fc**i,(y+aa>)+/(y-a0,
the functions being algebraic and integral.
3565
.-.

to ahiiy =

(cr,

w=(c?^-a^)->(*,y)
(3515)
= (2ad,)-> { (d.-ad,yl-(d.+ad,)-> ) <p (,, y)

(3470)

= (2a<i)-1 { eaxd' Je-"^ o (a;, y) ixe"*** |^ f (x, J/) ck }

(3470)

= (2a)-lJ{tf,(x, y+ax) *.,(x, y ax) ] ay,

since

eOId0 (x, y) = <j> (x, y + ax)

Here

*, (x, y) =

3566

If </>

(3492).

(as, y-ax) (fa + ^ (y),

v/) = 0, the solution therefore becomes

M = V'i 0/+OJ?) +xl

Oct')

[BooZe, ch. 16.

For the solution in this case by Monge's method, see

(3433).
3567

zxazy = emx cos ny.

z = (d,adl/)-1e""' cobiiij =

f e"1***

cob ny dx

(3470)

= e^f Je-'cosn (y-ax) dx (3492), and this by Parts, or by (1999), is

= erx,'vemx | m cos n(y ax) ansinn (y ax) j {m1 + a?n*yl + e* <p (y)
.. z = e'"-" | m cos ny an sin ny j (m' + aV)"'+^ (y + ax), by (3492).

3568

x-az^ = 0.

* = (dt-ad.u) -1 0 = e"'"^ (?), by (3472),

<p (x) taking the place of the constant C.
Therefore
z = p (x) + a^2,+as*>t, + &c. (3492)
Otherwise, to obtain z in powers of x, we have, putting i5 = o ,
Zi.bht = 0,
.-. ={(i+M})(<l.+64)}-,0 = e^(0+e-6^(<) (3518);
then expand by (150).
3569

Ztx+Ziy = cos nx cos my.

SYMBOLIC METHODS.

531

z = ((?>* + d*)"1 cos nx cos my.

Treating da> and cos my as constants, we have, by (352G), putting d2y for o,
x = corn* (dty 2) coamy+jl cosaa; + 5 sin ax, or by (3496),
= cos nx cos my ( m1 n') " 1 + <p (y) cos (z<Zy) + ^ (y) sin (xrfj,) ,
A and B becoming f (y) and >/< (i/).
3570

z.i4,+2zH+z.^+a2z = cos (m<+w/).

Therefore
(+*)(*- ia) z = i {e'<*'+> + e-o*+*M}1
where * = dt + d+. Therefore, by (3510), with x = e*, y = e*,
U -(i+)

a (m + n) j

a* (rc + n)
3571 Prop. I.To transform a linear differential equation
of the form
(a+bx+cx*...) M+('+6'.r+cV...) M(-d*+&c. = Q ... (1)
into the symbolical form
/.(D) u+fiiD) eu+f2(D)e*u+&c. = T

(2),

where Q is a function of .r, I7 a function of 0, x = e" and

D = de.
Multiply the equation by x* ; then the 1st term on the left
becomes, by (3489),
(a + be + ce2,,+ ...) D(D-l) ... (D-n + l) u.
This reduces, by the repeated application of formula (3476)
with the notation of (2451), to
3572

aZ)'" it+6 (D-l)w e(tt+c (D-2)w eMH+&c.

The other terms admit of similar reductions.

3573 Conversely, to bring back an equation from the sym
bolic form (2) to the ordinary form (1), employ formula (3475)
so as to transfer e* to the left of the operative symbol.
3574

Ex.:

z% (<ciuu+7xu,+ 5u) =e*{D(I>-l) + 7D+5}tt

= e" (Da + 6D + 5) m = <?'(D + l)(D+5)u
= (D-l)(D + 3) <Pu

(3476).

DIFFERENTIAL EQUATIONS.

532

For the converse reduction, the steps must be retraced, employing (3475).
See also example (3578).
3575

Prop. II.To solve the equation

u+arf (D) e'u + arf (D) <f> (D-l)eu ...

+<*<(/>)

...*(2>-*+l)-= V,

where U is a function of 6.
By (3491)
<p(D) <p(D-l) ...tf,(Z>-n + l)e"*M= {*()*]".
Putting p"u for this, the equation becomes
(1 + Oap + ^p2... +ap")= IT.
3576

Therefore

u= {A(l-?1p)-1 + ^(l-'Z!p)-1-+^.(l-?1.p)-1} ^,
where jg2 ... <? are the roots of the equation
?" + ffl1jn-1 + 2?"-2 +an = 0,
and

^r = 7
T7^
?
(?r-?l)(?r-?2) (?r-?)

The solution will then be expressed by

u = AiU^+AiUi ... +Anun,
where ur is given by the solution of the equation
3577

' ur-qr4>(D)e>ur=U.

3578 Ex.:
(a;' + 5*' + 6**) uu + (4a; + 25a1 + 3uV) u, + (2 + 20a: + 36V) = 20xt.
Putting x = e*, and transforming by (3489),
(1 + 5e*+ 6e") D (D - 1) + (4 + 25e* + 36e") Dm + (2 + 20e' + 36e") u = 20e*.
The first term = B (D-l) + 5 (D-l)(7)-2) e'+6 (D-2)(D-3) e*u
by applying (347G). The other terms similarly ; thus, after rearrangement,
(D+I)(H+2)i+5(jD + l),e,u + 6D(I> + l)eMu = 20e*.
Operating upon this with {(D + l)(Z) + 2)}"', we get

or
therefore
if

(l + 5p + 6p,)U = e",
if p = (D+ l)(D + 2) "* tf ;
= {3 (l + 3p)-,-2 (l+fy)"1} e" = 3y-2,
y = (H-3p)-'eM and * = (l + 2p)-le*.

SYMBOLIC METHODS.

Hence
therefore
or

533

(l + 3p)y = e or y + 3 (D+l)(D+2)-Vy = e" ;

(D+2) y + 3 (D + l) e'y = e"(3+2), by (3474),
(D + 2) y + 3e* (D + 2) y = 5eM, by (3475) ;

that is,
(* + 3-e') y, + 2 (1 + 3*) y = 5a;'.
Similarly
(a; + 2*') + 2 (1 + 2x) z = 5x*.
Solve these by (3210), and substitute in w = 3i/ 2z.
3579

Prop. III.To transform the equation

u+<}>(D)ereu= U into v+<j>(D+n) e^v V,

put

u = en9v

and

U = en$ V.

Proof.By (3474), because 0(D) e(ntr)'v = e"'p (D+n) f*v.

3580

Prop. IV.To transform the equation

u-\-<f>(D)er9u= U

3581

where P tiMl =

into

v+\ji(D) e^v = V,

(2>-r) (fl-2r) ...

Proof.Put m =/(D)t> in the 1st equation, and er*f{B)v=f{Br)er,v

(3476). After operating with /"' (D) it becomes
+ 0 (D)/(D-r)/-' (D) e" = /-> (D) IT,
therefore
<p (D) /(D r) /-1 (D) = i// (D) by hypothesis ;
therefore

/(D) = $fg/<I>-r> = tgi^/(D-20,

and so in in/.

Also U = / (D) F.

3582 To make any elementary factor xC^) f

De_
come, in the transformed equation, x (P i w')> where r is an
integer ; take $(D) = X (Dnr) Xi (#). See example (3589).

3583

To make any factor of * (D) of the form

disappear in the transformed equation, take 4>(D) = xi {D),

where Xi(D), in each case, denotes the remaining factors of
^(D). See example (3591).

DIFFERENTIAL EQUATIONS.

534

3584 In the application of Proposition IV., differentia

tion or integration will be the last operation according as
Pr i ( { (3581) has its factors, after reduction, in the
y \P)
numerator or denominator, and therefore according as \p (D)
is formed by algebraically diminishing or increasing the several
factors of <j> (D). However, by first employing Proposition III.,
the given equation may frequently be so prepared that the
final operation with Prop. IV. shall be differentiation only.
See example (1).
For further investigation, see Book's Diff, Eq., Ch. 17, and Supplement,
p. 187.
3585 To reduce an equation of the homogeneous class
(3531) to a binomial equation of the same order of the form
Vnx+qy = XThe general theory of such solutions is as follows.
the given equation be

Let

+</ {(/>+1)(Z)+a2) ... (D+a^-'e-u = U ... (1),

alt di, ... an being in descending order of magnitude.
u = e~"l*v, by Prop. III.,

Putting

v + q {D{D-a~^a~^ ... (D-~a~^an)}-lenev = eaU ... (2).

To transform these factors, regarded as <f> (D), by Prop. IV.
into r(D) =D(D I) ... (Dn + 1), we convert D into D + rn
(3582), r being an integer.
Hence for the pth factor we must have
D + T71 C^ + dp = Dp + 1,
3586 and therefore ! ap = rn+pl
If this relation holds for each of the constants
equation (1) is reducible to the form
3587
y+q{D{D-l) ... (D-n+l)}-le"y= Y

(3).
... a
(4),

which, by (3489), is equivalent to ynx+qy = Y = X.

y being found in terms of x from the last equation, and,
v being = P ^

y (3580), the solution will result from

while U and Y are connected by the same relation as u and y.

SYMBOLIC METHODS.

3589

Ex. 1: Given

535

aV+lSaV + 84xux+ 96u + 3xsu = 0.

Putting x = e' and employing (3489), this becomes

{D(D-l)(D-2) + 18Z> (D-l) + 84D+96} +3fl" = 0,
or
(D+8)(D+4)(D + 3) u+3e"u = 0,
therefore
w+3 {(D + 8)(I>+4)(D+3)}-1e!Mit = 0

(1).

Employing Prop. III., put

u = e~8'v,
therefore (3476)
+ 3 {D (Z>-4)(D-5)}-1e3' = 0
(2).
To transform this by Prop. IV. into
2/ + 3{D(i)-l)(D-2)}-1e3*2/ = 0
(3),
we have
P tS^. = J(P-l)(J-2)(D-3)(J-4)(D-5) ... _ fD_nffl_21
3^(D) D(D-4)(2)-5)(D-3)(D-7)(D-8) ... ~ 1
n
;'
.-. = (D-l)(D-2) y,
.: = e-"(D-l)(P-2)y
(4),
and the solution is obtained by differentiation only, performed on the value
of y as obtained by the solution of (3), that equation being equivalent to
D(D-l)(2?-2)y+3e"y = 0, or, by (3489), y+3y = 0.
If, however, Prop. IV. were used to pass directly from (1) to (3), we
should have
p <P () _ D (2?-l)(J-2)(J-3)(2?-4)(2?-5) ...
K2)) (D+8)(D + 4)(D+3)(D + 5)(D + 1) ...
1
(I>+8)(D+5)(D + 4)(D+3)(2)+2)(D+l),
and equation (4) would involve integrations of y as high as D~6y.
3590 Note.By the literal application of Rule IV., the right side of
equation (3) ought to be V = {(D 1)(L> 2)}"1 0 ; but no such term is
required when the original and transformed equations are of the same order,
for in such cases the arbitrary constants introduced by the operation upon
zero disappear with the terms containing them in the final differentiation.
The result is the same as if the operation upon zero had not been performed.
In the following example, V has to be retained.
3591 Ex.2:
(x-x>)uu + (2-Ux')u,-3Qxu = 0
(1).
Multiply by x, transform by (3489), and remove ew to the right of each
function of B by (3476), thus
u_(n+4XD + 3)euu = 0
D(D + 1)
Transform this by Prop. IV. into

We have

u =* P,^j~ * (D+4)(D + 2) v,
V= {(D+4)(D + 2)}~10 = Ae-v+Be-" (3518).

The operation upon zero is required in this example (see 3590), because (3)

536

DIFFERENTIAL EQUATIONS.

is of a lower order than (2) ; but only one term of the result need be retained,
becanse only one additional constant is wanted. Hence (3) becomes
(.D+l) -(D+3) e"t> = (D + l) Ae~u = -A<T.
Changing again to x, this equation becomes
(x*-x*) v.tatv+A = 0.
The value of v obtained from this by (3210) will contain two arbitrary con
stants. The solution of (1) will then be given by
u = (D+4)(D+2)v.
3592 Ex. 3 :
u^-n (n + 1) x-*it-q*u = 0,
[Boole, p. 424.
n being an integer.
Multiplying by x* and employing (3489), this becomes
-33{(D + n)(I>--l)}-1e" = 0.
This is changed by Prop. III. into
v-q* {D (D-2n-l)}-leuv = 0, with u = ev,
and this, by Prop. IV., into
y-q'{D(D-l)}-1ey = 0 or yu-qty = 0

(3489).

y being found from this by (3524), we then have

But, by (3484),
F(xd,-m) = xmF (xd.) x~m,
:. u = x~"x (xd,) x-1 . x* (xd.) x-1 ... x*-1 (xd.) aTJ" * 1 y,
or
=-<" u (x*d.)'
1y
= x-'-1 (xtdr)"x-u*1 (AtT+Be-") (3525).
This may be evaluated by substituting % =
Vol. xvn., p. 77.)

(See Educ. Timet Reprint,

3593 Ex.4:
M2x-a,t2-n(n + l)a!-3 = 0.
The solution is derived from that of Example (2), by putting q = ad
and arbitrary functions of y after the exponentials instead of A and B ; thus
u = a""-1 (x*d.)n x--' 1 {e"^ <p (y) + e' axd >^ (y) }
= x--' (ar,d.)'x-*+*{t (y+aa>) + ^(y+oa)}, by (3492).
[Boole, p. 425.
3594:

(l+aa*)utt+eumxniu = 0.

To solve this equation or its symbolical equivalent obtained

by (3489), viz.,
3595
+a(g^nV* = 0.

Substitute t = f
= in the solution of uun*u = 0, by (3523-4).
J '(1 + 0*')

SYMBOLIC METHODS.

3596

537

Similarly, to solve the equation

or, the same in its symbolical form,

dx c in the solution of w r?u 0.

Substitute t = fI 77-=
J x^/{x2 + a)
(3596) is obtainable from (3593) by changing 0 into 0.
3598

Pfajf's equation,
(a-t-&#) afUte+ic+ex") xu,+ (f+gxn) u= Q.

When Q = 0, the symbolical form becomes

b(B-n)(D-n-l)+e(D-n)+g
M+
aD{JD-l) + eD+f
6 U-

W'

If n be not = 2, substitute 20' = n0, and therefore 2de = nd9..

3599

Thus

+^:ffig:^ = o

where a a, are the roots of the equation

6 Oct )(Ja l)+e(|tt-n)+5r = 0
and /3j, /3, are the roots of
a ln(i (|}l/3-l)+c |/3+/ = 0.
Four cases occur

(2).

(3),

3600 I.If Oj a, and /3, /3, are odd integers, (2) can be reduced by
Prop. IV. (3581) to the form v + 1
ewv = 0,
and then resolved into two equations of the first order.
3601 II.If any one of the four quantities a, /3 c^ /3 a, /} n, /3,
is an even integer, (2) can be reduced by Prop. IV. to an equation of the
first order.
3602 HI.If /3, /3, and a^ + u,
/3, are both odd integers, then, by
Props, in. and IV., (2) is reducible to (3595).
3603 IV.If aj a, and a, + a, flx /3, are both odd integers, (2) is
reducible in like manner to (3597).
[Boole, p. 428.
Note.The integers may be either positive or negative, and when even
may be zero.

3 z

DIFFERENTIAL EQUATIONS.

538

SOLUTION OF LINEAR DIFFERENTIAL EQUATIONS

BY SERIES.
3604

Case I.Solution of the linear differential equation

f0(D)u-fl(D)e"u = 0

or /, (xda) u-f, (xd.) tfu = 0,

in which / (_D) , fx (D) are polynomial expressions of the form

a0 + aiD + a2D\..+anDa and/0(D) = (D-a)(D-b)(D-c) ....
3605
3606

Let <p (D) = f, (D) 4-f0 (D), and let

<t>(a) = l+^(a + ?-).'r+*( + 2r)4>(a+r);s*r
+ <p(a + 3r) <i>(a + 2r) $(a+r) ar"" + &c.

Then the solution will be

3607
= Axa<t>(a)+Bxb4>(b) + Cxc<P(c)+&c.
Proof.Operating with /0-1 (D) and writing p for <j> (D) e",
u-pu =f~i(D) 0 = Aeaa+Be'" + &o. (3518)
Therefore, by (3515), u = (1-p)'1 (Ae'"' + Beb9+ ...)
= (l+p+p'+ ...)
(1 +p+P5+ ...) Be" + Ac.
Now in each term substitute for p". the value in (3491), and remove D by
formula (3474).
Case II. Solution of
3608 /0(Z>) u+MD) e<u+MD) e**u ... +fn(D) e**u = 0
where

f0(D) = (D-a)(D-b)(D-c) ...

Let
= l + F1(a + l)x+Ft(a+2)a?+&c,
where the coefficients Fl(a + 1) or vu F1(a + 2) or vlt &c. are
determined in succession by the formula
3609 /o (m) vm+fx (ro)

...+/ ("0 - = 0 and , = 1

(2)The solution will then be expressed by

3610

u - Ar(a)+lto4(&)+CV(c)+&c.

Proof.From (1)
u = { 1 + (D) e" . . . + f (D) e" } -/,-' (D) . 0
(3),
where
fr(D) = /r(D)-8-/,(2>).
Here /."'(D) 0 = { (D-a)(D- 6) ... } " 0 = .4eM + 5e"+ ... (3518);
and

{l+f, (J?) a' ... +* (D) e"} _1 = 1+F, (D) e' + f, (D) e"+ ...

SOLUTION OF LINEAR DIFFERENTIAL EQUATIONS. 539

To determine Fv Fr &c, opei'ato upon each side with { 1 + <p (D) e + &c. } ,
and equate coefficients of powers of e; thus formula (2) is obtained. (3)
now becomes
u = {l + F1(D)e, + Fi(D)e1'+...}(Aea+Beb+...)
(4).
Multiply out; apply (3474), and put x for e*.
3611

Ex.:

xtu2x(a+bl)xux+abuqx>u = 0,

or, by (3489),
(D-a)(D-b) u-qeu = 0.
Here
/, (D) = (D-o)(D-6), /,() = 0,
(D) = -g.
Therefore (2) becomes (to a)(in b) vm = qvm.t,
therefore F Fs, &c. vanish, and Ft (a) = 1,
(a + 2-a)(a+2-&)
gir,(a + 2)
.F4(a+4) =
(a+4-a)(a + 4-6)

2(o + 2-6)'
4.2 (a + 4-6)(a+2-6) '

Therefore * (a)
2?!
+
+ ...
w = 1+
^2(a+6-2)^4.2(a-6
+9^.
4)(a-6 + 2)^
Similarly we find F% (6 + 2), Ft(b + 2), &c, and thence (i); and, sub
stituting in (3610), we have
M=

3612

Axa+

W +
+ ...
2 (<H-b-2) 4.2 (a-6 + 4)(a-6+2)

+lfe>+

^ +
+ ...
2(6-o+2) 4.2(6-o + 4)(6-o + 2)

The solution is arrived at more quickly by formula (3607).

producing the same series by the value of * (a).

We

Similarly with * (b).

3613 When /0(D) has r factors each =Da, the corres

ponding part of the value of u in equation (4) will produce
3614

A0+Ax log x+At (log^)2... +4,_1(log#)'-1,

where the coefficients A0, Au ... are each of the form

C0*+C1<ro+1+^o+3+ ...
3615 But if any one of the quantities Fr (a+r) = 0 (3608),
then Gr = 0 also.

DIFFERENTIAL EQUATIONS.

540

Peoof.-ffl (D) now contains a term of the form

...+er6r~l) = e"v, say.
The corresponding part of u in (4) is
{l + F^D) *+...} e"t
= {e"+ef1),J'1(I>+a+l)+e("*,),J'1(D+a+2) + ... } by (3475).
Expand each function F by Taylor's theorem in powers of D, operate upon v,
and arrange the result according to powers of 0.
In practice, proceed as in the following example.
3616

Ex. :

xute+ux+(fxu = 0.

Multiplying by x and changing by (3489), this becomes

D'u+tfeuu = 0. Dsm = 0 gives u = A+Bd.
Substitute this value and operate with D, considering A and B as variables,
and equate to zero the coefficients of the powers of d ; thus
D*A + ifPA + 2DB = 0, L>B + gV.B = 0.
Then change B into m, and er'A into am.r, to obtain the relations
tn'am + g'o^.j + 2mbm = 0 ; n?bm + ^bm_t = 0,
which determine the constants successively in terms of a, and b0 (which are
arbitrary) in tho equation
u = a0 + aix'+aix*+...+\ogx(bl>+brii, + b^+...),
which thnB becomes the solution sought.
{Boole, Diff. Eq., p. 439.

SOLUTION BY DEFINITE INTEGRALS.*

3617

is

Laplace's method.The solution of the equation

^(dl)++(d>=0

(1)

u = CjV^*?*^)-1} dt

(2),

the limits being determined by

<H>=0
(3).
Peoof.Assume = e je*'T<f<, and substitute in (1), putting f(dM)e*t
= ?(t) e" (3474), thus
\xe"<t> (0 Tdt+ f e") (0 IW = 0.
* This method of solution is merely indicated here, and the reader is referred to Boolo'i
Diff. Eq., Ch. xviii., for a complete investigation.

SOLUTION OF DEFINITE INTEGRALS.

541

Integrating the first term by parts, this becomes

^ (0 r-J* [dt ( * (o t]

(o r] a = o

(4),

an equation which is satisfied by equating each term to zero. The second

term thus produces a value of T by integration by (3209), and this value
substituted in the first term, and in the value of w, gives the results (3)
and (2).
3618

Ex. (1) :

xUte-\-auxq*xu = 0*

(4).

Here ^ (dx) d^q1, \j> (i) = adx, <p (t) =

q*, \p (<) = at.
(2) and (3) become
=cJe*<0,-S')"ld<; ^(f-g^O;

Hence

a being positive, the limits are t = q, and, putting t = q cos 8, we find

u = 0 [VMO"sin-,0<i0

(5).

3619 The solution in series by (3608) is as follows. Equations (1) and

(2) of that article are in this case
D (D+a 1) u oV'm = 0 and m (m+a 1) vm qtvnt.1 = 0.
Thus, a in (3608) =0, and 6 = 1 a.

Therefore (3610) becomes

I T2(o+l)T2.4(a + l)(a + 3)T

+^- { 1+ 2(1^ + 2.4(3-!aX5-a) +&- }

<

Both series are convergent by (239 ii.).

The results deduced by Boole are these
3620 (5) is equivalent to the particular integral represented by the first
series of (6).
3621 A second particular integral, by assuming = e(1"0),t>, is found to
be, when 2 o is positive,
u = <V"a re'ICO*W-a0Z0
(7).

3622

When a lies between 0 and 2, the complete integral is

u = C1{'<P*,Bina-10d0+Cizl-a["e''*"*' Biu*-a0d6

(8).

* The method by definite integrals is elucidated by Boole chiefly in the solution of this
important equation.

DIFFERENTIAL EQUATIONS.

542

3623

But, if a = l, the solution becomes

u = fV" { A + B log (x sin' 6)} d8

(9).

3624 If a does not lie between 0 and 2, then, if a be negative, put

a = a 2n, and replace the first term of (8) by
0, (xd, + a -l)(zd,+a'-3) ... (xd, + a'-2n+l) JV^'sin"'"'^ ... (10),
the transformation being effected by (3580).
3625 And if a be positive and > 2, put = e(1"0l* = x1'"v. This con
verts a into 2 a, a negative quantity, and the case is reduced to the last
one.
3626

Ex. (2).To solve by this method the P. D. equation

+%+* = 0

(1)

when r= ^(a^+y2).
Eliminating x and y, (1) becomes
rttir+ur+ruu = 0
(2).
Now the solution of this equation is number (9) of Example (1), if we
change x into r, q into idz, and A and B into arbitrary functions of *. We
thus obtain
= (V""1* { * () + 4> () log 0 sin' 0) } dd

(3),

or, by (3492),
=| ^{ + ircoB0j dfl + j \p (z +ir cob 6) log (r sin' 0) <?0

(4).

See (3551) for another solution.

3627 If u be the potential of an attracting mass at an external point,
and if = F (z) when r = 0 ; then, since log r = oo , i (z) must vanish ;
therefore

jF(z) = J ^ (2) a"0 = t ().

Hence (4) reduces to

= -i- f*p + 1> C08 6 J a"0.

Parseval's TJieorem.
3628

If5 for all values of ?<,

4 + Z?it + Ctt,+ ... =

and

4' + F-1 + C'-!+... = ^()

(1),

DIFFERENTIAL RESOLVENTS.

543

then
A A'+ BB+ . . . = i-J* [< (e) * (e) +

(c") * (o] rf*.

Proof.Form the product of equations (1), and in it put u = eiB and e''"
separately, and add the results. Multiply by dd, integrate from 0 to ir, and
divide by lie.

P. D. EQUATIONS WITH MORE THAN TWO

INDEPENDENT VARIABLES.
3629 By means of Fourier's theorem (2742), the solution
of the equation
iitt h2 (iite+ihy+Ute) = 0
may be deduced by a general method in the form
u = (l+dt) JJjJJJ<U+aM) * (

0 dadbdcifKdyLdv,

the limits of each integration being oo to oo , and the func

tion i// being arbitrary and different in the two terms arising
from the operator (l + dt).
Boole, Ch. xviii., and more fully in Cauchy's Exercice d'Analyse MathSmatique, Tom. I., pp. 53 et 178.
3630 Poisson's solution of the same equation in the form of
a double integral is
u=(l + dt)\ I t sin tyix + ht sin taint), y + ht sin cosy, + Af cosi)d<?ij
with the same latitude in the interpretation of \p.
[Gregory's Examples, p. 504.

DIFFERENTIAL RESOLVENTS OF ALGEBRAIC

EQUATIONS.
3631 Theorem I. (Boole)."If ylt yt ... yn are the n roots
of the equation
+l = 0
(1),
and if the mth power of any one of these roots be represented

DIFFERENTIAL EQUATIONS.

544

by u, and if a = e', then u as a function of 0 satisfies the

differential equation

and the complete integral of the same will be

u = Ciyr+Cy"+ ... +

3632 " Cob. I.If to = - 1 and if n be > 2, the differential

equation
D<- u-( Z)- -lY" "e"** = 0
n \ n
n
/
has for its general integral

y > y

being any n 1 roots of (1).

" If 6 be changed into 0, and therefore D into X>, the

above results are modified as follows :
3633

" Cob. II.The differential equation

has for its complete integral

= Ciy"+Ci}h +cyn,
Vu Vi

3634

Vn being the roots of the equation

ay*y*-x+a = Q

" Cob. III.The differential equation

u-n (D-2)*-" ^!L=lx)+l.y,,"1)J~>u = 0,

supposing n > 2 has for its complete integral

yi>y Vn-i being any n 1 roots of (2).

(2).

DIFFERENTIAL BESOL YEN TS.

545

" Theorem II. (Barley).

3635

" The differential equation

arLraxY >u I
\ n

>rx+

1 ) ofn
I
= 0
is satisfied by the mth power of any root of the equation
n

I jca.
\n

y'-Mf-'+a = 0,
u being considered as a function of x.
3636

" Cob.The differential equation

(^=r^-^)(""r,w<"

-(n-l)r (^xdx- ^-iy\vru = 0

is satisfied by the ma power of any root of the equation
yn nyn-r+(n 1) x = 0."
{Boole, Biff. Eq., Sup. 191199.
3637 See also Boole, Phil. Trans., 1864; Harley, Proc. of the Lit. and
Phil. Soc. of Manchester, Vol. II. ; Bawson, Proc. of the Lund. Math. Soc,
Vol. 9.

4 A

CALCULUS

OF

FINITE

DIFFERENCES.

INTRODUCTION.

3701 In this branch of pure mathematics a function <p (/)

is deuoted by ux, and <j>(x-\-h) consequently by nx+h. The
increment h is commonly unity . If Ax denotes the increment
h, and Anx the consequent increase in the value of uxi we have
3702
3703

Aw, = ux+Ax-t*x.
When Ax diminishes without limit, the value of
Aw*
Ax

Qr

wj+ax ux
Ax

.s

du,
dx '

3704 The repetition of the operation A is indicated as

follows :
AAux = A2ux, AA*ux = Asux, and so on.

3705

Ex.: Let ux = x\
x =1 2 3 4 5
1 4 9 16 25
<Xe' = 3 5 7 9
AV= 2 2 2

...
...

FORMULAE FOR FIRST AND nth DIFFERENCES.

If
3706
3707

ux = axn + bx"-1 + ex"-2 + &c,

AX = an (n-1) ... (n-r+1) af-'+c^^+Ac.,
ArS= an (n-1) ...3.2.1.

FORMULA FOR FIRST AND 11th DIFFERENCES.

54,7

3708 Hence the nth difference of a rational integral func

tion of the th degree is constant.
3709

So also

A,r" = 1.2.3. ..n.

3710

Notation.Factorial terms are denoted as follows :

3711
3712

Thus

x(x-l) ... (x-m+l)=z,

3713

1 3~n ...V(aj+m
T
n5*-1o;(a!+l)
1)
Hence [m, ml, and m(m' are equivalent symbols.

3714 According to (2452), as'"' would here be denoted by as'"'. The

suffix, however, being omitted, it may bo understood that the common differ
ence of the factors is always 1.
3715

Aa?<"' = mx{n-l\ AVW) =

An\vM = m{m) ,

and, if m < 11, AV"0 = 0, since Ac = 0 if c = constant.

3718

A*<> = m<r(~m_1),

AV"^ = (

-m-)

Attiffl = (ux+1
3720
Au-*' = K- wx+m)
3722 Ex.:
A (as +&)<"> =
3724

"* 1 .

A (a* + &)< -m) = - am (ax + by --!)_

3726

Alogw, = logh+^(, A log

= log i^2-.
<
W J
W.r-i+l
Aa* = (a- 1) ax,
AV* = (am- 1)"<T*.

3728

A- Z

Proof.

= (2 sto I)' j a*+i+ J+H j .

A sin (ax + b) = sin (aa: + a + 6) sin (ax + 6)

= 2 sin sin yax + b-\ J.

That is, A is equivalent to adding a~T * to the angle and multiplying the
c
sine by 2 sin .
u

548

CALCULUS OF FINITE DIFFERENCES.

3729 Conversely, the same formula holds if the sign of n

be changed throughout.
EXPANSION BY FACTORIALS.
3730

If A**(0) denote the value of A>() when x=0,

then <f> (#) = <f> (0) + A< (0) * +

*<2> +

.r(3) + &c.

3731 If Ax=h instead of unity, the same expansion holds

good if for An4>(0) we write (A"^(,c)-r/t")x=0; that is, making
x = 0 after reduction.
Proof.Assume ^ () = a0+a,:e + aJa:<z> + a^;!s, + <fcc.
Compute A0 (x), A'^> (a1), &c, and put z = 0 to determine au Oj,

3732

&c.

GENERATING FUNCTIONS.
If "Mx he the general term in the expansion of <p(t),

then <j>(t) is called the generating function of ux or <p(t) = Grux.

Ex.: (1 0"2= C?(as + 1), for x + 1 is the coefficient of t* in the
expansion.
3733

Gfu, = ^(0,

3734

GAux = (-L-l)<K0, - GA-u, = (j

Proof.

3735

GX+1 = M, ... Gux+n = *p..

(?Aw, = 6/u.,,! Ou &c.

THE OPERATIONS E, A, AND dx.

S denotes the operation of increasing x by unity,
Eux = u,+1 = h^+Au, = (1+A) *

The symbols 17 and A both follow the laws of distribution,

commutation, and repetition (1488-90).
3736

E = 1+A = ed*

or

eD*

Proof. Eux = uxtl = ux+dxux+\du.ux+ ^^d^^+Sus.

= (l + dx + ^d2x+ d^+Snc.) ux = e*xux.
By (1520), Aa being unity.
* Tho letter d is reserved as a symbol of differentiation only, and the suffix attached to
it indicates the independent variable. See (1487).

FORMULA FOR FIRST AND tt DIFFERENCES.

3737

Hence A = eD-l

and

3739

Consistently with (3735)

549

D = \ogE.
denotes the diminishing

x by unity ; thus
E~*ux = ux_v
For Eux.l = ux, .. ux.1 = E'1ux.
Ux+n in terms o/ux and successive differences.
3740 ux+n = x+Aux4- C (n, 2) A*,+C (n, 3) Aswx+&c.
Proof. (i.) By induction, or (ii.) by generating functions, or (iii.) by
the symbolic law :
(ii.)
<?ua+, = (-j)V(o = { i+ (4 -i) 5 VcoExpand by the Binomial theorem, and apply (3734).
(iii.)
nx.n = E*ux=(l + A)ux.
Apply the laws in (3735) by expanding the binomial and distributing the
operation upon ux.
Conversely to express A"iix in terms of ux, tix+u ux+2, &c.
3741 &"ux = ux+n-nux+n_1+C(n,2)ux+n_2 ... (-1)BX.
Proof.
A*wx = (E-l)nux (3736).
Expand, and apply (3735) as before. Putting x = 0, we also have
3742
3743

3744

A"0 =

C,2m_2 ... (-1)" u0.

= (x+n)m-n (a?+n-l)"
+ C(n, 2) (cr+n-2)m-&c.
A"0m = nn-n (n-l)"+C(n, 2)(-2)m
-C(n,3) (n-3)"+&c.

3745 Ex.: By (3717) A"0" = n!

obtained.
3746

Hence a proof of theorem (285) is

Axvx = {EE'-l) uxvx,

where E operates only upon ux and E' only upon vx.

Proof.

&uxvx = uxtlvxtl uxvx = Eux.E'vxvMva = (EE'l) uxvx.

Applications of (3746).
3747 Ex. (1):
Atv>.= (-l)''(l-EE')*u.vx.
Expand the binomial, and operate upon Hie subjects tix, vx; thus
3748

A,.= (-l)*j.,-n.tl,+ 0(,2).+Jt;.t,-4cj.

550

CALCULUS OF FINITE DIFFERENCES.

3749

Ex. (2) : To expand a' sin x by successive differences of sin a;.

A"a'sinas = jl?(l + A')-l j V sin as = { A+.EA' jVsina!

= { &n + n*kn-lEA'+C(n, 2) A""2 E1 A'' + &c. ] a* sin a

= A" a', sin * + /iA"- V" A sin x + 0 (, 2) A"J o*t! A' sina; + &c.
= a*{(a-l)"sina;+(<i-l)""1aA sina: + C(n,2)((i-l)''-VAssina; + &c. } ,
by (3727), while A'sina; is known from (3728).
3750 Ex. (3) : To expand A*^, in differences of m, and v, alone: put
-E = l + A, '=1 + A' in (3746), thus
Au,vM = (A + A'+AA')lt>
wliich must be expanded.
Anux in differential coefficients o/ux.
3751

&nux = d:ux+Ald:+i+A2dnx+2uJC+&c.

Proof.
A-m, = (e"'-l)nu, (3737).
Expand by (150) and (125) as if dt were a quantitative symbol.
(37G1).

See also

tt"u .

tjt
a
in successive differences
of
dxn
,u
J u.
3752

g={log(l+A)}V

The expansion by (155) and (125) will present a series of

ascending differences of u.
Peoof.
3753

e' = 1+ A,

.-. dx = log (1 + A).

EX,if = i, | = A-f + f -^+4o.

If C be a constant,
3754

(f>(D)C=<f>(A)C = <l>(0)C and

<f>(E) C = <j>{l) C,

Since every term of <p(I>), or of (j>(A)C, operating upon C, produces 0;

and every term of <j> (E) operating upon 0 produces 0.

3757
3758

HERSCHEL'S THEOREM.
<t>(et) = <f>(E)e-t
= *(i)+*()0.+*(tf)<H.I^+&c.

INTERPOLATION.

551

Proof.Let <j> (e<) = A0 + A^ ... +A e*'

= AJ'+A^t* ... +AnE"eot = (A^+A.E ... +AHEH) e-<
= <t>(E)e<=t(E){l + 0.t+^2+&c.},
and <f> (E) 1 = <p (1) by (3756).
A THEOREM CONJUGATE TO MACLAURIN'S (1507).
3759
<t>(t) = <f>(D)e-t
37 60

= * (0) + <j> (do) 0 . t +

Proof.

and

K) 02 . A + &c .

<p(t) = 1> (loge4) = tf> (log J?) e0' (3757)

= <f,(D)e<> t (3738) =^-(X) |l + 0J+ + &c.j,
? (D)l = 4>(0) (3754).

n being a positive integer,

3761
M

da" + 1 . 2 . . . (n+ 1) </a"+1 + 1 . 2 . . . (m+2) ete"+5! +

Proof.By (3758), putting <p (e<) = (e'-l)",

(e'-l)"= (^-1)"0.<+(^-1)0'.^+&c. = A0J+A,'0,-^- + &c.
1.^
1.2
Put < = <Z,, and employ (3736) and (3737).

INTERPOLATION.
Approximate value of ux in terms of n particular equi
distant values.
3762 If u is an integral algebraic function of the degree
n 1, A"x vanishes, and therefore by making x = 0, and
writing x for n in (3740),
3763

ux = M0+^A,+CX,,AX ... +C,,_1A-10.

This is formula (265). The given values are uQ, Aw0, A2z/0,
&c, corresponding to a, b, c, ...
3764 For an application of the formula to the problem of interpolation,
see (267), in which example x = 154 and ux = log 72 54.

CALCULUS OF FINITE DIFFERENCES.

552

3765 When the term to be interpolated is one of a set of

equidistant terms, employ (3741). &nUj 0, as in (3762);
therefore
3766

"-"-i+CM,2tt,_o-C,sMtt_3 ... + (-!)", = 0.

3767 Ex.: From sin 0, sin 30, sin 45, and sin 60', to deduce the value
of sine 15.
The formula gives sin 0-4 sin 15 + 6 sin 30" -4 sin 45" + sin 60 = 0,
or
-4sinl5+3-2v/2 + |v/3 = 0,
from which
sin 15 = (6 - 4 ^2 + ^3) = -2594.
The true value is -2588 ; the error "0006.
LAGRANGE'S INTERPOLATION FORMULA.
3768 Let a,b,c, ... k be n values of x, not equidistant, for
which the values of ux are known ; then genei'ally
3769
_
(g-6)(J-e)...(.r-ft)
(tr-)(.r-c)...(.r-AQ
K*
Wa(o-6)(a-c) ... (o-A-) ^ *(&-) (6-e) ... (6-A)
^ * (.r-a)(j--6)(^
(k-a){k-b){k-c)..:
c)...

Proof.Assume
m, = .4 (a> 6) (a c) ...(aft)
+
a) (a- e) ... (a> k) + 0(x-a)(xb)(x d) ... (xk) + &c.,
and determine A, B, C, &c. by making x = a, b, c, Sec., in turn.
If the values of a,b,c, ... k are 0,1,2, ... ], (3769)
reduces to
J770
_

* = .-i

1<2.3...(n-l)

.r(,r l) ... (a n+3)(cV n+l)

*-2
1.1.2.3... (n-2)

+M-S

2.1.1.2.3...(-3)

3771
^('"
f
w-i
*
(n-l)! U-n+l

C-u,-t +C-i.i",-_iic j
ctf-n+2
.r-n+3
'5

OF

MECHANICAL QUADBATUBE.

553

MECHANICAL QUADRATURE.
The area of a curve whose equation is y vx in terms of
n + 1 equidistant ordinates u, ult ... un, is approximately
3772

^ ^'+(^ -f)f!i +(x-3+"a)ff

. In'

3n* . lln'

, A A*u

+ (|'-2.+ 2|!!-^+12')^

Proof.The area is = f"

I uxdx. Take tlio value of x in terms of
Jo
u0, tt, ...
from (3763) and integrate.
3773

When n=2,

3774

= 3,

3775

n = 4,

ftt^ = w+4^+M-.
Jo
o

j,o,MlP = |(H+;i1+3,+H3),

Jo

= 14(u+^) + 64(l+tt;t) + 24^

4o

3776
n = 6,

= [M+2+K4+tt6+5 (w1+m6)+6m3] .

In the last formula, which is due to Mr. Weddle, the co

efficient of A6tt is taken as -po instead oi.-^g, its true value.
These results are obtained from (3772) by substituting for
each A its value from (3742).
COTES'S AND GAUSS'S FORMULA.
3777 These give the area of the curve directly in terms of
fixed abscissae.
They are obtained by integrating Lagrange's value of vx
(3769-71), and are fully discussed in articles (2995-7).
4 B

CALCULUS OF FINITE DIFFERENCES.

554

LAPLACE'S FORMULA.

3778

j\<&? = +,+.
- ^ (A- Atf.) + JL (AX-

the coefficients being those in the expansion of

f{iog(i+orProof.

Aw, = dx j *+
j

A3 ,A

j w
19 ... .

by (3736),
1

Hence, putting ux = Ajo^

and so on; then add together the n equations.

3779 Formula (3778) contains Alt,, A*2n, &c, which cannot
be found from u0} Mj . . . ti.
The following formula does not involve differences higher
than Att_i.
3780

fV* =

...+

- ^ (Aw..!- At*,) - 1 (A%.,- A20)-&c

Peoof.In the proof of (3778), change ,
into E,~AE I , ,
log(l + A)
log(l AB"1)
and put E~*icx = wx.i (3739) after expansion, and proceed as before.

SUMMATION OF SERIES.
3781

Definition:

Sx = wa4-Ha+1+Ma+8... + w,_1.

3782 Theorem:
2mx = A-1i/x+C,
where G is constant for all the assigned values of x.

SUMMATION OF SERIES.

555

Proof. Let (f>(x) be such that A0 (a;) = u then <f> (x) = A"'mx,
therefore = ^ (a + l) <j> (a). Write thus, and add together the values of
<m
Therefore, by (3781), 2ux = ^>{x)^(a) = A-'n, 9 (a),
and 0 (o) is constant with respect to x.
Taken between the limits x = a, # = 6 1, we have the
notation,
3783

'i%

or

X"a-lux = %ub-tua == A-X-A"1^.

Functions integrable in finite terms :

3784

3785

3786

3787

C7ss7.

^i-(M) = ^-^+C.
m+1

S(a#+6)<" = (g+jl^!!! +c.

a (m+l)

Glass II.

S*<-) = ^+C.

^SS^Vm.

S(^+6)^> = C-(g+^l).
a(m 1)

Formulae (3785) and (3786) are equivalent to the rules

(269) and (271).
They are the direct results of theorem
(3782).
3788

Sa* = -^7.
a 1

[By (3726).

Class III.If u be a rational integral function,

3789
$rx~V = {*+CXi2a+Cx,,aM- ...}..
Proof.By (3735) and (3736),
+! ... +atI-i = (l + -E + -Bs- +-EJ-')^ = ^^wa = (1 + ^r~1^
= the expansion above.
3790 The formnla has been given at (266) and an example of its appli
cation. The series there summed is 1 + 5 + 15 + 35 + 70 + 126 + to 100 terms.
The function ux which gives rise to these terms is found by (3763) to be
ux = (x* + 10x" + 35a2 + 50a + 24) 24.

556

CALCULUS OF FINITE DIFFERENCES.

3791 If this function be presented as u and 2*"'m, be required, we

first find = 1, ,= 5, ,= lo, &c. ; then the differences Aj A't^. A*u, =
1, 4, 6, 4, 1, and then, by (3789), the required sum, as in the example re
ferred to.
3792 For another example, let 2" z* = 1 + 2* ... + n* be required.
Here
Ax* = 3x* + 3x + 1, AV = 6* + 6, AV = 6,
therefore
AO1 = 1, A0 = 6, A>0=6
(1),
xs may now be expressed in factorials, and the summation may then be
effected by (3784). First, by (3730),
x"= x + 3x(x-l) + x(x-l)(x2) ;
therefore, by (3784),
2V = 2 (+ 1)" (3783),
, _ n(n + l) , 3(n + l)n(n-l) , (n + 1) n (n-l)(n-2) _ n*(n + lY
2lX
2~~ +
3
+
4
4
*
3793

Otherwise, by (3789), taking o = 0, we have

u, = x*, m0 = 0, A!t0 = 1, Asmu = 6, A*0 = 6, as above.

Therefore
_ n(w-l) 6n(-l)(n-2) C (-l)(n-2)(-3) _ w' (-!)'
0 '
2
+
1.2.3
+
1.2.3.4
~~
4
'
therefore, changing n into n+1, Sjw, = (*i2XjL.

3794 C/ass TV.When the general term of a series is a

rational fraction of the form

-,

where ux = ax+b,

and the degree of the numerator is not higher than x+m 2 ;

resolve the numerator into
A' + B'ux +- C'uxux+1 + ...+ D'uxux+1ux+i . . . ux+m_i}
by (3730). The fraction then separates into a series of frac
tions with constant numerators which can be summed by
(3787).
3795 If the factors ux ... ux+m are not consecutive, introduce
the missing ones in the denominator and numerator, and then
resolve the fraction as in the foregoing rule.

3796

Fx. : To sum the series r^j 4- jr^ + =; + to n terms.

1.4 2.5 3.0

SUMMATION OF SERIES.

557

The term is

1
=
0 + l)( + 2)
w(w + 3) n(ra + l)(n+2)( + 3)
r
w(w + l)+2n + 2
_
1
2
n( + l)(n + 2)(H + 3)
(n + 2)( + 3) ( + l)( + 2)(n + 3)

(n+l)(n + 2)(m+3)'
The sum of n terms is, therefore, by the rule (271),
\3

n + 3/

(n+2)(w + 3)/
3 \2.3 (n + l)(+2)(+8)
_ 11
3rt' + 12w + ll
18 3(n + l)(n + 2)(n + 3)'
If the form in (3787) is used, the total constant part 0 is determined finally
by making n = 0, which gives G = rrr.
18
3797

2 \2 3

Theorem.

f{E) ax$ (at) = a*f(aE) <f> (#),

f being an algebraic function.

Proof.Let a = e", then the left
= f(E)emx<j>(x)=f(e)e'*x </,(*) (3736) = e"V(eB"") f (a) (3475)
= o*/(aJ5) ^(a).
CZass F.If ^ (a;) be a rational integral function,
3798
So-* (,) = -L j + (,) - -2L A* (,) +

() - &c. | .

The upper limit is understood to be x 1, and a constant is to

loe added, (3781-2).
Proof 2ax^(a;) = A-la*p(a;) = CE?-l)-'a*0 (s) = ax (aE-iy1^ (x)
(3797) = a' { a (1 + A)-1 } "f () =

(l+ ^)

0).

Then expand the binomial.

2ax<f>(x) in successive derivatives of #(x).
3799

2**(*) =

[ l+4f(#)+ p| f

ProofBy (3757),
= $ (E) e"D; therefore (see last proof)
a*(aE
(x) (putting E = eD) = a* (aE-l)~l e0D$ (x)

558

CALCULUS OF FINITE DIFFERENCES.

3800 Ex.: To sum the series 2.1+4.8 + 8.27 + 16.64+ to n terms.

We require 2V + 2 2**" = 2V+2* (as-2Ax*+2,AV-2*As.s)
= 2V + 2I{'-2(3i, + 3x + l) + 4(6a! + 6)-8.6}
s= 2'{2*s-6z,+ 18a;-2G].
3801

If &~nux be known for all integral values of n, and if

vt be rational and integral,

%uxvx = t1ux.vx_1 %iux.Avx.i+^ux.^2vx.3&c.
Proof. Zuxvx = (EE -I)-' uxvx = (AiT+A')-1 uxvx (3746) and (3736).
Expand the binomial operator, observing (3738).
3802
Proof.
S*,a = (A'+AB')-t, as in (3801),
== A- vxux-n (A-"-l) e,A,+ 0,,,(A-"-,JIi) r^AS*,Ac,
producing the above by (3735) and (3782).
Observe that, in (3801) and (3802), two forms are obtainable in each case
by expanding the binomial operator from either end of the series.
3803

Ex. : To sum the series sin a + 2s sin 2a + 3' sin 3a + to x terms.

The sum is = x' sin ajs + Sx1 sin ax. Taking u, = sin ax and p, = a?, we
know A~" sin ax, by (3729) ; therefore (3801) gives
Sx'aiaax = (2 sin Ja)_l sin jaz | (a + >r) j (x l)1
-^sinia)-1 sin [ aaj-(a + T) ] (2x-3) + (2 sin a)-Jsin { aj:-f(a+x) } 2.

APPEOXIMATE SUMMATION.
3820

The most useful formula is the following

- r-u (' ,i ,o.M> i 1 du*- 1 rf3> +.

1
rf5*
~
\J * ^ 2 -1" 12 dx
720 rf^ ^30240 </.i5
1
Tu, ,
1
d*<_&c
1209600 rf.c7 ' 47900160 d.r>
Proof. 2m, = (eD- 1)~' it,. Expand by (1539) with D in the place
of x.

APPROXIMATE SUMMATION.

559

Ex. 1 : Tho value of Sx" at (2939) is given at once by the formula.

3821

Ex. 2 : To sum the series 1+ + ...+ approximately,

- +2- = - +C+loga)-f - -i-, + T^n-*0a>

as

a;

2a;

12a!*

120a;4

Put as = 10 to determine the constant ; thus

from which C = '577215, and the required sum is

3822

Ex.3:

l+^ + ^+^+Ao.,

x* ~

2x>

2x*

2x* "*" 2a;

2a* ^ 2as,'

1 a!8
2
2
4
12 12 20 12
The convergent part of this series, consisting of the first five terms, is an
approximation to tho sum of all the terms.
3823 A much nearor approximation is obtained iu this and analogous
cases by starting with the summation formula at a more advanced term.

= 2035
1728

_1 . , J_

3A_ A-B. +&c

2.5s

2.5*

2.5'

2.5"

2035
1
1 .
1
1
.
1728 50
250
2500 187500
"
The converging part now consists of a far greater number of terms than before,
and the convergence at first is much more rapid.
3824 Ex. 4: Tho series for logT(a! + l) at (2773) can bo obtained by
the above formula when x is an integer. For, in that case,
logT (a; + l) = logl + log2 + log3 ... +logx = loga + 2 logs:,
and (3820) gives the expansion in question, the constant being determined
by making x infinite.

3825

Formula (3820) may also be used to find

uxdx by the

process of summation, and thus answers the purpose of

Laplace's formula (3778).

CALCULUS OF FINITE DIFFERENCES.

560

2nux in a series of derivatives of ux.

3826

Lemma
n-l\ (*-l)- = (-l)"-1(</,+n- l)(d,+n -2)
...(dt+l) {e'-l}-1.

Proof.Put vH for n-1 ! (e'-l)-\ Then

, = (d,+n) = (d< + n)(J,+n 1)
S", may now be developed.
3827

Ex.To dcvelope S*u {Booh, p. 97)

2(e'-l)-= (dt + 2)(d(+l)je'-l} 1
= (^+3<it+2) Ji- --i +^ + J/ + 4c. J

with -42r = 0, and

4,,, = (-l)r^r+i + (2r + 2) ! = J- - | + | +(24 + 34,-1)
+ 2f { (r+2)(r+ 1) 4r+1+3 (r+1) Artl+2Ar) f.
Therefore, changing t into dz, we get
aV. = jjjW 1 ff ^ + J ^~ 1 .+ gQ -*c3828

2"ux in a series of derivatives of ux_a.

Let

af cosec" x = 1

+ C^b4 &c. , then

Sv.= />- Jl+C.(|)'+C.(f)+4c.j ,...

[.BooZe, p. 98.
3829
<A(0)-^(l) + ^(2)-&c. = I|l-|. + ^!-&c.(^(0).
By this formula, a series of the given type may often be trans
formed into one much more convergent.
PROOF.-The left =

, (0) = -

f (0) = | ^ (0),

the expansion of which is the series on the right.

3830

Ex.To sum 1 -i- + -g- -i- +&c. Summing the first six terms,

it becomes f| + i - -| + Ac.

Taking <p (0) = (0+ 7)"1, '

1
1 , .
HI,
1
2
,
2.3
,.)
7~ 8+fc--l(y + m + 4X8T9 + 8T8Jl0+*O-iThe sum after six terms converges rapidly by this formula, and more rapidly
than if the formnla had been applied to the series from its commencement.

PLANE COORDINATE GEOMETRY.

SYSTEMS OF COORDINATES.
CARTESIAN COORDINATES.

4001 In this system (Fig. 1)* the position of a point P in a

plane is determined by its distances from two fixed straight
lines OX, OY, called axes of coordinates. These distances
are measured parallel to the axes. They are the abscissa PM
or ON denoted by x, and the ordinate PN denoted by y. The
axes may be rectangular or oblique. The abscissa x is
reckoned positive or negative according to the position of P
to the right or left of the y axis, and the ordinate y is positive
or negative according as- P lies above or below the x axis
conformably to the rules (607, '8).
4002 These coordinates are called rectangular or oblique
according as the axes of reference are or are not at right
angles.
POLAR COORDINATES.

4003 The polar coordinates of P (Fig. 1) are r, the radius

vector, and 0, the inclination of r to OX, the initial line,
measured as in Plane Trigonometry (609).
4004 To change rectangular into polar coordinates, employ
the equations
,r = r cos 6,
y = r sin 0.
4005

To change polar into rectangular coordinates, employ

r = v^+7,

6 = tan-1 (A

* See the end of the volume.

4 0

PLANE COORDINATE GEOMETRY.

562

TRILINEAR COORDINATES.
4006 The trilinear coordinates of a point P (Fig. 2) are
o, |3, -y, its perpendicular distances from three fixed lines which
form the triangle of reference, ABC, hereafter called the trigon.
These coordinates are always connected by the relation
4007
4008

aa+tv8+cy=2,
or

a sin A +/J sin B +y sin = constant,

where a, b, c are the sides of the trigon, and 2 is twice its

area.
4009 If > V are the Cartesian coordinates of the point
a/3-y, the equations connecting them with the trilinear
coordinates are, by (4094),
a = x cos a 4-// sin aplt
B = cos B-\-y sin Bpi}
y x cos y-\-y sin y p3.
4010 Here a has two significations. On the left, it is the
length of the perpendicular from the point in question upon
the side AB of the trigon. On the right, it is the inclination
of that perpendiciilar to the a* axis of Cartesian coordinates.
Similarly (3 and y.
4011 The angles a, (3, y are connected with the angles
A, B, C by the equations
y B = irA,

ay = n~B,

aB ir-\-C,

only two of which are independent.

4012 Pu PuPi
the perpendiculars from the origin upon
the sides of the triangle ABC.

AREAL COORDINATES.
If A, B, C (Fig. 2) be the trigon as before, the areal co
ordinates a', /3', y' of the point P are
4013

a - -P"C
p~ ABC

o_ B _ PCA
P~J,~ ABC

,_y__PAB
y~p~ABC

SYSTEMS OF COORDINATES.

563

The equation connecting the coordinates is now

4014

a'+/8'+r'=l.

4015 To convert any homogeneous trilinear equation into

the corresponding areal equation.
4016

Substitute

aa = 2a',

b/8 = 2)8',

Cy = 2y'.

Also any relation between the coefficients I, m, n in the

equation of a right line in trilinears will be adapted to areals
by substituting la, mb, nc for I, m, n.
Similarly for a, b, c,
f, g, h, in the general equation of a conic (4656), substitute
aa2, btf, cc2, /be, gca, hab.
In either the trilinear or areal systems, a point is deter
mined if the ratios only of the coordinates are known.
Thus, if a : /3 : y = P : Q : R, then, with trilinear co
ordinates,
P2
P
4017 a = aP+bQ+cR ; and' With area1' = P+Q+R-

TANGENTIAL COORDINATES.
4019 In this system the position of a straight line is deter
mined by coordinates, and the position of a point by an
equation. If la+m^+ny = 0 be the trilinear equation of a
straight line EDP (Fig. 3); then, making a, /3, y constant,
and I, m, n variable, the equation becomes the tangential
equation of the point O (a, (3, y) ; whilst /, m, n are the co
ordinates of some right line passing through that point.
Let X, ft, v (Fig. 3) be the perpendiculars from A, B, C
upon EDP, and let pu pt, p$ he the perpendiculars from A, B,
C upon the opposite sides of the trigon ; then, by (4624), we
have
4020

iiX = lpL,

Up = mpt,

Rv = np3,

-where R = y/(P+m> + u2 2mcos A 2nl cos B 2lm cos G).

Hence the equation of the point O becomes
4021
X^+/AA+vX=0
Pi
Pi
Pi

or

A^+^^+v^ = 0,
Pi
Pi
Pi

where Pi = OA, 0, = /.BOG, Ac, and 2&BOC = PiP;ism9.

504

PLANE COORDINATE GEOMETRY.

Formula (4021) shows that, when the perpendiculars X, /*, v

are taken for the coordinates of the line, the coefficients be
come the areal coordinates of the point referred to the same
trigon.
4023 Any homogeneous equation in I, m, n as tangential
coordinates is expressed in terms of X, p, v by substituting for
I, m, to. , , respectively.
Pi Pi Pi

By (4020).

4024 An equation in
v of a degree higher than the first
represents a curve such that X, n, v are always the perpen
diculars upon the tangent. The curve must therefore be the
envelope of the line (X, fi, v).

TWO-POINT INTERCEPT COORDINATES.

Let X = AD, /x = BE (Fig. 4) be variable distances from
two fixed points A, B measured along two fixed parallel lines,
then
4025

X+6/i+c = 0

is the equation of a fixed point 0 through which the line DE

always passes. This may easily be proved directly, but we
shall show that it is a particular case of the system of threepoint tangential coordinates.
Let one of the vertices (C) of the trigon in that system be at infinity
(Fig. 3). Then equation (4022) becomes
X sin 0,1 + u sin 0,1 +sin
, COE
rim* sin
0,
a = 0.
n
Pi
Pi
For : p, = sin COE always. Divide by sin COE ; then X -f- sin COE = AD,
&c, and the equation becomes
1AD+ S?*AE+ sin 63 = 0.
Pi
Pj
The only variables are AD and AE. Calling these \ and /j, the equation
may be written
aX + bft + c = 0,
the form taken by aX+6/i + cV = 0 when v = ao and c vanishes.
ONE-POINT INTERCEPT COORDINATES.
4026 Let a, b be the Cartesian coordinates of the point O
(Fig. 5) ; and let the reciprocals of the intercepts on the axes

SYSTEMS OF COORDINATES.

565

of any line DOE passing through 0 be = -j^, v =

Then, by (4053),
4027

"t+bv = l

is the equation of the point 0, the variables being $, >j.

This is a case of the system of three-point tangential coordinates in which
two of the vertices (B, 0) of the trigon are at infinity. Equation (4022)
now becomes Xsmfll + sin BOD sin 0, + sin COE sin 0, = 0,
ft
sin 0, , sin 0, , sin 0. n
or
1 + TFT + -TTT = "l
p,
AD
AE
which is of the form a+br) = 1.
TANGENTIAL RECTANGULAR COORDINATES.
4028 This name has been given to the system last described
when the two fixed lines are at right angles (Fig. 6).
The coordinates , v, which are defined as the reciprocals
of the intercepts of the line they determine, have now also the
following values.
4029 Let x, y be the rectangular coordinates of the pole of
the line in question with respect to a circle whose centre is
the origin and whose radius is k ; then
=|[

and

*?=fr>

since x.OM = y .ON = h*; for M, N are the poles of y = 0,

4030 The equation of a point P on NM whose reotangular

coordinates are OB = a, OS = b, is
ag+br, = l, by (4053),
this equation being satisfied by the coordinates of all lines
passing through that point.
4031 In all these systems an equation of a higher degree
in , ij represents a curve the coordinates of whose tangents
satisfy the equation.

ANALYTICAL

CONICS

IN
CARTESIAN

LENGTHS

COORDINATES.

AND

AREAS.

Coordinates of the point dividing in the ratio n : ri the

right line which joins the two points xy, x'y' .
4032

f = 2^*.

v = "-p>.

Proof. (Fig. 7.) = x +AC = + -, (x'x).

= *<

Similarly for 9.

4033

Ifn = n',

i>=*.

4034

Length of the line joining the points xy, x'y'

= /{?x')i+{yy')i-

The same with oblique axes

4035

(y-yj+ 2 (x-x) (y-y) cos

Proof.By (Fig. 7), Euc. I. 47, and (702).

Area A of a triangle in terms of the coordinates of its
angular points x1yi, 2y2) x^js.
4036

A = i {.tj^-.r^! + r2y, a'3t/2 + .r3yi ^i#a} .

Proof.(Fig. 8.) By considering the three trapezoids formed by y,, t/ y,

and the sides of the triangle, we have
-1 = I (y, +ys) (*,-*,) + |(ys+i/s)(a!J-^)-i (yj+yi)^-^)-

LENGTHS AND AREAS.

567

Area of the triangle contained by the k. axis and the lines

y = m1x+c1,

y = m^c+Ci,

(4052)

Proof.(Fig. 9.) Area = (c^ cs)^, and^ is found from

jrm^pm1 = OjCj. The sign of the area is not regarded.
Cob.Area of the triangle contained by the lines

4038

V=
+ Ci,
(1 = m& +c2,
y = m& -f- c,
A
(Cl~Cz)2 |- (c2~cs)2 _|_ (c3-c,)a 1
2 (. tox m2

4039

m3 mj ) '

_ { Ci (%-ma) + c2 (ffl8 mO + c3 (wi m2) \ '

2 (m, i2) (m2 m3) (m3 mx)

4040
Af\A-\
mi

m2 ms

2^ (4A-iA) +

Square of Determinant {Ajfifi^\

~ 2 (AlBt-A1Bt)(AtBt-AiBJ{A3Bl-A1Ba)'

Pkoof.(Fig. 10.)

ABG = AEF+CDE-BED.

Employ (4037).

Area of Polygon of n sides.

First in terms of the coordinates of the angular points

4042

2A = fay,*4h)+(*tirr-<ffy)++(*yi=

(^2

(ys

(yi

0-

Secondly, when the equations to the sides are given, as in

(4037).
4043
2A
(ci cz)a _|_ (ciczf _j_
_|_ (p~Pi)2
! i2
4044

>/i2 m3

"

to,

Also three values similar to (4039, '40, '41).

Proof.By (4367), adding the component triangles.

4047 Each expression for the area of a triangle or polygon
will be adapted to oblique axes by multiplying by sin &>.

CARTESIAN ANALYTICAL CONICS.

568

TRANSFORMATION OF COORDINATES.
4048
Put

To transform the origin to the point hk.

x = x'+h,

y = y'-\-k.

To transform to rectangular axes inclined at an angle 0

to the original axes.
4049 Put
x = x' cosdy' siu9,

y = y costf-f^' 'shifl.

Pkoof.Consider ' as cos^ and y as sin^.

y = sin (0 + 0) (627, '9).

(Fig. 11.)

Then x = cos (<p + 0) and

Generally (Fig. 12), let u> be the angle between the original
axes ; and let the new axes of x and y make angles o and /3
respectively with the old axis of x.
4050
and

Put j: sin <u = x sin (a> o) +y sin (<a )

y sin <u =
sin o-fy siu /?.

Proof. (Fig. 12.) The coordinates of P referred to the old axes being
OC = x, PC = y, and referred to the new axes, DM =
PM= y, we have,
by projecting OCP and OMP at light angles first to CP and then to OG,
CD = MF-ME,
PN = ML + PK,
which are equivalent to the above equations.
To change Rectangular coordinates into Polar, hk being
the pole 0, a the inclination of the initial line to the x axis
(Fig. 13), and xy the point P.
4051

Put x = h+r cos (0+a),

y = k+r sin (0+a).

THE RIGHT LINE.

EQUATIONS OF THE RIGHT LINE.
4052

y = mx+c

(1),

4053

- + |-=1
a
o

(2),
'

THE BIGHT LINE.

569

4054

oc cos a-\-y sin a = p

(3).

4055

Ax+By+C = Q

(4).

Proof. (Fig. 14.) Let AB be the line. Take any point P upon it,
coordinates ON = x, PN = y. Then, in (l),- m = tan 0, where 8 = ]1AX,
the inclination to the X axis; therefore mx = 00, and c is the intercept
OB. In (2), a, b are the intercepts OA, OB. In (3), p = OS, the per
pendicular from 0 upon the line ; a = . AOS.
p = OR + LP = x cosa + y sin a.
(4) is the general equation.
4056

4060

4062

m tan 5 = = = cot o.
B
a

sin0 =

A
,
k/A^B1

cos e = -

11
s/A'+B1

p = c sin a =
Vl+ni1

k/A'+B2

Oblique Axes.
Equations (4052, '53, '55) hold for oblique axes, but
(4054) must be written
4065
x cos a+?/ cos $p.
(Fig. 14)
4066

tanfl=

wsiu<" = /Shl" ,
1 + mcoscj
A cos <oB
oi being tbo angle between the axes.
Proof.From m = sin 0 H-sin (w0).

4068

p =

c sin w

C siu a

v/l+2w cos u+m"

\/ 42+Ii3-2J /i cos w
Proof.From p = c sin (<> 0) and (4 iG6).
The equations of two lines being given in the forms (4052)
or (4055), the angle, <j>, between them is given by
1AHA
4070

,
,
mni
tan = ^
,

Proof.(Fig. 15)

AB' A'B
or

tan^ = tan (t)-O').

4 I)

.
Expand by (632).

CARTESIAN ANALYTICAL CONICS.

570

To oblique axes :
4072
tan 6 = z , V ,
;
l+(m+m ) cos + mm
Proof. As in the last, employing (4066).
Equation of a line passing through x'y:
4073

yy = m(x -x),

4074

or

y mx y mx ,

4075

or

Ax+By= Ax'+By'.

(Fig- 8)

Proof.From Figure (13), m being = tan 0.

Condition of parallelism of two lines :
4076

m = m,

or

AB"= A'B.

Hence the equations differ by a constant.

Condition of perpendicularity :
4078

mm - -I

or

AA'+BB'= 0.

(4070)

The same to oblique axes :

4080

1 + (m+m) cos <a+mm = 0.

4081

or

4082

or

(4072)

AA '+ BB' = (AB'+A'B) cos o>,

m =

.
m+cos tu

A line passing through the points sb^, a'2y2 :

4083

^Z#i =

Proof. (Fig. 16.)

= m.

By the similar right-angled triangles PCA, ADB.

4084

Or

u = mx+Xl!,,~*t91,
xx Xi

4085

or

(x-xJiyy,) = (x-x2)(y-yl).

Proof.This equation represents a straight line because it is of the first

degree; and the coordinates of each of the given points satisfy the equation.

TEE BIGET LINE.

571

A line passing through x'y and perpendicular to a given

line (m) :
4086
4087

y-y'= - (*-*')
or

(4073, -78)

BxAy Bx-

The two lines passing through x'y and making an angle

/3 (= tan_1m2) with a given line (m^ :
4088

2=K = r.1"2

and

iW'+m'.

(4073,70)

A line passing through hk and dividing the Une which

joins xlyl and se^y, in the ratio nx : n2 :
4089
/i

= h(.y.-fc)+*(yi-*)
ij (a?a /t)+n2 (iTj A)

(4073, -32)

Coordinates of the point of intersection of two lines :

4090

4092

= B^~Bfu
mt tn1
A^B2 A^BX
y = Cim2~c^ni = - At1^~A/C'.
m2
A^-iA.Bi

(4H6)

Length of the perpendicular from a point x'y' upon a

given line
4094
= 00 cos a+y' sin ap.
Proof.Let AB (Fig. 14) be the line, and Q the point x'y'. Then, by
(4054), x cos a + y sin u = OT, the perpendicular from 0 upon a parallel line
through Q, and p = OS.
Otherwise, the same perpendicular
4095

= A*'+By' + C_.
VA'+B*

The same with oblique axes

40Qfi
WO

{Ax'+By' + C) sino>
V/(^"+ii2-2^JBcosa.)'

obtained in a similar way from (4065-69).

(4060, '61, '94)

572

CARTESIAN ANALYTICAL CONICS.

Condition of three lines intersecting in one point :

4097

c1mic1mi+Comscimi+e3in1c1tn3 = 0.

The area in 1009 must vanish.

4098 Otherwise.If certain values of the constants /, m, n
make the expression
/ ( Axx+ B1y+Cl) + m (A^v+Bty+ Ct) + n
vanish identically, the three lines indicated intersect in one
point.
Proof.By (4099), for then values of x and y which make (1) and (2)
vanish also make (3) vanish.
A line passing through the point of intersection of the
lines Ax + By + C = 0 and Ax+By+C = 0 is
4099
4100

Ax+By+C = k {A'x+B'y+C),
or l(Ax+By+C)-m{A\v+B'y+C) = 0,

1c, I, and m being any constants.

4101 Rule.If the equation of a right line contains a third
variable k in the first degree, the line always passes through a
fixed point.
Proof.For the values of x and y, which satisfy simultaneously the given
equations, also satisfy (4099), whatever k may be. See (4004).
4102 If in the equation of a line Ax+By + C = 0, the co
efficients A, B, G involve x, y , the coordinates of a point
which moves along a fixed right line, then the first fine passes
through some fixed point.
Proof.By means of the equation of the fixed line, y may be eliminated,
and x' then remains a third variable in the first degree (4101).
4103 To find the point in which the line Ax + By + C inter
sects the line joining the points xy, xy; substitute
Ax+By+C forn, and A*'+By' + C for n in (4032).
Proof.By (4095), since the segments intercepted are in the ratio of the
perpendiculars from xy, x'y upon the line Ax + By + C.

THE BIGHT LINE.

573

Equations of the line -with I, m for direction-ratios, hh a

fixed point on the line, and r the distance of the variable
point xy from hh.
Atr\A

xh
I

4104
4105
4106

yk
,
in

r = j = ^
where
or

? = sin( 0),
sm&>
I = cos 0,

sin 0
m = - ,
sm to
m = sin 0.

(Oblique)
(Rectangular)

Polar Equation of a Straight Line.

4107

T cos (6 a) = p.

(Fig. 17.) Herep is the perpendicular to the line from the

pole 0, and a is the inclination of p to the initial line OA.
When the line passes through the pole, the equation is
4108

0 = constant.

A line passing through the two points r-fiu r20.2.

4109

vrx sin (0-00+nr, sin {6X- $t)+rtr sin (02-0) = 0.

PROOF.-(Fig. 18.) &POA + AOB-POB = 0.

Then by (707).

EQUATIONS OF TWO OR MORE RIGHT LINES.

The homogeneous equation of the nth degree,

4110

xT+p^y +p&*-y + . . . +pnyn = 0,

represents n right lines, real or imaginary, passing through

the origin.
For it is resolvable into n factors of the form (x ay), by (405).
For the case of two right lines represented by the general equation of the
second degree, see (4469).

Equation of two right lines through the origin :

4111

aa?+2hay+lnf = 0.

574

CARTESIAN ANALYTICAL CONICS.

If <j> be the angle between the lines,

4112

tan*=^(*'-fl6> or 2?SW-ab)
a+b
a+b 2h cos a
according as the axes are rectangular or oblique.
Proof.Assume (y mxx)(y m^e) = 0, and apply (4088).

Equation of the bisectors of the angle $ :

4113
ha?-(a-b)ay-hy* = 0.
Proof.Let y = fix be a bisector (ft = tanv/<) ; then, since 2\p = 9, + fl,,
r~i = ,
L=
V b7 (4111) ; and fx =
1 ^
1 mtm, a b
x
The roots of this equation are always real.

GENERAL METHODS.
APPLICABLE TO ALL EQUATIONS OP PLANE CURVES.

4114 Let F (*, y) = 0

(i.) and /(*, y) = 0
be the equations of two curves of any degree.

(ii.)

4115 To find the intercepts on the x and y axes.

Put y = 0 in (i.), then x becomes the intercept on the x axis.
Similarly, put x = 0 /or Me intercept on the y axis.
4116 To find the points of intersection of (i.) and (ii.).
Solve as simultaneous equations. Each pair of values ofx
and y so obtained gives a point of intersection. Imaginary
values give an imaginary point.
4117 To determine equation (i.) so that the line may pass
through certain fixed points, x^, x^y2, &c.
Substitute x,y!, x2y2, Sfc. for xy successively, so forming
as many equations as there are points. From these equations
the constants in (i.) must be determined in terms of xuy1} Xo,y2,
Sfc.
4118 The number of arbitrary points cannot exceed the
number of constants in the equation.

GENERAL METHODS.

575

4119 Condition that (i.) and (ii.) may touch.

At a point of contact hoo or more points of intersection
must coincide, and therefore the equation for x or y, obtained as
in (4116), must have two or more equal roots for each point of
contact. The contact is said to be of the second order when
there are three coincident points; of the third order when
there are four, and so on.
4120 To find the equation of the tangent at a point x'y' on
the curve f(x, y) = 0.
Form the equation to the secant through two adjacent points
xiju

(4083), and determine the limiting value of ^^1

u'hen the points coincide by means of the equations f (xl5 yx) = 0,

f (x2 y) = o.
4121

Otherwise

m = ^,

by (5101).

4122

For the equation of the normal, change m of the

tangent into (4086).

4123 To express the equation of the tangent, or normal, in

terms of m and the constants of the curve.
From the equation of the tangent or normal, the equation to
the curve, and the equationfurnished by the value of m, eliminate
x' y', the coordinates of the point of contact of the tangent.

THEORY OF POLES AND POLARS.

4124 Let F(x', y , x, y) = 0 represent the equation to the
tangent of a curve at the point x'y'.
Then F(x, y, x, y') = 0, the equation obtained by inter
changing the constants x',y' with the variables x, y, represents
the polar of any fixed point x'y' not on the curve.
Let xlyl,
(Fig. 19) be points A, B on the curve, and let the tangents
at those points intersect in x'y'. Consider the equations
F 0 y x, y) = 0 ... (1), F(x y x, y) = 0 ... (2), F(x, y, x, y') = 0 ... (3).
Here (1), (2) are the tangents, and (3) is some straight line or curve
according to the dimensions of x and y. Also (3) passes through the points

576

CARTESIAN ANALYTICAL C0NIC8.

of contact as,y .7y/2, and may therefore be called the curve of contact-; or, if a
right line, tho chord of contact of tangents drawn from x', y', i.e., the polar.

4125 Hence the coordinates of the points of contact of

tangents from an external point xy' will be determined by
solving (3) and the equation of the curve simultaneously.
4126 Again, let x'y (Figs. 20 and 21) be any point P not on the curve.
Then, from the equations
F(x, y, x, y) = 0 ... (4), F (x, y, x y,)=0... (5), F (x, y, xv yt) = 0 ... (6),
we see that (4) is some straight line ; that, if .r3y3 and xiyi are any two points
upon it, (5) and (6) are the curves of contact of tangents from those points ;
and that these curves of contact pass through the point x'y'.
4127 If the points xty}, xiyi are taken at A, B, where (4) intersects the
curve, (5) and (6) then become curves touching the given curve at A and B,
and passing through x'y. We may call these lines the curve tangents
from x'y'.
4128 Lastly, let xy in (ti) be a point within the given curve (Fig. 22),
then the equations
F(x,y,x',y")=0...(7), F(xt,yt, x,y) = 0... (8), F(xt,yt,x,y-) = 0 ... (9)
show that (7) is the locus of a point, the curve tangents from which have
their chord of contact always passing through a fixed point. When x'y' is
without the curve, as in Fig. (19), the same definition applies to every part
of the locus (3) from which tangents can be drawn.

4129 H the given curve be of a degree higher than the

second, the line of contact of the tangents from a point is a
curve, and the line of contact of the curve tangents from a
point is a straight line (Figs. 19 and 20). A similar converse
relation is exhibited in Figures (21) and (22).
If the curve be of the second degree, equations (3) and
(4) become identical. The line of contact or the polar is
always in this case a straight line, and so is the locus (7).
Figures (19) and (20) now become identical, as also (21)
and (22).
4130 The polar of the point of intersection of two right
lines with regard to a conic passes through their poles.
Proof.As in (4124). Let (1) and (2) be the two lines, (#,), (#,)
their poles, and x'y' their point of intersection.

GENERAL METHODS.

577

4131 To find the ratio in which the line joining two given
points xy, x'y' is cut by the curve /(x, y) = 0.
Substitute for x and y, the sujiposed coordinates of the
point of intersection, the values
p, S^!&
by (4032)
n-\-n
n+n
and determine the ratio n : n' from the resulting equation.
The real roots of this equation correspond to the real points of
intersection.
4132 To form the equation of all the tangents that can be
drawn to the curve from a point x'y'.
Express the condition for equal roots of the equation in
(4131), and consider xy a variable point.
4133 To form the equation of the lines drawn from x'y' to
all the points of intersection of two curves.
Substitute nx'-\-n'x, ny' + n'y for x and y in both curves,
and eliminate the ratio n : n'.
Proof.Take any other point xy on the line through x'y and a point of
intersection. The ratio n : ri (4131) is the same for each curve, and there
fore may be eliminated.

4134 To find the length, r = AP or AP' (Fig. 23), of the

segment intercepted between the point A or x'y' and the curve
f(x, y) = 0 on a straight line drawn from A at an inclination
0 to the X axis. That is, to form the polar equation with
x'y' for the pole and the initial line parallel to the x axis.
Substitute for x and y, the assumed coordinates of the point
of intersection, the values x = ON or ON, y = PN or P'N',
that is,
x = #'+rcos0,
y = y'+rshi#,
and determine r from the resulting equation. That is, put
a = 0 in (4051).
The real values of r are the distances of the points of inter
section from x'y'.
4135 When an equation has been obtained for determining
x the length of a line, important results may frequently be
arrived at by applying theorem (406) respecting the sum and
product of the roots.
4 E

CARTESIAN ANALYTICAL CONICS.

578

THE CIRCLE.
Equation with the centre for origin.
4136
a*+y2=rl.

(Kg- 24.)

Equations of the tangent at the point P or xij .

4137

y-y'=-^{x-x).
y
xx'+yy' = r*.

4138

(4120)

Also, by (4124), the polar of x'y', any point not on the curve.
f
4139
y = nuc+r y/l + m2 ;
m=- A(4123)
4140

x cos a+y sin a = r,

a being the inclination to the a; axis of the radius to the point

x'y'.
Equation of the circle with a, b for the coordinates of the
centre Q.
(Pig. 24.)
4141
(li-a)2+G/-&)2=ra.
Tangent at x'y', or Polar,
4142
(*-)W-")+(!/-b)(y'-b) = rii
4143

or

(4138)

(j? a) cos a+ (yb) sin a = r,

a being the inclination of the radius to the point x'y' .

General equation of the circle :
4144
4145

a?+y*+2gx+2fy+c = 0.
Centre (~g, -/).
Eadius v/(^+/s-c).

Proof. By equating coefficients with (4141).

Equation of the circle with oblique axes :
4146
4147

(Fig. 25.)

(x-ay+(y-b)2+2 (x-a)(y-b) cos o> = r\

or a*-\-2xy coscu+y 2 (a+6 cos <u) a?
2 (6+a cos u) y
+a2+2aft coso+62 = r*.

(702)

TEE OIBGLE.

579

General Equation.
^+2.^ cos a+yi+2gx+2fy+c = 0.

4148

The coordinates of the centre are

4149

4150

a =f"-Kt
sm* ot

b = Zs"-f
sin'- co

Radius = v/{g2-2/gcosa>+,f!-csin^}
sin to

Proof.By equating coefficients with (4147).

Polar Equation.
4151
4152

r2+P2rlcos(0-a) = c\
(Fig. 26)
or r2 2/ cos or cos 0 2/ sinar sin 0+f-ca = 0.

ProofBy (702), the coordinates of P being r and 0.

General form of the polar equation :
4153

r*+2gr cos 0+2/r sin 0-fc = 0.

4154

tan a =

l=</]?+F.

Proof.By equating coefficients with (4152).

4156 Equation of the circle passing through the three
points Xj.i/i, xflz, xsys.
x2y.2 1
x2ytl

-(*5+a a? ?/ 1
?/ 1

#3 ?/s 1
x y \

a--> y2 1

Proof.Eliminate g,f, and c from (4144) by (4117).

Equation of the chord joining x^y^, xty2, two points on the
circle x* + y2 = r2 :
4157

(ft+ft) =

(083, 4186)

4158 or a?cos(01+0a)+ysin(01+08) = rcos|(^ 03),

where

r cos 0j = xlt

r sin 0! = yu

&c.

580

CABTESIAN ANALYTICAL CONIC8.

4:159 Note.The coordinates x, y of a point on the circle xi + yt = i* may

often be expressed advantageously in this way in terms of 0, a single variable.
4160

Let

S = (<v-ay+(y-iy-r* = 0

be any circle (Fig. 27). Then, if xy be a point P outside the

circle, S becomes the square of the tangent from P. If xy be
a point P' -within the circle, S becomes minus the square of
the ordinate drawn through P' at right angles to the radius
through P1.

CO AXAL CIRCLES.
(See also 984 and 1021.)
4161

If

S = x*+tf+2gx +2fy +c = 0,
S'= a*+y*+2g'x+2fy+c' = 0

be two circles, the equation to the radical axis is

S-S'= 0.
If x = 0 be taken for the radical axis, the equation to any
circle (radius r) of the system of coaxal circles (1021) is
4162
xt+y*2kx& = 0 and k2-ri = ^,
+ in Figure (1), in Figure (2).
and k = 10 a variable.

Here S = IB a constant,

4164 The polar of xy' for any circle of the system passes
through the intersection of
aw' -\-yy 8s = 0 and x + x = 0.
Proof.Its equation is xz' + yy'-k (x + x) c5 = 0 (4121).
(4099).
4165 When i = 8, then k = IB = IU.
Poncelet's limiting points.

Then by

D and D' are

4166 The polar of D with respect to any of the circles

passes through U, and vice versa, by (4164).
4167 Tangents from any point on the radical axis to all
circles of the system are equal (4160, '61).

THE OIBOLE.

581

4168 The radical axes of three circles, Su S2, Ss, meet in a

point called their radical centre.
4169 The reciprocals with respect to the origin D or U of
the system of co-axal circles are all confocal conies (4558).
The equation of the circle, centre Q, cutting the system of
circles orthogonally is, putting IQ = h,
4170

2% 8* = 0.

(1230,1236)

This circle passes through D and D'.

The common tangents to the two circles
{x-a)l+{y-b)i = ri and (-')+

= r'\

(See also 1037.)

The equation for a in (4143) is
4171

(a a') cos a+ (6-6') sina+r^r' = 0.

Pboof.Assume (4143) in a, 6, r, a, and also in a', b\ r, a as coinciding

lines. Then tan a = tan a'; therefore a = a or ir + a. Take the difference
of the two equations.
The chords of contact are
4172

(a-o')(*-a)+(6-6)(y-6)+r(rTr') = 0,

4173

(a-a')(x-a')+(jb-b')(y-b')+r(r^r') = 0,

with for exterior tangents, + for transverse.

Pboof.For these are straight lines, and they pass through the points of
contact of each pair of tangents respectively, by (4171).
The centres of similitude 0, Q are the intersections of the
external and transverse tangents respectively.
4174

Coordinates of 0,

ar~aft
rr

b'r-br'
rr

4175

Coordinates of Q,

aV+T ,
r-\-r

hr+hJ'.
r+r

Proof. By equating coefficients in (4172) and (4142), the polar of O

or Q.
4176

The six centres of similitude of three circles lie on

582

CARTESIAN ANALYTICAL CONICS.

four straight lines called axes of similitude. See the figure

of (1046).
Proof.The coordinates of the three centres of the forms (4174, 75)
will in each case satisfy equation (4083).
4177 The equation of the external axis of similitude is, in
determinant notation (554),
(IrA)
(lV*s)y+0'i&iai) = 0.
Proof.By forming the equation of the right line passing through two
of the centres of similitude whose coordinates are as in (4174).
4178 The remaining three axes are found by changing in
turn the signs of ru rt ; r2, r, ; r r4.
4179 If one of the circles touches the other two, one axis of
similitude passes through the points of contact.
4180 The angle 0, at which the circle F(x, y) = 0, radius
r (Fig. 29), intersects the circle whose centre is hk, and
radius B is given by the equation
#2-2ilrcos0= F(h, k).
Proof: 6 = OQP,
(702) and (4160).
4181

E'-2Er cos 0 + r2 = PT + r3 - F(h, k)+r>,

Cor. 1.If the circles are given by the equations

a? + y* + 2r/x + 2f'y + c'= 0,

^ + f + 2gx+2fy + c = 0,

the equation for cos 0 becomes, since h = g, h= f,

2Rr cos 6 = 2gg'+ 2jf'-c-c.

(4145)

4182 Coif. 2. The condition that the two circles may cut
orthogonally is
2gg'+2j5r-c-c' = 0.
4183 Cob. 3.By solving three such equations, we can
find the circle cutting three given circles orthogonally (4186).
4184 Cor. 4. The condition that four
circles may have a common orthogonal
circle is the determinant equation

c
Cl
C2
C3

g f
gi fl
gi fi
"3 ft

1
1
1
1

= 0.

THE PARABOLA.

583

4185 Cor. 5.If the circle x*+y* + 2Gx + 2Fy + C = 0 cuts

three other circles at the same angle 0, we have, by (4081),
three equations to determine 0, F, G. The resulting deter
minant equation may be written
xi-\-yi x y 1
0 x y 1

gi

/. 1
ri
+2i* cosfl
/. 1
rt
/. 1
r

gi
g*
gi

f 1
/2 1
fs 1

= 0.

4186 The first determinant, put = 0, is the orthogonal

circle (4183), and the second, expanded, is the axis of
similitude.
4187 The locus of the centre of a circle cutting three given
circles at equal angles is a perpendicular from their radical
centre on any of the four axes of similitude.
Proof.By eliminating R and cos a between three equations, like (4180).
4188 Each of these four perpendiculars contains the centres
of two circles touching the three given circles.
Proof.Consider a = 0 or 180, in (4180).
To draw the eight circles which touch three given circles,
see (946) and (1049).
4189 The equation of the fourth degree of two of the
touching circles is
23 y/Si 31 ^/S2 12^S3 = 0,
where 23 signifies the length of the common tangent of the
second and third circles, &c.
Proof.By first showing that, if four circles are all touched by another
circle, the relation
4190
12.34X4. 23 31.24 = 0
will subsist, and then supposing the fourth circle to reduce to a point.

THE

PARABOLA.

4200 Def.'A conic is the locus of a point which moves in

one plane so that its distance from a fixed point S, the focus,

584

CARTESIAN ANALYTICAL CONICS.

is in a constant ratio (e) to its distance from a fixed right line

XM (the directrix).
When e = unity, the curve is a parabola. (See also p. 248,
et seq.)

Equation of the Parabola with origin of coordinates at the

vertex A.
4201
Here

y2 = 4a*.
a = AS, x = AN,
y = PN.

Proof.Geometrically, at (1229).
Analytically, from
PS1 = y* + (z-ay = PIP = (*+o)'.
The equations with the jf
origin at S and X respectively
are

4202 y2 = 4a (#+<*),
y = 4a(*-a). (4048)

Equations of the tangent at x'y' :

4204
4205
4206

y-y'=^(x-x').
y
yy = 2a (*+#')
y = mx+,
m

m = ^.
y

(4120)

(4123)

4207 (4204) is also the polar of any point x'y', by (4124).

Its intercepts are x and \y .

Equations of the normal at x'y :

4208

y-y =

j-(**)

4209

y'*+2ay- (yV+2ayO = 0.

42 10

y rmv 2am am3.

(4122)

(4123)

THE PARABOLA.

Equation of the parabola

with a diameter and tangent for
axes of coordinates.
4211
.Vs = 4*'*,
where
4212

585

Fft / /r
N/ /
i * /

a' = a cosec* 0 = SP;

x = PV;
y = QV.

Proof.Geometrically, at (1239). Otherwise, let VQ = VQ' be equal

roots of opposite signs of the quadratic (4221), V being the point x'y',
therefore
y* or r* = (y^ iax) cosec' 0 = 4a cosec2 0.x,
since y* = 4a X abscissa of P.
4213 Equations (4204-10) hold good for these axes, with a
written for a in each.
For the polar equation of the parabola, see (4336).
4214 Quadratic for nx : n2, the ratio of the segments into
which the line joining two given points a^/1} xty2 is divided by
the parabola y* 4ax = 0,
"i (yj 4a.r2)+2n1w2 {y^ 2a fa+x,)} +* Q/2

= (4131)

Equation of a pair of tangents from any point x'y' :

4215 (y'2-4ax')(y*-4ax) = {#y'-2a (,r+,r')}s = 0.
The condition for equal roots in (4214).
Quadratics for the coordinates of the points of contact of
tangents from x'y' :
4216

ax* -(y'2-2aa0 x-ax'2 = 0.

4217
y-%y'+W = 0.
Proof.Solve simultaneously the equations of the curve and the polar
(4205) and (4125).
Coordinates of the point of intersection of tangents at xxy
and xiy2 :
4218

x=m,

y = y^gh.
4 p

586

CARTESIAN ANALYTICAL C0NIC8.

Quadratic for m of the tangent from x'y' :

4220

mWmy'+a = 0.

(4206)

4221 General polar equation of the parabola, or quadratic

for r, the segment intercepted between a point, x'y, and the
curve on a line drawn from that point at an inclination 0 to
the x axis (4134),
r2 sin20+2r (y sin 0-2a cos0)+y'*4ax' = 0.

Quadratics for the coordinates of the points of intersection

of the line Ax + By + C and the parabola y* = 4ax :
(4116)
4222

AW-2 (2Br-a-AC) x+C = 0.

4223

Aif+ABay+ACa = 0.

Length of intercepted chord,

4224

4!^{{Biai-ACa){At+Bi)}^Ai.

(4034)

Equation of the secant through x1i/1, a^2, two points on

the parabola :
4225

y (yi+y*) = y^+^ax,

4226

or

4227

The subtangent

NT = 2x.

4228

The subnormal

NG = 2a.

(4083)

y (ttij+wij) = 2a-\-2in1mix.

Fig. of (4201)

Proof.Put y = 0 in (4205) and (4208).

4229
4230

The tangent PT2 = 4r (a+x).

The normal PGi = 4a (a+x).

The perpendicular y from the focus upon the tangent at xy :

4231
p = s/a (x+a) = <s/aaT.
(4212), (4095)

TEE ELLIPSE AND HYPERBOLA.

587

The part of the normal intercepted by the curve is equal to

4233
4234

= snr0cos0
. tf

(4221), (4135)

The minimum normal = 6av/3 and m = \/2.

Length of a chord through the focus

4235

= 4^ = 4a'.
snr 0

(4212)

Coordinates of its extremities, with the focus for origin :

AQotj

X~

2a cos 6
cos

2a sin 6
V~

Coordinates of its centre :

4239

*=2^^,

y = 2a cote.

THE ELLIPSE AND HYPERBOLA.

'

(See also p. 233, et seq.)

4250 Referring to the definition (4200) ; when e is less than

unity, the conic is an ellipse ; when greater than unity, an
hyperbola.
Equation of the ellipse with the origin of coordinates at X
and SX = p.
4251

yM-(.r-p)2 = eV.

Proof.By the definition in (4200).

Abscissae of vertices :
4252

XA = -?-,
1+e

(Supply A' in the following figure.)

XA =

P
1 e

(4ii5)

588

CARTESIAN ANALYTICAL C0NIC8.

V 6"

4254

C G ffS JA X T

Focal distances of vertices :

SA = W
1+e

4256

SA' ==

(4251)
1 e

SL = l = ep = a (1-e2) = ^.
62 = aa (1-e2) ;

(4251)

e2 = a2- ft2

4260

4262

4264
(4252)

4266
4268

If 6 = a tan a, then e = sec a in the hyperbola.

Equation with the origin at A

4269

tf=K(2a*-*>)

Ell.

4270

f = -(2a*+*)

Hyp.

Proof.By (4200),

+(3-8-4)' = J(x+AX)\ &c.

Equations with the origin at the centre C :

4271

f = -^-(2-*2) = (l-e2)(a2-^).

(4269)

THE ELLIPSE AND HYPERBOLA.

4275

Def.QCN is the eccentric angle,

589

of the point P.

x and y in terms of the eccentric angle :

4276

x = a cos <f>,

y = b sin <f.

(Ell.)

4278

x = a sec

y = 6 tan <.

(Hyp.)

Five forms of the equation of the tangent or polar of the

point x'y' :
4280

y-y' = -S (*-#').
ary

(20)

4281

fn-f^i-

4282

y = m^+-v/2m:!+62.

4283

Q^+.gj=1,
a
b

(EU.)

4284

^sec^_ytan^=1

(Hyp.)

(4123)
(4276)

(4278)
4285

x cos y +^ sin y = \/a2 cos2 y + 62 sin2 y,

7 being the inclination of p.

(4054) & (4372)

590

CARTESIAN ANALYTICAL C0NIC8.

Five forms of the equation of the normal at xy :

4286

y-y = ^

(4122>

4287

^-^-=at-b\
X
y

or

hx-ky = al-6s,

where h and k are the intercepts of the tangent.

4289

y = m.r- w(a2~y) .

(4123)

4290

ax sec <f> by cosec< = a3 A8.

(4276)

4291

x&-yxy = {x.x'-y^'),

(4352)

where

is the extremity of the conjugate diameter.

Intercepts of the tangent or polar on the axes :

2

4292

and

72
.
v

(4115), (4281)

Intercepts of the normal :

(4287)

4294

On the x axis,

a2 b2
; x
or

or

e2x.

4296

On the y axis,

^ y

or

y^p#-

Focal distances r, r' of a point xy on the curve :

4298

(aex) in Ell.

4299

(exa) in Hyp.

Peoop.From r> = (aezy+y', and (4272).

Perpendiculars from the foci upon the tangent :

4300

P bJ-n

p'=bj-.

(4095,4282)

TEE ELLIPSE AND HYPERBOLA.

4302

(p- 588) sili SPT = - = 4 = -JU = b

b'
r
r
rr
6a = pp.

591

(4365)

(4300)

4306
Segments of tangent and normal :

4307

PT = %LVW,
ox

pt =*^v/iy.
ay

4309

PG=-SW=^-,
a
a

PG' = -^v/57=$. (4294)

0
0

Bight Line and Ellipse.

Quadratic for the ratio Wj : n2, in which the hne joining
two given points a^, a%y2 is cut by the ellipse (4131).
4310

Equation of the two tangents drawn from x'y' :

4311

($+^_1)g+^_1)-(^+^_1)'=0.

Proof. By the condition for eqnal roots of (4310).

592

OARTESIAN ANALYTICAL OONIGS.

Quadratic for abscissae of points of contact of the tangent

from x'y :
4312 a (VaP+tfift-WVxaf+tf (6s -y'*) = 0. (4282, 4125)

Quadratic for m of the tangent from xy :

4313
to8 (x9~at)-2injpy+yt-bi = 0.

(4282)

General polar equation of the ellipse, or quadratic for r,

the segment intercepted between the point xy and the curve
on the right line drawn from that point at an inclination 0 to
the major axis and x axis of coordinates.
4314

(* sin20+62 cos20) r2

+2r (ay sin 6+Vaf cosfl) + (ay2+6V2-a262) = 0.

4315

Length of intercepted chord = difference of roots.

4316

Distance to middle point of chord = half sum of roots.

4317

Rectangle under segments = products of roots.

Cob.If two chords be drawn to a conic at two constant

inclinations to the major axis, the ratio of the rectangles under
their segments is invariable.
For, if x'y' be their point of intersection, the ratio in ques
tion becomes a2 sin20 + &2 cos20 : a2 sin2 0' -Hi'2 cos2 0J, which is
constant if 0 and
are constant.
Locus of centres of parallel chords :
4318

a*y sin d+tfjc cos 6 = 0.

(4314)

Quadratic for abscissas of points of intersection of the line

Ax+By + G = 0 and the ellipse Z*V + a2y2
= 0.
(4116)
4319

m1) a?+2A Ca*x+

Aoon

= 0.

- ~ACa2 BabV'AW+WP-C*
*

4321

For the ordinates transpose A, B and a, b.

THE ELLIPSE AND HYPERBOLA.

593

Length of intercepted chord

4.322
4=^22

2ab^/(A+B%AW+BW-C*)
AW+m

Hence the condition that the line may touch the elhpse is
4323

4W+J3262 = C*.

The chord through two points x$u x2y2 is

a ona

gfav+fO , yiyi+y*) _ nvVt yiy* , i .

or, denoting the points by their eccentric angles a, |3, the

chord joining aj3 is
Anne
4325

&
. V <* + /J
aB
cos -3^ + ^- sin -J^ = cos-^.

The coordinates of the pole of the chord or intersection of

tangents at Xjyu x2y2 (or a/3 as above).
4326

x = Xi^+A^ = ai(y^~y*) acos(a+fi)

yi+y*

4329

^1^2

cosi(a ^)'

V = <ri^2+^-yi Pi*!**) J Bin-(a+ft)

x^Xi
cos(a B)'

The following relations also subsist

mnnn

262

c? sin a sin B

__ b2 cos a cos B

__ 6 (sin a+sin B) _ a (cos a-j-cos B)

" which are of use in finding the locus of (x, y) when a, /3 are
connected by some fixed equation."
(Wolstenholme's Problems, p. 116.)
4 Q

594

CARTESIAN ANALYTICAL C0NIC8.

4334 If a, /3, y, 8 are the eccentric angles of the feet of the

four normals drawn to an ellipse from a point xy, then
*++y+8 = 3ir or 5ir.
Proof.Equation (4290) gives the following biquadratic in z = tan ty,
byz' + 2 (ax + a'-b') *+2 (aa-a' + i1) 2-6y = 0.
Let a, 6, c, d be the roots. Eliminate d from ab+ac + <fcc. = 0 and abed = 1
(406). Thus ab + bc+ca =
+ + r-; from which, since a = tana,
oe oa ao
Ac., we get
sin (/3 + y) + sin(y + a)+8in (a+/3) =0;
and, since
1 (ao + ac+&c.) + aoi = 0,
tan|(a+/3-ry+5) = oo,
.-. a+/3+y + a = 3x or 5V.
4335 The points on the curve where it is met by the
normals drawn from a fixed point xy are determined by the
intersections of the curve and the hyperbola
aVy Vtfx = <?xy.

(4287)

POLAR EQUATIONS OF THE CONIC.

The focus 8 being the pole (Fig. of 4201), the equation of
any conic is
4336

r(l+ecosfl) = i,

0 being measured from A, the nearest vertex.

For the parabola, put e = 1.
Proof.
r=8P; e = ASP;

l = SL; r = e(8X+SN) (4200) =J+ercos0.

The secant through two points, P, P, on the curve, whose

angular coordinates are a+/3 and a /3 (Fig. 28), is
4337

r {<?cos0+secj8 cos (a 0)} = I.

Proof.Let ASQ = a, PSQ = FSQ = /3.

Analytically. Take (4109) for the eqnation of TP. Eliminate r, and r,
by (4336), and substitute 2o for 0, + fl, and 2/3 for 0,-0,.
Geometrically. Let PP" cut the directrix in Z ; then
is a right angle,
by (U66). Take C any point in PP' ; SC = r; ASC = d. Draw CD, CE,
CF, CO parallel to SL, SP, SQ, SX, and EH parallel to XL. Then
l=SL= SE+HL.
orr
8L QT)
a CE 8P
SL EL EL
Sn=^SD = er cose. ^0 =
DX = CO'
.\ HL=0E = r sin CSF sec ft = r cos (a 6) sec fi,
Z = er cos/3 + r 6ec/S cos (a S).

THE ELLIPSE AND HYPEBBOLA.

595

The equation of the tangent at the point a is, conse

quently,
4338

r {e cos 0+cos (a-0)} = I.

A Focal Chord.
4339

21
~ l-^cos2**'

Length

(4336)

Coordinates of the extremities, the centre 0 being the

origin :
a (e cos 8)
Ism 6
X~ lecos0'
y ~ l ecos0'
4340
4342 The lines joining the extremities of two focal chords
meet in the directrix.
[By (4337)
Polar equation with vertex for pole :
4343

r2 (1 -e2 cos* 6) = 21 cos 6.

(4200)

Polar equation with the centre for pole :

4344
4345

r2 (a2 sin2 6+ 62 cos2 6) = aW,

or

r y/(1 -e2 cos2 6) = b.

Proof.By (4273).

Otherwise, by (4314), with x' = y' = 0.

CONJUGATE DIAMETERS.
dr.

AN

Equation of the ellipse referred to conjugate diameters for

coordinate axes :
4346

an +

- L>

CARTESIAN ANALYTICAL C0NIC8.

596

where
4347

a*a'y
_a2sin2a+62cos2a'

jV

a'6'
a2 sin2 +6* cos2

Here a' = CD, fc' = OP, a is the angle DCR, and /3 the angle
POP.
Proof.Apply (4050) to the equation (4273), putting = and
tan a tan /3 = - ^, by (4351).

When a'=b\ a+/3 =

4349

and equation (4346) becomes

a^+f = a'2 = i (a2+6*).

Let the coordinates of D be x, y', and those of P sc, y ; the

equation of the diameter CP conjugate to CD is

4350
4351

'+-f- = 0.
tan a tan 0

or

mm =

(4318)
a"

xy in terms of x'y', &c.

4352

* = -^y'>
o

4354

y = -xa

eh.

y = -^'-

Hyp.

Proof.Solve (4350) with (4273).

4356
4358
4360

X=(IR, , ^ X = pN.
^+*'2 = 2,
^-^=2,

y"-fy = 6.
y'i-yl=b\

(4274, 4352)
Ell.
Hyp.

(4352)
(4354)

4362

a2+62 = a'2+6'2.

Ell.

(4358)

4363

a2-62= a'2-6'2.

Hyp.

(4360)

4364

a'2 = V+eV.

4365

&'2 = o!-eV = rr.

(4271, '61)
(4298)

THE ELLIPSE AND HYPERBOLA.

597

The perpendicular from the centre upon the tangent at

xy is given by
4366
p*

= 4+ |ra*
o*

(4281> m4,)

The area of the parallelogram PCDL (Fig. of 4307) is

4367

P<*' = ab = a'b' sin w,

where p = PF, a' = CD, V = CP, u> =/.PGD.

Pboof.From (4366), and (4352), and a1=z'1+y'*.

Other values of p* :
4369

p^^-JL*-,.

4371

p* = a2 sin2 6+ 62 cos2 0.

(4362)

Proof.From (4344, '67), putting r = a'.

4372

p* = a2 cos2 y +62 sin2 y,

(4371)

7 being the inclination of p.

4373

p* = 2 (1 e2 sin2y).

(4372, 4260)

Equations to the tangents at P and P , the coordinates of

D being a/,
:
4374

xy'yai' .ab

(4073)

m = X-

DETERMINATION OF VARIOUS ANGLES.

4375

4377

pCd =

tan PCD = - ~,
c &y
where e = x/(a2-&2) = 08.

Fig. p. 595.

(4356)

(4070, 4352-3)

598

GABTE8IAN ANALYTICAL CONIC8.

4378

tan (SPT) = = 1+g.C.8 6,

(4070, 4256, 4336)
cy
esinfl
where 0 = PST.
[See figure on page 588.
If 4> be the inclination of the tangent to the x axis,

4380

tan* = _^ = ej^os0
a-y
sin 0

Proof : i> = 8+ SPT.

4382

4383

(4280)

Then by (631) and (4379).

tan SP& = r?^_.
o cy

tan 4P^' = -

(652, 4378)

tan CPG =

If OP, OP are tangents to an ellipse,

4385

cos POP1 = C01~al~~b*.

OS. OS

Proof.By figure and construction of (1180), POP'= MOS". Therefore

vnjy OS>+OS*-S'M* ZCO'+ICS'-W
.

If
4386

^' are the coordinates of 0,

tanPOi^-ifpg^B.

Proof.By (4311), taking terms of the second degree for the two parallel
lines through the origin and tan <j> from (4112).

It is worthy of remark that the substitutions (4276-8)

may also be usefully employed when the axes of reference are
conjugate diameters : though, in that case, the geometrical
signification of no longer exists.

TEE HYPERBOLA. V .

599

THE HYPERBOLA
REFERRED TO THE ASYMPTOTES.

4387

ay=i(a*+V).

Proof.By (4273) and (4050).

Here z = OK, y = PK.

Equations of the tangent at P, (x\ y).

4388
4389

xy'+x'y = i (2+62)y nuc+'Sm (aa+6*).

4390

m = -K.

4391

(20)
(4123)

Intercepts on the axes CI = 2x , CL = 2y'.

THE RECTANGULAR HYPERBOLA.

4392 Here a = b, e = y/2 ; and the equation with the

ordinary axes is
4393
4394

Tangent

x2f = a*.

(4273)

xafyy' = a*.

(4281)

CARTESIAN ANALYTICAL CONICS.

600

Equation with the asymptotes for axes :

4395
4396

Tangent

2xy = a\

(4387)

xy+x'y = a*.

(4388)

THE GENERAL EQUATION.

The general equation of the second degree is

4400
ax*+2hxy + byt + 2gx + 2fy + c = 0,
4401

or axi+bf + cz*+2fyz+2gzx+21ixy = 0, with 2 = 1.

The equation will be denoted by u or <j> (x, y) = 0.

THE ELLIPSE AND HYPERBOLA.
When the general equation (4400), taken to rectangular
axes of coordinates, represents a central conic, the coordinates
of the centre, Cf (Fig. 30), are
4402

V - hf-hS - G

,/ _ gft-<*/_ Z

(46651

Proof.By changing the origin to the point x'y' and equating the new
g and / each to zero (4048).
For the case in which ab h*, see (4430).
4404

The transformed equation is axt-\-2hjcy-\-btf-\-c' = 0,

4405

where c' = ax'i+2tuv'y+by'2+2gx'+2Jy'+c.

4406

= gJ+jy+c

4407

= ^+2/gA-af-^-^=A
ab ft
C

(4466)

The inclination 0 of the principal axis of the conic to the

x axis is given by
4408

tan 20 = -2A-.
ab

Proof.(Fig. 30.) By turning tho axes in (4404) through the angle 6

(4049) and equating the new h to zero.

THE GENERAL EQUATION.

601

The transformed equation now becomes

4409

aV+6y+c' = 0,

4410

in which a' - \ {+&+ y/U*+{a bf} ,

4411

b' = \ {a+6-V4/2+(a-6)2} ,

a and 6' are found from the two equations

4412

a'+b' = a+b,

a'V = ab-h*.

[See (4418).

The semi-axes and excentricity are

4414

V-4-,

and

6=^1-^). (4273) (4261)

For the coordinates of the foci, see (5008).

4416

Note.If 0 be the acute angle determined by equation

(4408), we have to choose between 0 and 0 +

for theinclina-

tion in question, since tan 20 is also equal to tan (20 + tt).

Rule.*For the ellipse, the inclination of the major axis
to the x axis of coordinates will be the acute angle 0 or 0-\-^v,
according as h and c' have the same or different signs. For the
hyperbola, read " different or the same."
Proof.Let the transformed equation (4409) be written in terms of the
semi-axes p, q ; thus qW+p'y1 = p'q%, representing an ellipse. Now turn
the axes back again through the angle 0, and we get
(g'cos'e+p'sin'O) Xt(p'q1) sin 20 xy + (gs sin'O+js' coss0) y* = p,q\
Comparing this with the identical equation (4404), axi + 2hxy + ly1 = c,
we have

(!>' 3s) sin 20 = 2h,

p^q* = c;

sin20 =

Hence 6 is <-
c pq
2
when h and c' have the same sign, p being >q. A similar investigation
applies to the hyperbola by changing the sign of q1.
This rulo and the demonstration of it are duo to Mr. George Heppel, M.A., of
Hammersmith.
4 H

602

CARTESIAN ANALYTICAL C0NIC8.

INVARIANTS OF THE CONIC.

4417 Transformation of the origin of coordinates alone

does not alter the values of a, h, or b, whether the axes, be
rectangular or oblique. This is seen in (4404).
When the axes are rectangular, turning each through an
angle 6 does not affect the values of
4418

ab-h\

a+b,

g>+f,

or c.

When the axes are oblique (inclination w), transformation

in any manner does not affect the values of the expressions
4422

^-

and

"+*-;.

sin2 (a

sin G>

These theorems may be proved by actual transformation by the formulas

in (4048-50). For other methods and additional invariants of the conic, see
(4951).

4424 If the axes of coordinates are oblique, equation (4400)

is transformed to the centre in the same way, and equations
(4402-6) still hold good. If the final equation referred to axes
coinciding with those of the conic be
4425

aV+&y+c' = 0,

and 6 the inclination of the new axis of x to the old one, we

shall have c' unaltered,
4426

tan 20 =

SAsinu-asin^
2/t costu a cos2w b '

4427
/ _ a+ b 2h cos <o+ y/Q

2sin2o

, , _ a+b 2h cos o ^/Q

2ih7^

where Q = a? + bi+2ab cos 2u> + 4h (a + b) cosw-f-4A*.

Proof.(4404) is now transformed by the substitutions in (4050),
putting /3 = 0 + 9O, and equating the new h to zero to determine tan 20.
a' and V are most readily found from the invariants in (4422). Thus, putting
the new h = 0 and the new = 90,
, ./ = a+b
2/icosw and, a,,,
a_ +b
r-i
6 = abh*
^,,
sin u
sin u>
equations which determine a' and b'.

TEE GENERAL EQUATION.

603

The eccentricity of the general conic (4400) is given by

the equation
44.90
e* - (a + b-2h coso))2
.
****
T=? ~ iab-h*) sin*o>
Proof.By (4415), and the invariants in (4422).
THE PARABOLA.
4430 When ab h2 = 0, the general equation (4400) repre
sents a parabola.
For x', y' in (4402) then become infinite and the curve has
no centre, or the centre may be considered to recede to
infinity.
Turn the* axes of coordinates at once through an angle 6
(4049), and in the transformed equation let the new coeffi
cients be a', 2h', b', 2g', 2/', c'. Equate h' to zero ; this gives
2h
(4408) again, tan 20 =
-. If 9 be the acute angle detera o
mined by this equation, we can decide whether 0 or O+jpr is
the angle between the x axis and the axis of the parabola by
the following rule.
4431 Rule. The inclination of the axis of the parabola to
the x axis of coordinates loill be the acute angle Oifh. has the
opposite sign to that of a, or b, and 0+%ir if it has the same
sign.
Proof. Since ab h% = 0, a and b have the same sign. Let that sign
be positive, changing signs throughout if it is not. Then, for a point at
infinity on the curve, x and y will take the same sign when the inclination is
the acute angle 0, and opposite signs when it is 8 +
But, since
ax1 + by* = + oo , we must have 2hxy = oo, the terms of the first degree
vanishing in comparison. Hence the sign of h determines the angle as stated
in the rule.

4432

sin 6 = J*
v a+6

cos 0 = J"
v a+o

Proof.From the value of tan 20 above, 8 being the acute angle obtained,
and from h1 = ab.
4434

Also

a =0

and

V = a+b.

For a'b' = abk1 = 0, and wo ensure that a' and not V vanishes by
(4431). Also a'+b'=a + b (4412).

CARTESIAN ANALYTICAL CONICS.

604

4436

g' = g cos e+f sin 6 = gVJ+l^y

4438

f' = f cos 0-g sin 6

4440

But if h has the same sign as a and b, change 6 into

(4431)
Proof.By (4418, 4432-3).
The coordinates of the vertex are

4441

rVc.

.__f_

Obtained by changing the origin to the point ar'y' and equating to zero
the coefficient of y and the absolute term. The coefficient of x then gives
the latns rectum of the parabola ; viz. :
4443

L.-^-^^f
6
-/(a+6)3

<*7>

METHOD WITHOUT TRANSFORMATION OF THE AXES.

4445 Let the general equation (4400) be solved as a quad
ratic in y. The result may be exhibited in either of the forms
4446

y = or+/? vV

4447

y = cuv+fi ^piix-pf+iq-p1)},

4448

y = cvc+p v/riGr-y)(,r-S),

4449

4456
4458

where

a = ^,
b'

2px+q),

ft
= *,
r~~~
b'

-ab
H= V~'
r~

y and S = p x/(pi q).

Here y = ax + 0 is the equation to the diameter DD

THE GENERAL EQUATION.

605

(Fig. 31), 7 and $ are the abscissae of D and Hf, its extremities,
the tangents at those points being parallel to the y axis. The
surd = PN = P'N when x = OM. The axes may be rect
angular or oblique.
When ab h2 = 0, equation (4446) becomes
4459

y = a,v +p j Vq'-2p'x,

4460

where

p' = bg-hf,

4462

In this case,

q' = f*bc.

is the abscissa of the extremity of

the diameter whose equation is y = ax + /3 and the curve has

infinite branches.

RULES FOR THE ANALYSIS OF THE GENERAL EQUATION.

First examine the value of ab h2, and, if this is not zero,
calculate the numerical value of c' (4407), and proceed as in
(4400) et seq. If ab h2 is zero, find the values of p' and q'
(4459), The following are the cases that arise.
4464

abh? positiveLocus an ellipse.

Particular Gases.

4465

A = 0Locus the point x'y'.

See (4402).
4466

For, by (4404), the conjugate axes vanish.

bA positiveNo locus.

By (4447-54), since qp* is then positive.

4467

h = 0 and a = bLocus a circle.

By (4144).
4468

In other cases proceed as in (4400-14).

abh? negativeLocus an hyperbola.

Particular Cases.

4469 A = 0Locus two right lines intersecting in the

point x'y' .
By (4447), since qp' then vanishes.

In this case solve as in (4447).

606

CARTESIAN ANALYTICAL CONICS.

4470

b& negativeLocus the conjugate hyperbola.

4471

a+b = 0Locus the rectangular hyperbola.

By (4414), since a = b'.

4472 a = b = 0 Locus an hyperbola, with its asymptotes
parallel to the coordinate axes. The coordinates of the centre
are now and -y-, by (4402).

Transfer the origin to

the centre, and the equation becomes

4473

In other cases proceed as in (4400-14).

4474

(ibh1 0Locus a parabola.

Particular Cases.

4475

p = 0Locus two parallel right lines.

By (4459).

4476 p' = q' 0Locus two coinciding right lines.

By (4459).
4477

p' = 0 and q negativeNo locus.

By (4459).

In other cases proceed as in (443043).

Ex.1.:

2z,-2xy+y1 + 3x-y-l =Q.

3 ,
1 1,
Here the values of a, h, b, g, f, c are respectively 2, 1, 1, ,
ab-V = l;

c' = = ^ + 2fgh-ap-bf-ch} = _ 9
{m&)
C
abh1
4
The locns is therefore an ellipse, none of the exceptions (4465-7) occurring
here. The coordinates of the centre, by (4402), are
, _ hf-bg _ _-.
, _ gh-af _ _ 1
ab-h1 ~
'
J
ab-h'~ 2'
Hence the equation transformed to the centre ia
2xt-2xy+y,--^ = 0.
Turning the axes of coordinates through an angle 8 so that tan 20= 2
(4408), we find the new a and b from
a' + 6' = 3,
a'b'=l- (4412)

THE GENERAL EQUATION.

607

therefore
a = | (3- -/5), 6' = |(3+ ^5),
and the final equation becomes 2 (3 ^5) ass+2 (3+ ^5) j/1 = 9.
The inclination of the major axis to the original z axis of coordinates is
the acute angle tan~' (2), by the rule in (4416).
Ex. (2) :
12x* + 60xy + 75f 12as-8|/-6 = 0.
The values of a, h, b, g,f, c are respectively 12, 30, 75, 6, 4, 6,
abh' = 0 ;
p' = bg-hf = -330;
q =f-bc.
(4460)
Since p' does not vanish (4475-7), the locus is a parabola. Proceeding,
therefore, by (4430-43), we have
tan20 = ^ = -g;

^6 = ^.,

cos 0 =

(4432)

By the rule (4431), we must take 0+\w for the angle, instead of 0. There-

and fe' = a + 6 = 87 (4435).

Consequently the transformed equation is

The coordinates of the vertex are computed by (4441), and the final
equation with the vertex for origin is y = 44/OQ x.
o7 v 29
Ex. (3):

x' + 6xy+9y' + bx + lby + 6 = 0.

5 , ,
15 6,
The values of o, h, b, g,f, c are respectively 1, 3, 9,
ab-h* = 0 and p = bg-hf = 0,
therefore, by (4475-7), if there is a locus at all, it consists of two parallel or
coinciding lines. Solving the equation therefore as a quadratic in y, we
obtain it in the form
(as + 3y + 2) (a; + 3y + 3) = 0,
the equation of two parallel right lines.
The equation of the tangent or polar of x'y' is
4478 Ujat+Uyy+UjZ = 0 or uxx'+ u9tf+uMat = 0 ;
(4401, 1405) obtained by (4120) in the form
4479

(ax'+hy'+g) x+QM'+by'+f)y+gaf+fi/+c = 0,

4480

or {ax+hy+g)

(hx+by+f) y'+gx+fy+c = 0,

608

CARTESIAN ANALYTICAL CONICS.

4481 or
axx'+h {xy'+xy)+byy'+g (x+x')+f(y+y')+c _ 0.
When the curve passes through the origin, the tangent at
the origin is
4482

gx+fy = 0.

(4479)

And the normal at the same point is

4483
4484

fx-gy = 0.
Intercepts of the curve on the axes,
a

4486 Length of normal intercepted between the origin and

the chord
_ VV +./**
(4483-4)
a+o

Bight Line and Conic with the general Equation.

4487 Quadratic for n : n, the ratio in which the line joining
xy, x'y' is cut by the curve.
Let the equation of the curve (4400) be denoted by
</> (x, y) = 0, and the equation of the tangent (4479) by
^ {x, y, x', y') = 0 ; then the quadratic required will be found,
by the method of (4131), to be
(<*', y) +2nn> (x, y, x', y) + n*+ (x, y) = 0.

The equation of the tangents from x'y' is

4488

+ (*', y) <f> (x, y) = {+ (*, y, x\ y')Y-

Peoof.By the condition for equal roots in (4487).

Cor.The equation of two tangents through the origin is
4489

Bx*-2Hxy+Cyt = 0.

The equation of the asymptotes of u (4400) is

4490

aul+2huxiiy+bul = 0.

(4665)

THE GENERAL EQUATION.

The equation of the

aa? + 2hxy + bif = 1 is
4491

equi - conjugates

609

of

the

conic

(a+b)(atf+2hxy+by2) = 2 (ab-h^+y2).

Proof.When the conic is ax' + by1 = 1, the similar equation is

(a + b) (atf+by2) = 2ab (*' + 2/2) or (oas'-ty") = 0,
given by the intersections of the conic and a circle. Transformation of the
axes theu produces tho above by the invariants in (4418).
4492 When the coordinate axes are oblique, the equation
becomes
(a 6)(cws3 by*) + 2x (lue + by)(h a cos o) + 2y (ax + hy)(h icos t) = 0.

General polar equation :

4493

(a cos2 6 + 2h sin 6 cos 6+ b sin2 6) r2

+2 (g cos 0+/sin 6) r+c = 0.

Polar equation with (x, y) for the pole :

4494

(4134)

(a cos2 6+ 2h sin 6 cos 0+ b sin2 6) r8

+2 {(ax+hy+g) C0S 6+(by+hx+f) sin 6) r+F(.vy) = 0.

Equation of the line through x'y parallel to the conjugate

diameter :
4495

(x-*)(ax'+hy'+g) + (y-y')(luv'+by'+f) = 0.

Proof.By the condition for equal roots of opposite signs (4494).

Equation of the conic with the origin at the extremity of

the major axis, L being the latus rectum.
4496

y* = Lx (1 - e2) X*.

(42G9, '59)

Equation when the point ab is the focus and

Ax + By-\-G = 0
4497

the directrix :

v/{0t-a)2+0/-ft)2} = e ^fj*+y ^00, 4095)

4 i

610

CARTESIAN ANALYTICAL CONICS.

INTERCEPT EQUATION OF A CONIC.

The equation of a conic passing through four points whose
intercepts on oblique axes of coordinates are s, 8 and t, t', is
4498

^+2^+^-,(I+:>.)-,(i-+|)+l = 0.

Equation of a conic touching oblique axes in the points

whose intercepts are s and t :
4499

4500

+*J,+ J_^_38L+l = 0>

or

(^.+^._iy+^ = o.

Comparing with the general equation (4400), we have

4501

* = -A
g

t = -
f

v = 2h-^ = 2h^ZSf.
st
c

Perpendicular p from xy, any point on the curve, to the

chord of contact :
4505
p2 = **'f*y8in'*> .
(4096, 4500)
sr-\-t-~ '1st cos <w
Equation of the tangent at x'y' :
4507

2(^-+^-l)(i+i-l)+r(^+* = 0.

4508

The equation of the director- circle is

(1 +%stc)(x' + y* + 2xy cos u) h (z + y cob <o) k (y + xcos w) -f-Afccosw = 0.

The parabola with the same coordinate axes as in (4499) :

Proof.From (4500), putting h (4i74), and therefore r = 4

THE GENERAL EQUATION.

Equation of the tangent at

4510

611

y' :
= 1,

(4509)

m = J1^-,.
V so?

(4123)

Equation of the normal at x'y' :

iMn
. sHmssf
4512
y = mx+
,

jsx'
m = ./-7.

(4122)

4513

x\/s = yVt.

4511

+
or

y = mx+ mS*,
ms-\-t

Normal through the origin

The equations of two diameters are, with any axes,

4514

-.lL _ 1
s
t

and

JL_#_ = _1.
s
t

Proof.Diameter through Of, 2 = by the property OB = BQ, in

the figure of (4211).
*
s
Coordinates of the focus :
4516

* = *+t+2tt cos*,'

V =

(509)

Equation of the directrix :

4518

cos ft>)+#

cos o>) = st cos <a.

Proof.Expand (4509), and form the equation of the polar of the focus
by (4479) and (4516).
When the axes are also rectangular, the latus rectum
4519

L = ,4^^i(r+r)*

(4095, 4516-8)

4520 Locus of the centre of the conic which touches the

axes at the points sO, Ot :
toe = sy.

(4500, 4402)

4521 To make the conic pass through a point x'y' ; substi

tute x'y' in (4500), and determine v.

612

CARTESIAN ANALYTICAL C0NIC8.

SIMILAR CONICS.
4522 Definition. If two radii, drawn from two fixed
points, maintain a constant ratio and a constant mutual
inclination, they will describe similar curves.
4523 If the proportional radii be always parallel, the curves
are also similarly situated.
If there be two conies (1) and (2), with equations of the
form (4400), then
The condition
situated is

of their

4524

being

similar

and

similarly

= T = 77ant)

Proof.By (4404), changing to polar coordinates, r : r = constant.

The condition of similarity only is
4525

g+ff-mV
nao
h- o

(1UM)

or, with oblique axes,

4526

(+6--2fccog)' = (^-a;w.y
hrao
h a b

(4422_3}

CIRCLE OF CURVATURE.
CONTACT OF CONICS.
4527 Def.When two points of intersection of two curves
coincide on a common tangent, the curves have a contact of
ike first order; when three such points coincide, a contact of
the second order ; and so on. To osculate, is to have a con
tact higher than the first.
4528

The two conies (Fig. 32) whose equations are

ax1 + 2hxy + bif + 2gx = 0

(1),

a'x*+2h'xy + b'y2 + 2g'x = 0

(2),

CIEOLE OF CURVATURE.

613

touch the y axis at the origin, 0, by (4482). Eliminate the

third terms from (1) and (2), and we obtain x = 0, the line
through two coincident points, and
4529

(ab'-a'b) jc+2 (hb'-h'b) y-2 (bg'-b'g) = 0,

the equation to LM, the line passing through the two remain
ing points of intersection of (1) and (2).
(4099)
Again, eliminate the last terms from (1) and (2), and we
obtain
4530

(ag'-a'g) x*+2 (hg'-h'g) xy+tyg'-Vg) f = 0,

the equation of the two lines OL, ON.

[By (4111) and (4099)

4531 If the points L, M coincide, the conies have contact

of the first order. The condition for this is that (4530) must
have equal roots ; therefore
4532

(ag'-a'g)(bg'-b'g) = (hg'-h'g)*.

4533 If the conies (1) and (2) are to osculate, M must

coincide with 0. Therefore, in (4529), bg' = b'g.
If in (4532) bg = b'g, the conies have a contact of the
third order.
CIRCLE OF CURVATURE.
(See also 1254 et seq.)
The radius of curvature at the origin for the conic
aa?+2hxy+bif+2gx = 0,
the axes of coordinates including an angle w, is
4534

P = --Z- .
b sin oi

Proof.The circle touching the curve at tho origin is

xi + 2xy cos to + if 2rx sin u = 0,
by (4148), and the geometry of tho figure, 2rsinw being the intercept on
the x axis. The condition of osculating (4533) gives the value of p.
p is positive when the convexity of the curve is towards the y axis.
Radius of curvature for a central conic at the extremity
P of a semi-diameter a', the conjugate being b'.

614

CARTESIAN ANALYTICAL C0NIC8.

4535

,=^!_ = ^ = ^ = J.
a sin ai
p
p*
ab

Proof.Take the equation and figure of (4346) (a = CP).

to parallel axes through P. Then by (4534).

w
Transform

The same in terms of x, y, the coordinates of the point P.

4539

=(*V+2li

Proof.By (5138), or from (4538) and the value of 6 at (4365).

The coordinates of the centre of curvature 0 for P, the
point xy, are
4540

= C^r,

V = ^(-t

where <?=aibt.

PBOOF.-(Fig.
" and OP =
K 6 33.)' From g
OP = *=*
P = PG
with the values of p, PG, and PG' at (4535) and (4309).

= PG

Radius of curvature for the parabola.

Taking the diameter and tangent through the point for
axes,
4542
p=M=J^=
(Fig. of 4201)
sin0
sin3 a
SY
By (4534), and equation (4211).
Coordinates of the centre of curvature at xy (rectangular
axes) :
4545

= ar+2,

17 = -^.

Proof.From y ij : p y : PG and p=2a cosec* 0, PG = 2a cosec 8

and y = 2a cot 0.

The evolute of a central conic (Fig. 33) :

4547
4548

(or)*+(%)*= (2-62)*,
or

(aV+fty-c4)s+27a262cVy2 = 0,

where c2 = a2 ft2.

OONFOGAL GONICS.

615

Proof.Substitute for z, y in the equation of the conic (4273) their

values in terms of 4, ij from (4540). Otherwise as in (4958), or by the
method of (5157).
The curve has cusps at L, H, M, and K.
The evolute of the parabola :

Proof.As in (4548), from the equations (4201) and (4545).

CONFOCAL CONICS.

4550

and

+ =1

are confocal conies, if

aa-a'2 = 62-6'2,
or the sign of

may be changed.

For the confocal of the general conic, see (5007).

4551

Confocal conies intersect, if at all, at right angles.

Proof.If u, u are the two conies in ^4550), changing the sign of V* to

make the second conic an hyperbola, u =0 will be satisfied at their point
of intersection ; and (by a2 a1 = b' + b"') this proves the tangents at that
point to be at right angles (4078, 4280).
Otherwise geometrically by (1168).
4552 Tangents from a point P on one conic to a confocal
conie make equal angles with the tangent at P. [Proof at (1291)
4553 The locus of the pole of the line Ax+By + C with
respect to a series of confocal conies in which a* b1 = A, is
the right line perpendicular to the given one,
Bav-ACy+AB\ = 0.
Proof.The pole of the line for any of the conies being xy ; Aa1 = Cju
and Bb' =-Cy (4292) ; also a'-fc1 = X. Eliminate os and bK
4554 Cob.If the given line touch one of the conies, the
locus is the normal at the point of contact.

616

CARTESIAN ANALYTICAL CONICS.

4555 Graves' Theorem. The two tangents drawn to an

ellipse from a point on a confocal ellipse together exceed the
intercepted arc by a constant quantity.
Proof.(Fig. 132.) Let P, P* be consecutive points on the confocal from
which the tangents are drawn. Let fall the perpendiculars FN, P'N'. From
(1291), it follows that /.PP'N= P'PN', and therefore P'N = PN'. The
increment in the sum of the tangents in passing from P to P' is
Rli'-QQ' + P'N-PN' = RR'-QQ'.
But this is also the increment in the arc QR, which proves the theorem.
4556 If the tangents are drawn from a confocal hyperbola,
as in (Fig. 133), the difference of the tangents PQ, PR is
equal to the difference of the arcs QT, BT.
The proof is quite similar to the foregoing.
4557 At the intersection of two confocal conies, the centre
of curvature of either is the pole of its tangent with respect
to the other.
Proof.Take r + ?= = 1 (i.) and- fr = 1 (ii.) for confocal conies.
a
b
a
o
At the point of intersection, a'2=
and y"1= ^ - (where <?= a* b1);
c
c
by a1 a"1 = V + b'2. The coordinates of the centre of curvature of x'y' in
(i.) are * = ^, y" =
^- (4540-1). The polar of this point with.
respect to (ii.) will be

= 1. Substitute the values of x", y"; and

we see, by the values of

>/, that this is also the tangent of (i.) at P.

4558 A system of coaxal circles (4161), reciprocated with

respect to one of the limiting points D or If, becomes a
system of confocal conies.
Proof.The origin B is one common focus of the reciprocal conies, by
(4844). The polar of B with respect to any of the circles is the same line,
by (41CC). B and its polar (both fixed) reciprocate (4858) into the line at
infinity and its polar, which is the centre of the conic. The centre and one
focus being the same for all, the conies are confocal.

ANALYTICAL

CONICS

IN
TRILINEAR

COORDINATES.

THE RIGHT LINE.

For a description of this system of coordinates, see (4006).

The square of the distance between two points a/3y, a'fi'y' is,
with the notation of (4008),
4601
{a 08-j80(y-y')+f> (y-y)(a-a')+t(a-a')(fi-P)},
4602
= ^ [a cos 4 (a-a'^+l) cos 08-/87+1 cos C(y-y')8}.
Proof.Let P, Q be the points. By drawing the coordinates fly, fl'y',
it is easily seen, by (702), that
PQ3 = [(/3-/3')2+ (y-y'y + 2 &-fl')(y-y') cosA] cosecM
(I).
Now, by (4007), o (a- a') + b (fl-fl') + c (y-/) = 0,
from which 6 (fl-fl')2 = -a (u-a')(j}-fl')-c (fl-fl")(y-y),
and a similar expression for c (y y'f. Substitute these values of the square
terms in (1), reducing by (702).
Coordinates of the point, which divides the straight line
joining the points a/3y, a'/3'y' in the ratio I : m:
4603
l+m

l+m

4=
t+m

By (4082).

ABG being the triangle of reference, and a = 0, (3 = 0,

7 = 0 the equations of its sides, the equation of a line passing
through the intersection of the lines a = 0, j8 = 0 is
4604

/a-r = 0

or
4 K

a-k,3 = 0.

618

TRILINEAB ANALYTICAL CONICS.

Pboof. For this is the locus of a point whose coordinates a, /3 are in the
constant ratio to : I or k (4099).
When I and m have the same sign, the line divides the external angle C
of the triangle ABC ; when of opposite sign, the internal angle C.

The general equation of a straight line is

4605

la+mB+ny = 0,

and it may be referred to as the line (I, m, n).

Proof: la + mfi = 0 is any line through the point G, and (la + mfi) + ny
= 0 is any line through the intersection of the former line and the line y = 0
(4604), and therefore any line whatever according to the values of the arbi
trary constants I, m, i.

The same straight line in Cartesian coordinates is

4606
(I cos a+m cos B-\-n cosy).r
+ (/ sin a+m sin /?+ sin y) y ((pi+mj)2+Mps) = 0.
Proof. By substituting the values of a, fi, y at (4009).

Or, if the equations of the sides of ABC are given in the

form AyV + Biy + Ci = 0, &c., the line becomes
4607

(lA^mAi+nAJx+W+mBt+nBJy
+/G',+mC4+nCs=0.

Proof.By (4095), the denominators like v/'(^1 + -B1) being included in

' the constants 1, m, v.

4608 If = 0, v = 0, w = 0 are the general equations of

the lines a, j3, y, then it is obvious that lu-\-mv = 0 is, like
(4604), a line passing through the intersection of u and v, and
lu + mo + nw = 0 represents any straight line whatever.
To make an equation such as a = p (a constant) homo
geneous in a, /3, y; multiply by the equation 2 = aa + b/3+c-v
(4007), thus
(Op - 2) a + bp3 + Cpy = 0,
which is of the same form as (4605).

THE BIGHT LINE.

4610

619

The point of intersection of the lines

la +mp + ny = 0

and

I'a + m'/3 + n'y = 0

is determined by the ratios

A-t- x= -Jr. = . J ,
mnmn
nlnl
Imlm

and (4017).

The values of a, /3, y are therefore

Ac-t-t

%(mrim'n)

% (nV n'l)

X(lm'l'm)

where

D a (mn-m'n)+b (nl'n'l) + t (Im'l'm).

Proof. By (4017), or by solving the three equations

Oa+6/3 + Cy = 2,
Za + m/3 + iy = 0,
l'a + mfi + n'y = 0.

The equation
4612

aa+I)+ry = 0

or

a sin A +0 sin B+y sin C = 0

represents a straight line at infinity.

Proof. The coordinates of its intersection with any other line
la+mP + ny = 0 are infinite by (4611).
4613 Note: an + 6/3 + Cy = 2, a quantity not zero. The equation
fla + b/3 + Cy = 0 is therefore in itself impossible, and so is a line infinitely
distant. The two conceptions are, however, together consistent; the one
involves the other. And if, in the equation la+mfi+ny = 0, the ratios
I : m : n approach the values a : b : c, the line it represents recedes to an
unlimited distance from the trigon.
4614 The equation corresponding to (4612) in Cartesian coordinates is
0a: + 0y+(7 = 0, the intercepts on the axes being both infinite. Cartesian
coordinates may therefore be regarded as trilinear with the x and y axes for
two sides of the trigon and the other side at an infinite distance.

4615 The condition that three points

ai/3i7u aifiiii> a8/3s7s may ue on the same
straight line is the determinant equation,

i
*
s

A
A
A

7i
y.2 | = 0.
7s

Proof.For it is the eliminant of the three simultaneous equations,

Inj+mft + nyjSsO,
la, + m/J, + ny
Ja,+wt/3, + ny, = 0. (583)

TRILIXEAR ANALYTICAL COXICS.

620

4616 Cor.The above is also the equation of a straight line

passing through two of the fixed points if the third point be
considered variable.
4617 Similarly, the condition that the three following
straight lines may pass through the same point, is the deter
minant equation on the right,
l1a+mlfi-\-nly = 0
= 0.
/3a+ms+3y = 0,

4618

The condition of parallelism of the two straight lines

/ a+n /3+ n y = 0,
ia+m'p+n'y = 0,

is the determinant equation

I m
n
/tit
m
n
a

= 0,

Proof.By taking the line at inBnity (4612) for the third Hue in (4617).
4619 Otherwise the equations of two parallel lines differ hy
a constant (4076). Thus
la+mB+ny+k (Za+bp+ty) = 0
or

(4007)

(/+*a) a+(m+kb) P+(n+kt) y = 0

represents any line parallel to la + m8-\-ny = 0 by varying

the value of k.

The condition of perpendicularity- of the two lines in

(4618) is
4620

W+mm'+nri{mri+m'n) cos A (nl'+riI) cosB

(Im'+l'm) cosC = 0,

4621

or I' (lm co$C n co$B) + m' (mn cos A IcosC)

-+-n' (n l cos B m cos A) = 0.

Proof.Transform the two equations into Cartesians, by (4606), and

apply the test AA' + BB' = 0 (4078), remembering that
cos(/3-y) = -cob A, 4c. (4vll).

THE IIIGUT LINE.

621

When the second line is AB or 7 = 0, the condition is

4622

n = m cos A +1 cos B.

It also appears, by (4676), that (4620) is the condition that

the two lines may be conjugate with respect to the conic
whose tangential equation is
4623

F+m2+n2-2mn cos A 2nl cosB-2lm cos C = 0.

The length of the perpendicular from a point a'/3'y' to the

line la + m(i+ny = 0 :
4624

la'+tn? n

V {P-\-m2-\-n2 2mn cos A 2nl cos if 2/wi cos C\

Proof.By (-1095) the perpendicular is eqnal to the form in (4006), with
as', y' in the place of x, y, divided by the square root of sum of squares of
coefficients of x and y. The numerator = la + m /3' + ny'. The denominator
reduces by cos(/3 y) = cos^., &c.
4625

Equation of the same perpendicular :

a

lm cos Cn cos B

/3
y

ft
y

mncosAlcosC
n I cosB mcosi

=0.

Proof. This is the eliminant of the three conditional equations

La + Mfl + Ny = 0, La' + Mp + Ny' = 0, and equation (4621).

4626 Equation of a line drawn through a'P'y' parallel to the

line (I, m, n) :
a a' tm ijn
= 0.
y

blHm

Proof. It is the eliminant of the three conditional equations

la + mfi+ny = 0, la' + m(i' + ny' = 0, and the equation at (4618).
4627 The tangent of the angle between the lines (I, m, n)
and (/', to', n ) is
(mn'm'n) sin A -\-(nl' n't) sin 2?+ {lm I'm) sinC
U'+mm'+nn'- (mn'+m'n) cosA (nl'-\-n'l) cosB(jtm'-{-l'm) cosC
Proof.By (4071) applied to the transformed equations of the lines,
(4606), observing (4007).

TRILIXEAB ANALYTICAL C0X1CS.

622

EQUATIONS OF PARTICULAR LINES and COORDINATE RATIOS

OF PARTICULAR POINTS IN THE TRIGON.
4628

Bisectors of the angles A, B, C :

B-y = 0,

4629

y-a = 0,

a-8 = 0.

Centre of inscribed circle (or in-centre)* 1:1:1.

The coordinates are obtained from their mutual ratios by the formula
(4017).
4630

Bisectors of the angles A, n B, v C:

B-y - 0,

y+a = 0,

a+j8 = 0.

Centre of the escribed circle which touches the side a (or

a ex-circle)
4631

1:1:1.

Bisectors of sides drawn through opposite vertices :

B sin B = y sin C,
4632

y sin C = a sin .4,

a sin A B sin B.

Point of intersection (or mass-centre) :

cosec .4 : cosecU : cosec C.

Proof.Assume vift ny = 0, by (4604), as the form of the equation of

a line through A, and determine the ratio m : n from the value of y : /3
when o = 0.
The coordinates of the point of intersection may be found by (4610), or
thus :
a : (i = Bin B .: Bin A = cosec.4 : cosecB,
ft : y = sin C : sin B = cosec B : cosec G,
therefore
a : ft : y = cosec A : cosec B : cosec 0.
4633 Perpendiculars
vertices :
BcosB = ycosC,
4634

Orthocentre :

to sides

drawn

y cos C= a cos .4,

through

opposite

a cob A = BcosB.

sec A : sec B : sec C.

* This nomenclat ure is suggested by Professor Hudson , who proposes the following :
" In-circle, circum-cirele, a ex-circle ... mid-circle for inscribed circle, circumscribed circle,
circle escribed to the side a, and nine-point circle ; also in-centre, circum-centre, a excentre, ... mid-centre, for the centres of these circles ; and in-radim, circum-radius, a exradius, ... mid-raditu, for their radii ; central line, for the line on which the circum-centre,
mid -centre, ortho-centre, and mass-centre lie ; and central length for the distance between
the circum-centre and the ortho-centre."

THE RIGHT LINE.

623

If the Cartesian coordinates of A, B, G be xxylt a^, x3y3i

the coordinates of the centre of the inscribed circle are
4635

x = eM?i+kg+ftr

a.yi+%+cjy3

Proof.By (4032). Find the coordinates of D where the bisector of

the angle A cuts BG in the ratio 6 : c (VI. 3), and then the coordinates of E
where the bisector of B cuts AD in the ratio b + c : a.
4636 For the coordinates of the centre of the a ex-circle,
change the sign of a in the above values of x and y.
4637

The coordinates of the mass-centre are

* = s fa+x*+x)

y = i (yi+yz+ys)

4638 The coordinates of the orthocentre are obtained from

the equations of the perpendiculars from x2y2, xsys, viz.,
fa x3) x+{yly3) y = xi fa

(yiy3),

fa #2) x + fayjy = x3 fax2)+y3 (#!#,).

Perpendicular bisector of the side AB :

4639

asin^-jSsinB+y sin (4-2?) = 0,

4640

or a cos 4-/8 cos I? ^-sm(A-B) = 0,

4641

or

/ . asini?sinC\
.
la. 6sinCsin4\

n
a+
.
)cos4
_____ cosU = 0.
V
2 sin 4
/
\
2 sin/? /
4642

Centre of circumscribed circle (or circum-centre) :

cos A : cosi? : cosC.

Proof. A line through the intersection of y and a sin A (3 sin B (4631)

is of the form o sin.4 /3 sinU+ny 0, and, by (4622),
n si u B cos -4 + sin .4 cos 2? sin (-1 Z?).
Otherwise, by (4633) and (4619),
a cos Ap cosB + k = 0
is any line perpendicular to AB; and the constant k is found by giving a '. /3
the value which it has at the centre of AH.

624

4643

Centre of the nine-point circle (or mid-centre) :

cos (B-C) : cos (G-A) : cos (A-B).

Proof.By (955) the coordinates are the arithmetic means of the corres
ponding coordinates of the orthocentre and circum-centre. Therefore, by
(4034, *42) and (4017).
a
^C
see A
_j
cos A
")
I sin^l sec.l +sinfi secB + sin(7 6ecC sin.4 cos^4 + sin.Bcos.B-(-sinCcosC ) '
which reduces to cos (B C) x constant.
4644 Ex. 1.In any triangle ABC (Fig. of 955), the mass-centre E, the
orthocentre 0, and the circum-centre Q lie on the same straight line ;* for the
coordinates of these points given at (4632, '34, '42), substituted in (4615),
give for the value of the determinant
cosec A (sec B cos C cos B sec C) + &c,
which vanishes.
Similarly, by the coordinates in (4643), it may be shown that the midcentre A7 lies on the same line.
Equation of the central line :
Ex. 2.To find the line drawn through the orthocentre and masscentre of ABC. The coordinates of these points are given at (4632, '34).
Substituting in the determinant (4616) and reducing, the equation becomes
o sin 2^1 sin (B-C)+ft sin 2B sin ((7 A) + y sin2C sin {A-B) = 0.
Ex. 3.Similarly, from (4629, '42), the line drawn through the centres
of the inscribed and circumscribed circles is
a (cos B cos C)+ft (cos C cos A) + y (cos A cos J5) = 0.
Ex. 4.A parallel to AB drawn through C:
a sin A + ft sin B 0.
For this is a line through aft, by (4604), and the equation differs only by a
constant from y = 0, for it may be written
(a sinA+fi sinB + y sin C)y sin (7 = 0.
Ex. 5.A perpendicular to BC drawn through C is
a cos C + ft 0.
For a perpendicular is ft cos B y cos C = 0 (4633)
(1),
and a line through C is of the form la + m/i = 0. Hence, by (4619), the
constant k (a sin A + ft sin B + y sin C) must be added to (I) so as to elimi
nate y. Thus
ft sin C cos B + a sin A cos C + ft sin B cos C = 0,
ft sin (B + C)+a B\nA cos C = 0 or ft + a cos C = 0.
* The central line. See note to (4629).

ANHARMONIC RATIO.

625

ANHARMONIC RATIO.

For the definition, see (1052).

4648 The three ratios of that article are the values of the
ratio h:k' in the three following pencils of four lines respec
tively
a = 0,

a-A/3 = 0,

=0,

a+'y3 = 0... (i.) (Fig. 34),

o = 0,

a-A = 0,

a-fc'/8=0,

a = 0,

= 0,

a+& = 0,

0 = 0... (ii.) (Fig. 35),

a+#/8 = 0...(iii.) (Fig. 36).

4649 The anharmonic ratio (i.) becomes harmonic when

1c = k'. Hence the lines a + /cj3, a k(i form a harmonic pencil
with the fines a, /3, the first dividing the external and the
second the internal angle between a and /3 (Fig. 37).
4650 Similarly, the anharmonic ratio of four lines whose
equations are
a /Aj/J = 0,

anzP = 0,

is the fraction

a/x3 = 0,

= 0,

0*i fh) 0*3 /*)_

0*1 /**) 0*2 Ms)

Proof.Let OL be the line a = 0, and

OB, 0=0.
/i,/U, = difference of perpendiculars from
^4 and upon OL, divided by p.
Similarly, ftt i*t, &c. These differences
are proportional to the segments AB, CD,
AD, BC, and p is a common divisor.
4651 nomographic pencils of fines are those which have the
same anharmonic ratio. Thus the two pencils

and

'*i0

Msft

<* A*4^

a fifl,

a' piP,

a' fifi',

a' fifi',

are nomographic pencils.

4 L

626

TBILIXEAB ANALYTICAL CONICS.

THE COMPLETE QUADRILATERAL.

4652 Def.Any four right lines together with the three,
called diagonals, which join the points of intersection, make a
figure called a complete quadrilateral.
4653 Let 0 be any point in the plane of the trigon ABC.
Draw AOa, BOb, COc, and complete the figure. The equa
tions of the different lines may be written as under, with the
aid of proposition (4604), the ratios I : m : n being arbitrary
and dependent upon the position of 0.

Aa,

mfi ny = 0,

AP,

mfi+ny = 0,

Bb,

ny -la =0,

BQ,

ny +la = 0,

Cc,

la -111/8=0,

CR,

la +mfi=0;

be,

mfi+ny -la

=0,

OP,

mfi+ny -2la

ca,

ny+la -mfi = 0,

OQ,

ny +la -2mfi = fr,

ab,

la +mfi-ny = 0,

OR,

la +mfi'2ny = 0,

PQR,

= 0,

la+ mfi+ny = 0.

Proof.Aa, Bb, Gc are concurrent by addition, he is concurrent with

lib and ji, and with Cc and y, by (4604). AP and OP are each concurrent
with be and a. PQB is concurrent with each pair of lines 6c and a, ca and
ft, ab and y. Similarly for the rest.
4654 Every pencil of four lines in the above figure (supply
ing AP, BQ, CB) is a harmonic pencil.
Proof.By the test in (4649), the alternate pairs of equations being the
sum and difference of the other two in every case.
Otherwise by projection. Let PQBS be the quadrilateral, with diagonals
BP, QS meeting in C. (Supply the lines AC, BC in the figure.) Taking the
plane of projection parallel to OAB, the figure projects into the parallelogram
pqrs ; the points A, B pass to infinity, and therefore the lines AC, BC become

THE GENERAL EQUATION.

627

lines harmonically divided by the sides of the parallelogram, the centre,

and the points at infinity.

4655 Theorem (974) may bo proved by taking a, fl, y for the lines BG,
CA, AB, and Va + mfi + ny, la + mfl + ny, lu + mft + n'y for be, ca, ab, the last
form being deduced from the pi-eceding by the concurrence of Aa, Bb, and Cc.

THE GENERAL EQUATION OF A CONIC.

The general equation of the second degree is

4656

aa2+6/S2+ey2+2//8y+2^ya+2/a^ = 0.

This equation will be denoted by <j> (a, (5, y) 0 or u = 0.

Equation of the tangent or polar :
4657

Uja+Upfi+Urf = 0

or

tt.af

= 0,

the two forms being equivalent and the notation being that of
(1405). The first equation written in full is
4659
(aa'+hfi+gy') a+{ha'+bfi'+fy') p+(ga' +&+<*/) y = 0.

628

TBILINEAB ANALYTICAL GONIOS.

Proof. By the methods in (4120). Otherwise by (4678) ; let a/3y

be on the curve ; then <j> (a, ft, y) = 0. Next let the point where the line cuts
the cnrve move up to afty. Then the line becomes a tangent and the
ratio n : n vanishes; the condition for this gives equation (4658).
Cor.The polars of the vertices of the triangle of refer
ence are
4660

aa+h/3+gy = 0, ha+bfi+fy = 0, ga+ffi+cy = 0.

4661 The condition that u may break up into two linear

factors representing two right lines is, by (4469), A = 0,
where
4662
A = abc+2fgh-af-bgt-ch\
(4454)

4663 The general tangential equation

of the conic (4656) expresses the condi
tion that the line Xa + A*/3 + *y niay touch
the curve and is the determinant equation
annexed. The same written in full is
4664

a
h
g
X

h
b
f
ft

g A
f p
C v
v

= 0.

{bc-f) \*+(ca-g*) iS+iab-h*) v*

+2 (gh-af) /tv+2 (hf-bg) v\+2 (fg-ch) \p = 0,

4665

or A\.2+BH.2+Cvi+2Fnv+2Gv\+2H\n = 0;

writing, as in (1642),
A = bc-f,

B = ca-g2,

C = ab-h\

F = gh-af,

G = hf-bg,

H = fg-ch.

The tangential equation will be denoted by <t> (X, n, v) = 0 or

U = 0, to correspond with (4656).
Pkoof. The determinant is the eliminant of the equation of the line
Xa + fifi + vy = 0, and the three equations obtained by equating X, fi, v to the
coefficients of a, ft, y in (4659).
Othencise.Assume \a + pft-\-vy 0 for the tangent. Substitute the
value of the ratio ft : y obtained from it in the equation of the curve, and
express the condition for equal roots (4119).
4666 Conversely, if the line Xa 4-^/3
+ vy has the coefficients X, p, v con
nected by the equation of the second
degree V = 0 (4664), then the enve
lope of the line is the conic in the

A
H
G
a

H
B
F

G
F
C
y

a
p
y

= 0.

THE GENERAL EQUATION.

629

determinant form annexed corresponding to (4663), or in full

4667

(BC-F8) a,+(C4-Gi)/5*+(^-/P)ys

+2(GH-AF) fiy+2(HF-BG) ya+2 (FG-CH) a/8 = 0.

4668

or A (aa2+fy8*+cy2+2/y+2g7a+2/jaj8) = 0.

Proof.Eliminate v from 77 = 0 and the given line. The result is of

the form L\* +2EX^ + M/u* = 0, and therefore the envelope is LM = 2i!J, by
(4792). This produces equation (4667). The coefficients are the first
minors of the reciprocal determinant of A (1643), and therefore, by (585),
are equal to aA, 7>A, &c.
4669 The condition that U may consist of two linear factors
is, as in (4661), D = 0, where
4670 D = ABC+2FGH-AFi-BGi-CH* = tf. (1643)
In this case U becomes the equation of two points, since the
line Aa-f-Ai3 + vy must pass through one or other of two fixed
points. See (4913).
4671

The coordinates of the pole of

are as

A\+Hfi+Gv : HX+Bp+Fv : G\+Fp+L\

4672

or

Ui-.Ut-.U,.
aa + hfi + gy = k\
ht + b/i+fy =k/t
ga+fft + cy = lev.

A\ + H/x + Gy

nx + Bfi + F*

G\+Fp + Cf

A'

Hence the tangential equation of the pole of \'a + p'fi + v'y,

i.e., the condition that \a-\-nfi-\-vy may pass through the
pole ; or, in other words, that the two lines may be mutually
conjugate, is
4674

\Ux+pUH'+vU, = 0

or \'UK+^+v'U,= 0t

The coordinates of the centre a0, /30, y0 are in the ratios

4677 Aa+m+Gt : m+Bb+Ft . Ga+Fb+Ct,
where a, I), C are the sides of the trigon.

630

TBILINEAR ANALYTICAL GONICS.

Proof.By (4671), since the centre is the pole of the line at infinity
aa + &/3 + ey = 0 (4612).
The quadratic for the ratio n : n' of the segments into
which the line joining two given points a(iy, a'/3'y' is divided
by the conic is, with the notation of (4656-7),
4678
<f> (a', ?, y) u2+2

nn'+i (a, ft, y) * = 0.

Proof.By the method of (4131).

The equation of the pair of tangents at the points where 7
meets the general conic u (4656), is
4679

u?+2Au.up+6u|[= 0.

Proof.The point a'/3', where y meets the cnrve, is fonnd from

aa'- + 2ha'iV + bft'1 = 0 [y = 0 in (4656)]. The tangent at snch a point is
.a'+Mp/3' = 0 (4658). Eliminate a, p.
The equation of a pair of tangents from a'fi'y is
4680

+ (W) <f> Mb) = UJ+Mr+k/Y.

Proof.By the condition for eqnal roots of (4678).

By actual expansion the equation becomes
(JV+C/P-2F/3y) a'J+(Ca5+^y1-2(?ya)/3'J + (.4/3, + Ba5-2Ha/3) y"
+ 2 ( - Ajiy + Hya + G/3 - Fa*) /3'y'
+2 (jff/3y-.Bya + fty3-(?,35) y'a'
+ 2 {Gfiy + Fya-Cafi-Hy%) a'/3' = 0.
In which either a', /3', y' or a, /3, y may be the variables, for the forms are
convertible.

4681

Otherwise the equation of the two tangents is

4682

<t>({3y'-/3?y, ya'-y'a, a^-a'p) =0.

(4665)

Proof.By substituting liy'p'y, &c. for X, 11, v in (4664), the condition

that the line joining a'fi'y' to any point a/3y on either tangent (see 4616)
should touch the conic is fulfilled. The expansion produces the preceding
equation (4681).

The equation of the asymptotes is

4683

* (a, fry) = 4 (a,, ft, y.) = A2

where a0, |30, y0 are the coordinates of the centre.

(1).

THE GENERAL EQUATION.

Otherwise
a, ft, y, is
4684

the

equation,

in a form

631

homogeneous

(nao+bfa+ty,) $ (a, fi,y) = k (aa+bP+ty)2

in

(2),

where 3, i), C are the sides of the trigon.

And, finally, if the tangential equation (4664) be denoted
by 3> (X, ^, v) = 0, the equation of the asymptotes may be
presented in the form
4685

* (a, I), r) <f> (, 0, y) = (aa+I)+r7)2A

(3).

Proof. (i.) The asymptotes are identical with a pair of tangents from
the centre ; therefore, put a0, fta, y0 for a', ft', y in (4680) ; thus
f (a, ft, y) 0 (a0, ft0, 7o) = V (aa + 6/3 + Cy)1 = V3>
(4),
since the polar becomes the line at infinity.
Now, multiplying the three equations in (4672) by a, ft, y respectively,
and adding, we obtain <p (a, ft, y) = k (Xa+pft + vy), and therefore
<Ma0,fty0) = k(an + W + Cy) = Ja
(5),
since the line at infinity (4612) is the pole of the centre.
From (4) and (5), by eliminating k, equation (1) is produced ; and by
dividing (4) by (5), we get equation (2).
Again, taking the values of a, ft, y from (4673), we havo
* + rf + >>r = *JW), and therefore ao-L^+c^ = * <a' b> c>.
k
A
k
A
By the last equation, (2) is converted into (3). See also (4960).
Cor. Since the centre (a0, /30, y0) is on the asymptotes,
we have
4686

t K A, yo) =

- * (a, b, 0-

4687 The semi-axes of the general conic (4656) are the

values of r obtained from the quadratic
/ , 5_cos_4\
h,
g,
a
V ^
r*
/' /
bs cosB\
f,
h
h,
\ +
r*
/'/ Lc*cosC\
*
f>
\C+)'
a,

fr,

t,

where a, I), t are the sides of the trigon, and

s = aI)rA-r-*(aur).

= 0,

632

TEILINEAB ANALYTICAL CONICS.

Proof.The centre being a^ftoy^ put aa0 = x, 0fio = y, y y0 =

afly being a point on the conic, and r the radius to it from the centre, we
have, by (4602),
r = ~ (as'a cos A + y,b cos B + z>c cos G)

(1).

Also (4656),
;. (a,/3, y) = f (a0 + x, fi0 + y, y0 + z) = 0.
Expand and write I, m, n for aa0 + hfi0+gya, ha0+ bfa+fy,,, ?"o+/A,-r c>v
The terms in x, y, z become
lx + wy + nz= l(n-a0)+&c. = 2-2 = 0 (4007)
(2),
and we obtain <p (x, y, ) = -$ (o0, ft, y) = 2'A-4-* (a, i, c) (4686)
(3).
The maximum and minimum values of r1 and therefore of
a?acoaA + y*b cos U + 'c cos 0
(4)
are required, subject to the equations (2) and (3). By the method of unde
termined multipliers (1862), the quadratic above is found.Ferrers ' Tril.
Coord., Ch. 4, Art. 18.

4688

The area of the conic =

irSaftfA

{*(a,fc,r)j*
Proof. If the roots of the quadratic (4687) are T\ i*-!*, the area
will be iri\rr The coefficient of r~* reduces by trigonometry to SV, and
the absolute term is * (a, 6, c). Hence the product of the roots is found.

4689 The conic will be an ellipse, hyperbola, or parabola,

according as $ (a, I),f) (4664) is positive, negative, or zero.
Proof. The squares of the semi-axes have opposite signs in the hyper
bola. Therefore the product of the roots of the quadratic (4687) must for
an hyperbola be negative, and therefore * negative in (4688).
* (a, 6, c) = 0 makes the curve touch the line at infinity (4664), a pro
perty which distinguishes the parabola.

The condition that the general conic (4656) may be a

rectangular hyperbola is
4690
a+b+c = 2fvoaA+2g cosJB+2A cos C.
Proof.Let the asymptotes be
la + mfl + ny = 0,
Va+m'fi+n'y = 0.
Forming the product, equating coefficients with (4685), and denoting
>j> (a, b, c) by <t>, we get the proportions
11'
a<j> a'JA

mm'
nn'
_ mn' + m'n
b(f> b-A cf c2A ~ 2(f<p beA)
nl' + n'l _ Im + I'm
2 (gf cab) ~ 2(ty abA)'

THE GENERAL EQUATION.

633

We may therefore substitute these denominators in (4620) for the condition

of perpendicularity of the asymptotes. The result reduces to the equation
above, by (837).
For another method, see (5002).

4691 The general conic (4656) will become a circle when

the following relation exists between the coefficients :
b sin*C+c mxiB-2fsmB sinC
= c sin2.4+a sin2C 2g-sinC sin4
= a sin2.B+6 sin2.4 2/t sin.4 sinB.
Peoop. Equate coefficients of the equation of tho conic (4656) with
those of the circle in (4751).

4692 The equation of the pair of lines drawn from a point

a'/3'-y' to the points of intersection of the conic <j> and the fine
L = Xa+^/3+vy = 0 is, writing L' for \a'-\-p(5' + vy, with the
notation of (4656-7),

L*4> (, A y) - 2LL\fa'+ W+M) + L*<f> (a', ?, /) = 0.

Proof.By the method of (4133).

4693 The Director-Circle of the conic, that is, the locus

of intersection of tangents at right angles, is, in Cartesians,
C (^+yi)-2Ga;-2Fy+A+B = 0.
Proof.Let the equation of a tangent through xy be
m jj + (y mx) = 0.
Therefore in the tangential equation (4665) put X = m, ft = 1, v = ymx,
and apply the condition, Product of roots of quadratic in m = 1 (4078).
The equation of the same circle in trilinears is
4694 (B+C+2FcoBA)a% + (G+A + 2GcoBB)^+(A + B+2HcosG)yt
+ 2 (4 cos A - H cos B - G cos 0- F) /3y
+2 (-H cob A+B cos B-F cos G-G) ya
+ 2 ( -G cos A-F cos B+ G cos O-H) a/3 = 0 ;
or, in the form of (4751),

4695
4 M

634

TBILINEAR ANALYTICAL GONICS.

Proof.The equation of a pair of tangents (4681) through a point a/3y

in trilinears, when the tangents are at right angles, represents the limiting
case of a rectangular hyperbola. Therefore the equation referred to must
have the coefficients of an, yS'1, &c. connected by the relation in (4690),
which thus becomes the equation of the locus of the point a/3y ; i.e., the
director-circle.

4696 When the general conic is a parabola, G = 0 in (4693)

and $ (a, b, c) = 0 in (4695), by (4430) and (4689), and these
equations then represent the directrix.

PARTICULAR CONICS.
4697 A conic circumscribing the quadrilateral, the equa
tions of whose sides are a = 0, /3 = 0, y = 0, 8 = 0, (Fig. 38)
ay = k/38.
Proof. This is a curve of the second degree, and it passes through the
points where a meots fi and S, and also where y meets /3 and S.

4698 The circumscribing circle is ay = 8; + or , as

the origin of coordinates lies without or within the quadri
lateral.
Proof. Transform (4697) into Cartesians (4009) ; equate coefficients of
x and y and put the coefficients of xy equal to zero.

4699 A conic having a and y for tangents and /3 for the

chord of contact :
(Fig. 39)
Proof.Make I coincide with /3 in (4698).

4700 A conic having two common chords a and /3 with a

given conic 8 :
(Fig. 40)
S = ka/3.
4701 A conic having a common chord of contact a with a
given conic 8 :
(Fig. 41)
S = kai.
4702 Cor.If BPQ be drawn always parallel to a given
line, PN* oc RP.PQ, by (4317).

PARTICULAR GONICS.

635

4703 A conic having a common tangent T at a point x'y'

and a common chord with the conic 8:
(Fig. 42)
S = T{lx+my+nz).
4704 A conic osculating 8 at the point x'y' where T touches
at one extremity of the common chord I (xx') + m (yy') :
(Fig. 43)

S= T(lx+my Ix'-my').
4705 A conic having common tangents T, T' at common
points with the conic 8 :
(Fig. 44)
S = kTT.
4706 A conic having four coincident points with the conic
8 at the point where T touches :
(Fig. 45)
S = kT\
4707

The conies S+L* = 0, S+M2 = 0, S+N* = 0,

(Fig 46) having respectively L, M, N for common chords of

contact with the conic 8, will have the six chords of inter
section
LM=0,
MN=0,
NL=0,
passing three and three through the same points.
ProofFrom (S + Jtf)-(6f+^) = (M+N)(M-N), &c.
By supposing1 one or more of the conies to become right lines, various theorems
may be obtained.

4709 The diagonals of the inscribed and circumscribed

quadrilaterals of a conic all pass through the same point and
form a harmonic pencil.
Proof. (Fig. 47.) By (4707), or by taking LM=R? and I'M' = E"
for the equations of the conic by (4784).

4710 If three conies have a chord common to all, the other

three chords common to pairs pass through the same point.
Proof.(Fig. 48.) Take S, S + LM, S+Ltf for the conies, L being the
chord common to all ; then M, N, MN are the other common chords.

636

TBILINEAB ANALYTICAL G0NIG8.

4711

The hyperbola oey = (0x+0y+p)2

is of the form (4699), and has for a chord of contact at infinity

Ox + Oy+p = 0, x,y being the tangents from the centre.
4712

The parabola y2 = (Ox+tiy+p) x

has the tangent at infinity Ox + Oy+p = 0.

4713 So the general equation of a parabola may be put in
the form of (4699). Thus
(o*+ity)+(^+2^+c)(0*+0y+l) = 0.
Here ax + fly is the chord of contact, that is, a diameter; 2gx+2fy+e is
the finite tangent at its extremity, and Oz + Oy + l the tangent at the other
extremity, supposed at infinity.
4714

The general conic may be written

(aa?+2hxy+by2) + {2gx+2Jy+c)(0x+()y+l) = 0.
For this is of the form ay + kflS, 2 being at infinity.
4715

The conies

S and S-k (0^+Oy+l)8

have double contact at infinity, and are similar.

4716

The parabolas

S and

Sk2

have a contact of the third order at infinity.

Proof.For S and S (Ox + Oy + k)* have the line at infinity for a chord
of contact; and, by (4712), this chord of contact is also a tangent to both
curves.
4717 All circles are said to pass through the same two
imaginary points at infinity (see 4918) and through two real
or imaginary finite points.
Proof.The general equation of the circle (4144) may be written
(x + iyXx-iy) + (2gx+2fy + c)(0x + 0y + l) = 0;
and this is of the form (4697). Here the lines xiy intersect Oa + Oy + l
in two imaginary points which have been called the circular points at infinity,
and 2gx+2fy + c in two finite points P, Q ; and these points are all situated
on the locus xi+y' + 2gx + 2fy+c = 0.
4718 Concentric circles touch in four imaginary points at
infinity.

PARTICULAR CQNIOS.

637

Proof. The centre being the origin, equation (4136) may be written
(x+iy)(x iy) = (0x + 0y+ry, which, by (4699), shows that the lines x iy
have each doable contact with the (supplementary) carve at infinity, and the
variation of r does not affect this result. Compare (4711).

4719

The equation of any conic may be put in the form

Here x = 0, y = 0 are two sides of the trigon intersecting at

right angles in the focus ; y = 0, the third side, is the directrix,
and e is the eccentricity.
The conic becomes a circle when e = 0 and y = oo , so that
ey = r, the radius, (4718).
4720 Two imaginary tangents drawn through the focus
are, by (4699),
{x+iy){x iy) = 0.
These tangents are identical with the lines drawn through
the two circular points at infinity (see 4717). Hence, if two
tangents be drawn to the conic from each of the circular
points at infinity, they will intersect in two imaginary points,
and also in two real points which are the foci of the conic.
All confocal conies, therefore, have four imaginary com
mon tangents, and two opposite vertices of the quadrilateral
formed by the tangents are the foci of the conies.
4721 If the axes are oblique, this universal form of the
equation of the conic becomes
a?+2xy cosai+y2 = e*y2.
The two imaginary tangents through the focus must now
be written
{oc-\-y (cos<u+* sin <u)}

(cosw i sin<u)} = 0.

4722 Any two lines including an angle 0 form, with the

lines drawn from the two circular points at infinity to their
point of intersection, a pencil of which the anharmonic ratio
is e'*-"*.
Proof.Take the two lines for sides /3, y of the trigon. The equation
of the other pair of lines to the circular points will be obtained by elimin

TBILINEAB ANALYTICAL C0NIG8.

G38

ating a between the equations of the line at infinity and the circnm-circle,
viz.,

aa + 6/3 + cy = 0 and

- + -- + - = 0.
(4738)
a
fl
y
The result is
/? -f 2/3y cos 6 + y* = 0 ;
or, in factors,
(/3 + e"y ) (/3 + e'^y) = 0.
The anharmonic ratio of the pencil formed by the four lines j3, /3 + e" y,
y,/He-*y is, by (4648, i.),
e" : e"to = ef = J[U).
4723 Cob.If 9 = \ir, the lines are at right angles, and the
four lines form a harmonic pencil. [Ferrers Tril. Coords., Ch. VIII.

THE CIRCUMSCRIBING CONIC OP THE TRIGON.

4724

The equation of this conic (Fig. 49) is

iPy+mya+nafi = 0

or

1+^+^ = 0.

p
y
Peoof.The equation is of the second degree, and it is satisfied by
a = 0, /3 = 0 simultaneously. It therefore passes through the point a/3.
Similarly through fly and ya.
The tangents at A, B, and C are
4726

+-=0,
p
y

*+l = 0,
y
a

i-+ = 0.
a
p

Puoop.By writing (4724) in the form

mya + fi {ly + na) = 0,
ly + na = 0 is seen, by (4097), to be the tangent at ay ; for the intersections
of a and y, with the curve, now coincide, and d (now Jy + ua) passes through
the two coincident points.
4729

The tangent, or polar, of the point afi'y is, by (4659),

(my'+np") a+(na'+ly') $+(l^+ma') y = 0.

4730 The tangents at A, B, G (Fig. 49) meet the opposite

sides respectively in P, Q, B on the right line
* .Ail _ 0
I
m
n
4731

The line ^

By (4604).

passes through (D), the intersection

of the tangents at A and B.

PARTICULAR C0NIC8.

639

4732 The diameter through the intersection of the tan

gents at A and B is
naa nt)/J+(Za mb) y = 0.
Proof.The coordinates of the point of intersection are I : m : n, by
(4726-7), and the coordinates of the centre of AB are 6 : a : 0. The
diameter passes through these points, and its equation is given by (4616).
4733

The coordinates of the centre of the conic are as

/( Za+mh+nf) : m(/a mh+nt) : n (Za+ml) nC).

Proof.By (4610), the point being the intersection of two diameters
like (4732). Otherwise, by (4677).
4734 The secant through (aj/3^!), (a2/32y2), any two points
on the conic, and the tangent at the first point are respectively,
JfL + ^+J9L = 0
PA
yiya

and

i| + ^ + ^ = 0.
i
Pf
y?

Proof.The first is a right line, and it is satisfied by a=al, Ac., and

also by a = a2, &c, by (4725). The second equation is what the first be
comes when a, = a1, &c. For the tangential equation, see (4893).
4735

The conic is a parabola when

Fa'+ni'b'+n'f 2mnfr 2n/ca 2/mafa = 0,

4736

or

y(/a)+v/("b)+y(ni) = 0.

Proof.Substitute the coordinates of the centre (4733) in aa + + cy = 0,

the equation of the line at infinity (4612).
Otherwise, the conic must touch the line at infinity ; therefore put a, (', c
for \,fi, v in (4893).
4737

The conic is a rectangular hyperbola when

I cos.4+m cosU+n cos C = 0,

and in this case it passes through the orthocentre of the

triangle.
Proof.By (4690), and the coordinates of the orthocentre (4634).
THE CIRCUMSCRIBING CIRCLE OF THE TRIGON.
4738

/3y sin A + ya sin B+ afl sin C = 0,

sin^4 . s'mB . sinC

TBILINEAB ANALYTICAL CONIG8.

640

Proof.The values of the ratios I : m : n, in (4724), may be found geo

metrically from the equations of the tangents (4726-8).
For the coordinates of the centre, see (4642).
THE INSCRIBED CONIC OF THE TRIGON.
4739

Fct+rrfF+ny^mnpy^nlya-'Hmap = 0.

4740

or

y/(k) + y/(m(3) + y/(ny) = 0.

Proof.(Fig. 50.) The first equation may be written

ny (ny-2Za-2m/3) + (Za-m/3)1 = 0.
By (4699) this represents a conic of which the lines y and ny 2la m(i are
the tangents at F and/, and lam.(3 the chord of contact. Similarly, it may
be written so as to shew that a and fl touch the conic.
4741

The three pairs of tangents at F, f, &c, are

2m/3+2ny-la)
and a
>'

2n7+2/a-m/3 ?
and
j'

2la+2mp-ny)
and y
)'

and they have their three points of intersection P, Q, B on

the right line la-\-mfi-\-ny.
By (4604).
4742

The coordinates of the centre of the conic are as

wh + mC : Zt+nfl : ma+Zh.

Proof.By putting a and (i = zero alternately in (4739), we find, for

the coordinates of the points of contact,
atD, 0= 2^!L_. andatS, = 2An
ni + mc '
'
na + k'
therefore the equation of the diameter through G bisecting BE is, by (4603),
q
= P
nb + mc k + na
Similarly the diameter bisecting DF is 5L_ = .
ma + lb nb + mc
Therefore the point of intersection, or centre, is defined by the ratios given
above.
Otherwise, by (4677), and the values in (4665), writing for a, b, c,f, g, h
the coefficients in (4739).
4743 The secant through a^y,, a$tyt any two points on
the curve.
a v/Z (\/fr7+ V^y,) +P\/m (\/y1a2+ y/y2o,)
+7>/ (v^+v^A) = 0.

PARTICULAR CONICS.

641

Proof.Put a1/31yi for a/3y, and shew that the expression vanishes by
(4,740).
4744

The tangent at the point a^-/! :

Proof.Put a,= a &c, in (4743), and divide by 2^/(^(3^).

4745 The equation of the polar must be obtained from
(4739) by means of (4659).
4746

The conic is a parabola when

a
b
Proof.Similar to that of (4736).

THE INSCRIBED CIRCLE OP THE TRIGON.

4747

aW^+0w|-+rW c2

2/Sy cos2^cos2-^ 2ya cos2-^ cos2-^ 2a/J cos2^-cos2-^.

4748

or cos^- yo+cos^- v'/S+cos-^ s/y = 0.

4749

The a-escribed circle :

A

/- .

. B

(4629)
/0 i

cos
Proof.At the point of contact where y = 0, we have, in (4740), geo
metrically, r being the radius of the circle,
A sin A : r cot
JB sin B = cos' \A : cos* %B ;
I : m 5= $ : a = r cot
-f- for the inscribed; for the escribed circle and wB instead of B.
4750

The tangent at a'P'J, by (4744), is

A a .
B ft
,
C y
n
cost 7? +cos T VP +cos a- 3t = "

The polar is obtained as in (4745).

GENERAL EQUATION OF THE CIRCLE.
4751

(/a+m+ny) (o sin A

sin B+y sin C)

+k Q3y sin A+ya sin B+afi sin C) = 0.

4 N

642

TRILINEAB ANALYTICAL CONICS.

Proof.The second term is the circumscribing circle (4738), and the

first is linear by (4609) ; therefore the whole represents a circle. By varying
h, a system of circles is obtained whose radical axis (4161) is the line
la+mp + ny, the circumscribing circle being one of the system.

4752 If l'a + m?fi-\-n'y be the radical axis of a second system

of circles represented by a similar equation, the radical axis
of any two circles of the two systems defined by k and k'
will be
tf(/a+m/8+ny) fc(ra+wi'+n'y) = 0.
Pkoof.By eliminating the term
0y sin A + ya sin B + a/3 sin ft

4753 To find the coefficient of s+y* in the circle when

only the trilinear equation is given.
Rule.Make a, /3, y the coordinates of a point from which
the length of the tangent is known, and divide by the square of
that length ; or, if the point be within the circle, substitute
" half the shortest chord through the point" for "the tangent."
Pkoof.If S = 0 be the equation of the circle, and m the required co
efficient ; then, for a point not on the curve, S -j- m = square of tangent or
semi-chord, by (41 60).
THE NINE-POINT CIRCLE.

4754

a2 sin 2 A +2 sin 2/i+y- sin2C

2 (0y sin^+ya sin B+afi sin C) = 0.

PARTICULAR CONICS.

643

Proof.The equation represents a circle because it may be expressed in

the form
(a cos.il-f-/JcosB+y cos C) (a smA+(S sinB + y sin 0))
2 (/3y sin^ + ya sinB + a/3 sin C) = 0.
See Proof of (4751). Now, when a = 0, the equation becomes
(0 sinB-y sin 0)(/3 cosB-y cosC) = 0,
which shews, by (4631, '3), that the circle bisects BC and passes through D,
the foot of the perpendicular from A.
4754a The equation of the nine-point circle in Cartesian
coordinates, with the side BG and perpendicular on it from A
for * and y axes respectively, is
a?+y*-R sin (B-G) x-B cos (B-G) y = 0,
where B is the radius of the circum-circle.

THE TRIPLICATE- RATIO CIRCLE.

47546 *Let the point S (Fig. 165) be chosen, so that its
trilinear coordinates are proportional to the sides of the trigon.
Draw lines through 8 parallel to the sides, then the circle in
question passes through the six points of intersection, and the
intercepted chords are in the triplicate-ratio of the sides.
[The following abbreviations are used, a, b, e, and not a, 6, c, being in
this article written for the sides of the trigon ABC.~\
KEEa' + b' + t?; X = /(W+A'+oV) ; A = ABC;
fi = ~;
By hypothesis,

-> = BFD = DE'F', &c. ;

=|=X=

6 = DFD' = DE'D', &c.

(4007) = f

(1),

BIX _aL_cL m _ BF
~BF
c
y K '
BD'
therefore BF.BF' = BD.BI/, therefore F, F', D, D' are concyclic.
If AS, BS, GS produced meet the opposite sides in I, m, n,
Bn _ o sin BCn _ aa _ a1 , ,-. *
IH - b sin ACn -bp- V 7 { }

/Q,
( ;<

The theorems of (1 to 36) are for the most part due to Mr. R. Tucker, M.A.
The original articles will be found in The Quarterly Journal of Pure and Applied Mathematics,
Vol. xix., No. 76, and Vol. xx., Nos. 77 and 78.
Other and similar investigations have been made by MM. Lcmoinc and Taylor and
Prof. Neuberg, Mattetia, 1881, 1882, 1884.

644

TRILINEAR ANALYTICAL CONICS.

SF = BB = -3- = T'2cA (1) = 2r. Similarly BF' =~, Ac. ... (3).
smB
KsinB w
K
J
K
v }
m = BP^A = ca? a=a>&c
sin C
K c
K
BD' = BD + DD' = a(c>+a')| &c

FD = v/(BD,+J,,-2D.5J,cos5) = ^, by (2) and (5)

XL

(4
..(5).

(6).

Hence D27F and D'E'F' are triangles similar to -42?(7, and they are equal
to each other because ESF = E'SF = E'SF', Ac. (Euc. L 37.)

Hence

BF' = J(BLa+BF"t-2BB.BF cosB) =^

JX.
BF' = FE' = EB'.
B'F =*_BB'= h(c' + a') &o
a
BF'+FB'-BB2 o'+c' + ft1
.
K
C0SU=
2BF.FB
=-2T~ (5&6)=2X
rin- = V(1-S)=TL(708)

(7).

(8).
/m
(9)'
(10)-

j <>),
\ &c.
f = <*2 cos ^4 + be
cos o0 = cos /{A
(11)A
AFE' + BBF' + CEB' = AF AE' B'mA + Ac. = /i'A = BEF, by (6) ... (12).
Or, geometrically, by Euclid I. 37.
Radius of T. R. circle,

p = pR, by (6) (E = circum -radius)

(13).

The trilinear equation of the T. R. circle is

ale (a' -f/32 + y2) = ^ (an + 6/3 + cy)2 + a*/Jy + &'ya + A/3
or

(&, + 6V+(6,+a1)/?+(ai+6,)y, = {
+ j (V+c*Kb>+al)+cW)
v
I ca

(14),

(a'+c*) +6V) &

\ (ct + ai)(c'+V) + a1b1) ^ ... (15).

1
' ao

Obtained by substituting the trilinear coordinates of B, E, F, through

which points the circle passes, in (4751), to determine the ratios I : m : n
and h. The coordinates of B are
n a(a? + tf) sin C ac* sinB
K
'
E
Similarly those of 17 and F.

PARTICULAR GONIGS.

645

THE SEVEN-POINT CIRCLE *

4754c Let lines be drawn through A, B, G (Fig. 165)
parallel to the sides of the triangles DEF, I/E'F', as in the
figure, intersecting each other in P, P, L, M, N. Let Q be
the circum-centre ; then the seven points P, P', L, M, N, Q, 8
all lie on the circumference of a circle concentric with the
T. R. circle.
(16)
The proof depends on Euclid III. 21, and the similar triangles DEF,
D'E'E".
The radius p' of the seven-point circle is
P
obtained from

1-3 tan2 u

1(18),

p'2 = p2 + SDP-lpSD cos (B- TDD') .

Expand and substitute cos TDD' =

=
by
2p
ZpK
m-r>T\>
o/1n
t> c2 + a2 6' . -r,
Bin TDD = cos 0 (11), cosi? =
, sinP =
2ca
3p + p' = E2, by (17) and (13) ;

(3) and (5),

2A
,
62 + cs a2
, cos A -
.
ao
2bo

by (17) and (9) ... { gjj>'

The trilinear equation of the seven-point circle is

abcia' + F + y1) = (tPy + Vya + faP
or

afiy+bya + eafi = ^(bca + ca(i + aby)(aa + bl3 + ey)

(21),
(22).

If the coordinates of P are a /3 y and those of P" a'lt fi'u y[ ; then

,,; = a/3; = Yly[
The equation of STQ is, by (4615),
a sin (B-0)+(3 sin (C-A) + y sin (^-P)
And the equation of PP' is
(a-6V) + 4 (&4- eV) + - (c4-a26s) = 0
aba

(23).
(24).
(25).

The point # has been called the Symmedian point of the

triangle. It has also this property. The line joining the mid This circle was discovered by M. H. Brocard, and has been callod " Tho Brocard Circle,"
tho points 1', f being called the Brocard points.

646

TMLINEAR ANALYTICAL GONIGS.

point of any side to the mid-point of the perpendicular on that

side passes through S.
Proof.Let X, Y, Z (Fig. 166) be the feet of the perpendiculars ; x, y, z
the mid-points of the same, and X', Y', Z' the mid-points of the sides. Now
the trilinear coordinates of X', S, and x in order are proportional to
0, c,
b |
This determinant vanishes ;
therefore the three points are on
a, b,
c
the same right line, by (4615).
1, cos G, cosB
That the three lines X'x, Y'y, Z'z are concurrent appears at once by (970),
since CX=z 2Y'x, &c.
The Symmedian point may also be defined as the intersec
tion of the three lines drawn from A, B, G to the corresponding
vertices of the triangle formed by tangents to the circumcircle at A, B, G.
Let Ba, (7/3, Ay be taken = CX, AY, BZ respectively.
Then Aa, B(3, Gy meet in a point 2, by (976), and this point
by similarity of figure is the Symmedian point of the triangle
formed by lines through A, B, C parallel to the sides BC,
CA, AB.
If the sides of X'Y'Z' be bisected, similar reasoning shews
that <r, the Symmedian point of the triangle X'Y'Z', lies
on #2.
It can also be shewn that, if A'B'G' be any triangle having
its sides parallel to those of ABC and its vertices on SA, SB,
SG, the sides of the two triangles intersect in six points on a
circle whose centre lies midway between the circum- centres
of the same triaugles. When A'B'G' shrinks to the point S,
the circle becomes the T. R. circle.
A more general theorem respecting the triangle and circle
is the following
Take ABC any triangle, and let BUEE'FF' be the points in order, in
which any circle cuts the sides.
Let
BB = pc, CE qa, AF = rb ")
f6->
CD' = p'b, AE' = q'c, BF' = r'a )
1 h
From BD.BD' = BF. BF', &c, Euclid III. 35, we can write three equations
which are satisfied by the values
p = r = tac,
1= P tab,
r = q' = tbc
(27),
and from these equations it appears that
BF=<jc; D'F' = aa, &c., where <r = J(f\i-tK+ 1)
so that DBF and B'E'F' are both similar to ABC.

(28),

SELF-CONJUGATE TRIANGLE.

647

Also DF' = talc, therefore DF = FE' = ED'

From sin BFD = *ac Sln we can obtain
a
cot BFD = cot <t = =f
4tA
The radius of the circle
= <rE
and the coordinates of its centre are
a = B J cos A + <(ga'~^~fc<~c<) J . Similarly /3 and y

(32).

The equation of the circle is

a/3y + bya + caj3 = t (aa + 6/3 + cy) { abc (1 - ta2) + &c. ]

(33),

or

(29).

(30).
(31),

tabo { a1 ( 1 - ta1) + /32 (1 - tb*) + y' (1 - *c2) }

= a/3y { (l-fPXl-t/HAV) + &c

(34).

When < = 0, <r = 1 and the circle is the circum-circle

(35).

When tK= 1, <r = \ =-^r and the circle is the T. R. circle

(36).

CONIC AND SELF-CONJUGATE TRIANGLE.

When the sides of the trigon are the polars of the opposite
vertices, the general equation of the conic takes the form
4755

ZV+m^-ny = 0.

Proof. (Fig. 51.) The equation may be written in any one of the
three ways,
Pa' = (ny + wj3)(y-m/3),
m*& = (ny + la)(ny-la),
n'y' = (la + imP)(laimP).
Hence, by (4699), a or BC is the chord of contact of the tangents
nyml3(AQ, AS) drawn from A, and /3 is the chord of contact of the
tangents ny la (BE, BP) drawn from B. Hence a, /3 are the polars of A, B
respectively; and therefore y or AB is the polar of C (4130). Also y may
be considered to be the chord of contact of the imaginary tangents Zaim/3
drawn from 0.
4756 If the points of intersection of a and /3 with the conic
be joined, the equations of the sides of the quadrilateral so
formed are
QR,

la+?nB+ny = 0,

SP, la+mfi-ny = 0,

PQ, /a+//8+y = 0,

RS, la-m$+ny = 0.

Hence QR, SP and PQ, RS intersect on the line y in A'

and R'.

648

TBILINEAB ANALYTICAL CONICS.

4757 Each pencil of four lines in the diagram is a harmonic

pencil, by the test in (4649).
4758 The triangle A'B'C is also self-conjugate with regard
to the conic.
Proof.The equations of its sides Clf, CA', A'B' are
la mB = 0,
la + mfi = 0, y = 0.
Denote these by a, B', y, and put a, B in (4755) in terms of a', B". The
equation referred to A'B'C thus becomes a'2 + B^ 2n'y* = 0, which is of the
same form as (4755).
4759 It is clear that the triangles AQS and BPB, formed
by a pair of tangents and the chord of contact in each case,
are also self-conjugate.
4760 Taking A'B'C for the trigon, and denoting the sides
by a, /3, y, the equations of the sides BS, PQ, QB, SP of the
quadrilateral become respectively
nyla = 0,

mfi i ny = 0.

Ex.As an example of (4611), we may find the coordinates of P from

the equations
Ua + hf}+ ty = 2^
fa = 2Zm-i- (amn + fcZ + cZm)
0 + mB ny = 0 / from which < B = 2,mn-i-(amn + bnl + ilm')
la+ 0 +ny = 0)
ly= 2nl-h(amn + toil + tlm).
To obtain the coordinates of Q, B, and S, change the signs of m, n, and I
respectively.
ON LINES PASSING THROUGH IMAGINARY POINTS.
4761 Lemma I. The right line passing through two con
jugate imaginary points is real, and is identical with the line
passing through the points obtained by substituting unity for
V 1 in the given coordinates.
Proof.Let (a + ia, b + ib') be one of the imaginary points, and therefore
(a ia, bib') the conjugate point. The equation of the line passing through
them is, by (4083) and reducing, b'x a'y + a'bab' 0, which is real.
But this is also the line obtained by taking for the coordinates of the
points (a + a', 6 + 6') and (aa', 6 6').
Lemma II.If P, S and Q, B are two pairs of conjugate
imaginary points, the lines PS and QB are real, as has just
been shown, and, therefore, also their point of intersection is

SELF-CONJUGATE TRIANGLE.

649

real. The other pairs of lines PQ, BS and PR, QS are

imaginary. But the points of intersection of each pair are
real, and are identical with the points which are obtained by
substituting unity for y/ 1 in the given coordinates, and
drawing the six lines accordingly.
Proof.Let the coordinates of the four points be as under
P

a + ia,

b + ib',

a + ia,

ft + ift',

aia',

b ib',

aia',

ft ift'.

The equations of PR and QS, by (4083), are L + iM and LiM, where

L = (bft) x (a a) y + aft ab +a'ft' a'b',
M= (b' + ft')x-(a'-a')y + a'ft-a'b-aft' +ab'.
Now the lines Zrt'lf=0 intersect in the same real point as the lines
L M = 0, because the values L = 0, M = 0 satisfy both equations
simultaneously. Hence, to determine this point, we have only to take i as
unity in the given coordinates.

Lemma III.If P, S are real points, and Q, B a pair of

conjugate imaginary points, the lines PS and QB are both
real, by Lemma I., and consequently their point of intersec
tion is real. The remaining pairs of lines PQ, BS and PB, QS
and their points of intersection are all imaginary. But the
line joining these two imaginary points of intersection is real,
and is identical with the line obtained by substituting unity
for s/ 1 in the given coordinates and drawing the six lines
accordingly.
Proof.Let the coordinates of the four points be as under
. . P

i2/

a + ia,

ft + ift',

a,y

aia',

ft iff.

Since the coordinates of R are obtained from those of Q by merely changing

the sign of i, the equations of the four imaginary lines will take the forms
...

PQ
PR

A-iB,
A + iB,

SQ
SR

0-%D,
O+iD.

Now let the coordinates of the point of intersection of PQ and SR be

L + iM, L' + iM ', then will L iM, L' iM' be the coordinates of the intersec
tion of PR and SQ, for the equations of this pair of lines are got from those
of PQ and SR by merely changing the sign of t. The points of intersection
are therefore conjugate imaginary points, and the line joining them is real,
by Lemma I. Also, since that line is obtained by writing 1 for i in the co
ordinates of those points, it will also be obtained by writing 1 for i in the
original coordinates of Q and R and constructing the figure as before.

4 0

TBILINEAB ANALYTICAL CONICS.

650

4762

To find a common pole and polar of two given conies :

(i.) If the conies intersect in four real points P, Q, R, S,

construct the complete quadrilateral (4652).
Then A'RC
(Fig. 51) is a self-conjugate triangle for each conic, by
(4758), and therefore each vertex and the opposite side form
a common pole and polar to the conies.
(ii.) If the conies do not intersect at all in real points, the
triangle A'WG is still real, by Lemma II. (4761), and can be
constructed in the manner shown.
(iii.) If two of the points (P, S) are real, and two (Q, R)
imaginary, then, by Lemma III., the vertex A' and the side
&C are real, and may be constructed, and they form a common
pole and polar of the given conies.
Returning to the triangle of reference ABC,
4763 Let la = ny cos <j>, w/3 =
joining two points fa, fa is

sin </> ;

la cos i (<i+<k) +m& sin \

then the chord

= ny cos \ (a\ &),

and therefore the tangent at the point ^ is

4764

la cos <'+i/3 sin <j> = ny.

4765 Putting P = L, m2 = M, n*=-N, the conic (4755)

becomes
La'+Mp+Ny* = 0
(1).
4766

The tangent or polar of a'p'y is

Laa'+Mpp+Nyy' = 0

4767

Hence the pole of Xa+ft^S-fvy = 0

(t**)

4768

(2).

The tangential equation is

X*

u*

l*2

r+&+=

<4>-

and this is the condition that the conic (1) may be touched
by the four lines
\anpvy = 0.

SELF-CONJUGATE TRIANGLE.

4769

651

In like manner,
LeP+MF+fy* = 0

(5)

is the condition that (1) may pass through the four points
(a, pt y').
4770 The locus of the pole of the line Xa+/i/?+vy with
respect to such conies is

a + fi ^ y
Proof.By (3), if (a, /3, y) be the pole, a = j- &c, .-. = ~, in (5),
the equation of condition.
4771 The locus of the pole of the line la-\-m$-\-ny, with
respect to the conies which touch the four lines \aiil3vy
is

+ -4
I
m

'- = 0.
n

Proof.By (3), if (a, (i, y) be the pole, a =

(4), the equation of condition.

&e., .". L = , 4c., in

4772 The locus of the centre of the conic is given in each

case (4770, '1) by taking the line at infinity
a sin.4+/3 sin_B-|-y sin C
for the fixed line, since its pole is the centre.
4773 Thus the locus of the centre of the conic passing
through the four points (a'/8'y/) is
a,2sin^ , ff2sin , y**inC__n
a
+

+
y
4774

The coordinates of the centre of the conic (1) are

given ,by

La
Ny
= Mfi = -J.

Proof.Let the conic cut the side a in the points (0/3,7,), (fy3jya)- The
right line from A bisecting the chord will pass through the centre of the
conic, and its equation will be /? : y = /3, + /?3 : yi + yj. Now /81+/32 is the
sum of the roots of the quadratic in B obtained by eliminating y and a from
the equations La' + MB' + Ny* = 0, a 0, and aa + fy3 + Cy = 2. Similarly
for y, + y, eliminate a and /8. The equation of the diameter through A being
found, those through B and O are symmetrical with it.

652

TKILINEAR ANALYTICAL GONICS.

4775

The condition that the conic (1) may be a parabola is

Proof.This is, by (4), the condition of touching the line at infinity

aa+fy? + cy = 0.
4776 The condition that (1) may be a rectangular hyperbola
is L+M-\-N = 0, and in this case the curve passes through
the centres of the inscribed and escribed circles of the trigon.
Peoof.By (4690), (a, b, c are now L, M, N).
a = /3 = y, the four centres in question.

4777

(1) is now satisfied by

Circle referred to a self-conjugate triangle :

o2 sin 2A +/8* sin 2#+y* sin 2C = 0.

Peoof.The line joining A to the centre is

= (4774). Therefore
M
N
be
;bcosU = ccosC
-, the condition of rperpendicularity
r
J to a byJV(4622).
' SimilarlyJ
N = L -, therefore (1) takes the form above.
c cos C a cos A

IMPORTANT THEOREMS.

CARNOT'S THEOREM.
4778 If A, B, C (Fig. 52) are the angles of a triangle, and
if the opposite sides intersect a conic in the pairs of points
a, a' ; b, b' ; c, c ; then
Ac.Ac'.Ba.Ba'.Cb.CV = Ab.Ab'.Bc.Bc'.Ca.Ca'.
Proof.Let a, /3, y be the semi-diameters parallel to BC, GA, AB ; then,
by (4317), Ab . Ab' : Ac : Ac = /3* :
Compound this with two similar
ratios.
4779

Cob.If the conic touches the sides in a, b, c, then

A<*.Bu\ Cb2 = Ab2. B<*. Ca\

THEOREMS.

653

4780 The reciprocal of Carnot's theorem is : If A, B, G

(Fig. 52) are the sides of a triangle, and if pairs of tangents
from the opposite angles are a, a' ; b, b' ; c, c' ; then
sin (.4c) sin (Ac') sin (Ba) sin (Ba) sin (Cb) sin (Cb')
= sin (Ab) sin (Ab') sin (Be) sin (Be') sin (Ca) sin (Ca')5
where (Ac) signifies the angle between the lines A and c.
Proof.Reciprocating the former figure with respect to any origin 0,
let A, B, G (i.e., BQ, QP, PB) be the polars of the vertices A, B, G. Then,
by (4130), Q, B will be the poles of AB, AG; and 6, V , the polars of the
points 6, V, will intersect in B and touch the reciprocal conic. Similarly, c, c'
will intersect in Q. A, V are perpendicular to OA, Ob', and therefore
Z Ab' = i. AOb', and so of the rest.

PASCAL'S THEOREM.

4781 The opposite sides of a hexagon inscribed to a conic

meet in three points on the same right fine.
Pboop. (Fig. 53.) Let a, fi, y, y, ($', a be the consecutive sides of the
hexagon, and let u be the diagonal joining the points aa and yy'. The
equation of the conic is either ay kfiu = 0 or a y'k'(Z'u = 0, and, since
these expressions vanish for all points on the curve, we must have ayk(iu
= ay' k'fi'u for any values of the coordinates.
Therefore ay ay
= u (i/3 k'fi"). Therefore the lines a, a' and also y, y' meet on the line
kfik'fi'; and /3, j3' evidently meet on that line.
Otherwise, by projecting a hexagon inscribed in a circle with its opposite
sides parallel upon any plane not parallel to that of the circle. The line at
infinity, in which the pairs of parallel sides meet, becomes a line in which
the corresponding sides of a hexagon inscribed in a conic meet at a finite
distance (1075 et seq.).

4782 With the same vertices there are sixty different

hexagons inscribable in any conic, and therefore sixty dif
ferent Pascal fines corresponding to any six points on a conic.
Peoop.Half the number of ways of taking in order five vertices B, G,
D, E, F after A is the number of different hexagons that can be drawn, and
the demonstration in (4781) applies equally to all.

BRIANCHON'S THEOREM.

4783 The three diagonals of a hexagon circumscribed to a

conic pass through the same point (Fig. 54).

654

TRILINEAR ANALYTICAL CONIOS.

Proof.Let the three conies S + L1, S + AP, 8+N\ in (4707), become

three pairs of right lines, then the three lines L M, M N, NL become
the diagonals of a circumscribing hexagon.
Pascal's and Brianchon's theorems may be obtained, the one from the
other, by reciprocation (4840).

THE CONIC REFERRED TO TWO TANGENTS AND

THE CHORD OF CONTACT.
Let L = 0, M = 0, B = 0 (Fig. 55) be the sides of the
trigon ; L, M being tangents and B the chord of contact.
4784

The equation of the conic is LM = R3.

4785

The lines AP, BP, and CP are respectively

IlL = R,

pR= M,

fSL = M.

(4699)

[By (4604).

Since the point P on the curve is determined by the value

of n, it is convenient to call it the point p.
4788 The points p and p (P and Q) are both on the line
ft*L = M drawn through G.
4789

The secant through the points

p' (P, F) is

lifi'L- (fi+p') R+M = 0.

Proof. Write it fi{n'LR) (ix'RM), and, by (4604), it passes
through the point ft'. Similarly through fi. Otherwise, determine the co
ordinates of the intersection of pLR and fiRAT, and of (t'L R and
fi'RM by (4610), and the equation of the secant by (4016).
4790 Cos.The tangents at the points fa and p (P, Q)
are therefore
!i2LT2nR+M=0.
4791

These tangents intersect on B.

[Proof by subtraction.

4792 Theorem.If the equation of a right line contains an

indeterminate n in the second degree, it may be written as
above, and the line must therefore touch the conic LM = R*.
4793

The polar of the point (L', M', B') is

LM'-2RR'+L'M=0.

THE GONIC LM = R.

655

Proof. For p+p' and pit', in (4789), put the values of the sum and
product of the roots of p2L'2pB' + M' = 0 (4790).

4794 Similarly the polar of the point of intersection of

aLB and bB M is
abL2aR+M=0.
4795 The line GE joining the vertex G to the intersection
of two tangents at p and p', or at p and p, is
Wi'L-M=0.
Otherwise, if two tangents meet on any line aLM, drawn
through C, the product of their p's is equal to a.
Proof. Eliminate B from the equations of the two tangents (4790).

4796 The chords PQ', FQ and the line GE all intersect in

the same point on B.
Proof.The equations of PQ', P'Q are, by (4789),

pp'L (p -p) B-M = 0,

and, by addition and subtraction, we obtain pp'L M = 0 (4795), or i? = 0.

4797 The lines pp'L + M (GD) and B intersect on the chord

PP/ which joins the points p, p' ; orThe extremities of any
chord passing through the intersection of aL + M and B have
the product of their pB equal to a.
4798 The chord joining the points p tan <j>, p cot <f> touches a
conic having the same tangents L, M and chord of contact B.
Proof.The equation of the chord is, by (4789),
p'L pB (ten <p + cot <j>)+M = 0,
and this touches the conic LM 6in'2<> = li" at the point p, by (4792).

4799 The tangents at the points p tan $, p cot <j> intersect

on the conic LM = B* sinE 2$.
Proof.Write the equations of the two tangents, by (4790), and then
eliminate p.
4800 Ex. 1. To find the locus of the vertex of a triangle circumscribing
a fixed conic and having its other vertices on two fixed right lines.
Take LM = B* for the conic (Fig. 56), aL+M, bL+M for the lines GD,
GE. Let one tangent, BE, touch at the point p ; then, by (4795), the others,

656

TBILINEAB ANALYTICAL GONICS.

PD, PE, will touch at the points , , and therefore, by (4790), their
equations will be
f- ^
^L-^B + M,

^L-^B+M.

Eliminate /j, and the locus of P is found to be (a + 6)JLlf = 4a6E2.

[Salmon, Art. 272.
4801 Ex. 2.To find the envelope of the base of a triangle inscribed in
a conic, and whose sides pass through fixed points P, Q.
(Fig.57.) Take the line through P, Q for B; LM-B' for the conic ; aLM,
bL M for the lines joining P and Q to the vertex G. Let the sides through P
and Q meet in the point on the conic ; then, by (4797), the other extremi
ties will be at the points and , and therefore, by (4789), the
M
I1
equation of the base will be abL + (a + b) fiB + ^M = 0. By (4792), this
line always touches the conic 4o6 LM = (a + b)1 R1.
[Ibid.
4802 Ex. 3.To inscribe a triangle in a conic so that its sides may pass
through three fixed points. (See also 4823.)
We have to make the base abL + (a + b) fiR+^M (4801) pass through
a third fixed point. Let this point be given by cL = R, dB = M. Elimi
nating L, it, B, we get ab+ (a + b) pc+n'cd = 0, and since, at the point fi,
l*L = B, p'L = M, that point must be on the line abL + (a+b) cR + cdM.
The intersections of this line with the conic give two solutions by two posi
tions of the vertex.
[_Ibid.

RELATED CONICS.
4803 A conic having double contact with the conies 8 and
S' (Fig. 58) is
fi3E2 -2n(S+S')+Ft = 0,
where E, F are common chords of S and S', so that
8-S' = EF.
Proof.The equation may be written in either of the ways
(pE+F)1 = 4/iS or (jiE-FY = 4^5',
showing that fiE F are the chords of contact AB, CD. There are three
such systems, since there are three pairs of common chords.
4804 Cob. 1.A conic touching four given lines A,B, C, D,
the diagonals being E, F (Fig. 59) :
ft22-2/* (AC+BD)+Fi = 0.
Here S = AC and S' = BD, two pairs of right lines.

ANHABMONlG PENCILS.

657

Otherwise, if L, M, Nbe the diagonals and L + M+N the

sides, the conic becomes
4805

fiW-n-ilS+AP-^+AP = 0.

For this always touches

(Li+M*-N1y4JJ'AP or (L+M+N)(M+N-L)(N + L-M)(L + 1I-N).
[Salmon, Art. 287.]
4806 Coe. 2. A conic having double contact with two
circles G, C is
l**-2n (C+C')+{C-Cy = 0.
4807

The chords of contact become

fi+C-C = 0

4808

and

f*-C+C' = 0.

The equation may also be written

which signifies that the sum or difference of the tangents

drawn from any point on the conic to the circles is constant.

ANHARMONIC PENCILS OF CONICS.

4809 The anharmonic ratio of the pencil drawn from any

point on a conic through four fixed points upon it is constant.
Proof.Let the vertices of the quadrilateral in Fig. (38) be denoted by
A, B, C, D, and let P be the fifth point. Multiplying the equation of the
conic (4697) by the constants AB, CD, BC, DA, wo have
j AB.CD _ ABa . OBy _ PA . PB sin ABB . PC . PD sin GPP
BG.BA
BOp.DAi
PB.PC sin BPO.PB .PA sin BPA
_ sin APB. am GPB
~ sin BPG. Bin BPA'
Compare (1056).
4810 If the fifth point be taken for origin in the system
(4784, Fig. 55), and if the four lines through it be
LfaR,

LiiiR, LfiiR,
4 p

L^R,

658

TBILINEAB ANALYTICAL CONICS.

the an harmonic ratio of the pencil is, by (4650),

f*4)(/*s IhY
4811 Coe. 1.If four lines through any point, taken for
the vertex LM, meet the conic in the points nu fit, p3, ^ the
anharmonic ratio of these points, with any fifth point on the
conic, is equal to that of the points filt fi2, ns,
in
which the same lines again meet the conic.
4812 Cos. 2. The reciprocal theorem is If from four
points upon any right line four tangents be drawn to a conic,
the anharmonic ratio of the points of section with any fifth
tangent is equal to the corresponding ratio for the other four
tangents from the same points.
4813 The anharmonic ratio of the segments of any tangent
to a conic made by four fixed tangents is constant.
Proof.Let ft, fix, fJt, /is, fit (Fig. 60)
anharmonic ratio of the segments is the
lines from LM to the points of section ;
HfisLM, ffjj- M, a pencil homographic

be the points of contact. The

same as that of the pencil of four
that is, of fj^LM, fi^LM,
(4651) with that in (4810).

4814 If P, P' are the polars of a point with respect to the

conies 8, S', then P + kP will be the polar of the same point
with respect to the conic S+kS'.
4815 Hence the polar of a given point with regard to a
conic passing through four given points (the intersections of
8 and S') always passes through a fixed point, by (4101).
If Q, Q' are the polars of another point with respect to
the same conies, Q + JcQ is the polar with respect to 8+kS'.
4816 Hence the polars of two points with regard to a
system of conies through four points form two homographic
pencils (4651).
4817 The locus of intersections of corresponding lines of
two homographic pencils having fixed vertices (Fig. 61) is a
conic passing through the vertices ; and, conversely, if the
conic be given, the pencils will be homographic.
Proof.For eliminating k from P + kP'= 0, Q+kQk', we get FQ'= PQ

CONSTRUCTION OF CONICS.

659

4818 Coe.The locus of the pole of the line joining the two
points in (4816) is a conic.
Proof.For the pole is the intersection of P+kP' and Q + JcQ'.
4819 The right lines joining corresponding points AA', &c.
(Fig. 62) of two homographic systems of points lying on two
right lines, envelope a conic.
Proof.This is the reciprocal theorem to (4817) ; or it follows from
(4813).

4820 If two conies have double

harmonic ratio of the points of
four tangents to the inner conic
set of four points (a, b, c, d) or
tangents meet the other conic.

contact (Fig. 63), the ancontact A, B, G, D of any

is the same as that of each
(a', b', c', d') in which the

Proof.By (4798). The ft's for the points on the latter conic will be
equal to the /i's of the points of contact multiplied by tau <p for one set, and
by cot^ for the other, and therefore the ratio (4810) will be unaltered.
4821 Conversely, if three chords of a conic aaf, bb', cc' be
fixed, and a fourth Amoves so that {a&cd} = {a'b'c'd'}, then
dd' envelopes a conic having double contact with the given
one.
For theorems on a right line cut in involution by a conic, see (4824-8).

CONSTRUCTION OF CONICS.

THEOREMS AND PROBLEMS.

4822 H a polygon inscribed to a conic (Fig. 64) has all its
sides but one passing through fixed points A, B, ... Y, the
remaining side az will envelope a conic having double contact
with the given one.
Proof.Let a, b, ... z be the vertices of the polygon, and a, a, a", a"
four successive positions of a. Then, by (4811),
{ a, a', a", a'" } = { b, b', b", b'" j = &c. = { z, z, z", z'" ) .
Therefore, by (4821), the side az envelopes a conic, &c.

660

TBILINEAB ANALYTICAL G0NIG8.

4823 Poncelet's construction for inscribing in a conic a

polygon having its n sides passing through n given points.
Inscribe three polygons, each of n + 1 sides, so that n of
each may pass through the fixed points, and let the remaining
sides fee aV, a"z", a"V", denoted in figure (65) by AD,CF,EB.
Let MLN, the line joining the intersections of opposite sides of
the hexagon ABCDEF (4781), meet the conic in K; then K
will be a vertex of the required polygon.
Proof. [B.KAGE J = iA.KBFBS, each pencil passing through
K, P, N, L; therefore the anharmonic ratio { KAGE } = { KDFB \ for any
vertex on the conic, by (4809) ; i.e., { Ka'a'a" } = { Kz'z'z" j . But, if az
be the remaining side of a fonrth polygon inscribed like the others, we have
by (4811), as in (4822), { aaa"a" } = { zz'z'z" } . Hence K is the point
where a and z coincide.

4824 Lemma.A system of conies passing through four

fixed points meets any transversal in a system of points in
involution (1066).
Proof.Let w, u be two conies passing through the four points ; then
u+ku will be any other. Take the transversal for a! axis, and put y = 0 in
each conic, and let their equations thus become az? + 2gx + c = 0 and
a'aP + 'Zg'x+c" = 0. These determine the points where the transversal
meets u and u. It will then meet u + ku in two points given by
ax' + Zgz + c+k (oV + 2g'x + c) ='0, and these points are in involution with
the former, by (1065).
Geometrically (Fig. 66),
{a.AdbA') = [e.AdbA') (4809),
therefore { AGBA' } = { AB'G'A' } = { A'G'B'A } , therefore by (1069).

4825 Cor. 1.One of the conies of the system resolves

itself into the two diagonals ac, bd. Hence the points B, B',
G, C are in involution with D, H, where the transversal cuts
the diagonals.
4826 Cob. 2.A transversal meets a conic and two tan
gents in four points in involution, so as to meet the chord of
contact in one of the foci of the system.
For, in (Fig. 66), if b coincides with c, and a with d, the
transversal meets the tangents in G, G', while B,B' , D, I/, all
coincide in F (Fig. 67), one of the foci on the chord of
contact.

661

CONSTRUCTION OF CONICS.

4827 The reciprocal theorem to (4824) isPairs of tangents

from any point to a system of conies touching four fixed fines,
form a system in involution (4850).
4828 The condition that \x+ny-\-vz may be cut in involu
tion by three conies is the vanishing of the determinant
i
('2

4
A.2
A,

B,
H*
Ha

Bx
Ih
Ba

X
0
0

ci

2/j

2?!

2/,
4/s
0

2<7,

c*
cs
0

0
A

h
h
h

2^3

2ftt
2/>2
2/*3

P
A
0

where Au Hu Bx belong to the first conic and have the values

in (4988).
Proof.The quadratic Alx1 + 2Exy + B1yi = 0, obtained in (4987), deter
mines the pair of points of intersection with the first conic. The similar
eqnation for the third conic will have Aa = Ax + \Ait <fcc, if the points are
all in involution (1065). The third eqnation is therefore derived from the
other two; therefore the determinant vanishes, by (583).
By expanding and dividing by *', the second determinant above of the
sixth order is obtained.

Newton's Method of Generating a Conic.

4829 Two constant angles aPb, aQb (Fig. 68) move about
fixed vertices P, Q. If a moves on a fixed right line, b de
scribes a conic which passes through P and Q.
Proof.Taking four positions of a, we have (see 1054),
{P.bb'V'b"') = {p.oo'oV'j = {Q.aa'a"a'"} = {Q.bb'b"b'"} .
Therefore, by (4817), the locus of 6 is a conic.

Maclaurin's Method of Generating a Conic.

4830 The vertex V of a triangle (Fig. 69), whose sides pass
through fixed points A, B, C, and whose base angles move on
fixed fines Oa, Ob, describes a conic passing through A and B.
Proof.The pencils of lines through A and B in the figure are both
homographic with the pencil through O, and are therefore homographic with
each other. Therefore the locus of V is a conic, by (4817).

f^vA'i.i uxy)

662

TEILINEAB ANALYTICAL CONICS.

Otherwise, let a, /3, y be the sides of ABC ; la + nifi+ny, l'a+m'fl+n'y

the fixed lines Oa, Ob ; and a = fifi the moving base ab.
Then the equations of the sides will be
(lfi + m) ft + ny = 0, (l'fi + m) a + n'py = 0.
Eliminate /i ; then Zm'a/5 = (??!/3 + ny)(ra+n'y), the conic in question, by
(4697).

4831 Given five points, to find geometrically any number of

points on the circumscribing conic, and to find the centre.
Let A, B, C, D, B (Fig. 70) be the five points. Draio any
line through A meeting CD in P. Draio PQ through the inter
section of AB and DE meeting BC in Q; then QE will meet
PA in F, a sixth point on the curve, as is evident from Pascal's
theorem (4781).
To find the centre, choose AP in the above construction
parallel to CD, and find two diameters, as in (1252).
4832 To find the points of contact of a conic with five right
lines.
Let ABCDE {Fig. 71) be the pentagon, Join D to the
intersection of AC and BE. This line will pass through the
point of contact of AB, and so on.
Proof.By (4783), supposing two sides of the hexagon to become one
straight line.

4833 To describe a conic, given four points upon it and a

tangent.
Let a, a', b, b' (exterior letters in Fig. 52) be the four
points. Then, if AB is a tangent, c, c' coincide, and Garnot's
theorem (4778) gives the ratio Ac2 : Be2. Then by (4831).
Since there are two values of this ratio, + (Ac : Be), two
conies may be drawn as required,
4834 To describe a conic, given four tangents and a point.
Let a, a', b, b' (interior letters in Fig. 52) be the four
tangents. Then, if Q be the given point on the curve, the lines
c, c' must coincide in direction, and (4780) gives the ratio
sin2 (Ac) : sin2 (Be), by which the direction of a fifth tangent
through Q is determined, Then by (4832). The two values
+ (sin Ac : sin Be) furnish two solutions.
Otherwise by (4804), determining fx by the coordinates of
the given point.

CONSTRUCTION OF CONICS.

663

4835 To describe a conic, given three points and two

tangents.
Let A, A', A" be the points (Fig. 67, supplying obvious
letters). Let the two tangents meet AA' in the points C, C.
Find F, F', the foci of the system AA', CC in involution (1066)
determining the centre by (985). Similarly, find G, G', the
foci of a system on the line A A". Then, by (4826), the chord
of contact of the tangents may be any of the lines FG-, FG',
F'G, F'G'.
There are accordingly four solutions, and the
construction of (4831) determines the conic.
4836 To describe a conic, given two points and three
tangents.
Let AB, BC, CA (Fig. 167) be the tangents, and P, P' the
points. Draw a transversal through PP' meeting the three
tangents in Q, Q', Q". Find F, a focus of the system PP', QQ'
in involution (1066, 985) ; G a focus for PP', QQ", and H/or
PP'j QQ" Construct a triangle with its sides passing through
F, G, H, and with its vertices L, M, N on BC, CA, AB,
by the method o/(4823), ivhich is equally applicable to a recti
lineal figure as to a conic. L, M, N will be the points of
contact.
The reason for the construction is contained in
(4826). There will, in general, be four solutions.
If the conic be a parabola, the foregoing constructions
can be adapted by considering one tangent at infinity always
to be given.
4837

To draw aparabola through four given points a, a', b, V.

This is problem 4833 with the tangent at infinity.

In figure (52), suppose cc to coincide and AB to remove to infinity so as
to become the tangent at c, the opposite vertex at infinity of a parabola, and
therefore to be perpendicular to the axis. Cc then becomes a diameter of
the parabola, and Carnot's theorem (4778) shows that
Ca.Ca _A^_B^_ sin2 ACc
Gb.CV
AbyBc*
sin2 BOc
since the points C, a, a, b, V are all on the axis of the parabola relatively to
the infinite distance of AcB. This result, however, is at once obtained from
equation (4221), Ca.Ca' : Cb.Cb' being the ratio of the products of the roots
of two similar quadratics. Thus a diameter of the parabola can be drawn
through C by the known ratio of the sines of ACc and BCc.
Next, describe a circle round three of the given points a, a', b. By the
property (1263) and the known direction of the axis, the other point in
which the circle cuts the parabola can be found.
Five points being known, we can, by Pascal's theorem, as in (4831),

664

TRILINEAR ANALYTICAL CONICS.

obtain two parallel chords, and then find P, the extremity of their diameter,
by the proportion, square of ordinate oc abscissa (1239).
Lastly, draw the diameter and tangent at P, and then, by equality of
angles (1224), draw a line from P which passes through the focus. By
obtaining in the same way another pair of parallel chords, a second line
through the focus is found, thus determining its position.

4838

To draw a parabola when four tangents are given.

This is effected by the construction of (4832, Fig. 71). Let AB, BC, AE,
ED be the four tangents, and CD the tangent at infinity. Then any line
drawn to C will be parallel to BG, and any line to D will be parallel to ED.

4839 To draw a parabola, given three points and one

tangent.
This is effected by the construction of (4835, Fig. 67). Let hC be the
tangent at co ; then the centre of involution 0 must be at C, so that
CC. CO' = O.oo = CA.CA' = CF, determining F. F', another point on
the chord of contact, being found by joining AA" or A'A", FF will be the
diameter through a, since the other point of contact b is at infinity.

4840 To draw a parabola, given one point and three

tangents.
This is the case of (4834), in which one of the given tangents V is at
infinity. B must therefore be at infinity, and QP, PR and the tangent 6,
since they all join 11 to finite points, must be parallel. The ratio found
determines another taugent, and the case is reduced to that of (4838).

4841 To draw a parabola, given two points and two

tangents.
This is problem (4836). Suppose AC in that construction to be the
tangent at infinity. F, G, H will be determined as in (4839) by mean
proportionals. The chords LM, NM will become parallel, since M is at
infinity ; and we have to draw LN and the parallel lines from L and N to
pass through F, G, H in their new positions, so that the vertices L, N may
lie on BC and AB.
Otherwise by (4509), the intercepts s and t can readily be found from tho
two equations furnished by the given points.

4842 To describe a conic touching three right lines and

touching a given conic twice.
Let AD, CF, EB (Fig. 65) be the three lines as they cut
the given conic. Join AB, AF, BC, BE, and determine K by
the Pascal line MLN. K will be one point of contact of the
two conies, by (4822) and the proof in (4823) ,"since AD, CF,
EB, and the tangent at K are four positions of the " remaining
side " in that proposition. The problem is thus reduced to

RECIPROCAL POLARS.

665

(4834), since four tangents and K the point of contact of one of

them are now known.
4843 To describe a conic touching each of two given conies
twice, and passing through a given point or touching a given
line.
Proceed by (4803), determining fi by the last condition.
To describe a conic touching the conies S+L2, S+M*,
8 + N2 (4707) and touching 8 twice.
[Salmon, Art. 387.

THE METHOD OF RECIPROCAL POLARS.

Def.The polar reciprocal of a curve is the envelope of
the polars of all the points on the curve, or it is the locus of
the poles of all tangents to the curve, taken in each case
with respect to an arbitrary fixed origin and circle of recipro
cation.
4844 Thus, in figure (72), to the points P, Q, R on one
curve correspond the tangents qr, rp, and chord of contact pq
on the reciprocal curve ; and to the points p, q, r correspond
the tangents QR, RP, and chord PQ.
The angle between the tangents at P and Q is evidently
equal to the angle pOq, since Op, Oq, Or are respectively per
pendicular to QR, RP, PQ.
4845 Theorem.The distance of a point from a line is to its
distance from the origin as the distance of the pole of the line
from the polar of the point is to its distance from the origin.
Proof.(Fig. 73.) Take 0 for origin and centre of auxiliary circle, PT
the polar of c, pt the polar of C, CP perpendicular on polar of c, ep perpen
dicular on polar of C. Then
r> = OC. Ot =Oc.OT ) Therefore, by subtraction, OC.mt = Oc.MT,
and
OO.Om=Oc.OM)
or
OC.cp =Oo.CP;
that is,
CP : CO :: cp : cO.
Q. E. D.

Cob.By making CP constant, we see that the reciprocal

of a circle is a conic having its focus at the origin and its
directrix the polar of the circle's centre.
4 Q

666

TBILINEAB ANALYTICAL CONIGS.

GENERAL RULES FOR RECIPROCATING.

4846 -4 point becomes the polar of the point, and a right line
becomes the pole of the line*
4847 -4 line through a fixed point becomes a point on a fixed
line.
4848 The intersection of two lines becomes the line which
joins their poles.
4849 Lines passing through a fixed point become the same
number of points on a fixed line, the polar of the point.
4850 A right line intersecting a curve in n points becomes n
tangents to the reciprocal curve passing through a fixed point.
4851 Two lines intersecting on a curve become two points
whose joining line touches the reciprocal curve.
4852 Two tangents and the chord of contact become two
points on the reciprocal curve and the intersection of the tan
gents at those points.
4853 -4 pole and polar of any curve become respectively a
polar andpole of the reciprocal curve; and a point of contact and
tangent become respectively a tangent and its point of contact.
4854 The locus of a point becomes the envelope of a line.
4855 -4" inscribed figure becomes a circumscribed figure.
4856 Four points connected by six lines or a quadrangle
become four lines intersecting in six points or a quadrilateral.
4857 The angle between hoo lines is equal to the angle sub
tended at the origin by the corresponding points.
(4844)
4858 The origin becomes a line at infinity, the polar of the
origin.
4859 Two lines through the origin become two points at
infinity on the polar of the origin.
4860 Two tangents through the origin to a curve become two
points at infinity on the reciprocal curve.
4861 The points of contact of such tangents become asymptotes
of the reciprocal curve.
4862 The angle between the same tangents is equal to the
angle between the asymptotes.
(4857)
* That is, with respect to the circle of reciprocation, and so throughout with the excep
tion of (4853).
*

RECIPROCAL POLARS.

667

4863 According as the tangents from the origin to a conic are

real or imaginary, the reciprocal curve is an hyperbola or
ellipse.
4864 If the origin be taken on the conic, the reciprocal curve
is a parabola.
For, by (4860, '1), the asymptotes are parallel and at infinity.

4865 A. trilinear equation is converted by reciprocation into

a tangential equation.
Thus ay = kfiS is a conic passing through four of the intersections of
the lines a, /3, y, S. Reciprocating, we get a tangential equation of the same
form AC = kBD, and this is a conic touching four of the lines which join
the points whose tangential equations are A = 0, B = 0, O = 0, D = 0.
See (4907).

4866 The equation of the reciprocal of the conic ay + b'2x2

= a2b* with the same origin and axes is

aW+by = tf,
where k is the radius of the auxiliary circle whose centre is
the centre of the conic.
Proof.Let p be the perpendicular on the tangent, 0 its inclination ;
then A;4r-2=/ = ascosJe + 62sin,e (4732).

4867 The same when the origin of reciprocation is the

point y,
{xx'+yy' +k>)2 = aW+by.
Phoop:

Vr-l=p= vV cosa0 + 62 sin20-(a)' cos 0+y' sin0).

4868 The reciprocal curve of the general conic (4656), the

auxiliary circle being x*+y2 = k- or x*+y2 + z2 = 0 in trilinears, will be symmetrically
A? +Bi? +C?+2Fvt+2G+2Hn= 0,
replacing I by k2.
Proof. Let Eri be a point on the reciprocal curve, then the polar of ij,
namely, *+t// fc* = 0, must touch, the conic, by (4853). Therefore, by
(4665), we must substitute , ij, k2 for A, fi, v in the tangential equation

A\* + &c. = 0.
4869 From the reciprocal of a curve with respect to the
origin of coordinates, to deduce the reciprocal with respect to
an origin x'y', substitute in the given reciprocal equation
,-

T-r-TT, for x

and

t-ff, , , tor y.

668

ANALYTICAL C0NI08.

Proof.Let P be the perpendicular from the origin on the tangent and

PR = k1. The perpendicular from x'y is P x cos 0 y' sin 0,
fc1
V
,
a . a
. ka xx' + yy' + k*
= -5- * cosfl y Bind,
.: - =
;
P
B
Bp

= j^-;
Vp cos 0
:. Rcos6
xx +yy +1?

TANGENTIAL COORDINATES.

4870 By employing these coordinates, theorems which are

merely the reciprocals of those already deduced in trilinears
may be proved independently. See (4019) for a description
of this system.
The following proposition serves to transform by recipro
cation the whole system of trilinear coordinates of points and
equations of right lines and curves, into tangential coordinates
of right lines and equations of points and curves.

THEOREM OP TRANSFORMATION.
4871 Given the trilinear equation of a conic (4656), the
tangential equation of the reciprocal conic in terms of X, p, v,
the perpendiculars from three fixed points A', B, C upon the
tangent (Fig. 74) will be as follows, 0 being the origin of
reciprocation and OA', OB', OC = p, q, r:
4872

5S+4+^+Sk + a + ?e.= OL
p*
qr ' r* 1 qr
rp
pq

Proof.Let a = 0, (3 = 0, y = 0 be the sides of the original trigon ABC.

The poles of these lines will be A', B', C, the vertices of the trigon for the
reciprocal curve. Let BS be the polar of a point P on the given conic ;
a, /3, y the perpendiculars from P upon BC, OA, AB ; i.e., the trilinear co
ordinates of P. Let \, fi, v be the perpendiculars from A', B', C upon ES;
i.e., the tangential coordinates of the pOlar of P referred to A', B', C'. Then,
by (4845), itF = oh"fp=m'irr = w- Sub8titnte these ^
of a, (3, y in (4656) and divide by OP1.
4873 The angular relation between the trigons ABC and
A'B'C is
B'OC = n-A,

COB' = w-B,

A'OB=n-C.

TANGENTIAL COORDINATES.

669

4874 If ABC be self-conjugate with regard to the circle of

reciprocation, it will coincide with A'B'C.
4875 Now let 0 be the circura-centre (4629) of A'B'C
(Fig. 74), then it will be the in-centre of ABC, and, by (4873),
2A' = rr-A,

2B = ir-B,

2C'=it-C.

Also p = q = r in (4872), which becomes <j> (X, n, v) = 0, so

that the conic and its reciprocal are represented by the same
equation. Consequently any relation in trilinear coordinates
has its interpretation in tangential coordinates.
We have
then the following rule :
4876 Rule. To convert any expression in trilinears into
tangentials, consider the origin of the former as the in-centre of
the trigon, change a, (3, y into X, p, v, and interpret the result
by the rules for reciprocating (4846-65). If the angles of the
original trigon are involved, change these by (4875) into the
angles of the reciprocal trigon, of which the origin will now be
the circum-centre.
4877 Referring trilinears and tangentials to the same trigon
ABC, the equation of a point, as shown in (4021), becomes

Pi
4878

P2,

Pz

or, by multiplying by 2,
BOC\+ COAp+AOBv = 0.

(Fig. 3)

The equation of a point can generally be obtained directly

from the figure by means of this formula.

EQUATIONS IN TANGENTIAL COORDINATES.

For direct demonstrations of the following theorems, the reader may
consult Ferrers' Trilinear Coordinates, Chap. VII.
4879 The point dividing AB in the ratio a : b, that is, the
intersection with the internal or external bisector of C, is
aX V = 0.

Centre of AB X+^t = 0.

The point O in (4878) is now on the side AB.

4881

Mass-centre,

\+n+v = 0.

[For BOC=COA = AOB.

ANALYTICAL CONIGS.

670

4882

In-centre,

aX+I)/t+Cv = 0.

4883

By (4878), for
B0C ^COA^AOB
a ex-centre, aX+l)/t+rv = 0. L a
b
c '

4884

Circum-centre X sin 2A +ft sin 2B + sin 2C = 0.

Proof.For BOG = 11* sin 2A, &c. in (4878). Otherwise.By recipro

cation (4876), a sin A +/3 sin B + y sin G = 0 is the line at infinity referred
to the trigon ABC ; therefore
X sin A + p sin B + v sin C = 0
is the equation of the pole of that line referred to A'B'C' ; that is,
X sin 2A'+fi sin 25'+ v sin 2C, by (4875).
4885

Foot of perpendicular from 0 upon 45,

Xtan^-f/i tanfi = 0.

4886

Orthocentre X tan A +/* tan B+v tan C = 0.

4887

Inscribed conic of ABC,

[Proof below.

Lpv+Mv\+N\n = 0.
4888

Point of contact with AB,

AfX+L/* = 0.

4889

In-circle (4629),
(*-a) i*+(*-t>) vX+ ($-1) Xp = 0.

4890

Point of contact with AB, (S b) X+(S-a) /* = 0.

4891

a ex-circle,

X/*+(S-r) vX Sifw = 0.

Proof.Since the coordinates of AB of the trigon are 0, 0, v, the equa

tion of the inscribed conic must be satisfied when any two of the coordinates
X, (i, v vanish, therefore it must be of the form (4887). Otherwise by
reciprocating (4724).
If the circle touches AB in I) (Fig. 3), X : -/j = AD : BD = 8 a : 8 b
(Fig. of 709), which proves (4890).
(4888) is the equation of the point of contact, because the line (0, 0, v)
passes through it and also touches the conic (4887).
(4889) is the in-circle by (4887) and (4890) and what precedes.
4892

Circumscribed conic,

[By (4876) applied to (4739, '40).

L2\i+M2ft2+X2v9-2MNiiy-2NLv\-2LM\n = 0, (4740)

TANGENTIAL COORDINATES.

4893

or

4894

Tangent at A,

4895

Circum-circle

671

y/(L\)+y/Mn+^Nv = 0.
Mp = Nv.

a4x2+i)V+tV-21)2fVv-2c2a2x-2a2l)V = 0;
4896

or

a v/X+li v//t+r y = 0.

Proof. By (4876) applied to (4747, '8), and by cos4 = sin A' (4875)

4897

Relation between the coordinates of any right line :

a(x-f*)(x-i,)+i>(ft-F)Ot-x)+^(i'-X)(^-/*) = S2.
4898

Coordinates of the line at infinity :

X = /x = v.

Peoof.The trilinear coordinates of the origin and centre of the re

ciprocal conic are a = y8 = y, (4876). It is also self-evident.
4899 The point l\ + m/x-\-nv = Q will be at infinity when
l-\-m + n -- 0.
Proof.By (4876), for the line la + mft + ny = 0 will pass through the
origin a = (3 y when l + m + n = 0.

4900 A curve will be touched by the line at infinity when

the sum of the coefficients vanishes.
Proof.By (4876), for this is the condition that the origin in trilinears,
a = ft = y shall be on the curve.

4901

The equation of the centre of the conic <f> (A, n, v) is

<\+<^+<^= 0,

4902

or (a+h+g)k+(h+b+f)n+(g+f+c)p=:0.

Proof.The coordinates of the in-centre of ABO (4876) are a'= /3'= y',
therefore the polar of this point with regard to the conic f (a, /3, y) is
^a + ^ + ^y= 0 (4658). This point and polar reciprocate into a polar and
point, of which the former, being the reciprocal of the in-centre, or origin, is
the line at infinity, and therefore the latter is the centre of $> (X, /x, v), while
its equation is as stated.

672

ANALYTICAL CONICS.

4903 The equation of the two points in which the line

(X', fi, v) cuts the conic is
4> (V, p\ v') <f> (X,

v) = (^X'+^/i'+^f')8-

(4680)

4904 The coordinates of the asymptotes are found from the

equations
<f>(\, p,v) = 0 and
+
<k= 0.
Proof.These are the conditions that the line (X, ft, v) should touch the
curve and also pass through the centre (4901).
4905

The equation of the two circular points at infinity is

a8 (x-F)(x-v)+b2 (ft-v)(/4-X)+ri (v-x)(v-/t) = o.

Proof.Put X'= p'= v in (4903) to make the line at infinity, and for
the conic take the in-circle (4889).

4906

The general equation of a circle is

a* (X-/t)(X-v)+62 (ft-v)(ft-X)+t8 (v-X)(v-/t)

= (/X+m^+nv)2

(1),

where l\+mfi+nv = 0 is the equation of the centre.

Proof.The general equation of a conic in trilinears may, by (4601), be
put in the form
a(/3-/30)(r-y.)+Hr-7)(a-0)+c(a-0o)(/3-/30) = (Ja + WIj3 + y),
where la + mfi + ny = 0 is the directrix, and a0/30y0 the focus. Now let the
focus be the iu-centre of the trigon, and therefore a0 = /30 = y0= |SS~' (709).
By this relation and aa + b/3 + Cy = 2, the equation is expressed as
0(-a)(a-/3)(a-y) + &c. = (Z'a + m'/3+n'y)',
or
(a /3)(a-y) coss|4 + &c. = (Z'a + m'/3 + n'y)\
Reciprocating by (4876), this becomes
(Xp)(X-y) sinM' + Ac. = (ZX+m/ +ny,
the constant factor introduced on the right being involved in I, m, n ; and
sin A' = cos 4,-4, by (4875). And we know that this is a circle by (4845
Cor.), and that the directrix of the conic reciprocates into the centre of the
circle.
Otherwise.The left side of (1) represents the two circular points at
infinity (4905), and, if for the right we take the equation of a point, the
whole represents a conic, as in (49(j9), of the form AG = B2. In this case,
A, G, the points of contact of tangents from B, being the circular points, the
conic must be a circle with B = 0 for its centre.

TANGENTIAL COORDINATES.

673

Abridged Notation.
4907 Let A = 0, B = 0, (7 = 0, D = 0 (Fig. 75) be the
tangential equations of the four points of a quadrangle, where
A = a^+bi/u + c^v, B = a,z\ + b2n + c2v, and so on. Then
the equation of the inscribed conic will be AG = kBD.
Proof.The equation is of the second degree in X, /i, v ; therefore the
line (X, n, v) touches a conic. The coordinates of one line that touches this
conic are determined by the equations -4 = 0, B = 0. That is, the line
joining the two points A, B touches the conic, and so of the vest.

4908 If the points B, D coincide (Fig. 76), the equation

becomes AC = kB2 ; and A = 0, G = 0 are the points of
contact of tangents from the point B = 0.
4909 Referring the conic to the trigon ABG (Fig. 78), and
taking AG = k2B2 for its equation, let a tangent ef be drawn,
and let Ae : eB = k : m. The equations of the points e and
/ will be
mA+kB = 0,
mkB + G = 0.
Proof. The first equation corresponds to (4879). For the equation of/,
eliminate A from mA + kB = 0 and AC = feaZJ2.

4910 Let e, h (Fig. 77) be two points on AB whose equa

tions are mA-\-kB = 0, m'A + kB = 0. The equation of the
point p, in which tangents from e and h intersect, is
mm'A + (m+m') kB+C = 0.
Proof. The equation may be put in the form
(mA + kB)(m'A + l-B) = 0,
because k-Bl=AC if the line touches the conic. The equation being of the first
degree in A, B, O, must represent same point. That is, the relation between
X, /u, v involved in it makes the straight line \a + /ifi + vy pass through a
certain point. But the equation is satisfied when mA + IcB = 0, a relation
which makes the straight line pass through e. Hence a tangent through e
passes through a certain fixed point. Similarly, by m'A + kB = 0, another
tangent passes through h and the same fixed point.

4911 Cor. Let m' = m, then the equation of the point of

contact of the tangent joining the points ma + kB and mkB + G
(4909) (e and/, Fig. 78) will be
m2A + 2mkB+C = Q.
4912

If m Fig. (78) the trilinear coordinates of the points

4 R

674

ANALYTICAL CONlCS.

A, B, G are a^, yu zu xit yi} z,, a;3, ys, zs, the coordinates of the
point of contact p of the tangent defined by m will be
m2x1-\-2mkx2+xs, m2y1 + 2ynkyi + ys, m\ + 2mfcz2 + z3,
and the tangent at p divides the two fixed tangents in the
ratios k : m and mk : 1, by (4909).
4913 Note. The equation U or * (A, p, v) = 0 (4665) expresses the
condition that \a+fi(3 + vy shall touch a certain conic. When U is about
to break up into two factors, the minor axis of the conic diminishes (Fig. 79).
Every tangent that can now be drawn to the conic passes very nearly
through one end or other of the major axis. Ultimately, when the minor
axis vanishes, the condition of the line touching the conic becomes the con
dition of its passing through one or other of two fixed points A, B. In this
case, U consists of two factors, which, put equal to zero, are the equations
of those points. The conic has become a straight line, and this line is
touched at every point by a single tangent.

4914 If U and U' (Fig. 80) be two conies in tangential

coordinates, kU+U' is then a conic having for a tangent
every tangent common to ZJand V ; and kU+AB is a conic
having in common with U the two pairs of tangents drawn
from the points A, B.
The conic U' in this case merges into the line AB, or,
more strictly, the two points A, B, as explained in (4913).
4915 If either kU+U' or kU+AB breaks up into two
factors, it represents two points which are the opposite ver
tices of the quadrilateral formed by the four tangents.

ON THE INTERSECTION OF TWO CONICS.

INTRODUCTORY THEOREM.

Geometrical meaning of ^( 1).*

4916 In a system of rectangular or oblique plane coordinates, let the
operator / 1 prefixed to an ordinate y denote the turning of the ordinate
about its foot as a centre through a right angle in a plane perpendicular to
the plane of xy. The repetition of this operation will turn the ordinate
* [The fiction of imaginary lines and points is not ineradicable from Geometry. The
theory of Quaternions removes all imaginariness from the symbol V 1, and, as it appears
that a partial application of that theory presents the subject of Projection in a much clearer
light, I have here introduced the notion of the multiplication of vectors at right angles to
each other.]

ON TEE INTERSECTION OF TWO GONICS.

675

through another right angle in the same plane so as to bring it again into
the plane of xy. The donble operation has converted y into y. But the two
operations are indicated algebraically by v 1 . v l.y or ( v l)s?/ = y,
which justifies the definition.
It may be remarked, in passing, that any operation which, being per
formed twice in succession upon a quantity, changes its sign, offers a con
sistent interpretation of the multiplier v 1 .
4917 With this additional operator, borrowed from the Theory of
Quaternions, equations of plane curves may be made to represent more
extended loci than formerly. Consider the equation a? + y2= dr. For values
of x < a, we have y = Ja*x%, and a circle is traced out. For values of
x > a, we may write y = t \/as' a8, where = v 1. The ordinate
-/x% a* is turned through a right angle by the vector i, and this part of the
locus is consequently an equilateral hyperbola having a common axis with
the circle and a common parameter, but having its plane at right angles to
that of the circle. Since the foot of each ordinate remains unaltered in posi
tion, we may, for convenience, leave the operation indicated by i unperformed
and draw the hyperbola in the original plane. In such a case, the circle may
be called the principal, and the hyperbola the supplementary, curve, after
Poncelet. When the coordinate axes are rectangular, the supplementary
curve is not altered in any other respect than in that of position by the
transformation of all its ordinates through a right angle ; but, if the coordi
nate axes are oblique, there is likewise a change of figure precisely the same
as that which would be produced by setting each ordinate at right angles to
its abscissa in the xy plane.
In the diagrams, the supplementary curve will be shown by a dotted line,
and the unperformed operation indicated by i must always be borne in mind.
For, on account of it, there can be no geometrical relations between the
principal and supplementary curves excepting those which arise from the
possession of one common axis of coordinates. This law is in agreement
with the algebraic one which applies to the real and imaginary parts of the
equation x1 (iy)* = a*. When y vanishes, x = a in both curves.
If either the ellipse b^x1 + a'y1 = aV or the hyperbola 6V aSf = a%2
be taken for the principal curve, the other will be the supplementary curve.
It is evident that, by taking different conjugate diameters for coordinate
axes, the same conic will have corresponding different supplementary curves.
The phrase, "supplementary conic on the diameter DD," for example, will
refer to that diameter which forms the common axis of the principal and
supplementary conic in question.
4918 Let us now take the circle x*+y2 = a' and the right line x = b.
When b is > a, the line intersects the supplementary right hyperbola in two
points whose ordinates are i \/bl aJ. By increasing b without limit, we
get a pair of, so-called, imaginary points at infinity. These lie on the
asymptotes of the hyperbola, and the equation of the asymptotes is

{x+iy)(x iy) = 0.
We can now give a geometrical interpretation to the statements in (4720).
The two lines drawn from the focus of the conic 6V + aSf = a262 to the
" circular points at infinity " make angles of 45 with the major axis, and
they touch the conic in its supplementary hyperbola 6V a'2 (iy)2 = a'63.
An independent proof of this is as follows.

676

ANALYTICAL OONIGS.

Draw a tangent from S (Fig. 81) to the supplementary hyperbola, and

let x, y be the coordinates of the point of contact P. Then

by the value of x.

Also

SN = x-CS=

...

... - v/a'-i1 =

vV-i2)

/('-&')

Therefore j/ = &N", therefore SP makes an angle of 45 with GN.

The following results are required in the theory of projec

tion, and are illustrated in figures (82) to (86). Two ellipses
are taken in each case for principal curves, and the supple
mentary hyperbolas are shown by dotted lines. As the planes
of the principal and supplementary curves are really at right
angles, the intersections of the solid lines with the dotted are
only apparent. The intersections of the solid lines are real
points, while the intersections of the dotted lines represent
the imaginary points.
4919 Two conies may intersect
(i.) in four real points (Fig. 82);
(ii.) in two real and two imaginary points (Fig. 83) ;
(iii.) in four imaginary points (Fig. 84).
[When the two hyperbolas in figures (83) and (84) are similar and
similarly situated, two of their points of intersection recede to infinity (Figs.
85 and 86). Hence, and by taking the dotted lines for principal, and the
solid for supplementary, curves, wc also have the cases]

(iv.) in two real finite points and two imaginary points at

infinity ;
(v.) in tivo imaginary finite points and two imaginary
points at infinity ;
(vi.) in two imaginary finite points and two real points at
infinity ;
(vii.) in hvo real finite points and two real points at
infinity.
4920 Given two conies not intersecting, or intersecting in
but two points, to draw the two supplementary curves which
have a common chord of intersection conjugate to the

THE METHOD OF PROJECTION.

677

diameters upon which they are described, or in other words,

to find the imaginary common chord of the conies.
Poncelet has shewn by geometrical reasoning (Proprietes des Projectives,
p. 31) that such a chord must exist. The following is a method of deter
mining its position
Let
(abcfghjxyl)* = 0 and (a'b'c'f'g'h'^xyiy = 0
(i.)
be the equations of the conies 0, C (Pig- 89), the coordinate axes being
rectangular. Suppose PQ to be the common chord sought. Then the
diameters AB, A'B' conjugate to PQ bisect it in D, and the supplementary
curves on those diameters intersect in the points P, Q. Now, let the coor
dinate axes be turned through an angle 0, so that the y axis may become
parallel to PQ, and therefore also to the tangents at A, B, A', B' . This is
accomplished by substituting for as and y, in equations (i.), the values
x cosO y sin 0 and y cos 0 + x sinO.
Let the transformed equations be denoted by (ABCFGH~$xyiy = 0 and
(A'B'CF'G'H'~xy\y = 0, in which the coefficients are all functions of 0,
excepting c, which is unaltered. Solving each of these equations as a quad
ratic in y, the solutions take the forms
y = aa; + /3 \/fi (x22px + q), y = a'as + /3' \Zfi (%* 2p'x + q') ...(ii.),
with the values of a, j3, /j,, p, q given in (4449-53), if for small letters we
substitute capitals. Thus, a, /3, /x, p, q are obtained in terms of 0 and the
original coefficients a, h, b, f, g, h.
Now, the coordinates of D being = ON, n = DN, we have i; = a+/3
and 7; = a'$ + /3', therefore
a + fi = a'$ + /3'
(iii.).
The surd in equations (ii.) represents the ordinate of the conic conjugate
to the diameter AB or A'B'. For values of x in the diagram > OM and
<OR, the factor / 1 appears in this surd, indicating an ordinate of the
supplementary curve on AB or A'B'. Hence, equating the values of the
common ordinate PB, we have
H (?-2pt+q) = n' (?-2p'H + q')
(iv.).
Eliminating between equations (iii.) and (iv.), we obtain an equation for
determining 0 ; which angle being found, we can at once draw the diameters
AB, A'B'.

THE METHOD OF PROJECTION.

4921 Problem.Given any conic and a right line in its
plane and any plane of projection, to find a vertex of projec
tion such that the line may pass to infinity while the conic is
projected into a hyperbola or ellipse according as the right
line does or does not intersect the given conic ; and at the
same time to give any assigned proportion and direction to
the axes of the projected conic.

678

ANALYTICAL C0NIG8.

Analysis.Let HCKD be the given conic, and BB the right line, in

Fig. (87) not intersecting, and in Fig. (88) intersecting the conic. Draw
HK the diameter of the conic conjugate to BB. Suppose 0 to be the
required vertex of projection. Draw any plane ECQD parallel to OBB,
intersecting the given conic in CD and the line HK in F, and draw the
plane OHK cutting the former plane in E, F, 0 and the line BB in A ; and
let the curve ECQD be the conical projection of HCKD on the plane parallel
to OBB.
By similar triangles,
HF~ HA

FK

AK'

" HF.FK

HA.AK

( '"

Let a, j3 be the semi-diameters of the given conic parallel to HK and CD ;

th"'"

0F% - &HF.FK
as'

GF*
= F-HAAK
"' EF.FO
a2.OA'

(9\
{)'

Now, since parallel sections of the cone are similar, if the plane of HCKD
moves parallel to itself, the ratio on the right remains constant ; therefore, by
(1193), the section ECOK is an ellipse in Fig. (87) and an hyperbola in
Fig. (88). Let a, b be the semi-diameters of this ellipse or hyperbola
parallel to EQ and CD, that is, to OA and BB ; then, by (2),
* = *A^,
a'
a'
0Al

,. OA> = %*HA..AK
Far

(3).

But ^ HA.AK = AB\ where AB in Fig. (88) is the ordinate at A of the

a
given conic, but in Fig. (87) the ordinate of the conic supplementary to the
given one on the diameter conjugate to BB. Therefore
AOt=^AB1

(4).

Hence AO, AB are parallel and proportional to o and b. And, since AB

is given in magnitude and direction, we have two constants at our disposal,
namely, the ratio of the semi-conjugate diameters a and b and the angle
between them, or, which is the same thing, the eccentricity and the direction
of the axes of the ellipse or hyperbola on the plane of projection.

4922 The construction will be as follows :

Determine the point A as the intersection of BB with the
diameter HK conjugate to it. Choose any plane of projection,
and in a plane through BB, parallel to it, measure AO of the
length given by equation (3) or (4), making the angle BAO
equal to the required angle between a and b. 0 will be the
vertex of projection, and any plane LMN parallel to OBB will
serve for the plane of projection.
4923 Coe. I.If AO = AB, the projected curve in Fig. (88)
will in every case be a right hyperbola.

THE METHOD Of PROJECTION.

679

4924 Coe. 2.If BAO is a right angle, the axes of the pro
jected ellipse or hyperbola are parallel and proportional to
AO and AB. Hence, in this case, the eccentricity of the
hyperbola will be e = OB : OA.
4925 Cor. 3. If AO = AB and BAO = a right angle,
the ellipse becomes a circle and the right hyperbola in Cor. 1
has its axes parallel to AO and AB.
4926 To project a conic so that a given point in its plane
may become the centre of the projected curve.
Take for the line BB the polar of the given point, and con
struct as in (4922). For, ifP be the given point, and BB its
polar (Fig. 87 or 88), p the projection ofV tvill have its polar
at infinity, and will therefore be the centre of the projected
ellipse or hyperbola, according as P is within or without the
original conic.
4927 To project two intersecting conies into two similar
and similarly situated hyperbolas of given eccentricity.
Take the common chord of the conies for the line BB (Fig.
88), and project each conic as in (4922), employing the same
vertex and plane of projection. Then, since the point A and
the lines AB and AO are the same for each projection, corres
ponding conjugate diameters of the hyperbolas are parallel and
proportional to AO and AB ; therefore, Sfc.
4928 To project two non-intersecting conies into similar
and similarly situated ellipses of given eccentricity.
Take the common chord of a certain two of the supple
mentary etirves of the conies (4920), in other words, the
imaginary common chord of the conies, for the line BB, and
proceed as in (4927).
4929 To project two conies having a common chord of
contact into two concentric, similar and similarly situated
hyperbolas.
Take the common chord for the line BB, and construct as in
(4922). The common pole of the conies projects into a common
centre and the common tangents into common asymptotes.

680

TBILINEAB ANALYTICAL OONICS.

4930 To project any two conies into concentric conies.

Find the common pole and polar of the given conies by
(4762), and take the common polar for the line BB in the
construction of (4922). The common pole projects into a
common centre.
4931 Ex. 1. Given two conies having double contact with each other,
any chord of one which touches the other is cut harmonically at the point of
contact and where it meets the common chord of contact of the conies.
[Salmons Conic Sections, Art. 354.
Let AB bo the common chord of contact, PQ the other chord touching
the inner conic at C and meeting AB produced in D. By (4929), project
AB, and therefore the point D, to infinity. The conies become similar and
similarly situated hyperbolas, and C becomes the middle point of PQ (1189).
The theorem is therefore true in this case. Hence, by a converse projection,
the more general theorem is inferred.
4932 Ex. 2. Given four points on a conic, the locus of the pole of any
fixed line is a conic passing through the fourth harmonic to the point in
which this line meets each side of the given quadrilateral. [Ibid., Art. 354.
Let the fixed line meet a side AB of the quadrilateral in D, and let
ACBD be in harmonic ratio. Project the fixed line, and therefore the point
B, to infinity. C becomes the middle point of AB (1055), and the pole of
the fixed line becomes the centre of the projected conic. Now, it is known
that the locus of the centre is a conic passing through the middle points of
the sides of the quadrilateral. Hence, projecting back again, the more
general theorem is inferred.
4933 Ex. 3. If a variable ellipse be described touching two given
ellipses, while the supplementary hyperbolas of all three have a common
chord AB conjugate to the diameters upon which they are described ; the
locus of the pole of AB with respect to the variable ellipse is an hyperbola
whose supplementary ellipse touches the four lines CA, CB, C'A, CB, where
C, C are the poles of AB with respect to the fixed ellipses.
(Salmon, Art. 355.)
Proof.Project AB to infinity and the three ellipses into circles. The
poles P, C, C become the centres p, c, c of the circles. The locus of p is a
hyperbola whose foci are c, c. But the lines Ac, Ik now touch the supple
mentary ellipse of this hyperbola (4918). Therefore, projecting back again,
we get AC, BC touching the supplementary ellipse of the conic which is the
locus of P. Similarly, AC', BC touch the same ellipse.

4934 Any two lines at right angles project into lines which
cut harmonically the line joining the two fixed points which
are the projections of the circular points at infinity.
Proof.This follows from (4723).

INVARIANTS AND COVAMANTS.

681

4935 The converse of the above proposition (4931), which is the theorem
in Art. 356 of Salmon, is not universally true in any real sense. If the lines
drawn through a given point to the two circular points at infinity form a
harmonic pencil with two other lines through that point, the latter two are
not necessarily at right angles, as the theorem assumes.
The following example from the same article is an illustration of this
Ex.Any chord BB (Fig. 88) of a conic HCKD is cut harmonically by
any line PKAH through P, the pole of the chord, and the tangent at K.
The ellipse BKB here projects into a right hyperbola ; B, 11 project to
infinity. The harmonic pencil formed by PK and the tangent at K, KB and
KB projects into a harmonic pencil formed by pk and the tangent at k, kb
and kb, where b, b are the circular points at infinity : but pk is not at right
angles to the tangent at k of the right hyperbola. The harmonic ratio of the
latter pencil can, however, be independently demonstrated, and that of the
former can then be inferred. (Note that k is 0 in figure 88.)
If we may suppose the ellipse to project into an imaginary circle having
points at infinity, the imaginary radius of that circle may be supposed to be
at right angles to the imaginary tangent. The right hyperbola, however, is
the real projection which takes the place of the circle in this and all similar
instances ; and it is only in the case of principal axes that the radius is at
right angles to the tangent.

INVARIANTS AND COVAMANTS.

4936 Let u ={abcfghjxyz)\ u' = {a'b'c'fg'h'Jxyzy
be two conies as in (4401) with the notation of (1620).
The three values of /;, for which Icu+u' = 0 represents two
right lines, are the roots of the cubic equation
4937
4938

Atf+e/p+e'fc+A' = o,
where &=abc+2fghaf2bg2-ch\

4939

= Aa'+Bb'+Ce'+2Ff+2Gg'+2Hh',

and

A=bcf*,

F=ghaf, &c.

(4665)

For the values of A' and 0' interchange a with a, b with V,

&c.
Proof.The discriminant of ku + u, which
must vanish (4661), is evidently the determi
nant here written, and it is equivalent to the
cubic in question.

4940

ka + a, kh + h', kg + g'
kh + h', kb+b', kf+f
kg + g', kf+f, kc+c

A, 9, 9', and A' are invariants of the conic ku+u'.

4 s

ANALYTICAL CONICS.

682

That is, if the axes of coordinates be transformed in any

manner, the ratios of the four coefficients in (4937) are
unaltered.
Proof.Tho transformation is effected by a linear substitution, as in
(1794). Let w, u thus become v, v. Then ku + u becomes kv + v', and
k is unaltered. If the equation ku + u = 0 represents two right lines, it will
continue to do so after transformation ; but the condition for this is the
vanishing of the cubic in k ; and k being constant, the ratios of the coeffi
cients must be unalterable.

4941 The equation of the six lines which join the four
points of intersection of the conies u and u' is
Au,3-0K,2+0Vi-A,tts = 0.
Proof.Eliminate k from (4937) by ku + u' 0.
4942

The condition that the conies u and u' may touch is

(00-9AA')2 = 4 (W-'dAQ ^SA Q),

4943

or 4A0s+4A'0,+27A2A'i-18AA,00'-020'i.

Proof.Two of the four points in (3941) must coincide. Hence two oat
of the three pairs of lines must coincide. The cubic (4937) must therefore
have two equal roots. Let a, a, ft be the roots ; then the condition is the
result of eliminating a and ft from the equations
A(2a + /5) = -9, A(al + 2a/3) = 0',
= -A' (406).
4944 The expression (4943) is the last term of the equation
whose roots are the squares of the differences of the roots of
the cubic in k, and when it is positive, the cubic in k has two
imaginary roots ; when it is negative, three real roots ; and
when it vanishes, two equal roots.
Proof.By (543) or (579). The last term of/ (x) in (543) is now
= 27F(a) F(ft), a, ft being the roots of 3AxJ + 29a; + e' = 0. When this
term is positive, /(as) has a real negative root (409), and therefore F(x) has
then two imaginary roots ; for, if (a ft)1 = c, a b = ic, and a and b are
both imaginary. When the last term of / (x) is negative, all the roots of
/ (x) are positive, and therefore the roots of F (%) are all real.

INVARIANTS OF PARTICULAR CONICS.

4945

When u = aar + by2 + c and

A = abc,

0 = bc+ca+ab,

u = af+if + z1,

0'= a+6+c,

A'= 1.

INVARIANTS AND GOVAMANTS.

4946

When u= (abcfghXxyzy and u' = x' + if+s?,

@ = A+B+C,

4947

& = a+b+c,

A' = 1.

When u = a? + yi-r2 and v! = (i-a)2 + (y-/3)2-s2,

A = -r2,
0 = a2+p2r2-s\

4948

683

A' = -*2,
0' = o2+i82-r2-2*2.

The cubic for A; reduces to

{k+1) {sW+^+^-a2-?) fc+r2} = 0.

4949

When

u = 6V + ay - a!62
A = -a464,

and

m' = (as - a)8 +

0 = a262 {o2+/82-a2-62-r2},

0' = a^+ftV-a'V-r8 (a*+&2),

4950

- /3)2 - r2,

A' = -r2.

When m = y24m:c and ?t' = (a: a)2 + (y /3)a r2,

A = -4ma, 0 = -4m(a+m), 0' = ^-^ma-r2, A'=-r2.

4951
A,

When u = (abcfghXxyzy and

A'= 0,

= x2 + 2xy cos &>+v/,

e=e(a+b)-f-tf+2(fg- c^cosw,
0' = c sin2o>.

Hence the following are invariants of the general conic,

the inclination of the coordinate axes being <o.
aqm

abc+2fgh-~af2~bg2-ch2 _ A
csin2u
~~ 0'

m
v ;'

4953

e (q+6)-/2-g2+2 (fg-ch) cos 6. = 0

csin2cu
0'

(g)

4954

5^
surcu

(4).

(3),

and
sin'' en

For these are what (1) and (2) become when the axes are
transformed so as to remove / and g.

684

ANALYTICAL CONIC8.

If the origin be unaltered, c is invariable, and transforma

tion of the axes will then leave invariable
4956

*h-<f-W
sin" oi

and f
sin o>

as appears by subtracting (3) from (1) and (2) from (4).

4958 Ex. (i.)To find the evolute of the conic fiV+aY

= a2b2. See also (4547).
Proof.Denote the conic by w, and by u the hyperbola c'xy + b'y'x a'x'y
(4335), which intersects in the feet of the normals drawn from x'y'. Two of
these normals mnst always coincide if x'y' is to be on the evolute. u and u
must therefore touch. We have
A = -a*b\ 9 = 0, G'= -fl?*(aW+jy-^), A'= -2aWzy.
Substitute in (4942), and the equation of the evolnte is found to be
(aV+&y-C4),+27a2&V.iy = 0.
4959 Ex. (ii.) Similarly the evolute of the parabola is
obtained from
m - i/2 4mx, u = 2xy + 2 (2m x') y4my',
A = 4m', 9 = 0, 0'= -4 (2m-x), A' = 4my,
producing the equation 27mj/2 = 4 (a: 2m)*. See also (4549).

4960 Ex. (iii.) The locus of the centre of a circle of radius

R, touching the conic b2x^ + a2y2a2b2, is called a, parallel to
the conic. Its equation is
BV-2.Ry {c2 (a! + i2) + (a2-2Z>!) x* + (2a1-b2) y2}
+ E4 {c4 (a4+4a,6+&4)-2c (a4-a2&2+3&4) x'+2<? (3a4-a2&2 + 14) y2
+ (a4 - Ga2i2 + 664) x> + (6a4 - 6a262 + M) y4 + (6a4- 10a26> + 6&4) *y }
+ E2 { -2aW (a2 + 62) +2C2 (3a4-a2&2 + &4) ai,-2c2 (a4-a263 + 364) if
-(6a4-10a262 + 664)(6V + ay) + (4a8-6a462-6a2i4 + 4fe4) x2y2
+ 2 (a2-2&') &V + 2 (&2-2a2) aV~2 (4-a262 + 3fc4) x'y1
-2(3a4-a2i2 + 64)a!2//4}
+(5'fl!2 + ay-aV)2{(a;-C)2 + 2/2} {(x + c)2+y2} = 0,
Peoof.If the curves in (4949) be made to touch, aft will be a point on
the curve parallel to u at a distance r. Therefore put the values of A, 6, 0',
and A' in equation (4942).
[Salmon, p. 325.

INVARIANTS AND COVABIANTS.

4961 When u
vanishes, and

685

of (4936) represents two right lines, A'

4962 0' = 0 is the condition that the two lines should

intersect on u ;
4963 0 = 0 is the condition that the two lines should be
conjugate with regard to u.
Proof.Transform u = 0 into 2xy = 0, bo that the axes x, y are the
right lines. This will not affect the invariants (4940). We now have, by
(4937),
A' = 0, e = 2(fg-ch), e' = -c.
e = 0 makes u pass through the origin xy ; fg = ch makes x and y conjugate.
For in (4671), if Xx + fty + y becomes y = 0, then X = v = 0, and the pole
is given by H : B : F.
But x = a = 0 at the pole, therefore H =
fg-eh = 0.
4964 The condition that either of the lines in u' should
touch u is, by (4943),
02 = 4A0'

or

AB = 0,

with the above values of 0 and 0'.

4965 The equation of the two tangents to u, when \x+fiy-\-v
is the chord of contact, is, with the notation of (4665),
(X, ii, v) = (kv+ny+vzy A.
Proof.The conic of double contact with u, ku + ^x + iiy + vy (4699),
must now become two right lines. In (4987) A' = 0 and 9' = 0, therefore
+ 8 = 0. But 9 = * (X, ft, y). Hence eliminate k.
4966 Cor.Taking the line at infinity ax + by + lz, we obtain
the equation of the asymptotes (4685).
The invariant 0 of the conic ku+u' vanishes
4967 (i-) Whenever an inscribed triangle of u' is self-con
jugate to u.
4968 () Whenever a circumscribed triangle of u is selfconjugate to u'.
4969 0' vanishes under similar conditions, transposing u
and u in (i.) and (ii.)
Proof. (i.) u becomes ax1 + by1 + cz1 (4765), and f=g = h = 0. There
fore 9 in (4937) vanishes if a'=b'=c'=0; i.e., if u is of the form
f'yt+g'zx + h'xy (4724).

ANALYTICAL C0NIG8.

686

(ii.) In this case, /'= g'= h'= 0 and 0 vanishes if be =f, Ac., i.e., if
the line x = 0 touches u, Ac.
4970 If ' be two conies, and if 62 = 4A8', any triangle
inscribed in u' will circumscribe u, and conversely.
Proof.Let u = x,+y*+z,2yz2zx2j-y and u = 2fyz + 2gzx + 2hxy,
both referred to the same triangle, (4739) and (4724). Then
A=-4, e = 4(/+? + A), 9'= -(/+? + A)', A'=2fgh;
therefore 63 = 4A0', a relation independent of the axes of reference (4940).
4971 Ex. (i.)The locus of the centre of a circle of radius r, circum
scribing a triangle which circumscribes the conic b'z,+a1y% = aV, is
(x1+f-ai-b2 + r,y+4l{b,x2+aY-atb1-t3 (a' + b*)} = 0,
from G5 = 4AG' and the values in (4949).
4972 Ex. (ii.)The distance between the centres of the inscribed and
circumscribed circles of a triangle is thus found, by employing the values of
6, G', and A in (4947), to be D = v/(r'J 2rr'), as in (936).

4973 The tangential equation of the four points of intersec

tion of the two conies u = 0, u' = 0 is
V2 = 4*717',
with the meanings
4974

U=(AB CFGHJ\iLvf ;

(4C64)

U'= {A'B'C'FG'H'X^vf.
4976
4977

V s (A"B"C"F"G"H"X^yy.
A = bc-f, &o.;

A'^b'c'-f*, &c,

as in (4665), and
4978

A" = bc'+b'c-2jr,

F" = gh'+g'h-af-af,

4979

B" = ca'+c'a-2gg',

G" = hf+h'f-bg'-b'g,

4980

C" = ab'+a'b-2hh',

H"=fg'+fg-ch'-c'h.

Proof.The tangential equation is the condition that \a + fifi + vy may

pass through one of the four points of intersection of u and u . The tan
gential equation of the conic u + ku is obtained by putting a + ka for a, &c.
in U (4665), and is U+kV+ti'U' = 0. The tangential equation of the
envelope of the system is V3 = 4sUU' (4911). This is the condition that the
line (X, /j, v) may pass through the consecutive intersections of the conies
obtained by varying k. But these conies always intersect in the same four
points. The above is therefore the tangential equation of the four points.

INVARIANTS AND COVARIANTS.

681

4981 The equation of the four common tangents of two

conies u, v! is
F2 = 4AAW,
where
and

F = (a"b"c"f'g"h"Xa/3yy,
a' = BC'+B'C-2FF', &c,

f" = GH'+G'H-AF'-A'F, &c,

as in (4978-81).
Proof.This is the reciprocal of the last theorem. Z7+ kll' is a conic
touching the four common tangents of the conies U and Z7'. The trilinear
equation formed from this will, by (4607), be A + AT + fcVA' = 0. The
envelope of this system of conies is the equation above, which must therefore
represent the four common tangents.
The curve P passes through the points of contact of u and u with the
locus represented by (4981).
4982 Hence the eight points of contact of the two conies
with their common tangents lie on the curve F.
4983 The reciprocal theorem from equation (4973) is
The eight tangents at the intersections of the conies envelope
the conic V.

4984 F = 0 is the locus of a point from which the tangents

to the two given conies u, u form a harmonic pencil.
Proof. Putting y = 0 in (4081), we get a quadratic of the form
oa^^-2/iay3+^)y8, = 0, which determines the two points in which the line y is
cut by tangents from a, /3', y'. Let the similar quadratic for the second conic
be a'cC + 2h'aP + b'P' = 0. Then, by (1064), ab' + a'b = 2hh' is the condition
that the four points may be in harmonic relation. This equation will be
found to produce P = 0.
4985 The actual values of a, h, b, suppressing the accents
on a, /3', yt are
Cp+Bf^F&y,

Gpy+Fya-Ca-Hy\

and similarly for a, h', b', with A' written for A, &c.
4986 If the (inharmonic ratio of the pencil of four tangents
be given, the locus of the vertex will be F2 = leuu'. If the
given ratio be infinity or zero, the locus becomes the four
common tangents in (4981).

ANALYTICAL CONICS.

688

4987 V = 0 is the envelope of a conic every tangent of

which is cut harmonically by the two conies u, u' ; i.e., the
equation is the condition that
+
should be cut har
monically by the two conies.
Proof.Eliminate y between the line (X, p, v), and the conies n and '
separately, and let Aa> + 2Ha(l + Bp = 0 and A'at + -lR'aji + B'f? = 0 stand
for the resulting equations. Then, by (10G4), AB' + A'B = 2HH' produces
the equation V = 0, which, by (4666), is the envelope of a conic.
4988

The actual values of A, H, B are respectively

a2+cX2_2gl,xj

hv'-giiy-fvX+eXii,

c/i'-f&p*2//**;

and similarly for A', E', B', with a' for a, &c.

4989

F2 = 4AA'mm' is a covariant (1629) of the conies u, u'.

For the four common tangents are independent of the axes of reference.
4990 U=0 and V = 0 (4973) are both contravariants
(1814) of u and v!.
Proof.For U = 0 is the condition that \a + pfi + vy = 0 shall touch the
conic u ; and V = 0 is the condition that the same line shall be cut har
monically by u and it'; and if all the equations be transformed by a recipro
cal substitution (1813, '14), the right line and the conditions remain
unaltered.
4991 Any conic covariant with u and u' can be expressed in
terms of u, vl', and F ; and the tangential equation can be
expressed in terms of U, U' and V.

4992
is

Ex. (1).The polar reciprocal of u with respect to n

0m' = F.

Proof.Referring u, u to their common self-conjugate triangle,

w = ax1 + by*+cz', u = + j/5 + z1,
P = a (b + c) z'+b (o + a) y* + c (a+b) **.
The polar of f, n, K with respect to u is & + r)y + b, and the condition that
this may touch is be I? + ca t)* + ab i* = 0 (4664), or, which is the same
thing, {bc+ca + ab)(x% + y2+z1) = Por6it' = P (4945).
4993 Ex. (2).The enveloping conic V in (4987) may also
be written
'+' = F.

INVARIANTS AND GOVAMANTS.

689

Proof.With the same assumptions as in Ex. (1), V in (4973) becomes

(b + c)\2+(c + a) u2+(a + b) v* = 0. The trilinear equation is, therefore,
by (4667),
(c + a)(a + b) x' + (a+b)(b+c) y2 + (b + c)(c + a) * = 0,
or
(bc + ca + ab)(xi + y2 + zi) + (a+b + e)(ax' + by2 + cz2) = P.
4994

Ex. (3).The condition that F may become two right

lines is

A A' (00' AA') = 0.

Proof.Referring to Ex. (1), A = be, B = ca, C = ab, F=G=H=Q,

A'= B'= G'= 1 ; therefore, in (4981), a" = B+G = a(b+c), &c. Hence
the discriminant A of P = abc (b + c)(c + a)(a + b),
or

abc j (a + b + c)(bc+ca+ab) abc } = the above, by (4945).

4995

To reduce the two conies u, u' to the forms

rN-y+z8 = 0,

ax^+W+jZ1 = 0.

By (4945), a, /3, y will be the roots of the cubic

AF-e^-f e'jfc-A' = o

(i),

and #*, y2, z2 will be found in terms of u, v! and F, by solving

the three equations tf + if + z1, = u, ax^+fiif + yz* = ii! and (by
4994),

(i3+y)r+^(7+)/+7(+P)^ = F

(2).

4996 Ex. (1) : Given a' + 1/2 + 2y + 2s + 3 = 0 ; x2+ 2y2+4,y+2x + 6 = 0 ;

to be reduced as above. To compute the invariants, we* take
a b c f y h
= 1 1 3 1 1 0 in the first equation.
and = 1 2 6 2 10 in the second.
therefore
A B G F G H
= 2 2 1 1 1 1 in the first equation,
and = 8 5 2 2 2 2 in the second.
Therefore (4938, '9) A= 1, 0 = 6, 6'= 11, A'= 6. The roots of equa
tion (1) are now 1, 2, 3. Therefore (2) becomes 5XJ + 8F! + 9Za = P.
Computing P also by (4981) with the above values of A, B, &c, we get the
three equations as under, introducing z for the sake of symmetry,
Z*+ Y2 + '/r = x2+ tf+ 3z2 + 2yz+ 2zx,
X2+2Y2+3Z2 = x2 + 2y2 + 6r+ 4yz + 2zx,
bX2 + 8 Y2 + 9Z2 = 5x* + 8y2 + 22a2 + 16^ + lQzx,
The solution gives X = x + 1, Y=y + 1, Z 1, and the equations in the
forms required are (z + l)2+(y + l)2 + l = 0, (a; + l)J + 2 (y + l)! + 3 = 0.
4997 Ex. (2).To find the envelope of the base of a triangle inscribed
in a conic u so that two of its sides touch a.
4 T

690

ANALYTICAL CONICS.

Let
u = z% + y, + r2yz-2*x2xy2hlcxy,
and
u'=
+ 2gzx + 2Ajy,
a and y being the sides touched by . Then u + ku will be a conic touched
by the third side z. By finding the invariants, it appears that 9*4A9'
= 4AA'Ar, whence A- is determined, and the envelope becomes
(9'-4A9') ' + 4AA' = 0.
Compare (4970).
4998 The tangential equation of the two circular points at
infinity (4717) is
Xl+/t* = 0.
Proof.This is the condition that \z+fty + v should pass through either
of those points, since x iy = c is the general form of snch a line.
4999 U = 0 being the tangential equation of a conic, the
discriminant of k U+ V is
A2^+ Ae'j^+A'eH A'1.
Proof.The discriminant of kU+TJ' is identical in form with (4937),
but the capitals and small letters must be interchanged. Let then the dis
criminant be Af + ef + e'fc + A' = 0. We have
A = A' (4070), 9 = (BC-^A'+icc. = A'ab + Sx. (4668) = AG'.
Similarly 9' = A'9, A' = A'*.
5000 If 9, ' be the invariants of any conic TJ and the pair
of circular points Xs -f ^s (4998) ; then 9 = 0 makes the conic
a parabola, and 6'= 0 makes it an equilateral hyperbola.
Proof. The discriminant of kU+\'+ft' is Wb' + k (a + b) A + oi h1.
For, as above, d = A' ; 9 = ^'aA + ^fcA = (a + b) A since A' = = 1,
C &c. = 0 ; 9' = (A'B'-IP) C=G = ab-V; and A' = 0. The rest fol
lows from the conditions (4471) and (4474).
5001 The tangential equation of the circular points is, in
trilinear notation (see the note at 5030),
\2+/t2+v2-2fii' cos A -2v\ cosB 2Afi cos C.
Pkook : Xs + /is = 0, in Cartesians, shows that the perpendicular let fall
from any point whatever upon any line passing through one of the points is
infinite. Therefore, by (4(524).
5002 The conditions in (4689) and (4690), which make the
general conic a parabola or equilateral hyperbola, may be
obtained by forming 0 and 0' for the conic and equation
(5001) and applying (5000).

INVARIANTS AND COVARIANTS.

691

5003 If 6'2 = 49, the conic passes through one of the

circular points.
5004 When uf in (4984) reduces to X2+^2, that is, to the
circular points at infinity, F becomes the locus of intersec
tion of tangents to u at right angles, and produces the equa
tions of the director-circle (4693) and (4694).
5005

The tangential equation of a conic confocal with U is

kU+\2+ita=0;

5006 And if the left side, by varying k, be resolved into

two factors, it becomes the equation of the foci of the system.
Proof.Since k'+fi* represents the two circular points at infinity (4998),
liU+\3 + fi2 = 0, by (4914), is the tangential equation of a conic touched by
the four imaginary tangents of U from those points. But these tangents
intersect in two pairs in the foci of U (4720) ; and, for the same reason, in
the foci of kU+\'+fi\ which must therefore have the same foci.
If kU + X'+p* consists of two factors, it represents two points which, by
(4913), are the intersections of the pairs of tangents just named, and are
therefore the foci.

5007 The general Cartesian equation of a conic confocal

with u = 0 (4656) is
k*Au+k {C(o!2+yi)-2Gx-2Fy+A+B}+l = 0.
Proof. (5005) must be transformed. Written in full, by (4664), it
becomes (kA + 1) \*+(A;J5 + l) S + kCy*. Hence, by (4667), the trilinear
equation will be
{(IcB+l) W-k*F2} a2 + &c. = V (BG-F1) a' + kCa' + Ac.
=kIaAa' + kCa1+&c.,
(4668)
and so on, finally writing *, y, 1 for n, /3, y.

TO FIND THE FOCI OF THE GENERAL CONIC (4656).

{First Method.)

5008 Substitute in kU + Xa-f/i2 either root of its discriminant

kW+k (a-f-b) A+ab-h* = 0 (5000), and it becomes re
solvable into two factors (Xxi-|-juyi + v)(Ax2-|-/iy2 + v). The
foci are x^ and x2y2, real for one value of k and imaginary
for the other.
Proof.By (5006) the two factors represent the two foci, consequently
the coordinates of the foci are the coefficients of X, ft, v in those factors.

692

ANALYTICAL GONTGS.

(Second Method.)
5009 Let xy be a focus; then, by (4720), the equation of an
imaginary tangent through that point is ( x)+i (nj) = 0
or S+irj (x+iy) = 0- Therefore substitute, in the tangential
equation (4065), the coefficients X = 1, fi = i, v = (x+iy),
and equate real and imaginary parts to zero. The resulting
equations for finding x and y are, with the notation of (4665),
5010

2(ttr-G)2 = A [a_5+v/{4A2+(a-6)8}].

5011

2(CV-FY = A [6_+v/(4A2+(a-6)i}].

5012 If the conic is a parabola, (7 = 0, and the coordinates

of the focus are given by
(F*+G*)a; = FH+\{A-B) G,
(7^4- G2) y - GH-\{A-B) F.
5013 Ex. To find the foci of 2x1+2xy + 2y2 + 2x = 0. By the first
method, we have
a,
b,
e, f,
g,
h T from which A = 2. The quad= 2,
2, 0, 0,
1,
1 I ratio for k is
and A, B, G, F, G, H\ ^'+4^ + 3 = (2fc-3)(2fc-l) = 0,
= 0, -1, 3,
1,-2, OJ therefore = | or i.
Taking f, kU+X'+p' = f (-/^'+3, + 2^^v-4<A)+X,+/l, = 0,
or
2\1-12v\-^, + 9va + 6/i' = 0.
Solving for X, this is thrown into the factors
{ 2X + /lv/2-3 (2+ v/2) v } { 2X-/i v/2-3 (2- V2) v } .
Therefore the coordinates of the foci, after rationalizing the fractions, are
2-./2
3'
v/2-1
8~ aDd,
2+3~'
>/2
^/2 + l
5014: Otherwise, by the second method, equations (5010, '1) become, in
this instance, (3a; -t-2)1 = 2, (3y 1)* = 2, the solution of which pro
duces the same values of x and y.
5015 When the axes are oblique, the coordinates x, y of a
focus are found from the equat ions
{ C (.r+y cos <a) - F cos - G } 2 = iA ( y/P-4J+2a - 1)
(Cy-F)1 sin2<a = |A (<//2-4J-2a+/),
where I and J are the invariants (4955) and (4954) respec

693

INVARIANTS AND GOVARIANTS.

lively. The equations may be solved for x' = x +y cos u> and
y' J sinw, ivhich are the rectangular coordinates of the focus
with the same origin and x axis.
Proof.Following the method of (5009), the imaginary tangent through
the focus is, by (4721), x-\- (if ?/)(oos u + i sin w). The two equations
obtained from the tangential equation are, writing Aa for BCF1, &c.
(4668),
X1 Y*= A (a + b 2h cos w2a sins), XT = A (h sin u a sin ia cos u) ;
where X= C(x+ycoaw) Fcosio O and Y= (Gy F) sinw.
5016

ijf ^Ae equation of the conic to oblique axes be

axi+2hxy+byi+c = 0,

the equations for determining the foci reduce to

y (.r-f-y cos a) _ x(y-\-xcosu>) _
c
acosah
6cos<u h ~ abhr

5017 The condition that the line \x-\-py+vz may touch the
conic u + (\'x+ fi'y + v'z)2 is
U+<f> (fiv'ii'v, vk'v% X/*'- ty) = 0.
5018 or
(A+U')U=ni,
where

(4656, 4936, 74)

(4938)

2Tl = \'UK+ii'Ull+vUl,.

(4674)

Proof.Put a + X'1 for a, &c. in U of (4664). The second form follows

from the first through the identity
A^Oik'-^V, &c.) = uu'-n\
5019 Otherwise, let F = ux,x-\-uy/y + u2,z, the polar of
x', y, z' (4659), then the condition that F may touch u + F'*
becomes, in terms of the coordinates of the poles,
5020

(i +") ' = W+M'

(See 4657)-

Proof.If we put ur, uy.,

from (4659), for X, ft, v in U to obtain the
condition of touching, the result is Am' ; and similar substitutions made in
II give A
+ &c.), therefore (5018) becomes (l + 'i")' = (f*" + &c.).
5021

The condition that the conies

u + (X'* + Jy + v'z)\

u + (\"x+n"y + v"z)'

may touch each other is

(A+ U')(A+^ U") = (A n)2.

(4938-74)

ANALYTICAL CONIGS.

694

Proof. Make one of the common chords

(Xx+ft'y+r't) (\"x+p"y + r"*)
touch either conic by substituting X' X" for X, &c. in (5018). The result
is (A+ U')(U' 2U+U") = (U' Uf, which reduces to the form above.
5022 The condition, in terms of the coordinates of the poles
of the two lines, is found from the last, as in (5019), and is

5023 The Jacobian, J, of three conies u, v, w, is the locus

of a point whose polars with respect to the conies all meet in
a point. Its equation is
Uix+ihy+giz,

<h*+h%y+gt*>

w+fhy+gs
= 0.

gix+fiy+ci z>

giV+fzy+ciZy

gsv+fsy+csz

Pkoop.The equation is the eliminant of the equations of the three

polars passing through a point
viz., ?tI|+u,if + tf=0,
+ tyj+t>1 = 0,
w^+wt,+w,i=-0. See (4657) and (1600).

5024 The equation of a

conic passing through five
points a^iji, a.,(3y2, &c. is
the determinant equation
annexed ; and the equation
of a conic touching five
right lines X,ju1v1, X^v^, &c.
is the same in form, X, n, v
taking the place of a, /3, 7.

a2 fl2 y2 fly y a afi

! ft y* Ayi yiai aiA
& yl Ay* yi. "iA
a* #; y* Aty3 yso-j 3&

= o.

$ y! Ay* y^a a*ft

aj >8s yl Ays yisa5 *5&

Proof.The determinant is the eliminant of six equations of the type

(4656) in the one case and (4665) in the other. By (583).
5025 If three conies have a common self-conjugate triangle,
their Jacobian is three right lines.
Proof.The Jacobian of a^ + b^f + c^2, a^ + brf + c^z2, afC1 + bsy2 + c1z1
is, by (5023), xyz = 0.
For the condition that three conies may have a common point, see
Salmon's Conic*, 0th edit., Art. 389a, and Proc: Lond. Math. Soc., Vol. IV.,
p. 404, J. Walker, M.A,

INVARIANTS AND COVARIANTS.

695

5026 A system of two conies has four covariant forms

u, u', P, J, connected by the equation
J* = F_F2 (0it'-0'it)+F (AW+A0V2)
+Fuu' (00'-3AA')-AA'V-A'AVs
+A'u*u (2A0'-02)+Au'2 (2A'0-0'2).
Proof.Form the Jacobian of u, u, and P. This will be the equation of
the sides of the common self-conjugate triangle (4992, 5025). Compare the
result with that obtained by the method of (4995).
5027 By parity of reasoning, there are four contravariant
forms U, U' V, I1 where T is the tangential equivalent of J,
and represents the vertices of the self-conjugate triangle. Its
square is expressed in terms of U, U', V and the invariants
precisely as /2 is expressed in (5026).
5028 The locus of the centre of a conic which always
touches four given lines is a right line.
Proof.Let V = 0, U' = 0 be the tangential equations of two fixed
conies, each touching the four lines ; then, by (4914), U+kU' = 0 is another
conic also touching the four lines. The coordinates of its centre will be
GJrh.Q' and ~^7, by (4402). The point is thus seen, by (4032), to lie
on the line joining the centres of the two fixed conies and to divide that line
in the ratio kC' : G.
5029 To find the locus of the focus of a conic touching four
given lines.
In the equations (5010, '1) for determining the coordinates of the focus,
write A+ kA' for A, &c, and eliminate k. The result in general is a cubic
curve. If 2, 2' be parabolas,
is a parabola having three tangents in
common with 2 and 2'. If G = C = 0 the locus becomes a circle. If the
conies be concentric, they touch four sides of a parallelogram, and the locus
is a rectangular hyperbola.
Note on Tangential Coordinates.
5030 It must be borne in mind that a tangential equation in trilinear
notation (that is, when the variables are the coefficients of a, /3, y in the
tangent line la+m/3 + ny) will not agree with the equation of the same locus
expressed in the tangential coordinates A, ft, v of (4019). Thus, to convert
equation (5001), which, for distinctness, will now be written
P + m'+n1 2mn cos A2nZoo8 B 2lm cos (7 = 0
into tangential coordinates, we must substitute, by (4023), oX, bfi, cv for
I, m, n. The equation then becomes
at\, + b'fS + ?v'-2U cos Afiv-2ca cos By\ 2ab cos C\p = 0.
Put 26c cos A = fc' + c' a2, &c, and the result is the equation as presented
in (4905).
Corrigenda.Jn (4678) and (4692) erase the coefficient 2 ; and in (4680) and (4903) supply
the factor 4 on the left of the equation.

THEORY

OF

PLANE

CURVES.

TANGENT AND NORMAL.

5100 Let P (Fig. 90) be a point on the curve AP ; PT, PN,

PO, the tangent, ordinate, and normal intercepted by the z
axis of coordinates. See definitions in (1160). Let PTX
= 4.
5101

tan V = ^, by (1403) ;

5104

Sub-tangent NT = yxt,

sin i/r =

cos* =

Sub-normal NG = yyx.

5106

PT = y^0T^)f

PT'=xV(l+yi).

5108

PG=y v/(I+^),

PG'=^V(1+^).

Let 0P = r (Fig. 91), ti = r~\ AOP = 0,

Arc AP = a. Then, by infinitesimals,
5110
5113
5114

5115

s'md> = r^,
as

cos<&=^,
as

(dxy+(dyy = (d*y,

OPT=<$;

tan i = r
dr

*x = v/(i+^)-

tan* = r'9iu*+rc'8*
rt cos 0 r sin 0

(1768)

Intercepts of Normal 0(i = r^-, OG'= r$. (Fig.90)

TANGENT AND NORMAL.

Proof:

697

QG = r sin 0PO = rC08 =r- = r^-.

sin PGN
sin NPT
x,
dr,

5116

se = </{r*+rl),

*, = ^(1+^).

Peoof.By r0, = sin ^ and tan <p = rOr (5110).

EQUATIONS OF THE TANGENT AND NORMAL.
The equation of the curve being y =/() or = 0
= 0, the equation of the tangent at xy is
5118

v-y = ^(f-*)

5119

or

5120

or

f</x 17 =

<412)

y,

fw,+T?w, = xux+yuy.

(1708)

5121 If 0 (a;, y) = vH + vn.i+ ... +v0, where vn is a homogeneous function

of x and y of the nth degree, the constant part forming the right member of
equation (5120) takes the value
_, 2t>_a . . . (n 1) t), 7ii>0.
By Enter's theorem (1621) and <p (x, y) = 0.
The equation of the normal at xy is
5122

V~y = -^(t-x),

5123

or

frt+y = xxy+y>

5124

or

^uyqux = xuyyux.

(4122)

(1708)

POLAR EQUATIONS OF THE TANGENT AND NORMAL.

Let r, 0 be the coordinates of P (Fig. 91), and B, 9 those
of 8, any point on the tangent at P ; and let u = r-1, U= B'1,
r = 0 0; the polar equation of the tangent at P will be
5125

il = -5-;
;
ae (r sin t)

Or

1/ = U COS T+ Ut sin T .

The polar equation of the normal is

5127

R = -r-r^r,
afl(r cost)

or

U = u cos r - u*0 sin r .

r>D.I,

_
OP _
OSP- _
,
. * = r0r
..
ifkoof.JJrom
-r =
= sin
.
= sin (* r)' , and, from
tan
Jtt
US sin OPS
sin <p
(5112). Similarly for the normal.
4 U

THEORY OF PLANE CURVES.

698

Let OY=p be the perpendicular from the pole upon the

tangent, then
5129

P = r sin

= (u2+t$-.

5131

vm = M^r

(5112)

OS, drawn at right angles to r to meet the tangent, is

called the polar sub-tangent.
5133

Polar sub-tangent

= r20r.

(5112)

RADIUS OF CURVATURE AND EVOLUTE.

Let , v be the centre of curvature for a point xy on the

curve, and p the radius of curvature ; then
5134

(x-sy+(y-vy = P*

(i).

5135

(*-Z) + (y-rj)yx = Q

(2),

i+yi+b/-v)y** = o

(3).

Proof. (2) and (3) are obtained from (1) by differentiating for x, con
sidering , jj constants.
The following are different values of p :
5137

p-

5139

=
yn xt **2< yt

5144

5146

(#+#)f

1
yu&

^uVt

=
= /*+jp2+ = rr,.

Proofs.For (5137), eliminate x and w ij between equations (1),

(2), and (3).
(5138) is obtained from the preceding value by substituting for yz and

RADIUS OF CURVATURE AND EVOLUTE.

699

the values (1708, '9). The equation of the curve is here supposed to be
of the form <f> (x, y) = 0.
For (5139) ; change the variable to t. For (5140) ; make t = s.
For (5141-3) ; let PQ = QR = ds (Fig. 92) be equal consecutive elements
of the curve. Draw the normals at P, Q, R, and the tangents at P and Q to
meet the normals at Q and R in T and S. Then, if PN be drawn parallel
and equal to QS, the point N will ultimately fall on the normal QO. Now
the difference of the projections of PT and PN upon OX is equal to the pro
jection of TN. Projection of PT = dsx, ; that of PN or QS = ds (x, + xu ds)
(1500); therefore the difference = dsxuds = TN cos a. But TN : ds =
ds : p, therefore pxu = cos a. Similarly py = sin a.
For (5144) ; change (5137) to r and 6, by (1768, '9).
(5145) is obtained from p = rr = -f and (5129) ; or change (5144)
from r to w by r = a"1.
(5146.) In Fig. (93), PQ = p, PP' = ds, and PQP' = <ty.
(5147.) In Fig. (93), let PQ, P'Q be consecutive normals ; PT, FT'
consecutive tangents ; OT, OT', ON, ON' perpendiculars from the origin
upon the tangents and normals. Then, putting p for OT = PN, q for
PT = ON, and d+ for / TPT' = PQP', &c, we have
q = %>
(5148.)

QN=%'

and p = fQ=p+(1n = p+p*+-

dp = r cos <p d\p and cos</> = r,.

Eliminate cos^li.

5149 Def.The evolute of a curve is the locus of its centre

of curvature. Eegarding the evolute as the principal curve,
the original curve is called its involute.
5150

The normal of any curve is a tangent to its evolute.

Proof.By differentiating equation (5135) on the hypothesis that { and

7i are variables dependent upon x, and combining the result with (3), we
obtain yxt)( = 1.
In (Fig. 94), the normal at P of the curve AP touchos the evolute at Q.
Otherwise the evolute is the envelope of the normals of the given enrve.
If xy and are the points P, Q, we have the relations
5X51

t^di+v_zJLd
p
p

di_=*n_^dp
i\-y
p

Peoof.Take Qn = di, and ns = dt), then Qs = dp.

Qn, ns gives dp in (5151) and proportion gives (5152).

The projection of

5153 The evolute and involute are connected by the for

mulae below, in which r' , p', s' in the evolute correspond to
r, p, s in the involute.
5154

ps' = constant;

p'^^-p2;

r'2 = i*+p'-2pp.

THEORY OF PLANE CURVES.

700

Proof.From Fig. (94), dp = ds', &c, * being the arc RQ measured

from a fixed point R. Hence, if a string is wrapped upon a given carve, the
free end describes an involute of the curve. (3155, '6) from Fig. (93).

5157 To obtain the equation of the evolute ; eliminate x and

y from equations (5135, '6) and the equation of the curve.
5158 To obtain the polar equation of the evolute ; eliminate
r and p from (5156) and (5157) and the given equation of the
curve r = F(p).
5159

Ex.To find the evolute of the catenary ? = -( + ). Here

m
yx=\ {e~c -e'7) = v/^'~c'^ ; yu = i (e7 + e ' ) = ^ ; so that equations
c
2c
<f
(5135, '6) become
0-Q + (yt) ^(y1-6*) = 0
C

and

l + ^ + (y-,)^ = 0.
c
<r

From these we find y=^, x = {

4c1).

Substituting in the

equation of the curve, we obtain the required equation in and q.

INVERSE PROBLEM AND INTRINSIC EQUATION.

An inverse question occurs when the arc is a given func

tion of the abscissa, say s = <p (x) ; the equation of the curve
in rectangular coordinates will then be
5160

y =j

1) dd\

[From (5113).

5161 The intrinsic equation of a curve is an equation inde

pendent of coordinate axes. Let y = <p (x) be the ordinary
equation, taking for origin a point 0 on the curve (Fig. 95),
and the tangent at 0 for x axis. Let s = arc OP, and \f* the
inclination of the tangent at P ; then the intrinsic equation of
the curve is
5162
where

* = j sec fjcf rfi/r ;

is found from tan \f* = <(>'(x).

ASYMPTOTES.

701

. To obtain the Cartesian equation from the intrinsic equa

tion :
5163 Let s = Fty) be the intrinsic equation.
between this and the equations
x = J cos t/r ds,

Eliminate 4>

y j" sin V ds.

5165 The intrinsic equation of the evolute obtained from

the intrinsic equation of the curve, 8 = F(\p), is
ds

*'=/,

a constant.

(5154)

5166 The intrinsic equation of the involute obtained from

s'= F(\f,), the equation of the curve, is
* = j{Z-F(*)}#.
For d\p is the same for both curves (Fig. 94), ^ only differing
by \ir, and s =pdip.

ASYMPTOTES.

5167 Def.An asymptote of a curve is a straight line or

curve which the former continually approaches but never
reaches. (Vide 1185).
GENERAL RULES FOR RECTILINEAR ASYMPTOTES.
5168 Rule I.Ascertain if yx has a limiting value when
x. = oo . If it has, find the intercept on the x or y axis, that
is, x yxy or y-xyx (5104).
There will be an asymptote parallel to the y axis when yx is
infinite, and the x intercept finite, or one parallel to the x axis
when yx is zero and the y intercept finite.
5169 Rule II. When the equation of the curve consists of
homogeneous functions of x and y, of the mth, nth, Sfc. degrees,
so that it may be written
^(^)+^X(|-)+&c. = 0

(1);

702

THEORY OF PLANE CURVES.

put ftx+ftfor j and expand f(n+

&c., by (1500).

Divide

(1) by xm, and make x infinite; then

= 0 determines ft.
Next, put this value of (i in (1), divide by xm-1, and make x
infinite; thus ft<l>' (p) + \ (p) = 0 determines ft.
last equation be indeterminate, then

Should the

W00+rV00+*00=o
gives two values for ft, and so on.
When n is < m 1, ft = 0, and when n is > m 1, /3 = oo.
5170 Rule III. If <j> {x, j) = 0 be a rational integral equa
tion, to discover asymptotes parallel to the axes, equate to zero
the coefficients of the highest powers of x and j, if those co
efficients contain j or x respectively.
To find other asymptotesSubstitute px + ft for j in the
original equation, and arrange according to powers ofx. To
find p, equate to zero the coefficient of the highest power of x.
To find ft, equate to zero the coefficient of the next power of x,
or, if that equation be indeterminate, take the next coefficient in
order, and so on.
5171 Rule IV.If the polar equation of the curve be r = f (0)
and if r = oo makes the polar subiangent r*0r = c, a finite quan
tity, there is an asymptote whose equation is rcos(0 a) =c;
where a + \v = f-1 (oo ) = the value of 0 of the curve when r is
infinite.
5172 Asymptotic curves.In these the difference of corres
ponding ordinates continually diminishes as x increases.
As an example, the curves y = f (x) and y = f (z) + are asymptotic.
5173

Ex. 1.To find the asymptotes of the curve

(a + SzXz'+y1) = 4x
(1).
The coefficient of i/', a + 3x = 0, gives an asymptote parallel to the y axis.
Putting y = fix + fi, (1) becomes
(a + 3a!) (3? + /iW + 2/i/3 + 1?) - 4a> = 0
(2).
The coefficient of as*, 3 (1

4 = 0 gives fi = ^. Substituting this

v
6/3 and this, equated
value of fj. in (2), the coefficient of as1 becomes j-;
to zero, gives /3 = =F 2<i . Hence the equations of two more asymptotes
o ^/o
are 3y^3 = (3x-2a).

SINGULARITIES OF CURVES.

703

Ex. 2.To find an asymptote of tbe curve r cos 6 = a cos 28. Here
^ d$ _
o1 cos 28
dr
a cos 28 sin 8 2a sin 26 cos 8'
When r = co , 8 = Jt, and r5 6r = a. Hence the equation of the asymp
tote is r cos 8 = a.

SINGULARITIES OF CURVES.
5174 Concavity and Convexity.A curve is reckoned convex
or concave towards the axis of x according as yy^ is positive
or negative.

POINTS OF INFLEXION.
5175 -A- point of inflexion (Fig. 96) exists where the tangent
has a limiting position, and therefore where yx takes a maxi
mum or minimum value.
5176

Hence yte must vanish and change sign, as in (1832).

5177 Or, more generally, an even number of consecutive

derivatives of y = <p (x) must vanish, and the curve will pass
from positive to negative, or from negative to positive, with
respect to the axis of x, according as the next derivative is
negative or positive.
[See (1833).

MULTIPLE POINTS.
5178 A- multiple point, known also as a node or crunode,
exists when yx has more than one value, as at B (Fig. 98). If
^ (x, y) = 0 be the curve, $x and $y must both vanish, by
(1713). Then, by (1704), two values of yx determining a
double point, will be given by the quadratic
^+2^+^=0

(1).

5179 If <p-ix, <piy, <pxy also vanish; then, by (1705), three

values of yx, determining a triple point, will be obtained from
the cubic
<f>Syfx+S<f>2yxyl+3^x+<f>Sx =0
(2).
5180

Generally, when all the derivatives of <f> of an order

704

THEORY OF PLANE CURVES.

less than n vanish, the equation for determining yx (put = z)

may be written
(*(VW*(*%y) = 0Proof.Let ab be the multiple point. Then, by (1512),
<p (a + h, b + k) = . (W.+W,)"f (x, y)
nl
+ terms of higher order which vanish when h and A; are small.
And
=
=
in the limit.
/t
0y
ax
CUSPS.
5181 When two branches of a curve have a common tan
gent at a point, but do not pass through the point, they form
a cusp, termed also a spinode or stationary point.
5182 In the first species, or ceratoid cusp (Fig. 100), the
two values of yte have opposite signs.
5183 In the second species, or ramphoid cusp (Fig. 101),
they have the same sign.
CONJUGATE POINTS.
5184 A conjugate point, or acnode, is an isolated point whose
coordinates satisfy the equation of the curve. A necessary
condition for the existence of a conjugate point is that <j>x and
$y must both vanish.
Proof.For the tangent at such a point may have any direction, there
fore -x is indeterminate (1713).

5185 There are four species of the triple point according as

it is formed by the union of
(i.)
. (U.)
(iii.)
(iv.)

three crunodes, as in (Fig. 102) ;

two crunodes and a cusp, as in (Fig. 103) ;
a crunode and two cusps, as in (Fig. 104) ;
when only one real tangent exists at the point.

5186 Ex. The equation y* = (x a)(x b)(x c)* when

a < b < c represents a curve, such as that drawn in (Fig. 97).
* Salmon's Higher Plane Curvet, Arts. 39, 10.

705

When b = c the curve takes the form in (Fig. 98). But

if, instead, b = a, the oval shrinks into a point A (Fig. 99).
If a = b = c the point A becomes a cusp, as in (Fig. 100).
A geometrical method of investigating singular points.
5187 Describe an elementary circle of radius r round the
point x, y on the curve <j> (x, y) = 0, intersecting the curve in
the point x + h, y+k. Let h = r cos 0, k = r sin 0. Expand
<p (x+h, y+k) = 0 by (1512), and put <f>x = K siny, <py =
K cos y. We thus obtain
K sin (y + 0) + | (fc, cos2 0 + 2<j>xy sin 0 cos0 + <p2y sin2 0) + = 0
R being put for the rest of the expansion
According as the quadratic in tan 0,

(1)>

02* + fyxy tan 0 + <j,2y tan2 0 = 0,

has real, equal, or imaginary roots ; i.e., according as
flyfufiy is positive, zero, or negative, xy will be a crunode,
a cusp, or an acnode. By examining the sign of R, the
species of cusp and character of the curvature may be deter
mined.
Figures (105) and (106), according as R and <f>iy have
opposite or like signs, show the nature of a crunode ; and
figures (107) and (108) show a cusp.
Proof.At an ordinary point the circle cuts the curve at the two points
given by 8 = y, 8 = ir y. But, if fx and tyy both vanish, there is a
singular point. Writing A, B, C for <p2x, <j>X!/, ty-^, equation (1) now becomes
Ccos'ejtan'fl+^tane+^l +^=0

(2).

(i.) If -B,>-401this may be put in the form

9 7?
C cos1 0 (tan 0 - tan a) (tan 0 - tan #) + ~ = 0,
r
and the points of intersection with the circle are given by 0 = a, (i, ir + a,
andjr+/3. (Figs. 105 and 106.)
(ii.) When B* = AC, we may write equation (1)
C cos* 0 (tan 9 -tan a)s+ ^ = 0.
r
If B and C have opposite signs, there is a cusp with a for the inclination of
the tangent (Fig. 107). So also, if i2 and C have the same sign, the inclina
tion and direction being n + a (Fig. 108). The cusps exist in this case
because B changes its sign when ir is added to 0, R being a homogeneous
function of the third degree in sin 0 and cos 0.
(iii.) If B*<AO, there are no real points of intersection, and therefore xy
is an acnode.
4 X

THEORY OF PLANE CURVES.

706

CONTACT OF CURVES.
5188

A contact of the ntb order exists between two curves

when n successive derivatives, yx, ... yM or rfl, ... r^, corres

pond. The curves cross at the point if n be even. No curve
can pass between them which has a contact of a lower order
with either.
Ex.The curve y = <f> (z) has a contact of the n"1 order, at the point
where as = a, with the curve y = f (a) + (x a) ^> (a) + ... + * p*- ^"(a).
5189 Cob.If the curve y=f(x) has n parameters, they
may be determined so that the curve shall have a contact of
the (n l)th order with y = <f> (x).
A contact of the first order between two curves implies a
common tangent, and a contact of the second order a common
radius of curvature.

Conic of closest contact with a given curve.

5190

Lemma.In a central conic (Fig. of 1195),

tan CPG = i
3 as

Pkoof. Putting PCT = 9, CPT = <p, GP = r, CD = R, we have, by

(1211),
r, + R* = a* + b\ .:rr, = -RR,
(i.).
Also
Rr sin <p = ab, by (1194), .-. R^B, = ab, by (5110)
(ii.).
Now
But

tan CPO = -cot = - It (5112) =

=
by (i.) and (ii.).
r
rd,
ab .
P= ~ (4538),
ab

.-. 4- 4r = ^ = tan CPG.

6 as
ab

5191 To find the conic having a contact of the fourth order

with a given curve at a given point P.
If 0 be the conic' s centre, the radius r = OP, and the
angle v between r and the normal are found from the equations
.
1 dp
tan v = -A
3 ds
and these determine the conic.

cos*
1
=
r
p

dv
ds

ENVELOPES.

707

Proof.In Fig. 93, let 0 be the centre of the conic and P the point of
contact. The five disposable constants of the general equation of a conic
will be determined by the following five data : two coordinates of 0, a com
mon point P, a common tangent at P, and the same radins of curvature PQ.
Since v-POT, dO = POP', d^=TOT, and ds = PP', we have, in
passing from P to P", dv = P'OTPOT = d^-d6. Now rdd = ds cos v,
therefore 008 v = = ~- : and tan v has been found in the
lemma.
r
dscosd
*
ds
The squares of the semi-axes of the same conic are the
roots of the equation
(9+62-3ac)sa?a-9a2 (18+262-3ac)(9+62-3ac) *+729o*
= 0,
a, b, c being written for p, p pu.
from

The eccentricity is found

9(e-2)' _ (18+26*-3aC)*
1-e1 ~
9+6*-3ac '

Also the equation of the conic referred to the tangent and

normal at the point is
Aa*+2Bay+Cy* = 2y,
where A = \ B = -f, tf = I + |??_^.
p
op
p
yp
6
Ed. Times, Math. Reprint, Vol. XXI., p. 87, where the demonstrations by
Prof. Wolstenholme will be found.

ENVELOPES.

5192 An envelope of a curve is the locus of the ultimate

intersections of the different curves of the same species, got
by varying continuously a parameter of the curve ; and the
envelope touches all the intersecting curves so obtained.
5193 Rule.If F (x, y, a) = 0 be a curve having the para
meter a, the envelope is the curve obtained by eliminating a
between the equations
F (x, y, a) = 0

and

daF

y, a) = 0.

THEORY OF PLANE CURVES.

708

Proof.Let o change to a + h. The coordinates of the point of intersec

tion of F (x, y, a) 0 and F (x, y, a+h) = 0 satisfy the equation
F(x, y, a+h)-F(x, y, o) _ Q ^ dF(x, y, a) _ Q
(1404)
h
'
da

5194 If F(x, y, a,b,c, ...) = 0 be the equation of a curve

having n parameters a,b,c, ... connected by n 1 equations,
then, by varying the parameters, a series of intersecting
curves may be obtained. The envelope of these curves will
be found by differentiating all the equations with respect to
a, b, c, &c, and ehminating da, db, ... and a, b, ...
5195 Ex.In (2) of (5135), we have the equation of the normal of a
curve at a given point xy ; 4, i/ being the variable coordinates, and x, y the
parameters connected by the equation of the cnrve F (as, y) = 0. By differ
entiating for x and y, (5136) is found, and the elimination as directed in
(5157) produces the equation of the evolute which, by (5194), is the envelope
of the curve.

INTEGRALS OF CURVES AND AREAS.

FORMULA FOR THE LENGTH OF AN ARC *

5196

*=^d*= jV(l+i) dx = jVl+i; dy

(5113)

5200

=y/ty+tf)dt=fy(t>+7>&d$

(5116)

5201

=$^+l)dr=$v0^-ry

(5111)

5203

Legendre's formula, s = pt+^pdiji.

5204

The whole contour of a closed curve = J pdty.

Proof.In figure (93), let P, F be an element ds of the curve ; PT, P'T'

tangents, and OT, Of the perpendiculars upon them from the origin ; OT=.p,
PT = q. Then ds + P'T'-PT = TL, i.e., ds+dq = jxty; therefore s + q
= \pd\js. But qd*j> = dp; therefore s p++\pd>j/. Also, in integrating
all round the curve, P'TPT taken for every point vanishes in the summa
tion, or

= 0.

Therefore J ds = | pdip.

INVERSE CURVES.

709

FORMULAE FOR PLANE AREAS.

5205 If y = (*) be the equation of a curve, the area
bounded by the curve, two ordinates (x = a, x = b), and the x
axis, is, as in (1902),
A = \ <f> (of) da?.
5206 With polar coordinates the area included between two
radii (9 = a, 9 = (3) and the curve is

Proof.From figure (91) and the elemental area OPP'.

5209 The area bounded by two circles of radii a, b, and the
two curves 0 = f (r), 0 = i/, (r) (Fig. 109).
J =rr<rWrd0 = fr {^(r)-<^(r)} dr.

Here r {^(r) <f>(r)} dr is the elemental area between the

dotted circumferences.
5211 The area bounded by two radii of curvature, the curve,
and its evolute (Fig. 110).
A = jVety = ij^d*.
Proof.From figure (93) and the elemental area QPP'.

INVERSE CURVES.
The following results may be added to those given in Arts. (1000-15).
5212 Let r, v' be corresponding radii of a curve and its
inverse, so that rr' = k2 ; s, s' corresponding arcs, and <j>, <f>'
the angles between the radius and tangents, then
4^- = r
as
r

and

d> = <'.

Proof.Let PQ be the element of arc ds, P'Q' the element els', and O
the origin.
Then OP. OF = OQ.OQ', therefore OPQ, OQ'P' are similar triangles;
therefore PQ : FQ' :: OP : OQ' = r : r ; also Z OPQ, - OQ'F.

THEORY OF PLANE CURVES.

710

5214

If p, p be the radii of curvature,

^ + 4= 2 sin 6.
P
P
Pboof.From p = r sin </>, p' = r sin tf>, we have
p'=
therefore
= k'^~2rP
Also

r=

therefore

(L).

= -5-

(ii.).

Now p' = r' (5148), therefore 1 = 1*'^ = *--^, by (i.) and (ii.).
dp
p
r dr dr
rr pr
Therefore

I7 + - = ? = 2sin$>.
p
p
r

5215 To find the equation of the inverse of a curve in

rectangular coordinates, substitute
gg

.and

fe2.y

for a; and y in the equation of the given curve.

5216

The inverse of the algebraic curve

where un is a homogeneous function of the ntb degree, will be

ft*u+*,<-X_a (*+y)+*(-1>u_1
...+Q(^+y2)n=0.

...

5217

The inverse of the conic u2 + u1+u0 = 0 is

5218

If the origin be on the curve, this equation becomes

tfut+u^xt+y*) = 0.

5219 The angle 4 will also be unaltered in any curve,

r =/(0), if the inversion be effected by putting
r = &r'n and 6 nff.
Proof.
tan <p' = r'8% (5112) = r'^rr = r'^knr'-1 = fcr'0r = r0T = tanf

PEDAL CURVES.

711

PEDAL CURVES.
5220 The locus of the foot of the perpendicular from the
origin upon the tangent is called a pedal curve. The pedal of
the pedal curve is called the second pedal, and so on. Re
versing the order, the envelope of the right lines drawn from
each point of a curve at right angles to the radius vector is
called the first negative pedal, and so on.
5221

The pedal and the reciprocal polar are inverse curves

(1000, 4844.)

AREA OP A PEDAL CURVE.

5222 Let G, P, Q be the respective areas of a closed curve,
the pedal of the curve, and the pedal of the evolute ; then
P-Q=C,

P+Q = jV#,

2P = C+ijWf

Proof.With figure (93) and the notation of (5204), we have, by (5206),

P=i
Q = \ J q'd+ ; therefore P+ Q = J j (p' + 2*) <ty = | J r*ty.
Also, taking two consecutive positions of the triangle OPT = A, we get
OPT- OP'T = SA = SC+SQ-SP. Therefore, integrating all round,
IdA = 0 = G+Q-P.

5225 Steiner's Theorem.If P be the area of the pedal of a

closed curve when the pole is the origin, and P the area of
the pedal when the pole is the point xy,
P'-P = &r(a+yt)-ax-by,
pcosddO and
o
0 being the inclination of p.

b= \ psinOdO;
Jo

Proof.(Fig. 111.) Let LM be a tangent, S the point xy, perpendiculars

OM=p and 8R=p'. Draw 8N perpendicular to OM, and let ON = p, ;
then P' = i j jAty = i J (p -Piy <ty = \ J fty + i j p\ d+ - ^ppx cty
= P+ OS1- Pp (z cos 8 +y sin fl) dd, by (4094), and dfl = d+.
2
Jo
And | I pjd^ = twice the area of the circle whose diameter is 08.

712

THEORY OF PLANE CURVES.

5226 Cob. 1.If P be given, the locus of xy is a circle

whose equation is (5225), and the centre of this circle is the
same for all values of P, the coordinates of the centre being
and .
7T

TT

5227 Cob. 2.Let Q be the fixed centre referred to, and

let QS = c. Let P" be the area of the pedal whose origin is
Q; then
P'-P" = brc\
For a and b must vanish in (5225) when the origin is at the
centre Q. and x2+y2 then = c8.
5228

Cob. 3.Hence P" is the minimum value of P\*

ROULETTES.
5229 Def.A Roulette is the locus of a point rigidly con
nected with a curve which rolls upon a fixed right line or
curve.
AREA OF A ROULETTE.

5230 When a closed curve rolls upon a right line, the area
generated in one revolution by the normal to the roulette at
the generating point is twice the area of the pedal of the
rolling curve with respect to the generating point.
Proof.(Fig. 112.) Let P be the point of contact of the rolling curve
and fixed Btraight line, Q the point which generates the roulette. Let R be
a consecutive point, and when R conies into contact with the straight line,
let P'Q' be the position of RQ. Then PQ is a normal to the roulette at Q,
and P is the instantaneous centre of rotation. Draw QN, QS perpendiculars
on the tangents at P and R. The elemental area PQQ'P', included between
the two normals QP, Q'P', is ultimately equal to PQR+QRQ'. But PQR
= dO, an element of the area of the curve swept over by the radius vector
QP or r round the pole Q ; and QRQ' = %r'd4> ; therefore, whole area of
roulette = O+i I ><fcj = 2P, by (5224).

5231 Hence, by (5228), there is one point in any closed

curve for which the area of the corresponding roulette is a
* For a discussion of the pedal curves of an ellipse by the Editor of the Educ. Timet and
others, see Reprint, Vol. I., p. 23 ; Vol. xvi., p. 77 ; Vol. xvii., p. 92 ; and Vol. xx., p. 106.

ROULETTES.

713

minimum. Also the area of the roulette described by any

other point, distant c from the origin of the minimum roulette,
exceeds the area of the latter by irc\
5232 When the line rolled upon is a curve, the whole area
generated in one revolution of the rolling curve becomes

where p, p are the radii of curvature of the rolling and fixed

curves, and G is the area of the former.
Proof.(Fig. 113.) Instead of the angle d\fi, we now haveWhe sum of
tho angles of contingence at P of the rolling curve and fixed curve, viz.,
d^ + #' = #(l + |^) =#(l+ A),
since pd^i = ds = p'dx)/', hy (5146).
LENGTH OF THE ARC OF A ROULETTE.
5233 If " and I be corresponding arcs of the roulette and
the pedal whose origin is the generating point ; then, when
the fixed line is straight, a = I; and when it is a curve,
5234

Jd<r = J(l+A)rf{.

Proof. (Fig. 112.) Let B be the point which has just left the straight
line, Q the generating point, N, S consecutive points on the pedal curve.
Draw the circle circumscribing BQNS, of which HQ = r is a diameter, and
let the diameter which bisects NS meet the circlo in K. Then, when tho
points P, B, P' coincide, KN and BQ are diameters, and SKN = SPN = d\p
= QBQ' ; therefore SN or d = rd\\i = QQ' or da. When the fixed line is a
curve, d* = rdf (l+
as in (5232).

RADIUS OF CURVATURE OF A ROULETTE.

5235 Let a (Fig. 113) be the angle between the generating
line r and the normal at the point of contact ; p, p' the radii
of curvature of the fixed and rolling curves, and B the radius
of curvature of the roulette ; then,

714

THEORY OF PLANE CURVES.

Proof. Let consecutive normals of the roulette meet in 0 ; then

OQ=R, PQ = r, MPT=a.
Rr PH ds cos a
, ,
,,. . j ,^
I ds . ds\
from which R is obtained. If the curvature of the roulette is convex to
wards P (Fig. 114), we must write R + r instead of Rr above.
5236 The curvature is convex towards P when B is posi
tive, that is, when the carried point Q falls within the circle
whose diameter measured on the normal of the rolling curve
= p? ,. When Q falls without this circle, the curvature is
P+P
concave ; and when Q falls upon the circumference, the point
is one of inflexion. The circle has for this reason been called
the circle of inflexions.
5237 In figure (163) let PA = P, PB = P', PQ = r, OQ=B,
as in (5235). Draw PCD, the circle of inflexions, with its
diameter PC = pp and therefore PD = pp ;COSa. From
P+P
P+P
these values and proportion it follows that BG : BP : BA and
QD : QP : QO. Also, if the circle on diameter PE = PC be
drawn, AE : AP : AB and OF : OP : OQ.
5238 A simple construction for the centre of curvature of
the roulette is the following. (Fig. 164, with letters as in
5237.) At P draw a perpendicular to PQ to meet QB in N.
Join NA, which will meet QP produced in 0, the required
point.
Proof.From equation (5235), assuming 0 to be the centre of curvature,
we can deduce the relation (BA : AP)(PO : OQ)(QN : NB) = 1, therefore,
by (908), A, 0, N are collinear points.
THE ENVELOPE OF A CARRIED CURVE.
5239 When a curve is rigidly connected with a rolling
curve, it will have an envelope. The path of its point of
contact with the envelope is a tangent to both curves, and
therefore the normal, common to the carried curve and its
envelope, passes through the point of contact P of the rolling
and fixed curve.
5240 The centre of curvature of the envelope is obtained as
follows.

TRAJECTORIES, tfc.

715

In Fig. (163), from P draw a normal to the carried curve meeting it in

Q, and let 8 on PQ be the centre of curvature of the envelope for the point
Q ; and 0 that of the carried curve. Then PS is found from
1 jl
7 = COS a /( 1 ,Y 1 V
1 .
P p
\PS T POJ

5241 When the envelope is a right line, the centre of curva

ture lies on the circle of inflexions (5236). When the carried
curve is a right line, the same point lies on the circle PEF
(Fig. 163), and if the right line always passes through a fixed
point, that point lies on the circle PEF.
5242 If p be the perpendicular from a fixed point upon a
carried right line whose inclination to a fixed line is ^ ; the
radius of curvature of the envelope is p = ^> +j>2+> by (5147).

INSTANTANEOUS CENTRE.

5243 When a plane figure moves in any manner in its own

plane, the instantaneous centre of rotation is the intersection
of the perpendiculars at two points to the directions in which
the points are moving; and a line from the instantaneous
centre to any point of the figure is the normal to the path of
that point.
Ex.Let a triangle ABO slide with its vertices A, B always upon
the right lines OA, OB. The perpendiculars at A, B to OA, OB meet in Q,
the instantaneous centre, and QO is the normal at G to the locus of G.
Since AB and the angle AOB are of constant magnitude, OQ, the
diameter of the circle circumscribing OAQB, is of constant magnitude.
Hence the locus of the instantaneous centre Q is a circle of centre 0 and
radius OQ.

5244 Holditch's Theorem.If a chord of a given length LM

moves completely round a closed curve, the area enclosed
between the curve and the locus of a point P on the chord is
equal to vcc' where c = LK, c' = MK.
5245 If the ends of LM move on different closed curves
whose areas are A, p, while the area described by K is k, then
Xc'+ixc
/
c-\-c
Proof.(5244). Let the innermost oval in figure (134) be the envelope
of LM, e its area, and E the point of contact. Let EL= I, EM = m,

THEORY OF PLANE CURVES.

716

EKk, l+m = a = c + c' ; 0, the inclination of LM. Then, integrating

in every case from 0 to 2t,
ijpifl =\-e) .: i$(P-m,)dO = (l-m)dO=\-m.
ijm'dfl = /i-e j
.-. ajtaa = jraJ + X-/z

Also | j(Z+m) dfl = to^

(i.)-

Similarly c JtaO = *c*+X-k

(ii.),

the last being obtained from -J

i2) dfl = X r. k is then fonnd by
eliminating the integral between (i.) and (ii.).
(5245.) If the curves X, p coincide, X = y. and therefore Xk = rcc.

TRAJECTORIES.
5246 Def.A trajectory is a curve which cuts according to
a given law a system of curves obtained by varying a single
parameter.
The differential equation of the trajectory which cuts at a
constant angle /3 the system of curves represented by
^ (x, y, c) = 0 is obtained by eliminating c between the
equations
<Ar>,, c) = 0

and

tan =

the derivatives of <j> being partial, and yt referring to the

trajectory.*
Proof. At a point of intersection we have for the given curve
m = <^x~-^y, and for the trajectory m' = yx. Employ (4070).
If the trajectory is to be orthogonal, tan /3 = co , and the
second equation becomes
fafayx = 0.
Ex.To find the curve which cuts at a constant angle all right lines
passing through the origin.
Let y = cx represent these lines by varying c ; then, writing n for tan /3,
the two equations become ycx = 0 and n (l + cyx) = yxc. Eliminating
c, *yx-y = n(yyx+x). Divide by z'+y1 and integrate; thus
tan-> ^ = n log y(*' + j/') + C,
z
9
which is equivalent to r = ae, the equation of the logarithmic spiral (5289).
* For a very full investigation of this problem, see Euler, Ifovi Com. Petrop., Vol. xrv.,
p. 46, xvii., p. 205 ; and Nova Acta retrop., Vol. L, p. 3.

CAUSTICS, Sfc.

717

CURVES OF PURSUIT.
5247 Def.A curve of pursuit is the locus of a point which
moves with uniform velocity towards another point while the
latter describes a known curve also with uniform velocity.
Let f(x, y) = 0 be the known curve, xy the moving point
upon it, >j the pursuing point, and n : 1 the ratio of their
velocities.
The differential equation of the path of j is
obtained by eliminating x and y between the equations
f(x, y) = 0

(i.),

yq = r/(

(5+10 = (1+10

(h\),
(Hi.).

Proof.(ii.) expresses the fact that xy is always in the tangent of the

path of i,t\.
(iii.) follows from 1 : n= /(<K' + oV) : </(dx% + dy') ; the elements of
arc described being proportional to the velocities.

Ex.The simplest case, being the problem nsually presented, is that in

which the point xy moves in a right line. Let x = a be this line, and let
the point ij start from the origin when the point xy is on the x axis. The
equations (i.), (ii.), (iii.) now become, since xc = 0,
x = a, y = ri + n,(al), y, = y(l + ?J)From the second ye = >?2t (o 4), therefore (a ) ij2t = v^l +

Integrating by (1928), we find

log(jj+vT+p) = ralog(a $)+loga,
so that p and 4 vanish together at the origin ;
therefore

+p*+p

an0^ therefore </l +p*p

i=H(A)"-(-?n'

the equation of the required locus, the constant being taken so that r\ = 0
together. If, however, n= 1, the integral is
42-2a4 - 77
a ,lg
a-l
1 = .4a
2
a

718

THEORY OF PLANE CURVES.

CAUSTICS.
5248 Def.If right lines radiating from a point be reflected
from a given plane curve, the envelope of the reflected rays is
called the caustic by reflexion of the curve.
Let $ (x, y) = 0, $ (x, y) = 0 be the equations of the
tangent and normal of the curve, and let hk be the radiant
point; then the equation of the reflected ray will be
<f> (h, k) * (0% y)+* (h, k) <f>

y) = 0,

and the envelope obtained by varying the coordinates of the

point of incidence, as explained in (5194), will be the caustic
of the curve.
Ex.To find the caustic by reflexion of the circle as'+y* = r*, the radiant
point being hk.
Taking for the tangent and normal, as in (4140), xcosa + y sin a = r,
and x sin a y cos a = 0, the reflected ray is
(7i cos a + 7; sin o r) (x sin a y cos a)
+ (h sin a A; cosa)(x cos a + y sin a r) = 0.
Reducing this to the form
A cos 2a + 2? sin 2a + G sin a D cos a = 0,
and differentiating for a,
-2A 8in2a + 2J5cos2a + Ocosa + Dsina = 0.
The result of eliminating a is
{4 (h, + kiXxi+yt)-7'(x + hy-r,(y + kyy = 27 (kx-hyYix' + if-h'-k1)1,
the envelope and caustic required.
5249 QueteleVs Theorem. The caustic of a curve is the
evolute of the locus of the image of the radiant point with
respect to the tangent of the curve.
Thus, in the Fig. of (1178), if S be the radiant point, W is the image in
the tangent at P. The locus of W is, in this case, a circle, and the evolute
and caustic reduce to the single point 8'.
Since the distance of the image from the radiant point is
twice the perpendicular on the tangent, it follows that the
locus of the image will always be got by substituting 2r for r
in the polar equation of the pedal, or for r in the polar
equation of the reciprocal of the given curve with respect to
the radiant point and a circle of radius k.

TRANSCENDENTAL AND OTEEB CURVES.

719

TRANSCENDENTAL AND OTHER CURVES.

THE CYCLOID.*

(Fig. 115)

5250 Def.A cycloid is the roulette generated by a circle

rolling upon a right line, the carried point being on the cir
cumference. When the carried point is without the circum
ference, the roulette is called a prolate cycloid ; and, when it is
within, a curtate cycloid.
5251

The equations of the cycloid are

,v = a (0+sin 6),

y = a(l cos 6),

where 0 is the angle rolled through, and a the radius of the

generating circle.
Proof(Fig. 115.) Let the circle KPT roll upon the line DE, the
point P meeting the line at D and again at E. Arc KP KD ; therefore
arc PT = AK = OT. Also 0 = PCT, the angle rolled through from A, the
centre of the base ED. Then
x= OT+TN = ad + asinO; y = PN = a a cosfl.
5253

If s be the arc OP and p the radius of curvature at P,

s = 2PT = v/(8ay),

p = 2PK.

Peoof.(i.) The element Pp = Bh = 2 (OB -Ob) ultimately ; therefore,

by summation, s = 20B. Also OB = PT = ^/(TK. TR) = </(2ay).
(ii.) Let two consecutive normals at P and p intersect in L. Then PL,
pi are parallel to BA, bA ; therefore PLp is similar to BAi. But Pp = 2Bi ;
therefore p or PL = 2BA = 2PK.
5255 Cob.The locus of L, that is the evolute of the
cycloid, consists of two half-cycloids as shown in the diagram.
5256 The area of a cycloid is equal to three times the area
of the generating circle, and the curve length is four times
the diameter of the same circle.
Proof. (i.) Area PptiN = PprR = BbqQ ultimately. Therefore, by
summation, DE . AO cycloid = ira'. Bnt DE.AO = 2ira.2a = iira1; there
fore cycloid = 3wa*.
(ii.) Total curve length = 8a, by (5253).
* The earliest notice of this curve is to be found in a MSS. by Cardinal de Cusa, 1461
See Labnitn, Optra, Vol. m., p. 96.

720

5257

THEORY OF PLANE CURVES.

The intrinsic equation of the cycloid is

* = 4a sin /.

Proof : = 2PT = 4a. sin PET, and PKT = PTN x/,.*

THE COMPANION TO THE CYCLOID.

5258 This curve is the locus of the point B in Fig. (115).
Its equation is
y
Pboof.From x = aO and y = a (1 cos 6).
5259 The locus of S, the intersection of the tangents at P
and B, is the involute of the circle ABO.
Proof : BS = BP = arc OB.
PROLATE AND CURTATE CYCLOIDS.
5260

(5250)

The equations in every case are

x = a (0+ m sin 6) ,

y = a (1 rat cos 6).

The cycloid is prolate when m is > 1 (Fig. 116), and curtate

when in is <1 (Fig. 117), m being the ratio of CP to the
radius a.

EPITROCHOIDS AND HYPOTROCHOIDS.

(Fig. 118)

5262 These curves are the roulettes formed by a circle

rolling upon the convex or concave circumference respectively
of a fixed circle, and carrying a generating point either within
or without the rolling circle.
The equations of the epitrochoid are
5263

= (a+b) cos 6-mb cos ^ 6,

5264

y = (a+b) sin 6-mb sin^ 6,

* For other properties, see Pascal, Hittoire de la Roulette ; Carlo Cati, History of the
Cycloid; Wallis, Traite de Cycloide ; Groningius, Historia CycUtidie, Bibliothecn Univ.; and
Lalouerc, Geometria promqfa in ttptem de Cycloide librie ; Bernoulli, Op., Vol. it., p. 98;
Euler, Comm. Pet., 1766 ; and Legendre, Exercice du Caleul. Int., Tom. n, p. 491.

TRANSCENDENTAL AND OTHER CURVES.

721

where a, b are the radii of the fixed and rolling circle

(Fig. 118), 0 is the angle OCX, Q is the generating point
initially in contact with the x axis, and m is the ratio OQ : b.
The dotted line shows the curve described. For the hypotrochoid change the sign of b.
Proof:
x= CN+MQ; CN = (a + b) cos 6;
HQ = OQ cos OQM = - OQ cos (<f> + 0), where <f> = FOR and b<b = ad.
5265

The length of the arc of an epitrochoid is

s=(a+b)^l+m2-2mcos^dd,

which is expressed as an elliptic integral E (k, <f>) by substi

tuting afl = 2b<p.
For the arc of a hypotrochoid, change the sign of b.
Proof : s = s,dd = j J(z] + yl) dO (5113). Find as, and y, from (5268-4).

EPICYCLOIDS AND HTPOCYCLOIDS.

(Fig. 118)

5266 For the equations of these curves make m = l, in

(5263, '4). P is then the generating point, and the curve is
shown by a solid bne in Figure (118).*
5267 If 4> be the inclination of the tangent at a point P on
any of these curves,

tan xfi =

5268

cos 6 m cos ^ 0
a+2b
j = tan i_ $} if m 1.
sin 0m sin 4 0

Hence, in the epicycloid, rjt =

05

and the equation of the tangent is

x sin ^i^B-y
2b
9 cos ^|^
2b 0 = v(a+2b)y sin 4
2b ft
5269

The equation of the normal will be

u-\-2b a .
. a+2b Q
a a
x cos 0+w sin-i, V = a cos-^0.
2b
9
2b
2b

* Prof. Wolstenholme has investigated these curves considered as the onvelopos of n

chord whose extremities move on a fixed circle with uniform velocities in the ratio m : or
m : ( ).Proc. land. Math. Soc, Vol. IV., p. 321.
4 z

722

TBEOBY OF PLANE CURVES.

5270 The length of the arc of an epicycloid or hypocycloid

included between two successive cusps is
(a 6). and the included area is (3a 26).
a
a
Pboof.Putting m = 1 into (5265) and ad = b<p, the length becomes
(a o) sin J- df = (a 6).
a
J0
2
a
Otherwise by (5234) ; the pedal being the cardioid whose perimeter = 8o
(5333).
(ii.) The area, by (5232), is x62 + ^j 461 sin1 -| ( 1 +

d+ ; since, in

Fig. (118), dif> of (5232) = dPOB = df and r = PB = 26 sin |-.

5271

The evolute of an epicycloid is a similar epicycloid.

Proof.The equation of the tangent referred to an x axis drawn through

the summit of the curve will be (by turning axes through an angle br-i-a),
a + 2b a .
. a + 26/,
.. ,
a a
0S 2b~
y 8m ~^Zb~
(o+26)cos 0.
Comparing this with (5270), which is the equation of the tangent of the
evolute, we see that the epicycloid and its evolute are similar curves having
their parameters in the ratio a + 26 : a; and that the radius drawn through
a cusp of either of the curves passes through a summit of the other.
5272 When b = \a, the hypocycloid becomes a straight
line, namely, a diameter of the fixed circle.

THE CATENARY.

(Fig. 119)

5273 Characteristic.The perpendicular TP from the foot

of the ordinate upon the tangent is of a constant length c, and
therefore equal to OA, the perpendicular from the origin on
the tangent at the vertex, c is the parameter of the curve.
The equation is
5274

* = (*+"*).

Pboof: tanPCT = ^ =

,c

(1928), since x = 0 when y = c.

ef=i {y+ Jtf-<?)}
c

.-. x = c flog(y + /t^?)-logc}

Therefore
therefore e"* = [y-Jtf-c1)}.
c

TRANSCENDENTAL AND OTHER CURVES.

5275

If s = arc 40,

Proof:
=J
5276

723

s = (eZ-eS) = CP.

s = ^(l+yl)dx (5197)
+ ^) * = J** = f
e-7) = y(^-cs) = CP.
The area 04O27 = cs.

(5205)

5277

The radius of curvature at 0 = *-t and is therefore

c
equal to the tangent intercepted by the axis of x.
Proof:

cos i/f = , .. sin tpip, = jy .*. p =

y
y

(5146).
q

5278 The catenary derives its name from a chain, which,

when suspended from its extremities, takes the form of this
curve.
For the equation of the evolute of the catenary, see (5159).

THE TRACTRIX.

(Fig. 119)

5279 Characteristic.The length of the tangent intercepted

by the x axis is constant. This curve is the involute of the
catenary, being the locus of P in Figure (119).
The equation of the tractrix is
5280

x = c log {c+v/^-y2)} -c\ogy-y/{f*-f).

Proof.Let the tangent FT = c, then the differential equation of the

cnrve is therefore yx = VC*y'. Substitute z = </e'y2, and integrate
by (1937).
5281

The area included by the four branches = ire2.

Proof.Area = 4,\ydx = -4> ["/c'-tfdy = ire1, by (1933).

THE SYNTRACTRIX.
5282 This curve is the locus of a point Q on the tangent of
the tractrix in Fig. (119). Let QT be equal to a given con
stant length d ; then the equation of the syntractrix will be
5283

* = c log {d+ y/^-y8)} -c log y- J{<P-tf).

724

THEORY OF PLANE CURVES.

THE LOGARITHMIC CURVE*

(Fig. 120)

5284 Characteristic. The subtangent is constant.

The equation of the curve is either
5285

y = aen,

or

v
x=n\og^-,

where n = NT, the constant subtangent, and a is the intercept

on the y axis.
5287 If n be an even integer, y may take negative values.
The most general form of the equation may perhaps be
assumed to be
/ cos 2rv .\-i. sm
. 2r7r\
y = enl
j.f,

THE EQUIANGULAR SPIRAL.

(Fig. 121)

5288 Characteristic. The angle OPS between the tangent

and radius is constant. The equation of the curve is either
5289
5291

t ae*
tan

or

= n,

r
8 = n log .
tt
s = r sec ^,

measuring s from the pole.

Proof.By (5112) and (5200).
5293 Hence the length of the spiral measured from the pole
0 to a point P (Fig. 121) is equal to PS, the intercept on the
tangent made by the polar subtangent OS.
5294 The locus of S is a similar spiral, and is also an invo
lute of the original curve.
5295 The pedal curve, which is the locus of Y, is also a
similar equiangular spiral.
Proof.The constancy of the angle <j> makes the figure OPY8 always
similar to itself. Therefore P, Y, and S describe similar curves. Hence, if
ST is the tangent to the locus of S, OST = <p = OPS ; therefore PST is a
right angle ; therefore the locus of S is an involute of the original spiral. X
* Originated by James Gregory, Geometric Part Universalis, 1668.
t See Eiiler, Anal. Infin., Vol. n., p. 290 ; Vincent, Ann. de Bergonne, Vol. xv., p. I ;
Gregory, Camb. Math. Journal, Vol. I., pp. 231, 264 ; Salmon, Higher Plane Curvet, p. 274.
X For additional properties, see Bernoulli, Opera, p. 497.

TRANSCENDENTAL AND OTHER CURVES.

THE SPIRAL OF ARCHIMEDES *

725

(Fig. 122)

5296 Characteristic.The distance from the pole is propor

tional to the angle described. Hence the equation is
5297

r = ad.

Also tan < = 0.

By (5112).

5299 The intercept, PQ, on any radius between two succes

sive convolutions of the spiral, is constant and = 2av.
5300 The area swept over by any radius is one third of the
corresponding circular sector of that radius.
5301 This curve is one of the class the general equation of
which is
r a0",

with

tan d> .
n

THE HYPERBOLIC OR RECIPROCAL SPIRAL.

5302

The equation is

(Fig. 123)

r=
u

5303

An asymptote is the line y = a.

5304

The spiral is also an asymptote to itself.

(5171)

For when the radius is of the first order of smallness, the distance
between two successive convolutions is of the second order. Hence the
distance to the pole measured along the curve is infinite.
The area between the radiants ru rg is = \a (rxrj).
5305

The equation of the Lituus is r =

THE INVOLUTE OF THE CIRCLE.

5306

(Fig. 124)

The equation is
^/(r2-8) = rt^+cos-1|-).

Proof: <f, = OPY= cos"1- and J(f-a?) =BP = arc AB = a (0+?).

r
5307

The pedal of the involute is the spiral of Archimedes.

^Invented by Conon, B.C. 250.

THEORY OF PLANE CURVES.

726

Proof.Let r, 6' be the coordinates of Y on the pedal carve.

r' = BP = arc AB = a (0' +
(See 5297).
5308

Then

The reciprocal of the involute is the hyperbolic spiral.

Proof.(Fig. 124.) Let P' on OY correspond to P, and let r, 0' be the

polar coordinates of P'. Then r = OP' =
But

OY=BP = arc AB = a ('+ K),

.-. r = w^-r.

THE CISSOID*

See (5302).

(Fig. 125)

5309 Characteristic. A line drawn from the end, 0, of

a fixed diameter of a circle to the end, Q, of any perpendicular
ordinate intersects the parallel ordinate equidistant from the
centre in a point, P, whose locus is the cissoid. The equation
of the curve is
5310

and ^ =

Proof.By similar triangles, y : x = /(2ax x1) : 2a x. Two mean

proportionals between the radius a and CS are given by the curve, for it
appears that a1 : CT* :: OT : CS, and therefore a : CT : VCS.GT : C8.
5311 The tangent of the circle at B, the other end of the
diameter, is an asymptote to both branches of the cissoid.
5312 The area between the curve and its asymptote is equal
to three times the area of the circle.
ydx substitute x = 2a sin* 0.

THE CASSINIAN OR OVAL OF CASSINI.

(Fig. 126)

5313 Characteristic.The product PA.PB of the distances

of any point on the curve from two fixed points A, B is con
stant ; the equation is consequently

or

(a?8+y*+a2)2-4oV = m*,

where 2a = AB.

The equation in polar coordinates is

r* 2aV cos 26+a'-m* = 0.

* Diocles, a.d. 500.

TRANSCENDENTAL AND OTHER CURVES.

727

5314 If a be > to, there are two ovals, as shown in the

figure. In that case, the last equation shows that if OPP'
meets the curve in P and P1, we have OP. OP" = v/(a4 m4) ;
and therefore the curve is its own inverse with respect to a
circle of radius = \/(a*m*).
5315 0 being the centre, the normal PG makes the same
angle with PB that OP does with PA.
PfiOOF.From (r + dr)(r'dr) = m2 and rr ' m1 ; therefore rdr = r'dr
or r : r' = dr : dr' = sin 6 : sin 6', if 0, 0' be the angles between the normal
and r, r. But OP divides APB in a similar way in reverse order.
5316 Let OP = B, then the normal PG, and the radius of
curvature at P, are respectively equal to
m2R
,
2miRs
.
R*+a2

3#4+a4+m*"

THE LEMNISCATE. t

(Fig. 126)

5317 Characteristic. This curve is what a Cassinian be

comes when m = a. The above equations then reduce to
(xi+y*y = 2a*(x*-y2)

and

r2 = 2a2 cos 20.

5318 The lemniscate is the pedal of the rectangular hyper

bola, the centre being the pole.
5319

The area of each loop = a2.

(5206)

THE CONCHOID. +

(Fig. 127)

5320 Characteristic. If a radiant from a fixed point 0 in

tersects a fixed right line, the directrix, in B, and a constant
length, BP = b, be measured in either direction along the
radiant, the locus of P is a conchoid. If OB = a, be the per
pendicular from 0 upon the directrix, the equation of the
curve with B for the origin or 0 for the pole is
5321

-*y = (a+y)2 (V-y1)

or

r = a sec 0 b.

* B. Williamson, M.A., Educ. Timet Math., Vol. xxv., p. 81.

t Bernoulli, Opera, p. 609.
X Nicomedea, about Jl.d. 100.

THEORY OF PLANE CURVES.

728

5323 When a < b, there is a loop; when a = b, a cusp;

and when ay b, there are two points of inflexion.
5324 To draw the normal at any point of the curve, erect
perpendiculars, at B to the directrix, and at 0 to OP. They
will meet in S the instantaneous centre, and SP will be the
normal at P (5242).
5325 To trisect a given angle BON by means of this curve,
make AB = 20N, and draw the conchoid, thus determining
Q; then AON = 3A0Q.
Proof.Bisect Q Tin S; QT=AB = 20N, therefore SN = SQ = ON;
therefore NOS = NSO = 2NQ0 = 2A0Q.
5326 The total area of the conchoid between two radiants each making
an angle 0 with OA is
a,tanfl + 26,fl + 3av/(6s-a') or astan0 + 26!0,
according as b is or is not >a.
The area above the directrix 7 _
j
I * 0\ ,
between tho same radiants )

\4
2/
The area of tho loop which exists when b is >a is
6'
_2a& log ^+^-a^ +ay(6'-a5).
o
a v(o a)

THE LIMAgON*

(Fig. 128)

5327 Characteristic.As in the conchoid, if, instead of the

fixed line for directrix, we take a fixed circle upon OB as
diameter. This curve is also the inverse of a conic with
respect to the focus. The equation, with OB for the initial
line and axis of x is
5328

r = acos06

or

(tf+tf-axy = b* O^+y2),

where a = OB, b = PQ.

5330

With b > a, 0 is a conjugate point.

With b < a, 0 is a node.

5331

[For m = a, see (5332).

The area = it (W+b2)-

When a = 2b, the limacon has been called the trisectrix.

Blaise Pascal, 1643.

TRANSCENDENTAL AND OTHER CURVES.

THE VERSIERA*
(Or Witch of Agnesi.)

729

(Fig. 130)

5335 Characteristic.If upon a diameter OA of a circle as

base a rectangle of variable altitude be drawn whose diagonal
cuts the circle in B, the locus of P, the point in which the
perpendicular from B meets the side parallel to OA, is the
curve in question. Its equation is
5336

oey = 2a v/(2alr .v2),

where a = 00 the radius.

5337 There are points of inflexion where x a.
The total area is four times the area of the circle.

THE QUADRATRIX.t

(Fig. 131)

5338 Characteristic.The curve is the locus of the inter

section, P, of the radius OB and the ordinate QN, when these
move uniformly, so that x : a :: 0 : ^v, where x = ON,
a = OA, and 9 = BOD. The equation is
,

(a oo

ir\

5339 The curve effects the quadrature of the circle, for

OC-.OB :: OB : arc ABB.
Proof: OC : OB :: CP : BD. But CP = x in the limit when it is small,
therefore CP : BD :: a : ADB.
5340

The area enclosed above the x axis = 4a3ir-1log 2.

Proof.In the integral j x tan ^ -

dx substitute jt (a x) = 2ay,

and integrate yt&nydy by parts, using (1940). The integrated torms

produce log cos
logcosijr at the limit \w, which vanishes though of
the form oo oo . The remaining integral is J log cos ydy, and will be found
at (2635).
THE CARTESIAN OVAL.

(Fig. 134)

5341 Characteristic. The sum or difference of certain fixed

multiples of the distances of a point B on the curve from two
Donna Maria Agnesi, Itutituzioni Analitichc, 1748, Art. 238. t Dinostratus, 370 B.C.
5 A

730

THEORY OF PLANE CURVES.

fixed points A, B, called the foci, is constant.

of the inner and outer ovals are respectively
5342

mr1+lr2 = ncs,

The equations

mr^ lr2 = ncs,

where i\=AP, r2 = BP, c3 = AB, and >ra>Z.

5343

To draw the curve, put = u and =-a ; therefore r, ftr. = a,

mm
where o is > AB and ft < 1 (1). Describe the circle centre A, and radius
AR = a. Draw any radiant AQ, and let P, Q be the points in which it cuts
the ovals, then, by. (1),
5344
PR=pPB and QR = pQB
(2).
Hence, by (932), we can draw the circle which will cut AR in the required
points P, Q. Thus any number of points on the oval may be found.
5345 By (2) and Euc. vi. 3, it follows that tho chord RBr bisects the
angle PBQ.
Dr;iw Ap through r, and let PB, QB produced meet Ar in p and q. The
triangles PBR, qBr are similar, therefore qr = p-qB ; therefore q is on the
inner oval. Similarly p is on the outer oval. By Euc. Tl. B., PB . QB
= PR.QR+BR1; therefore, by (2), (1-/*') PB . QB = BRK Combining
this with PB : Bq = BR : Br, from similar triangles, we get
5346

BQ.Bq =
1 fi

= p
I ft

(3).

5347 Draw QG to make Z BQG = BAq ; therefore, A,Q,G,q

being concyclic, we have, by (3),
BQ .Bq = AB.BC = ^5
1 p

(4).

Hence G can be found if a, fi, and the points A, B are

given. G is the third focus of the ovals, and the equation of
either oval may be referred to any two of the three foci.
Putting BO = Cj, AO = c AB cs, the equation between I, m, n is
obtained from (4)thus: ^,(1/u1) = a}cs; therefore Cj(c, + c,) = a'+^'c,^,.
But c. + Ci = c,, a = K u , and the result is
TO
TO
5348

Pcx+n2c3 = m2c2

or l2BC+m*CA+n2AB = 0...(5),

where GA = AG.
Putting ru
r3 for PA, PB, PC, the equations of the
curves are as follows

TRANSCENDENTAL AND OTHER CURVES.

731

5349

Inner Oval.
mrt+lrz = nes ... (6),

Outer Oval.
m^ lr2 = nc3 ... (7),

5351

+lr3 = mc% ... (8),

nrx lrs = mc2 . . . (9),

5353

mrs nr2 = Id

--.(lO),

wr2 mr3=ic1

...(11).

That (6) and (7) are equations of the curve has been shown. To deduce
the other four, we have Z APB = AqB = ACQ (5347) ; therefore ACQ,
APB are similar triangles. But, by (6), mAP + WP = nAB, therefore
mAC+lOQ = nAQ or nAQWQ = mAC, which is equation (9). Again,
ABQ, APO are similar. But, by (7), mAQlBQ = nAB ; therefore
mACWP = nAP or nAP+lCP = m^lC, which is equation (8).
Equations (10) and (11) are obtained by taking (6) from (8) and (7)
from (9), and employing (5).
AP.AQ = AB.AC = constant.
5355
Proof.Since A, Q, C, q are concyclic, t QCA = QqA = ABB ; there
fore P, Q, C, B are concyclic; therefore AP.AQ = AB.AC = constant (12).
CP. CP' =CA.CB = constant,
5356
Proof : Z PCB = PQB = Bpq = BCq. Hence, if CP meets the inner
oval again in P', CBq, CBP' are similar triangles. Again, because Z BPG
= BQG = BAq = BAP', the points A, B, P , P are concyclic ; therefore
CP. CP1 = CA.CB = constant.
Q. E. D.
Hence, by making P, P' coincide, we have the theorem :
5357 The tangent from the external focus to a series of triconfocal Cartesians is of constant length, and = ^/(GB.OA).
5358 To draw the tangents to the ovals at P and Q. De
scribe the circle round PQGB, and produce BR to meet the
circumference in T; then TP, TQ, are the normals at P
and Q.
The proof is obtained from the similar triangles TQR, TBQ, which show
that smTQA : sin TQB = I : m, by (2), and from differentiating equation (7),
which produces : ^ = I : m.m

5359 The Semi-cubical parabola y2 = aa? is the evolute of a

parabola (4549). The length of its arc measured from the
origin is
For the length of an arc of a Cartesian oval expressed by Elliptic Functions, see a paper
by S. Roberts, M.A., in Proc. Loni. Math. Soc., Vol. v., p. 6.

732

THEORY OF PLANE CURVES.

5360 The Folium of Descartes, ,z?3axy-\-if = 0, has two

infinite branches, and the asymptote x+y-\-a = 0.
For the lengths of arcs and for areas of conies, see (6015), et seq.

LINKAGES AND LINKWORK.

5400 A plane linkage, in its extended sense, consists of a

series of triangles in the same plane connected by hinges, so
as to have but one degree of freedom of motion ; that is, if
any two points of the figure be fixed, and a third point be
made to move in some path, every other point of the figure
will, in general, also describe a definite path. With two points
actually fixed, the linkage is commonly called a piece-work,
and if straight bars take the place of the triangles, it is called
a link-work.

THE FIVE-BAR LINKAGE.

5401 Mr. Kempe's fundamental five-bar linkage is shown
in Figure (135). A, B, U are fixed pivots indicated by small
circles. C, D, B', G', in the same plane, are moveable pivots
indicated by dots. The lengths of the bars AB, BC, CD, DA
are denoted by a, b, c, d. The lengths of AB', WC, CD1, UA
are proportional to the former, and are equal to ka, kb, Ice, kd,
respectively.
Hence ABCD, AB'G'D1 are similar quadri
laterals, and Z AUG' = ADC. P being any assigned point
on BC and BP = A, P' must be taken on DC so that
UP' = X ^

Draw PiY, P'N' perpendiculars to AB.

Then,

throughout the motion of the linkage in one plane, NN' is a

constant length.
Proof: NN' = BD'-(BN+N'D'). But BD'= a-kd, and
BN+ N'B' = \ cos B-\ c4 cos D =
(2ab cosB-2i cosD)
ab
2ab
= -\ (o' + 61 - c' - d?) (702). Hence

LINKAGES AND LINEWOBK.

5402

5403

733

NN' = a-kd- ^(tf+tf-c'-d2).

Case L- (Fig. 136.)

If A =

then

JftP = ^ ; consequently, if the bars PO = BB and 7*0 =

P'D' be added, the point 0 will move in the line AB.
If, in this case, d = lea and b = c, then X = b and P coin
cides with G, P with C, and B' with D, 0 as before moving
in the line ylP.

5404 Case II. (Fig. 137.) If, in Case I., led = a and
a2 + b2 = c2 + d2, X is indeterminate; that is, P may then be
taken anywhere on BG. D1 coincides with B, and NN' = 0.
PP is now always perpendicular to AB. If the bars PO,
P'O be added, of lengths such that P02-P'02 = BBS-P'B2,
O will move in the line AB. If, on the other side of PP, bars
Pa= PB and PO' = PB be attached, then O7 will move in a
perpendicular to AB through B.

5405 Case III. (Fig. 138.) If, in Case I., led = a,b = d,
and c = a, the figure A. BCD is termed a contra-parallelogram.
a2
BP X is indeterminate, BG' = lec =
and BP' = A.
CI
Hence BG' and BP' are measured in a reversed direction ;
PP1 is always perpendicular to AB, and if any two equal bars
PO, P'O are added, 0 will move in the line AB.
5406 If three or more similar contra-parallelograms be
added to the linkage in this way, as in Figure (139), having
the common pivot B and the bars BA, BG, BE, BG in geo
metrical progression ; then, if the bars BA, BG are set to any
angle, the other bars will divide that angle into three or more
equal parts.
5407 If, in Figure (138), AD be fixed and DG describe an
angle ADG, then RG' describes an equal angle in the opposite
direction. Mr. Kempe terms such an arrangement a reversor,
and the linkage in Figure (139) a multiplicator. With the aid

734

THEORY OF PLANE CURVES.

of these, and with a translator (Fig. 140), for moving a bar

AB anywhere parallel to itself, he shows that any plane curve
of the nth degree may, theoretically, be constructed by linkwork.*
5408 Case IV. (Fig. 141.) If, in the original linkage
(Fig. 135) led = a, D' coincides with B. Then, if the bars
BPO, BP'O be added by pivots at P, F, and B ; and if
OP = PB = BP' and OP' = P'B = BP ; the points 0, C
will move in perpendiculars to AB. For by projecting the
equal lines upon AB, we get NL = BN' and BN = N'L',
therefore BL = BL' = NN = a constant, by (5402).

5409 Case V. (Fig. 142.) Make ha = d and X = b. Then

B' coincides with D, P with G, and P with C. Eeplace DC,
CD by the bars DK, KH equal and parallel to the former.
Also add the bars GO DK and OK = CD. Draw the per
pendiculars from 0, G and G' to AB.
Then by projection,
NL = N'U ; therefore BL = BN+NL = BN+N'D1 = BIT
NN' = constant. Hence the point 0 will move perpendi
cularly to AB.
5410 Case VI.(Fig. 143.) In the last case take k = 1.
Therefore d = a,D' coincides with B, BK = BC, and CDKO
is a rhombus. This is Peaucellier's linkage.

5411 Case VII. (Fig. 144.) In the fundamental linkage

(Fig. 135), transfer the fixed pivots from A, B to P, S, adding
the bar SA, so that PBSA shall be a parallelogram. Then,
since NN is constant (5402), the point P1 will move perpen
dicularly to the fixed line PS.
5412 Join AG cutting PS in U, and draw UV parallel to AD.
Then UV : AD = PU : AB = CP : CB = constant ; there
fore P U and UV are constant lengths. Hence it follows that
the parallelism of AB to itself may be secured by a fixed pivot
at U and a bar UV instead of the pivot S and bar SA.
5413

In Case VII. (Fig. 144), with fixed pivots P and 8

* Proc. of the Lond. Math. Soc, Vol. Til., p. 213.

LINKAGES AND LTNKWOBK.

735

and bar SA, make b = a, d = c, ka = d, \ = b. Then F coin

cides with D, N' with N, P with C and L, and F with 0' ;
and we have Figure 145. DC, DC are equal, and they are
equally inclined to AB or CS; because, in similar quadri
laterals, it is obvious that AB and CD and the homologous
sides DC and AD" include equal angles. Therefore CC is
perpendicular to CS, and C moves in that perpendicular only.

5414 If two equal linkages like that in (5413), Figure (145),

but with the bars AS, CS removed, be joined at D (Fig. 146)
and constructed so that CDy, yDC form two rigid bars, then
AB, a/3 will always be in one straight line. Let A, B be made
fixed pivots, then, while C describes a circle, the motion of
the bar a/3 will be that of a carpenter's plane.
5415 On the other hand, if the linkage of Figure (145), with
AS and CS removed as before, be united to a similar inverted
linkage (Fig. 147), with DC, DC common, then, with fixed
pivots A, B, U, the motion of the bar a/3 will be that of a lift,
directly to and from AB.
5416 The crossing of the links may be obviated by the
arrangement in Figure (148). Here the bars (7/3, CD, CD'
are removed, and the bars FD, FE, FG added in parallel ruler
fashion.

5417 Case VIII. (Fig. 149.) In Case VII., substitute the

pivot U and the bar UV for S and SA. Make d = a, and
therefore k = 1 . Then b' = b and c' = c, making BCDC a
contra-parallelogram ; D' coincides with B, and B' with D.
The bars AB, AD are now superfluous. Take BP = A ; then
BF = ^j\ therefore PF is parallel to CC, therefore to BD,
therefore to PF (5412) ; therefore V, P, F are always in one
right line. P, as in Case VII., moves perpendicularly to PU
and AB. This arrangement is Hart's five-bar linkage.
5418 When a point P (Fig. 152) moves in a right line PS,
it is easy to connect to P a linkage which will make another
point move in any other given line we please in the same

736

THEORY OF PLANE CURVES.

plane. Let QB be such a line cutting PS in Q. Make Q a

fixed, pivot, and let OQ, OP, OB be equal bars on a free pivot
0. Then, if the angle POP be kept constant by the tie-bar
PR, PQB, being one half of POB (Euc. in. 21), will also be
constant, and therefore, while P describes one line, B describes
the other.
If the bar PO carries a plane along with it, every point in
that plane on the circumference of the circle PQB will move
in a right line passing through Q.

THE SIX-BAR INVERTOR,*

5419 If in the linkwork (5410, Fig. 143) the bar AD be
removed, and D be made to describe any curve, 0 will describe
the inverse curve, just as, when D described a circle, 0 moved
in a right line which is the inverse of a circle.
Proof.Let BOD and OK intersect in E. Then BO.OB = BEl0Et
= BG%OC% =- a constant called the modulus of the cell.

THE EIGHT-BAR DOUBLE INVERTOR,

5420 Two jointed rhombi (Fig. 150) having a common
diameter AB form a double Peaucellier cell termed positive
or negative according as P or Q is made the fulcrum. We
have PQ.PB = PQ.QS = AP^-AQf, the constant modulus
of the cell.

THE FOUR-BAR DOUBLE INVERTOR.

5421 lf on the bars of a contra -parallelogram ABCB
(Fig. 151) four points p, q, r, s be taken in a line parallel to
AC or BD, then in every deformation of the linkage, the
points p, q, r, s will lie in a right line parallel to AG ; and
pq.pr = pq.qs = a constant modulus. Thus, if p be a ful
crum and r describes a curve, q will describe the inverse
curve. If q be the fulcrum, p will describe the inverse curve.
Proof.Let Ap=mAB, therefore pq = mBD, and pr = (1 to) AC,
therefore pq .pr =
to) AC.BD = m (lm)(AD1AB1) = constant.
* Since tho curve described is the inverse and not the polar reciprocal of the guiding curve,
it seems better to call this linkage an invtrtor rather than a reciprocals.

LINKAGES AND L1NKW0BK.

737

THE QUADRUPLANE, OR VERSOR INVERTOR.

5422 Let the bars of the contra-parallelogram invertor
(5421, Fig. 151) carry planes, and let P, Q, B, S be points in
the planes similarly situated with respect to the bars which
contain p, q, r, s respectively, so that Z.PAp = QAq and
AP : Ap = AQ : Aq ; and similarly at G. Then, if P be the
fulcrum and B traces a curve, Q will trace the inverse curve
and the angle QPB will be constant.
Pboof.Let PA = nAB and PB = n'AB, therefore, by similar triangles,
PAQ, BAD, PQ = nBD. Also, by the triangles PBB, ABG, PB = n'AO ;
therefore PQ.PB = nn'AG.BD = FA.'B (AW-AB2), a constant.
Again, the inclination of PQ to BD = that of AP to AB, which is con
stant. Similarly, by the triangles PBB, ABO, the inclination of PB to AO
= that of BB to BO, which is also constant ; therefore QPB, the sum of
these two inclinations, is a constant angle.
THE PENTOGRAPH, OR PROPORTIONATOR.
5423 Let ABGD (Fig. 153) be a jointed parallelogram, A, B
fixed pivots, q a tracer placed at any assigned point in BO
produced ; then a pencil at p will evidently reproduce any
figure traced by q diminished in linear proportions in the ratio
of Bq to BG.

THE PLAGIOGRAPH, OR VERSOR PROPORTIONATOR.

5424 In the same figure, make an angle qBQ = pDP,
BQ = Bq, and DP = Dp, and let a tracer Q and pencil P be
rigidly connected to the arms BG and DG. Then P will pro
duce a similar reduced figure as before, but no longer similarly
situated. It will be turned round through an angle QBq.
This is Prof. Sylvester's Plagiograph.
Proof.Let BO = k.Bq; therefore AD = kBQ, DP = kAB, and i. ABQ
= PDA ; therefore (Enc. vi. 6) AP = kAQ. Also PAQ is a constant angle,
for PAQ = BAD-BAQ-PAD = BAD-BAQ-BQA = BAD-(*-ABQ)
= BADir + ABG + QBq = QBq.
THE ISOKLINOSTAT * OR ANGLE-DIVIDER.
5425 This linkage (Fig. 154) accomplishes the division of
an angle into any desired number of equal parts. The diaInvented and so named by Prof. Sylvester.
5 B

738

TEEOBY OF PLANE CURVES.

gram shows the trisection of an angle by it. A number of

equal bars are hinged together end to end, and also pivoted
on their centres to the same number of equal bars which
radiate, fan-like, from a common pivot. The alternate radial
bars make equal angles with each other.
The same thing is accomplished in a different way by
Kempe's Multiplicator (5406, Fig. 139).

A LINKAGE FOR DRAWING AN ELLIPSE.

5426 In the arrangement of (5413, Fig. 145) the locus of
any point P, on DC, excepting D and G', is an ellipse.
ProofTake CS, CC for as and y axes ; P the point xy ; 8CD = 8,
and therefore CDC = 20; PD = h. Then we have x = (c-h)coad,
y = (c+h) sinO, therefore
r-rj + y , = 1 is the equation of the locus.
(c h)
(c + A)
Any point on a plane carried by DC also describes an ellipse round C ; but
if the point lios on a circle whose centre is D and radius DC, the ellipse be
comes a right line passing through C, as appears from (5418).

A LINKAGE FOR DRAWING A LIMAQON, AND ALSO A

BICIRCULAR QUARTIC *
5427 (Fig. 155.) Let four bars AF, AQf, BG, CD be
pivoted at A, B, G, D, and let AB = BG = BQ = a ; AD
= DG = DF = b. Take a fulcrum F on BG, a tracer at P,
and a follower at Q, so that PQ is parallel to BD. Let FP = p,
FQ = r; then, if P traces out a circle passing through F, Q
will describe a limacon.
Pboof.Let BQ = ma, therefore PD = mb; r = 2m. BN, p = (m+l).DN
+ (1-ot) BN. Also BN^-DN' = a'-V. Eliminate BN and DN, and the
eqnation between r and p is
r'+(l-m)rp-V = *(*+l)l(o,-62) = V.
If P describes the circle p = c cos 0, Q describes the locus
r2 + (1 to) cr cos 6 mc? cos3 0 = V,
which is the inverse of a conic, that is, a limacon (5327).
If C be made the fulcrum, the equation reduces to r1p1 = 4 (a' &')
5428 With the same fulcrum F, drawing FH parallel to
AG, if a tracer at H describes the circle, then a follower at K
on CD will trace out a bicircular quartic.
W. Woolsey Johnson, Km. of Math., Vol. v., p. 169.

LINKAGES AND LINKWOBK.

739

Proof. Draw FL, LK parallel to BA, AD. Let FH = p, FE = r,

CK /3, CF=a = nFB, and therefore CL = np. Now
2 (a2 + /F) = r+y+ (a*-f?.
T
Therefore, if ff moves on the circle p = c cos#, K will describe the curve
l+W cos20-2 (a2 + /32)
(a2-02)2 = 0,
or
(,+?/),+ (V-2a2-2/32) z*-2 (a2 + /3s) j/ + (a2-/32)J = 0.

A LINKAGE FOR SOLVING A CUBIC EQUATION*

5429 Let the three-bar linkwork (Fig. 156) have the bars
AB, DC produced to cross each other. Let AB = AD = a,
BC = b, CD = c ; and let b and c be adjustable lengths.
Suppose s^qx+r = 0 a given cubic equation.
Make c = -j \Z^2+"~) & =

~j ^en deform the

quadrilateral until 7(7 = (72) ; D-E will then be equal to a real

root of the cubic.
~
Proof :

w = a;2+c2-62
cos E
3= (x
v+ a)2 + 4^-a2 ,

from which
x>- 2 (c2 + 62) a) + 2a (c2- 62) = 0.
Equate coefficients with the given cubic.f

ON THREE-BAR MOTION IN A PLANE.

5430 If a triangle ABG (Fig. 157) be connected by the
bars AO, BO' to the fulcra 0, Cf, the locus of G is called a
three-bar curve.
OA, O'B meet in Q, the instantaneous centre of rotation of
the triangle, since QA, QB are perpendicular to the movements
of A and B respectively. Therefore OQ is the normal to the
locus of 0.
5431 If a triangle similar to ABG be placed upon 00'
(homologous to AB), the circum-circle of the triangle will
pass through the node, and the vertices of the triangle are
called the foci of the curve.
* M. Saint Loup, Comptcs Rendu*, 1874.
t The foregoing account of linkages is taken chiefly from a paper hy A. B. Kempo,
F.R.S., in the Proc. of the Royal Soc. for 1875, Vol. xxm. Other results by the same author
will be found in the Proc. of the lond. Math. Soc., Vol. ix., p. 133 ; and by H. Hart, M.A.,
ibid., Vol. vi., p. 137, and Vol. vni., p. 286. See also The Messenger of Mathematics, Vol. v.

740

THEORY OF PLANE CURVES.

Figures (158) and (159) exhibit different varieties of the

curve according to the relative proportions between the lengths
of the bars.*

MECHANICAL CALCULATORS.
The Mechanical Integrator.^
5450 This instrument computes not only the area of any
closed plane curve, but the moment and also the moment of
inertia of the area about a fixed line. The principle of its
action is shown in Figure (160). OP is a bar carrying a
tracer at P, and a roller A at some point of its length. The
end 0 is constrained to move in the fixed line ON. When
the tracer P moves round a closed curve, the length OP mul
tiplied by the entire advance recorded by the roller is equal to
the area of the curve.
Proof. Let the motion of the tracer from P to a consecutive point Q be
decomposed into PP' and P'Q parallel and perpendicular to ON. Let
OP = a and PON = 6. When the pointer moves from P to P', the roll
accomplished is PP' sin 8. The roll due to the motion from P" to Q will be
neutralized by the exactly equal and opposite roll in the motion of the pointer
from q to p', since the bar will there have again the same inclination. Con
sequently the product of the entire roll and the length a is equal to the sum
of such terms as aPP' sin 6. But this is the area OPP'O' = NPP'N'. The
algebraic addition of such rectangles gives the entire area, and the instru
ment effects this, for the area SN is subtracted, by the motion of the roller,
from the area QN which is added.

5451 The instrument itself is shown in Figure (161). A

frame moving parallel to OX by means of the guide BB
carries two equal horizontal wheels geared to a central wheel
which has two circumferences, such that its rate of angular
motion is half that of the lower wheel and one third of that of
the upper. The latter wheels carry two rollers, M and I, on
horizontal axles ; and the middle wheel carries an arm OP, a
pointer at P, and a roller A. In the initial position, the
* The curve is a tricireular trinodal sextic, and is completely discussed by S. Roberts,
F.R.S., and Prof. Cayley, in the Proe. of the Lond. Math. Soc., Vol. vn., pp. 14, 136.
t Invented and manufactured by Mr. J. Amsler-Laffon, of Schaffhausen. The demon
strations (which in clearness and elegance cannot be surpassed) of the action of this instru
ment, and of the Planimeter which follows, were communicated to the author by Mr. J.
Macfarlano Gray, of the Board of Trade,

MECHANICAL CALCULATOES.

741

rollers A and J are parallel, while M is at right angles to A.

The frame is thus supported above the paper on the three
rollers ; and if the arm OP be moved through an angle AOA',
the axles of the rollers M and I will describe twice and three
times that angle respectively. Putting OP = a as above, and
A, M, and I for the linear circumferential advances recorded
by the three equal rollers respectively, we have the following
results
I.The area traced out by the pointer P = Aa.
a*
II. The moment of the area about OX
= M.
4
s

III. The moment of inertia about OX

= (3.4+ J) .

Proof.I. Since 0 moves in the line OX, while the pointer P moves round
a curve, the roller A will, as shown above, make the rolling 2A sin 6, where
h = PP in Figure (160), and the area of the curve = aXh sin 6 or a X roll.
II. The moment of the area about OX
= 2 (afc sin 0 X ?L^1\ = ^ (Xh-lh cos 26).
Now 2A vanishes when P returns to the starting point, and S/i cos 28 is the
roll recorded by M. For, when OP makes an angle 0 with OX, the axis of
M will make an angle (90 + 26) with OX. In this position, while P makes a
parallel movement h, the roll produced thereby in M will be h sin (90 + 26)
= h cos 20. Therefore ~ X roll of M = moment of area.
4
III. Lastly, the moment of inertia of the area about OX
= 2 (oAsinfl x a's20\ - |L2 (3fc sinO-h sin 30).
Now, when OP makes an angle 6 with OX, the axis of I makes 36 ; there
fore %h sin 36 is the entire roll of I. Hence the moment of inertia
= ^ X roll of A+ x roll of I.
4
12

The Planimeter.

(Fig. 162)

5452 This instrument * is a simpler form of area computer.

0 is a fixed pivot ; OA, AP are two rods having a free pivot
at A; G is the roller, and P the pointer. The area of a closed
curve traced by the pointer is equal to the total roll multiplied
by the length AP.
* like The Integrator, the invention of Mr. Amsler.

THEORY OF PLANE CURVES.

742

Proof.Decompose the elementary motion PQ of the pointer into Pf,

effected with a constant radios OP, and P'Q along the radius OP', and so all
round the curve. The roll of 0 accomplished while P moves from P to Q
will be neutralized by the equal contrary roll when P moves from q top' on
the radius Op' = OP. Thus the total roll recorded will be the snm of the
rolls due to the movements PP', QQ', &c.
Draw OB perpendicular to AP, aud, when P comes to R, let B1 be the
altered position of B. The area PQSR = \ (OP1- OR2) u, where u = POQ.
But 0P1 = OA2 + PA*-2PA.AC-2PA.BC (Euc. n. 13); therefore, since
BC is the only varying length on the right, we have PQSR = PA(BC~B(/)u.
But BCto is the roll of 0 due to the angular motion u of the rigid frame OAF,
and the subtraction of the area OSR from OPQ is effected by the instrument,
since when the pointer moves from 8 to R the direction of the roll must be
reversed. Hence the total area = PA x the total recorded roll.

APPENDIX ON BIANGULAR COORDINATES*

5453 In the figure of (1178), the biangular coordinates of a

point P are defined to be 6 = PSS' and <p = PS'S, or
o = cot 0 and |3 = cot <t>.
5454

The equation of a right line YY' is

aa+6/8 = 1,

where a = cot SYS' and b = cot SY'S'.

Proof.Supplying the ordinate PN in the figure and denoting the angle
S'SY by the equation is obtained from ON cos 4> + PN s'mif/ = p the per
pendicular on the tangent, SS' sin <p = YY' and SS' cos 41 = SYS'Y .

5455

cot/ = ab.

5456

Equation of a line through G :

o /8 = const.

5457

Equation of the line at infinity :

a-|-/8 = 0.

5458

Let SS' = c, then the distance between two points

_J

M2i

* Quarterly Journal of Mathematics, Vols, 9 and 13 ; W. Walton, M.A.

APPENDIX ON BIANGULAR COORDINATES.

5459

743

The equation of a line through the two points is

PPi ~ ft &
5460 The length of the perpendicular from a'|3' upon the
line aa + Ifi = 1 is
p
c
aa-\-bf 1
"7+^ V{(a-6)8+l}"
5461 Cor.The perpendiculars from the poles 8, 8' are
therefore
SF=_
_
S'Y'=
"c
<S{(a-by+l}'
5463

^{{a-by+1}'

When the point a'/3' is on SS' at a distance h from #,

(a b) h+bc
P~ V{(a-by+l\'

With two lines aa + 6j3 = 1, a'a + b'fl = I, the condition

5464

of parallelism is

ab = ab',

5465

of perpendicularity (a b){a' b')+l = 0.

5466 The equation of the line bisecting the angle between

the same lines is
qa+fy8-l
_
a'a+b'P-l

5467 The equation of the tangent at a point a'j3' on the

curve F(a, /3) = 0 is

5468

And the equation of the normal is

o-o^
P-P
(a^-l^+a + a2)^ ~ (a'ff-1)
+ (!+") /y
5469

The equation of a circle through S, S' is

aj8-l= m(a+j8),

where m = cot SPS' the angle of the segment.

THEORY OF PLANE CURVES.

744

5470

If G be the centre, the equation becomes

a/9 = 1.

5471 And, in this case, the equations of the tangent and

normal at a'/3' are respectively
4 +4 = 2
a
fS

and

a-P = a'-p.

5472 The equation of the radical axis of two circles whose

centres are S, S', and radii a, b, is
(e?-a2+62) a = (cs+aJ-6i) .
Proof. By equating the tangents from aft to the two circles, their
lengths being respectively
i^ffty-a

and ih#-tf,lv(5*58)-

5473 To find the equation of the asymptotes of a curve

when they exist,
Eliminate a and /3 between the equations of the line at
infinity
the curve

= 0,
F(o, /?) = 0,

and the tangent (a -a') F + (-')

= 0.

Ex.The hyperbola a*+(? = m* has, for the equation of its asymptotes,

a ft m \/2.

SOLID COORDINATE GEOMETRY.

SYSTEMS OF COORDINATES.
CARTESIAN OR THREE-PLANE COORDINATES.

5501 The position of a point P in this system (Fig. 168) is

determined by its distances, x = PA, y = PB, z = PC, from
three fixed planes YOZ, ZOX, XOY, the distances being
measured parallel to the mutual intersections OX, OY, OZ of
the planes, which intersections constitute the axes of coordi
nates. The point P is referred to as the point xyz, and in
the drawing x, y, z are all reckoned positive, ZOX being the
plane of the paper and P being situated in front of it, to the
right of YOZ and above XOY. If P be taken on the other
side of any of these planes, its coordinate distance from that
plane is reckoned negative.
POUR-PLANE COORDINATES.

5502 In this system the position of a point is determined

by four coordinates a, j3, y, B, which are its perpendicular
distances from four fixed planes constituting a tetrahedron of
reference. The system is in Solid Geometry precisely what
trilinear coordinates are in Plane. The relation between the
coordinates of a point corresponding to (4007) in trilinears is
5503
Aa+Bp+Cy+DS = 3V,
where A,B,0,D are the areas of the faces of the tetrahedron
of reference, and V is its volume.
TETRAHEDRAL COORDINATES.

5504 In this system the coordinates of a point are the

volumes of the pyramids of which the point is the vertex and
5 c

746

SOLID GEOMETRY.

the faces of the tetrahedron of reference the bases : viz., %Aa,

\Bfi, ^Gy, ^DS. The relation between them is
5505

a'+#+r'+8' = V.
POLAR COORDINATES.

5506 Let 0 be the origin (Fig. 168), XOZ the plane of

reference in rectangular coordinates, then the polar coordi
nates of a point P are r, 0, <f>, such that r = OP, 0 = Z.POZ,
and f = /_XOG between the planes of XOZ and POZ.

THE RIGHT LINE.

5507 The coordinates of the point dividing in a given ratio
the distance between two given points are as in (4032), with a
similar value for the third coordinate I.
5508

The distance P, Q between the two points xyz, x'y'z' is

PQ =^{(x-Xy+{y-y'y+(z-zy}. (Euc.i.47).

5509 The same with oblique axes, the angles between the
axes being X, p, v.
PQ = J{(x-x'y+ty-y'y+(z-zy+2{y-y')(z-z) cosX
+2 (zz'){x-x) cos/t+2 {xx')(yyr) cos v\ . (By 702).

5510 The same in polar coordinates, the given points being

rfy, r'ey,
PQ = </[r2+r'*-2rr {cos 6 cos tf'+sin 6 sin ff cos (<-<') } ].
Proof.Let P, Q be the points, 0 the origin. Describe a sphere catting
OP, OQ in B, G and the z axis in A; then, by (702), PQ} = OP, + 0(?
20P.OQ cos POQ and cosPOQ, or coso in the spherical triangle ABC, is
given by formula (882), since 6 = 0, c = 6', and A = <p<f>'.
DIRECTION RATIOS.
5511 Through any point Q on a right line QP (Fig. 169),
draw lines QL, QM, QN parallel to the axes, and through any
other point P on the line draw planes parallel to the coordi

THE BIGHT LINE.

747

nate planes cutting the lines just drawn in L, M, N; then the

direction ratios of the line OP are
5512

l-Q

m-QM

n-?

The angles PQL, PQM, PQN are denoted by a, /3, y; and the
angles YOZ, ZOX, XOY between the axes by X, fi, v.
5513 When X, /x, v are right angles, the axes are called
rectangular, and the direction -ratios are called directioncosines, being in that case severally equal to cos a, cos/3,
cos y.
5514 When L, M, N are the direction-ratios (or numbers
proportional to them) of a line which passes through a point
abc, the line may be referred to as the line (LMN, abc), or, if
direction only is concerned, merely the line LMN.

EQUATIONS BETWEEN THE CONSTANTS OP A LINE.

5515 The relation between the constants of a line with
rectangular axes is
P+m2+n2 = 1 ;
and with oblique axes, it is
5516

I cos a+w cos #+n cosy = 1.

Proof.The first by (Enc. i. 47). The second by projecting the bent

line QLOP (Fig. 169) upon PQ, thus PQ = QL cos a+iCcos/3+ CP cos y,
and QL = PQ.l, Ac, by (5512).
5517

Also, when the axes are oblique,

cos a =

Z+m cos v+ n cos /*,

cos # = m+ n cos X+ I cosv,

cosy= n+ I cos/t+wl cosX.
Proof.By projecting QP in figure (169) and the bent line QLOP upon
each axis in turn, and equating results; thus PQcosa= QL + LG cos/3
+ CP cos y, applying (5512).
5518

The relation between I, m, n and X, fi, v is

t2-\-tn2+n2-\-2mn cosX+2nZ cos/*+2fm cosv = 1.

Proof.By eliminating cos a, cos/3, cosy between (5516) and (5517).

SOLID GEOMETRY.

748

5519

The relation between cos a, cos /3, cos 7 and X, p, v is

cos1 a sin2 X+cos* /8 sin* /t+cos1 7 sin* v

+ 2 cos /8 cos y (cos ft cos v cos X)
+2 cos 7 cos a (cos p cos X cos ft)
+2 cos a cos /3 (cos X cos /t cos i)
= 1 cos*X cos* /1 cos*i'+2 cosX cos/i cos v.
Pkoof.By eliminating I, m, n between the four equations in (5516) and
(5517).
5520 The angle 0 between two right lines Imn, Vm'n', the
axes being rectangular :
cos 6 = U'-\-mm'+nn'.
Proof.In Figure (169), let QP be a segment of the line Imn. The
projection of QP upon the line Vm'n' will be QP cos 0. And this will also
be equal to the projection of the bent line QLCP, upon Vm'n, for, if
planes be drawn through Q, L, G, and P, at right angles to the second line
Vm'n', the segment on that line intercepted between the first and last plane
will be = QP cos 0, and the three segments which compose this will be
severally equal to QL.V, LG.m, GP.n, the projections of QL, LC, CP.
Then, by (5512), QL=QP.l, Ac.
5521

sin* 6 = (mri-m'ny+(nl'- n7)*+(/m' -I'm)*.

Proof.From
I-cob'0 = (P+m'+n'W+m'i + n^-ilV +mm +nn'y (5515, '20).
5522

With oblique axes,

cos 0 = ll'+mm'-\-nri-\-(mri-\-m'n) cosX

+ (n/'+n7) cos ii+(lm'+l'm) cos*.
Proof.As in (5520), substituting from (5517) the values of cos a, Ac.

EQUATIONS OF THE RIGHT LINE.

5523

Z^ = ^ = =_<
L
M
N

or

fzf = jf= = =e
Imn

Here abc is a datum point on the line, and if r be put for the
value of each of the fractions, r is the distance to a variable
point xyz. L, M, N are proportional to the direction ratios of

749

THE BIGHT LINE.

the line, which ratios must therefore have the values

5524

l== y/^+BP+N*)'

m== JW+M'+N*)'
N

n=
5525 Note.Instead of a, b, c in the equation we may use
kL+a, JcM+b, hl + c, where k is an arbitrary constant.

5526

The equations of a line may also be written in the

forms

x Xs+a,

y fiz-\-/3.

5527 These are the equations of the traces on the planes of

xz and yz, and are equivalent to
oo a _ y fi _ z 0

5528 If the line is determined as the intersection of the two

planes Ax+By + Cz = D and A'x-\-B'y-\-G'z = Z7, we may
write equations (5523) by taking
L = BC'-B'C>
=

M=CA'-C'A,

DB'-D'B
^
,

N=AB'-A'B,

,
DA-DA
b =
w,

n
e = 0.

Proof. Eliminate z between the equations of the planes, then the

reciprocals of the coefficients of x and y will be L and M.
5529 The projection of the line joining the points xyz and
abc upon the line Imn is
I

a)-\-m (yb)-\-n (z c).

5530 Hence, when the line passes through abc, the square
of the perpendicular from xyz upon it is equal to
(x- ay+(y-by+(z-c)*- {l (x-a)+m (y-b)+n (z-c)}*5531

Condition of parallelism of two lines LMN, L'M'N' :

L : L' = M : W = N : A'.

750

5532

SOLID GEOMETRY.

Condition of perpendicularity :
LL'+MM'+NN' = 0.

(5520)

5533 Condition of the intersection of the lines (LMN, abc)

and (L'M'N', a'b'c') (5514) :
(a -a') (MN' - M'N) +(b- b') (NL'- N'L)
+ (c-c')(LM'-L'M) = 0.
Proof.Eliminate as, y, z between the equations
_
- Z_Z_? _ r and x-a.' y-V

z-c _ .

by subtracting in pairs, and then eliminate r and r .

5534

The shortest distance between the same lines is

(a-a')(MN'-M'N) + (b-b')(NL'- N'L) + (c-c')(LM' -L'M)

J { (MN'-M'Ny+(NL'~ N'Ly+iLM'-L'Mf]
Proof.Assume X, fi, v for the dir-cos. of the shortest distance. Then,
by projecting the line joining abc, a'b'c upon the shortest distance, we get
p = (a - a) X + (b b') n + (c c) v. Also, by (5520), L\ + Mfi + Nv = 0 and
L'X + M'p+N'v = 0, giving the ratios X : n : v = MN'M'N : NL'-N'L
: LM'L'M; and (5524) then gives the values of X, ft, v.

5535 The equation of the line of shortest distance between

the lines (Imn, abc) and (I'm'ri, a'b'c') is given by the inter
section of the two planes
l{x-a)+mty-b) + n (z-y) = "^J*

(i-)>

t {x-a') + m' (y-6')+' (z-y') = *+"$ ... ("),

sin p
where

u = I (a'a)+m (b'b)+n (c'c),

u'= l'(a-a')+m'(b-b')+n'(c-c'),

and

cos 0 = 11' + mm' + nri.

m

Proof.(Fig. 170.) Let 0 be the point xyz on the line of shortest dis
tance AB ; P, Q the points abc, a'b'c on the given lines AP, BQ. Draw BR
and PR parallel to AP and AB ; RT perpendicular to BQ ; and QS, Til per
pendicular to BR. Then L RBQ = d, RN = u, QT = ', therefore Kit
= m'co80 and RH = RN+NM = u + u cos 0, and in the right-angled tri
angle RTB, 22Af cosec'0 = RB, the projection of OP upon AP, that is, the

THE BIGHT LINE.

751

left member of equation (i.). Similarly for equation (ii.). It should be

observed that (i.) and (ii.) represent planes through AB respectively per
pendicular to the given lines AP and BQ.
5536 Otherwise, the line of shortest distance is the inter
section of the two planes whose equations are
r(x-a)+m'(y-b)+n'(z-e) _ CQ8 0
I (x
(y 6)+n (zc)
I (x-a')+m (y-b')+n (z-c')
V (x-a')+m' (y-b') + n' {z-c')'
For these equations state that cos 6 is the ratio of the projections of OP
or of OQ upon the given lines, and this fact is apparent from the figure.
5537 Equations of the line passing through the two points
abc, a'b'c' :
xa
yb _ zc
a a'
6 6'
cc
5538 A line passing through the point abc and intersecting
at right angles the line Imn :

where

x a
yb _ zc
~L~ ~ M ~ ~W~'
L = Im (bb')+nl(cc') (mi-\-nt)(aa'),

and symmetrical values exist for M and N.

Proof.The condition of perpendicularity to Imn is
Ll+Mm+Nn = 0;
(5520)
and the condition of intersecting the line is
(a-a)(Mn-mN) + (b-b')(Nl-nL) + (c-c')(Lm-lM) = 0.
These equations determine the ratios L : M : N.
5539 Equations of the line passing through the point abc,
parallel to the plane Lx + My+Nz = D, and intersecting the
line {I'm'n', a'Vc') :
xa _ yb _ zc
I
m
n '
where I, m, n are found, as in the last, from
LI + Mm + Nn = 0,

and

(a-a')(mn'-m'n) + (b-b')(nl'-ril)+(c-c'){lm'-l'm) = 0.

SOLID GEOMETRY.

752

5540 Equations of the bisector of the angle between the

two lines ^m^, Z2m2n2 :
x

Proof. Taking the intersection of the lines for origin, let 3,^2,, sBjt/,2, be
points on the given lines equidistant from the origin ; then, if xyz be a point
on the bisector midway between the former points, a; = (Xj-t-x,), <fcc.
(4033) ; and the direction-cosines of a line through the origin are propor
tional to the coordinates.
5541
are

The equations of a right line in four plane coordinates

a~a' -

_ tZjL - 8~8'

(i )

where a/3-yS is a variable point, and a'fi'y'S' a fixed point on

the line. The relation between L, M, N, B is
5542

AL+BM+CN+DR = 0

(ii.).

Proof.For, since equation (5503) holds for a/3y and also for a'fl'y'i',
we have
A (a-a') +B (/3-/T) + C (y-y')+D (S-S') = 0.
Substitute from (i.) a a' = rL, (3f}' = rM, &c.
5543 In tetrahedral coordinates the same equation (i.) sub
sists, but the relation between L, M, N, B becomes, by
changing Aa into a, &c,
5544

L+M+N+R = 0.

THE PLANE.
5545

General equation of a plane :

Ax+By+Cz+D = 0.

5546

Equation in terms of the intercepts on the axes :

5547 Equation in terms of p, the perpendicular from the

origin upon the plane, and I, m, n, the direction-cosines of p :
te+my+nz = p.

Th?E PLANE.

753

Peoof.If P be any point xyz upon the plane, and 0 the origin, the pro
jection of OP upon the normal through 0 is p itself; but this projection is
Ix+my + nz, as in (5520).
5548 The values of Z, m, n, p for the general equation
(5545) are
/4
&c
n ~D
y/iA'+JP+C*)*
'
P ~ ^/(A'+B'+C2)'
Peoof.Similar to that for (4060-2) : by equating coefficients in (5545)
and (5547) and employing P + mf + n1 = 1.

5550

The equation of a plane in four-plane coordinates is

Za+m/8+ ny+rS = 0,

with

I : m : n : r = ::: ,
Pi Pi Ps Pi

where a19 ft, yu c\ are the perpendiculars upon the plane from
A, B, G, D, the vertices of the tetrahedron of reference, and
2>i, p.i, p3, Pi are the perpendiculars from the same points upon
the opposite faces of the tetrahedron.
Proof.Put y = = 0 for the point where the plane cuts an edge of the
tetrahedron, and then determine the ratio i : to by proportion.
See Frost and Wolstenholme, Art. 81.
5551 The equation of a plane in tetrahedral coordinates is
also of the form in (5550), but the ratios are, in that case,
I : m : n : r = at : ft : y1 : 81 .
The relation between the three-plane and four-plane coor
dinates is

5552

a = p Lv my nz .

The equation of a plane in polar coordinates is

r {cos0 cosfl'+sinfl siuff cos

^')} =/).

Peoof.Here p is the perpendicular from the origin on the plane, and

p, 0', <p' the polar coordinates of the foot of the perpendicular. Then, if i is
the angle between p and r, we have p = r cos ^ and cos i/< from (882).
5553

The angle 0 between two planes

lx-\-my+nz = p

and I'x+m'y + n'z = p'

5 D

SOLID GEOMETRY.

754

is given by formula (5520), and the conditions of parallelism

and perpendicularity by (5531) and (5532), since the mutual
inclination of the planes is the same as that of their normals.
5554 The length of the perpendicular from the point x'y'z'
upon the plane Ax+By-\-Cz-\-D = 0 is
, Ax+By+Cz+D
, ,
vw+w+n = p~Lv ~my ~nz ProofAs in (4094).
5556

The same in oblique coordinates

Q
= (Ax+By+Cz+D)

!- = px coso y cospz
cosy,
P
where p is found from (5519) by putting A, B, C for p cos a,
P cos (3, p cos y. This gives
cecQ

( A* sin' X + B* sin* n + C sin! v + 2BC (cos p cos v cos X) )

\ +2CA (cos v cos X cos n) + 2AB (cos \.cos ft cos v) )
1 cos2 X cos* p cos1 y + 2 cos X cos fi cos v

5559 The distance r of the point x'y'z' from the plane

Ax+By + Cz-\-D = 0, measured in the direction Imn, the axes
being oblique :
_ _ Ax'+By'+Cz'+D
T
Al+Bm + Cn
'
Proof.By determining r from the simultaneous equations of the line
and the plane, viz.,
!LZ" = U=2l - =f! = r and Ax + By + Cz+B = 0.
{
m
n
Otherwise, by dividing the perpendicular from x'y'z' (5554) by the cosine of
its inclination to Imn, viz., y^^^^y

EQUATIONS OF PLANES UNDER GIVEN CONDITIONS.

5560 A plane passing through the point abc and perpen
dicular to the direction Imn :
I {x a)-\-m (y b)-\-n (z c) = 0.

THE FLANE.

5561

755

A plane passing through two points abc, a'b'c' :

xa .
yb , ze
A
r^+^fT7+1'
a
a
oo
cc7 = 0,

5562

with

X+fi+v = 0.

Puoof.By eliminating n between the equations

l(x a)+m(y b)+n(z c) = 0, I (a a) + m (b 6') +n (cc) = 0,
and altering the arbitrary constant.
5563 A plane passing through the point of intersection of
the three planes w = 0, v = 0, w = 0:
lu+mv-\-nw = 0.
5564 A plane passing through the line of intersection of the
two planes u = 0, v=0 :
lu-\-mv = 0.
5565 A plane passing through the two points given by
u = 0, v = 0, iv = 0 and = a, v = b, w = c:
lu-\-mv-\-nw = 0

with

la+mb-\-nc = 0.

5566 The equation of a plane passing

through the three points xxyxzu ^j^, xayAz3
or A, B, G, is given by the determinant
annexed, in which the coefficients of x, y, z
represent twice the projections of the
area ABC upon the coordinate planes.

x
xx
Xi
<*3

y
yx
y2
y$

z
zl
z,
%3

1
1
1
i

= o.

Proof. The determinant is the eliminant of Ax + By + Cz = 1 , and

three similar equations. Expanded it becomes
x (y^iy^i+y^iy^s+yii'iyr'd+y (&c-)+* (Ac.)
*- = o.
Hence, by (4036), we see that the coefficients are twice the projections of
ABC, as stated.
5567 The sum of squares of the coefficients is equal to
four times the square of the area ABC.
Proof.For, if I, m, n are the dir-cos. of the plane, and ABC = S, the
coefficients are = 2SI, 2Sm, 2,Sn, by projection.
5568 The determinant (xu yi} zs), that is, the absolute term
in equation (5566), represents six times the volume of the
tetrahedron OABG, where 0 is the origin.

756

SOLID GEOMETRY.

Proof.Writing the equation of the plane ABC, Ax+By + Cz+D = 0,

we have for the perpendicular from the origin, disregarding sign,
P = ^(A'+B'+C) = 28 (556?)'
therefore D = 2p8 = 6 x the tetrahedron OABC.
5569 If xyz be a fourth point, P, not in the plane of ABC,
the determinant in (5566) represents six times the volume of
the tetrahedron PABC.
Proof.By the last theorem the four component determinants represent
six times (OBCP+OCAP+OABP+OABC) for an origin 0 within the
tetrahedron.
5570 A plane passing through the points abc, a'b'c', and
parallel to the direction Imn :
xa yb ze
a a bb' cc'
Imn
aa
bb' cc'

= 0.

Proof. Eliminate X, ft, v between the equations (5561-2) and

= 0, the condition of perpendicularity between
a a
b6
cc
and the normal of the plane (5561)
5571 A plane passing through the point abc and parallel
to the lines Imn, I'm'ri :
xa

yb

m' | = 0.

z c

11

Proof.The equation is of the form X (x a) +fi (y 6) + v (a c) = 0,

and the conditions of perpendicularity between the normal of the plane and the
given lines are l\ + mfji+nv = 0, l'\ + m'fi+n'v = 0. Form the eliminant
of the three equations.
5572 A plane equidistant from the two right lines (abc, Imn)
and (a'b'c', I'm'ri) :
(+')

'

l'

* i(c+c')

ri

By (5571).

TRANSFORMATION OF COORDINATES.

757

5573 A plane passing through the line (abc, Imn) and per
pendicular to the plane l'x+my+n'z = p :
The equation is that in (5571).
For proof, assume X, fi, v for dir-cos. of the normal of the required plane,
and write the conditions that the plane may pass through abc and that the
normal may be perpendicular to the given line and to the normal of the
given plane.

TRANSFORMATION OF COORDINATES.
5574 To change any axes of reference to new axes parallel
to the old ones :
Let the coordinates of the new origin referred to the old
axes be a, b, c; xyz and x'y'z', the same point referred to the
old and new axes respectively ; then
# = <2/+a> y=y'+b,

z = z'-\-c.

5575 To change rectangular axes of reference to new

rectangular axes with the same origin :
Let OX, OY, OZ be the original axes, and OX', OT, OZ'
the new ones,
Zi m^j the dir-cos. of OX' referred to OX, OY, OZ,
^m^
do.
OY'
do.
do.
ls ms ns
do.
OZ'
do.
do.
xyz, vl the same point referred to the old and new axes
respectively. Then the equations of transformation are

5576

*=UHHU
y = mg+mfii+mtl

(i.).
(ii.),

* = i+2i?+s
(iii.).
And the nine constants are connected by the six equations
5577 fi+ml+nl = 1 ...(iv.), Vs+7ri27n3+n2n3 = 0 ... (vii.),
ll+ml+n2 1 ... (v.), yi+TO3m1+ n^ = 0 ... (viii.),
fs+ml+ns = 1 ...(vi.), ^2+w1ma+,2 = 0 ... (ix.),
so that three constants are independent.

SOLID GEOMETRY.

758

Proof.By (5515) and (5532), since OX', OY', OZ' are mutually at
right angles.
5578

The relations (iv. to ix.) may also be expressed thus

h
mi
ni
_ i j
m2ns m5n2 2ZS n3/2 /273 /3/n2
_

1n2 m2W!

mi

Hh
x/2 2/x

__

n2

__ _J_ J

!h
= 1
/iWij /2mx

/x \

/XJ \

(xii.).

Obtained by eliminating the third term from any two of equations

(vn.ix.). Also, by squaring each fraction in (x.) and adding numerators
and denominators, we get

5579 If the transformation above is rotational, that is, if it

can be effected by a rotation about a fixed axis, the position
of that axis and the angle of rotation 0 are found from the
equations
kkoa
5580

. cos2 aj

2 cos 0 = Zi+w2+3 1 ,
r = r~,cos2/?
-t = j. cos2y
L

\,

where a, /3, y are the angles which the axis makes with the
original coordinate axes.
Proof.(Fig. 171.) Let the original rectangular axes and the axis of
rotation cut the surface of a sphere, whose centre is the origin 0, in the
points x, y, z, and / respectively. Then, if the altered axes cut the sphere in
I, v, i, we shall have fl = L xli in the spherical triangle ; Ix = IS, = u ; Iy =
J; = (i ; Izz= J; = y, and by (882) applied to the isosceles spherical triangles
xIS, &c., ^ = cos xl = cos' a -+- sin* a cos 0, mt = cos yt) = cos5/3 + sin* /3 cos 0,
n, = coszf = coss y + sin* y cosO. From these cosfl, cos a, cos/3, and cosy
are found.
5581

Transformation of rectangular coordinates to oblique :

Equations (i. to vi.) apply as before, but (vii. to ix.) no

longer hold, so that there are now six independent constants.

tHE SPHERE.

759

THE SPHERE.

5582 The equation of a sphere when the point abc is the

centre and r is the radius,
(*-a)2+(y-&)2+(*-c)2=r*.
5583

The general equation is

x?+y*+z2+Ax+By+Cz+D = 0.

ABC
The coordinates of the centre are then , , ;
and the radius = ^^/(A'+^+C^-W).
Proof.By equating coefficients with (5582).
5584 If xyz be a point not on the sphere, the value of
( a)2 + (y iy+(z cf r2 is the product of the segments
of any right line drawn through xyz to cut the sphere.
Proof.From Enc. hi., 35, 36.

THE RADICAL PLANE.

5585 The radical planes of the two spheres whose equations
are u = 0, u' = 0, is
u-u'= 0.
5586 The radical planes of three spheres have a common
section, and the radical planes of four spheres intersect in the
same point.
Proof. By adding their equations, and by the principle of (4608)
extended to the equations of planes.

POLES OF SIMILITUDE.
5587 Dep.A pole of similitude is a point such that the
tangents from it to two spheres are proportional to the radii.
5588 The external and internal poles of similitude are the
vertices of the common enveloping cones.

SOLID GEOMETRY.

760

5589 The locus of the pole of similitude of two spheres is a

sphere whose diameter contains the centres and is divided
harmonically by them.

CYLINDRICAL AND CONICAL SURFACES.

5590 Def.A conical surface is generated by a right line

which passes through a fixed point called the vertex and
moves in any manner.
5591 If the point be at infinity, the line moves always
parallel to itself and generates a cylindrical surface.
5592 Any section of the surface by a plane may be taken
for the guiding curve.
5593

To find the equation of a cylindrical or conical surface.

Rule. Eliminate xyz from the equations of the guiding

curve and the equations x - - = ^ ^ = ^ of any generating
line ; and in the result put for the variable parameters of
the line their values in terms of x, y, and z.
5594 Ex. 1. To find the equation of the cylindrical surface whose
generating lines have the direction Imn, and whose guiding curve is given
by bV + ay = aW and z = 0.
At the point where the line ^-r = *^ = meets the ellipse, * = 0,
Imn
x = a, y = (i. Therefore fc'a'+a'/j' = a'6J. Substitute in this, for the
vuriablo parameters, a, /3, a = x, fi = y ; and we get, for the
n
n
cylindrical surface 61 (nxky + a? (nynu)* = o'iV.
5595 A conical surface whose vertex is the origin and
guiding curve the ellipse iV+a4^ = oV, z = c, is

* + b*

c*

Proof.Here the generating line is 4 = ^- &b the point of interimn

CONICOIDS.

761

section of the line and carve z = c, = -, = ; .-. ^ +

= a2b\
n
n

n
Substitute for the variable parameters I : m : n the values x : y : z, and the
result is obtained.

CIRCULAR SECTIONS.
5596 Rule. To find the circular sections of a quadric curve,
express the equation in the form A (x2 + y2 + z2 + c2) + &c. = 0.
If the remaining terms can be resolved into two factors, the
circular sections are defined by the intersection of a sphere and
two planes.
5597

Generally the two quadrics

az'i + by2 + cz* + 2fyz + 2gzx + 2hxy = 1

and (a + \)x* + (b+\) y*+(c + \) z2 + 2fyz + 2gzx+2hxy = 1

have the same circular sections.
Proof.Let r, p be coincident radii of the two surfaces having Imn for a
common direction. Then \ = aP + brn2 + cn2 + 2fmn + 2gnl + 2hlm and
r"
'
p= the same +\. Therefore, if r has a constant value throughout any
section, p is also constant throughout that section.

5598 Ex.An oblique circular cone whose vertex is the point a, 0, b,

and guiding curve the circle z1+y1 = c' ; z = 0 ; is
(az-bzy + by = c1 (z-b)'.
The equation may be written
b% (x*+yt+zt-c1) = z {2abx + (b' + ci-ai) z-2bc>},
and therefore the cone has two series of parallel circular sections, z h and
2abx + (bi + c'-ai) z-2bct = p1 (5583).
{Frost and Wolstenholme.)

CONICOIDS.

5599 Defs.A conicoid is a surface every plane section of

which is a conic.
The varieties are the ellipsoid, the one-fold and two-fold
hyperboloids, the elliptic and hyperbolic paraboloids, the
spheroid of revolution, the cone, and the cylinder.
5 E

762

SOLID GEOMETRY.

In any of the following equations of a conicoid, by making one of the

variables constant, the equation of a section parallel to a coordinate plane is
obtained, and the equation of the surface is by that means verified. Thus,
in the equations of (5600) or (5617), Figs. (172) and (173), if z be put
= ON, we get the equation of the elliptic section RPQ, the semi-axes of
which are NQ = - ^/(c,-0^',) and NR = -vV-CW'), a, b, c being the
c
c
principal semi-axes of the conicoid ; that is, OA, OB, OC in the figure.

THE ELLIPSOID.
5600 The equation referred to the principal axes of the
figure is

5601 There
equations are

are two planes of

circular

section whose

with a > b > c.

,(i-ii)+,(I-i)+,(A-i) = o
is a cone having a common section with the conicoid and a sphere of radius
r. If the common section be plane, one of the three terms must vanish
in order that the rest may be resolved into two factors.
Since a > b > c, the only possible solution for real factors is got by
making r = 6.
5602 Sections by planes parallel to the above are also
circles, and any other sections are ellipses.

5603 The umbilici of the ellipsoid (see 5777) are the points
whose coordinates are
c

Proof.The points of intersection of the planes (5601) and the ellipsoid

(5000) on the xz plane are given by x' = a W

z' = c W -j-g-

Since, by (5602) the vanishing circular sections are at the points in the m
plane conjugate to x' and z', we have, by (4352), x = z', t as'.
c
a

CONIGOIDS.

763

5604 If a = 6, in (5600), the figure becomes a spheroid, and

every plane parallel to xy makes a circular section. Hence
the spheroid is a surface of revolution. It is called prolate
or oblate according as the ellipse is made to revolve about its
major or minor axis.

THE HYPERBOLOID.
5605 The equation of a one-fold hyperboloid referred to its
principal axes is
4
a2 + ^-4
bl
c2 = 1-

(Fig. 173)

5606 The planes of circular section, when a>b>c, are all

parallel to one or other of the planes whose equations are

Proof.As in (5601), putting r = a.

5607 The generating lines of this surface belong to two
parallel systems (i.) and (ii.) below, with all values of 0.
5608
x
a
z a
= cos v sin v

= cos 0+ sin 6 ^
a
c
(i.),
4-= sin 6 cos 6
b
c

k..(n.).
-f= sin0+ cosfl

For the coordinates which satisfy either pair of equations,

(i.) or (ii.), satisfy also the equation of the siirface. The
equations may also be put in the forms
rCin
ODlO

x a cos 6
ybs'm0
, z
a = -;
n =
asinc*
ocos0
c

5612 If z = 0, x = a cos 0 and y = b sin 9.

Hence 9 is
the eccentric angle of the point in which the lines (i.) and (ii.)
intersect in the xy plane.
5613 Any two generating lines of opposite systems intersect,
but no two of the same system do.

764

SOLID GEOMETRY.

5614 If two generating lines of opposite systems be drawn

through the two points in the principal elliptic section whose
eccentric angles are 0 + a, 0 a, a being constant, the coordi
nates of the point of intersection will be
x = a cos 0 sec a, y = b sin 8 sec a, z = c tan a,
and the locus of the point, as 0 varies, will be the ellipse
5615
2 ll 2 i + -V2 , = 1 ; * = e tan a.
or sec2 o o2 sec2 a
Proof.From (i.) and (ii.), patting 0a for 0*

5616

The asymptotic cone is the surface given in (5595).

Proof.Any plane through the z axis whose equation is y = mx cuts the

hyperboloid and this cone in an hyperbola and its asymptotes respectively.

5617

The equation of a two-fold hyperboloid is

4j i = l.
a?

6*

(Fig. 174)

c2

and the equation of its asymptotic cone is

5618

4-^-4
= 0ar
62
c2

Proof.Any plane through the x axis, whose equation is y = mz, cuts

the hyperboloid and this cone in an hyperbola and its asymptotes respec
tively.
There are two surfaces, one the image of the other with regard to the
plane of yz. One only of these is shown in the diagram.

5619 The planes of circular section when b is > c are all

parallel to one or other of the planes whose joint equation is

Proof.As in (5601), putting r3 = b*.

5620 If b = c, the figure becomes an hyperboloid of revo

lution.
THE PARABOLOID.

5621 This surface is generated by a parabola which moves

with its vertex always on another parabola ; the axes of the
two curves being parallel and their planes at right angles.
* The surface of a one-fold hyperboloid, as generated by right lines, may frequently be
seen in the foot-Btool or work-basket constructed entirely of straight rods of cane or wicker.

CENTRAL QUADBIG SURFACE.

765

The paraboloid is elliptic or hyperbolic according as the

axes of the two parabolas extend in the same or opposite
directions.
5622

The equation of the elliptic paraboloid is

Zr + = x,
0
c

(Fig. 175)

b and c being the latera recta of the two parabolas.

Proof: QiP = b.OM; PN* = c.QN; ;. ^ + = OM+ QN = x.
o
c
If b = c, the figure becomes the paraboloid of revolution.
5623

Similarly the equation of the hyperbolic paraboloid is

Jr-^r = x-

5624

(Fig. 176)*

The equations of the generating lines of this surface

are
\/o

* = OT
y/c

and
y/b

T -4- = ,
y/ c
m

the upper signs giving one system of generators and the lower
signs another system.
5625

The equations of the asymptotic planes are

y/b * y/e

CENTRAL QUADRIC SURFACE.

TANGENT AND DIAMETRAL PLANES.

5626

Taking the equation of a central quadric -j + + -=

a1
b2
c2
= 1 to include both the ellipsoid and the two hyperboloids

The curvature of this surface is anticlmtic, a sort of curvature which may be seen in
the saddle of a mountain ; for instance, on the smooth Bward of some parts of the
Malvern Hills, Worcestershire,

766

SOLID GEOMETRY.

according to the signs of Ir and c2, the equation of the tangent

plane at xyz is

& +f + ^=L

By(5679)-

5627 If P he the length of the perpendicular from the origin

upon the tangent plane at xyz,
p2 ~ a* ^ b'^ c*'
Proof.From (5549) applied to (5626).

5628 The length of the perpendicular let fall from any

point ij upon the tangent plane at xyz is

5629

at xyz,

Direction cosines of the normal of the tangent plane

l = lf'

m = ^f>

n = ^-

Proof.By (5548) applied to (5626) and the value in (5627).

5630

If h w, n are the direction cosines of p,

p= lx-\-my-\-nz and p2 = uHi-{-btml-\-^ni.

Proof. (5G30) By projecting the three coordinates x, y, z upon p.

(5631) By substituting the values of x, y, z, obtained from (5029), in
(5630).

5632

The equation of the normal at xyz is

since the dir-cos. are the same as those of the tangent plane
at (5626).
5633 Each term of the above equations

or p multiplied into the length of the normal.

VENTRAL QUADRIG SURFACE.

Proof.Each term squared

767

= iLjZL = (* )' = ^

a*
fc*
Add numerators and denominators, and employ (5627).

c4

5634 Equation (5631) is the condition that the plane

lx+my+nz=p may touch the conicoid ; and if p = 0, we
have for the condition of the plane Ix -\-my-\-nz = 0 touching
,,2
?/2
z2
the cone -z + fi + ~i = 0,
a2
tr
r
5635
a*P+bim*+cini = 0.

5636 The section of the quadric made by a diametral plane

conjugate to the diameter through the point xijz has for its
equation

& + |Z + & = 0.

By (5688).

5637 Hence the relation between the direction cosines of

two conjugate diameters is
//'

mm

nri

ECCENTRIC VALUES OF THE COORDINATES.

5638

These are defined to be

x = a\,

y = b/i,

z = cv,

with

X2+/t2+v2 = 1.

5640 ^> p> v are the dir-cos. of a line called the eccentric
line; and = r\, = r/i, X, = rv are the coordinates of the
corresponding point upon an auxiliary sphere of radius r.
5641 The eccentric lines of two conjugate semi-diameters
are at right angles.
By (5637).
5642 The sum of the squares of three conjugate semidiameters is constant and = a2 + 62 + ca.
Peoof. Let a', h', o be tbe semi-diameters, and x^y^, rc22/jZs* 5!/3~3 their
extremities. Pat the eccentric values in the equations a^+y,+'J = a l> &c-i
and add. By (5641), X' + ^ + A^ = 1, Ac.

768

SOLID GEOMETRY.

5643 The sum of the squares of the reciprocals of the same

is also constant.
Proof. Put r, cos a r, cos /? r, oos y, for as,, y a, in the equation of the
quadric. So for z i/s, z3 and xs, </ zs. Divide by ru rt, r and add the results.

5644 The sum of squares of reciprocals of perpendiculars

on three conjugate tangent planes is constant.
Proof. For each perpendicular take (5627), and substitute the eccentric
values as in (5642).

5645 The sum of the squares of the areas of three con

jugate parallelograms is constant.
Proof.By the constant volume of the parallelopiped p^ = ptA, = p,A
(5648) and by (5644).

5646 The sum of the squares of the projections of three

conjugate semi-diameters upon a fixed line or plane is
constant.
Proof.With the same notation as in (5642), let (Imn) be the given line.
Substitute the eccentric values (5638) in (lxl + myi + nzl)--\-(lQs1 + 7nyt + nz,y
+ (lxi + myi + nzsy. In the case of the plane we shall have
a"1 (Za?! + myl + nz^1 + &c.

5647 Cor. The extremities of three conjugate semidiameters being x^z^ xiy.,z.i, x^i/3z3, it follows that, by pro
jecting upon each axis in turn,

5648 The parallelopiped contained by three conjugate semidiameters is of constant volume = abc.
Proof.By (5568), the volume

x{

yt

z, I = abc

x, yt zt j
*k

by the eccentric values (5638).

(584, I.).

y>

But the

z,

last determinant

=1 by

5649 Cor.If a, V, c' are the semi-conjugate diameters,

to the angle between a' and V', and p the perpendicular from
the origin upon the tangent plane parallel to a'b', the volume
of the parallelopiped is pa'b' sinw = abc.

CENTRAL QUADRIG SURFACE.

5650

769

Hence the area of. a central section in the plane of ah'

in
abc
irab sin co = is
.
P

5651

Quadratic for the semi-axis of a central section of the

g,2 = 1 made by the plane Ix + my + nz = 0 :
quadric -j + y% + ~ji
a
o
c
aH*

bhri2

c'n2

Proof. The equation is the condition, by (5635), that the plane

Ix + my + nz 0 may touch the cone

as in the Proof of (5600). For another method, see (1863).

5652
form

When the equation of the quadric is presented in the

aa?+by'+cJ+2Jyz+2gzx+2hxy = 1,

the quadratic for is takes

a
the form of the determi
nant equation annexed.
Or, by expanding, and
writing A' for the same
determinant, with the
fraction \ erased, the
r
equation becomes

b-Ar

m
= 0.

r
r

AV+ { {b +c) P+ (c+a) *+ ( + b) s - 2/ron - 2gnl - 2hlm] r

P-m-n = 0.
Proof. The equation of the cone of intersection of the sphere and
quadric now becomes
(a- I) x<+ (b- I) y+ (c- I) *2 +

+ 2gzx + 2hxy = 0,

and the condition of touching (5700) produces the determinant equation.

5 p

SOLID GEOMETRY.

770

5654

To find the axes of a non-central section of the

Let PNQ (Fig. 177) be the cutting plane. Take a parallel

central section BOC, axes OB, OC, and draw NP, NQ parallel
to them. These will be the axes of the section PNQ, and XQ
0 N2
NO2
wdl be found from the equation ^ T. +
= 1.
OA2
OC2
5655

The area of the same section

irabc / ^ _ p'2\

where p' and p are the perpendiculars from 0 upon the cutting
plane and the parallel tangent plane.
Proof.The area

= irNP.NQ =

OB. 00
OC

= , (l_ <g) OB. OC = ^ (l- ), by (5650).

SPHERO-CONICS.
Def.A sphero-conic is the curve of intersection of the
surface of a sphere with any conical surface of the second
degree whose vertex is the centre of the sphere.
Properties of cones of the second degree may be investi
gated by sphero-conics, and are analogous to the properties
of conies.
A collection of formula will be found at page 562 of Routh's Rigid
Dynamics, 3rd edition.

CONFOCAL QUADRICS.
5656
a2

Definition.The two quadrics whose equations are

b2

are confocal.

c2

a+\

oa+X

r-f A

We shall assume a > b > c.

5657 As A decreases from being large and positive, the

third axis of the confocal ellipsoid diminishes relatively to the

CENTRAL QUADRIC SURFACE.

771

others until A = c2, when the surface merges into the

focal ellipse on the xy plane,

X still diminishing, a series of one-fold hyperboloids appear

until X = b2, when the surface coincides with the focal
hyperbola on the zx plane,
vL
At
<|
=b2 ~ P=? =
The surface afterwards developes into a series of two-fold
hyperboloids until X = a2, when it becomes an imaginary
focal ellipse on the yz plane.
5658 Through any point xyz three confocal quadrics can be
drawn according to the three values of X furnished by the
second equation in (5656). That equation, cleared of frac
tions, becomes
5659

X8+(i+68+c,-#8-y8-*8)tf

{6V+cV+aW- (8* +e)

(c*+a2) y2- (a'+fr2) z2} X

+ aW - bW - c*ay - aW = 0.
These three confocals are respectively an ellipsoid, a one
fold hyperboloid, and a two-fold hyperboloid. See Figure
(178); P is the point xyz; the lines of intersection of the
ellipsoid with the two hyperboloids are DPE and FPU, and
the two hyperboloids themselves intersect in HPK.
Proof. Substitute for X successively in (5659) aJ, b1, c2, oo ; and the
left member of the equation will be found to take the signs + , , +,
accordingly. Consequently there are real roots between a1 and b2, V and r,
c2 and oo .
5660 Two confocal quadrics of different species cut each
other everywhere at right angles.
Proof.Let a, b, c ; a, V, c be the semi-axes of the two quadrics ; then,
at the line of intersection of the surfaces, we shall have
'(*-?M*-W+'(7-?)-*

772

SOLID GEOMETRY.

which, since a'1 a1 = V* I1 =

\, becomes the condition of per
pendicularity of the normals by the values in (5629). Thus, in (Fig- 1'8),
the tangents at P to the three lines of intersection of the surfaces are
mutually at right angles.
5661 If P be the point of intersection of three quadrics
aJhci> dil>-iC2, a3b3cs confocal with the quadric abc; the squares
of the semi-axes, dt, d^, of the diametral section conjugate to
P in the first quadric are (considering a1>a4>as, and writing
the suffixes in circular order)
dt = a\ ~ o\

d\ = a% 4,

In the second,

dl = ai ~ al

dl = a\ ~

in the third,

d$ = a\ ~ !

d\ = a\ ~ 4

Or, if for a\, a\, a\ we put a* + A1} a2 + A2, o2 + X3, the above
values may be read with A in the place of a and the same
suffixes.
Proof.Put cf a* = /*; then
+ 7i + ~5 = 1 and ~%
+ TT~ + 1
1
a\ bl
oJ-M V-P c|-/<
are confocal quadrics. Take the difference of the two equations, and we
obtain, at a common point x'y'z', -=;
+ &c- 0. Comparing this with
a* (a p)
(5651), the quadratic for the axes of the section of the quadric by the plane
lx+my +nz = 0, we see that, if Z, m, n have the values Ac., ft is identical
a
with ; the plane is the diametral plane of P ; and the two values of ft are
the squares of its axes. Let dj, be these values ; then, since there are bat
three confocals, the two values of p must give the remaining confocals, i.e.,
a, d, = oj and a* c/~' = aj.
The six axes of the sections are situated as shown in the
diagram (Fig. 179). Either axis of any of the three sections
is equal to one of the axes in one of the other sections, but
the equal axes are not those which coincide. 0 is supposed
to be the centre of the conicoids, and the three lines are drawn
from 0 parallel to the three tangents at P to the lines of in
tersection.

5662 Coordinates of the point of intersection of three con

focal quadrics in terms of the semi-axes :

CENTRAL QUADRIC SURFACE.

._
I i

2 2 2
Oi2"8
l*\f 2

2\>

_
&
/ 2

1*1*1*
P1O2P3
l?\/l*

773

2\

222
Cl C2 CS

2 _

(a!-C?)(6;-cf)The denominators may be in terms of any of the confocals

since a\ b\ = a\ b\ = a] bl, &c.
Proof.The equation of a confocal may be written + -j>rj + s
1
1
1
= 1, producing a cubic in aj, the product of whose roots a*, aj, a* gives sr.
5663 The perpendiculars from the origin upon the tangent
planes of the three confocal quadrics being plt p2, p3 :
2i2 2
22 2
,_
aA^i
2 _
a2b2c2
Pl
(al-aD(al-aiy
P>
(af-s)(*-*)'
nt
P>

2i2 2
a33C8
(al-al)(al-aiy

Proof.By (5649), pid^ = afiiC, ; then by the values in (5661).

EECIPROCAL AND ENVELOPING CONES.
5664 Def.A right line drawn through a fixed point always
perpendicular to the tangent plane of a cone generates the
reciprocal cone.
The enveloping cone of a quadric is the locus of all
tangents to the surface which pass through a fixed point
called the vertex.
5665
tively

The equations of a cone and its reciprocal are respec

A^+Bf+Cz' = 0

(i.),

and

! + ^+^ = 0

(ii.).

Proof.The equations of the tangent plane of (i.) at any point xyz, and
of the perpendicular to it from the origin, are
Axl+Byn + CzZ = 0

(iii.), and

i- =
=
Ax By Cz

(iv.).

Eliminate x, y, z between (i.), (iii.), and (iv.).

5667

The reciprocals of confocal cones are concyclic ; that

774

SOLID GEOMETRY.

is, have the same circular section ; and the reciprocals of concyclic cones are confocal.
Proof.A series of concyclic cones is given by
Ax' + B,/+Cz! + \(x*+y*+z') = 0
by yarying \ ; and the reciprocal cone is
TS + 51* 4,x
= 0.
A+ \ B+\ C+\

(5665)

5668

The reciprocals of the enveloping cones of the series

2^
2^
of confocal quadrics -=- + ^ _ -|- . _ = 1, with fnh for
ar + X
o~ + A
r+A
the common vertex, P, of the cones, are given by the equation

Proof.Let Imn be the direction of the perpendicular p from the origin

upon the tangent plane drawn from P to the qnadric. Equate the ordinary
value of 2>s at (5631) with that found by projecting OP uponp; thus
(a' + X) P+t&' + XW+Oj' + X)**' = (fl + gm + hn)1.
Now p generates with vertex 0 a cone similar and similarly situated to the
reciprocal cone with vertex P, and I, m, n are proportional to x, y, z, the
coordinates of any point on the former cone. Therefore, by transferring the
origin to P, the equation of the reciprocal cone is as stated.
5669 Cob. These reciprocal cones are concyclic ; and
therefore the enveloping cones are confocal (5667).
5670

The reciprocal cones in (5668) are all coaxal.

Proof. Transform the cone given by the terms in (5668) without X to

its principal axes ; and its equation becomes Ajt + By} + CV = 0. Now, if
the whole equation, including terms in X, be so transformed, j'! + y2 + z* will
not be altered. Therefore we shall obtain
(A+\) x2+(B + \) y-+(C + \) z2 = 0,
a series of coaxal cones.
5671 The axes of the enveloping cone are the three normals
to the three confocals passing through its vertex.
Proof.The enveloping cone becomes the tangent plane at P for a con
focal through P, and one axis in this case is the normal through P. Also
this axis is common to all the enveloping cones with the same vertex, by
(5670). But there are three confocals through P (5658), and therefore
three normals which must be the three axes of the enveloping cone.
5672

The equation of the enveloping cone of the quadric

THE GENERAL QUADRIC.

775

7r
7/
Z
-=r- +
. 4- -5r- = 1 is, when transformed to its prin
ts + A
w 4- A
c+A
cipal axes,
iV

If

55

(\

tXf

If

<2>

II

where X1} X2, X3 are the values of X for the three confocals
through P, the vertex, and d,, ds are the semi-axes of the
diametral section of P in the first confocal (5661).
Pboof.Transform equation (5G08) of the reciprocal of the enveloping
cone to its principal axes, as in (5670). Let X A,, A, be the values of A
which make the quadric become in turn the three confocal quadrics through
P. Then the reciprocal (A+\) x2 + (B + \) ^3 + (C+A) z! = 0 must become
a right line in each case because the enveloping cone becomes a plane.
Therefore one coefficient of x1,yt, or za must vanish. Hence .4 + A1 = 0I
jB + Aj = 0, C+A, = 0. Therefore the reciprocal cone becomes
(A-A,) x>+ (A- A,) /+(A-As) z> = 0,
and therefore the enveloping cone is
_!_+_!_ + *' =0.
X Xj X X2 X Xj

THE GENERAL EQUATION OF A QUADRIC.

5673 This equation will be referred to as f{x, y, z) = 0 or

U = 0, and, written in full, is
iu?+byl+czl+2ft/z+2gzx+2hxy+2px+2q!/+2rz+d = 0.
By introducing a fourth quasi variable t = 1, the equation
may be put in the homogeneous form
5674

*2+ btf+ c*+ dwt+2fyz+2gza:+2hxy

+2pxt+2qyt+2rzt = 0,

abbreviated into
(a, b, c, d,f, g, h, p, q, r)>, y, z, t)2 = 0,
as in (1620).
Transforming to an origin x'y'z and coordinate axes
parallel to the original ones, by substituting + y' + v, z' + E
for x, y, and z, the equation becomes, by (1514),

SOLID GEOMETRY.

776

5675

f +6^+^+2/^+2^+2/^
+(uM+vu,+tu.+u=o,

where U =f(x, y', z') (omitting the accents).

5676

The quadratic for /, the intercept between the point

x'y 'z and the quadric surface measured on a right line drawn
from xyz' in the direction linn, is
r8 (a/i+6?n2+cn2+2/inn+2g-/i/+2A/m)
+r (lUx+mU,+nUz)+U = 0.
Obtained by putting = rl, i = rm, 1 = rn in (5674).
5677 The tangents from any external point to a quadric are
proportional to the diameters parallel to them.
Proof.Prom (5676), as in (1215) and (4317).
5678 The equation of the tangent plane at a point xyz on
the quadric is

5679

or

R^+^^+C^+t^ = 0,

with t and i made equal to unity after differentiating.

Proof.From (5676). Since xyz is a point on the surface, one root of
the quadratic vanishes. In order that the line may now touch the surface,
the other root must also vanish; therefore lUx + mUll + nUz = 0. Pn'
rZ = i x, rm = nij, rn = ( z;
being now a variable point on the
line, and therefore on the tangent plane.
5680 Again,
zUx + yU, + zU. + tUt=2U, by (1624),
therefore
x JJX + y Uy + z Uz = tUt,
which establishes the second form (5679).
5681 Equation (5679) also represents the polar plane of
any point xyz not lying on the quadric surface. Written in
full it becomes
(ax+hy+gz+p)

or

x {a^+hrj +gt+p)

+17 (hx+ by+fz+q)

+y (ht+bii+fl+q)

p.v+qy+rz+d =0,

pH+qy+rt+d = 0.

TEE GENERAL QUADBIG.

5683

777

That is, the forms

(Ua+JiU,+tUt+U= 0

and

xUi+yUy)+zUi+ U= 0

are convertible, U standing for f(x, y, z) in the first, and for

/(, t), I) in the second.
5685 The intersection of the polar planes of two points is
called the polar line of the points.
5686 The polar plane of the vertex is the plane of contact
of the tangent cone.
Proof. If
be the vertex and xyz the point of contact, equation
(5683) is satisfied. If x, y, z be the variables and {ijf constant, the second
form of that equation shows that the points of contact all lie on the polar
plane of the point t)Z.
5687 Every line through the vertex is divided harmonically
by the quadric and the polar plane.
Proof.In equation (5684) put x = t + Rl, y = r) + Em, z = +Rn to
determine R, the distance from the vertex to the polar plane. This gives
R=
, employing (5680).
Now, if r, r are the roots of the quadratic (5076), with , i/, Z written for
2rr'
x, y, z, it appears that
-, = R, which proves the theorem.
r+r
5688 Every line (Imn) drawn through a point xyz parallel
to the polar plane of that point is bisected at the point, and
the condition of bisection is
Wx+mUy+nU. = 0.
Proof.The equation is the condition for equal roots of opposite signs in
the quadratic (5676). Since I, m, n are the dir. cos. of the line and Ux, Uy,
Uz those of the normal of the polar plane (5683), the equation shows that
the line and the normal are at right angles (5532).
5689 The last, when x, y, z are the variables, is also the
equation of the diametral plane conjugate to the direction hnn.
Expanded it becomes
(al+fun+gn) <v+(hl+bm-\-fn) y+{gl+fm-\-cn) z
+pl-\-qm-\-rn 0.
5 G

778

SOLID GEOMETRY.

For the point xyz moves, when x, y, z are variable, so that every diameter
drawn through it parallel to Imn is bisected by it, and the locus is, by the
form of the equation, a plane.
If the origin be at the centre of the quadric, p, q, and r of course vanish.
5690 The coordinates of the centre of the general quadric
U = 0 (5673) are
A'
A'
A'

Proof.Every line through xyz, the centre, is bisected by it.

tion for this, in (5688), is Ux =0, TJy= 0, and U, = 0, in order
pendent of Imn. The three equations in fnll are
h 9
ax + hy+gz+p
= 0"
z+p=0-)
f
hx+by+fzA =
'< + q = 0 ; and A' =
c
gx+fy + cz! + r = 0)
Solve by (582).

5691

The condi
to be inde
h
b
f

The quadric transformed to the centre becomes

Proof. By the last theorem, the terms involving {, jj, ( in (5675) vanish.
The value of U or / (x, y, z), when xyz is the centre, appears as follows :
U = \Ut (5680) =px+qy + rz + d =^&l*k+d (5690) = (1647).
The last equation, being again transformed by turning the
axes so as to remove the terms involving products of coordidinates, becomes
5692
5693

orl+j8y+ya>+^ = 0,
where a, (3, y are the roots of the discriminating cubic

RR* (a+b+c)+R (be+ea+ab-figiht) A = 0,

or (R-aXR-tyiR-^-iR-^ft-iR-fygt-iR-^h
-2fgh.= 0.
Proof.It has been shown, in (1847-9), that the roots of the discrimi
nating cubic ^multiplied in this case by
;j are the reciprocals of the
maximum and minimum values of x' + y,+z1. But such values are evidently

THE GENERAL QUADRIG.

779

the squares of the axes of the quadric surface. Let the central equation of
the surface be
+ if? + -7 = 1. Therefore -j =
&c, producing
a
b
e*
a'
A
the equation above.
5694 The equations of the new axis of x referred to the old
axes of $, ?, I are
(F+af) w = (G+ag) y = (H+ah)z;
and similar equations with /3 and -y for the y and z axes.
Proof.When Imn, in (5689), is a principal diameter of the quadric, the
diametral plane becomes perpendicular to it, and therefore the coefficients of
z, y, z must be proportional to I, m, n. Putting them equal to Rl, Rm, Rn
respectively, we have the equations
(aR) l+hm + gn = 0
(1) ")
The eliminant of these equations is
hl+ (6 R) m+fn = 0
(2) > .
the discriminating cubic in R algl+fm + (c-R) n = 0
(3) )
ready obtained in (5693).
From (1) and (2), I: m = hf-g (b-R) : gh-f(a-R),
and from (2) and (3), m : n =fgh (cR) : hf-g (bR) ;
therefore (gh-af+Rf) I = (hf-bg + Rg) m = (fgch+Rh) ii,
which establish the equations, since * : y : z = I : m : n and F = ghaf,
&c., as in (4665).
5695 The direction cosines of the axes of the quadric.
If the discriminating cubic be denoted by <f> (R) = 0, and
its roots by a, (3, y ; the direction cosines of the first axis are
/~gg
V
<k(a)

J~Ma)
v
^{ay

JZES
v
^(a)-

For the second and third axes write (3 and y in the place of a.
Proof.Let F+af=L, G+ag = M, H+ah = N
(i.),
(a-b)(a-c)-/' = X, (a-c)(a-a)-gl = ti, (a-a)(a- b)-h* = v...(ii.).
Then the equation ^ (a) = 0 may be put in either of the forms
L* = uv, M* = f\, N' = \u
Now the dir. cos. of the first axis are, by (5694), proportional to

(iii.).

\ if : \ = vx : ^ : ^ by (Hi )Their values are, therefore,

y/X
J(\+fi + vy

v/(X+/u + v)'

y/l'
^(X+fx + y)-

But X =_^M and \+(i + v = 'Qj?l, by actual differentiation of the

da
da
cubic in (5693).

SOLID GEOMETRY.

780

5696 Cauchy's proof that the roots of the discriminating cnbic (5693)
are all real will be found at (1850).
5697 The equation of the enveloping cone, vertex xyz, of
the general quadric surface U = 0 (5673) is
4 (abcfghXlmnyU = {Wx+mU,+nUz)\
with x, vy, tz substituted for I, m, n.
Proof.The generating line through xyz moves so as to touch the
quadric. Hence the quadratic in r (5676) must have eqnal roots. The equa
tion admits of some reduction.
5698 When U takes the form aof + bif+cz1 = 1, equation
(5697) becomes
(aP+bnf+ctfXaaf+btf+cJ-l) = (aLr+bmy+cnzy.
5699 The condition that the general quadric equation mayrepresent a cone is A'=0; that is, the discriminant of the
quaternary quadric, (5674) or (1644), must vanish.
Proof.By (5692). Otherwise A' = 0 is the eliminant of the four
equations Um = 0, U = 0, Ut = 0, U = 0, the condition that equation
(5675) may represent a cone.
5700 The condition that the plane
Ix+my + nz = 0 may touch the cone
(abcfgh^jxijzy = 0 is the determinant
equation on the right.

a
h
g
I

h g
b f
f c
m n

I
m
n

= 0.

Proof.Equate the coefficients Z, m, n to those of the tangent plane

(5681), p, q, r being zero, and xyz the point of contact. A fourth equation
is Ix + my + nz = 0, which holds at the point of contact. The eliminant of
the four equations is the determinant above.
5701 The condition that the
plane lx+my + nz + t = 0 may
touch the quadric
{abcdfcjhpqr\xyz\y = 0
(5673) is the determinant equation
on the right.
Proof.As in (5700).

* / q
for
q r d
m n
t

I
m
n
t

= 0.

RECIPBOCAL POLABS.

781

5702 If the origin is at the centre, p = q = r = 0. In that

case, transposing the last two rows and last two columns, the
determinant becomes
a
h
g
I
0

5703
planes

h g
b f
f c
m n
0 0

I
m
n
0
t

0
0
0
t
d

= 0,

or,

a
A b fm
g f c n
I m n

= t2

a h g
h b f
g f e

The condition that the line of intersection of the

Ix+my+nz+t = 0... (i.)

and

l'x-\-m'y-\-n'z-\-t' = 0 ... (ii.)

may touch the general quadric (abcdfghpqryxyzi)* = 0, is the

determinant equation deduced below.
Multiply equation (i.) by and (ii.) by n to obtain the piano
(lt+1'r,) x + (mt+m'ij) y+(nt+n'r,) z + tl + t',, = 0
(iii.),
passing through the intersection of (i.) and (ii.). The line of intersection
will touch the quadric if (iii.) coincides with the tangent plane at a point
seyz, and if xyz be also on (i.) and (ii.). Therefore, equating coefficients of
(iii.) and the tangent plane at xyz (5681), we get the six following equa
tions, the elimiuant of which furnishes the required condition,
ax+ hy+ gz + pw = K +
hx+ by + fz+ qw = ml + m'n
gx + fy + cz+ rw = ni + n'lj
px + qy + rz + dw = ti, + t'n
lx + my + rtz + tw = 0
Vx + my + n'z + t'w 0

h
b
f
1
m
m'

'J
f
c
r
n
u

I
in
u
t

I'
m'
ii
t' = 0.

RECIPROCAL POLARS.

5704 The method of reciprocal polars explained at page 665

is equally applicable to geometry of three dimensions.
Taking poles and polar planes with respect to a sphere of
reciprocation, we have the following rules analogous to those
on page 666.

782

SOLID GEOMETRY.

RULES FOR RECIPROCATING.

5705 A plane becomes a point.

5706 A plane at infinity becomes the origin.
5707 Several points on a straight line become as many planes
passing through another straight line. These lines are called
reciprocal lines.
5708 Points lying on a plane become planes passing through
a point, the pole of the plane.
5709 Points lying on a surface become planes enveloping the
reciprocal surface.
5710 Therefore, by rules (5708) and (5709), the points in
the intersection of the plane and a surface become planes
passing through the pole of the plane and enveloped both by
the reciprocal surface and by its tangent cone.
5711 Wlien the intersecting plane is at infinity, the vertex
of the tangent cone is the origin.
5712 Therefore the asymptotic cone of any surface is
orthogonal to the tangent cone drawn from the origin to the
reciprocal surface. The cones are therefore reciprocal.
5713 The reciprocal surface of the quadric is a hyperboloid,
an ellipsoid, or a paraboloid, according as the origin is without,
within, or upon the quadric surface.
5714 The angle subtended at the origin by two points is equal
to the angle between their cm-responding planes.
5715 The reciprocal of a sphere is a surface of revolution of
the second order.
5716 The shortest distance between two reciprocal lines
passes through the origin.
5717 The reciprocal surface of the general quadric
(abcdfghpqr\xyzVf = 0 (5674), the auxiliary sphere being

xi+yi+zi = V, is
a
h
g
p

h
b
f
q

g
f
c
r

or, if p = q = r = 0,
p
q
V
d a h g -k* a h g
r
I = 0,
hb fr,
h bf
d -k*
gfc
gfc C
S V CO

THEORY OF TORTUOUS CURVES.

783

Proof.The polar plane of the point ij with respect to the sphere is

& + vy + & k* = 0. This must touch the given surface, and the condition is
given in (5701).
5718 The reciprocal surface of the central quadric
sb2
v*
z%
+ ^ + = 1, when the origin of reciprocation is the
point x'y'z', is

or, with the origin at the centre,

5719
osr+ftV+cV = **Proof.Let p be the perpendicular from x'y'z' upon a tangent plane of
the quadric, and
the point where p produced, intersects the reciprocal
surface at a distance p from x'y'z'. Then
fcV1 =p = Ix'+my'+nz- ,/(a,P + &,m, + cW).
(5030)
Multiplying by p produces the desired equation.

THEORY OF TORTUOUS CURVES.

5721 Definitions. The osculating plane at any point of a
curve of double curvature, or tortuous curve* is the plane
containing either two consecutive tangents or three consecu
tive points.
5722 The principal normal is the normal in the osculating
plane. The radius of circular curvature coincides with this
normal in direction.
5723 The binormal is the normal perpendicular both to the
tangent and principal normal at the point.
5724 The osculating circle is the circle of curvature in the
osculating plane, and its centre, which is the centre of circular
curvature, is the point in which the osculating plane intersects
two consecutive normal planes of the curve.
5725 The angle of contingence, d\p, is the angle between two
consecutive tangents or principal normals. The angle of torsion,
dr, is the angle between two consecutive osculating planes.
* Otherwise named " space curve."

784

SOLID GEOMETRY.

5726 The rectifying plane at any point on the curve is per

pendicular to the principal normal ; and the intersection of
two consecutive rectifying planes is the rectifying line and
axis of the osculating cone.
5727 The osculating cone is a circular cone touching three
consecutive osculating planes and having its vertex at their
point of intersection.
The rectifying developable is the envelope of the rectifying
planes, and is so named because the curve, being a geodesic
on this surface, would become a straight line if the surface
were developed into a plane.
5728 The polar developable is the envelope of the normal
planes, being the locus of the line of intersection of two con
secutive normal planes. Three consecutive normal planes
intersect in a point which is the centre of spherical curvature :
for a sphere having that centre may be described passing
through four consecutive points of the curve.
5729 The edge of regression is the locus of the centre of
spherical curvature.
5730 The rectifying surface is the surface of centres (5773)
of the polar developable.
5731 An evolute of a curve is a geodesic line on the polar
developable. It is the line in which a free string would lie if
stretched between two points, one on the curve and one any
where on the smooth surface of the polar developable.
5732 In Figure (180) A, A', A", A'" are consecutive points on a curve.
The normal planes drawn through A and A' intersect in CE ; those through
A' and A" in G'E', and those through A" and A'" in C"E". CE meets CE'
in E, and G'E' meets C"E" in E'. The principal normals in the normal
planes are AG, A'G', A"G", and these are also the radii of curvature at
A, A', A", while C, G', C" are the centres of curvature. / ACA' = d\j/ and
GA'G' = dr.
The surface ECC'C'E' is the polar developable, CC'C" being the locus
of the centres of curvature, and EE'E" is the edge of regression.
EA is the radius and E the centre of spherical curvature for the point
A. hH, IIH', H'H" are elemental chords of an evolute of the curve, AhH
being a normal at A, and A'HH' a normal at A', and so on. The first
normal drawn is arbitrary, but it determines the position of all the rest.

THEORY OF TORTUOUS CURVES.

785

PROPERTIES OF A TORTUOUS CURVE.

5733 The equation of the osculating plane at a point xyz on
the curve is
(-)X+(i-y) /*+(-*) " = 0.
5734 X, A*> v ar6 the direction cosines of the binormal, and
their complete values are
p{y,z*.yi,*.\

p(z,.v2,zitx,),

p{petyuxuy,).

5735 The angle of contingence

dxjt = \/ { (y,Zuyu%.y+ (z.XuZi.x.)t+ {pc,yuxuy,y} ds.
Proof.Let the direction of a tangent be Imn, and that of a consecutive
tangent l + dl, m + dm, n + dn. Since the normal of the plane mast be per
pendicular to both these lines, we shall have, by (5532),
l\+mfi+nv = 0 and (l+dl) \ + (m + dm) p + (n + dn) v = 0,
therefore
\ : fi : v = mdn ndtn : ndlldn : Idm mdl,
and the denominator in the complete values of A, fi, v is
/{(mdn ndm,y + &c] = sin d\j/,
by (5521) ; that is, =
Also I, m, n = x y z, and dl = x^ds, &c.
ds Similarly, /i and v; and e+ = p, by (5146).
Therefore X = (y,zu 2/aiz) -37.
ay
5736

The radius of curvature p at a point xyz.

i _ 2 _i_,.2
_ Kt+y2t+zli~slt
~~r a'i*~Tyis'Tzi,
~*

p
st

Proof:
d<p= y{ (i/,z, yj,*,)'+ &c.} ds, in (5735),
therefore ^, = V { (<#,+yl+ *)
y*u+ <) - (xjh, + yj/u + ***)*}
= +lt+<4) i since *?.+y)+Z = i ;
and differentiating1 this equation makes xg XiB + &c. =0.
Otherwise, geometrically, precisely as in the proof of (5141), we find the
direction cosines of the principal normal to be
5737

cos a = pxu,

cos = pyUy

cos y = pz^.

T herefore
p1 ( x2^ + y2u + z^) = cos' a + cos1 /3 + cos' y = 1 .
The change to the independent variable < is made by (1762).
5738

The angle of torsion, in terms of X, /i, v of (5734), is

dr ^/{K+pl+v*.) ds = (Xa^+iufu+vzu) pds
= v/{(^ _fl,,)+(x,-F.x)+(x^,-Kp-y}
5 H

786

SOLID GEOMETRY.

Proof.By (574.5), we have (<Zr)' = (iX)'+ (dhy+ (dy)'

(i.),
which gives the first form. The third reduces to this by the method in
(5736). For the second form put = y,Zu~yitz$i &c-> then
X
n
v
1
ds
j.
du udK .
-=- = = -~ = (5734), d\= - - , &c.
u
v
w
K d\fi
K
A*
Substitute in (i.), reducing by E} = u'+v'+w' and EdK= udu+vde+icdv.

CURVATURE AND TORTUOSITY.

5739

Radius of curv., p = 4tJ

ay

Radius of torsion, <r =

or

Curvature = =
p
as
Tortuosity = =
aas

If t, changes sign while passing through the values

zero or infinity, there is a point of inflected torsion or
a cuspidal point, respectively. If r without changing sign,
passes through zero or infinity, there is a point of suspended
torsion or infinite torsion respectively.
If rt is zero, identically, the curve is plane.

5740

The radius of spherical curvature,

R = ^/(p*+pT).

Proof.In Fig. (180) R* = p1 + ECt and EC = pT by analogy with q=p^

n a plane curve (see proof of 5147).
5741 The element of arc of the locus of centres of circular
curvature is
ds' = Rdr, and therefore R = s'T.
Proof.In Fig. (180) ds' = 00' = pdr &ec<f> = Rdr.

5742

The radius of curvature of the edge of regression

= S" = RRP = p+ptr ,

S" being the arc of the edge of regression.

Proof.An inspection of Figure (180) shows that R and p stand in the
same relation to the edge of regression that r and p occupy with regard to a
curve in the standard formula. In fact we may substitute R for r, p for p,

THEORY OF TORTUOUS CURVES.

787

<j> for 8, t for </>, and <j> remains <p. The chosen line of reference AB being
always parallel to the tangent EG, then AEG = BAE = <f>. Also the angle
of contingence CEO' = GAC = dr, by the right angles at C and C. Ac
cordingly, we have the formula p = + = rrp =p +pi+ from (5146-8), and the
values above corresponding to them.
5743 -4 method of estimating the variation in direction of a
right line whose position is given as defending upon the form
of a tortuous curve at every point.
Let x, y, z be the direction cosines of the line referred to a
fixed principal normal, tangent, and binormal of the curve
[x, y, z may either be constants with respect to the varying
principal normal, tangent, and binormal, or they may be func
tions of the angle between the binormal and the spherical
radius].
5744 The complete changes in x, y, z, with respect to the
fixed origin and axes, will be
S.r = dx+ydtyzdr,
Sy = dyxd^i,
Sz = dz-\-ocdr .
Proof.In Figure (180) AG, AB are the fixed axes of x and z. Let a
line AL of unit length be drawn always parallel to the line in question ; then,
if x, y, z be the coordinates of L, x, y, z will also be the direction cosines of
AL, and therefore of the given line.
Now, suppose A to move to A', and consequently AL to take the position
A'L'. Then the changes in x, y, z will be the changes dx, dy, dz relatively to
the moving axes, plus the changes due to the rotations d\j/ round the
binormal and dr round the tangent. With the usual notation, we shall have
fix = cZaj + WgZ a;,?/, Sy = dy + <asx w1z, Sz = dz + wly wsa;,
with <i>, = 0, <i, = dr, u)t = dx//.
5745 If dx be the angular change in the direction of the
right line,
dx = V{(8xy+(8yy+(8z)*}.
For d\ = LL' since AL is a unit length.

Examples.
5746 The angle between two consecutive radii of circular
curvature being dt,
(def = (d+y+idry.

788

SOLID GEOMETRY.

Proof.Here, in (5744), x = \, y = 0, z=Q, therefore Sx = 0,Sy = dip,

Sz = dr. Substitute these values in (5745).
5747 The angle, dn, between two consecutive radii of
spherical curvature, <f> being the inclination to the binormal,
{dvY = (#.sin^)2+(rf^-dr)s.
Proof.In (5744) the direction cosines of R (Fig. 180) are x = sin
y = 0, z = cos f,
therefore Sx = cos^> (d$ dr), Sy = ttysin^, Sz = sin^ (cty dr).
Substitute in (5745).

5748 The angle of contingence of the locus of the centres

of circular curvature,
(dXy = (d+.costy+w+dTf.
Proof.The dir. cos. of the tangent at 0 to the locus (Fig. 177) are
x = cos f, y = 0, z = sin f ;
therefore Ix = sin^ (df + dr), $y = dt/ cos^, Sz = cos^ (dq + dr).
Substitute in (5745).
5749 The osculating plane of the same curve has its
direction cosines in the ratios
^ sin* cos*: -(*+*:): -**W*.
d\
\dx
dx!
dX
Proof.As in the Proof of (5735), the dir. cos. of the normal to this
plane are proportional to ylzzly, ztxx&z, xlyylx. Substitute the
values in last proof.
5750 The angle of torsion of the same curve is found from
(5745) and (5744) as above, x, y, z being in this case the dir.
cos. of the normal of the osculating plane as given in (5749).
5751

The direction cosines of the rectifying line are

q
'

dr
de '

rfaV
dc

Proof.The rectifying plane at A' (Fig. 180) is perpendicular to the

normal A'C. Therefore its equation is x ydi/z + zdr = 0. The ultimate
intersection of this plane with the rectifying plane at A (that is, the plane
of yz) is the rectifying line. Hence the equation of the latter is ydift = zdr
and the dir. cosines reduce to the above by (5746).

THEORY OF TORTUOUS CURVES.

5752

789

Cor.The vertical angle of the osculating cone

= 2tan-^.
ar

5753

The angle of torsion of the involute of the curve is

Proof.This angle is also the angle between two consecutive rectifying

lines. Therefore, taking the dir. cosines from (5751), we must put in (5744)
n
dr
dtb

therefore

5754

Bx = dip -- dr = 0 ;

Sy = ri(de ;

$z = \f/if de.

The angle of torsion of an evolute of the curve

= d\jt sin (a r).

Proof. (Fig. 180.) Let EH'H" be an evolute of the curve, AE the

tangent to it in the normal plane of the original curve at A, and let a = CAH,
the inclination of AE to the principal normal. At any other point H" of
the evolute, where its tangent is A 'H'H", let the corresponding angle be
0 = G"A"H". Then 6 = a T, r being the sum of the angles of torsion
between A and A", or the total amount of twist of the osculating plane. Now
the normal of the osculating plane of the evolute at H ". is perpendicular to
HH' and E'H", two consecutive tangents. Therefore its dir. cosines in
(5744) must be
as = sin (a r), y = 0, z = cos (a r) ;
therefore
Sx cos (a r) dr + 0 cos (a r) dr = 0,
iy = sin (a r) d\p ; Iz = sin (a r) dr sin (a r) dr = 0.
Hence the angle required = fy = dij/ sin (ar).

5755 Approximate values of the coordinates of a point on a

tortuous curve near to the origin in terms of the arc, the axes
of x, y, z being the principal normal, tangent, and binormal,
and the arc s being measured from the origin :

y = s

6p*^ 8p '

'

6p<r

24>\p&

p and a being respectively the radii of circular curvature and

torsion.

790

SOLID GEOMETRY.

Proof.By Taylor's theorem (1500), since x, y, z, $ are the same as

dx, dy, dz, d$ initially, we have x = x^ + lx^+^x^^+ic, and similar ex
pansions for y and z. The dir. cosines of the principal normal at the point
xyz will be, from (5737),
cos (-<W = P*, cos (| +f) = pry*, cos
= P*u;
4> = difi and r = dr being estimated positive as drawn in Figure (180) for
positive values of x, y, z.
Differentiate these eqnations for s, and in the results put the initial values
x, = z,= 0,

y, 1,

4> = r = 0, ^, = , r, = , Ac.,
p
a
to determine the derivatives in the above expansions.

THE HELIX.
5756 The helix is a curve traced on a cylinder of radius a,
so that its tangent preserves a constant inclination, = %ira,
to the axis. Taking the axis of the cylinder for the z axis of
coordinates, the equations of the helix are
x = a cos 6,

y = a sin 6,

z aO tan o.

5757

The radius of curvature p = a see4 a.

5758

The radius of torsion a = 2a cosec 2a.

Proof.p from (5806) ; since p, = a, p, = ao , and 6 = a at every point.

By (5739), a = 8T. But dz = dssiaa and adr=.dzcosa.
5759 The helix of closest contact with a given curve may
be found as follows.
Determine the constants a and a from equations (5757-8), with the
known values of p and a for the given curve ; then place the helix to have a
common tangent with the curve at the point, and make the osculating planes
coincide.

GENERAL THEORY OF SURFACES.

5770 Definitions.A tangent plane passes through three

consecutive points on a surface which are not in the same
right line.
5771 The normal at any point of a surface is perpendicular
to the tangent plane.

THEORY OF SURFACES.

791

5772 A normal plane is any plane through the normal.

5773 A line of curvature on a surface is a line along which
consecutive normals to the surface intersect. At every point
of a surface there are usually two lines of curvature at right
angles to each other ; and to these correspond two principal
radii of curvature. The two lines of curvature coincide with
the principal axes of the indicatrix at the point. See (5778).
5774 The surface of centres is the locus of the centres
of principal curvature. There are two such surfaces, for
there are two centres on each normal, and the normal is a
tangent to both surfaces. Either surface may be regarded as
generated by the evolutes of the lines of principal curvature.
5775 A geodesic is a line traced on a surface along which
the osculating plane at every point contains the normal to
the surface. See (5779).
5776 The radius of geodesic curvature * of a curve traced
on a surface is measured by the ratio of the element of arc of
the curve to the angle between consecutive normal sections
of the surface drawn through consecutive tangents of the
curve. Geodesic curvature, being the reciprocal of this, is
therefore the rate of angular deviation of the normal section
per unit length of the curve.
5777 An umbilicus is a point on a surface where a section
parallel to and close to the tangent plane is a circle ; in other
words, the indicatrix is a circle.
For a definition of Indicatrix, see (5795).
5778 In Figure (182) OCT) is the normal at 0 to a carved surface;
AQA', BOB' are the lines of cnrvature, therefore the normals to the surface
at A and 0 intersect in the centre of curvature radius pi (5773), and the
normals at B and 0, in the centre, radius p2. The normals to the line of
curvature BOB' at B and 0, drawn in the osculating plane BOB", intersect in
K, and those at B' and 0 intersect in II. HOD is the angle between the
osculating plane of the line of curvature and the plane of normal section.
Similarly for the line of curvature AOA'.
5779 I? POP' be a geodesic, its osculating plane POP' contains OB the
normal to the surface at 0, and therefore p = OD, the radius of cnrvature
of this section at 0 ; but PE, the normal to the surface at P, does not inter
sect OD, the consecutive normal at 0, unless the geodesic coincides with one
of the lines of curvature, OA or OB. The angle DPE is the angle of torsion
which vanishes in the latter case.
Not to be confounded with the radius of curvature of a geodesic.

792

SOLID GEOMETRY.

GENERAL EQUATION OF A SURFACE.

5780 Let the general equation of a surface be represented
by i> (x, y, z) = 0.
5781

The equations of any tangent at a point xyz are

tz = a = ^,
tr
lit,
ll

with

l^+mh+n*. = 0.

Proof.At an adjacent point x + rl, y + rm, z+rn, we have

<t> (x + rl, y+rm, z+rn) = 0,
therefore, by (1514), $ (x, y, z) +r (l<px+m(j>a + n^i) 0,
the rest vanishing in the limit. Bat <p
y, z) = 0, therefore
l<f>x+m$y+n<r-t = 0.
But I, m, n are the direction cosines of the line joining the two points, which
becomes a tangent in the limit ; and if
be any point on this line distant p
from xyz, t,x = pi, rj y = pm, (z = pn, &o.
5782

The equation of the tangent plane at xyz is

(f-*)
*,+(-) 4b = 0.
Proof.Eliminate I, m, n from lfx + mft/ + n^>s = 0 by lx = pi, Ac., as
above.
TANGENT LINE AND CONE AT A SINGULAR POINT.
5783 If, in the expansion in (5781) by Taylor's theorem, all the deriva
tives of ^ (x, y, z) of an order up to n inclusive vanish, we have
<j>(x+rl,y + rm, z + rn) = <p(x,y,z) +
(ldx + rrult + ndz)n* 1 <p (x, y, z) = 0.
w ~t~ 1
There are in this case n + 2 coincident points at xyz in the direction Imn,
and since the equation (ldx + mds/ + ndz)n*1 <p (x, y, z) = 0 is of the n + 1"1
degree in I, m, n ; n + 1 tangents to the surface at xyz can, in general, be
drawn in any given plane through that point. This equation now takes the
place of the conditional equation in (5781).
5784

Equation (5782) is now replaced by

{(-*)

<*,+({-*)

(*, y, z) = 0,

the equation of the locus of all tangents at the point xyz, and
representing a conical surface generated by the motion of
those tangents.
5785

The equation of the normal at xyz is

THEORY OF SURFACES.

793

5786 The equation of the tangent at a point x'y'z' on the

curve of intersection of the tangent plane at xyz with the
surface is
-x' _y-y' ^l-z'
X
ft
v '
with the two conditions

For these are the conditions of perpendicularity to the normals of the

tangent planes at xyz and x'y'z' respectively.
There are three exceptional cases in which the ratios
X : fi : v have more than one set of values ; namely
5787 I-When fa <j>y, <t>z vanish simultaneously, there is a
tangent cone at xyz.
5788 II- When $x., ^y, $z. vanish simultaneously, x'y'z' is
a singular point on the surface.
5789

=
=
In this case the point
fr'
fV
<Pz'
x'y'z' coincides with xyz, and the tangent there meets the curve
in more than two coincident points, the condition for which is

with

HI.When

(\dx+pd+vd,y<l> Or, y, z) = 0

(i.),

\xf>, +/*^ + v$, = 0

(ii.) .

These equations furnish two sets of values of the ratios

X : fx : v, giving thereby the directions of two inflexional
tangents (tangents to the curve of intersection) at xyz, each
meeting the surface in three coincident points. If all the
derivatives of an order less than n vanish at xyz, equation (i.)
will be replaced by (Xdx+fidy + vtL)n(t> (x, y, z) = 0, which,
together with (ii.), will determine n inflexional tangents at
the point.
5790 The polar equation of the tangent plane at the point
rO<f>, r' , &, <j>' being the variables, is, writing u for r-1,
' = (u cos 6 u$ sin 6) cos & + ( sin 0 + u$ cos 6) cos (^'<f>)sin6/
cosec 6 sin (<{>' <f>) sin ff.
5 i

SOLID GEOMETRY.

794

Proof.Write the polar equation of the plane through pa/3, the foot of
the perpendicular on the plane from the origin ; thus
pu = cos 8 cos a + sin 6 sin a cos (<j> ft).
Differentiate for 6 and <p to find pue and pu^, and eliminate p, or, and ft. This
elimination is troublesome.
5791 The length of the perpendicular from the origin upon
the tangent plane at xyz,

the second form being the value of p when the equation of

the surface is $ {x, y, z) = c, a constant, and when ^ is a
homogeneous function of the th degree (1624).
5793

In polar coordinates,
1
i , 2 , 2
/>
r2+rj+r! cosec2^
j = uf+ul+ul cosec-0 = !*\
.
pr

Proof.Add together the values of the squares of pu, pue, and pu, found
in (5790).
For a geometrical proof, see Frost and Wolstenholme, Art. (314).

THE INDICATRIX CONIC.

5795 Def.The indicatrix at any point of a surface is the
curve in which the surface is intersected by a plane drawn
parallel to the tangent plane at that point and infinitely near
to it.
5796 The following abbreviations will be employed
The derivatives of <p (x, y, z), ^ <t>is, fc,, $fS, >, f
a>,, a>,, f
will be denoted by

a,

b,

c,

f,

g,

h,

I, m, n.

5797 Pbop. The indicatrix at a point xyz of a surface

<p (x, y, z) = 0 is the conic in which the elementary quadric
surface
5798

I.

P
is intersected by the tangent plane at xyz, whose equation is
5799

H.

l{+mr)+nl+iN=0.

THEORY OF SURFACES.

795

The origin of coordinates is the point xyz in both equa

tions. B is an indefinitely small radius from the centre of
the quadric (I.) to a point \vl on the indicatrix, and p is the
radius of curvature of the section of the surface <j> by a normal
plane drawn through B ; the ratio B2 : p being constant for
all such planes.
Peoof.Let 0, in Fig. (181), be the point xyz on the surface f; x+,
y + n, z + an adjacent point P. Then
<t> (x + 1, y + V, z + 0 = <f (, y, z) + U + wij + nt+ * (a?+ . . . + 2hln) + &c.
With xyz for origin, draw the quadric surface
a? + bn> + c(,+2fvt; + 2gtf+2hiTi = N
(i.)
and the plane
U, + mr) + nZ+N = 0
(ii.).
Since i, i}, '(, are very small, N is likewise. Also the unwritten terms in the
above expansion may be neglected in the limit. Hence, any point
lying
on the intersection of the quadric (i.) and the plane (ii.) will also lie on the
original surface f + y + v, z + () = 0.
To determine N, we have the perpendicular from xyz upon the plane (ii.),
N
rr- (5549).
rp = 2-
y(P + TO*z + n)
v
' The radius of curvature of the section of the
surface $ made by a normal plane at 0 drawn through P being p, we have
p = , and therefore N = ^(P+wi'+n*).
In the Figure, ii = OP, p = OL, and the intersection of (i.) and (ii.) is
the conic PSQ. Since p is indefinitely small, we may put N = 0 in equa
tion (ii.). This amounts to taking the parallel section of the quadric by the
tangent plane at 0 instead of the section PSQ. But these two will be equal
in all respects, since the section of the quadric is a central one.
5800 If m = 0, equation (II.) becomes l + nl = 0, and if
the inclination of the indicatrix plane to the plane of xy be a,
tan a = -. To obtain, in this case, the equation of the
n
indicatrix in its own plane, put = ' cos a, l = t sin a, and
t) = ', in equation (I.).
5801 When none of the three constants I, m, n are zero,
the quadric (I.) simplifies as follows
From (II.) we have l^+mri = nl and two similar equa
tions. Square these, and by the results eliminate the terms
in nl,
S>? from (I.), which then becomes
5802
where

HI.

H*+KV*+LF=N,

H = a+ (lfmgnli),
K = b+
(mgnhlf),
ran
nl
L = c+j- (nhlfmg).

SOLID GEOMETRY.

796

This is the equation of another quadric intersecting the

plane (II.) in the indicatrix.

5803 The equation of a surface for points near an origin 0

(Fig. 182), the normal at 0 being taken for z axis, is

Pi

r- = 2a,
Pi

where px, p2 are the radii of curvature of the normal sections

through the x and y axes, and those sections will be proved to
be the lines of curvature at 0.
Proof.Let AC = a and BC = b be the semi-axes of the indicatrix conic
at a small distance z from 0 (5795). The equation of the conic will there
in' + u* = 1 ; but
o* = 2p, and
6* = 2p giving the equation
fore be j
a
o
z
%
required.
Secondly, on a line of curvature, the normal to the surface at a point ryz
will intersect the z axis (5773). The condition for this, by (5533) [with
xyz for abc, the origin for a'b'c,
L,M,N= fn <t A, (5785) = , % -2, and If, M', N' =0,0,1],
Pi P:
gives xy [
J = 0, therefore * = 0ory = 0ona line of curvature.
yP>
P'7
Q. E. D.

5804

If the equation of the surface with the same axes be

z = aa?+2hxy + by2 + 2fyz + 2gzx + cz* + higher powers,

then

Pl - ,

Pi = 1.

Proof.Put y = 0 and divide by z, therefore l = a +2gx + cz+&c.

z
When x and z vanish, we have 1 = lapi.
5805

For a normal section making an angle 0 with AC,

-i- = 2 (a cos2 6+2h sin 6 cos 6+b sin8 6).

Proof.Turning the axes in (5804) through the angle 6 by (4049), the

coefficient of x'2 becomes a cos' 0 + as above.
5806

Euler's Theorem.If p be the radius of curvature of

TEEOBY OF SURFACES.

797

any other normal section at 0, making an angle AGP = 6 with

AC (Fig. 182),
1
cos2 6
sin2 e
p ~

Pi

pi

Proof.Let r = CP; then x = r cos 6, y = r sin 6, and i3 = 2pz, which

snbstitnte in (5793).
5807 Cob. The sum of the curvatures of two normal
sections at right angles to each other is constant ; or, if />, p
be the radii of curvature for those sections, and pa, pb the
radii for the principal sections,

Pa

Pi,

5808 The radius of curvature of a normal section varies as

the square of the radius of the indicatrix in that section.
Peoof.Prom r1 = 2pz, in Figure (182).
5809 Meunier's Theorem. The radius of curvature of an
oblique section of a surface is equal to the radius of curvature
of the normal section through the same tangent multiplied by
the cosine of the inclination of the planes.
t>
. 1Q,^
PN> therefore
,
p
NO = cos <p,
Proof.(Fig.
183.) p , = PN'
-, p = ,
-=-when NO and NG vanish.
5810 Quadratic for yx at a point on the surface z = <j> (x, y)
giving the direction of the principal normal sections, and,
therefore, of the lines of curvature (notation 1815).
{ pqt - (1 +92) s} jrl+ { (1 +/) t - (1+ <f) r) y,
+ {(l+p*)s-pqr] =0.
Proof.(i.) The equations of the normals at the consecutive points xyz
and x + dx, y + dy, z + dz of the surface <p (x, y, z) = 0 are
lx _ r/y _ (z an(j (x + dx) _ v(y + di/) _ ((z + dz)
fx
<t>t
'%
$x + d<px
f^ + d^
f. + df,
5811

The condition of intersection is, by (5533),

dx dy dz
1
y*
=
0,
or
fx
P
?
f.
r + syx s + tyx

p+w*
-l
= 0,
0

798

SOLID GEOMETRY.

by dividing the first row by dx, and putting zx= <t>x + <t>,yx, <fyx = *tr+0jy
&c. The form of <p (x, y, z) being, in this case, <j> (*, y) z, <fz becomes 1,
and d<px becomes zero. The determinant equation produces tbe quadratic.
(ii.) Otherwise. Consider ij; the point of intersection of consecutive
normals. The equations of a normal being
l - ^ =
or - = t>(*-0 and r,-y = q (i-Q.
p
q
-1
Differentiate both equations for x, considering , ij, ( constant and p, q func
tions of x and y ; the results are
l + (r+syx)(z-i;)+p<J>+qy*) = 0 and yx+(s+tyx)(z-o+q(p+qyx)=0.
Eliminate z to obtain the quadratic in yx.
5812 If the equation of the surface be in the form <j> (x, y, z) = 0, the
quadratic for yx may be obtained in the same way. The requisite substitu
tions in the first determinant are found from <px-r <pvyx + <t>zzx 0, giving zt;
d<px = <t>2z + <pIl/yx + </>X2zx, &c, and with the notation of (5796) the determin
ant equation and quadratic for yx becomes
n
ny,
-G + myx)
I
m
n
= 0.
an gl + (hn gm) yx hnfl + (bnfrn)yx gncl + (fncm)yx
5813 The above determinant, or the corresponding one in (5810), is the
differential equation of the lines of curvature.

5814 The radii of curvature of the principal normal sections

of the surface <p (x, y, z) 0 at a point xyz are given by the
following quadratic, in which A' is the bordered determinant
in (5700), and the notation is that of (5796) and (1620).
A>2+ {(a+b+cW+m'+n^-iabcfghXlmny} kP-hA = 0,
where lcl = P+mP + n2.
Proof.The quadratic in (5653) applied to a section of the quadric (I.)
(5798) by the plane (II.), becomes
b'R* + { (b + c) V + (c + a) m' + (o + 6) n' - 2/mn - 2gnl - 2Mm } NR*
whose roots, being the two values of R', are the squares of the principal
semi-axes of the indicatrix. Put R* = ~, as in the Proof of (5797).
K
5815 Otherwise, the quadratic in (5651) might be applied to a section of
the qnadric (III.) (5802) by the plane (I.).
5816

If the equation of the surface be given in the form

THEORY OF SURFACES.

799

z = <p(x, y), the quadratic becomes [writing, as in (1815),

p, q, r, s, t for zx, zy, zu, z, z2y],

(r*_^ f-{(i+pF) t-2pq*+(l+tf) r) kp+k* = 0,

where 7c2 = p* + q1 + 1 .
Otherwise, this equation may be found from the two equations obtained
in the second proof of (5810), by eliminating yx instead of z .

5817 The radius of curvature at a point xyz on the surface

rp (x, y, z) = 0 of the normal section whose tangent has the
direction cosines X, p, v is, with the notation of (5796) and
(1620)'

y(/2+m2+ft2)

(abcfghXW

Proof.From equation (I.) (5798), since J, n, I are respectively equal

to R\, R/i, and Rv.

5818 The curvature at any point of a surface $ (x, y, z) = 0

is termed elliptic or synclastic, hyperbolic or anticlastic, and
parabolic or cylindrical, according as the indicatrix is an
ellipse, hyperbola, or two parallel right lines, or according as
the principal curvatures have the same signs, opposite signs,
or one of them vanishes ; and this will be according as the
determinant A', in (5814), or sirt, in (5816), is negative,
positive, or zero.
Pkoof.The rule follows at once from the consideration that the two
values of p in the quadratic of (5814) must have the same sign in the first
case, different signs in the second, and that one value must be infinite in the
third case.

5819 The condition for an umbilicus is that the indicatrix

must be a circle ; therefore, either (III.) (5802) must be a
sphere, or, if it be a quadric surface, the plane (II.) must
make a circular section of it, and therefore either I, m, or n
must vanish.
5820 Otherwise, the quadratic in (5814) or (5816) must
have equal roots.
5821 Otherwise, the conditions for an umbilicus on the sur
face 0 (x, y, z) = 0 are the two equations
bn2+cm22fmn __ cP+ari1 2gnl _ ami-\-bP2hlm

800

SOLID GEOMETRY.

Proof.The radius of the indicatrix, and therefore also p in (5817), is

constant for all values of X, fi, v. Now, by (5817),
k;
a\3 + &c. =
P
.: (a-^\,+ (b--y)iA*+(c-^ *, + 2ff*v + 2gv\ + 2h\p = 0,
and l\ + mfi + nv = 0, since \/tv is always tangential, and Imn is normal to
the surface. As these equations are true for all values of X, p, v, the second
expression must be a factor of the first. The quotient, by division, is therefore
(0_!U + (6_*)jl + (c_*w.
\
p I I
\
pI m
\
pin
Equating to zero each of the three coefficients of the remainder, and elimi
nating p, we obtain the above conditions.
5822 If a common factor of the three fractions in (5821)
exists, that factor equated to zero is the differential equation
of a line of spherical curvature at every point of which there
is an umbilicus. If the fractions are identically equal, the
surface has an umbilicus at every point, and must therefore
be a sphere.
5823 The number of umbilici on a surface of the nth degree
cannot exceed n (I0ns 25w + 16). Salman, p. 208.

5824 The condition that the indicatrix may be a rectangular

hvperbola is
(a+6+c)(Zs+m4+n2) = (abcfghXlmny.
Proof.The quadratic in (5814) must have equal roots of opposite
signs.
Similarly, when z = <f> (x, y) is the equation of the quadric, the condition
becomes
(l+p1) t2pqs+(l + q') r = 0.
(5816)
5825 The condition that the indicatrix may become two
coinciding lines.
Here equation I. (5798) must represent a cone, and the plane (II.)
must touch it. Hence N = 0, and, if ( be eliminated, the quadratic for the
ratio ; : q obtained is
(on' + cP 2jnZ)? + 2 (clm-fnlgmn + hn1) ij -Kin1 -rem1 2/ntn) ij1 = 0,
and this must have equal roots.

CURVATURE OF A SURFACE.
5826 Defs.Integral curvature of a closed surface is equal
to the area of that part of the surface of a sphere of unit
radius which is intercepted by radii drawn parallel to the

THEORY OF SURFACES.

801

normals at all points of the given surface. This area also

measures the solid angle of the cone generated by the radii.
The curve on the sphere is called the horograph of the curve
on the original surface. In other words, integral curvature
of a closed surface is the area of the horograph of its
boundary.
5827 Average curvature is the integral curvature divided by
the area of the surface.
Specific curvature is the average curvature of a small
element at the point; i.e., * ' --(ds)1 = -.
P1P2
PiPi
5828 The last is the usual measure of curvature at a point,
and its value in coordinates of the point is given by
1

A'

rts*

(579 .

according as (j> (x, y, z) = 0 or z = <j> (x, y) is the form of the

equation to the surface.
Proof.From the product of roots of the quadratics in (5814) and
(5816).
5829 In a plane curve integral curvature is the plane angle
contained by the terminal normals, and average curvature is
the integral curvature divided by the length of the curve.

5830 Another measure of curvature at a given point of a

surface is the ratio of the area of the indicatrix to the area of
the indicatrix cut off by the same plane on a sphere of unit
radius .which touches the surface internally at the point. This
measure is = \/ pipt.
Proof.Putting AG Rl, BC = Rt, in Fig. (182), and OG = z, the area
of the indicatrix of the surface is irB,Ji, at an ellipsoidal point. But
R\ = 2pj* and R\ = 2p,z, therefore 7ri?,E, = 2irz v/(PiP3)- Also the indicatrix
of the sphere = 2tt since p, = pa = 1 for the sphere.

5831 The radius of curvature of any normal section at a

point P of an ellipsoid (Fig. 184) is equal to the square of
the semi-diameter parallel to the tangent of that section,
5 K

SOLID GEOMETRY.

802

divided by the perpendicular from P upon the diametral

plane conjugate to OP.
Proof. Let AOB be the plane parallel to the tangent plane at P;
OA = d, the semi-diameter in it parallel to the given tangent PT. Draw
PR perpendicular to OA and PN=p perpendicular to the plane AOB. The
radius of cnrvatnre at P of the elliptic section PA =
(4536). Therefore,
PR
by (5809), the radius of curvature of the normal section through the same
tangent PT, will be P = ^ x |g = .

5832 The principal radii of curvature at P, viz. pu pt, are

found from their sum and product thus : putting y for OP,
and a, b, c for the semi-axes of the ellipsoid,
Pi+Pt =

Pi Pi =

PnooF.Let a, p be the semi-axes of the section AOB (Fig. 184), then

a'+fi + y* = a' + b' + c' (5642) and pafi = abc (5648). By these values
eliminate a, /3 from pi+P, = and p,p. =
(5831).
V
F
5833 The lines of curvature on a quadric surface are its
intersections with the confocal quadrics.
Proof. Let the quadric and confocal be the ellipsoid and one-fold
hyperboloid in (Fig. 178) intersecting in the line DPE, and let their equa
tions be, as in (5656),

At any point P on the line of intersection z, y, z satisfy the three following

equations :
First, the differential of (ii.),
+
+
= 0.
Second, the difference of (i.) and (ii.),
*
+
y'
+

=0

Third, the difference of their differentials,

xdx
,
ydy
,
zdz

a2(a' + X) b>(b* + k) c'(cl+\)~ '

The eliminant of these equations in x, y, z pro
duces the determinant equation here annexed, which,
by (5811), is the condition for the intersection of con
secutive normals. Hence this condition holds for
every point of the line of intersection of (i.) and (ii.).

dx dy dz
X i z
V?
= 0.
dz
dx
a' f

THEORY OF SURFACES.

803

The general method of determining the lines of curvature

of a surface from the differential equation in (5811) is here
exemplified in the case of an ellipsoid.
5834 The determinant just written gives for the differential equation of
the lines of curvature
(i'-c5) xdydz + (c'-a') ydzdx + ia'-V) zdxdy = 0
(i.).
To solve this, multiply by -= and substitute for z and dz from the equation
c
of the quadric. The result is of the form
Azyyl+tf-AifB) yx-xy = 0,
in which A =

B=

^p ' ; or, multiplying by

A ^ (xyyt-y*)-Byy* +(xyy,-y*) = 0,
x
x
which is of the form in (3236). Solving by that method, we find that the
two equations
= <x and xyy, y1 = /3 have the common primitive
x
ax*y, = /3, which, with the relation Aa/3B,+f}=0, constitutes the
solution. The result is that the projections of the lines of curvature upon
the xy plane are a series of conies coaxal with the principal section of the
ellipsoid, and having their axes a, b varying according to the equation
a2(a2-c')
_ ,
oa(o2-6s) b\b'-dl)
At an umbilicus y = 0, therefore, equation (i.) becomes [(62 c')xdz
+ (a2 bi) zdx] dy = 0. Here dy = 0, being a solution, gives y C = 0,
showing that the plane of zx contains a line of curvature. The other
factor, equated to zero, taken with the differential equation of the curve
<?xdx + a*zdz = 0, gives the coordinates of the umbilicus, as in (5603).
OSCULATING PLANE OF A LINE OP CURVATURE.
5835 Let (j> be the angle between the osculating plane and
the normal section through the same line of curvature, ds an
element of the other line of curvature, and p, p their radii of
curvature respectively : then
as

p p

ProofFig. (185). Let OA, OB be the lines of enrvature; OP, AP

consecutive normals along OA and OS, BS the same along OB. Also, let
BQ, OQ be consecutive normals along the line of curvature BC. Then,
ultimately, OP = p, OS = p, BQ = p + dp. Also, let QP produced meet the
osculating plane of AO in R. Join RO and RA, and draw QN at right angles
to PS. Since the tangent to AO at O is perpendicular to the plane OBQP
and that at A to AGQP, it follows that both tangents are perpendicular to
QP, which must therefore be perpendicular to the osculating plane ARO.
Hence <j> or ROP = PQN.

804

SOLID GEOMETRY.

Now m = ^R
ds

SB

t^=^, ... tan, = ^ = . +- ultimately.

p

NQ

as

p p

5836 At every point on a line of curvature of a central

conicoid pd is constant, where d is the semi-diameter parallel
to the tangent at the point and p is the perpendicular from
the ceutre upon the tangent plane.
Proof.Let the first and third confocals in (56G1) be fixed, and there
fore ci] and a, constant. Draw the second confocal through the point of
contact P of the tangent plane (Pig. 178). Then, by (5663), pxa\ and ptd\
are constant along the line of intersection of the first and third surface,
because, by (5661), da = a' a\ and d\ = ^ a^

GEODESIC LINES.

5837
are

The equations of a geodesic on the surface <p (x, y,z) = 0

= &*. = %2 .
9x
&
<f>;

Proof.The osculating plane of the curve contains the normal to the

surface (5775) ; therefore, by (5737) and (5785).

5838 -A. geodesic is a line of maximum or minimum distance

along the surface between two points.
Proof. The curve drawn in the osculating plane from one point to a
contiguous point is shorter than any other by Meunier's theorem (5809),
for any oblique section has a shorter radius of curvature and therefore a
longer arc. A succession of minimum arcs, however, may constitute a maxi
mum curve distance between the extreme points ; for example, two points on
a sphere can be joined by either of two arcs of a great circle, the one being
a minimum and the other a maximum geodesic.
5839 A surface of revolution such as the terrestrial globe affords a good
illustration. A meridian and a parallel of latitude drawn through a point
near the pole are the two lines of curvature at the point. The meridian is
also a geodesic, but the parallel is evidently not, for its plane does not
contain the normal to the surface.

5840 A geodesic is the line in which a string would lie if

stretched over the convex side of a smooth surface between
two fixed points.
Proof.Any small arc of the string POP' (Fig. 182) is acted upon by
tensions along the tangents at P and P', and by the normal reaction of the
surface at 0. But. these three forces act in the osculating plane (5775) ;
therefore the string will rest in equilibrium on the surface in that plane.

THEORY OF SURFACES.

805

Cor.Two equal geodesies drawn from a point and in

definitely near to each other are at right angles to the line
which joins their extremities.
5841 If a geodesic has a constant inclination to a fixed line,
the normals along it will be at right angles to that line.
Proof.Let Imn be the fixed line and a the constant angle ; then
lx, + my, + nz, = cos a, and therefore Zasj, + myM + nzu = 0.
Therefore, by (5837), the principal normal is at right angles to Imn.
Example.The helix, the axis being the fixed line.

5842 On any central conicoid pd is constant along a geo

desic, where p is the perpendicular from the centre upon the
tangent plane and d is the semi-diameter parallel to the
tangent of the geodesic.
Proof. (Fig. 186.) Let AT, BT be the tangents at the two extremities
of a small geodesic arc AB, and let the tangent planes at A and B be ADO
and BOB. AT and BT make equal angles with OB, by the property of
shortest distance, for if the plane BOB be turned about OB until it coincides
with the plane ABO, ATB will become a straight line, and therefore
*ATB = BTC = i, say.
Let id be the angle between the tangent planes ; let the perpendiculars
upon those planes from A, B be A M = q, BN= q, and from the centre of the
quadric p, p ; and let xyz and x'y'z' be the points A, B. Then
q = ATe\ni sinw, q = BT sin i sin u>, .. q : q' = AT : BT
(i.),

,-/+$ +-i) am, ,--#($+0+$-i).

therefore
q '. q ' = p' '. p
(ii-)Again, let d, d' be the semi-diameters parallel to AT and BT. Then, by
(5677), AT : BT = d : d' ; therefore p' : p = d : d' or pd =p'd' ; that is,
pd is constant.

5843 If a line of curvature be plane, that plane makes a

constant angle with the tangent plane to the surface.
Proof. Let PQ, QR, RS be equal consecutive elements of the line of
curvature. The consecutive normals to the surface bisect PQ and QR and
meet in a point. Therefore they are equally inclined to the plane PQR.
Similarly the second and third normals are equally inclined to the plane QRS,
and so on. Hence, if the curve be plane, all the normals are equally inclined
to its plane. Hence also the following theorem.

5844 Lancret's Theorem.The variation in the angle be

tween the tangent plane and the osculating plane of a line of

806

SOLID GEOMETRY.

curvature is equal to the angle between consecutive osculating

planes.
5845 Cor. If a geodesic be either a line of curvature or a
plane curve, it is both, but a plane line of curvature, as in
(5839), is not necessarily a geodesic.
GEODESIC CURVATURE.

Theorem.The square of the curvature at any point of a

curve traced on a surface is equal to the sum of the squares
of the normal and geodesic curvatures (5776), or
5846

7 = ^ + ^'

where p is the radius of curvature of the normal section and

p" the radius of geodesic curvature. Also, if <f> be the angle
between the plane of normal section and the osculating plane,
5847

p = p" sin <f> = p cos <f>.

Proof. Let PQ = QR (Fig. 187) be consecutive elements of any carve

traced on a surface. Produce PQ to S, making QS = PQ. Let QT = PQ
be the consecutive elements of the section of the surface drawn through
PQS and the normal at Q. Join RS, ST, TR. PQSR is the osculating
plane of the curve PQR. PQST is the plane of normal section, and there
fore PQT is a geodesic. QRT is the tangent plane, and STR is a right
angle.
Then, putting SQR = <fy, SQT = <ty', RQT = <fy", R8T = ^, we have
ds
,
ds
,,
ds
/tr?-c\

p=w p =w p =w

Therefore

{hlt)

._..__. mf.

Also

^ = ^4t = || = cos *, as in (5809).

p
ds.dy
aa
Thus both theorems are proved. Note that p' is the radius of curvature of
the geodesic PQT, while p ' is the radius of geodesic curvature of PQR.

RADIUS OF TORSION OF A GEODESIC.

5848 If 9 be the angle between the geodesic and one of the

lines of curvature ; plt p.z the principal radii of normal curva
ture, and a the radius of torsion,
= (
<r

\py

) sin 0 cos 6.
pi/

THEORY OP SURFACES.

80?

Proof.(Fig. 182.) Let OP = ds be the geodesic, OA, OB the lines of

curvature, and 0 = AGP. The angle of torsion dr measures the rotation of
the normal to the surface round OP = ds. But this angle is equal to the
sum of the rotations of the normal round OA and OB resolved along ds.
For, in travelling along each of the lines CN and NP, which are in the direc
tions of the lines of curvature, the normal rotates only about the other.
Therefore, if w wi be the rotations round OA, OB, dr = w, cos 0 + , sin 0.
ds sin 0 , w, = ds cos 8 ; .-.
1 = -=(
dr
/ 1
1 \ sin
afl cos a.
a
But w, = '
Pt

Pi

ds

\ft

ft/

5849 The product pd has the same value for all geodesies
which touch the same line of curvature.
Proof.By theorems (5836) and (5842), since the product where they
touch it must be the same as that for the line of curvature.

5850 The product pd has the same value for all geodesies
drawn through any umbilicus on a conicoid.
Proof. The semi-diameter d, in this case, is the radius of a circular
section, and therefore equal to the mean semi-axis b for all the geodesies ;
and p is the same for all.

5851 The geodesies drawn through any point on a conicoid

to two umbilici make equal angles with either line of curva
ture through the point.
Proof. pd is the same for each geodesic, by the last, and p is the same
for each ; therefore d is the same, that is, the diameters parallel to the two
geodesies at the point are equal ; therefore they are equally inclined to each
axis of their section ; but these axes are parallel to the lines of curvature
(5803) ; therefore, Ac.

5852 Hence the geodesies joining any point to two opposite

umbilici lying on the same diameter are continuations of each
other.
5853 The sum of the distances of any point on a line of
curvature from two interior umbilici is constant ; and the
difference of the distances from one interior and one exterior
umbilicus is constant.
Proof.Geometrically, as in the analogous theorem for the focal distances
in a conic, if r, r are the distances and r + dr, r' + dr' the distances for a
consecutive point on the line of curvature, it follows from (5851) that
dr = dr' for interior umbilici and dr = dr' for exterior ones.

5854

A system of lines of curvature and the umbilici on a

808

SOLID GEOMETRY.

quadric surface has therefore analogous properties with a

system of confocal conies and their foci in a plane, the geo
desies corresponding to straight lines.
5855 In the same way, every surface has a geodesic geo
metry proper to itself ; spherical trigonometry, for instance,
being the geodesic geometry of the sphere.

INVARIANTS.

INVARIANTS OF A SINGLE FUNCTION.

5856 The constancy of the ratio Bi : p in equation (5798)

gives rise to the following invariant forms. Since the quadric
surface I and the tangent plane II are the same for all posi
tions of the coordinate axes, they have been called respec
tively the invariable quadric and the invariable plane. As a
consequence,
5857
#+4i+ti
is an invariant of tj> (x, y, z).
Proof. By (5791), since the perpendicular from the origin upon the
invariable plane is constant. Also, the coefficients of the discriminating
cubic (5693) of the invariable quadric will not be altered by transformation
of axes. Therefore the following are also invariant forms :

5858

&+&+&>

5860

&&&+2<rSf<rVr& &*$* <ta#&f#fir

5861 A similar theorem applied to a function <j> (x, y) of

two variables gives the invariable conic and invariable line ;
namely,
f*<fc*+2f#+i?2<fc, = l and $M-i^ = lj
and from these the invariants,
5863
5866

#+ #>

<Ptx + <P>y>

<Ptx<Piy <t>%,>

xj>yy<f>xi xfa+yf..

QUADRATURE AND GUBATURE.

809

Proof.The last two invariants are obtained from the cosine of the
angle between the invariable line (5862) and the fixed line ylxt) = 0,
joining the point xy with the origin, or the fixed line xZ+yy = 0.

INVARIANTS OF TWO FUNCTIONS.

5868

An invariant of the two functions <(> (x, y), \p (x, y) is

Proof. Form the cosine of the angle between the invariable lines
tyw+trfr = 1 and i^a,+ i>i/'J, = 1, observing (5863).
Also the two following expressions are invariants,
5869

KK+M*-tyj*

5870

<M*,+ <k,<r%+2<k,rV

Proof.From the invariable couics (5861) of ^ and \f>, we get

Xfc.) P + 2 (fm +
, +
+ X*)
invariable for any value of X. Hence the coefficients of the several powers
of X in the invariant
(ft. + Mu) (fcr + XK) ~ tin + XKY
are also invariants. This gives (5869). Subtracting the latter from the
invariant (<t>u + 92!/)(4'a + 4'2i,) produces (5870).

INTEGRALS FOR VOLUMES AND SURFACES.

5871 If F be the volume included between the surface

z = ^ {x, y), three rectangular coordinate planes, the cylindri
cal surface y = i/> (x), and the plane x = a,
Fig. of (1906)
5872

V = ^dxdydz = ^zdxdy.

For the limits and demonstration, see (1906).

5874 The area of the surface $ (x, y, z) = 0 or z =/(, y)
will be
S = $^W'+fi+Wdxdij or S=fjs/(l+zl+zl)dxd!f.
5 L

810

SOLID GEOMETRY.

Pboof.The area of the element whose projection is dxdy will be

dxdy secy, where y is its inclination to the plane of xy, and therefore the
angle between the normal and the z axis. Therefore
secy = vT>l + f!+fl) +. = ^(1+4 + 4). by (1708).

5875 Let the equation of a surface APB (Fig. 188) in polar

coordinates be r=f(9, <p), and let V be the volume of the
sector contained by the planes AOB, AOP, including an angle
<j> = PHG, the given surface APB, and the portion OPB of
the surface of a right cone whose vertex is 0, axis OA, and
semi-vertical angle 0 = AOB or AOP ; then
sin 0dddd>.
(I. I!

Proof.Through P, any point on the surface, describe a spherical sur

face POD, with centre 0 and radius r = OP. The volume of the elemental
pyramid, vertex 0, base Pe, = \r.Pf.Pg = \r.rdd .rs'mOdf. Here the
error of the small portions, like PE, ultimately disappears in the summation,
since the volume of PE, being equal to \dr.rd0.r sin 6d<j>, is of the third
order of small quantities ; and so in similar instances.

5876

The area of the same surface APB (Fig. 188) is

S = f \*ry/{(i*+r]) sin'tf+r^} ddd<f>.
Jo Jo

Proof.Let the perpendicular from 0 upon the tangent plane at P to

the given surface be ON = p. The element of
area PE = area Pe . ^- = r dO . r sin Bd<t> . = r*ame d8d<p.
ON
p
p
Substitute the value of p in (5793).

SURFACE OF REVOLUTION.

If y =/(*) (Fig. 90) be the generating curve, and the x

axis the axis of revolution, V the volume, and 8 the surface
included between the planes x = a, x = b;
5877
V = fyUv,
S = fb2iry v/(l +yl) dx.

Proof.The volume of the elemental cylinder of radius y and height dx

is Ky'dx. The element of the surface of revolution is
2wy ds = 2*ysxdx = 2*-y ./(l +t) dg>,
(5113)

QUADRATURE AND OUBATURE.

811

Guldin's Rules.When the generating curve of a surface

of revolution is a closed curve, and does not cut the axis of
revolution, a solid annulus, or ring, is formed.
5879 Rule I.The volume of the solid ring is equal to the
area of the generating curve multiplied by the circumference of
the circle described by the centroid * of the area.
5880 Rule II. The surface of the ring is equal to the
perimeter of the generating curve multiplied by the circum
ference described by the centroid of the perimeter.
Pkoop.Let A be the area of the closed curve, and dA any element of A
at a distance y from the axis of revolt!tion. The volume generated
= j 2vydA = 2jt j ydA = 2*yA,
by the definition of the centroid (5885), y being its distance from the axis.
Similarly, if P be the perimeter, writing P instead of A.

Quadrature of surfaces bounded by lines of constant

gradient.
5881 Defining the curve (y) as the locus of a point on the
given surface at which the normal has the constant inclina
tion y to the 2 axis ; let F (y) be the projection of the area
bounded by the curve (y) upon the xy plane ; then the area
itself will be found from the formula,
S=(ysecyF'(y)dy.
PfiOOF.The element of area between two consecutive curves (y) and
(y + dy) projected on the xy plane will be dF(y) = F'(y) dy ; and, since the
slope is the same throughout the curve (y), this projected element must be
equal to the corresponding element of the surface multiplied by cos y.

5882 Rule. Equate coefficients of the equation of the

tangent plane with those of l+mj-l-n = p, and eliminate 1
and mfrom l2-|-m2+n2 = 1. The result will be an equation in
x, y and n = cos y, representing the projection of the curve (y)
upon the xy plane. From this F (y) must be found.
5883 ^x- Taking the elliptic paraboloid
h*~ =2z; the tangent
plane at xyz is ~ + - ( = z. Equating coefficients of the last with
l+im) + n =p, and substituting for I and m in P+m' + 5=l, we obtain
as* + tt
* = tan'y. The area of this ellipse
for the projection on the xy plane, j Centre of m<u, or gravity.

SOLID GEOMETRY.

812

is F (y) = jrab tan* y, and therefore F' (y) = 2nab tan y sec1 y. Consequently,
by (5881),

8 = lirab j\an y sec'ydy = \*db (sec'y-l).

CENTRE OF MASS.

5884 Definitions.The moment of a body with respect to a

plane is the sum of the products of each element of mass of
the body and the distance of the element from the plane.
5885 The distance (denoted by x) of the centre of mass *
from the same plane is equal to the moment of the body
divided by its mass.
5886 Note.If the body be of uniform density, as is supposed to be the
case in all the following examples, assume unity for the density, and read
volume instead of mass in the above definitions.
The definition gives the following formulas for the position
of the centre of mass of a uniform body :
5887

For a plane curve,

-_$xds
( t i
jib

JjVq+jflfto
fr cos gy/(r*+Qrffl

HZ
^
. fOllO)
f
**
Wif+O &

For y, change x into y and cos 0 into sin 0 ; but observe that in all cases,
if the body be symmetrical about the axis of x, y vanishes. The formula
gives the centre of volume of the portion of the curve included between the
limits of integration.
For a plane area,
5890

x = ^xdxdy = $xydx

The area is bounded by the curve, the x axis, and the ordinates x = a,
x = b, if such be the limits of integration.
For a plane sectorial area bounded by two radii SP = r,
SP = r' (Fig. 28) and the curve r = F{B) ;
* Also called centre ofgravity or inertia, and more recently centraid.

CENTRE OF MASS.

jJrrfWr

813

5893

J^rf*

'

.rq^
5894

7, - tfr* sin 0d0dr

jVsingtffl
V~
tirdOdr
=
jrW
"

The second forms for i and y give the centroid of an area like SPP'
(Fig. 28). The double integrals applied to that figure require the limits of
the integration for r to be from 0 to F(d), and afterwards for 8, from
0, = ASP to 0, = ASP". But, if applied to the area in (Fig. 109), the order
of integration must be reversed, as explained in (5209).
For a surface of revolution round the x axis,
5896

x - hi

dx - ^ sin 6 cosg v/(y2+Q dd

W(l+3)d*

jVsin<V(r*+r')</0

Proof.By (5885), for the moment = j" a; . 2iryd and the area = ^2xyds ;
the second form by (5116). If x = a, x = b are the limits of integration, the
surface is bounded by the parallel planes x = a, x = b ; and in the second
form, the corresponding values of 0 are the limits defining the same parallel
planes.
For any surface,
5898
- = W(l+4+4)^fr
(5874)
JM1+4+-S)
For a soZid of revolution round the a; axis,
5899

_
f.tyd.y
HV sin 5 cosddddr
a? =
=^-^

.
Itfdw
Mr- sin 0d6dr

Proof. By (5885), for the moment =^x.wy*dx and the volume

= f vy'dx. The limits as in (5896).
5901 For any solid figure bounded as described in (5871),
the coordinates of the centroid are given by
Vx = y^x dx dy dz = \jxz

dy,

Vy =^ydxdydz =^yzdxdy,
Vz =

|* z dx dy dz \^zidxdy,

SOLID GEOMETRY.

814

where

V 1 j j dxdydz \\ zdxdy,

as in (5872-3), and the limits are as defined in (1906).

5902 For the wedge shaped solid (OAPB, Fig. 188) defined
by the polar coordinates r, 0, <j>, as in (5875),

where
Proof.By (5875); multiplying the elementary pyramid \r* sinddOdf
separately by the distances of its centroid from the coordinate planes ; viz.,
frsinflcos^, frsin0sin^>, and frcosfl.

MOMENTS AND PRODUCTS OF INERTIA.

5903 Definitions.The moment of inertia of a body about
a given right line or axis is the sum of the products of each
element of mass and the square of its distance from the line.
5904 The square of the radius of gyration of the body about
the given line is equal to the moment of inertia of the body
divided by its mass.
5905 The moment of inertia of a body with respect to a
plane or point is the sum of the products of each element of
mass and the square of its distance from the plane or point.
5906 The product of inertia of a body with respect to two
rectangular coordinate planes is the sum of the products of
each element of mass and its distances from the two planes.
5907 Let -4> B, G be the moments of inertia of a body
about three rectangular axes ; A', B1, C the moments of
inertia with respect to the three planes of yz, zx, and xy ; and

MOMENTS OF INERTIA.

815

F, G, H the products of inertia with respect to the second

and third planes, the third and first, and the first and second
respectively. F, G, H are more frequently called the products
of inertia about the axes of yz, zx, and xy respectively.
By the definitions we have the values
5908

5914

A = 2m

F = Zmyz,

B = Zm (V+.r2),

G = tmzx,

C = Xm (a?

H = Xmocy.

A' = Xmx* = SA
where S = '
B = %my* = S-B

C = tmz*= S-C ,

5920 Theorem I. The M. I. of a lamina about an axis per

pendicular to its plane is equal to the sum of the two M. I.
about any two axes in its plane drawn through the foot of the
perpendicular axis and at right angles to each other.
Proof.By the definition (5903), and Enc. I. 47.
5921 Theorem II.The M. I. of a body about a given axis,
plane, or point is equal to the M. I. about a parallel axis or
plane through the centroid, or about the centroid itself
respectively, plus the M. I. of the whole mass, supposed col
lected at the centroid, about the given axis, plane, or point.
Proof.In the figure, p. 168, let the given axis be perpendicular to the
paper at B ; let A be the centroid, and m an element of mass at C ; then, for
every thin section of the solid parallel to the paper,
M. I. = Xm .BO1 = 2m (AC*+AB>-2AB.AD)
= 2lm.AC, + 2,m.AB%-2AB.'2lm.AD.
But 2m. AD = 0, by (5885), since A is the centroid of the body, which
proves the proposition. Similarly for the plane or point.
Cor. I.Hence, if the M. I. about any axis is known, that
about any parallel axis can be found without integration. For,
let 7X be the M. I. about a given axis, whose distance from
the centroid is a, and let It be the required M. I. about an
axis whose distance from the centroid is b ; then, by Theorem
II.,
Ia = I1-m(ai-b2).

816

SOLID GEOMETRY.

Cor. II. The M. I. has the same value for all parallel
axes at the same distance from the centroid.
5922 Theorem III. The product of inertia for two assigned
axes is equal to the product for two parallel axes through the
centroid of the body plus the product taken for the whole
mass collected at the centroid with respect to the assigned
axes.
Proof.Let x = +x', y = y + y' be the coordinates of an element of the
body with respect to the assigned axes ; x, y being the coordinates of the
centroid, and x', y' the coordinates of the same element with respect to
parallel axes through the centroid, all axes being parallel to z. Then
2mxy = 2ot (x + x')(y + y') xy2m+2mx'y' +x2my' +y"Zmx
= xy 2m + Sroz'y'.
Since Smas' and %my' vanish by the definition of the centroid.
5923 The M. I. of a body with respect to a point is equal
to the M. I. for any plane through the point plus the M. I.
about the normal to the plane through the point.
Proof.For the origin and yz plane,
2mx1 + 2m (y1 + z1) = 2mr.

(5908, '14, '19)

5924 Given the moments and products of inertia, A, B, G,

F, G, E, as above, about three rectangular axes, the moment
of inertia of the body about a line through the origin, whose
direction cosines are I, m, n, will be
I = AF+Bmt+Cnt-2Fmn+2Gnl+2Hlm.
Proof.(Fig. 11.) Let xyz be a point P of the body, OM the line imn,
and PM the perpendicular upon it. Then the M. I. about OM
= 2m (OF-OAf1) = 2,m{(z' + y' + z3)(t' + m'+nt)-(lx+my+nz)i} (5530)
producing the above result, by (5908-13).
ELLIPSOIDS OF INERTIA.
5925

The equation of the Momental Ellipsoid is

obtained by putting Ir* = Mt*. M being the mass of the body,

and * a constant to make the equation homogeneous. Hence the
square of the radius of the momental ellipsoidfor any point varies
inversely as the moment of inertia of the body about that radius.

MOMENTS OF INERTIA.

817

5926 If the products of inertia vanish, the axes are called

the principal axes of the body.
5927 At every point of a body there are always three
principal rectangular axes.
Peoof.These are evidently the principal axes of the momental ellipsoid
of the point ; for if the coordinate axes be made to coincide with the former,
F, 0, M will vanish.

5928 The equation of the momental ellipsoid referred to its

principal axes will be
5929

The moment of inertia about a line Imn will now be

I=AF+Bm*+Cn2.
THE ELLIPSOID OF GYRATION.

5930 The equation of the Ellipsoid of Gyration referred to

principal axes is
A +B + C ~ M
It is the reciprocal surface of the momental ellipsoid (5719),
and its property is
5931 The moment of inertia about the perpendicular from
the origin upon the tangent plane varies as the square of the
perpendicular.
5932 For any other rectangular axes through the point, the
equation of the ellipsoid of gyration is, by (5717),
0, being the reciprocal surface of
A -H -G x
the momental ellipsoid,
-E B -F y
(A,B,G, -F, -G, -HJxyzy
-G -F
G z
= M,
1
with the radius of the sphere of
*
~ M
reciprocation = 1. The equation when expanded becomes

A H G
5933

(BC-Fi)**+...+2(FG+CH)ay= H
H B
B F
F \.
G F C
5 M

SOLID GEOMETRY.

818

LEGENDRE'S EQUI-MOMENTAL ELLIPSOID.

5934

The equation is

A' + B' ^ C ~ M'

with the values in (5914).
5935 The mass of this ellipsoid is taken equal to that of
the body, and it has the same principal moments of inertia.

THE MOMENTAL ELLIPSOID FOR A PLANE.

5936 If A', B', C be the moments of inertia for the three
coordinate planes, as in (5914), the M. I. for a plane through
the origin whose dir. cos. are /, m, n, will be
V = A'F+Bm2+Cri1+2Fmn+2Gnl+2Hlm.
Proof:
5937

V = 2m (lz + my + nz)* = 2nue,.Z1 + &c. = AT + &C.

The momental ellipsoid for this plane will be

A'x*+B'yt+C'zl+2Fyz+2Gzji'+2Hxy = Me*,

and its property is

5938 The M. I. for any plane passing through the centre of
the ellipsoid is equal to the inverse square of the radius per
pendicular to the plane.
5939 If t be a radius of this ellipsoid, and a, b, c its semiaxes, the M. I. about r
-J-+J-+!_!
~ a* ^ 6- + c1
r*
Proof. (Fig. 11.) M. I. about r, plus M. I. for the plane OM perpen
dicular to r
= 2mOP' = Irnx' + Smtf + Smz' = \ + i + ~, by (5938).
a
o
<r

EQUI-MOMENTAL CONE.
5940 The equation of the equi-momental cone at any point
of a body, referred to principal axes of the body at the point,

MOMENTS OF INERTIA.

is

819

(A-I)A*+{B-I)f+(C-I) s2= 0,

its property being that

5941 The generating line passes through the given point, and
moves so that the M.I. about it is a constant = I.
Proof.Let Imn be the generating Hue in one position, then
AP + Bm'+Cn* = /(P + r^ + zi2). Therefore, &c.
5942 Theorem.If two systems have the same mass, the
same centroid, principal axes and principal moments of inertia
at the centroid, they have equal moments of inertia about any
right line whatever, and are termed equi-momental. By (5906)
and (5929).
5943 If two bodies are equi-momental, their projections are
equi-momental.
Proof.If the projection be from the xy plane in the ratio 1 : n, the
coordinates x, y, z of a particle become x, y, nz, and the mass m becomes nm.
The conditions in (5942) will then be fulfilled.
MOMENT OF INERTIA OF A TRIANGLE.
5944 The M. I. of a triangle ABD (Fig. 190) about a side
BD, distant^ from the opposite vertex A, is
j _ mp*
Proof.Let BD = a and EF = y ; I = fMp-y) ,fjy = ^ = mP*
"
J
p
9
12
0
5945 The M. I. of a triangle ABC (Fig. 190) about a
straight line BD passing through a vertex B, and distant p
and q from the vertices A and G, is

b
Proof.By (5944), taking difference of M. I. of the triangles ABD, CBD.
5946 The M. I. of a triangle ABC about an axis through
its centroid parallel to BD, is
/ = m?-Pj+<f.

By (5921)

820

SOLID GEOMETRY.

5947 Cor.If the triangle be isosceles, so that p = q, the

last two moments of inertia become
and
2

18

5949 The M. I. of the triangle about axes perpendicular to

ABC through B and through the centroid, respectively, are
mirrir1

and

(5920)

5951 The M. I. about GF of the trapezoid ACGF (Fig.

190), is
m

5952 The moments and products of inertia of a triangle

about any axes are the same for three equal particles, each
one-third of the mass of the triangle, placed at the mid-points
of its sides.
Proof.(Fig. 190.) The M. I. of the three particles at the mid-points
of AB, BO, OA about BD, any line through a vertex, will be

which is equal to that of the triangle, by (5945).

MOMENTAL ELLIPSE.
5953 If a, /3 be the radii of gyration of a plane area to
principal axes Ox, Oy, where 0 is the centroid, the equation
of the momental ellipse for the point 0 will be
aV+j9y = 2a22.
5954 Also the area is equi-momental with three equal
particles, each one one-third of its mass placed anywhere on
the ellipse so that 0 may be their centroid.
Proof.Let xy, x'y', x"y" be the coordinates of three equi-momental
particles : then
2-(**+x* + x"^ = mfP |l(j,' + ^ + y") =
xy + xy' + x'y = 0;

MOMENTS OF INEBTIA.

821

and the two systems have the same centroid ; therefore

x + x + x" = 0 and y +y" + y'" = 0.
Eliminating
y', x", y" between the five equations, we find the equation of
(5953) for the locus of xy.
5955 The momental ellipse for the centroid of a triangle is
the inscribed ellipse touching the sides at their mid-points.
Pkoof.(Fig. 189.) The inscribed ellipse, which touches two sides at their
mid-points, also touches the third side at its mid-point, by Carnot's theorem
(4779). Now DF is parallel to AO, the tangent at E ; therefore BE, which
bisects DF, passes through the centre 0 of the ellipse. Similarly, AD
passes through it ; therefore 0 is the centroid of the triangle.
Let OE = a, and let 6' be the semi-diameter parallel to AO ; then
Q*jl + E^l = l. But ON =
therefore FN* = ft'2. The M. I. about
OE, by (5954), = \m%b'2 sin'w = mj, where a, b are the semi-axes.
Hence the M. I. about OD, OE, OF varies inversely as the squares of those
lines, and therefore the ellipse in the diagram is a momental ellipse, since it
has six points which fulfil the requirements.
5956 The projections of a plane area and its momental
ellipse form another plane area and its momental ellipse. (5943)

5957 The M. I. of a tetrahedron ABCD about any plane

through A is

where a, /3, -y are the perpendiculars on the plane from B, G, D.

5958 The tetrahedron is also equi- momental with four
particles, each one -twentieth of the mass, placed at the
vertices, and a particle equal to the remaining mass placed at
the centroid (5942).
5959 The equi-momental ellipsoid of a tetrahedron has the
same centroid, and touches each edge at its middle point.
Peoof. By projecting a regular tetrahedron and its eqm-momental
sphere (for the centroid) of radius = </3 X radius of inscribed sphere.
5960 To find the point, if it exists, in a given right line at
which the line is a principal axis, and to find the other prin
cipal axes at the point.

822

SOLID GEOMETRY.

Let 0 be a datum point in the line. Take this for origin,

the given line for axis of z, and OX, OY for the other axes.
Then, if h be the distance from 0 to the required point O',
and 0 the angle between OX and the principal axis O'X',
5961

,1=^VE = ^
Amy
z,inx

and

tan2* = -^L,
AB

where A, B, H are the moments and product of inertia about

OX, OY.
Proof. At the point 0, 0, h, 2m (z h) x = 2m (z h) y = 0, from
winch h is fonnd ; and the equation for 6 is that for determining the prin
cipal axes of the elliptic section of the momental ellipsoid, whose eqnation
is Ax' + 2Hxy + Bif = Me\ as in (4408).
5964 The equality of the two ratios in (5961) is the condi
tion that the z axis should be a principal axis at some point of
its length.
5965 If an axis be a principal axis at more than one point
of its length, it passes through the centroid of the system ;
and, conversely, if it be a principal axis at the centroid, it is
so at every point of its length.
Proof.For h must be indeterminate in (59C1).
TLmy = 0, Xmzx 0, 2mx = 0.

Therefore Zmy: = 0,

5966 The principal axes O'X', OY' are parallel to the

principal axes of the projection of the body in the original
plane of xy. By (5962-3).
5967 Given the principal axes of a body at its centroid, to
find the principal axes and moments of inertia at any point in
the principal plane of xy.
Let C in the Figure of (1171) be the centroid, CX, GY
principal axes, A, B the M. I. about them, and P the given
point. Find two points 8, S', called foci of inertia, such that
the X and Y moments of inertia there are equal, and therefore
B + m . CS- = A ;

giving

CS = OS' = JA~ . . . (i.).

'
m

The internal and external bisectors of the angle SPS' will be

two of the principal axes at P, and the third will be the normal
to the plane.

MOMENTS OF INEETIA.

823

Proof.The X and Y principal moments being equal at S, the moment

about every line through S in this plane is the same. [For I = AP + Bm*
+ Cn* and n = 0 and A = B, therefore I = A.~] Therefore the moments
about SP and S'P are equal. Therefore the bisectors PT, PO of the angles
at P will be principal axes.
5968 Let SY, S'Y' be the perpendiculars on PT, and SZ,
S'Z' those upon PG ; then the M. I. about PT and PG will be
respectively,

Pkook.Draw OE perpendicular to SY. The M. I. abont OR (0 = SCR)

= Acoste+BBmie (5929) = A-(A-B) sin20
= A-m OS* sinJ 0 (by i.) = A-mSB?.
Therefore M. I. abont PT = A-mSE? + mRY- (5921)
= A+m(RY+SR)(RY-SR) = A + mSY.S'Y'
= ^ + mJ3Cs (1178) = B+mAC1 (by i.) = B + m ^ggg[j^',
Similarly for the M. I. about PG.
5969 Hence, if an ellipse or hyperbola be described with
S, S' for foci, the tangent and normal at any point of the
curve are principal axes, and the M. I. about either is constant
for that curve.
5970 Similarly, for a point P in any plane through the
centroid 0, it may be shown that the same construction will
give the axes PT, PG about which the product of inertia
vanishes, OX, OY being the axes at 0 in the given plane about
which the product of inertia vanishes.
5971 The condition for the existence of a point in a body
at which the M. I. about every axis through it shall be the
same, is
There must be two principal a,xes of equal moment at the
centroid, and the M. I. about each must be less than the third
principal moment.
Two such points will then exist situated on the axis of
unequal moment, and equi-distant from the centroid.

824

SOLID GEOMETRY.

5972 Given the principal axes at the centroid of a body and

the moments of inertia about them, to find the principal axes
and moments at any other point.
[See (5975) for the result.]
Let A, B, G be the given principal moments, and let the
mass of the body be unity. Then the ellipsoid of gyration at
the centroid 0, and a quadric confocal with it, will be
_._l_ ^ _L.fl- 1 and
^
l
yi
I
** - 1

5973 Prop. I.The M. I. is constant for all tangent planes

of the confocal, and is equal to the
M. I. for the origin 0

= S+X.

(5919)

Proof.Let I, m, n be the dir. cos. of the tangent plane of the confocal,

p the perpendicular on the plane from 0. The M. I. for this plane
= M. I. for a parallel plane through 0+p' (5921)
= A'F+B'rrt + C'it+p1 (5936)
= (S-A)P+(S-B)m* + (S-C)nt+(A + \)r+(B + \)mt+(C+k)nt
(5914, 5G31) = S+\, which is independent of I, to, n.
5974 Prop. II.All these planes are principal planes at
their points of contact, and if the three confocals be drawn
through any point P, the tangent planes at P to the confocal
ellipsoid, two-fold hyperboloid, and one-fold hyperboloid, are
respectively the principal planes of greatest, least, and mean
moments of inertia. The normal to the confocal ellipsoid is
the axis of least moment, and the normal to the two-fold
hyperboloid is the axis of greatest moment.
Proof.Draw any other plane through P. The perpendicular on it
from 0 is less than the perpendicular on the parallel tangent plaue to the
confocal ellipse, and greater in the case of the two-fold hyperbola. Then,
by (5921).
The solution of the problem at (5972) is now given by Proposition III.
5975 Prop. III.The principal moments of inertia at P are
OP2 Xx, OP2 X2, OP2 X3, and the normals to the three con
focals at P are the principal axes.
Proof.The M. I. about the z axis at P
= M. I. for the origin P M. I. for the yz plane
= S+OP,-(S+\1) = OP'-X, (5921-73).

MOMENTS OF INERTIA.

825

5976 The principal moments of inertia above, expressed in

terms of Xx of the confocal ellipsoid and dit ds, its principal
semi-diameters conjugate to OP, will, by (5661), become
OPa-A

Opt-K+dt,

OP^-^+dt.

5977 The condition that the line abc, Imn, referred to prin
cipal axes at the centroid, may itself be a principal axis at
some point of its length, is
a
b
b
c
c
a
I
m
m
n
n
I
1
A-B ~~ B-C ~ C-A ~ p'
Here abc is any point on the line, and if a confocal quadric
of the ellipsoid of gyration at the centroid be drawn through
the stated principal point of the given line, p is the perpen
dicular from the origin upon the tangent plane of the confocal
at that point.
Phoof.The given line - , a = ^- = -Imn
)? = 1
must be a normal to the confocal o? + ?/'
y + A-\-\ x> + X U + A
Therefore, by (5629), I =

m=

n=^

(i.)
(ii.).
(iii.).

Eliminate x, y, z from (i.) by means of (iii.), and from the resulting equa
tions eliminate p, and the condition above is obtained.
Also, by (5631),
p* = (A + X) V + (B + X) m' + (0+ X) ' = AP + Bm> + On* + X . . . (iv.) .
The principal point xyz is now found by eliminating X and p from equa
tions (iii.)> by means of (iv.) and (5977).

INTEGRALS FOR MOMENTS OF INERTIA.

By the definition (5903), the following indefinite integrals
for moments of inertia are obtained :
5978 For a plane curve, y =f (x), the M. I. about the x and
y coordinate axes are
^y'ds and ja^ds;

and therefore J (x2 +y*) ds = |rads

5 N

826

SOLID GEOMETRY.

is the M. I. about an axis perpendicular to both the former

through the origin (5920).
5980 Observe that ds may be changed into dx, dy, or d6
by the substitution formulae (5113, '16).
5981 For a plane area bounded by the coordinate axes, the
ordinate y and the curve y =f(x), the M. I. about the x and
y axes are
I j yidxdy = I y*dx

and

j \ afdxdy = \ ot?ydx.

5983 And the M. I. about an axis perpendicular to both

the former drawn through the origin,

= ff(rf+tf) dxdy = fft*drd$ = 4 (Vd0,

but in the last two integrals the area has the boundaries
described in (5894).
5986 The M. I. of a solid bounded by three rectangular
coordinate planes and the surface z =f(x, y) about the z axis,
will be
JjV-t-y) zdxdy = (TjV sin3 ddrddd^,
but in the last integral the solid is bounded as described in
(5875).
5988 The volume, which represents the mass in all these
cases, has already been expressed (5205, 5871) ; and by
dividing by the volume, the square of the radius of gyration
of the solid is found (5904).
Proofs. Formulae (5981-3) are directly obtainable by geometry from
figures 90 and 91, and formulae (5986-7) from figures 168 and 188. The
transition to polar coordinates may also be effected by the formula of
(2774).
5989 In expressing moments of inertia, the factor m will stand for the
mass of the body, and the remaining factor will therefore be the valne of the
square of the radius of gyration.

PERIMETERS, AREAS, VOLUMES, $-c.

827

PERIMETERS, AREAS, VOLUMES,

CENTRES OF MASS, AND MOMENTS OF INERTIA
OF VARIOUS FIGURES.

RECTANGULAR LAMINA AND RIGHT SOLID *

For a rectangle whose sides are a, b, the moments of inertia
about the sides, and an axis perpendicular to both where they
meet, are respectively
6015

m,

Proof :

m ,

m .

["ax1 db> = ^ = m^-.

Jo
3
3

The third by (5920) .

6018 Hence, for a right solid, whose dimensions are 2a,

2b, 2c,
M. I. about the axis of figure 2c = m.a
.

ARC OF A CIRCLE.
6019 Let AB (Fig. 191) be the arc of a circle whose centre
is 0 and radius r. Let the angle AOB = 9 ; then
Length of arc AB = rd.

(601)

6020 Huygens' Approximation.Rule.From 8 times the

chord of half the arc take the chord of the whole arc, and divide
the remainder by 3.
Proof.The rule gives

(l6 sin j 2 sin V

Expand the sines by (764) as far as 6s, and the result is rd.

6021 Taking an axis OX through the mid-point of the arc

with origin O, the centroid of the arc is given by (5889)
x=

2rsinA0
. * .
v

. . , 2r
Hence tor a semi-circle x .
ir

* For M. I. of a triangle, see (6944-52).

SOLID OEOMETBY.

828

6023

Also, for the centroid of BX, y = ^rs'D

a
where a = /_X0B.
6025

The M. I. of the arc AB about OX and OY are

-2" I1H

and

-r{1+-r}

(5978)

6027 -W J. about axes perpendicular to XOY, through 0 and

X the mid-point of the arc respectively, are
mr2

and

m. 2r* (l-^^j.

(5979)

6029 Cob. The M. I. of a) _ mr*

circular ring about a diameter ) ~ 2

SECTOE OF CIRCLE.
anon
6030

i
^0
Area ,

4r sin 40
oc =
.

^10B (Fig. 191)

_ nT) - 4rsin2|a
For X0J3, y
^ 8 .

Proof.x, y are respectively f of x, y in (6021, '3) ; since the centroid

of each elemental sector is distant -fr from 0. Otherwise, by (5893, '5).
6033

The M. I. about OX and OY are

Proof.By (5981-2) ; or integrate (6025-6) for r from 0 to r.

SEGMENT OF CIRCLE.
6035

*-*<--.#>.

6037

For CM,

ABX (Fig. 191)

i-^Sy.

y = -(2-3co8.+Co3')
o(a Bin a cos a)

Proofs.From the sector and triangle ; otherwise, the centroid, by

(5893, '5).

PERIMETERS, AREAS, VOLUMES, 8rc.

6038

M. I. about OX and OY,

|1 (30-4 sin 0+sin0 cos 0)

6040

829

(5981-2)
and

^ (0-sin 0 cos 0).

Cob.Hence, for a semi-circle, He = 4-

6041 Also, tbe Jlf. J. o/ a circZe about a diameter, and about

a central axis perpendicular to its plane, are respectively
and
4

^.
A

(5920)

THE BIGHT CONE.

If h be the height, r the radius of the base, and I the slant,
6043

Curved surface = vrl.

Volume = ^irr^h.

6045

Distance of centroid from vertex = fA.

6046

-3^ I- about axis of figure = m -j^r8.

6047
1- about cross axes through the vertex and centroid
respectively.
m ^ (r2+4!hi) and m& (4r2+A2).

FEUSTUM OF CYLINDER.
Let 0 be the inclination of the cutting plane to the base,
and c the length of the axis intercepted.
6048

The distance of the centroid from the axis is

a2 tan 0
x
.
4c

a2
The M. I. about the axis = m-^, being the same as
M
that of a cylinder of height c. Hence, by (5921) and the value
of x above, the M. I. about any line parallel to the axis can be
found.
6049

SOLID GEOMETRY.

830

SEGMENT OF SPHERICAL SURFACE.

(Fig. 191)

Let 0 be the origin of coordinates ; OA = r the radius ;

and OC = x the abscissa of AB the plane of section.
6050 The curved area of AB = 2irr (r a?) = the area of its
projection on the enveloping cylinder of the sphere.
Proof:

Area =

2*y dx = 2*r (ra;).

6051

For centroid of surface, x = | (r+r).

6052

The M. I. about the axes OX, OY are

^-(2r2-rx-xi)
o

and

(5878)

(4^++**).
o

HEMISPHERICAL SURFACE.
6054

Area = 2nr\

(6050-1)

6056

M. I. about OX or OY = m r*.

(6052-3)

SEGMENT OF SPHERE.
6057

Volume = ^{2r+x)(r-x)\

* = l%+f)\-

6059

M. I. about OX = ^ (r-a?)s (8^+^+^^).

6060

3f. I. about OY
= JL (r-^^lBr3-!- 17^+18^+9^).

Proof.As in (6146-7) ; or put o = 6 = c in the results.

HEMISPHERE.
6061

Volume = fur3,

x = fr.

(6064)

Proof.Vol. = surface (6054) X , by elemental pyramids having their

o
common vertex at the centre of the sphere. Otherwise, make x = 0 in
(6057).

PERIMETERS, AREAS, VOLUMES, $c.

6063

M. I. about OX or OY = m fr2.

831

(6059-GO)

SECTOR OF SPHERE.
6064

Volume = fwr2 (r-h),

x = (r+h).

Proof.Vol. = surface (6050) x

x = \ of x in (6051), since the

o
centroid of each elemental pyramid is distant' fths of r from the centre.
6066 For the M. I. add together the M. I. of the cone and
segment (6046, '59).

THE PARABOLA, y* = tax.

6067

Rad. of curv. p =

6069

Coordinates of centre of curvature

(Fig. of 1220)

= 2a (l + | )f.

&v+2a,

(4542)

(4545)

6071

ArcAP = s = y/(ax+a?) + a\og^'v+x//{a+ *\

6072

= a [cot 6 cosec 0+log cot

Peoof:

=jy(l +

(5197,4206)

Substitute v/, and integrate by (1931).

6073

Arc AL = a ^/2+a\og (1 + ^2).

Centroid of arc AP with above value of s.

6074

2x+a
-Q-log
a2 ,
2,r+a+2v/(,i?!+a.r)
r -v/(d72+ac')\ +
^
t v \

4s
o
a

6075

^ = |{v/(^+a)3-a2}.

6076

For centroid of arc AL, putting x = a,

. _ 6y/2 + log (3 + 2 v/2)
* 8{^2+log(l + y2)} a'

._ 4
(2 ^2-1) a
y
3 - v/2 + log (1+^/2)"

832

SOLID GEOMETRY.

Half-segment of parabola ANP.

6078

Area = %xy,

x = %x,

6081

The M. I. about the x and y axes are

to fax

and

jf = $y.

to fx2.

THE ELLIPSE.
6083 The equation being BV +aY = a!62, the length of the
arc AP (Fig. of 1205), putting </> for the eccentric angle
of P, is
* = ai ^/(l e* cos2 <) d<.
Jo
Proof. In (ds)1 = (daj),+ (rft/)! (5113), substitute e&s = a sin fify,
dy = b coBfdf, by (4276), and use (4260).
6084

The length of the elliptic quadrant AB is

tta ft
2 1

e*
4

3c*
2!2!2*

32.5e6
3I3I28

ff-S'^e8
4! 4! 2s

\
V

Proof.Expand the binomial surd above, and employ (2454) and (2472).
Similarly, from (5887) and (5978) the three following values are found.
6085

Tor the centroid of the same quadrant,

r= 1_|gl_^4

6086

approximately.

The M. I. about the x and y axes are approximately,

m62 l-jg-jfrg*
2
l-ie2-^4

^ 1-fg2-^
2 ' l-^-^

6088 Fagnani's Theorem.(Fig. 192.) Let P be any point

on the ellipse, GY the perpendicular on the tangent at P;
/.ACY=6; Q the point whose eccentric angle =^w 0.
Then
6089

PY+AP = ay/(l-Jain'0)dO = BQ;

and in the hyperbola (Fig. 193)

6090

PY-AP = ojV(l-e" sn20) dd-

PERIMETERS, AREAS, VOLUMES, %c.

833

6091 Cor. The difference between the lengths of the

infinite curve and asymptote = a [ ^/(l el sin20) dO, where
a
'Jo
tan a = .
b
Proofs.By (5203),
AP + PY or s + q = ^Pd6=a^J(\-e1B,mid)dB = BQ, by (6083).
In the hyperbola we have qs

6092 Draw the tangent at Q and the perpendicular CJJ upon

it. Let as, x be the abscissae of P, Q. The following relations
subsist,
PY=

= QU,

CY.CU=ab,

CF+CU* = a8+6*.

a
Proof.Let <p = the eccentric angle of P, and let ACT! = 0'.
tan f =
ox

= - tan 0.
a

Similarly for Q,

tan ^ ^ 0 j = cot 9 = tan 6',

therefore

tan^ = cot 0'

or

<p = - 0'

Then
(4276-80)

(i.).

The relation therefore between P and Q is reciprocal. Now PT = x sin 6"

(4295) and as' = a sin 6, therefore PY = = QU, by the reciprocity.
Again, 0 U* = a2 cos! 6' + 62 sin' 0' (4372) = a' sin2 ^ + 6s cos>
(ii.).
Put <]> in terms of 6 by the above, and we find
prjj
aV
_ a86'
~ a>cos> 0 + b1 sm' O
GY1'
Lastly, GP*+CUi = x2 + y* + a* Bin* <p + b* cos2 <p, by (ii), = a2 + 62 (4276-7).
6095 When P coincides with Q, the point is called ' ' Fagnani's
point," CY = y/(ab), PY = a-b, and x = a*
6096 Griffith* Theorem* If an ellipse of eccentricity e,
and a hyperbola of eccentricity e_1, be placed as in the figure
of 1205 (the circle representing the ellipse), P, p being con
sidered corresponding points ; then, calling PQ, in (6088), a
Fagnanian arc, we have the following theorem :
J. Griffiths, M.A., Proc. land. Math. Soc, Vol. v., p. 95.
5 0

834

SOLID GEOMETRY.

The ratio of the difference of two Fagnanian arcs on the

ellipse to the difference of the two corresponding arcs on the
hyperbola is equal to the product of ez and the four abscissae
of the points on the ellipse.

SECTOR AND SEGMENT OP ELLIPSE.

6097 The formulae for the sector and segment of a circle
may be adapted to the ellipse by writing a for r and multi
plying linear dimensions parallel to the minor axis by b : a.
But a will then represent the eccentric angle of the semi-arc,
and 0 twice that angle. Thus, in the figure of (1205), if
AGP be the half sector, a = ACp, 6 = 2AGp.
Sector of eUipse (2 AGP in fig. of 1205) :
anno
6098

a
aW
Area
,

4a sin kd
oe = ^ * ,

the last being for the half sector AGP.

x and y axes are
emi
6101

m62/i
-^1

sin 6\
j-)

,
and

46 sin2 4a /cr.rt 0.
y =
* , (6030-2)
The M. I. about the

ma% ( * . sin 6\
(6033)

Segment of ellipse (2ANP in same figure) :

6103

Area =^{6-sm6),
2

= 4a ain^
3(8 sin 0)

(6035-6)

6105 For y of the half segment ANP, and for the M. I.

about the x and y axes, replace r by b in (6037-8) and by a
in (6039).
6108

For the whole ellipse, the area = vab.

(6103)

6109

For the half ellipse, x = |.

Ait

(6104)

6110 The M. /. about the x and y axes, and a third central

axis perpendicular to both,
&
4

mtf
4

and

m (a>+y)<
4

(6041_2)

PERIMETERS, AREAS, VOLUMES, %c.

6113

835

The area of the ellipse whose equation is

{abefghXxyiy=0,

is

= j^-V)*

Proof.If a, (3 be the semi-axes of the conic, the area irafi takes this
value, by (4414) and (4407).
6114 Lambert's Theorem.The area of a focal sector of an
ellipse, as PSP' (Fig. 28), in terms of <j>, <f>', the eccentric
angles of P, P', is
_

(sin <-sinf)} = ^{x-X-(sinX-sinx')}-

In the second value, sin

and sin
are = i \ r"^r c
Z
2
v
a
respectively, where r = SP, r' = SP1, and c = PP'* a result
of use in Astronomy.

THE HYPERBOLA.
6115 The length of an arc of the hyperbola 6V aV = a2ft2
and the abscissa of its centroid may be approximated to, as in
(6084) for the arc of an ellipse, by the substitutions from
(4278),
| ds = a J sec <f> </(e2 sec2 <f> 1) d<f>
and

[vds = a1 fsec2 <f> y/(e* sec2 $1) d<f>.

6117 Landen's Theorem. This theorem gives any arc of an

hyperbola in terms of the arcs of two ellipses, as follows :
j x/(a2+bi+2ab cos C) dC =
Jv/(2-62sin2^)rf^ + (V^2-"2 s\n*B)dB+2asmB+ const.,
that isArc of ellipse whose semi-axes are a+ft and a b
= Arc of ellipse whose major axis is 2a and eccentricity b : a
+ difference between a right line and, the arc of an hyperbola
whose major axis is b and eccentricity a : b.f
Williamson's Integ. Cafe., Art. 137.

t Hid., Art. 167.

SOLID GEOMETRY.

836

6118
curve

Area ANP (Fig. of 1183) bounded by x, y, and the

=A

6119
6120

vV-a') -a* log *+ ^(^t-gl) | .

= | [^-a61og(^+|)|.

(1931)

(4271)

-4r<?a o/ sector between GA, CP and the curve

6121 4rea between two ordinate* yu y2, when the asymptotes

are the coordinate axes
2
Proof :

sin 2ACO \ydx =

e
j ^ ^.

(4387)

6122 The centroid of ANP, A being the area (6118), is

given by

6124

The M. I. of ANP about the * and y axes are

^L(W-5a>*) v/(^-2) + ^log^^^!).

6125

A(2^-a^) (-) - ^ log^b^=^.

THE ELLIPTIC PARABOLOID.

6126
6127
6129

Equation,

+
= 2s.
a
o
Fbi. o/ segment = ir y/(ab) zi,

z = .

iW- about the axes of x, y, and z respectively,

laz , z2\
ibz , zi\
a +6

PERIMETERS, AREAS, VOLUMES, Sfc.

6132

837

The surface 8 of the same segment may be found from

s=4i

J.

W+f+W**

(5874)

6133 If the surface of the paraboloid be bounded by a

curve of constant gradient y (5881), the area becomes
8 = firab (sec3y-l).

(5883)

THE PARABOLOID OF REVOLUTION.

6134

Equation, <r2+#2 = 2az or r! = 2az.

6136

Surface of segment, S = fn- y/a { (2z+a)% a'} . (5880)

6137

Volume = ttaz% = \,m^z,

6140

M.I. about axis of figure =

z = z.

(5887, '99)
(6131)

o
6141

For M. I. about OX and OF put a = b in (6129-30).

THE ELLIPSOID.
6142

Equation, ^ + +^ = 1, semi-axes a, 6, c (5600).

6143 The surface of the segment cut off by the plane whose
abscissa is x, will be found from
8 = 4 f f*ViW^MW
J*J0
\
<**&* a*6Va46y
/
*
Proof.By (5874) and (5629, '7), eliminating by means of the equa
tion of the surface.
6144 The volume of the solid segment and the centroid are
given by
r=i? <*.+*).-*). *=i$
Proofs.Let (Fig. 177) represent one octant of the ellipsoid; OA, OB,
00 being the principal semi-axes. The elemental section
4PArQ = vNP.NQdx = v /a2^? VcC'-a?dx.
a
a
Therefore Vol. = ^ ["(a8-*') dx = ^ (2as-Sci1x+x*) = <fcc.
a J,
3a2

SOLID GEOMETRY.

838

The moment with respect to the plane of yz

and division by the volume gives x as above.

6146

The M. I. of the solid segment about the axis a

Proof.(Fig. 177.)
M. I. = ["rNP.NQ^ + W * (6112) =
6147

[>-*)' (lx = Ac.

The M. I. about the axis b

= ^ j- (a-tr)s(8a2+9a^+ar2)+2a8-5V+at-5|.

Proof :

AT. /. = ^NP.NQ
= ^ ["(a'-*')1
4a J,

6148

+ ON1} dx
^ ("(a'-*1)
a Jx

(5921)
= Ac.

The volume of the whole ellipsoid = irabc.

Proof.By making x = 0 in (6144).

Otherwise : Let ) be the point on the auxiliary sphere of radius r cor
responding to xyz on the ellipsoid. By (5638-9), rz = a, ry =
rz = cC
Therefore

6149

^dzdydz = ^^didvdi = ^$*r*.

(6061)

For the centroid of the semi-ellipsoid x =

(6145)
o

6150

The M. I. about the axis a = m 1*I!).

5

(6146)

6151 The volume of a segment cut off by any plane PNQ

(Fig. 177), where OA = rZ is the semi-conjugate diameter, and
AN=h, is
lr(3d-h)

Proof.Taking the area of the section from (5655), the volume of the
segment will be
irabc sin
j
l
-o6 = ^
P
0 1 /I i1 *
-rr \I rfjr,
where
sin
p
Jx\
rf* /
J
0 being the inclination of d to the cutting plane. Integrate, and put
x = d-h.

PERIMETERS, AREAS, VOLUMES, 8fc.

839

PROLATE SPHEROID.
Put c = b in equation (6142) of the ellipsoid ; then a will
be the semi-axis of revolution.
6152 The surface of the zone between the plane of yz aud a
parallel plane at a distance x is
S = nb \ sin"1 + y/(a2-M ] .
t e
a ' a
)
Proof.By (5878).
Then by (1933).
6153

S = ^ \'y/{? ~ *') dx-

Otherwise, make b = c in (6143), and reduce.

Cor.The lohole surface = 2irb (b+

sin"1^.

6154 The centroid of the surface of the zone in (6152) is

given by
-_2vbe(a3
/a*
All

Proof.From
6155

Sx =

J" yj

') dx.

The ilf. J. of the same zone is

6156 And for the whole surface, by making x = a and

doubling,
/2
1 \
/
1 \

Pkoof; Jf. J. = 2f

^) J. = ^

>/(')<fe

-!?W(*-*)*-!5sWG-*)fc
The first integral by (1933). For the second, by Rule VI. 2048, we obtain
the formula
6157
f *a y(3-*'-') dx = sin"1 1 +
SW-z*),
in which must now be written for a.
e
6158 For the volume, moment of inertia, and abscissa of
centroid of the solid prolate spheroid, make c = b in (6144-51),
a being the axis of revolution.

SOLID GEOMETRY.

840

OBLATE SPHEROID.
6159 Put b = a in the equation (6142) of the ellipsoid ;
then c will be the semi-axis of revolution.
The surface of the zone between the plane of xy and a
parallel plane at a distance z, is
S = * v/(c + aVar) H

Proof.By (5878). 8 =
6160

log

v v .

jy (j? +*') *

i.

Then bv (1931)-

Cor.The whole surface = 2ira8+ ?^!logi

e
1 e

6161 The centroid of the surface of the zone in (6159) is

given by

Proof.As in (6154). z for the surface of half the spheroid is obtained

in this case by making z = c, but in (6154) put x = a.
6162

The M. I. of the same zone is

, ffc2(42-3ca) 1
es-r-v/(ci+aW)
+
4e3
g
c*
6163 And for the whole surface, by making z = c and
doubling,

p,o,- , . I. = &f ^

*=

f (^-O

The first integral involved is given at (1931), and the second is obtained in
the same way as in the Proof of (6155), giving
S7+Zdx =
v/Or'+o.')- log {x+ /(*+<**)}.
6164
o
o
6165 For the volume, moment of inertia, and abscissa of
centroid of the solid oblate spheroid, make b = a in (6144-51),
c being the axis of revolution.

JOINT

INDEX
TO THE

SYNOPSIS
AND TO TIIE
PAPERS

ON

PURE

MATHEMATICS,

CONTAINED IN TIIE UNDERMENTIONED

BRITISH AND FOREIGN JOURNALS
AND
TRANSACTIONS OF SOCIETIES.

"There is an immense amount of knowledge lying scattered at the present day,

and almost useless from the difficulty of finding it when wanted."
#
Profeator J. D. Everett.

KEY

TO

THE

INDEX.

Prefixed to each title will be found the symbol by which the work is
referred to in the Index. The words "with Vol." or "with Year," signify
that any number following the symbol in the Index denotes, respectively,
the Volume or Year of the journal. The year is given in all cases in which
the work consists of more than one series of volumes. In order to connect
the volumes with the years of publication, a Chronological Table is prefixed
to the Index ; in which table successive series of numbers in any column
indicate successive series of volumes of the publication.

A. with Vol.Archiv der Mathematik; 1843 to 1884; 70 vols.

[B. M. C. : P.P. 1580.] *
Ac. with Vol.Acta Mathematica, Zeitscbrift Journal, herausgegeben
von G. Mittag-Leffler; Stockholm, 1882 to 1885; 7 vols.
AJ. with Vol.American Journal of Mathematics ; Baltimore. Editor:
J. J. Sylvester, F.R.S. ; 1878 to 1885 ; 7 vols. [B. M. C. :
P.P. 1575. &.]
An. with Tear. Annali di Scienze Matematiche e Fisiche, compilati da
Prof. Barnaba Tortolini ; Rome, 1850-57 ; afterwardsAnnali
di Matematiche pura et applica; Rome, 185865. Series II.,
Annali di Matematiche para et applica, compilati da F. Brioschi
e L. Cremona; Milan, 1868-85 ; 23 vols, in all. [B. M. C. :
P.P. 1573 and 952.]
At. with Year. Atti della Reale Accademia delle Scienze di Napoli;
1819 to 1878 ; 15 vols.
[B. M. C. : for 1819-55, 8 vols.,
Acad. 2813 ; for 1863 to 1878, 7 vols., Acad. 96.]
C. with Vol. Comptes rendus liebdomadaires des seances de l'Academie
des Sciences ; Paris, 1835 to 1885 ; 100 vols. [B. M. C. :
Acad. 424 and B.B. 2099. c] f
CD. with Vol. Cambridge and Dublin Mathematical Journal. Editor,
W. Thomson, B.A. ; 1846 to 1854; 9 vols.
[B. M. C. :
P.P. 1565.]
CM. with Vol. Cambridge Mathematical Journal ; 1839 to 1845 ;
4 vols. [B. M. C. : P.P. 1565.]
CP. with Vol. Cambridgo Philosophical Transactions; 1822 to 1881 ;
13 vols. [B. M. C. : Acad. 3008.]
* i.e., British Museum Catalogue, Press mark P.P. 1580.
t .. signifies Heading-Room volumes within reach.

844

KEY TO THE INDEX.

E. with Vol.Educational Times Reprint of Mathematical Questions

and Solutions, with additional papers; London, half-yearly,
1863 to 1885; 44 vols. Editor: W. J. C. Miller, B.A.
[B. M. C. : 2242. c]
G. with Vol.Giornale di Matematichead uso degli student! delle universit Italiane, pubblicato per cura del professore G. Battaglini ;
Naples, 1863-85 ; 23 vols. [B. M. C. : P.P. 1572.]
I. with Vol. Journal of the Institute of Actuaries, or, The Assurance
Magazine; London, 1850-84; 24 vols. [B. M. C. : P.P.
U23.g.g. and 126.]
J. with Vol. Jourual fr die reine und angewandte Mathematik,
herausgegeben von A. L. Crelle ; 1826-1856 ; and Journal
als Fortsetzung des von A. L. Crelle gregrundeten Journals von
C. W. Borchardt; Berlin, 1856-1884 - 97 vols. [B. M. C. :
P.P. 1585 and R.B. 2022.?.]
JP. with Vol. Journal de l'Ecole Polytechnique ; Paris, 1 796 to
1884 ; 34 vols. [B. M. C. : T.G. 1.6.]
L. with Year.Journal de Mathematiques pures et appliquees, ou
Becueil mensuel de memoires sur les diverses parties des
Mathematiques, publie par Joseph Liouville; Paris, 1836 to
1884 ; 49 vols. [B. M. C. : P.P. 1575 and B.B. 2022. g.~]
LM. with Vol.London Mathematical Society's Proceedings; 1866 to
1885 ; 16 vols. [B. M. C : cad. 4265, 2.]
M. with Vol.Mathematische Annalen, in Verbindung mit C. Neumann
begrndet durch R. F. A. Clebsch unter Mitwirkung der
Herren Prof. P. Gordan, Prof. C. Neumann, Vols. 1-9 ; and
Prof. K. V. Mhl, gegenwrtig herausgegeben von Prof. F.
Klein und Prof. A. Mayer, Vols. 10, &c. Leipsig, 1869-1885 ;
25 vols. [B. M. C. : P.P. 1581.5.]
Man. with Tear.Manchester Memoirs, or, Memoirs of the Literary
and Philosophical Society of Manchester; 1805 to 1884;
23 vols. [B. M. C. : 255.d., 9-12, and Acad. 1360.]
Me. with Year. The Oxford, Cambridge, and Dublin Messenger of
Mathematics; 1862 to 1871; 5 vols. Continued asThe
Messenger of Mathematics. Editors : W. A. Whitworth,
M.A., C. Taylor, D.D., R. Pendlebury, M.A., J. W. L.
Glaisher, F.R.S. ; Cambridge, 1872 to 1885; 14 vols.
[B. M. C. : P.P. 1565.6. and 463.]
Mel. xoith Vol.Melanges mathematiques de l'Academie des Sciences
dc Saint Petersburg; 1849 to 1883; 6 vols. [B. M. C. :
Acad. 1125/9.]
Mem. with Year.Memoires de l'Academie Imperiale des Sciences de

KEY TO THE INDEX.

845

Saint Petersburg ; 1809 to 1883 ; 51 vols.

[B. M. C. :
1809-30, T.O. 9.a., 14-20; 1831-59, Acad. 1125/2; 1859-83,
Acad. 1125/3.]
Mo. with Year.Monatsbericht der Koniglich-Preussichen Akademie
der Wissenscbaften zu Berlin; 1856 to 1881 ; continued as
Sitzungsberichte der Koniglich, &c. ; 1881 to 1885; 30 vols.
[B. M. C. : Acad. 855.]
N. with Year.Nouvelles Annales de Mathematiques. Journal des
Candidats aux Ecoles Polytechnique et Normal. Redige par
MM. Terquem et Gerono ; Paris, 1842 to 1882; 41 vols.
[B. M. C. : P.P. 1544.]
No. with Year.Nova Acta Regiae Societatis Scientiarum Upsalieusis ;
1775 to 1884; 26 vols. [B. M. C. : for 1775-1850, 14 vols.,
T.O. 6.6'., 13-19; for 1851-1884, 12 vols., Acad. 1076.]
P. with Year.The Philosophical Transactions of the Royal Society of
London; 1781 to 1884; 104 vols: i.e., vols. 71 to 174.
[B. M. C. : R.R. 202 1.?.]
Pr. with Vol.Proceedings of the Royal Society of London; 1800 to
1885; 39 vols. [B. M. C. : for vols. 1 to 23, Acad, 3025, 21 ;
for vols. 24, &c, R.R. 2101. d.]
Q. with Vol,Quarterly Journal of Mathematics ; 20 vols. Cambridge,
1857-78; Editors: J.J. Sylvester, F.R.S., N. M. Ferrers,
F.R.S., G. G. Stokes, F.R.S., A. Caylcy, F.R.S., M. Hermite,
F.R.S. ; 1878-85, N. M. Ferrers, F.R.S., A. Cayley, F.R.S.,
J. W. L. Glaisher, F.R.S. [B. M. C. : P.P. 1566 and 25L]
TA. with Vol.Transactions of the American Philosophical Society ;
Philadelphia, 1818-71; 14 vols. [B. M. C. : Acad. 1830/3
and T.O. 18.b. 11.]
TB. with Vol. Transactions of the Royal Society of Edinburgh ; 1788
to 1880; 29 vols. [B. M. C. : T.C. 15.6. 1 and R.R. 2099.;/.]
TI. with Vol. Transactions of the Royal Irish Academy; 1786 to
1879; 26 vols. [B. M. C. : Acad. 1540 and B R. 2099.?.]
TN. with Vol.Transactions of the Royal Society of New South Wales ;
Sydney, 1867-83; 17 vols. [B. M. C. : Acad, 1971.]
Z. with Vol.Zeitschrift fur Mathematik und Physik, herausgegeben
von Dr. O. Schlomilch und Dr. B. Witzschel ; Dr. O. Schlomilch, Dr. E. Kahl, und Dr. M. Cantor; Leipzig, 1856-1883.
28 vols. [B. M. C. : P.P. 1581.]

Year.
1S01
3
3
4
5
r.
7
s
J
10
11
12
13
14
15
Hi
17
IS
19
2n
21
23
23
21
25
J.'.
27
2S
2y
30
31
32
33
34
35
30
37
3S
3!l
4U
41
42
43
44
45
40
47
4S
4'.i
51)
51
52
53
54
55
50
57
5S
59
00
01
02
03
04
65
00
07
OS
09
70
71
72
73
74
76
70
77
7S
7:>
-ii
81
S2
.S3
S4
1SS5

Ac AJ

An

At

CD

CM

CP

...
...

...

...

...
...

...

...
.
...

...

...
...
...

...
!!!

...
...

...

...
i7i
...

...
...
1

...
...

...

8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
33
34
...

1
2
3
4
5, 6
7
8,9
10
11, 12
13, 14
IS
16, 17
18
19
20, 21
33
23, 24
25, 26
27, 23
29, 30
31, 33
34, 35
36, 37
38
39, 40
41, 42
43, 45
46, 47
48
49. 50
51, 52
53, 54
55
56, 57
58
59, 60
61
62
63
64
65, 66
67
68, 69
70
71, 72
73
74, 75
76
77,78
79, 80
81
83-84
85. 86
87, 88
89, 90
91, 93
93, 94
95
96,97
...

3
...
...
...

4
i
s, 3
4, 5
6, 7
8, 9
10, 11
12, 13
14, 15
16, 17
18, 19
20, 21
22. 23
24. 25
26, 27
28, 29
30, 31
32, 33
34, 35
36, 37
38, 39
40
41, 42
43, 44
45, 46
47
48
49, 50
SI
52
S3
54, 55
56, 57
58
59, 60
61
62
63,64
65, 66
67
68
69
70
A

...

i"i
2
3
4
5
6
7
8
II. 1
2
3
4
5
6
7

6
II. 1
a

iiii'i
a
3
4

mil
2
3
4
S
6
7
8
; ...
1 ...
1,2
3.4
5-7

1
2
3
4
5
...
6
7

'-Ac 1*

5
6
"'
...
...

Ad

At

...
1
2,3
4, 5
6, 7
8, 9
10
11-13
14, 15
16, 17
18, 19
20, 21
22, 23
24, 25
26, 27
28, 29
30, 31
32, 33
34. 35
36, 37
38, 39
40, 41
42, 43
44, 45
46, 47
48, 49
50, 51
52, 53
54, 55
56, 57
58, 59
60, 61
62, 63
64, 65
66, 67
68, 69
70, 71
72, 73
74, 75
76, 77
78, 79
80, 81
82, 83
84, 85
86, 87
88, 89
90, 91
92, 93
94, 95
96, 97
98, 99
100
C

4
5
...
6
1
2
3
4
1
2
3
4
5
6
7
8
9
...
...

...
9
...
...
9
...

...
...
...
...

...
...
...

...
...

10

11

...

...
...
...

...
...
...

13
13
...
...

CD

CM

CP 1

1
2
i, 2
3
3,4
4
5,6
5
7,8
6
9, 10
7
11, 13
8
13, 14
9
15, 16
17, 1 10
19, 20
21, 22 ia
23, 24 13
25, 26 14
27, 28 15
29, 30 16
31, 32 17
33, 34 18
35, 36 19
37, 38 I 24)
39, 40 21
41, 42 23
43, 44 23
K

1
2
3
4
5
6
7

JP Yew.
3 ISO!
1
3
4 4
S
5,6 6
7 7
8
8 9
In
11
12
9 IS
14
10 15
16
...
17
18
19
...
11 20
22
13 23
24
35
26
...
27
...
38
29
30
13 3.
33
...
14 33
34
15 35
3r
3;
16 3:>;
4>>
17 41
42
43
44
45
46
18 47
19 43
49
...
50
51
-.
30 53
54
...
55
...
21 50
r:::
o:<
...
00
...
33 Pi
f'2
as 03
04
24 65
66
35 67
68
69
z
36 70
71
72
73
37 74
75
76
77
78
38 79
29 SO
30, 31 81
33 S2
33 83
34 84
... i 1885
t

JP

LM

Me

Man. Mem. Mel. Mo

i.\

No

i.' 1
2
3 12
4
5
6 13
7
8
9 14
10
11
12
13
14 I.'i
10
16
2
17
18
19
20
3
II. 1
2
4
3
4
5
5
6
6
7
8
9
7
10
11
12
8
13
14
9
15
16
17
18 in
19
20 11
III. 1
12
No

...

...
2

i. 1
2
3
4
5

...
I. 7
...

"6
"7
8
9
10

9
5

1T1
2
3
4
5
6
7
8
9
10
11
12
13
14
IS
16
17
18
19
20
II. 1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
III. 1
2
3
4
5
6
7
8
9
10
I,

11
II. 1
2

10

3
11
4
6
5
7
8
"6
9
10
11
12
13
14
15

i. i i'i
2
2
3
3
4

a
3
4
5
6
7
8
9
10
11
12
13
14
15
16

1
2
3. 4
5
5 II. 1
6
2
7
3
8
4
9, 10
5
11, 12
6
13
7
14, 15
8
16, 17
9
10
16
19
11
20,21
12
22,23
13
24, 25
14

LM

M 1 Me

5
6
"7
8

"' 7
8
9
III. 1
2
3
4
5, 6
7
8
9
10
11
12, 13
14, 15
16
17, 18
19, 20
21
22
23, 24
25
26
27
28, 29
30
31

"2

5
6

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
2(1
27
28
29
30

Man. Mem. Mel. Mo

P
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
V2<J
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174

...

...

...

Pr

Q TA TE TI TN Z Year.
8
9
5
10
11
6

1
"7 12
1

8 13

9
u

2
10
2

15
16

3
4

5
6

a
18

7
8

19
20

9
21
5
22

10
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25, 26
27
28, 29
30
31, 32
33
34, 35
36, 37
38, 39
Pr

23
1
2
3 11
4
5
12
6

23

24
13 25
10 14
26
8
9

13
14
15
16
17
18
19
20

1
2
3
4
5
6
7
24 8
25 9
27
10
11
12
28 26 13
29
14
15
16
17

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

a TA TE TI TN Z

1801
2
3
4
5
6
7
8
9
10
11
IS
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
1885

EXPLANATION OF ABBREVIATIONS, &c.

i. eq. - indeterminate equation.

ureal coordinates.
a. c.
algebraic.
imag. = imaginary.
alg.
= application.
1. e. m. = lotC^St CQttttfl&tt tttttlttpltf m
ap.
num. = numerical.
anal. = analytical.
o. c. = oblique coordinates.
ar. p. = arith. progression.
= Cartesian coordinates.
p. c. = polar coordinates.
c. c.
p. d. e. = partial difference equations.
construction.
en.
cond.
condition.
p. e. - polar equation.
perpendicular.
perp.
curv. = curvature.
= plane.
pL
d. o. = differential calculus.
--- differential equations.
problem.
pr.
d. e.
= definite integral.
d. i.
q. c. = quadriplanar coordinates.
= radius.
rad.
equation.
eq.
= example or exercise.
= solid or 3-dimensional.
sd.
ex.
= extension.
= solution.
sol.
ext.
= formula.
sym. = symmetrically.
f.
table.
f. d. c. = finite difference calculus.
to.
= trilineor coordinates.
t. c.
f. d. e. = finite difference equation.

= geometrical.
tangential coordinates.
tg. c.
geo.
geometrical progression.
tg. e. - tangential equation.
g- P=
=
theorem.
th.
general.
gn.
- treatise (i.e., more than 50 pages).
generalization.
tr.
gz
=
highest common factor.
transf.
transformation.
h.c.f.
integral calculus.
i. c.
15a.The suffix meann that three articles under the same heading will be found in the volume.
1 82 1 .Means that one article on the subject will befound in each of the four consecutive volumes.

References to the Synopsis stand first, and are the numbers of Articles,
not of Pages. An asterisk (*) is prefixed where such numbers will be found.
The unclassified references following a principal title commonly refer to
papers on the general theory of the subject ; but some papers are occa
sionally included amongst these of which the titles are too long for inser
tion, and do not admit of abbreviation.
Subjects which might well have been included under the same head
ing appear sometimes under different ones, for the following reasons :
Exigencies of space have decided the insertion of the number of the volume
only of the particular work in question, and a subsequent examination of
the Index of that volnme is required in order to find the page. It, there
fore, became desirable not to change the original title of the paper when
there was danger, by so doing, of making it unrecognizable. When, how
ever, the same subject appears in two parts of this Index under different
names, cross references from one to the other are given. Some changes,
however, have been made when the synonym was perfectly obvious ; for
instance, when a reference to a journal, published fifty years ago, is found
under the heading of " Binary Quantics," the actual title of the article will,
in all probability, be " Homogeneous Functions of Two Variables," and so
in a few other instances.

INDEX.

Abacus of the Pythagoreans : L.39.

Abelian cubics and symmetrical equa
tions : Q.5.
of class
31) Mo.82.
Abelian equations: A.68: C.95: J.93 :
M.18: Mo.77,92.
Abelian functions : see " Hyper-elliptic
functions."
Abel's formula for F(x+iy)+F{xiy) :
2705. Me.73.
Abel's theorems: 1572: C.94: J.9,24,
61,90: LM.12: M.8,17: P.81,
83: Pr.30,34.
<#> (z) +<t> (y) = f W(y)+yf(x) } = An.
57.
Abscissa: 1160.
Acceleration: Me.tr 65.
Algebra: 1-380: A.tr20.
application to geometry : JP.4.
foundation, limitations : AJ.6 : CP.7,
8,: Q.6.
history of, in Germany : Mo.67,70.
Algebraic :
Calculus : N.81.
definitions : C.37.
forms: C.84,94: M.15.
coordination of: J.76.
whoseHessianvanishes identically :
M.10.
in theory of cubics : M.8.
formulas: G.12: Q.5.
functions: A.10,31: J.92 : L.50,51 :
M.10: N.62.
applied to geometry : G.22 : M.7.
number of constants : J.64.
as partial fractions : Z.9.
rationalization of : A.69.
representation by : J.77,78.
resolution into factors : A.46.
theorems : J.82 : M.1,6.
synthesis : C.163.
Algorithmic geometry : N.57.
Algorithm :
re definition of ( ) :
J.27.
V6>
of higher analysis : Mo.75.
of arithmetical functions : G.23.

Alternants of 4th order, co-factors of :

AJ.7.
Alternate numbers : LM.lOj.
Alternating functions : AJ.7 : C.12,
22 : J.83 (Vandermond's) : Me.
82.
Altitudes, determination of: A.12,19:
Mem.15: N.45.
Amicable numbers : A.70.
Anallagmatic curves and surfaces :
0.87 : N.64 (quartic surface).
Anallagtnatic pavcmonts : E.10.
Analysis : A.l : An.50 : 0.3,11,12 :
J.f7: P.14: Q.6,7.
ap to geometry : G.23 : JP.4.
Analytical:
aphorisms: A.5 : J.10.
combination theorem : J.ll.
functions : Ac.6 : thsAn.82 : La
grange, trJP.3.
system of, and series from it: An.
tr 84,.
geometry : 40016165 : A.2,11,38 :
C.6: JP.9: L.72: M.2 : Mem.
13 : Z.9: tr 11,12.
theorems and problems : A.8,52: J.46.
plaue and solid in homogeneous co
ordinates : Z. 15,16.
of three dimensions : CM.4.
metrics : Q.7,8.
theorems: A.8. treatise : C.13.
Angles :
conterminal : Me.74.
of a central conic : 4375.
division into n and n-f 1 parts : A.70.
of five circles or six spheres : Me.79.
of two circles : 4180.
problems on : P.1791.
two relations botween five : A.20.
trisection of: 5325: A.4,34 : C.2,66,
81 : G.15 : Mc.72 : N.56,76.
Anharmonics : LM.2,3.
Anharmonic pencils of oonics : 4809
21.
Anharmonic ratio: 1052,4648: GM.12.
corresponding to roots of a biquad
ratic : N.60.
of a conic: Q.4: of four tangents: 4986.
5 <*

850

INDEX.

Anharmonic ratio {continued) :

Arithmometer : I. 1618.
of 5 lines in space : Me.76.
Aronhold, theorems of : gzJ.73.
of 4 points in a plane : A.l : C.77.
Associated forms :
systems of :
sextic : LM.2,60 : Q.37,38.
gzAJ.l : C.86 : CD.6.
systems : cnM.10.
and spherical harmonics : Me.85.
Annuities: 302: A.2o: Ac.l : CP.3:
Astroid of a conic : A.64.
J.83: 1.124: P.1788, -89, -91, Astronomical distances : p 5.
Asymptotes : 5167 : A.p. c 15,17 : CM.
-94, 1800, -10 : N.47.
Anticaustic (by refraction) ofa parabola:
4 : M.ll : N.68 : thsN.48,. and
N.83,85.
73.
Anticlastic surface : 5623,5818.
* of conies: 1 182, 4490, t.c 4683 : tg. c
Aplanatic lines, lemniscates, caustics,
4904, -66 : Me.71 : Q.3,8.
Ac: thL.50: N.45.
of intersections of quadrics : N.73.
of imaginary branches of curves :
Apolarity of rational curves : M.21.
Apollonius's problem : A.37 : M.6.
CM.2.
Approximation : J.13 : N.66.
Asymptotic :
chords : A. 12.
algebraic : J.76.
* cone of a quadric : 5616 : E.30, g.e 34.
to functions represented by integrals : curves : 51 72.
C202.
lines of surfaces : A.60,61 : R.84.
of several variables : C.70.
law of some functions : Mo.65.
successive : Mm.38.
methods : M.8.
planes of a paraboloid : 5625.
Apsidal surfaces : Q.16.
Arabs, mathematics of: C.39,60.
planes and surfaces : CD.3.
Arbitrary constants : C.15 : L.80 : in
Atomic theory and graphical represen
d.e and f.d.e, TI.13.
tation of invariants and covari
Arc, area, &c. :
quantification of :
ants of biuary quantic3 : AJ.lj.
1244,5205, 5874,6015.
Attraction :
of confocal ellipsoids :
relations of: G.16: C.80,94: L.46.
Me.82.
Arcs with a rectifiable difference and
of ellipsoids: CD.4,9 : J. 12,20,26,31 :
areas with a quadrable difference :
JP.15: L.40,45: M.10: N.76:
L.46.
Q.2,7,17.
Areas :
and volumes in t.c and q.c :
of ellipsoidal shell : J.12 : JP.15 : Q.17.
of paraboloids : L.57.
4688 : Q.2.
* approximation by ordinates : 2991 7:
of polyhedra : J.66.
of a right line and of an elliptic arc :
C.78: CD.9.
between 3 lines : 4038 : Me. 75.
An.59 : CD.3.
ext. of meaning : CD.5.
of a ring and of elliptic and circular
Arithmetic : A.5,18 : L.59.
plates : G.21 : Z.ll.
ancient : C.71 : N.51.
of spheroids : J.12 : JP.8 : L.76 :
degenre, ext. of the notion : C.94.
Mem.31 : P.(Ivory) 12.
higher : J.35.
of solids of revolution, <fcc. : An.56 :
history of : C.17.
CD.2.
of Ibn-Esra: trL.41.
solid of maximum : TE.6.
of Nicomaque de Gerase : An.57.
theorems : Q.4,17.
Arithmetico-geometric mean : Mo.58 :
theory of : L.44,6.
Auxiliary circle : 1160.
Z.20.
Arithmetic mean : 91 : CD.6 : of n
Averages : 1.7,9.
Axes :
of a conic : gn.eq4687 : A.30 :
quantities : 332 : L.39.
Arithmetical progression : 79 : thL.
E.36: G.12: Q.q.c4; t.c 5,8,15
39: Pr.10.
and 20 : Me.ax 64,71 : N.43,48,58:
and g. p when n (the number of terms)
t.cQ.12.

construction of : 1252 : Me.66.,.

is a fraction : A.35.
when the terms are only known ap *
cn. from conj. diameters: 1253:
proximately : C.96.
A.13,20 : Me.82 : N.67,78.
Arithmetical theorems : A.10: C.93,97.: of a quadric : 5695 : A.30 : Au.77 :
G.9: J.2,64,82: N.43,51, cn 68,
CD.6: G.7,18: L.63: of I.e. m'.,
N.57 : Gergonne, C.5.
69,74.
Arithmetical theory of algebraic forms : * rectangular, nine direction cosines
J.92,93.
for two systems : 55778 :
Arithmographe polychrome : Col,53.
L.44.

INDEX.
Axis :

of perspective or homology :
975.
* of reflexion: 1007.
* of similitude : 1046, 4177.
Axonometry and projective collineation
in space : M.25: Z.12,21.
Babbage's calculating machine : C.992.
Barycentric calculus and right line con
struction : J.28.
Battement de Monge : L.82.
Beltrami's theorem : A.44.
Bernoulli's numbers : 1539 : A.3 :
AJ.5,7: An.59,77: C.54>,583)81- :
G.9: J.20,21 ,28,58,81,84,85 (first
62),88,92 : LM.4,,7,9 : Me.75 :
gzMem.83: N.76: Q.6,22.
application to series : see " Series."
and interpolation : C.86.
and their first 250 logarithms : CP.12.
and secant series: A.1,3,35: C.4,32.
bibliography of : AJ.5.
indeterminate representation of : Z.
19.
new theory of : C.83.
theorems on: E.2,8 : N.77.
Bernoulli's series : 1510.
Bessel's functions (see also " Integrals
of circular functions ") : J.75 :
M.3,4,9,14,16 : Q.20,21.
representation of arbitrary functions
by : M.6.
squares and products : M.2.
tables : Z.2.
Bonnet, two formulas of : G.4.
Bicircular quartics : LM.3,99 : P.77 :
Pr.25: Q.192: TI.24.
focal conies of : LM.ll.
with collinear triple and double foci :
LM.12,14.
nodal, mechanical cn. of : LM.3.
Bifocal variable system : M.16.
Bilinear forms: J.68.,84,86 : L.74:
Mo.6(j,68-.,74.
congruent transformation of : Mo.
74.
four variables : G.21 : Mo.83.
relation between two and their
quadric and quartic system: M.l.
reciprocals : G.22.
reduction of : C. 78,92.
Bilinear functions : GM.ll.
polynomials : C.77.
trilinear and quadrilinear systems :
E.5.
Billiards, theory of : L.83.
Bimodular congruences : G.21.
Binary and ternary quadratics: N.64,65.
Binary cubics : 1631 : A.17 : C.92 : G.17 :
J.27,53,41 : Q.1,11.

851

Binary cubics{continued) :
automorphic transf. of : LM.14.
and quadratic forms : G.21.
system of two : E.7 : G.17 : LM.13 :
M.7.
resultant : Q.6.
tables and classification of : A.31.
transformation by linear substitution :
J.38.
Binary forms: An.56,77 : At.65: G.
2,3,10,163: J.74: M.2,3,20: Q.14.
and their covariants, geo. : M.23.
ap. to anal, geometry : L.75.
ap. to elliptic functions : AJ.5.
ap. to Euler's integrals : 0.47.
canonical : J.54 : M.21 .
evectant : Q.ll.
geo. interpretation or ap. : C.78 : G.
17: M.9,223.
having the same Jacobian : C.94.
having similar polar forms : M.8.
in a cubic space-curve : J.86.
in two conj ugate indeterminates : C .97.
most general case of linear equations
in: C.99.
{q) groups of: M.23.
with related coefficients : M.12.
transference of, when not of a prime
degree : M.21.
transformation of : M.4,9.
typical representation of : An.68,69.
Binary nomographics represented by
points in space, applied to the
rotation of a sphere : M.22.
Binary nonics, ground forms : AJ.2.
Binary octics : thC.96 : M.172.
Binary quadrics: C.47 : G.3 : J.27:
L.59,77 : M.15,172.
construction of, through a symboli
cal formula : C.57.
indeterminate, integral sol. : J.45.
for a negative determinant : No.81 :
C.60 (table): L.57 : M.172,21,25.
partition table : AJ.4.
representing the same numbers: L.59.
transformation of : C.41.
with two conjugate indeterminates :
C.96.
Binary quantics : 1636 : An.56 : C.52 :
CD.9.
(2m l)-ic, canonical form of: Q.20.
derivatives of : Q.15.
* discriminant of : 1638 : Q.10.
reduction of : J.36 : L.52 : Q.7.
transformation of : CM.l : thE.28.
in two polynomials U, V, prime to
each other and of tho same de
gree: N.85.
Binary quartics :
and their invari
ants- A.18: G. 14: J.41 : M.19:Q.7.

852

INDEX.

Binary quartics(continued) :
condition for perfect square : E.36.
or quintics, with three equal roots or
two pairs of equal roots : P.68.
and ternary cubic, correlation be
tween: An.76.
Binary quintic : G.14.
canonical form for: Q.19.
Binary sextics : taAJ.4: C.64,87,963:
G.14: M.2,76,77.
syzygies of : AJ.7.
Binet's function : L.75.
Binodal quartic with elliptic function
coordinates : AJ.5.
Binomial :
coefficients : 283,3667 :
A.1,2: G.14: Mem.24: N.th60,
61,70,85: Z.25.
sum of selected : Mo.85.
* equations : 480 : A.10 : At.65,68 :
C.10,44: G.52,10: L.57: LM.ll,
12,16.
irreducible factors of : An.69.
zT1 = 0 : see " Roots of unity."
equivalences to any modulus : C.25.
* theorem: 12536: A.8,geo.61: C.
45: CD.7: CM.3: G.12: J.1,4,5,
28,55: Me.71: N.423,47,50,71,78 :
P.l 6,16,95,96: TI.12.
generalization of : J.l : N.572.
Binomial : 5723.
Bipartite functions and determinants :
LM.16.
Bipartition, repeated : Mel.4.
Biquadratic:
equation (see also
"Cubic and biquadratic"): 492
501: A.1,12,16,23,31,39,40,41,
45,69: AJ.l: CM.1,46.,,47: G.5
7,13,21: J.90: Me.62:'N.44,58,,
59,60,63,78,81,83: Q.7,28: Z.6,8,
18.
cond. for two equal roots: E.44.
and elliptic functions : C.57 : L.58.
reduction to canonical form : G.5.
reduction to a reciprocal equation :
L.63.
solution of: A.51,56: C.49,82 : L.
73: Q.19: numerical, C.61: with
out eliminating the 2nd term,
A.39 : 4 variables, J.27 : trigono
metrical, A. 19,70.
and sextic eqs. in the theory of
conies and quadrics : J.53.
*
values which make it a square :
496 : G.7 : E.22.
function with four variables : An. 59.
involutions : C.98.
Biquaternions : AJ.7 : LM.4.
Biternary forms with contmgredient
variables : M.l.
Borchardt's functions : J.82.

Brachistochrone : 3037, 3044: L.48:

Man .31: Me.80: Mem.22: N.
77,80.
Brahmins, trigonometrical tables of the :
TE.4.
Brianchon's theorem : 4783 : A.53 : C.
82: CM.4: G.2: J.84,93: gzN.
82 : Z.6.
and analogues : CD.7 : Q.9.
on a quadric surface : C.98 : E.19.
on a sphere : A.60.
Brocard circle and Lemoine's point :
4754c : gzN.85.
*Burchardt"s factor tables p. 7: erra
tum, A.23.
Burmann's theorem, d.c : 1559.
Calculating machine : Pr.37.
Calculus :
algebraic, which includes
the calculus of imagrnaries and
quaternions : C.91.
of chemical operations : Pr.25.
of direction and position : AJ.6 : M.
pr6.
of enlargement : AJ.2.
of equivalent statements : see"Logic."
of forms (Invariant theory) : CD.7j,
8,9,.
of infinitesimals, third branch of :
viz., given y and yx, to find x :
TE.24.
of limits, ap. to a system of <Le: C.15*.
ofVictorius: Z.16.
of other subjects : see the subject.
Calendar: J.3,9,prs 22.
Jewish: J.f28.
Canal surfaces : A.1,10.
Canon-arithmeticns of Jacobi . C.39,63 .L.54.
Cantor's theorem : M.22.
Cardioid : b=a in 5328, Fig.129 :
A.59,63,68: LM.4: Me.64: N.81.
and ellipses : Pr.6.
Carnot's theorem : 4778.
Cartesian oval : 53415358 : A.69 : C.
97 : LM.l,3j,99 : Me.7o : Q.l : Me.
74: Q.12,cnl5.
area of: E.21.
eq. with triple focus as origin : E.9,23.
foci: E.7.
functional images : Q.l 8.
mechanically drawn : LM.5,6 : Q.13.
perimeter: E.21.
rectific. by ellip. functions : C.80,87 :
LM.5.
through 4 points on a circle : LM.12.
with 2 imaginary axial foci: LM.3.
Cassinian oval: 5313: Me.77,83: N.57.
analogous surfaces : An.61.
radial of : E.26.

INDEX.

Cassinoid with n foci, rectif. of : L.48.

Catacaustic and diacaustic of a sphere :
JP.17.
Catalecticant of a binary quantic : E.
37,38.
Catenary: 5273: Me.64,66,68.
by parabolic trigonometry : Pr.8.
revolving: E.22.
Cauchy's formula i.c : 2712.
closed carve theorem : Mo.85.
various formulas : C2710.
Caustics : 52489 : Ac.4 : L.46 : P.57,
67: Pr.8,14,,15: Q.l,2,3cn8,9,12.
by successive reflexion from spheres :
J.13.
identity with pedals : Z.14.
of a cardioid : Me.83.
* of a circle: 5248 : CD.2.
of a cycloid : A.30.
of an ellipse, focus at centre : J.44.
of infinitely thin pencils : J.98.
of refraction at a plane surface : N.47.
radii of curvature : N.65.
surfaces and singularities : J.76.
Cells of bees : N.56 : Q.2.
Cell structure : LM.16.
Centimetre - gramme - second Bystem :
p.l.
Centrals, theory of : J.243.
Centre of:
curves and surfaces : L.
46: Q.10.
a circle touching two : A.24.
geometrical figures : A.16.
harmonic mean : J.3.
mean distance of curves and surfaces :
N.ths85: Q.33.
mean distance of points of contact of
parallel tangents, ext. of th. : L.
44.
* similarity: 947.
* similitude: 1037.
* similitude of 3 circles : 1046 : 4176.
similitude of 2 quadrics each of which
circumscribes the same quadric :
J.31.
Centres, theory of : J.24.
Centro-baric methods in anal, geometry :
J.5j.
Centroid :
formulae for : 58845902,
6015: JP.26: L.43.
and its use in stereometry : A.39.
of common points of two conies : A.3.
* of circular arc : 6021 : E.13.
of a dice : N.63.
* of frustrums : 6048, &c. : A.33.
of a gauche curve after development
on a right lino, locus of : C.88.
of algebraic curves and surfaces :
An.68.
of a frustrum of a prism : L.39.

853

Centroid(continued) ;
of a frustrum of a pyramid : Me.79 :
N.762.
of oblique frustrum of a cone : E.33.
of a perimeter: A.51.
* of plane curves : 5887 : J.21 : G.12.
* of spherical and other areas : 5898,
6051 : J.50 : L.39,422.
* of surface and solid of revolution :
58969: AJ.3: L.39.
of a trapezium : Q.9.
* of a triangle: 951: A.52,58.
Change of independent variable : 1760
1816 : AJ.3 : CM.l : G.2 : L.40,58 :
Q.1,2,10: Z.17.
* from x, y to r, 6 : 1768.
* from x, y, ztor,6,<t>: 1783.
* in a definite multiple integral : 2774
9: A.22,41.
in transcendent definite integrals :
C.23.
in the theory of isotropic means : C.34.
Characteristics : E.5 : J.71 : M.6.
of conies : A.l : C.67,72,76,83..: JP.28:
LM.9: M.15: N.66,,71.
of conies of 5-point contact : E.27.
of cubic systems : C.742of curves and surfaces : C.73.
of quadrics : C.67 : JP.28 : N.68.
relation between two characteristics
in a system of curves of any de
gree: C.62.
surface groups defined by two : C.79 :
M = l, = 1, C.78.
Chart construction : MeX2: K.60,783.
Chemico-algebraic theory: AJ.l.
Chemico-graphs : AJ.l.
Chess board, ths and prs : A.56: E.34,
42,44.
Chessmen, relative value of: E.39 :
Mel3.
Chineso arithmetic and algebra : C.51 :
N.63.
Chord :
joining two points on a
circle: 4157: A.43,44: E.22.
* of contact for two circles : 4172.
Chronology : J.26.
Circle: 413690c.c: A.l ,3 th 4,9,25,27,
th47: C.94: J.14,17: Q.19: TE.
th6.
* approximate rectification and quadra
ture: 6019, &c: A.2,6,geol3,14,
43: J.32: Me.75,85: N.45,47 :
Q.4.
arc of : see Circular arc.
area of segment : 0035 : A.27,39,44.
* chord of : 4157 8.
* chord of contact: 1017:4138,-72.
* coaxal: 102136: A.23.
configuration of : C.9:>2.

854

INDEX.

Circle(continued) .Circle(continued) :
* touching 3 circles, cn : 946,1049 :
* cn. from 3 conditions : 937 : JP.9.
A.24,26,28,35 : An .68: C.60: Me.
* Cotes's properties : 821: A. 11, ext. to
ellipse 30 : P.13 : TI.7.
62: N.63,65,66,84 : Q.8.
touching the 4 circles which touch
* cutting three at given angles : 4185 :
the sides of a spherical triangle :
LM.3,5: N.83-.
A.4.
division of: A.27,37,41 : At.19: E.l,
and triangle, ths and prs : A.30.57 :
31: G.6: C.85,93: J.27,54,56 :
LM.15: Q.4.
N.53,54.
and theory of numbers: J.30,84j,87 : * two ; eq. for angle of intersection :
41801.
N.56.
*
theorems: 98-t1045: Q.ll : see
ths. on sum of sqs. of perpendicu
" Radical axis " and " Coaxal cir
lars, Ac. : 10948.
cles."
* eight circles touching three, cn:
and two points ; Alhazen's pr : AJ.4.
4189: Mel.3: Q.5.
Circulants:
final expansion of: Me.85.
eight through 6 points of intersection
of odd order : Q.18.
of 3 conies : Q.10.
arc : length, centroid, &c. :
* equation of: 413648, p.c 4151 : Circular :
6019.
Pr.27: TI.26.
with real tangents: Z.l.
*
general eq. : t.c, 4691,4751 : tg.c,
graphic rectification and trans
4906.
position of : Z.2.
Euler's th. extended to ellipse : A.51.
cubics, involution of : LM.1,7.
five-point th. : E.5.
chord of curvature of : E. cn 36.
four pairs of circles through 6 points
and elliptic functions in continued
common to 3 circles : Q.9.
fractions : CD.4.
four points concyclic, condition : A.
* functions: 606: A.17 : J.16.
44: N.84.
points at infinity : see " Imaginary
geometry of : A.67 : Z.24.
ditto."
groups of points on : A.14.
relation of Mobius : LM.8 : N.76 : Q.2.
in tri-metric point-coordinates : Z.27.
* segment : arc, chord and area : 6035 :
* and in-quadrilateral : 733: CD.9.
lines of equi-different powers in two
N.63.
Circulating functions : P.18.
circles : A.19.
* of curvature: 1254,5134: A.31,63 : J. *Circum-centre of a triangle : 4642 : tg.
45: p.cN.84.
eq 4883.
*Circum-circle :
of a triangle: 713,
* polar of x'y' : 4138,64.
4738: tg.eq4895: A.51.58.
rectangles of : Z.14.
coordinates of centre : 4642.
ring of, touching two fixed : J.39 :
hypocycloidic envelope of Ferrers :
Me.78.
N.70.
six points th. : Q.8.
and in-conic : N.79.
and self-conj. triangle: A.41.
* of a polygon : 746 8 : A.19.
and sphere, geo. : Mo.82.
system through a point on a plane or * of a quadrilateral : 743 : N.79.
Circum-cone of a quadric, locus of
sphere: G.16.
* tangents : 413743,4160 : L.56 : P.14.
vertex : N.52.
of a triangle : 4724 :
common to a circle and conic : *Circum-conic :
tg.eq4892: An.57.
cnA.69.
*
common to two circles: 953,4171: * of a quadrilateral : 4697 : At.54.
locus of centre : E.l.
cnA.34.
Circum cubic of a complete quadri
*
locus of a point, the tangents from
lateral : G.10: Q.5.
which to two circles have a given
*Circum - parallelogram of an ellipse :
ratio : geo.9656.
4367.
* three: 9979,1036,104651.11837.
Circum-pentagon of a conic : M.o : N.67.
*
pra.{Gergonne) : 1049 : At.19 : L.46.
Circum - polygon :
of a circle:
through mid-points of sides of a tri
7468: CD.l: Me.80: N.66.
angle : see Nine-point circle.
of a conic : M.25.
* through 3 points : 4156,4738.
of a parabola : CM.2.
through 3 points on a conic : A.2 :
of a cuspidal cubic: LM.13: ditto
J.39. ,
quartic : LM.14.
touching a conic twice : J.56 : N.UO.

INDEX.

Circum-quadrilatoral :
of a circle :
E.35: N.48.
of two circles : N.67.
Circum-rhombus of an equil. triangle :
A.45.
Circum-triangle :
of a conic : N.70.
*
locus of vertex : 4800 : E.35.
of a triangle : J.30.
Cissoid : 530912 : A.62,69 : N .43,85.
tangents of : LM.2.
Cissoidal curves : A.56.
Glairaut's function and equations: Pr.
25.
Clinant geometry: Pr.l0,ll2,12,15.
Closed curves : Me. 77 : geo th Q.4.
* andmovingchord.Holditch's th: 5244:
gzMe.78: Q.2.
ext. to surfaces : Me.81.
quadrature of : A.61 : N.43 : Crofton's
ths A.55: C.65,68: E.36V
JJs'^^tZxcZi/, < = 0, i' = 0 being the
tangents from xij : LM.2.
* rectification of : 5204.
Closed surfaces : JP.21.
Coaxal circles : 102136 : 416170: Q.
52 : reciprocated, 4558.
Poncelet's limiting points: 4165:
thJ.86.
Coaxal conies : Q.10.
Cochleoid, (x2 + y~) tan " 1 " = n ry : A.70.
Coefficients:
detached: 28.
* differential : 1402.
* indeterminate : 232.
Cogredients : 1653.
Collinear and concurrent systems of
points and lines : 967 : A 69 : G.21.
Collineation and correlation : M.22 :
prM.10.
and reciprocity : M.23.
" gleichstimmigkeit " of, in space :
Z.24,28.
multiple c. of two triangles : A.2,70.
of plane figures, ground forms : J.74.
paradox : Z.28.
Combinations: 94107: trA.15: CP.8:
G.18 : J.5,13,212,34,38,th53 : M.5 :
Z.2.
ap. to determinants : JP.28.
complete, i.e., with repetitions : C.92 :
N.42,74.
compound : Man. 79.
* C (, r) an integer : 366 : L.42.
C (n, r) when n is fractional : A.70.
* C (n, r) = 0 (n-1, r-l) + C (n-1, r) :
102.
0 (m+m',p) = 2* C {m, r) . C (m,pr) :
L.42.

855

Combinations(continued) :
* problem or theorem : 1057 : A.21 :
C.97 : CD.2,4,7,82 : 1.5 : J.3,45,56 :
L.38: Mdm.ll: N.53,73 : Z.15.
of Eulerand its use in an eq. : L.39.
of 1,2, ... n, each c. having a sum
>: G.19,20.
of dominoes : An.73.
of n dice each with p faces : TE.21.
of n points in space : L.40.
of observations : L.50.
of planes through a system of
points : N.57.
Combinatorial :
products : A.34.
systems : L.56.
Combinatory analysis : A.2,50,70: J.ll,
22: Mdm.50: N.80.
Commensurable quadratic divisors : N.
47.
Commensurables : TE.23.
Commutative law : 1489.
Companion to the cycloid : 5258.
Complanation formula : A.48.
Complementary functions: C.19: J.ll.
Complete functions : C.86 : J.48j.
Complete numbers : Mo.62.
Complete primitive : 3163.
Complex axes of a quadric : Z.19.
Complexes: L.44,47 : M.2,4.
of axes of a quadric : N.83.
in combinations and permutations :
A.21.
of 1st and 2nd degrees and linear
congruences : An. 76 : L.51 : M.2,
9: N.85: trZ.27.
linear : N.85 : Z.18 : of an in-conic of
a quadrilateral : G.21.
of 2nd degree: G.8,17,18: cnj.93:
M.7 : N.72.
of 2nd degree with a centre : L.82.
of 2nd degree of right lines which cut
two quadrics harmonically : M.23.
quadratic ray- & web-complexes: J.98.
of nth degree, singularities : M.12.
and congruences, spherical of 2nd
degree, their circles and cyclides :
J.99.
and spherical complexes, ap. to linear
p. d. e : M.5.
tetrahedral in point space : Z.22.
Complex numbers: A.28: C.90,99: G.ll:
J.22,35,67,93 : L.54,75,80: M.22:
Mo.70: Q.4.
from the 31st roots of unity : Mo.70.
from the ith roots of unity : J.40 :
Mo.70.
index and base of a power, geo : Z.5.
prime and from roots of unity : J.35 :
taMo.75.

856

INDEX.

Complex numbers(continued) :
prime and from the 5th roots of unity:
Mo.ta 59.
resolution of A'+fr+C" = 0, and
when n = 5 : L.47.
from the roots of unity ; class num
bers : J.65: Mo.61,633,70.
in theory of residues of 5th, 8th, and
12th powers : L.43.
Complex unities : C.96,99 : J.53.
Klein's groups : J.50.
Complex roots of an algebraical eq :
M.l : N.44s: of SB" = 1, Mo.57.
Complex variables, functions of: An.
59,68,7182,83: G.3,6 : J.54,73,83 :
M.19 : Z.82,10.
especially of integrals of d.e : J.
75,76.
Composite functions of a higher order :
G.2.
Composite numbers :
for construc
tion of factor tables : A.45.
groups of: J. 78: LM.8 : Me.79.
" Compteurs logarithmiqnes " : C.40.
Concavity and convexity : 5174.
Concentric circles : LM.14.
3 quadrics, intersection of : E.39.
Conchoid: 5320: A.55: N.43*.
Concomitants of a ternary cubic :
AJ.4.
Concurrent lines and collinear points
967-76: A.69.
th. on conic and triangle : B.35.
Concyclic conicoids : 995 : Q.ll.
Cone : 115059, 6043 : A.16 : L.61 : Me.
62.
and cylinder, auperficcs, tr: An.57.
general d.e of : E.18.
intersection of two : N.64.
* oblique: eq 5598 : J.2: Me.80.
* sections of: 11509.
* and sphere : 5652 : thsHe.64.
superfices of oblique frustrum : J.2.
through m points and touching 6m
lines : LM.4.
volume of frustrum : Ac.41 : N.13.
Configuration of 16 points and 16 planes:
J.86.
(3, 3)10 and unicursal curves : M.21.
Confocal conies : 45508, 5007, tg.e
5005: J.54: LM.12,13 : Me.66,68,
73: N.80: Q.10: TE.24: Z.3.
* Graves' theorem: 4555: Griffith's ext.
LM.15.
* tangents of : 4555.
Confocal quadrics : 565672 : A.3 : CD.
4,5,93: G.16: M.18: thMe.72: Q.3.
relation to curves and cones : CD.4,9.
volume bounded by three and the co
ord, planes : A.36.

Confocal surfaces : Me.66.

Conformable figures : A.59: LM.10: M.
19:Z.17.
Congruences : C.51,88 : th and apJ.19 :
thMe.75: N.50: P.61.
binomial : AJ.3 : C.61 : expon. to base
3, Mel.4.
classification of roots : C.63.
Cremonian : LM.14.
irreducible: J.40.
and irreducible modular functions :
C.61.
linear: LM.4: of circles in space : C.93.
numerical : An.60 ; multiplication of,
61.
of 1st degree : A.32.
in several unknowns : L.59.
sol. by binomial factorial : Mem.44.
transformation of modulus : Mel.
2: N.59.
with composite modulus : E.30.
of 2nd degree: 0.62*: Mem.31 : re
duced forms : C.74.
of 3rd order and class : LM.16.
higher, with real prime modulus :
An.83: J.31,54,99.
resultant of systems of linear : C.88.
and trigonometrical functions : J. 19.
z"+y2 = 1 (mod. p) : J.19.
as' = 1 (mod. p) : J.31.
Congruent divisors of a number, no. of :
A.37.
Conical functions : M.18,19.
Conical surfaces : 5590 : A.63 : Ac.5 :
LM.33: M.3.
through 6 points, locus of vertex : J.92.
Conicoids : 5599 : A.48 : Q.tg.o 9, q.c
and t.c 10.
50-point : Me.66.
Conies: 40325030: A.l..,5,17,31,32,60,
68: C.83: G.1,2,3,21: J.20,30,32,
45,69,86: M.17,19: N.42,43s,44a,
45,7L,753,82 : P.62 : Q.8,tg.c9:
Z.18,21,23.
anharmonic correspondents, problem
of 5 conies and 5 lines : N.56.
of Apollonius : L.58.
* angles connected with : 4375.
arcs similar to: N.44.
* areas of (see also " Sectors ") : 4688,
60976121 : N.46 : t.cQ.2.
* auxiliary circle : 1160.
* centre :
coordinates of 4402, 4267,
t.c 4733 and 4742 : tg.eq of 4901 :
ths and prs N.45.
*
locus of 4520, 5028.
* chords of: 4315,4322; p.c 4337 and
CD.l: Me.66: see also " Focal."
cutting an ellipse at a given angle :~
E.28.

INDEX.

Conies :
chords of[continued) :
intersecting: 1214,4317.
moving round an ellipse : A.43,44 :
E.22.
* chord of contact : 4124,4281 ; 4699
4721.
* and circle, intersection th : 1263 :
A.59 : N.64.
collinear relation to circle : Z.l.
and companion quadric : An.60 : JP.7.
conjoint lines of : L.382.
* conjugate diameters: 1193 1213,
4346 ; ths 127885 : CM.l : L.37 :
N.42, ths 44,69 : Q.3.
*
parallelogram on : 1194: 4367.
relation to ellipse when equal : A.18.
conjugate points : Q.8.
* construction of : 1245,4822 : A.28,43 :
E.29: Q.4: N.59,73: with help of
circle of curv. A.24.
*
from conj. diameters : 1253: A.52.
* contact of: 452733: A.1,60: C.78:
Pr.34.
at 2 points : Q.3 : N.74.
*
do. with each of 2 conies or circles :
4803 -6: CD.5,6: B.31,34.
4-pointic with a quartic : M.12.
*
5-pointic : 5190-1 : C.782 : E.5a,23 :
J 21 P.59.
with surfaces : C.912 : P.70,74.
convexity from focal property : N.56.
criterion of Mobius : J.89.
* criterion of species : 446477 ; t.c
4689: 5000.
* curvative:
centre and radius of:
453449 : geo 1254-66 : thsMe.
73 : N.79,85.
*
geocn: 1265: A.17.
*
chord of: 1259,64: Q.6: locus of
mid-point A.70.
as curves in space : tr A.37 : M.64.
* definitions : 1160.
degenerate forms : LM.2,.
* diameters : 1214,35 : eq 4458 : cn
Me.66: N.65.
* director circle : geo 1217, eq 4693 5 :
o.cE.40: LM.13: N.79.
* directrices: 1160: trA.63: LM.ll.
* eccentric values of coordinates : 4275 :
CM.4.
* eccentricity: 1151,4200.
elementary formula : G.9.
* ellipse and hyperbola : 425096.
* equations of: 4251,4273; p.c 4336;
tg.c 4663,4870 : J.2.
*
general: 4400,4714,4719; t.c 4755
and 4765; tg.c 4664 and 4872;
p.c 4493 : t.c A.51 : CP.4,5 : t.c
G.6,7: Mem.52: N.43,45,65. (See
also " Conies, general equation.")
*

857

Conies :
equations of(continued) :
*
intercept : 4498.
* equations of parabola : 4201 ; t.c
4775; p.c 4336.
*
general: 4430,4713; t.c 4656 with
4689 ; tg.c 4974 with 5000.
* ay = A/3S and derived equations : 4697
4719: Q.4.
* ay = fe/35 or LM= IP : 4699, 4784 : N.
44: 8+1? = 0, &c: 4707.
* La*+Mp-+Nr = 0: 4755, '65 : Me.62.
equation in p and p -. Q.13.
* equi-conjugates, gen.eq : 4491.
formulas: J.39: N.62.
from oblique cone : L.38.
* general equation ; cond. for a circle :
*
4467: t.c 4691 and Me.68 : Q.2:
from eq. of axes : N.57.
*
cond. for an ellipse : 4464 ; t.c 4689.
*
cond. for a hyperbola: 4468; t.c
4689: A.39.
*
cond. for a rectangular hyperbola:
4737 ; t.c 4690 ; tg.c 5000.
*
cond. for a parabola : 4430 ; t.c 4696,
4735, 4746 and 4775 ; tg.c 5000.
*
cond. for two right lines : 4469,
4475, t.c 4662.
generation of : N.75 : Z.23.
by a moving chord of a circle : A.34.
*
Maclaurin's method : 4830 : LM.4.
*
Newton's method : 4829.
graphic problems : N.80.
Halley's pr : N.76.
harmonically in- and circum-scribed :
Q.18.
intersecting in 4 points : J.23.
intersecting a surface in 5 points :
C.63.
with Jacobian = 0: M.15.
limiting cases : 446577 : Me.68.
* normals: 1171,sd562932 : A.16,24,32,
cn 43,47 : An. cn 64,78 : C.72,84 :
J. cn 48,56,62 : Me.66 : Mel.2 : N.
70,81 : Q.8: Z.11,18,26.
circle through feet of: N.80.
cutting off the min. or max. arc or
area : N.44.
dividing ellipse most unequally :
E.29.
*
eccentric angles of the feet of four,
th: 4334.
*
equations of : 4286, 4483, 4512.
*
intercepts: 4294: segments; 4309,
4486.
least distance between two : A.21,38.
number of real : J.59 : N. 70,72..
number cut by 8 lines in space : J.
68.
number under double conditions : C.
59.
5 R

$58

IS'LEX.

Conic*(vmtinwA) :
octagram : LM.2.
* passing through given points and
touching given lines : cn 4831
40.
5 points: cn4831, eq 5024: A27:
A.9,24,64: An.50: N.57.
4 foci of a conic : Q.5.
4 point* : N.66: Q.2,8: Z.9.
4 point h, envelop of : E.28 : Do. of
axes, N.79.
*
4 points, locus of pole of a line :
4770.
*
4 points and touching a line: cn
4833: A.65.
3 points and touching a circle
twice: N.80.
3 points and touching a line : Q.6.
3 points with given focus : cn A.54.
2 points and touching a line : Q.2.
2 points and touching 2 lines, locus
of centre : 0 7
four such conies, th : Q.8.
1 point and touching 3 lines : Q.8.
parameter of : N.43.
* perpendicular from centre on tan
gent : 1195, 436673.
*
ditto from foci : 1178, 4300.
of 8 points : J.65.
of 9 points : A.43 : G.l.
of 9 points and 9 lines : G.7,8.
of 14 points : Me.66.
pencil of : M.19.
and polar, Uesarqnes' th: N.64.
* pole of chord joining xfjv se,yj : 4326 :
parabola, 4218.
* properties of: 1274: A.4,25,70 : p.cJ.
38.
quadrature of : TE.6.
and quadrics : A.30 : L.42 : N.58j,665,
ths 73 : geo interpretation of
variables, N.66.
* rectification of: 6071, C084: P.2 :
TI.16: Z.2.
* reduction of , u' to the forms
3e>+y*+z* = 0, ax,+Py2+yz'> = 0:
_ 4995.
scries of : A. on 67 and 68.
* soven points of: C.94.
* similar : 4522.
six points of : A.62: J.92.
systems of : C.62,65 : M.6 : N.66 : of 2,
Q.7: of 4, L.54.
and orthogonal lines : N.71.
and quadricB : Z.6.
* tangents or polar: 11679, 42805,
47f0: gn.eq 4478: A.61 : cnCD.
3 : CM.l, p.eq 4 : N.79.
intercepts on the axes : 4292.
*
two at the origin, eq : 4489.

Conks :
two from iy-. 4311:
4+.
llsl:
4965, A-57 : tx 46802,
A^3j.
do. for parabola: 4215,
1332:
ratio of lengths, 1243.
quadratic for m : 4313 : parabola.
4220.
quadratic for abscissa? of points of
contact : 4312 : parabola, 4216.
subtend equal angles at focus :
1181, 1234 : C1L2.
locus of *y : A.69.
segments of : 4307.
at the points /, tan <, pcotp:
4799.
tangent curves of : Z.15.
theorems: 1267: A.54: J.16: L.44:
M.3 : N.45s,4cV72,84 : Q.4.6,7 tx :
by Pascal, Desarques, Caraot, and
Chasles, A.53.
conic and triangle, Q.5.
3 circles touch a conic in A,B.C and
all cut it in D; A,B,C,D are
concyclic J.36.
three : 4707, 4710; in contact, 4803.
Jacobian of : 5023.
touching :
a conic and line, cond. :
5017.
a curve twice : J.45.
curves of any order : C.59.
five curves : C.584.
four lines: 4804: locus of centre.
4772, 5028 : locus of focus, 5029 :
N.45,67.
n lines : N.61.
a group of lines, and having a given
characteristic and focus: A.49.
a quintic curve in 5 points, no. of:
X.66.
two circles twice : 4806.
two conies twice : 4803.
two sides of the trigon : 4784
4808.
transformation of : G.10.
two : 4936-5030 : N.58.
with common chords or tangents :
47005.
common elements, cn : A.68.
with common points and
47017 : LM.14 : Z.18.
at infinity : 47156.
common pole and polar, cn : 4762.
condition of touching : 4942, t.c
5021.
intersecting in 4 points : 4700 : A.
32 : at oo, A.16.
points of intersection : tg.e 4973 :
cn A.69.
reciprocal properties : E.29.

INDEX.

Conies :
two(continued) :
reduction to a?+y*+z2 = 0 and
<wJ+j3i/2+yzs = 0: 4995.
*
six chords of, eq : 4941.
*
tangents of, four: eq 4981 : J.75:Q.3.
* under5 conditions: 482243: L.10,59.
6 conditions, cn : C.59.
7 conditions in space : C.61.
Conjugate :
functions : apLM.12 :
TI.17.
lines of surfaces : CD.9.
* points : 1066, 5184 : in ellipse, A.38.
point-pair of a conic : Z.17.
tetrahedrons of a quadric, each ver
tex being the pole of the opposito
side: N.60,83.
triangle of a conic : N.83.
Connexes (1, n) corresponding to d.e :
A.69.
in space : Mel.6.
linear: M.15.
of 1st order and class in Bimple invo
lution: G.20.
of 2nd order and class : G.19.
analoguein anal. gco.of space: M.14.
Cono-cunei : A.2.
Constant coefficients, theory : 1.23.
Constant functions and their deriva
tives : A.15.
Contact and circle of curvature : 4527,
5188, 5134 : A.30.
Contact of curves :
5188 : cir. of curv.
A.53 : C.32 : J.66 : P.62 : Pr.ll :
Q.7.
with a parabola : Z.2.
with faisceaux of doubly infinite
curves : C.83.
with surfaces : L.78 : with triangles,
M.7.
Contact :
of lines with surfaces : L.
37: Q.1,17: Z.12.
of an implexe with an alg. surface :
C.84.
of quadrics : LM.5.
4-pointic on an algebraic surface: M.9.
of spheres : JP.2.
of surfaces: J.4: JP.15: P.72,74,76 :
Q.12.
of 3rd order between 2 surfaces: C.74;.
problem: of Apollonius, A.66: spheri
cal, At.l92.
transformation : C.85 : M.23.
Contingence angle of : see " Torsion."
Continuants : Aj.l : Me.79.
Continued fractions : 16087 : A.18,33,
55,69 : An.51 : gzC.99 : CM.4 : th
E.30: G.10,15: J.6,8,114 and ap,
18,53,80 : L.50,68,65 : LM.5 : Me.
77: Mem.aptoi.c9,13:Mo.66:N.
42,tr49,56,66 : Q.4 : geoZ.12.
*

859

Continued fractions (continued) :

11 and,1111
a.ot-J
;
, : A.42
: J.3.
a+ a+
a+b+a+b +
^\ b>a+l: Mel.l.
a+ <x + l + a +2 + &c.
1
1
and
X+X+1+ x+2+
1
1
A.30.
2.c+l+2j;+3+2x+5 +
2+LL and
E-36-332+2+
<Z+2+
ascending: A.60 : Z.21.
algorithm ^> = ap i + 6p,,-2 : 168 : J.
69,75.
do. ap. to solution of trinomial eq :
A.66.
combinator representation of the ap
proximation : A. 18.
development : Ac.4 : C.87 : Mem.9.
for e* and \og(l+x): CM.4.
for e exp. a+bx+cx2+ : C.87.
x-1-<e-i+x-x-+
J.27,28.
for
lx-1+x-,x-'+
for Usina+Vcosaj+W; U.V.W,
polynomials in x : J.76.
for (m+ Vn)+p : N.45.
for powers of binomials : CM.4 : M^m.
18.
infinite : A.33j.
numerical values : Q.13.
Eisenstein's, TE.28: Wallis's, Mem.
15.
periodic : A.19 (2 periods), 33 : G.16 :
J.53 : N.42,43,45,462.
reduction of : J.46.
* reduction of a square root to a : 195 :
A.64: J.31.
do. of a cube root : A.8.
do. of an nth root : A.64.
Contingence, angle of : 5725.
Continuity, principle of : CP.8: in rela
tion to Taylor's and Maclaurin's
theorems, L.47.
Continuous functions: 1401: A.15:
Ac.5: : C.18,20, and discontinuous
40, of integrals of d.e 23: TA.7.
Continuous manif'oldness of 2 dimen
sions : LM.8.
Contragredients : 1813.
Contraposition : E.29.
Contravariants : 1814 : G.12, of 6th deg
19.
of two conies : 4990 and 5027.
Convcrgents : 16087 : gzC.98 : CD.5 :
J.37,67,58: N.46.
Convex polygon, intersection of diags :
N.80.

860

INDEX.

Coordinates, transformation of : 4048

51,4871,557481.
Coordinate systems: 400131, 5453,
55016 : G.16 : J.5,45,50 : LM.12.
ap. to caustics : An.69.
* areal: 4013: Me.80,82, gn eq 81.
axial : N.844,85 : Q.10.
* biangular: 545373: Q.9,8,13.
* bilinear : ap 5341 : A.32.
bipolar : N.82.
bipunctual : AJ.l.
Boothian : see tangential.
* Cartesian : 4001.
dual inversion of c.c and t.c : Q.17.
central: Z.20.
curvilinear : A.34 : An.57,64,683,70,
73: C.48,54: CP.12: G.10, and
i.c 15 : L.40,51,82 : Mem.65 : Q.19 :
on a surface and in space, An.692,
71,73: including any angle, J.58.
* eccentric values of : 4275, 5638 : CM.4.
elliptic: A.34,40;: J.85 : Q.7 : Z.20.
Fuchsian functions of a parameter :
C.92.
hyperelliptic coordinates : J.65.
* one-point intercept : 4026.
* two-point intercept : 4025.
linear : Z.21.
mixed coordinates : A.13.
parabolic : C.50.
parallel : N.84.,,85.
pedal : Me.66.
pentahedral : Me.66.
of a plane carve in space : LM.13.
* polar : 4003,sd5506 : Me.76.
polar linear in a plane : Z.21.
quadrilinear : Me.62,642.
* quadriplanar or tetrahedral : sd5502,
apA.53: Q.4.
surface: C.65,81.
* tangential: 4019, 48704915, 5030:
Me.81 : Pr.9 : Q.2,8.
* tangential rectangular, or Boothian :
4028.
tetrahedral point coordinates : Z.8.
triaxial : A.64.
trigonal: G.14 : Q.9.
* trilinear : 4006 : A.39 : Mc.62,64 : N.
63: Q.4,5,6 : conversion to tan
gential, 4876.
trimetrical point: A.67.
Coplanation :
Z.ll.
of central quadric surfaces : Z.8.
of pedal surfaces : Z.8.
Coresolvents : Q.6,10,14 : non-linear,
TN.67.
Correlation of planes : J.70 : An.75,77 :
LM.5,6,8,10.
Correlative figures, focal properties:
LM.3.

Correlative or reciprocal pencils : M.12.

Correspondence :
principle of : geo
thAn.71 :
extC.78,80,ex83,85, :
N.46.
of algebraic figures : M.2,8.
application to Bezout's th C.81 ; to
curves, C.72 ; to elimination, N.
73 ; to evolute and caustic, N.71.
complementary theorem to : C.81.
determination of the class of the en
velope and of the caustic of a
curve : C.72.
determination of the degree of the
envelope of a curve or surface of
>i parameters with n2 relations :
C.83J.
determination of no. of points of inter
section of 2 curves at a finite
distance : L.73 : C.75.
determination of the number of solu
tions of n simultaneous algebraic
equations : C.78.
determination of the order of a geo
metrical locus defined by alge
braic conditions : C.82,845.
forms : M.7.
multiple in 2 dimensions : G.10.
of curves: C.62 : P.68: Q.15.
of two planes : LM.9.
of points : LM.2 : C.62 on a curve :
Q.ll on a conic : M.18 on two
surfaces.
for groups of n points and n rays :
M.12.
of two variables (2,2) : Q.12.
of 2nd deg. between 2 simply infinite
systems : An. 71.
Correspondent values, method of : P.
1789.
Corresponding points :
in some in
volutions : LM.3.
on two curves : M.3.
on two surfaces : C.70.
Corresponding surface elements : M.ll.
Cosmography, graphic method: N.82,79.
Cotes's theorem: circle,821; areas, 2995.
analogous ths : CM.3.
Counter-pedal surface of ellipsoid: AJ.4.
Ooupures of functions : C.99.
Covariants : 1629, 49365030 : C.80,81,
th90: G.1,20: Q.5,16: J.47,87,
90 : thM.5 : N.595 : ap to i.c C.56.
binary : G.2.
of binary forms : An.58 : C.82,86,87 :
G.17 : L.76,79 : M.22.
of binary quadrics, cubics, and quartics : An.65 : J.50 : Q.10.
of binary quantics : E.31 : Q.4,5,17.
of a binary quartic: J.53: quintics.An.
60.

INDEX.

Covariants(continued) :
and contravariants of a Bystem of
simultaneous forms in n varia
bles ; to find the number of: C.84.
of quantics : An.58 : binary, Me.79.
of a septic, irreducible : C.87.
sextic : G.19.
of a system of binary cubo-biquadratics ; number of irreducible covariants: C872.
of a system of 2 binary quadratic
forms ; number of : C.84.
of ternary forms : G.192.
* of two conies : 4989, 5026 : of three,
Q.10.
Covariants and invariants : An.60.
of binary forms : An.58,59,61,83 : C.
66,69 : reciprocity law, C.86.
of a binary octic, irreducible system :
C.86,93.
of a binary quintic : C.96.
of a binary sextic : C.96V
as criteria of roots of equations : An.
68.
Cribrum or sieve of Eratosthenes : N.
43,49.
Critical functions : see" Seminvariants."
Criticoids and synthetical solution : E.
9,26,32.
Cubature of solids : see " Volumes."
Cube numbers, graphic cn of : E.43 :
Me.85 : tables to 12000, J.42.
Cube root extraction: A.22,64: J.ll :
N.44,58,,: TI.1.
Cube roots, table of (2 to 30): p.6.
Cubes of sums of numbers : N.71.
Cube surfaces : Pr.17.
Cubical parabola : Q.6,9.
metrical properties : M.25.
Cubic and biquadratic equations : 483
501 : A.45 : No.1780 : An.58:
C.geo41,90: JP.85: L.59 : M.3 :
N.79: Z.8; Arabic and Indian
methods, 15.
Sturmian constants for : Q.4.
mechanical construction : LM.22.
Cubic and biqaadraticproblems: An.702.
Cubic classes which belong to a deter
mining quadratic class, number
of: A.19.
Cubic curves: An.73 : AJ.5,15,27: C.
37: thCD.7: CP.ll: G.l;,2,14,
23 : J.ll,32,34,ths42,63,geo78,90:
L.13,44,45 : M.15 : t.cMe.64 : Mo.
geo56: N.50,67,geo723 : P.57,58 :
Pr.8,9: Q.4,,5: Z.geol5,22.
classification of : Q.16.
and conies which touch them : J.36.
coordinates, explicit functions of a
parameter : J.82.

861

Cubic curves(continued) :
48 coordinates of : P.81 ii.
with cusps : A.68.
degeneration of : A.4 : M.13,15.
derivation of points : LM.2.
with a double point : M.6 : with two,
M.3.
with double and single foci : Q.14.
Geiser's: J.772.
generation of : G.ll : J.36 : M.52,6 :
by a conic pencil and a projective
ray pencil, Z.23 : linear, J.52.
and higher curves : A.70.
mechanical construction of : LM.4.
number of cubic classes which belong
to a determining quadratic class :
A.19.
nodal, tangents of : LM.12.
of third class with 3 single foci : Q.
14.16J.
18 points on : E.4.
inflexion points of : J.38 : M.2,5 : N.
73,th83,85.
12 lines on which they lie in threes :
E.29.
rational: A.58 : G.9.
referred to a tetrad of corresponding
points : Q.15.
represented by elliptic functions :
JP.34.
and residual points : E.34_..
resolved into 3 right lines : M.14.
and right lines depending on given
parameters : J.55.
16 cotangential chords of : Q.9.
and surfaces : J.89.
synthetic treatment : Z.21.
tangential of: P.58: Pr.9.
tangents to : cnE.257,ths28 : LM.3 :
Q.3.
with a double point or cusp : M.l.
forming an involution pencil : LM.
13j.
their intersections with cubics or
conies : C.41 : TI.26.
through 9 points : C.cn36,37 : L.54 :
two cubics do., CD.62.
through 8 points : Q.5.
through2circularpointsat 00 : cnZ.14.
transformations of: C.91.
Cubic equations: 48391: A.1,37,11,
22, 25a, 32, 37, 41 , 42,44* prs47, 68 :
An.55: C.num46,85: CD.4,6: E.
35: G.12,16: J.27,56,90: LM.8 :
M.3 : Mem.26 : N.45,th52,th56,64,
66,70*75,78,81,84: TE.24: TI.7.
See also " Cubic and biquadratic
equations."
and division of angles : A.15.
in a homographic prof Chasles : C.54.

862

INDEX.

Cubic equations(continued) :
irreducible case: A.30,39,41,42: No.
1799: AJ.la: C.58: 3.2,7: L.79:
N.67 : in real values, A.49: by
continued fractions, A.2.
roots :
condition of equality : E.30.
definition of: thA.31.
geometrical construction of: C.44,
45 : Aral), m.s.s, J.40.
integral : N.75j.
power-sums of : C.54,55.
squaresof differencesof : An.56: Q.3.
* solution of:
Cardan's formula:
484: A.14,40,gz22 : Me.85.
Cockle's: CM.2,3.
*
trigonometrical : 489 : A.19 : N.61,
67,71 : TE.5.
by differences of roots : J.42.
by continued fractions : A. 10,39.
by logarithms : C.66.
by x = +h: C.60.
2
*
mechanical : 5429 or C.79.
numerical : C.44.
Cubic forms (see also " Binary cubics "):
thsJ.27.
ternary: J.28,29 : JP.31,32.
quaternary: transfJ.58 : P.60: Pr.10:
division into five, J.78.
Cubic surfaces : Ac.3,5 : thsAn.55 : C.
97o,98 : G.22 : J.53,65,68,69,88,89 :
M.6: Mo.56: N.69: P.69: Z.20.
classification of: M.14.
double sixers of : Q.10.
with 4 double points : M.S.
" gobbe": G.14,17,21.
hyperboloidal projection : G.2.
27 lines of : C.52,68,70: J.62 : L.69:
M.23: Q.2.
27 lines and 45 triple tangent planes
of : An.84 : and 36 double sixers
Q.18.
locus of centre of quadric through 8
points : Me.85.
model of : CP.12. '
polar systems : M.20.
properties of situation : M.8.
in quaternions : AJ.2.
reciprocal of Steiner's surface : N.72s,
73.
singular points of : P.63.
triple tangent planes : CD.4,.
Cubo-biquadratic eqs., no. of irreducible
forms: C.87.
Curvature: 12548: 5134,5174: A.l,
28,43 : J.81 : Me.62.,,f64,72,75 :
N.th60,69: Q.t.c8,12.
* circle of: 12545, 51 3'4: A.cn30,37:
J.C8.

Curvature(continued)
* at a cusp : 5182 : N.71.
* at a double point : 5187 : Q.3.
dual, evolnte and involute : Q.10.
of an evolute of a surface : C.80.
of higher multiplicity (Riemann) :
Z.20,24.
* of higher order : 518891 : M.7,16.
of third order : C.26.
of intersection of 2 quadrics : An.63.
mean: C.92.
* at a multiple point : 5187 : C.68.
of orthogonal lines : JP.24.
* parabolic: 5818.
of a plane section of a surface : C.78 :
Z.17.
spherical : A.25.
Curvature of surfaces : 581826 : A.4,
20,41,57 : An.f and ths61,64 : C.
th25,49,60i.ths66,67,68.geo74,84 :
J.1,3,7,8: JP.13: L.44.72-: p.c
Me.71 : Mcl.3 : Q.12 : Z.27.
* average, specific, integral, &c. : 5826
30.
axis of curvature of envelope of a dis
placed plane: C.70.
approach of 2 axes of finite neigh
bouring curves : C.86.
circular and spherical : see "Tortuous
curves."
constant : J.88 : G.3 : mean, C.76, L.
41,53; neg.,C.60,M.16; pos.,G.20;
total, C972.
Euler's theorem : gzC.79.
Gauss's
Q.
16. thC. 42 : analogy M.21 : V
ap. to aneroid barometers : C.86.
indeterminate : CD. 7.
and inflexion : trA.19.
* integral : 5826.
and lines : An. 53.59 : L.41.
mean = zero throughout : Mo.66.
and pencils of normals : C.70.
and orthogonal surfaces : P.73.
of revolution : L.41 : Z.21,22.
skew : Z.26.
sphere of mean curvature of ellipsoid :
A.43.
Curves (see also " Curves algebraic "
and " Curves and surfaces ") :
5100: A.2,16,32,66: An.53,54. :
C.geo72,91 : J.14,31,34.,63.643,70 :
L.38,44.ths57 and 61 : M.lti : N.
p.c61, 71,77,853 cn from p.c.
from Abel's functions, p 2 : M.l.
Aoust's problem : A.2,66.
arcs of, compared with lengths : JP.
23.
of " allineamento " : G.21.
analytical method : LM.9,16.

INDEX.

Curves (continued) :
whose arcs and coordinates are con
nected by a quadratic equation:
J.62.
whoso arcs are expressible by elliptic
or hyperelliptic functions of the
1st kind: Z.25.
argument of points on a plane curve :
LM.15.
bicursal : LM.4,7.
with branches : imaginary, CM.1.Q.7:
infinite, Q.3.
2 characteristics defining a system
of algebraic or transcendental
curves : C.78.
least chord through a given point :
A.23.
class, diminution of : N.67.
closed : see " Closed curve."
of 2-point contact with a pencil of
curves : M.3.
of 3-point contact with a triply infinite
pencil of curves : M.1U : 4-point
do., LM.8.
whoso coordinates are functions of a
variable parameter : Me.85 : ellip
tic, J.64 : N.68.
cutting others in given angles or in
angles whoso bisectors have a
given direction : C.58,83.
and derived surfaces : An.59,61.
derived from an ellipse : A.10.
determination from their curvature:
P.83,84.
from property of tangents : A.51.
determination of the number of curves
of degree r which have a contact
of degree n<mr, with an jfi.-tic,
and which satisfy \ r (r+3) n
other conditions, and similar
problems : C.635.
defined by a differential equation : C.
81,90,93,98: L.81,82.
do. algebraic and an analogous
space theorem : L.76.
<fc*+<hi = 0, Pr.15: p oc r, Mem:
24 : p" = A sin a, N.76.
diameters of : L.49 : N.71 : and sur
face, C.60.
eq obtained from tangent : N.45.
whose equations are : y = x, A.14,
16: v3 = u (it1) (x) (uy),
x and y constants, C.93.
*
r <p = a sin <, A.48 : y = T (x), 2323.
linear functions of the coordinates :
N.65.
equidistant, tangents to : cn Z.28.
whose evolute and involute are equal :
C.84.
extension to space : C.85.

863

Curves(continued) :
a family of: N.72.
four, with two common points : Q.9.
generation of : geoJ.58,71 : M.18.
by intersections of given curves :
Z.14.
by collinear ray-systems : Z.19.
geometrical : A.37 : two laws, C.84.
relation to harmonic axes : C.73<.
" gobbe": of zero kind, G.ll : rational,
G.9,12,
higher plane : A.70 : L.61,63.
homofocal : N.81.
defined by intersecting conies : C.37.
intrinsic equation : CP.8 : Q.5.
joining two points : pr L.63.
with multiple points : C.62 : L.69.
with three of higher degrees, cn
An.58.
n.-tic with m.p of n1th order : C.
80: N.76.
network of : C.67.
pencils of : A.65 : of 3rd order, Z.13.
2>+P2+ = +: E.ll.
with a constant polar subtangent : N.
62.
with several " points d'arret " : N.
60.
in a power-series of sines : J.3.
* of pursuit : 5247 : C.973 : N.83.
of " raccordement " : JP.12.
rational: A.56 : G.thl5,16: M.9,18.
generation of : G.12.
reciprocal of : J.42.
of section : A.43.
of a series of groups of points, ths :
C.73,.
with similar evolutes : Me.66.
* singularities of : 5187 : An.71 : C.78,
80 : CP.9 : J.64 : JP.7 : L.37,45 :
LM.6: M.8IQ.,16: N.50,80,813 :
Q,2,7 : higher singularities, J.64:
L.70.
of the species 1 : C.973.
sextactic points of : P.65.
on surfaces : see " Surface curves."
systems of : An.61 : G.13 : Mo.82 :
theory, C.635,94.
and surfaces : A.73 : Ac.7 : L.65.
tangential polar eq of : Q.l.
theorems or problems : A.prl,313,prs
37 and 42: G.l: J.l : M.14: Q.3.
re arc CP and chords CP, PM, CM
Mem.10.
to describe curves which shall have
equal arcs cut off by a fixed pen
cil of lines : Mdm.10.

864

INDEX.

Curves(continued)
re lines drawn at all points of a
curve at the same inclination to
it: C.74.
tracing apparatus : LM.4.
transformation of : CD.8 : LM.l :
scalene Q.13 : M.4,20,21 : of 14ics which cut a quartic in the
points of contact of its double
tangents: J.52.
and transversals : J.47.
under given conditions : P.68.
Curves algebraic (see also " Curves and
surfaces ") : C.99,ths60 and 80 :
CM.4 : G.1,4,5 : J.12,47,59 : N.50 ,
81,on83.
represented by arcs of circles : JP.20.
with axes of symmetry : N.80.
of 2nd class and 2nd order : G.l.
of 3rd class and 4th order : G.4.
of 3rd class and curves of 3rd order :
J.38 : L.78.
of 4th class with a triple and a single
focus : Q.20i.
of 6th class : Ac.2.
of class n and order m, two laws :
C.85.
common points, a system of : J.54.
generation by right lines : J.42.
and homothetic conies, ths : J.53.
lemniscatic : An.68.
manifoldness of : M.10.
* mechanical construction of an n-tic :
5407: LM.7.
with a mid-point : J.47.
number of points of contact : C.82.
number of intersections : C.76 : M.15.
projective involution : M.3.
remarkable group of : M.16.
species determined : M.23.
symmetrical expression of constants :
Q.5.
theorems : two metric, 11.11 : Maclaurin's, N.50.
Curves and surfaces :
M.8 : N.59 :
gnthsC.45 and G.4 : algebraic,
An.77 ; J.49 and L.55.
" arguisiane ": G.12.
curves having the same principal
normals and the surface which
the normals form : C.85j.
of same degree, a common property :
G.8.
satisfying conditions of double con
tact: C.89.
Curvilinear angles, ths : L.44,45.
Curvilinear triangles : A.612 : N.45.
Curvital functions : C.60.
Curvo-graph : A.l.
Cusps : 5181 : Mem.22 : Q.10.

Cusps(continued) :
construction of 8 cusps of 3 quadric
surfaces when 7 are given : J.26.
* keratoid: 5182.
* ramphoid : 5183.
Cyclic :
curves : A.37 : Z.cu2,26.27.
functions : A.69, and hyperbolic 37.
interchanges (higher algebra) : Man.
62.
projective groups of points : M.13,20.
number of do. in a space transf. :
C.90.
surfaces : Z.14.
systems : C.76.
Cyclides : N.66,70 : Pr.19 : Q.9,12.
reducible : LM.2.
and sphero-quartics : P.71.
Cycloids : 5250 : A.13 : N.52,82.
and trochoids on surface of sphere :
Mem.22: Q.19.
surface of, th of Archimedes : Me.
84.
Cycloidal curves : Z.9.
Cyclosis in lines : LM.2.
Cyclotomic functions : C.90j:
Cylinder, frustrum of : 6048.
circumscribing a torus of revolution:
C.45.
and cones, intersection by spheres,
ths : J.54.
and hemisphere : P.12.
Cylindrical functions : A.56 : An.73 :
M.5,16 : and d.iM.8.
of 1st and 2nd class : M.l.
J (x) analogous to the spherical func
tion P" (cos 6) u : M.3.
representing a function of 2 variables:
M.5.
Cylindrical surfaces : 5591 : LM.3.
quadrature of : A.9.
Cylindroids : At.19,39 : Me.80 : Z.25.
Decimal fractions :
approximation
by : N.51.
error in addition of non-terminating :
C.40: N.56.
repeating: A.16,33,56: G.9: Me.85:
Mel.5: N.42,49,74.
- where p is one of the first 1500
P
primes : A.3.
Definite integrals : see " Integrals."
Deformation : of conies : Z.25.
of a cache-pot < N.81.
of a one-fold hyperboloid : E.30.
of surfaces : C.68,70 : G.16 : JP.22 :
L.60.
De Moivre's theorem: 756 : A.6,11.
Demonstrations, reduction to simplest
form: C.83.

INDEX.

865

Derivation :
applied to geo prs :
Determinants(continued) :
An.54.
function in analysis for a certain de
of analytical functions, gz : G.22.,.
terminant of n quantities : C.70.
of a curve : An.52.
gauche (a^ = a,p) : C.88,89. : CD.9 :
Derivatives : see " Differential coeffi
J.32,38,50,th55 : L.54.
cient."
involving t/l, Ac. : Q.l 5,16,1 7.
* Arbogasfs: 1536: CD.6 : 1.12: L.
of lower determinants : J.61.
82 : extMe.78 : P.61 : Pr.ll : Q.
of minors of given determinant : C.86.
4,7.
* minor : 554 : G.l.
Schwarzian : CP.13j.
* multiplication of: 562, 570 : A.14,59 :
Descriptive Geometry : An. 63: L.39 :
L.52.
N.52j,56.
number of terms in : LM.10.
Detached coefficients : 28.
partial: C.97.
Determinants : 554 : A.44,56,65 : apAn.
persymmetric : Me.82.
57 : J.22,tr51o,72,73,74,89 : C.86 :
with polynomial elements : Me.85.
CP.8: G.1,4,8,9,10 : L.84: LM.
of powers : AJ.4.
10: Me.62,78,79,,83 : N.51,69,ap
quadratic forms of : J.53,89 : L.56 :
70: Pr.8: Q.8: TE.28: Z.16.
N.52.
and algebraic " clefs " : 0.36.
ditto of negative dets. : J.37 : L.60 :
of alternate numbers : LM.ll.
M.22: Mo.62,75.
application to :
algebra and
of rational fractions : Me.82.
geometry, A.51,50,53.
resolution into quadratic factors of a
contact of circles and spheres : N.
det. formed from two circulants :
60.
Me.82.
cylindrical surfaces : A.58.
of the 16 lines joining tho vertices of
equations : Q.19.
two tetrahedrons : J.62.
geometry : J.403,492,77.
of sixth order: Me.84.
surfaces of revolution : A.582.
* signs of the terms : 557 : E.29 : Me.
arithmetical : G.23 : LM.10 : Me.78 :
80.
Pr.15.
skew:
Q.8,18.
of binomial coefficients : Z.24.
of squares of distances of points : Q.
catalogue of papers and treatises : Q.
11.
18.
Sylvester's det. and Euler's resultant:
of Cauchy ("aleph") : G.17a.
An.59.
combination of : CD.8.
symmetrical : G.l : J.82 : M.16 : th
combinatory analysis of : C.86.
Me. 85 : Q.14,18.
* composite : 555 : J.88,89.
and Lagrange's interpolation : LM.
* compound : 555 : AJ.6 : LM.14 : Me.
13.
82.
ap to a pr in geo : Z.20.
in conies : J.89,92.
of nth order and n1th power X
with continued fractions : J.69.
sq. of a similar determinant :
cubic: G.6: LM.13 : and higher, 11.
AJ.4.
cycle of equations : G.ll.
theorems and problems : AJ.3 : An.
of definite integrals : L.52 : Z.ll.
pr60: G.2,45,62,12,16 : J.pr66,pr
development of : An.58 : N.85 : in
84 : L.51,54 : M.13 : Me.79 : N.65 :
binomials, G.10: in polynomials,
Q.l,pr2,15 : Z.7,prsl8.
G.13,15: and ap to resultant of
transformation of: An.73: G.10,f 16 :
2 eqs, G.21.
of product, L.60.
division problems : A.59.
unimodular,
cn : Z.21.
double orthosymmetric : Z.26.
for verifying a system of d.e : C.23.
and duadic disynthemcs : AJ.2*
with a diagonal of zeros : Me.73.
elements of : G.10,15.
Developable cylinders, motion of : Man.
equation in which apq = avv: C.41.
84.
of even order, analogy between a class
Developable surfaces : A.69 : M.18 : Me.
of: J.52.
of figurate numbers : G.9.
77: Q.6.
circumscribing given surfaces : Z.13,
functional: CD.9: J.22,69,70,77,84 :
15.
M.1,18 : Me.80 : Q.l : of binary
circumscribing 2 quadrics: C.67,ths
forma, C.92 : of a system of func
54,gz63: CD.5.
tions, L.51.
5 s

866

INDEX.

Developable surfaces(continued) :
of a conical screw : A.69.
edge of regression : L.7'2.
of first 7 degrees : J.64.
through a given curve which develops
into a circular arc : L.56.
through a gauche curve : C.97.
mutual : J.19.
quintic: C.54 : CD.5 : G.3,.
of surfaces having principal lines of
curvature plane : C.36.
Development of tortuous curves : prs
Mem.
Diacaustic of a plane : N.75.
Diagonal scales : LM.6.
Diametral curves : CM.2.
of constant sectional area, prs : N.43.
Didon, proposition of : C.86.
Differences :
and differential-quo
tients : A.49,53 : N.69.
equations of mixed : JP.6 : N.85.
parameters of functions : C.95.
Differential Calculus: 14001868: J.
ll12)13jl143,15J,16 : Me.662 : Q.4.
reciprocal methods : CD.7,8.
Differential coefficients or differential
quotients or derivatives (see also
" Differentiation ") : 1402, 1422
46 : Pr.12.
of algebraic functions: M61.1.
of logic and a" : A.l ; of x and a' : N.
53.
of {/(*)}": M.3; of ycxp(z'): A.22.
calculated from differentials: AJ.16.
* of a composite function : 1420 : nth,
G.13.
* of exponentials and logarithms : 1422
7: A.11: N.60,52,85.
* of a function of a function : 1415 : A.9:
in terms of derivative of inverse
function, Mem.57.
of a function of two independent
variables : 1815.
of irrational functions : P.16.
of products whose factors are con
secutive terms of a series : Me.31.
ratio to the function at the limit oo :
J.74.
successive or of nth order : 1405, 1460
72, 285291 : A.l,4,7 : An.57 :
G.18: M.4: Z.3.
*
of a sum, product.or quotient: 1411.
independent repres. of: M.4.
*
of a function of a function : G.13 :
n 4, 1419.
of functions of several variables :
C.93.
of a logarithmic function : A.8.
of a product : 1460, 1472.
and summation symbols : J.33,ths32.

Differential coefficients of nth order

(continued) :
*
of (as+*')": 2860;
, 1467.

*
*
*
*
*
*
*

(1*)"- (Jacobil:
-/(a?fc2*5), A.3.

1471,

A.4;

(a + bx+cx")", 2858;
(x"-+ax+b)-m, A.8.
tan*: A.12 ; cos"'*, A.9; cat,
1461, N.62.
Sln *3, 2862 ; sin"1*, 28545 ;
COS '
>
'
tan"1*, 1468, apN.9.
e", e"y : 14634 ; e"xm, CM.2.
e"*: 2861, A.30; e"cosbx, 1465.
it"-1 logs : 1466; e*c**cos (* sin a),

of nth order with * = 0 in the result

tan-1*, sin-1*, (sin"1*)3, 28659 ;
003 msin"1*, ^nicos-1*, 28717;
' sin
sin
*
(l+*a)^smm tan-1*, 2883-7;

*
#

x'e"eosbx, 288991 ;
e*1
sin (x or y) + cos * ^ 3
l+2y cos *+i/3
Differential equations (D.E.) : p.460,
3150 3637: A.1,52,67: AJ.4:
An.50a: C.8,15,23,29 ,42,54,70,83 :
CM.3: E.9: J.l,36,58,64,66,74,75,
76,78,86,91 : L.38,52,56: LM.4.10:
M.8,12,25 : Man.79 : Me.81 : Mem.
30 : Mo.84 : N.72,80 : Pr.7 : TI.13 :
Z.4,16,27.
Abel's theorem : J.90.
algebraic : An.79 : C.86.
with algebraic integrals : J.84j.
approximate solution : C.5.
by equations of differences : L.37.
*
by Taylor's theorem : 3289.
of astronomy: C.9,29 : P.4.
asymptotic methods : C.94 : Q.5.
Bes8el's numerical solution : Z.25.
* Complete primitive : 31636 : J.25.
no. of constants : CP.9.
with complex variables : Mo.85.
of a conic : E.38.
continuous and discontinuous integ
rals of : C.29.
for a conical pendulum : A.84.
relation between its constants and
the constants of a particular
solution : C.92.

INDEX.

867

D. E.(continued) :
D. E. (continued) :
of curves having the same polar Bur- * Parseval's theorem : 3628.
and p.d.e of first order : J.23.
face : An.76.
particular integrals of: CM.2 : alge
* depression of order by unity : 32629.
braic, C.86.
with different, total integrals : L.84.
of dynamics: C.5,26,402: CD.2 : G.l,
relations of the constants : C.93 :
4: L.37,49,52,553,72,74 : M.2,17,
J.10: JP.6.
25 : Mel.4 : Pr.12, : P.54,55,63.
in problem of n bodies : An.83.
ap. to engineering : JP.4.
of perturbation theory : Mem.83.
and elliptic functions : L.49.
with quadratic integrals : J.99.
for roots of algebraic equations : P.
elliptic: G.19: M.21.
elliptic multiplier : M.21.
64: Pr.13.
* exact: 3187,3270-5: G.12: C.1,10,11. * rule for equivalence of two solutions
of families of surfaces : Me.77.
3167.
with fractional indices : JP.15.
* singular solution of : 316978, 3301
6, 3401-3 : C.19,94 : CM.2 : JP.
of functions of elliptic cylinders : M.
22.
18 : M.22 : Man.83,84 : Q. 12,14.
general methods : L.81.
of sources : A J.7b.
of a surface : G.2.
* generation of : 3150.
satisfied by the series l2g + 2o4
geometrical meaning of : Q.14.
22+&c. . 2y2 + 2t/(/+2y24s+
* homogeneous: 3186, 3234, 32628:
C.13: CM.4: J.86.
&c. : L.49 : J.36.
satisfying Gauss's function F(aJ3,y,x):
hyperelliptic : J.32,55: Mo.62.
of hypergeometrieal series : J.56,57j,
L.82.
73.
synectic integrals of : C.40.
integrability of : Z.12 : immediate,
and tortuous curves : L.53.
transformation of : An.52 : CD.9 : in
C.82.
whose integrals satisfy relations of
curvilinear coords : J.85.
the form F [<as] = fx Fx : C.93.
D. E. linear : A.28,35,40,41 ,43,45,46,53,
59,65,69 : Ac.3 : AJ.7 : An.50,85 :
whose integrals are satisfied by the
periodicity modulus of elliptic
At. 75 : C.7S,293,58,73,84,88,903,913,
92,,94: CD.3,42,9: CP.9,10 : G.
integrals of the first kind : J.83.
integrating factors : pp468 471,
15 : J.23,24,25,40,42,55,63,70,76,
79,80,81,83,87,88,91,98 : L.38,64:
3394: C.68j,97.
M.5,11,12: Me.75: P.48,50,51 :
ofPdx+Qdy+Rdz: Q.2.
Pr.55,iaJ;193,20 : Q.8: Z.3,7,9.
integration : by Bessel's function :
without absolute term, condition of
Me.80.
solutions in common : C.95.
by Gamma function : TE.20.
with algebraic integrals : C.96.97 : J.
*
by definite integrals: 361728:
C.17 : J.74.
80,90: M.21.
determination of arbitrary constant :
by differentials of any index : C.17 :
L.44.
At.65: Q.192.
argument & parameter interchanged
by elimination : CP.9.
in the integral : J.78.
by elliptic functions : An.79,82 : C.
41 : JP.21.
bibliography of : AJ.7.
with coefficients that are algebraic
by separation of operative symbols :
functions of an independent varia
Z.15.
*
by series: 360416: C.10,94: LM.
ble : C.92,94.
* with constant coefficients : 323850 :
6: Me.79: Q.19: TL7.
An.64: CM.1:,2: E.34 : JP.33 :
by theta-functions : C.90.
irreducibility of: J.92.
L.42: N.47,84.
with periodic coefficients : C.91,92 :
isoperimeters, pr : H<5m.50.
doubly periodic : C.90.,,92,982 : J.
of Lame : J.89.
of light: M.l.
90.
with rational coefficients, algebraic
in linear geometry : M.5.
integrals of : C.96 : JP.32,34.
of motion : C.55 : of elastic solids,
with rational coefficients, upon
Q.13 : of fluids, UP.7 : of a point,
whose solution depends the quad
C.26.
rature of an irrational algebraic
with integrals " monochrome and
product : C.913, 92s.
monogene " : C.40.

8(38

INDEX.

D. E. linear(continued) .with variable coefficients : C.92 :

J.66,68,76 : L.80,81.
which connect a complete function of
the 1st kind with the modulus :
C.86.
homogeneous: Ac.l : J.90 : Mo.82.
integrating factors of : C.97,98.
integration by Abelian functions, C.
92 : J.73 ; by finite differences,
Q.l ; by series, 3.76.
which admit of integrals whose loga
rithmic differentials are doubly
periodic functions : L.78.
whose particular integrals are the
products of those of two given
linear d.e : A.41.
irreducibility of: J.76.
Landen's substitution, geo : J.91.
Malmsten's theorem : J.40.
singular solution : J.73,83,84.
transformation of: C.91,96.
* n variables, 1st order : 332032 : C.
14,15 : G.13 : J.20,80 : L.38.
n variables, 2nd order : L.37 : 2 varia
bles, C.70.
n variables, any order : Me'm.13.
* Pdx + Qdy+Rdz = 0:P,Q,Rii\vol\mg
x, y, z, 3320 ; geoM.16 ; Z.20 : P,
Q, R, integral functions of x onlv,
Q.19,: P = (ax" + bS-, + &c.)~ ,
Q, R similarly with y and z, Q.202.
* Xdx + Ydy+Zdz+Tdt = 0 : condition
of being an exact differential,
3330.
zdt+ydx+zdy + tdz = 0: A.30.
*D. E. of first order, linear: p467 : C.86 :
G.13: algG.18; M.23.
* exact: 3187.
* homogeneous : 3186.
integration by a particular integral :
C.86.
reduction to a continued fraction of a
fraction which satisfies a: C.98.
* separation of variables : 3185: GM.1.
* Mdx+Ndy = 0 : 3184 : N.74,77.
* (ax+by+c) dx+(a'x+b'y+c')dy = 0:
3205,p471: L.59.
(cut + by+c)"dx + (a'x + b'y + c')"dy : A.
64.
~Tp + ^Q = ()> P,Q being quartics in
*,y. C.66: LM.8 : ME.79.
f(x)dx . f(y)dy _ 0 / (at) of 1st deg..
F(x) T F(y)
' F\r) of 5th deg.'
C.92.

D.E. of first order, linear(continued) :

* Pldx+Ptdy+Q(xdy-ydx) = 0; P
Ps being homogeneous and of the
pth deg. in z, y ; Q homogeneous
and of the gth deg. : 3212.
P P., Q different linear functions
of x, y : C.78,83 : L.45 : J.24.
* yx+Py = Q. where P, Q involve x
only: 3210.
* y,+Py = Qy : 3211.
yy +Py+Q-. Mem.ll.
y'+y- = (P+2fr+R*-)f Wbere F'
Q, R are functions of x : Mem.ll.
y,+a+by+y* = 0: J.25.
y,</(m+x) = \y\/mx ; A.42.
V'=f(y) J.9; yi=f(x,y) : An.73:
L.55.
C.88.
y* = '- x (8y-l)
V'+f(x) sin y+F(x) cos y+<f>(*) = 0 :
L.46.
*
= ex' (Eiccati's eq.) : 3214:
A.40: C.11,85: (tn=-6)E.7:
ext28 : JP.14 : L.41 : P.81 : Q.7,
11,16.
allied eqs : L.51 : Me.78 : Q.12.
sol. by continued fractions: Mem.18.
by definite integrals : J.12.
transformation of : Me.83.
*D. E. of first order : 322136 : A.29 :
C.40,45,66 : M.3.
two variables : An.76 : J.40: Mem.
62 : N.50 : singular solution, J.38.
* Clairaut's equation, y = px+f(p) :
3230: CM.32: Me.77.
integration by second order d.e: A.46.
homogeneous in * and y : 3234.
reduction to alinear form with respect
to the derivatives of an unknown
function: C.87.
related transcendents : Ac.3.
separation of variables : CD.9.
* singular solution : 3230 : A.56,58 :
CP.9 : J.48 : Me.73,77.
* solution by differentiation : 3236.
* solution by factors : 3222.
transf. by elliptic coords : J.65.
verified by a reciprocal relation be
tween two systems of values of
variables : C.15.
dz2+dy3 = ds2 and analogous eqs :
L.73.
* adx+bdy ds : 3287.
dz*+dy*+dz* = dsS: L.48.

INDEX.

D. E. of first order(continued) :
dz*+dy2+dz2 = \(<jUP+dfi2+dy2): L.
50.
F(u,u,) = 0: C.93.
* x<t>(p)+y+(p)=x(p): 3226.
D. E. of second order, linear : A.29,
32.55,64 : An.68,79,82a : C.822,90,
91,9397: J.51,74: L.36 : Me.l4 :
M.ll : Mo.64 : Z.5.
with algebraic integrals : C.90 : J.81,
85 : L.76.
with doubly periodic coefficients :

*
*
*

*
*

homogeneons : M.22.
integration by Gauss's series : Z.19.
transformation of : An.52.
y" = a : 3288.
y" = Py : C.9.
y"+a*y = Q : 3522,'5 : geoMe.66 :
Q = cos we, 3526: Q = 0,35234.
y" = Ay (a+2bx+cx)-* L.44.
y" = ax+by: 3281.
xy" = y : Z.2.
(1T*2) y" my = 0, Ac. : CM.3J.
x*(y"+q*)=p(p-l)y. CM.2.
y" = {h+n (n+1) k*$n*x} y (Lamp's
eq.): C.85.
y" = ^>(l+y(j)1+y^+&c., when <p &c.
are trigonometrical series : C.98.
y" = y (e*+e-*)-s: L.46.
y" + a/xmy =/(m) : E.6.
y" = ay-\-ty (*) : A.45.
y"=f(y)- 3257.
y" =f(x, y) (Jacobi) : 3285.
y" + py'+ry = 0 : G.85,90 : Q.19.
y" = xy'nxy : A.53.
xy"+my'+nxy = 0 : L.45,78.
xy"+y'+Axmy = 0 : C.39.
x*y"+rxy' = (bx"+s) y : An.51 : CD.
5.
z*y"+2xty'+f(y) = 0 : A.28,30.
y"+f(x) y'+F(y) y'2 = 0: 3284: L.42.
(a+bx)y" + (c+dx)y'+{e+fx)y = 0:
A.58.
(a+bx")x^y"+ (c+exn) xy'+ (f+gxn)
u == Q (Pfaff) : 3598 : J.2,54 : and
like eqs., Z.2,3 : with 6 = 0, A.38.
x* (abx) y" 2x (2a bx) y'
+2 (3a bx) y = 6a? : A.28,30.
xy"+ y'+y (x+A) = 0 : Me.81,84.
*y"+y'+y (-A) = d,
Me.
82.
x*y" 2xy*+2y = x*yf-a~ : A.28,30.

869

D. E. of 2nd order, linear(continued) :

sy" + (r+qx)y' + (p-\-nx + mx~) y = 0 :
A.23: Z.8,9.
* (1x) y" xy+qh) = 0 : 3282.
* (1 +ax2)y"+axy'qh) = 0: 3283,3594.
2x(lx^)y"y'+n(n+l)y = 0: Q.18.
x (1-x) y"+ $-ix) y'+Jv y = 0 : Me.
82: Q.17.
(m+a:) (n+x)y"+(vin) y'
X2 (m+xfy = 0 : A.42.
(mx2+nx+p) y"+ (qx+r) y'+sy = 0 :
JP.13: Z.4.
\py"+A\y'+Bfiy = 0, )iy"+A\y'
+B\ny = 0, and ny" +A\y' +B/i
= 0; with \ = a + bx+cx% and
li = b+2cx : A.423.
dx {(x-rf) y,}-xy = 0 : L.54.
* y"+Py*+Qy+B = 0, P, Q, B being
functions of x : 3280.
Py"+Qy'+By = 0-. Ac.l.
zy"+az,yK+f(y) = 0: Me.71.
D. E. of second order : Ac.l : An.79 :
JP.29 : C.67,69,80,91 : J.90: L.39:
LM.11,12,13,16 : Z.15.
with algebraic integrals : C.82.
derived from linear eq : Me. 73.
with elliptic function coefficients :
Ac.3j.
in the neighbourhood ofcritical points:
C.87.
polynomials which verify : Ac.6.
solution by definite integrals : A.27.
by factors : C.68.
by Challis's method, and application
to Calc. of Variations : A.65,66.
yy" = \y'-+2py1: L.73.
Myy"+Ny'*=f(x): N.79.
* y"+Py'+Qy'i=0, P,Q functions of
x . 3276.
* y"+Py' + Qy" = 0 : 3278.
* y"+Py'2+Qy'" = 0: 3279.
* y" + Qy'2+B = 0 : 3277.
I*+I= 0, where I is Bessel's
x
function : J.56.
D. E. of third order, linear: C.881 : Q.
7,8,14: M.24.
y' = y'"\ JP.15: ut = UiI: C.3.
x3y"'y = 0: Z.8.
y"'=. 3mx'2y"+6m(li+2)xy'+3m(n+2)
(?+\)y. A.42.
y'" = xm (AxY'+B-cy'+Cy) : A.58.
D. E. third order: An.832: C.98,: M.23.
*D. E. of higher order, linear : 323750 :
A.65 : C.97- : J.16 : M.4 : Q.18.

870

INDEX.

D. E. of higher order, linear(cont.) ;

of orders p and m+p, th : C.43.
yu = *y,y A.l.
* 2/=/(x): 3256.
V = ie"y : L.39.
y = (a+fr) y : J.10.
^"Vrnx =
: A.32.
s*"yi.=y: A.42.
xijn, = 1/ : A.26.
a:*"i/=j/: by definite integrals:
C.482,49: J.57.
ynr =
: by definite integrals : J.19.
*""iy{tm + i)z = y : by Bessel's func
tion : M.2.
?/ = *"'y + A + Bx + Cx*+ . . . +Nx" :
Z.10.
i/ = zlj;mi/,+Bz"'-1y : A.28,38.
x^y^ =
+By : A.33.
y xy = ijn_r+abxny : by definite in
tegrals : J. 17.
y = Axryit+Bvyt+Cy: A.53: M.3.
Axv,yu+Bxm-,y,+ Cxm-'y : A.
29,30,33,38.
2yx+ay{n-i)x = bxy : J.2 : Z.10.
*2/h^(i,~i)x-r% = 0: A.40.
2/x+X!/;n-i)* = * (y+Ay) : A.86.
AzyM+Byi.-i),=:xm(Aiey,+By) : Z.8.
xhfmM+qmxh/ = p (jj1)
: J.2.
= * (tw^jx+t^x-r Cy) : A.38.
* y+oiy(.-i).+...+.y = 0 : 3239: A.
40.
* ditto =/(): 3243 , 3516: o, ... a
functions of at, 3237 : J.39.
n fractional and all lower orders
integral : L.36.
(d.+afy =/(): CM.4.
* ditto = e* : 3528 : ditto = sin mx :
3529.
* (p+qx)ny,i*+ai(p+qx)"-*y:n-r*+...
+any =/(*) : 3250 : with p = 0,
C.96.
^(.)x+... + (a+a-) (/ + ...
+ ay = 0 : A.47.
zn-i(a+bx)yx+x"-'2(e+d)y[-])r+...
+ty = 0: J.39.
yu+myI+ny!L+py = q: L.44.
I
2
*D. E. of higher order : 325169.
* y = F(y(K.1).): 3258.
* 2/nx = ^(2/(-2,x): 3260.
Pyx+Q = Q, where P, Q are func
tions of it, y, and the first n 1
derivatives of y : J.31.

D.E., simultaneous system of: A.n.69,

82,84: C.10,432,47,923 : CM.l: LM.
14: Me.13,80: Pr.12.
Harailtonian : Q.14.
integration and inversion of the in
tegrals : C.23.
Jacobi's: CD.3.
* method of multipliers : 3353.
number of arbitrary constants: Me.
77.
* reduction of order : 3350.
reduction to a P.D. eq : C.44.
theorem of Abel : C.24.
theorem analogous to Lagrange's
in the Perturbation theory : L.
51.
theorem of a new multiplier : J.27.
transformation and integration of :
L.45.
* xu = ax + by and yv = cx+dy : 3354
and a similar example.
D.E., simultaneous linear: 334059:
AJ.4: C.9,92: E.5: N.66,84.
Pfaff's method : C.14 : J.2 :
transformation of : J.98.
* $ = = |:3346: Q.14.
*
= _#_ dz - 3347
PixP PtyP P3zP'
* x,+P(ax+by) = Q and y,+P(ex+dy)
= B : 3348.
* tx,+2(xy) = t and ty,+(x+5y) = P:
3349.
* equations in x, x?t, x, &c. . y, yu, yit,
&c. : 3357.
* homogeneous in sr, y, z...and their
second derivatives only : 3358.
D.E., simultaneous first order : 3340
49: C.43: J.48: Pr.62.
D.E., symbolic methods: 34703636:
CD.l: P.61: Q.3,17j.
* F(d.)u = Q: 3515.
* tijx+a^it = Q and similar : 3522.
* exceptional case of the inverse pro
cess : 3526.
* reduction of an integral of the nth
order : 3530.
* azym,+ bx"yKr+&c.= Q: 3531.
* aym,+byn,+&c.=/I(e',sin0,cos0):3o35.
* air1mz + kri"z+&c. = Ui+Hl+Ac :
3540.
* F(ir)u = Q: 3541.
* Eeduction to the form (IT+.l.n"-1-!. . . +An)u = Q, where n = Mdx+
Ndy+&c. : 3546.

INDEX.

D.B., symbolic methods(continued) :

* F[?d yd,, ...) = SAxmyn... : 3558.
* to transform (a + bt + . . . ) u+("' +
+ ...)(_!),+ + . ..&c. = Q intothe
symbolic form : and the con
verse : 3571, 3573.
* u+a,0 (D) CU+&C. = U: 3575.
* to transform u+<p (D) er'u = U :
3579-30.
* to reduce a homog. eq. to the form
ynJr+qy = X: 3585.
Differential expressions : 1407 : prAn.
85.
algebraic : An.79 : M.9, by homog.
coords.
transf. of: J.85 : Q.I62 : Mo.69.
Differential :
formula), theory of : L.
52.
functions, theory of : C.60.
expressions, linear : J.85.
parameters of functions : C.66,78.
quadratic forms : An.84 : transfor
mation of, A.16.
* resolvents of alg. eqs : 36317 : An.
83 : C.91 : LM.1,9,14 : M.geo4,18 :
Me.75,82: Man.65,84 : Q.6,11.
*
of y ny"-r+(nl)x = 0, &C. :
36336: Man.65.
of y+by'+at = 0: Q.17.
of 12 {y*+ay*+xy) a? : Me.82.
Differentiants in terms of differences of
roots of parent quantities : AJ.l.
Differentiation : (see also ' Differential
coefficients ') 140282 : CM.l.
* formulas: 141172: An.59 : CD.2 :
CM.l : Pr.9.
by the method of " Hates " : Me.75.
general, i.e., with any index fractional
or imaginary : An.58 : AJ.3 : CD.
32,4,5 : J.12 : JP.13 : L.55 : N.84 :
Q.3,4: TE.14,15: Z.16 : change
of independent variable, JP.15.
* under the sign \ : 2253 : A.17.
successive : A.20 : Q.12.
when the function becomes infinite :
C.88.
Digits :
calculus of : Sbonimsky's
th: J.30.
frequency of in numbers : AJ.4.
origin of: L.39.
Diophantine analysis : see " Partition
of numbers."
Di-polar geometry : Z.27.
Direction ratios and cosines : 5511 14.
Directive algebra : N.68.

871

Directrix:
of a conic: 1160: gn.eq
A.25: E.36: Me.80 : t.cQ.13.
of in-parabola of a triangle : Me.80.
of a curve : A.20 : J.2.
of a parabola : gn.eqE.29.
of a quadric : N.74,75.
Discontinuity :
in curves : CM.4 :
Z.26.
in fractions : Man.48.
in maxima and minima: CD.3.
Discontinuous functions : A.7 : C153,
28: G.19: J.7,10: LM.6 : Man.
48: TI.21.
Discriminant : 1627, 16389, 1644 : Ac.
1: J.90: LM.2: M.12 : N.59 :
Pr.14: Q.10,11.
of an algebraic d.e of 1st order in 4
variables, and of its complete
primitive : An.84.
of alg. eqs., resolution into factors :
M.24.
of alg. functions : J.91.
applied to conies and quadrics : A.
58.
of binary quantic : An.56.
of a binary sextic : An.68.
of a quartic : N.83.
of a ternary quadratic form : Me.68.
Discriminating cubic : 1849, 5693 : G.
16 : J.26,71.
* proof of real roots : 1850 : A.29.
Displacement :
theory of : N.82.
of plane figures : C.80 : N.73.
of an invariable dihedron : Me.85 :
infinitesimal ditto, C.84.
of an invariable figure : C.51,52,,66,
92,: JP.26: L.74,75.
of a figure, two of whose points slide
on two curves : C.82.
of a solid : L.40 : determination of
tho normals to the lines or sur
faces described : C.62.
of a system of points : C.78.
virtual displacement : J.ll.
infinitesimal : of an alg. surface : 0.
70.
of bodies only defined by 4 coordi
nates : C.73.
of a parallelogram : C.97.
Distance :
between 2 points : 40345,
t.c4601 : Q.7 : sd5508, 5510: some
relations, G.9.
correspondences for quadric surfaces :
LM.16.
of a point from a line and from a
plane : A.57.
* of a point from a plane in a given di
rection : 5559.
relations : Z.27.
Distributive law : 1488.

872

INDEX.

Divisibility : C.96 : E.8 : M.39 : N.67,74.

of decadic numbers : Z.22.
of numbers of the form 2in + 1 : Mel.5j.
of a quotient by the powers of a fac
torial : C.94.
of (*+y)"+(*)"+( y)" : Me.79.
Division : prJ.47 : prL.56.
* abridged: 28: arithj.31 : N.452,46j,
52,54,57,81 : algN.42.
by 73 or47; rule for remainder: E.22.
* effected by determinants : 581.
Fourier's rule : K.52.
of planes and spaces : J. 1,2.
of a rectilineal figure and of a spheri
cal polygon : J.10.
* of a right line into equal parts : 950.
and transformation of plane figures :
A.+j.
of trapeziums, pyramids, and spheres :
All.
of triangles: A.11,17.
Divisors :
of an integer, number of :
374: C.96: N.68.
ofintegral rational functions : Mem.57.
* Newton's method of : 459.
of a polynomial with commensurable
coefficients : N.75.
rational, of 2nd and 3rd degree : N.45.
* sum of : 377 : Ac.f4: L.56,57.
sum of powers of: L.58.
of a+Ay* : L.49.
Double algebra : LM.15.
Double function, laws of change of
higher order : A.21.
Double points : 5178 : n-tic with
$n{n3), C.60; with I (n 1)
(u 2),M.2; Clebscli's thsof these
qualities, C.84: n-tic with t(i 1)
(m2)2 double points, L.80.
of plane curves in cubic space : Z.28.
in a locus defined by alg. conditions :
C.88.
of a pencil of curves : An.64.
of plane curves in cubic space : Z.28.
in the projected intersection of 2 quadrics : N.84.
of tortuous curves : M.3.
Double relations : A.60.
Double tangents : An.51 : J.49 : P.59 :
Q.4: Z.21.
of a Cartesian : E.30.
of an n-tic : M.7 : number = in (n2)
(n*-9), J.40,63; N.53.
of a quartic : C.37 : J.49,55,68,72 : M.
1: N.67: P.61: Pr.ll : with a
double point, M.4,6 : reciprocity
of 28 double tangents,
to the surface of centres of a quadric :
C.78.
Drilling, shape of hole : Pr.35.

Dual relation between figures in space :

J. 10.
Duplication of the cube, approx. : Pr.20.
*e (see also " Expansion ") : 151 : N.67,
68: : geo meaning E.4 : N.55.
combinatorial definition of : A.lj.
incommensurable: 295 : CM.2 : L.40 :
Me.74.
and it, numerical th : Q.15.
e ' in fractions : L.80-.
e **, &c. : CP.6.
e^e4*: E.37.
e** : 766 : AJ.7 : in transformations :
CM.4.
(J"*": A.33.
Edge of regression : 5729: tg.eC.71.
radius of curvature of : 5742.
Eisenstein's theorem : G.16.
Elastic curve : C.18,19: JP.34.
Elementary calculation : N.45.
Eliminants : 583, 1626.
and associated roots : LM.16.
of two cubics : J.64.
degree of: G.ll, two eqs 12 : J.22,31.
Elimination: 58294: A.2, : Ac.6,7:
C.12,87,90: CD.3,,6: CM.3: G.
15,17 : J.34,43,60 : JP.4 : L.41,44 :
LM.11: M.5,11: N.42,45,,46,80,
82,83-: Q.7,thl2: Z.23.
problems: C.84,97: J.58 : M.12 : Q.
8,11 : in metrical geo, A.63.
Elimination of x between two equations:
58694: C.12: CM.2: J.15,27:
Mo.81: N.43i,76j,77.
by Bezout's method : 586 : A.79 : J.
53: Me.64: N.74,79.
bv cross multiplication : CM.l.
by the dialytic method : 587 : N.79.
in geodetic operations : Z.3.
by h.cf : 593 : JP.8.
by indeterminate multipliers : CM.l.
by symmetrical functions : 588.
degree of the final equation: J.27:
L.41.
Elimination :
ap to alg. curves : M.4.
ap to in- and circum-conics of a poly
gon : At.63.
calculation of Sturm's functions : C.
80.
of functions : 3153 : C.84,87 : Me-73,
76.
with linear equations : At.53.
with linear differentials : L.36.
with n variables : CP.5.
resultants, comparison of : J.57 : and
interpolation, J.57.
transformation and canonical forms :
CD.6.

INDEX.

873

Ellipse (see also " Conies ") : 1160,

Elliptic functions(continued) :
4250: cnl245: geoQ.9.
of first and second kind : C.10.
theorems: A.30,47,prs49 : N.76 : Mc
as functions of their amplitude : JP.
Cullagh's, N.72.
14.
eq. r+r' = 2a: A.2.
representation in a simple form :
equal chords : tg.eE.22.
Z.21.
of maximum surface : N.65.
series by which they are computed :
* as the projection of a circle : 4921 :
J.16,17.
N.75s.
of second kind : J.93.
* rectification of : 608396 : A.3,22,27,
mechanical representation : Me.75.
30: At.39: graphic: Me"l.l : N.
reduction to first kind from same
43: TE.4 : Z.6: when e is very
modulus : A.56.
small, TI.9.
of second and third kind, expression
representation by a circle : An. 70.
by 6 function : Z.10.
* quadrature of : 6108,6113; t.c4688:
of third kind: C.94,96 : CD.8: J.14,
A.46 : of sector 6098, A.20 : of
47: LM.13.
segment, 6103.
addition of: A.47,geo64 : AJ.7 : C.59,
and triangle : thQ.4.
78: J.35,41,44,54,8g2,90: LM.13:
Ellipse and hyperbola : 1160, 4250 : A.
M.17: Me.80,84: Q.18: Z.l.
24,28.
of 1st kind, Z.26 : 3rd kind, Me.81.
theorems : A.23 : CM.3 : N.85.
2nd kind by q series : Me.83.
sectors : TE.14.
application of : 0.857,865,892,90,93,948.
Ellipsoid: 5600,6132: A.28 : thCD.2 :
to algebra : J.7.
prC.20 : L.38 : cnM.20 : P.9.
to arithmetic : C.98 : L.622.
centro-surface : CP.12 : LM.3.
to confocal conies : Z.7S.
cubature of some derived surfaces :
to geometry : G.12 : J.38,53.
A.12.
to in- and circum-circles of a poly
* and enveloping cone : 566472 : Q.6.
gon : L.45.
generation of (Jacobi) : CD.3.
to rectification : L.45.
* of gyration : 5930 : of inertia : 5925
to spherical conies : Z.22.
39.
to spherical curves and quadrature:
and plane of constant segment, th :
An.50.
E.32.
to spherical polygons with in- and
* of revolution : 5604 : area, N.42.
circum-circles : L.46.
a locus in space : Q.16,172.
to spherical trigonometry : Q.20.
* quadrature of: 6143: J.17: Z.l : of * approximation to : 212732 : P.60,62.
zone, A.22.
arithmetical consequences : Ac.5.
* volume : 6144,8 : A.46.
arithmetico-geometric mean : J.58,85,
Ellipsoidal geometry : A.10: LM.4.
89.
surfaces : G.17.
arg sn a and (arg sn a)-, as def. inte
Elliptic functions : 2125: A.l,122,16,21,
grals : Q.19.
35,48: trAc.5,62,ths72 : An.61,84:
in complex regions : Z.282.
C.46,506,90,96,97 : CD.2,32,5 : CM.
development of : 212732 : J.81 : 1st
32 : E.23 : G.l4 J.2,34,44,6,8,16,26,
and 2nd kind, C.92 : with respect
27,302,32,35,37,39,46,72,832 : JP.
to the modulus of X (x), y. (z) and
25 : L.55,56,61 : LM.7,10,29 : M.
their powers, C.86.
3,ll,12,pr25 : Me.79,80,81,82, :
development of an imaginary period
Mo.81,823,832,85 : N.773,784,792 :
when the modulus is small
P.31,34,75,78 : Pr.6,9,10,12,23, :
enough : An.70.
Q.11,17,19 : Z.2ll,27.
differentiation by periods and invari
of first kind : A.12,21 : C.16 : J.93 :
ants : J.92.
L.43.
discriminant of modular equations :
with complementary moduli exten
M.8,9.
sion of a theorem of Lagrange :
An 833.
double substitution : J.15.
normal forms of 3rd and 5th de
A lr am - cos * am sin
gree : M.172.
replaced by one of second kind : J.55.
represented by gauchebiquadratics :
x4*A2am * dx: J.37.
C.83.
Jo
T
5 T

874

INDEX.

Elliptic functions(continued) :
eqs. for the division of : Mo.75.
formulae: AJ.5 : J.15,36,50: LM.13 :
Me.78,80,85 : Jacobi Mt51.1 : Q.16,
19: from confocal conics,LM.14 ;
diflerential,Mo.82 ; for sn, cn, dn,
of u+u + ro, Me.82.
Galois' resolvent : M.18.
goo. problems : M.19.
geo. properties : L.43,45 : P.52,54.
geo. representation : A.22,61 : An.60,
61 : At.53 : C.19,213 : J.63 : L.44,
78 : in solid geo, M.9 : of 1st kind,
An.53: C.70: CD.l : JP.28 : L.
43,45^46,78 : of 3rd kind, A.24.
identities : Me.77.
imaginary periods : AJ.6.
infinite double products, A.14 : with
elliptic functions as quotients,
J.35.
inversion of : J.4 : JP.34 : L.69.
KK+KEKK'=\n: Me.75.
of iff: Me.85.
modular equations : C.47a : J.58 :
LM.9,10: M.12.
modular functions : A.11, 13: G.12 :
J.72,83: L.402: M.17,18: differen
tiation for modulus of am, LM.
13 : expansion in powers of modu
lus, J.41 : formulae, L.64: relation
between the modulus and the
invariant of a binary quartic, Z.18.
multiplication of : C.88, : J.14,39,41,
74,76,81,86,88,: M.8: Mo.57,83,:
and division, Z.7., : formula?, trA.
36 : C.59 : J.39,48 : complex, M.
25 : Mo.62 : Q.19,20 : mod.
J.48.
periodicity moduli of hyper-elliptic
integrals as functions of a para
meter: J.71,91.
subsidiary, pm (u, k) : LM.15.
products of powers: Mem. 71.
quadriqnadric curve : M.25.
g-formula for sin am : LM.ll.
5-series : and f853: for
+ coeffs.:
Q.21.
reduction of: An.64: in canonical
forms, J.53.
relations : A.67 : J.56 : between E'
(A;) and J" (A;): J.39.
representation by power series: J.54.
representation of quantities by sin
am (u + w, k) : J.45.
series : C.95.
sn 8u, cn 8u, dn 8u in terms of sn u,
tables: Pr.33.
sn.cn, and dn of u+v+w : LM.13.

Elliptic functions(continued) :
spherical triangle of : Q.19.
and spherical trigonometry : Q.17.
substitution of 1st order: J.34.
and theory of numbers : L.58.
transformation : An.57,-,,58,60 : Ac.3 :
C.49,f79,f80,82 : CD.3,5: J.3,34,
35,f55,5o,87,88,89 : LM.9,11 : M.
14,193,22: Me.83,tr84: Pr.27 : Q.
13,20.
of 1st kind : A.33.
of 1st and 2nd kind as functions of
the mod : h.4>\.
of 3rd order : J.60 : Me.83.
of 7th order, square of mod : J.12 :
LM13.
of 11th order : At.5.
of the orders 11, 13, 17, and 19: J.
12,16.
cubic: C.64: Q.13.
and division : J.76 : M.25.
of a double integral, &c. : Me.75.
Hermite's ; tables : J.72.
Jacobi's: LM.153,16: J.87.
linear : J.91.
modular, of Abel : ap to geom : C.
58 : to conies, C.79.
modulus of in a function of the
quotient of the two periods : An.
70.
pertaining to an even number : J.
14.
quartic : Q.12.
by roots of unity : J.6.
of rectangular coordinates : LM.15.
and of functions in theory of Cate
nary : A.2.
triple division of and ap. to inflex. of
cubics : A. 70.
Weierstrass's method : AJ.6.
Emanents : 1654.
Empirical formulae, calculation of : Me.
73.
Engrenages : L.39,40^
" Ensembles," theory of : Ac.2.
Envelope : 5192 : A.24,prs56 : C.45,86.
p.d.eCM.4: G.ll: M.84: Me.64,72: N.
44,59,68,74.
application to perspective : A.9.
class of (Chasles), th : C.85.
from ellipse and circle : LM.15.
* of a carried curve : 5239.
of conies, theorems : N.45.
of chords of a conic: N.48: subtending
a constant angle at the focus,
CM.3.
of chord of a closed curve : E.28 : cut
ting of a constant area, E.31.
of curves in space : L.83.
* of a curve with n parameters : 5194.

INDEX.

Envelope (continued) :
of directrix of a parabola : E.34.
of geodesies : M. 14,20.
imaginary, of the conjugates of a plane
curve : C.75.
of pedal line of a triangle : Q.10 : do.
of in-triangle of a circle, Q.8,9.
of perpendiculars at extremities of
diameter of an ellipse : N.46.
of a plane : C.35.
of planes which cut a quartic gauche
curve of the 2 in 4 points of a
circle : An.71.
of planes perpendicular to radiants of
an ellipsoid at the surface : An.
59 : Pr.9.
of plane curves : G.11,12 : singulari
ties of, LM.2j.
of polars of a curve : J.58.
of a quadric : Q.ll.
of a right line : N.63,79,83 : Q.13.
cutting two circles harmonically :
N.85.
sliding on two rectangular axes :
N.45.
of a Simson line : E.29,34.
of a sphere : C.67 : J.33 : touching 3
spheres, N.60.
of a surface : CM.l : M.5 : degree of,
N.60.
of a surface of revolution : L.65.
of tangent of 2 variable circles : N.
51.
Enveloping asymptotic chords and
polars : A.14,16,17.
Enveloping cone : 566472.
of an n-tic surface : CD.4.
* of a quadric : 5697 : th of Jacobi, J.
12 and CD.3.
of a twisted hexagon, locus of vertex :
A.10.
Enveloping line of class cubic : invo
lution th, E.29.
Epicycloids: J.l : Mem.20 : N.45,46,60 :
TE.24: thsZ.15.
centre of curvature : N.59: plane and
spherical, JP.14.
double generation of : N.69.
reciprocal polar of : geoK.19.
*Epi- and Hypo-cycloids : 526672 : LM.
4: Z.18.
and derived curves : Z.17.
tangential properties of : absPr.34.
*Epi- and Hypo-troelioids : 52625 :
LM.4.
Equality and similarity of figures : J.52.
Equations (see also " Linear equa
tions ") : 5067, 211-222, 400
594: A.6,18,57,58,60,61,652,67: tr
Ac.32: AJ.6: An.51,54,: C.44,

875

Equations (continued) :
47,59,62,68,91,97,99s: CM.3: CP.
4: G.l : J.13,16,34: L.672,69 : M.
14,21 : Me.76 : Mo.79,f80 : N.67,
68,ths55,67, and 80 : P.1799.
(For Binomial, Biquadratic, Cubic,
Cubic and biquadratic, Linear,
Quadratic, Quiutic, and Trans
cendental equations, see those
headings. Other kinds will be
found below.)
Abel's properties : C.91.
algorithms for solving : M.3.
whose coefficients are rational func
tions of a variable : J. 74.
of degree above the 4th not soluble :
J.83.
whose degree is a power of a prime :
An.61: C.48: L.68.
* derived: 424 31: A.22 : in d.c,
170812.
developments : An.61.
differential operators in : LM.14.
Eisenstein's theorem : LM.7.
extension of theory of : C.58.
fundamental principles or theorems
A.1,11: C.96,97: L.39,40: J.23.
Galois' theory : C.60 : G.12 : M.18,23.
of geometry : C.68 : homogeneous,
N.64.
generic : Q.4,5.
Hariot's law of: J.2 : extC.98.
homogeneous, reduction of a princi
pal function which verifies a
characteristic homog. eq. : 0.13.,,
14,.
identical: J.27.
impossible: Man.51.
in geo. mean of roots : N.45.
in quotients of roots : N.45.
in sums of the 0 (,2) roots of another
eq. : N.43.
insolubility of quintics, &e. : J.l.
irrational : Man.51.
* with integral coefficients : 503 : 0.24:
J.53 : complex, J.53.
irreducible: An.51 : Mo.80 : of prime
degrees, AJ.7.
* linear: see " Linear equations."
* miscellaneous : 214.
numerical: C.10,123,32,78,81 : G.13 :
J.10: L.36,38,4l,83.
and commensurable quadratic fac
tors : L.45.
of nth degree with two real roots :
C.98.
from observations : A.21.
* with only one positive root: 411:
C.98.
of payments : A.34,36 : CD.l : CM.2.

876

INDEX.

Equations(continued) .* reciprocal : 466 : A.44 : 0.16 : of a

qnartic, N.66.
reduction of : C.97 : CD.6 : to recipro
cal eqs., A.35.
relation to linear d.e and f.d.e : L.36.
roots of: see " Roots of equations."
* simultaneous (see also " in two or
three variables") : 59, 211, 582 :
C.25 : LM.6 : thsN.48 and 81 :
quadratics, N.60.
deducible the one from the other :
C.22.
of the form zm+y"+zm = a : N.46.
* solution of : 45,54,59,211,466533,
582: A.64: trAn.52: C.3-,5,,62,
64,: J.4,272,87: Mo.56,61.
*
by approximation : 50(5533 : A.30:
Ac.4: C.11,17,45,60,79,,82 : E.4:
G.8 : J.14,22 : Me.66 : N.51,62,78*
80,84: No.58: P.5: Q.3 : TI.7:
Z.23.
*
Horner's method: 533: P. 19.
*
Lagrange's method : 525 : C.91.
*
Newton-Fourier method : 5278 :
AJ.4: G.2: Me.66: N.46,56,60,
69,79.
Weddle's method : Z.7,8.
by continued fractions : J.33.
by definite integrals : Me.81 : P.64 :
Z3.
by diminishing the powers of the
roots : C.41.
by elimination of integers: N.70.
by geometry : C.87.
by imaginary values : J.20.
by infinite series : J.33.
by interpolation : C.5.
by logarithms : C.95.
the one by the other: C.72 : L.71.
by radicals : C.58 : Q.15.
by series: An.57 : C.49,52 : J.6:
Mem.33.
by transcendents : An.63 : Q.5.
*
by trigonometry : 480: A.l.
a nonic eq. which has this charac
teristic : A given rational sym
metrical function 8(a,fi) of two
roots, gives a third root y, such
that a = 0{fj,y), P = 6(y,a),
y = 6 (a,/3): J.34.
symbolic, non-linear : C.22.
systems of: C.67 : G.11,18: LM.2,8 :
Q.ll: M.19: Z. 14,18 (see also
" Linear equations.")
transformation of : C.64.
* in one variable : 4558, 21416, 400
550 : approx A.20.
graphic solution : C.65.
x-k = 0: An.59.

Equations(continued) :
xmpr+q = 0 . number of real
roots: C.98.
z^+q^+p' = 0 and derivatives:
N.65*
x"-1+xn-*+...+\ = 0 irreducible
if n be a prime : L.56.
ax*m*,+bz"""+ "+i = 0: G.14.
(z-l)!+l = z": L.56.
(z'-a') + (z) = 0: N.82.
(1-f r)",(l+bz) = 0 when z is small:
A.2.
in two variables : 5967,211,2178 :
A.20,25: CM.2: J.14: N.47,,48,
63: Pr.8: Q.18.
of any degree with a variable para
meter: L.59.
implicit : Mem.30.
numerical solution : Z.20.
a;3-|-y3 = a and x^y+xy'2 = 6 : A.48.
in three variables : gn.sol, 60 : A.1,64:
N.47: M.37: byacubo-cycloid.C.69.

(y c)(zb) = a2, sym in z,y,z:

219.
*
yt+z*+yz = a2, &a, sym : 220.
*
x1yz = a2, Ac.,sym, and x cy + bz
&c, sym : 2212.
wyz = aj{(\yi)(\zi)}, Ac.,
sym : A.35.
az + by + cz = I, a'x+b'y+c'z = V,
x3+y3+z2 1, by trigonometry :
A.6.
Equiangular spiral: 5288: Me.62* N.69,70.
Equilateral hyperbolic paraboloid and
derived ray-system : Z.23.
Equimultiples in proportion : Gl.
Equipollences, method of : N.69..,70i,737,
74,.
Equipotential corves : Me.82 : Pr.24.
Equipotential surfaces : G.20 : geoJ.
42: M.8.
of ellipsoid : L.82..
Equivalence of forms : C.88,90 : JP.29.
Equivalent representation : Z.23.
Equivalents, theory of : A.44.
Eratosthenes' crib or sieve : N.43,49.
Error in final digit of decimals : C.40 :
Me.74: N.56.
Errors of observation : A.18,19: An.58:
C.93 : JP.13 : N.56 : P.70 : TE.24.
Errors of constants : Mo.83.
'Euclid, enunciations : p. xxi.
axiom 11 : J.l ; 1.47 : new proof, C.60.
II. 12 and 13: Me.80; 71.7: Q.ll
new proof, Q.9.
XI., Ac. Me.71 : XI.28 : A.10.
XII. , Ac, G.9 ; criticism on : Q.7,9.

INDEX.

Enler's algorithms : A.67.

Euler's constant: 2744: Pr.15,16,18,19,
20, Table 27.
and Binet's function : C.77 : L.75.
Enler's equation : N.72 : integration of
it by the lines of curvature of a
ruled hyberboloid, N.75.
Euler's equations of motion solved by
elliptic integrals : Q.14.
Euler's formula for (1+as)": L.44.
Euler's integrals: 2280-2323: A.41 :
Ac.1,2: An.54: C.9,17,94,95,th96:
J.15,212,45 : fL.43 : Me.83 : Z.9.
* B(l,m): 2280: An.69 : G.9.
r (n) : see " Gramma function."
ap. to series and functions of large
numbers : JP.16.
sum formula and quadratic residues :
An.52.
Euler's numbers : AJ.5 : An.77 : C.66,
83 : J.79,89 : prsL.44 : Me.78,80.
Evectant of Hessian of a curve: E.
32.
*Even and odd functions : 1401 .
Evolute: 5149 59, 5165: An.53,61 :
C.30: Q.3,11.
analogous curves : L.76.
* of a catenary : 5159.
of a cubic curve : Q.ll.
of a cycloid : A.30.
of an evolute, in inf. : L.59 : Me.80.
* of an ellipse: 4547,4958: C.84: N.52,
632.
and involute in one : L.41.
of the limacon, rectif. and quadr. of:
E.40.
of negative focal pedal of a parabola :
E.29.
oblique, direct and inverse of differ
ent orders : C.85.
* of a parabola : 4549,4959 : Q.5 : N.65.
rectification and quadrature of : A.4.
of surfaces : C.74.
of symmetrical bicircular quartics :
Q.18.
* of a tortuous curve : 5731 : A.25.
*
angle of torsion of evolute : 5754.
integrable equations : L.43.
Evolution : 35.
E (x) = integral part of as : O.50 : L.57.
Ex-circle of a triangle : 711,953 :
4749: A.54: tb.N.6'0.
locus of centre, th : Q.9.
Expansions of a function in a series
(see also " Series " and " Sum
mation"): A.31: Au.7: thsAJ.
3 and 4: C.7,13,17,20 : CM.4:
J.90 : L.38,46,76 : M.16 : M<51.3 :
Mem.33 : N.82,83 : num. Q.3 :
Z.2.

877

Expansions of a function in a series

* of circular functions: 2955: A.ll :
GM.3: J.43: L.36: Q.12: of
imag. arcs, J.6.
coefficients of : gn form, C.85 : gn
property, J.41.
connected with a 2nd order d.e : C.5 :
L.36,372.
of denominators of convergents :
C.46: JP.21.
of exponentials : J.80.
* of explicit functions : 150047.
extended class of : C.82 : approxi
mating to functions of very large
numbers, L.782.
of faculties of the variables : M<5m.31.
of implicit functions : 155073.
of Jacobian functions: An.82.
of Legendre's functions, X : An.75.
with limits : C.34.
of another function : 1559 : C.95,96.
of periodic quantities : 0.52,53 : JP.ll.
of powers of the variable : At.57 :
0.19: L.46.
of powers of a polynomial : C.86 : J.53,
88,.
of powers of another function : Me'm.
33: N.74.
within a given interval according to
the mean values of the function
and of its successive derivatives
in this interval : C.90.
by Bessel's function : J.67 : M.10,17 :
Z.l.
* by binomial theorem : 125.
* by factorials : 3730.
* by generating functions : 3732.
* by indeterminate coefficients : 232,
152734: A.3.
by logarithmic method : C92.
* by Maclaurin's th : 1524.
by a series : C.93,95.
Expansion of :
alg. functions : C.89
Z.45.
Eisenstein's th : J.45.
11, alg. functions from n eqs : O.ll.
_i
(1 + ax)" in an integral series : A.65.
(l-x)(l-z*)(l-x*) ... : C.92: J.21 :
L.42.
(1)(1e)(l-as*)(la*) ... : Me.80.
(l-H)(l+2as) ...(1+n-lx): C.25: J.
43.
(l+aa!-|-6*2 + ... + ia!"-1)-1: AJ.6.
{(x-ai)3+...+(x-a,)*}-1: C.95,.
(l+az + bx*)" . Q.18.
(12o(B+(eS)-: L.372.
21.1
(12ax+a?a2y ~* : C.86j.

INDEX.

878

Expansion of(continued) :
(l-oaj-fce*) : J.43.
J.40.
(xz)" in powers of z*l : CM.
(x+2/)<">: CM.3.
nth derivative of </(a2hh$) : A.4?.
(1~^21~4^)"in powers of t, when
V=

: 1565.

* Bernoulli's numbers : 1545.

circular functions : J.24 : Q.5.
an arc in linear functions of sines or
tangents of fractions of the arc
in g.p: L.43.
powers of arc in powers of sines : J.11.
* it 29312, 2945, 29002 : He.78.
*
powers of n : Me.78 : ir-, 858 : sr"1,
Me.83.
* sin 6 and cos 6 in powers of 8 : 764,
1531 : A.5,29 : C.16.
* Bin"# and cos"# in sines or cosines of
multiple arcs : 7724 : A.24,55 :
C.12 : OD.3: J.1,5,14: N.71 : TI.7.
* sin n$ and cos n6 in powers of sine or
cosine : 758, 775-79, 1533 : C.82 :
CM.2: Me.76: M<5m.l3,15,18 : N.
73.2,83 : Q.4 : couvergency of the
series, J.4.
*
cos u6 in powers of cos 6 : 780 : Q.
12.
*
sin-1* : 1528,-64 : J.25 : N.74: re
mainder, Z.I..
cos"u,e: A.ll.
* tan
1525, 2913,-17: A.16 : C.88 :
N.57.
* 0Ot: 2911,16: C.88: Q.17.
* sec x : 1526 : A.16 : C.88 : J.26 : N.57 :
Q.17.
* cosecu; : 2914, 8.
* tan-'*: 791.
* tannfl: 760.
* a'sin * in differences of sin x : 3749.
sin nx I cos x" : A.4.
(1 p cos <j>) _1 = 2a cos 2ii<f> : A.21.
(a^ + 62 2a6cos <*>)exp. (m + J):TE.5.
(a+b cos <p + c cos <p')~" in cosines of
multiples of <t> and <P' : J. 15.
cos k cos-1 (cosa+a) : C.15.
cos {0 + 6l + ... + 0.i): A.34.
y in powers of x when x = sin y
sin (y+a)'
796,1558.

Expansion of(continued) :
* do. when x = '-^O . 1570.
y
6 cot 6 in powers of sin20: Q.6.
cot"' (m1) cot-1 (m+l) : A.47.
differential coefficients by f.d.c and
the converse : J.16.
elliptic functions : A.19.
andof their powers : C.83: cos amx,
L.64.
equations : L.50.
exponential functions : N.82.
j_
* e = limit of (1 +x) '-. 1590 : A.3,23 : Q.7.
* e': C.30: N.48.
*

^4= 1539, 15434.

* e* in powers of ye*> : 1571.

(oe*l)-1: At.57.
* -e -. 2962.
e"'e-"
eexp.
= Z.3.
* e" cos bx . 798.
e~" cos 6a; dx, and summation of
the series : J.41.
e exp. sin x -. 1529.
e exp. a sin-1* 1535.
c exp. log (z+* sin y) : 1557.
e ezp.<p(x,y,z. ..): C.58j.
x exp.[a; exp.[a: exp, &c. : J.28.
fractions : 248.
functions :
Al^^ll^x), Ah(x) (Weierstrass's func
tions) in powers of the modulus :
C.82s,85,86: L.792.
by Taylor's th : CM.4.

j:

f(x+h) (see Taylor's th) : 15009,

1520: Abel's th, 1572; Stirling,
1546 ; Boole, 1547 : AJ.3.
f(x) (Maclaurin): 1507: 3759.
f(x + h,y+k): 1512,1521.
f(x,y) : 1516, 1523 : Me.3.
/(0) (Bernoulli) : 1510.
<t>(a+bx + cx2+...) (Arbogast): 1536:
CD.1,6.
-fr (Cayley) : 1555.
f {z+x <p(y)} in powers of x (La
grange) : 1552 : Laplace's th>
1556.

INDEX.

Expansion of(continued) :
* f(y) in powers of ^(y) (Burmann) :
1559.
* f{+-l(*)} and
15613.
* <f (e') (Horschel) : 3757.
* uxtH: 3740; A"tl: 3761.
* A"u, and A"u0 . 37412.
*
in differential coefficients of w :
3751.
* A"zm and A"0'" : 37434.
* , in differences of u : 3752.
* j^WiCfe in terms of o> j, &c. : 3778.
a function of a complex variable : M.
19.
a function of a function : AJ.2,3.
functions of infinitesimals : G.12.
a function of a rational fraction : At.
65.
a function of n variables : C.60j : J.
66.
a function of y,y in ascending powers
of x,x' when y = z-\-x<p(y) and y =
z'+x'<p(y') as in 1552: J.48.
holomorphic functions : M.21 : by
arcs of circles, C.94.
* implicit functions : 551,1550: L.81.
integrals : A.l : of linear d.e, An.71 :
of log x, A.4.
* logarithms : 1529 : N.82.
* log (1 a0, log i?, &c. : 155-9.
1 x
* log y and log y" in powers of a-1 when
ifay+S = 0: 1553-4.
* log (<*+&+*+...): 1537.
* log (1 +2 a cos x+a2) . 2922.
* log (l+ncosai): 2933.
* log 2 c?s (%) : 2927.
6 sin \ 2)
* logr(l+*): 2294, 2773.
higher integrals of log x : A.4.
numbers : M.21.
* a polynomial : 137 : Z.26.
a quartic function : A.35.
Exponents: N.57 : P.1776.
reduction for d.i : C.16.
Exponential, th : 149 : N.52.
functions : P.16.
replaced by an infinite product : G.99.
Exponentials, successive of Euler : L.45.
Factorials:
calculus of: L.57 : N.
60: Pr.22: Q.12: Q.f8.
geom. i.e(l+aO(l+rac)(l+r*ss)...: C.
17.
* notation : 94, 2451 : Q.2.

879

Factorials(continued) :
reciprocal : C.17.
treatment by limits : J.39.
1,22,33.. .n": Me.78.
* w! = r(l+n) : 2290.
approx. to when n is large : C.9,50 :
J.25,27 : L.39.
n ! = n"e-V(2njr) (Stirling) : Q.15.
{2J"1^! ot"""}-1 : CM.3.
* l'3/l "' theorem: 339.
2.4.6 ...
0 (n, r) when n a+ij3 : J.43.
Factors: 127.
in analysis of integral functions : M.
15.
application to rotations to indicate
direction: J.23.
* of composite numbers : 274: J.ll.
complex : 0.24.
equal, of integral polynomials : C.42 :
L.56.
* of an equation : 400 : J.3 : condition
for a factor of the form x"a",
A. 55,63.
irreducible, of an integral function ac
cording to a prime modulus p
C.86.
linear, resolution into : N.82;.
of polynomials and geo.ap : J.29,89.
product of an infinite number of : A.
59.
x cos
to cos
m ... : -Br
-A
cos
N.tO.
2
4
o
radical, of numbers : C.24,25.
of Ax2+Bif+Cz2, th of Lagrange:
AJ.3.
otx*fgy* = l : A.33.
of (x+y)nx y" : thQ.15,16-2.
* of
2j;"!/ncos^+f/2": 807.
of
2re cos n6+x-n : CP.ll : Me.76.
(1-e) (l-) (1
C.96.
* tables of (Burchardt's) p.7 : to 4100,
J.46.
geo. properties : J.22.
transformation of: A.57.
of 100...01 : Me.79.
Faculties, analytical : J.7,1 1,33^35,40,
44,51.
coefficients of : A .9,11 : At. 75.
divisibility of : A.48.
numerical, of 2nd order : He'm.38.
series : Z.4.
Fagnani's theorem: 6088: A.26: LM.
5,13,23 : Z.l.
onrves having Fagnanianaros: LM.ll.
stereometric analogy : Z.17.

880

INDEX.

Faisceaux :
of binary forms having
the same Jacobiau : C.93.
of circles : C.76.
of conies : Z.20.
curvature relations : Z.15.
formation of : C.45 : CM.3.
intersections of : N.7'2 : degree of the
resulting curve, J.71.
of lines and surfaces : N.53,83.
plane : N.53 : defined by a first order
d.e, C.86.
of tortuous cubics in connection with
ray-complexes : Z.19.
Fan of Sylvester : E.33.
Faure's theorems : G.1,19 : and Painvin's, N.61.
Format's theorems :
of (N"'1!)
+p : 369 : A.32 : AJ.3 : J.8.
of ss"+y" = z" being insoluble when
n is an odd prime, &c : An.57 :
C.gz84 and 965,91 : J.40,87 : TE.
21.
analogous theorem: J.3.
case of n = 14 : J.9.
and periodic functions : Me.76.
z+y = , x2+y3 = D3: M6m.26.
of the semicircle : A.27,30,31 : gzA.31.
method of maxima : C.502.
Fcuerbach's : th of the triangle, Me.84 :
circle, A.59.
Fifteen girl problem : E.34,35 : Q.8,9.
" Fifteen" puzzle : AJ.2...
Figurate numbers : 289: A.5,69.
Finite differences, calculus of : 3701
3830: A.13,18,24,63: C.70: J.ll2,
12,13fcl4fcl5,.16: Me.82: M61.5 :
Mem.13 : N.69 : thsP.16,17.
ap. to complex variability : An.82 :
ap. to i.eq, An.50.
* first and th differences : 3706.
A*tt = 0 : An.73.
* A0" : 3744: Q.5,8,9: Herschel's table,
N.54.
A"0"+n(w), table of: CP.13.
8it0, fi2M0, Ac, in a function of Au0,
tihto, &o- : N.61.
* A"u in successive derivatives of u :
3761 : N.73.
Alp and Bernoulli's numbers : An.59.
hu, = Aux Au,+ Au, &c. :
2t
12
Ac.5.
A sin* and A cos*: CAI.l.
Finite difference equations : AJ.4: An.
59: CD,2: CM.1,3,4: CP.6: JP.
6: L.83: P.60: Pr.10.
of integrable form : C.54.
of mixed differences : Q.10.

Finite difference eqs.(continued) :

of the kind u,it u,-,,^,, : CM.4.
linear: AJ.7: An.50: At.65: Q.l.
first order, constant coefficients :
C.8.
determination of arbitrary con
stants : A.27 : At.65 : G.7.
integration to differences of any
order: J.12.
with variable coefficients : C.17.
partial :
constant coefficients : C.8.
linear of 2nd order : C.98.
of physics : C.73.
Finite differences :
exercises : No.
44,47.
formulaa: CD.9 : Q.2.
sum and difference: J.58.
of functions of zero: TI.17.
H (n, r) value of : Q.9.
integrals : C.39,57 : JP.4, : L.44.
expressed by definite integrals :
An.53.
2e*i/: A.6: No.44.
inverse method : C.74 : P.7.
involving 1/1 : Me. 78.
of powers converted into d.i : JP.17.
Fleflecnodal planes of a surface: Q.15.
Flexure : AJ.2, : Me.2.
of ruled surfaces : An.65.
of spaces : LM.9.
of spherical surfaces : Me.77.
Fluctuating functions: 2955a: LM.5 :
M.20: TI.19.
Fluents: P.1786.
of irrational functions: P.16.
Focal: chords of conies: 1226, 4235,
4339.
circle of conies : Mel2.
distances: 4298: N.64.
pedal of a conic : N.66.
Focal properties:
of conies: 1163,
11679, 1181, 1286-8, 4298
4306, 433645, 4378, 4382, 4516,
455058, 4719 21, 500816:
CD.7.
of curves: CD.7.
of nomographic figures : N.71.
of a parabola : 1220, 12236, 123 04,
4231, 42358 : G.22.
of a quadric surface : An.59 : N. 58.
Focal quadrics of a cyclide : Me.85 .
Foci: J.64: N.42,44,53,85 : Q.2,a.c9.
of conies : 1160: trA.25,63,64,cn69 :
gzC.22andL.39: CP.3 : N. 69,74,
78,81,82: t.cQ.8,13,12and 45: gen.
eq, N.48.
analogous points in higher plane
curves: J.10.

INDEX.

Foci :
of conies(continued) :
coordinates of : 4516.
eq. of : LM.ll : o.cE.40.
exterior: N.79.
*
to find them: Q.25: from gn.eq,5008.
through four points : N.833.
*
of four tangents : 5029 : N.83.
locus, a cubic : M.5.
negative : A.64.
under three conditions : Q.8.
of curves: C.82; nth class, 86 : N.59,79.
of cones : N.79.
of differential curve of a parabola :
A.58.
of in- conic of an -tic, locus of : E.21.
of lines of curvature of an ellipsoid :
Z.26.
of quadrics : N.42,66,74,75,78.
of quartics : J.56.
of the section of a quadricbyaplane :
N.64,70.
by another quadric : N.47.
of surfaces : C.74 : of revolution, N.
74.
Folium of Descartes : 5360 : N.44.
Forms, theory of : M.18 : of higher de
gree, Mo.83 : Pr.38.
reciprocity principle : An.56.
Formulae: G.15,19.
for log 2, &c. : Me.79.
in the Fund. Nova : Me.76.
* of reduction in i.c : 1965 : Me.3.
Four colors problem : AJ.23.
Four-point problem : E.5,6,8*
Four right lines not 2 and 2 in same
plane: J.5.
Fourier- Bessel function : J.69 : M.3.
Fourier's formula in i.c: 2726 42:
CM.3: J.36,69: L.36: M.4 : Me.
73: Q.8: gzZ.9.
ap.to calculationof differentials : J.13.
Fourier's theorem : 518 : 528 : An.50,
75: J.13: M.19: Me.77,82,83.
ap. to a function of a complex varia
ble: M.21.
Fractions : AJ.32 : G.9,prl6 : J.88 : L.10.
continued, decimal, partial, vanishing,
&c. : see each title,
number expressible by digits 5> n :
C.96.
reduction to decimals : A.1,25.
transformation into decimals: A.ll.
Frullani's formula: 2700: LM.9.
Fuchsian functions : 0.927,93,94j,95,96.
Fuchs's theorem on F{z,y,y:) = 0 : C.99.
Functional equations : CM.3 : J.90 :
TE.14.
f.*x=fa: C.88.
f.<px = \+fx: C.99.
<t>.fx = F. fx, to find <t> : Mem.31.
*

881

Functional equations{continued) :
<p*+<py = pfafy+yfo) J-2.
fa <t>&+fiy .<p#s+&c.
= fix <Piy+fyt <hy+&c. . J.5.
^+^=^fy-Fx+fx-Fy: j.46.
b
x i<*y)
/(, 6) <p(x + 8) dd = F(x), to find <p :
Pr.8.
fc,-*^| = &c: Q.15.
cx + d
/(*)=/( sin ^) : C.88.
Functional images in ellipses : Q.17.
in Cartesian ovals : Q.18.
Functional powers : Mem.38.
symbols : Q.4.
Functions: A.28 : AJ.6 : An.79 : C.43,
91 : CP.l : J.16,prs71,74,84,87,
91 : L.45 : Me.7 : Mo.80,81 : P.15,
16,17,62: Pr.ll2: prsZ.26.
algebraic, alternating, analytical, cir
cular, circulating, conjugate, con
tinuous, curvital, cyelotomic, de
rived, discontinuous, elliptic, even
and odd, exponential, Fuchsian,
gamma, generating, hyperbolic,
implicit, infinite, imaginary, in
tegral, irrational, irreducible,
isotropic, iterative, monodrome,
monogenous, monotypical, lionuniform, periodic, polyhedral,
quantitative, rational, representa
tive, transcendental, trigono
metrical : see the respective
headings.
analogous :
to algebraic functions :
C.89.
to circular functions : C.84.
to Euler's : C.89 : M.19.
to functional determinants : J.75.
sine and cosine : Q.16.
to modular functions : Ac.2 : C.93.
connected by a linear cq. : C.17.
condition of /(.<,?/) boing a function
of <j>(x, y) : A.21.
development of : see " Expansion."
defined by d.e : JP.21,28.
differing very little from zero : L.74.
errors of geometricians : J.16.
expressed by other functions, remain
der: C.98.
fractional : J.8 : the variable being
the root of an equation, N.56.
from functional equations : M.24.
from Gauss's equation : C.92.
with lacunas : C.96.
Lagrange, tr : JP.5,7.
5 U

882

INDEX.

Functions(continued) :
linear: C.90.
with linear transformations inter se :
M.19,20.
whose logarithms are the sums of
Abel's integrals of the 1st and
3rd kind : C.92.
with non -interchangeable periods :
M.2021,25.
number of values of : C.48.
do. through permuting the variables :
JP.10j.18: L.50,60.
of two variables : Ac.3 : C.90,96V
made constant by the substitution
of a discontinuous group : C.97.
which arise from the inversion of
the integrals of two functions :
C.92,.
whose ratio has a fixed limit : G.5.
f(x, y) such that / {zf (x, y)} is sym
metrical : J.l.
of three variables satisfying the d. e,
&F = 0: Ac.4.
of three angles, th. re 1st derivatives :
J.48.
of 4 and 5 letters : L.56.
of 4, 5, and 6 letters : L.50.
of 7 letters : C.57,953.
of 6 variables which take only 6 diffe
rent values through their permu
tation, not including 5 symmetri
cal permutations : A.68.
of n variables : C.21 : Mo.83 : with
2w systems of periods, C.97.
analogous to sine and cosine : Q.16.
number of values : J.85. : do. by
permutationofthe variables: C214.
obtained from the inversion of the
integrals of linear d. e with
rational coefficients : C.90,: J.89.
of an analytical point, ths : C.95,.
of a circular area from a given inte
gral condition : Z.26.
of imaginary variables : C.32,48 : JP.
21 : L.58,59S)603,61j,62 : LM.geo
8.
of large numbers, appro*. : C203.
of a real variable, connexion with
their derivation : M.23,24.
of real arguments, classification ac
cording to their infinitesimal
variation: J. 79.
of the Bpecies zero and unity : C.95.
of a variable analogous to the poly
nomials of Legendre : C.95.
allied to Ffaffians : Q.16.
rationally connected : L.59.
with recurring derivatives : LM.4 :
TE.24.

Functions(continued) :
which relate to the roots of the equa
tion of division of a circle or of
n'-l = 0: J.17.
representation of : C.92j : M.17 : onevalued, Z.25.
approximate : Z.3.
by an arbitrary curve : M.22.
by Bessel's functions : M.6.
by definite integrals : Ac.2.
by elliptic functions : An.82.
by Euler's sum-formula: J.56.
by Fourier's series : Mo.85,.
by graphic methods : A.2 : imag.,
J.55.
by infinite products : Z.24.
y = e"^, constant and r a positive
integer: A.42.
y = zV: A.52.
reproduced by substitution : C.19.
resolution into factors : Ac.65: C.19,
30 : CP.U : J.18.
satisfying the eq. Af= 0 : C.96.
singularities of : M.19.
whose successive derivatives form an
arith. prog. : An.71.
systems of: Mo.78 : of two inter
connected, C.98.
of two systems of quantities, cor
relative and numerically equal :
C.98.
which are neither rational nor reduci
ble to irrational algebraic ex
pressions : C.l&V
which are of use in elliptic functions
and logarithms : No.58.
which take a given value in a given
position : An.82.
which have no derivative throughout a
certain interval : An.77.
which vanish with their variables :
TI.16,.
x*: An.63.
{(")"}" and so on, and the corres
ponding inverse function : J.42.
arising from *^(42xz-r-z*): J.2.
f()=^|; LM.9.
cx+d
f(u,z), u being an implicit function
of an imaginary variable : Pr.42.
/(*), formula of analysis : J.53.
f(x) = 0,y=f (*), th. re f (y) : B.36.
*(*)=<Uogr(*): 2743.
^ (a) of Jacobi : J.93.
Q(): Ac.2.

INDEX.

Functions(continued) :
Bessel's, I(x) = ['cos (x cos 6) d8
IT Jo

23 22.42 22.42.6Cauchy's numbers ; N-njj,

(e<"e-")''du.
cosine integral ; Ciq = [' co^ - dx :
J. x
taP.70.
Dirichlet's function,
F(x) = z{^)^ . Z.27.
\ n I n'
elliptic ;
far-*-1 g (a") {E (**)}*<&! : J.23.
* Euler's ; B (I, m) : 2280.
expon-integral ; Ei q = - cfc: A.
10:taP.70.
1'" *
E = E" f sin2" u cos (e cos o>) <2o> : An.
70.
M<$1.6.

= -915965,594177 ... : Mem.83.

r (x) . see " Gamma function."
Jacobi's

C.592,60: L.47,50.
Laplace's Y(n) : M.14.
log-integral; Liq = 1*-^-.
Jologa
L(l+x)^xg + |i-&c.
Legendre's X, : see " Legendre."
P.wheref.expf-g) =
Q.10.
P" (cos 7), n = 0: G.22.
P(a^,x): Z.14.
n (z) = a;'e-*<fa ; ^ (z) b i log n (z) :
Qi.
sine-integral : Sia= ynnx fa taP.
70.
Jo *
2 J e exp (z') P(z) dz = 0 : C.93.

883

Functions(continued) .JVfl+aa3} dx : J.92.

E(p,<p) dp
r(p, =
Jo1 -v/{l sin^}"
X, Y, &c, such that SXYd<r = 0 and
and that any function can be exEanded in the form aX+/3 Y+ &c. :
M.10.
Gamma function, T (n) : 2284 : A.4,6,61 :
An.69 : C.35,92,96 : J.35,82,90,
ap57: L.42,46,52,55 : Q.9: Z.l,
25.
application of this and other trans
cendents : C.86.
of a complex : Me.84.
* curve y = T(x): 2323.
* deductions: 22862316: A.10.
derivatives of : Q.6.
of equidifferent products : J.36.
of an infinite product : J.39.
* = limt. of
: 2293 : A.30.
* logarithm of : 2294, 2768 : C.9.
*
numerical calculation : 2771.
*
as a definite integral : 2768.
n, negative : CD.3.
* numerical calculation of : 2317.
* the function + (x) = d* log r (x) : 2743
70.
reciprocal of : Z.25.
reduction of : J.40.
* transformation of : 2284,2318: J.57.
r(+l) = V (2tt) e-"nn+(l-f-e) (Stir
ling): C.502.
*r(m)r(l-w)=
: 2313: Ac.3.
sin mn
*r(x) r (+- ) ... r ((6+2=1): 2316=
L.55,56.
T(x) = P(x)+Q(x): Ac.2.
2 -i-, : G.6.
r(x)
Gauche cubics : C.82 : J.60 : N.623.
3rd class, theorems : J.58.
number of common chords of two :
An.70.
through five points : N.83.
through six points, cn : N.66.
Gauche curves : C.70,77,903.
Mo.82: thsN.53j.
classification of : JP.32.
on a cubic surface : C.62.
of a developable surface, singulari
ties: An.70.
differential invariant? of: JP.28.

884

INDEX.

Gauche curves {continued) :

intersection of two surfaces having
common multiple points, singu
larities of : C.80.
on a one-fold hyperboloid : An.l : C.
52,53.
metric properties of, in linear space of
n dimensions : M.19.
representative curve of the surface of
principal normals of : C.86.
of the zero species : C.80.
Gauche helicoids : rad. of curv. : N.45.
in perspective : JP.20.
Gauche :
in-polvgons of a quadric:
C.82.
perspective of algebraic curves : C.80.
projection : N.65.
quadric : N.67 : and orthogonal tra
jectory of generatrices : thN.48.
quartic : A.62 : C.82 : L.70.
9 points of, 7 points of a gauche
cubic and 8 associated points :
C.98.
unicnrsal, a class of : C.83.
surface : N.61.
sextic curve : C.76.
surface: JP.17: L.37,72.
deformation of : C.57.
which can be represented by a p.f.d.e
of the 2nd order: C.61.
Gaussian periods of congruent roots
corresponding to circle division :
J.53.
Gauss's function : see " Hypergeometric
function."
Gauss's theorems : J.3.
General methods in anal, geometry :
4114.
General numerical solution of any
problem : LM.23.
Generating functions : 3732 : J.81 : N.
81: Pr.5
and ground-forms of binary qnantics
of first ten orders : AJ.23,3.
do. of binary 12-ic and of irreducible
syzygies of certain quantities :
AJ.4.
of some transcendental series : At.55.
for ternary systems of binary forms,
ta: AJ.5.
Geodesies: 5775, 583755: A.39 : C.
402.41,9<).p.c97: CD.5 : G.19 : J.
50,91 : M.20 : Mo.71 : N.45,65 : Q.
1,5: Z.18,26.
of cubic surfaces, loci : CD.6.
curvature: 5846: C.42,80 : L.12.
on an ellipsoid : M.20.
*
radius of: 5776,5846.
duals of: Q.12.
equations of: 5837.

Geodesies(continued) :
flexure of: C.66.
forms from cn of their polar systems :
Z.24,.
geometry : 5855.
problems: A.8 : An.65: CP.6: J.53,:
L.49 : TN.73;.
on a quadric or ellipsoid: C.22-: CD.4:
J. 19: L.4,.,41,44,48,57: M.35: N.
76.
Joachimsthal's theorem: J.42.
and corresponding plane corves
C.50.
and lines of curvature : L.46, : N.
82.
*
j>d constant along such : 5842.
shortest lines : .T.26.
*
through an umbilic : 5850.
on a right cone : A.69.
radius of torsion of: 5848: RR'PU +
P = 0, Me.75.
sections : Z.2.
shortest lines on surfaces : 5838 : J.
20.
on a spheroid : J.43.
triangles : Mo.82.
best form of : N.55.
reduction of arc of a small one : An.
50.
Geodesy, spherical problems : A.25j,63.
representation of one surface upon
another: An.70.
Geodetica, integration of its eq : An.
53.
Geography, comparative : A.57.
Geometrical conies: 1150 1292 (see
Contents p. xviii) : G.l : Me.62,
64,71,73 : Q.10.
Geometrical :
constructions : LM.2.
definitions : J.l.
dissections and transformations : Me.
75.
drawing : A.23.
figures, general affinity of : J.12.
forms : G.l : of 2nd species, G.3,4.
* mean : 92 : CD.8 : Pr.29.
approx. to by a series of arith. and
harm, means : N.79.
paradoxes : Z.24.
* progression: 83: A.pr2,6 : G.ll : N.
54.
a property of 1,3,9.27... : A.33.
proportion, theory of pure : A.62.
quantities and algebraic eqs : C.29.
reckoning (Abzahlenden) : M.10.
relation of the 5th degree : M.2.
relations, ap of statics : J.21j.
signs : Z.14.
theorems and problems : 9201102.

INDEX.

Geometrical(continued) :
theorems: J.ll : L.46 : N.743: Lawson's, Man.13.
method of discovering : J.8.
from a principle in alg. : LM.ll.
problems: At.25,32.
transformations : A.32.
and ultra-geometric quantities : C.52,
55.
Geometry :
of the Ancients : At.22.
comparative, ap. to conies : N.653.
of derivation : An.543.
* elementary : 9201102 : J.6,10 : A.2 :
N.623.
principles of: A.40: C.56: G.11,14,
20: LM.16: Me.62: Z.20.
enumerative : Ac.l.
higher : A.20 : No.73 : prsA.55.
instinct of construction : N.56 : Q.2.
der Lage : Z.6.
linear : A.27 : ap. to quadrics, M.10.
linear and metrical : M.5.
of masses : JF.21.
and mechanics, on their connection :
L.78.
organic, of Maclaurin : L.57.
plane, new anal, foundation : M.6.
plane and solid, analogies : L.36 : N.
6.5,.
plane and spherical, ths : Mem.15.
of position : J.50 : Q.l : analTE.9.
theorems: An.55: J.31,34,38,413 :
TB.28.
5 points in space : CM.2.
in lieu of proportion : CP.10.
of space, abstract : Pr.14,18 : aphor
isms, J.24j.
* of three dimensions : 55016165 : N.
63.
Glissettes :
problem: Q.11.
centre of curv. : JP.21 : rad. of curv.,
L.45.
Golden section : A.4.
Goniometrical problems : Q.7,15.
Graphic calculus : C.89.
Graphs (Clifford's) : LM.10.
application to binary quantics : LM.
17,.
to compound partitions : AJ.96.
Grassman's life and works : M.14.
Greatest common measure : see " Highest
common factor."
Greatheed's theorem, D.C : CM.l.
Grebe's point : A.58.
Green's theorem, &c. : J.39.44,47 : TE.
26.
Griffith's theorem (Conies) : 6096.
Ground figures, single and double rela
tions : J.88.

885

Groups : AJ.l : LM.9 : M.13,20,22.

cyclic, in Cremona's transf. of a plane
An.82.
in a quadratic transformation : An.
82.
discontinuous : C.94.
of linear substitutions : Ac.l.
of finite order contained in a group of
quadratic substitutions : C.97,98.
of finite order contained in the semicubic groups of Cremona : C.99.
Fuchsian : Ac.l,.
formed from a finite number of
linear substitutions : C.83.
of interchangeable elements : J.863.
introduction to the theory of : Me.62-.
Kleinean: C.93.
of many-valued functions : Man.62.
modular eqs. (Galois) : M. 14,18.
non-modular : Man.653.
of points G'4 on a sextic with 5 double
points : M.8.
primitive: C. 72,78,96 : L.71.
for the first 16 degrees : C.75
degree of, containing a given sub
stitution : J.79.
(P)j6o (njjso of the figure of six linear
complexes of right lines two and
two in involution : An.83.
principal, classification of : C.73.
of substitutions: C.67,84,94: M.5:
isomorphism of, G.16.
of 168 substitutions and septic equa
tions : M.20.
transitive: G.22: J.83 : N.84.
Guldin's theorems : 5879 : Me.85.
Harmonic axes : of curves : C.743.
of a system of right lines and planes :
G.4.
Harmonic centre for a system of 4 points
in relation to a given pole : Z.20.
Harmonic division :
of a conic : G.10.
of a quadric : G.10.
* of a right line by a circle and chord of
contact : 948.
Harmonic :
hexhedron and octahe
dron : Z.18.
* pencils and ranges : 933, 4649 : Q.6.
*
of 4 tangents to two conies, locus of
vertex : 4984.
* points, system of four: 1063: N.51.
polar curves : A.50 : M.2.
* progression or proportion : 87 : ext
of th,A.31,43,tr41: C.43: Me.82 :
N.85: Z.3,14: sum of, Pr.20.
divergency of : A.l.
* section by a quadric and polar plane :
5687.
Harmonics in a triangle : A.57,

886

INDEX.

Hermite's <p function, linear transf. of :

M.3.
Helix : 5756 : A.64.
conical : N.13,53 : rectif. of : N.45.
on a twisted cone: A. 16.
relation with cycloid : C.51.
Hemisphere, volume, &c. : 6061.
Herpolode of Poinsot : C.99.
Hesse's surface, &c. : Z.19.
Hessian : 1630 : J.80 : curve, M.13.
covariant of binary quintic form : M.
21.
of a quaternary function : Q.12 :
cubic, Q.7.
of a surface : nodes of, J.59 : con
stant of, M.23.
Hexagon: thN.65.
Pascal's : see " Pascal."
in space : J.85,93.
Hexahedron : 907.
Higher algebra : An.54 : N.55j (Serret) :
Q.45.
Higher :
analysis : A.25 : G.14.
arithmetic : J.6,9 : N.81.
geodesy: Z.19: trZ.13..
geometry: A.10: N.57: Z.6,17.
planes : A.47.
variation of simple integrals : Z.22.
Highest common factor : 30 : A.3 : M.
72: N.42,44,452.
of 2 complex numbers : no. of divi
sions : L.46,48.
of 2 polynomials : CM.4.
remainder in the process : C.42.
Holditch's theorem : see " Closed
curve."
Holomorphic functions : C.99 : G.22.
development in series : C.94: M.21.
Homalographic projection: N.61.
Homaloidal system, n-tic surface and
an (n l)-ple point : G.13.
Homofocal :
conies : tbN.49,: loci
relating to parallel tangents, C.
62,63,.
quadrics: C.50 : L.th51,60: N.th64,
79: Pr.332.
paraboloids : A.35.
and conjugate surfaces, tr : Z.7j.
common tangents of : C.22 : L.46.
quartic surfaces, triple system of, in
cluding the wave surface : N.85.
sphero-conics : L.60.
surfaces, and p &in~i'+v sin2t" = a~ :
C.22.
Homogeneity of formulae : C.96 : thsN.
49.
Homogeneous coordinates : G.l3,8: Z.15.
metrical relation : G.ll.
Homogeneous functions : see " Quantics."

Homogeneous products, E(n,r) : 98

9: Q.6,9,10.
and sums of powers : 538 : E.39,40.
Homographic division of three tangents
to a conic : Mel.2.
Homographic figures : threeC.94 : thQ.
3 : N.58,68,pr61.
corresponding points, th : L.45.
focal properties : LM.2j.
relation of roots : N.73.
Homographic:
pencils: 4651.
systems of points : 105873.
on quadric scrolls : Q.9.
theorem of a conic : N.48,49.
transformation : N.70 : of angles,Q. 14.
Homography : Me.62 : N.60 : Z.21.
and perspective : N.69.
and rotations, correspondence of : M.
15.
Homological polar reciprocal curves :
ET.44.
Homology : 975 : G.3,8 : N.44 : E.24.
conic of: C.94.
of sets : Q.2.
of triangles : 975 : Me.73.
Horaothetic conies : 4523 : N.64,th68.
with the same centre : C.66.
Horograph : 5826.
Hyperbola :
theorems : A.27,46 :
CD.l : N.42v
with asymptotes for coord, axes :
4387: Me.73.
asymptotic properties : 1182.
conjugate : 1160.
construction : 1247, 1289.
eccentric circles : A.44.
quadrature of,&c. : 6118 : A.25,26,27 :
N.44 : TI.7 (multiple areas).
rectangular : 4392 : Z.26 : under 4
conditions, A.3.
segment of : 6118 : N.61.
Hyberbolic arc, rectification of: 6115:
J.55 : P.2,11,59.
Landen's theorem : 6117: LM. 11,13..
Hyperbolic functions : 2180: A.19 : G.
15 : Mem.30 : N.64.
analogy with the circle: An. 51.
ap. to evolution and solution of eqs. :
A.38.
construction of tables of : J. 16.
generalization of: A.35.
Hyperboloid : 5605 : J.85 : Me.66.
theorems : geoG.4 : J.24,86.
one-fold: 5605: of rotation, A.70:
L.39: M.18: N.58.
parameter of a parabolic section of :
N.75.
two-fold : 5617 : A.18,ths27.
conjugate : CD.2.
equilateral and of revolution : Ac.5.

INDEX.

Hyperboloid(continued) :
* generating lines of : 5607.
and relation to ruled surfaces : Z.23.
of revolution : N.72.
Hyperboloidic projection of a cubic
" gobba " : An.63.
Hypercycles : 0.94s.
Hyperdeterminants : CD.9 : J.34j,42.
Hyper-elliptic functions : A.16 : AJ.53,
7 : An.70 : C.408,622,67,92,94,97 :
CD.3 : J.25,27,30,40,47,52,54,75,
76,81,85: L.54: M.3,11,13 : Q.15,19.
of 1st order : J.12,162,35,98.
containing transcendents of 2nd
and 3rd kind : J.82.
multiplication of : Ac.3 : M.17,20.
transformation of : Ac.3 : {p = 2),
M.15.
transf. of 2nd degree : M.9 : Mo.66.
transf. of 3rd degree : M.l,19s.
transf. of 5th degree : M. 16,1 7,20.
of 3rd order (p = 4) : M.12.
of 1st order and 3rd kind : J.65,68,88.
of 1st and 2nd kind : An.582: J.93 :
in series, M.9.
of 3rd kind, exchangeability of para
meter and argument : J.31.
of nth order, algebraic relations : C.
993.
Gopel's relation : An.82.
addition theory : M.7.
addition th. for 1st order in a system
of confocal quadrics : M.22.
approximation to : P.60,62.
choice of moduli : C.88.
division of: C.68,98 : L.43 : M.l.
bisection: 0.702 : trisection, An.76 :
M.2.
generalisation of : C.84,98.
geo. representation : L.78.
inversion of : C.99 : J.70.
in logarithmic algebraic functions :
M.ll.
and mechanics : J.56.
periodicity moduli : A.68 : An.70.
periodic: J.32: of the 1st class, LM.
12; with four periods, An.71.
with quartic curves, 4 tables : M.10.
reduction of, to elliptic integrals :
Ac.4: C.856,93,99: J.55,76,79,86,
89: M.15: TI.25.
transformation of 2nd order, which,
applied twice in succession, pro
duces the duplication : C.88.
transformation : M.7,prl3.
of two arguments, complex mult, of:
M.21.
Hyper-elliptic ^-functions, alg. charac
teristics : M.25.

887

Hyper-Fuchsian functions from hypergeometric series of two variables :

U.99.
Hyper-Fuchsian groups : C.98.J.
Hyper-Jacobian surfaces and curves :
LM.9 : P.77 : Pr.26.
Hyper-geometric functions or series :
291: A.55,57: J.152,75 : M.3 : Q.
16: Z.8,26,27.
as continued fractions : 2912 : J.66.
of two variables : C.90,,91,95 : L.82.2,
84.
extension of Riemann's problem :
C.90.
of nth order: C.96 : J.71,72 : M.2.
and Jacobi's polynomials : 0.89.
square of : J.3.
Hyper-geometric integrals : J.73 : Z.22.
Hypocycloid : 5266.
with 2 cusps : Z.19.
with 3 cusps : J.64 : Me.83 : N.703,75.
Hypsometric tables of Bessel : Pr.12.
Icosahedron : 907: M.12,25 : and star
dodecahedrons, Z.18.
Icosian game : Q.5.
Imaginary :
quantities : 223 : A.20,
22: 8 square, AJ.4 : 0.18,24,25,
88,94: JP.23: N.63,64: P.1,6,31.
ap. to primitive functions of some
derived functions : N.63.
*
conjugates : 223 : modulus of, 227.
*
logarithm of : 2214: LM.2.
curves : Q.7.
exponents : A.6.
integrals of d. e : 0.23.
prime factors of complex numbers
formed from the roots of irreduci
ble rational equations : Z.10.
transformation of coordinates : Q.7.
variables : C963 : polygons of, 0.92.
^^0+16 in the form x + iy : A.55.
tan-1 (f
in the form x+iy : A.49.
<t>(x,y)+i<l>(x,y) = F (x + iy), to deter
mine <p and 4> : A.10.
* geometry : 4916 : A.32,,61 : C.61 :
CD.7,8: J.55,70: M.ll: Me.81:
N.70,,,725: TE.16.
of Lobatschewski : G.5: J.17 : N.
683.
of Stand t : M.8.
use in geometrical drawing : J.l.
* circular points at infinity : 4717, 4918,
4935 : tg.eq4998, 5001 : Me.68 : Q.
3,8,32.
* coordinates : 4761 : 0.75 : Man.79 :
homog.Q.18.
elements in geometrical constructions,
and apparent uncertainty there
from : Z.12.

888

INDEX.

Imaginary(continued) :
* lines through imaginary points: 4761,
47223.
problem, Newton-Fourier : AJ.2.
* tangents through the focus of aconic:
47201,5008: A.22.
variables, generating polygons of a
relation between several : JP.30.
x + iy = V-X+ilT, and the lemniscatic coordinates of the nth order.
Implexes of surfaces : C.80..
Implicit functions :
of one independ
ent variable, <p (xy) ; values of <f>, ,
<P, <PSl : 1700-6.
*
the same when (x,y) = 0 is also
given: 17189.
*
Vir, Vu when <p (x,y) = 0 : 170 7
16 : ynx, An.58.
*
*.(.*')> fcKy,): 1720l.
*
<Pt (*, y< z> ) when 3 eqs. connect
x,y,z,: 1723.
* of two independent variables :
yTZ when <p (as, y, z) 0: 1728.
*
<p, (x y z), fjr, <p'--, when + (as, y, z)
= 0: 172932.
*
<f> (x, y, z, , i)) when 3 eqs. connect
*,y,<t,(,V- 1735.
* of n independent variables : 1737 :
An.58.
trausf. into isotropic means and trig.
series : C.38.
transf. into explicit functions : C.38.
denned by an alg. eq. : C.47.
determined by the infinitesimal cal
culus : C.34.
Increment : 1484.
Incommensurable :
numbers: JP.
15: N.43.
limits of numbers : N.81.
lines : N.44: in ratio i/S : 1, A.3.
Indeterminate coefficients: 232,1527:
A.3: J.5.
Indeterminate equations (see also
" Numbers " and " Partition ") :
18894: C.10,th78,88 : G.5 : J.92.
Mem.44 : N.44,45,pr57,5971,78,
81,prs81,85 : TE.2.
ap. to a geo. problem : M6m.20 : Z.20.
impossible class of : N.63.
linear : JP.13 : L.41 : N.43 : P.61 :
Z.19.
*
with 2 unknowns : 188 93 : A.3,7 :
J.42: L.63,69: Mem.31.
*
with 3 unknowns : 194 : G.2.
with n unknowns : C.94 : N.52.
xi+2xi+...+n%n = m: G.l.

Indeterminate equations{continued) :
and congruences : Pr.ll.
quadric : J.45 : in n unknowns, N.84.
quartic : geo.cnL.63.
quintic: J.3.
quadratics in two unknown integers :
se2ay" = 1 : A.12,52 : E.23,28 :
J.17 ; by trig. : L.64-66 : N.78.
x*-ay2=b : C.69 : L.37,38 : Mem.
28.
*>ay' = 4, a= 5 (mod 8) : J.53.
x1 + y, = (a? + 5')*: C.36: An.53.
x? + y" = 0 : geoA.55.
(n + 4)x--nif =4: N.83.
axr + bx = i/:: L.76.
ax? + bx + c =*t/2: G.7.
2*s + 2* + l y': N.78.
ax" + bxy + cif = 0 : geoC.9.
x' + nxy-ny^l : N.83.
x1if = xy impossible: N.46.
ax3 + bxy + cy" + dx + cy +/'= 0 : C.87.
quadratics in three unknown integers :
x? + y1 + 2* = 0 : geoA.55.
x-> + y" = z* : A.22,33 : E.30 : G.19 :
solution prior to Diophantus,
C.28s.
x> + ay~z': N.78.
a*' + by2 = z': G.8
x* + ay" = z : N.78.
a;5ai/-'=4z : N.72.
x' + a(x + b)'=y: N.78,.
(x- + hy") z ax + bky : J.49.
(o, b, c, d, e,f3tyz)- = t : Pr.13.
*s+f/5+16z!=n5: Mel.4; =4n + l,
L.70.
quadratics in four unknown integers :
flj' + j/'iz^f5: C.66:N.48; x* + 2y*
+ 3z3 = <, L.69.
i/5 = z' + <(z+/3);: N.78,.
quadratics in five uukuown integers
x: + by1 + cz3 + dP = u, with the follow
ing values of b, c, d : 1,1,1 ; 2,3,6
L.45: 1,1,2; 1,1,4; 1,1,8; 2,2,2
2,4,8 ; 4,4,4 ; 3,4,12 ; L.61 : 1,2,4
1,4,8; 2,2,4; 2,4,4; 2,8,8; 4,4,8
4,4,16; 4,16,16; 8,8,8; 8.8,16
8,16,16; 16,16,16; L.62 : 2,3,3
3,a,3a; L.66 : 1,3,3; L.60 63
1,1,3; 1,2,6; 2,2,3; 2,4,6; 4,4
12; 1,1,12; 2,2,12; 1,4,12; 1
3,4; 3,4,4; 4,12,16; 3,6,6; 3,3,3
3,3,12; 3,12,12; 12,12,12 ; L.63

INDEX.

Indeterminate equations{continued) :
1,1,5; 2,3,6; 5,5,5; L.64 : 1,5,5;
1,6,6; 1,9,9; l,n,n; 2,,2re ; L.65,
59 : with c = a&, 0.42 and L.56.
ax- + by- + cz3 + <ii-=,with the followingvalues of o, b,e,d: 2,2,3,4 ; 2,3,
3,6; 3,3,3,4; 1,2,6,6; 2,3,4,4;
L.66: 2,2,3,3; L.65: 3,4,4,4;
3,4,12,48; L.63.
ic3 + J/2 + 23 + <2 = 4lt: L.56.
ax- + ft;/3 + cz3 + dt- + exy +fzt = U, with
the following values of a, b, c, d,
e,f: 1,2,-2,2,1,2; L.63: 1,2,3,3,
1,3; 2,2,3,3,2,3 ; 1,1,6,6,1,6 ; L.64 :
2,3,2,3,2,2; 2,5,2,5,2,2; L.64: 1,1,
2,2,0,2; 1,1,1,1,0,1; 1,1,1,1,1,1;
2,2,3,3,2,0; 1,1,3,3,1,0; L.63; 3,5,
10,10,0,10; 2,3,15,15,2,0; 2,3,3,3,
2,0; L.66.
*3 + 2y + 2z3 + 3*3 + 2yz u: L.64.
2.c3 + 3i/3 + 3z3 + 3*3 + 2>/z=M: L.66.
z?+y'2 + zZ + 2u? + 2uv + 2v!S + P = w.
L.64.
z3 + y + 2z3 + 2zl + 2& + 3ii3 + 3u3 =w :
L.64.
2 (a3 + xj/ + 2/2) + 3 (za + p + J43 + 2)=MJ :
L.64.
aj/ + j/z + z< + <M u : C.622 : L.67.
y3 = Zi+x\+ ...+Xn-. G.7.
quadratics in soven unknown inte
gers :
x3 + ay2 + bz' + ct~ + du2 + ov- = w, with
the following values of a, b, c,d,e:
4,4,4,4,4; 1,4,4,4,4; 2,2,4,4,4; 1,1,
4,4,4; 1,2,2,4,4; 1,1,1,4,4; 1,1,2,
2,4; 1,1,1,1,4; 4,4,4,4,16; L.65:
1,1,1,1,1; 1,1,1,1,2; 1,1,1,2,2;
1,2,2,2,2; 2,2,2,2,2; 2,2,2,2,4; 3,3,
3,3,3: L.64.
higher degrees :
cubic : AJ.2.
a$ = y*+a: N.7a2,83: with a = 17,
N.77.
-x+y3 = as . N.78j,80.
x3 + j/3 + z3 + 3 = 0: a.49.
ax*+by* = z2-. C.87,91,94: N.79:
a = 7, 6 = -5, L.79.
icrl2"'j/4 = z3 and similar eqs : L.53.
x*+ax2yi+y* = z3 : Mcm.20.
axi+bxhjt+cyi+dxhj + exy* =/z2 :
C883.
x -f j/5 = az5 : L.43.
x"+yl = zl, impossible: C823 :
L.40v

889

Indeterminate equations(continued) :
w-tic solution by alg. identities :
0.87,.
"+?/" = z" impossible if n>2 (Fermat's last th.) : A.26,58 : An.64:
C.24,89,902,98 : J.17.
x2"if" = 2x": L.40.
axm+bym = cz" : N.794.
xm y*+l, impossible : N.50,70,71a3aj/3 = z: C.99.
simultaneous :
x = u-; a; + l=2y3; 2x+l =3m;2 :
N.78.
a^4-a=^; j/3-a = a2 : An.55: C.78.
afi+x+2 = y*; &x-2z=z~; N.76.
aa;+&)/+cz = 0 ; Ayz+Bzx+Cxy
= 0: A.28.
xP+ytz"~ = a ; x*2/2+z3= ;
x?+y*+z* = : E.20.
x2+axy+y*= ; j/3+aj/z+z3= ;
z2+azx+x2= : B.20,21.
x+y+z = a ;
*2+i/34-z3= ;
Z>i+y3+z3=n : E.17.
six eqs. in nine unknowns : N.50.
exponential, xy y' : A.6 : Z.23.
a'-b* = l: N.57.
Indeterminate forms : 158093 : A.26 :
AJ.l exponential : J.l : Me.75 :
N.48,77 : ~, A.21 y'L.l:; L.41,
0
00
*
42 : N.46
when x = 00 , 1592:
0, J.11,12: Oexp 0*, J.6.
* with two variables : 1592a.
Indeterminate multipliers : 213, 1862,
3346: N.47.
Index law : 1490.
Indian arithmetic, th : L.57 : calculation
of sines, N.54.
Indicatrix: 5795: C.92 : Mo.72: N.74.
* a rect. hyperbola, condition : 5824.
* two coinciding lines : 5825.
* an umbilicus : 5819.
to determine its axes : L.78,82.
determination of a surface from the
indicatrix : A.59.
Indices : 29 : N.766,77.,,78.
in relation to conies : N.722.
of functions, calculus of : JP.15.
Induction : 233 : C.39 : G.15 : L.48.
Inequalities: 33041: A.1,24.
in integrals: Mdl.3 : f.d .c. Mdm.59.
(|n)3>7: N.60.
5 x

800

INDEX.

Inequalities(continued) :
a* >x . A.14.
if a+ya 22, xm+ym < z*: A.20.
geo. mean of n numbers < arith.
mean : 332 : N.42.
Infinite :
equalities : M.10 : prG.22.
functions : An.71 : J.54.
from gnomonic projection : Me.66.
linear point-maniloldness : M.15,17,
20,212,23.
point-mass : M.23.
products: J.27 : N.69.
value expressed by r functions :
Ac.3.
exhibiting circular arcs, logarithms
and elliptic functions of the 1st
kind : J.73 : Ac.4.
use of in mathematics : C.73.
Infinitesimal calculus : 1407 : M.11,18.
Infinitesimal geometry : An.59 : C.82.
of a surface, formula? : G.13.
Infinity, points at on alg. surfaces : C.
59.
Inflexional tangents : 5789 : A.35.
of a cubic curve : E.30 : J.38,58.
Inflexion curves : Z.10.
Inflexion points : 5175: CM.4: J.41.
of cubic curves : J.28 : axis, B.31.
Herse's equation : N.81.
Inscribed figures :
In-circle :
of a quadrilateral, locus of
centre: A.52.
of a triangle : 709, 953, 474750 : tg.e
4889: CM.l.

in-centre : 709, t.c4629, tg.c4882.

In-conic :
of a circle : thj.91.
four of a couic : prJ.39.
of a developable quartic : An.59.
of a polygon : M.25.
of a quadric : J.41.

of a quadrilateral : tg.e4907 : N.63 :

four, N.56.

of a triangle : 473946, tg.e4887 :

A.2: N.50; max,A.8: (J.2 : lat.
rect., E.34.
In-cubic of a pencil of six lines : Q.9.
In-hexagon :
of a circle : A.22.
of a conic : 4781 : N.57,82.
In-parabola of a triangle : CD.7.
In-pentahedron of a cubic : Ac.5 : M.5.
In-polvgons :
of a circle: 746: CM.l:
'J.35: N.50.
regular of 15,30,60,120, Ac., sides :
A.62.
do. 9 and 11 sides : LM.10.
do. 17 sides: TI.13.
do. four of 30 sides : N.78.
do. 5, 6 and 10 sides, relation : A.40,
43,45,48.

In-polygons{continued) :
two stars, one double the other :
A.61.
of a circle and conic (Poncelet) : G.l.
* of a conic : 4822 : thsN.47 : cnTN.69.
*
with sides through given points,
cn : 4823 : An.51-.
semi-regular: N.63.
of a cubic (Steiner) : M.24.
of a curve : Q.7.
of a polygon, th : CD.5.
of a quadric with sides through given
points : LM.2s.
In-quadrics :
of a developable : Q. 10 :
quartic, An.59.
6 of a quadric, 2 touching 4 : An.69.
In-quadrilateral :
of a circle : 733 :
A.5 : cnE.21 : area, N.44 : P.14.
* of a conic : 4709.
of a cubic : N.84.
In-sphere of a tetrahedron : A.61.
In-spherical quadrilateral : N.49.
In-square : of a circle : J.32.
of a quadrilateral : A.6.
In-triangles :
of a circle : P.71.
with sides through given points :
J.45: N.44.
of a conic : J.7 : Maccullagh's th, N.65.
with given centroid : G.23.
similar : A.9.
of a triangle : thsQ.21.
two (Steiner's " Gegenpunkte ") :
J.62.
In- and circum-circlcs :
of a poly
gon : N.45.
distance of centres : A.32.
of a quadrilateral : Fuss's prMel.3.
* of a triangle: 935: A.38.
*
distance of centres : 936, 4972 : eq,
4644.
In- and circum-conics : of a pentagon:
N.78^
of a polygon: J.64,70: regular, Z.14.
of a quadrilateral : 60 theorems, N.
45,: N.76.
of a self-conjugate triangle : Me.81.
* of a triangle: 4724, 4739: An.52:
G.22,.
In- and circum-heptagons of a conic :
A.3.
In- and circum-pentagon : of a circle :
A.22,43.
In- and circum - polygons (see also
"Regular polygons"):
of a
circle : L.16 : N.80 : P.ll : Q.ll.
*
sum of squares of perps., &c. : ths
1099.
difference of perimeters, ths : N.43j.
of two circles, respectively : C.63 :
G.21 : L.78.

INDEX.

891

In- and circum -polygons(continued):

Integrals or Integration (continued) :
of a conic : A.4 : ellipse An.52 : An.
from orthogonal surfaces' theory :
57 : J.64 : N.57,84
L.38.
of two conies : C.90.
by Pfaff's method : A.47.
of a curve : C.78.
by series : Me.71.
of a homonymous polygon : A.50.
by substitution : A.18.
by Tchebychef's method: L.74 : M.5.
In- and circum-quadrics of a tetra
hedron : eqsN.65.
comparison of transcendents : Me.79 :
In- and circum-quadrilaterals :
of a
Pr.8.
complex, representing products and
circle : A.48.
powers of a definite integral : J.
of a conic : 4709 : and pentagon ths,
N.48.
48.
connected with trinomial integrals :
In- and circum-spheres :
of a tetra
hedron: N.73.
L.55.
of a regular polygon : A.32.
convergency of : M.13.
* of a regular polyhedron : 910.
definite :
applied to Euler's, &c. :
In- and circum-triangles :
of a circle
J.16.
(Castillon's pr) : Q.3.
with finite differences : J.12.
equilateral, of another triangle : Me.
from indefinite : J.41,51,52.
whose derivatives involve explicit
74.
functions of the same variable :
and square of an ellipse : A.30.
* of two conies : 4970 : N.80.
C.12.
*
envelope of base : 4997.
determination of functions under the
respectively of two conies having a
sign J: JP.15.
common pole and axis : CD.4.
Instantaneous centre : 5243.
* difference between a sum and an
Interest : 296301 : and insurance, A.
integral: 2230: G.9.
division into others of smaller inter
26.
Integrability of functions : An.50,73 :
vals : A.4.
* eight rules for definite integration :
0.28 : J.59,79 : JP.17 : L,492.
criterion for max. and min. values of
2245.
equations for obtaining functions as
a primitive : An.52.
Integral calculus : 19002997 : A.ext
integrals : J.79.
expressible only by logarithms : An.
18,26: Euler's, A.20 : C.14,42:
Newton, CD.8: G.19: L.47: Me.
76.
extended independently of the con
72,74,75: Mem.18,36.
ception of differentials : A.61.
paradoxes : C.44.
formulas of: A.l : J.18,19 : M.4: Mo.
* theorems, &c. : 270042: A.45: C.
76 : Mo.85 : N.85 : failure of f,
13 : L.geo.ap50,56 : Me.77 : Mem.
prsl5,30.
CM.2.
and gamma-function : LM.12 : Z.9,12.
Integral functions : C.88,89,98 : G.4,22..:
higher, of composite functions : A.20.
h.c.fofG.2.
with imaginary limits : C.23: J.37.
with binomial divisors : J.70.
use of imaginaries in : M.14.
and continued fractions : An.77.
inverse method: CM.4: CP.4,5: L.78.
reciprocal relation of : A.67.
Integrals or Integration : 1908 : A.1,2,
involving elliptic functions : Pr.29 :
4,5,6,10,23 : Ac.1,32,44 : C.90 :
Q.19.
CD.9: CP.3: J.2,4,8,17,25,36,39, * limits of: 2233-44: L.74.
multiplication of : Pr.23.
61,92 : JP.9,10,11 : L.39 : M.6,16,
number of linear independent of 1st
73,75: Mem .31 : P.14,36,37 : Pr.
kind: An.82.
7,39: Q.11,13: Z.7,ll,15,18>2-2;,23.
of alg. differentials by means of
* approximation to : 2127, 2262, 2991 :
logarithms: Mo.57: An.75: C.
A.9,14: C.97: CM.2: G.3: J.l,
90;: J.12,24,78,79: Mo.84: N.81
16,18,37,48: L.80.
(see " Integrals ").
*
Gauss's: 2997: C.84: M.25.
of algebraic surfaces: C993: J.26 :
by the principle of Abel's derivative :
' octic, An.52 : cubature, C.80.
J.23.
* by differentiating under the sign of * of circular functions : 193897 : 2453
2522: No.1799: LM.4: M.6:
integration : 2258.
by elliptic functions : G.ll : L.46.
Mcm.9.

892

INDEX.

Integrals or Integration(continued) :
sine and cosine : G.6 : M.U.
* of exponential and logarithmic func
tions : 2391-2431: E.17,18.
* of circular, logarithmic and expo
nential functions : 2571 2643
(see " Integrals ").
of a complex function: A. 66: th of
Cauchy, Ac.84.
* of a closed curve : 5204 : C.23 : E.28 :
Z.17.
of differentials containing the square
root of a cubic or biquadratic :
Me.57.
* of discontinuous functions : 2252 :
C.23 : LM.6.
of dynamics : L.52,55,58.
of explicit functions, determination of
algebraic part of result : An.61.
* of functions which become infinite
between the limits : 2240 : J.20 :
JP.ll: Q.6.
of infinite relations : M.14.
of irrational alg. curves by loga
rithms : An.61.
* of irrational functions : 211020:
AJ.2 : An.56 : C.32j,89 : L.53,64 :
Hem.30.
* limits of : 1903,-6, 2233, 2775.
* for quadrature of curves : 520511 :
C.68,70 : circle, J.21,23 : JP.27.
triple integrals : J.59.
* of rational functions : 2021- 32, 2071
2103 : L.27 : N.73.
* of rational fractions : 1915 : CD.3 :
Mdm.33;.: N.72.
of total differentials : C993.
of transcendental functions : JP.26.
of two-membered complete differen
tials : J.54.
periods of : C.36,38,75;. : G.753 : JP.274.
* principal values of infinite definite :
2240: A.68.
properties by elliptic coordinates :
L.51.
quotient of two d.i of the form
JcZas dy,..dz : J.67.
reduction to elliptic functions : An.
60 : LH.12 : Me.77,78.
residues of : JP.27.
Kiemann's of first kind : An.79.
singular values of: A.ll.
* successive : 2148 : L.62 : 2nd order,
M.20 and Z.ll.
* summation of : 2250 : J.47 : JP.12,21.
tables of definite, by B. de llaan, note
on: 0.47.
and Taylor's theorem : Me.84.
theorems : L.48 : P.55 : Q.10,12.

Integrals or Integration(continued) :
* transformation of : 224552: A.10 :
CM.4 : J.fl5,22,36 : L.36 : Mel.3 :
Ql.
* variation of arbitrary constant : 2247 :
J.33.
whose values are algebraic : J.10 :
JP.14: L.38.
algebraic FUNCTIONS. Indefinite :
unclassified: An.75 : C.903: J.12,24,
78,79: Mo.84: N.81.
* simple functions of x2a- : 192637 :
A.4 : V{a*-tf), A.38 :
N.82:

An.68.

fractions involving a binomial surd :

200819.
1
: Mem.13.
V(2**-l)
(z"a) </(x"b) : Q.18.
y/(l+a*) : 20 K>: L.80: Z.8.
a*-)2
Mem.10.
(lss)(l + 6a:2-|-at*)
(*S+8) V(&-1) ' A"3'
and deductions : 20218.
!C"1
zm (a+btry-. 203560 : A.36 : Mem.ll.
1
20615: J.36.
x*+a?
functions of (a + bx+cx-) : 207180:
21039.
f
^ 55
S/(a + bx + eS-)"'
functions of (a + bx*+cx*) : 20815.
functions of (a + bx'+cx2") : 2086
2102.
rational algebraic functions of irra
tional ones :
integrated by rationalizing: 2110
20.
reducible to elliptic integrals, viz. :
rational functions of v/X, y/A'j, and
where X is a quartic in x ; 2121
41.
F{x, yx,): 2121 : LM.8.: L.57.
^: 21336: J.36: LM.14.

INDEX.

Integrals :
algebraic(continued) :
^ : L.64 : MeX3.

893

Integrals :
algebbaic(continued) :
* x**F{(a*~
(Cauchy): 2712:
A.9.

4=r : C.59 : CD.l : E.36 : J.10.

-: A.12: E.41.
l+2a;cos$4-a;2'
deductions from this involving in
to grills of the forms

-ip: 2141 : J.17: Mati.79.

VAl
^(v/X3): L.57.

Me.82.
(1-a?)1
y-iM^1 &c- reduced t0

dx and
J

n_ . A 35
J " sin nir

OiAer limits :
f1
f(!)& Jf4 F(x)dx
J-,^(l-*2)'
(a;-c) : 23689.

Jacobi's functions : Q.18.

M
. j.32.

n /a
(a 6a:) '^-'da: : A.35.

sundry: CM.2 : L.47.

Limits 0 to 1 :
Euler's : see " Euler's integrals."
* it'"1 (1-a)"'-1 : 2280: A.40 : C.16 :
J.ll,173.
* g'-'+g 1 2341 o 55
(l+s)''
* similar forms: 23424, 2352, 2356
-67.
': 2367: A.10.
1X 1 B"
xm (!)"
L.59.
x"(lx)"

L.57.

sin a
d=f(a): geoN.85.
il2a; cosa+a;2
CIRCULAR FUNCTIONS. Indefinite :
* sin x, sin_1a!, &c: 193849: sin"a:,&c.,
19547 : cos"a:, N.74.

*
*
*
*

s^'<1-^l:L.56-7.
1
: J.42.
(abx)" (1a;)1-"^"
Limits 0 to <x>:
*

-^-,&c.:230912, 234555: A.38:

1 as
J.24: L.41 : Q.12: Z.19.
1
: 2364.
('^+a2)
X"x-': Me.83.
1+S^,2n >a>0: N-48l+x-+...+x*i=T,' *" A-16K"->(1
"1 -J-a;2) ' and 12 similar ' A.35.

1
1951: J.9;
1
(a+ftcosx)"'
a+6cosas'
1958 : Me.80.
products and quotients of sine and
cosine and their powers : 1959
80, 206670 : A.49.
binom. funcs. of sin and cos : 198292.
ditto of tangent : 1983, 1991.
F
,<*>*) . 1994_7. A-12 .
a cos x + b sin x+c
J.19,32.
(a-f2&cosas+ccos2a;)-1: 1993;
, 2029.
a2"2,c*' cos n8+l
F (cos a:)
1996.
(a, + ii cos x)(a.2+b2 cos x)...&c.
a function of sine or cosine in a ra
pidly converging series, and suc
cessive integration of it : J .4,15.
sinnd>
: A.ll.
(l-k sin </>) v/fl-fc2 cos^)
(sin arnxf" -. J.81.
^{1fc2 sin2 i (a+x) sin2 i (*)} .
J.39.
sin^ : A.17 : with m = 1, G.7.

894

INDEX.

Integrals :
circular. Indef.(cont.):
m
sin he /() (Fourier): 272642:
z or Bin *
A.3&2.
Limits 0 to :
2
* sin" as: 2453- 5, 2458, 2472: E.29:
n = i, E.28.
* tan1 'as: 2457.
* sin" x cos" x : 245965.
* ma' x cos m : 2481 2484_92 : L-43sin _ * an(j 8jrai]ar. a.59.
cos"
sinz
* x oos" " 2 z sin nx . 2494.
1
and similar: 2496
a2 cos2z + fc2 sin2x
2501, 2344.
* cos,,""~1zsin""1 x cos rpz:and
sin'-'asinpz: 25858.
* at cot ax : A.34 : No.19 ;
xtan4ir(l bx): 5340.
*
*
*
*

Limits 0 to n :
sin"z cos^x : 2459.
sin nz8lnux: 24679.
cos-^
f sin Yx sin nx : 247482, 2493.
V cos / cos
sin!"3id/(cosz) : 2495.
cos n (xa sin x) : L.41 : 0 to 2>r, C.39.
cos (a+nx) cos (b+px) cos (e+qx) : C.
54.
-: Pr.25: with a = l, 2506
l+a2cos2z'
and Q.11.
z sinz or sing sinrx or cos rz
-9.
1 2a cosx+a2
sin2"ic or cos
px Q-H:LM.11:=1,
, , rn-ii
i
nS
T^i:
(12a cosx+a2)"
L.74.
zF(sinz cosz) : JP.27.
about 250 integrals with limits chiefly
from 0 to ir, some from 0 to oo :
Pr.25,,26,27,29,30,31 ,32,,33.
Limits 0 to oo :
-, cosx2, cos (ox)2: 25079: Q.
12:
(sinY z: Q.13; ^z": Me.72.
\cos/

Integrals:
circular. 0 to oo (cont.):
sin bx: 2579-81.
cos
cos bx .alsof'^rfz: A.10.
x+a
J. X
2510: A.30: E.26: Z.5: n = 2
x"
and 3, 25112.
# cosgz-cospz, 2513_5
z or z2
sinaxsin&z and, similar
.
: nE1
2ol422.
.
# cos2az suVx 2722 5
* (sin az, cos fcz), reduction of : J.I5.
(p (sin ax, cos fcz) . j 0g 25
: 2572 : L.49 : Z.7,8.
cosrx : 2573: A. 10,1 1,59: J.33 : Me.
a2+*'
72,76: Z.7,8.
: 2575 : A.10 : J.33.
a2+x2
!-: L.40: Q.18.
(a2+z2)'
<b(n) am rx
r v ' cos
: All : with <(*) = *-,
a2+z2
L.46: <t> (x) = tan-1 cx, All.
and similar : Z.13.
a2+*2
sin ax
and related integrals :
(l+x2)sin6z
Ac.7.
1cos"
_Q
z2 x = Z7,a
1 or z sin car
( 1 + z2) (1 2a cos + a2) ' 26302.
* tan"1 da: 0rAq 1 4 _i
i
ztl+z2)
z2
a
b
2504.
# tan"' ax tan-1 bx : 2505.
Other limits :
\TT <fa, : . Hm. 0 to 2ir:
Jo a + b cos k+c sin x
A.55.
* f'tan,"'a;az, deduced from (2416):
Jo 1+z
L.69.

INDEX.

Integrals :
circular(continued) :
['tan-1 ax dx 2502.
Jo x^(lx?)
xJ(\
E^1
m X"^ dx, where F (x) is a rational
integral circular function : OM.3.
sine-integral, &c. : see " Functions."
exponential FUNCTIONS. Indefinite :
* <?,a': 1924;
2004.
* ez {<p(x) + <t>'(x)}: 1998.
eexp(^): E.34.
X being a rational integ. funct. of x :
e*X : J.13;
e*, Mdm.83 ; e exp
(-*?), Q.l.
cexp(*p
th
x or
^/(a+bx^)
+ (c+dx2) : L.52.
Limits 0 to co :
* e-u x" : 228491 : sec " Gamma
function."
* e exp (fcx2) : 2425 : evaluation by a
continued fraction, J.12.
* other forms : 2426-31, 2595, '8, 2601.
eexp (^-x^-~ ) : 26045: L. 56.

895

Integrals :
logarithmic. Indef.
(continued) :
x log a: .
j several: M6m.l8.
(1*")'' JP
Limits 0 to 1 :
* log-^-: 2284; a* flog V, 2291.
x
\
x j
* involving log x or log (1 as) : 2391
2403, 241622.
* !3|(l2l. 2416: C.59: J.6: L.43,44.
1+x3
togs . 06% O 12 g""'loKic
^(1
and many cases, J.34.
* _L
x logB 12
1x : 2403 : L.73.
"(g-1>': Z.3; f=*,A.37.
log x
log to
about 540 expressions chiefly formed
from log (1 x) or log (1 x)
orlog(l + z+x2) orlog(l+x2+x4),
joined to a single factor of the
form x or (1 as") or \ g
1 =F x'"
with integral values of m and n :
A.39,40.
about 280 expressions, nearly all
comprised in the form

: Z.6 ; ditto X xm, L.46.

l+x?
, and the same X x : A. 10.
-*<(): M61.2.
e exp (ax") . Ei ( ax') : Q.18.
Other limits :
expon-integral : see " Functions."
| e exp (x2) [ J?r/C f (Glaisher)] :

a-U-x")5 (logx)',
with integral values of to, n, p, q,
and t : A.40.
about 130 expressions of the forms,
xm (1x")' (logx)2<"3,
xm(l x") p logx, and

Me. 76.
f' 8-*2rd: C.774.

values of to, ,
and r : A.40.
Zimite 0 to oo : 24234.
Other limits :
log (1-a!) : 240812 : L.73 : Zl.
x
i log |-K, limits 0 to
: 2415.
x
1X
log-integral : see " Functions."
circular-exponential FUNCTIONS. In
definite :
eM (am Ybx: 1999; C sin" x cos" x,
\cos )
2000.

^ log x, with integral

oxp(-^-g):0.12.
*
Flre'Oe-"*': A.15.
*
logarithmic functions. Indefinite:
* logx: 1950; as" (logx)**, 20036.
x-loga+xj.&c.: A.39.
F(x)\ogx: G.12.
Z<(log^'" and a similar form : 20302.
x 1

INDEX.

896

Integrals :
circular-exponential
(continued) :
(e" sin ic)""1 : L.74.

Integrals :
circulajl logarithmic
(continued) :
log sa x and log ^/(l-& sin2 *), each

Limits, 0 to M :
e""ain": 2571,2591.
as
g sin 6k : 2583 : 0 to 1, Mcm.30.
e~"x" sin bx : 2577, 2589 : J.33 : Z.7.
cos
e-" lsin Vx: 2608-11 : A.7 :
\cos /
with limits- to
2612.

with the above denominator : J.92.

Limits 0 to ir :
* log (12a cos a;+a2) : 26202 : L.38 :
Q.11.
* cosrv:log(l 200032+02): 2625: A
13.
* * log sin x : 2637 ; x log sin2 x, 2638.
Limits 0 to 00 :
cos a* n
...

e"+e sin mas, &c. : 25932600.

e"** cos nus sin" a; : 271720.
* cexp(a-j;2)^0^ 2&j;, and similar:
2614-8: Z.1,10.
* c cxp ^as + ^ )cos 0 j X &c. :
2606: Q.12.
* COS*-6- . 261Q
as
CIRCULAR-LOGARITHMIC FUNCTIONS :
*
'(logg)2033.
as*'2as" cosntf+1
Limits 0 <o 1 :
* cos ^ x^ ^
x' an<^ s'm^ar : ^41
3.
tog(l-2*cog+sS) 4o_i ^ . A 34
x
,log 1 2.C ;COS tJ-1,&
a + 353 iq

T n
lo similar
: Ij./<32.
a; (I*2)
log Bins; log cos x, &c. [ limits Qto^-y.
E.22.
V

* log(l 2ocoscj!+a2) 2g3i

1+a;2
Other limits :
'a+b cos x dx: A.53.
1 a 6 cos a;
F (cos nx) log sin ~ dx : Z.10.
)cos,!"-1 ]x log tan x dx : A.16.
fl
"log(l2(icosa;+a;2)<it, and
similar: J.403.
EXPONENTIAL-LOGARITHMIC FUNCTIONS :
[~er'-e~"log(x*+u2)dx, &C: J.38.
] e"
J"
'- log (12a cos x+a?) dx: J.40.
c
MISCELLANEOUS THEOREMS :
* formal of Frullani, Poisson, Abel,
Kummer, Cauchy, &c. : 270013.
/ =/(6)-/(a): 1901-3: A.16.
(* t M~* <H dx: Me.81 : n =0,2700.
Jo
a:"
l4>(x){+(x)}**dx: 2001-2.
fZMoV
(5 of Taylor's th) : LM.13.
Jo
xr

Limits 0 <o -jj-:

J
* log Bin as : 2635 : CM.2 : E.23 : Q.12 :
0 to jut and similar integrals, A.
16.
* log(l+ccos*) . 2633 : 0 to tt, 2634.
cos a;
tana; log coseca: : E.27.
Iog(l+2 sin2 a;). L 4g
v/(l-fc2sin2a:) '

J" (a:) da;, approx. to: C.97.

//(*-)
J

= f/fo) dy : CD.4.

0 ^a: + ^-&-dx, and similar:

Me.75,76,77.
\f(x)<t>(x)dx: L.49.
f ifvdV. * I udx [ udV : AJ.7.
Jo
Jo
Jo

INDEX.

Integrals :
miscellaneous
(continued) :
(u). du : Z.25.

th.

jo (u+k)kti du = (1)* J (u)k*tdu:

A.38.
Jb P (a;) Pn (x) dx (P, a; a Legendre's
coefE.) : Pr.23.

<J = pos integer:

! ^"(ae+iJ+j)'"1'
C.78.
fJ ? (sin 2*) cos * dx
= I > (cos2 *) COS 3! <fo : A.21
J
L.&32.
J/(a;)sino>jra!(Z(log): Mo.85.
c?s x*sin2, , transf. of:
cos2
jo tt-r..- sin
J.36.
integrals deduced from
z'+Fz'+Qz = 0, and y"+
(P+2B)y'+{Q+B(P+B)+B'}
y = 0: J.18.
\n(x,y)dx: J.61 : Mo.61 : Q.7.
if f(x+iy) =P+iQ,ih: 2710.
[Upwards of 8,000 dofinito integrals have
been collected and arranged in a 4to
volume by D. Bicrens de Haan ;
Leyden, 1867 (B.M.C. : 8632. ft.)]
Integrating factors of d.e : pp.468471,
3394.
Integrator mechanical : 5450.
Intercalation : CM.3.
Intercepts, to find : 4115.
Interest: 296: N.48,61,64.
Interpolating functions : C.ll.
Interpolation : 3762 : A.32,61,62,70 :
AJ.2 : C.19,48,68,92 : J.5 : L.37,
46 : Me.78 : N.59,76 : Q.7,8.
of algebraic functions, Abel's th : J.28.
Cauchy's method : A.2 : C.37a : L.53.
by circular functions : N.85.
by cubic and quintic equations : C.25.
formulas: J.2 : C992 : Mo.65.

Lagrange's : 3768 : J.1,84 : N.57,61.

Newton's : N.57,61,71.
for odd and even functions : C.99.
and mechanical quadrature : 3772 :
A.20.

897

Interpolation (continued) :
by method of least squares : C.373 :
Mem.59.
* by a parabolic curve : 2992 : C.37.
Stirling's series : Me.68.
and summation : I.ll2,12,14.
tables : I.llj.
of values from observation : Mel.2 :
tr, Mem.59.
Intersection :
of circles and spheres :
L.38.
* of 2 conies : 4916 : CD.5,6 : N.66.
* of 2 curves : 4116, 4133 : CM.3 : J.15 :
L.54 : by rt. lines, Me.80.
* of 2 planes : 5528.
of 2 quadrics : C.62 : N.684.
of right line and conic : see " Right
line."
of successive loci, ths : N.42.
of surfaces : J.15: L.54: by rt. lines,
_ Me.80.
Invariable line, plane, conic.and quadric :
585666.
Invariants : 1628: An.54.,: C853: E.42 :
G.1,2,15: J.62,68,69: L.55,61,76:
M.3,5,17,19: Me.81: N.58,59j,69,
70 : P.82 : Pr.7 : Q.12 : Z.22.
of binary cubics : An.65.
of binary forms : of 8th deg., C.84 :
G.2 : M.5 : simultaneous, M.l.
of higher transformations : J.71.
superior limit to number of irre
ducible : C.86.
of a binary quadric : M.3.
* of a binary quantic: 1648: E.40 :
Me.79 : of two, 1650.
of a binary quartic : M.3 : Q.10.
of a binary quintic, table of irreduci
ble: AJ.l.
of a bi-ternary quadric : J.57.
* of a conic : 4417 : 49365030.
* of two conies : 4936 : N.75.
of three conies : Q.10.
* of particular conies : 4945.
of a correspondence : G.20.
differential: M.24 : of given order
and degree belonging to a binary
10-ic, C.89.
of d.e linear : C.88 : of 4th order, Ac.
3.
and covariants of /(*3, y2) relative to
linear transformation : G.17.
of linear transformations : M.20.
mutual relation of derived invariants :
J.85.
of an orthogonal transformation : J.
65: LM.13.
of a pair of homog. functions : Q.l.
partial : LM.2.
of points, lines, and surfaces : Q.4.
5 v

898

INDEX.

Invariants (continued) :
of a quadric : J.86 : of two, M.24 : Q.6.
of a quintic : of 12th order, Q.l : of
18th order, C.92 : J.59.
related to linear equations : C.94.
of sixth order: G.19.
skew, of binary quititics, soxtics, and
nonics, relations : AJ.l.
of ternary forms : G.19j.
transformation theorem : M.8 : Me.85.
Inverse calculus of differences : N.51.
Inverse equation of a curve or inverse
method of tangents : 5100 : No.
1780: J.26: Mem.9,26.
Inversion: 1000, 5212: An.59 : C.94,pr
90: LM.5: Me.66: thsN.61 : Pr.
34 : problems by Jacobi, J.89 :
geo. ths, Q.7.
formulae: An.53: Lagrange's, J.42,54.
of arithmetical identities : G.23.
* of a curve: 5212: U.4 : J.14 : Pr.14.
*
angle between radius and tangent:
5212,5219 : E.30.
of 2 non-intersecting circles into con
centric circles : E.39.
* of a plane curve : 5212 : G.4 : Pr.14.
of a quadric : J.52,76 : Q.ll.
of a system of functions : An.71.
and stereographic projection : E.35.
Involute : 5149,-53,-66.
* of a circle : 5306 : C.26 : successive,
E.34.
and evolute in space : CD.6.
integrals of oblique : C.85.
* of a tortuous curve : 5753.
Involution: 1066: A.55,63 : gzAn.59 :
At.63: CD.2: thCP.ll: thE.33 :
G.10,20: J.63: N.53,64,65.
and application to conies : A.4,5.
of a circle : Me.66.
of a cubic space and the resulting
complex : Z.24.
of higher degrees : C993 : JM.72 : of
3rd and 4th, An.84 : Z.19.
of numbers, machine for : P.15.
of ?i-tic curves : C.87.
relation between a curve and an n-tic,
the latter having a multiple point
of the n1th order: C.96.
pencils with problems in conies and
cubics : N.85.
of points on a conic : N.82.
of 6 lines in space : C.52.
of right lines considered as axes of
rotation : C.52.
* systems of points in : 4826, 4828.
ditto, marked on a surface : C.99.
Irrational fractions : decomposition of,
J.19 : irreducibility of, Mem.41 :
rationalization of, A.18.

Irrational functions : M.4 : of the 2nd

degree, 0.98.
Irreducible functions with respect to
a prime modulus : C.70,90,93 : L..
73,.
Isobaric:
calculus, N.85: homog.
functions, G.22 : algorithm, N.84.
Isogonal relations : A.60 : Z.18, 20.
do. represented by a fractional func
tion of the 2nd degree : M.18.
representation of x = " 'X and
: Z.26.
\cX"+d
transformation of plane figures: N.69.
Isomerism, pr : AJ.l.
Isoperimeters, method of : N.47,74,82.
problems: J.18: M.13: Mel. 5.
triangle with one side constant, and
a vertex at a fixed point : C.84.
Isoptic loci : Pr.37.
Isosceles figures : C.87 : JP.30.
Isotherms, families of: Z.26.
Isotropic functions : 0.262,278.
Iterative functions : L.84.
Jacobian : ths 16009 : AJ.3 : thZ.10.
* of3conicB: 5023: LM.4.
formulae: d.c.1471 : J.84;: Mo.84.
function : one argument, G.2.
of several variables : Mo.82*.
modular eq. of 8th degree : M.15.
and polar opposites : Me.64.
sextic equation : Q.18.
system, rnultiplicator : M.12.
Jacobi-Bernoulli function : J.42.
Kinematics: A.61 : AJ.3: G.23: L.
53,80: LM.thl7: N.82, ths 83
and 84.
of plane curves : A.55 : N.82 : caus
tics, Z.23.
paradox of Sylvester : Me.78. .
of plane figures : Mdl.26 : N.78,80, :
of a triaugle, N.53.
of a point : N.49,82 : barycentric
method, MtM. 5.
of sliding and rolling solids : TA.2.
Kinematic geometry :
of space : J.
90.
of similar plane figures : Z.19;,20,23.
Knight's move at chess : C.31,52,74 :
CD.7 : CM.3: E.41: N.54: Q.14j.
Knots : TE.28 : with 8 crossings, E.33.
Kummer's equation, i.e. : 2706.
rational integrals of : M.24.
an analogous eq. : C.99.
Kummer's 16-nodal quartic surface :
C.92 : J.84.
figures of : M.18.
lines of curv. of : J.98.

INDEX.

Lamp's equation : An.79 : C.86,90,91,

ths92.
Lamp's functions: C.87 : J.56,60,62 :
M.18.
Lagrange's theorem (d.c) : 1552 :
C.60,77: CD.6: CM.3: Mel.2:
gzC.96 and Me.85 : gzQ.2.
Lambert's th. of elliptic sector: 6114:
of a parabolic sector, A.16,33,48 :
Me.78: Q.15.
Landen's th. of hyperbolic arc: 6117:
E.21.
Laplace's coefficients or functions : see
" Spherical harmonics."
Laplace's equation: and its analogues,
CD.7 : and quaternions, Q.l.
Laplace's th. (d.c) : 1556.
Last multiplier, Jacobi's th. of : L.45.
Lateral curves : A.58.
Latus rectum : 1160.
Law of reciprocity : N.72 : d.e,3446.
ext. to numbers not prime : C.90.
Least divisors, table of, from 1 to 99000 :
page 7.
Least remainder (absolute) of real
quantities : Mo.85.
Least squares, method of: A.ll, 18,19 :
AJ.l:
C.34,37,40:
CP.8,11 :
G.18 :
J.26 :
L.52,53,67,75 :
Me.80,81 : Mdl.1,4: gzZ.18.
Legal algebra (heredity) : N.63.
Legendre's coefficient or function, X :
2936: C.47,91: L.76, gz79 :
Me.80: Pr.27.
and complete elliptic integral of 1st
kind: Me.85.
rth integral of and log integral of:
Me.83.
product of any two expressed by a
series of the same functions :
Pr.27.
Legendre's symbol
) ' Mel.4,5.
Leibnitz's th. in d.c : 1460 : N.69 : a
formula, Mo.68.
Lemniscate: 5317: A.55,cn3: At.51 :
thsB.4: L.46,47: Me.68 : N.45.
chord of contact, cn : Z.12.
division of perimeter: C.17 : L.43 :
into 17 parts, J. 75 : ineducibility
&c. of the partition equation,
J.39.
tangents of : J.14 : cnZ.12.
Lemniscatic geometry: Z.2U: coordi
nates, Z.12 ; of th order, J.83.
Lemniscatic function : biquadratic
theorem, multiplication and
transformation of formula), J.30.
Lexell's problem : LM.2.

899

Limacon : 5327 : C.98 : N.81.

Limited derivation aud ap. thereof :
Z.12.
Limiting coefficients : C.37.
Limits : theory of, Me.68.
of functions of two variables : M.ll.
of ^1 + ^ when x = oo : L.40:
N.85: Q.5.
Life annuities : A.42 : cn of tables, P.
59.
Linear :
associative algebra : AJ.4.
construction : Man.51.
coordinates in space: M.l.
dependency of a function of one vari
able : J.55.
dependent point systems : J.88.
forms : L.84 : with integral coeffi
cients, J.86,88.
function of n variables : G.14.
Z7a = V- whore U, V are products of n
linear functions of two variables :
CD.5.
geometry, th : M.22.
identities between square binary
forms : M.21.
systems, calculus of : JP.25.
Linear equations : A.512,70 : Ac.4 : C.
81,th942: G.14: J.30:JP.29: L.f
39,66: Mo.842: N.51,75,80s: Z.
15*22.
analogous to Lame's : C.98.
with real coefficients : M.6.
similar : N.45.J.
solution by roots of unity : C.25.
* systems of: 582: A.10,22,52,57 : G.
15: C.81,96: L.58: N.46;.
in one unknown : C.90 : G.9.
of nth order : J.15.
*
standard solution : 582 : Q.19 : gen.
th, A.122.
symbolic solution in connexion with
the theory of permutations: C.21.
whose number exceeds the number of
variables: N.46.
Lines :
alg. representation of : C.76.
" de faits et do thalweg " in topo
graphy : L.77.
generated by a moving plane figure :
C.86.
of greatest slope : A.29 : and with
vertical osculating planes, C.73.
loxodrornic : J. 11.
six coordinates of : CP.11.
Lines of curvature : 5773 : A.34,37 :
An.53: C.74,: CD.5: L.46: M.2,
32,76: N.79: Q.5.
of alg. surfaces : Z.24.
* and conies, analogy : 5854 : Me.C2.

900

INDEX.

Linkage and linkwork(continued) :

Lines of curvature(continued) :
dividing a surface into squares : C. * the Proportionator : 5423.
74: LM.4: Mo.83.
* the Quadruplane : 5422.
of an equilateral paraboloid : N.84.
* the Reciprocator : 5419.
of an ellipsoid: A.38,48 : An. 70: ths * the Reversor : 5407.
* the Translator: 5407.
CD.3 : CM.2,3,4 : JP.l : N.81.
comparison of arcs of, by Abel's th : * the Versor-invertor : 5422.
* the Versor-proportionator : 5424.
An.69.
Lissajons' curves : A.70: M.S.
and of its pedal surfaces : Q.12.
Lituus : 5305.
projection of : Z.2.
rectification of : An.73.
Loci, classification of: C.83.85 : P.78:
generation of surfaces by : J.98 : N.
Pr.27.
Locus of a poin t :
the centre of a circle
77.
cutting 3 circles in equal angles :
and geodesies of developables : L.59.
* near an umbilic : 5822 : A.70 : Q.10.
N.53.
the centre of collineation between a
* osculating plane of : 5835.
quadric surface and a system of
* plane, condition for: 5843: An.68 :
C.363,42:,96 : G.22 : Me.64.
spherical surfaces : A.65.
plane or spherical : An.57 : C.46 : JP.
dividing a variable line in a constant
20: L.53.
ratio : gzAJ.3.
of intersection of common tangents
* of a quadric : 58334 : C.22j,49,51 :
to a conic and circle : N.63,79.
G.ll: 3.26 : Me.l: N.633: Pr.32:
TI.14.
of intersection of curves : CM.2.
*
pd constant along it : 5836.
ditto of two revolving curves : N.64.
projected from an umbilic into conon a moving right line : L.49.
focal Cartesians : E.19.
on a moving curve : Mcm.18.
the product of2 tangents from which,
*
quadratic for y giving the direc
tions : 5810.
to 2 equal circles is constant :
of two homofocal quadrics : L.45.
An.64.
of quartic surfaces : C.59 : L.76.
at which 2 given lengths subtend
of ruled surfaces : C 78.
equal angles : A.68.
whose sum of distances from 3 lines
spherical: C.362,42..,.
is constant: A.17,46: from 2
and shortest distance of 2 normals
lines, N.64 : from lines or planes,
one of which passes through an
umbilic : L.55.
A.192,prs and ths31.
of surface of the 4th class, correlatives
whose distances from 2 curves have a
constant ratio: An.58: or satisfy
of cyclides which have the circle
at infinity for a double line : C.92.
a given relation, A.33 symbolic f.
Locus :
of pole of one conic with
of the tetrahedral surfaces of Lame,
Ac: C.84.
respect to another : N.42.
of remarkable points in a plane tri
and triple orthogonal systems : M.3.
Linkage and linkwork : 540031 : B.
angle : A.43.
of vertex of constant angle touching
28,30 : Me.75 : N.75,78.
* 3-bar: 5430, B.34; 4-bar, Me.76.
a given curve: N.61.
of vertex of quadric cone passing
conjugate 4-piece : LM.9.
* for constructing :
an ellipse: 5426.
through 6 points : N.63.
Logarithmic:
curve, 5284: quadra
a lemniscate : AJ.l.
*
a limacon : 5427 : Me.76.
ture, N.45.
integral : A.919 : J.17 : Z.6.
x" and x'" : AJ.l and 3.
numerical determination : A.ll.
*
root of a cubic equation : 5429.
of a rational differential : J.3.
* Hart's: 5417: LM.6,8,.
* Kempe's : 5401 : Pr.23.
parabola: CD.7.
* Peaucellier's : 5410 : B.21 : LM.6 : N.
potentials : M.3,4,8,13,16 : Z.20.
82.
rational fractions : A.6.
the Fan of Sylvester : E.33.
systems : A.14.
* the Invertor : 5419.
transcendents : P.14.
* the Multiplicator : 5407.
waves : LM.2j.
Logarithms: 142: P.1792, 1787, 1806,
* the Pentograph : 5423.
* the Plagiograph : 5424.
17 : TE.26.
* the Planimeter : 5452.
and anti-logarithms, cn : 1.124,24.

INDEX.

Logarithms(continued) :
* calculation of : 688: A.24,27,42: LM.
1,5: Me.74: N.51 : Pr.31s.32:
TE.6,14 : TI.6,8.
Hnyghen's method : 0.665,682.
and circular functions from definite
integrals : A.65.
common or Briggean : A.24.
constants in integral : $4>0.
of different orders of numbers : L.45.
higher theory of : trA.15.
impossible : CM.] .
natural, or Napierian, or hyperbolic :
A.25,26,57.
of commensurable numbers or of
algebraic irrationals : C.95.
base of : see " e."
*
modulus of : 148 : A.3.
of negative numbers : No.1784.
new kind of : J.70.
powers of : OM.2.
of sum and difference of 2 numbers :
A.45.
with many decimal places : N.67.
of 2, 3, 5, 7, 10 and e all to 260 decimal
places: Pr.27.
of 2, 3, 5, 10 and e to 205 places : Pr.
6,20.
* of primes from 2 to 109 : table viii.,p.6.
tables of log sines, &c. cn : Q.7.
Logic, algebra or calculus of : A.6 : AJ.
3,72 : CD.3 : M.12 : Man.71,76,823 :
Q.11.
of equivalent statements: LM.ll.
Logic of numbers : AJ.4.
Logocyclio curve : Pr.9 : Q.3.
Longimetry applied to planimetry : J.52.
Loto, game of : L.42.
Loxodrome : eqA.21 : N.61s: Z.5.
of a surface of revolution : N.74.
of cylinder and sphere : A.2.
of ellipsoid and sphere : A.32.
of paraboloid of rotation : A.13.
Loxodromic triangle upon an oblate
spheroid : A.27.
Lucas's th : G.14 : analogous f., G.13.
Ludolphian number : Mo.82.
Lunes : J.21.
Maclaurin's th. (d.c) : 1507: A.12 :
CM.&i: J.84: N.70.
symbolic form : CM.4.
Maclaurin-sum-formula : J.12.
Magical equation of tangent : Q.6.
Magic :
cubes, Q.7 : CD.l. cyclovolute : TA.5,9.
parallelopiped : A.67.
rectangles : A.65,66.
squares : A.21,57,66 : CD.l : CM.4 :
B.8 : J.44 : Me.73 : Pr.15,16 : Q.
6,10,11 : TA.5,9.

901

Malfatti's problem, to inscribe 3 circles

in a triangle touching each
other: A.15,16,20,55: J.10,452,76,
77,84,89 : LM.7 : M.6 : P.52 : Pr.
6 : Q.1 : TE.24 : Z.21.
Malfatti's resolvents of quintic eqs : A.
45.
Malm's surfaces, th : J.84,88.
Mannheim, two theorems : G.8.
Martin's measure of distance : A.19.
Matrices : E.42 : LM.4,16 : thMe.85 :
P.58,66 : Pr.9,14.
and function /(*) = ^6 :
. I
cx+d
Me.80v
Cayley's th: LM.16: Me.85.
equation, px xq: C992.
of 2nd order : linear eq.( C.992 : quad
ratic, Q.20.
of any order : linear eq., C.99.,.
notation of : J.50.
persymmetrical, th : E.34.
product of : Gr.5,11.
roots of a unit matrix : C.94.
whose terms are linear functions of * :
J.50.
Maximum or minimum : 58, 1830 : A.4,
13,22,35,49,53,602,70: C.17,24 : J.
48 : JP.25 : Me.l,geo5,72,76,81,
83 : N.43 : Z.13.
* problems on: 183540, 1847: A.2,
geol9,38,39: geoL.42: Mem.ll,
a paradox, N.63.
of an arc as a function of the abscissa :
J.17.
* continuous : 1866.
of a definite integral : Z.21.
discontinuity in : CD.3.
distances between points, lines, and
surfaces, geo : At.65.
duplication of results : Me.80.
ellipse which can pass through 2
points and touch 2 right lines :
A.14.
elliptic function method : Mel.5.
of figures in plane and in space : CM.
3: L.41 : J.24,: Z.ll.
* functions of one variable : 1830 : ditto.
with an infinity of max. and min.
values : J.63.
* functions of 2 variables : 1841 : M<5m.
31 : Q.5,6 : Lagrange's condition,
CM.2.
* functions of 3 variables : 1852 : CD.l :
prsl8605.
* functions of n variables : 1862 : L.43 :
Mem.59 : Q.12 : symmetrical,
M61.2.

902

INDEX.

Maximum or minimum(continued) :
of in- and circam-polygon of a circle :
A.29,30: do. of ellipse, and analo
gous th. for ellipsoid, An.50.
indcterminates : CM.4.
in-polygon (with given sides) of an
ellipse: A.30.
by interpolation, f : A.25.
method of substitution : A.23.
of multiple definite integrals : Mel.4.
planimetrical groups of : A.2.
of single integrals between fixed
limits : J.54,69 : M.25.
of the sum of the distances of a point
from given points, lines, or
planes : J.62.
of the sum of the values of an integral
function and of its derivatives :
_ L.68.
solids of max. vol. with given surface
and of min. surface with given
vol. : C.63.
* of J F(as, y)ds, &c, to find 8 : 30702.
* of j J F(x, y, z) dS, &c, to find S : 3078
80.
Maximum :
ellipse touching 4 lines :
A.12.
ellipsoid in a tetrahedron : Z.14.
of a factorial function : Mo.73.
polyhedron in ellipsoid : A.32.
of a product : N.44.
of a sphere, th : N.53.
* solid of revolution : 3074.
tetrahedron :
in ellipsoid, A.32 :
whose faces have given areas, C.
54,66 : N.62,.
* volume with a given surface : 3082.
of sin*. A36j. Qf!!/x &0w A42.
x
of ax+ky + &c, when a;3+2/2+&c.=l :
N.46.
Mean centre of segments of a line cross
ing throe others : A.40.
Mean distance of lines from a point : Z.
11.
Mean error of observations : A.25 : C.
377 : TI.22.
in trigonometrical and chain measure
ments : A.46 : Z.6.
Mean proportionals between two lines :
A.31,34.
Mean values: C.18,20,23,26,27-, : L.67:
LM.8 : M.6,7 : Z.3.
of a function of one variable : G.16 :
of 3 variables, C.29.
and probabilities, geo : C.87 : L.79.

Measures of length <tc. : page 4.

exactitude of : Z.6: do. with chain,
Z.l.
Mechanical calculators : C.28 : 1.16 :
P.85.
for *' least squares " : MeL2.
Mechanical construction of :
curves :
M.6 : N.56.
Cartesian oval : AJ.l.
conies : An.52 : three, N.43.
ellipse: A.65: Z.l.
lemniscate : A.3.
conformable figures : AJ.2.
cubic parabola : N.58.
curves for duplication of roots : A.48.
(a?3fi)ly : E.18.
surfaces of 2nd order and class : J.34.
Mechanical :
division of angles : Q.4.
measurement of angles : A.61.
* integrators : 5450 : C.92,94,95 : Pr.24<.
for Xdx+ Ydy : Me.78.
involution : AJ.4.
* quadrature : 3772 : A.58,59 : J.6,63
gzA.66 and C.99.
solution of equations : Me.73 : N.67.
linear simultaneous : Pr.28.
cubic and biquadratic, graphically :
A.l.
Mensuration of casks : A.20.
Metamorphic method by reciprocal
radii : N.54.
Metamorphic transformation : N.46.
Metrical :
system : E.30.
properties of figures, transf. of : N.
582,59,60 : J.4.
properties of surfaces : AJ.4.
Meunier's theorem : 5809 : gzC.74.
Minding's theorem : Quaternion proof :
LM.10.
Minimum :
theory of : L.56: prM.20.
angle between two conj. tangents on
a positive curved surface : A.69.
area : J.67.
of circum-polygon : CD.3.
of a hexagonal " alveole," pr : N.
43.
circum-conic of a quadrilateral : A.
13: An.54.
circum-tetrahedron of an ellipsoid :
Z.25.
circum-triangle of a conic : Z.28 : of
an ellipse, Z.25.
curves on surfaces : J.5 : see " Geo
desies."
distance of 2 right lines : G.4 ; of a
point, ths : A.8.
ellipse through 3 points and ellipsoid
through 4 : L.42.
ellipsoid, th : Mo.72.
N. G. F. of a binary septic : AJ.2.

INDEX.

Minimum :
theory of(continued) :
numerical value of a linear function
with integral coefficients of an
irrational quantity : C.53,54.
perimeter enclosing a given area on a
curved surface : J.86.
questions relating to approximation :
Mel.2: Mem.59.
* sum of distances from two points :
920-1.
sum of squares of distances of a point
from three right lines : Z.12.
sum of squares of functions : N.79.
Minimum surfaces :
eqA.38 : G.14,
22: J.81,85,87,ext78: Mo.67,72 :
projective, M.14 : metric, M.15.
algebraic : M.3 : lowest class-number,
An.79.
not algebraic and containing a succes
sion of algebraic curves : C.87.
arbitrary functions of the integral eq.
of: C.40.
between 2 right lines in space : C.40.
generation of : L.63.
representation of by elliptic functions :
J.99.
of a twisted quadric : At.52.
limits of (Calc. of Var.) : J.80 : on a
catenoid, M.2 : determined by
one of the edges of a twisted
quadrilateral, Mo.65.
variation of surface, capacity of : Mo.
72.
Minimum value of
J' i(A+Bx+Cz*+&c.)dx: N.73.
of J y/(dtfi+dy*+...) when the varia
bles arc connected by a quadric
equation : J.43.
Models : LM.39 : of ruled surfaces,
Me.74.
Modular :
equations : An.79 : of 8th
degree, 69: C.48,,493,66 : M.1,2 :
Mo.65 : see also under " Elliptic
functions."
degradation of : M.14.
factors of integral functions : C.24.
functions and integrals : An.51 : J.
184,193,20,21,23,25 : M.20.
indices of polynomials which furnish
the powers and products of a
binomial eq : C.25.
relations : At.65.
Modulus:
of functions, principal:
C.20.
of series : C.17.
* of transformations : 1604: A.17.
Momental ellipse : 5953.

903

Momental ellipsoid : 5925, 5934 ; for a

plane, 5936.
Moment of inertia : 5903: An.63 : At.
43 : M.23.
of ellipsoid : 6150 : CD.8 : J.16.
by geometry of 4 dimensions : Q.16.
* principal axes : 5926, 5967, 5972 : At.
43.
* of a quadrilateral: 5951: Q.ll.
of solid ringB of revolution : Q.16.
* of a tetrahedron : 5957.
* of a triangle : 5944 : Me.4 : Q.6 :
polar, N.83.
* of various lamina; and solids : 6015
6165.
Monge's theory " des De'blais et des
Remblais " : LM.14.
Monocyclic systems and related ones :
J.98.
Monodrome functions : C.43 : G.18.
Monogenous functions (Laurent's th) :
Ac.4j= C.32,43.
Monotypical functions : C.32.
Monothetic equations : C.99.
Mortality: A.39.
" Mouse-trap " at cards : Q.152.
Movements : JP.15.
elliptic and parabolic : JP.30.
groups of : An.692.
of a plane figure : thAn.68 : JP.20,
28: LM.3.
of an invariable system : C.43.
of a point on an ellipsoid : AJ.l : J.54.
relative : JP.19 : of a point, L.63.
of a right line : C.89 : N.63.
of a solid : JP.21.
transmission of and the curves result
ing : JP.3.
of "annlich - veranderlicher " and
" affin-veranderlicher " systems :
Z.24 and 19.
Multinomial theorem : 137 : Me.62.
Multiple-centres, geo. theory : L.45.
Multiple curves of alg. surfaces : An.73.
Multiple Gauss sums : J. 74.
Multiple integrals : 1905, 2825 : A.64 :
An.52: C.8,11: thsCD.l : thE.
36: J.69: L.39,43a,45.,46,th48,56:
LM.8,: Me.76j: Z.l3,3.
* double: 2710, 273442: A.13: Ac.5:
An.70 : J.27j : G.10 : L.58j : Mem.
30.
approximation to : J.6.
Cauchy's theory, ext. of : C.75j.
*
change of order of integration :
2775: A.19.
expressing an arbitrary function :
J.43.
reduction of : J 45 : Z.9.
residues of: C.75j.

904

INDEX.

Multiple integrals, double(continued):

(a-*if) dxdy
t

Multiple integrals(continued) :
volume integral of

= f = L.38.
same with log of numerator : L.50.

{('-)+...}
= 4mjr : C.66.
transf. of ff
d+d+
J J ^(sin3!/sin-<4 cos3^)'
J.20.
j-p'f*
cs i* cosj> (fx dy
c 96
2 ^a^LCOSa *oey) } '

JJJ e exp (z3y3z3), x*y*z dxdydz:

N.54.
\[\ ...i(ax- + bya + &c.) *>y .. dxdy ...
with limits 0 to oo in each
(Pfaff) : J.28.
dx dy .
{(a-x)3+(6-J,)3+...} by discontinuous functions : TI.21.
do. with n = and with a numerator
(a-x)F(g + g+ .) : CM.3.

JJ F (a+te+cy) dx dy : A.37.
JJi^(a+ty)^rf!/: J.42.
ff^d*d: C.96.
JJ G (u, a)

*
*

37.
evaluation : A.5 : by Fourier's th.
CM.4.
expansion of : Q.8.
Frullanian : LM.lo.
limits of : LM.16.
reduction of : An.57 : L.41,39.
by transf. of coordinates : C.13.
J"F (z3+y'+ . . . ) f (cu + by+ . . . ) dxdy
... : L.57.
of theory of attraction : CD.7.
transformation of : 2774 : A.10 : An.
53 : No.47 : CM.4 : M<51.2 : Mem.
38: Q.4,12.
an indef. double : J.8.,10.
a triple integral : 27749 : J.45.
J/]rf-fi + ... + i/dx: LM.ll.
triple: 2774: A.30: J.45.
which are unaltered in form by trans
formation of the variables : J.15,
91.
ffl...d*dyd*...: 0,23.
JJJ ...af-*y1*-l...dzdydz ...with
different limiting equations :
2825 : CM.2 : L.51.
some other integrals evaluated by r
functions: 282634.
\\\F (ax + by+cz, ax + b'y +e'z,
a"x+b"y+c"z)dxdy dz, limits
oo : A30.

JJ . . . F (x, y, z. . . ) PQ dx dy dz .. . where
P = (i-*)-l(i-y)*_ Q **'... . L.59.
arising from (2604), viz :
eexp^-xS-^(ix: pr.42.
Multiple points: 5178: CM.2: thG.15:
Me.2: Q.2,6.
on algebraic curves : An.52 : L.42 :
N.51,59,81, at * 64,.
on two curves having branches in
contact : C.77.
on a surface : J.28.
Multiplication : 28 : J.49 ; abridged, N.
79.
by fractions : Me.68.
Multiplicator equations : M.15.
Multiplicity or manifoldness : J.84s,86 :
thAc.5: Z.20.
Music: E.27,,28: Pr.37.
Nasik squares and cubes : Q.8,15j.
Navigation, geo. prs.of use in : A.38.
Negative in geometry: No.1792.
Negative quantities : At.55 : N.44j,67 :
TE.1788.
Nephroid: LM.10.
Net surfaces : J.1,2 : M.l : any order,
An.64.
quadric: J.70,82 : M.ll.
quartic : M.7.
and series : C.62.
trigonometrical : Z.14.
having a 3-point contact with the
intersection of two algebraic sur
faces : G.9.
Newton, autograph m.s.s of : TE.12.
Nine-point circle : 954, 4754 : A.41 : E.
7,30,th35,pr39 : G.l,ths4: Me.64,
68: Q.58.

INDEX.

Nine-point circle(continued) :
an analogous circle : A.51.
* contact with in- and ex-circles : 959 :
Me.82: Q.13.
and 12-point sphere, analogy : N.63.
Nine -point conic of a tetrahedron:
Me.71.
Nonions (analogous to Quaternions) :
C.97,98.
Non-uniform functions : C.88.
Nodal cones of quadrinodal cubics : Q.
10.
Node cusps : Q.6.
Nodes, two-plane and one-plane : M.22.
Normals: 1160: 41223,5122: A.13,
53: LM.9: p.eMe.66: Z.cn2and3.
of envelopes : He.80.
* plane of a surface : 5772.
* principal : 5722 : condition for being
normals of a second curve, C.85.
of rational space curves : J.74.
section of ellipsoid (geodesy) : A.40.
* of a surface : 5771, 5785 : G.52 : CD.3 :
CM.2: L.39,47,72: M.7.
coincident : L.48.
transformation of a pencil of : C.882.
Notation (see also " Functions ") :
A,B,C,F,G,H : 1642.
* A.P., G.P., H.P. : 79,83,87.
* (016303) : 554.
()
Jacobi's function (sec
" Functions ").
* ab or o"/ ; 2451.
a+b+-c+ 160.
*
Bernoulli's nos. : 1539.
* C (n,r) or C,r : 96. Otherwise
G (,3) = number of triads of n things,
&c.
^ ^ = rth coeff. of nth power of (1 +*) :
also Jacobi's function (see " Func
tions ").
* D 3489 : dt,dw Ac. : 1405.
* ePy.dzS^&c: 1407.
dx
* d(uvw) . lg00
d (xyz) '
*A: 582, 1641, 3701 ; a', 1645.
*E : 902, 3735.
*e: 150, 1151.
A
A
eexpaH
or e
=e *
x
I
a; I
/(as) : 400, 1400 ; /-' (a;), 430.
*/ (*) f" (*) : 424, 1405.

905

Notation(continued) :
*<p (a/3y) = u : 4656 ; * (A/xe) or U, 4665.
*Gu,: 3732.
*H(n,r); 98.
*/: 1600.
N.G.F = numerical generating func
tion.
N=b(modr) signifies that Nb is
divisible by r.
* [n = ("! = ! : 94,3713.
* it as operator : 3500.
* P(n,r) or P,r = re(r> : 95. Also,
P (n, r) = number of triplets of n
things, &c.
* () or Z'(x) = dr log r (as) : 2743.
R,r,r : 90913.
8m,8~,: 534; ST., 2940.
sin-'.&c. : 626; sinh, &c, 2180.
2 : 37813.
w : 3499.
(1 (n) = sum of divisors of n.
^| or '
or E ^~ j = integer next
n.
<
d
1 integer next
d
(j} =rth coeff. of (1+*)".
<fc = not less than ;
= not greater
than.
(+) = denominator to be stated after
wards.
( I ) ^ ( % ) : 1620.
algebraic : CP.3.
for some developments : C.98.
continuant = contd. fraction determi
nant.
median = bisector of side of a triangle
drawn from the opposite vertex.
aubfactorial : Me.78.
suggestions : Me.73.
Numbers (see also " Partition of," and
" Indeterminate equations") : 349 :
A.2,16,26,58,59 : Ac.2 : AJ.4,6 :
C.f 12,43,44^,45,^60 : CM.l : G.16,
32 : J.93,39,404,48,77 ; tr,273,28and
292 : JP.9 : L.3739,41,45,586,59,
60 : LM.4 : Mem.22,24; tr(Euler),
30: N.443,62,79: Q.4: TE.23.
ap. of algebra, JP.ll ; of r function,
No.81 ; of infiuitesimal analysis,
J.19,21.
formula? : L.64-,652.
relation of the theory to i.c : C.82.
5 z

906

INDEX.

Numbers(continued) :
representation of by forms : C.92 ; by
infinite products, A.l.
square having prime factors of the
form 4n + l : N.78.
squares of : J.84 : M.13 : Pr.6j,7.
three in ar.p : N.62.
sums depending upon E if) : L.60*.
sums of digits : Me.66 : TE.16.
sum and difference of two squares :
thsN.63.
* sums of divisors : 377 : Ac.6 : G.7 :
L.63,: Mel2: Mem.50.
* sums of powers of (see also " Series "):
Dirichlet's f. for class numbers as
276, 2939: An.61.65: thCD.o :
positive determinants : L.57.
Me.75: N.42,56,70: Q.8.
division of : A.26: J.13: Hel.3: Pr.
of cubes: An.65: L.66 ; of the odd
7,10 ; by 7 and 13, A.25,26 ; by
nos., A.64.
vu+mf-,l&.6m.\b: P.17,88 : Q.
of ?i primes : N.79. ; 4th powers,
19,20.
A.54.
divisors of y2+Az- when .4 = 4n + 3
of squares : A.67.
a prime : J.9.
of uneven orders : Mo.57.
divisors arising from the division of
symmetrical functions of: Q.7.
the circle : L.60.
systems : Z.14 ; history of, by Hum
4m + 1 and 4m + 3 divisors of a num
boldt, J.4.
ber : LM.15.
theorems: A.7,10,20,49 : An.70: C.
factors of: Mem.41.
25j,43,83: CM.2: G.8 : L.48,52:
Gauss's form : L.56.
N.75; Cauchy's, gzC.53 ; Eisenintegral quotients and remainders :
stein's, J.27,50,83 : LM.7 : Q.5,6.
An.62.
Gauss's on X =
. C.98.
largo, analysis of : A.2 : C.2,29.
xy
method with continuous variables :
Lagrange's
arithmetical
: A.47.
J.41.
p"+q'' in terms of pq : N.75.
multiples of : C.2.
ou2"l: C.85,86: Me.78.
non-pentagonal th : J.31.
2, biquadratic character of : C.57,
number of integers prime to n in
66: L.59.
! = <*>(): L.57.
on (n+l)~n": N.44; ft(n),L.69.
odd: A.lo : and prime to all squares, *
on n' n (n 1)' + , &c. : 285 : A.30.
C.67.'
on 2m positive numbers : N.43.
Pellian equation : prA.49 : LM.15 :
sol. by ell. functions, Mo.63.
on P(m) + E [~) P(l)+ : Q-3.
perfect : C.81 : N.79.
polygonal, Fermat's th. of : P.61.
on the greatest product in whole
polynomials having determinate prime
numbers of given sums : J.57.
divisors : C.98.
on au odd sum of 12 squares : L.60.
powers of, 12 theorems : N.46.
on products of sums of squares: G.2.
prime to and < N : A.3,29 : E.2& :
on 4 squares : N.57.
J.31 : N.45.
on 2, 4, 8, and 16 squares : Q.17.
prime to and < the product of the
on <p(a) + <p (a') + , &c.= n, where a,
first )i primes : A.66.
a',<Scc.,arcthe divisors of n: Cm.3.
prime with respect to a given ar.p :
Numeration, ancient decimal : C.6,84.
C.54.
Numerical approximations : N.42j,53.
prime to the radix having multiples
Numerical functions : L.57 : Me.62.
made up of repeating digits :
simply periodic : AJ.l.
Me.76.
sums of, approximately : C.96.
products of divisors of : Q.20.
which express for a negative deter
quadratic forms of : Mdm.53.
minant the number of classes of a
rational linear functions taken with
quadratic form, one at least of
respect to a primo modulus, and
whose extreme coefficients is odd :
connected substitutions : C.483.
C.62.,.

Numbers{continued) -.
approximation :
to -/N, E.17 ; to
functions of large numbers, C.82.
binomial eqs. with a prime mod : C.62.
cube : Q.4.
cubic binomials : z3j/s: C.612.
determined by continued fractions :
LM.29.
digits, calculus of, th : J.30.
digits terminating a power: A.58 :
N.46.
Dirichlet's th. 2>
\ =((m) : L.57.

INDEX.

Obelisks: A.9,11.
Oblique :
bevilled wheels, cn : J.2.
coordinates : 4050, 55119 : fN.54.
cyclic surface : TI.9.
and oscillating circle of a conic : G.22.
Octahedron function : Q.16.
Octahedron, centroid of: LM.9.
Octic equations : G.7,10 ; and curves,
M.los.
Octic surface : G.13 : M.4 : Q.14.
Operative or symbolic calculus: 1483,
34703628: AJ.4: C.17: G.202:
J.5,59: LM.123: Me.82,85: P.37,
44,6063: Pr.l0,,ll12,,134 : Q.
4,5,8.
applications: G.19: Me.82.
algebraic : TE.14 ; ap. to geometry,
CM.1,2; ,CM.3.
expansions: Pr.l42.
formula? : C393.
index symbol : 1485 : CD.H.
integration : CD.3 : exMe.76.
representation of functions : C.43 :
CD.2.
seminvariant operators : Q.20.
on the symbols x'J, log/,!?, sin x, cos x,
8in"'ir, cos-'k : A.9,11.
theorems: A.57 : CD.R, : LM.ll : Q.
3a,15,163 ; from Lagrange's series,
Q.16; from irpupau pu, CD.5.
Operators : d 1405; e'"', 15201, Q.9S.
{d,m)-\ &c. : 347085 ; gz of 3474,
C.43.
D(D-l) ...(D-n+l) : 3489.
= xdT+ydy+&c. : 3500.
expansions and formula) for :
F (xd.) U, where U = f (as) = a + hx +
cx*+ : 3486.
f(D) uv, 3494: vf{B)v, 3495, with/
as above.
D"f(xD)U: E.36.
{+(!)) er"}Q: 3491.
e-*vF(x),&c.: E.34.
e"*7F(x)='^F(x+l): E.36.
X
QF(a,x): Q.13.
lmtl/{a v/vfcy(u-i)}, &c. : C.96.
/(as+AD).l: E.39.

= <M<*,+)
3498.
reduction of F (jr,) : 3503.
F(w)U and F-1 (it) U : 350910.
C(i(,m)/m! : 3514.
transformation of Vdr. Udt..., &c : G.
21.
Orthocyclc: Q.17.

007

Orthogonals, algebraic system of : C.69.

Orthogonal :
circles : 4170, 41824 ;
of in- and circum-circles of a tri
angle, Q.18.
circle and conic : E.7.
coefficient system : A.2,61.
coordinates : C.60 ; curvilinear, JP.
26.
conies : N.84 ; families of, A.63.
curves : J.35 : N.52,81.
system from logarithmic repre
sentation: Z.16.
* lines of a triangle : 4633.
lines and conies : C.72.
* projection : 1087.
in metrical projective geometry :
GM.14.
of a circle into an ellipse : A. 2.
of a triangle: E.30,31,36,37.
substitution: J.67 : M.13: Z.24.
surfaces : C.29, spheres 36,54>,59,72,
79,87,thsl7 and 21 : J.84:JP.17:
L.43,44,46,47,63 : N.51 : P.73; :
Pr.21 : Q.19.
cubic eq. for : C.76j.
with elliptic coordinates : C.53 :
J.62.
and isothermal: C.84: JP.18: L.
43,49.
systems : C.67,754 : M.7 ; condition,
J.83 ; quadric, J.76 ; parallel,
M.24.
triple : A.55 58 : An.63,77,85 :
C.alg58,67 ; cyclic, G.21,22 : J .84 ;
quartic, 82: Z.23.
trajectories: L.45 : Me.80: Z.17.
of circles : Me.85.
of circular sections of an ellipsoid :
L.47.
of a moveable plane : Pr.41.
of a moveable sphere : C.42.
of a surface : Mem.20.
Orthomorphic projection of an ellipsoid
on a sphere : AJ.3.
Orthomorphosis of a circle into a para
bola: Q.20.
Orthoptic :
lines of a conic: A.57.
loci of : LM.13 : Pr.37 ; of 3 tangs.
to a quadric, E.40.
surface of a quadric : J.50.
Orthotomic circles : Me.64,66 : Q.2.
Osculating:
circle: 5724: L.39.
of conies : A.70 : N.60.
of a family of curves : N.70.
of a parabola, ths : N.66.
of quadric curves N.43.
of tortuous curves : N.81.
* cone : 5727 : angle of, 5752.
conic : L.39 : triply, A.69 : Z.19.
of a cubic curve : J.68: : Z.17.

908

INDEX.

Osculating {continued) :
curves : Q.ll.
helix : N.71.
line of a surface : C.82 : J.81.
parabola : N.81.
paraboloid: JP.15: N.82: ofaquadric,
L382.
* plane: 5721, eq 5733: and radii of
curv. at a multiple point of a
gauche curve : An.71 : C.68.
of a tortuous curve : C.96 : J.41,63.
sphere :
M6m.20 : N.70 : of curve
of intersection of two surfaces,
en : C.83.
of two curves having a common
principal normal : LM.16.
surfaces : C.79>, degree of 98 : L.41,
80 : of quadrics, N.60.
Oval of Cassini : see " Cassinian oval."
Oval of Descartes : see " Cartesian
oval."

Paraboloid(continued) :

quadrature of : 6127 : An.55.

segment of : 612733 : A.29.

hyperbolic : 5623 : A.11.
of revolution : 6134.
Paradoxes of De Morgan : J.ll3,124, 13,
16.
Parallels : A.8,47 : At.51 : J.11,73 : Mel.
1,3 : Mem.50.: Z.21,22 : Thibaut's
proof, A.15.
in analytic geometry : A.44.
Parallel curves : J.55,ths32 : LM.3 : Q.
11 : Z.5.
closed : A.66.
of ellipse : 4960 : A.39 : An.60: N.44-:
Q.12.
Parallel surface: C.54: LM.12.
of surface of elasticity : An.57.
of ellipsoid: A.39: An.50,60: E.17 :
J.93.
Parallelogram with sides through four
given points : A.39.
Pangeometry : G.5,15.
Parallelogram of Watt : A.8 : L.80.
Pantograph : 5423 : Mem.31 : TE.13.
Parallelopipeds : on conjugate diame
Paper currency : A.42.
ters : 5648.
Pappus, prs in plane geometry : A.38 :
diagonals, &c. : CM.l.
Z.5.
equality of : A.4.
Parabola: geo. 122044 : anal. 4200
analogues of parallelograms : LM.2 :
39: eqCM.2: N.424,54,70 : geo
Me.68.
CM.4 : Me.71 : cnl249 : thsN.60,
on a spherical base : N.45.
63,71,76,80,.
system of : LM.8.
circum-hexagon and triangle : CM.l.
Partial differences: question in analysis:
* circle of curvature of : 1260: A.61.
J.16.
chords of : 1239, 4224.
Partial differential equations (P.D.E.) :
*
two intersecting : 1242.
33803445 : C.34,ll,16,78,95,96 :
determination of vertex and axis :
thsCD.3 : J.58,80,prs26 : JP.7.10,
11: L.36,80,83: M.ll: Z.6,8,18N.58.
eq. deduced from eq. of ellipse : 1219. P.D.E., first order : 3399-3410 : A.33,
focus and directrix: N.49.
tr50: An.55,69: C.14,,5354s :
* focal chord : 42359.
CD.7: CM.l.: J.2,17: trJP.22 :
latus rectum : 1222 : Me.75.
L.75: M.9,ll,th20: Z.22.
complete primitive connected with
* normal, length of : 42334.
plane and spherical : A.60.
any solution : 3405.
quadrature of : 1244 : A.32.
derivation of the general primitive
* and right line : see " Right line."
and singular solution from the
segment of : 6078 : A.26,29.
complete primitive : 34ol.
derivation of a singular solution from
solid generated by it : N.42.
the differential equation : 3403.
sector : E.30 : N.57 : Lambert's th,
with a general first integral : Me.78.
J.16.
in space : A.3.
integration by :
Abelian func
tangents : see under " Conies."
tions : C.94.
through 4 points, cn : 4837 : J.26.
Cauchy's method : C.81.
Charpit's method : 3399.
* triangle of 3 tangents : 1237,68 : A.
Jacobi's first method: C. 79,82: and
47: Me.75.
trigonometry of : CD.8.
ap. to Pfaff's pr, J.59.
Paraboloid : 5621, 6126-41 : N.61 : Q.
Jacobi-Hamilton method : M.3.
Lie's method : M.6,8.
13.
generating lines of : 5624.
Weiler's method : M.9.
law of reciprocity : 3446.
of eight lines : C.84.
* elliptic: 5622: A.11 : L.58: P.96.
and Poisson's function : C.91.

INDEX.

P.D.E., first order(continued) :

simultaneous: C.68,76 : L.79: M.4,5.
* singular solution : 34013 : J.66.
systems of : A.56 : M.11,17.
theorem of Jacobi : C.45.
3 variables : J.64.
* n variables : 3409 : A.22 : J.60 : LM.
10,11with coxistant coefficients : M<51.5.
integration by calc. of variations :
C.14.
z = px+qy+F (p,q): Z.5.
xfy"z'pmqn=A: CM.l.
(l+Pi+..,+P-$dz,dv)<>Z= Q: Z.13.
P.D.E., first order, linear: 338195:
reduction to, C.15 : J.81 : Me.78.
* Pzx+Qzg=:R: 3383: extension to
n variables, 3384.
L(px+qyz)MpNq+B = 0 :
C.83.
a (yuz zuy) + b (zuxxuz ) + c(xuyyur)
= 1: Q.8.
Pt,+ Qk+ ... +Bt, = S: 3387.

* (xa)zr + (yb)z, = cz: 3392.

* aj+2/2+z3 = 2a: 3393.
* simultaneous : 3396 7 ; cs. 3398 :
J.81.
u = vy and w = v* : J.70.
P.D.E., second order : 342045 : A.33
C.544,702>78,98 : JP.tr22 : L.72
M.15: Me.76,,77: Mdl.3: N.83
P.46 : transf. of, C.97.
in two independent variables : trA.
54: trNo.81 : C.92 ; transf. of, 97 :
M.24.
in 4 and 5 variables : Mem.13.
* Br+8s+Tt=V, Monge's method:
3423: CM.3: N.76 : Q.6.
* Br+ Ss+Tt+U (r<-2) = V : 3424,
343440: J.61.
* Rr+Ss + Tt+Pp+Qq + Zz = U:
3442.
r+t = 0: A.2: CM.l: J.59,73,74 :
L.43.
r+t+Wz = 0: M.l.
* r-aH = 0: 3433.
* rcfit = <f> (x,y), &o. : 3565.
x (raPt) 2np : E.13.
r-o?ylbt = Q: E.27,28.
r=q2mt: C.98.
r = q : J.72.

909

P.D.E., second order(continued) :

dx(p sin x) + t+n (n + l)zsin2se = 0 :
L.46.
r (l+22) = /(l+l>2): J.58,.
q%r-AH+ ^ i,, q = 0 : E.22.
n
d,(px)+d(qz) = 0: Z.28.
construction of explicitly integrable
equations of the form 8 = z\ (x,y) -.
JP.28.
*- {f(*)+F(y)Y
o2i,logXX = 0: L.53.
8+Pp + Qq+Z=0: Me.76.
* s+ap + bq+abz= V : 3444.
(ax + by+ c) 8+ cikq+bnp 0 : A.33.
(x+yf 8+a (x+y)p + b (x+y) q+cz
= 0: A.33,38.
2 (Z8pq)2+q = F(y): A.70.
rt2: geoQ.2.
* P = (rt8*)" Q, Poisson's eq. : 3441.
(l+r)t+(l+t)r2pqs = 0: An.53*.
* q (1+g-) r+p (l+p) t-(p+q+2pq)
8 = 0: 3432.
4s2+(rf)3 = 4fc2 : approx. integn.,
C.74.
As+Bq+f(r,p,q,x,y,z) = 0: 0.93.
(logz)ly+az = 0: 0.36.
u* = Rr- where t = f
^
' r
J y(2Br2+^3)
L.38.
* uta+it2>+M2>= 0 (see also " Spherical
harmonics ") : 3551, 3626, J.36 :
Mo. 78.
* ur+u,+us = xyz : 3552.
* aux + bny +cur = xyz, &c. : 3554.
* asMjr+au, g*asu = 0: 3618.
* a?(ulT+uy+u3,)=u1, : 3629: C.7 :
LM.7,.
integration bv change of variables :
0.74.,.
P.D.E., third order, two independent
variables : LM.8: N.83.
P.D.E., fourth order : aa = 0 : C.69.
P.D.E., any order: No.73 : C.80,89:
M.l 1,13.
_
x 2zx = z, : Z.7.
two independent variables : C.75 :
CP.8.

910

INDEX.

P.D.B., any order{continued) :

any number of functions and inde
pendent variables : C.80.
and ap. to physics : JP.13.
of cylinders : Me.77.
and elliptic functions : J.36 : hyperell., J.99.
integration by definite integrals : An.
59 : C.94 : L.54.
of dynamics : C.5 : J.47.
Hamiltonian: M.23: Z.ll.
of heat : L.48 ; of sound, L.38.
integrated in series : C.15,16.
of Laplace: G.23.
linear: An.77 : C.13,90 : CD.92 : CM.
2: J.69: JP.122: L.39.
of orthogonal systems of surfaces :
Ac.4: C.77.
with periodical coefficients : C293.
of physics : L. 72,47.
P.D.E., simultaneous : C.92,th78: LM.9:
M.23 : Z.20.
linear : J.65.
P.D.E., system of: C.13,74,81.
amz, = x^Zn, : A.30,31 , by z = e'/(x).
+ Fl(y)+xFAy)+ ...+
x^Fm(y): A.51.
Aznl+{illr+d^+...)"z = 0: C.94.
dz = Hdx+Kdy + Ldp+Mdq+Ndr+
Ac.: J.14.
Partial differentials of : J.ll.
x^+y"
Partial fractions : 235, 1915 : A.30,66 :
C.46,49,,783 : CM.l : G.2 : J.1,5,9,
10,11,22",32,50., : JP.3 : L.46 : Mem.
9: N.452,64,69: Q.5.
Partition of numbers (see also "Num
bers" and " Indeterminate eqs.") :
AJ.2,5,6 : An.57,59 : At.ti5 : O.80,
86,90.91: UP.8: J.13,61,85: M.
14: Man.55: Me. 78,79 : Mem.13,
geo.ap20,44 : Mel.l : N.69,85 : P.
50,56,58 : Pr.7,8 : Q.l2,2,7,15 : Z.
20,24.
by Arbogast's derivatives : L.82.
of complex numbers in Jacobi's th :
C.96.
by elliptic and hyper-elliptic func
tions : J.13.
formula of verification : Pr.24.
into 2 squares : An.50,52,54: C.87 : J.
49 : LM.8,9 : N.54,78,alg65 : odd
squares, Q.19.
into 3 squares : J.40 : L.59,60.
into 4 squares : C.99 : L.68 : Pr.9 : Q.l.
4 odd, or 2 even and 2 odd : Q.19,20.
into 5 squares : C.97,98 : J.35.

Partition of numbers(continued) :
into ten squares : C.60 : L.66.
into squares : C.39,90 : L.61 : N.
54.
and an integer : L.57.
into the product of two sums of sqs. :
L.57.
into parts, the sum of any two to be a
sq. : Mdm.9.
into 2 cubes : L.70.
into sum or difference of 2 cubes :
AJ.2.
into 4 cubes : N.79.
into maximum nth powers : C.95.
into 10 triangular numbers : U.62.
formation of numbers out of cubes :
J.14.
2 squares whose sum is a sq. : E.20 :
N.50.
3 squares, the sum of every two being
a sq. : E.17.
4 squares, the sum of every two being
a sq. : E.16.
3 nos., the sum or diff. of two to be a
sq. : Mem.18.
2 snms of 8 sqs. into 8 sqs. : Me.
78.
a sum of 4 sqs. into the product of 2
sums of 4 sqs. : TI.21.
n nos. whoso sum is a sq. and sum of
sqs. a biquadratc : E.18.22,24.
a quadric into a sum of squares : N.
80,81.
nS ms into 3/>2+ q* : N.49.
a square into a sum of cubes : N.67.
a cube into a sum of cubes : E.22,
23.
into 4 cubes : N.77 ; into 3 or 4
cubes, A.60.
n139w12 or its double into 2 cubes :
N.81.
3 nos. whose sum is a cube, snm of
sqs. a cube, and sum of cubes a
sq. : E.26.
5 biquadrates whose sum is a square
E.20.
of n into 1, 2, 3, &c. different num
bers: E.34.
of pentagonal numbers : C.96.
a series for the : AJ.6.
tables, non-unitary : AJ.7.
theorems: AJ.6: C.40,96 : Me.76,80,
83 : pr. symm. functions, G.10.
Partitions :
in theory of alg. forms :
G.19.
number of for n things : E.10.
in planes and in space: J.l.
Sylvester's theorem : Q.4.
trihedral of the X-ace and triangular
of the X-gon : Man.58.

INDEX.
Pascal's theorem : 4781 : AJ.2 : CD.3,
4: CM.4: J.34,41,69,84,93: LM.
8: Me.72: N.44,52,82: Q.1,4,5,9:
Z.6,10.
extension of and analogues in space :
C.82,98: CD.4,5,6: G.ll : J.37,75:
M.22: Me.85.
ap. to geo. analysis : A.18.
on a sphere : A.60.
Stei tier's " Gegenpunkte " : J. 58.
Pascal lines : B.30.
Pedal curve : A.35,36,52 : J.48,50 : M.6 :
Mc.80,81 : Q.11 : Z.52,21.
circle and radius of curvature : C.84 :
Z.14.
of a cissoid, vertex for pole : B.l.
of a conic : A.20 : LM.3 : Z.3.
central : A.9 : Me.83 : N.71.
foci and vector eq. : LM.13.
negative central: E.20,29 : TI.26.
negative focal : E.16,17,20.
nth and n 1th : E.18.
of evolute of lemniscate : E.30.
inversion and reciprocation of : E.21.
of a parabola, focal and vector eqs. :
LM.13.
rectification of difference of arcs of :
Z.3.
which is its own pedal : L.66.
Pedal surfaces : A.22,35,36 : M.6 : J.50 :
Z.8.
counter: AJ.5.
volumes of: C.55: A.34 : An.63: J.
62: Pr.12.
Pentagon, ths : A.4 : J.5,56: N.53.
diagonals of: A.57.
Pentagonal dedecahedron : A.25.
Pentahedron of given volume and mini
mum surface : L.57.
Periodic continued fractions : A.62,68 :
C968: J.20,33,53: N.68 : Z.22.
closed form of : A.62.
representing quadratic roots : A.43.
with numerators not unity : 0.96.
Periodic functions : A.5 : J.48 : N.67 :
gzC.89: Mo.84.
cos xj cos 3z+\ cos 5x : CD.3j.
doubly: C.32n,40,70,90 : J.88;, : L.54.
of 2nd kind : C.90,98 : gzL.83.
of 3rd kind: 0.97.
monodromic and monogenous :
C.40.
with essential singular points: C.89.
expansion in trig, series: N.78.
4-ply, of 2 variables : J.13.
2n-ply, of n variables : Mo.69.
multiply: C.57,58j.
integrals between imaginary limits:
A.23.
real kind of: Mo.66,84.

911

Periodic functions(continued) :
of 2nd species : M.20.
of several variables : C.43 : J. 71.
in theory of transcendents : J. 11.
of 2 variables with 3 or 4 pairs of
periods : C.90.
with non-periodic in def. integrals :
C.18.
Periodicity theory : M.18.
Periods :
cyclic, of the quadrature
of an algebraic curve : C.80,84.
of the exponential e* : C.83.
of integrals : see " Integrals."
law of: C963.
in reciprocals of primes : Me.733.
Permutations: 94: Al. : C.22 : (JD.7:
L.39,61 : LM.15: Me.64,66,79 : N.
44,71,763,81 : Q.1 : Z.10.
alternate : L.81.
ap. to differentiation and integration
A.21.
of n things : C.95 : N.83 ; in groups,
L.65.
of :.!./ and 2q letters, 2 and 2 alike :
N.74,703.
number of values of a function
through the permutations of its
letters: C .20,21 ,46,47 : L.65.
successive (" battement de Monge ") :
L.82.
with star arrangements : Z.23.
Perpendicular from a point :
upon
a line : length of : 4094, t.c4624 :
eq4086, t.c4625 : sd5530.
* upon tangent of a conic : 4366 73.
* upon a plane : 5554.
* upon tangent plane of aquadric : 5627.
* ditto for any surface : 57913.
Perpetuants : AJ.72.
Perspective: 1083: A.692: G.3 : thsL.
37: Me.75,81.
analytical : A.ll.
of coordinate planes : CM.2.
* drawing : 10836.
figures of circle and sphere : A.57.
isometrical : CP.l.
oblique parallel : Z.16.
projection : A.16,70.
relief: A.36,70: N.57.
* triangles: 974: E.29 : J.89 : M.23,16:
in a conic, Al.
Petersburg problem : A.67.
Pfaff's problem: A.60: C.94 : J.61,82:
M.17 : th of Jacobi, J.57.
Pfaffians, ths on : Me.79,81.
n (see also "Expansion of"): C.95: E.
30 : N.42,45.
calculation of: A.6,18: E.27 : G.2 :
cnJ.3: Me.73,74: N.50,56,66.
by equivalent surfaces : N.48.

912

it :

INDEX.

calculation of(continued) :
by isoperi meters : N.46.
by logarithms : N.56.
to 200 decimal places : J.27 ; to 208,
P.41; to333,A.212; to400,A.22;
to 500, A.25 ; to 607, Pr.6,11,22.
formula) for, or values of : A.12 :
J.17 : L.46 : M.20.
ir = 3 +
approx. : Me.66 ;

= - log i, J.9.
z
functions of : p.6 : A.l : C.56,74.
it"1 : E.27; to 140 places, LM.4.
hyperbolic logarithm of : LM.14.
powers of ir and of w~l : LM.8.
* incommensurable : 795.
series for: Q.12 : TE.14.
theorem on n and e : Q.15.
U(x)s 1(1 +*)... {l+(n-l)}: A.12:
J.43,67.
n (x) and imaginary triangles and
quadrangles: A.51.
Piles of balls and shells : N.72.
Pinseux's theorem : Mo.84.
Plane: J.20,45 : Mem.22.
* equations of : 5545, q.c 5550, p.c 5552 :
Z.l.
*
under given conditions : 556073.
* condition for touching a cone : 5700.
*
ditto for a quadric : 5635, 5701.
cond. for intersection of two planes
*
touching a quadric : 5703.
figtires, relation between : A.55 : J.52 :
M.3.
kinematics of : Q.16.
and line, problems : CD.2 : J.14.
motion of : JP.2 : LM.7.
point-systems : J.77 ; perspective,
Z.17.
representing a quadric : N.71.
Plane coordinate geometry : 40015473.
Planimeter : A.58 : Mcl.2,3 ; Amsler's,
5452,0.77; Trunk's, A.44; polar,
A.51 : N.80.
Planimetrical theorems : A.37,60.
Pliicker's complex surfaces : M.7.
Pliickerian characteristics of a curve
discriminant : Q.12.
Pliickerian numbers of envelopes : 0.78.;.
Point-pair, absolute on a conic : Q.8.
harmonic to two such : Z.13.
Point-plane system : M.23j.
Points :
in a plane, relation between
four: A.2,26.
*
tg.eq of two : 4069, 4i13.
on a circle and on a sphere : N.82.
of equal parallel transversals : A.61.

Points (continued) :
four, or lines, ths : CD.8.
at infinity on a quadric : N.65.
roots in a closed curve : N.68.
in space, represented by triplets of
points on a line : LM.2.
systems: JP.9 -. M.6,25: N.58.
of cubic curves : Z.15.
three : coordinates of, N.42; pr, A.8.
Polar: 1016,4124: A.28: J.58: gzLM.2 :
Me.64,66: N.72,79.
of conies : 4762 : thsN.58.
of cubic curves : J.89 : L.57 : M61.5 :
Q.2.
curves, tangents of : N.43.
developable : 5728.
inclined : N.59.
line of two points with respect to a
quadric : 5685.
plane: Q.2: Z.22.

of a quadric : 5678, 5687 : An.71 ;

of four, LM.13.
of a quartic : L.57.
of 3 right lines : A.l.
subtangent: 5133.
Polar surface:
of a cubic: J.89;
twisted, Z.23o,24.
of a plane : C.60 : N.66.
of a point : N.65.
tetrahedron: J.78: N.65: Z.13.
of a triangle : A.59 ; perspective, J.
65.
Pole :
of chords joining feet of nor
mals of conic drawn from points
on the evolute : N.60.
of the line Xa+fifi + vy: 4671.
of similitude : 5587.
Pole and Polar : 1016, 4124.
Political arithmetic : trA.3638.
Pollock's geo. theorems : Q.l.
Poloids of Poiusot : CD.3.
Polyacrons, A-faced : Man.62.
Polydrometry : A.38,39.
Polygonal numbers: 287: Pr.10.11,12-,
13.
Polygonometry : thsAn.52 and J.2,47 :
Mem.30.
Polygons (see also " Regular polvgone"): An.cn53,63: JP.4,9 : N.
74: Z.ll ; theorems: A.1,2 : C.
26a: prsCD.5: Mel.2 : N.58.
area of : 748, 4042 : J.24 : N.48,52.
articulated and pr. of configuration,
tr: An.84.
centroid : N.77.
of circular arcs, cn : A.3 : J.76.
classification : Q.2.
division into triangles : A.1,8 : L.38-,
39, : LM.13 : Pr.8.
of even number of sides : LM.1.

INDEX.

Polygons(continued) :
family of : N.83.
maximum with given sides : J.26.
of n+2t sides, numbers related to:
A.62.
of Poncelet, metrical properties : L.79.
semi-regular: JP.24; star, A.59 and
L.79,80.
sum of angles of : A.52 : N.50.
Polyhedral function (Prepotentials) :
CP.13,.
Polyhedrons : 906 : 0.462,6062 : J.3 :
JP.4,9,,15,24 : L.66 : Man.55 : N.
83 : P.56,57 : Pr.8,9,112,12 : Q.7 :
Z.11,14 ; theorems : E.3942 : J.
18: N.43; F+S = E+2: 906:
A.24: E.20,273.
classification of : C.51,52.
convex : angles of, C.74 ; regular, A.
59.
diagonals, number of : N.63.
Euler's theorem : J.8,14 : Mem.13 :
Z.9.
minimum surfaces of : A.58.
maximum : regular, C.6L; for a given
surface, M.2 : Mel.4.
regular: ellipsoidal, 0.27; self-conj.,
A.62 ; star, A.62 : thsC.26.,.
symmetrical: J.4: L.49.
surface of : A.53 ; volume, J.24 : N.52.
Polynomials : geometry of, JP.28 : th
Q.14
determined from its partial differen
tial : A.4.
product of two: N.4k
system of : L.56.
of two variables analugousto Jacobi's:
A.16.
value when the variable varies be
tween given limits : 0.98.
Polyzonal curves, ^/U+ </V+= 0 : TE.
25.
Porisms L.59 : P.1798 : Q.ll : TE.3,4,9.
of Euclid: C.29,48,5659 : L.55.
of two circles : Me.84.
Feriuat's fourth : A.46.
of in- and circum-polygon : Me.83 :
P.61.
of in- and circum-triangle : LM.6,9 :
Q.l.
Poristic equations : LM.4,5.
Poristic relations between two conies :
LM.8.
Position, pr. relating to theory of num
bers : M(51.2.
Potential: thsAn,82 : C.88: G.15 : J.
20,32,63,70,th81,85: M.2,3: N.70.,:
Z.17.
of a circle : J.76.

913

Potential(continued) :
of cyclides : 0.83 : J.61 ; elliptic, Me.
78.
cyclic-hyperbolic, tables : J.4,63,72,83,
94.
of ellipsoids: J.98 : Me.84: Q.14;
two homog., J.63,70.
elliptic: J.472.
of elliptic disc, law, r~3 -. Q.14.
Gauss's f. and theory : Z.84.
gz. of first and second : L.79.
history of : J.86.
of homogeneous polyhedra : J.69.
Jellett's eq. and ap. : Q.16.
Newton's: M.11,13,16.
one-valued: J.64.
p.d.eof: 0.90.
Poncelet's ths : Z.3.
a related integral : L.45.
of a right solid : J.58.
of a sphere : Me.81 ; surface of, Me.
83: Q.12 : Z.7.
surfaces : J.54 ; conicoids, &c, 79.
vector : Me.80.
Pothenot's problem of the sphere : A.
44,47,54,.
Powers :
angular functions, &c. : J.72.
and determinants, relation : Z.24.
of negative quantics : Me. 73.
of polynomials : JP.15.
Power remainders : M.20.
Prepotentials : P.75.
Prime divisors of quartics : J.3.
Prime factors ofnumbers : J.51 : N.71,76.
Prime-pairs : Me. 79.
Primes : 349378 : A.2,19 : AJ.7 : fAn.
69: C.thsl3,49,50,f 63,962 : G.5:
J.thl2,thl4,20 : L.52,54,th79 :
LM.2 : M.21 : Me.41: N.46,56:Pr.
5 : Q.5 : up to 10, M.3.
in ar.p : Z.6.
calculation of: J.10: in 1st million,
M.25.
in a composite number : C.32.
distribution of, ap. of recurring series :
0.82.
division of a prime 4 + l into sum of
squares: J.50 : ditto of 8?i+3,
7+2, and 7n+4, J.37.
even number = sum of 2 odd primes
(?) : E.10: N.79.
Fibonacci's problem : LM.ll.
of the following forms and theorems
respecting them: 4/i + l, 4fc + 3 ;
L.60: 6fe + l; .1.12: Sk + 1; L.61,
62: 8^+3; L.58,60,614,622: 8fc+5
or 7; L.60,61: 12fc + 5; L.61,63 :
16fc+3; L.61 : 16fc+7; L.60,61:
16fc + ll; L.6062: 16& + 13; L.
61 : 20fc+3, 20fe+7 ; L.63,64:
6 A

914

INDEX.

Primes(continued) :
24&+1, 5, or 7 ; L.61, : 24k + \l
or 19; L.60 : 24fc + 13; L.61 :
40fc+3; L.61: 40fc + 7or23; L.
60j,613: 40A-+11 or 19; L.60:
40fc+27; L.613: 120A-f31,61,79,
or 109; L.61: 168/fc+43,67,orl63;
L.63 : m* + knt with k = 20,36,44,
56, or 116, and n an odd number;
L.65 : 4in2+5n- with m odd ; L.
66.
general formula for : C.63.
of a geo. form, limit of : C.74.
irreducibility of l+te+.-.+ss*-1, when
p is a prime : J.29,67 : L.50 : N.49.
law of reciprocity between two, ana
logue : J.9.
* logarithms of (2 to 109) : page 6.
* number infinite : 357 : Mo.78.
number within given limits : A.64 :
C953: M.2: Z.5.
number of digits in their reciprocals :
Pr.223,23.
number in a given quantity : Mo.59.
product of n, th : N.74.
* relative : 34950, 355,-8, 373 : J.70:
L.49.
tables of : from 10s to 100001699 : of
153 of the 10th million :
of cube roots of to 31 places : Me.78.
of sums of reciprocals and their
powers : Pr.33.
M=a'V for a prime or composite
modulus : Q.9.
totality of within given limits : AJ.4 :
L.52.
transformation of linear forms of into
quadratic forms : C.87.
on that prime number X for which the
class-number formed from the
Xth roots of unity is divisible by
X: Mo.74.
Primitive numbers : thC.74.
Primitive roots : C.64 : JP.ll : fL.54 :
taJ.45.
of binomial eqs. : thsL.40 : N.52.
of primes : J.49.
product of, for an odd modulus : J.31.
of unity: C.92.
Abel's theorem : An.56.
divisors of functions of periods of :
C.92.
period, Jacobi's method : C.70.
of primes : J.49 : and their residues,
Me.85.
sum of, for an odd modulus : J.31.
table, for primes below 200: Me"m.
38.
table of, for primes from 3 to 101 :
J.9.

Principal axes of a body : 5926,'60,

72,77 : J.5 : JP.15 : L.47 : N.46.
Prismatoids : A.39 : volume of, Z.23.
Prismoids : A.39.
Prism : volume of, A6.
Probability: 309: A.l3,19,47 : C.65,97:
CP.9: E.274,30: G.17: 1.15: J.
262,30,,33,34,36,42,50 : L.79: N.
51,73: Z.2.
theorems : Bernoulli's : LM.5 ; P.62 ;
TE.9,21.
problems: A.61,64: CD.6 : E. all the
volumes: G.16 : L.37 : Me.4,6:
Q.9.
de l'aiguille, &c. : L.60.
in analysis : J.6.
on decisions of majorities : L.38,42.
duration of life : C'M.4.
rouge et noir : J.67.
drawing black and white : 1.14 : L.41.
errors in Laplace, p. 279, and Poisson,
p. 209: CP.6.
games: A.ll : head and tail ; C.94.
of hvpothesis after the event: 324:
MeM.3.
local : P.68 : exE.7.
notation : LM.12.
position of double stars : Pr.10.
principal term in the expansion of a
factorial formed from a large
number of factors : C.19.
random lines : Pr.16 : E. frequently.
repeated trials : 31721 : C.94.
statistics : L.38.
testimony and judgment : TE.21.
Products :
continued : Me.77s.
of differences : thQ.15.
of 4 consecutive integers : N.62.
of inertia: 5906.
infinite, convergency of : A.21 : transf.
of, C.17.
of linear factors : C.9.
of n quantities in terms of sums of
powers : Me.71.
of 2 sums of 4 squares (Euler) : Q.
16,17.
systems of: L.56.
Progressions: 7993: An.64 : C.20:
G.6,,7,11 : of higher order, G.12-.
with n = a fraction : N.42.
Projection : 1075, 4921 : A.3,6,12 : prsJ.
70: Q.21,prl3.
and new geometry : A.l.
central : G.13 : derivation from or
thogonal, A.62.
central and parallel, of quadrics into
circles : A.52.
of conies: 492135: A.66: J.37,86.
of a cubic surface : M.5.

INDEX.

915

Projection(continued) :
Pythagorean triangles : taA.l : E.20.
Quadratic equations : 45 : A.242 : with
of curves : 3.66: Z.IS^: loci of cen
tres, A.6.
imaginary coefficients, A.8.
on spheres : Q.14.
graphic solution : Me. 76.
tangents to : cnL.37.
real roots of : J.61.
of a curved surface on a plane : J.67.
solution by continued fractions : L.40.
of an ellipsoid on a plane : J.59.
by successive approximation : N.74.
of figures in one plane : A.l.
Quadratic forms (see also " Quadric
gauche : N.66.
functions"): trA.15 : Ac.7 : C.
of Gauss: M61.2.
85 : J.27,f39,54,56,76,86 : L.51*,
* map : Mercator's, 1093 : A.50 : G.18 :
73: M.6,23: Mo.683,748: thsAn.
JP.24: N.79,.
54: J.53: M.20 and Z.16,19.
* of one line on another : 5529 : on a
Dirichlet's method : Mo.64.
plane, A.6.
having one at least of the extreme
* orthogonal: 1087.
coefficients odd : L.67,69.
plagiographic : A.8.
Kronecher's : L.64.
of ruled quartics : M.2.
multiplication of : An.60.
of shadows : N.56.
number of the genera of : J.56.
of skew hyperboloid : Me.75.
number which belong to a real deter
of solids : Man.84.
minant in the theory of complex
* of the sphere : 1090 : AJ.2.
numbers : J.27.
* stereographic : 1090: A.3032 : L.
odd powers of sq. root of 12ijU+>7z:
42,54.
M61.5.
of surfaces on a plane : M.5.
positive : A.II2.
of surface of tetrahedron on a sphere :
reduction of : J.39 : L.48,56,57.
J.70.
relation, anal, investigation : Z.14.
* of two rectangular lines : 4934.
Quadratic loci, intersection of : AJ.6.
Projective :
correspondence between Quadratrix:
5338.
two planes and two spaces : G.22.
tangent,
cn
: N.76.
equations of a surface, relation to
of a curve : C763.
tg.e: E.2.
Quadrangle : prA.55 : C.95.
figures on a quadric : A.18.
of chords and tangents : A.2.
generation of alg. surfaces : A.1,2 : of
differential relation of sides : Me.77.
curve forms, An.14: J.54: M.25 :
dualism in the metric relations on the
Z.18.
sphere and in the plane : Z.6.
geometry : A.8 : An.75 : At.752,78 :
and groups of conies : A.l.
G.1214,17: J.84: M.17,18: ths
of two intersecting conies, area :
N.77: ths of Cremona, G.132,14:
LM.8.
thZ.ll.
metrical and kinematical properties :
loci and envelopes : G.19.
C.95.
and P.D. eqs. : A.l.
Quadrature: 5871-83: A.26 : An.50g:
point series : J.91.
CD.ths5,9: E.6: J.34: L.54:
and perspectivity of higher degree in
Mem.24,41: N.42,f55,54: N.75 :
planes : J.42.
TI.l.
Proportion : 68 : A.8 : G.6,13 : TE.4.
approximate : C.95 : N.58.
Pseudosfera : G.10.
of the circle : Me.74 : Pr.7,20.
Ptolemy's theorem : A.2,67 : At.19 : J.
Cote's method : N.56.
13: LM.12.
with equal coefficients, f. : C.90.
ext. to ellipse : A.30: inverse of, A.5.
Gauss's method: A.32 : C.84,90s:
Pure mathematics, address by H. J. S.
J.gn55,56.
Smith, P.R.S. : LM.8.
from integrals of differentials in two
Pyramid : triangular : A.l,3,21,28,32,
variables : M.4.
36 : J.3 : volA.14.
Laplace's formula, f.d.c : 3778.
vertices of : A.2.
of a small geodesic triangle, Gauss's
and higher w-drons : prsA.9.
th: J.16,58.
and prism sections, collineation, Ac. :
by lines of equal slope : 5881.
A.9.
quadrics : CD.l : L.63 : formed by
Pythagorean theorem: A.11,1 7,20,24:
intersecting cylinders, An.65.
gzJ.26 and N.62.
sphero-conic : J.14.
spherical analogue : A.44 : N.52.

916

INDEX.

Quadratures(continued) :
which depend upon an extended
class of d.e with rational coeffi
cients : C.92.
Quadric cones : 5590,5618,'54,'97 : N.66.
locus of vertex : C.52.
through six points : C.52.
Quadric functions or forms : A.13j,38 :
C.44,55,78,892,95: J.472: JP.28,
arith 32, positive 99 : algL.74 :
Mo. 58 : Bee also " Quadratic
forms."
bipartite: P.58.
in coefficients and in indeterminate
complexes : J.24.
equivalence of: C.93.
in n variables : Me.2 : disappearance
of products, N.55.
quaternary : An.59 : L.54 : M.5,13 :
whose det. < 0, C.96; and corres
ponding groups of hyperabelians,
C.98.
reduction of : C.91,93,96 : G.5 : to
sum of squares, Mel.5.
represented by others : 0.93.
transf.of : C.86 : CD.4 : G.l : LM.16 :
N.66: Q.17.
reciprocal: J.50.
invariability in number of pos. and
neg. Bquares : J.53.
with two series of variables : C.94.
Quadric surfaces : 55825703 : A.32,4,
12,16,45,56 : An.52,geo60 : At.51 :
C.76: CD.3: G.13,14: J.1,18,38,
f42,63,69,89 : JP.6 : L.39,43,50 :
M.2,23: geoMe.74: N.56,57,583,
59s,60,61,77 : Z.5,cnl3.
theorems: An.54 : CD.5 : J. 54,85 :
L.43 : Mo.79 : N.632,642 : Q.2,4.
problems: An.61 : C.60: J.73: N.
58: Q.10.
analogy with conies : Me. 72.
anharmonic section of : G.12.
bifocal chords of : CD.5.
* central equations : 55995672.
* central sections : 5650 : area, 5650 :
axes, 5651.
locus of focus : N.66.
* non-central sections : 5654.
*
centre : N.75 : area, 5655.
* centre, coordinates of : 5690: A. 16.
* circular sections : 5596, 5601,'6,'19 :
C.43: CD.l: CM.l: E.30: J.47,
65,71,85: N.51.
common enveloping cone : G.6.
* condition for a cone : 5699.
* conjugate diameters : 5637 : N.42,61 ;
parallel, G.l ; rectangular sys
tem, L.58.

Quadric surfaces :
conjugate diame
ters (continued) :
*
parallelopiped on them : vol. 5648 :
sum of squares of areas of its
faces, 5645 : do. of reciprocals of
the perpendiculars on its faces,
5644.
*
sum of squares of their reciprocals :
5643: ditto of their projections
on a line or plane, 5646.
construction and classification by
projective figures : A.9.
correlation of points and planes on :
CD.5.
* cubature of : 612665 : A.14.
* diameters : 5677,-88.
* diametral plane : 5636 ; gn.eq, 5689.
of constant sectional area : N.43.
director-sphere of : Q.8.
duplicate: CD.6.
* enveloping cones of : 566472,5697.
equation between the coeffs. : J.45.
without foci : L.36.
focales, a property from the theory
of: L.45.
* general eq. : 5673 : CD.5 : CM.4 : LM.
12,13 : M.l : Me.64.
*
condition for a cone : 5699.
condition for a sphere : Q.2.
coefficients : A.l.
* generation: 560724, N.47,75; Jacobi's, J.73.
bomofocal : L.60.
indices of points, lines, and planes,
theory: N.704.
* intersection of two : 5660, C.64- : G.
6 : M.15 : Q.10 ; ruled, N.83 ; with
a sphere, ths, N.64.
parameter representation of: M.15.
tangent to : LM.13.
* intersection of three : 5661, J.73 :
TN.69.
loci from : A.27.
* normals of : 5629-32 : C.78 : J.73,83 :
N.63,78: Q.8.
oblique coordinates : N.82.
* polar plane of : 5681 8.
Pliicher's method : L.38.
of revolution : N.72,81 ; through 5
points, 66 and 79.
self-reciprocal : M.25 : Z.22.
sections of : J.74.
similarity of two : CD.8.
with a " Symptosen-axe": A.60,61.
system of: Q.15; reduc. and transf.,
L.74: Z.6.
* tangents: 5677: G.12.
* tangent planes : 5626,78 : CM.l :
cnJ.42: LM.ll: N.46.

INDEX.

Quadric surfaceB(continued) :
locus of intersect, of three : LM.15.
through 9 points or under 9 condi
tions : A.17: CDA: J.24,62,68:
L.58,59: Q.8: Z.25.
under 8 conditions : C.62.
through 12 points : G.17,22.
through a twisted quintic curve : E.
38.
transformation of two, by linear sub
stitution into two others, in which
the squares only of the variables
remain : J.12.
principal axes : see " Axes."
principal planes of : L.36 : N.67,71 :
Z.24.
and their umbilics : C.64.
volume of segment : A.27 ; of oblique
frustum, E.19.
Quadricuspidals : L.70.
Quadrilateral: A.6,48 : Me.66 : thG.5:
prsN.43.
area : N.48 ; as a determinant, N.74.
between two tangents to a conic and
the radii to the points of contact :
A.53*
bisectors, a property : N.75.
* and circle, geo. : 733 : Q.5.
* complete : 4652 : A.24,69 : At.63 ; mid
points of 3 diagonals, collinear,
Me.73.
and conic : 4697 : N.75,76.
*
and in-conic : tg.c4907: Q.ll.
cn. with given sides and equal diags. :
A.5.
convex : A.66 ; area, Q.19.
Desargues' theorem : Z.24.
plane and spherical : A.2 : Z.6.
and quadrangle : A.l.
right-angled : A.2,3.
with sines of angles in given ratio :
A.2.
* sum of sqs. of sides : th 924.
th. extended to 3 dimensions : J.56.
Quadruplane : LM.14.
Quantics: 1620: C.47: J.56: LM.6:
N.48: Cayley P. 54,56,58,59,61,
67,71 and 78: Pr.7o,8,9.,,ll,15,17,
18,23 : thsCD.6 : J.53 : N.53 :
Q.14; Cauchy, Pr.42.
derivatives, relation between 1st and
2nd: E.39.
derivation from another by linear
substitution : C.42.
of differentials : J.70,71 : Mo.69.
index symbol of : CD.5.
integration of a rational : C.97.
in linear factors : C.50 : L.ths61 : Q.6.
transf. of ax? +ly2+cz~+dws,by linear
subst. : J.45.

917

Quantitative function, transf. of : CD.3.

Quartic equations : see " Biquadratic."
Quartic curves : An.76,79 : At.52 : C.37,
ths64,65,77,98 : CD.5: G.14,16 :
J.59: M.thl,4,7,12: prN.56.
aV+6pH = 0: M.l.
and Abel's integrals : M.ll.
binodal, mechanical cn. of : E.18.
characteristics of a system : C.75.
chord of contact, eq. : H.17.
classification by inf. branches : L.36.
with cuspidal conies : M.19.
with 3 cusps of 1st kind : An.52.
degenerate forms : LM.2.
developable reciprocated : Q.7.
with a double line : A.2 : C.75.
with a double point : M.19; two, C.97. ;
three, Q.18; three of inflexion,
N.78.
with double tangents : J.66.
and elliptic functions : J.57,59.
of 1st kind and intersections with a
quadric : An.692.
generation of C.45 : J.44 ; 3rd class,
J.66 and Z.18.
16 inflexion points of 1st species of :
trZ.282: E.32.
and in-pentagon, th : M.13.
oblate: C.74.
parameter representation of : M.13.
penultimate : Me. 72.
with quadruple foci : Q.18;;.
rectification : C872.
and residual points : E.34j.
and secants : M.12.
singularities of : L.75 : M.14.
synthetic treatment of : Z.23.
through which one quadric surface
only can pass : An.61.
trinodal: thE.30.
unicursal twisted : LM.14.
Quartic surfaces : A.12: C.70 : G.11,12 :
LM.3,19: M.1,7,13,18,20: Mo.66,
72 : N.70,87, : Q.10,11.
containing a series of conies : J.64.
with a cusp at infinity : LM.14.
with double conic: A.2: M.1,2,4:
Mo.68.
with eq. Sym. det. = 0 : Q.14.
generated by motion of a conic : J.61.
Hessian of : Q.15.
and 2 intersecting right lines : M.3.
with 12 nodes : Q.14.
with 16 nodes : J.65,73,83,84,85,86,87,
88: Mo.64; principal tangent
curves of, M.23 and Mo.64.
Steiner's : C.86 : J.64,.
with a tacnode at infinity at which
the line at infinity is a multiple
tangent: LM.13.

918

INDEX.

Quartic surfaces(continued) :
with triple points : M.24.
Quaternions: C.86,98,: CD.4 : G.20 :
M.11,22: Me.62j*,64,81 : P.52 : Q.
6 : TE.27,28 : T1.21.
ap. to linear complexes and congru
ences : Me.83.
ap. to tangent of parabola: AJ.I2.
elimination of a/3y from the conditions
of integrability of Suadp, &c. :
TE.27.
equations: C.98; linear, 99a: of sur
faces, Joachimstahl's method,
E.43.
vp<t>p=0: TE.28: qQqQ' = 0,
Me.805.
finite groups : AJ.4.
f. for quantification of curves, sur
faces, and solids : AJ.2.
geometry of : CD.9.
integration ths : Me.85.
transformations : Man.82.
Quetelet's curve : 5249.
Quintic curves : cnM.25.
Quintic equations : AJ.6-,7 : An.65,68j :
C.462,48,50,61,6273,803,85 : J.59 :
M.13,,14,15: P.64: Pr.ll : Q.3 :
TI.19 : Z.4.
auxiliary eq. of : Man.l53 : P.61 : Q.3.
condition of transformability into a
recurrent form: E.35.
irreducible : AJ.7 : J.34.
functions of difference of roots : An.
69.
reduction of : Q.6.
resolvents of : 0.632.
whose roots are functions of a varia
ble: Q.5.
solution of : An.79 : J.59,87 : N.42 : Q.
2,18.
Descartes' method : A.27.
Malfatti's : An.63.
in the form of a symmetrical deter
minant of four lines : An.70.
Quintic surfaces : LM.3.
having a quintic curve : An.76.
Quintics, resolution of : Q.4.
Radial curves : LM.l : of ellipse, Q.18.
of conies, catenary, lemniscate, &c. :
E.24.
Radiants and diameters of a conic : C.
26.
Radical axis (see also " Coaxal Cir
cles ")|: 908,98499,4161: LM.
2: Me.66.
of symm. circle of a triangle : A.63.
of two conies : Q.15.
Radical centre : 997.
Radical plane : 5585.

Radii of curvature of a surface : A.ll,

55 : Q.12 : Z.8.
principal ones : L.47,82 : M.3 : Me.80 :
K55.
Radii of curvature of a surface : 5795
5817: A.11,55: Q.12: Z.8.
ellipsoid: 5831.
flexible surface : L.48..
principal: 58146: L.47,82: M.3:
Me.80: N.55.
constant : Me.64.

for an ellipsoid : 5832.

equal and of constant sign : C.41 :
JP.21 : L.46-,50.

Euler's theorem : 5806.

one a function of the other : An.65 :
C.84: J.62.
product constant : An.57.
reciprocal of product : An.52.
sum constant : An.65.
sum = twice the normal : C.42.
Radius of curvature of a curve : 5134 :
A.cn4,9,31,33 : CD.7: J.2,45: ths
M.17: N.62,74: q.c and t.c Q.12 :
Z.3.
absolute : CM.1.
circular : 5736 ; ang. deviation, 5746.
of conies : 1259 : A.9 : CM.l : J.30 :
L.36 : Me.66 : Mei.2 : N.45,68-.
at a cusp or inflexion point : N.54.
in dipolar coordinates : Me.81.
of evolutes in succession : N.63.
of gauche curves : N.66.
of a geodesic : L.44 : on. an ellipsoid,
An.51.
and normal in constant ratio : N.44.
of normal section of 0 (*, y,z) = 0 :
5817.
of a parabola : 1261, 4542.
of polar curves : A.51 : CM.2.
of polar reciprocal : N.67.
of projection of a curve: N.61 ; of
contour of orthogonal projection
of a surface, C.78.
of a roulette : 5235 : N.73.
transf. of properties by polar recipro
cals : L.66.
of tortuous curves : Mem.10.
of circum-sphere of a tetrahedron in
terms of the edges : N.74.
Radius of gyration : 5904.
Radii of two circles which touch three
touching two and two : A.55.
Radius vector of conic : J.30 : N.47^.
Ramifications : E.30s,33,40,pr 37,th27.
sol. by a diophantine eq. : Mo.82.
Randintegral : J. 71.
Ratio and proportion : 68 ; compound,
74.
of a* : br : geo.cnN.44.

INDEX.

Hat in and Proportion(continued) :

of differences of geo. quantities : C.40.
* limits of : 753.
* of segments of lines and triangles,
geo. : 92932.
* of two distances, geo. : 9268.
Bational :
derivation, cubic curve :
AJ.3,.
divisors of 2nd and 3rd degrees : N.
45.
functions, development of : AJ.5.
Rationalisation of:
alg. fractions:
A.13,33,35.
alg. equations: A.13: CD.8 : J.14 :
P.14: TI.6.
alg. functions : A.69.
a series of surds (Fermat's pr.) : A.35.
Rational functions :
of n elements :
M.14.
infinite form systems in : M.18.
Ray systems (see also "Congruences"):
J.57: L.60,74: Me.66: N.60,613,
622: Z.16.
of 1st order and class and linear pen
cils : J.67,692: Z.20.
1st and 2nd order : Mo.65 : M. 15,17.
2nd order and class : J.92.
3rd order and 2nd class : J.91.
6th order and 2nd class : J.93.
2nd class and 16 nodal quartics : J.86 :
Mo.64.
complex of 2nd degree and system of
2 surfaces : M.21.
and refraction theory : Q.14. : TI.15
17.
forming a group of tangents to a sur
face : Z.18.
infinite geometry of : Z.17.
Reciprocal polars : 4844, 5704 : A.36 :
gzE.24: J.77: LM.2: N.48,49:
num.fG.21.
Reciprocal : of a circle : 4845.
* cones : 5664, 5670.
* of a conic : 48668.
* of a quadric surface : 57178.
radii: M.13.
relations : J.48,79,90 : M.19,20.
* spiral: 5302.
* surfaces : 570419 : J.79 : M.4,10 : P.
69.
curvature of : L.77.
degree of : CD.2 : TI.23.
of Monge : C.42.
*
of quadrics : gn 5705 ; central, 5706.
of surface of centres o a quadric :
Q.13.
of the same degree as their primi
tives : Mo.78.
theorems on conies and quadrics : L.
61.

Reciprocal (continued) :
transformation, geo. : L.71.
triangle: th Q.l ; and tetrahedron,
Q.l.
Reciprocants : LM.172.
Reciprocity : anal., A.7 : geo., CD.3.
Reciprocity law : 3446 : AJ.l.th 2 : C.
902 : J.282,39 : LM.2 : Mo.58 : d.e,
3446 and A.33.
in cubics : M.12.
history of: Mo.75.
for power residues : C.84; quadratic,
C.24,88: J.47 : L.47j and Mo.80,
84,85 ; cubic, in complex numbers
from the cube roots of unity, J.
27,28.
quadratic F% system of 8th degree :
J.82.
supplementary theorem to : J.44,56.
Rectangle : M. I. of, 6015.
Rectangular hyperbola : 4392 : Me.62,
66,72: N.42,65.
Rectification of curves : 5196 : A.26 :
Ac.5: An.69: CR.95: CD.9: G.
11 : J.14: L.47: N.ths 53,64.
approximate : M.4 : M61.4.
by circular arcs : C.77,85: L.50: M<5m.
302.
by elliptic arcs or functions : J.79 :
Mem.30.
by Poncelet's theorem : C.94.
mechanical : Z.16.
on a surface : Mdm.22.
Rectifying :
developable : 5727.
line : 5726,-51 : N.73.
plane: 5726.
surface: 5730.
Reflexion :
from a revolving line :
TE.28.
from plane surfaces : A.60.
from quadric surfaces : J.35.
Refraction curve : A.51.
Regie a calcul : C.58 : N.69.
Regular polygons : 746 : A.21,cn24,39 :
L.38 : M.cn6,13 : N.42,44,47.
convex : Me.74.
eqs. of and division into eqs. of lower
degrees, tr. : A.46.
in and circum : N.45 : Q.2.
in space : Me.75.
spherical : N.60,67.
star: J.65 : Me.74: N.49.
funicular : N.49.
5-gon : M.83.
7-gon: A.17.
7-gon and 13-gon : M.6.
8-gon : A.6.
12-gon : complete, C.95.

920

INDEX.

Regular Polygons(continued) :
17-gon: A.o: J.cn24,75 : N.74; and
division of the circle, A.42.
eqs. for sides and diagonals : A.40.
Regular polyhedrons : 907 : A.ll3,47 :
Pr.34: Me.66 : Q.15.
* relation of angles : 909 : Me. 74.
volumes by determinants : A.57.
Related functions : M.25.
Relationship problems : E.35,38-2.
Relative motion : N.66.
Rents: A.40.
Representative functions : M.18.
Representative notation : Q.6.
Reproduction of forms : 0.97.
Reptation : N.54.
Residues: A.26: C.12,,13,32,ap32.,>41.44,
49: CM.l: J.252,31,89: L.38: M61.
4: N.46o,Mem 70.
ap. to infinite products : C.17.
ap. to integrals whose derivatives in
volve the roots of alg. eqs. : C.23.
ap. to reciprocity law of two primes
and asymptotes : C.76.
of complex numbers : Mo.80.
primes of 5th, 8th, and 12th powers :
J.19.
biquadratic : C.64 : J.28,39 : L.67.
cubic: A.43,63: 0.79: J .28,32 : L.76.
quadratic: Ac.l : J.28,71 : Q.l.
ext. of Gauss's criticism : Mo.76.
of primes, also non-residues : L.42 :
Mel.4.
and partition of numbers : J.6I2.
quintic: 0.76: Z.27.
septic : C.80.
of 9exp.(9exp.9) by division by
primes : A.35.
Residuation in a cubic curve : Me.74.
Resultant alg. : M.16: extC583.
and discriminants and product of dif
ferences of roots of eqs., relation :
Me.80.
of two equations: J.30,50,53,54 : M.3:
P.57,68.
of two integral functions : Z.17.
of n equations : An. 56.
of covariants : M.4.
of 3 ternary quadrics : J.572 : N.69.
Reversible symbolic factors : Q.9.
Reversion of angles : LM.6.
Rhizic curves : Q.ll.
Rhombus : quadrisection by two rect
angular lines : Mem.ll.
circumscribing an equil. triangle : A.
45.
Riemann's surface : LM.8 : M.6,182 :
thsZ.12.
of 3rd species: M.17.
new kind of: M.7,10.

Riemann's surface (continued) :

irrationality of : M.17.
Riemann's function : A.68 : J.83 : M.21.
ext. to hyper-geo.-f unctions of 2 vari
ables : C.95-2.
5-formula, gz : Ac.3.
Right-angled triangles : 718 : prs A.2.
with commensurable sides : B.33.
Right cone : Me.72,73, 75,76.
Right line: A.49,57 : fQ.15: t.cMe.62.
and circle : ths N.56.
coordinates of: G.10.
and conic : Q.7 ; en. for points of sec
tion, A.59,66 : N.85.
*
quadratic for abscissas of the points :
4319.
*
tg.e of the points : 4903.
*
condition of touching: 4315,4323,
t.c 5017.
* drawn from x'y' across a conic : quad
ratic for the segments in an el
lipse, 4314; parabola. 4221 ; gen.
eq., 4494 ; method, 4134.
* joining two points, coordinates of
point dividing the distance in a
given ratio : 4032, t.c 4603, 5507.
*
tg. eq. of the point : 4879.
* joining two points and crossing a
conic : quadratic for ratio of seg
ments in an ellipse, 4310 ; para
bola, 4214 ; gn. eq., 4487, t.c 4678 ;
method, 4131.
* constants, relations between : sd.
5515.
* coordinates of, relation between the :
4897.
* and curve : 4131 5 : ths. in which
pairs of segments have a constant
length, C836; a constant product,
0.82.1,832; a constant ratio, 0.83;
ths. in which systems of 3 seg
ments have a constant product,
C.83,.
crystallography : A.34.
* equations of: 405266, p.c 4107, t.c
46058 ; sd 5523, q.c 5541.
geometry of : A.64 : thsJ.8.
* at infinity : 46124, tg.c4898.
*
cond. for touching a curve : 4900.
pencils of: C.70 : L.72 ; quadruple,
J.67.
and plane : prs CD.l and CM.2 : t.o
and q.c Q.5.
pole of: t.c 4671 : tg.e 4674.
* and quadric : 5676 ; harmonic divi
sion, 5687.
and quadric of revolution : N.82.
six coordinates of : CP.ll.

INDEX.

Bight line(continued) :
system of : At.68 : G.9,10,16.
of 1st degree, G.6 ; of 2nd, G.7.
in space: G.113,12 : L.46.
and planes, geo. of 2nd kind : At.
65.
* three, condition of intersection: 4097:
t.c4617.
* three points lying on, cond. : 4036,
t.c4615.
* through a point : 4073, 40889, 4099,
t.c4608.
*
condition: 4101.
*
and perp. or paral. to a given line :
sd5538-9.
* through two points : 4083, sd5537 ;
t.c4616, 4789 ; p.c4109 : on a
conic, equation of, ellipse 4324,
parab. 4225.
through four lines in space: A.l :
CM.3 : Gergonne's pr. J.2 and
N.17.
* touching a surface : condition, 5786.
*
quadric : 5703.
planes or points through or on given
points, lines, or planes, number
of such : Z.6.
* two:
angle between them: 4112;
sd5520,5553: CP.2 : N.66.
*
bisector of the angle : 4113, sd5540,
q.c5543.
*
cond. of parallelism : 4076 ; t.c4618 ;
sd5531.
*
cond. of perp. : 4078 ; t.c4620 : sd
5532.
*
cond. of intersecting on a conic, gn.
eq : 4962.
*
cond. of either touching the conic :
4964.
*
cond. of intersection : sd5533.
coordinates : 4090, t.c4611.
*
shortest distance : sd55346.
drawn to the points of section of a
right line and conic, eq. of : A.69.
*
through origin, eq. of: 4111.
under given conditions : C.73,74 :
under four, C.68.
Right solid : M. I. of, 6018 : tbA.34.
Rodrique's th. : Me.80.84j.
Rolling cones : L.53.
Rolling and sliding solids : geo thsC.46.
Rosettes: N.48.
Rotation : CM.3 : LM.3j : infinitesimal,
C.78.
of system of lines drawn through
points on a directrix, modulus
of: C.21.
Roots of algebraic fractions : N.46.

921

Roots of an equation (see also "Equa

tions ") : 50,402 : A.14 : CM.2 :
0P.8 : E.36 .- J.20,31 : N.42,56 :
P.1798,37,64: Q.1,5,6.
of a biquadratic, cn : K.44.
by parab. and circle : N.87.
* commensurable : 502 : N.45,th57.
limits to the number : N.59.
* common: 462: C.80,88: N.55,69.
as continued fractions : CM.3.
continuity of : N.76.
in a converging series : C.23,38.
of cubic : L.44 : N.42.
of cubic and biquadratic : An. 55 : L.
55.
as definite integrals : J.2.
y"xy11 = 0 : Me.81 : P.64.
as determinants of the coefficients :
A.59,61.
* discrimination of : 409 : A.46.
* equal : 43247 : CD.5 : E.33 : Mdl.l :
P.1782: Q.9,18.
with equal differences : G.15.
existence of: A.15: CD.2:'CP.10:
E.36: G.2: J.5,44,88: LM.l : Q.
11 : TI.26.
expanded in power series : J.4S.
of the form a+
-/c+ : N.45.
forms for quadrics, cubics, and quartics: Z.24.
* functions of the roots of another eq. :
425,430 ; products in pairs, Q.13.
*
squares of differences: 541 : ap,N.
50: E.40.
as functions of a variable parameter :
C.30.
functions of : similar, L.54 ; relation
to coefficients, TE.28.
geo. cn of: JP.10.
in g.p : N.89.
in a given ratio : J.10.
* imaginary: 408: geo.cn, A.15,45 : C.24,
8688: JP.ll: L.50: N.46,47,
68j: approx. N.45,53 : Q.9.
between given limits : A.21 : L.44.
Newton's rule : Me.80 : N.67 : Pr.
13.
Newton-Fourier rule : Q.16.
*
Newton-Sylvester rule: 530: C.
994: LM.l: Me.66: Pr.14: Q.9 :
TI.24.
* incommensurable : 506 (see " Sturm's
th.")
infinite : N.440,45.
in infinite series : A.C9.
* integral, by Newton's method of
divisors : 459.
least : M.9 : TE.28.
* limits of : 448 : C.58,60,93 : geo CP.
12 : N.43,45,59o,72,802,81.
6 B

922

INDEX.

Boots of an equation(continued) :
number between given limits : A.l :
G.9: J.52: L.40.
the eq. containing only odd powers
of x : N.63.
*
Rolle's th. : 454: AJ.4: N.44: ext
L.64.
Rolle, Fourier, and Descartes : A.l.
number satisfying a given condition :
C.40.
product of differences : Me.80 : P.61.
* of a quadratic : 503.
of a quartic and of a Hessian, rela
tion: E.34.
ofquintics: C.59,60: LM.14 : TI.18.
rationalization of : P.l 798,14.
real: A.36.58: C.61 : J.50: JP.10 :
N.50: P.57.
of a cubic: N.72 : Z.2.
Fourier's th. : N.44.
developed in a series : L.78 : N.56.
limits of: J.l: N.53,79.
series which give the number of :
Z.2.
to find four : An.55.
* rule of signs: 41623: A.34: C.92,
982,993 : N.43,46,47,67,69,79.
separation of: A.28,70 : J.20: N.68-,
72,74,75,80,.
by differences : N.54.
for biquadratics : A.47.
for numerical : C.89,92 : G.6.
simultaneous eqs. : C.5.
* squares of differences : 541 : CM.l :
N.42,44: Q.4.
* sums of powers : 534 : E.38 : thJ.9 :
N.53,75 : gzMe.85 : Q.19.
in sums of rational functions of the
coefficients : Ac.6.
surd forms of : CM.3.
* symmetrical functions of : 534 : A.16 :
AJ.l : An.54,55,60 : C.44,45 : G.
5,11: J.19,54,81: Me.81: N.48,
50,55,66,84: P.57: Pr.8 : Q.4:
TI.25.
do. of the common roots of two eqs. :
N.60: Z.15.
do. of differences of roots : C.98.
which are the binary products of the
roots of two eqs. : An.79.
with a variable parameter : C.12.
which satisfies a linear d.e of 2nd
order: C.94.
Roots of numbers : 108 : A.17,26,35 :
C.58: B.36: Me.75: N.61,70.
* square root: 35: C.93 : N.45s,46,61,
70.
*
as a continued fraction : 195 : A.6,
12,49: CM.2j: L.47 : Me.85 :
Mem.10: TE.5: Z.17.

Roots of numbers(continued) :
to 25 decimal places : Me.77.
* cube root : Horner's method : 37 :
A.67.
of 2 to 28 decimal places : Me.76,78.
and 8q. root, limit of error: N.48.
fourth root : A.30.
nth root as a fraction : A.46.
Roots of unity : 47581 : C.38 : J.40 :
L.38,54,59: Me.75: N.43, : TE.
21 : Z.22.
cubic roots, alg. and geo. deductions :
C.84.
function theory : Z.22.
23 roots, composition of number 47 :
J.55,56.
Roulettes: 5229: Ac.63 : C.70: CP.7 :
J.65 : L.80,81 : N.56: TE.16 : Z.28.
areas of, and Steiner's transf. : E.35.
generated by a circle rolling on a
circle : JP.21.
by focus of ellipse rolling on a right
line: A.48.
by centre of curvature of rolling
curve: L.59.
Ruled surfaces: An.68 : CD.8 : G.3 :
J.8: L.78: N.61.
areas of parallel sections : Z.20.
and guiding curve : A.18.
of minimum area : L.42.
octic with 4 double conies : C.60.
P. D.eq. of: Me.77.
quadric : Me.68.
quartic: A.65; with 2 double lines,
A.65.
quintic: J.67.
represented on a plane : C.85j.
of species, p = 0 : M.5.
symm. tetrahedral : C.62.
torsal line : M.17.
transformation of : C.88.
Scales of notation : 342 : J.l : L.48,5ry,
10ry,20ry: Phil. Soc. of Glasgow,
vol. 8.
Screws: TI.25.
Scrolls: A53: CD.7: CP.ll : J.20,67:
M.8 : cubic, M.l : P.63,64,69 :
Pr.12,13,16: Z.cn28.
condensation of: LM.13.
cubic on a quadric surface : Me.8o.
flexure and equilib. of : LM.12.
ruled : A.68 : z = mxtfi, A.55.
tangent curves of : M.l2.
Sections of the cone : 1150.
Sectors and segments of conies and
conicoids : 6019 6162 : G.l :
thsZ.l.
Secular eq. has real roots : J.88.
Self-conjugate triangle : 4755, 4967 :
G.8 : N.67 : Q.5,10.

INDEX.

Self-conjugate triangle(continued) :
of 2 conies : Me.77 ; of 3 conies, 5025.
and tetrahedron in conies and quadrics : Z.6.
Self - enveloping carves and surfaces :
Z.22.
Self-reciprocal surface : Mo. 78.
Self-reciprocal triangle : 1020.
Seminvariants : AJ.7 : E.th42,6 : Q.
1921.
critical and Spencian functions: Q.4,6.
and symm. functions : AJ.6.
Septic equations : Mo.58.
Series: _(see also "Summation" and
" Expansions ") : 125 9, 149
59, 24895, 756817, 1460, 1471
2, 150073, 27089, 274360,
2852-64, 2880, 2*1168, 3781,
3820 : A.4,52,9,14,18,23,60 : No.39,
472: C.29,pr92: CP.9 : G.10: J.
3,17 ,34,38,th53 : L.th56,813 : Me.
64: N.59,th62,70 : Q.3: Z.15,16,
23.
Useful summations :
, X% , fcS
= log, 1
156.

923

Series(continued) :
anb+C(n,2)c&c.: J.31.
n'-n(n-l)'+C(n,2)(n-2y
285; r = n,CM.l.
J.37; with
<b = 1, J.2.
1
(a+nd)k : N.59.
deductions from
f?^*" V' = 2

\ a^n\J
(a+l)i"'2*nr
LM.9.

29404 : A.41 : LM.8 : 0.7 : with

n = 1,2...8; 2945 ; E.:!2,39 ; G.10 ;
N.79 ; Z.3. Note that by (2391),

2 : Mem.U : with x = l, J.5.

= log(l+): 155.
a +Y~-

2^,:A.61.
157.
= tan" : 791.

2 T 3!

= e*-l : 150.

1 e*
2! 3!
. Xs , X6 ,
"+Sl + 51 +
Xs , X5
X
. = sin x : 764.
3!^5!
__ e"+e
2! 41 6! '
2
:
a_a^ ,x _ = 1cos a; : 765.
2! 4!+6! '
+ ... +n": 2939 : A.65: Me.78.
j> = 1,2,3, or 4 : 276 : A.64 : E.34.
V2'+3p ... &1''y+S" ... :
J.7.
2 n*x" : A.27.
2(a+n)V: N.56.
2(an + fcxe)*~'': Z.15.
l_*-3 y (4)(n-5) .
. J20
2
2.3

2^, A.41
a"
X~f\+Wi~ -' E-44; witha;=1.
J.5.
2 (n-l)g
to! a""1 ' A.50.
IB"
2(-l)
(a+2) (1-*)"
. 1 E.36.
,()

Q.6.

2 a*i"' : 2709.
x(H)
: A.2
2 ^ TT.
2 1.3 ... 2n1 : L.60.
m!2" (2n + l) '
!+ (2_i)(3,_i)a! +
(g-1) (2s-l) (<r-l) (3'*-l)
+ ... : J.32,70.
'{(*-?)() + }"
2 anm+ain"-l
. + ... +a : A.35.

INDEX.

924

Series(contintted) :
2 A

, a pos. integer < a,

no+P
A",, = the general term of some
recurring series : C.86.
: A.34.

2"J,* -E(n^-) : 0.60.

* l(""2 + 10)/n:J.64;withfc = 2,
2960-1, J.8.
2 sin* (2n+l) </> / (2n+l)* : E.39.

* 2 (a"S*)/! 29223.
2 sin ^ 2r? L ^
COS
y
* 2w8in n6-. 2962: M.6.
2 Ancoa" 6c?Bn6: Z.l.
sin
* scc>+n<9): m: Q"3;
* 2c-^ns(a+/3),783;
* 2eo>+^>'7882/() Smntf: J.42: L.52.
'
cos
2 IT (m,n) z am nd> : J.41.
v ' ' cos r
2 i- tan : A.44.
2
2"
2/(nx) : L.51. 2An(j> (n) z, J.25,28.
if/(.+ i),A.a
from J as (1at)" das : A.47.
from f ' cos2 * log } + smx dx : G.10.
Jo
1sins;
of Abel: C.93: N.85.
application : thA.48 ; to aritb, G.7.
in a.p : see " Arithmetical progres
sion."
analogous series : N.69.
a.p and g.p combined : A.9.
with Bernoulli's nos. : An.53 ; and bi
nomial coefficients, A.23.

Series(continued) :
binomial (see " Binomial theorem") :
analogous series : E.35 : J.32 : N.
82 ; with inverse binomial co
efficients, Me.80.
coefficients independently deter
mined : A.18.
whose coefficients are the sums of
divisors of the exponents, sq. of
this series : Me.85.
combination : A.26.
* convergent: 239: A.2,6,8,14,26,41,67,
69: No.44: C.10,ll3,28j,40,43j: J.
2,3*11,13,16,22,42,45,76: L.39
42: M.10,2022: Me.64: N.45,
46,67,69,70? : P.87 : Z.10,11.
and of d.i with a periodic factor :
L.53.
power-series : A.25.
representing integrals of d.e : C.40j.
representing functions : M.5,22.
in Kepler's problems : Ac.1799.
multiplication of : M.24.
and products, condition : M.22.
whose terms are continuous func
tions of the same variable : C.36.
with constant ultimate differences
Pr.5,.
converted into continued fractions
J.32,33 : Mem.9 ; into products
of an infinite no. of factors, J. 12
L.57 : N.47.
in cosines of multiple angles : C.44
M^m.15.
and definite integrals : L.82 : Man.
46.
derived : A.22 ; from tan" 1 6, A.16.
developed in elliptic integrals of 1st
and 2nd kind : An.69.
* difference: 264: A.23,24.
differential transf. and reversal of :
Pr.7.
and differentiations : A. 10: J.36.
C.21: L.
Dirichlet's f. for 2
\p I n
46.
discontinuous : CP.6 : L.54: Me. 78,
82 N 85
divergent: A.64 : No.68: C.17.20:
CP.8,10 : J.11,13,41 : M.10 : Z.10.
division of : AJ.5.
double: C.63.
doubly infinite : CD.6 : M.24.
ext. of by any parameter : A.48.
factorial : 268 : M*m.20 : N.67 : TE.
20.
of fractions : L.40.
Fourier's : A.39 : C.91,92,96 : CM.2 :
Z.27.
of Gauss and Heine : C.73 : G.9.

INDEX.

Series (continued) :
* Gregory's : 791.
harmonic periodic : J.23,25.
of Hermite, a th. : E.29.
from infinite products : Me.733.
integration of infinite : A.3.
irrationality of some : J.37.
involving two angles : L.74.
Klein's higher : An.71.
of Lagrange : C.23,343,522 : L.57 : N.
76,gz85 : Q.2 ; remainder, C.53.
an analogous series : C.99.
of Lambert ; 2
: A.10 : An.68 :
1as"
J.9: Z.6.
Laplace's (d.c) : C.68.
of Laplace's functions : 2Y, C.882 ;
2n Y, C.44.
in Legendre's function X : An.76 :
CM.
of Leibnitz : J.89.
limits of : A.20 : Me.76 ; remainders,
C.34 ; by the method of means,
J.13.
from log (1 -fas), (1+aj)" and e* by in
termitting terms in the expan
sions : A.21.
modular : C.19.
multiple: C192; "regulateur" of,
neutral: CP.ll.
obtained by inversion from Taylor's
series : Mem.ll.
of odd numbers : A.64.
a paradox : He.72.
periodic, critical values of : CP.8.
of polynomials : C.96.
of posterns : G.6.
* of powers (see also " Numbers ") :
277: C.87: G.2: cubes, L.64 and
65 : M.23 : Mo.78 : Q.8 : Z.l.
approximate fractions : J.90.
of a binomial : Mdm.13.
in a convergent cycle, constants
in: M.25.
like numbers : N.71,77.
or multiples of 3 : A.272.
of terms in ar.p : L.46.
products of contiguous terms of :
Mem.18.
of reciprocals : Q.8.
* recurring : 251 : doubly, An.57 : J.33,
38 : Me.66 : M6m.24.26 : N.84 : of
circles and Bpheres, N.62 : f, Z.14.
represented by rational fractional
functions : J.30.
* reversion of: 551: J.52,54: LM.2 :
TI.7.
of Schwab : N.59.

025

Series{continued) :
self-repeating : CP.9.
of spherical functions : An.75.
of Sterling, for transformation : .1.59.
with terms alternately positive and
negative : C.64.
whose terms are the coefficients of
the same power of a singlo vari
able in a multiple integral : C.20.
in theory of numbers : 0.89-2.
transformation of : C.59 : J.7,9 : into
a continued fraction, M6m.20,
Z.7 j of 2f/(a:, t)dt and others,
C.13.
in a triangle problem : A.64.
trigonometrical (see above) : A.63 :
Ac.2: C.95,97: M.46,162,17,22,
24 : J.71,722 : representing an
arbitrary function between given
limits, J.4 ; conversion in mul
tiples of arc, L.51 ; symbolic
transf. of, Q.3.
triple: G.9.
two infinite, multiplication rule : J.79.
Seven-point circle : 4754c.
Seven planes problem : N.56.
Sextactic points of plane curves : Pr.
13,14.
Sextic curves : axi + byl+c* = 0, Q.15 ;
mech.cn, LM.2.
bicursal : LM.7.
and ellipse, pr : J.33.
Sextic developable : Q.7,9.
Sextic equation : C.64: M.20.
irreducible : J.37.
solution when the roots are connected
by (a-/3) (b-y) (c-a) +
(ab) lfie) (y-a) = 0: J.41.
Sextic torse : An.692.
Sextinvariant to a quartic and quartinvariant to a sextic : AJ.l.
Shortest distance :
between two
lines : 5534 : A.46 : G.5 : N.49,66.
between two points on a sphere : A.
14: N.14,67,68.
from the centre of a surface : A.63.
of a point from a line or plane : N.44.
Shortest line on a surface : A.23,37,64 ;
in spheroidal trigonometry, A.40.
Signs: CP.2,11: J.12 : Me.73 ; (=),
Me.75; (), CD.6,7: Me.85 : N.
48,49.
Similarity:
of curves and solids:
A.13.
Similarly varying figures : LM.16.
Simson line of a triangle : E.29.
Simpson's f. in areas : 2992 : 0.78.
Sines of higher orders : C.91<)922 ; ap.
to d.e, C.90s,91.

INDEX.

Sines, natural, limit of error : N.43,.

Sin6<6-^--. geo.N.75.
4
Sin-'(a!+ti/) : Q.15.
Sine and cosine :
extension of mean
ing : A.31 : C863.
* in factors : 807 : A.27 : J.27 : L.54.
of infinity: CP.8: Me.71 : Q.ll.
of multiple arcs (see also " Expan
sions "J : CM.4 : TI.7.
* of particular angles: 690; 3,6,...to
90, N.53.
sums of powers : An.l.
* tables, formation of : 688 : A.66 : N.
42,.
values near 0 and 90 : G.9.
* of (a6): 627; A.6,21,36.
Six-point circle of a triangle : Me.82,
83,.
Six points on a plane or sphere : LM.2.
Skew surfaces : see " Scrolls."
Sliding rule: LM.6.
Small quantities of second order : 1410.
Smith's Prize questions, solutions : Me.
71,723)4, 77.
Solid angle : A.42 ; section of, No.19.
Solid harmonics : Me.80.
Solid of revolution : A.60,67.
between two ellipsoids : A.2.
* cubature and quadrature of : 5877
80 : A.68 : N.42.
Space homology : G.20.
Space theories: An.70: LM.14: P.70 :
Z.17,18.
absolutely real space : G.6.
continuous manifoldness of two di
mensions : LM.8.
space of constant curvature : An.69,
73: J.86: M.12.
Plucker's " New geometry of " : G.8:
L.66: P.ll : Z.11,12.
Grassmann's " Ausdehnungslehre " :
AJ.l: CP.13: M.7,12: Z.24; ap.
to mechanics, M.12.
non Euclidean or n-dimensional : A.
6,29,58: ths64: A J.4,5 : An.71 :
C.75: G.6.10,12,23 : M.4,5,-7:
Me.ths68,72: Pr.37.
3-dim., J.83; 4-dim., J.83, M.24; 6dim., G.12.
angles (4-dimen.) : A.69.
areas and volumes : A.69 : CD.7.
bibliography of: AJ. 1,2.
circle: G.12,16,18.
conies : AJ. 5.
curves : C.79 : M.18.
Feuerbach's points : G.16.
hyperboloid : Z.13.

Space theories(continued) -.
imaginary quantities : Z.23.
loci (anal.) : C.24.
planes (4-dimen.) : A.68.
plane triangle : A.70.
point groups : thsAc.7.
polars and alg. forms : J.84.
potential function : An.82,83.
projection; M.19; 4-dim. into 3-dim..
AJ.2.
quadric, super lines of (5-dim.) : Q.12.
quaternions : CP.13.
regular figures : AJ.3.
reversion of a closed surface: AJ.l.
representation by correlative figures :
C813.
simplicissimum of nth order : E.44.
screws, theory in elliptic space : LM.
15,16.
21 coordinates of : LM.10.
Sphere: geo,C92 : ths and prs M.4:
q.cMe.62.
and circle : geo,A.57.
cn. from 4 conditions : JP.9.
cutting 4 spheres at given angles :
An.51: N.83.
cutting a sphere orthogonally and
touching a quadric, locus of cen
tre : TI.26.
diameters, no. of all imaginable: A.24.
* equation of: 5582.
5 points of: J.23: N.84.
illumination of : Z.27.
kinematics on a : LM.12.
sector of (eccentric) : A.65.
small circle of : Me.85.
touching an equal sphere : E.31,32 ;
as many as possible, A.56.
4 spheres, pr. : L.46.
4 touching a 5th : At.19.
8 touching 4 planes : E.19 : N.50.
16 touching 4 spheres : J.37 : JP.10 :
Me.cn82: Mem.10: N.44,47,65,
66,84: Z.14-,.
* volume, 4c. of segment and zone :
6050: A.3,32,39: An.57 : P.l.
Spherical :
areas : 902.
catenaries : J.33.
class cubics with double foci and
cyclic arcs : Q.15.
conies : thQ.3 ; and quadrangle, Q. 13 ;
homofocal, L.60.
coordinates : CD.l : CM.l,ap2 ; ho
mogeneous, G.6.
* curvature : 5728,'40,'47 : thE.34.
curves : A.35,36 : Mera.10.
of 3rd class with 3 single foci : Q.
17.
of 4th class with quadruple foci :
Q.18.

INDEX.
Spherical :
curves(continued) :
of 4th order : J.43.
with elliptic function coordinates :
J.93.
equidistant : An. 71 : J.25.
and polars : No.63.
rectification of : An.54.
ellipse : t.cQ.8.
quadrature, &c. : L.45 : N.484,54.
epicycloid : G.12.
excess: Mel2 : cn,N.46: f,Z.6.
of a quadrilateral : Me.75.
figures, division of : J.2j.
geometry : G.4 : J.6,8,13,thsl5 and
22 : M.15 : N.48,586,59 : Q.4 : TI.8.
harmonics or " Laplace's functions " :
An.68a: C.86,99 : CD.l : CM.2:
J.26,56,60,geo 68,70,80,82,90 : L.
45,48 : LIO -. Me.77,782,85: P.57 :
Pr.8,18: Q.73: Z.24.
analogues of : J.66 : LM.ll.
and connected d.i : Q.19-.
as determinants : Me.77.
and homogeneous functions : CM.2.
and potential of ellipse and ellip
soid : P.79.
and ultra-spherical functions : Z.12.
P* (cos y), to = oo : J.90.
f Pj/i-i/i, &c. : Q.17.
b',t] by continued fractions : JP.28 ;
q.Q.: P.70.
loci in spherical coordinates : TE.122.
oblong : An.52 ; area, J.42.
polygonometry : J.2.
polygons in- and circum-scribed to
small circles of the sphere, by
elliptic transcendents : J.5.
quadrilateral : A.4,40 : th,E.28 : N.45.
surface of : A.34j,352.
quartics : foci, Q.21 ; 4-cyclic and
3-focal, LM.12.
representation of surfaces: C.68,75,
944,96: M.13.
surface represented on a plane : Me.73.
triangle: A.9,ll,20,ths50,65 : B.f30 :
J.10,pr28: JP.2.
ambiguous case : Me.77,85j.
angles of, calculated from sides :
A.51.
by small circles, area : N.53.
*
and circle : 898 : A.29,33.
oos (A+B+C), i. : Me.72.
and differentials of sides and
angles : A.10.
and ex-circles : 898 : E.30.
graphic solution : AJ.6.

Spherical :
triangle (continued) :
and plane triangle : A.l ; of the
chords, A.33 : An.54 : Z.l.
*
right angled : 881 : A.51 ; solution
oy a pentagon, A.ll.
of very small sides : N.62.
two, relations of sides and angles :
A.2.
* trigonometry: 876: A.ths2,ll,13,28,
37: J.prse.ISs: LM.ll: N.42 :
Z.16.
d.e of circles : Q.20.
*
cot a sin b : Me.64 ; mnemonic, 895 :
CM.3.
derived from plane : A.26,27.
*
formula) : 882 : A.5,16,24,26 : N.45,
46,53 : graphically, A.25: ap. in
elliptic functions, A.40.
geodetic reduction of a spherical
angle: A.51s.
*
Cagnoli's th. : 904.
*
Gauss's eqs. : 897 : A.13,17 : J.7,12
LM.3,13.
Legendre's th. : C.96 : J.44 : L.41 :
M.l : N.56 : Z.20.
*
Llhuillier's th.: 905: A.20.
*
Napier's eqs. : 896: A.3,17: CM.3:
LM.3,13.
*
Napier's rules : 881.
supplement to, and geodesy : A.36.
Sphero-conics : 5655a: tg.c,Q.8,93: Me.3:
Z:6,23.
homofocal : C.50.
mechanical cn : LM.6.
Sphero-conjugate tangents : An.55.
Sphero-cyclides : LM.16.
Spheroidal trigonometry : J.43 : M.22.
Spheroidic transformation f. of Bessel :
A.53.
Spheroids : 5604, 6152 ; cubaturo, 6158,
A.2.
Spieker's point : A.58.
Spiral : A.28 : L.692 : N.79 : Z.14.
* of Archimedes : 5296 : A.65,66.
conical : N.45.
* equiangular : 5288.
* hyperbolic: 5302.
* involute : of circle, 5306 ; of 4th order,
C.66s.
Squares : J.22.
whose diagonals are chords of given
circles : A.64.
maximum with given sides : J.25.
maximum in a triangle : J.15.
whose sides and diagonals are
rational : J.37.
whose sides pass through 4 points :
A.43.
64 transformed into 65, geo. : Me.77.
sum of three : G.15 ; of four, L.57.

928

INDEX.

Standards of length : Pr.8,21.

Statistics : Z.26."
Steiner's ths. and prs. : A.53 : J.13,14,16,
18,23,25,36,71.73 : N.48,562,593,62 :
Hexagon, A.6 : Ray - systems,
CD.6.
Stereographic projection : A.39 : JP.
16: L.46.
Stereograms of surfaces : LM.2.
Stereometry: JP.l, : thsA.10,31,43,57 ;
J.1,5.
multiplication : J.49.
quadric and cubic eqs. and surfaces :
J.492.
Sternpolygon and sternpolyhedron : A.
13.
Stigmatics : see " Clinant."
Stewart's geo.ths. : CM.2 : L.48 : TE.
2,15.
Striction lines of conicoids : Z.28.
Strophoids : AJ.7 : N.752.
Sturm's functions : 506 : A.62 : C.36,
62,683: G.1,20: J.48: L.46,48,
67: Mo.58,78: N.43,46,52,54,66,
67,81: Q.3.
and addition th. for elliptic functions
of 1st kind: Z.17.
ext. to simultaneous eqs. : C.35.
and H. C. F. : Pr.6.
and a quartic equation : A.34.
and their reciprocal relations : Mo.
73.
remainders : J.43,48.
tables: P.57 : Pr.8.
ap. to transcendental eqs. : J.33.
ap. to transf. of binomial eqs. : L.422.
Subdeterminants of a symm. system :
J.93: M.82.
Subfactorial : Me.78.
Subinvariants = seminvariants to bin
ary qualities of unlimited order :
AJ.5.
Subnormal: 1160.
Subsidiary angles : 726, 749.
Substitutions : A.62 : C.ths66 and 67,
74,76,79 : G.9,103,ll,14,19 : L.65,
72: M.13.
ap. to functions of six or fewer vari
ables : C.21.
ap. to linear d.e : C.78.
by approximation, of the ratio of the
variables of a binary quantic to
another function of the same
degree : C.80.
canonical forms of : L.72.
and conjugate substitutions : C.21s,22.
n-i
of the form 9 (r) e e (r"-J+ar ) :
M.2.
linear: C.98: J.84 : M.19,20..

Substitutions :
linear (continued) :
of a determinant : An.84.
and integral : M.24.
powers and roots of : G.94.
reduction of : C.90 : JP.29.
for reduction of elliptic functions
of 1st kind : An.58.
successive : A.38.
which transform quadric functions
into others which contain only the
squares of the variables : J.57.
of n letters : geo. for n = 3,4,5,6, and
mystic hexagram : An.83.
no. of in a given no. of cycles : A.68.
permutable amongst themselves :
C.21s.
of equidistant numbers in an integral
function of a variable : N.51.
of systems of equations : N.81.
of six letters : C.63.
a th. of Sylvester : AJ.l.
which admit of a real inversion : J. 73.
which do not alter the value of the
function : C.212.
which lower the degree of an eq. in
two variables and their use in
Abelian integrals : C.15.
Sum and difference calculus : A.24*
Sum of squares of lines drawn from a
point to cut a curve in a given
angle : J.ll.
Summation of series (see also
"Series" and "Expansions"):
3781 : A.1013,262,30,55,62 : No.
84: C.87,88: J.10,31,33: LM.4,
72: Mem.20,30:P.1782,17847,
91,-98, 1802,6,7,11,19 :
Pr.14.
* approximate : 3820 : J.5 : 1.24.
of arcs : A.63.
Bernoulli's method : J.31.
Cauchy's th. : M.4.
C (n,r) products of a, a+6, ... a +
(n-l)b: Q.18.
by definite integrals : A.4,6,38 : J.17,
38,42,46,74: Man.46 : M6ni.ll.
of derivatives and integrals : C.44j.
* by differences : 264 : Mem.30.
by differential formulae : L.31 : M6m.
15.
by + (x) : 2757.
formuhie : A.47 : An.55 : J.30.
Maclaurin's, C.86. : f(A,D)DF{n),
Q.8: Lagrange's, J.34 : P.60 :
Vandermonde's, L.41.
of terms of a high order : C.32.
Rummer's method : C .64.
Lejeune Dirichlet's : CD.9.
periodic : J.15.
selected terms : Me.75.

INDEX.

929

Surface or surfaces (continued) :

Snmmation of series (continued) :
of sines and cosines : J.4.
?! + ^+2- = l= A.35;
a
0
c
* theorem re Xf (x, h) dx : 2708.
a = b = c = l, A.21.
of transcendents in alg. differentials :
areas of: G.22 : N.52.
J.19.
argument of points on : LM.16.
trigonometrical and infinite : Me.64.
complex: M.5 ; of 4th order and
Superposition: TE.21,23.
class, M.2.
* of small quantities : 1515.
whose coordinates are Abelian func
Supplementary :
angles : 620.
tions of two parameters : C.92 :
* chords: 1201.
M.19.
cones : thN.48.
of corresponding points : M.4.
* curves : 491720 : Man.54.
cut orthogonally by spheres : 3393 :
Supplement integrals : J.98.
C.36.
Surds: 108: N.47.
of : M.3.
quadratic (see " Boots of numbers.") deficiency
definitions: 5770.
V (a2+&3), V (o262) approximately :
determined from two surfaces of cen
J.13.
tres : A.68.
* V(aV6): 121: A.3,13: J.17,20 :
Dirichlet's problem : An.71.
MtSm.10: N.46,48.
Dupin's theorem : C.74 : CM.4 : Q.12.
* */ (a Vi) : 122 ; */ (a Vb), 124.
doubly circumscribing an n-tic sur
Surface of centres : 5774 : A.682 : An.
face : J.54.
of elliptic cone : An.51.
57: C.70,71": E.30: L.41,58:
of equal slope : C.98 : N.65.
LM.4 : M.5,16.
curvatures of the two ; relation : C. equation of: gn5780: A.3; for points
near origin, 5803.
74,79.
the two focal conies of a system of
of even order : A. 70.
families of : C.70 : Me.72.
homofocal quadrics : C.532.
of a quadric : Q.2.
flexure of : J.18.
Gaussian theory of : LM.12 : N.52.
bitangents of : Q.13.
generation of : C.94,97 : G.9 : J.492 :
model of : Mo.62.
principal axes of : N.48.
L.56,83 : M.18.
Surface curves : A.39,60 : An.54 : C.21,
implexes of : C.79,82.
80j : L.66 : G.19 : J.2 : N.54,84.
of minimum area : C.57 : J.8,13 :
an an alg. surface : An.63.
L.59: Q.14.
of nth order : cnM.23 ; 2nd, 3rd and
on cubic surfaces : M.21.
curvature of : N.65.
4th, McSl.6.
on a developable : An.57 ; the oscu
of normals : N.59.
lating plane making a constant
whose normals all touch a sphere or
conical surface : JP.4 ; do. for
angle with it, L.47.
on an ellipsoid : An.51 : CD.3 : geoL.50.
surface of revolution, JP.5.
groups of rational : M.3.
number under 9 conditions : C.62.
of intersection : M.2 : N.68 ; of 2
octic of zero kind : G.12.
quadrics, A.16.
order determined : Me.83.
multiple : G.142 ; singularities of, M.3.
one-sided : A.57.
on a one-fold hyperbola : CD.4.
parabolic points of : CD.2.
and p. d. e: C.13: CD.2: Z.7.
on quadrics : N.70.
rectification of : A.36.
and plane curves : J.54,72 : M.7.
and point moving on it : L.70,77.
relation to their tangents : C.82.
on surfaces of revolution : Z.18,28.
and point at 00 on it : J.65.
and osculating sphere : C.73.
relation of in Riemann's sense : M.7.
Surface or surfaces : 5770 : A.14,f32,41,
representation of : J.83.
59,60,62 : An.51 3,55,<50,,Gl,ti5,
on a plane : An.68,71,76.
71: At.57: 0.17,33,37,49",th858,
one upon another : An.77.
of revolution : C.86 : G.4.
61,64,67,69,80,86,992 : G.3,21 : J.
areas and volumes : 5877,9 : A.48.
9,thsl3,58,63,64,85,98 : JP.19,24,
25,33: L.44,47,513,60: M.2,4,7,9.,,
of a conic about any axis in space :
cnl9: Mo.82,833,84 : N.53,65-,68,
JP.23 : L.63.
72 : TI.14 : Z.7,8,20.
of constant mean curvature : L.412.
6 c

930

INDEX.

Surface or surfaces (continued) :

meridian of : a lemniscate, G.21 ;
generation of, 0.85.
meridian and contour curves of :
Z.21.
oblique : Q.6.
passing through a given line, tan
gent plane of : N.84.
in perspective : JP.20.
of reg. polygon about a side, vol. :
A.57.
quadric : A.55 : L.60.
shortest line on : A.38.
superposable : C.80: N.81.
Riemann's symmetrical; and perio
dicity modulus of the related
Abelian integral of the 1st kind :
Z.28.
ruled, with generators part of a linear
complex : C.84.
ruled octic with 5 quartic curves :
C.61.
screw, parallel projection of : cnZ.18.
section of, homogeneous eq. : LM.15.
self-reciprocal : LM.2 ; quadric, E.36.
sextic of first species : M.21.
singularities of: A.25: An.79 : CD.7:
CM.2: J.72: M.9: N.64: Q.9.
cubics: M.4; and quadrics, A.17.
16 singular points : C.92j.
solutions by infinites of 3rd order :
geoC.802.
Steiner's : LM.5,14 : M.5.
touching a plane along a curve:
CD.3: CM.2.
touchinga fixed surface always : 6.20.
transformation of: M.19.
and transversal, th. : N.49.
trapezoidal problem : Z.14.
two series of: prsAn.73.
web-system : J.82.
which cuts the curve of intersection
of two alg. surfaces in the point
of contact of the stationary oscu
lating planes : L.63.
Surveying : N.502 : geo. th.A.37.
Symbolic geometry : (Hamilton), CD.
1-4.
Symbolical language : E.282.
Symmedian line : E.42 : N.835 : Q.20s.
Symmedian point : 4754c.
Symmetrical :
conies of triangles :
A.59.
connections by generating functions :
J.53.
* expressions : 219.
figures : J.44.
functions : C.76 : J.69,93,98 : LM.13 :
Q.20: Z.4.
Brioschi's th. : C.98.

Symmetrical(continued) :
of the common sol. of several eqs. :
An.58.
multiplication of : Me.85.
of any number of variables : C.82.
simple and complete : M.18,20.
tables of : AJ.5 ; of 12-ic, AJ.5.
points : of 1st order : A.60 : cnA.64.
of tetrahedrons : A.60.
of a triangle : A.58j.
products : P.61 : Pr.ll.
with prime roots of unity : Me.85.
tetrahedral surfaces : Z.ll.
Symmetry ; plane and in space : N.47.
Synthesis : C.18.
Synthetic division : 28 ; evolution,
Me.68.
Syntractrix : 5282.
Tables : mathematical ; Sect. I., con
tents, p. xi.
of Bernoulli's nos., logarithms, Ac.,
calculation of : J.2.
for empirical formulas : AJ.5.
in theory of numbers : Q.l.
of e*, e", logio*, log^e"1 : CP.13.
Tac-loci: Me.83.
Tamisage: LM.14.
Tangencies (circles, points, and lines) :
937: Pr.9: Q.2,8.
Tangential eq. with the intercept and
angle of inclination for coor
dinates of the line : P.77.
Tangent cone at a singular point : 5783.
Tangents : 1160 : cnA.4,33 : J.56-,73 :
N.42: Z.23.
and contact-point in loci and en
velopes : C.85: cnN.57.
conjugate (and Dupin's th.) : An.60.
construction of : M.22 : N.80.
double: Q.3.
eq. of ; to find it : 4120,-32; Me.66.
faisceaux of: N.50.
cut by lines at a constant angle : A.43.
locus of intersections of three : LM.
15.
at a multiple point, cn. : JP.13.
and normals : 5100; relations, A.51.
parallel : N.45.
segments of : 4307 ; equality of, C.8I4.
of a surface : 5781 ; at singular points :
5783: M.11,15.
Tangent planes : 5770,-82 ; p.c5790 :
CM.4: N.45.
and surface ; intersection of : 5786
9 : L.58.
to equidistant surfaces : cnZ.28.
triple: C.77.
tan 120 : P.8,18.
tan z as a continued fraction : Z.16 ;
do. tan nse, Me.74.

931
tan nx : f. in N.53.
Tetrahedron(continued) :
tan"1 (x+iy): Z.14.
of given surface : J.83.
*tanh 8 : 2213.
with opposite edges; equal, N.79 ;
tantochrone : L.44 : An.53.
at right angles, Me.82.
Taylor's theorem: 1500,-20,-23: A.
with edges touching a sphere : N.74.
and four spheres : LM.122.
8,13: AJ.4, extl: C.58 : CM.l :
homologous : J.56.
J.ll: L.37,38,45,58,gz64 : M.21*
and quadric : CD.8 : thN.71 ; 2 quad23: Me.72,,73,75: Mem.20: N.
rics, Q.8.
52,63,70,743,79: TA.7,8: Z.gz 2
and 25.
6 dihedral angles of, en. : N.46,67.
analogues of: J.6.
theorems : A.9,10,31 : N.61,.,66-2 : Q.3,5.
convergence of : C. 60,746 : J.28 :
two: M.19.
L.73.
* volume : 5569 : A.452,57 : LM.2 : N.
Cox's proof: CD.6.
67: Z.ll.
deductions : G.12.
volumo and surface in c.c and g.c :
for an imaginary variable : Me.79.
A.63 : CD.8 : N.68.
kinematic meaning of : J.36.
volume and normal, relation : N.54.
reduced forms of : C.84.
Tetratops : A.692.
* remainder: 1503 : An.59 : C.13: J.17:
Theoretical value function : J.55.
Theta-functions : Ac.3 : C.ths90 : J.61,
N.60,63: Z.4.
a transformation of : C.78.
66,74: M.17: Me.ap78: P.80,82:
thsZ.12.
Terminology: LM.14.
analogues of: C.93.
Ternary bilinear forms : G.21.
addition theory : J.88,89 : LM.13.
Ternary cubics : AJ.2,3 : C.56,90 : CD.l :
ap. to right line and triangle : A.3.
J.39,55: JP.31,32: L.58 : M.1,4,9.
argument : J.73 : 4thM.14 : M.16.
in factors : An.76 : Q.7.
characteristic of: complex AJ.6 : C.
as four cubes : LM.10.
88.: J.28.
parameter of the canonical transf. :
th. of Riomann : J.88.
LM.12.
as a definite integral : Me.76.
reduction to canonical form : C.81.
double : LM.9 : Q. transf. 21.
transformation of : J.63.
and 16 nodal quartic surface : J.83,
Ternary forms : At.68: G.1,9,18.
85,87,88.
order of discriminant : LM.3.
formula of Riemann : J.93.
with vanishing functional deter
Jacobian, num. value : M.7,11.
minants : M.18.
modular
integrals : trAn.52,54: J.71.
Ternary quadric forms : At.65 : C.92,
multiplication of : LM.l : M.17.
94,96: G.5,8: J.20,40,70,77; transf.
quadruple : AJ.62 : J.83.
of, 71,78 and 79 : L.59,77 : P.67.
reduction of, from two variables to
and corresponding hypeifuchsian
one: C.94.
functions : Ac.5.
representation of : M.6: Z.ll.
indefinite: J.47 ; with2 conjug. indetransf. of: A.l : An. 79 : J.32 : L.80 :
terminates, C.68.
M.25j.
representation by a square : An.75.
linear: Me.84 : Q.21.
simultaneous system : J.80.
triple : J.87.
table of reduced positive : J.41.
in two variables: C.92o: J.84: M.
Ternary quartic forms : C563: An.82 :
14,24.
M. 172,20.
Theta-sories : constant factors of, J.98 :
Terquem's th. : Q.4.
Me.81.
Tesselation pr. : LM.2.
m-tuple : J.48.
Tesseral harmonic analogues : CP.13.
3a = a2 (c-.r) / {c {c-x)-V} : Q.15.
Tetradrometry, differential f. : A.34 :
Z.5.
Three - bar curves : 5430 : LM.7,9 :
Tetragon, analogue in space : CD. 7.
Mo.76.
Tetrahedroid : J.87.
triple generation of: Me.83.
Tetrahedron : 907 : A.3,16,232,51,56 :
Toothed wheels : cnTE.28.
J.65 : LM.4 : Me.62,662,82 : Mel.
Topologv with tables : M. 19,24.
2 : N.pr74,80,81 : Q.5 : geo Z.27.
Tore: section of: G.10: N.59,61,64i,652
determined from coordinates of ver
circular, 56 and 65.
tices : J.73.
and bi-taugent sphere : N.74.

932

INDEX.

Toroid : A.8 : rectif. Ac, A.9 : N.44.

Toroidal functions : P.81 : Pr.31.
Torso: sextic : CP.llj : Q.14.
circumscribing two quadrics : Me.72.
depending on elliptic functions: Q.14.
and curve: Q.ll ; and sphere, G.12.
Torsion : A.19,62,65 : J.60 : angle of,
5725 ; nf involute, 5753 ; of evolute, 5754.
* inflected, infinite and suspended :
5739 : N.45.
constant : L.422.
Tortuous curves : 5721: Ac.2 : C. 19,43,
58,cn62 : G.4,5,21 : J.16,59,93 :
JP.2,18: L.50,,th51,522: M.5,19,
th25: Mo.82: N.60: No.79: Q.
4,5,7.
* approximate coordinates of a point
near the origin in terms of the
arc : 5755.
* circular curvature of : 5722 : An.60 :
CD.9: J.60: JP.15;: Q.6.
*
locus of centre of do. : 5741, 5748
50.
with circular and spherical curvature
in a constant ratio : L.51.
cn. upon ruled cubics and quartics :
C.53.
with coordinates rational functions of
a parameter : M.3.
cubic : J.27,80,: M.20: Z.27.
* definitions: 5721.
determined from relation of curva
ture and torsion : A.65.
cubics : An.58,59 : C.45 : L.57 : of
3rd class, J.5ti ; four tangents,
M.13.
and developable surfaces : CD. 5 :
L.45: M.13.
elements of arc : N.73s.
generation of : C.94 : L.83 ; by two
pencils of corresponding right
lines, G.23.
with homog. coordinates : Q.3.
the loci of similar osculating ellip
soids ; d.e of same : C.78.
with loops : M.18.
with a max. or min. property: Mem. 13.
on a one-fold hyperboloid : C.53.
with the same polar surface, d.e :
C.783.
quadrics : J.20.
quartics : C.543 ; 1st species, J.93.
cn. of two : C.53.
intersection of quadrics : C.54.
qnintics : C.54j,58.
* radii of curvature and torsion : 5739 :
L.48: Mem.18.
* and right line method : 5743.
sextics, classification of : C.76.

Tortuous curves(continued) :
singularities of: C.67 : J.42.
spherical curvature : 5728,40,i7 :
A.19.
triple, and their parallels : A.65.
Tractrix : 5279 ; area, E.35.
Trajectory: 5246: Mo.80.
of a displaced line ; oscul. plane, Ac, :
C.70,76.
of 3 homofocal conies : An.64.
of meridian of surface of revolution:
L.46.
surface of points of an invariable
figure whose displacement is
subject to 4 conditions : C.76,77.
of a tortuous curve : L.43.
Transcendental :
arithmetic : J.29.
curves : 5250 : An.76 : M.22.
equations: C.5.59,72,94 : G.6,10: J.
22 : L.38 : N.55,56.
mechanical solution : CP.4.
separation of roots by " compteurs
logarithmiques " : C.44.
without a root : J.73.
Transcendental functions (see also
" Functions ") : 1401 : A.37 : C.
86: J.3,th9,20: L.51.
of alg. differentials : J.23 : L.44.
arithmetical properties : M.22.
classification of : L.37.
connected with elliptic : C.17.
decomposition into factors by calcu
lus of residues : C.32.
and definite integrals : P.57.
expansion of : J.16.
integral: C.94,95 : G.23: J.98.
reduction of: Me.72.
squares of : Me.72.
theorem of Sturm : L.36s.
whose derivatives are determined by
cubic eqs. ; summation of the
same : J.ll.
which result from the repeated integ
ration of rational fractions : J.30.
Transformation:
bilinear: M.2.
birazionale : G.7 ; of 6th deg. in 3
dimensions, LM.15.
contact : M.8.
of coordinates : pl.4048 ; cb.557481 :
A13: A.26: CM.lj: J.2: JP.7 :
N.63.
in 3 dimensions : A.13 : Q.2 : J.8.
rectangular into elliptic : M61.4.
Cremona's : M.4.
of curves : L.49,50.
of differential equations : Me.82.
of equations : A.40 : N.64j ; 3 vari
ables, G.5.
of a characteristic eq. by a discri
minant : Au.56.

INDEX.

Transformation(continued) :
qnadric: CM.2: N.75 : M.23.
simultaneous: Q.ll.
of figures in a plane : A.4 : N.64,774.
in space : C.94,95a,96 : N.79,80,.
double reciprocal by normals to a
sphere : C.56.
formulae : J.32.
of functions : An.50 ; in an inf. series,
A.3.
by substitution : Mdm.31.
quadric : Q.2 ; quadric differential,
J. 70.
(aybx)2 + (bzcy)*+(cx azf :
CM.1,2.
two variables : Mem. 11.
homographic plane : L.61.
in geometry : C.71 : J.67 : JP.25.
descriptive : G.13 : Z.9.
linear: 582, 1794: CD.1,6: CM.3,4:
M.2: Mo.84.
groups of: M.12,16.
methods of : C.92 ; which preserve
an invariable relation between
derivatives of the same order,
C.82: L.76.
which preserve the lines of curva
ture : C.92.
* modulus of : 1604.
* orthogonal: 584, 1799.
for equations of dynamics : C.67.
plane : M.5 ; and in space, G.16.
of powers into binomial coefficients :
No.75.
of product of n factors : An.51.
quadratic, of an elliptic differential :
M.7.
in rational space : An. 73 : LM.3.
rectangular : LM.14.
reciprocal: C.92: G.17: N.64,82,83.
of rectilineal space coordinates : J.63.
by series doubly infinite : An.56.
space, for representation of alg. sur
faces : M.3.
of surfaces : M.21 : N.69 : Q.12.
of symbolic functions into isotropic
means : C.43.
of tan _ 1 / -- + symmet. y,
V c+*
CD.9.
Tschirnhausen's : An.58 : P.62,65 :
Pr.11,14.
ext. to quintics and higher : B.l3,4 :
J.58,.
* unimodular: 1605.
of variables : G.5 : No.78.
Transitive function : of 24 quantities :
L.73.
doubly, of 5 or 6 variables : C.

933

Transitive function(continued) :
reduction to intransitive: C.21.
Transversals: 96774: A.13,18,27,30,
56: CD.5: CM.l: J.84 : G.22 :
Me.t.c68j,75 : N.432,48: TE.t.c24.
orthogonal : AJ.3.
of plane alg. curves : Z.19.
of two points : A.66.
parallel: A.13,57.
of spherical triangle and quadrangle :
A.45.
Trees, analytical : AJ.4.
Triads of once-paired elements : Q.9.
Triangle: 700: A.17,19,22,29,33,36,43,
cn46,61 : G.21 : J.50 : M.geol7 :
Me.q.c62: N.42,,43.
angles of: 677: 73845: A.65 : P.28:
division, A.51,58 : sum, geo960 :
C.69,70 : difference, Q.8.
area : 707,403641 : A.45,57 : CM.2
Mem.13.
bisectors :
of angles : 709, 742,
932, 4628,-30.

of sides : 738, 922t, 951, 4631 :

Mero.prl0,13.
central line : 957, 4644.
and circle, ths. : A.9,40,47,60 : Q.7,8,
10.
and 3 concentric circles : Me.85.
of 3 intersecting circles : Q.21.
circle and parabola : Q.15.
and conic : N.70.
of constant species : 977.
construction of : 960.
equilateral, sum of sqs. of distances
of any point from its vertices :
923 : A.69 : gzl094.
formed by joining the feet of bisec
tors of a triangle : A.64,.
Gauss's equations for a plane tri
angle : A.5.
notation: 4629: E.44.
orthocentre : 952, 4634.
pedal line of : Me.83.
perpendicular bisectors of sides : 713,
4639.
perpendiculars on sides : 952, 4633 :
Mem.13.
and point : Q.5.
and polars: circle, G.ll ; conio, Q.7.
of mid-points of sides, t.cQ.8.
quadrisection of : Mdm.9.
rational: A.51,56.
remarkable points of (see also " Incentre,"&c): 9559 : A. four, 47 ;
two, 48 ; five, 52 ; 66,67 : E.28,
30,40 : LM. nine, 14a : N.70,73,83 :
Z.11,15.
of reference in t.c : 4006.
and right line : A.59.

934

INDEX.

Underdeterminants of a symm. deter

minant (i.e., successive first
minors) : J.91.
Undeterminants, &c. : A.594.
Unicursal curves: A.60: C.78,94,96:
LM.4.
cubics, ths. re inflexion : E.28.
quartics: LM.16: N.84: Z.28.
surfaces, transf. of : M.66.
Uniform functions : C.92,947,95,96j.
with an alg. relation : C.912.
of an anal, point xy : Ac.l : C.94.
two points : C.96,97.
with " coupures " : C.96.
with discontinuities : C.94.
doubly periodic : C.94.
from linear substitutions : M.19,20.
with a line of singular points, decom
position into factors : C.92.
monogenous : Ac.4.
in the neighbourhood of a singular
point: C.89.
of two independent variables : C.94,95.
Units of elasticity, electricity, and
heat : p.2.
Vanishing fractions : 1582: M.15: Q.l.
Vanishing groups : CD.2,3,6,7,8.
ap. to quantics : CD.2,3,6.
Variation: 76: C.17.
of arbitrary constants : L.38.
* calculus of : 302891 : A.3,prs42 :
An.52.: C.16,50: CD.3: CM.2 :
J.13,prl5,41,54,65,prs74,82 : JP.
17: L.41: M.2,,15.: Me.prs72 :
Mo.57 : N.83 : Q.prlO : and d.e,
L.38 and C.49.
history : N.51.
*
immediate integrability : 3090.
and infinitesimal analysis : C.17,40.
*
relative max. and min. : 3069 : L.42.
of multiple integrals : J.15,56,f59 :
Mem.38.
transformations : J.552.
*
two dependent variables : 3051.
cos3 a sin4 0+2 sin a cos a sin 0 cos# *
two independent variables : 3175.
+ 8in3acos4$ = 0:
integral, of functions : C.40a.
Trihedron :
about a parabola, locus * of parameters : th2714, 3243: N.77.
of vertex : N.63.
* of second order : 3087 : A.4 : J.55 :
and quadric : N.71.
Q.14: Z.23.
and tetrahedron : A.57.
* of higher order : 3089 : A.27 : Z.26.
Versiera or Witch of Agnesi : 5335.
Trilinear forms : C.92,93.
of solids : 587183 : A.
Trilinear relation of plane systems : Volumes :
J.98.
31,32,36: J.34: N.f57,80.
Triplicate-ratio circle : 4754b : E.40,42,
approximate : C.95.
436,44.
of frustums : A.33 ; of conicoids,
Twisted surface : see *' Scrolls."
J.44.
Twist of a bar : Me.80.
of right cylinders and cones in abso
Umbilics : 5777, 581923 : J.65 : JP.
lute geometry : A.59.
of surface loci of connected points
18 : M.9 : Q.3.
* of qua dries : 5603, 5834 : C.96.
LM.14.

Triangle(continued) :
scalenity of : Me.66,68.
aides of : bisectors : A. 17.
division of : A.63 : N.83.
containing conjugate poles with
respect to four conies : An.69.
cubic eq. for in terms of A, B, and
r : J.20.
similar : C.79a ; under 3 or 4 condi
tions, C.7&,.
* solution of : 718, 859 : A.3,51 : J.44. :
TE.10.
six-point circle : Q. 1.5.
symmetrical properties : A.57 : Pr.ll.
theorem
A.9,43,45,S5,57,60,613,63,
W J.6o.71,pr28 : Q.9,t.c7 and 8,
geol,'2, and 4.
relating to triangles of same peri
meter under four other condi
tions : C.84.
Triangular :
numbers : 287 : E.30 :
J.69: L.ths63.
prism : thN.42.
pyramid : At.19 : No.pr32.
Tricircular and tetraspheric geometry :
An.77.
Trigon. in t.c : 4006.
Trigonometry: 600: A.l2,2,8,ll,13,30 :
G.13: extM.35.
formula : A.27,65 : Me.81,gzl : N.77:
geoQ.5,10.
* formulae : 627,700,823 : A.33 : N.46,80.
functions : P.1796 ; in factorials,
A.43.
in binomial factors : L.43.
in partial fractions and products :
Z.13.
functions closely allied to : A.27.
tables, cn : A.l,23:,25.
theorems : A.2,50,51 : Q.pr7.
on product of 4 sines : Me.81.
on (l+?)(l+f)-<WM.

INDEX.

Volumes (continued) :
and surfaces by curvilinear coor
dinates : C.16.
*Wallis's formula : 2456 : A.39,gz39 :
JP.28.
Waring's identity: N.49,62 : extMe.85.
Wave surface : C.13,47,78,geo82,92 :
CD.7,8: CP.6: L.46: Me.66j.73,
76,78,79 : N.63,82 : Pr.32 : Q.2A,
4,5,9,1 5j,l 7.
asymptotes : C.97.
and cone : Q.23,26.
cubatureof: An.61.
generation and cn : C903.
lines of curvature : An.59 : C.97.
normals and centres of curvature :
geoC.64.
umbilics, geo : C.88j.

935

Wear of gold coins : E.43.

Web surfaces : see " Net surfaces."
Weierstrass's function 2nb" cos a*xir
o
with a > 1 and 6 < 1 : G.18 : J.
63,90.
expansion in powers of the modulus :
C.822,85,86 : L.79,.
Wilson's theorem : 371 : A.48 : CD.9 :
J.8,19,20 : Me.83 : N.43.
generalisation : J.31 : Me.64 : M<51.2 :
N.45.
Wronski's methods : C.92 : L.82,83.
formula of 1812: N.74..: Q.thl2.
Zetafuchsian functions : Ac.5.
Zonal conies of tetrazonal quartics :
Q.10.

Bate

Hate 2.

16-3C

27.

30.

31-45

Place 4.

46--51

48.

I i: ^ > "

Plate 5

52 62

PJate 6

~5

63-75

Plate 7.

-v

76-86

tea

-\

87-6

Ha

\0

-X'

'

I I a

TV

JV

/*

116 -12c

Pta*i8

130

\i

124-135

N
"

Plate 14

0'
/

133-14"

"

'late 15.

\
I \

'

142-152

1 53 -1 b'

V v.
r ,

165-167

It

168-177

178-182

82

x-jdte 20

183-1 93

uJt

RETURN TO the circulation desk of any

University of California Library
or to the
NORTHERN REGIONAL LIBRARY FACILITY
Sldg, 400, Richmond Field Station
Jniversity of California
Richmond, CA 94804-4698
\LL BOOKS MAY BE RECALLED AFTER 7 DAYS
- 2-month loans may be renewed by calling
(510)642-6753
- 1-year loans may be recharged by bringing
books to NRLF
Renewals and recharges may be made 4
days prior to due date.
DUE AS STAMPED BELOW

MAY 2 fi 2001

it"

OCT 3 0 2007

'000(11/95)

HQ3

YD 07347

II I1'