Course of Modern An 00 Whit

^
UNIVERSITY OF CALIFORNIA
AT LOS ANGELES
A COURSE OF
MODERN ANALYSIS
CAMBRIDGE UNIVERSITY PRESS

CLAY, Manager
C. F.
LONDON
FETTER LANE,
E.G. 4
NEW YORK THE MACMILLAN

:
CO.
BOMBAY
CALCUTTA I MACMILLAN AND CO., Ltd.
MADRAS
TORONTO THE MACMILLAN CO. OF
\
j
:
CANADA,
Ltd.
TOKYO MARUZEN-KABUSHIKI-KAISHA
:
ALT.
RIGHTS RESERVED
A COURSE OF
MODERN ANALYSIS
AN INTRODUCTION TO THE GENERAL THEORY OF
INFINITE PROCESSES AND OF ANALYTIC FUNCTIONS;
WITH AN ACCOUNT OF THE PRINCIPAL
TRANSCENDENTAL FUNCTIONS
BY
E. T.
WHITTAKER,
Sc.D., F.R.S.
PROFESSOR OF MATHEMATICS IN THE UNIVERSITY OF EDINBURGH
AND
G. N.
WATSON,
Sc.D., F.R.S.
PROFESSOR OF MATHEMATICS IN THE UNIVERSITY OF BIRMINGHAM
THIRD EDITION
CAMBRIDGE
AT THE UNIVERSITY PRESS
1920
First Edition 1902
Second Edition 1915
Third Edition 1920
PEBFACE
Advantage
has been taken of the preparation of the third edition of
add a chapter on Ellipsoidal Harmonics and Lame's Equation,
this
work
and
to rearrange the
to
chapter on Trigonometrical Series so that the parts
which are used in Applied Mathematics come at the beginning of the

A number of minor errors 'have been corrected and we have
chapter.
endeavoured to make the references more complete.
Our thanks
proofs
and
are due to Miss
Wrinch
for
reading the greater part of the
to the staff of the University Press for
much
courtesy and con-
sideration during the progress of the printing.

E. T.
G. N.
July, 1920.
W.
W.
CONTENTS
PART
Comijlex
...........
.........
.....
Numbers
The Theory
II
THE PROCESSES OF ANALYSIS
I.
III
Continuous Functions and Uniform Convergence
IV
The Theory
and
11
.......
.........
......
41
of Rieinann Integration
The fundamental
VI
of Convergence
properties of Analytic Functions
Liouville's
Theorems
The Theoryof Residues; application
61
Taylor's, Laurent's,
82
to the evaluation of Definite Integrals
VII
The expansion
VIII
Asymptotic Expansions and Summable Series
111
of functions in Infinite Series
125
150
IX
Fourier Series and Trigonometrical Series
Linear DiflFerential Equations
194
Integral Equations
211
XI
PART
XII
The
...........
THE TRANSCENDENTAL FUNCTIONS
II.
Gamma
160
Function
XIII
The Zeta Function
XIV
of
235
Riemann
265
The Hypergeometric Function
281
XV
Legendre Functions
302
XVI
The Confluent Hypergeometric Function
337
XVII
XVIII
Bessel Functions
355
The Equations
386
of Mathematical Physics
XIX
Mathieu Functions
XX
Elliptic Functions.
XXI
404
General theorems and the Weierstrassian Functions
The Theta Functions
XXII
XXIII
The Jacobian
Ellipsoidal
Elliptic Functions
429
462
491
Hamionics and Lame's Equation
APPENDIX
LIST OF AUTHORS QUOTED
GENERAL INDEX
[Note.
536
579
591
595
The decimal system of paragraphing, introduced by Peano, is adopted in this

The integral part of the decimal represents the number of the chapter and the
work.
fractional parts are arranged in each chapter in order of magnitude.
pp. 187, 188, 9-632 precedes
97
because 9-632
< 9-7.J
Thus,
e.g.,
on
PART
CHAPTER
COMPLEX NUMBERS
Rational numbers.
11.
The
idea of a set of
numbers
derived in the
is
first
instance from the
consideration of the set of positive* integral numbers, or positive integers;
that
to say, the
is
nnmbL-rs
1.
8,
2,
4, ....
many
Positive integers have
Theory of Integral
Numbers; but at a very early stage in the development of mathematics
it was found that the operations of Subtraction and Division could only be
performed among them subject to inconvenient restrictions and consequently,
properties,
which
will
be found
in
treatises
on
the
in elementary Arithmetic, classes of
numbers are constructed such that the
operations of subtraction and division can always bB performed
among them.
numbers among which the operation of subtraction

can be performed without restraint wc construct the class of integers, which
consists of the class of positive f integers (+1, +2, +3, ...) and of the class
of negative integers (-1, -2, -3, ...) and the number 0.
To
obtaiti a class of
class of numbers among which the operations both of subcan be performed freely:):, we construct the class of
division
and
of
traction
Symbols which denote members of this class are |, 3,
rational numbers.
To obtain a
0,
-Y--
We
have thus introduced three classes of numbers,
the integers,
(ii)
It
is
(iii)
(i)
the signless integers,
the rational numbers.
not part of the scheme of this work to discuss the construction of
the class of integers or the logical foundations of the theory of I'ational
numbers^.
The extension
of tbe idea of
number, which has just been described, was not
without some opposition from the more conservative mathematicians.
effected
In the latter half
and Frend (1757-1841) published works

which the use of negative numbers was disallowed,
although Descartes had used them imrestrictedly more than a hundred years before.
of the eighteenth century, Maseras (1731-1824)
on Algebra, Trigonometry,
* Strictly speakinpr, a
etc.,
in
more appropriate
epithet would be, not positive, but signless.
t In the strict sense.

With the exception of division by tbe rational
:*:
number
0.
Such a discussion, defining a rational number as an ordered number-pair of iuteêrs in a

similar manner to that in which a complex number is defined in 1'3 as an ordere number-pair
of real numbers, will be found in Hobsou's Functions of a Real Variable, 1-12.
12
rational
number x may be represented
[CHAP.
to the eye
in
the following
manner
and a fixed segment OPi
we take an origin
a length OFx such that
we can measure from
the ratio OPx/OPi is equal to x the point P^ is taken on the right or left of
We may regard either
according as the number x is positive or negative.
on a straight
If,
line,
(Pi being on the right of 0),
the point
P^ or the displacement OP^ (which

number x.
Avill
be written OPx) as repre-
senting the
All the rational
but the converse
is
numbers can thus be represented by points on the line,

For if we measure off on the line a length OQ
not true.
equal to the diagonal of a square of which OPi

that
is
one
side, it
can be proved
does not correspond to any rational number.
the line which do not represent rational numbers may be said to represent
numbers
thus the jjoint Q is said to rej^resent the irrational number
But while such an explanation of the existence of irrational numbers
,^/2 = l 414213....
satisfied the mathematicians of the eighteenth century and may still be sufficient for
those whose interest lies in the applications of mathematics rather than in the logical
Points
oil
irrational
upbuilding of the theory, yet from the logical standpoint
it
is
improper to introduce
geometrical intuitions to supply deficiencies in arithmetical arguments
shewn by Dedekind
in 1858 that the
aud
was
it
theory of irrational numbers can be established on
a purely arithmetical basis without any appeal to geometry.

1"2.
Dedekind' s* theory of irrational numbers.
The geometrical property
of points on a line which suggested the starting
point of the arithmetical theory of irrationals was that,
if all
are separated into two classes such that every point of the
the right of every point of the second

point at which the line
is
class,
points of a line
first class is
on
there exists one and only one
thus severed.
Following up this idea, Dedekind considered rules by which a separationf

or section of all rational
(which will
be. called
numbers
into
two
classes can be
made, these classes
the Z-class and the P-class, or the
left class
and the
right class) being such that they possess the following properties
member of each class

Every member of the X-class is
At
(i)
(ii)
least
one
exists.
less
than every member of the
P-class.
It
is
*
is
obvious that such a section
either the greatest
The
number
is
made
b}' an}^
rational
of the Z-class or the least
number x and
number of the
;
was not published before the appearance of Dede1872. Other theories are due to
von Dantscher, Die Weierstrasa'sche Theorie der irrationulen Zahlen (Leipzig,
theory, though elaborated in 1858,
kind's tract, Stetigkeit
Weitrstrass [see
und
irrationale Zahlen, Brunswick,
1908)] and Cantor, Math. Ann.
v.
(1872), pp. 123-130.
t This procedure formed the basis of
mathematicians in the sixth aud
fifth
the treatment of irrational numbers by the Greek
centuries b.c.
The advance made by Dedekind consisted
observing that a purely arithmetical theory could be built ap on
it.
in
1'2]
COMPLEX NUMBERS
But sections can be made in which no rational number x plays this

Thus, since there is no rational number* Avhose square is 2, it is easy
i?-class.
part.
we may form a section in which the i?-c]ass consists of the positive

numbers whose squares exceed 2, and the Z-class consists of all
to see that
rational
other rational numbers.
Then
this section
such that the i?-class has no least member and the

member for, if x be any positive rational fraction,
is
Z-class has no greatest
and 2 are
in order of
magnitude; and therefore given any member x of the

find a greater member of the Z-class, or given any
we can always
Zr-class,
member x
of the 7?-class, we can always find a smaller member of the

such numbers being, for instance, y and \j where y' is the same
/i- class,
function of x' as y of
If a section
/>
is
x.
made
in
has a greatest
class
number, which
it is
If a section
is
which the
i^-chiss has a least member Jo, or if the

J,, the section determines a rational-real
member
convenient to denote by the samef symbol
made, such that the
Z-class has no greatest
member,
i^-class has
the section
Ao_ or
A^.
no least member and the
determines an irrational-real
number I.
If X, y are real
than
if
3/
defining
numbers (defined by
sections)
we say that x is
members of the
the Z-class defining x contains at least two
greater
i^-class
y.
Lot a, ^, ... be real numbers and let .4j, i?,. ... be any members of the
corresponding Z-classes while A., B... are any members of the corresponding
The classes of which A^, Ao,

denoted by the symbols {A^), {A.^), ....
i^-classes.
Then the sum (written

number (rational
the real
{A^
+ B,) and
It
is,
numbers.
For
+ ^)
are respectively
of two real
numbers a and
or irrational) which
the E-class {A.
members
is
will
be
defined as
/3 is
determined by the Z-class
+ B.,).
of course, necessary to prove that these classes determine a section of the

rational
It is evident that A + 5,
A, + B., and that at least one member of each of the
i
<
classes {Ai
*
...
if
+ B^),
{A.,-\-B.,) exists.
It
remains to prove that there
piq be such a number, this fraction being in
(2'if-i')/(p-9) is another such
number, and
0<p-q<q,
its
is,
at most, one rational
lowest terms,
so that pjq
is
it
not in
may
its
be seen that
lowest terms.
The contradiction implies that such
a rational number does not exist.

+ This causes no confusion in practice.
X B. A, W. Russell defines the class of real numbers as actualhj being the class of all L-classes
the class of real numbers whose i classes have a greatest member corresponds
to the class of
rational numbers, and though the rational-real number x which corresponds to a rational
number
.r is conceptually distinct from it, no confusion
arises from denoting both by the same symbol.
If
member
the classes had only one

of the JL-class of
.r
and the
member in common, that member might

member of the i^-class of y.
least
be the greatest
6
number which
is
greater than every Ai
that tliere are two,
and y
.v
+ B^ and less than

Let
(?/>.r).
cfj
be a
every
member
A.,
of
[CHAP.
+ B.,;
(.Ij)
suppose,
and
let
of (Ao); and let lY be the integer next greater than

(a2-i)/{i(y-.^)}-
the numbers
ai-|-
them which belongs
to (Ao)
these two
let
^2 -'î
Choose
c/i,
do in a similar
C2
numbers be
maimer from the
+ cL - f
+ d.,î/, Ci+di^x, and
therefore
member
Take the
last of
first
of
- ^)-
(3/
<y -
C2
possible,
Then
Cj, c,.
classes defining
c?i
if
be a
which belongs to (A^) and the
1, ... lY),
= Jr {ao - i) 
/3;
then
A'.
+ c?2-Ci-c/i > i/-.?-; we
arrived at a contradiction by supjjosing that two rational

neither to ( Jj + B^) nor to ( ^2 + ^2)-
numbers
a;,
have therefore
exist belonging
If every rational number belongs either to the class (J +

^1) or to the class (^2+^2),
1
then the classes (Jj + î), (^2 + ^2) define an irrational number. If one rational number.)exists belonging to neither class, then the Z-class formed by x and (Jj+^j) and the
i2-class (^2
is
+ -52)
called the
The
Z'-class
sum
define the rational-real

a
number
In either
x.
ca.se,
the
number
defined
/3.
difference a-/3 of
two
numbers
real
is
defined by the Z-class
(J1-Z2) and the
(^2-^1).
The product
of two positive real
and the Z-class of
all
numbers
a,
defined
is
/3
by the 7^class
{A^B-i)
other rational numbers.
The reader will see without difficulty how to define the product of negative real numbers and the quotient of two real numbers; and further, it may be shewn that real
numbers may be combined
in accordance
with the associative, distributive and commuta-
tive laws.
The aggregate
of rational-real and irrational-real
aggregate of real numbers
numbers are
real
1"3.
numbers is called the

numbers and irrationalnumbers respectively.
for brevity, rational-real
called rational
and
irrational
Complex numbers.
We have seen that a real

along a definite straight line.
plane, the displacement
PQ
number may be
If,
however,
visualised as a displacement
and
needs two real numbers
instance, the differences of the coordinates of
are any two points in a

for its specification
for
and Q referred to fixed

rectangular axes. Ifthe coordinates of P be (^, 77) and those oi
Q(^ + cc,7}+i/),
the displacement PQ may be described by the symbol [x,
We are thus
y].
led to consider the association of real numbers in ordered* pairs. The natural
definition of the sum of two displacements [x,
y\ [x, y'] is the displacement
which is the result of the successive applications of the two displacements
it
is
therefore convenient to define the
sum
of two number-pairs
by the
equation
[.c,
The order
number-pair
[?/,
of the
,r].
y]
+ [x',
y']
= [x + x\ y
-f y'].
two terms distiuguishes the ordered number-pair
[.r,
ij]
from the ordered
COMPLEX NUMBERS
rS]
The product
of a number-pair and a real,
number x
is
then naturally
defined by the equation
x X
We
\x,
y\
x'y\
\x'x,
are at liberty to define the product of two number-pairs in any
convenient manner; but the only definition, which does not give rise to
results that are merely trivial,
[x,
It
is
that symbolised by the equation
is
and
0] X [x,
[0, y]
The geometrical
y ] = [xx,
is
it
= xx
x'y]
[x, y']
= y x [- y',
interpretation of these results
number x; but the
to multiply
It
- i/t/', xy + xy\
xy'^
= [- yy',
x [x, y]
multiplying by the displacement

the real
[xx'
then evident that

[x,
is
y] X [x, ij]
by a
real
[x, 0] is
effect
multiplying a displacement by
effect of
convenient to denote the number-pair
it
of
[0, y]
through a right angle.
[x,
y]
now conveniently
is
the
that
the same as that of multiplying by
number y and turn
symbol x + iy; and a number-pair
is
x'].
by the compound
called (after Gauss)
a complex number; in the fundamental operations of Arithmetic, the complex
number x+ iO may be
+ il, we have i- = [0,
The reader
number x
replaced by the real

1] x [0, 1]
= [
1,
0]
and, defining
to
mean
and so r may be replaced by
1.
will easily convince himself that the definitions of addition
and multiplication of numbei--pairs have been so framed that we may perform

the ordinary operations of algebra with complex numbers in exactly the same
way as with real numburs, treating the symbol i as a number and replacing
the product
Thus he
it
by
wherever
it
will verify that, if a,
occurs.
b, c
ab
b(i,
a +
(a
if
It
-\-
then either a or 6
we have
a,
+ b) +c = a +(b
ab .c = a.bc,
a {b
and
are complex numbers,
c)= ab +
-\-c),
ac,
ab
is zero,
is
found that algebraical operations, direct or inverse, when applied to
is zero.
complex numbers, do not suggest numbers of any fresh type the complex
number will therefore for our purposes be taken as the most general type
;
of number.
The introduction of the complex number has led to many important developments in
Functions which, when real variables only are considered, appear as
essentially distinct, are seen to be connected when complex variables are introduced
mathematics.
[CHAP.
thus the circular functions are found to be expressible in terms of exponential functions
of a complex argument, by the equations
cosa;=-
(e'^
+ e^^*),
sin
a'=.(e'^ e~").
most important theorems of modern analysis are not true if the

thus, the theorem that every algebraic
restricted to be real
equation of degree n has % roots is true in general only when regarded as a theorem
concerning complex numbers,
Again,
many
of the
numbers concerned are
Hamilton's quaternions furnish an example of a
of number.
extension of the idea
still fiu'ther
quaternion
iv+xi+1/j + zk
is
formed from four
numbers w, .i; y, z, and four number- units 1, i, j, l\ in the same

number x-\-iy might be regarded as being formed from
and two number-units 1, i. Quaternions however do not obey
real
way
that the ordinary complex
two
real
numbers
x, y,
the commutative law of multiplication.
The modulus of a complex number.
1'4.
Let X
iy
be a complex number, x and y being real numbers. Then

4- y- is called the modulus of {x + iy), and is
the positive square root of xwritten
\x
+ iy\.
Let us consider the complex number which

complex numbers, x + iy and u + iv. We have
{x
The modulus
of two given
+ iy) + ( + iv) = {x + u) \- i {y {- v).
of the
sum
of the
\{X
or
sum
the
is
[{x-
two numbers
+ Uf +
((/
+ y-) + {ur + V-)
therefore
V)^} 2,
+ ijv)]^.
2 {xu
-+
is
But
{\x-\-iy\-\-\u
+ iv\]-=
[{x-
/y"')^
(w-
= (^2 4- y^) + (^<2 +

= (x- + y-) + (u^ +
and
this latter expression is greater
{x-
We
y'^)
v'^yf
^;2^
tj2)
2 {x'
2 {(xu
t/'f {tC
+ V')^'
+ yv)- -h (xv - yu)-]^,
than (or at least equal to)
+ (u" + V-) +
2 (xu
+ yv).
have therefore
X +'iy
I
\u
iv\'^\ (x
iy)
+ {u
+ iv)
the moduhis of tJie sum of two complex numbers cannot be greater than the
sum of their moduli and it follows by induction that the modulus of the sum
of any number of complex numbers cannot be greater than the sum of their
i.e.
moduli.
COMPLEX NUMBERS
14, 1-5]
Let us consider next the complex number which

given complex numbers, x + iy and u + iv we have
the product of two
is
{pG 4-
iy) {u
= {xu yv) + i {xv + yu),
iv)
and therefore
{x
j
+ iy)
(u
+ iv) =
{{xu
- yvf + (xv +
{(^
\x-\- iy
?/-)(u'
yuf'^
+ v2)j^
+ iv
'
The modulus of the product of two complex numbers (and hence, by induction, of any number of complex numbers) is therefore equal to the product
of their moduli.
The Argand diagram.
15.
We
may be represented in a geometrical

diagram by taking rectangular axes Ox, Oy in a plane. Then a point P
whose coordinates referred to these axes are x, y may be regarded as
representing the complex number x + iy. In this way, to every point of
have seen that complex numbers
the plane there corresponds some one complex
every possible complex number
number
and, conversely, to
there corresponds one, and only one, point
of the plane.
The
iy may be denoted by a single letter* z.
then called the representative point of the number z we shall
also speak of the number z as being the ajfx of the point P.
The complex number x +
point
is
If we denote (jf + /)^ by r and choose 6 so that r cos 6 = x, r sin 9 =y,

then r and 6 are clearly the radius vector and vectorial angle of the point P,
referred to the origin
and axis Ox.
The representation
Argand diagram^.
of complex
By
the definition already given,
The angle
We
numbers thus afforded
is
it
is
evident that r
is
is
the modulus of
called the argument, or amplitude, or phase, of
write
arg
often called the
z.
z.
z.
From geometrical considerations, it appears that (although the modulus of a complex

number is unique) the argument is not unique ; if ^ be a value of the argument, the
other values of the argument are comprised in the expression 2mr + d where a is any
integer, not zero.
The principal value of arg 2 is that which satisfies the inequality
- 7^<arg^^ TT.
;
* It is convenient to call .r and

y the real and imaginary parts of z respectively. We frequently write x = R(~), y = I(z).
t .J. E. Argand published it in 1806; it had however previously been used by Gauss, and
by Caspar Wessel, who discussed it

published by that Society in 1798-9.
X See the Appendix, A-521.
in a
memoir presented
to the
Danish Academy in 1797 and
10
[CHAP.
and P2 are the representative points corresponding to values z-^

z, then the point which represents the value z-^ + z.^ is
the terminus of a line drawn from P^, equal and parallel to that
If P,
and
respectively of
Z.2
clearly
which joins the origin
To
Zo
to P.2.
which represents the complex number
find the point
we
are two given complex numbers,

z^
z.2,
then,
=
=
notice that
^1
(cos 6^
sin
?-2
(cos
sin ô)
f^2
z^Zo,
where
z^
and
if
6^),
by multiplication,
Z\Z%
= n^2 [cos ((9i +
^2)
+ ^ sin (^j +
The point which represents the number

measured by the product of the
angle equal to the
sum
z^z^
6.^].
has therefore a radius vector
Pj and P2, and a vectorial
radii vectores of
of the vectorial angles of Pj
and Pg.
REFERENCES.
Tke^gical foundations of the theory of number.
u^. X.
'B.
A.
Whitehead axd B. A. W. Russell, Principia Mathematica (1910-1913).

W. Russell, Introduction to Mathematical Philosophy (1919).
Onirrational numbers.
R.
''V.
Dedekixd, Stetigkeit unci irrationale Zahlen. (Brunswick, 1872.)

VON Dantscher, Vorlesungen ueber die Weierstrass'sehe Theorie der irrationalen
Zahlen.
(Leipzig, 1908.)
E^.W. HoBSON, Functions of a Real Variable (1907), Ch. i.

T.'*?. I'a. Bromwich, Theory of Infinite Series (1908), Appendix
i.
Onômplex numbers.
H. Hankel, Theorie der complexen Zahlen-systeme.
O. Stolz, Yorlesungen iiber allgemeine Arithmetik,
G. H.
Hardy,
course of
Pure Mathematics
(Leipzig, 1867.)
(Leipzig, 1886.)
II.
(1914), Ch. in.
Miscellaneous Examples.
Shew
1.
numbers
that the representative points of the complex
+ 4/,
4-7?',
34-10/,
are coHinear.
2.
Shew that a paralwla can be drawn to pass through the representative points of
the complex numbers
2 + V,
10 + 2.5i.
44-4t,
6 + 9/, 8+^16/,
3.
Determine the
number
number
unity by aid of the Argand diagram
of integers (including unity) less than
Prove that
p m
nth. roots of
and shew that the
of primitive roots (roots the powers of each of which give all the roots)
if ^j, ^2? ^3,
71;
and
that,
the constituent primes.
when
is
the
it.
he the arguments of the primitive roots, 2 cosjo^=0 when
& positive integer less than -7
primes of
n and prime to
where
p (toe
,
-?
m iC
a, b,
c, ...
k are the difterent constituent
'f n
p6= ,-^
(
2 cos
ccoc
, ,
where
/x
is
the
number
fc
(Math. Trip. 1895.)
of
CHAPTER
II
THE THEORY OF CONVERGENCE

The definition* of
2"1.
Let
Then,
number
of a sequence.
be an unending sequence of numbers, real or complex.

I
exists such that, corresponding to every positive f
no matter how small, a number ??o can be found, such that
Zi, Zo,
if
the limit
e,
...
z-i,
number
\Zn-l'.<
for all values of n greater
as n tends
to
than
sequence
o. the
(z,,) is
said
tend
to
to the limit
infinity.
Symbolic forms of the

tends to infinity,
is
/
'
statement;]:
the limit of the sequence
'
{Zn),
as n
are
\unzn=l,
lim Zn=l,
2,1*
^"
If the sequence be such that, given an arbitrary
->
number
(no matter
how large), we can find 7?o such that ^,i > iV for all
/lo, we say that '\Zn\ tends to infinity as n tends to infinity,' and we
values of n greater than
kn -> ^
write
In
i/;,j
the
oc
when x,i>N when n>
corresponding case
n^
we say
that
If a sequence of real
the sequence
211.
is
numbers does not tend
to a limit or to co or to
said to oscillate.
Definition of the phrase
'
of the order
of
and {zn) are two sequences such that a number n^ exists such that
jK" whenever n > n^, where
is independent of n, we say that ^n is
(Ku/^n)
of the order of Zn, and we write
If (^)
<
'
15!il^ =
thus
If lim(^/^)
*
= 0, we
n"^
write ^n
definition equivalent to this
was
first
0fl
\n-
o{Zn).
given by John Wallis in 1655.
[Opera,
i.
(1695),
p. 382.]
t The number zero is excluded from the class of positive numbers.

J The arrow notation is due to Leathern, Camb. Math. Tracts, No.
I.
This notation
(1909), p. 61.
is
due to Bachmann, Zahlenthcorie (1894),
p. 401,
1.
and Landau, Primzahlen,
12
[CHAP.
II
The limit of an increasing sequence.
22.
be a sequence of real numbers such that Xn+i'^Xn for all values

of w; then the sequence tends to a limit or else tends to infinity (and so it does
not oscillate).
Let
(a7)
Let X be any rational-real number

Xn
(i)
^X
the value of
Or
If
(ii)
(ii)
then either
number
n greater than some
for all values of
/?o
depending on
x.
Xn
<X
is
not the case for any value of x (no matter
for every value of n.
how
large),
then
XnÔO.
But if values of x exist for which (ii) holds, we can divide the rational
numbers into two classes, the Z-class consisting of those rational numbers x
for which (i) holds and the i^-class of those rational numbers x for which (ii)
This section defines a real number
holds.
rational or irrational.
a,
And if be an arbitrary positive number, a-ê belongs to the i-class

which defines a, and so we can find n^ such that Xn>oi \e whenever n > n^
and a + ê is a member of the i?-class and so Xn<a + ê.
Therefore,
;
whenever n >
n^
\CC-Xn\<.
Therefore Xn >
Corollary.
Example
1.
For, given
e,
(i)
Let
?ii
a.
decreasing sequence tends to a limit or to
If lim4:,
we can
^,
find
= r,
lims,'
n and
'
n'
oo
then lim(^, + 3,') = ^ + ^'.
such that
whenHi>?i, \z^ l\<\i,
be the greater of n and
then,
(ii)
whenm>;i', \z,,l-V \<\(.
m>
when
(2m + ^m') -{l + l')\^\
{Z,n
"
|
?ii,
{^m -l')\,
<f
and
this is the condition that Urn
Example
lim{zjz,')
+ Sm')='i +
if A-
Prove simikrly that lim(s,- 2,,/) = ;-/",
3.
If
<x<
a >
1,
= (l+a)-i,
x"
by the binomial theorem
Let
= ll',
and,
if
/'
+ 0,
0.
rfM
(!
number
2'21.
l[m{z,z,')
and
<
positive
i'-
= l/l'.
Example
For
2.
{z,n
f,
for
+ )! + '
a positive integral index.

no such that (l + na)-'-
we can choose
And
<
it is
obvious that, given a
when n
>
7l^^,
and so
,)"
^0.
Limit-points and the Bolzano- Weierstrass* theorem.

(xn) be a sequence of real
numbers.
* This theorem, frequently ascribed to Weierstrass,
bohmischen Ges. der Wiss.
v.
(1817).
seems to have been known
to
Cauchy.
If
any number
exists such
was proved by Bolzano, Ahh. der
[Eeprinted in Klassiker der E.mkten Wiss., No. 153.]
k.
It
1-
2*2-2-22]
13
no matter how small, an unlimited number

of terms of the sequence can be found such that
that, for every positive value of
e,
G 6 < Xn < G +
G
then
is
e,
called a limit-point, or cluster-point, of the sequence.
Bolzano's theorem
that, if
is
$^'<cp, ivhere X, p are independent oj n,
then the sequence {x^^ Juts at least one limit-point.
To prove the theorem, choose a

of
the rational
all
number
there are only a limited

Z-class
and
Xn>A; and
G+
2'211.
(ii)
the
such that x^ ^ a
be an arbitrary positive
if e
L and
ê are members of the
classes respectively,
an unlimited number of terms of the sequence satisfying
G- e<
G
a-',i
of the Z-class.
ê and
so there are
and so
be any one of them,
of terms Xn satisfying
This section defines a real number G; and,
number,
the i^-class consists
(i)
that, if
such that there are an unlimited number of terms
is
members a
for all
section in which
numbers which are such
G - he^Xn^G + ^<G-{-
satisfies the condition that it
e,
should be a limit-point.
Definition of 'the greatest of
tlie
limits.'
The number G obtained in 2'21 is called the greatest of the limits of

the sequence {xn)' The sequence {x^) cannot have a limit-point greater
than G] for if G' were such a limit-point, and e = i ((r' (r), G' e is a
member of the jR-class defining G, so that there are only a limited number of
'
terms of the sequence which satisfy
Xn>G'
sistent with G' being a limit-point.
We
This condition
is
incon-
write
G= \imxn.
The
'
least of the limits,' L, of the
2-22.
sequence (written lim x^)

lim (-
is
defined to be
Xn).
CaUCHY's* theorem on the necessary and sufficient CON-
DITION FOR THE existence OF A LIMIT.
We
shall
now shew
that the necessary and sufficient condition for the
existence of a limiting value of a sequence of numbers
any given positive number

possible to find a number n such that
corresponding
for
to
all positive integral values
and fundamental theorems of
of j)-
e,
This result
analysis.
It is
hoiuever
is
z-i,
small,
... is
it
that,
shall
be
one of the most important
sometimes called the Principle
of Convergence.
*
z^, Zn,
Analyse Algebrique (1821),
p. 125.
14
[CHAP.
II
we have to shew that this condition is necessary, i.e. that it is

whenever a limit exists. Suppose then that a limit I exists then
corresponding to any positive number e, however small, an integer n
First,
satisfied
( 2'1)
can be chosen such that
for all positive values of
therefore
Zn+p
i
-l\ + \Zn-l\<^,
which shews the necessity of the condition

I
and thus establishes the

Secondly,
it
we have
-2'rt
<
^>
half of the theorem.
first
prove* that this condition
to
then a limit
satisfied,
is
îi+p
is
suficient,
i.e.
that
if
exists.
Suppose that the sequence of real numbers {xn) satisfies Cauchy's

that is to say that, corresponding to any positive number e, an
integer n can be chosen such that
(I)
condition
for all positive integral values of p.
Let the value of
Let
Xj, pi
corresponding to the value
n,
be the least and greatest of
Xi-1 <Xn<
values of n
for all
Then,
write Xj
for all values of n,
= X,
pi
X < Xn <
e,
be
7n.
av; then
...
a;i, a-g,
p,
1 of
l,
+ 1 = p.
p-
IVierefore by the theorem of 2'21,
the sequence (xn) has at least one liniit-point G.
Further, there cannot be more than one limit-point; for
two,
H (H < G),
and
n exists such that
and
Xn+p
take
Xn
if
there were
< I {G H). Then, by hypothesis, a number

< e for every positive value of p. But since G
e
|
H are limit-points, positive numbers q and r exist such that

G X,^+q
l-nen
But, by
X-^^q
Cr
1'4,
Therefore
the
-j|
Xn^q
sum on
GH<
4e,
X^
is
^^
-||
the left
which
\H Xn+r <
\<,
X^
^n+r
is gi-eater
~r
I
Xji^f.
Ji
<
'6.
than or equal to
contrary to hypothesis
GH
so there
is
\.
only
Hence there are only a finite number of terms of the sequence

outside the interval {G - ^, (j + S), where 8 is an arbitrary positive number;
one limit- point.
This proof
is
given by Stolz and Gmeiner, Theoretische Arithmetik,
ii.
(191)2), p. 144.
2 '3]
15
number of such terms, these would have a

would be a limit-point of the given sequence and which
would not coincide with G and therefore G is the limit of {x^).
for, if
there were an unlimited
limit-point which
Now
(II)
sequence
the
let
Cauchy's condition
and
let Zn
^ ^n+p
of real or complex
{Zn)
= Xn
numbers satisfy
where Xn and yn are real then for
iyn,
-f-
values of n and p
all
^n-irp
-^
2'n
yn+p
~ yn ^
\
^n\-
2'n+p
Therefore the sequences of real numbers {x^) and (yn) satisfy Cauchy's
condition
and so, by (I), the limits of (.r) and {y^) exist. Therefore, by
;
2*2
example
Let
Wj,
The
the limit of (2) exists.
Convergence of an infinite
2"3.
the
1,
?/2, u.i,
...
Kit, ...
result
is
therefore established.
series.
be a sequence of numbers, real or complex.
Let
sum
Ml 4- Uo
-I-
tin
be denoted by SnThen,
if
*S^,i
tends to a limit
Hi
"i
is
as
tends to infinity, the infinite series
/;
-f-
^<4
said to he convergent, or to converge to the

series
infinite
is
said
to
SSn, which
expression
be divergent.
sum
the
is
called
In other cases, the
S.
the
series converges,
the
of the series
"+l+ "+2+ Un+-,+

is
sum
When
the remainder after n terms, and
...,
is
frequently denoted by the
symbol R^.
The sum
will
Un+^
11^+..
...-\-
Un+p
be denoted by Sn,p.
by combining the above definition with the
It follows at once,
results
of the last paragraph, that the necessary and sufficient condition for the
convergence of an infinite series
is
we can
e for
find
n such that
/S'_^
|
<
Since Un+i
= ^^n,l,
it
that, given an arbitrary positive
follows as a particular case that lim Un+i
words, the 7?th term of a convergent series must tend to zero as
e,
in other
//
tends to
though necessary, is not sufficient in itself

ensure the convergence of the series, as appears from a study of the series
infinity.
to
But
number
every positive value of p.
In this
series,
this last condition,
Sn,n
'
= n+
+
+ n-^
^,i + n^,
+2
+6
The expression on the right

>^
each term, and so Sn,
,1
is
.^
2n
diminished by writing
(2?i)~^ in
place of
16
Therefore
S^n+i
=1+
/Si,
+ ^2, +
2
>
^ (
>S'8_ g
->
3)
>Si6, le
We may
it will
tained
is
arrive at a series
S=
by some formal process, e.g. that of

by a series, and then to justify the
Simple conditions
convergent.
Given an expression
i(,n
in 1673.
usually have to be proved that the series thus formally ob-'
such circumstances are obtained in

(ii)
+ S.n^
problems which we are called upon to
classes of
solving a linear differential equation

process
II
investigate in connexion with the convergence of series

(i)
was noticed by Leibniz
so the series is divergent; this result
There are two general
'S14, 4
[CHAP.
may be
S, it
for all values of n
Rn, valid
establishing convergence in
for
2'31-2"61.
possible to obtain a development
and, from the definition of a limit,
00
follows that, if
it
we can prove that Rn -^
then the series 2
0,
u^,,
converges
m=l
and
its
sum
is
An
S.
example of
this
problem occurs
in 54.
Infinite series were used* by Lord Brouncker in Phil. Trans. 11. (1668), pp. 645-649,
and the expressions convergent and disergent were introduced by James Gregory, Professor
Infinite series were used systematically
of Mathematics at Edinburgh, in the same year.
by Newton in 1669, De anali/si per aequat. num. term, inf., and he investigated the convergence of hypergeometric series ( 14- 1) in 1704. But the great mathematicians of the
eighteenth century used infinite series freely without, for the most part, considering the
Thus Euler gave the sum
question of their convergence.
...
+ -,+
1
-^
of the series
+ -+l+2 + 2-, + r'+

,
..
{a)
as zero, on the ground that

2
+ ,2 + ,3 + ...^^
1
1+-
and
(6)
\-z
1
-t-
-,
+ ... =
z~
(c).
The eiTor of course arises from the fact that the series {h) converges only when
and the series (c) converges only when s > 1, so the series (a) never converges.
|
2
j
<
1,
For the history of researches on convergence, see Pringsheim and Molk, Encyclope'die
i. (1) and Keifi", Geschichte der unendlichen Reihen (Tiibingen, 1889).
des Sci. Math.,
2*301.
AheVs inequality\.
"*
I
Let fn^fn+\>^ for all integer values of
is
,
i
^ Af,
ivhere
of the sums
the greatest
1
Then] 2 /
n.
+ ^2
o-i
>
I
0.2
a,
.,
rto
+ am
* See also the note to 2*7.
t Journal
Corollary
(i),
fiir
Math.
i.
(1826), pp. 311-339.
also appears iu that
memoir.
particular case of the theorem of 2-31,
2-31]
2* 3 01,
For, writing Ui
ao
a^
= Sn, we
have
rii
t
11
ilnfn
= 5i/i +
- Si)fo + (Ss - S^jfs +
{S.2
/o,/2 fs,
Since /i
(/2
~y 3) +
...
?Pi,
Wj,
sums
the greatest of the
alsO
',
S,n
m,
fm < ^fm,
M' +
is
+ Si_i \Jm-i ~Jm) + Smfmwe have, when n = 2, 3, ...
are any numbers, real or complex,
...
2 aw \<,A\ 2
where J
we get
1'4,
n=l
If i, a^^
Corollari).
(fn-i -fn) ^ ^ (fn-i - fn)
Sn-i
I
S-2.
are not negative,
summing and using
so,
.+ (s^ - Sm-i)fm
=1
Si (/i J 2) +
and
17
2 a
w,^
i
{p = \,
u-
'/'
(Hardy.)
m).
2, ...
n=l
DiHchlet' s*
2"31.
Let
and lim/
< K,
O-f*,
test
for convergence.
luhere
Then, if fn >fn\-\ >
independent of p.
is
((nfn converges.
M=l
For, since lim/,i
<
such that /,+!
= 0,
that,
an\ î
by Abel's
e,
we can
find
m+q
m+q
Then
given an arbitrary positive number
e/2/i
inequality,
a
j
we
cirt
have, for
<
all
2/1', for all
positive values of ^
so
positive values of j;;,
m+p
^ -^Jin+i
CLnJn
n=m+l
where
>
A <2K.
m+p
Therefore
Corollary
cinfn
n = m+l
Jfte^'s test
(i).
< 2Kfm+i <
for convergence.
and
If
so,
by
2-3,
cinf71 converges.
2 converges and the sequence
(m) is
n=l
nionotouic
(i.e.
independent of
,i^?t +
/(,
always or else M^Wrt + i always) and j?(|<k, where
is
then 2 i/ converges.
For, by 2-2, tends to a limit u;
therefore 2 ^' converges.
But,
if
let
(m) is
Then /,i-^0
|w-|=/.
steadily;
and
an increasing sequence, /^=w-m, and
so
rt=i
2 {li-nÛn converges; therefore since 2

=1
ua^^ converges,
2 converges.
n=\
ii=\
If (?<J is
a decreasing sequence / = e<,i , and a similar proof holds.

* Journal de
Math.
(2), vii.
(1862), pp. 253-255.
of Jordan's Cours d' Analyse (1893), Dirichlet's test
Before the publication of the 2nd edition

test were frequently jointly described
and Abel's
as the Dirichlet-Abel test, see e.g. Pringsheim, Math. Ann. xxv. (1885), p. 423.
t In these circumstances, we say/,j^-0 steadily.
W. M. A.
[CHAP.
18
Corollary
Taking a = (-)-i
(ii).
and lim / = 0, /i -ft +/3 -/i +
Example
Shew that
1.
0<^<27r,
if
test,
it
follows that,
p
2 sin(9 <coseci(9
is
Shew
2.
tt.
2 (-)"/' cos (9 converges
that, if fn-^0 steadily,
not an odd multiple of
+^
example
for ^ in
if
and that 2 / cos nd converges
for all real values of 6,
7r
that,
"=i
not an even multiple of
Example
and deduce
n=l
if
/^/;î
if
converges.
. .
2 /sin nO converges
f^^-^Q steadily,
Dirichlet's
in
II
and 2 {-)"-fnS\nne converges
tt
if
is
real
and
=\
for all real values of
6.
[Write
1.]
Absolute and conditional convergence.
2'32.
QO
In order that a
series
complex terms may converge,
Un of real or
it is
n=l
(but
sufficient
moduli
the series of
that
not necessary)
Un
should
n=\
00
For, if
converge.
<Tn,p
Un+i
Un+2
i
Un+p
and
if
71
\ve
can find
n,
corresponding to a given
But
values of jj.
Sn,p
cr_^
<
e,
and
number
so
e,
converges,
|
=1
such that
cTn^p
<
e for all
converges.
=i
The condition
we
see that t
i_}-41
is
not necessary
9 + o~4+---
for,
writing /,i
converges, though
= 1/n
in 2"31, corollary
( 2'3)
(ii),
the series of moduli
+...is known to diverge.

-i
In this case, therefore, the divergence of the series of moduli does not
entail the divergence of the series itself.
Series,
which are such that the
series
formed by the moduli of their terms
are convergent, possess special properties of great importance, and are called
absolutely convergent series.
lutely convergent
(i.e.
Series which though convergent are not abso-
the series themselves converge, but the series of moduli
diverge) are said to be conditionally convergent.
"
The geometric
2"33.
series,
and
the series
2
w=l
The convergence
by the
following articles)
which
of a particular series
direct consideration of the
is
known
by
to
sum
a. comparison of
is
Sn^p,
in
^^
most cases investigated, not
but (as
will
appear from the
the given series with some other series
be convergent or divergent.
We
shall
now
investigate
the convergence of two of the series which are most frequently used as
standards for comparison.
'
THE THEORY OF CONVERGEXCE
2-32, 2-33]
(I)
'
'
The geometric
The geometric
19
series.
series is defined to be the series
+ z--\-z^ + z'+
\-\-z
....
Consider the series of moduli
l+\z\ +
for this series
example
Sn,p =\z\''^'^
=
Hence,
\z\-'
if
<
.g-
\z, "+-
:^
given any positive number
3,
l-\z\P
l-\z\
izr+^
then S,i,p<
1,
+ \z'^+
...->r
z ^^v
'
for all values of
e,
...\
we can
j),
and, by
22,
find n such that
[^|n+i{l_|2i|-i<e.
Thus, given
by
is
we can
e,
find n such that, for all values of p,
Sa,p<.
Hence,
the series
2'22,
convergent so long as 2
<
1,
and therefore
the geometric series is absolutely
convergent if\z\<\.
When
terms of the geometric series do not tend to zero as n
'^1, the
tends to infinity, and the series

VV
Ihe
TT
(II)
therefore divergent.
is
.11111
- -
series
^^
+ ^- + - +
"
Consider now the series
= 2
^,i
7n
^^'e
....
where
112
=
+
have
2"
3^-
<
and
Thus the sum
so on.
sum
so the
6'
2^1
J_
J_
J_
2*~i
4*~i
8^~^
1
]^s-i
and
of
of ani/
number
1.
5*'
greater than
2î
2*-
<=
11114
4^
s is
= l i*
7*
4*-i
terms of the series
is less
of terms
2(2>-i) (s-1)
is less
than
than
1
(1
2^~*
2^~*)~^
Therefore
S m~^
the increasing sequence
cannot tend to infinity; therefore, by
2'2,
w=l
=0
the series
M
positive,
=1
is
convergent if
s>\\ and
since its terms are all real
and
'i
they are equal to their own moduli, and so the series of moduli of
the terms
is
convergent
that
is,
the convergence is absolute.
22
20
If s
+1+1+1+
...,
which we have already shewn to be divergent

divergent, since the effect of diminishing s
<
The
series
n=l
and when
<
a fortiori
1, it is
terms of the
to increase the
is
therefore divergent if s
is
1.
''
The Comparison Theorem.
2'34.
We
II
the series becomes
1,
series.
[CHAP.
now shew
shall
vergent, provided that
independent of
n,
and
that
u^
is
;î
i^,
than
less
2/3+
absolutely
is
...
\vn\, ivhere
con-
some number
is
known
nth term of another series which is
is the
v,i
series
always
to
be absolutely convergent.
we have
For, under these conditions,

I
'^?i+i
'i'i+2
where n and p are any
convergent, the series
<C
Un^p
Vn
Vn+i
But
integers.
|
f ,iJ-2
Si',i
given
so,
...
since the series
convergent, and
is
\
V^+p
1
\
is
absolutely
we can
e,
find
%Un
is
such that
for all values oi p.
"^n+i
^/i+2
.-..+
Vn-\-p
I
Un+i
for all values of p,
the series
i.e.
Wn+2
I
Un
e/ C,
we can
It follows therefore that

1
<
Un+p
find
<
/;
e,
The
convergent.
is
\
such that
series
therefore absolutely convergent.
Corollary.
term of a
series is absolutely convergent if the ratio of its th
which
series
known
is
term
to be absolutely convergent is less than
to the nih.
some number
indej)endent of n.
Example
1.
Shew
that the series

32 +
COS42+...
+2COS22 + .-J5COS
4"
o"
COS,Z
-T,
li"
is
absolutely convergent for
all real
values of
z.
TliZ
iCOS5
n^
the terms of the given series are therefore less than, or at most equal
The moduli
of
ji-
to,
the corresponding
terms of the series

n
which by
2-33
2.
2-2
.3-^
absolutely convergent.
is
convergeut.
Example
Shew
The given
is
+ 7^,
+ ^,
+ 42(2-24)
+ ,
^^,
22(2-2,)
32(2-23)
where
2
all
series
1111
^
convergent for
that the series
l2(2-2i)
is
4^2+-'
values of
2,
= e'",
which are not on the
^s
(?i=l,
circle
2, 3,
= 1.
therefore absolutely
2*34, 2-35]
The geometric representation
of
complex numbers
Let values of the complex number
this kind.
is
21
helpful in discussing a question of
be represented on a plane
then the
numbers Z\t z^, Zz, ... will give a sequence of points which lie on the circumference of the
circle whose centre is the origin and whose radius is unity
and it can be shewn that
;
every point on the circle
For these
a limit-point ( 2-21) of the points z^. -
is
special values
of
z-,^
the given series does not
^,
nator of the nth term vanishes when

for
any point
exist, since the denomiFor simplicity we do not discuss the series
^,,.
situated on the circumference of the circle of radius unity.
Suppose now that
some value
of c
+ l. Then
[zj
values of
for all
7i,
- [2
(1
'}
|>c~i, for
so the moduli of the terms of the given series are less than the corre-
sponding terms of the series

c
which
known
is
convergent for
The given
to be ab.solutely convergent.
all
values of
z,
series is therefore absolutely
except those which are on the circle
is
divided into two parts, between which there
1^1
= 1.
Example
Shew
3.
converges absolutely for
on
2),
^^-fit
.J
values of
all
(2/8")
such that 3" sin
= 2, we
i'
235.
i/'
-*.
For
(2/3") \<i:
]""
<
test
...
-l-2"sin
.fo
...
independent of n (but depending
/,
and therefore
jj
can find
independent of
for absolute
1,
Ave
<Û)
V3.
?/,i
converges absolutely.
=1
such that, when n^ni,
Then, when n >
u.
convergence']'.
ni,
'.
Un \<
p'^]
m^
|^'"
^p <
follows
from
2-34!
that
Un (and therefore
1,
and since
n
it
circle
/2\"
i^) converges, the given series converges absolutely.
Gauchy's
lim
>i
series converges
no intercommunication, by the
can find a number
3"
^
= 1.
2.
2sin^
Since 2
that the series
SsinJ-f 4sin |-f8 sin
Since* lim 3" sin
is
which the
It is interesting to notice that the area in the i-plane over
where p
is
p" converges,
= >n+l
m,J
converges ab-
solute I}'.
[Note.
If lim \u,^\^'^>\,
u^ does not tend to zero, and, by
converge.]
*
This
is
evident from results proved iu the Appendix.
t Analyse Algehrique, pp. 132-135.
2-3,
2 m does not
D'Alemhert's* ratio
2-36.
We
now shew
shall
that a series
U2+
u-i
1/4+
...
absolutely convergent, provided that for all values of n greater than
fixed value
the ratio
r,
II
for absolute convergence.
test
?/i+
is
[CHAP.
22
independent of n and
^^
'
less
For the terms of the
than
less
is
p, tvhere
a positive number
is
some
than unity.
series
Ur+i
Ur+-2
I
Ur+i
...
are respectively less than the corresponding terms of the series
which
is
when
absolutely convergent
<
therefore
Un (and hence
= / + !
the given series)
is
particular case of this theorem
is
that
if
lim
(Un+Jun)
= <1,
l
the
series is absolutely convergent.
For,
by the
!^
11
>
_4 
r,
and then
when
we can
definition of a limit,
<l(l+l)<l,
P-'
r.
[Note.
If lim \u,^ + lî<,,^
% does
>1,
not tend to zero, and, by
2-3,
?<,,
does not
H=l
converge.]
Example
If
1.
|<1, shew that the series
/t=i
all
values of
z.
[For w + i/?i,i = c( + i)--''-e^^c-" + ie^-^0, as ??-^x,
Example
2.
Shew
(a-
Z+-2J-.-+
[For
^^^^1
Un
=^
;Cj<l.]
that the senes
a-6
converges absolutely
if
if
(a-
6)
26)
3j
{a-h){a-2h)ia-^h)
j^
+...
\z\<\b~'^.
'1-z-^-hz, as k-ôo
n+\
^+
'
;
'
so the condition for absolute convergence is
\hz\<\,\.Q.\z\<h-^.'\
* Opuscules,
t.
V. (1768),
pp. 171-182.
2-36, 2-37]
Example
Shew
3.
when ;2|<1,
[For,
2-(l 4-n-i)"
but this latter series
I;
is
_^,^ converges absolutely if i2|<l.
^ (1 +-!)"-
moduli of the terms of the series are

2 n Is""!
that the series
23
^ 1 + 1 +^^ +
2"
i
...
- 1>1,
so the
than the corresponding terms of the
less
series
absolutely convergent, and so the given series con-
verges absolutely.]
2'37.
It
is
is
i/-,i+i
I
general theorem on series for
obvious that
greater than
if,
for all values of
luhicli
|-^ =1.
lim
n greater than some fixed value
r,
Un\, then the terms of the series do not tend to zero as
?i
> X and
,
the series
On
therefore divergent.
is
the other hand,
if
^^
j
is
than some number which

(when n > r), we have shewn
less
of n
The
vergent.
is itself less
in 2-36 that the series is absolutely con-
which, as n increases,
critical case is that in
the value unity.
In this case a further investigation
We shall now shew that* a series u^ +
u..
For, compare the series
y1 is
a constant
Vn
\n
Un
u.i+
..
'
tends to
necessary.
is
.,dn
-^^
which lim
c exists
-"^ =
such that
with the convergent series Ivn, where
we have
1/
nj
V
\'Vn+i
and hence we can
number
will be absolutely convergent if a positive
and
than unity and independent
find
\n
n
^
such that, when n
> m,
u,
By a
all
suitable choice of the constant
values of n
we
shall
Si'n is
convergent,
for
< V,v
Un
i
As
A, we can therefore secure that
have
Un
is
also convergent,
and
so
Sm,i is absolutely
convergent.
* This is the second (D'Alembert's theorem given in 2-36 being the first) of a hierarchy of
theorems due to De Morgan. See Chrystal, Algebra, Ch. xxvi. for an historical account of
these theorems.
[chap.
II
independent of
n,
24
Corollary.
then the series
If
is
1 H
^-
where î
],
<-
if J.i
00
Example.
^
is
\n-J
Investigate the convergence of 2
''
exp
n=l
"
/ -^-2
\), when r>k

1
and when
"v
r<k.
Convergence of the hijpergeometric
2'38.
series.
The theorems which have been given may be
illustrated
by a discussion
of the convergence of the hypergeometric series
a(a+l)6(6+l)
a.h
+ 17^^+ 1.2.c(c + l) ^
which
is
If c
"^
a(a+ l)(a + 2) 6(6 + 1)(6 + 2)

1.2.3.c(c + l)(c + 2)
generally denoted (see Chapter

is
a negative integer,
denominators;
and
either a or 6
if
terminate at the (1
all
F {a,
b; c;
is
"^'
z).
c)th
have zero
a negative integer the series will
term as the case may be. We shall

so that a, h, and c are assumed not to be
a)th or (1
suppose these cases set aside,
XIV) by
the terms after the (1
6)th
negative integers.
In this series
i
Un+i
,
as
??
>
We
00
Ufi
l)(b + n 1)1
n{c + n l)
("+ n
-,
<s
^^ II
*
>
see therefore,
by
2"36,
\z\<l, and divergent ivhen ^

|
When U = 1, we
I
>
have *
0-1]
that the series is absolutely convergent ivhen

1.
2-38-2-41]
Hence
ivhen \z\
= \, a
sufficient condition*
the absolute convergence of
for
the hypergeometric series is that the real part of a
25
+bc
shall be negative.
Effect of changing the order of the terms in a series.
2*4.
In an ordinary sum the order of the terms
is
of no importance, for
can be varied without affecting the result of the addition.

series,
however, this
example.
cr
and
will
--->
and Sn denote the sums of their first n terms. These infinite

formed of the same terms, but the order of the terms is different,
let 2,i
so
Let
appear from the following
-.
,S'=l-2 + 3-4 + 5-6 +
series are
and
no longer the casef, as
it
infinite
^,11111111
= H-3-2+5 + 7-| + + n-o+---
T.
Let
and
is
In an
-^ and Sn are quite
distinct functions of
n.
26
be an absolutely convergent
and
series,
let
[CHAP.
II
S' be a series formed by the same
terms in a different order.
Let
be an arbitrary positive number, and
let
n be chosen so that
^
'
for all values of p.
Suppose that
terms of S'
if
to take
k > m,
= ^n + terms
^k
n terms of S we have
in order to obtain the first
then
of
with
suffices greater
than
n,
so that
>SV
Now
S = Sn S + terms
sum
the modulus of the
with suffices greater than
number
of any
sum
greater than n does not exceed the
than
of
of terms of
of their moduli,
n.
with suffices
and therefore
is
less
2 e-
Therefore
>Sâ;'
'S'
< Sn- S\ + ^.
But
j
^ - >Sf < lim

I
Un+i
11
Un+-2
Uri+p
i
p-*-x
1
Therefore given
we can
find
such that
\Si:-s\<
when k >
therefore
/S^,,/>
S,
which
the required result.
is
If a series of real terms converges, but not absolutely,
sum
of the first
positive terms,
and
if
terms, then 8p>cc, cr> oo; and lim

are given
some
Riemann
that
relation
it
is
between
Let
iii,n
cr,j)
be the
negative
does not exist unless
It has, in fact,
n.
>S^^
we
been shewn by
relation,
to
make
given real number*.
be a number determinate
for all positive integral values of
consider the array
?? ;
* Ges.
(>Sp
if
first n
Double series\.
2'5.
and
p and
by choosing a suitable
possible,
lim(>Sf^4- (7n) equal to cnii/
and
an be the sum of the
Werke,
^],l)
WjÔ)
'**I,3j
'^'2,1
^'2,2)
*'2,3)
^';i, 1
^h, 2
^^3, 3
p. 221.
complete theory of double
series,
on which this account
is
based,
is
given by Pringsh&iui,
Miinchcner Sitzunysberichte, xxvii. (1807), pp. 101-152.

See further memoirs by that writer,
Math. Ann. liii. (1900), pp. 289-321 and by London, ibid. pp. 322-370, and also Bromwich,
Infinite Series, which,
ments
of the subject.
Sac. (2),
I.
addition to an account of Pringsheim's theory, contains
many
develop-
Other important theorems are given by Bromwich, Proc. London Math.
(1904), pp. 176-201.
2-5, 2-51]
27
Let the sum of the terms inside the rectangle, formed by the
first n columns of this array of terms, be denoted by S,n,n-
rows of the
is
possible to find integers
number
If a
any arbitrary positive number

and n such that
exists such that, given
*S'
whenever both
/a
>
general element
is
u,,.^^
and
ru
>
we say* that the double series of

to the sum S, and we write
n,
converges
^=
lira
we say that the given double

Since
w^
is
and
Stolz necessary
is
]
w^,^
|,
is
convergent,
is
it
easily seen
is
Uf^^,
0.
for convergence.
suffi^cientf condition
condition for
2'22) is that, given
obviously necessary (see
and n such that S^+p, y^^ *%, / < e whenever fi > m and
The condition is also
p, a may take any of the values 0, 1, 2, ....
for, suppose it satisfied
then, when fM> m + n, >S'^+p,^+p - S^^^ < e.
find
p >
and
way that
v>
An
Corollary,
n; that
moduli, then, given
p,
is
+p=
*S';
a,
f,
sufficient;
and then making p and a tend to

see that \S 8,,^^, ^ e when-
v,'e
to say, the double series converges.
absolutely convergent double series
series converges absolutely
> n and
Therefore, by 2*22, S^^^ has a limit

infinity in such a
we can
e,
ever
that, if
convergent, then
lim
convergence which
luliich the
series is absolutely convergent.
= (/SV,^ >SV,-i) ('S'^-],-'Sm-i,^-iX
the double series
e, it
S.
*S'
fi,V
which the general element
If the double series, of
first
and
we can
if
find
But \Sp^q-S,^n\îp,q-tm,H and
ty,t^n
fx
he the
sum
of
For if the double

is convergent.
rows of n columns of the series of
i,j,q t,n,n<^q>n>fi; and this
such that, when p>ra>fi and q>/i>fi,
so \Sp^g-S,^n\<e
when
jc??i>/x,
the condition that the double series should converge.
is
Methodsl of summing double
2'51.
series.
00
CC
Let us suppose that S
sum by rows
called the
OC/X\
is
5^
li^,
u^^^,
of the double series
Similarly, the
).
sum
converges to the
sum
that
by columns
is
is
That these two sums are not necessarily the same

Su
'^'
and
*
=
S
p+
in
which the sum by rows

-^
is
1,
the
Then S
S^.
to say, the
defined as
is
shewn
b}^
is
sum by rows
30/00\
2
^V,*')-
the example
sum by columns
-^
>S'^
is
+1
does not exist.
This definition
is
practically due to CsiU.chy, Analyse Algehrique, p. 540.
t This condition, stated by Stolz, Math. Ann. xsiv. (1884), pp. 157-171, appears to have
been
first
proved by Pringsheim.
J These
methods are due
to
Cauchy.
28
Pringsheim's theorem*
exist,
If S
and
exists
the
[CHAP.
II
sums by rows and columns
then each of these suyns is equal to S.
For since
then we can find
exists,
*S^/i,
S <
if
e,
yu.
such that
>
7u
> m.
And
^ Sp S
In
lim
therefore, since
when
iê
fx
>
>Sf^
{Mvn
m, and so
2"52.
We
Absolutely convergent double
can prove the analogue of
222) the sum by rows converges
to S.
is
to S.
series.
2 '41 for double series,
that
say,
8^,^^)
manner the sum by cohimns converges
like
S %e;
to
exists,
namely that if
the
terms of an absolutely convergent double series are taken in any order as a

simple series, their sum tends to the same limit, provided that every term occurs
in the
summation.
Let
cr^
sum
be the
fx rows and v columns of the

and let the sum of this double
and n such that o- cr^ < e
of the rectangle of
double series whose general element
is
m^,
;
|
series
be
Then given e we can

fi > m and v> n.
cr.
whenever both
Now
find
suppose that
it is necessary to take iV terms of the deranged series

which the terms are taken) in order to include all the terms
and let the sum of these terms be ty
(in ^the order in
of
>S^j/-fi,3/+i,
Then #a' 'Sj/-t-i, j/+i consists of a sum of terms of the type

> m and
> n and therefore
p > m, q >n whenever
Also,
S Sji+iîjjî
S - Sjfi+ijf+i ^
any given number

Example
1.
Sm+1,3T+1
e,
we can
2 ^
'
O"
0-j/+i_ 3/+1
which
f-
> m, q>
j)
Sty <
therefore
X; and
<
which
in
iip^q
find
Up,q in
consists of terms
o-jf+^^M+i <
<r
to
h'
n; therefore
and, corresponding
therefore ty-^S.
Prove that in an absolutely convergent double
series,
!<(,
exists,
and
let
be the
sum
H=l
thence that the sums by rows and columns respectively converge to S.
sum
[Let the
of
fi
rows of
columns of the
series of
moduli be
t^^ ^,
and
of the series of moduli.
Then 2
?i^, ^
\<t,
and so 2
\bi\
m^,
;,
+ \b.i\ +
and so 2 b^ converges absolutely.

exists,
converges
...
let its
+ \b^\ ^
exists
be 6^
then
lim t^^^^t,
Therefore the
and similarly the sum by columns
sum
sum by rows
of the double
series
and the required result then follows from
Pringsheim's theorem.]
* Loc. cit. p. 117.
2 52-2 6]
Example

Shew from
2.
first
principles that
29
the terms of au absokitely convergent
if
double series he arranged in the order
M + (2,l + l,2) +
series converges to
tliis
2"53.
+ 2,2 + "l,3') + (*/4,l + --- + '<l,4) + ---5
(:i,l
^S*.
CaucJnj's theorem* on the
multiplication of absolutely convergent
series.
We
now shew
shall
that if two series
S=
+ V. + +
T =Vi + Vo + V3+
and
u^
(/3
...
are absolutely convergent, then the series
P = U{Vi +
formed by the products of their
vergent, and has for sum ST.
Sn
Then
by example 2 of
"i
V-i
terms, written in
Ui
Now
SnTn =
Ui
I'l
double series
this
U.,
<7"
rn
cTnTn is
U>Vi
l/il'2+
U.,V.^
known
to
sum converges
Shew
Example.
and rearranging
=
=
"1
\Vi
have a
absolutely con-
...
4-
HnVi
UnV.^
is
/2
I
+ n
+
+ ...+\Vn\,
\v.,\
Therefore, by 2o2,
limit.
these terms are
if
a^Tn, where
is u,nVn,
if
the elements of
be taken in any order,
that the series obtained by multiplying the two series
22
23
^4
+^+
2;.
+ 24 +
accoi'ding to
powers of
z,
,111
+ +^+^+
-'
-'
converges so long as the representative point of
the ring-shaped region bounded by the circles
\z\
=l
and
= 2.
Fower-Seriesf.
series of the type

Uo
in
is
to ST.
2"6.
.,
any order,
the double series, of which the general term
lies in
absolutely convergent; for
is
replaced by their moduli, the result
their
2'2.
+
+
and
^1
+ Un,
+ +
Tn=V, + V.,+ ...+ Vn.
=
ST lim S,, lini T = lim (SnTn)
Let
But
1^2
which the
aiZ
a^z-
coefficients a^, a^,a2, a^,
proceeding according
to
+ a-iZ"
...
-\-
...,
are independent of
z, is
called a series
ascending powers of z, or briefly a poiver-series.
Analyse Algebrique, Note
The
vii.
results of this section are due to
Cauchy, Analyse Algebrique, Ch.
ix.
THE PROCESSE>S OF ANALYSIS
30
We
it ivill
[CHAP.
II
that if a power-series converges for any value z^ of z,

he absolutely convergent for all values of z whose representative points
shall
now shew
are luithin a circle luhich passes through
and has
z^
centre at the origin.
its
00
For, if z be such a point,

anZo^
must tend
we have
to zero as ?i->oo
<
^'o
Now,
since
and so we can find
V^ converges,
M (independent of n)
such that
0.nZ,''
\<M.
anz''
\<M
Thus
^\
00
Therefore every term in the series
term in the convergent geometric
n^"
is less
than the corresponding
series
2i)/ the series
therefore convergent
is
and so the power-series
convergent, as the series of moduli of
the result stated
Let lim
absolutely
a convergent series
is
is
~^"*
|
terms
its
is
=r
then, from 2-35,
anZ^ converges absolutely
when
00
\z\<r\
The
\z\>r, anZ" does not tend to zero and so
if
circle
\z\=r, which includes
a^z^^ diverges ( 2'3).
the values
all
of z for which the
power-series
tto
aiZ
a.iZ"-
ag^^
-f-
converges, is called the circle of convergence of the

the circle is called the radius of convergence.
In practice there is usually a simpler way of finding
r is lim (/ + 1) if this limit exists.
test ( 2-36)
r,
series.
The
radius of
derived from d'Alembert's
power-series
may
converge for
all
values of the variable, as happens, for
instance, in the case of the series*

z^
which represents the function

whole ^r-plane.
On
zero
sin z
in this case the series converges over the
the other hand, the radius of convergence of a power-series
may be
thus in the case of the series

I
+ 1\ z + 2\
The
z- -\-Z\ z^
+ 4>\
z' -^ ...
= n\z
we have
*
z^
U
series for c^, sin
log z will be
z,
cos z
assumed throughout.
in the Appendix.
and the fundamental properties of these functions and of
brief
account of the theory of the functions
is
given
2 61]
"
31
which, for
all values of n after some fixed value, is greater than unity when
any value different from zero. The series converges therefore only at
the point z = 0, and the radius of its circle of convergence vanishes.
z has
power-series
may
may
or
the periphery of the circle
-'-
z-
+p+
25
whose radius of convergence

according as s
Corollary.
limit
is
is
is
z*
z'-'
+ 3 + 4i +
'
unity, converges or diverges at the point z
greater or not greater than unity, as was seen in
If () be a sequence of positive
equal to lim
261.
not converge for points which are actually on
thus the series
2-33.
terms such that lim(a+i/a)
exists, this
'"'.
Convergence of series derived from a poiuer-series.
Let
cify
i2
+ a.iZ- +
a^z^
a^r*
be a power-series, and consider the series

ai
which
is
1a.,z
-\-
+ 4a42-' +
â^z"
.,
obtained by differentiating the power-series term by term.
now shew
shall
We
that the deHved series has the same circle of convergence as the
original series.
For let 3 be a point within the circle of convergence of the power-series

and choose a positive number 7'i, intermediate in value between \z\ and r the
X
radius of convergence.
terms must tend to zero as n

positive
of
Then, since the series
> x
number M, independent
and
of
it
nî" converges absolutely, its
must therefore be
such that
<
possible to find a
Mr^~^^ for all values
n.
Then the terms
"~^
are less than the corre-
2'36, since \z\<r-^.
Therefore, by 2'34, the
of the series
?i
21
=i
sponding terms of the series
But
this series converges,
by
71:2,"-'
=i
n"~'
series
^ncin
\z\''-^
converges; that
is,
the series
^la^^""^ converges absolutely for all points z
n=l
situated within the circle of convergence of the original series
OnZ''^.
When
n=
> ?^
^
I
anZ^^
does not tend to zero, and a fortiori
na^z^''
does not tend to
zero
and so the two
series
have the same
circle of
convergence.
[CHAP.
32
Corollary.
The
2 -
series
term by term, has the same
obtained by integrating the original power-series
>
convergence as 2
circle of
II
a^z""-.
n=0
Infinite Products.
2"7.
We
next consider a class of
Let
vanish.
+ ai,
If,
as n
oo
+ as,
a.2,
known
limits,
as infinite products.
be a sequence such that none of
its
members
the product
(1
ai) (1
2) (1
+ as)
(1
+ a)
(which we denote by Tin) tends to a definite limit other than

is called the value of the infinite product
zero, this limit
n = (l + a0(l + a,)(l + a3)...,

and the product
is
said to be convergent
condition for convergence
that lim
is
It
*.
cin
is
= 0,
almost obvious that a necessary
since lim Un-i
= lim Un + 0.
00
The
limit of the product
written II (1
is
4- cin).
n-l
Now
(l+a) =
exp-^
w=l
exp
andf
if
the former limit exists
lim
,]
rn
log(l +a,,)K
n-\
lim {exp
ii,n]
hence a sufficient condition that the product
00
should converge
is
that
log(l
a) should converge
when
the logarithms
n=l
have their principal values. If this series of logarithms converges absolutely,

the convergence of the product is said to be absolute.
The
condition for absolute convergence
in order that the infinite
given by the following theorem
is
(l+a,)(l+a2)(l +
may
product
he absolutely convergent,
it is
tti
necessary
+ 02 + as +
a3)...
and
.
sufiicient that the series
sJiould be absolutely convergent.
For,
by
definition, 11
is
absolutely convergent or not according as the
series
log (1
is
+ Oi) + log (1 + Oo) +
log (1
+ ag) +
...
absolutely convergent or not.

*
The convergence
of the product in
as 1655.
t See the Appendix, A-2.
which
rt,i_i=
- l/n^ was investigated by WaUis as early
27, 271]
Now,
THE THEORY OF COXVERGEXCE
= 0, Ave
since lim a
can find
33
such that, when n >
?n,
|
<
and
then
"' log (1
Un)
23+--- =
+ 2^+
<2^.
2-^
And
thence,
when
n> m, ^^
^h
-^^^
theorem, the absolute convergence of

conversely, provided that a
4=
1 for
therefore, by the comparison
log (1
) entails that of
any value of
Sa^ and
n.
This establishes the result*.

If,
number
in a product, a finite
of factors vanish,
the resulting product converges, the original product
a product as
n (!-"')
and
is
if,
when
these are suppressed,
But such
said to converge to zero.
said to diverge to zero.
is
n=2
of
Corollary.
Since, if Sn-^'l, exp (S')--exp ^, it follows from 2-41 that the factors
an absolutely convergent product can be deranged without aftecting the value of the
product.
Example
Shew
1.
that
if
(1
+) converges, so does 2 log
H=l
(1
+a),
if
the logarithms
)l=l
have their principal values.
Example
Shew
2.
that the infinite product

sin z sin \z sin \z sin ^z
z
is
all
\z
'
values of
[For (sin-j /(-) can be written

lêndent of n
it
with 2
w=l
-^^.
and the
The
2 ^
series
infinite
product
'
is
'
the form
in
is
\z
\z
'"
z.
|,
where
X |<X- and /
aVjsolutely convergent, as is seen on
i.s
inde-
comparing
therefore absohitely convergent.]
'^
Some examples of infinite
2"71.
products.
Consider the infinite product
{^-^){^-m-)
which, as will be proved later
In order to find whether

series
product
Now
-,
is
or -
it is
7 '5), represents the function z
absolutely convergent,
sin
we must
this series is absolutelv convergent,
let
all
values of
z.
consider the
and so the
z.
the product be written in the form
discussion of the convergence of infinite products, in which the results are obtained
without making use

(1889), pp. 119-154,
W. M.
A.
of the
and
logarithmic function,
also by
Bromwich,
is
given by Pringsheim, Math. Ann.
xxxm.
Infinite Series, Ch. vi.
34
The absolute convergence
But
is
[CHAP.
of this product depends on that of the series

z
IT
IT
27r
27r
this series is only conditionally convergent, since its series of
divergent.
\z\
\z\
\z\
\z\
IT
IT
27r
27r
In this form therefore the infinite product
convergent, and
so, if
II
the order of the factors
is
is
moduli
not absolutely
deranged, there
is
a risk of altering the value of the product.

Lastly, let the
in
same product be written
in the form
which each of the expressions

1
e^mn
miTj
is
counted as a single factor of the infinite product.
of this product
depends on that of the
series of
The absolute convergence

which the (2?n l)th and
(2m)th terms are

1
But
it is
\m-/
infinite
convergent, the
series in question follows
+ l+2^ +
product in
2,
this
3.
+ 3, +
last
form
adjunction of the factors
c is
+ p+....
is
'*""
part of the factor theorem of Weierstrass
1.
Prove that
n
n=i
z, if
4.
](l
l.\
c-f-?i/
e"
therefore
is
is
infinite
product converges absolutely with the series
n=i t\
+ nj
a particular case of
( 7'6).
a constant other than a negative integer.
For the
again absolutely
having changed the con-
This result
vergence from conditional* to absolute.
Example
by comparison
111111
with the series
first
1.
mTTJ
and so the absolute convergence of the
the
easy to verify that
The
mn
all
values of
271]
Now
the general term of this series
is
je"-!-
converges, and so, by 2-34, 2 ](l

n=\ l\
and therefore the product converges absolutely.
But 2
n=l
Example
Shew
2.
n jl-H--]
that
outside a circle whose centre
For the
infinite
product
is
term of the
But lim (l
series to the Jith
+ %/
converges absolutely,
z~'\ converges for
all
points
situated
the origin and radius unity.
absolutely convergent provided that the series
is
=o
is
35
term
is -
yn
S-"
=e, so the limit of the ratio of the (w + l)th
-- )
there
is
therefore absolute convergence
when
z
1
Example
Shew
3.
<
when
i.e.
1,
>
1.
that
1.2.3...(m-l)
(2+l)(2 + 2)...(2 + 7ft-l)
tends to a
finite limit as ??j--x
unless 2
is
a negative integer.
For the expression can be written as a product of which the nth factor
This product
is
is
+ 7i
\ n
~\
n)
22
is
^""U'
therefore absolutely convergent, provided the series
When
this is the case.
and a comparison with the convergent
2 is
series
* 1
2 -^ shews that
a negative integer the expression does not exist because one of
the factors in the denominator vanishes.
Example
Prove that
4.
=e
--log
^
sin
2.
For the given product
,|o..(,-i)(i-i)(i.|)...(.-^,^)(i-^)(i.)
(,77
^"''2
' 2k-l
2k
k/
lim
lim
--
^-27'^^
e"-V 2+3"-+2A-i
2k)
Ji_^^\
1-
,2kn
2/C7
en fi
+ ^\
e~ ^
2 \
+ klT
-1
(l-ê^^ (1+^ e
<!-...,
32
36
[CHAP.
II
since the product whose factors are

1
is
absohitely convergent,
and so the order
this shews that the given product
of its factors can be altered.
equal to
is
--logs
e
sin
"
2.
Infinite Determinants.
28.
and
Infinite series
known
e^
l-HJ-i + *---M
log2 =
Since
infinite
by any means the only
not
products are
The
cases of limiting processes which can lead to intelligible results.
researches of G. W. Hill in the Lunar Theory* brought into notice the

possibilities of infinite determinants.
The actual
investigation of the convergence
la Soc. Math, de France, xiv. (1886), p. 87.
H. von Koch, Acta Math,
is
due not to Hill but to Poincare, Bull, de

shall follow the exposition given by
We
xvi, (1892), p. 217.
Let Aik be defined

and denote by
for all integer values (positive
the determinant formed of the numbers Aik{i,k

as
m -^ cc
and negative) of
m,
...
+m);
we
the expression D,n tends to a determinate limit D,
i,
then
k,
if,
shall say
that the infinite determinant

[-^>i-J?,i-=-<...+oo
is
convergent and has the value D.
minant
T^he elements
An, (where
diagonal of the determinant

takes
is
all values),
fixed
A-iy; is
does not exist, the deter-
and
takes
D;
all values),
are said to form the principal
the elements Aik, (where
2'81.
fixed
i is
and k
are said to form the 7'ow i; and the elements A^c, (where k
takes
all
values), are said to form the
column
k.
called a diagonal or a non-diagonal element, according as
The element
We
If the limit
in question will be said to be divergent.
udo.o is
Any element
= A; or i $ k.
called the origin of the determinant.
Convergence of an infinite determinant.
shall
now shew
that
an
infinite
determinant converges, provided the product of the

the non-diagonal elements converges
diagonal elements converges absolutely, and the sum of

absolutely.
For let the diagonal elements of an infinite determinant I) be denoted by l+a^,

and let the non-diagonal elements be denoted by ajj., {i=k), so that the determinant is
*
Reprinted in Acta Mathematica,
viii.
Infinite determinants
(1886), pp. 1-36.
occurred in the researches of Fiirstenau on the algebraic equation of the
had previously
7ith degree,
Darstellung
der reellen Wurzeln alyebraincher Gleiclnmgen durch Determinanten der Coeffizienten (Marburg,
Special types of infinite determinants (known as continuants) occur in the theory of
1860).
infinite
continued fractions; see Sylvester, Math. I'apers,
i,
p.
504 and in,
p.
249
2-8-2'82]
Then, since the
series
37
i,k=-
is
convergent.
Now
form the products

m
m
2
P,=
1+
au-
P,=
1+
m
2
= -n( \
fc=-m
terms are replaced by zero and certain other
terms have their signs changed, we shall obtain i), thus, to each term in the expansion
of 2>, there corresponds, in the expansion of P,, a term of equal or greater modulus.
fc=
i"=-ni\
then
if,
in the expansion of Pi, certain
Now An + p - An
when
sum
represents the
i), + p which vanish

by zero; and to each of
of those terms in the determinant
the numbers ai^{i, k=,{m-\-\)
...
(hi+jo)} are replaced
these terms there corresponds a term of equal or greater modulus in
Hence
,-D^\<R,
Pm + p^m-
-P.
Therefore, since P,n tends to a limit as ni-*-cc
so also
Z),
This
tends to a limit.
establishes the proposition.

2"82.
We
The rearrangement Theorem for convergent
infinite determinants.
now shew
that a determinant, of the convergent form already co)isidered,

remains convergent when the elements of any row are replaced by any set of elements whose
shall
moduli are
all less
than some fixed positive mimber.
Replace, for example, the elements

-"^0,
row through the origin by the elements
.../!_,,...
A,
'0
>i
of the
jiQ ... ix,n ...
which satisfy the inequality

I
M>-
< Mj
where /x is a positive number and let the new values of Z)i '^nd D be denoted by
Dm and D'. Moreover, denote by /*/ and P' the products obtained by suppressing in
P,n and P the factor corresponding to the index zero we see that no terms of 2)^' can
have a greater modulus than the corresponding term in the expansion of nP^' and
consequently, reasoning as in the last article, we have
;
which
is sufficient
Example.
to establish the result stated.
Shew
that the necessary and sufficient condition for the absolute conver-
gence of the infinite determinant

lim
a,
...
...
a.,^
^2
,0
is
...
as
...
/3,
that the series

ai/3i
shall be absolutely convergent.
+ ao/32 + 03/33 + ...

(von Koch.)
38
[chap.
II
REFERENCES.
Convergent
series.
A. Pringsheim, Math. Ann. xxxv. (1890), pp. 297-394.

T. J. I'a. Bromwich, Theory of Infinite Series (1908), Chs.
Conditionally convergent
Ii, ill, iv.
series.
G. F. B. RiEMANN, Ges. Math. Werke, pp. 221-225.

A. Pringsheim, Math. Ami. xxii. (1883), pp. 455-503.
Double
series.
A. Pringsheim, MUnchener Sitzungsherichte, xxvii. (1897), pp. 101-152.

Math. Ann. hill. (1900), pp. 289-321.
G. H.
Hardy, Proc. London Math.
Soc. (2)
(1904), pp. 124-128.
i.
"""#), lim
1.
Evaluate
2.
Investigate the convergence of
litn (e
(?i~
ôgn) when
a>0, b>0.
(Trinity, 1904.)
3.
Investigate the convergence of
(l.3...2n-l
- 1
=i\
4.
.
2. 4, ..2%
Find the range of vakies of z
for
Shew that the
'
series
+ (-) + i2sin2'^s+...
series
1
l_
2+1
1_
_1
2
+2
+3
Jl_
2+1
+ p\
11
11
Z+p + l
Z+p
7.
but that the series
+ 2p + ^-l + 2 + 2^ + +
..
5-
is
divergent.
(Simon.)
Shew that
li 1 + 1-1- 1+1Shew that the
series
=ilof^2
1111
1"
23
3"
(Trinity, 1908.)
(l<a</3)
4^^
convergent, although
2n + l/W2n-â>8.
Shew
(Ceskro.)
that the series

a
is
(Peterhouse,
1906.)^
^
which (p + q) negative terms always follow p positive terms,

6.
is
^.
conditionally convergent, except for certain exceptional values of z
in
+3
+ 2j
convergent.
5.
is
2>i
which the
2sin2s-4sin-2 + 8sin''s-...
is
4/1
+ /3- + a3 + /3< + ...
(0<a</3<l)
convergent although
(Cesaro.)

Shew that the
9.
39
series
^ z"-i{(H-%-i)-l}
=l(3"-l){3-(l+->)}
converges absokitely for
all
values of
except the values
z,
Z(\ j^^gtkKilm
(a
Shew
10.
1; k = 0,
when s>
that,
series
= l,
2, 3, ...).
j-ji
r^ +
Lii
<s<
on the right converges when
Mem. de VAcad. de
(de la Vallee Poussin,
In the series whose general term
11.
m-\
1, ...
1,
i=^+
and shew that the
= 0,
1.
Belgique,
denotes the
number
lim
and that the
series is convergent,
Shew that the
u^^ = q,
although lim Mn+i/wn==c
series
= j'
the
+ l)th
where
(0<5'<1)
""'"""',
!?v
convergent, although the ratio of
is
(0<9<1<^)
of digits in the expression of n in the ordinary decimal scale
shew that
of notation,
12.
when n
term to the nth
(?t
is
greater than unity
not a triangular number.
is
Shew
13.
(1896), no. 6.)
is
= y-^yi'"'+l',
where
liii.
(Ceskro.)
that the series
2
=o(<'
where
taken
in its
is
is real,
and where {w + nY
arithmetic sense,
If
M,i^
w,i
15.
convergent for
all
mean
eiog(w + n)^
values of
s,
when x
the logarithm being
when 1
{x) is positive,
whose
real part is positive,
Un>0, shew that
if 2?<
converges, then lim {nu,^ = 0, and that,
then lim
i,
understood to
convergent for values of

14.
is
is
+ )*'
(?i?<)
is real
and not an
if
and
integer.
in addition
= 0.
mn
If
om+n
2m+n
"MLn
am,o = 2-'",
{m + n
1)
m\ n\
ao,=-2-,
(m,
'
n>0)
ao,o=0,
shew that
(Trinity, 1904.)
n!=0 \?i=0
16.
By converting
=0 \m=0
the series
.2
l-q + 1+^2
(in
which
g|
< 1),
into a double series,
X2
{\-qf
"f"
l3 +
1
l-jS
shew that
2
24g3
16g2
1+:
+ (1+^2)2
it is
equal to
8(73
(1_^3)2
(Jacobi.)
40
17.
sin2 = 3
Assuming that
shew that
?-^qo
if
and
3t-*-co iu
n'
If Uq-=Ui
= U2 = 0,
and
if,
(1 +?<)
n=0
is
= k,
where k
is finite,
then
omitted.
(Math. Trip., 1904.)
when n>\,
W2-i then
(in\n)
=F''^
the prime indicating that the factor for which r =

18.
II
n (I--5,),
such a way that lim
hm
[CHAP.
^7l'
si 11
n y/n
converges, though 2 and 2 2 are divergent.

n=0
?i=0
(Math. Trip., 1906.

19.
Prove that
n
where k
20.
is
If
any
2
-^
--
exp
positive integer, converges absolutely for all values of
a,i
z.
be a conditionally convergent series of real terms, then
n=l
(l4-,i) con-
n=l
verges (but not absolutely) or diverges to zero according as 2 a2 converges or diverges.

11=1
(Cauchy.)
21.
Let 2 dn be an absolutely convergent

11=1
A{C)-
(C- 4)2-^0
series.
Shew
that the infinite determinant
CHAPTER
III
CONTINUOUS FUNCTIONS AND UNIFORM CONVERGENCE

The dependence of one complex number on another.
3"1.
The problems with which Analysis
is
mainly occupied relate to the

If z and ^ are two complex
dependence of one complex number on another.
numbers, so connected that,
if
given any one of a certain set of values,
is
corresponding values of ^ can be determined, e.g. if ^ is the square of z, or if

for all other values of z, then ^ is said to be a
^ = 1 when z is real and ^ =
function of
z.
This dependence must not be confused with the most important case of
it,
which
If f
may
is
will
be explained later under the
real function of a real variable
z,
title
o{ analytic functionality.
then the relation between ( and
2,
which
be written
can be visualised by a curve in a plane, namely the locus of a point whose coordinates
No such simple and convenient
referred to rectangular axes in the plane are (s, ().
geometrical method can be found for visualising an equation
considered as defining the dependence of one complex
complex number
for real variables

^,
fj,
X,
One
is
now
=x+
i)/.
f=|
on another
i'7
would require four-dimensional space, since the number of variables
four.
is
to use a doubly-manifold
system of
quadruply-manifold totality of lines in three-dimensional space.
Another suggestion
A third
number
representation strictly analogous to the one already given
suggestion (made by Lie and Weierstrass)
lines in the
is
to represent ^
suggestion, due to Heffter*,
then draw the surface
function and on
possible to modify
on the surface r=r
3'2.
it
= r{x,
to express
y)
is
;;
separately by
means
of surfaces
to write
which
may
be called the modular-surface of the

It might be
the values of 6 by surface-markings.
this suggestion in
(.r,
and
various ways by representing 6 by curves
drawn
y).
Continuity of functions of real variables.
The reader
will
have a general idea (derived from the graphical represen-
tation of functions of a real variable) as to

* Zeitschrift
fur Math,
tind
what
is
Phys. xlix. (1899),
meant by
p. 235.
continuity.
42
We
now have
which
to give a precise definition
[cHAP.
shall
embody
this
Ill
vague
idea.
when a ^x ^b.
Jjetf{x) be a function of x defined
Let
be such that a ^x^ ^b.
x^
corresponding to an arbitrary positive number
number
rj
number
If there exists a
we can
e,
such that,
find
a positive
such that
\f{a:)-l\<e,
Xi\<
whenever \x
^ x^, and
rj,
may happen that we can

that \f(x) l+\< when
It
^x ^ b,
find a
then
number
is
called the limit of /(a;)
(even when
1+
does not exist)
We call Z+ the limit of f{x)

x-^ < x < Xi +
when X approaches x-^ from the right and denote it by/(a;i + 0); in a similar
manner we define f{x^ 0) if it exists.
such
If f(xi
continuous at x^
7]
0), f{xi), f{x-^

;
rj.
0) all exist
and are equal, we say that
so that \if{x) is continuous at
x-^
then, given
e,
/(a-) is
we can
find
such that
\f{x) -f(x,)
whenever \x
Xi\<
and a
rj
<
e,
^x ^h.
If 1+ and l_ exist but are unequal, f{x)
discontinuity* at Xi] and if 1+

discontinuity at
lif{x)
where g
is
From
,
is
For when \f(x) f{x^ <
(x).
the result stated
Example.
have an ordinary
^ 2*2
so are f{x)
The popular
= g(x) + i h
(x)
(x) are real, the continuity of f(x) at x^ implies the con-
h{xi)\ < e and
tinuous at Xi
f{^)
to
said to have a removable
a complex function of a real variable, and \if{x)
tinuity of g (x) and of A
and h (x)
/_ 4=/(î),
said
is
x-^.
and h
(x)
examples
f{x) x
{x),
is
and 2 deduce that
{x)
and,
if
e,
then g (x) g(xi)

|
<
obvious.
^ (.rj)
=t=
if
f{x) and
0, f{x)/(l)
cf)
(x) are
con-
(.*).
i-eal variable f{x) depending

from that just considered, and may
perhaps best be expressed by the statement "The function f{x) is said to depend continuously on X if, as x passes through the set of all values intermediate between any
two adjacent values Xi and X2, f{x) passes through the set of all values intermediate
between the corresponding values /(^j) and/(.r2)."
idea of continuity, so far as
on another real variable
The question thus
x, is
arises,
somewhat
how
it
relates to a
different
far this popular definition is equivalent to the precise
definition given above.
Cauchy shewed that

definition,
then
it
if
satisfies the precise

a real function f{x), of a real variable
what we have called the popular definition this result
* If a function is said to
is
implied that
it is
.^,
also satisfies
have ordinary discontinuities at certain points of an interval
continuous at
all
other points of the interval.
it
3*2 1]
43
be proved in ^ 3"63. But the converse is not true, as was shewn by Darboux.
may be iUustrated by the following example*.
vill
fact
x= I
Between
It can
and
x= +1
(except at x=0), let f{x)
= sin
This
and let/(0)=0.
then be proved that/{.r) depends continuously on x near .r=0, in the sense of

in the sense of the precise definition.
is not continuous at a; =
the popular definition, but
Example.
If f{x)
be defined and be an increasing function in the range
(a, b),
the
limits /(.r + 0) exist at all points in the interior of the range.
f{x) be an increasing function, a section of rational numbers can be found such

A be any members of its Z-cla.ss and its ^-clas.s, a<f{x + h) for evei-y positive
value of h and A '^f{x-\-h) for some po.sitive value of h. The number defined by this
[If
that, if a,
section is/(a; + 0).]
Simple
3"21.
Continua.
curves.
Let X and y be two real functions of a real variable t which are continuous
We denote the dependence of x and y
t such that a-^t %h.
for
every value of
on
by writing
.'^
^(0.
3/
= i/(0-
{a.^t%h)
The functions x (t), y (t) are supposed to be such that they do not assume the
same pair of values for any two different values of t in the range a < t < b.
set of points with coordinates (x, y) corresponding to these values
Then the
of
is
called a simple curve.
X
the simple curve
Example.
The
is
If
(a)
said to be
circle x^
+ 7/-= 1
= X {h),
is
(a)
=y
(b),
a simple closed curve
x=coiit,
closed.
two-dimensional continuum
is
for
we may write t
y = smt.
(0^^^27r)
a set of points in a plane possessing the
following two properties

(i)
number
8 (depending on
distance from
(ii)
entirel}''
any point of it, a positive

x and y) can be found such that every point whose
If (x, y) be the Cartesian coordinates of
{x,
y)
Any two
is less
than S belongs to the
set.
points of the set can be joined by a simple curve consisting
of points of the set.
Example.
The points
for
which .r-+3/-<l form a continuum.
point inside the unit circle such that

points inside the circle
may
OP=r<\, we may
take
For
if
be any
8=1-?-; and any two
be joined by a straight line lying wholly inside the
circle.
The following two theorems | will be assumed in this work simple cases
them appear ob\dous from geometrical intuitions and, generally, theorems
;
of
of a similar nature will be taken for granted, as formal proofs are usually
extremely long and

*
Due
difficult.
Mansion, Mathesis, (2) xix. (1899), pp. 129-131.

t For a proof that the sine and cosine are continuous functions, see the Appendix, A-41.
+ Formal proofs will be found in Watson's Complex Integration and Cauchy's Theorem.
to
(Cambridge Math. Tracts, No.
15.)
44
(I)
interior
'
'
'
If
27r,
be a point not on the curve,
and Ox increases by + 27r or by zero, as P describes

Q is an interior point or an exterior point. If the
QP
the curve, according as

is
').
be a point on the curve and
the angle between
Ill
simple closed curve divides the plane into two continua (the
and the exterior
'
(II)
increase
[CHAP.
said to describe the curve
is
continuum formed by the
'
counterclockwise.'
interior of a simple curve
is
sometimes called
an open two-dimensional region, or briefly an open region, and the curve is

such a continuum with its boundary is then called a
closed two-dimensional region, or briefly a closed region or domain.
called its boundary;
simple curve
simple curve with
sometimes called a closed one-dimensional region; a

is then called an open one-dimensional
is
its
end-points omitted
region.
Continuous functions of complex variables.
3"22.
Lety"(^) be a function of
^^
defined at
points of a closed region (one- or
all
two-dimensional) in the Argand diagram, and let
Then f{z)
we can
is
said to be continuous at
z^
number
find a corresponding positive
if
77
z^
be a point of the region.
given any positive number
e,
such that
\f{z)-f{z,)\<e,
whenever
3'3.
\z
Zx\<r)
and
2-
is
a point of the region.
Uniformity of convergence.
Series of variable tei-ms.
Consider the series
(1+
a;2
ieO'
This series converges absolutely

If 8n (^) be the
sum
(1
2*33) for
+ ^T
all real
= 0, and
x.
'
'^"^">=^^+"^-(TT^^'
lim Sn {x) = \^x'^;
and so
but Sn{0)
values of
of n terms, then
therefore lim Sn (0)
>
^/
{xrf 0)
= 0.
M-*-00
Consequently, although the series
continuous functions of
x,
the
sum
is
is
an absolutely convergent
a discontinuous function of
series of
x.
We
naturally enquire the reason of this rather remarkable phenomenon, which
was investigated in 1841-1848 by Stokes*, Seidelf and Weierstrassj, who

shewed that it cannot occur except in connexion with another phenomenon,
that of non-uniform convergence, which will
*
Cnmb. Phil. Trans,
viii. (1847), pp.
t Mi'mchener Abhandlungen,
X Ges. Math. U'erke,
i.
v.
533-583.
(1848), p. 381.
pp. 67, 75.
now be
explained.
[Collected Papers,
i.
pp. 236-313.]
3-22-3'3l] CONTINUOUS FUNCTIONS

Let the functions
u^ (z), Wg (z),
Argand diagram.
of the
AND UNIFORM CONVERGENCE
be defined at
...
45
points of a closed region
all
Let
= U, (z) +
Sn (z)
+ ...+ tin {Z).
M2 (2)
'
oc
The condition that the
2 Un {z)
series
n
value of z
that, given
is
number
e,
for all positive values of
^n+p
n should exist such that
{z)
Sn (z)\<
the value of
jo,
should converge for any particular
=l
7i
Let n have the smallest integer value
of course depending on
e.
which the condition
for
is satisfied.
This integer will in general depend on the particular value of z which has
been selected for consideration. We denote this dependence by writing
place of
n (z) in
INDEPENDENT OF
Now
?i.
Z,
it
mag happen
n{z)<N
for all values of z in the region
number
If this
we can
that
SUch that
find a
number N,
^'"^
"'
under consideration.
exists, the
series
is
converge uniformly
said to
throughout the region.

If no such
number
iV exists, the convergence
said to be non-uniform*.
is
Uniformity of convergence
is thus a property depending on a whole set of

whereas previously we have considered the convergence of a series
various particular values of z, the convergence for each value being con-
values of
for
z,
sidered without reference
We
to the
define the phrase
that there
is
'
3'31.
If
uniformity of convergence near a point z
common
to the circle \z
On
the condition
mean
z^\^h and
the region in
{z)
Un+i {z)
for uniformity of convergence'^.
+ iin+2 (z) +
...
necessary and sufficient condition that

in a region is that, given
+
S
Un+p
WO have Seen
(z),
that
the
Un (z) should converge uniformly
any positive number
N INDEPENDENT OF z (but depending on

I
for
to
series converges.
Rn,p
choose
'
a definite positive number 8 such that the series converges
uniformly in the domain
which the
other values.
e,
e)
it
should be possible to
such that
Rn, p(z)\<
ALL positive integral values of p.
* The reader who is unacquainted with the concept of uniformity of convergence will find it
made much clearer by consultinf; Bromwich, Iiijinite Series, Ch. vii, where an illuminating
account
of
Osgood's graphical investigation
t This section shews that
means
it
is
is
given.
indifferent
whether uniformity of convergence
of the partial remainder Rj^^p(z) or by iJ(2).
as fundamental.
Writers
differ in the
is
defined by
definition taken
[CHAP.
46
If the condition
is satisfied,
by
Ill
8n{z) tends to a limit, S{z), say for

e is independent of p,
2-22,
each value of ^ under consideration; and then, since
and
when n >
therefore,
iV,
S (2) - Sn (Z) =
Thus (writing ê
condition
is
R^rp{z)
<
2e,
Shew
1.
for if it
that, if
x be
real,
n-N (z),
The sum
;
of the first
when
a*
4=0, the
+ l)
sum
of the series
{'ix+l)^
n terms
follows as in 2-22 (I)
it
the condition for uniformity.
is
sum
the
{x + l)'^ {x
discontinuous at .r=0 and the series
is
E^v.
satisfied
is
which, by definition,
X
\
sum
n>N
e,
also sufficient
Example
is
i^^v, p {2)}
a necessary condition for uniformity of convergence

the
is independent of z
and
whenever
for e)
Sn (z) <
that \S{z)
that
lim
\S(z)-Sîz)\<2.
and so
is
is
is
--'^
'{{n-\)
x+l}{nx + \y
'"
non-uniformly convergent near
easily seen
be
to
nx+i
^'
so
= 0.
when x Q the
is 1.
'
The value
of
RnXx) = S{x)-Sn{x)
-1
is
xÔ;
if
so
:r=one-hundred-milliouth, the remainder after a million terms
when x
is
is
small,
say
or l-TTyTj ^^
1
100 +
the
first
And
million terms of the series do not contribute one per cent, of the sum.
make
general, to
<
nx +
e, it
is
in
necessary to take
7i>-(--l
X \e
n<N
for
exists, independent of x, such that
Corresponding to a given e, no number
for by taking x sufficiently small we can
values of x in any interval including x =
which is independent of x. There is therefore nonmake n greater than any number
all
uniform convergence near
Example
2.
,r
= 0.
Discuss the series
x{n.{nV)x'''-\]
nZx {\+n^x^}{\
in
which x
r.^,
The
7ith
term can be written
nx
55 l+n\v^
,
^W
[Note.
+ {7i-\-\fx^y
is real.
In this example the
sum
(n-\-\)
,^.^
{n
+ \y X''q
+(,i + 1)2^2
of the series
But (taking 6<i, and ^4=0), \Rn{x)\<e
?i+l>|{-i + Ve-2-4}|.T|-l
if
or
is
so
(x)
and
l+x''
not discontinuous at ^ = 0.]
e-i(?i
?t
+ l)
\x\<l-ir{n+\f x^
l<J{e-i-v'e = 3-4}|:p|-.
i.e.
if
3*32]
Now
not the case that the second inequality
it is
is satisfied for all
47
values of n greater
than a certain value and for all values of x and the first inequality gives a value of
n(x) which tends to infinity as x-^0 so that, corresponding to any interval containing the
point ^=0, there is no number N' independent of x. The series, therefore, is non-uniformly
;
convergent near
The reader
= 0.
:i,
n{x)
will observe that
discontinuous at
is
.r
= 0;
for n{x)-^'x> as
.r-^-0,
but n(0) = 0.
Connexion of discontinuity
3'32.
We shall now
shew that
luith
non-uniform convergence.
a semes of continuous functions of z
if
sum
convergent for all values of z in a given closed domain, the
is
uniformly
a continuous
is
function of z at all points of the domain.
For let the series be f{z) =

where Rn (z) is the remainder
Since the series
it,
{z)
Wo
(2^)
Un {z)+ ... Sn (z)
uniformly convergent, given any positive number
is
can find a corresponding integer n independent of
Sn
(2^),
n and
being thus
find a positive
z,
such that
we
e,
R^ (^) <
|
^ e
domain.
for all values of z within the
Now
+ Rn {z),
after n terms.
fixed,
number
rj
we
can, on account of the continuity of
such that
\Sn(z)-S{z')\<l,
whenever
We
z z' <
\
r}.
have then
l/(^)
- Sn{z')] + lRn(2)- Rn{^)

< Sn{z) - Sn(z') + Rniz) + Rn{z')
-/V) =
I
[Sni^)
<
'
which
is
1.
Shew that near x =

Ui {x)
z.
the series
+ U<i {x) + M3 (^0 +
where
and
real values of
shew
(a;)
i(vP)=a;,
In this example
that, given
small as
we
value of
n,
values of x
are concerned,
it is
6,
the condition for continuity at
Example
is
= ;p^"~* :p^"~^,
discontinuous and non-uniformly convergent.
convenient to take a slightly different form of the test
an arbitrarily small number
f,
it
is
we
shall
possible to choose values of x, as
please, depending on n in such a way that R^ [x) is not less than e for any
no matter how large. The reader will easily see that the existence of such
|
is
inconsistent with the condition for uniformity of convergence.

i_ ^
The value of Sîx) is

ing as x is positive, zero,
values of
or,
.r"^"-i
as n tends to infinity, S^ {x) tends to
or negative.
and has a discontinuity at
The
.t'
1, 0,
or
1,
accord-
series is therefore absolutely convergent for all
= 0.
48
[cHAP.
Ill
(2" - 1)
In this series R^ {x) = l-x^''-\ {x > 0) however great n may be, by taking* x = esmall.
The
not
arbitrarily
we can cause this remainder to take the value l-e'^ which is
series is therefore non-uniformly convergent near .r = 0.
;
Example
Shew that near z =
2.
il
is
non-uniformly convergent and
the series
{H-(l+2)-l}{l+(l
its
sum
is
The nth term can be written

l-(l+3)"
so the
sum
than -
of the
1, it is
first
n terms
seen that the
There
or positive.
is
1-(1+Z)"-^
H-(l+z)-i'
+ (H-3)
is
sum
^,
Thus, considering real values of
to infinity is
thus a discontinuity at
or
1, 0,
= 0.
when
1, according
as z
This discontinuity
fact that the series is non-uniformly convergent near

in the series
+ 2)"}
discontinuous.
2=0
for the
is
is
greater
negative, zero,
explained by the
remainder after n terms
z is positive is
-2
r+rr+2)"'
and, however great

2
than -^^
which
is
n may
by taking
be,
not arbitrarilv small.
= ~,
this can be
The
series is
made numerically
greater
therefore non-uniformlv con-
1-1-e'
vergent near
= 0.
The distinction between absolute and uyiiform convergence.
3'33.
domain does not necessitate

any points of the domain, nor conversely. Thus
The uniform convergence

its absolute
convergence at
the series
vz
(1
^r
+ ^")"
of a series in a
converges absolutely, but (near z
= 0)
not uniformly
while in the case of the series

=i
the series of moduli
z^
1
=
=i \n
is
'
is
5",
which
+n
divergent, so the series
Z'\^
only conditionally convergent; but for
is
all
real values of z, the terms of the series are alternately positive and negative
and numerically decreasing, so the sum of the series lies between the sum of
its first n terms and of its first (n -f 1) terms, and so the remainder after
n terms
finite
is
numerically
less
number (independent
values of z the remainder

series is
than the
term.
nth.
Thus we only need take a
of z) of terms in order to ensure that for

is
less
than any assigned number
e,
and
all
real
so the
uniformly convergent.
Absolutely convergent series behave like series with a finite
number
terms in that we can multiply them together and transpose their terms.
* This value of x satisfies the condition
< 5 whenever 2rt - 1 > log 5~i.
a;
of

and uniform convergence 49
3'33-3"341] CONTINUOUS functions
Uniformly convergent series behave like series with a finite number of

terms in that they are continuous if each term in the series is continuous
and
we
(as
334.
condition,
may
due
Weierstrass* for uniform convergence.
to
though not necessary, condition
sufficient,
of a series
If,
can then be integrated term by term.
shall see) the series
be enunciated as follows
the uniform convergence
for
within a domain, the moduli of the terms of a series
for all values of z
*S'
Ui (z)
i<2
(^)
W3 (^)
are respectively less than the corresponding terms in a convergent series

of positive terms
where
M^
independent of
is
then the series
of T, and therefore the

*Si,
is
uniformly convergent in
being convergent,
is
independent of
formly convergent
The
Example.
by
first
n terms
modulus of the remainder after the first n terms

and since the value of n
e
than an assigned positive number
less
is
thus found
follows ( 3-31) that the series
it
z,
234, the series
is
uni-
also converges absolutely.
.scries
1
cos Z + ^,
is
always possible to choose n so that the remainder after the
it is
of
z,
This follows from the fact that, the series
this region.
uniformly convergent for
all re<il
.,
CO.S-
,
CO.S-* Z-\- ...
2+57,
values of
r,
because the moduli of
its
terms are not
greater than the corresi)onding terms of the convergent series
whose terms are positive constants.

3 341
Uniformity of convergence of infinite products t.
convergent product
{1
+ ?< (z)}
is
said to converge uniformly in a
domain of values
M=l
of z
if,
given
e,
we can
tind
m independent of z such
n
{1
+ (z)} - n
that
{i+u^{z)} \<
for all positive integral values of p.
The only
is
condition for imiformity of convergence which will be used in this
that the product converges uniformly
if
m(s)
< J/ where J/
is
independent of
work
and
2 J/ convei-ges.
n
=l
Abhandlungen aus der Funktionenlehre, p. 70. The test given by this condition is usually
(e.g. by Osgood, Annals of Mathematics, iii. j[1889), p. 130) as the M-test.
The
t The definition is, effectively, that given by Osgood, Funktionentheorie, p. 462.
condition here given for uuiformity of convergence is also established in that work.
*
described
W. M. A.
50
To prove the
we observe
validity of the condition
that
[CHAP.
+ J/)
(l
converges
M=l
and
so
we can choose
Ill
( 2-7),
such that
Vl+p
VI
{l
M,,}-
n=l
{l
+ M}<;
=1
)(
and then we have

m+p
{l+un (.-)} -
{1
+t<
=1
(s)}
M=l
11=1
^n(i + 14)
3 "35.
independent of
?h is
Hardy's
tests
{i
~]
ni+2>
+u (z)} -
{i+i/}-i
L n = m+l
m=l
and the choice of
~]
\_n=m+i
n=l
I
I
ni+p
+ (z)}
{1
z.
for uniform convergence*.

p
The reader
real
and k
is
oo
domain.
in a given
and independent
of
then 2 a
converges uniformly.
finite
uniformly as w --
from
if,
will see,
2-31, that
{z)
f^
{)
and
and
2,
if
2 a
^k
{z)
/(2) ^/,t +
where a (2)
is
and fn{^)-^^
(s)
t=i
Also that
where k
is
if
and 2 a
converges uniformly, then 2 a,j {) w

M=l
[To prove the latter, observe that m can be found such that
independent of
Ii
verges uniformly.
m + l(2)+7 + 2(2),
7n + l(2),
are numerically less than e\k
(2)
m + 1 () + m + 2 ()+
and therefore
2
Example
1.
and
Shew
hi is
con-
(2)
+ m + p
(2)
( 2-301)
< e<,n+i (2)//(-<
(s)
n=jn+l
and the choice of
(s)
=l
e,
independent of 2.]
that, if
S>0, the
^
series
cos nQ
""
,
7t=l
'*
%^
sin
2
=l
'i
converge uniformly in the range

S
^ ^ 27r
(9
S.
Obtain the corresponding result for the series

^
(-)"cos?i<9
2i
by writing O + n for
Example
2.
'^
n=l
- ) sin n6
,
n=\
^.
If,
when aâ,'^6,
co,i (.1;)
<
^-j
and 2
<
(.r)
co
(.r)
|
<j('2,
where
)(=i
^1, k.>
are independent of
n and
.r,
and
if
2 a^
is
a convergent series independent of
x,
n=i
then 2 a,tC>(:*;) converges uniformly when a ^.r^

n=l
* Proc.
(Hardy.)
These results, which are generalisathough well known, do not appear to have been published
From their resemblance to the tests of Dirichlet and Abel for convergence,
London Math.
tions of Abel's theorem

before 1907.
/^.
Bromwich proposes
to call
Soc. (2) iv. (1907), pp. 247-265.
3-71, below),
them
Dirichlet's
and Abel's
tests respectively.
"
3"35, 3"4]
m+p
[For we can choose m, independent of
such that
.v,
2
a
n=m+l
<
e,
and then,
51
2-301
l)y
m+p
corollary,
we have
3 "4.
<
2
n=m+l
ttnOin (^)
(^'l
'^'2)
f ]
Discussion of a particular double senes.
Let
and
(1)1
be any constants whose ratio
w.^
is
not purely real
and
let
a be positive.
The
series
r-
which the summation extends over
in
positive and negative integral and zero values of in and n, is of great

importance in the theory of Elliptic Functions. At each of the points
all
= 2mcoi 2/10)2
uniform
m and
where
n,
8 is
such that
all
if ni
and
n are
integral values of
pending on B but not on
for all integral values of
not both zero, and
and
2,
if
^^
an arbitrary positive number.
Let S' denote a summation

which 7n = n = being omitted.
Now,
if
2m(o^
and
n,
the term for
2nco.2\ "^
then we can find a positive number
n,
+ 2w&)i + 2a)a)''
Consequently, by
>
for
C. de-
such that
1
(z
be shewn that the series
> 2, and the convergence is

2 + 2niQ)i + 2nco2 ^ 8 for all integral
other values of
for all
for those values of z
values of
It can
the series does not exist.
converges absolutely
3"34,
{2m(Oi
the given series
convergent in the domain considered
is
+ 2710)2)"
absolutely and uniformly*
if
2'
mcoi
I
no)2
converges.
To
discuss the convergence of the latter series, let

0),
where
a^,
the series
a.,,
/3i, /Sa
are real.
CTj
4- 1/3,
Since
0)0
=
is
co../ coi
Qfo
not
z'/So
real, a^/S.
ou/3i 4= 0.
Then
is
2'
{(a,m
This converges
+ aojiy- + {/3,m + /SoTi)^}^*
( 2"5 corollary) if the series
S=
^'
(m-+n2)^
converges
for the quotient of
The reader
corresponding terms
will easily define uniformity of
is
convergence of double series (see
3-5).
42
52
where
/z
= njm.
[cHAP.
Ill
This expression, qua function of a continuous real variable
minimum*
can be proved to have a positive
and so the quotient
(not zero) since
ofi/3o
jx,
=|=
K (independent
always greater than a positive number
is
ao/3i
of/x).
We
have therefore only
study the convergence of the series S.
to
PI
V'p,q ^
Let
S'
00
00
-^,
+ 71^)2 *
^4 S
S'
m = n=0 (m^
m = n, m
Separating Sp^g into the terms for which
>
n,
and
m < n,
re-
we have
spectively,
IS.
pi
= S
I-
p m-l
s S
'%^
But
r-
n=o
{ni"
IS ^ t
Therefore
+ n^)^"-
1
.
ml
<
-J +
n-1
+ s s
(m^)^"
+ i
^.
known to be convergent if a 1 > 1. So the series S

is convergent if a > 2.
The original series is therefore absolutely and uniformly convergent, when a > 2, for the specified range of values of z.
But these
last series are
Example.
Prove that the series

1
2
(mj^
in
values of mj, m2,

if
...
+ 7112^ +
all
. . .
+ my?y^
positive
and negative integral values and zero
except the set of simultaneous zero values,
wi^,
is
(Eisenstein, Journal fur Math, xxxv.)
fi>ir.
The concept of uniformity.
3'5.
There are processes other than that of summing a

of uniformity
series in
which the idea
of importance.
is
Let e be an arbitrary positive number; and let f{z, ^) be a function of

two variables z and ^, which, for each point z oi a, closed region, satisfies the
inequality \f{z, ^) < 6 when t, is given any one of a certain set of values
which will be denoted by ( ^z) the particular set of values of course depends
on the particular value of z under consideration. If a set (^)o can be found
|
such that every
member
function f{z, ^)
is
The reader
question
is
will
of the set (^)o
is
member
of all the sets
find
no
difficulty in
verifying this statement
the
minimum
given by
K^"' = h
(^j),
the
said to satisfy the inequality uniformly for all points z of
W+
a^^
+ ^.^ + ^r-
{(a,-/3,r-+(a2
+ ,3i)=l^
{(ai
+ /32)2+
(a^
-îP}^]-
value in
3"5, 3'6]
And
the region.
value of
if
a function
(f>
{z) possesses
some property,
53
for every positive
in virtue of the inequality \f{z,^):<e,(f) (z) is then said to possess
e,
the property uniformly.
In addition to the uniformity of convergence of series and products, we shall have

and also uniformity of continuity thus
to consider uniformity of convergence of integrals
a series
uniformly convergent when \R,^{z)\ <e, t( = ^0 assuming integer values in-
is
dependent of
only.
Further, a function f{z) is continuous in a closed region

whenever
Iôsitive number r/^ such that 1/(2 + ^2) fi^)
\
if,
given
e,
we can
find a
<
0<\C^\<r}^
and
+f
a point of the region.
is
The function
be uniformly continuous
will
pendent of z, such that
rjKr}^
and \f{z + C)~f{^)
we can
if
i
<
find a positive
number
inde-
whenever
0<UI<'7
and 2 + f is a point
moduli are less than
We
in
of the region, (in this case the set (f)o
is
the set of points whose
r)).
shall find later ( 3-61) that continuity involves uniformity of continuity; this is
marked contradistinction
to the fact that convergence does not involve uniformity
of convergence.
The modified Heine-Borel theorem.
36.
The
following theorem
of uniformity
Given
(i)
each point f
is
of great importance in connexion with properties
we give a proof
for
a one-dimensional closed region*.
a straight line CD and (ii) a latv by which, corresponding to

of CD, we can determine a closed interval I{P) of CD, P being
an interior^ point of I (P).
CD
Tlien the line

Ji, Jo,
can be divided into a
finite
number of
point) Pr, such that no point of Jr lies outside the interval

(by
means of
A
If
the given law)
luitli
an end
associated
that point Pr.
and
be said to satisfy condition {A).
will
CD satisfies condition {A
),
what
is
required
either or both of the intervals into which
bisect
I (Pr)
closed interval of the nature just described will be called a suitable
interval,
if
closed intervals
Jk, such that each interval Jr contains at least one point {not
...
it
is
CD
proved.
is
If not, bisect
divided
is not-
CD
suitable,
or them||.
A formal proof of the tlieorem for a two-dimensional region will be found in Watson's
Complex Integration and Cauchy^s Theorem (Camb. Math. Tracts, No. 15).
t Examples of such laws associating intervals with points will be found in 3'61, 5'13.
*
t Except
when P
is
at
or D,
when
it is
an end point.
This statement of the Heine-Borel theorem (which
is
sometimes called the Borel-Lebesgue
due to Baker, Proc. London Math. Soc. (2) i. (1904), p. 24. Hobson, The Theonj of
Functions of a Real Variable (1907), p. 87, points out that the theorem is practically given in
Goursat's proof of Cauchy's theorem {Trans. American Math. Soc. i. (1900), j). 14) ; the ordinary
form of the Heine-Borel theorem will be found in the treatise cited.
theorem)
II
is
suitable interval is not to be bisected
might not be suitable.
for
one of the parts into which
it
is
divided
54
[CHAP.
Ill
This process of bisecting intervals which are not suitable either will
terminate or
it will
If
not.
it
does terminate, the theorem
is
proved, for
CD
have been divided into suitable intervals.
will
Suppose that the process does not terminate
and
an
let
interval,
which
can be divided into suitable intervals by the process of bisection just described,
be said to satisfy condition (B).
CD
Then, by hypothesis,
does not satisfy condition {B)
one of the bisected portions of

one which does not
(if
CD
therefore at least
does not satisfy condition {B).
Take that
neither satisfies condition {B) take the left-hand one)
which does not satisfy condition {B).

This process of bisection and selection gives an unending sequence of intervals
and
bisect it
5o, Si, S2, ...
select that bisected part
such that
(i)
The length
(ii)
No
(iii)
The
point of
of s
is
s,i+i is
2-" Ci).
outside
Sn-
interval s^ does not satisfy condition (-4).
Let the distances of the end points of s from G be Xn, yn\ then
Therefore, by 2*2, x^ and yn have limits and, by the
Xn < a?+i < 2/,i+i ^ Hncondition (i) above, these limits are the same, say ^ let Q be the point whose
;
distance from
is
^.
But, by hypothesis, there
every point of CD, whose distance from
is
less
is
number
than
5q, is a
hq such that
point of the
Choose n so large that ^t^^CDk 8q then Q is an

internal point or end point of Sn and the distance of every point of Sn from
Q is less than Sq. And therefore the interval 5 satisfies condition (A), which
associated interval /(Q).
is
contrary to condition
(iii)
The hypothesis
above.
that the process of
bisecting intervals does not terminate therefore involves a contradiction

therefore the process does terminate and the theorem
is
proved.
In the two-dimensional form of the theorem* the interval CD is replaced by a closed

two-dimensional region, the interval I{P) by a circlet with centre P, and the interval
Jj.
by a square with sides

3'61.
From
parallel to the axes.
Uniformity of continuity.
the theorem just proved,
function f(x) of a real variable x

is
it
is
follows without difficulty that if a
continuous
when a^x^b, then f(x)
unifurmly continuous^ throughout the range a^^x ^b.
For
let e
be an arbitrary positive number
then, in virtue of the con-
tinuity of f{x), corresponding to any value of

number S^, depending on x, such that
1/(^0 -/(Ô
for all values of x'
*
The reader
such that
\x'
will see that a proof
x\ <
may
x,
we can
find a positive
<ê
Sx-
be constructed on similar lines by drawing a square
circumscribing the region and carrying out a process of dividing squares into four equal squares.
t
:J:
Or the portion of the circle which lies inside the region.

This result is due to Heine; see Journal fiir Math. lxxi.
p. 188.
(1870), p. 361,
and lxxiv.
(1872),
CONTINUOUS functions and uniform convergence
3'61, 3'62]
Then by
3"G
we can
that \fioc')
Let
f{xx) <
whenever x
-,
number of closed
number Xi such
divide the range (a, b) into a finite
intervals with the property that in each interval there

lies in
is
the interval in which
be the length of the smallest of these intervals
So
55
and
a;,
let
lies.
|'
be
< ô- Then

any two numbers in the closed range (a, 6) such that
if they lie in adjacent intervals
^, f lie in the same or in adjacent intervals
Then we can find numbers x^, Xo, one in
let ^^, be the common end point.
|
'
^'
each interval, such that
'
\f{^)-f{^.)\<\e,
,/(?o)-/(-^0 <^f>
/(r -/(ê) \<\^,
fit)
-fu^
<
6,
so that
i/(i)
-/(r) = {/(B
,
-fM] - [fit) -fM]

-{/(r)-/(^-.)}
<
If
same
in the
^, ^' lie
{/(ô)-/(^.)}i
6.
interval,
we can prove
similarly that
i/(^)-/(r)i<2În either case we have shewn that,
(iny
for
number ^
in the range,
we have
\f(^)-f{^+0 <e
^+
whenever
The uniformity
Corollary
(i).
is
in the
range and
of the continuity
From
is
Bo< ^<
where
Bq,
So is
independent of
^.
the two-dimensional form of the theorem of 3"6 we can prove

all jwints of a closed region of the
that a function of a complex variable, continuous at
Argand diagram,
Corollary
hounded
\f{x)
is
uniformly continuous throughout that region.
(ii).
in the range
<K
a^x^b
is
function f {x) which is continuous throughout the range
that is to say we can find a number k independent of x such that
for all points
[Let n be the
number
Let
we can
, î, ^2>
find
in the range.
of parts into which the range
In-ij ^ be their
numbers
l/()-/(-^'i)|<^,
Xi,
x-i, ...
end points
then
is
if
divided.
x be any point
of the rth interval
Xn such that
l/(.i-i)-/'(li)|<i^,
l/(î)-/('^2)|<ie,
l/(.^2)-/(^2)|<if,...
\f{^ryi)-f{x)\<h.
Therefore \f{a)-f{x) |< ire, and so
which
is
the required result, since the right-hand side
The corresponding theorem

3'62.
for functions of
is
independent of
complex variables
x."]
is left to
the reader.
real function, of a real variable, continuous in a closed interval,
attains its upper bound.
Let f{x) be a real continuous function of x when

section in which the i?-class consists of those
a^x^b.
Form a
numbers r such that r >f{x)
56
x
for all values of
in the range (a,
h),
[cHAP.
and the X-class of
Ill
other numbers.
all
This section defines a number a such that f{x)â., but, if h be any positive
number, values of x in the range exist such that f(x)>a 8. Then a is
upper bound oi f{x); and the theorem states that a number
called the
in the range can be found such thai f{x)
For, no matter
|/(^)
aj"^ >'8~^;
but since
for
which
the range;
in
not continuous at some point or points of the
(ii)) it is
f{x) a
of
therefore
therefore ( 3'61 cor.
range
may be, we can find values

{/(.^) - a| |~^ is not bounded
how small
x'
= a.
continuous at
is
all
points of the range,
ciprocal is continuous at all points of the range ( 3*2
=a
those points at which f(x) = a', therefore f{x)

range the theorem is therefore proved.
at
its re-
example) except
some point of the
(i).
The lower bound of a continuous function may be defined
manner; and a continuous function attains its lower bound.
Corollary
in a similar
Corollary
3'63.
If /(^) be a function of a complex variable continuous in
(ii).
a closed region,
f{z)
attains its upper bound.
real function, of
attains all values between its
Let
X,
31,
by
-v,
x) into
real variable, continuous in a closed interval,

loiuer bounds.
m be the upper and lower bounds off{x)

= M,f{x) = m;
362, such that/()
m< fjb< M.
(x,
a.
upper and
'A
let
then we can find numbers
//.
be any number such that
Given any positive number e, we can (by 3-61) divide the range
number, r, of closed intervals such that
finite
l/(.r,'-')-/(*2'^')|<6,
where
aî""*,
iCjC*
are any points of the rth interval;
the end points of the interval

for
which /(*'!<'') -
f^,f(x.J^^)
/x
then there
is
take
^-i**"',
x./''
be
to
at least one of the intervals
have opposite signs
and since
|[/(^,<'-')-/z}-{/(^,"-))-^j|<6,
it
follows that
/(aa""')
/ii\<
e.
Since
we can
find a
number
i'''*
to satisfy this inequality for all values
no matter how small, the lower bound of the function \f(x)-fA,\

zero since this is a continuous function of x, it follows from 3-62 cor.
that/(^) vanishes for some value of .^.
of
6,
is
(i)
yLt
The fluctuation of a function of a real variable*.
3'64.
Let/(^;) be a real bounded function, defined
a^Xi^X2^
Then /(a) -f(x,) +
\
fluctuation oi
f{x)
...
\f(x,) -f(x,)
when a^x^b.
Let
:^Xn^b.
...
+ l/C^J -f(b)
is
called the
in the range (a, b) for the set of subdivisions x^, X2,
...
Xn.
The terminology of tliis section is partly that of Hobson, The Theory of Functions of a Real
Variable (1907) and partly that of Young, The Theory of Sets of Points (190(5).
3'63-37l] CONTINUOUS FUNCTIONS AND UNIFORM CONVERGENCE

bound
If the fluctuation have an upper

...
iCi, j-o,
(a, h).
then f{x)
Xn,
If f{x) be
1.
n, for all
monotonic* in the range

'>i
is|/(a)-/(6)|.
(a, b), its total
fluctnation in the range
Example 2. A fimction with limited total fluctuation can be expressed

ence of two positive increasing monotonic functions.
may
[These function.s
Example
f{x0)
li
4.
as the differ-
be taken to be | {Fa'^+fix)}, h {Fa'-f{x)}.]
have limited total fluctuation in the range
If f{x)
3.
{a, b),
then the limits
[See 3*2 example.]
exist at all points in the interior of the range.
Example
choices of
said to have limited total fluctuation in the range
called the total fluctuation in the range.
Fa!' is
Example
is
independent of
FJ*,
57
total fluctuation in the range (a, b) so has
f{x\ g{x) have limited
f{x)g{x).
[For
\f{x')g{x')-f{x)g{x)\^\f{:>f)\.\g{x')-g{x)\
and so the total fluctuation of f{x) g

upper bounds of |/(^) |, \g (x) .]
{x)
+ \g{^)\-\f{^')-fi.^)l
cannot exceed g FJ'+f. G^K where

.
/,
g are the
Uniformity of convergence of power
-^3'7.
Let the power
.series
+ tti^;-!when z = Zo.
ao
converge absolutely
Then,
if
series.
^'o
+an2"+
a^^" ^ aZo"
...
'
CO
00
But
since
anZj"
converges,
'
it
follows,
by
that
3-34,
'
w=
S
71
uniformly with regard to the variable z
when
converges
ttn^"
\z\-^\Zq\.
Hence, by 3"32, a power series is a continuous function of the variable

throughout the closed region formed by the interior and boundary of any
with the
circle concentric
3'71.
circle of
convergence and of smaller radius
Abel's theoreni-'r on continuity
up
( 2"6).
of convergence.
to the circle
00
Let
be a po^ver
a,j^"
series,
whose radius of convergence
is
unity,
and
M=
cc
let it
be such that
>i
(OC
S
n=0
a^ converges
and
^ a; ^
let
then Abel's theorem
=
00
'*]
= S
an.
M=0
For, with the notation of 3-35, the function x^ satisfies the conditions
00
laid
on
when 0â;^l; consequently
u,^{x),
/(a:;)
= 2
On^" converges uni-
M=
*
The function
different values of
.r
monotonic
and x'
is
if
{f{x)-f{x')\\{x-x')
is
one-signed or zero for
all pairs
of
Abel's proof employs directlj' the

t Journal fiir Math. i. (1826), pp. 311-339, Theorem iv.
arguments by which the theorems of 3-32 and 3-35 are proved. In the case when S a
converges, the theorem is obvious from 3-7.
I
58
formly throughout the range
â;:^ 1
by
therefore,
it is
function of x throughout the range, and so
[CHAP.
Ill
a continuous
3"32,
lim f{x)=f{l), which
the
is
a;-*l-0
theorem
stated.
Abel's theorem* on multiplication of series.
3'72.
This
a modification of the theorem of
is
2"53 for
series.
Let
Then
the convergence of
Cn
= ttohn + i &_i +
"^
an,
M=0
and 2
bn
+ an ^0
c is
sufficient condition that
w=0
n=0
^00
00
00
For, let
,
B(x)= 2
^(a;)= 2 an^
n=0
C(a.')-=
6na^",
n=0
CnX''.
M=0
Then the series for A (x), B{x), C(x) are absolutely convergent when
< 1, ( 2-6) and consequently, by 2-53,
;
A(x)B(x)=C{x)
when
< ^<
therefore,
by
example
2-2
lim A{x)]{
a;^.l-0
2,
B(x)}
lim
a;--l-0
lim C(x)\
x^-l-O
provided that these three limits exist; but, by

00
00
a,
n=0
bn,
'Z
w=
3'73.
3-71,
these three limits are
30
and the theorem
Cn',
proved.
is
w=0
Power
which vanish identically.
series
If a convergent poiver seynes vanishes for all values of z such that \z\
where r^ > 0, then all the coefficients in the power series vanish.
For, if not, let
Then am +
a^ be the
'7/t+i^
which does not vanish.
first coefficient
+ m+2'^"+
...
^r-^,
vanishes for
values of z (zero excepted)
all
and converges absolutely when \z\^r<r^; hence,

have
= a^+i + a,n+2Z +
if s
.,
. .
we
00
Z,
and so we can findf a
=l
S^r
number
positive
(ljn-|-7i
I
>j
such that, whenever
^
|
and then
|
a^
* Journal filr
s
|
Math.
(^m+i ^
^ m |
i.
+ (^f"m+2^ _L
+
2
'
!
'
>
a.m\,
proof,
S,
1^2'
and
Theorem
by the use
'
^'"
'
I
so
a,
}
(1826), pp. ;:ili-339,
some text-books a more elaborate
vi.
of Cesaro's
This
sums
+
is
=f
when
Abel's original proof.
( 8-43), is
given.
<
S.
In
372,373] CONTINUOUS FUNCTIONS AND UNIFORM CONVERGENCE 59
We
Corollary
sums
some
have therefore arrived at a contradiction by supposing that
coefficient does not vanish.
"We
1.
may
Therefore
Corollary
We may
2.
when
the coefficients vanish.
'equate corresponding coefficients' in two power series whose
are equal throughout the region
equal only
all
1^
|<S, where 8>0.
also equate coefficients in
two power
series
which are proved
z is real.
REFERENCES.
T. J. 1'a.
Bromwich, Theory of Infinite
Series (1908), Ch. vii.
E. GouRSAT, Cours d'Analyse (Paris, 1910, 1911), Chs.
J.
DE LA Vallê Poussin (Louvain and
Introduction and Ch.
i,
xiv.
Paris, 1914), Cours d^ Analyse Infinitesimale,
viii.
G. H.
Hardy, A
VV. F.
Osgood, Lehrbiich der Funktionentheorie (Leipzig, 1912), Chs.
course of
Pure Mathematics
(1914), Ch. v.
ii, iii.
Watson, Complex Integration and Cauchy's Theorem (Camb. Math.
G. N.
No.
15), (1914),
Chs.
Tracts,
I, II.
1.
Shew that the
series
=l(l-2)(l-2^l)
is
equal to
,-,
{\-zy
when
<1
and
is
equal to
Is this fact connected with the theory of

2.
Shew
,,
-,
z{\-z)-
when
>
uniform convergence
1.
that the series
2sini + 4sinl + ... + 2"sin^^ + ...

all
values of
z {z
excepted), but does not converge uniformly
near 2=0.
3.
Un {x)=-2{n-\f .re-'"-'''^' + ^n^xe'''^'''',
If
shew that 2
?<
{x)
does not converge uniformly near x=0.
n=\
4.
Shew that the
by Abel's
series pr
+ r^...
is
convergent, but that
its
square (formed
rule)
T
is
j-
^2+1,73 +
2;
U'4'^vW
divergent.
5.
If the
convergent series
5= +-+... (r>0)
be multiplied by
itsel,
the teiTus of the product being arranged as in Abel's result, shew that the resulting series
(Cauchy and Cajori.)
diverges if r ^| but converges to the sum s^ \i r>\.
60
6.
If the
two conditionally convergent
^ -
where
and
s lie
and
between
Ill
series
and 2
W^
n=i
[CHAP.
n'
=i
be multiplied together, and the product an-anged as in
1,
Abel's result, shew that the necessary and sufficient condition for the convergence of the
resulting series is r + s
7.
Shew
that
be multiplied by
if
>1.
(Cajori.)
the series
itself
-3+ 5-t+
any number of times, the terms of the product being arranged as
in Abel's result, the resulting series converges.

8.
Shew
that the
qth.
power of the
! sin
is
convergent whenever
5-
(1
d+a^
- r)< 1,
(Cajori.)
series
sin 2^
...
+a,i sin n6 +
r being the greatest
...
number
satisfying the relation
for all values of n.

9.
Shew
that
if
is
or a multiple of 2n, and
not equal to
sequence such that u^-Ô steadily, then the series
Shew
also that, if the limit of
t< is
S
series is oscillatory if
is
rational,
but
2i/ cos {nd
not zero, but
?i
is still
that, if - is irrational,
is
if %0) Ui, u^,
monotonic, the
the
...he a
convergent.
sum
of the
sum may have any value
TT
TT
between certain bounds whose difference
+ a)
is
a cosec ^6, where
a=
lim u^-
CHAPTER
IV
THE THEORY OF RIEMANN INTEGRATION

4"1.
Tlie concept
The reader
is
of integration.
doubtless familiar with the idea of integration as the
operation inverse to that of differentiation
and he
is
equally well aware that
the integral (in this sense) of a given elementary function

expressible in terms of elementary functions.
is
not always
In order therefore to give
a definition of the integral of a function which shall be always available,
even though
function
is
it
is
not practicable to obtain a function of which the given
the differential coefficient,
we have
recourse to the result that the
integral* o{ f{x) between the limits a and h
is
the area bounded by the
a, x = b.
We proceed to
curve y =f(x), the axis of cc
frame a formal definition of integration with this idea as the starting-point.
and the ordinates x =
4"11.
Upper and lower
integrals'f.
Let f(x) be a bounded function of x in the range {a, b). Divide the
Let U, L be
... Xn-iia^x^-^x^^ ... ^Xn-i^b).
the bounds of /(a;) in the range (a, b), and let Ur, L^ be the bounds of f{x)
interval at the points Xi,Xo,
where
in the range (xr-i, Xr),
Consider the
Xq
= a,
Xn=b.
sums:|:
Sn
U,
Sn
L, {X^
(iCi
a)
+Uo{Xn-X,)+
- a) + Z/2
(^2
- î) +
U(b- a) ^ Sn ^
Then
s.,
...+ Un (b
- Xa-,),
+ Ln (b - Xn-i).
^L{b- a).
For a given n, Sn and s are bounded functions of x^, x^, ... ^n-i- Let
and upper bounds respectively be S^, Sn, so that Sn, s depend
only on n and on the form of f{x), and not on the particular way of dividing
their lower
the interval into n parts.

Defined as the (elementary) function whose differential coefficient is/(x).
The following procedure for establishing existence theorems concerning integrals is based
on that given by Goursat, Cours d' Analyse, i. Ch. iv. The concepts of upper and lower integrals
are due to Darboux, Ann. de I'Ecole norm. sup. (2) iv. (1875), p. 64.
+ The reader will find a figure of great assistance in following the argument of this section.
*
Sn and
s represent the
sums
of the areas of a
number of
x-a,
greater and less than the area bounded by y=f(x),
assumed to exist.
The bounds
of a function of
of a function of a single variable
n variables are defined in

( 3-62).
rectangles which are respectively
x-h
just the
and
?/
= 0,
if
this
area be
same manner as the bounds
[CHAP. IV
62
Let the lower and upper bounds of these functions of n be

Sn >
We
proceed to shew that
Let the intervals
new
S,
(a, x^),
{x-^, x.^,
Then
s.
^ S.
Sn
most equal to
s is at
S,
S'^s.
i.e.
be divided into smaller intervals by
...
points of subdivision, and let
a,y^,y^,
...
yk{=î), yk+i,
yu-i,
be the end points of the smaller intervals;
yi{=^2), yi+1,
yi-i,
U/,
let
L,.'
Vm-i, h
be the bounds of /(^)
in the interval {y,i, yr)-
m
Let
T,,^
= X
{yr- yr-i) UJ,
= t
t,n
Since Ui,
Now
U.2, ...
Uk do not exceed
subdivision
consider the
{y,.
r=l
r=\
JJi, it
of (a, b)
- yr-i) L,!.
follows without difficulty that
into intervals
by the points
and also the subdivision by a different set of points

Let S'n',s'n' be the sums for the second kind of subdivision which correspond to the sums *S^, Sn for the first kind of subdivision.
Take all the points x^, ... Xn-i] x^, ... x'n'_y as the points y^, y.^, ... y^.
x-y,
X2,
a?/,
x^,
Xn-\,
...
...
x'n'-i-
Then
and
T,
t,,
Sn,
T^n
tm
s'n'
S'n'
Hence every expression

expression of the type
[For
g^^
8n ^
\i
S<r],
s'n'
of the type Sn exceeds (or at least equals) every

and therefore S cannot be less than s.
and s S 27] we could find an Sn and an
s
s'n'<'n and so s'n'>Sn, which is impossible.]
S<s
The bound S
called the
is
upper integral off(x), and
is
such that
s'n'
written
s is called
the lower integral, and written
J a
S = s,
If
common
their
value
the limits* of integration a and
The
integral
f{x)dx.
rb
ra
We
define
|
Exam'ple
f{x)dx, when
1,
{/ (*')
+^
a<
(*-')}
dx
2.
By means
to
b,
J a
Example
f(x) taken between
b.
written
is
/ {x) dx.
called the integral of
is
f(x) dx
of
example
1,
f {x) dx+
mean
f(x)dx.
(x) dx.
J a
define the integral of a continuous complex
function of a real variable.

*
'
would be a more appropriate term but limits has the sanction of

'Termini' has been suggested by Lamb, Infinitesimal Calculus (1897), p. 207.
Extreme values
custom.
'
'
'
4*12, 4" 13]
Riemann's condition of integrahility*.
412.
function
said to be
is
integrable in the sense of
'
notation of 4"11) Sn and

infinity in such a
necessa7-y
way
common
number
the
Riemann
if
(with the
limit (called the
Riemann
of intervals
sufficient condition that
Sn
should
(xr-i, Xr) tends to infinity in
way
such a
x^) tends to
{x,._-^,
to zero.
a hounded function should he
numher of
zero luhen the
teiid to
'
them tends
that the length of the longest of
and
integrahle is that S^
have a
s,i
when
integral of the function)
The
63
intervals
that the length of the longest tends
to zero.
The
condition
^n Sn
that
if
as
lim (Sn
71
obviously necessary, for
is
>
30
And
Sn) = 0,
for,
since
s,i
have a
common
Sn^S'^s^ s,i,
it
limit
follows
then
lim
Note.
sufficient
it is
S^ and
if
*S'
continuous function f{x)
= S = s.
lim s
For, given e, we can find 8 such

x"\<8. Take all the intervals {Xg_i, x-g)
and so >S' s<e therefore <S' s-^-0 under the
'integrable.'
is
that \f(af) f{x")\<(l{b-a) whenever \x'
than
less
8,
and then Ug- Lg</{b a)
circumstances
Corollary.
If
<S'
and
the condition of integrahility.
.specified in
have the same limit
into intervals of the specified kind, the limits of
for
.S'
one mode of subdivision of (a, h)

s for any other such mode of
and of
subdivision are both 8.
Example
I.
Example
2.
discontinuities
\li
given
The product
of
two
function which
integi'able functions is
an integrable function.
continuous except at a
is
finite
number
of ordinary
integrable.
is
f{x) have an ordinary discontinuity at c, enclose c

we can find 8 so that f{x')f{x) < e when x' -x
f,
in
\
an interval of length
<8 and
x, x' are
S,
not in this
interval.
Then
*S'
X, x' lie in
When
-s,j^f (6-a-8i) + /'Si, where k
is
the greatest value of \f{x')f{x)\,
when
the interval.
Si-^0, y(--^|/(c
Example
413.
3.
discontinuities
is
+ 0)-/(c-0)
and hence lim
(.S'-s)=0.]
-*
function with limited total fluctuation and a finite

(See 3"64 example
integrable.
number
of ordinary
2.)
general theorem on integration.
Let /(a;) be integrable, and
let
be any positive number.
Then
it
is
possible to choose S so that

n
rh
{xp
- Xj,_i)f(x'p_i) -
p=\
provided that
*
sum
Biemann
Xy
(Ges.
occurring in
Xp_i-^h,
Math. Werke,
-i-lS
but
f{x)dx <
6,
J a
it is
p. 239)
then
general definition of integration (which
Xp_i^x'p_i%Xp.
bases his definition of an integral on the limit of the
difficult to
is of
prove the uniqueness of the limit.
Functions of Real Variables) has been given by Lebesgue, Annali di Mat.

pp. 231-359.
A more
very great importance in the modern theory of
See also his Lecons sur V integration (Paris, 1904).
(3)
vn. (1902),
64
To prove the theorem we observe

of the longest interval,
B,
that, given
S^
so small that
<
Sn
[CHAP. IV
we can choose the length
e,
e.
Sn> S
Also
(^^
^^_i)/(a?'^-i)
Sn,
p=l
Therefore
rb
2
9
{xp
- Xp_^)f(a;'p_,) -
=1
f{x) dx ^s.,
^ f-Jn
<
As an example*
'^n
J a
6.
of the evaluation of a definite integral directly
"^
of this section consider
from the theorem
X<1.
where
Jo (1-^2)2
Take S= - arc
sin
X and let ,?,= sin
^,
.^c
-^^s-i
(s
'^= 2
2 -^
s=i(i_y2^_j)i .=1
Then
+ i)
so that
8< S
|S cos (s+^)
sin
= sin
Xg
also let
<s8 <h tt),
s8, (0
^s+i-A's=2
8.
sin sS
sin
(5
1)S
^
^
\,.
cos(s-A)S
= 2/9 sin 2 S
= arc sin X. {sin |S/(JS)}.
By taking p
we can make
sufficiently large
dx
Jo
Xg-Xs-i
,s=l(l_.^'2^_j)4
(l_,,.2)i
arbitrarily small.
We can
also
make
arc sin
<
-j-|
arbitrarily small.
That
is,
given an arbitrary
number
f,
we can make
dx
arc sin
i~
<e
by taking p sufficiently large. But the expression now under consideration does not
depend on p and therefore it must be zero for if not we could take c to be less than it,
and we should have a contradiction.
;
rx
That
is
to say
Example
1.
^=arc
f^g
'-
Jo
Shew
that
I+COS- + COS
,.
Iim
2x
2.
If
fb
J a
where the limit
is
(n
f-...+cos^
l)x
= sm^
.^^n^^^^^^^n
71
f{x) has ordinary discontinuities at the points

(
fu^-S,
f{x)dx = \\m\\
\J
taken by making
+
81, S2,
* Netto, Zeitschriftfilr
...
a, +6,
, ei, tî
...
, then
+...+
J
aj, 02,
[b
fa.,-S.,
-^
*''
n-*-'x>
Example
sin X.
(i_^.'^)2
f{x)dx\,
J ax + <c
f*
Math, und Phys.
tend to
+0
xl. (1895).
independently.
THE THEORY OF RIEMANX INTEGRATION
4-14]
Example
If /(^) is integrable
3.
d_
Bxample
jjix) dx=
a continuous function of
-y-
a and
when
if,
Oj
^ a < 6 < 6i we write

,
b),
b,
-I
-f{a),
Prove by differentiation that,
4.
and
<^(,& + fi)->^K&)^^(^ô)^
that, i{f(x) is continuous at
da
dx
^ x ^ î
then
exists,
lim
Deduce
f{x)dx = {a,
and if/(6 + 0)
when
65
if
jjix) dx=f{b).
(f>
{x) is
./;
and
then
t,
at
fxi
fir
ft
^^<^{x)dx=\^^^^{x)'j^dt.
Example
If /' {x)
5.
and
<^'
continuous when
{x) are
a^x^b, shew
from example 3
that
r / (x)
Example

{o;)
dx +
{x)f{x)
4>'
dx=f {b)
If/(.r) is integrable in the range (a,
6.
<^ (b)
-f{a)
cf>
(a).
J'l
c)
and
^6 ^c, shew
that
J a
is
414.
J/c?i
The two
Fa^we Theorems.
following general theorems are frequently useful.
U and L
Let
(I)
range
f {x) dx
b.
be the upper and lower bounds of the integrable function /(.r) in the
(a, b).
Then from the
definition of an integral it is obvious that
U-f{x)]
dx,
'
fj(x)
^ L) dx
J'^
are not negative
and so
U{b-a)-^l f{x)dx^L{b-a).
This
li'
is
f{x)
known
is
Mean
as the First
contimious
we can
Value Theorem.
find a
number
U and L
any given value lying between
| .such that
( 3-63).
a-^^^b and such that/(^) has
Therefore we can find | such that
rf{x)dx = {b-a)f{$).
J a
If
F{x) has a continuous
differential coefficient
F'
(x) in
the range
(a, 6),
we
have, on
writing F' (x) for f(x),
F{b)-F{d) = {b-a)F'{^)
for
some value
of ^ such that
Example.
lif{x)
is
a^^^b.
continuous and
'
fix)
a
W. M. A.
cf>
(x)
(^
{x)'^0,
dx =f (I)
shew that ^ can be found such that

/
%
a.
(x) dx.
[CHAP. IV
66
Let
(11)
and
/(.?)
that a
number
^ exists
b) and let ^(a-) be a positive

Second Mean Value Theorem is
be integrable in the range
4>{.v)
decreasing function of
Then Bonnefs* form of

such that a ^ | ^ 6, and
{" f{x)ci>{x)dx^4>{a)
(o,
the
.r.
\y'
\^ f{x)dx.
J a
J a
sum
For, with the notation of 4'1-4'13, consider the
2 {Xs-x,_i)f{x,_^)(i>{x,_{).
;S'=
s=l
Writing {x^- x^y) f{x,_i) = a,_i, ^{Xs-i) = 4>s-\, o + i +
Each term
the summation
in
have
^'e
increased by writing b for 6g_i and decreased
is
be the greatest and least of
ftg_i, if b, b
writing b for
+ 08 = ^s,
---
...
6o, 6i,
6p_i
and
by
so b(j)(,^S ^b(Po-
Therefore
between the greatest and
lies
sums
least of the
2 (xg-Xg_i)f{Xg_i)
({){xq)
s=l
where
m = l,
2, 3,
...
But, given
p.
e,
we can
- x,_i)f{Xsî)
{x,
(t>
find 8 such that,
(.r,_i)
f^p
f{x)
S=l
b for
a,
<
e,
Cxra
{X,
-Xs-i)/ (X, _ i) - <^
/ (^) <^*- < f

[
(ô)
s=l
writing
dx
so,
(x)
(f)
J ^0
<^ (:ro)
and
when Xg-Xg_i<d,
^ a^
a'q, .^^p,
we
find that
f{x)(j>{x)dx
lies
between the upper and
J a'
lower bounds ott
Let
C and L
<^ (a)
'
f{.v)d.v2e, where
may
^j
be the upper and lower bounds of
(a)
</>
fh
U+ 2e ^
/'(.r) <^ (^0
dx'^L-2e
r^
Since
(j){a)
\f {x) dx qua function
all
values between a and
for a^^ jjositive values of
f (x)
cfi
(x)
dx
?>.
f{x) dx.
Then
take
therefore
^ L.
of ^j takes all values between its upper
and lower
bounds, there
is
some value
proves the Second

E.rarnple.
By
writing
value theorem, .shew that
such that a
^$^b
^, .say,
which
of |i for
is
it
equal to
f (x)
(f>
(x) dx.
This
Mean Value Theorem.

(^
{x) -(f>{b)\ in place of
if
<^
(x)
is
(.v)
(f)
in Bonnet's
a monotonic function, then a
form of the mean
number
| exists
and
f{x)(l>{x)dx = 4>{a)
f{x)dx + 4){b)
f{x)dx.
(Du Bois Reymond.)

* Journal de Math. xiv. (1849), p. 249.
The proof given
due to Holder, Gdtt. Nach. (1889), pp. 38-47.
By
413 example
6, since /(.r) is
bounded,
'
f(x)
is
a modified form of an iuvestigatiou
d.v is
a continuous function of fj.
4 '2]

Differentiation of integrals containing a parameter.
42.
The equation*
f{x,
ace
Riemann
For
of
But, for any given

tinuity of fa
number
Taking
<
a +
[H{x,
^-^-!
mean
first
second
the
\fa {x,
whenever a'
j
= ^\is
.'
a,
e,
fa.
value theorem, since /
integrand
8 independent of
<
-fix,
h)
a) -fa
(x, a)
I
<
e/(6
see that
a)
/
'
dxJ
6h
,
/a
<
and
8,
exists (since the con-
a),
(^, a)
[^
dx
<
.
I
fa
(x,
8,
m - /
^w(x, a)m dx
^
+ Oh)
x
e.
( 3"2),
+ h)-f(,^.a)^^
i'f(-.
lim
hÔ
J a
I"*
equal to
fadx.
a
Example
so whenever \h
<
is
is
where
6h),
fa
Therefore by the definition of a limit of a function
and
S.
we
exists
{x,
is
uniform]: with respect to the variable x) such that
is
x and
But, by the
function
true if f(x, a) possesses a
is
vCL
= lim f .Aj;._ + A)-/(^.) ^^

h
h-f-O
a
<fa
limit exists.
(I
a.
I- {'/(., )
eta
continuous
^ dx
a)dx=\
(I
x and
this
integral with respect to
hoth-f the variables
if
67
If a, b be not constant.s but functions of a with continuous differential
1.
shew that
coefficients,
^|'/(.Example
If /(.r, a)
2.
is
a)d.v=f{b,
a)^-/(a,
^ +/^ fj:.:
a continuous function of both variables,
f{x, a)dx
is
J a
continuous function of
*
t
5
a.
This formula was given by Leibniz, without specifying the restrictions laid on/(.r, a).
(p {x, y) is defined to be a continuous function of both variables if, given e, we can find
such that
4> {x',
y')
(/>
(x,
that
if
(x, y)
is
y)\< whenever
a Cartesian diagram,
it
is
x)'^
(y'
- y)'}-<S.
It
can be shewn by
uniformly continuous throughout the region (the proof
identical with that of 3-61).

of each variable,
{(x'
3-6
a continuous function of both variables at all points of a closed region in
it is
It
should be noticed that,
if ^(.r, y) is
not necessarily a continuous function of both
^[x,y)=.^^p!^,
is
almost
a continuous function
as an example take
0(0,0)^1;
and of y at (0, 0), but not of both x and y.

would have been sufficient to assume that /^ had a Riemann integral
and was a continuous function of a (the continuity being uniform with respect to x), instead
of assuming that / was a continuous function of both variables.
This is actually done by
Hobson, FuJictions of a Real Variable, p. 599.
this
is
a continuous function of x
X It
is
obvious that
it
52
[CHAP. IV
68
Double integrals and repeated integrals.
4"3.
'Letf{x, y) be a function which

variables x and y, when a ^x ^h,
continuous with regard to both of the
is
a^y -^ /3,
By
4'2
example 2
both
exist.
it is
clear that
/(^' y) dy\ dx,
These are called repeated
U J{x,
y) dx\
dy
integrals.
Also, as in 3"62, f(x, y), being a continuous function of both variables,
attains its upper
and lower bounds.
Consider the range of values of x and y to be the points inside and on a

rectangle in a Cartesian diagram divide it into nv rectangles b}- lines parallel
;
to
the axes.
Let
Z7i,^,
L,n,^
be the upper and lower bounds of f{x, y) in one of the
smaller rectangles whose area
Then
^ m,n
jj.
-"-m,!!-
^^
^n,v
=1
*S',^ >?,^_^,,
and, as in
and
Ayn,,j.',
71
w
j/t = 1
say,
is,
4"11,
-i
=1
;u.
let
-^j,n -^m,/u ^^ ^h,
I'-
=1
we can
are the lower and upper bounds of Sn,v,
find
numbers
^h,.-, s,,,^
respectively, the
V"
which
values of
depending only on the number of the rectangles and not on their

We then find the lower and upper bounds {S and s)
shapes; and ^n,^^s,^.
respectively of ^, , Sn,v qua functions of n and v and S^ v^ S^s^ s^^,, as in
Sn,v, Sn,v
411.
Also, from the uniformity of the continuity of f(x, y), given
e,
we can
find
B such that
'J m,iJ.
(for all values of
and
-^Hij/i
whenever the
/i)
'"^
f>
sides of all the small rectangles are
than the number h which depends only on the form of the function f{x, y)
and on e.
less
And then
>S', ^
Sa, ^
Ss<
and so
But S and
s are
of
S and
<
e (6
e{h
independent of
The common value

is
e,
a)
(/3
a)
(/3
a).
and so S
s is called
a),
s.
the double integral of f{x, y) and
written
f(x, y) (dxdy).
It is easy to
when
f{:c,
shew that the
reijeated integrals
and the double integral are
y) is a continuous function of both variables.
all
equal
4 '3, 4*4]
For
as
Then
2 Y,
"1
69
be the uppei- and lower bounds of
let Yj, A,
and
varies between x^-i
.V
(x,
a:,.
x,-i)
=1
>
But*
2
M=l
f {x,y)dy\dx^
\3
<'
a.
^ A,
=l
Ura,,.i^y.-y,.-i)^\,n^.\n^ 2 Z,,m
/n = l
Multiplying these last inequalities by
x^-Xm^x,
(y/x
(a;,,,
x,,,_^.
-^m-i)-
using the preceding inequalities and
summing, we get
2
(
and
so,
=1
pi
f{x,y)dy\dx^2
2 r.M^,.^^/ ]/
=l
J a \J
a.
m=
2 L,,Â,^^;
l
/x
=l
proceeding to the limit,
'S'^ [jy{x,y)dy^dx^s.
But
and
lfix,y){dxdy),
'^=s=l
so one of the repeated integrals
repeated integral
Similarly the other
a continuous function of both variables,
Infinite integrals.
If lini
question
equal to the double integral.
is
equal to the double integral.
If/(.v, y) be
Corollary.
44.
is
is
/(.r)c?j;j exists,
we denote
called
an
d'x
,.
(I
(^'
+ a')' "
By
integrating by parts,
by
it
f{x)dx; and the
limit in
infinite integral^;.
Examples.
^,.
^^
(3)
\\
2 (62 + a')
+ 2a2y'
shew that
'
2a^
t"e~^dt = n\.
(Euler.)
Similarly
we
define
/f{x)dx
-00
of a
*
is
is
-X
defined as
f(x)dx
mean
f(x)dx+l
to
-<x>'
lim
/
f(x)dx,
a-^ -o J a
In this
f{.v)dx.
if this
limit exists;
last definition
and
the choice
J a
a matter of indifference.
The upper bound
of f{x, y) in the rectangle
of /(.r, y) on that portion of the line
.r
=|
t This phrase, due to Hardy, Proc.
which
lies in
^ is not less than the upper bound

the rectangle.
--i,,
London Math.
analogy between an infinite integral and an infinite
Soc. xxxiv. (1902), p. 16, suggests the
series.
70
4"41.
of continuous
integrals
Infinite
[CHAP. IV
Conditions for
functions.
con-
vergence.
f{x)dx
necessary and sufficient condition for the convergence of
any positive number
that, corresponding to
exist such that
f{x)
The condition
it is satisfied
is
then,
dec
<
to prove that
it is sufficient,
f(so),
.
Hence, by
Sn tends to a
2*22,
Sn+p
Sn\<
limit,
S and
then,
-a
'
if
>a+
S-i^f{x)da;\^ S-\''^''f(a;)dx\+ t
J
suppose
ra+n
n^X -a and n be a positive integer and Sn =
we have
should
whenever
obviously necessary
ii
number
a positive
e,
is
n,
f(x)da;\
a+n
<26;
and
f{x) dx
lim
so
so that the condition
Uniformity of convergence of an
4"42.
The
= S;
f{x, a) dx
integral
J
in a given
number
e,
is sufficient.
infinite integral.
said to converge uniformly with regard to a
is
domain of values of
there exists a
number
corresponding to an arbitrary positive
if,
independent of a such that
/(^> a)dx\^<
J
for all values of a in the
The reader
4"41
will see
domain and
all
values of x
^ X.
without difficulty on comparing
that a necessary and sufficient condition that

.'
2'22 and 3'31 with
f {x,
a)
dx should
converge uniformly in a given domain is that, corresponding to any positive

independent of a such that
number e, there exists a number
f(x, a)dx
for all values of a in
4'43.
x'
<
the domain whenever x"
Tests for the convergence of
an
^ x ^ X.
infinite integral.
There are conditions for the convergence of an infinite integral analogous

to those given in Chapter II for the convergence of an infinite series.
The
following tests are of special importance.
4-41-4*43]
Absolutely convergent integrals.
(I)
may be shewn
It
71
f{x)dx
that
a
certainly converges if
\f(a;)
dx does
The comparison
and the former integral
The proof
said to be absolutely convergent.
Example.
so
is
then
similar to that of 2-32.
and
If \f{x)\-^g{x)
test.
is
{x)
dx
converges, then
f{x) dx converges absolutely.
[Note.
was observed by Dirichlet* that
It
f{x)dx that f{x)-a'0 as x-^cc
may
the reader
not necessary for the convergence of
it is
see this
/(^) =
(^n^x^n + l-{n + l)-'^),
f(x) = {n + iyin +
where n takes
For
all
l-x){x-{n+l) + {n+l)-^}
{n +
-(n + l)-^^x^n + l),
integral values.
f(x)dx
with ^ and
increa.'ses
without difficulty that
f{x)dx
and so f{x) does not tend
"00
f{x)dx=l{n+l)-^
But when
converges.
a-
= n + l -i
whence
(?n-l)-2,
it
follows
y'(.t>)
=^
to zero.]
The Maclaurin-Cauchyf
(II)
by considering the function
Tf/(A-)>0 and/(x')-*0
test.
steadily,
00
f{x) dx and S /() converge or diverge together.

J
=l
fm +
For
A^--
/(0
>
J
and SO
2
m=
The
first
fix) dx
^f{m + 1
f{x)dx'^ 2 /(m).
( 2-2)
when
),
n+1
fn+]
f{m)^l
l
inequality shews that,
f{x)dx converges
wi=2
the series converges, the increasing sequence
if
--oo through integral values,
and hence
it
follows
fx'
without difficulty that
f(.v)dx converges when .r'-*-x
also if the integral diverges,
so does the series.
The second shews that
if
converges so does the series
(III)
X<
Bertrand'sX
and \if{x)
the series diverges so does the integral, and
test.
1^
f{x)
f{x) dx converges when
Maclaurin
result.
example was/(.r) = sin

{Flit.vions,
Cauchy's result
X Journal de Math.
is
i.
.r'-
pp. 289, 290)
vii. (1842),
Journal
pp. 38, 39.
<
0.
fiir
test given in (I).
Math.
xvii. (1837), p. 60.
makes a verbal statement
given in his Oeuvrcs
X,
These results are particular cases of the comparison
the integral
f{x)dx converges when
0{x^~'^),
{x~'^ {loga;}^"^),
.
* Dirichlet's
if
( 2'2).
(2), vii. p.
269.
practically equivalent to this
,,
Chartiers test^ for integrals involving periodic functions.
(IV)
^x
steadily as
If /(^) -^
and
if
<^
then
^ (x) dx
f{x)
is
For
the upper bound of
if
.X
f{x)
6 (x) dx =\f {x')
a;'
which
x^<x>
X such that f{x)<e/2A
dx \he A, we can choose
(.r)
J a
and then by the second mean vahie theorem, when
> A';
/""
_/
as
convergent.
bounded
is
when
dx
{x)
'
[CHAP. IV
72
f
J
dx =f{x')
{x)
x'
.v"^.v'
(f){x)dx-
cf)
{x)
^ A', we
^ 2Af {x') < f
dx
J a
J a
have
the condition for convergence.
is
**
I
Example
I.
2.
dt' converges.
a;
sin {x^
- ax) dx
converges.
an
Tests for uniformity of convergence of
4-431.
De
(I)
*'
Example
Sill
Vallee Poussins
la
The reader
test^.
infinite integral f.
will easily see
by using
("00
the reasoning of
to a in a
f{x, a) dx converges uniformly with regard
3-34 that
domain of values
of a if \f{x, a)
<
where
fi{x),
fee
of a and
/jl
(x)
independent
r"
[For, choosing
dx converges.
fM{x) is
so that
rx"
<
f{x, a)dx
when x'^x'^X, we have
and the choice of
e,
fx(x)dx<e
is
inde-
J x'
pendent of
a.]
/oo
Example.
The method of change of variable.
(II)
This
converges uniformly in any interval {A, B) such that
x'^~'ê~''dx
may
Consider
be
ilkist rated
by an example.
dx where
'-
is real.
X7-ax
y^" sin
We have
3c'
^^^ dy converges we
'^^
/"^" sin
Y such
can find
~^7~ ^^
J ax'
that
dx
<6
whenever
a^'
|
/"*"
I
F;
if
sin
ax
I
I
a;'
-^
^ S > 0, we
dy <e when y"^y' ^
Y.
y
therefore get
dx \<f
-'
Journal de Math, xviii. (1853), pp. 201-212. It is remarkable that this test for conditionally
convergent integrals should have been given some years before formal definitions of absolutely
*
convergent integrals.
t
The
results of this section
and
of 4-44 are
Scientifique de Bruxelles, xvi. (1892), pp. 150-180.
X This
name
is
due to Osgood.
due to de
la Valine Poussin,
Ann. de
la
Soc.
4-431, 4-44]
when .'."^.//^X= Y/8; and
^S>
uniform when a
Example.
values of
independent of
is
So the convergence
a.
is
^ - 8 < 0.
sm{&â::^)d^> dxîs uniformly convergent
j/
this choice of
and 'when
73
any range of
in
real
(de la Vallee Poussin.)
a.
2- i sin zdz
does not exceed a constant inde!
and
lêndent uf a
.v
since
z-i sin
dz converges.]
(III)
method of integration by parts.
T/ie
If
and
/ (x,
a)
dx ^<p{x, a)+
a)-*-0 uniformly as x^-X3
if (^f.r,
and
to
then obviously
a,
f{x, a)
;( (.r,
|J
d-''
converges uniformly with regard
<t
sin (a^+l)^
a^^^^^|J
to
a.
a= 1
sin (o^- l)a.
in
any closed domain of
f{x, a) dx converge uniformly luhen a
Let
real values of
infinite integrals.
lies
a domain
in
S.
a continuous function of both variables
'Then, if f{x, a) is
f{x, a)dx
in S,
is
a continuous function* of
x^ a
and
f(x, a)dx
<e
ivJien
a.
For, given
whenever
are excluded.
Theorems concerning uniformly convergent
4"44.
^^
^^.^^|J
loth of the latter integrals converge uniformly

a from which the points
lies
a)o?jp
dx converges uniformly with regard
c_os.r^sin
Example.
(*>
The method of decomposition.
(IV)
(I)
e,
we can
find
X independent of
a,
such that
'
r*
I
^^X.
Also we can find 8 independent of x and
a,
such that
\f{x,a)-f(x,a')\<el{X-a)
whenever
That
is
<
a.'
B.
to say, given
we can
independent of
find 8
a,
such that
{f{x,a)f{x,a')]dx\
f{x,a)dx $
f{x,a.')dx\
.
e,
\Jx
f{x,
a')
dx +\
\Jx
fix, a) dx
<3e,
whenever
*
of a
a'
This result
if it
does not
is
'
<
due
and
to Stokee.
this is the condition for continuity.

His statement
converge infinitely slowly.'
is
that the integral
is
5^
[chap. IV
74
If f {x,
(II)
when
A <a<B,
a) satisfies the
by
(I),
and if
a.
lies in
then
f{x, (x)dx\doi=
For,
same conditions as in
fix, a)da[dx.
4-3,
f{x,
If
A
a) dx^r
da=
f(x, a) day dx.
[J a
Therefore
<
But, from
and
exist,
f{x, a)
da)-
dx
eda<e{B-A),
^.
4"41, this is the condition that
lim
should
\i
for all sufficiently large values of

2'1
da
f{x, a) dx[
and be equal
a)da\dx
fix,
to
f{x, a) dx\ da.
dx is true if the integral on the

rb{.v, a)dx=l
-r- \
da J a
J a va
right converges uniformly and the integrand is a continuous function of both variables,
and if the integral on the left is convergent.
when x'â and a lies in a domain
The equation
Corollary.
>S',
Let
be a point of S, and
and
by
so that,
a),
va
Then
S=f{x,
let
/(.r, a)
therefore, since
cf)
f{x, a) da = ^
a)
(.r,
day dx converges, that
a)dx converges,
{x,
is
so does
da
{4> {x,
{x,
a)-(}){x, A)}
dx converges,
A) dx.
J a
J a
Then
3,
A).
{x,
example
4-13
{x, a)
(f)
dx \=j-
{0
(*-',
d^
da
= T
daj
\l
n)
(-^j
-'1
)}
^^
f{x,a)daydx\
f(x.a)dx\da
[J a-^
'
'
=^fjix,a)dx=fy^dx,
which
is
the required result
the change of the order of the integrations has been justified
above, and the differentiation of
example
3.
with regard to a
is justified
by
4'44 (I)
and
4-13
4*5, 4-51]
-^
Imjiroper integrals.
4*5.
/(x) \>cc as X ^ a
If
I
Principal values.
then
0,
f{x) dx
f(x) dx
lim
i-^ +
written simply
may
exist,
abd
is
a+&
O J
an improper integral.
this limit is called
75
J a
00
If \f{x)
as
a;
> c,
a<
where
<b, then
rb
/e-5
lim
S^+O
may
/()
lim
/(-Ô c?
S'^+oJ
J a
this is also written
exist;
c?j:^
C+&'
f(x)dx, and
is
also called
an improper
J a
integral
8, S'
>
might however happen that neither of these
it
limits exists
when
independently, but
f{x)dx+i
lim ]/
f(x)dxy
exists; this is called 'Cauchy's principal value of
f{x)dx' and
is
written
J a
for brevity
f(x) dx.
J a
Results similar to those of
But
integrals.
convergence,
all
is
4-4-4-44
may be
required in practice
is
obtained for improper
(i)
the idea of absolute
the analogue of Bertrand's test for convergence,
(ii)
analogue of de
that
(iii)
Vallee Poussin's test for uniformity of convergence.
la
construction of these
is
reader, as
the
to
left
of integrals in which the integrand has an
is
also
the
The
the consideration
infinite limit at
more than one
point of the range of integration*.

Examples.
(1)
cos
.v <ilr is
an improper
integral.
- b
r ./"^ (1 -.rf "^
(2)
It does not
dx
1
is
dx
is
an improper integral
if
<X < 1, </i< 1.
converge for negative values of X and
/x.
the principal value of an improper mtegral
when
A'
0<a<l.
.
?r
The inversion of
4-51.
the order of integration of
a certain repeated
integral.
General conditions for the legitimacy of inverting the order of integration when the
integrand
The
is
not continuous are difficult to obtain.
following
is
a good example of the difficulties to be overcome in inverting the
order of integration in a repeated
improper
integral.
For a detailed discussion of improper integrals, the reader is referred either to Hobson's or
The connexion between infinite integrals and
of a Real Variable.
improper integrals is exhibited by Bromwich, Infinite Series, 164.
*
to Pierpont's Functions
[CHAP. IV
'^Q
and
Let f{x,y) he a continuous function of both variables,
<v<1
0<X^1, 0</x^l,
let
then
This integral, which was
first
employed by Dirichlet, is of iraijortance in the theory of

which we shall give is due to W. A. Hurwitz*.
integral equations; the investigation
Let
x''~î/'^~^ (1
-x-yy~'^f{x,y) = (li{x,y)
Let S be any positive
Draw
number
less
x = b, y = b,
the triangle whose sides ave
this triangle
(^ {x,
y)
is
and
M be the upper bound of \f{x,y)
let
|.
than \.
x+y = l-b\
at
points on and inside
all
continuous, and hence, by 4-3 corollary,
Now
r~''dx 11^"''
{X,
ct>
dx
y) dy^
where
/i
But
=l'~'^
</>
/i
<^ (^,
V) ^l^
L=
y) dy,
{x,
Jo
I
{|P"'
{x,
(f>
hdx+j'^
'*
Ldx,
f]"'^
y) dy.
J i-a;-6
< r
i/.r^ -
y^- 1
- .r - y)" - 1
(1
o^y
(l-^^-?/r-i<(l-^'-S)''-^.
since
Therefore, writing
T""^
}
x = i\
/i dr
^ J/S'^ ,x- p~^
Hence I
will
./ - 1
&
.^
- S)" - 1 c^^
Jo
î/r
The reader
we havet
-S)a'i,
^-1
(1
- bt^'-^C x,^-^
(1
-A-i)""'
dx
prove similarly that Ly-^0 as 8-0.
o?a'
4){x,y)dy\=
hm
lim
t/^
1-25
dx
fl-2S
C
J:
4>{x,y)dy\
fl-x-S
(}){x,
y)
dy
fi-y-S
lim
5H..0
Annals of Mathematics,
ix. (1908), p. 183.
a.-i^~^(l-.ri)''-^ dxi
= (A,
exists
^)
t The repeated integral exists, and
is,
if
\>0, j'>0 (4-5example2).
in fact, absolutely convergent; for
/"i
writing
2/
= (1-
a.-)
s;
and/
il/^^-^ (1 -
.r)'^
+ ''-i
cZx
^"
integral exists, its value which
ri'
.
s'^-1 (1 -s)''"^
?s
-^'
lim
is
5,
fl-e
I
eÔ J
may
fl-ZS
be written lim
SÔ J
exists.
And
since the
4 6]
by what has beeu already proved

integral
We
but,
by a precisely similar
7*7
work, the last
jjiece of
is
have consequently proved the theorem in question.
Corollary.
Writing ^ = a + {h-a)
x,
rj
= h {b-a)y, we
see that,
if
4> i$j
^)
i^
con-
tinuous,
//I {f^ (|-;^~' ib-vf' iv-$r''^ a,
V)
dv}
-J/V {fli^-af-Hb-rir-Hn-^)"-' c}>{^,

This
is
[Note.
rj)d^Y
called Dirichlet's formula.
What
now
and improper integrals Avere defined by Cauchy,

though the idea of infinite integrals seems to date from
Maclaurin (1742). The test for convergence was employed by Chartier (1853). Stokes
(1847) distinguished between 'essentially' (absolutely) and non-essentially convergent
are
called infinite
Leco?iS sur le calc. inf. 1823,
integrals though he did not give a formal definition.
Such a definition was given by

and 1858 (see his I'orlesiingen, 1904, p. 39). In the early part of the
nineteenth century improper integrals received more attention than infinite integrals,
probably because it was not fully realised that an infinite integral is really the Iwiit
Dirichlet in 1854
of an integi'al.]
Complex inteyration*
4"6.
Integration with regard to a real variable x
along a particular path (namely part of the

\^Qtf{z),
(=
7^ -f iQ),
may be
regarded as integration
the Argand diagram.
real axis) in
be a function of a complex variable
z,
which
is
continuous
along a simple curve î^ in the Argand diagram.
Let the equations of the curve be

x
Let
X (a)
Then
if-f x(t),
= x (t),
iy {a)
tj
= Zq,
(a
it)
x (b) + iy (b)
^t^b).
= Z.
y(t) have continuous difierential coefficients J Ave define
f{z)dz taken along the simple curve
AB
mean
to
dx
dy
{F + iQ)(^^ + i'^]dt.
dt
Mt
.
The 'length'
of the curve
It obviously
exists if
^
-7-
dt
AB
will
be defined as
-~ are continuous
\/ (--f) +(^)
we have thus reduced the
dt
discussion of a complex integral to the discussion of four real integrals,
!
*
'^^
\A-' />!- />
viz.
dt
treatment of complex integration based on a different set of ideas and not making
so many assumptions concerning the curve AB will be found in Watson's Complex Integration
and Cauchy's Theorem.

f This assumption will be
X Cp.
4-13 example 4.
made throughout
the subsequent work.
78
By
4'13 example
AB
when
integral
Examples,
dz= -
f{z)
definition
4, this
happens
to
"
[cHAP. IV
consistent with the definition of an
is
be part of the real
axis.
f{z) dz, the paths of integration being the
same (but
in
opposite directions) in each integral.
/:
/.w:f-s-4f--(4'-4)}*_^
The fundamental theorem of complex integration.
4*61.
From
the reader will easily deduce the following theorem
4"13,
Let a sequence of points be taken on a simple curve z^Z and let the first
n of them, rearranged in order of magnitude of their parameters, be called
;
^gC",
î<">,
and
that,
z<"'
(iTo"*'
when n > N,
whose parameter
arbitrarily small
An
4*62.
Let
= z
^+i'"'
= Z)
let their
the sequence be such that, given any
let
- ^^<"' <
^+i<"'
lies
B,
for
r=
- Zr^fiC/-^) -
by taking n
upper limit
h,
^j"*', ^J"',
we can
...,n; let
find
. . .
^<'",
N such
be any point
^,.""
then we can make
between
(^,+i'")
number
0, 1, 2,
<^<^'',
^r+i*"'
parameters be
\f{z) dz
sufficiently large.
to the
value of a complex integral.
M be the upper bound of the continuous function \f{z)
\.
jJVw,.j^:/(.)sj(| + 4y)],,
Then
^Ml,
where
is
the
'
length
'
of the curve
z^yZ.
That
is
4'7.
Integration of infinite series.
We
to say,
shall
f{z) dz cannot exceed Ml.
now shew
that
if
S {z) =
u^ {z)
vergent series of continuous functions of
some
z,
ii.^
{z)-\- ... is
for values of z
a uniformly con-
contained within
region, then the series
III
(where
all
{z)
dz
lu {z)
dz
...,
the integrals are taken along some path
vergent, and has for
sum
S (z) dz.
in the region) is con-
'
4 "6 1-4 7]
THE THEORY OF EIEMANN INTEGRATION
79
For, writing
>Sf
{Z)
= Ml {Z) + 2 (^) +
+ Ihi {Z) + Rn {Z),
we have
S {z) dz =1
Now
since the series
there corresponds a
Rn (z) <
f>
u^{z)dz
is
Un
...^-\
(z)
dz
R,, (z) dz.
uniformly convergent, to every positive number
number
r independent of
z,
for all values of z in the region considered.
Therefore
if
be the length of the path of integration, we have
modulus
the
Therefore
r
Um
(z)
dz can be made
4'62)
Rn{z) dz <el
C
the
of
difference
between
such that when n ^ r we have
{z)
and
dz
than any positive number, by giving n any
less
m = lJ c
This proves both that the series
sufficiently large value.
OT = 1
convergent, and that
sum
its
is
As
Corollary.
if
may
"
be shewn that*
"^
,
the series on the right converges uniformly and the series on the
Example
"
which x
The
left is
convergent.
Consider the series
1.
'ix{n{n +1)^11x^x^-1} cos x"^
?i
in
in 4-44 corollary, it
0?
is
S(z)dz.
Um{z)dz
J
{1+ 2 sin2 x^]
{1
+ {n + 1)2 sin-^ .r^}
is real.
Jith
term
is
2x (n + 1) cos x^
2.rw cos x^
1
and the sum of n terms
is
+ 71^
sin"-^
X-
2x(n + l) cos x^
2x cos x'^
'
therefore
1+Sin2^2~
Hence the
where ?>i = 1 2,
+ (n + 1 )2 sin- x^
+ (7i+l)2sin2a;2*
series is absolutely convergent for all real values of

. .
x except
^/(mrr)
but
r>
,_ 2x(n+l)cosx^
^^-'"''-l+(K + l)2sin2,^:2'
and
if
n be
any
integer,
by taking x = {n + l)~'^
therefore non-uniformly convergent near
* -^-^'
is
means
lira
"^^
this has the limit 2 as n^<X)
The
series is
x=0.
^^^^^ where h-^0
along a definite simple curve; this definition
modified slightly in 5-12 in the case when/(z)
is
an analytic function.
80
Now
sum
the
to
.V
to infinity of the series is -^
sum
tbe
'
n terms of the
first
/i-^
Example
.r^}
sum
the other hand, the
- arc tan
this tends to arc tan {sin x-]
Therefore the integral of the

the terms by
and so the integral from
x^
+ sm-'^,
to
,r
of
of the integrals from
series is
arc tan {sin
and as
On
of the series is arc tan {sin^r^}.
of the
[CHAP. IV
sum
{(
+ !) sin x'^},
- hir.
of the series difiers
from the sum of the
integi-als of
^tt.
Discuss, in a similar manner, the series
2.
2e.p{l-(e-l) + e" + U'2}
==
for real values of x.
Example
Discuss the series
3.
7(1
11-2
+ U3+...,
where
Ui
for real values of
The sum
of
is
z.
= ze-'\
Un=nze-'"'^ - (
1) se-l"-!)^^,
z.
of the first
n terms
is ?i2e~"^",
so the
Since the terms Un are real and ultimately
sum
to infinity is
same
of the
all
for all real values
sign, the
convergence
absolute.
In the series
/
the
is
sum
of
Ji.
terms
is
-g
(1
Uidz+
e""^^),
and
not equal to the integral from
The explanation
convergence near
and by taking
series is therefore
Example
4.
v.3dz
ti.2dz+ I
+ ...,
this tends to the limit | as n tends to infinity
to 2 of the
sum
this
of the series 2?<n-
of this discrejjancy is to be found in the non-uniformity of the
= 0, for
7i~'^
the remainder after
we can make
n terms
this equal to
e'^'"-,
non-uniformly convergent near
Compare the values
/\U=l2
in the series Ui
which
is
+ 112 + ...is -aze~^"^
not arbitrarily small; the
= 0.
of
tin[
dz
and
l)
{l
2 /
n=lj
Undz,
where
"
2n^z
{l+n'z^)\og{n +
2{n-\-l)^z
+ {n + -iyh'-}log{n + 2)'
(Trinity, 1903.)
REFERENCES.
RiEMANN, Ges. Math. Werke, pp. 239-241.
P. G. Lejeuxe-Dirichlet, Yorlesungen.
(Brunswick, 1904.)
F. G. Meyer, Bestimmte Integrale.
(Leipzig, 1871.)
E. GoDRSAT, Cours d^ Analyse (Paris, 1910, 1911), Chs. iv, xiv.
C. J. DE LA Vallê Poussix, Cours d' Analyse In fiyiite'stmale (Payis and Louvaiu,
G. F. B.
Ch.
E.
VI.
W. HoBSON,
T. J. I'a.
Functions of a Real Variable (1907), Ch.
Bromwich, Theory of Infinite
Series (1908),
v.
Appendix
ill.
1914),
81
Shew
1.
that the integrals

sin {x^)dx,
cos
dx,
(.r-)
x exp ( -
Jo
Jo
converge.
dx
and
(Dirichlet
Du
Bois Eeymoud.)
If a be real, the integral
2.
is
x^ sin- x)
Jo
cos (ax)
Jo
1+^^
(Stokes.)
a.
Discuss the uniformity of the convergence of
3.
x sin
ax) dx.
{x^
Jo
{x^-ax)dx= -f
3 /.rsin
- -l-ir-j) cos (x-^-ax)

/"/I
Shew
4.
of values of
that
hr,
hir)
(Stokes.)
r
x'^dx
"
Jo l+JT" |smjp|P
when
u, p, p
)/-
^-
are positive.
r
(Hardy, Messenger, xxxi. (1902),
Examine the convergence
6.
~\
^""-J
(de la Vallee Poussin.)
ex\)[-e'<^{x^-nx)]dx converges unifonuly in the range
x^
a.
Discu.ss the convergence

of
*
5.
1 fsin(xâx)
a\
,
,
:^j^"^^'^-"^)^'^'+3"j
,
JU^ +
Jo
p. 177.)
of the integrals
l-e'J
.r
X"
jo
'
(Math. Trip. 1914.)
Shew that
7.
-^ exists.
J "
Shew that
8.
.r
(sin x)^
x'^
-"e""^' sin
2.rc/.r
converges
if
>0, a >0.
(Math. Trip. 1908.)
J a
2 (c^-
If a series (7(2)=
9.
(?^
v=0
in
an interval, shew that g
+ i)sin
TT
{z)
Tri:,
(in
which
C(,
= 0),
converges uniformly
^
the derivative of the series/ (2)= 2
.
is
-.
sm
(2i/+l)
sin 2vTrz.
v=i V
irZ
(Lerch, Ann. de VEc. norm. sup. (3) xii. (1895), p. 351.)

10.
Shew that
r...
|Lî
tff2-^^n
converge when a>hi' and a~i + /3~i + ...+X~'

11.
<1
^d
r r
dx,dx,...dx^
(Math. Trip. 1904.)
respectively.
Iff{x,^) be a continuous function of both.r andy in the ranges (a^x^b), (a^?/^6)

it has ordinary discontinuities at points on a finite number of curves, with
except that
continuously turning tangents, each of which meets any line parallel to the coordinate axes
only a finite
fb
number
of times, then
/a, Sj
Sj, 2
fi, f2?
is
a continuous function of y.
[b
Ca.2-&2
1
dx
+...+
{fC^'j
y + h)f(x, y)]dx, where the numbers
are so chosen as to exclude the discontinuities ot f{x,
range of integration
W. M. A.
f{x, y)
Oj, 02,
...
being the discontinuities off{x,
y).]
y + h) from the
(Bocher.)
CHAPTER V
THE FUNDAMENTAL PROPERTIES OF ANALYTIC FUNCTIONS
TAYLOR'S, LAURENT'S AND LIOUVILLE'S THEOREMS
Property of the elementary functions.
51.
The reader
used
be already familiar with the term elementary function, as

on Algebra, Trigonometry, and the Differential Calculus)
will
(in text-books
depending on a variable z, the

symbols involved therein being those of elementary algebra together with
exponentials, logarithms and the trigonometrical functions examples of such
to denote certain analytical expressions*
expressions are
Z-,
arcsm^'-.
iogz,
e~,
Such combinations of the elementary functions of analysis have

a remarkable property, which will now be investigated.
Take
as
an example the function
Write
Then,
if
2^
in
common
e^.
e'=f{z).
be a fixed point and
_ e~ -e' _
z z
f{z')-f{z)
be any other point, we have
if z'
last series in
brackets
-1
z
e'^'-^'
z'
f,
and since the
(z
z)-
uniformly convergent
is
for all values of
/, it follows ( 3*7) that, as z'>z, the quotient
tends to the limit
e^,
This shews that
uniformly
tlie
for all
values of arg {z
z).
limit of
f{z:)-f{z)
z
is
in this case independent of the path by which the point z
coincidence witJt
It wall
be found that this property
elementary functions
*
The reader
( 3"1) in this
Thus
is
shared by
many
of the well-known
namely, that iif(z) be one of these functions and

uot the sense in which the term function
will observe that this is
work.
tends towards
z.
e.g.
.r
hj
and
are functions of z
(x + iy)
but are not elementary functions of the type under consideration.
is
h.
be
defined
in the sense of 3-1,

FUNDAMENTAL PROPERTIES OF ANALYTIC FUNCTIONS
5 1-5 "12]
any complex number, the limiting value
exists
and
The reader
then
will,
however, easily prove that,
IXJ-
lim'^-^^^
5'11.
of
independent of the mode in luhich h fends
is
is
83
ii
to zero.
f(z)=x iy, where

mode
not independent of the
in
which
= x-\- iy,
A>0.
Occasional failure of the jwoperty.
For each of the elementary functions, however, there

the function l/(^
a)
}i^Qh\z
does not exist
when
and z^ at the point
= a,
at the point z
= a.
z = Q.
z
be certain points
it
does not hold for
since
a+h
Similarly
will
Thus
z at which this property will cease to hold good.
za
does not hold for the functions log z
it
These exceptional points are called singular points or singulaHties of the

function f{z) under consideration
at other points f{z)
The property does not hold good

5'12.
any point
at
is
said to be analytic.
for the function \z\.
Cauchy's* definition of an analytic function of a complex variable.
The property considered

definition of
in
b\\
will
be taken as the basis of the
an analytic function, which may be stated as
follows.
Let a two-dimensional region in the ^-plane be given

function of z defined uniquely at
all
points of the region.
values of the variable z at two points, and u, u

of
u.
Then,
if,
at
any point z within the
area,
+ Bu
and
Let
let
z,
w be a
z-\-hz be
the corresponding values
-^ tends to a limit when 8x>0,
By*0, independently (where 8z = 8x + iBy), u is said to be a function of z^

which is monogenic or analytic j" at the point. If the function is analytic and
one-valued at
all
we say that the function
points of the region,
is
analytic
throughout the region\.
We
shall frequently use the
word function alone to denote an analytic

work will be almost exclusively
'
'
function, as the functions studied in this
analytic functions.
*
See the memoir cited in
5 "2.
The words regular and holomorphic are sometimes used. A distinctiou has been made
by Borel between monogenic and analytic functions in the case of functions with an infinite
number of singularities. See 5*51.
t
'
'
'
'
'
'
'
'
X See 5-2 cor. 2, footnote.
62
[CHAP. V
84
In the foregoing definition, the function u has been defined only within
As will be seen subsequently, however, the
a certain region in the ^-plane.
function u can generally be defined for other values of 2 not included in this
region
and
may have
elementary functions already discussed)

which the fundamental property no longer holds,
(as in the case of the
singularities, for
at certain points outside the limits of the region.
We
now
shall
more
state the definition of analytic functionality in a
arithmetical form.
Let f{z) be analytic at z, and let e be an arbitrary positive number;

then we can find numbers I and h, {h depending on e) such that
whenever \z' z\<h.

Vi f{z) is analytic at
consequently write 1 =
all
/(/) = f{z) +
Hence
where
v is a function of z
Example
(i)
2^.
ez
(j
and
cosec2
(ii)
= 0).
Example
of a region,
^^
obviously depends on z
- z) f
(V)
If z = x
2.
-\-
iy,
we
z such that
+ v{z'- z\
{z)
<
when
\z
z\<t>.
Find the points at which the following functions are not analytic
1.
1
(iv)
points
{z).
= /itt,
(2
w any
{{z-\)zf
(2
f {z) = u \-ii\
integer).
(iii)
-^
\
-
(3
= 2,3).
= 0,1).
where
u, v, x,
are real
and /
is
an analytic
shew that
function,
8m
^:r
ex
5*13.
An
"bv
= ^^
oy
cu
cv
:^=
;^.
oy
/T^-
(Kiemann.)
ex
application of the modified Heine-Borel theorem.
Let f(z) be analytic at all points of a continuum and on any point z of

the boundar}' of the continuum let numbers
(z), S (S depending on z) exist
;
such that
\f(z')-f(z)-{z'-z)Mz)<e,z'-z\
whenever
[We
c
\z
z <
write /i
approaches
iz)
and z
instead of /'
a point of the continuum or
{z)
is
its
boundary.
as the differential coefficient might not exist
from outside the boundary so that/j
The above inequality

as well as
is
(2) is
when
not necessarily a unique derivate.]
obviously satisfied for
all
points z of the continuum
boundary points.
Applying the two-dimensional form of the theorem of 3"6, we see that

the region formed by the continuum and its boundary can be divided into
a jinite number of parts (squares with sides parallel to the axes and their
5" 13, 5-2]
such squares) such that inside or on the boundary of
interiors, or portions of
any
there
jârt
one point
is
such that the inequality
z^
fiz') - f{z,) is satisfied
by
85
(/
- z,)f, {z,) .<e\z'-z,
points z inside or on the boundary of that part.
all
CaUCHY'S theorem* on the integral OF A FUNCTION ROUND A
5*2.
CONTOUR.
simple closed curve
a contour
ii
A, B,
in the plane of the variable z
is
often called
he points taken in order in the counter-clockwise sense
along the arc of the contour, and
if
f{z) be a one-valued continuousf

all points on the arc, then the
function of z (not necessarily analytic) at

integral
f{z)dz
or
f{z)dz
taken round the contour, starting from the point

is
and returning
called the integral of f{z) taken along the contour.
integral taken along the contour
other than
We
shall
is
unaltered
if
again,
some point
in the contour
taken as the starting-point.
due to Cauchy, which may be stated as

analytic at all points on^ and inside a
a function of
now prove a
If fiz)
follows.
is
to
Clearly the value of the
is
result
z,
contour G, then
f{z)dz =
For divide up the
interior of
axes in the manner of
.5"13
C by
0.
lines parallel to the real
then the interior of Cis divided into a number
of regions whose boundaries are squares Cj, C^,
whose boundaries D,, D,,

of
consider
...
and imaginary
Dy
Mr
Cm
and other regions
f(z)dz+ S
...
are portions of sides of squares and parts
n = lJ(,C)
f{z)dz,
n=lJ(.D)
each of the paths of integration being taken counter-clockwise
in
the
complete sum each side of each square appears twice as a path of integration,
and the integrals along it are taken in opposite directions and consequently
cancel ; the only parts of the
*
Memoire sur
here given
is
les
sum which
survive are the integrals oi f{z)
integrates definies prises entre des limites imarjinaires (1825).
that due to Goursat, Trans. American Math. Soc.
t It is sufficient for
f {z)
to be continuous
when
i.
The proof
(1900), p. 14.
variations of
along the arc only are
considered.
:J:
It is
not necessary that f(z) should be analytic on
(it
is sufficient
that
it
be continuous
on and inside C), but if /(^) is not analytic on C, the theorem is much harder to prove. This
proof merely assumes that /' [z) exists at all points on and inside C. Earlier proofs made more
extended assumptions; thus Cauchy's proof assumed the continuity of f'{z). Eiemann's
proof made an equivalent assumption.
Goursat's first proof assumed that f{z) was uniformly
differentiable throughout C.
See
4G, example.
[CHAP. V
86
taken along a number of arcs which together make up
taken in the same sense as in
f\z)dz
G,
each arc being
these integrals therefore just
make
hc)'
up
f{z)dz.
Now
consider
f{z)dz.
With the notation
- z,)f' (.-,) + {z- z,) v]
of 5-12,
J (Cn)
/
J (On)
f{z) dz
{f{z,)
[
-'
(^
{/(î)
But
f?^
dz +/'
(^01 f
= Mc = 0,
(z,)
zdz
by the examples of
let In
4*6, since
z,)
vdz.
0,
_l^n
the end points of C coincide.
An
be the side of Cn and
Then, using
(z-
zdz+l
l_
Now
dz
(C)
the area of C.
4-62,
f{z)dz\
Zi)vdz ^1
(z
^\
\(z
Zj)vdz\
I/,
!^ (C)
< e^ V2
\dz\=
eln \/2
4^,,
= 4e^ sj'2.
J Cn
In like manner
f{z)dz\î
J (Dn)
\{z-z^)vdz\
J (Dn)
^ 4>e (An +
Avhere
An
In Xi)\^'2,
the area of the complete square of which
is
Dn
is
part, In is the
side of this square and \n is the length of the part of G which lies inside this
Hence, if \ be the whole length of G, while I is the side of a square
square.
which encloses
all
the squares Gn and Dn,
f(z)dz
|J(0
k
I
S
(
<4eV2
< 4e V2
e is
arbitrarily small,
N
M
^ An+ S An' +
and
{I'
I,
f{z)dz
n = \\J{Dn)
(m = 1
Now
f{z)dz \+
n = \'J(Cn)
m=1
iV
l
tXn\
n=l
^X).
\ and
f{z)dz are independent of
1
I
It therefore follows from this inequality that the only value
can have
is
zero
and
this is
Cauchy's
result.
e.
iC)'
which
f(z) dz
'
5*2]
Corollary
function of
If there are
\.
Thus,
and Zq,BZ from 2o to Z, and if /(z) is a

and throughout the domain enclosed by
same value whether the path
f{z) dz has the
This follows from the fact that ZqAZBzq
ZqBZ.
(jv
taken round
jjaths z^ÂZ
analytic at all points on these curves
these two paths, then

Z(iAZ
two
87
(which
it
is
of integration is
a contour, and so the integral
is
the difference of the integrals along zqAZ and ZoBZ)
/(2) be an analytic function of
the value of
is
zero.
f{z)dz is to a certain extent

AB
independent of the choice of the arc AB, and depends only on the terminal points A and B.
It must be borne in mind that (his is only the case whenf{z) is an analytic function in the
if
z,
sense of 5*12.
Corollary
Suppose that two simple closed curves (7^ and Cj are given, such that Co
e.g. would be the case if Cq and Ci were confocal ellipses.
2.
completely encloses Cj, as
Suppose moreover that/ (2) is a function which is analytic* at all points on Cq and Cj
and throughout the ring-shaped region contained between Cq and C^
Then by drawing a
network of intersecting lines in this ring-shaped space, we can shew, exactly as in the
theorem just proved, that the integral
.
//(^ dz
integration is taken round the whole boundary of the nng-shaped space;
boundary consisting of two curves Co and C^, the one described in the counter-clochvise
is zero, ivhere the
this
direction aiid the other described in the qlockwise direction.
Corollary 3.
In general, if any connected region be given in the 3-plane, bounded by
any number of simple closed curves Co, Ci, Co, ..., and if /(z) be any function of z which
is analytic and one-valued everywhere in this region, then
I
is zero,
where the integral
is
f{z)dz
taken round the whole boundary of the region ; this boundary

..., each described in such a sense that the region is kept
consisting of the curves Co, Ci,
uhvays on the right or always on the
either
round
An
is
extension of Cauchy's theorem
at the
Morera, Moid, del
Amer. Math.
Soc.
f{z) dz = 0, to curves lying on a cone
II.
1st.
(1896), pp.
Lombardo, xxii.
(1889),
191,
p.
whose vertex
(3) xvi.
(1897),
and Osgood, Bull.

f{z)dz =
296-302, have shewn that the property
be taken as the property defining an analytic function, the other properties being
deducible from
Example.
2-plane.
(See p. 110, example 16.)
it.
ring-shaped region
Verify that the value of
of this region,
*
of a person walking in the sense in question
has been made by Ravut (N'ouv. Annales de Math.
origin,
pp. 365-7).
may
left
boundary.
the
is
bounded by the two

,
circles
where the integral
is
=1
and
=2
in the
taken round the boundary
is zero.
The phrase
'analytic throughout a region' implies one-valuedness ( 5-12); that is to say

has described a closed path surrounding Co, f{z) has returned to its initial value. A
function such as log z considered in the region 1 ^ ^ 2 will be said to be analytic at all
that after
'
points of the region.
[CHAP. V
88
For the boundary consists of the circumference

the circumference
direction, together with
Thus,
direction.
for points
if
on the
the second circumference we write
jo
described in the clockwise
the counter-clockwise
= e^,
circumference
we
then 6 and
are real, and the integral becomes
'2e<'t>,
jo
e'
= 1,
described in
cp
write
and
for points
on
2e'*
of an analytic function at a
2'he value
5'21.
iirst
]2|
= 2,
|s|
'point,
expressed as an integral
taken round a contour enclosing the point.
Let
Then,
he a contour within and on which /(^)
a function of
the point z
an analytic function of
z.
a be any point within the contour,
if
z
is
is
z,
which
e,
we can
is
analytic at
points within the contour
all
except
= a.
Now, given
find 3 such that
\fi^z)-f{a)-{z-a)f'{a)\^^\z-a\
whenever ^
|
7'
<
8,
a <
with the point a as centre describe a circle 7 of radius

7 lies wholly inside
r being so small that
Then
between 7 and
in the space
G f{z)\{z
a)
is
analytic,
and
so,
by
we have
5-2 corollary 2,
f{z)dz_[f{z)dz
c z
where
and
G and 7
denote integrals taken counter-clockwise along the curves
respectively.
But, since
2^
-a <
|
f(z)
Jy
where \v\<
e;
Now,
if
and
on
7,
we have
dz_f f{a)+iz- a)f (a) + v(z-a) ^^^

a
Z-a
z be on 7,
\lyZ
we may
"2t
a
jyZ
y^ ^
also,
j^,
Jo
dz=
by
4"62,
dz^^f () { dz^-\ vdz.

-a
Jy
Jy
-^
write
the radius of the circle
and
so
z
is
J y
m^^^y^^j
^
where r
Jy z
a = re^^,
7,
and consequently
ire^^dO
re'"
de
Jo
ire'^dd
vdz <
27rr.
0;
2771,
FUNDAMENTAL PEOPERTIES OF ANALYTIC FUNCTIONS
5"21, 5 -22]
c z
But the left-hand

is
arbitrary
that
is
/'
r
n.7:
f(z)d2
r
(
Thus
5: 1
side
2îf(â)
dz\^
J y
independent of
is
89
e,
2irre.
and
must be
so it
zero, since e
to say
2771 ] c
This remarkable result expresses the value of a function f{z\ (which is

(7) at any point a within a contour C, in terms of an
analytic on and inside
integral which depends only on the value
of/(^)
at points on the contour
itself.
Corollary.
If f{z) is
bounded by two curves
an analytic one-valued function of
C and
C is
and a
is
2 in
a ring-shaped region
a point in the region, then
2ni J c ^-(i
'
where
C",
"tti } c' Z
the outer of the ciu-ves and the integi-als are taken counter-clockwise.
The derivates of an analytic function f{z).
5"22.
The function/' (2'), which
is
the limit of
f{z
+ h)-f{z)
h
as h tends to zero,
f'{z)
is
called the derivate of
is itself
z,
/ {z).
We
shall
and consequently
now shew
that
itself possesses
derivate.
For
be a contour surrounding the point

is analytic, we have
if
a,
and situated entirely
within the region in which f{z)
(a)
f(a+h)-f(a)
= hm --^
hô
,,^0 2TTih [J c
= Km
-I-
h^Q
liri
while
C,
f{z)
we can take
is
(j
z-a
/(^) dz
\
Itti J c {z
Now, on
z-a-h
c{z-a){z- a~h)
- af
h^
27ri J c {z
- ay {z-a- h)
continuous and therefore bounded, and so

less
than the upper bound of
^ U a
|.
is (z
a)~^
' '
'
90
Therefore
Then,
if
bounded
is
a)-{z a h)
(z
[chap. V
upper bound be K.
let its
be the length of C,
lim
I
f{z)dz
h^Q lirij c {z
<lim
-af{z-a- h)
f (a) = Ziri
J
and consequently
^^
c {z
!A|(27r)-^ 7^:^
a)-
0,
a formula which expresses the value of the derivate of a function at a point

as
an integral taken along a contour enclosing the
From
this
formula we have,
f{z)dz
27ri
the points a and a
if
f(a + h)-f'(a) ^ J_
dz
...f(z)
and
it is
easily seen that
ay
c \Z
+h
(z
a hy
(z
a hy {z ay
{z
ay
+hAh,
a bounded function of z
J./^ is
are inside C,
f{^)dz.
9:Z;f
iTTl J c
ZTTi J
point.
Therefore, as h tends to zero, A~' {/'(-
+ A) /'
when \h\<^\z
(a)}
a\.
tends to a limit,
namely
J_
f(z)dz
ay
27ri J c {z
Since /' (a) has a unique differential coefficient,

of a; its derivate, which
denoted
by/" (a), and
Similarly
it
is
is
it is
an analytic function
represented by the expression just given,
is
called the second derivate of /(a).
can be shewn that /"(a)
is
a,
possessing
a derivate equal to
2^
27ri
this is
denoted by
f" (a),
and
is
f(z)dz_
Jc{z-ay'
called the third derivate of /(a).
And
general an nth derivate/*"' (a) of /(ct) exists, expressible by the integral
n^r
r.
P^
and having
itself
f(z)dz
2'rTiJc{z-aT+^'
a derivate of the form

(n
+ 1)!
27ri
f{z)dz
Jciz-
a)"+^
the reader will see that this can be proved by induction without difficulty.
'^
in
5 '23, 5 3]
function which possesses a
variable z at
all
derivate with respect to the complex
first
points of a closed two-dimensional region in the ^-plane
therefore possesses derivates of
orders at
all
Caiichys inequality for f
5'23.
^"^^
points inside the region.
all
(a).
Let f(z) be analytic on and inside a

be the upper bound oif{z) on the
circle
Let
91
with centre a and radius
Then, by
circle.
?\
4*62,
!/""()i;4/c;^'i''^'
M.n\
Example.
l{f{z)
is
analytic,
V2 log 1/(2)
unle.ss/(2)
or/'
{z)
V- = ;^^,
= x-\-iij and
1
+ ^-2,
.shew that
= 0; andy2|/(e)|>0
= Q.
(Trinity, 1910.)
Analytic functions represented by uniformly convergent
5"3.
series.
X)
Let X fn
(z)
be a series such that
(i)
converges uniformly along a
it
=o
contour C,
(ii)
(z) is analytic
throughout
C and
its interior.
00
Then 2 fni^) converges, and the sum*
of the
series
an analytic
is
n=
function throughout
C and
its interior.
00
For
a be any point inside C; on
let
Then
^-.
27ri J
^dz
a = ^j
27ri
c z
C, let
^ fn(^)\
c Im=o
= ^(z).
fn{z)
\2mlc 2-a
.o
-a
'
00
by*
But
47.
this last series,
by
5-21, is
fiid)',
the series under
n=
consideration therefore converges at
(as well as
ha,s
a unique differential coefficient at
But
if
points inside C; let
all
Then the function
on C) be called ^(z).
a and a + h be inside
A
5'22,
sum
inside
analytic
if it
points inside G.
all
G,
^(a + h)-^(a) _ J^
and hence, as in
its
is
^ (^) (^^
27rt J
lim W^{a-\-h)
c {z
a){z
^{a)]
a h)'
A~^] exists
and
is
equal to
7t-*0
of
S
n=0
Since
{z)l[z
-a
|~i is
bounded when a
a) follows
is fixed
from that of 2 /
n=0
[z).
and
z is
on C, the uniformity
of the convergence
'
92
^^^ dz
c{z
27rt J
and therefore

(z)
00
we
Further, by
analytic inside G.
is
a)
transforming the last integral in the same way as

one,
[CHAP. V
see that
<!>'
= S
(a)
we transformed
the
first
ao
2 fn (a) may
/' (a), so that
n=0
be
'
differentiated
n=0
term by term.'
If a series of analytic functions converges only at points of a curve
which
is
not closed
nothing can be inferred as to the convergence of the derived series*.

'^
Thus 2
COS
)"
H=l
series
H
7hJ7
2
converges uniformly for real values of x
But the derived
( 3*34).
'i
sin
)""*
îj^
converges non-uniformly near
=l
A'
= (2m+1)
(m any
tt,
integer)
and
'ii
the derived series of this,
viz.
- )"~^
cos
n.v,
does not converge at
all.
(1=1
By
Corollary.
sum
3-7, the
of a
power
series is analytic inside its circle of con-
vergence.
531.
Analytic functions represented hy integrals.
Let f{t,
z) satisfy
of integration (a, h)
^f
(i)
and
are continuous functions of
The continuity
(in)
when
any point of a region
is
lies
on a certain path
>S'
t.
dz
/ is
(ii)
the following conditions
and z
the variable
z.
df
of ~- qua function of z
is
uniform with respect to
t.
rb
Then
f(t, z)dt
!*
unique derivate
is
z)
f* dt\t,
-^^
z.
Analytic functions represented hy
From
4*44 (II) corollary,

all
it
infinite integrals.
follows that
points of a region
an analytic function of z when
(iii)
is
has the
4*2, it
J a
is
by
dt.
5*32.
function of z at
For,
is
>S'
if (i)
f (t,
z) dt
an analytic
is
the integral converges,
(ii)
on the path of integration and z
a continuous function of both variables, (iv)
dz
f{t, z)
is
^^
on
S,
dt
OZ
converges uniformly throughout 8.
For
if
f{t, z) dt has the unique derivate
these conditions are satisfied

J
dz
J a
*
This might have been anticipated as the main theorem of this section deals with uniformity
of convergence over a two-dimensional region.
5 -3 1-5 -4] Taylor's,
Laurent's and liouville's theorems
case of very great importance
afforded by the integral
is
93
e~'^/(0
dt,
Jo
where /(t)
continuous and \f{t)\
is
<
Ke'^ where K, r are independent of
t;
obvious from the conditions stated that the integral is an analytic

function of z when
{z) ^ r, > r.
[Condition (iv) is satisfied, by 4-431 (I),
it
is
since
fe"""''*' rf^
converges.]
Jo
Taylor's Theorem*.
5-4.
Consider a function f{z), which is analytic in the neighbourhood of a

= a. Let (7 be a circle with a as centre in the ^-plane, which does
not have any singular point of the function f{z) on or inside it so that f{z)
point z
is
analytic at
Then, by
the circle C.
Itti J c
5*21,
by
on
is
3*61 cor. (ii), will
be any point inside
z-a-h
'
+h
we have
'
^-rriJc^^^^^^Xz-aîz-af^
But when
Let z = a
points on and inside C.
all
'
{z
ay-^'
^ {z - a)''^' {z - a - h)]
f{z)
modulus of ^ j is continuous, and
z a h
not exceed some finite number M.
C, the
so,
Therefore, by 4-62,
1
f(z)dz.h''+^
27riJc{z-ay'+'{z-a-h)
where
is
"^
the radius of the circle C, so that 'IttR
of integration in the last integral, and
cumference of
[rJ
27r
is
R = \z a\
'
the length of the path

for points z
on the
cir-
C.
The right-hand
side of the last inequality tends to zero as
??
>
oo
We
have therefore
/(a + /0=/(a) + /^/'(a)-h|^,/"(a)+...-f^|/-'(a) +
...,
which we can write
f(z)=f{a) +
This result
Cauchy.
*
is
(z
- a)/'
known
(a)
The formal expansion was
...
+^ri^/<)
(a)
....
and the proof given is due to

radius of convergence of a poiuer series is always
as Taylors
It follows that the
Incrementonim.
+ ^~^~ff"{a) +
first
Theorem
published by Dr Brook Taylor (1715) in his Methodus

94
from
at least so large as only just to exclude
[CHAP. V
the interior of the circle
of con-
vergence the nearest singularity of the function represented by the series.
by
the radius of convergence
5*3 corollary, it follows that
Hence the radius
than the number just specified.
of convergence
And
not larger
is
just such
is
as to exclude from the interior of the circle that singularity of the function
which
nearest to
is
At
a.
we may
this stage
introduce some terms which will be frequently
used.
= 0,
If f{a)
the function f(z)
If at such a point
is
(a) is different
said to have a zero at the point z
from
zero, the zero of
f(a)
= a.
said to be
is
however,/' {a),f"{a), .../"*"'' (a) are all zero, so that the Taylor's
expansion of f{z) at z = a begins with a term in (z a)", then the function
simple;
f{z)
is
if,
said to have a zero of the nth. order at the point z
= a.
Find the function / (s), which is analytic throughout the circle C and
1.
whose centre is at the origin and whose radius is unity, and has the value
Example
interior,
a cos 6
a^-2acos6 + l
(where
a> 1
and 6
is
its
sin 6
a^-2acos0 + l
the vectorial angle) at points on the circumference of
C.
[We have
/()(0)
'
'
-^-^dz
= ^.f f{z)
2mjc
n
z"^^
/"St
-7- ^ ^~;r
a- -2a cos ^ + 1
e-^'O.idd.
27nJo
_n\ n^ e~"i9d6 _
,
'
(puttuig
^^
= e*e)
'
n
f
dz
f d"" 1
a-e<9 ~2iriJcz''(a-z)~~\_dsâ-zJ
~2ffjo
"1
Therefore by Maclaurin's Theorem*,
)l=0
or/(2) = (a-2)~' for
all
points within the
"
circle.
This example raises the interesting question, Will it still be convenient to define f{z)
as (a-2)~^ at points outside the circle ? This will be discussed in 5-51.]
Example
2.
Prove that the arithmetic mean of
the circumference of the circle \z\
= l,
all
values of 2"" 2
a, if Sc?^" is analytic
is
cv,
for points z
throughout the
circle
on
and
its interior.
[Let 2 v2''=/(2), so that
a^,
/(") (0)
;-
Then, writing
277
2"
jo
2ni j c 2"*^
z'i
The re8ult/{2) =/(0) +2/'
(0)
+ -/"
usually called Maclaurin's Theorem;

Maclaurin (1742) in his Fluxions.
is
= e'^,
and
calling
the circle
"
K=0
(0)
it
...,
n\
"""-'
obtained By putting a =
was discovered by
in Taylor's
Stirling (1717)
Theorem,
and published by
Taylor's, Laurent's
5-41]
Example
and
'L&tf{z) = z^; then /(s + A)
3.
ltouville's theorems
an analytic function of A when |Aj<|2
is
'y (fp
for all values of r
when A
I
<
and so
This
|,
{z
95
+ hy=z^ + rz''-'^h-{--^-
\\
z'"'^ k^
+ ...,
this series converging
the binomial theorem.
is
Example
Prove that
4.
if
is
a positive constant, and
- 2zh-\-h?) ~
(1
in
expanded
in
the form
\
(where P(2)
so long as z
is
+ hP^{z) + h''P.2{z) + h^P^{z) +
(A),
easily seen to be a polynomial of degree n in
is in
then this series converges
z),
the interior of an ellipse whose foci are the points
whose semi-major axis

Let the series be
regarded as a function of
first
It is a
A.
of this circle is the point A

(1
= 0,
22A
so the singularities of (1
and
its
which
2zh-\-h'^)~^
But
= {A - 2 + (22 - 1
A2
-1-
)5>. {/i
_2_
ellipse in the 2-plane passing
(^2
is less
which gives
(22
1 )i
1)2,
i.e.
so long as A
and having
its foci at
+L
it.
sin
6),
= a + (a2_i)4.
series (A) converges so long as
and a-{a^-
sin 9,
1 )^
= {a + (a2 _ i)^} (cos ^ +
i2(22-l)i
Thus the
1 )ij,^
|.
through the point
= a cos ^ -f i (a^ -
so
a-|-(a2- 1)^
and 6 the eccentric angle of 2 on

2
centre
than both
|2-(22-l)^| and |2-f(22_l)^
Then
The
5'7).]
Draw an
axis,
and
series in A,
-plane.
h=z {z'^-\)'^ and h=z + {z^ \)^
22A-|-A-)~2 are the points
series (A) converges so long as
semi-major
/;
nearest to A = 0.
Thus the
its
and 2= 1, and
circumference will be such as to pass through that
is
[These singularities are branch points (see
Let a be
=\
power
therefore converges so long as the point A lies within a circle in the
singularity of
^ (A + A"').
is
than the smaller of the numbers
is less
is less
than a-(a2_i)5.
But A = a (a2-
1)^
when a = |(A-f-A~i).
Therefore the series (A) converges so long as
and whose semi-major
1,
axis
is
Forms of the remainder
5'41.
2 is
within an ellipse whose foci are
Let f{x) be a real function of a real variable
It
hm
li.
7! (1
and
let it
*>}
f\n-\
/in
- ')"/'""
^îfr
Integrating this between the limits
and
n-l hm
/!
hn
f(a + h)=f{a)+^ -,f^^(a)+

ni = im:
Jo
Rn =
Let
and
have continuous
a^xâ + h.
when
Oî^l, we have
(71-1
fl
and
in Taylor's series.
differential coefficients of the first n orders
If
h {h + h~^).
let
j-j
^J~iyi ]
\l -
0'^-
be a positive integer such that
/'"'
1,
we have
/I
f\ni
'^,
]
{nL)l
V'"* (
p^
+ "'> - ''/' ( + "')
n.
f'Ha + th)dt.
+ ih)
dt
96
Rn = r^^-
Then
Let U,
" ^y~"
f ' (^
'
(^
"
O""^/'"* (
be the upper and lower bounds of
th) dt.
- ^)'^-p/("'
(1
[cHAP.
(a
th).
Then
X (1 -
f'
0^"'
dt<\\\- t)P-'
(1
- O'^-P/"''
(a
+ th) dt<[ U(l-
Since (1
t)'^~P /^'"^
(a
+ th)
U and
L,
and hence we can
values between
{I- ^)-y
t)P-' dt.
Jo
Jo
Jo
(a
<'^'
is
th) dt
it
passes through
find 6 such that
=^-1(1-
ey-Pf^"^ (a
^ ^^
1,
all
and
Oh).
Jo
R^ = .J^\y
Therefore
Writing
p = n, we
Me remainder and
;
which
is
get
Rn =
writing
Cauchys form for
Taking n = \ in this
result,
j9
;(1
^f-^/*"' (a
6h).
A"
/"" (a + Oh), which
1,
we get Rn =
is
A"
-
(1
Lagrange s form for
^)''-'/''" (a
Â),
the remainder.
we
get
f{a^h)-f{a) = hf{a + 6h)

\i
a^xâ + h;
is continuous when
Value Theorem (see also 4-14).
f {x)
Mean
this
result
is
usually
known
as the First
Darboux gave in 1876 {Journal de Math. (3) ii. p. 291) a form for the remainder in
Taylor's Series, which is applicable to complex variables and resembles the above form
given by Lagrange for the case of real variables.
55.
The Process of Continuation.
Near every point P, Zq, in the neighbourhood of which a function f{z) is

analytic, we have seen that an expansion exists for the function as a sei'ies
of ascending positive integral powers of {z
Zq),
involve the successive derivates of the function at
Now
circle
let
the coefficients in which
z^^.
be the singularity of f{z) which is nearest to P. Then the

is valid has P for centre and PA for
within which this expansion
radius.
Suppose that we are merely given the values of a function at
all
points of
the circumference of a circle slightly smaller than the circle of convergence
and concentric with
it
together with the condition that the function
theorems enable us to find
its
value at
all
is
to
be
Then the preceding
analytic throughout the interior of the larger circle.
points within the smaller circle
and to determine the coefficients in the Taylor series proceeding in powers

of z
Zq.
The question
arises, Is it possible to define
outside the circle in such a
way
that the function
a larger domain than the interior of the circle
the function at points

is
analytic throughout
,
and
Taylor's, Laurent's
5*6]
liouvtlle's theorems
97
In other words, given a potver series which converges and represents a

function only at poiiits within a
circle,
of the function at points outside the
For
this purpose choose
We
PA.
the series,
define hy
to
means of
it
the values
circle.
any point Pi within the
know the value of the function and all

and so we can form the Taylor series
circle,
its
not on the line
derivates at Pj, from
(for the same function)

with Pi as origin, which will define a function analytic throughout some
Now
circle of centre Pj.
which
new
this circle will extend as far as the singularity*
may
nearest to Pi, which
is
or
may
not be
but in either case,
circle will iisuall}'! lie partly outside the old circle of
for jjoints in the region which is included in the

circle, the
new
series
may
do
the old series failed to
Similarly
new
this-
convergence, and
circle but not in the old
he used to define the values of the function, although

so.
we can take any other point Po, in the region

now known, and form the Taylor
values of the function are
for
which the
P^
which will in general enable us to define the function at other
at which its values were not previously known and so on.
series with
as origin,
points,
This process
is
By means of it,
we can
called continuation^.
representation of a function by any one power series

of other power series, which between
them
starting from a
find
any number
define the value of the function
any point of which can be reached from P without

passing through a singularity of the function
and the aggregate of all
the power series thus obtained constitutes the analytical expression of the
at all points of a domain,
function.
It is
give the
important to know whether continuation by two different paths fBQ, PB'Q
same
power
final
series
it
will
will be seen that this is the case, if the function
have no singularity inside the closed curve PBQB'P, in the following way Let P^ be any
point on PBQ, inside the circle C' with centre P; obtain the continuation of the function
with Pi as origin, and let it converge inside a circle Ci let P^ be any point inside both
circles and also inside the curve PBQB'P; let S, Si, Si be the power series with P, Pi,
Pi as origins then|| *S'i = S'i' over a certain domain which will contain Pi, if Pi' be taken
sufficiently near Pi
and hence Si will be the continuation of Si
for if Ti were the
continuation of Si, we have Ti = Si over a domain containing Pj, and so ( 3"73)
corresponding coefficients in î and Ti are the same. By carrying out such a process a
sufficient number of times, we deform the path PBQ into the path PB'Q if no singular
:
point
is
PBQB'P.
inside
The reader
will
the process can be carried out in a finite

*
Of the function defined by the new
convince himself by drawing a figure that
number
of steps.
sei'ies.
+ The word usually must be taken as referring to the cases which are likely to come
under the reader's notice while studying the less advanced parts of the subject.
'
'
X French, prolongement German, Fortsetzung.

Such an aggregate of power series has been obtained for various functions by M.
;
by purely algebraical processes, Proc. London Math. Soc. xxxv.

II
Since each
W. M. A.
is
J.
M.
(1903), pp. 388-416.
equal to S.
Hill,
98
The
Example.
[CHAP. V
series
1
22
^3
a^
a"^
a*
a
represents the function
/('-)
only for points
within the circle
But any number of other power
=
a
type
series exist, of the
1
z-h
{z-hf
{z-hf
a-b'^ {a-hf"^ {a-bf'^ {a-bf^'-'
if
'
not real and positive these converge at points inside a circle which is partly
and partly outside s = a
these series represent this same function at points
b/a
is
inside
outside this circle.
.On functions
5-501.
which the continuation-process cannot be applied.
to
not always possible to carry out the process of continuation.

the function /(2) defined by the power series
It is
which clearly converges in the interior of a

is
whose radius
circle
is
Take as an example
unity and w^hose centre
at the origin.
Now
it
obvious
is
^^-1-0, /(2)^+qc
as
that,
But
and
if
the
+1
point
is
therefore a
of/ (2).
singularity
/(2)=22+/(^2)^
z--^\-0, f(z^)-^x and so /(s)-^x, and hence the points
singularities oi f{z)
the point
2= -
1 is
for
which
z-
=l
are
therefore also a singularity oif{z).
Similarly since
we
of
see that if 2
any
is
such that
2*
= 1,
then
a singularity
z is
of/ (2)
and, in general, any root
of the equations
22=1,
2*
= 1,
28
= 1,
2l=l,
...,
But these points all lie on the circle 2 = 1 and in any arc
of this circle, however small, there are an unlimited number of them.
The attempt to
carry out the process of continuation will therefore be frustrated by the existence of this
imbroken front of singularities, beyond which it is impossible to pass.
is
a singularity of f{z).
In such a case the function f{z) cannot be continued at all to points

the circle
a limiting
=1
such a function
fmiction,
and the
situated outside
circle is said to
The identity of tivo functions.
The two
be
.,
series
1
and
(^
do not both converge

Nevertheless,
2)
for
- (^ -
^'
2y-^
2'
(^
any value, of
2)^
z,
- (2 - 2^ +
and are
we generally say that they represent
strength of the fact

1
expression
called a lacuvary
the function.
circle for
551.
is
...
distinct
expansions.
same function, on the

that they can both be represented by the same rational
the
-"
>
DEPARTMENT OF
V MATMEMATIC
5501, 5"51] Taylor's, Laurent's and liouville's THEOREl|^^k)l)B R A R

This raises the question of the identity of two functions.
different expansions
be said to represent the same function
When
can two
We
might define a function (after Weierstrass), by means of the last

one power series together with all the other power
series which can be derived from it by the process of continuation.
Two
different analytical expressions will then define the same function, if they
represent power series derivable from each other by continuation.
article, as consisting of
Since
a point
power
if
a function
analytic (in the sense of Cauchy,
is
can be expanded into a Taylor's
it
series has a
series,
unique differential coefficient
definition of Weierstrass
5"12) at and near

and since a convergent
that the
( 5'3), it follows
really equivalent to that of
is
Cauchy.
important to observe that the limit of a combination of analytic

functions can represent different analytic functions in different parts of the
It is
This can be seen by considering the series
plane.
5('
The sum
The
circle
+ j)\!,('"i)(TT7.-r;i^)
of the first n
terms of this series
1\
zJ' I
+ z'^'
But, as w
1.
when \z\<
>
oo
>
or
is
the series near
^^
|
>
^^
|
C which
according as
is less
to infinity of the series
is
This series therefore represents one
1.
circle.
^ = 1, and an entirely different

The reader will see from 5"3
j
1.
if
a region
C is
taken and a set of points
are arbitrarily near every point of
a function which has a unique differential coefficient

of
oo
connected with the non-uniformity of the convergence of
has been shewn by Borel* that
points of the set
sum
see that the
function at points in the interior of the circle

that this result
-^
"
i
we
and - when
1,
2:'*^
function at points outside the same
It
on the
series therefore converges for all values of z (zero excepted) not
2^
or greater than unity; hence

z
is
do not belong to
*S';
but the function
is
it
6',
(i.e.
is
may
such that
be possible to define
monogenic) at
not analytic in
all
points
in the sense of
Weierstrass.
Such a function
is
.,
00
exp(-expw*)
^
f{z)= 2 2 2
;
^ \.
n=ip=oq=o z-{p + qi)jn
* Proc.
Math. Congress, Cambridge (1912), i. pp. 137-liJ8. Leqons sur les fonctions monoThe functions are not monogenic strictly in the sense of 5'1 because, in the
example quoted, in working out {f{z + h) -f(z)]jli, it must be su^Dposed that R{z + h) and I{z + ]i)
genes (1917).
are not botli rational fractions.
[CHAP. V
100
Laurent's Theorem.
5-6.
very important theorem was published in 1843 by Laurent*
to expansions of functions to
it
relates
which Taylor's Theorem cannot be applied.
is the inner
be two concentric circles of centre a, of which
Let G and
and let f{z) be a function which is analytic i* at all points on G and G' and
throughout the annulus between G and C. Let a + A be any point in this
ring-shaped space. Then we have ( 5"21 corollary)
ZTTt j
a-
cz
liTi j
cz
where the integrals are supposed taken in the positive or counter-clockwise

direction round the circles.
This can be written
We
the proof of Taylor's Theorem, that
find, as in
f(z)dz.h-^^
f(z)dz(z-ar+^
Jc {z-a-h)h^+'
c(z-a)"+'{z-a-h)
tend to zero as n^cc; and thus we have
/(a +
where +
cio
a^h
/ ^K^,
= ^^-
This result
h)
+ ajr +
and 6 = ^5
Laurent's Theorem
is
...
-~
+ j^^+
{z
- ay'-'f(z) dz.
...,
changing the notation,
it
can be
expressed in the following form: If f(z) be analytic on the concentric circles

G and G' of centre a, and throughout the annulus between them, then at any
point z of the annidus f{z) can he expanded in the form
f{z)
where
An
i {z
tto -h
-a) + a.2{z-ay-^
= ^. ^^^^
j^
a.nd b^
= ^.
[^
J^
(t
-f
^y:^^y
the circle
- aT'^f^t) dt.
important case of Laurent's Theorem arises when there
is
only one
namely at the centre a. In this case

G' can be taken as small as we please, and so Laurent's expansion
singularity within the inner circle G'
is
...+
valid for all points in the interior of the circle C, except the centre a.
*
Comptes Rendus,
xvii. (1843), pp.
t See 5'2 corollary

X
We
inside
C.
cannot write
2,
a-^^
348-349.
footnote.
= fW (a)ln
as in Taylor's
Theorem
since /(2)
is
not necessarily analytic

and liouville's theorems
Taylor's, Laurent's
5*6]
Example
Prove that
1.
= J^{x)+zJ^{x) + z'^J^{x) + ...-irz-J,,{x)-\-...
^^
e^^'
/"St
Jn (*) = s~
where
^^^
^TT J
[For the function of
z=Q
include the point
n = s
where
6'
~ ^ ^^"
^'^^
^) ^^
z under consideration is analytic

and so by Laurent's Theorem,
z) = a,, + axZ + aoz'-+...
g2V
and where
101
+ -^->r^ + ...,
n and G =
any domain which does not
in
i,
and C" are any circles with the origin as

= e'^, we have
z^
^dz,
Taking C to be the
centre.
circle of
radius unity, and writing

1
Z"'^""
-=-,
27ri.
suice
e:'^sin.e-'îrf^
27r
may
sin (/i^-î'sin ^) tZ^ vanishes, as
/'-'f
1
= jr-
Jo
be seen by writing in-cf) for
a = J(A'), and & = ( -)", since the function expanded

for z, so that 6 = ( - )"Ai(.:i^), and the proof is complete.]
Example
Shew
2.
that, in the
- X H\n 0) dB,
COS (n6
annulus defined
is
unaltered
by|a|<|3|<i6|,
if
9.
ThiLS
l)e
written
the function
î
bz
r
I'
\{z-a){b-z)}
can be expanded in the form
*"+.?,*'(? + 6-.)
* 1.3. ..(2^-1). 1.3... (2i + 2>i-l) /a\'
c.
"^^^^
'^=
2^^^^.ll{l + n)l
,!o
The
function
is
[b)
one- valued and analytic in the annulus (see
5'7), for
the branch-points
a neutralise each other, and so, by Laurent's Theorem, if C denote the

where o < < 6 the coefficient of 2" in the required expansion is
0,
'
circle
\z\=r,
Putting
= re'^,
[-'
this
dz
27riJ
bz
1^
c^'-^\{z-a){b-z)}
'
becomes
,, 1.3... (2i-l) ?*<*" 1.3... (2i-l)a'-''
the series being absolutely convergent and uniformly convergent with regard to
The only terms which

So the coefficient of z"
1
r^'T
2ir J
Similarly
it
6.
give integrals different from zero are those for which k = l + n.
is
*
1=0
(2^-1)
2Kl\
3 ...(2^
+ 2/i-l)
2* + .(/+m)
can be shewn that the coefficient of
is
S a'\
a^
*"
_Sn
'
b"
102
Example
Shew that
3.
where
rsT
'"
ZTT
î
cos {(
- v) sin
(9
- ?i(9}
o?^,
/"-"^
= 5-
e'" +
")'='^cos{(i'-?Osi"^-'<9}<^<9-
ZTry
nature of the singularities of one-valued functions.
T}ie
Consider
'' "> '''"
Z''
and
5"61.
[cHAP.
first
a function f{z) which
is
analytic
throughout a closed
region 8, except at a single point a inside the region.
Let
it
be possible to define a function
^ {z)
(ii)
when ^â, /(^) =
is analj^tic
By
h
a.
(z
</,(^)
+
^-T^
a)^
...
-.
(z
^^
aY
Any
(z
is
^"
{z
a)-
ay-
n at a'; and the terms
are called the principal

near
r
rpart of jf(z)
\ /
the definition of a singularity
the singularity
-^4-A^.^ + ...+
said to have a 'pole of order
is
that
throughout S,
(i)
Then f(z)
^ {z) such
5"12) a pole
is
a singularity.
If n
a.
1,
called a simple pole.
singularity of a one-valued function other than a pole
is
called
an
essential singularity.
If the essential singularity, a, is isolated

interior point, can be found containing
(i.e. if
a region, of which a
no singularities other than
a),
is
an
then a
Laurent expansion can be found, in ascending and descending powers of a
when dk>\z a\>h, where A depends on the other

Hence the principal
function, and 8 is arbitrarily small.
valid
'
near an isolated essential singularity consists of an infinite

It should
that
all
singularities of the
part
'
of a function
series.
is, by definition, an isolated singularity, so

which are not isolated (e.g. the limiting point of a
be noted that a pole
singularities
sequence of poles) are essential singularities.

There does not
singularity in the
Corollary.
exist, in general,
If
f{z)
has a
i\r{a)= lim {z-a)'^f{z), then
Example
[Prove that
of 2
1.
a would
if
all
an expansion of a function valid near a non-isolated

is valid near an isolated singularity.
way that Laurent's expansion
function
pole of
y\r{z) is
is
order
analytic at
at
a,
and ^{z) =
{z
aYf{z) {zâ),
a.
not bounded near an isolated essential singularity.
the function were bounded near z=a, the coefficients of negative powers
vanish.]
5-61, 5 -62]
and
Taylor's, Laurent's
liouville's theorems
c
Example
At
= 0,
the numerator
is
analytic,
the function has a simple pole at

Similarly there
denominator
A.t z
and the
= a,
is
z_
Find the singularities of the function e^~7{e -
2.
is
103
1}.
and the denominator has a simple
zero.
Hence
= 0.
a simple pole at each of the points ^mria
analytic and does not vanish for other values of
(= +
+2, +3,
1,
...);
the
z.
the numerator has an isolated singularity, so Laurent's Theorem is applicable,

Laurent expansion may be obtained from the quotient
coefficients in the
c^
z- a
(z
aV
Ml+'--" + ...)-l
which gives an expansion involving
an essential singularity at 2 = a.
Example
all
positive
Shew that the function
3.
and negative powers of
by the
defined
{z
a).
So there
is
series
wg"-^{(l +)t-^)"-l }
I
=1 (2-l){2-(l+/i-l)}
has simple poles at the points
= (1 + ~i)e-^'''^/
(^=0,
1, 2, ...
\;
y
The 'point at
5'62.
^^
'
If
we
values of
and
complete
the
is
write z
in
z'
-,,
the ^'-plane, there
it is
point at
Let/(2^)
near z
\
is
is
zero
is
sometimes convenient
can be treated in a similar
a one-one correspondence between

;
and
make
when
to
to say that
the correspondence
z
is
the origin, z
Then
really a proposition
it is
<^{z') is
its
<^(0) as lim
not defined at z
0,
but
behaviour
its
Taylor (or Laurent) expansion in powers
(/)
if
For instance
that limit exists.
(/) may have a zero of order m at the point

expansion of ^(^') will be of the form
Taylor
the
case
2'
so the expansion of f{z) valid for sufficiently large values of
of the form
is
is
careful to observe that this
the function
In this case,/(0)
by small
0.
determined by
and we define
oo
and any proposition about
/=
4>{z').
limit.
But the reader must be
infinity.'
not a definite point,
and
2, 3, ...).
so that large values of z are represented
provided that neither
concerning the point
of z
= l,
infinity.'
The behaviour of a function /C^') as

way to its behaviour as z tends to a finite
?i
(Math. Trip. 1899.)
said to have a zero of order
at
'
infinity.'
in this
.0
|
will
be
'
[CHAP. V
104
Again, the function
may have
(f)(2')
a pole of order
at the point
= 0;
z'
in this case
and
so, for sufficiently
large values of \z\, f(z) can be
expanded
form
in the
N P
f{z)
= Az"' + Bz'''-' + Cz'"-- +...+Lz + M+- + -^+

is
Similarly f{z)
said to have
is
an
/=
0.
'
infinity.
Thus the function
essential singularity at infinity, since the function
_1_
at
essential singularity at infinity, if
has an essential singularity at the point
e~'
e'
^ {z)
has an
or
= 0.
has an essential singularity at z

Example.
said to have a pole of order
In this case,/(2^)
....
Discuss the function represented by the series
(>1).
55,
:;
The function represented by
this
series
has singularities at
2=
and 2=
1
-î
since at each of these points the denominator of one of the terms in the
These singularities are on the imaginary axis, and have 3 = as a limiting
point so no Taylor or Laurent expansion can be foi^med for the function valid throughout
any region of which the origin is an interior point.
{n=\,
2, 3, ...),
series is zero.
;
For values of
z,
other than these singularities, the series converges absolutely, since the
limit of the ratio of the
(?i
+ l)th term
to the ?ith is lim (?i+ l)~i
an even function of z (i.e. is unchanged

2 -^ Qc
and is analytic on and outside a
if
the sign of
The function
is
be changed), tends to zero as
circle
of radius greater than unity
So, for points outside this circle,
it
a~^ = 0.
and centre
at the origin.
h + ^j + b+
where, by Laurent's Theorem,
'"
This double
of 2
it
.series
27nJ
=o! a-2 + 22
converges absolutely when
> 1,
and
if it
be rearranged in powers
converges uniformly.
Since the coefficient of
2~
Ms
2
H=o
zero integral
is
the term in z~^,
and the only term which furnishes a non-
we have
(_)fc-ia-2A- dz
b'k
Itti
J c =o
,^fo
n\ a2*
when
Therefore,
2
|
> 1,
The function has a
ill
e^
e*
zero of the second order at infinity, since the expansion begins with
z~'^.
Liouville's Theorem*.
5-63.
Let f{z) he analytic for all values of z and
of
f{z)
is
Let
inside
is
a constant
(so that \f{z)
\f{z)\<K for
let
hounded as
is
^r
|
all values
> x
Then
).
constant.
be any two points and
z, z'
Then, by
it.
take
where
z,
105
the function can be expanded in the form
e"'^
a term in
and
Taylor's, Laurent's
5-63, 5-64]
to
be a
write \^= z
whose centre
circle
\- pe'^
let
be a contour such that
z,
are
5'21,
since
z and whose radius
is
/|^2P when
jf
is
on
it
is
/s
on
follows from 4-62
that
= 2\z -z\Kp-K
Make p-^
that
is
As
to say,
cc
keeping z and
f(z)
is
next
will 1)0 seen in the
article,
volume (Chapters xx, xxi and xxii),

proofs of some of the most important
shall
now shew
its
expansion at the pole
Let the principal part of
will
*
be
results in Analysis.
is
jjoint (including oo )
Cj,
(Z
Cr
its
let its singularities in
...
c-j,
are rational functions.
and
Ck'.
let
the finite part
the principal part
5"61)
be
Cr
::;
Crf
+ (Z
'"
CyJ^^
expansion at the pole at infinity be
a-^z
there
= 0;
and again frequently in the latter half of this

theorem furnishes short and convenient
f(z) be such a function
let
if
obvious that/(/) f(z)
Liouville's
any
of the plane be at the points
of
it is
that the only one-valued functions luhich have no
singidarities, except poles, at
For
then
Functions with no essential singularities.
564.
We
fixed;
z'
constant.
a-iZ^
not a pole at infinity, then
...
all
anZ^^\
the coefficients in this expansion
zero.
This theorem, which
was given
this
is
really
due to Cauchy, Comptes Rendus,
name by Borchardt, Journal fvr Math,
in Liouville's lectures in 1847.
xix. (1844), pp. 1377, 1378,
lxxxviii. (1880), pp. 277-310,
who heard
it
106
Now
therefore
everywhere and
analytic
Theorem,
Liou\dlle's
where
[CHAP.
the function
has clearly no singularities at the points
/(.)
is
+ .. + a..= +
constant
is
.
.
f(z)
is
a constant; that
+ ,,.. +
is
c^,
c.j,
...
bounded
Ck,
or at infinity;
\z\>cc, and
as
is
it
by
so,
is,
I lî + (^^. +
therefore a rational function,
(i^J
and the theorem
is
established.
theorem (combined with
It is evident from Liouville's
that a function which
is
361
analytic everywhere (including
corollary
oc )
is
Functions which are analytic everywhere except at
constant.
considerable importance; they are

of such functions are
(ii))
merely a
oc
are of
known as integral functions*. Examples

From 5-4 it is apparent that there is no
e^, sin z, e^
convergence of a Taylor's series which represents an integral
and from the result of this section it is evident that all integral
.
finite radius of
function
functions (except mere polynomials) have essential singularities at oo
Many-valued functions.
5"7.
In
all
the previous work, the functions under consideration have had a
unique value (or limit) corresponding to each value (other than singularities)
of
^.
But functions may be defined which have more than one value
value of z
thus
if
r* (cos 1(9
2-
= r (cos 6 + i sin
+ t sin ^
(9)
Arc tan x + mr, where
^ tt
the
r^ jcos \ {0
and the function arc tan x (x
6),
real) has
< Arc
function 2-
lir)
sin
tan x <-^'tt and n

z,
is
except at
0,
and we can apply to them
the two functions are
called
'
change in
Thus
z-
27r)|
any integer
has a branch-point at
z^ as z describes a circle counter-clockwise

* Vxench, fonction entilre; Germa,n,
viz.
further
however, analytic
is.
branches of the many-valued function
coincide or at which one branch has an infinite limit

branch-points.'
the theorems of this chapter
game
and
z'-.'
two or more branches
There will be certain points in general at which
'
\ (6
sin {z~).
Either of the two functions which z^ represents
has the two values
an unlimited number of values,
examples of many- valued functions are log
^^
each
for
these points are called

and,
if
round
Funktion.
we
0,
consider the
we
see that 6
TAYLORS, Laurent's and ltouvtlle's theorems
57]
107
r remains unchanged, and either branch of the function passes

This will be found to be a general characteristic
over into the other branch.
increases by
27r,
It is not the
of branch-points.
purpose of this book to give a
properties of many-valued functions, as
of the
we
full
discussion
always have to
shall
consider particular branches of functions in regions not containing branchpoints, so that there will be comparatively little difficulty in seeing
or not Cauchy's
Theorem may be
Thus we cannot apply Cauchy's Theorem

iutegi-ation is a circle
the branches of z^
surrounding
when
out the contour and
Prove that
Example.
to such a function as z^
origin
but
the path of integration
its interior
are represented on an
tlie
if
whether
applied.
it is
is like
when the path
permissible to apply
it
of
to one of
that shewn in 6 '24, for through-
the function has a single definite value.
the different values of
a^,
corresponding to a given value of
Argand diagram, the representative points
equiangular polygon inscribed in an equiangular

independent of a.
spiral,
will
z,
be the vertices of an
the angle of the spiral being
(Math. Trip. 1899.)
The
idea of the different braiickes of a function helps us to understand such a paradox
as the following.
y = 'î
Consider the function

for
which
~=x^{\-\- log x).
AVhen x
is
negative and real,
not
is
real.
But
if
(where p and q are positive or negative integers), y
is
negative and of the form
is real.
Z'j^ -f- 1
If therefore
we have
we draw
the real c\irve
for negative values of
./
a set of conjugate points, one point corresponding to each
x with an odd denominator
and then we might think of proceeding to

form the tangent as the limit of the chord, just as if the curve were continuous and
rational value of
thus -, when derived from the inclination of the tangent to the axis of
The question thus
to be real.
give a non-real value for

arise
^ ?
arises,
Why
would appear
does the ordinary process of differentiation
The explanation
is,
from the same branch of the function y = x'^.
x,
that these conjugate points do not
We
have in fact
all
To each value of k corresponds one branch of the function y.

Now in order to get a real value of y when x is negative, we have to choose a suitable
value for k and this value of k varies as we go from one conjugate point to an adjacent one.
So the conjugate points do not represent values of y arising from the same branch of the
where k
any
is
integer.
function y=x^, and consequently
of
when evaluated
branch to be given by the tangent of the inclination to the axis of x of the

two arbitrarily close members of the series of conjugate points.
for a definite
line joining
we cannot expect the value
[CHAP. V
108
EEFEREXCES.
E. GouRSAT, Cours dJ Analyse,
J.
Hadamard, La
E. LiNDELOF, Le
C. J.
ii.
(Paris, 1911), Chs.
Serie de Taylor et son prolongement analytique (Scientia, 1901).
Calmd
des Residus (Paris, 1905).
DE LA Vallee Poussin, Covrs
1914),
Ch.
xiv and xvi.
d' Analyse Infinitesimale,
(Paris
i.
and Louvain,
X.
E. BoREL, Lecons
stir les
Fonctions Entieres (Paris, 1900).
G. N. Watson, Complex Integration
and Cauchy's Theorem (Camb. Math.
Tracts,
no. 15, 1914).
Obtain the expansion
1.
/(.)=/() + 2
1^-/ (-^^ +
(^-2-j
-3-3-r/
and determine the circumstances and range of
its
-25751-^^'
Obtain, under suitable circumstances, the expansion
3.
Shew that
(Corey, Ann. of Math.
....
'
vaUdity.
2.
\^r-]
(2),
i.
(1900), p. 77.)
for the series
)i=o
-r~
the region of convergence consists of two distinct areas, namely outside and inside a circle
of radius unity,
it
and that
in each of these the series represents one function
and represents
completely.
(Weierstrass, Berliner Monatsherichte., 1880, p. 731
4.
Shew
2
tends to infinity as 2--exp {i-niplm
integer
Ges.
Werke,
(1895), p. 227.)
11.
that the function
and p takes the values
!)
0, 1, 2,
s"'
along the radius through the point

..'.
(ni
where
is
any
1).
Deduce that the function cannot be continued beyond the unit circle.
(Lerch, Sitz. BOhm. Acad., 1885-6, pp. 571-582.)
5.
Shew
that, if z-
1 is
not a positive real number, then
-".^.s:)"<'-")-^/:'=-('-'-^'-**(Jacobi
and Scheibner.)
and
Taylor's, Laurent's
Shew
6.
that, if s
,,
1 is
109
not a positive real number, then
^2^"" +-+^(m +
=l+-2 + m(m+\)
m,
(1-2)
to
1)
...
(to
+ 1)
?i
-1^.
(Jacobi and Scheibner.)
Shew
7.
that, if z
and 12 are not negative
m+l
Jo
+ ^^"')
numbers, then
real
m+3
(m+3)
...
(m + 27i-
^^
(m+l)(w + 3)...(m + 27i-l)Jo^
^^
1)
'^^-
(Jacobi and Scheibner.)

8.
after
in the expansion of (a
If,
+ i2 + a22^)'"
by the multinomial theorem, the remainder
n terms be denoted by R^ (2), so that

(a4-i2 + a2S-)'" = ô + -î2 + -'-l2-"" +
shew that
/4(.) = ( + a.. + ..r
j,
---
(a +
+ ^n-iS"'"' + ^(s),
.M+a,^r^^"
'^^
(Scheibner.)
If
9.
(ao
/'(ao + i<
+ i2 + a22^)""'"'
+ 2-)'"o?<
be expanded in ascending powers of
the form
2 in
Jl2 + .4222+...,
shew that the remainder

(ao
after
+ ai2 + a22^)~'""^
n-\
(ao
terms
is
+ a, + a2^^)'"{o^.i-(2?'i +
?^
1)
2^-i^} ^""'c^C
(Scheibner*.)
Shew
10.
that the series
X(2)= -
where
and where
the
sum
(2) is
analytic near
+2- f, +
= 0,
I",-
- + (-
)"
,^'
convergent near the point
is
and shew that
if
of the series be denoted by/(2), then/(2) satisfies the differential equation
/'(--)=/(^)-0(4
(Pincherle, Rend, dei Lincei
11.
Shew
the series
"2
that the arithmetic
0,^^
on
a circle |2|
mean
= r,
(5), v.
of the squares of the moduli of
(1896), p. 27.)
all
the values of
situated within its circle of convergence,
is
equal
i=0
to the
sum
of the squares of the moduli of the separate terms.
(Gutzmer, Math. Ann. xxxii. (1888), pp. .596-600.)

*
The
results of
examples
5,
and 7
are special cases of formulae contained in Jacobi's dis-
sertation (Berlin, 182.5) published in bis Ges.
Werke, in. (1884), pp. 1-44.
were generalised by Scheibner, Lelpziger Berichte, xlv. (1893), pp. 432-443,
Jacobi's formulae
110
Shew
12.
[CHAP. V
that the series
2 e-2(am)^-m-l
m=
l
converges when
<1
be represented when
and that, when a > 0, the function which

< 1 by the integral
/ay
it
represents can also
f" e-"" dx
'
Jo ex_2 x^
\-n-J
and that
has no singularities except at the point z=l.
it
(Lerch, Monatshefte fiir Math,
Shew that the
13.
in
(i'
+z
z_
= 0, = 0),
v'
that
its
sum

is
+1
or
all integral
all
values of
according as the real part of
1,
v, v',
except the combination
values of z except purely imaginary values

z is positive
Shew
which the
and
p. 735.)
that sin \u (^ + -)[ can be expanded in a series of the type
a^,
in
or negative.
(Weierstrass, Berliner Monatsherichte, 1880,
14.
viii.)
zj^
2
r
\
)-V- 2
\(i_2^_2y'zi)(2v + iv'zif'^{\-iv-2v'z-î){2v + 2v'z-'^xf]'
__,,
{z
und Phys.
series
coefficients,
+ aiZ + aoy-+...+-^ + j, + ...,
both of
s"
sm
2ir
and of z~'\ are
/2t
J
(2m cos
^)
cos Jiô?^.
{)
n=i n-z' + a-
shew that/ (2)
is finite
a Maclaurin's series in
and continuous for all real values of z, but cannot be expanded as

ascending powers of z and explain this apparent anomaly.
;
[For other cases of failure of Maclaurin's theorem, see a posthumous memoir by Cellerier,
Bidl. des Set. Math. (2), xiv. (1890), pp. 145-599
pp. 126-138; Pringsheim, Math.
Miinehener Sitzungsberichte,
16.
region,
If f{z)
and
Ann.
Lerch, Journal fiir Math.
XLii. (1893), pp.
153-184
and
Du
cm.
(1888),
Bois Reymond,
vi. (1876), p. 235.]
be a continuous one- valued function of
throughout a two-dimensional
if
hf{z)dz =
^
for all closed contours
lying inside the region, then f{z)
is
throughout the interior of the region.

[Let a be any point of the region and
let
F{z)=\'f{z)dz
J a
from the data that F{z) has the unique derivate f{z). Hence F{z) is
analytic ( 5*1) and so ( 5*22) its derivate /(z) is also analytic.
This important converse
of Cauchy's theorem is due to Morera, Rendiconti del R. 1st. Lomhardo {Milano), xxil.
It follows
(1889), p. 191.]
CHAPTER VI
APPLICATION TO THE EVALUATION OF
DEFINITE INTEGRALS
THE THEORY OF RESIDUES

Residues.
6"1.
-^'-^
If the function /(^r) has a pole of order
at 2
= a,
then,
by the
definition
of a pole, an equation of the form
where ^
The
(z) is analytic
near and at
a, is
true near
expansion
coefficient aî in this
a.
called the residue of the function
is
f{z) relative to the pole a.

Consider now the value of the integral
f{z)dz, where the path of
J a
integration
small that
is
a circle*
a,
whose centre is the point a and whose radius p

and on the circle.
m
We
r=
J a'
(f){z)dz
by
52
dz
f{z)dz= S a_r
have
Now
a\2 ~
{z - ay
JA^-aV
J
Jo
,,
when
1,
p'-e"^
p''e
and (putting
^{z)dz.
a.
z-a = pe'^) we
Jo
'^
have,
ie'
r^l,
if
27r
-r
0.
we have
r
Hence
so
w+ J
(i)
gd-r)
But,
is
{z) is analytic inside
_d^ ^
finally
p"
f(z) dz
idd
27rz.
Now let C be any contour, containing in the region interior to it a number

of poles a,h,c,... of a function f{z), with residues a_i, 6_i, c_i, ... respectively and suppose that the function f{z) is analytic throughout C and its
:
interior, except at these poles.
Surround the points
a, b,c, ...
by
circles a,
/3,
7, ... so
small that their
respective centres are the only singularities inside or on each circle

function f{z) is analytic in the closed region bounded by C, a, /3, 7,
*
The
existence of such a circle
singularity.
is
then the
....
implied in the definition of a pole as an isolated
[CHAP. Yl
112
Hence, by
5-2 corollary 3,
f{z)dz=\ f{z)dz+\ f{z)dz+...
^ a
27rm_i
/3
27n'6_i
Thus we have the theorem of residues, namely that if f{z) he analytic

throughout a contour C and its interior except at a number of poles inside the
contour, then
f
where
XR
denotes the
sum of
f{z)dz = 27ri^R,
c
the residues
of the function
f (z)
at those of its
poles which are situated within the contour G.
This
Note.
6 "2.
is
an extension of the theorem of
If a is a simple pole oi f{z) the residue of f{z) at that pole is lim {{z-a)f{z)}.
The evaluation of definite
We
5*21.
shall
now apply the
of definite integrals
integrals.
result of 6*1
to evaluating various classes
the methods to be employed in any particular case
may
usually be seen from the following typical examples.
The evaluation of
6"21.
and
between the limits
An
the integrals of certain periodic functions taken
27r.
integral of the type

2n
R (cos e, sin 6) dd,

where the integrand
a rational function of cos 6 and sin 6 finite on the
is
range of integration, can be evaluated
by writing
- 1 (^ + z-%
cos ^
the integral takes the form
sin ^
e'^
=z
since
(2 - ^-1),
.
S{z)dz, where 8{z)
is
a rational function of z
J c
finite
on the path of integration
C, the circle of radius unity
whose centre
is
the origin.
Therefore, by
of
{z) at those
Example
1.
61, the integral
of
If
its
p-^
The only
equal
to
^wi times the sum of the residues

circle.
cos^+/>2
dz
j f;i{l-pz){z-p)'
pole of the integrand inside the circle
is
a simple pole at
is
lim
-r
,î{\-pz){z-p)
^(l-^2)
p and
;
the residue
6 '2-6 -22]
de
Hence
<p < 1,
If
2.
277
1^
jo
Example
113
- '2p
-J^
cos 2^+^52
iz\2' '^2'
'
(1 -jt)22) (1
_p5-2)
= 27r2i?,
2^
where
(26+1)2
snm
denotes the
of the residues of
,^
4s-"'
^
sr^-s
pz^)
{z^ p)
and the residues at them are
at its poles inside
l+p^ + p*
poles are 0, -p^, p^
and hence the
(1
^-o
(3+l)2
^C
these
(3+XY2
,
"C,-.
;tt
integral is equal to
Tr{l-p+p^)
Example
If
3.
n be a positive
integei",
/2t
Example
If
4.
a>
pT
i
622.
00
We
+ 6cos^)2
(a2_t2)l'
finite
evaluate
analytic
when x
Q {x) dx and
Given
whenever
e,
^;
jo
0?^
{a
+ hcoii^6f
Q (x) dx, where Q {z)
when the imaginary
of poles),
zQ{z)^>Q uniformly
Jo
= 0.
7r(2a + 6)
a^{a + hf'
number
that (iv)
sin (?i<9- sin ^)rf^
of certain types of integrals taken between the limits
(i) it is
e*^"^^
pT
2na
de
now
shall
/"2t
/
> 0,
T/ie evaluation
and + 00
2ir
e'99cos(n^-sin^)o?(9=
(ii) it
real,
Q {x) dx
is
a function such that
positive or zero (except at a
has no poles on the real axis and
(iii)
as
2^
|
both converge.
we can choose p^ (independent

> po and ^ arg z îr.
of arg2^) such that \zQ{z)
<
Q(z)dz taken round a contour C consisting of the part
+ p (where p > po) and a semicircle F, of radius

centre at the origin, above the real axis.
Then, by
6"1,
residues of Q{z) at
*
The condition xQ
consider
(x)
its
(x)
Q (z) dz
27ri'ER,
A.
p,
where 2i2 denotes the sum of the
poles above the real axisf.
is
not in
itself sufficient to secure
the convergence of
(x)
= (x log .t)~i.
t Q(z) has no poles above the real axis outside the contour.
W. M.
ejir
of the
real axis joining the points

its
Consider
having
00
arg^ such that Oârg^^^Tr; provided

xQ{x)>0, as ic> + x in such a way* that
for all values of
is
J -00
part of z
is
dx
114
Therefore
f'
J -p
In the
dz
(z)
27riSi?
'
= pe^^, and
then
= \rQ (pe*) pe^'idd
Q(z)dz\
11
dz
Q (z)
I
J
last integral write z
\j
[CHAP. VI
Jo
< Wei'JT)de
Jo
= e,
by
4-62.
Hence
lim
-p
p--oo J
But the meaning

lim
(r--oo
Jo
same
Q (a;)
of
(x) da;
Q (x) dx
and lim
rf^
Q {z) dz =
îri^R.
lim
is
both
exist, this
Q (x) dx
and since
double limit
is
the
p-^x> J -p
^
as lim
p-ôo
-J
Q (x) dx.
Hence we have proved that

Q(x)dx=^2'7rilR.
This theorem
is
particularly useful in the special case
when Q(x)
is
rational function.
Even
[Note.
condition
if
(iv) is
{Q{x) +
not satisfied, we
Exaynple
p^f:o J
The only
1.
residue there
- ^ttt i-
R,
2.
If a
6221.
By
3.
R + ai,
/
dx
> 0, 6 > 0,
e""'^
a pole at
= ^ with
_3
shew that
integrating
R + ai and making
x*dx
+ bx^)^
e-^^- dz
IT
I6aib^
round a parallelogram
R-^qc shew
e-''^'^coii{2Xax)dx = e-^<>-''
that, if X
> 0,
whose corners are
then
e->'x-dx = 2X-h e-^<^-
e-<^^dx.
Certain infinite integrals involving sines and cosines.
If Q(z) satisfies the conditions

(z)
is
Therefore
'-^(a
- R,
Q{x)dx = 2Tri2R.]
lb
r=^-
Example
have
pole of {z^+\)~^ in the upper half plane
/"^
Example
Q(-a;)}dx=\im
./
still
(i), (ii)
also satisfies those conditions.
and
(iii)
of 6-22,
and
ni
>
0,
then
6-221, 6-222]
Hence
{Q
means the sum

and so
If
(i)
(a;) e'^'''
115
+ Q {- x) e'"'''']
2E'
of the residues of
Q {x)
is
If
Q (x)
is
i.e. if
Q ( x) Q (x),
dx
ttiSR.
dx
ttSR'.
(x) cos (jyix)
an odd function,
Q (x)
sin (inx)
Jordan's lemma*.
6'222.
The
the upper half plane
CO
Jo
(ii)
equal to ^ttiIR', where
is
{z) e^^^ at its poles in
an even function,
/
dx
results of 6'221 are true if
condition Q(z)>0 uniformly
be subject
{z)
when Oârg^^Tr
to the less stringent
as \z:>oo in place of the
condition zQ{z)-^0 uniformly.
To prove
this
If Q(z)^>0
and if Q
we
T is a
Given
then,
if
e,
>
\z'.>c{a constant) and

e'""
(\
dz)
Q (z)
viz.
Oârg^^Tr,
^ arg z ^w, then
v)hen
= 0,
choose po so that \Q{z)\<
when \z\>
ejir
p,^
and
^ arg
z ^tt;
po.
|emipeose|
Q {z) dz
= i^
and
so
e""^
Q {z) dz
{el'Tr)pe-'^p^''">dd
= (2/7r)
Now
j^
lemma,
semicircle of radius p above the real axis with centre at the origin.
e"""^
But
as Jordan's
argz as
iDiiformly with regard to
(z) is analytic luhen both
liin
where
known
require a theorem
sin 6
2^/7r,
e''^
whenf
Q {z)
* Jordan, Cours d'Aiiahjse,
dz
ii.
<
^ ^
tt,
(2e/7r)
and
'"
pe-''>i"f/6'.
so
pe'-'^p^l'^dd
Jo
(2e/7r).(7r/2m)
<
e/m.
_ g-2mp0/7r
(1894), pp. 285, 286.
t This inequality appears obvious
proved by shewing that
"S
'
when we draw the graphs
(sin 6)1$ decreases as 6 increases
from
)/
= sin.r,
y = 2xjir;
it
may
to iir.
82
be

[CHAP. VI
116
Hence
lim
This result
(z)
da;
e'"''
dz
0.
Jordan's lemma.
is
Now
[" |e*^
Q (x) + e-'*^ Q (- a;)]
Jo
and,
making p>oo, we
{e"^^
is
Shew
1.
Q (x) +
Q (- x)] dx =
e-'"''^
that, if a
> 0,
Shew
that, if a
eos^r
^ 0, 6^0,
cos
,_
x^ + a^
'o
2.
e""'^
Q (z) dz,
l-rri^R,
6"221.
then
/""
Example
I
J
the result corresponding to the result of
Example
see at once that
Jo
which
2TrilR'
_^
?r
2a
then
2ax cos
'ihx
dx=n {h-a)
X'-
i:
(Take a contour consisting of a large semicircle of radius p, a small semicircle of

8, both having their centres at the origin, and the parts of the real axis joining their
radius
ends
then
make p-^
Example
3.
cxj
Shew
4.
S-^0.)
that, if 6
{x^
Jo
Example
Shew
T^TZTi
+ o'^y
that, if
> 0, m ^ 0,
Shew
that,
if
^'
> 0, a > 0,
.r(j??-
6.
Shew
- a^ mb
+ a^)|.
(36-
then
,
mx
+ a2)2
(Take the contour of example
Example
{Sb'^
^ sin ax
ax=i7re~'"K
,
,,
m ^ 0, a > 0,
sin
j^~
46-'
5.
cos mxclx =
/" ""
Example
then
then
tt
Tre""*" /
4a^
2a''
2.)
that, if the real part of
be positive,
{e-*-e-'')j=\ogz.
[We have
^
Ji>
'
sô,p^
lim
a-.0, p-*-x
=
since i~' e~'
is
lim
[Js
-!
dt
I./ 5
"
I /
Js
I
du
J 5c
-dt-]^- dt\
analytic inside the quadrilateral whose corners are
8, 8z, pz, p.
UA
6"23, 6*24]
Now
t~^ e~^dt-^-0
~
/
a,s,
117
when^(2)>0; and
p-^cc
= \ogz-
t-^ e-^dt
"
t-^{l-e-f) dt-^\ogz,
since
i''^ {1
Principal values of integrals.
6'23.
It
axis
in 6'22, 6 '221, 6*222 that the function Q {x) had no poles on the real
the function has a finite number of simple poles on the real axis, we can obtain
was assumed
if
-e-*)-^l as t-^0.]
theorems corresponding to those already obtained, except that the integrals are all principal
means the sum of
( 4'5) and 2/2 has to be replaced by S/^ + ^S/s'q, where
the residues at the poles on the real axis. To obtain this result we see that, instead of
2^
values
the former contour,
we have
to take as contour a circle of radius p
and the portions of the
real axis joining the points
-p, a-8i;
and small semicircles above the

a, b,
call
c, ...
are the poles of
these semicircles yj,
+ hi,h-b~2\
the real axis
Then
y.,,
Let
a'
Ynn
-2
be the residue of Q{z) at a
...
...
witli centres a, b,
and then we have
to
Q (2) dz+
then writing
Q (2) dz=
(^
Q (2)
Sj,
c, ...,
where
8.^, ...
-^0
= a + 8ie'^
Q{a + 8i e')
= -Irri^ It.
dz
on
y^
we
get
Sj e'6 id0.
But Q {a + 81 6^^)816'^^- a uniformly as 8,-^0; and therefore lim
we thus
make
Q (2) dz = 27ri2R,
Q (2) dz +
c-8^,
instead of the equation
Q{z)dz+
get
+ 8o,
real axis of radii Sj, So,
Q (2) on
we
Q{z)dz=
-iria'
get
P y Q{z)dz+(
-p
and hence, using the arguments
P
The reader
will see at
of
i^
(5-22,
Q{z)dz = 27ri2R + ni2R,,,
we
get
Q{x) dx = 2m {2R + ^2 /?o)
once that the theorems of ^^ 6"221, 6-222 have precisely similar
generalisations.
The process employed above

of the contour
is
of inserting arcs of small circles so as to diminish the area
called indenting the contour.
.^
6"24.
Evaluation of integrals of the fovTii
"^
af'~^Q{x)dx.
Let Q{x) be a rational function of x such that

positive part of the real axis
it has no poles on the

and x^Q{x)^ both when a;>0 and when
118
Consider l(
[chap. VI
Q{z)dz taken round the contour G shewn
zY~'^
in the figure,
consisting of the arcs of circles of radii

S and the straight lines joining their
end points ( zY"'^ is to be interpreted
p,
as
exp \(a-1) log (- z)]
and
log (- ^)
log
+ ^ arg (- z),
'
tt ^ arg ( ^) ^ tt
where
with these conventions the integrand
is
one-valued and analytic on and within

the contour save at the poles of
Hence,
if S?'
denote the
residues of ( zY~'^
at all its poles,
Q (z)
On
(- zy-'
the small circle write
Q (z) dz =
z=
Se^^,
27ri'Zr.
and the integral along
it
becomes
( zyQ{z)id6, which tends to zero as 8>0.
On
Q {z).
sum of the
{ sT
On
z = pe'^,
the large semicircle write
Q (2) idO, which
tends to zero as
p> x
z = xe"\
one of the lines we write
(-zy~^ becomes
and the integral along
it
becomes
z = xe~'^'
on the other
and
a;-ê'"-i>'''.
Hence
r [w''-' e-
lim
(S-.0,
pôc )
and therefore
af^~'^
.'0
Corollary.
'^*
Q (x) - x^-'e'<^-^> "' Q (x)] dx =
l-Trilr
Q (x) dx = tt cosec (air) Sr.
Q{x) have a number of simple poles on the positive part
it
may be shewn by
a;"~^
Q (x) dx
tt
indenting the contour that
cosec
(citt) S?'
tt
cot
((/tt)
2r',
where 1r'
Example
If
of the real axis,
<"-''
J 5
is
the
1.
If
sum
of the residues of
z"'-^
Q (z)
at these poles.
< a < 1,
I
Jii
\+X
dx=ir cosec arr,
jn
\-x
dx=iv
cot
air.

6'3, 6*31]
Example
<2<
If
2.
and -
tt
<n<
tt,
^gi(2 l)a
fz-l
.'di=.
119
(Minding.)
/,
Example
Shew
3.
that,
if
< s < 3,
then
^
/;
Example
Shew
4.
that,
if
dx^-^^-'^
{\-\-x'^'f
<p< 1
'
~4cos^7r2'
and -
x~P dx
/:
6 "3.
+2x cos X + X-
<X<
tt
tt,
then
TT
sin^X
sin pn
sin X
(Euler.)
Cauchy's integral.
next discuss a class of contour-integrals which are sometimes found useful
^V'e shall
in analytical investigations.
Let
Let
Cbe
a contour in the 2-plane, and let/(s) be a function analytic inside and on C.

be another function which is analytic inside and on G except at a finite luimber
(2)
of poles
let
C be
the zeros of <^{z) in the interior* of
multiplicity be
rj, ro,
...
degrees of multiplicity be
and
5,
let its poles in
of--^}^'
and
C be
let their
6j, ho, ...,
degrees of
and
let their
2)
Then, by the fundamental theorem of residues,

of the residues
a,, 02, ...,
the interior of
dz
fiz)
2ni J
c'
is
equal to the
sum
(f>{z)
at its poles inside C.
{z)
Now -
- can have singularities only at the poles and zeros of 0(s).
of the zeros, say
otj
we have
0(2) = iI(2-ai)'-i-|-5(2-ai)'-. + i-f-....
= Ar^ {z - a-^Yi -1 + 5 (r, + 1 ) (s - ai)-i +
Therefore
and
Therefore
<^' (2)
. .
/(2)=/(i) + (2-i)/'(i) + ....
P^
Thus the residue
Near one
- '^^^]
of
j^^^
is
analytic at a,
at the point
= ,,
is rif{ai).
Similarly the residue at 2 = 61
is Si/(6j); for near 2 = 61, we have

= C(2-6,j-. + Z>(2-6,)-*, + l-l-...,
f{z)=f{bi) + {z-b,)f'{b,) + ...,
(^(2)
and
so^^f!'^ + '-^^
2-61
0(2)
is
Hence
analytic at
i^.
i^J/^'^ f^
the summations being extended over
all
clz
= 2r,fia,) - 2s,f{b,),
the zeros and poles of
(f)
(2).
The mimher of roots of an equation contained loithin a contour.

The result of the preceding paragraph can be at once applied to find how many roots of
6'31.
an equation
cj)
{z)
lie
within a contour C.
For, on putting /(2) =
in the preceding result,
we obtain the
resvilt
that
z)
27rilc^{z)'^'
is
equal to the excess of the
number
of zeros over the
number
of poles of
the interior of C, each pole and zero being reckoned according to

*
(f)
(^)
must not have any
its
zeros or poles on C.
(2)
contained in
degree of multiplicity.
120
Example
Shew
1,
that a polynomial
Let
(^ (2)
= ao2 + ais'"~^ +
^) _
rpj^gj^
ao2'
Consequently, for large values of
if
zeros of
Example
(\^
...
then the contour contains
2itI
J c
VJ
dzÔ
C,
the total
number
of
= a,2'^(l +
â <
Therefore the
C the
inequality
roots of the equation

1
2"'1
. . .
+ ! 2 + o = 0.
+
. . .
+ ai2 + ao
+ ...+a..,i2^^^-fa,-i2>^-' + ...+ao\
tO,
on the contour, a being independent* of
number
of roots of f{z) contained in
2T!-i;
and, since
C/'|<1,
2.
l+U
2injc\s
27^^jc/(2)
^-^(^2 = TO.
+ ,_! 2' - +
/(2) = ,2^ î^^^--
Then
\z^/
we have
+ ai-i~^*^~^ + fc+i2'^î + ... + a,2'"j
/(
/ (2) = a^z''' + a^ _
For write
f/"!
If at all points of a contour
2.
a, 2'"
But
at the origin,
is
has no poles in the interior of
lfcS*|>lo + i2 +
is satisfied,
(o=t=0)-
(2) is
lim
where
27ri
c 2
j
(f){z)
roots.
c (p{^)
_/
27rt J
27r^
+ ,,
has
+ + -1
+ ...+,
a circle of radius p whose centre
and hence as
(f)
..
C be
But, as in 6-22,
as p-âo
>og'"~^
(^(2)
Thus,
of degree
<^ (2)
[CHAP. VI
we can expand
(1
dzj
U)~'^ in the uniformly cou-
vergent series
Therefore the
Example
3.
number
of roots contained in
Find how mauy
I'oots of
26
C is
equal to
i:
the equation
+ 62+10=0
each quadrant of the Argaud diagram.
lie in
*
C/ is a continuous function of
upper bound a must be less than 1.
1
on C, and so attains
(Clare, 1900.)
its
upper bound
( 3-62).
Hence
its
6 "4]
>
121
Connexion between the zeros of a function and the zeros cf
6 "4.
Macdonald* has shewn that
its
a fvMction of z analytic throughout the interior of

equation \f{z) = 3/, where
is a constant, then the
if fiz) he
a single closed contour C, defined hy
the
of f{z) in this region exceeds the number of zeros of

in the same region hy unity.
number of
f {z)
derivate.
the derived
zeros
On C \etf{z) = 3fe'e
then at points on
/=*'"-|-
function
'"=^'""{'S-(I)}-
Hence, by 6'31, the excess of the number of zeros oi f{z) over the number of zeros
of/'(s) inside +
is
27^^
Let
be the arc of
C makes
Ox
with
~ 27ri j
j c f{z)
C measured
cf (2)
from a
U^'
27rt j c
and
fixed point
let y^
dz)
'^^
he the angle the tangent to
then
1
1
f (dH /de\
r de']
^^= - 2^?rS dz]c
-2îja [dJ^ I dz)
,
r,
d0
dzl
= -2îr^ds-^''SdsjcNow
dB
log
dz
log
=i\j/
is
change
ordinary curve,
curve
value
>//
in
y^rj'iiT
by
27r
the same as
its final
oi f{z) over the
zei'os
in describing the curve
increases
is
C and
;
it is
value
number
obvious
that
and
of zeros of
if
is
any
as the point of contact of the tangent describes the
this gives the required result.
Example
n has n
its initial
hence the excess of the number of
/' (s) is the
purely real and
Deduce from Macdonald's
1.
theorem that a polynomial of degree
result the
zeros.
Example
Deduce from Macdonald's result that if a function/ (2), analytic for real
and all its zeros real and different, then between
two consecutive zeros oi f{z) there is one zero and one only of/' (2).
values of
z,
2.
has
all its coefficients real,
KEFERENCES.
M.
C.
Jordan, Cours
E. GouRSAT, Cours
dJ Analyse,
11.
(Paris, 1894), Ch. vi.
d' Analyse (Paris, 1911),
Ch. xiv.
E. LiNDELOF, Le Calcid des Residus (Paris, 1905), Ch.
Froc.
t /'
where
London Math.
Soc. xxix. (1898), pp. 576, 577.
does not vanish on
and
<A
^
^
ox
dy
,
(2)
~,
ox
xL
unless
are real, since
^
ay
all
il.
vanish
J-
ax
C has
/dy
a node or other singular point

it
and these are
follows
that
sufficient
X For a formal proof, see Proc. London Math. Soc.
(2),
if
/'(2)
at
for, if
f= <p + i\p,
any point, then
conditions for a singular point on
xv. (1916), pp. 227-242.
"
122
[chap. VI
1.
2
I
^1
function
when
zero
(2) is
(f)
f(x, y) is the coefficient of
if
= 0,
and
in
{x
when
real
is
+ iy),
'27r
'ix
cos
-^/{aoa
if
de = n4>
sin 6)
e,
z is real,
prove that
-1
and
is
analytic
when
< ^< 1,
(^).
(Trinity, 1898.)
By
2.
integrating ^-^
e"'^'
0,
R,
B+
i,
round a contour formed by the rectangle whose corners are

1
sin
By
3.
ax
3ttX
/.
1 e"
ax =
and making R-^xi
and
(the rectangle being indented at
shew that
(Legendre.)
4 e-l
2a
round the contour of ^ 6-24, where Q{z) is a rational

|-^0 and as 2 -^ x shew that if Q (2) has no poles
integi-ating log (-2) Q{z)
2^(2)^0
function such that
as
on the positive part of the
real axis,
Q (.r) dx
sum
equal to minus the
is
of the
.'
residues of log
between
Shew
4.
{-z)Q (2)
at the poles of
that,
where the imaginary part of log ( - 2)
lies
> 0, 6 > 0,
if
f
Shew
dv
hx)-^ =-|7r(e-l).
ecos6xgii^ (a sin
5.
Q (z)
tt.
that
-
a sin 2x
-,
l-2acos2.r + a2
xdx = -
TT
log
b\(1 +0),
j,
-7rlog(l
(
\
1< a < 1;
+ a-i),
{a?>r,
(Cauchy.)
Shew
6.
f
/
J
if <î)
that
sva.iX
2_!_
sintboX
02) ^ni oi)
a-i)
snia^r
sind).r
-^--^
^--= ...
COS
ai^r
. . .
COS
a,a7
a.r
"
=
m, (Do
.
'I
(,
...
^-
be real and a be positive and
o-m
>1<^1
+ |<^2i+---+l0nl + lai|+
...+|aJ.
(Stormer, Acta Math, xix.)
7.
C and
If a point 2 describes a circle

its interior
except at a
number
of centre a,
that
if
r,
where r
is
the
sum
u=f{z) will
y be attributed
the centre of gravity of y is the
flz)
of the residues of
^^-!-
at its poles in the interior of C.
(Amigues, Noiiv. Ann. de Math.

8.
(.3),
xii. (1893), pp.
Shew that
7r{2a + b)
dx
/ - (.2 + 62) {x^ + a2)2

9.
/(2) be analytic throughout
to each element of
a mass proportional to the corresponding element of C,

point
if
of poles inside C, then the point
Shew
describe a closed curve y in the w-plane.
and
Shew
<ia?h (a
+ 6)2
that
dx
/:
+ 6^2)
IT
^n^h
1.3...(2?i-3)
l.2...(H-l) a""^'
142-148.)

n n
Fn {z)=
If
10.
m=l
- 2'P), shew
(1
123
that the series
]i=l
^"(^''^")
fiz)=- I (2"??~"-l)%""^
1=2
is
an analytic function when
sum
not a root of any of the equations
z is
whose centres are at the
circles
+ 1,
a radius between n and n
is
= %"
and that the

two
and
If -4
11.
of the equation /(j)
whose radius
Shew
12.
that, if
Argand diagram two given
lAB cot -
is
{z)
(1899), pp. 234-241.)
roots (real or imaginary)
within a circle whose centre
shew that there
is
(Grace, Proc. Camb. Phil. Soc. xi.)
0<i'<l,
g{2v-l)ziri
^2
s\nirz
round a
lim
circle of radius
+5
and make
/i--x
(Kronecker, Journal
Shew
1 3.
that,
is
the middle point
=s
[Consider
(3), xviii.
of degree n, with real or imaginary coefficients,
at least one root of the equation/'
J^ and
Aim. de Math.
A'ôiiv.
represent on the
one having a small radius and the other having

number of prime numbers less than n + l.
origin,
equal to the
(Laurent,
of
2"
of the residues of f{z) contained in the ring-shaped space inchided between
if
m > 0,
fiir
.]
Math, cv.)
then
* sin" mt
dt
2^^g^{'"--'T<"-^>-^"'2i^'"'-^>'-- ""'"3V''""
Discuss the discontinuity of the integral at
If
14.
A + B + C+
...=0 and
a, b,
m = 0.
are positive,
+ ^cosâ;
^cosa.r+ficos6.r+...
1:
c, ...
shew that
log
dx= -A.,log a - 5dilog b- ,..- Ai-i^^/.
,
7,
that,
if
(Wolstenholme.)
By
k
considering
-j
lim
pâa
-.
J -P
-T
dt=ni+
'^
^'
axis (according as
a;
or
.v
< 0),
according as
x>0, x =
[This integral
Shew
is
or x
known
that, if
PI
J -p
example
2,
dt,
or its reflexion in the real
that
Um
p-.oo
lim
p^x
of 6*222
and thence deduce, by using the contour
6.
dt taken round a rectangle indented at the origin, shew
> 0,
I
a:.
'*'
/gX(*+)
15.
('- ''>'- + }
/"p
px(k +
ti)
-P k +
dt = 2,
or 0,
ti
< 0.
as Cauchy's discontinuous factw.']
< a < 2, 6 > 0, / > 0,

.r-i sin (âTT -607)
then
l-,^-l abr
-5- = i7rr
124
Let
17.
>
and
let
e-n-'^f
[CHAP. VI
= ylr{t).
}l=-oo
r
By
considering
-dz round
g-Z^TTt
-^^
N is an integer, and making N-*- qo
By expanding
a rectangle whose corners are
{JV+^)i, where
shew that
these integrands in powers of e~^^'^,
e-""^
respectively and integrating
term-by-term, deduce from 6"22 example 3 that
{TTty
Hence, hy putting
=l
---=
shew that
v.(o=^-H(i/o.
due to Poisson, Journal de VEcole foly technique, xii. (cahier xix), (1823),
see also Jacobi, Journal fiir Math, xxxvi. (1848), p. 109 [Oes. Werke, ii. (1882),
(This result
p.
420
is
p. 188].)
18.
Shew
that, if i;>0,
J(
2 e
= - 00
ntTTt
- 2nnat
Mem. de I'Acad. des

403-404 [Ges. Werke,
(Poisson,
(1828), pp.
Math. CXI. (1893), pp. 234-253
~^
e'"''-^
\\
ll
Sci. vi. (1827), p.

i.
c - "' W<
592
(1881), pp. 264-265]
see also 21-51.)
COS 2n7ra
Jacobi, Journal fur Math.
ill.
and Landsberg, Journal fur
CHAPTER
VII
THE EXPANSION OF FUNCTIONS
A formula
7"1.
let
Darhoux*
to
Let f{z) be analytic at all points of the straight

(^ {t) be any polynomial of degree n in t.
Then
dt
due
IN INFINITE SERIES
(_)m
^^^
if
(2
1,
we have by
rt)' (^
(-")
(f)"'>
Noting that
and 1 of
limits
<^')(0){/(^)
and
(2
a))
</>'"'
(0),
</>
(0/"'"^"' {a
+ t{z- a)).
and integrating between the
we get
-/()]
(-)'-n2-r{<^"'""''(i)/"'"(^)
+ (-)" (z - a)"+'
is
constant =
<^"" (t) is
t,
= t
which
(a
+ t(z- a)) + (-) {z - )"+'
(t)/' (a
z,
"
= -{z-a)
a to
differentiation
/' ''
(^)
=l
,n
line joining
(0/*"^" (
</)
I
-</>'"""''
(^
- a))
(0)/ <''()}
c^^,
the formula in question.
may be
Taylor's series
4>{t) = (t
1 )"
obtained as a special case of this by writing
and making >ao
Example. By sub.stitutiug 2?i for n in the formula of Darboux, aud taking (^ (0 = t^{t
obtain the expansion (supposed convergent)
^
f{z)-f{a)=
and
l^
tind the expression for the
7'2.
The
function ^ z cot
z, it
is
analytic
,1
Journal de Math.
(3), ii.
series.
the Bernoullian polynomials.
when \z\<
and, since
27r,
it
is
an even
thus
series,
p
=1 - B, ^ ~ B.~B,-4!
'61
2!
r>
^2 =
30
>
^3
2-'
Z'
called the /ith Bernoullian

-Si
t These
+ (- )-!/('-) (a)},
can be expanded into a Maclaurin

1
is
{/(") (.)
remainder after n terms in this
-zcot ^j
^''~'^- ^1"
The Bej'noidlian number's and
function of
then Bn
1)",
It is found that J
numbevf.
= 42
'
^^
= 30
'
-^5
= 6g
(1876), p. 271.
numbers were introduced by Jakob BernouUi
in bis Ars Conjectandi, p. 97 (published

posthumously, 1713).
X Tables of the first sixty-two Bernoullian numbers have been given by Adams, Brit. A.is.
ReiJorts, 1877.
126
These numbers can be expressed as definite
We
have,
"^
l"
(p.
sin pxdx
J;
^ + ^ cot tp
=-
integi'als as follows
122) of Chapter VI,
by example 2
-.
[CHAP. VII
(
a.'"
sin {px-\--^mT\
Since
e'^^
converges uniformly (by de
444
la
Vallee Poussin's tegt) near
p = 0\ doing
and writing
so
f"'*-i
jo
we may, by
both sides of this equation any number of
corollary, differentiate
times and then put
dx
we obtain
2t for x,
dt
proof of this result, depending on contour integration,

v.nd Phys. v. (1894), pp. 321-4.
is
given by Carda, Monatshefte
fur Math,
Shew
Example.
that
"~7!^^2,i 1) jp
e^*
Now
consider
the
function
Maclaurin series in powers of
valid
~1
sinh.r
It is
when
denoted by
,
V
ot_l
which may be expanded mto a
The Bernoullian polynomial* of order

in this expansion.
<
lir.
v is defined to
<^ (z),
-i
be the coefficient of
so that
n=l
This polynomial possesses several important properties.

preceding equation and subtracting, we find that
Writing
z+1
for z in the
te^'=
On
which
equating coefficients of
.[,^(^
t"
+ l)-0,(^)}^,.
on both sides of this equation we obtain
a difference-equation satisfied by the function
is
The name was given by Raabe, Journal filr Math.
(pn (z)-
xlii. (1851), p. 348.
THE EXPANSION OF FUNCTIONS IN INFINITE SERIES
7 '21]
An
explicit expression for the Bernoullian polynomials can
We
as follows.
127
be obtained
have
B^t*
Bjt-
Hence
From
this,
{z)
</)
by equating
coefficients of
= z''-\
+ nG,B, z'^-- -
nC,B,z^-'
nz''-'
the last term being that in z or

coefficients
When
z is
an
integer,
it
may
Example.
that,
,162,
nCeB,z"-' -...,
being the binomial
n(^\,
nth Bernoullian polynomial.
for the
+ 2"-' + ...+(2-l)"--.
series for the expression
Shew
and
be seen from the difference-equation that
/(0)/=l-'
The Maclaurin
z-
Maclaurin series
this is the
we have
( 3'73),
on the right was given by Bernoulli.
when n>\,
The Euler-Maclaurin expansion.
7"21.
In the formula of Darboux
( 7'1)
write </>n(0
^^^' 4'{^)>
where
(f>n{t) is
the
Bernoullian polynomial.
?ith
Differentiating the equation
times,
we have
*-'
</)n'"
Putting
(^
in this,
1)
we have
Now, from the Maclaurin

<^n<"-^*-^'
- </>u'""*'
(0)
^,^(n-i)(0)
= 0,
history of the formula
was discovered by Euler
is
(0)
</,,<'
^n*"~*' (1)
n {n
(1)
</),<"-^)
= -l.n:,
obtain the Euler-Maclaurin

*
It
(^'"-*'
kt''-\
0"'"*' (0).
we have
series for ^,1 (z),
Substituting these values of
we
{t)
k >
^^ (-)'-'B,,
(0)
and
if
;!.
^,i'"~**
(0) in
given by Barnes, Proc. London Math. Soc.
(1732), but
Darboux's result,
sum formula*
was not published
at the time.
(2), iii.
(1905), p. 253.
Euler communicated
(June 9, 1730) to Stirling who replied (April 16, 1738) that it included his own theorem (see
1233) as a particular case, and also that the more general theorem had been discovered by
and Euler, in a lengthy reply, waived his claims to priority. The theorem was
Maclaurin
published by Euler, Comm. Acad. Imp. Petrop. vi. (1732), [Published 1738], pp. 68-97, and by
Maclaurin in 1742, Treatise on Fluxions, p. 672. For information concerning the correspondence
it
between Euler and Stirling, we are indebted to
Mr
C. Tweedie.
128
- a)f (a)=f(z) -f(a) -
(z
{/' (z)
dV*^"*"'
(z
i^
'^'''
"
term tends
last
obtain an infinite series ior f {z)

If
we
write
"
a and
for z
F{a;)
dx =
to zero as
+ {z- a) t]
l(o {F{a)
dt.
n^ oo and we can thus

,
f {a).
F{a;) for/' (x), the last formula
J a
^^)/'"^" [a
Jo
(271)!
In certain cases the
-f {a)]
{^ni)i
tu=\
[cHAP. VII
+ F{a +
becomes
o))]
+ i
Writing a
we
(o,
a+
2co, ...
+ (r
1)
co
for a in this result
and adding
up,
get
"^
F{x) dx=oj
i^
F{a)
+ F{a + w) + F{a-\-2co)+
F(u + rco)i
\în)\
m-l
E = --
where
...^\
<^^, (t)
F'-"> {a
mco
wt)
dt.
This last formula is of the utmost importance in connexion with the

numerical evaluation of definite integrals. It is valid if ^(a;) is analytic at
all points of the straight line joining a to a + rco.
Example
1.
Example
2.
expansion of -
Iif{z) be an odd function of
z,
shew that
Shew, by integrating by parts, that the remainder after n terms of the

s cot - 2
may
be written in the form

(
)n +
/o"\
(2)
I
!
22ft
sm2
fl
/
/o
02i (0 COS (zO

^
<^^-
(Math. Trip. 1904.)

7*3.
Burmanns
theoiêm*.
We shall next consider several theorems which have for their object the
expansion of one function in powers of another function.
*
Memoires de VInstitut,
(1902), pp. 151-153.
ii.
(1799), p. 13.
See also Dixon, Proc. London Math. Soc. xxxiv.
7-3]
Let (z) be a function of z which

a is an interior point and let
(f)
is
analytic in a closed region
129
of
which
Suppose
that
also
(a)
cf)'
Then
4= 0.
theorem
Taylor's
the
furnishes
expansion
<l,{z)-b
and
if it is
= cf^'{a)(z-a)+'^(z-ay + ...,
we obtain
legitimate to revert this series
which expresses z as an analytic function of the variable [(j>{z)b], for

a
If then /(2:) be analytic near z = a, it
sufficiently small values of |^
follows that/(2^)
small,
and so there
f{z) =/{a)
The
actual
theorem, which
[<^
b] when \z a\is sufficiently
(z)
will
be an expansion of the form
{</)
a,
(z)
generally
6}
the
in
coefficients
is
is
known
(z)
1^ {0
^-J^
|^
{</>
{z)-bY+....
expansion are given by the following
as
Bilrmanns
theorem.
Let "^(z) be afunctiGn of z defined by the equation

z
tlien
-a
an analytic function f{z) can, in a certain domain of values of
expanded
in the
z,
be
form
f(z) =f(a)
+ 'X\
i^%^"
1^ [/' (a) t (a)l-] +
m
'4>{z)-b
dtdz
f
{
R>.,
''-'
Rn =
where
(t)
z)
(f>'
'
27ri
<\>{t)-f(z)
l<l>{t)-b_
and 7 is a contour in the t-plane, enclosing the points a and z and such that, if
^ be any point inside it, the equation (f)(t)=:
(^) has no roots on or inside the
(f)
contour except* a simple root
To prove
this,
^.
we have
f{z)-f{a)=
f'{K)d^ =
f'(t)<f>'(Odtd^
-2 (0(0-6
f'{t)^{^)dtd^
27ri
+
It is
cf>{t)-b
0(0-6
{0(o-6r-M^(o-0(r)]j
assumed that such a contour can be chosen
W. M. A.
_m =
ii
\
- a\ be
sufficiently small; see 7'31.
r
130
But, by
4-3,
-cf>(^)-by\f'(t)<t>'(Odtd^_{<}>{z)-bY-^^
(f)
[CHAP. VII
27ri(m
{t)
+ l)
(l){t)-b
6J
"
(t-ar^^
jy
m1
Therefore, writing
/{.) =/()
27ri
'"^
"i'
for
1) ly
{(f)
(t)
~(mTl)r 5^ L/
1^ [/
()
- b]'^+'
()
It
W(
1-
It ())"]
XO-6"
f'{t)dt
??i,
<!;'''"'
m=l
{m+
27ri
-^f'(t)cf>\Odtd^
cj>{t)-b]
</)(0-<^(r)
we may
If the last integral tends to zero as n^cc,
^VTite
the right-hand
side of this equation as an infinite series.
Example
1.
Prove that
z
= a+
n=l
where
To obtain
this expausiou, write
cf>(z)-b = {z-a)e^'-^-\
f{z)=z,
in the
above expression of Biirmann's theorem
But, putting
^{r{z)
= e'^-^',
we thus have
= a-|-<,
= (7i -
1)
= ( - 1)
X the coefficient of i""
X the
= (n-l)!x
'
in the expansion of (i~"'C^"*')
coefficient of t--^ in
+ 0"
(-)'''^''^''P
'
'
The highest vahie of r which gives a term in the summation is = -!. Arranging
summation in descending indices r, beginning with r n- 1, we have
/
therefore the
which gives the required
Example
2.
result.

,
1..
2.41
SUl''0+
....
SI]
Example
3.
Let a line
be drawn through the origin in the ^-plane, perpendicular to
the line which joins the origin to any point
a.
on the same side of the line p as the point a
is,
If z be any point on the 3-plane which

shew that
logs = loga + 2 2
7'31.
Teixeira's extended
In the
section
last
relation to the
theorem
viz.,
now
is
form of Biirmanns
series, for
theorem.
investigated closely the conditions of
much more
the reason that a
general
same
theorem just given that Laurent's theorem bears to Taylor's
in the last paragraph we were concerned only with the
will
next be stated
expansion of a function in
shall
\-
we have not
convergence of Bilrmann's
form of the theorem
131
this generalisation bears the
powers of smother function, whereas we
'positive
discuss the expansion of a function in positive
and
negative powers
of the second function.
The general statement of the theorem

we shall follow in this section.
is
due
whose exposi-
to Teixeira*,
tion
Suppose (i) i\in,t f{z) is a function of z analytic in a ring-shaped region A,

bounded by an outer curve G and an inner curve c (ii) that 6 {z) is a function
analytic on and inside G, and has only one zero a within this contour, the zero
;
being a simple one

points ^ of
and
(iii)
that
a;
is
a given point within
(iv) that for all
C we have
for all points z oi c
we have
\e{x)\ >\e{z)\.
The
e(z)-d{cv)=0
equation
has, in this case, a single root z
=x
in the interior of G, as
seen from the
is
equation f
j_r
e'(z)dz
rr
27riJc
of
^).
^...
e'(^)
.
,
d'(z)dz
'
9{z)
which the left-hand and right-hand members represent respectively the

of roots of the equation considered ( 6 "31) and the number of the
number
roots of the equation 6 {z)
contained within G.
Cauchy's theorem therefore gives
f^-L
*
Journal
The expansion
fiir
Math,
is
f{z)&{z)dz
f{z)^'{z)dz-
"
e{z)-d{x)
}cd(z)-d(x)
Jc
Jo
e{z)-d{x)_
cxxii. (1900), pp. 97-123.

justified
by
4-7, since
S
H=l
-!
j-
converges uniformly when
z is
\.o[z))
92
on C.
132
integrals in this formula can be expanded, as in Laurent's theorem,
The
in
[CHAP. VII
powers of 6 (x), by the formulae

i
f{z)d'{z)dz
]cd{z)-d{x)
ro^^
^^
f{z)e^
[d{z)\ n+i
ic
L
We
thus have the formula
Avhere
Integrating by
we
jârts,
get, if rt=t=0,
This gives a development of f{x) in positive and negative powers of

6{a;), valid for all points a; within the ring-shaped space A.
and poles off(z) and d{z) inside C are known, A^ and

5'22 or by 61.
If the zeros
5,i
can
be evaluated by
Example
1.
Shew
that, if
_1
^'~2
Shew
that,
Example
when
2.
If
< 1,
a'
j
> 1,
ni at a time,
sum
denote the
2.y
1.3
Y "^2.4.6
U+W
member
the second
2,
taken
then
V1+W'^2T4
.r
S"
2^
represents
sin2
is
7'32.
sin 2
,ito (27i
y ^'
.r~i.
numbers
G^...{2n-2)%
shew that
+ 2j!
the expansion being valid for
equation
2x
of all combinations of the
42,
.(1)
\l+xy
=l
all
\2?i
+3
\n C<('0
2?H-1^""'^
values of
/^
represented by points within the oval whose
and which contains the point
= 0.
(Teixeira.)
Lagrange's theorem.
Suppose now that the function f{z) of 7 '31 is analytic at all points in
Then 6^ {x) is analytic and
the interior of C, and let B {x) = (x a) 6^ {x).
not zero on or inside G and the contour c can be dispensed with therefore
the formulae which give An and Bn now become, by 5'22 and 01,
;
_i_r
liTi J c
Bn=0.
f'{z)dz
di{z)
^1
-^
c^-^
"
|/>)|
.,
7*32]
The theorem
we
write
di (z)
133
of the last section accordingly takes the following form,
= l/<^ (z)
Let f(z) and

(z) he functions of z analytic on and inside a contour
surrounding a point a, and let t be such that the inequality
(f)
\t^{z)\<
is satisfied at all
if
\z
points z on the perimeter of
a\
equation
tJien the
^=a + t<f>{a
^, has one root in the interior of G ; and further
any function of ^ analytic on and inside G can he expanded as a power series
in t hy the formula
regarded as an equation in
This result was published by Lagrange* in 1770.
Example
where a
|
1.
<^
Within the contour surrounding a defined by the inequality

a the equation
\z (z
a)\>\a\,
has one root
f,
the expansion of which

C
= +
is
given by Lagrange's theorem in the form
2 ^-4
,,
a".
Now, from the elementary theory of quadratic equations, we know that the equation
z
=
z
has two roots, namely 9]l
+ A/(l + -^)f
an
^î^
l^
/\./(^
example of the need
represents the former \ of these only
"^
'^/l
'
^>^d
our expansion
for care in the discussion of
these series.
Example
If
2.
which tends to
y be that one
of the roots of the equation
when z~^0, shew that
2/^=l+nz+
(n + 3)^2
'
^^,
z^+
,
w(n+4),(?i + 5)
g,
n(n + 5){n
'n + 6){n6){n + 7)
4
so long as
Example
,
z*+
"
,
?i (;i
(n + 8) (h + 9)
+ 6) (m + 7)
.
z
'
+..
< j.
3.
If
x be that one
of the roots of the equation
x=l +yx"'
which tends
to 1
when
3/--0,
shew that
log.^=3/ +
2a -1
2-
,
2/2
(3a
^
-1) (3a -2)'3/'+->
^^
the expansion being vahd so long as

I
<
(a
1
- l)-i -
(McClintock.)
I
Mem. de VAcad. de Berlin, xxiv.; Oeuvres,

The latter is outside the given contour.
ii.
p. 25.
134
V^
^
The expansion of a class of functions
7'4.
of the plane are simple poles a^
in rational fractions*'
a.,, fts,
where
.,
be the residues at these poles, and
bs,
l>2,
sequence of
the finite
in
aj
let it
fig
|
=$
Os
part
let
be possible to choose a
being R,n)
circles Ci (the radius of Ci
[CHAP. VII
function f{z), whose only singularities
Consider a
bi,
vvith centre at 0,
bounded on
not
(The function
cosec z may be cited as an example of the class of functions considered, and
we take Rm = {m + ^) tt.) Suppose further that Rmôc as m>oo and that
the upper boundf of \f(z) on Cm is itself bounded as| 7n >oo so that, for all
is independent of m.
points on the circle C^, \f{z)\< M, where
passing through any poles, such that \f(z)\
is
Cj.
Then,
if
a;
be not a pole oi f{z), since the only poles of the integrand are
the poles off{z) and the point z

1
= x, we
f(z)
But
2-771]
~!
f(^dz =
c^z-x
//
where the summation extends over
have,
all
by
6'1,
^r
poles in the interior of C,.
fi^' + JL[
27riJ
^TriJcm
^fi^dz
c^z(z-x)
^^''^^7a,.^2'jri]c^z{z-x)'
if
we suppose the function f{z)
Now
as
w> 00
f(z)dz
'^-^-^
,
be analytic at the
to
- *)
Cm ^ (^
(Rm~^),
is
and
origin.
so tends to zero as
tends
to infinity.
Therefore,
making m>ao, we have
=/(.) -/(O) +
-^
n=l
- i) - Urn ^.
(^~
=f^^
JCmZ(z-x)
W-iC aJ môo^TTl
f
f(x) =/(0)
i.e.
which
is
+ i
{-^ + -1
bn
an expansion of f(x) in rational fractions of x
and the summation
extends over all the poles of f{x).

this series converges uniformly throughout the region given by
where a is any constant (except near the points ). For if R,^ be the radius
of the circle which encloses the points ai'\, ... a |, the modulus of the remainder of the
terms of the series after the first ?i is
If
A'
a|<|n + i|
^ a,
*'
I
by
4'62
and, given
* Mittag-Leffler,
t,
/"
/ {^) ^^
^')
2iri J Cm ^ (z
^^'*
I
Rm-(^^
we can choose n independent
Acta Soc. Scient. Fennicae,
of
a;
xi. (1880), pp.
such that Maj{Rm

273-293.
CL)
<
e.
See also Acta Math.
iv.
(1884), pp. 1-79.
Which
is
a function of m.
X Of course 2? need not (and frequently must not) tend to infinity continuously;
example taken i?OT = ("^ + i) ^> where m assumes only integer values.
e.g. in
the
7 '4]
135
The convergence is obviously still uniform even if < + 1 provided the terms of
the series are grouped so as to combine the terms corresponding to poles of equal moduli.
|
If,
\<M, we have the condition z~'>f{z) < M, where M is

and ^ is a positive integer, then we should have to
instead of the condition \f{z)
independent of
expand
ra
when
on
z is
(7,,
by writnig
1
z-x
XP"
.V
-, + ...++ z^^'"^
zJ'+îz-xy
z^
and should obtain a similar but somewhat more complicated expansion.
Example
Prove that
1.
cosec2=- + 2(-)"(
z
\^~ "
the summation extending to
To obtain
all
positive
this result, let cosec
points z=nir, where
The
Ji
'
residue of f{z) at the singularity
see that \f{z)
is
bounded on the
),
The
n.
singularities of this function are at the
'
any positive or negative
is
1^'^
and negative values of
-- = f{z).
z
rnr is
circle \z\
integer.
therefore (-)", and the reader will easily
= {n + ^)
as --x
ir
Applying now the general theorem
/W=/(0)+.o,.[^-L.+l],
where
c is the residue at the singularity a
we have
f(^)-m+^(-r{7^^ + i]/(0)= lim 14^-^=0.
But
Therefore
which
is
cosec
= - + 2 - )
+
nnj
\_zmr
z
(
Example
2.
If
<a<
e"^
Example
3.
Prove that
1
27r;j;2
(cosh
^- cos
.<)
shew that
^22 cos
4n7r
27iair
111
27r.t'^
e'^-e-'^ Tr^+i^r*
sin 2na7r
^ e^-e--^""
{^nY+^x*
3
e37r_
The general term
of the series on the right is
(grTT
which
is
6-3^(3^)4 + 1-^
_ e-rTT)
1(^^)4
+ 1^.4}
the residue at each of the four singularities
(ttV+j^)
r,
(e'^- e-'"2) sin
'
-r,
nz
ri,
- ri
of the function
136
The
[CHAP. VII
which are not of the type
singularities of this latter function
r,
r,
7-i
ri,
are
at the five points
{li)a:
At 2 =
the residue
is
at each of the four points
2=
the residue
{27r^2 (cos
^j
(-l)*-?grn-
_ g-rn-
^.
(;.^)4
(cOsh
jr_^.2
whose radius
is
+ -,
But, at points on C, this integrand
is
0{\z
circle
.V
- COS ^)
nzdz
/"
,.
hm
,;J^ J c (7rV + J^*)
Stti
where Cis the
'.
^^
+ ^_^4
is
_ cosh .r)} ~
12
Therefore
an
(?i
\~^)
- e-'^^)
(e'^
and whose centre
integer),
is
the limit of the integi'al round
'
sin
tt^
the origin.
is
there-
fore zero.
From
the last equation the required result
now
is
obvious.
Ea:ample
4.
Prove that sec.r=4.
(^^^^-^-A^^ +
Example
5.
Prove that cosech x =
Example
^
6.
Prove that seeh .v=47r
Example
7.
Prove that coth x = - + 2x( -x
Example
8.
Prove that
m=-x
)!
2.v
-^
.
--
Vtt +4.t-2
(
-.
-5
<,
+ tt-s
rirr
ix^
257r2
^-^^
= -cc {m^ + a^) [n^ +
+ 5 + r +
^r-^
Q-n-^
^^J-^^- ..}j
= -7
ao
0^)
...).
,-...).
/
4A-2
+ _ -^
+ .).
coth tto coth nb.
(Math. Trip. 1899.)

7"5.
jTAe
expansion of a class of functions as infinite products.
The theorem
of the last article can be applied to the expansion of a certain
class of functions as infinite products.
For
Oi,
of
tts.
3,
f{z) be a function which has simple zeros at the points*

> where lim
an is infinite and let/(2') be analytic for all values
let
z.
Then /' (z)
is
analytic for
all
values of z
singularities only at the points a^, a^, a^,
( 5*22),
and so
Consequently, by Taylor's theorem,
f{z)
/'
and
*
/
/
={z{z)
ar)f' (.)
^-^^V"
M+
= /' {a.r) + {z - a,) /" (a,) +
These being the only zeros oif(z)
and a4=0.
"
f'iz)
,.
can have
It follows
immediately that at each of the points
has a simple pole, with residue
we can
If then
expansion given in
bounded on C^
is
an
n = \ [Z
f{0)
Since this series converges uniformly
from the
Doing
are suitably grouped

so,
and taking the
we get
/(^)
independent of
= 0, we
Clr.
when the terms
integrate term-by-term ( 4-7).
exponential of each side,
Putting z
follows,
it
[1,1
-
c is
ni>cc,
as
7 '4, that
f{z)
where
+ 1.
f(^)_f'(0)
we may
the function
a^,
find a sequence of circles C\n of the nature described in
f (z)
^;
such that
( 7*4),
137
7'5, V'G]
% 7"4,
ce/()
"
e4,
(1
z.
see that /(O)
= c, and
thus the general result becomes
This furnishes the expansion, in the form of an infinite product, of any

function /(^) which
Example
rrr,
where
1.
the conditions stated.
Consider the function f{z) =
any
r is
fulfils
which has simple zeros at the points
positive or negative integer.
/ (0) = 1
In this case we have
/' (0) = 0,
and so the theorem gives immediately

sins
]e
1-1
for
it is
n-l IV
mr
''
îTT
IV
easily seen that the condition concerning the behaviour of -fFS ^^
'
^ 1"*"
^'^
fulfilled.
Example
2.
Prove that
H^) {-G.-^J} H^r\ H^)) H^))

_ cosh k cos X
1
cos X
(Trinity, 1899.)
7"6.
The factor theorem of
The theorem
of 7"5
is
Weierstrass*
very similar to a more general theorem in which
the character of the function f{z), as

*
Berliner Alh. (1876), pp. 11-60
^^
j
j>x
is
Math. Werke,
not so narrowly restricted,

11.
(1895), pp. 77-124.
138
[CHLAP. VII
Let f{z) be a function of z with no essential singularities (except at

point infinity'); and let the zeros and pol6s oi f{z) be at
<
ai
I
If the
ja,i
>X),
a.2
]
Let the zero* at be of (integer) order
number
aj, ag, as,
of zeros
as ?i>oo;
for,
and poles
if not,
is
unlimited,
it
is
'
the
....where
??^.ft.
necessary that
the points a would have a limit point i*,
which would be an essential singularity oi f{z).
We
proceed to shew
first
gn{z) such that

00
of
all
that
it
is
possible to find polynomials
77]
Now
F(2)= U
let
Then,
ii
and has no
f{z)
\(l-)
F {z) = Gi{z),
Gi{z)
eOn
i^)
an integral function
is
139
5*64) of z
zeros.
It follows that
^
p
\Ti \Z J
,
-Jctz
G^ {z)
is
analytic for
all finite
values of z
and
00
so,
by Taylor's theorem,
converging everywhere
where G{z)= S
this function
integrating,
and
6,1^"
nbnZ^''~^
follows that
it
a constant
c is
can be expressed as a series
this series converges everywhere,
M=l
and
G (z)
so
is
an integral function.
Therefore, finally,
/(^)=/(0)e'-)
G (z)
where
The
[Note.
f(z) is
due
is
r|(i_f),.n..)
some integral function such that

presence of the arbitrary element
G (0) = 0.
G (2)
which occurs
to the lack of conditions as to the behaviour of /(s) as

If wi=l,
Corollary.
it is
sufficient to take k
The expansion of a
7'7.
class
n,
by
'
in this formula for
s |-9-oo
.]
2"36.
of periodic functions in a series of
cotangents.
Let f{z) be a periodic function of

of simple
poles;
z,
convenience, let
for
analytic except at a certain
number
be the period of f{z) so that
tt
f{z)=f(z + 'rr).
= x + iy
f{z)>l uniformly with respect to x as

when O^x^tt; similarly let/(^)^^ uniformly as ^> x
Let z
and
let
y>+ x
Let the poles of f(z) in the strip

residues at
them be
Further, let
IT
+ ip and
< x ^tt he
nt a^,
and
be a rectangle whose corners are* ip,
let
the
ir
ip,
(a,.
z),
ip in order.
^
ZlTl
taken round this rectangle
and the residue at z
is
f{t) cot
{t
- z) dt
the residue of the integrand at
Also the integrals along
DA
and
while as p'>oo the integrand on
CB
CD
of the poles are

...
on x = 7r,
AB
tends uniformly to
I'i,
tends uniformly to li; therefore

c,
cot (a,
r=l
a-i,
cot
cancel on account of the periodicity
Ul'-l)=f{z)+i
taken so large that ai,
a,, is c,.
f{z).
of the integrand; and as p>y:>, the integrand on
any
an',
Ci, Cj, ... c,i.
ABGD
Consider
* If
a.^, ...
-4
shift the rectangle slightly to the right;
a are inside the rectangle.
p, p'
are to be
140
That
we have the expansion
to say,
is
[CHAP. VII
f{z)
= \^ {V -0+1
c,
cot {z
- ar).
r=l
Example
1.
cot {x -
ftj)
a.,)
cot {x
. .
cot {x
aj = 2
cot (a^ ai)
. .
.*.
. .
cot (a^ a) cot
(a;
a^) + - )^
(
r=l
n
=2
or
cot
(c?^
ai)...*...cot
(,.
-a)cot
(.V
a,.),
r=l
according as n
Example
even or odd
is
the * means that the factor cot (a^ a,)
is
omitted.
Prove that
2.
sin {x bi) sin

sin {x a-i) sin
{x -h.^
...
sin {x 6)
02)
...
sin
(0^
(a'
a)
_ sin
(!
sin (aj
6i)
^2)
sin (ai
6)
sin (!
a)
sin {a^
- &)
sin (a2
- a)
. . .
cot (a;-ai)
(q.2
- ^1)
sin (a2
- i)
cos (ai
+ a2 + ...+a-6i-62- ... -^n)-
sin
cot
(.r
- 02)
+
+
7"8.
BoreVs theorem-^.
Let/(2-)
= S
be analytic when \z\^r, so that, by
ttji^"
-5"23,
awr**
< M,
=o
where If
is
Hence,
independent of
if
and similarly
Now
of z
<^{z)=
</)<'^'
|
{z)
consider f^ {z)
when \z\<r, by
Also, if
we
n.
CtnZ"
^^
<^{z) is
an integral function, and
< il/el^l/''/r^
=1
e~^
Jo
^ {zt) dt
this integral is
5'32.
integrate by parts,
/ {z) =
-e-'(f> (zt)
+z
e-'
4>'
{zt) dt
JO
= 2
+ ^"+1
e-^
(f>^''+'^
(zt) dt.
t=0
But lim
-<
Therefore
</)">
(^)
= ai
and,
when \z\<r, lim
f,(z)= t
a.nZ'^
e"'
</>'*"'
(zt)
= 0.
+ Rn,
m=0
t Lemons sur les series divergentes (1901), p. 94.
See also the memoirs there cited.
7"8, 7'81]
141
/no
\Rn\<\ ^""^^
where
e'K Me\ '^
''
'
r-""-^
dt
Jo
=
Consequently,
when \2\<
zr-' I"+iif {1
r-i}-i->0, as n->oD.
^^
I
r,
m=
f{z)=\
ancZ 50
where <^{z)= 2
l
-^
2 and
^*
<^(2^) is
e-^4>{zt)dt,
called Borel' s function associated with
an^:**.
M=
-^ and if we can establish the relation S=
e~'0 (^) o?^,

Jo
the series <S is said ( 8-41) to be summable (BY; so that the theorem just proved
shews that a Taylor's series representing an analytic function is summable (B).
If *S'=
(f){z)=
n=0
re=0
''^
'
BoreVs integral and analytic continuation.
7'81.
We
next obtain Borel's result that his integral represents an analytic function in
a more extended region than the interior of the circle \z\
= r.
This extended region is obtained as follows take the singularities a,b,c,... of f{z) and
through each of them draw a line perpendicular to the line joining that singularity to the
origin.
The lines so drawn will divide the plane into regions of which one is a polygon
:
with the origin inside
Then
it.
Borel's integral
represents
an analytic function (which, by
5*5
and
7'8,
is
obviously that defined by f{z) and its continuations) throughout the interior of this
polygon.
The reader will observe that this is the first actual formula obtained for the
analytic continuation of a function, except the trivial one of 5 '5, example.
For, take any point P with affix ( inside the polygon; then the circle on GF as
diameter has no singularity on or inside it*; and consequently we can draw a slightly
*
The reader
will see this
from the figure
ing side of the polygon would pass between
for
if
there were such a singularity the correspond-
and
i.e.
would be outside the polygon.
142
larger concentric circle*
with no singularity on or inside
"
( 3-34)
independent of
by
where
S is
therefore,
c^
(C0 = 2^-
real part of (jz
on
since f{z)
is
bounded and
^ 8 > 0,
2
l
4-7,
^-\f{.z)
exp
{Ctz-') dz,
where
F (0 is bounded in any closed region lying

t
and \
is
the greatest value of the
C.
we draw the circle traced out

when z is at the extremity
1C|.{ICI + S}-^<1.
If
gi-eatest
We
on
independent of
is
5-4,
< ^ (0 <^^^
so, when t is real, ^ (CO

wholly inside the polygon and
and
Then, by
it.
CZ"*^
2771 J
but 2 C^-Llf^ converges imiformly
[CHAP. VII
l)y
the point
2/^,
we
see that the real part of f/2 is

(, and so the value of X is
of the diameter through
can get a similar inequality for
(^'
(CO and hence,
by
5-32,
e~'^(CO
^t
is
.'
analytic at ( and
This
is
7"82.
A
and
is
obviously a one-valued function of
C-
the result stated above.
Expansions
mode
of
Stirling:]:
in series
of inverse factorials.
development of functions, which, after being used by Nicole

in the eighteenth century, was systematically investigated by
Schlomilch in 1863,
that of expansion in a series of inverse factorials.
is
To obtain such an expansion

the function he
f{z)= X
"h^"",
when \z\>
of a function analytic
and use the formula /(^)
=o
r,
we
let
ze~^^^{t)dt,
'0
00
where
(f>{t)=
anV^j^n
!)
this result
may be
obtained in the same
way
as
=o
that of
7-8.
Modify this by writing
e"'
</>
(0
F (^)
then
Jo
Now
if t
= u + iv
and
if
be confined to the strip
7r<w<7r, Hsa
one-
valued function of | and F{^} is an analytic function of ^; and | is restricted

so that TT < arg (1 |) < w. Also the interior of the circle ^ = 1 corresponds
|
The
Mem
difforeuce of the radii of the circles beioR, say,
5.
de VAcad. des Sci. (Paris, 1717); see Tweedie, Proc. Edin.
Math. Soc. xxxvi.
(1918).
J Methodus Dijferentialis (Londou, 1730).

More recent investigations are due to Kluyver, Nielsen
Compendium der h'dheren AnalysU.
and Pincherle.
sup.
See Comptes
(3), XIX., XIII., xxiii.,
liendiis, cxxxiii. (1901), cxxxiv. (1902),
JRendiconti del Lincei, (5),
xi.
(1902),
Annales de I'Ecole norm,
and Palermo Rendiconti, xxxiv.
Properties of functions defined by series of inverse factorials have been studied in an

important memoir by Norlund, Acta Math, xxxvii. (1914), pp. 327-H87.
(1912).
7 "8
2]
by the point
to the interior of the curve traced out
(writing ^
= exp [i (6 +
ir)})
and
when
It follows that,
M2
is
cos ^6)
^ id,
(t)Y
TT'f
- R (0-^0,
dependent of
Now
= log (2
inside this curve
\t\-R(t)^[{R
R (^)>oc
as
143
and
||
1,
^^<
suppose that
i^(^)
F{i) < M^
so
(1
\
then,
< Me""^*^ < M,
e*-*
where
|,
J/j is in-
^^l
by
5-23,
F"*' (|)
<
ifa
n\p-'\ where
the upper bound of \F{z)\ on a circle with centre ^ and radius
P<l-l
=
Taking p
=-(1
and observing that*
^)
(1
i-^f+Tlf
<M,e{n + iy .111(1Remembering
by
that,
e^+O
Jo
integrations
by
''^
^0
we
find that
have,
by repeated
r\^-^y^'(^)d^
1
-(i-^yîD
"^^+1
h
+
lim
e^+O
we
.'0
parts,
lim
f{z)^
<
^)->-'\
means lim
4*5,
4- ;i~^)'^
+
1
{z
+ l)(z + 2)
...+
{z
l)(z+2)...{z + n)
+ Rn,
l-e
where
if
bn
the real part of ^
Rn\^
(2
(l
say, putting
7'
+ l)(z +
2)...(z
i.e.
if
R (z) > r
lim f
eÔ J
n)
further
'|(l-^y+i^<"+i'(|^)|cZ^
+ 2y.nl
+ 8)...{r + n + 8).S'
M,e(7i
(r+l +
= R{z
+ x-i)^
= F^""^ (0),
n > 0,
\{z+l){z
<
lim
M,e(n + 2y.n\
+ 2)...(z + n)\.R{z-r)
<
where 8
B){r
r).
increases with x; for
y-'^
l + x,
a;
log (l
:;
>ey,
+ a;~^) = ]og
when
?/<l, and so log
(l+a;~^) -z
>0.
:,
)>'/
That
is
to
+J
144
[CHAP. VII
>+
n^(i + '-^^le
n|(i
Now
tends to a limit
to zero
n->oo
2-71) as
and
so
Rn\->0
if
(n
+ 2ye
^'"^^^^'^^"'
tends
but
+i
[/m >
1 l/m>
dx
^^
log(n
l),
by 4-43 (II), and (n + '2y(n+ l)-'-^-*0 when 8 >0; therefore

n^x and so, when R (z) > r, we have the convergent expansion
K->0
as
Example
1.
Obtain the same expansion by using the results
Example
2.
fi^=.[
dt
j^f{t){i-uy-'-^du.
iog(i+J)=!--"'
z{z+l)
where
u=
{I-
t)
z{2
+ l){z + 2)
{2-t) ...{n-l-t)
'
dt,
and discuss the region
in
which
it
(Schlomilch.)
converges.
REFERENCES.
E. GouRSAT, Cours d' Analyse (Paris, 1911), Chs. xv, xvi.
E. BoREL, Lecons sur
les series
divergentes (Paris, 1901).
Bromwich* Theory of Infinite Series (1908),

Schlomilch, Compendium der hoheren Analysis, ii.
T. J. I'a.
Chs.
0.
(Dresden, 1874).
viii, x, xi.
y -x (i>{y) = 0, where
If
1.
is
a given function of
its
argument, obtain the
expansion
/(,)-/(.)
where
denotes any analytic function of
validity.
2.
its
1,)7(.^),
argument, and discuss the range of
(Levi-Civitk, Bertd. dei Lincei,
(5),
xvl
its
(1907), p. 3.)
Obtain (from the formula of Darboux or otherwise) the expansion

/(.)
find the
*
1 H wr (i^
-/()=
/~J;"^jy" {/"' (2) -
'"/<"> ()}
remainder after n terms, and discuss the convergence of the
series.
The expansions considered by Eromwich are obtained by elementary methods,
the use of Cauchy's theorem.
i.e.
without

Shew
3.
145
thcat
13
(2m
OT=1
\)
h^
^
V"'
where
and shew that y (^)
is
ascending jiowers of
t.
By
4.
the coefficient of
?i
in the expansion of {(1
- t.v) (1+ ^ - Lv)}
2 [î
taking
'
+ l) = ,-^
<^(^^
in the formula of Darboux,
f(l-r)e^'
d"-
(^
l-re-
shew that
+ A)-l/M(4
/(^ + A)-/(.r)=- I ,^;|/(-)(,;
r
Vl.
m=l
r (ji(t)f(n+^(x+ht)dt,
_i.(_)A"+i
where
_^^-=i-a, ~ + a,~-asy, + ....
Shew that
5.
/(.)-/()
J^
(-)!
-^^^^
^/'"""" ()+/<''"-^'
2^!
(^)}
Prove that
6.
/(^2) -/(--l)
= ^l (^2 - --1)/' {Z2) + C2 iZ2-Ziyf"
-<^4(^2-^l)V*M2l)
(^l)
C3
{Z2
"
ÔV" (^2)
+ --- + (-)M^2---l)^^/^'{|^(e'''SechH)|^^_^/( + l)(2l + to.-^.-l)^^;
in the series phis signs
and minus signs occur
and the
in pairs,
term
last
l)efore
the
integral is that involving {Z'>-h)"", also Cn is the coefficient of 5" in the expansion of
cot
7.
all
If Xi
values of
'"
ascending powers of
and x^ are
z
such that
integers,
(Trinity, 1899.)
z.
and
<f)
x^^R (z) ^X2,
(z) is a function which

shew (by integrating
is
analytic and'bounded for
f <t>(^)dz
je=^27r^_l
round indented rectangles whose corners are
Xi, ^2?
^2 <^
i,
Xico
i)
that
i(^(a-i) + <^(*-i + l) + (^(î + 2)+... + 0(^2-l) + 2<^(.^2)

+ ^3/) - (-^1 +iy)-(i> (^2 - ^.y)
iy)
1 /'" (^ {x^
(^2 +iy)-(pi-vi+iy)-(p
(Aî - ^y)
^y) + <^ (^î
A. i~\ ^,^
'"'^
(/)
<?>
Mo
e2f i/
^..
^"
Hence, by applying the theorem

/'
4?i
W. M. A.
^,2n-l
-1^
-^du
= Bn,
10
'
146
where Bi,
C is
(where
shew that
are Bernoulli's numbers,
B.^, ...
[CHAP. VII
a constant not involving w), provided that the last series converges.
(This important formula

pp. 403-418
is
due to Plana, Mem. della R, Accad. di Torino, xxv. (1820),
a proof by means of contour integration was published by Kronecker,
Journal fur Math. cv. (1889), pp. 345-348. For a detailed history, see Lindelof, Le Calcul
Some applications of the formula are given in Chapter xii.)
des Residus.
8.
""
,
1.3...(2;i-3)
.r
n\
2^
"=2+!.(->"~^
x = 2u + u\ and shew that
for one root of the equation
If
9.
'
denote the smii of
.S'' ,
2n.+l
taken
at a time,
cos._
jp
|
<
1.
5%...{2n-ir,
32,
shew that
(_) +
|_(-)^|2^(^)_
+ 2)
oc
1
1
sins
converges so long as
combinations of the
th numbers
all
12,
it
_^
U,
+r-)n^<)
^-Isin^-i.
=o(2?i
(Teixeira.)
If the function f(z) is analytic in the interior of that one of the ovals
10.
equation
points
is
sin s
= C (where C< 1),
within this oval,
which includes the

be expanded in the form
0.
/(.)=/(0)+ S-
^-^
/'^"-^' (0)
*S'
2n
is
the
sum
+4;;^^/(2n-i)
of all combinations of the

22,
taken
at a time,
'
and
*S
"
mi at
11.
...
+<v, /' (0)
all
combinations of the numbers
52,...(2;i-l)2,
(Teixeira.)
a time.
Shew
that the two series

2z^
2^+32"
2_
2z
r^
^"^
represent the
same function
into each other
2z
1z^
+ "52
+'
2^
/ 2z
1.32^1-^2; +3.52V1-1
in a certain region of the z plane,
and can be transformed
by Biirmaun's theorem.
(Kapteyn, Nieuw Archief,
12.
sin^^z
numbers
32,
(0)+
(2), iii.
(1897), p. 225.)
and is analytic at all points in the

between the two branches of the curve jsin2| = C
points in the strip it can be exjianded in an infinite series
If a function f{z) is periodic, of period 27r,
infinite strip of the plane, included
(where
C> 1),
shew that
at all
of the form
/(2) = ô + ylisin2+...+^sin"2
+
+ cos2(5i + 52sin2+... + ^sin"-is + ...)
and
whose
can, for all
(2/i-2)2,
+ 1 denotes the sum of

12,
taken
62, ...
42,
2(1
shew that f{z)
/(2)(o)+Mi'/C'^-2)(o)+...+>s^;'"V'(o)
^
"^
I
where
origin,
find the coefl&cients -4
and B^.

13.
of
If
(^
147
and /are connected bj the equation
which one root
shew that
is a,
j<^'
^(^)=^-T^'/^'+i^^3
0"
{PP')
the elements of the
)'"
",
,
^"
row are
first
iiaultiplied
(^f\imim +
^', {(^^)',
i)
..., (<^'"~^)',
{(f)^)',
(/3/")
{ct>T
ipF')"
+ ...,
1! 2! 3
I
X"'
the general term being
((^2)'
by a determinant
and each row
{/'" P')
the jjreceding one with respect to a; and P,
diflferential coefficient of
in
/",
P',
...
which
is
the
denote
F{a),f{a\P'{a)
(Wronski, Philosophie de la Technie, Section
ii.
p.
Lagrange, Brux. Mem. Couronnes,

14.
If
the function W{a,

T.r
W{a,
x)
b,
xlvii. (1886), no. 2.)
4",
x) be defined
b,
shew that
x\<
-p
and shew that
W (a,
x)
b,
= l+{a- b) W {a - b,
b, .v)
.r),
then
this function are
Jr(l,0,
lf(0,
= e*-l,
A-)
= log(l+.r),
1, .r)
W'(a, \,x) =
15.
series
y= W {a, b,
x= yV {b, a, y).
if
Examples of
by the
a-b
(a b)(a~2b)
= x + ^j-x^
â-^+...,
+
^^
which converges so long as
For proofs of the theorem

Ann. de Math. xill. (1874j, and
381.
see Cayley, Quarterly Journal, xil. (1873), Transon, Nouv.
{\+xY-\
(Jezek.)
Prove that
{-Yx^ ^
a:ir"
n=0
where
Gn =
2ai
4a2
3ai
2ao
Gag
5ao
4ai
(2-2)a_i
(-l)ao
nan
(?i
and obtain a similar expression
3ao
- 1 a_x
a^
for
h
2 a^.xA'
U=o
(Mangeot, Ann. de VEcole norm. sup.

16.
Shew
(3),
xiv.
that
1_ cSrî
=o?'+l cai
- 2
2
a^x"^
r=0
102
148
where Sr
is
sum
the
[cHAP. VII
of the rth powers of the reciprocals of the roots of the equation
r=0
(Gambioli, Bologna Memorie, 1892.)
denote the nth derivate of f{z), and if /_ (z) denote that one of the nth
integrals oif{z) which has an w-ple zero at 2 = 0, shew that if the series
17.
If /
(z)
is
convergent
it
represents a function oi
the origin in the .r-plane, the series
+x
and
if
the domain of convergence includes
equal to
is
2 /_(2 + .t-)5r(0).
n=
Obtain Taylor's series from this result, by putting g{z) =

18.
Shew
that, if
jr
be not an integer,
,=-;,
as
1/
-^ QC
provided that
(Guichard.)
1.
all
n=-v{x-mf{x + nf
terms for which
in
=n
are omitted from the summation.
(Math. Trip. 1895.)

19.
Sum
the series
^
n^-q
where the value n =
is
1'
\{-Yx-a-n
n)
'
omitted, aûd p, q are positive integers to be increased without
limit.
(Math. Trip. 1896.)

20.
If i^(.r)
= Jo*"^*(^")''^,
shew that
F{x)=<^:tîJ^
and that the function thus defined
satisfies
F{-x) = ^^,
Further,
if
L,
the relations
F(x)F{l-x) = 2sm.V7r.
^ {^=^ + ^2 + ^' + -=
logC^-Oy,
shew that
(Trinity, 1898.)
ll-e-2'^'^Kl.
when
21.
Shew that
[-i-(^-.)i-(i^.)"][-(^-...)i'-G,y"]
n {l-2e-aCOs(a; + /3) + e-2p}^{l-2e-a(,cos(^-i3)+e-2(,}^
^ g=i
2*'*(l-cos.r)*e-'^''''*'''"
where
and
,20-1
ag=k?,\r\-^
TT,
2af-l
Pg=KCoa-^
0<a.'<27r.
tt,
(Mildner.)
THE EXPANSION OF FUNCTIONS

If
22.
0.'
I
<1
and a
=i
where C
is
149
IN INFINITE SERIES
not a positive integer, shew that
is
n~a ~ 1 - e2'^'
- e2a7r/
'
^~^J
j ^
a contour in the ^plane enclosing the points
0, x.
(Lerch, Casopis, xxi. (1892), pp. 65-68.)

If (^1(2),
23.
and
function,
if
(z), yjr^ (z), ...
J a
f F(^)
in
,
(x)
F{z) be any integrable
if
(x)
'l>^-{z)-<^
(^,_, (^0
...
r ^(f )^ = Ml) + J2j^ +
^^^^
V-3(^)
^1
2 (2)
^^^^^^^
J a
Shew that
and
z,
be polynomials defined by the equations
(x)
cî
any polynomials
are
<^2 (2), ..
yjî
(2) (^2 (2)
^3 (2)
24.
system of functions
as
=1
(S)
^0
where a and
p^, (z),
pi
po
(2),
(z), ... is
defined
Pn + l{z) = (2- + n2 + K) Pn
6 are given functions of n,
dx
x'
by the equations
{z\
which tend respectively to the limits
and
71-*- cx)
Shew
that the region of convergence of a series of the form 2epn{z), where
are independent of
Shew
2, is
a Cassini's oval with the foci +1,
that every function / (2), which
inside the oval, be
expanded in a
is
ej, 621
1.
analytic on and inside the oval, can, for points
series
/(2)=2(c + 20f'n(2),
where
Cn
= 27^. j ( + 2) ? (2) f (2) dz,
Cn'
= ^.
qn {z)f{z) dz.
the integrals being taken round the boundary of the region, and the functions q^
defined by the equations
?0
(^)
= 2^
yl^n
+ ao2Â-h
+ Oo
^" +
'
^")
^ XTT^
+ 0
+ rt+i ^TTT
2''
Let C be a contour enclosing the point

on or inside C. Let < be so small that
25.
2 is
on the periphery of
By expanding
in ascending
powers of
<,
and
let (f){z)
\t(f){z)
when
a,
\<.\ z
a\
C.
Stti
j c
shew that
l-^<^^(2)
f(.)
it is
- a - 0
^.
(2)
equal to
f{a)+ I ^^[/'(a)(0(a)}].
Hence, by using
6"3, 6"31,
obtain Lagrange's theorem.
being
2'n (^)-
(Pincherle, Rend, dei Lincei,
z is
(2)
(4), v.
and/ (2)
(1889), p. 8.)
be analytic
when
CHAPTER
VIII
ASYMPTOTIC EXPANSIONS AND SUMMABLE SERIES

Simple example of an asymptotic expansion.
8"1.
Consider the function f{x)
and the path of integration
By
t~^ e*~* dt,
where x
real
is
and
positive,
the real axis.
is
repeated integrations by parts, we obtain
lAy
\Aj
\AJ
iAj
CC
In connexion with the function /(a;), we therefore consider the expression
'^ni
and we
X^
shall write
11,2!
K
J^J
Then we have
iAj
t'--'
(J
Um/Um-i
= mx~^ >
(-rn\
iM^
tAj
oo
as
mcc.
The
series Si/m is there-
fore divergent for all values of x. In spite of this, however, the series can
be used for the calculation off(x) this can be seen in the following way.
;
Take any
We
fixed value for the
number
n,
and calculate the value

'"
f{x)-Snix) = {-r+^(n + l)lf

J
and
of Sn-
have
e^-^ dt
therefore, since
^ 1,
le e^~*
e^-Kl,
'"^
For values of x which are

equation
is
very small.
e'^-^dt
_.
r dt _n\
sufficiently large, the right-hand
Thus,
if
we take x ^
2/i,
member
of this
we have
\f{x)-Sn{x)\<^^^^:^^,
which
for large values of n is very small.
It follows therefore that the value
of the function f {x) can he calculated with great accuracy for large values of x,
by taking the sum of a suitable number of terms of the series Xum.
Taking even
fairly small values of
aS5(10)
= 0-09152,
x and
and
?i
</(iO)- ^5 (10)
< 0-00012.
ASYMPTOTIC EXPANSIONS
8 "1-8 21]
The
series
function f{x).
is
151
on this account said to be an asymptotic expansion of the
The
an asymptotic expansion
precise definition of
now
will
be given.
8 "2.
an asymptotic expansion.
Definition of
divergent series
z
in
which the sum of the
z^
first (n
I)
z^'-
terms
is
Sn
(^), is
said to be
an asymptotic
expansion of a function f(z) for a given range of values of arg^;,

expression
Rn (z) = z"'
Sn (z)]
{/(z)
Rn (2) =
lim
even though
When
this
is
lim
the case,
satisfies
|i?(2r)|
if
the
the condition
{n fixed),
=x
fixed).
(2;
we can make
\z-{f{z)-Sn{z)}\<e,
where
e is arbitrarily
We
small,
by taking
denote the fact that the series
is
sufficiently large.
the asymptotic expansion off{z) by
writing
f{z)'^
The
AnZ-"".
which has just been given is due to Poincare*. Special

asymptotic expansions had, however, been discovered and used in the
eighteenth century by Stirling, Maclaurin and Euler. Asymptotic expansions are of great importance in the theory of Linear Differential Equations,
and in Dynamical Astronomy some applications will be given in subsequent
definition
chapters of the present work.
The example discussed in 8-1 clearly satisfies the

given for, when x is positive, x" {/{x) - Sn (x)} < n ^-^ ->
!
definition just
as
a;
00
For the sake of simplicity, in this chapter we shall for the most part consider
asymptotic expansions only in connexion with real positive values of the argument.
The theory for complex values of the argument may be discussed by an extension of the
analysis.
8"21.
Another example of an asymptotic expansion.
As a second example,
consider the function
f{x)= 2
i
where
=l
/(.r),
represented by the series
-,,
x-\-h,
x>0 and < c < 1.

*
Acta Uathematica, vm. (1886), pp. 295-344.
152
The
term of
ratio of the ^th
the series converges for
values of
We
x.
therefore,
is less
than
c,
and consequently
shall confine our attention to positive
00
00^
00^
T*^
allowable* to expand each fraction
vkêre
it
We
when x > k,
have,
00 '\'lc
If,
{k- l)th
this series to the
positive values of x.
all
[CHAP. VIII
rearrange the series for / {x) in descending powers of
-4-
-1
in
this way,
and
we should obtain the formal
to
series
A2
Ax
x,
-,
^^
-i
x'^
-4-
.Z'"
OC
where
But
J,j
= (-)-i
and
this procedure is not legitimate,
shew that
it is
an asym]3totic expansion
Then
k=
so that
I
Now
('-
S^{x)= i
k"^c''
f{x)
/f--^c*^.
in fact
xj
Sn {x) = 2
^4n-'t"~"
...
+ izyi^)
We
diverges.
can, however,
of/(.r).
-^4 + ^lf +
l\
--
x + k'
j\<x-"--^ 2
converges for any given value of n and
is
Z:c*.
equal to C, say
and hence
fc=i
\f(x)-S{x)\<CnX-^-'-.
/(^)~ 2 A^x"^.
Consequently
Example.
If
f{x)=
where x
e^^-''^ dt,
is
positive
and the path of integration
is
the
J X
real axis,
prove that
.,
1_
'^'^^~2^
[In fact,
it
was shewn by Stokes

o
,0
.,
1.3
1.3.5
2%'
22.^3 +23.^'5
"
in 1857 that
/I
1.3.5
1.3
/
the upper or lower sign
8"3.
We
is
to be taken according as
- ^tt
<arg x<îr
or |^7r<arg
< 77.
Multiplication of asymptotic expansions.

shall
now shew
that two asymptotic expansions, valid for a
range of values of arg^', can be multiplied together in the

ordinary series, the result being a
new asymptotic
let
* It is
f(z)^ 2
not allowable, since
common
same way
expansion.
cc
OC
For
^,>^^-^
k>x
for all
</)(^)~
B,^2-'\
terms of the series after some definite term.
as
8*3-8 '32]
and
ASYMPTOTIC EXPANSIONS
Sn{z) and Tn{z) be the sums of their
let
153
(n
first
1) terms;
so that,
n being fixed,
f(z)
Then,
if
- Sn (Z) = (Z-X
C^ = ô^m +
îB,n-i
<t>
(Z)
Tn
+ ^mô.
{z)
(z^).
obvious that*
it is
SJz)Tn{z)= i C,nZ--^+0(z-).
But
f{z)
Cf>
(Z)
{Sn (Z)
= Sn {Z)
Tn
(Z-)] [Tn
(Z)
{z)
(^-)}
(Z-^)
TO =
This result being true
for
any fixed value of n, we see that
f(z)<f>{z)'^
Integration of asymptotic expansions.
8'31.
We
C\nZ-'^.
now shew
shall
that
it
permissible to integrate an asymptotic
is
expansion term by term, the resulting series being the asymptotic expansion
of the integral of the function represented
For
f{x)'^
let
^m*"'^ and
î
m=
let
by the
Sn{x)= % A^x-"'.
III
'l
Then, given any positive number

\f{x)
original series.
e,
we can
Sn (x)\<
=2
find Xq such that
\x\ ~"
when x>Xo,
and therefore
f{x)dx\
Sn{x)dx %\
\f(x)-Sn(x)\dx
(n-l) x""--^
\^
But
and therefore
On
8'32.
the other hand,
expansion
this
may
A;
A,
X
Sn{x^dx=^^ Ix" +
it
is
fix^dx*^
1.
...
'"
A
(n-l)a;""^'
.,'"
-.
not in general permissible t to diflFerentiate an asymptotic
be seen by considering e~^ sin
(e^).
Uniqueness of an asymptotic expansion.
question naturally suggests
itself,
as to
whether a given
series can
be
as 2 -*- x
See 2-11 we use o (2~") to denote any function
(z) such that 3"
(2) -For a theorem concerning differentiation of asymptotic expansions representing analytic
functions, see Kitt, Bull. American Math. Soc. xxiv. (1918), pp. 225-227.
*
i/-
i/-
154
[CHAP. VIII
The answer
the asymptotic expansion of several distinct functions.

is
To shew
in the affirmative.
L {x) which
zero,
a;
function e~^
is
x^L {x) =
for
all
of
whose terms are
every fixed value of
The
n.
-* 00
such a function when x
of a function J{x)
to this
observe that there are functions
first
are represented asymptotically by a series
functions such that lim
i.e.
we
this,
is positive.
The asymptotic expansion*
therefore also the asymptotic expansion of
is
J{x) + L(x).
On
the other hand, a function cannot be represented by more than one distinct
asymptotic expansion over the whole of a given range of vabies of
f{z)~ 2
for, if
f{z)~ 2 B^z-^^\
m=0
A,z-'-,
)n=0
tben
lim.(jo + 4^ + ... + 4r-5o-^^-...-^'VO'

^
^
^
\
r-*-oo
which can only be
if
A(^
A^ B^,
= Bq\
Important examples of asymptotic expansions will be discussed later, in connexion

with the Gamma-function (Chapter xii) and Bessel functions (Chapter xvii).
Methods of summivg
8"4.
'
We
have seen that
'
series.
possible to obtain a
it is
f(x)= i
A,,x~'"
development of the form
+ R,,{x),
m=
00
where Rn(x)^
co
as ?i> oo
and the
series
S)
A.,nX~^ does not converge.
m=
We
now
consider what meaning,
That
a non- convergent series.
we wish
is
numbers
'
sum
'
of
ao, a^, a^, ...,
by which we can obtain from them a
to formulate definite rules
number 8 such that S =
any, can be attached to the
to say, given the
00
00
'S,
if
an
=0
when
if
a converges, and such that
aS'
exists
)i=0
this series does not converge.
Borel'sf method of summation.
8'41.
We
have seen
7-81) that
/"
00
anz""
n=0
where
(tz)
= X
"
,
e-^(f) (tz) dt,

J
the equation certainly being true inside; the circle
00
of convergence of
circle,
* It
there
is
we
define the
a^z'^.
'
Borel
If the integral exists at points z outside this
sum
'
of
a,i2" to
mean
the integral.
has been shewn that when the coefficients in the expansion satisfy certain inequaUties,
only one analytic function with that asymptotic expansion. See PJiil. Trans. 213, a,
(1911), pp. 279-313.
t Borel, Leô)is
stir les
Series Diveryentes (1901), pp. 97-115.
SUMMABLE SERIES
8-4-8-43]
R(z)<
Thus, whenever
the
I,
'
sum
Bore]
'
155
of the series
2''^
is
=o
e-*e''dt
= {l-z)-\
If the
sum
Borel
'
exists
'
we say
that the series
A method, practically due

sum
of
'
may
a,i
'
'
Cesdro's-f
(1
when
ctn^'^
this limit exists.
Un
X+
cc-
...)
1+
lim
then if S
'summable
is
...
(1
would be
+ x)~^ =
i.
method of summation.
= ! + ao +
say that
be defined as lim
of the series 1
lim
Sn
3*7l
2:^-1-0 n=0
Thus the sum
8'43.
00
n=0
Let
(B).'
by the theorem of
to Euler, is suggested
00
'
summable
'
Elder's* method of summation.
8"42.
the
is
lim
and that
(C'l),'
(s,
Sj
+ ^n)
sum (CI)
its
is
exists,
S.
It
S=
we
is
necessary to establish the 'condition of
when
consistency;!:,'
namely that
this series is convergent.

n
00
To obtain the required
result, let
= s, S
m=l
to prove that
^S^^
s^
= ^Sn
',
then we have
m=l
> s.
n+p
Given
e,
we can choose n such that
a,
<
e for all
values of p, and
m=n-rl
so
Then,
S,
<
e.
if
i^
= i + J 1
Since
1, l
>
we have
n,
--")
v~\
+
l
...
2v~^,
from Abel's inequality

/-,
ftn+i
w\
U~^
+ a,, (1 -
...
is
M _
^^
^ + +,
^j
4.
. .
+ ^
a positive decreasing sequence,
_ 'i__\
it
follows
2'301) that
/
+ ^^n+2 (1
/?
+ l\
j +
.
.
/,
1
+ aJ
v-l
<"-7J^-
Therefore
S^-\a, + a,(\ - 1) +
...
* Instit. Cale. Diff. (1755).
+.(l -
V)}h
(^
See Borel, loc. cit. Introduction,

t Bulletin des Sciences Math. (2), xiv. (1890), p. 114.
+ See the end of 8-4.
^)
156
Making v^x, we
of S^
see that,
S be any
if
[cHAP. VIII
one of the limit points
( 2'21)
then
n
S %
Therefore, since
5,1
<
a,
e.
we have
e,
This inequality being true for every positive value of

that
S =^s;
had
to
that
is to
say S^ has the unique limit
we
infer, as in 2"21,
this is the
theorem which
be proved.
Example
1.
Frame a
2.
If
definition of 'uuiforui .summability
(Cl) of a series of variable
terms.'
Example
if
2 a,=s, then
&,,
hm
^^^ +
i,
i,
when
<v, and
when n
if,
It follows
J/
is
said to be
'summable
where
is
the particular case of this result
when
= 0.
The method of summation of Rieszi.
A more extended method
of
'
summing a
'
hm
X,i is
any
85.
Hardy's:]:
series
than the preceding
- ^-
cin
is
by means of
which tends to infinity with n.

'summable {Rr) with sum-function X.'
real function of n
this limit exists is said to be
Let
a,t6,^ exists,
from 8-43 example 2 that the 'condition of consistency' is satisfied in

proved* that if a series is summable (C/) it is also summable {Cr) when
r>r'; the condition of consistency
which
(Cr)' if lim
fact it can be
in
and
2 aJ),^\=S.
00
2 a
series
8"44.
= 1,
Cesdrds general method of summation.
8"431.
infixed, lim 6,^
series for
which
convergence theorem.
an he a series tuhick
is
s
summable {G
(
1).
Then if
w=l
a,i=0(l/?i),
the series
n converges.
n=\
*
Bromwich,
Infinite Series, 122.
t Comptes Rendus, cxlix. (1910), pp. 18-21.
X Proc. London Math. Sac.

indebted to Mr Littlewood.
(2),
viii.
(1910), pp. 302-304.
For the proof here given, we are
SUMMABLE SERIES
8-431-8'5]
Let
Sn
a.^
cin
then since
Un
157
is
summable {G
1),
we have
M=l
Si
where
s is the
sum
let
ti
this notation, it
independent of
and
n,
is
if
...
Sn
n[s
(l)j,
Sm-s =
With
is
((71) of
Let
and
+ So+
{m =
t,a,
to+
...
l, 2, ... n),
+tn =
sufficient to
On = n.o (1),
(Tn
shew
then
that, if
tn >
sis
a < Kn~^, where

|
n >
cc
Suppose first that a^, a^, ... are real. Then, if tn does not tend to zero,
is some positive number h such that there are an unlimited number of
the numbers t,i which satisfy either^ (i) t^ > h or (ii) tn < h.
We shall shew
there
that either of these hypotheses implies a contradiction.
and choose n so that

Then, when r
tn
0,
>
Take the former*,
h.
1,2,
< K/n.
Now
diagram.
than
whose coordinates are (r, tn+r) in a Cartesian

Since tn^r+i-tn+r = an+r+i, the slope of the line PrPr+i is less
arc tan (K/n).
plot the points P,.
Therefore the points Pq, Pj,
Let Pk be the
so that k
Draw
P.,,
^ h cot
lie
...
last uf the points P, Pj,
...
above the
which
rectangles as
shewn
""h+A;
The area
in the figure.
O"?! 1 ^^ tn
'
n+i
>|-A'-cot^
The reader
will see that the latter
left
of
xtan
6.
hcot
6,
a;'=
of these rectangles
= h x tan 6 and
+
the axes
Cn+fc
= i/i2ir-i7^.
hypothesis involves a contradiction by using arguments
of a precisely similar character to those
hypothesis.
line
on the
6.
exceeds the area of the triangle bounded by y

that is to say
lie
which
will be
employed in dealing with the former
158
But
(Tn+k
I
[CBAP. VIII
0"7i-i < 0'7i+A; + ^n-i

= (n + k).o{l) + {n-l).o(l)
!
= n.o{l),
since k ^ hnK~^,
and
h,
K are independent of
Therefore, for a set of values of
n tending
n.
to infinity,
^h-K~^n <n.o{l),
which
is
This
impossible since ^h'-K~^

is
h,
we
Similarly, by taking the corresponding case in
0.
arrive at the result lim tn
lim
we have
and
t^
so
That
to say Sn
is
>
If an be complex,
00
that
the contradiction obtained on the hypothesis that lim tn> h
therefore Urn tn ^
tn^
(1) as /i->
is )iot o
S R{an) and S
s,
and
we
so
lim
tn
> 0.
a,i is
consider
Therefore since lim
0.
0;
which
> lim
tn,
= 0,
tn
convergent
R (a)
/(c/) converge
tn
>
and
its
sum
is
s.
and / (a^) separately, and
find
by the theorem just proved, and
so
=i
w=l
00
an converges.
M=l
in
The reader will see in Chapter ix that

the modern theory of Fourier series.
this result is of great
importance
a function of | such that 2 (^) is uniformly summahle {C

n=l
A'"~^, where
is independent of
throughout a domain of values of |, and if a (|)
Corollary.
If
(I) be
<
\)
|,
a (^) converges uniformly throughoxit the domain.
n=i
For, retaining the notation of the preceding section,
uniformly, we can find a positive
number
if ^n(^)
does not tend to zero
h independent of n and | such that an infinite
<-h
^,j
The contradiction implied
in the inequality
for some point

sequence of values of n can be found for which t^ (^) > A or i (^)
of the domain* the value of ^ depends on the value of n under consideration.
;
We
then
the original theorem,
find, as in
WK-^n<n.o{\)
for a set of values of
n tending to
shewst that h does not
exist,
and
infinity.
so ^rt(|)-9-0 uniformly.
assumed that (^) is real the extension to complex variables cau be made as in the
If no such number h existed, < (|) would tend to zero uniformly.
t It is essential to observe that the constants involved iu the inequality do not depend on |,j.
For if, say, K depended on ^, K~^ would really be a function of n and might be o (1) qua function
of n, and the inequality would not imply a contradiction.
*
It is
former theorem.
ASYMPTOTIC EXPANSIONS AND SUMMABLE SERIES
159
REFERENCES.
H. PoiNCAR^, Acta Mathematica,
viii. (1886), pp.
295-344.
E. BoREL, Lecons sur les Series Divergentes (Paris, 1901).
Bromwich, Theory of Infinite Series (1908), Ch. xi.

W. Barnes, Phil. Trans, of the Royal Society, 206, a (1906), pp. 249-297.
G. H. Hardy and J. E. Littlewood, Proc. London Math. Soc. (2), xi. (1913),
G. N. Watson, Phil. Trans, of the Royal Society, 213, a (1911), pp. 279-313.
S. Chapman +, Proc. London Math. Soc. (2), ix. (1911), pp. 369-409.
T. J. Pa.
E.
pp. 1-16*.
e~^
/"
when X
is
real
and
:idt~
:;
4
~ + -^
...
positive.
Discuss the representation of the function
2.
f{x)={'' ^<^{t)e^-dt
(where x
is
supposed
and
real
by means of the
ditions)
positive,
and
//^^_<^W
Shew
that in
is
a function subject to certain general con-
series
certain cases
(e.g.
^'(0)
(^(;)
= e'")
<^"(0)
the series
is
absolutely convergent, and
represents /(.r) for large positive valu&s of x\ but that in certain other cases the series
is
the asymptotic expansion off{x).
Shew that
3.
,7
for large positive values of
-l
(a-l)(a-2)
z^
z^
z.
(Legendre, Exerdces de Calc. Int. (1811),
Shew
4.
that
if,
p. 340.)
when x>0,
,xa
du
^^'^=rK^*''^KT:^)}
Shew
also that/(.r) can be
expanded into an absolutely convergent
2
f{x)=
'^ '
^,
series of the
(Schlomilch.)
,r-
A.=i(^+l)(.r-i-2)...(^ + ^-}
Shew
5.
precede each
its
sum
that
if
1+0 + 0-1+04-1 + + 0-1 + ...,
zero precedes each
in which two zeros

+1, be 'summed' by Ceskro's method,
(Euler, Borel.)
is f.
Shew that the
6.
series
the series
1 and one
series
l+0-2! + + 4! + ...
This paper contains
memoir.
form
1-21 + 4!
...
cannot be
summed by
Borel's
method, but the
can be so summed.
many
references to recent developments of the subject.
bibliography of the literature of summable
series
will
be found on p. 372 of this
CHAPTER IX
FOURIER SERIES AND TRIGONOMETRICAL SERIES
Definition of Fourier series*.
91.
Series of the type

Ôo
+ (oî
cos
+ b^ sin x) + {a.2 cos 2x +
=
where
an, bn are
They
gations.
If there
is
independent of
x,
i^(7o
62
sin 2ic)
+ S
( cos
Via;
bn sin w^),
many
investi-
Riemann
integral
are of great importance in
are called trigonometrical series.
a function/(^.) such that
f{t) dt exists as a
J IT
or as
an improper integral which converges absolutely, and such that
iTan=\
f(t) COS ntdt,
then the trigonometrical series
is
7r6
f(t) sin ntdt,
called a Fourier series.
first appeared in analysis in connexion with the investigations

on vibrating strings d'Alembert had previously solved the equation of
Trigonometrical series
of Daniel Bernoulli
motion y = a^ -r^în the form
y=^ {f {x-\-at)-\-f {x -at)}, where 3/=/ (a-)
of the string starting from rest
is
the initial shape
and Bernoulli shewed that a formal solution
the fixed ends of the string being
(0,
0)
nnat
niTX
?/= 2 bn sui
and
-J{I,
is
cos
0)
and he asserted that
this
was the most
This appeared to d'Alembert and Euler to be impossible,

since such a series, having period 2^, could not possibly represent such a function ast
cx{l x) when t = 0. A controversy arose between these mathematicians, of which an
general solution of the problem.
account
is
given in Hobson's Functions of a Real Variable.
Fourier, in his Theorie de la Chaleur, investigated a
and shewed
to the
sum
that, in a large
number
number
of trigonometrical series
of particular cases, a Fourier series actually converged
Poisson attempted a general proof of this theorem. Journal de VEcole
f{x).
poly technique,
xil.
VAcad. R. des
Sci. vi.
(1823),
pp. 404-509.
Two
proofs were given by Cauchy,
Men. de
(1823, publi-hed 1826), pp. 603-612 {Oeuvres, (1), n. pp.
12-19)
and Exercices de Math. 11. (1827), pp. 341-376 (Oeuvres, (2), vil. pp. 393-430) these proofs,
which are based on the theory of contour integration, arQ concerned with rather particular
The second proof has been investigated by
classes of functions and one is invalid.
Harnack, Math. Ann. xxxii. (1888), pp. 175-202.
;
Throughout
this chapter (except in 9-11)
it is
supposed that
all
the
real.
t This function gives a simple
form
to the initial
shape of the
string.
numbers involved are
FOURIER SERIES
911]
9"1,
In 1829, Dirichlet gave the
161
rigorous proof* that, for a general class of functions,
first
the Fourier series, defined as above, does converge to the sum/(j;).
proof was given
The
if
result of -Dirichlet is that| if/(0 is defined
f{t) has only a finite
number
tt,
tt),
7ra,j=
and bounded
is
- tt,
(
number
in the range
maxima and minima and a
of
continuities in this range and, further, iff{i)
outside the range
modification of this
by Bonnet +.
later
finite
and
tt)
of dis-
by the equation
defined
then, provided that
Trh=
f{t) cos ntdt,
the series !(,+ 2 (Un cos 7uv + b niu
no;)
f{t) sin ntdt,
converges to the
sum
{/(^ + 0)+/(^ 0)}.
and Cantor developed the theory of trigonometrical series generally,

more recently Hurwitz, Fejer and others have investigated properties of Fourier
series when the series does not necessarily converge.
Thus Fejdr has proved the remarkable theorem that a Fourier series (even if not convergent) is 'summable (Cl)'
kjat all points at which f{x0)
exist, and its sum (C 1) is h {f{x + 0)+f{x 0)},
Later, Rieraann
while
still
provided that
f{t) dt is
of the convergence of Fourier series
One
integral.
which we shall give
of the investigations
later ( 9 "42) is
based on this
result.
For a fuller account of investigations subsequent to Riemann, the reader is referred to

Hobson's Functions of a Real Variable, and to de la Vallee Poussin's Cours dWnalyse
Infinitesimale.
Nature of
911.
the region within
which a trigonometrical
series converges.
Consider the series

1
- ao
where
may
be complex.
If
we
This Laurent series
where
a, b
But,
write
=^
o +
*
2 (a cos nz -J-
(1
^2^ I2
if z
= x-\-iy,
|
will converge, if it
C =^~''i
case which
is
+ 2 (" + *'^^ ^
converges at
Example
1.
in a region in
all,
which a
f ^b,
[
^'i*^
so
we
get, as
the region of convergence of the trigono-
plane defined by the inequality
a^- y ^
log
b.
line,
namely
a=b = \,
and
tlie real axis.
Let
fiz)^
= sin z - sin
2z
+ - sin Sz - - sin 4z
3
-f-
z=^x-\-iy.
Journal J iir Math.
iv. (1829),
pp. 157-169.
t Meiuoires des Savants etramjers of the Belgian
ploys the second
mean
Academy,
xxiii. (1848-1850).
Bonnet em-
value theorem directly, while Dirichlet's original proof makes use of
arguments precisely similar to those by which that theorem

X
of the greatest importance in practice is that in which
the strip consists of a single
the series becomes
("n - ib) f "
log
where
= f,
sin nz),
are positive constants.
metrical series, the strip in the
The
e'^
b,,
The conditions postulated
tov f[t) are
known
is
proved.
See
9"43.
as Dirichlet's conditions; as will be seen in
9-2, 9 42, they are unnecessarily stringent.
w. M. A.
11
162
[CHAP. IX
Writing this in the form
we
notice that the

"Writing
.v
we
in place of z (x being real),
(rsin.r
fix)= lim
converges* only
first series
-T-^ sin2.r
^ 0,
if 3^
by Abel's theorem
see that
+ -?'3sin 3,f ...
and the second only
if
( 3'7l),
lim \--iir&='--rê^'i^ + -)'^(^''-...\
+ ^ i (>-e-^-| ?-2e-2'^+i 7-36-3'^- ...U

This
is
and as r-^\
Xow
-tt
(if
Nil 1
<x<it),
yju
Since \x
= 0, we
is
>-e'^)
+\i log
a'
tt
where S
tt
see that ^
1)
integer.
and
is
therefore con-
any positive constant.
is
Tr<x<
lies in
the range
and putting
f{x) = ^x.
it,
Stt,
f{x)=f{x-2Tr) = ^
and generally,
some
= 0.
<x<
when n
is
example
( 3'35
converges uniformly
+ 8 ^ ^ S,
+ /-e-*^),
(1
\x + kn, where k
this tends to
continuous, k has the same value wherever
Therefore, v)heii
But,
nx
^^
tinuous in the range
A'
the limit of one of the values of
- U log
^
^ ^0.
if {2n 1)
< a' <
{2n + 1)
tt
(a- 27r)=-i.f-
77,
tt,
f{x)=-^x-n7r.
We
have thus arrived at an example in which
(,/-)
is
not represented by a single
analytical expression.
must be observed that
phenomenon can only occur when the
strip in which the

For if the strip is not of zero breadth, the
associated Laurent series converges in an annulus of non-zero breadth and represents an
analytic function of f in that annulus and, since {" is an analytic function of z, the Fourier
series represents an analytic function of z
such a series is given by
It
Fourier series converges
this
a single
is
line.
r sin
where
<r<
its
sum
is
X ^r^ sin 2x + J?'^ sin 3.>;
arc tan
l+/'COSA-
. .
representing
the arc tan always
a an angle
i,
J
f
between |7r.
Example
2.
When tt ^x^rr,
(-)-! cos
=^
n=i
The
series converges only
when x
.;
is
.,
real
by
.,
12
n-
3*34
the convergence
is
then absolute
and uniform.
Since
|.i-
= sin x - ^ sin
2x + 1 sin
3.-/;
. .
- tt + S ^ x ^ tt - 8,
> 0),
and this series converges uniformly, we may integrate term-bv-term from

and consequently
(-)"-^(l-C0Sn.r)
1^2
I
;5N
-2
2^-=
8^x^77 -d).
2
{-71 +,;i^,,^
*
The
series
do converge
if
0,
see 2'31
example
2.
to
( 4"7),

FOURIER SERIES
9-12, 9*2]
That
is
is
163
when -tt + S^x^tt-S,
to say,
^
where
nx
)"~^ cos
a constant, at present undetermined.
since the series on the right converges uniformly throughout the range - tt ^ :r $ tt,
sum is a continuous function of x in this extended range and so, proceeding to the
limit when x-^ 7r, we see that the last equation is still true when x= + it.
But
its
To determine
each side of the equation
C, integrate
( 4"7)
between the limits tt,
rr
and we get
D
Example
3.
By
t:
?l.
writing tt
^ sin^
Values of
tJie
2.>;
no;
/i^
71=1
9'12.
7r^--:F''= 2
Consequently
for
ttâ'^tt).
'*'
=: 1
.r
in
example
(Oâ'^tt),
-i
shew that
2,
= ^^ (tt d?)
| = {tt A'l -A-}
(-TT^.r^Tr).
coejjicients in
terms of the
sum of a
trigonometrical
seines.
00
Let the trigonometrical
^Co+ S
series
(c cos
?? a;
c?,j
sin n^)
be uniformly
n=l
convergent in the range (
tt,
tt)
and
let its
sum be
Using the obvious
f{x).
results
f
we
find,
sin
7/ia;
{m = niÔ),
[=7r
.!_
sin
^1=0
nxax
{m^n),
|-
on multiplying the equation ^Co
+ 2
aa;
(c,iCos?ia^
= ztt,
.^
C'^nsinna;)
=/(a;)
n=l
by*
cos nx or
by
nx and integrating term -by-term i*
sin
7rc,i=
J -TT
Corollary.
f (x)cosnxdx,
irdn
( 4*7),
f{x) sin nxdx.
J TT
trigonometrical series uniformly convergent in the range (-tt,
is
tt)
Fourier series.
Note.
Lebesgue has given a proof
{Series trigonometriques, p.
124) of a theorem
00
communicated
verges for
9"2.
to
all real
On
him by Fatou that the trigonometrical

values of
( 2'31
example
Dirichlet's conditions
1), is
series
2 sin nxllog
n,
which con-
not a Fourier series.
and Fourier's theorem.
theorem, of the type described in
O'l,
concerning the expansibility of

is usually described
a function of a real variable into a trigonometrical series

*
Multiplying by these factors does not destroy the uniformity of the convergence.
and 27r), Nova Acta Acad. Petrop. xi. (1793).
t These were given by Enler (with limits
112
164
On
as Fourier's theorem.
[CHAP. IX
account of the length and difficulty of a formal
when the function to be expanded is subjected

we defer the proof until 9"42, 9"43. It
proof of the theorem (even
unnecessarily stringent conditions),
however, convenient to state here certain
sufficient conditions
a function can be expanded into a trigonometrical
7r^^<
Let f{t) he defined arbitrarily wlien

real values oft by
and defined* for
all other
an improper
integral,
27r)=f{t),
a periodic function with period
Let f{t) be such that
f{t) dt exists
J
let it
under which
series.
means of the equation

f{t
so thatf{t) is
tt
to
is,
lir.
and if this
is
IT
be absolutely convergent.
Let a,
7r
b,i
be defined by the equations^
7r6=
f(t)cosntdt,
f(t) sin ntdt
Then, if x he an interior point of any interval

limited total fluctuation, the series
X
g-ao + 2 (a cos
n=l
is
convergent,
at
x,
this
= 0,
(;i
1, 2, ...).
J IT
J -IT
and
sum
its
sum\
is
nx +
(a, b)
in which f{t) has
hn sin nx)
^{f{x + 0)-\-f{x
Iff(t)
0)].
is
continuous
reduces tof(x).
This theorem will be assumed iu
j$
9'21-9"32; these sections deal with theorems con-
cerning Fom'ier series which are of some importance in practical applications.
It should
be stated here that every function which Applied Mathematicians need to expand into
Fourier series satisfies the conditions just imposed on f{t), so that the analysis given later
in this chapter establishes the validity of all the expansions into Fourier series which are
required in physical investigations.
The reader
will
observe that in the theorem just stated,/(^)
is
subject to less stringent
conditions than those contemplated by Dirichlet, and this decrease of stringency

considerable practical importance.
Thus, so simple a series as 2
)"~^ (cos
nx)ln
is
is
of
the
n=l
expansion of the function log 2 cos l^;*'

|
;
|
and
this function does not satisfy Dirichlet's
condition of boundedness at +7r.

QC
It is convenient to describe the series
âo+ S
(on cos nx
hn sin nx) as
n=l
Fourier series associated
the
*
This description must, however, be
This definition frequently results in f{t) not being ex^^ressible by a single analytical ex-
pression for
t
tuith f(t).
all real
The numbers
values of
a,^, fc
t.
Cf. 9'11
example
1.
are called the Fourier constants of f(t),
and the symbols
a,^,
b will be
used in this sense throughout i)*2-9"5. It may be shewn that the convergence and absolute
convergence of the integrals defining the Fourier constants are consequences of the convergence
and absolute convergence
of
f(t)dt.
/"-/'
|
Cf. 2-32, 4-5.
limits / (.r 0) exist, by 3-64 example 3.

example 6 at the end of the chapter (p. 190).
The
Cf.
FOURIER SERIES
9-21, 9'22]
165
taken as implying nothing concerning the convergence of the series in

question.
The representation of a function hy Fourier
9'21.
than (
series
for ranges
otJter
TT, tt).
Consider a function f(oo) with an (absolutely) convergent integral, and

with limited total fluctuation in the range aôs^b.
Then
= l(a + b)-l(a-b)'7r-W,
Write
it is
known
F{x').
that
( 9-2)
[F {x \-0)
f{x)
+ F{x' - 0)} = \
fto
+ t
(an cos nx
+ b,, sin nx),
n=l
and so
l{f{^ + 0)+f{x-0)}
5
2
tto
+ S
n=i
f
-^
ab)
nir (2x
a cos
h bn
sm
mr (2x a
b
b)
where by an obvious transformation

r/
cos
^{b-a)an = jr fi-i-)
1
1/7
^ (b
^
9 "22.
The
- a - b)
dx,
-jz^
nTT (2x
n7r(2x-a-b)j
a)\;6,1 = f%/
ax.
/ (x) sm
\
J a
cosine series
and
the sine seines.
Let f(x) be defined in the range (0, I) and let it have an (absolutely)
let it have limited total fluctuation in that range.
Define f (x) in the range (0, I) by the equation
convergent integral and also
f(-x)^f(x).
Then
2{f(x + 0)+f{x-0)]=-ao+ 2 jacos-j^ + 6sin-pk
where, by
9'21,
la,,
=r
fit) cos
'^ dt = 2 [ /(O cos '^ dt,
lbn=f f{t)sm'^dt = 0,
so that
when l^x^l,
l{fix + 0)+f{x-0)] = lao+iâncos'^',
this is called the cosine series.

If,
however, we define f(x) in the range
(0,
I)
by the equation
166
we
when
get,
I "^
[CHAP. IX
cc-^l,
where
ixnrf/
f(t) sin-.- dt;
lhn='2
J
I'
this is called the sine series.
Thus the
series
o
where
/lai'g f/ie
saie
f{t) cos -j-
On
Sill
nnx
y-
M=l
dt,
^ ^^"
sum ivken O^x ^l; but

when O^x'^
their
''
/(^) ^"^
^^'
sums are numerically equal and
l.
was given by Clairaut, Hist, de I'Acad. R. des Sci. 1754 [published,

memoir dated July 9, 1757; the sine series was obtained between 1762 and
cosine series
1759], in a
1765 by Lagrange, Oeuvres,
Example
[We
'
opposite in sign
The
j-
cos
Cin
n=\
^lan
mra;
2
+ ^
1
.-
1.
p. 553.
l.
Expand h {" ^v)
sin
^x^n.
in a cosine series in the range
.v
by the formula just obtained,
have,
^)
(tt
sin.r
= |o+
a,iC0s?2^,
n=l
j7ra=
where
x)
[ir
x cos nxdx.
sin
But, integrating by parts,

2{'7T
if 7i 4= 1,
x)smx cos nxdx
/:
(tt a;) {sin
(?i
+ l) A'-sin(?i- 1)
(^.r
.a;}
.'
_r
if
+l
1,
we
I'T
f^ (cos{n + l)x
eos(
l).r"i
+ \){n-l)'
{n
l).r"l
-27
n-lj
get
cos(?i
= 7rf
Whereas
rcos( + l)A'
(tt
- x)
sin
.r
xdx=^Tr.
cos
Therefore the required series
11
- + - cos X .
It will be observed that
proved to be |(7r
the sum of the series
is
2x
5 COS
1.0
it is
o cos 3x
only for values of
x) sin
series is
1 cos Ax.
0.0
x between
thus for instance when
.v
and n that the sum of this

and
has a value between
sin x, but | {tt+x) sin x

when x has a value
happens to be again h {n x) sin x, but this is a
mere coincidence arising from the special function considered, and does not follow from
TT,
between
tt
and
277,
the
sum
not i
is
(tt
- ^)
of the series
the general theorem.]
Example
rmi
I
The
2.
Expand
series is sin
îrx
sin
-\
(tt
Sx
^.
x)
sin
-\
in a sine series, valid
5x
-^
!-...]
when
^.v^tt.
FOURIER SERIES
9-3]
Example
Shew
3.
that,
^ x ^ tt,
when
COS 3^
COS hx
o
7r(7r-2.r)(7r2
+ 27r.r-2.r-) = cos.r+p- +^ +
1
\ /
-'N
[Denoting the left-hand side hy f{x), we have, on integi-ating by parts and observing
= 0,
that/'(0)=/'(7r)
fix) cos
Jo
nx dx=-\ fix)
L
= If-/
o
sin
nx
(^')
cos
7,1
Example
Shew
4.
|_-^
T''
f" ix) sin nx
/'" (x) cos
'/I*
+ -^
nx dx
"
(.r)
cos
dx
?a;
sin
/"'" (.f)
Vo
Jo
nx
Jo
'
- cos
(1
nxdx
??7r).]
-
4/i-*
x between
that for values of
sin
';o
Jo
^L
?i.;
'L
(a-)
'iJo
"!"
'
/'
Jo
and
tt,
^ can
be expanded in the
cosine series
2s
, ,
COS 4r
cos 2x
and draw graphs of the function
Example
Shew
5.
e**
25
and of the sum of the
that for values of
/cos
, ,
x between
cos ^x
J
j;
series.
and
tt,
the fimction
^tt
(n--
2.r)
can
be expanded in the cosine series

cos
COSA-
and draw graphs of the function
The nature of
93.
3.;
+ -5 +
cos
").,
-,-
+...,
- 2x) and of the sum of the
^tt (tt
series.
the coeffi^cients in a Fourîer series*.
Suppose that (as in the numerical examples which have been discussed)
interval ( tt, tt) can bo divided into a finite number of ranges
( TT, ki), (ki, ko) ... (k, tt) such that throughout each range f(x) and all its
differential coefiScients are continuous with limited total fluctuation and that
the
they have limits on the right and on the
left ( 3'2)
at the
end points of these
ranges.
Then
TTo,,,
'
f{t) cos nitdt
f(t) cos
mtdt + ...+
f(t) cos mtdt.
Integrating by parts Ave get

7ra,
*-^
m~^f(t) sin mt
m~^f{t) sin
7nt
r
... 4-
m-'^f
it)
sm mt
kn
pi
m~M
f^'i
f {t)Q\nintdtn\r^\ f (t) sin mtdt ...
J -tt'
The analysis
'
of 9-31
is
contained in Stokes' great memoir, Camb. Phil.
538-583 [Math. Papers,
i.
pp. 236-313].
of this section
Tratis. VIII. (1849), pp.
f'(t)sinmtdt,
'
k
.'/I-/
SO that
-m~^
f"^
and
[CHAP. IX
168
where
7r^)=
and bj
is
sin'nikr{f{k;.
a Fourier constant of /'
Similarly
h,
O)f(kr + 0)},
(x).
= ^ + -^
where
n
7ri?,
and
=- S
ai' is
cos
mkr
a Fourier constant of/''
Similarly,
IT
0)
-/(- tt + 0)|,
(a;).
'
"
Ti
are the Fourier constants
A,: =
sin 7AV
n "
'
Oj)j
/', &/'
of/"
(a;)
>
and
r (kr - 0) -f (kr + 0)},
r=l
7rB:
cos m-rr {/(tt
we get
Wjjj
where
- 0) - f(k, + 0)} -
{f{k\.
= - i cosw^,{/'(A^,-0)-/'(Av + 0)}
)
=!
- cos
miT {/'
(tt
'
- 0) -/' (- TT + 0)}.
Therefore
A
Now
as vi<xi,
Ti
n "
'
m-
we
see that
Ti
nr
AJ = 0{1),
'
nv
"
iiv
a; = 0(1),
and, since the integrands involved in a," and hj' are bounded,
it is
evident
that
,/=0(l),
Hence
if
^j = 0,
and uniformly, by
, = 0,
= 0(1).
the Fourier series for f{x) converges absolutely
3 "34.
The necessary and
6,/
sufficient conditions that A-,n
= B.,n
for all values of
are that
f{kr
that
is
to say that* /(a;) should
9'31.
The
- 0) =f{k, + 0),
- 0) =/(- TT + 0),
be continuous
for all values oi x.
Differentiation of Fourier series.

result of differentiating
"
term by term
tto
is
Of course /(x)
4-
is
2 {am cos mx
jn
9)i
/(tt
-i-
hm sin mx)
=l
{?7i6iCosma;
7710^ sinma;}.
=l
also subject to the conditions stated at the beginning of the section.
9 "3 1-9 '4]
FOURIER SERIES
With the
notation of
H o'
^
provided that
J.,,,
9"3, this is
+ S
the same as
cos
(ai
169
mx + hj sm mx),
m=\
= i?, =
and
/' (x) dx
jr
these conditions are satisfied
Consequently
continuous for
if f(x) is
sufficient conditions
all
values of
a;.
the legitimacy of differentiating
for
a Fourier series term by term are that f{x) should be continuous for all
values of x and /' (x) should have only a finite number of points of discontinuity in the range (
both functions having limited total fluctuation
tt, tt),
throughout the range.

Determination of points of discontinuity.
9"32.
b,n which have been found in 9-3 can frequently be applied

examples to determine the [loints at which the sum of a given Fourier series
may be discontinuous. Thus, let it be required to determine the places at which the sum
The
expressions for a, and
in practical
of the series
sin u'+
Assmning that the
3.';
+ 4 sin
")./
. .
and not <i/ij/ trigonometrical series and

{l-coa nin), we get on considering the formula found in
series is a Fourier series
0^=0,
observing that
!^
sin
discontinuous.
is
b^
= {2m)-
'^
9-3,
^,,,
Hence
if
y(-, ,
k.^, ...
= 0,
= ^ -h cos
B^
nirr,
a,'
= 6,' = 0.
are the places at which the analytic character of the
sum
is
broken,
we have
= 7r.4, = [sin ml'i

Since this
there
and
- 0)
true for all values of
is
-/(/(,
?,
^'3,
+ 0)} +sin mk.> {/{h, - 0) -f(L;, + 0)} + ...].
the numbers
it
is
true for
all
values of
?n,
This shews that,
open range
94.
We
must be multiples
of
tt
but
î
we have
i7r=/( + 0)-/(-0),
any
...
in the range
/{,
Since this
niT {n
ki, h^,
7r<.r^7r, namely zero. So = 0,

= in the equation 5m = i-i cos mir, we have
Substituting
... do not exist.
7r(^-^COSJ7r)=-[cosm7r{/(7r-0)-/(-7r+0)}+/(-0)-/( + 0)].
only one even multiple of
is
X^)
{f(i-^
integer),
(
tt,
0)
if
the series
|7r=/(7r-0)-/(,r + 0).
a Fourier
is
series, /(.r)
has discontinuities at the points
and since a,' = 6m' = 0, we should expect* /(^) to be constant

and to be another constant in the open range (0, tt).
in the
Fej^r's theorem.
now begin the
discussion of the theory of Fourier series by proving
the following theorem, due to Fejer+, concerning the summability of the
Fourier series associated with an arbitrary function, f{t):
Let f{t) he a function of the real variable

TT ^t< TT, and defined by the equation
f(t
*
lu point of fact
f(x)=-^Tr
f(x)
= îr
t Math. Ann. lviii. (1904), pp. 51-69.
27r)=f(t)
(-7r<:x<0);
(0<x<?r).
t,
defined arbitrarily luhen
170
oft; and
all other real values
for
let
f{t)dt
Fourier
the
at all points
And
the two limits
which
at
associated with the function
ser'ies
sum (CI)
its
Let
an improper
an, bn, (n
f(x
f (f)
is
summable* (CI)
0) exist.
is
i{/(*'
and
{if it is
ahsolutely convergent.
let it he
Then
and
-n
integral)
exist
[CHAP. IX
= 0,
2,
1,
+ 0)+/(^-0)}.
denote the Fourier constants (9*2) of /(^)
...)
let
= Ao,
âo
S An(x) = S^ni^)-
anCosnx + bnSinnx = An{x),
n=0
Then we have
to prove that
- {A, + S,{x) + S,{x) +
lira
...
^W
in-*-x
+ S,^_,{x)] = ^
provided that the limits on the right

If
we
exist.
substitute for the Fourier constants their values in the form of
integrals ( 9'2),
^0
{f{x + 0)+f{x -0)],
it is
easy to verify thatf
+ 2
>S (x)
= mAo + (m -l)A^(x) + (m -2) A.,{x) +
...
+ A,n-i {x)
n=\
/""
= -1
"j
[^m + {m -
1) cos
{x-t) + (m
- 2) cos 2{x-t)+
...
-:r
r sinîM^
- i_
~27rj_.
sin^
-27rj_.+,
// -x
1(^-0
-^
7,
-^^^
1(^-0
sin^
(m - l){x - 1)] fit) dt
cos
'
the last step following from the periodicity of the integrand.

If
now we
bisect the path of integration
the two parts of the path,
A+ 1
Consequently
1
=-
Sn (x)
r^^sin'^wi^
-^-^r^ f{x
it is sufficient to
_^.
,.
^x
write X for
e(^-')
+ 2d)de + that, as
prove
-/
,^
and write
a;
+ 2^ in place of
is
obvious that,
m + (m - 1)
(X
if
we
+ X-i) +
(hi
m^cc
f-^'sin-m^
then
..
^, ,^
2) (X2
in the second line, tlien
+ X-2) +
(X'-i
+ Xi-')
= (l-X)-i{\i- + X2-'+...+X-i + l-X--X'-i-...-X'}

= (1
in
-^r,' j{oc- 26) dO.
* See 8-43.
t It
we get
- X)-2 {xi-v - 2X + X'^+i}
= (X^'"
- X~^'"f /(X^ - X"^)^
.,
..
FOURIER SERIES
9 4]
Now,
if
we
integrate the equation

!
1 sin^ ni0
2 sin-
we
171
^m + {iu l)cos2^+
+ cos 2{m
...
1)6,
6*
find that
'^"sin^m^
and so we have
r^'^sin^m^
.
,.
^^/i 4>(6)(i6>0 as
1
(f)(0)
siTi'd
mx,
stands in turn for each of the two functions
fix + 26) -fix +

Now, given an
fix - 26) -fix - 0).

number e, we can choose
0),
arbitrary positive
(^
<
(^)
0<
6 ^^S.
This choice of B
sin-
m6
,^
whenever
^TTin,
to prove that
mJo
where
da
sin
S so that*
obviously independent of m.
is
Then
-1
fi""
-^^n
mJo sin-6
,^,
4>id)de
'
sin- m6
1 f^
f^sm-mO
,^
^
^
---^rQ- 4><6) d6+
^
mj sin-6
.
ri*sin-?^
1
711
f?^
,^.
,^
'.--4>i6)\d6
-r \ /
sin^f^
-^
..
\4>{6)
d6
ri"
...J \i>i6)\d6
sm- ^6 Jo '^^ ^'
-îjl"
/(O
[i"
/"i"
-.--
sm-6
i7re+
the convergence of
fi'^sin^m^
sm"^ ni6
f^"
-.^
msm^ôjji
msmHSJw'^
^r7^f^^+
})i\
Now
sin-^
sin-c^
fi'sin^/^^
^-6
=
-r-r^dc)+
m.'o
'o
//i'js
(f>(6)d6.
entails the convergence of
J -1T
<f>{6)
d6,
and
given
so,
(and therefore
h),
îrem
by taking
lh>
\'\<\>i6) dO,
Jo
sufficiently large,
I
-1
fi'^
\in J
e is
sin'
sufficiently large.
Hence, by taking
where
we can make
sin-??i^
sm- 6
we can make
//ix j/i
<7re,
4>i6)d6
^
.
an arbitrary positive number
that
is to say,
a limit,
môo
and
so Fejer's
theorem
*
is
'
sm- e
established.
On
the assumption
that/(x0)
exist.
from the definition of
172
[CHAP. IX
Let V and L be the upper and lower bounds o{f{t)

Corollary 1.
whose length does not exceed 77, and let
P
Then,
if
a +r) ^.v^b-rj, where
;;
in
any interval
{a, b)
\f{t)\dt = nA.
any positive number, we have
is
^-^(^^^^^
^-.-^rMx-'"^ = 2T-^{j-... + j.-,^j..J sinH(i-0^
iP-
^/-^-|-sinH>(^-0
./a;+J sin^ ^ (x -
r2rw7rij_^+.,.
-^
^-^
so that
i 1^0+"'i'
U \+U}l{m sin2 i;,}.
Û-^{\
'S' (.x-)|
Similarly
{ô+'î
'^"
^^^]
>L-{\L'H^]l{r>i
sinH^}.
Corollary 2.
Let/(^) be continuous in the interval a^t^b. Since continuity implies
uniformity of continuity ( 3-61), the choice of S corresponding to any value of x in (a, b)
is independent of x, and the upper bound of |/(.r0) |, i.e. of \f{x) |, is also independent
of X, so that
/
'"
j
(^(^)
d6=
i
and the upper bound of the
Hence the choice of
x,
and
sinînd
x.
...
J-
,. (^{6)d6\ <7re,
sm^ 6 ^
consequently
uniformly throughout the interval a
"'-1
n=\
\A(^+
in
m--a:
independent of
last expression is
mJ
independent of
f{xU)-f{xQ)\de
which makes
??z,
"i"2 p.
is
['"
Jo'
I
Jo
2 S.n{x)\
tends
to
the
limit f{x), as
^x%b.
The Riemann-Lehesgue lemmas.
9"41.
In order to be able to apply Hardy's theorem

vergence of Fourier series from Fejer's theorem,
( 8"5) to
we need
deduce the con-
the two following
lemmas
(I)
Let
-x^
(
{6)
dd
exist
and
it
{if
an improper
is
integral)
let
it
he
Then, as
X-^ x>
yl/{e)sm{\e)de
(II)
If, further,
i/r
is
0(1).
(6) Juts limited total fluctuation in the range {a, h) then,
as X><X)
J a
/
yfr
(6) sin (XO)
de
is
0{1/X).
FOURIER SERIES
9-41]
173
Of these results (I) was stated by W. R. Hamilton* and by Riemannt in the case of
bounded fiuictions. The truth of (II) seems to have been well known before its importance
was realised
it is a generalisation of a result established by Dirksenf and Stokes
;
(see 9 '3) in the case of functions
The reader should
with a continuous differential
coefficient.
obser\'e that the analysis of this section
by
sines are replaced throughout
remains valid when the
cosines.
lemma first in the case in which

bounded
in
the
range
(a,
b).
In
this
case, let
be the upper bound
(0)
of r\fr{d)\, and let e be an arbitrary positive number. Divide the range (a, 6)
into n parts by the points x,, x.2, ... a'n-i, and form the sums Sn, Sn associated
with the function i/r (0) after the manner of 41. Take n so large that
It is convenient^ to establish this
(I)
is
yjr
^n Sn<e;
this is possible since
In the interval
integrable.
is
[6)
-yjr
write
(a-v_i, a^,)
y}r{d)
SO that
+ co,(e),
Ur L,-,
ylr,(xr-,)
\(0r(6)\
<z
where Ur and Lr are the upper and lower bounds of -^{6)
in the interval
yXfi, Xf).
It is then clear that

6
yjr
(6) sin
(\d)de
J a
= t
i/r,.
sm{\d)(W+
(av-i)
{$) sin (kO)
(o,
:i
dO
r=l
sm{\0)dd + S
O)r(O)\d0
r=l
Xr-i
<{2nK/X) + .
By taking X
may be made
less
2e, so
has been chosen),
that
= O,
this is the result stated.
When
4'5,
yfr{6) is
we may
* Trans.
unbounded,
(1906), Ch.
if it
has an absolutely convergent integral, by
enclose the points at which
Dublin Acad.
unbounded
in a finite
|j
number
For Lebesgue's investigation
see his Series trigonometriques
III.
X Journal fUr Math. iv. (1829), p. 172.

For this proof we are indebted to Mr
Hardy
other writers, e.g. de la Vallee Poussin, Cours
The
it is
xix. (1843), p. 267.
t Ges. Math. IVerke, p. 241.
li
than
ylr(0)sm(xe)dd
lim
K-*-ac J
and
remaining fixed after
sufficiently large (n
the last expression
finiteness of the
integral, 4-5.
number
it
seems to be neater than the proofs given by
cV Analyse Infinitesiniale,
of intervals
is
assumed
ii.
(1912), pp. 140-141.
in the definition of
an improper
174
of intervals
Sj, So, ...
[CHAP. IX
Bp such that
d6<.
If
denote the upper bound of
and
intervals,
to 8^, Bo,
...
we may prove
8p
=\t
J a
^\^X
I
Now
if
yjr
the choice of
less
(6)
V,.
f
= lJ Sr
\ylr(d)sm{\d)ide
26.
n and K, so that the last expression
e fixes
than 3e by taking
'
S,-
f{e)sin{Xe)de + S
r=l.
sufficiently large.
X.
That
may be
to say that, even
is
be unbounded,
lim [
J
A.-.oo
provided that
The
ylr{e)sm{\e)de\
r=lJ
< {2nKj\) +
made
ylr(e)sm{\e)d0+ t
\r=lJyr
'
which
as before that
?3+i r
rb
ylr{e)sm{\e)de
outside these
7^+1 denote the portions of the interval (a, b)
if 71, 70, ...
do not belong
for values of
\ylr(6)\
first
(II)
example
yjr
When
2,
{6} has
lemma
yjr
f{e)sm(ke)dd = 0,
an (improper) integral which
is
therefore completely proved.
is
{6) has limited total fluctuation in the range (a,
we may
b),
by
3'64
write
^^W = %:(^)-%2(^),
where Xi
(^)>
%2 (^) are positive increasing
bounded
Then, by the second mean-value theorem

that a-^^^b and
functions.
number ^
4"14) a
f Yi (6) sin (XO) dd\ = \xAb)l' sin (Xd) dO
exists such
^2xAb)IX.
If
we
treat
yjr
X2.
(^) in ^ similar
(d) sin (XO)
d6
manner,
it
follows that
X, (0) sin (Xd)
dd +
X',{0)^\n{xe)dd-
^2te(6) + ;^,(6)}/X
=
and the second lemma
Corollary.
If f{t)
is
0(1/X),
established.
such that
lie
f(t)
exists
and
is
integral, the Fourier constants a, bn oi f{t) are o(l) as n--oo
limited total fluctuation in the range (-tt,
it),
;
and
if,
further, /(<) has
the Fourier constants are 0{l/n).
[Of course these results are not sufl&cient to ensure the convergence of the Fourier
with/(i) for a series, in which the terms are of the order of magnitude
series associated
of the terms in the
harmonic
series ( 2'3), is not necessarily convergent.]
FOURIER SERIES
9*42]
The proof of Fourier's theorem.
942.
We
175
now prove
shall
the theorem enunciated in
ir^tKir, and
Let f(t) be a function defined arbitrarily luhen

the equation
exist
and
f{t +
it is
(if
=f{t) for
27r)
all other real values
an improper integral)
namely
9*2,
oft; and
let
defined by
fit) dt
let it be
Let an, bn be defined by the equations
7ran=
f{t) cos ntdt,
7rb,i=
f(t)sinntdt.
J TT
Then, if OS be an interior point of any interval
(a, b) luithin tuhich
f(t) has
limited total fluctuation, the series

00
âo+ 2
is
convergent
and
its
sum
is
^ [f{x + 0)
It is convenient to give
two
+f{x 0)}.
one applicable to functions for which
proofs,
permissible to take the interval
it is
+ 6,1 sin ?ia;)
cos n^
(tt,j
(o, b) to
the other applicable to functions for which
When
(I)
from
the interval
9'41 (II)
theorem
( 9"4)
(C'l) is* ^
( 8'5),
(II)
tt
a-,
it
-\-
x),
not permissible.
may be taken to be ( tt +
+ bn sin nx is 0(1/??) as n> oo
(a, b)
that a cos nx
a?,
+ x), it follows
Now by Fejer's
tt
is summable (Cl) and its sum

by Hardy's convergence theorem
convergent and its sum (by 8"43)
the series under consideration
{/(^ + 0) +f{x 0)|.
Therefore.,
the series under consideration
isif/(*'
be the interval (
it is
is
+ 0)+/(.r-0){.
Even
if it
whole interval (
not permissible to take the interval
is
tt
o;,
tt
+ a;),
it
is
(a, b) to
by hypothesis,
possible,
be the
to choose a
number S, less than tt, such that/(^) has limited total fluctuation in
h, x h). We now define an auxiliary function g (t), which
equal to f(t) when xB%t^x + 8, and which is equal to zero throughout
positive
the interval {x
is
-\-
the rest of the interval {ir

Then g
(t)
x, tt
+ x);
+ 27r)
and g(t
is .to
be equal to g
satisfies
the
conditions postulated for the functions under
namely that
has an
integral which
consideration in
(I),
convergent and
has limited total fluctuation in the interval (tt
and
it
(t)
t.
it
is
absolutely
+ x,
tt
+ x);
denote the Fourier constants of git), the arguments used

prove that the Fourier series associated with g (t), namely
so, if a,^"*, 6"'
in (I)
cto"'
+ S
('" COS
nx
+ 6*^'
sin nx),
n=l
is
convergent and has the
sum
^ [g (x
1 {/(^
*
The
0)
+g
(x
0)},
and
0) +/(.. -0)}.
limits /(.r0) exist, by 3-64
example
3.
this is equal to
176
Now
let
Sm{^) and
denote the sums of the
Sm^^^ (x)
the Fourier series associated with f{t) and

easily seen that
Sm (^) =
first
m+1
...
terms of
Then
respectively.
(t)
c/
+ cos {x-t) + cos'2{a;-t) +
{2
[cHAP. IX
it
is
+ cos vi (x - t)} f{t) dt
~27rJ_/ ^m^{x-t) -^^^^"^^

_1_ /-+- sm(m + ^)(x-t)
IT
sm^
Jo
by steps analogous
In
e
= -1
(1)/
/^"
so,
to those given in
sin(2t
sm
Jo
-^
d'4>.
+ l) ^ g{x+2e)d6
-3^
sm
TT J
and
IT
manner
like
'Sfm'''(^)
^
*^
17
using the definition of g
- S,>
S>n (x)
(x)
=-
{t),
sin
W-5+
|"i'^sin(2m
+ TT
Sin
1)^
om^/a
g{x-2e)dd,
,
t/
we have
(2m +
1)6'^M^^ dd
smt/
TTJjs
+ - \'\m{2m+\)d-^^''~l^'> dd.
TT J ks
smu
Since cosec ^
f{x
and
is
a continuous function in the range {^8,
so,
by the Riemann-Lebesgue lemma of 9*41
right in the last equation tend to zero as
That
to say
is
Hence, since
it
^tt), it follows
that
20) cosec 6 are integrable functions with absolutely convergent integrals;
lim
{>S^^
{x)
(I),
both the integrals on the
m>oo
S,J'^ (x)]
0.
+ 0) +f(x 0)|,
lim S^^^^ (x)
=^
lim>Sf^(^)
=l{/(^ + 0)+/(.^'-0)|.
{f(x
follows also that
We
have therefore proved that the Fourier series associated
namely â^
+ 2 (a,iCos nx + 6 sin
i{/(^-
9"43.
It is of
is
its
sum
is
+ 0)+/(^-0)}.
The DiricJdet- Bonnet proof of Fourier s theorem.
making use
which
and
nx), is convergent
luith f{t),
some
interest to prove directly the
of the theory of summability
on the same general
theorem of
lines as the proofs
due
9*42,
without
we now
give a proof
to Dirichlet
and Bonnet.
accordingly
177
FOURIER SERIES
9-43]
As usual we denote the sum of the first m + 1 terms of the Fourier

by Sm {(jc), and then, by the analysis of 9'42, we have
.,
ri''sin(2Hi
l)6^
+ l)6'
r^'^sin(2m
Sin f
TT j
sm
TT J
series
om
^zj
(7
Again, on integrating the equation

sin
(2m +\)6
sin 6
=1+2
cos 2^
5''siu(2m
we have
^^^^^
+l)^
dd =
-x
sin^
4^
2 cos
+ 2 cos 2i 6,
l'n-,
^
so that
1 ri'^sin(2m + l)^
0)+/(^--0)}=i/^'^^^^^^
>S;(^)-H/(-^ +
1*'^
sin(2m
l)^
.
2^^
_ ^,^^.
^^y^^
In order to prove that

lim
it is
(.r)
<S',
\ {/(./
0)
^f{x - 0)),
therefore sufficient to prove that
where ^ (^) stands in turn
f{x
20)
for
each of the functions
-fix +
Now, by 3"64 example
4,
0),
f{x -
6 <t) (6) cosec 6
fluctuation in an interval of which ^
26)
is
is
-fix -
0).
a funcôn with limited total
an end-point*; and so we
may
write
^ <^(^) cosec ^
where
Xi C^)-
= Xi(^)-X.(^),
bounded positive increasing functions of 6 such that
%2 (6) are
X^(+0) = xA+0) = 0.
Hence, given an arbitrary positive number e, we can choose a
number 5 such that
whenever
We
positive
h8.
now obtain
inequalities satisfied
by the three integrals on the right
of the obvious equation
l)^
+
..^
6(6)
^
sm6
r^'^sin(2m
^-^
Jo
-,
+
*
f^^
,.
(16=1
.'
/
x
sm(2i
+ il)6
ks
zj
<^(^)
..
sin
jzj
a6
r>
psin(2m
,, ,^
^
'xA^^dd-j^
(^'sm{2m + l)6
The other end-point
is
^(b-x)
or d ^(x -a), according as
+ l)^
(p
(0)
,^,,.
X2(^)d6.
represents one or
other of the two functions.
W. M.
A.
12
178
The modulus
of the
sufficiently large
made
integral can be
first
from
this follows
9*41
[CHAP. IX
less
since
(i)
</>
than
by taking
(6) cosec 6 has an
integral which converges absolutely in the interval {hS, Itt).
Next, from the second mean-value theorem,
and
number ^ between
isin(2i
l)6'
isin(2m
= N:i (i S)
dd
X, (0)
+ l)^
'
/('+^)sinw
sin
Since
bound*
which
is
[i^
N
I
On
,-,,.,,
,,
independent of
^^ +
sin
(2m
and
/3,
dd
P sin u du
dt is convergent, it follows that
is
such that
follows that there
it
has an upper
then clear that
it is
1)6
'xi(0)d ^2Bxi(i8)<2B6.
treating the third integral in a similar manner,
we
see that
we can
make
+ l)^
^j^^^ .#.() c<(4B
^'^sin(2??i
f
Jo.
by taking
sufficiently large
and
so ive have
|-*''în(2m
sin (2m
lim
...
...
But
it
sm
sin
^
.00-0
'
has been seen that this
1).
proved that
l)^
+ 1)^
f/
4-
^^^p
/
a sufficient condition for the limit of
is
Sm
(oc)
be i {f(oo + 0) + f(x 0)] and we have therefore established the convergence of a Fourier series in the circumstances enunciated in 9'42.
to
NoTK.
The reader should observe that
in either proof of the convergence of a Fourier

mean-value theorem is required but to prove the summability of the
series, the first mean -value theorem is adequate.
It should also be observed that, while
restrictions are laid upon /{() throughout the range ( tt, tt) in establishing the summability/
at any point .r, the only additional restriction necessary to ensure convergence is a restriction on the behaviour of the function in the immediate neighbourhood of the point x.
The fact that the convergence depends only on the behaviour of the function in the
immediate neighbourhood of x (provided that the function has an integral which is
series the second
absolutely convergent) was noticed by
Riemann and has been emphasised by Lebesgue,
Series Trigonometriques^ p: 60.

It is obvious that the condition t that
in
which f{t) has limited
vergence of the Fourier series
and
,.
lim
The reader
will find
it
/'i'^
I
m^xjo
*
x should be an
total fluctuation is
it
may
merely a
interior point of
sufficient
be replaced by any condition which makes
sin(2TO-f 1)(9
^sm(9;r^(b(d)dd=0.
,
...
,-
interesting to prove that
,,
B=
Due
to
an interval
condition for the con-
Jordan, Comptes Rendus, xcii. (1881),
p. 228.
du = hTr.
FOURIER SERIES
9*44]
Jordan's condition
179
however, a natural modification of the Dirichlet condition that
is,
number
the function /(^) should have only a finite

not increase the difficulty of the proof.
Another condition with the same

namely that, if
efifect
of
maxima and minima, and
due to Dini, Sopra
is
le
does
it
Serie di Foiirier
(Pisa, 1880),
*(^) = {/(:p + 2^)+/(.r-2^)-/(ar + 0)-/(a.'-0)}/^,

then
* {&)dQ should converge
absolutely for
[If the condition is satisfied, given e
we can
some
positive value of
a.
find S so that
/
and then
<\fTi
sni
Jo
^^-^
the proof that
(f)(6)
dBl
<
for sufficiently large values of
follows
from the Riemann-Lebesgue lemma.]
more stringent condition than Dini's is due to Lipschitz, Journal f>ir Math.
namely
{B)\< 06", where C and k are positive and independent of
(1864), p. 296,
(f)
LXiil.
6.
For other conditions due to Lebesgue and to de la Vallee Poussin, see the latter's
d' Analyse In/lnitesimale, ii. (1912), pp. 149-150.
It should be noticed that Jordan's
condition differs in character from Dini's condition
the latter is a condition that the
series may converge at a point, the former that the .series may converge throughout an
Cours
interval.
9'44.
The xmiformity of
the convergence
of Fourier
Let f{t) satisfy the conditions enunciated in

having limited total fluctuation)
series associated
a + b^x
Let h
^b
{t)
b.,
with
convei'ges
(t)
where 8
is
uniformly
in
to
and further let it be continuous

an interval (a, b). Then the Fourier
9'42,
!^
(in addition to
series.
sum f{x)
the
be an auxiliary function defined to be equal to/(^) when
to zero for other values of
Also
constants oi hit).
^ '
series associated
Then, by
with h
.v
for which
a^t ^b
and equal
at all points
any positive number.
in the range (-tt,
S TO i.v)
let
denote the
sum
tt),
and
let a,
/3,i
m+1
of the first
'
denote the Fourier
terms of the Fourier
{t).
9"4 corollary 2, it follows that \af^-\-
summable throughout the
interval (o +S,
i
b-h)
a cos 1lX-\-^n sin
(a cos
;.i'
/3,i
sin j7) is xmiformly
and since
nx $ {an-^^n'r-,
\
which
is
independent of x and which, by
9*41
(ii),
nx + ^n
sin n.v)
is
from
{1/n), it follows
8-5
corollary that
âo+ 2
(an cos
n-l
converges uniformly to the
Now, as
sum
h{x),
which
is
equal
to/(.i').
in 9'42,
J hib-x)
smO
'
'
7rj^(^_a)
sm^
'
"
122
'
[CHAP. IX
180
As
we choose an arbitrary
in 9'41
positive
number
and then enclose the points at

p
which f{t)
If
unbounded
is
in a set of intervals Sj, 80,
K be the upper bound of >f{t)

S, {^)
i
... 8,,
outside these intervtils,
- S[^l {^)
<
(^1^^^
+ 2e)
uniformly,
as in 9-41,
of the function /(<).
S^{x) S^'^^{x) tends uniformly to
<e.
8,
Hence,
Since Sj{x)-^f(x)
zero.
then obvious that S^{x)-*~f{x) uniformly; and this
it is
\f{i) \dt
J Sr
we then have,
cosec
where the choice of n depends only on a and b and the form

by a choice of m mdependent of x we can make
arbitrarily small; so that
such that 2
r=l
is
the result to be
proved.
must be observed that no general statement can be made about uniformity

Thus the series of 9* 11 example 1
converges uniformly except near .r = (2 + l) n but converges absolutely only when x=nir,
Note.
It
or absoluteness of convergence of Fourier series.
whereas the series of

values of
9'11
example 2 converges uniformly and absolutely
for all real
x.
Example
{6) satisfies suitable
If
1.
conditions in the range
Example
Prove that,
2.
if
shew that
[''''2}L(^:^<i.in-e)d0
lim
> 0,
'
hm
(0, tt),
;'"
Siu(2+1)^
-a
7A
.^.^
= \Tr coth
e- aea de
hair.
^
(Math. Trip. 1894.)
[Shew that
^
e~^^
d6= hm
SI" d
sm
w-*-* J
sm 6
^ sin(27i+l)^
sin 6
and use example
Example
3.
e-a^dd
.],
,n^-^ J
e-a&dd
1
e-af'
l.J
Discuss the uniformity of the convergence of Fourier series by means of
the Dirichlet-Bonnct integrals, without making use of the theory of summability.

9"5.
The Hurwitz-Liajjonnoff*
tlieorein
Let f{dc) he hounded in the interval (
concerning Fourier constants.

tt,
tt)
and
let
*
it
Math. Ann.
(1898), p. 1024.
exist, so
Liapounoff discovered the theorem in 1896 and published

Math. Soc. of the Univ. of Kharkov. See Comptes Rendw, cxxvi.
lvii. (1903), p. 429.
in the Proceedings of the
f{x) dx
FOURIER SERIES
9-5]
that the Fourier constants , ^n
off{x)
â^+ i
181
Then
exist.
the series
+ h,^)
{an^
n=l
is
convergent
and
its
1 t"
is* -
sum
she^m
It will first be
fix)Ydx.
that, with the notation of S 9"4,
Sn(x)\ dx = 0.
\f{x)-- 1
lim
Divide the interval ( tt, tt) into 4r parts, each of length 8 let the upper and lower
bounds of /(.v) in the interval {{2p~l) 8-it, (2p + 'S)8 7r} be Up, Lp, and let the upper
bound of \f{x)\ in the interval ( - tt, tt) be K. Then, by 94 corollary 1,
;
"t-i
2 S^{x)
\f{x)-">
when X
lies
between 2pb and
Consequently, by the
fir
\f{^)--
(2/?
first
<
^-Zp + 2^/{OTsin2^8}
<
2A'[l
n=0
+ l/{wsin2^S}],
+ 2) S.
mean-value theorem,
w-l
12
2 >%{x)\
dx<2K\\ +
2r-l
4/f;.
{Up-Lp)^~^\
r!^T-.N2a 2
Since /(j;) satisfies the Rieraann condition of integrability
( 4'12), it follows
that both
r-l
r-l
48 2 {U2p-L2p) and 48 2 ((7.,p+i~

sufficiently large value.
When
^2iJ + i)
can be made arbitrarily small by giving
r (and therefore also 8)
has been given such a value, we
may choose mj so large that ?/{)! sin^ ^8} is arbitrarily small. That is to say, we can
make the expression on the right of the last iiiecpiality arbitrarily small by giving m any
value greater than a determinate value
Hence the expression on the
nti.
left
of the
inequality tends to zero as wi-s-jo.
But evidently
/(^)-- % Sn(x)\dx
=
r \f{x)-t''Â,,{x)Xdx
m1
ni
-1
Vj
)j
/(^)- S 4(^)+ S -^(^) dx

\f(x)- X An{x)\ dx+
JJ
/it
f(x)
J
\f{x)
/
n=0
- 1 An {x)\
A
4-12 example
[n = ofn
1.
dx +
^^t_
ill,
H=A
(m-l
- S An (x)\ At
*.
This integral exists by
-n
-TT
m-l
+2
1 -An(x)\ dx
.'
A,,{x)\dx
{n=--0
n^ (an'
K^
^-t
proof of the theorem has been given by de la
Valine Poussin, in which the sole restrictions on /(.r) are that the (improper) integrals of f {x)
and {f(x)}'^ exist in the interval {-ir, tt). See his Cours d' Analyse Infinitesimale, ii. (1912),
pp. 165-166.
182
/"T
,'n
since
f{w)Ar(x)dx=\
when
= 0,
-n
1, 2, ...
hn-l
]
-n
[CHAP. IX
S An
(.
"I
(i^'n
(x)
dx
m1.
Since the original integral tends to zero and since

equal to the sura of two positive expressions,
expressions tends to zero
that
it
has been proved
it
follows that each of these
to say
is
\f{x)- 2 An{x)\ dx-*0.
Now
the expression on the
left is
r {f{x)Ydx-2r
equal to
\f(x)-'''SA,,,{Â.\'''i'A,,{x)\dx
rn-
/'tt
=
-n
'
.'
rir
{
-n
m>oo
so that, as
This
is
-TT
[h =
/(^)P dx-irl-
tto"
")
d^
^,i(*')t
)
+ 2
(a,r
+ h,i)
n=\
(."'
\
)
the theorem stated.

l{f(x), F{x) both satisfy the conditions laid on/(.r)
ParsevaVs theorem*.
Corollary.
^, B^ be the Fourier constants of F{x),

by subtracting the pair of equations which may be combined in the one form
at the beginning of this section,
f"
J -TT
and
if
it
follows
{f{x)F{x)}^dx = n\\{a,A,f+ I {{a,,A,,f + {KB,,)^}\

=1
L
J
that
/ {x) F {x) rfx = TT
9"6.
c?
W-1
(^
in=0
I'm
TT
\f(x)Ydx-
"la
2 ^()^
J -TT
im-l
^ (a^ + 6 5,0 \
âo J
Riemann's theory of trigonometrical
The theory
series.
of Dirichlet concerning Fourier series
is
devoted to series
Important advances in the theory were

made by Riemarin, who considered properties of functions defined by a series
which represent given functions.
the
of
lim
{(In
typef ^ô+ S
cos
nx +
(a,i
6 sin nx)
Riemann's theorem that

;}:
Mem. par
divers senium,
i.
cos
= 0.
if
wa'
We
6 sin
/ia-),
where
it
is
assumed that
shall give the propositions leading
two trigonometrical
(1805), pp. 639-648.
bility of integrating the trigonometrical series
series converge
Parseval, of course,
up
and are equal
assumed the permissi-
term-by-term.
t Throughout 9-6-9-6i52 the letters , 6 do not necessarily denote Fourier constants.
The proof given
is
due to G. Cantor, Journal
fiir
Math,
to
lxxii. (1870), pp. 130-142.
TRIGONOMETRICAL SERIES
9-6, 9 '61]
at all points of the range
number
with the possible exception of a
two
of points, corresponding coefficients in the
finite
series are equal.
Rieinanns associated function.
9'61.
Let the sum of the series
any point x where
at
tt)
tt,
183
rrOa
^
+ 2
(a,i
cos
+ hn sin
nx
nx)
= A^-\- S An {x),
n-\
n=l
converges, be denoted by /(a;).
it
F{x)
Let
= \a,x'- i
n-Ân{x).
Then, if the series defining f{x) converges at all points of any finite interval,
the series defining F{x) converges for all real values of x.
To obtain
this result
Cantor's lemma*.
If lim
we need the following
Lemma
J {x) =
x such
for all values of
due to Cantor
that a
^x ^b,
then o^-0,
^. 0.
For take two points
x, x-i-8 of
Then, given
the interval.
we can
e,
find
?io
such thatt,
n> n
when
<
a cos nx + b sin nx
\a cos
e,
{x + 8)
+ 6 sin n {x + S)
<
e.
Therefore
ax + 6 sin nx) + sin n8
cos nb (a cos
I
Since
it
follows that
and
it is
obvious that
sin nx + 6 cos nx) < e.

\
cos n8 (a cos ?; + b sin nx)
n8
sin
7i8
(a cos
nx + bn sin nx)
<
< 2f,
< 2c.
6,
a sin nx + bn cos nx)
.sin
Therefore, squaring and adding,
(H62)*! siii8! <2ev'2.
Now
it
have not the unique limit
hypothesis involves a contradiction. For, by this hypothesis,
exists such that there is an unending increasing sequence /ii,
suppose that
,,
will be
shewn that
sotne positive
n.,, ...
this
number
e^
of values of n, for
which
Now
let
Take n{ the smallest

all its
/j of
the range of values of S be called the interval
>
length L^ on the real axis.
27r then sin /i/// goes through

of the integers ?v sch that Hi Zi
1 V2;
call I., that sub-interval; of /j in which .sin ^j'y
I^
phases in the interval
= Zo.
>
the smallest of the integers nr{>)ii) such that

n'Zo >27r, so that sin n^'i/ goes through all its phases in the interval I2; call I3 that .subWe thus get a
l/\^2; its length is 7r/(2n2') = -^3.
interval + of L in which sin ?j2'y
sequence of decreasing intervals Zj, /.,, ... each contained in all the previous ones. It is
its
length
is
7r/(2H/)
Next take
iio
>
obvious from the definition of an irrational

is
number that
not outside any of these intervals, and sin a^l/^/2
For these values of

*
n, {a,'^
Riemann appears
to
+ b,c) ^
sin na
> 2q ^1%.
there
a certain point a which

^2', ... (.'r
iv. (1871),
The value
X If there
is
of
/(^
depends on
.r
and on
more than one such
i>
The proof here given
pp. 139-143, and Journal
fiir
(1870), pp. 130-138.
'^/)-
But it has been shewn that corresponding
have regarded this result as obvious.
modification of Cantor's proof, Math. Ann.
is
when w=/,
5.
sub-interval, take that
which
lies
on the
left.
Math,
is
lxxii.
184
numbers
to given
since
and
we can
find
such that when n
?Jo
>
.o,
[CHAP. IX
{a
J + 6^) ^
< 2e ^f2
(sin na)
values of / are greater than n^, the required contradiction has been obtained,
some
may
because we
take
e<
e,,
Assuming that the
therefore Un-Ô, 5-*-0.
series defining /(a:) converges at all points of a certain
interval of the real axis,

real values of x,
we have
an cos nx
Then,
just seen that a,i>0, 6^,>0.
hn sin
+ 6,f)2-^0, and
nx ^ (a,/
\
by
so,
for all
the
3"84<,
00
series
Â^x- S n~-An{x) = F{x) converges
real values oi
therefore, ( 3"32),
F{x)
continuous for
is
all
real values of
all
^c.
Properties of Rienianns associated function; Rieiiianns first lemma.
9'62.
It is
absolutely and uniformly for
now
lemma
possible to prove Riemann's first
G {x,
a)
that if
- 2F {x)
= F{x-\-'2a) + F{x- 2a)

4a^
S An{x)
then limG(*', a.)=f(x), provided that
converges for the value of x
o-*-0
F{x
Since the series defining F{x),
them
rearrange
+ 2a) +
cos n (x
sin
it is
2a) converge absolutely,
we may
and, observing that
n {x
2a)
cos n (x
+ sin
n (x
2a)
2 cos nx
2a) 2 sin
nx
= 4 sin^ na cos nx,

= 4 sin^ na sin
nx,
evident that
/-,
/
\
G(x,a)
= Ao+ V
1
/sinna\''^
M=i
now be shewn that
It will
An{x).
na
this series converges uniformly with regard to

00
a for
is
all
values of
a,
S An{x)
provided that
then an immediate consequence of
and/
(0)
= 1,
then
/"() is
To prove that the
a,
continuous for
and therefore, by
that
X
n=\
'
fn+\ (a)
In fact*,
if s
be the integer such that
s-l
G (x,
/ + 1
*
(a)
-/. (a)
a,
^
j
and
(a
^ 0),
so G{x, a)
= lim G (x,
is
a).
converges uniformly, we employ
K
j
is
independent of
<(*
+ 1)
s-l
0)
result required
The expression corresponding to &> (x)

^1 it is therefore sufficient to shew
2.
where
i^,
f^ (a)
values of
all
a)
obvious that |/n(a)j
A() <
for, if
3*2,
G (x,
series defining
the test given in 3"35 example

is fn{oi), a-nd it is
3'32
The
converges.
2 (/
(a)
Since x~^ sin x decreases as
-/
.r
^. ,
(a))
sii2
SUi^
a.
when
a
o=t=0
Sin-.a
increases from
to w.
we have
9-62,
186
lim la-i [F{x^- 2a)
But
lim
(x)a
.4o
+ F{x -
+ ^{7r-a.)A, (x) + 2
2a)
[CHAP. IX
2F(.x)]
g^ (a) [An+i (x)
- An ix)}
w=l
and
this limit is the value of the function
value
zero since lim
is
Two
= 0.
when
must have corresponding
3"32
and
this
lim.
and are equal at all jôints of

a finite number of points,
the possible exception of
witli
tt),
tt,
by
see that lim
trigonometrical series.
trigonometrical series which converge
range (
= 0,
By symmetry we
Riemanns theorem* on
9'63.
the
An(x)
coefficients equal.
An
immediate deduction from this theorem is that a function of the type considered
be represented bv auy trigonometrical series in the range ( tt, tt) other
than the Fourier series. This fact was first noticed by Du Bois Reymond.
in 9'42 cannot
We observe
that
it is
certainly possible to have other expansions of (say) the form

oq
(a,
cos i rax + /3, sin ^ rax),
which represent /(a-) between - tt and tt for write .r=2^, and consider a function c^ (^),
- ^tt,
which is such that (^ (^) =/(2|) when -\ir <^<\tv, and
(|) =g (|) when - tt < | <
and when ^tt < ^ < tt, where g (^) is any function satisfying the conditions of 9-43.
Then if we expand (^ (|) in a Fourier series of the form
;
0,1
m^ + /3i cos m^),
2 (a^ cos
when - Tr<x<ir and clearly by choosing the function g (|)

ways an unlimited number of such expansions can be obtained.
this expansion represents /(.r)
in different
The question now at issue is, whether other series- proceeding in sines and cosines of
x exist, which differ from Fourier's expansion and yet represent / (.r)
integral multiples of
between
tt and
tt.
If possible, let there be
and
conditions,
two trigonometrical
let their difference
series satisfying the given
be the trigonometrical series
Ao+ t An{x)=f(x).
=1
Then f(x) =
exceptions
let
F (x)
at all points of the range

let î
if
Schwartz'
f (x) =
For
if
the value of F{x)
lemma f.
In
the
when ^j<
We
x<
finite
number
=1
or
if
<9=
of
proceed to establish a
fo-
range î<x<^.^, F{x)
is
a linear function of
-1
= e^F{x)-Fi^,)-l^^{F{i,)-F(i,)}y^,hHx-^,){^,-x)
a continuous function of x in the range
w^th a
tt)
in this range.
cl>{^)
is
( tt,
be a consecutive pair of these exceptional points, and
be Riemann's associated functiou.
lemma concerning
9"631.
ô
î
^j;
^ ^,,
^^nd
The proof we give is due to Gr. Cantor, Journal fit r Math,

Quoted by G. Cantor, Journal filr Math, i.xxii. (1870).
(î)
= (^ (^2) = 0.
lxxii. (1870), pp. 139-142.
x,
63-9 '632]
If the first
x=c
at which
trigonometrical series
term of
it is
not zero throughout the range* there will be some point

Choose the sign of $ so that the first term is positive at c,
(x) is
(f)
not zero.
and then choose h so small that

Since
cf)
(x) is
positive since
(f>
continuous
(c)
> 0.
Then, by Riemanifs
(c'l
+ a) ^
(f)
Let
(f>
(x)
we have
We
is
+ ) + <^(gi-)-20(<^ i)_^.
(ci-a) -^(p
(0|),
first
^1
must be negative or
SO this limit
term of
arrived at a contradiction.
a linear function of x in the range
9632.
upper bound ( 3"62), and this upper bound

attain its upper bound at Cj, so that Ci + |i Ci4=^2
</'(gi
Hence, by supposing that the

(^j, ^2)5
positive.
(c) is still
lemma,
first
(c']),
(f)
it "attains its
lini
But
187
not everywhere zero in the range
(x) is
Therefore
< x < ^2-
it
is
zero
lemma
^ li^J
zero.
and consequently F{x)
is
therefore proved.
is
Proof of Rieiiianns Theorem.
see that, in the circumstances under consideration, the curve y
= F {x)
represents a series of segments of straight lines, the beginning and end of
each line corresponding to an exceptional point

convergent,
is
x,
as F{x), being uniformly
these lines
But, by Riemann's second lemma, even
if
must be connected.
^ be an exceptional point,
aâ
Now
and
the fraction involved in this limit
is
the difference of the slopes of
the two segments which meet at that point whose abscissa
two segments are continuous

a single
line.
If then
we
is ^ therefore the
equation y = F{x) represents
cx-\-c, it follows that c and c' have
;
in direction, so the
write F{x)
the same values for all values of
Thus
^.
Â^x"^ ex
c'
= S
n~-A n (x),
=i
the right-hand side of this equation being periodic, with period
27r.
The left-hand
Hence
side of this equation
^4,,
(),
c' =
and
>i
Now
must therefore be
c
27r.
periodic, with period
= 0,
ir'-A,i(x).
=\
the right-hand side of this equation converges uniformly, so
multiply by cos nx or by sin nx and integrate.
This process gives

7rn~-a,i
= cj
cos
nxdx =
0,
sin
nxdx =
0.
7rn~^bn=
* If it is zero
throughout the range,
F (x)
is
a linear function of z.
we can
188
Therefore
This
the coefficients vanish, and therefore the two trigonometrical
all
whose difference
series
J.o+
is
follows from 9 "43 that, if
It
number
of discontinuities
range (
x x
lim'
corresponding coefficients equal.
Fourier s representation of a function hy an integral*.
9'7.
\=
S An(x) have
9"63.
the result stated in
is
+ l)(^-^) y(^) ^^ ^
siii(^^^^
Now let X
27n + 1 + 277
be continuous except at a
/ {ai)
11^ iTT^-^sin 6 [f{x
^ t; <
where
X{t- x) - sin {2m +

2 {cos (2i
so
that
(^
x)] (t
t;) (^
(t
- x)]
{sin
rj
- x)-' f (t) dt
{t
x)] {t
- x)-'f{t) dt
>x by
W)].
1.
J a.
??i
+ 26) +/(^ -
be any real number, and choose the integer
{sin
finite
and if it have limited total fluctuation in the

x be any internal point of the range ( a, /3),
then, if
),
Then
as
[CHAP. IX
9"41 (ii),
since
x)~'^
f (t) sin
7}
{t
x)
has limited total
fluctuation.
Consequently, fi'om the proof of the Riemann-Lebesgue

it is
obvious that
if
\f{t)' dt
and
and
-00
= h^{f(^ + ^)+f(^-^)]>
r-X ^^^^f(t)di
{t-x)
1"^
A-*-x
of 9"41,
1/(0 ^^ converge, thenf
Jo
lemma
so
lim
To obtain
cos u
(t
- x) du\
Fourier's result,
f(t) dt
we must
^tt
{f{x +
0)
+f(x - 0)}.
reverse the order of integration in
this repeated integral.
For any given value of \ and any arbitrary value of
number ^ such
e,
there exists a
that
r\f(t)idt<êl\J p
La
Theorie Analytique de
la
r
I
means
-X
to lim
p^x I'J -p
Chaleur, Cb.
the double limit
lim
p -
ix.
("^
cr
-*
If this limit exists, it is of
course equal
FOURIER SERIES
97]
writing cos
\r\
'
u(t
ir)
.f(t)
(f>(t,
a)
dul
I
^
dt-l
\t
-[
=
\l
[J
<
4>(t, u)
(fi
\-
u) dt\
{t,
dt
\j
(f>{t,ii)dtldu\
du
fi
j'^
u)\duldt+i''
u)dul
du
u) dt\
dul dt-\
'
{t,
(f>
{t,
<f)
\ i''\<t>(t,
du
u) dt]
(f>{t,
duldt+
a)
(f){t,
we have*
u),
(t,
(f>
189
(t>(t,
u)[dtdu
< 2\ ^\f{t)\dt<.
I
Since this
rk
/oo
=
J a
true for
is
r\ foD
;
Hence
all
/ Qo
r\
[f{x + 0)
Jo
i)
+/(./;
no matter how small, we infer that
e,
r\
similarly
I)
| ir
values of
{)
- 0)} =
lim
00
= /,
This result
known
JO.
for the function defined
(t
- oc)f{t) dtdu
-00
00
;"
COS u
{t
oe)f{t) dtdu.
.'
as Fourier s integral tlieorem\.
Verify Fourier's integral theorem dinx-tlj
Example.
(ii)
is
cos u
A-.
(i)
fur the function
by the equations
/(.r)=l,
{-\<x<\);
fix) = 0,
(|x|>l).
(Rayleigh.j
REFERENCES.
G. F. B.
RiEMANN,
Ges.
Math. Werke, pp. 213-250.
W. HoBSON, Functions of a Real
E.
H. Lebesgde, Lemons sur

C. J.
les Series
DE LA VallJê Poussix, Cours
1912), Ch.
Variable (1907), Ch.
Trigonometriques.
vii.
(Paris, 1906.)
d: Analyse Infinitesimale, ir.
(Louvain and Paris,
IV.
H. Burkhardt, Ena/clopddie der Math. WUg.
ii.
(Leipzig, 1914.)
(7).
G. A. Carse and G. Shearer, A course in Fourier's analysis and periodogram analysis

(Edinburgh Math. Tracts, No. 4, 1915).
The equation
is easilj' justified
by
4-.3,
by considering the ranges within
which /(x) is continuous.

t For a proof of the theorem when f(x] is subject to less stringent restrictions, see
Hobson, Functions of a Real Variable, %% 492-493. The reader should observe that, although
Um
A-*oo y
-00 7
exists, the repeated integral
7 -00
sin m
((
-x)
duj- /(<) dt
J
does not,
,,
the processes of analysis
190
[chap. ix
Obtain the expansions
1.
l-2rcos2 + r2
'
- log
(6)
- 2r cos z + r^) =
?'sinz
arc tan
(c)
^
\-r cos 2
2rsin2
and shew
that,
when
22
cos 2^ - r^ cos 3^
?:
_
/'^sin 3s
.
thej- are
02
convergent for
. .
values of
all
. .
+ ...,
= r sin z-\--r^, sin 3^

+ - r sm
r''
< 1,
3-
''^
1
= rsins + -?'^sin
.
arc tan
(a)
r cos J - ^
z in
certain strips
parallel to the real axis in the ^-plane.
Expand
2.
the value of the
and x
x^
sum
X ^
sin
for all values of
sin 2.r
+ -3 sin
3*'
-^
sin 4.r
. .
and hence
find
(Jesus, 1902.)
a.-.
3.
when - tt < .r < tt
in Fourier sine series valid
of the series
of
by 2 ~i
I'epresented
sin
sin^
/2a,
is
constant
n=\
(0
< X < 2a)
and zero
(2a
<x<
tt),
and draw a graph of the function.

(Pembroke, 1907.)
Find the cosine
4.
series representing /(.r)
where
/ (x) = sin X + cos X

/(.r) = sin.r-cos.r
Shew
5.
[x]
and
zero
is
6.
denotes 4-1 or
if
<
(in-
.^'
^ ^tt ),
^ .r < tt).
(Peterhouse, 1906.)
that
sin TTX +
where
(0
is
an
Shew that
+ + +
sin 3n-^
sinSjTA'
^^
sin
Ttt.^
r
= Jtt [x],
-,
...
according as the integer next inferior to x
is
even or uneven,
(Trinity, 1895.)
integer.
the expansions
1
2 cos -
log
= cos X- cos
COS X -
2x + - cos 3x
and
2
log
are valid for

7.
COS 2x - cos
3x
tt.
find the
8.
of x, except multiples of
all real A'alues
"(-)"*
cos m.r
~
and
.sin
1
\
^ = (cos X + cos o2x)Mlog /^2 cos -x\
-yT-.
range of values of
Prove that,
< <
if
a-
sin
a'^
+ Y' "^
.^'
for
which
27r,
then
2 sin
2x
a^+2-^
it is
Z sin
+-,:?; (sin
T
zx
applicable.
_
"2
tt
3.:*;
"a^T^^ +
'
'
"
cos x,
(Trinity, 1898.)
sinh a
'
sin x)
(tt
- x)
sinh aTr
'
(Trinity, 1895.)
FOURIER SERIES
Shew
9.
that between the values -tt and

2
sin
in.v
/ sin ^
.
r
\V-m''
=-
sin ran
IT
cos
mx = - sin ??l7r
.
e-mx^g-mx
X be a
Let
10.
Shew
^2
x the following expansions hold
o(
2 sin 2x
,
3 sin Zx
...
+ =
32-72
-l^-m^
m cos X - m cos 2.r m 5cos 3^'

^
-r^
^ +
P
22-77l2
32-J/l2
7rt-
??IC0S.
?/l
COS
and
real variable V)etween
and
1,
7?2C0S3.1-
2.r
n be an odd number
let
3.
that
(1)* = -
if
;;
\2m
TT
+n
191
is
not a multiple of -, where
+-
2 tan
cos2OTn-:r,
the gi-eatest integer contained in nx; but
s is
^1
0= - + -2^1
2 tan cos 2m7rx,
WITT
if jp is
11.
an integer multiple of
Shew
that the
sum
7r,=ii
(Berger.)
1/n.
of the series
X
^5
+ 47r ~
2 m~^ sin
^mn cos 2m7rx
m=1
is 1
when
12.
0<x<l,
and when fj<A'<l, and
If
shew
that,
when 1
=o
<x <\,
cosBttj:
s47r.r
cos 2irx+
22"
sin
8IU
2"^""*'^^"
\-i
i-
32
6nx
sin
Sin
47ra;
'iTTX
2/i
OTTX-
+
22"^2
-
= (-)"+!
sm27ra;+-j^;^^+-^;^^+.
^__,,
"32
"2-" +
2 + 1
.
(Trinity, 1901.)
a" V (x)
ae"'
e"
-1 when l<.r<|.
is
-,
,.
..
"*"
^^
.,
V,,,{x).
(Math. Trip. 1896.)

13.
If
7?i
is
an integer, shew that, for
cos2'a;=2
1.3.5...(2m2.4.6 ...2m
r,i
1) fl
(2
values of
all real
+l
cos 2x +
.v,
m{m-\)
-^ +~
(ot4-1)(7
cos 4x
2)
m{m-l){m - 2)
* (m l)(?tt 2)(w+3) cos 6.r + ...V
+
+
4 2.4.6...(27ft-2)
^_.
r.ns2m-l_;p|__
_
\__
^
_^
J^
,
1.3.5
14.
(1
-!
;;
...(2i-l) (2
f2m-l)(27n-3)
COS
2x+
(2to
any time
after the beginning of the
u
77
and that the
u,
at equal intervals
motion
ut
++V-
-;^C0S4a-+.
,^
+ ,,
1)(2h-3)
point moves in a straight line with a velocity which
receives constant increments, each equal to
at
2w-l
^
2/u + l
r.
and which
Prove that the velocity
is initially u,
is
2miit
sin
di-stance traversed is
2mTTt
lit
+ ^) + T7>-o32
2 -2 cos
(Trinity, 1894.)
192
[cHAP. IX
If
15.
f{x)= 2
ji=i '2n
- 3) j; - 2 2
sin (6?i
1
zn
71=1
3 v'3
-sm{2n-l)x
i
sin 5.r
sin ll^r
sin 7.r
/ + 0) =/ (tt - 0) = - ^tt,
shew that
/G^+0)-/(i7r-0)=-i7r, /(;:|7r+0)-/(37r-0)=*7r.
and
Observing that the
last series is
sin ^ (2ft-
TTsin
{2n-l)x
(Math. Trip. 1893.)
draw the graph of /(*)

16.
Shew
that,
^/
when
N
'f(,v)
^
.1
'^
< n,
<.r
2^/3/
'-
= sin
\
2x'
1
1
\
Ti
cos .r-- cos 0.^'+- cos /.<- cos 11.
r,'-f...
11
/
+ - sin
+ - sin 8.v + r5 sin
4.f
Find the sum of each
lO.r
where
/{^) = i^
(0<^<^7r),
f{x) =
(l7r<.r<g7r),
f{x)=-\v
series when .r = 0,
{%7T<X<1t).
\ir, qtt,
n,
and
for all other values of x.
(Trinity, 1908.)
17.
Prove that the locus represented by

2
^^
n=l
is
two systems of
area
''^"
'*''
^^^^
ny =
"
coordinate plane into squares of
lines at right angles, dividing the
(Math. Trip. 1895.)
tt'\
18.
Shew
that the equation
"
)"
~ ^ sin
represents the lines i/= mir,
semi-axes are
tt
and
(wi
ny cos nx _
n^
n=i
= 0,
1, 2,
...)
together with a set of arcs of ellipses whose
nlslZ, the arcs being placed in squares of area
19.
(Trinity, 1903.)
Shew
xyz =
Draw a diagram
27r"^.
of the locus.
Tr,
that,
if tlie
point {x,y,
z) lies
inside the octahedron
bounded by the planes
then
"^
,
sin 7ia;sin 7jwsin wj
^j-Y-^^
-^
=2-y(Math. Trip. 1904.)
20.
Circles of radius a are
of a regular hexagon of side a.
diawn having
their centres at the alternate angular jjoints
Sliew that the equation of the trefoil formed by the outer
arcs of the circles can be put in the form
nr- =
6s'3a
cos 3(9- -= cos 6^+

- cos9^+
22.4
8. 10
5.7
-;
^^^-^
...,
the initial line being taken to pass through the centre of one of the circles.
(Pembroke, 1902.)
FOURIER SERIES
21.
Draw the graph represented by
= 1 2m sm
.
where m
22.
193
rr
J1
[2
a=i
- )" COS nm6\

1 - {nm)-^
J
'
an integer.
is
(Jesus, 1908.)
With each vertex
formed by the
hexagon of side 2a as centre the arc of a
of a regular
radius 2a lying within the hexagon
Shew that the equation
drawn.
is
six arcs is
+ 2Si< -)-'^j-^f
^
f-=6-3,.i
4a
a=i {6n-l){bn+l)
co.6<>,
the prime vector bisecting a petal.

23.
Shew
circle of
of the figure
that, if
(Trinity, 1905.)
c> 0,
lim
'^
cot A" sin
{2ii
+ l)x dx=- n tanh -ctt.

.
(Trinity, 1894.)
24.
Shew
that
dx
r sin(2 +X l)^ l+x'^
lim
1
IT
Ooth
1.
sin
(King's, 1901.)
25.
Shew
when
that,
hm
< <1
.r
and a
is real,
/'sin(2n + l)^sin(H-.r)(9
^
-.
^^
sm^
;;
^.xjo
a^
-r,
+ 6-
sinha.r
smh a
,^
d6=
--TT
.,
(Math. Trip. 1905.)

26.
Assuming the
possibility of
the form Â^-ainkx, where ^
is
expanding f(x)
in a
a root of the equation /-cosa/t-Hft sin
summation
is
extended to
all
uniformly convergent series of
a^-
and the
positive roots of this equation, determine the constants
Jj..
(Math. Trip. 1898.)

""
27.
is
f{x) = -a^+ 2 {acosnx + b,t>im nx)

2
,(=1
If
a Fourier
series,
shew
=- P
7T
that, iif{x) satisfies certain general conditions,
f (t) cos nt ta.n -
'
--
'It
6 =
""yO
f{t)smnttn,n-t
'It
(Beau.)
W. M. A.
13
CHAPTER X
LINEAR DIFFERENTIAL EQUATIONS
Linear Differential Equations*
lO'l.
Ordinary points and singular points.
In some of the later chapters of this work, we shall be concerned with the
investigation of extensive and important classes of functions which satisfy
linear differential equations of the second order.
that
we should now
establish
some general
Accordingly,
it is
desirable
results concerning solutions of
such differential equations.
The standard form

will
of the linear differential equation of the second order
be taken to be
S-;+;'W5 + ?W = o
and
it will
be assumed that there
analytic except at a finite
Any
point of
*S^
which p
at
a domain
is
number
(A),
which both p {z), q
in
(z) are
of poles.
q (z) are both analytic will be called an

other points of 8 will be called singular
(z),
ordinary point of the equation
points.
8olutionf of a differential equation valid in
10'2.
ordinary
t/ie
vicinity
of an
point..
Let b be an ordinary point of the differential equation, and let St be the

domain formed by a circle of radius r^, whose centre is b, and its interior, the
and is
radius of the circle being such that every point of 8^ is a point of
an ordinary point of the equation.
^S',
Let z he a variable point of
8i).
In the equation write w = vexp
^^^^
/ (z) =
where
*
of solving differential
subject, such as Forsyth,
t This
method
to equations of
is
p{^)d^> and
,
it
becomes
+ J(zyv=0
q (z)
-1
(B),
-^ -Îp
{z)Y-.
mainly theoretical it consists, for the most part,

assumed that the reader has some knowledge of practical methods
equations; these methods are given in works exclusively devoted to the
The analysis contained
of existence theorems.
in this chapter is
It is
Treatise on Differential Equations (1914).
applicable only to equations of the second order.
any order, see Forsyth, Theory of
Diff'erential
Equations,
For a method applicable

iv. (1902),
Ch.
i.
lO'l, 10-2]
It is easily seen ( 5'22) that
195
an ordinary point of equation (A)
is
also
an ordinary point of equation (B).
Xow
consider the sequence of functions
i'(2),
analytic in
Sf>,
defined by
the equations
Vo (z)
= flo + i
(^
- h),
(^-2)J(0Vn-i{0d^,
Vn(z)=^
(^
where
, !
Let M,
Then at
= 1,
2, 3, ...)
are arbitrary constants.
IX
be the upper bounds of \J{z)\ and \vo{z)\ in the domain *SV

of this domain
all points
\Vn{z)\^fiM^\z-b\^l{n^).
For
when
this inequality is true
we have, by taking the path

\Vm{z)\
if it is
true
when w = 0, l,...m-l,
of integration to be a straight line,
= \\\i:-z)J{OVn,-AOdX
\J b
<
^
(m-1)!'^
ml
fiM"^
ri--*i
fiM'"-
\z-h'
f-'^-^-dt
Jo
\z-b
I-'".
'
Therefore, by induction, the inequality holds for

Also, since
verges,
it
Vn (z)
when
is
in
S^,
all
values of
n.
and S fiM^^ri^^Kn
!)
con-
n=0
follows ( 3"34) that
v{z)= 2
v (z) is a series of analytic functions
n=0
uniformly convergent in Sb
while, from the definition of Vn
j^v,,{z)
j^^Vn{z)
hence
it
= -^~^J{\;)Vn-,{t;)dt
(z),
{n
l, 2, 3, ...)
= -J{z)Vn-Az)\
follows ( 5'3) that
d^v{z)
dz^
^ d'v.iz)
dz^
d?vj^
nZi
dz^
= -J{z)v{z).
Therefore v{z)
is
a function of
z,
analytic in
Sj,,
tuhich
satisfies
differential equation
132
the
196
and, from the value obtained for
= ao,
v{b)
where
: i o,re
1021.
-j- Vni^:), it is
evident that
= |^^(^)j^
^'(^)
[CHAP.
=!'
arbitrary.
Uniqueness of the solution.
If there were two analytic solutions of the equation for
such that rj(6)
= Vo(6) = a^, Vi{b) = Vo'ib) = a^,
say
v,
v^
(z)
and
v.,
(z)
= Vi(z)-V2(z),
then, writing 2u{z)
we should have
^ +
d^W
(z)
Differentiating this equation
(z)
= 0.
n.
times and putting z

.
i\ (z),
and
v.,
= Aq,
(z)
= v (2)
that
exp j-
u
where
u'ib) = A-^,
ô = o,
Now
= 0;
by Taylor's theorem, w{z)
infer without difficulty that
that u{b)
/'*--'
(6)
get
w (b) = 0.
that
is
to say the
(z) are identical.
Writing
so,
= b, we
the differential coefficients
'2,
vanish at b
two solutions
we
(6)
Putting
lu
+ J{h) lu^-"-^^ (b) + n-oCWib) w^""-'^ (6) +

n=
8, 4, ... in succession, we see that all
?/;'"'
of
7i
J{z)iv{z)
we know that a
(z) is
A^
pi^dn,
the only analytic solution of (A) such
= a^-\p{b)aa.
solution of (A) exists which
is
analytic in
^'^,
and such that u{b), u' (b) have the arbitrary values Ag, Ai, the simplest
method of obtaining the solution in the form of a Taylor's series is to assume
u(z)= S An{z by^,
substitute this series in the differential equation and
equate coefficients of successive powers of ^

in order the values of ^2, -^3,
[Note.
it
...
determine
In practice, in carrying out this process of substitution, the re;vder will find
simple to have the equation 'cleared of fractions' rather than in the
Thus the equations
be treated in the form in which they stand

divided out.
From
The same remark
the
differential
examples
in
the factors
applies to the examples of
is
- s-,
^v^
analytic at all points of
The
equation.
analytic throughout
may
(z
not be one-valued
'
solution
( 5"2 cor. 2, footnote),

i.e.
it
may
describes a circuit surrounding one or
-2)
10".3,
the general theory of analytic continuation
the solution obtained
'
6 to zero ( 3"73) to
much more
canonical form (A) of ^ 10"1.
of
in terms of -4o, ^1.
and 2 below should

{z
3)
should not be
10-32.]
( 5*5) it
follows that
except at singularities
however
is
not,
in
general,
except at these points, as
it
not return to the same value when z
more
singularities of the equation.
10*21, 10-3]
[The property that the solution of a linear
differential equation is analytic
except at singularities of the coefficients of the equation

equations of
When
197
is
common
to linear
all orders.]
two particular solutions of an equation of the second order are not
constant multiples of each other, they are said to form a fundamental system.
Example
Shew
\.
that the equation
(l-s2)w"-2.-w' + |i =
has the fundamental system of solutions
_ U ?2 _ JUL ^4 _
1
i/
Determine the general

vergence of each series
Example
2.
in each series,
coefficient
and shew that the radius of con-
is 1.
Discuss the equation
{z-2){z-S)u"-{2z-5)u' + 2u =
in a
manner similar
10'3.
1.
Points which are regular for a differential equation.
have poles at
analytic at
c,
is
is
called a regular point* for the differential
such that, although p {z) or q {z) or both

the poles are of such orders that (z c)p(z), {z - c)'-q{z) are
Suppose that a point
equation.
example
to that of
of
Such a point
c.
Any poles
of p{z) or of q (z) which are not of this nature are called

The reason for making the distinction will become apparent
irregular points.
in the course of this section.
If c be a regular point, the equation
P (z
c),
Q {z
written
+ (z-c).P(z-c)'^^+Q{z-c)uÔ,
(z-cY'^^'',
where
may be
c)
are analytic at c
hence, by Taylor's theorem,
P{z-
c)
=p^+pi{z-c) + 2h {z-cy- +
...,
Q (z -
c)
...,
q,
(z- c)+
q.
{z-c)-+
and these series converge in the

domain S,. formed by a circle of radius r (centre c) and its interior, where r is
so small that c is the only singular point of the equation which is in Scwhere
Pq, pi, ..., g, qi,
are constants;
Let us assume as -^formal solution of the equation

u
where
a, a^, a.,, ...
= {z- cY
+ S
an (z
- cr
are constants to be determined.
* The name 'regular point' is due to Thome, Journal fiir Math. lxxv.
Fucbs had previously used the phrase point of determinateness.'
t Frobenius calls this the normal form of the equation.
'
(1873),
p.
266.
198
[chap.
Substituting in the differential equation (assuming that the term-by-term

differentiations
and multiplications of
(z-cY a(a-l)-l- S an{a +
series are legitimate)
we get
+ n-l)(z-cY
7i)(a
+ {z-cYP{z-c). a+ 1
an(a
+ n)(z-cy'
+ {z-cYQ{z-c) 1+ 1 ttniz-cT =
Substituting the series for
P(z c), Q{z c),
0.
multiplying out and equating
to zero the coefficients of successive powers of z
c, we
obtain the following
sequence of equations
+ {}h-l)a + qo = 0,
, {(a + iy + (po-'i-)(o( + l) + qo}+ ap, + q, = 0,
+ 2y + (po - 1) ( + 2 + ô] + a,{(a + l)p, + q^} + ap, + q, = 0,
oi-
a, {(a
an {(a
nf
-1-
+ (p,-l){ci-]+ 2
qo]
Qn-m
{(
n)
- m) p,n +
+ dpn + qn = 0.
?,}
m=l
The
of these equations, called the indicial equation*, determines
first
values (which may, however, be equal) for
The reader
a.
two
will easily convince
had been an irregular point, the indicial equation would have

and he will now appreciate the distinction
made between regular and irregular singular points.
himself that
been
if c
(at most) of the first degi'ee
= pi,
Let a
p.2
be the roots i* of the indicial equation
F{a) =
a?
+ {po-l)oi +
q,
0;
then the succeeding equations (when a has been chosen) determine

in order, uniquely, provided that
that
is
and,
if
to say, if a
Hence,
p..,
if
that
pi,
p^ is
that p^
F{a +
is
n) does not vanish
when
not one of the numbers
not one of the numbers p^
the difference of the exponents
-}-
1, p.,
pi
?i
-|-
a^, a^, ...,

1, 2, 3, ...
1, pi -h 2,
not zero, or an integer,
is
...;
2,
it is
always possible to obtain two distinct series which formally satisfy the
equation.
Example.
is
Shew
that, if ra is not zero or
an integer, the equation
formally satisfied by two series whose leading terms are
16(H-to)
7'
'
i^
determine the coefficient of the general term in each

^ converge for all values of
*
The name
is
-+
.*-'jl+.
"
16(l-m)
series,
.:,}
and shew that the
series
z.
due to Cayley, Quarterly Journal, xxi. (1886),
p. 326.
t The roots pi, po of the indicial equation are called the expotients of the differential
equation at the point c.
Convergence of the expansion o/ lO'S.
10"31.
If the exponents pi, p^ are not equal, let
5"23,
we can
find a positive
Mr-\
\qn\< Mr-'\
\pn\<
is
independent of n
Taking a
= pi we
P2 = s\
let pi
F (pi + n) = n(s+
Now, by
be that one whose real part
p-^
not inferior to the real part of the other, and
where
199
10-31]
then
71).
number
p.j^n
M such that
qn
< Mr-'\
convenient to take if ^
it is
is
1.
see that
M
aAJMl^,<
r s+1
\F{p, +
<
\)\
since |5
If
>1.
now we assume
a^-t
a,,,
I
<
{(pi
i/"r-"
vi-t)pi-it
ttm.
when n=\,2, ...m-
F(p,
+ piPm +
qt]
1,
we get
qr.
+ m)
m
a,^t \-\piPt
+ qt] +
piPm +
q,a
+ s (m -
\vt
m\8 +in\
m,M'^r-"'+
m"
Since
+ snr^ ^ 1,
2 {m-t)\
']
+ smr^
because
a,
and
so,
by induction, a <
|
<
M'^r
R (s) is
not negative, we get
m + 1 jpn ,.-m < j,/'r-'",

2m
M'^r'"^ for all values of
?2.
If the values
a/,
(iz, ...
of the coefficients corresponding to the exponent po be
we should
obtain,
by a similar induction,
\an\<M'Wr-'\
where k is the upper bound of |1 s|~\ \l
bound exists when s is not a positive integer.
We
^s\~\
|-s |~S ...
this
have thus obtained two formal series

Wi (z)
= {z C)P' 1+ S aniz-cr
M
^2
(2^)
= {^ CY
=1
1+ ian(z-cY
The first, however, is a uniformly convergent series of analytic functions

when \z-c\< rM~\ as is also the second when \z-c\< rM-'^K-\ provided
200
in each case that arg (z
c) is restricted in
[CHAP.
such a way that the series are
one-valued; consequently, the formal substitution of these series into the

left-hand side of the differential equation
a solution of the equation
is justified,
and each of the
pi
provided always that
p.,
is
series is
not a positive
integer or zero*.
With
we have
this exception,
therefore obtained a fundamental S3^stem of
And by the theory

in S of the equation
solutions valid in the vicinity of a regular singular point.
of analytic continuation,
are regular points, each

at all points of
we see that if all the singularities

member of a pair of fundamental solutions
is
analytic
except at the singularities of the equation, which are branch-
points of the solution.
Derivation of a second solution in the case when

is an integer or zero.
10"32.
difference
tlie
of the exponents
p^ po, = s is a positive integer or zero,

may break downf or coincide with Wi{z).
In the case when

Wo{z) found in
If
we
lO'Sl
= w^
write u
(z
cy
{z)
-r^^
dz-
^,
the equation to determine f
\2 {z
- cf
of which the general solution
^= A-rB\
ry-
{W,{Z)Y
A+B
A+Bj
(z-
larities of
exp
^
interior of
u\(z)
is
+ (z-c)Piz-c)\-^ = 0,
)
dz
is
(z-c)-P
where A,
'--^-{
the solution
^^^^- dz\.dz
2-C
c)-P-^p^g (z) dz,
are arbitrary constants and
any
whose centre
circle
P (^ c)
is
g {z) is analytic throughout the

which does not contain any singu-
c,
or singularities or zeros of {z
c)"''i w^ {z)
also
(c)
1.
00
g{z)=\+ S gn.(z-cy\
Let
n=l
Then, HsiÔ,
^=A+Br\l+i(z-cf\(z-c)-'-'dz
J
= A+B
=l
--(z- c)- - 'i

_
-^n
(z
- cy-' + gs log (z - c)
n=l ^
+
* If
Pi
The
-p2
is
a positive integer, k does not exist;
coefficient
will be a solution
a/
may
be indeterminate or
it
if p^
containing two arbitrary constants
factor will be a constant multiple of wi
(z).
= p2,
may
qq'
-^
s
(z
cY-'
n = s+l n
the two solutions are the same.
be infinite
and aj
in the former case
the series of which Ug
to-,
(z)
is
10 "3 2]
Therefore the general solution

analytic at
all
points of
Awi
where, by
differential
equation, which
is
(c excepted), is
{z)
+ B [gsti\ (z) log (z-c) + w (z)],
w (z) = {z - c)
2-53,
of the
201
-- + S
p\
h,,
{z
c)"l
the coefficients hn being constants.
When s =
0,
the corresponding form of the solution
Aw,(z) +
(z-c) + (z-
w, (z) log
cY^
hn (z
- c)"
n=l
The statement made

exceptional case Avhen s
is
end of
at the
is
10-31
now seen
is
to hold in the
zenj or a positive integer.
= 0,
In the special case when gg
the second solution does not involve
a logarithm.
solutions obtained, which are valid in the vicinity of a regular point
The
of the equation, are called regular integrals.
Integrals of an equation valid near a regular point c

practically
by
first
obtaining
w-^ {z),
and then determining
may be
obtained
the coefficients in
30
a function u\ (z)
= S
hn{z
c)''-+'*, by
substituting
lu^
(z) log (z
c)
n\
{z) in
the left-hand side of the equation and equating to zero the coefficients of the
various powers of z
due
to
Frobenius*
is
An
in the resulting expression.
alternative
method
given by Forsyth, Treatise on Differential Equations,
pp. 243-258.
Example L
Shew that
integrals of the equation
<Pu
regular near
and
..,(.)
log.-
l+
Shew
Verify
+ _+...+-j.
vahies of
all
z.
-.sdu
,^
are
''W = ^ +
and
(^_
that integrals of the equation
^,dû
regular near
^^.^^^
^2^
Verify that these series converge for

2.
are
Wi{z) =
Example
du
.!.( 2.4...2,. )-
4^j(^_^-^-^j (^__^^ + __...__J..

that these series converge when . < 1 and obtain integrals regular
.,(.)log. +
Journal fur Math, lxxvi. (1874), pp. 214-224.
near
3= L
202
Example
3.
^:i
dz^
by the hypergeometric
+ {c-(a->rh + \)z}^ -ahu = Q

dz
series of 2-38.
Obtain the complete sohition of the equation when

10*4.
Shew that the hypergeometric equation

z{\-z)
is satisfied
[CHAP.
= \.
Solutions valid for large values of\z\.
Let ^ =
l/5i;
then a solution of the differential equation
valid for
'
and
said that
large values of ^
said to be
valid for sufficiently small values of Ui
if it is
'
is
an brdinary (or regular or irregular)

point of the equation when the point ^j =
is an ordinary (or regular or
irregular) point of the equation when it has been transformed so that Zi is
the independent variable.
it is
the point at infinity
'
is
'
Since
we
2= oc should be (i) an ordinary

z~p {2), z*q{z) should be analytic
see that the conditions that the point
point,
a regular point, are
(ii)
and
at infinity ( 5*62)
Example
1.
Shew
(ii)
that 2z
(i)
that zp
{z),
z-q (z) should be analytic at infinity.
that every point (including infinity)
is
either an ordinary point or
a regular point for each of the equations
z{l-z)-jj^->r{c-{a-vh + \)z}-^^-ahu = 0,
(1-^2)
where
a, &, c,
Exam-pie
,^-^-2s^+?i(?i + l) = 0,
n are constants.
2.
Shew
that every point except infinity
is
either
an ordinary point or a
regular point for the equation
^^S-"+4"-^(^'-')='
where n
is
a constant.
Example
3.

,,
.,,
d'û
has the two solutions
i
7^'^
2_ 1
^"
the latter converging

10'5.
when
3'
s
>
î
277
du
^.
3.4.5.6
7^'^ 2. 4.
7. 9
1
z^
1.
Irregular singularities and confluence.
Near a point which
is
not a regular point, an equation of the second order
cannot have two regular integrals, for the indicial equation

the
We
first
shall
degree
there
may be one
see later (e.g. 16*3)
regular integral or there
what
is
is
at
most of
may be
none.
the nature of the solution near
such points in some simple cases.

is
203
10'4-10-G]
beyond the scope of
general investigation of such solutions*
this book.
It frequently happens that a differential equation may be derived from

another differential equation by making two or more singularities of the
Such a limiting process
tend to coincidence.
latter
and the former equation

seen in
is
called a confluent
The most general

every point except
may be
of a
more
latter.
equation of the second order which has
differential
a^, a^, as, a^
and
being regular points with exponents

at
It will be
The differential equations of mathematical physics.
10'6.
/j,2
of the
latter.
lO'G that the singularities of the former equation
complicated nature than those of the
fjbj,
form
called confluence;
is
X may
,
dz''
oo
as an ordinary point, these five points
a^, ySr at
ar{r
4)
1, 2, 3,
and exponents
be verified f to be
z-ar
Vti
)dz
\rZi{z-ayf
T{(z-ar)
r=\
where
is
such
that;): /î
and
î.,
are the roots of
and B, C are constants.

B6cher|| that
The remarkable theorem has been proved by Klein and

the
all
linear
equations
differential
which occur
in
certain branches of
Mathematical Physics are confluent forms of the special equation of this

type in which the difference of the two exponents at each singularity is |
a brief investigation of these forms will now be given.
If
we put
y9,.
a,.
+ ^,
(r
l, 2, 8,
4)
the last
and write ^
in place of
+ h)
2^r+C ^_Q
z,
written equation becomes
dî
h-^Âdu
(4
ar(ar
^?^ +
r=l
* Some elementary investigations are given in Forsyth's Differential Equations (1914).
Complete investigations are given in his Theory of Differential Equations, iv. (1902).
The
coefficients of
-^ and u must be rational or they would have an

4
some point; the denominators
at
essential singularity
dz
of p{z), q{z)
must be
{z-a^),
r=l
{z-a^)^ respectively;
J-=l
into partial fractions and remembering that p{z)
putting p(z) and q (z)

as 2 ^- 00 we obtain the required result without difficulty.
= 0(z-^),
q{z)
0(z~-)
X It will be observed that mi,
fJ-i
are connected by the relation
M1+M2+ 2
Ueher lineare Differentialgleichungen der zweiter Ordnung (1894),
p.
Uber Lamd^schen Funktionen.

II
Ueber die Reihenentwickelungen der Potentialtheorie (1894),
(a,.
+ /3^) = 3.
j=i
p. 193.
40; see also Vorlesung
204
where (on account of the condition
yu..,
fî
[chap.
2)
\r=l
This differential equation

It
evident,
is
is
on writing
called the generalised

i
= ao
confluence of the two singularities
exponents
a, /3
are given
/3
2 (!
equation.
throughout the equation, that the
a^,
a.,
yields a singularity at which the
by the equations
oo),
a^
a,
(oc,
+ h) + a, (a, +
D = (Aa^^- + 2Ba^ + C)/{(ai -
where
Lame
a.^)
(a^
|)
D,
a^)}.
Therefore the exponent-difference at the confluent singularity
is not ^,
have any assigned value by suitable choice of B and C. In like

manner, by the confluence of three or more singularities, we can obtain
but
it
may
one irregular singularity.
By
suitable confluences of the five singularities at our disposal,
obtain six types of equations, which
number
may
classified
by means
exhaustive*:
of the
following
number
we can
according to (a) the
of their singularities with exponent-difference |, (6) the
their other regular singularities, (c) the

larities,
be
number
of
of their irregular singu-
scheme, which
is
easily seen
to
be
lOT)]
Example
\.
205
Obtain Lame's equation
r=l
(where h and n are constants) by taking

n,
and making 04^- x
Example
2.
=0^ = 03 = 04 = 0,
8B = n('/i-\-l)
Obtain the equation
dc'
^\rc- y
dc
4c (c-
(where a and j are constants) by taking ai = 0,

Mathieu's equation ( 19-1)
-7^
by the substitution
Example
3.
4(7=/i4,
a^,
0-2
= 1,
^~
1)
'
and making
= c/4^-x.
Derive
2i) =
+ (a + 16jcos
f = cos^2.
Obtain the equation
^*4./K_JLW,
"
(r c- 1/
c^r^
f!L(^L)__!!!!_l
* I
c/c
f - ij
(c-
_n
~
1)
'
by taking
i
= 2=l,
3 = a4 = 0,
ai
a.j
= a3 = 0,
a4
= |.
Derive Legendre's equation ( 15-13, 15'5)
by the substitution
(=
z~\
By taking
Derive Bessel's equation
17 "11)
Example
4.
(/.2
= 0,
01
= 02 = 03 = 04 = 0,
and making
a3
= a4-*-x,
obtain
the equation
gdû
dû
dz^
by the substitution
Example
5.
du
^ " dz
C=z'^.
By taking i=0,
01
= 02 = 03 = 04 = 0,
and making
(/o
= 3 = a4-*-x
obtain
the equation
.
dhi
du
^,
f^+i5^+i(+*-if)=o.
Derive Weber's equation
( 16'5)
dû
^ + (+*-^-')=o
,
by the substitution f=2.
Example
6.
By taking 0^=0, and making
^+
By
0^-9-
(^
= 1,
2, 3, 4),
(îf+Ci) = 0.
taking
u = {B,C+C\r-v,
B,C+C\ = {IB,z)^,
shew that
d^v
dv
,.
obtain the equation
206
[CHAP.
Example 7. Shew that the general form of the generalised Lame equation is unaltered (i) by any homographic change of independent variable such that qo is a singular
point of the transformed equation, (ii) by any change of dependent variable of the type
'll
= {z-ar)^ V.
Example
generalised
7 that the various confluent forms of the

always be reduced to the forms given in examples 1-6.
Deduce from example
8.
Lame
may
equation
[Note that a suitable homographic change of variable will transform any three distinct
points into the points
10'7.
0, 1, qo
.]
ivith three singularities.
Linear differential equations

d?u
Let
^du
d^'^^^^^dz'^^^^^'^^
have three, and only three singularities*, a, h,

points, the exponents thereat being a, a! /3, /3'
;
Then p
{z) is
{z-^).
7,
these points be regular
7'.
a rational function with simple poles at
these poles being

is
c; let
-a-
- /3 -
a'
+ ^ + y8' + 7 +
a',
/S',
- 7 - 7' and
;
a, b, c, its
as 2
^ co
residues at
p{z)-
2z-^
Therefore
andf
7'
1.
In a similar manner
,,
^^
[aoL
{a-h){a-c)
l^
B^'ih-c){h-a)
yy' (c
z b
-a
a) {c
- b)
c
1
X
{z
and hence the differential equation

d-'u
dz'
oia'{a-b ) (a-c)
z
1-7-7
1-/3-/3^
U-ô-^'
a){z
/3^' (b
This equation was
-c)(b-a)
z
called
first
dz
77
(c
-a)(c- b)
z~c
\
I
= 0.
c)
given by Papperitzj,
fact that u satisfies
an equation of this type (which
will
be
Riemann's P-equation), Riemann wrote

fa
u = p}a
W
*
g)
du
{z a){z b){z-
To express the
b) {z
is
The point
/3
z[.
/3'
an ordinary point.
t This relation must be satisfied by the exponents.
+ Math. Ann. xxv. (1885), p. 21B.
at infinity is to be
It wiU be seen from this memoir that,

Abh. d. k. Ges. d. Wiss. zu Gottingen, vii. (1857).
although Kiemann did not apparently construct the equation, he must have inferred its existence
from the bypergeometric equation.
107, lO"7l]
207
The singular points of the equation are placed in the first row with the
corresponding exponents directly beneath them, and the independent variable
is placed in the fourth column.
Example.
Shew
that the hypergeometric equation

,,
is
d-u
.
7
du
c?
defined by the scheme

00
j'
pJ
[l-o
10 71.
2-.
c
a-h
Transformations of Riemanns P-equation.
The two transformations which

;
(I)
are typified
(^
iv^l)
z-h [v^k)^{^
by the equations
z\
= P{a^-k
^-k-l
y-^l
208
[CHAP.
The connexion of Riemanns P-equation with
10"72.
the hypergeometric
equation.
B)^
means
of the results of lO'Tl it follows that
where
= (z
(z
Hence, by
a'
/3
/3'
+a+y
7'
7'
-7
+7
a)(c b)
h){c a)'
example, the solution of Riemann's P-equation can
10"7
{a
always be obtained in terms of the solution of the hypergeometric equation

1
,7
whose elements
a, 0,
x are
c,
a+ + 7,
r^
n,'
a-t-p+7,
i+a
-1
a,
{z
a){c}))
a)
-
{z-b){c
respectively.
Linear differential equations
10"8.
If,
in
we make the point
10'7,
'
l_^_y=0,
,
divisible
by
Hence
^^
77'
c,
and
+ a + ^ + jS'
j 1 a a'
d-ii
of which the solution
p (z) and
q (z)
aa = yS/S', and
1 + a + a'] du
dz
zb
-D
to say, the solution involves
When
4.
(b
^^^1^
may
-_^Lv
c) (b - a)
be analytic at
atx
{a
{z of
j^^^^
^g
c.
is
by
{z
'
b)-
is
.
is
/3/3'
the equation
aY + B
u=A{fz
\z b)
that
we must have
a regular point,
-^
^^
in order that
two singularities.
aa(a-b)(a-c)
= 0,
dz-
tuith
a, the solution
.
fz(^
\z
bj
elementary functions only.
is
fz-aY
fz
aY
fz-a
bJ'
REFEREXCES.
FuCHS, Journal fur Math. Lxvi. (1866), pp. 121-160.
L. W. ThomiS, Journal fur Math. Lxxv. (1873), pp. 265-291, Lxxxvii. (1879), pp. 222-349.
L. ScHLESiNGER, Handbuch der linearen Differentialgleichungen.
(Leipzig, 1895-1898.)
G. Frobenius, Journal fur Math. Lxxvi. (1874), pp. 214-235.
L.
RiEMANN, Ges. Math. Werke, pp. 67-87.

Kleix, Ueber Uneai-e Differentialgleichungen der zineiter Ordnung.
A. R. Forsyth, Theory of Differential Equations, iv. (1902).
T. Craig, Differential Equations.
(New York, 1889.)
E. GouRSAT, Coxirs d\inalyse, 11.
(Paris, 1911.)
G. F. B.
F. C.
(Gottingen, 1894.)
10*72]
209
Shew
1.
are 2
3^7 2*
that two solutions of the equation
+ ..., 1-^2^ + ...,
and investigate the region of convergence of these
series.
Obtain integrals of the equation
2.
d"^^
regular near
= 0,
in the
,,
ON
form
î
.
^"r + r6 +
1024
+ -/'
2 =
Wll0g2-jg+....
3.
(n+l-l2^)=0
has the solutions
^~
2?i
+l
Shew
4.
96
4/i2
12
and that these
"''
'"+
+ 4?i + 7
480
series converge for all values of
z.
that the equation
dZ^
Zttr
tr=l
dz
a,.Y
\r=i{z
r=l*-rj
where
2
(a,
+ 3,) = -2,
r=I
is
Z),
= 0,
(a,
A-+ari3r) = 0,
the most general equation for which
(a,2/)^
+ 2a,a,i3,) = 0,
r=l
r=l
r=I
points (including
all
),
except aj,
ordinary points, and the points a^ are regular points with exponents
a^,
a-.,,
a, are
...
^r respectively.
(Klein.)
5.
Shew
that, if
/iJ
+y+
/3'
(0
[The
dififerential
Shew
22'
&
^2-
that, if y
+ y' = ^
and
rO
if
co,
differential equation in
A.
y'
r^
23'
^22 .
~|
,2_l
IJ
rl
CO
0)2
= p'y
ly
7'
23-
222
C22^23_l
W. M.
2-.
(Riemann.)
is
each case
d'û
""J
or are the complex cube roots of unity, tfeen
p'
dz
X
2/3
i-1
P\0
pJo
[The
then
equation in each case
c/22
6.
+ y' = ^,
X 1
^j
2-.
(Riemann.)
>'
is
du
^yy zu
dz^
{z^-\f
^^ ,
1^
210
Shew
7.
[CHAP.
that the equation
(l_.2)g_(2o + i)2^Vn( + 2a) =

is
defined
b\^
the scheme
pj
z^
-n
\jf-a
+ 2a
and that the equation
may
be obtained from
it
by taking a = l and changing the independent
variable.
(Halm.)
8.
Discuss the solutions of the equation

-.du
dii
dhi
'-U
,
^ + (^ + l+m)^
^^22
,
valid near s
9.
and those valid near
= oo
,11
+ l+^mj\ w = ^
,
(/i
(Cunningham.)
Discuss the solutions of the equation
valid near
2=0 and
those valid near z=<x,
Consider the following special cases

(i)
pi=
Prove that the equation
10.
(ii)
/i
5,
(iii)
/x
+ = 3.
(Curzon.)
i/
'
s(l-2)^ +-(l-22)^ + (a2 + fe)w =

has two particular integrals the product of which is a single-valued transcendental
Under what circumstances are these two pax'ticular integrals coincident ?
function.
If their product
where C
is
be F{z), prove that the particular integrals are
see 19-5.)
Prove that the general linear differential equation of the third order, whose
which has all its integrals regular near each singularity (the
1, oc
l\.
singularities are 0,
exponents at each singularity being
dhi
(2
d^^W
+
a
ny
ha^'e
1, 1,
\
^^)
\\
where
(Lindemann
a determinate constant.
1), is
d^_{\_
dz^
3cos2a
V
3sin2a
1^_"1
z{z-\y{z- If]
1
p - w^) " ^^' " ^m
any constant
value.
du
dz
_n
'
(Math. Trip. 1912.)
CHAPTER XI
INTEGEAL EQUATIONS
11"1.
An
Definition of
is one which involves an unknown function under

and the process of determining the unknown function
integral equation
the sign of integration

is
an integral equation.
called solving the equation*.
The introduction of integral equations

(1782) who considered the equations
f{x)
(where in each case
(f)
[e^f
(f)
(t) dt,
represents the
the solution of differential equations.
{x)
into analysis
[^^-^ 4> (t)
is
due to Laplace
dt
unknown function), in connexion with

The first integral equation of which
a solution was obtained, was Fourier's equation

f(.v)
cos (xt)
(f>
(t) dt,
of which, in certain circumstances, a solution isf

2 f^
(f)(^x)
=-
cos (ux)f(u) du,
TTJo
f{x) being an even function of a;, since cos
an even function.
{xt) is
Later, Abel+ was led to an integral equation in connexion with a mechanical

problem and obtained two solutions of it after this, Liouville investigated an
integral equation which arose in the course of his researches on differential
;
equations and discovered an important method for solving integral equations ,

which will be discussed in 11 '4.
In recent years, the subject of integral equations has become of some

importance in various branches of Mathematics such equations (in physical
;
problems) frequently involve repeated integrals and the investigation of them

naturally presents greater difficulties than do those elementary equations
which
will
be treated in this chapter.
To render the
analysis as easy as possible,
that the constants

*
a, h
and the variables
Except in the case of Fourier's integral
we
shall suppose
throughout
x, y,
^ are real and further that
we
practically always need continuom
9'7)
solutions of integral equations.

t If this value of
(p
be substituted in the equation
we obtain a
result
which
is,
effectively, that
of 9-7.
+ Solution de quelques problemes a Vaide d'integrales definies (1823). See Oeuvres, i. pp. 11, 97.
The numerical computation of solutions of integral equations has been investigated recently
by Whittaker, Proc. Roijal Soc.
xciv. (A), (1918), pp. 367-383.
U2
212
y, ^î>'^ also
O'^^,
that the given function*
[CHAP. XI
which occurs under the
(x, y),
integral sign in the majority of equations considered,
is
a real function of
and y and either (i) it is a continuous function of both variables in the

range (a ^x^b, aî/ ^b), or (ii) it is a continuous function of both variables
when y > x; in the latter case
(x, y) =
in the range a^y ^x^b and
distributed,
and in either case it is
regularly
discontinuities
its
K{x, y) has
a;
proved that, iff(y)
easily
continuous when
is
a^y^b,
f{y) K{x, y) dy
is
J a
continuous function of x
11 'll.
The
An
when a^x^b.
algebraical lemma.
algebraical result
which
will
now be obtained
is
of great importance in Fredholm's
theory of integral equations.
be three points at unit distance from the origin.

volume of the parallelepiped, of which the lines
joining the origin to these points are conterminous edges, is +1, the edges then being
perpendicular. Therefore, if Xr^ + 7/^^ + Zr^ = l {r = l, 2, 3), the upper and lower bounds of
Let
The
(a'i, 3/1, Zy), {.V2, i/2, 22)) (^'35
^3) ^3)
greatest (numerical) value of the
the determinant
^2
2/2
22
^3
^3
23
are 1.
lemma due
to
Hadamardt
generalises this result.
Let
a^n
21?
0^22,
^nl)
^)!2' <^n
...
where
let
a^,. is real
A be
and 2
a"^^r = l (i = l,
n); let
2, ...
the determinant whose elements are
Since Z)
is
ordinary theory of
air ('=!) 2,
...
maxima and minima
n) only,
its
A^j.,
A^j.
elements, and
and
is
if
"
is
stationary for such variations
if
D and
by a well-known theorem |,
so that,
applicable,
is
be the cofactor of a^r in
obviously bounded, the
we
consider variations in
dD
2 ^
8aij.
r=l VOlir
= 0,
where
Saj^,...
are variations subject to the sole condition 2 a.ir8air=0

r
but 2 airAir=I), and so X'2a\r = D
=l
therefore
therefore Air = Da.ir-
r=l
Bocher in his important work on integral equations (Camb. Math. Tracts, No. 10), always
more general case in which A' (x, y) has discontinuities regularly distributed,
The reader will
i.e. the discontinuities are of the nature described in Chapter iv, example 10.
see from that example that the results of this chapter can almost all be generalised in this
*
considers the
way.
To make
this chapter
+ Bulletin des Sci.
Math.
more simple we
(2), xvii.
shall not consider such generalisations.
(1893), p. 240.
J Burnside and Panton, Theory of Equations, ii. p. 40.

By the ordinary theory of undetermined multipliers.
INTEGRAL EQUATIONS
11-11, 11-2]
213
Considering variations in the other elements of D, we see that

is stationary for
all elements when Amr=Damj. (m=l, 2, ... n
r=l, 2, ... n). Consequently
A = D". D, and so I)'^'^^ D^~^. Hence the maximum and minimum values of I) are +1.
variations in
Corollary.
If
a^r be real and subject only to the condition
a^^
I
< ^j
since
2 {a,,/(ii/)}2^1,
r=l
we
maximum
easily see that the
Fredholms* equation and
11'2.
An
value of i9
is (/i^
J/)" = ?i^"i/".
its tentative solution.
important integral equation of a general type
is
J a
where f(x)
is
complex) and
a given continuous function,
(x, ^) is called
(oo,
a parameter (in general
is
subject to the conditions f laid
is
^)
down
in ll'l.
the nucleusX of the equation.
This integral equation
is
known
as
Fredholms equation
or the integral
equation of the second kind (see 11 "3).

It was observed by Volterra that an
equation of this type could be regarded as a limiting form of a system
Fredholm's investigation involved the tentative carrying
of linear equations.
out of a similar limiting process, and justifying
below in
it
by the reasoning given
Hilbert (Gottinger Nach. 1904, pp. 49-91) justified the
11-21.
limiting process directly.
We
and
now proceed
shall
to write
down the system
of linear equations in question,
then investigate Fredholm's method of justifying the passage to
the limit.
The
integral equation is the limiting

where x^- Xq_i = 8, XQ = a,

Since this equation
Xi, Xi,
...
.r
{x)
form (when 5-*-0) of the equation
=f {x) + \i
K {X, X,)
(t>
(.r,) 8,
= b.
is
to be true
{x.p
when a ^x^b,
it
is
true
when x takes
the values
Xni and so
n
-\8 2
x^)
(.rg)
(f)
(Xj,)
ip = l,2,...
=/ (Xj,)
n).
q=l
*
Fredholm's
first
paper On the subject appeared in the Ofversigt af K. Vetenskaps-Akad.

lvii. (1900), pp. 39-46.
His researches are also given in Acta Blath.
Forhandlingar (Stockholm),
xxvii. (1903), pp. 365-390.
The reader
will observe that
x), the equation
Tk
(x, I)
(^) d|.
called an equation with variable upper limit.

Another term is kernel French noyau, German Kern.
is
may
be written
214
[chap. XI
This system of equations for <^ (.rp), {p=l, 2, ... 7i) has a unique solution if the
This determinant is
{x^) does not vanish.
determinant formed by the coefficients of
Z> (X)
- X8 K (.^1 ,xi)
-\8K {x.,
=1 -X
A'l)
-\8K (.r
-X8K
xi)
2 8A' {x
.Vp)
p=l
;>> <I,
Making
{x.,
x.^) ...
-UK
{x^
.v)
X S A'
(.rg
.^)
x^)
.r.,)
...
- \8 K {x^,
x.,)
...
+
2
"
)i
on expanding* in powers of
S2
q=l
!p,
X3
'^
- XS
{.vi
1-X8K (x^
^
^
K {X^, Xp) A [Xq, Xg)
^''''' '^^^
("^'p' '^P'
V"^p
^^'"
'^
"^'9/
"^'^^
(."^p> "^r)
K{Xq,Xp)
K{Xg,.Vq)
K{Xg,Xr)
lK{x,.,Xp)
K{Xr,Xq)
K {Xr
r=l
+ ...
Xr)
X.
and writing the summations
S-3-O, ti-s-qo,
a-s
integrations,
we
are thus led to
consider the series

/6
= l-X|^
i)(X)
x2
^0^^1 + 2]
A' (li,
[^
[^
dîdi;-...
^(^2,^1)
/^ j^
/>:(^2, I2)
Further, if Z)(.r^, x^,) is the cofactor of the term in I)n{\) which involves
the solution of the system of linear equations is
/(î)i)(.r^, xi)+f{x.2) Z)(av,
.v.i)
Now
it is
easily seen that the appropriate limiting
with Da
{Xf^,
x^)
Dn (AV
-iV)
is
D{\)
= U\K {x^
also that,
x\)
if
- XS 2
+ ^,\^8^
So that the limiting form
(.r
,Xt,; X)
= X A" {x^>,
form to be considered in association
K(x^,x\)
K{x^,Xp)
A {Xpi
Xi,)
(A'p,
Xp)
^(Â'-^V)
^{Xf^,.Vp)
K{x^,Xg)
K {Xp
.r^)
K (Xp
.?7p)
A'
X\)
K (Xg
Xp)
^ (Xp
K {Xg
(.Pg
K{x^,x^)
Xn)
i',
for 8~^ I) {x^., x^) to

''
/n
+ ...+f{xJD{x^,
K {x\, x^J,
Av)-X2
be considered t
;^g)
Xg)
is
A'(.r^, ^1)
I
2!
ja Ja
Îl'^l2-----
A:(î,
K{^2,X;)
K{^-2,^l)
^(^2,6)
Consequently we are led to consider the possibility of the equation
<P(x)=f{x) +
J^^
\[d{.x,^; \)fii)d^
giving the solution of the integral equation.

*
p, q,
for
The
...
D
t
factorials
appear because each determinant of
take all the values
1, 2,
...
n,
whereas
it
rows and columns occurs
(X).
The law
times as
appears only once in the original determinant
of formation of successive terms is obvious
from those written down.
;;
] 1
INTEGRAL EQUATIONS
-21]
Example
Shew
1.
215
that, in the case of the equation
^{x) = x->r\
xii(l){y)dy,
we have
D{\) = \-1\,
and a solution
0(^) =
Example
(.r,
X)
y;
= \xy
is
Shew
2.
Zx
3-X'
<j>{x)
=x+ \
(xy+y^)<f>(i/)dy,
we have
D {x, y;
X)
=X
{xy
+ ?/2) +X^ {^xy^ -
},xy
- If + y\
and obtain a sohition of the equation.

11"21.
So
Investigation of
+ 2
=i
write the series, by which
been purely tentative
we now
and verify that we actually do get a solution of the equation
do this we consider the two functions
We
1
solution.
far the construction of the solution has
start ah initio
to
Fredholms
D (\), D{x, y; \)
D (^-)
was defined
arrived at in 11*2.
in 11 '2, in the form
so that
n\
rb fb
rb
d^,d^2
J
K
where M
since
{x,
is
y)
-1
continuous and therefore bounded, we have
is
d^n;
independent of x and y since

y)
( 11*11) and we see at once that
;
{x,
real,
is
\K {x,y)\< M,
we may employ
Hadamard's lemma
Write
n^^'M'' (b
a)"
=n
then
lim(6.WM=lim^f-"^-^f|(l+jYf
nj
(\
Môo (71+1)5
uôo
since
The
series
bnX"- is therefore absolutely
convergent
for all values of
X,
w=l
and so
2-34) the series
fore ( 5"64) represents
Now
Tfrite
1+2
-^~
converges for
an integral function of
the series for
D (x,
all
values of
X and
X.
y;X) in the form 2

'-
there-
[CHAP. XI
216
Then, by Hadamard's lemma
and hence
'
<
n\
Cn,
where
1111),
c is
of
independent
^
a;
and y and 2
Cn>^^^^'^ is
=o
absohitely convergent.
D {x, y; \) is an integral function of \ and the series for

\K
(x, y) is a uniformly convergent ( 3-34) series of continuous*
y\\)-
Therefore
D{x,
Now
a^x^b, a^y^b.
x and y when
functionsj|of
pick out the coefficient of A"
(a;,
y) in
D{x,y;X)^ \D {\)K{x,
y)
+ 2
D(x, y;X); and we get
(-)X+'
^^^^
where
Expanding
integral of the
in place of
we g^t Q
in minors of the first column,
sum
of
n determinants; writing
^ in the ??ith of them,
fi, ô, ...
the determinants + are equal and
Qn {x,y)
where
Pn=\
we
ô, ...
y) equal to the
{x,
^m-i,
|^,
^,n,
^n-\
see that the integrals of
all
so
K {x, y) Pnd^dî
= -n\
J
|î,
a J a
{X, I),
{X, ^,),
K{^Â
i^di.a
d^n-i,
...
K(x, |_0
...
A-(inf-o
It follows at once that
D(x,y\X) = \D(X)K(x,y)+\
Now
D (x, ^\X)K{ly)dl
take the equation
{^)=fi^)
+ ^\' K{ly)<f>{y)dy,
J a
D (x,
multiply by
X) and integrate, and
we get
l'f(BD(^,^;^)d^
J a
(I)
0^-.
? ^) ^?
;
r [' ^
-^
.'a
a J
(''
^ ^) ^^
;
<^^'
^)
^^> ^y*^^'
i(
the integrations in the repeated integral being in either order.

*
is
It is
easy to verify that every term (except possibly the first) of the series for
(i) or hypothesis (ii) of 11-1.
^
(x,
y;
\)
a continuous function under either hypothesis

t
The order
of integration
is
immaterial
( 4-3).
h
t
INTEGRAL EQUATIONS
11-21]
That
is
217
to say
J a
= r<j>(^)D(w,^;\)d^-r{D{w,y;\)-\D(\)K{x,y)]<j>(y)dy
J a
J a
= \D{\)\' K{x,y){y)dy
in virtue of the given equation.
Therefore
D{X)^0
if
and
if
Fredholm's equation has a solution
it
can be
none other than
=f{x)
{X)
<^
^^^^ d^
+jy(^)
and, by actual substitution of this value of
we
see that
This
it
actually
is
the integral equation,
<f>{x} in
a solution.
unique continuous solution of the equation
therefore, the
is,
if
i)(X)4=0.
Corollary.
If
we put /(^) = 0, the homogeneous equation

'
'
J a
has no continuous solution except
Example
1.
(j)
D{X) = 0.
(.i')=0 unless
By expanding the determinant
involved in $
y) in minors of its first
(.r,
row, shew that
D{x,y;
\)
= \D{K)K{x,y) + \\' K{x,^)D{^,y;
\)d^.
J a
Example
2.
2)(X) =
By
using the formulae
1+ i
/i=i
""fi^,
'I
D{x,y;
f " i>
3.
If
{x,
y) =
)"
(^,
(y
X) f^l
^ ^0)
^^^^
- X
i^^\
y)
=^
(y
4.
Shew
that,
if
(.r,
y)=fi
(x) .f^ [y),
and
>
if
-6
fi{^)f2{x)dx = A,
then
D{\) =
\-A\ D{x,y;\) = \h {x)f., (y),
and the solution of the corresponding integral equation
is
^"/''
'
"
D{\)=\-{b- a) X.
shew that
Example
^" '
= \D {X) K(x, y)+
11=1
shew that
Example
\)
'^)->
^\
218
Example
5.
Shew
[CHAP. XI
that, if
K {x, y) =/i {x) gi {y) +f^ {x) ^2 {y),
D (X)
then
and
D {x, y;
X) are quadratic in
and,
more
generally, if
II
K{x,y)= 2
m=l
f^{x)g,n{^),
then i)(X) and D{x, y, X) are polynomials of degree n in
11*22.
Two
Volterra's reciprocal functions.
functions
bounded
X.
K {x,
y),
k{x, y; X) are said to be reciprocal
in the ranges a-^x,
regularly distributed
( ll'l,
if
they are
y -^1),^ any discontinuities they may have are

footnote), and if
K{x,y)+k{x,y]\) = \\
k{x, ^\\)
{^,
y)d^.
J a
We
observe that, since the right-hand side
functions
is
Also, a function
{x,
their difference k^ {x, y)
y) can only have one reciprocal
is
By
We
X)
=X
use
the
first
>(-,(.r,
and by
to be regarded as a parameter),
solution of this equation
11 '21 corollary, the
only continuous
observe, from the relation
11*21 example
k {x, y;\)
A;(^,
D {x,
y X)
;
^ -X)
K H y) d^
is
1,
the equation
+ K{x,
y)
=X
f
a
{x, ^)
{^,
y;X)d^
M
is
evidently true.
Then,
if
we take the
integral equation
<}>{x)=f(x)
+ xl'K{x,^)<f>(^)d^,
J
when
a'>
an elegant
type.
-D{x,y;X)l[\D{\)],
and from
were two,
^)K{k,y)di,
functions, Volterra has obtained
11 "21, that the value of
for if there
between pairs of equations of Fredholm's
B{x,y;\) = \D{X)K{x,y) + \\
proved in
is zero.
of reciprocal
reciprocal relation
if Z) (X) 4=
would be a continuous solution of the homogeneous equation
h{x,y;
(where x
continuous*, the surn of two reciprocal
is
continuous.
x^h, we have, on multiplying the equation
By example
10 at the end of Chapter
iv.
by k
(x,
219
INTEGRAL EQUATIONS
11-22, 11 -23]
^ \) and integrating,
;
J a
= r k(x, ^:\)f(^)d^ + xf' I' k(x,

J a
J
J
a
^:\)
K{1
^,)<f>(^,)d^,dl
II
Reversing the order of integration* in the repeated integral and making
we get
use of the relation defining reciprocal functions,
\\{x,^:\)4>(^)di
J a
^:\)f(^)d^+r{K(x,
!'k(w,
and so
^,)
l^
k{w, ^,;X)}<^(|0^î,
J a
J a
-X
^X)/(^)fZ^ =
fa ^'(^,
(x, ^^)
(^,) d^,
<f>
= -<P(x)+f{x).
Hence
f(x)
+\
(^)
similarly,
fa
f \)/(^) d^
^' (a;,
from this equation we can derive the equation
4>{x)=f(x)
+ \f'K(x,^)4>(^)d^,
J
II
may be regarded
so that either of these equations with reciprocal nuclei
as
the solution of the other.
Homogeneous integral
11"23.
rb
The equation
<^
{x)
=X
)
-
We
{x, ^)
</>
(^)
d^
called a
is
homogeneous integral
1121 corollary) that the only continuous

the homogeneous equation, when D (X) 4= 0, is (^ {x) = 0.
equation.
of
equations.
The
have seen
of
roots
the
D (X) =
equation
are
therefore
importance in the theory of the integral equation.

characteristic
numbers^ of the
now be shewn
It will
of
They
solution
considerable
are called the
nucleus.
that,
when
D (X) = 0,
a solution which
is
not
identically zero can be obtained.
Let+ X
Since
= Xo
D (X)
be a root
is
m times repeated
an integral function, we
of the equation
may expand
it
D (X) = 0.
into the convergent
series
D (X) = Cm (X - Xo)" + c,n^, (X - Xor +1 +

*
The reader
will
have no
diflSculty in
. . .
extending the result of
(m >
0, c,
4-3 to
+ 0).
the integral under
consideration.
t French valeurs caracteristiques,

J It will be proved in 11-51 that,
root.
German
if
K (x,
Eigenicerthe.
ij)
= K{y, x),
the equation
D (X) =
has at least one
220
Similarly, since
D {x,
y X)
is
[CHAP. XI
\,
there exists
a Taylor series of the form
by
3'34
easily verified that the series defining g^
it is
(cc,
(n
y),
1, 1
1, ...)
when a^x^b, a^y
converges absolutely and uniformly
-^b, and thence that

y \) converges absolutely and uniformly in the same
domain of values of x and y.
D (x,
the series for
But, by
11-21 example
2,
L i)(f.f;X)df=-X^>
now
the right-hand side has a zero of order
side has a zero of order at least
and
I,
so
Substituting the series just given for
11'21 example
m1
we have
D (\)
at
A,o,
m\'^
and
while the left-hand

I.
D {x,y\ X)
in the result of
2, viz.
D{x,y-\) = XD
(X)
{x,
y)
+ \\''K {x,
^)
{I y \)
;
d^
J a
dividing by
(X,
XoY and making X

9i {^, y)
Hence
\J
-*
X,o.
we get
{x, I) gi (^,
y) d^.
y have any constant value, gi{x, y) satisfies the homogeneous

integral equation, and any linear combination of such solutions, obtained by
if
giving y various values,
is
a solution.
The equation
Corollary.
<l>
G^) =/(.^-)
+ Xo
f'
iT (.r, ^)
<^ (^)
d^
J a
has no solution or an infinite number.
For,
if (^ {x) is
a solution, so
is
{x)
+ ^c^gi (x,
y),
where
Cy
may
Example
be any function of y.
1.
Shew that
solutions of
<^{x)
=\
cos'<(.r-|)(^(^)c/^
I
are
</>
(.r)
= cos (?i - 2r)
.r,
and
(^ (.r)
= sin (% - 2?-) .r
where
assumes
all positive integral
values (zero included) not exceeding hi.
Example
2.
Shew that
^{x) =
\y
cos'' (.r-h I) (^ (^) o?|
has the same solutions as those given in example

values of X give all the roots of
D (X) = 0.
1,
and shew that the corresponding
ir3-ir4]
INTEGRAL EQUATIONS
221
Integral equations of the first and second kinds.
11*3.
Fredholra's equation
is
sometimes called an integral equation of the second
kind; while the equation
f{x)
= \\\{x,^),^{^)d^
J
is
called the integral equation of the first kind.
In the case when
{x, ^)
> x, we may
ii
write the equations of the
and second kinds in the respective forms
first
f{x)
= \[ K{x,^)^{^)dl
J
4>{a^)=f{x)
+ xrK{x,^)cf>(^)d^.
J a
These are described as equations with variable upper

11"31.
limits.
Volterra's equation.
The equation of the first kind with variable upper limit is frequently
known as Volterra's equation. The problem of solving it has been reduced
by that writer
to the solution of Fredholm's equation.
Assuming that
^ $ X, we have
{x, ^) is
f{x)
a continuous function of both variables
= \\^K{x,^)<\>{^)dl
J
The right-hand
-T exists
and
is
ox
side
This
is
has a differential coefficient
(x)
= \K (x,
x)
example
1) if
+ Xj^cf>{^) dl
{X)
{^)dl
a
and
so the solution of the
equation
The
if /(a)
Fredholm equation gives a solution of
Volterra's
0.
solution of the equation of the
first
kind with constant upper limit
can frequently be obtained in the form of a series *.

11"4.
The Liouville- Neumann method of successive substitutions "f.
method of solving the equation

cf>{x)=f(x)
+ \\'
J
which
*
is
of historical importance,
See example
7, p.
t Journal de Math.
see his Untersuchungen
231
ii.
due
to Liouville.
a solution valid under fewer restrictions
(1837),
ilber
is
K(x,^)cf^(^)d^,
iii.
(1838).
K.
Neumann's
is
given by Bocher.
investigations were later (1870)
das logarithmische und Newton'sche Potential.
[CHAP. XI
222
by the
which occurs on the right-hand side.
It consists in continually substituting the value of 4){oo) given
right-hand side in the expression
<^
This procedure
S{x)=f{x) + \\' K{x,

J
aA?)^^+ mS=
Since
K{x, y) and \f{x)
X2
K{x, ^0
fa
[*
J
K{^
I,)
are bounded, let their upper bounds be
(^)
gives the series
M, M'.
of the general term of the series does not exceed
Then the modulus
|\['l/'ir(6-a)'".
The
S {x)
series for
therefore converges uniformly
when
\\\<M-'{h-a)-^;
and, by actual substitution,
Kix,
If
term
y)
when y>x\ we
in the series for
S {x)
the integral equation.
it satisfies
find
by induction that the modukis of the general
does not exceed
Jf'"
M'
{x
- a)>"l{m
l)^\X\'>' i/' M' (6
ay>'jm
and so the
series converges
Fredholm's solution
uniformly for all values of X
is
and we
infer that in this case
X.
from the form of the solution that when X. < M~^

the reciprocal function k (x, ^ X) may be written in the form
It is obvious
(b
a)~\
k{x,
^X)=.-K{x, ^)- t X-m=2
K(x, ^0
f'
a
/l (I,,
eO
J a
J a
for
with this definitipn of k
{x,
A,),
we
see that
S{x)=f{x)-\\' k(x,^;\)f(^)d^,
J
so that k {x,
|;X)
is
11"22 there
is
B ^r - J^n+r (^,
^),
a reciprocal function, and by
only one
reciprocal function.
Write
(X, I)
= K, {X,
^),
f K {X,
r)
Kn (r,
and then we have
-k{x,^;X)= S
V^^K,+,(x,^),
m=
fa K^ (x, r Kn (r, e ^r = A^.+
while
(*'.
^)>
as
may be
The
seen at once on writing each side as an (m
functions K^^,
(x, f)
+n
are called iterated functions.
l)-tuple integral.
INTEGRAL EQUATIONS
ITO, ll'olj
11*5.
Symmetric
Let Ki
{x, y)
The
Kn {^,
Kn+r
functions
= Kn (y,
y)
(^,
y)
nuclei.
= K^ (y, x); then
iterated
223
x) for
all
= f K, {^,
J a
^)
the nucleus K{x, y)
such
of
Kn {I
y)
nucleus
values of n
for, if
d^ =
j^
Kn
K,
are
said to be symmetric.
i.e.
symmetric, then
K, {y,
(|, o:)
symmetric,
also
{x, y) is
'a
is
^)
d^
Kn(y,^)K,i^,x)d^ = Kn^,{y,x),
J a
and the required result follows by induction.

Also, none of the iterated functions are identically zero for, if possible, let
(x, y) = 0; let n be chosen so that 2""^ <]) -^ 2", and, since Kp (x, y) = 0, it
;
Kp
K^n {x, y) =
follows that
But then
0.
K^n
{x,
from the recurrence formula.
x)
^,_i
{x, ^)
K^^-i
(^,
d^
J a
[Kn-.i^.^rr-d^,
J a
and
Ki
- ^ continuing this argument, we

and the integral equation is trivial.
so K^n-i i^, ^)
(x,
y)
0,
Schmidt's* theorem that, if
11"51.
J) (\) =
tfoe
nucleus
is
find ultimately that
symmetric, the equation
has at least one root.
To prove
this theorem, let
Un =
Kn (X,
X) dx,
J a
when \\\< M~^
SO that,
(b
a)~\ we
dDiX)^ ^
D{\)
Now
since
dx
{fjLKn+i (x, |)
have, by 11-21 example 2 and 114,
"
ri
^,i_i (x, ^)]-
d^dx ^
J aJ a
we have
fi,
tl'U^^.
and
so
U^n+i U.n--
Therefore
Ui,
U.,,
Therefore
so,
by
*
when
5'4,
X-
U^n-,
?7o+2 C/2,1-2
^ v~\
the function
The proof given
U,n-,
> 0,
Uon-1
>
are all positive, and
...
duction from the inequality
and
+ 2/JLU.>n +
is
0.
if-^JZ^/C^,
i^,
it
that Unn+n/U.n
C^2i'.
>
follows,
by
in-
v^-
the terms of
^^

-7-^
Un'\.'^~^
do not tend to zero
has a singularity inside or on the
due to Kneser, Palermo Bendiconti,
xxii. (1906), p. 236.
224
\^
circle
v~^\ but since D(\)
of
sinerularities
-r.
1)
(X,)
d\j-^
By
[Note.
an integral function or
has a zero by 6"31 example
it
the former case
D (X)
Orthogonal functions.
The
real continuous functions
If
the range
u-^{x), Uo{x),
(x),
(jj^
has a zero
e^',
(f)o
else a
mere polynomial
the point of the theorem
which has no
(x),
is
in
that in
zeros.]
are said to be orthogonal
continuous linearly independent
real
we can form n
...Un(x),
D (X)
{a, b) if
given
are
Ave
cannot be such a function as
11'6.
for
D(X); therefore
11-21, Z> (X) is either
the latter case,
and normal*
an integral function, the only possible
is
are at zeros of
= v~^
inside or on the circle \\\
[CHAP. XI
linear combinations of
functions
them which
are orthogonal.
For suppose we can construct m 1 orthogonal functions </>!

is a linear combination of u-^, ii^, ... Up (where p = 1,
(f>p
that
we
shall
now shew how
to construct the function ^, such that
... (f>m-i
such
I);
cî, <p.2,
(f>m
2, ... ni
are all normal and orthogonal.
Let
^(f>m
(x)
Ci,
(f>i
(^)
C2,
so that i(^^ is a function of Mj, lu,
()
<^2
+ Cm-i <pm-i (i^) + Urn {^),
w,.
...
Then, multiplying by ^^ and integrating,

i<\>ni
{x)
{oc)
(t>p
= Cp, , +
dx
Hence
Um {^)
i4>m i^) 4>p {oc)
(x)
<j>p
dx
{p
<
m).
dx =
J a
if
Cp^
==
dx
Uin (jc) <pp ()

J a
a function
i^, (x),
orthogonal to
(f)^
(x),
(f>.
(x),
<pm-i (x), is therefore con-
structed.
rb
Now
choose a so that
a-
[i(f>r,i
(x)]-
dx=l
J a
and take
<p,n
Then
We
*
j^
</)^
(x)
(x) 4>p (x)
can thus obtain the functions

They
= a.
(/)i,
i<^,.
(x).
dx |~
^
c/)..,
...
^^
^ ^^'
in order.
are said to be orthogonal if the first equation only is satisfied
of such functions is due to

pp. 113-148, 31.5-394.
Murphy, Camb.
the systematic study
Phil. Trans, iv. (1833), pp. 353-408,
and
v.
(1835),
ir6, ir6l]
INTEGRAL EQUATIONS
The members
pendent.
For,
of a finite set of orthogonal functions are linear!}' inde-
if
a,(j),
we should
get,
(x)
(x)+
a.2(f>.^
on multiplying by
(f>p
...
It is obvious that
Example
tt,
1.
tt
~ ^ cost/u;,
tt
From
the functions
1,
2.
mx
1, .r, x-,
From
X, ^^
i,
the functions
1,
which are orthogonal (but not normal)
[A
0,
a^
therefore
all
nugatory.
is
form a
normal orthogonal functions
set of
tt).
which are orthogonal (but not normal)
Example
sin
an(f>n (^v)
and integrating,
(a;)
the coefficients a^ vanish and the relation
foifthe range
225
similar investigation
is
given in
construct the following set of functions
...
for the range
1, 1)
^x- + -^g,
x^-'jx, x*
x, x^,
....
construct a set of functions
...
for the range (a, h)
>5
where
15-14.]
The connexion of orthogonal functions with homogeneous integral
11 "61.
equations.
Consider the homogeneous equation

<l>{x)
= \,\%{^)K{x,^)dl
a
where
Xq
is
a real* characteristic
number
K(x,
for
^);
we have already seen
how to construct solutions of the equation let n linearly independent solutions

be taken and construct from them n orthogonal functions (pi,
... (/>.
;
<^._,,
Then, since the functions
rr
i <l>Uy)f
J a
^, are orthogonal,
K{x,^)cf>,(^)di\'dy
J a |_t = 1
and
it is
=i
f\<p.n(y)fK(x,^)<f>,{^)Ydi/,
m = lJ
J a
[_
easily seen that the expression on the right
may be
written in the
form
rb
i If K(x.^)4>,(^)d^
=
J
>H
H
(.
on performing the integration with regard
i
m.
Therefore,
if
r K {x,
y)
cf>,n
= \J a
we
(y)
write
K for K
{x, y)
i 4>Ay)\'
m=\
*
It will
dyi'
to
y; and this
{x, |)
,/,,
is
the same as
(^) rff
and
for
K{^,^)<l>.nX^)dl
J a
be seen immediately that the characteristic numbers of a symmetric nucleus are
real.
W. M. A.
15
all
226
rb
rb
so
KAdy,
J a
a
rb
rb
and
A?dy
we have
J
A-dij
la
[CHAP. XI
rb
(K - Kfdy.
K'dy J a
Therefore
\m = \
and SO
fô
Xo~'
J a
Integrating,
[K{x,y)Ydy.
{<^,^(^)}^^[
w=1
J a
we get
n
<: Xo'-
[K{x, y)Y-dydx.
J aJ a
This formula gives an upper limit to the number,

corresponding to any characteristic number
7i,
of orthogonal functions
Xo-
These n orthogonal functions are called characteristic functions (or autofunctions) corresponding to Xq.
Now
<^<i'
^'"* (x),
let
different characteristic
Then
</>'>
(x)
(x)
be characteristic functions corresponding to
numbers
</><!'
(x)
Xq, Xj.
=^xJ
{x, ^) (^c) {x)
</)'^'
{^) d^,
J a
and
so
[
(f>^'^>(x)(f)^'^(x)dx
= \,\
K(x,^)4>^'>^(x)cf)''^{^)d^dx
...(1),
and similarly
(/)''
(x)
</)
w (x) dx
= \of
= X[
J a
on interchanging x and
We
{x, ^)
0" (^)
</)(!>
(x)
d^dx
J a J a
la
...(2),
^.
from (1) and (2) that
infer
[ K(lx)cf>^''îx)(f>^'^(^)dxd^
a
'
if
\^\
and if K
{x, ^)
= K{^,
x),
4>^'^(x)(f>^'^(x)dx=0,
J a
and so the functions
<^"''
(x),
^'" (x) are mutually orthogonal.
the nucleus be symmetric and if, corresponding to each

number, we construct the complete system of orthogonal
the functions so obtained will be orthogonal.
If therefore
characteristic
functions, all
Further,
real
if
the nucleus be symmetric all the characteristic numbers are
for if Xq, Xj
be conjugate complex roots and

* V (x)
and w
(x)
being real.
if* Uq {x)
= v (x) + iw (x)
be
>t\
INTEGRAL EQUATIONS
117]
a solution for the characteristic number
then
X^,
number \;
a solution for the characteristic
227"
iii
replacing
(x)
= v (x) iiu (x)
0*'' (x), ^'i*
is
(x) in the
equation
(^'o'
(x)
</><'
(x)
dx =
Ja
by
+ iw (x),
V (x)
iw (x),
v {x)
obviously permissible), we get
= 0,
w (x) = 0,
v (x)
is
[\v(x)Y+ \w{x)Y]dx
f
J
which implies
(which
so that the integral equation has
no solution
except zero corresponding to the characteristic numbers Xq. îl this is

contrary to 11 "23; hence, if the nucleus be symmetric, the characteristic
numbers are
11
The development* of a symmetric
7.
Let
real.
<^j
{x),
satisfying the
(f).,
^3 (x),
(x),
...
nucleus.
be a complete set of orthogonal functions
homogeneous integral equation with symmetric nucleus

cf>(x)
= xfK(x,^)<f>(^)dl
J a
the corresponding characteristic numbers beingf Xj,

XT
Now
when
J.
a^x^
b,
Then
^y %b.
it luill
K(x.y)=i
=i
<^n (*) <î(v)

-^
^^^^
f'
supposel that the series -
X^, X^,
is
-r
umiormly convergent
shewn that
he
M^My).
^
For consider the symmetric nucleus
71
If this nucleus
characteristic
Let
ylr(x)
is
not identically zero,
number
=1
it will
/^n
possess ( 11 '51) at least one
/j,.
be any solution of the equation

^jr{x)=fJ,f H(x..^)ylri^)dl
'
which does not vanish

Multiply by
cf)n
(x)
and integrate and we get
= f,f Hxix,^)- I "t^îf^l^^l ^ (^) cjÛ^) dxd^;
ry!r(x)cf>,,{x)dx
J a
*
This investigation
is
due
t These numbers are not

characteristic
identically.
to
'
n=l
^7i
Schmidt, the result to Hilbert.
all different if
there
is
more than one orthogonal function
to each
number.
The supposition
is,
of course, a matter for verification with
any particular equation.
152
228
since the series converges uniformly,
f ir
Therefore
(x)
t/^
dx
{x)
(.v) 4>
^\\
0.
(^)
orthogonal
is
we may
(/>
to
integrate term by term and get
i^)d^-^r
<î
[cHAP. XI
<^ (I)
(x), <^2 (^),
f (^) d^
and
',
so taking the
equation
rb
we have
yjr
^n
n=l
(x)
y"-
K (x, ^)'^ {^) d^.
J a
Therefore
a characteristic
is
/i,
a linear combination of the (finite

to this
number
number of
number of)
(x, y),
and
functions
so
<^n (*')
-v/r
{x)
must be
corresponding
let
m
Multiply by
functions
(j)n
(j>m
(x),
and integrate
(x)
we
see that a,
then since
0, so,
The contradiction implies that the nucleus

zero
that
is to say,
expanded
y) can be
(x,
(x) is orthogonal to all the
-yfr
contrary to hypothesis,
(x,
y)
yjr
(x)
must be
0.
identically
in the given series, if
it is
uniformly convergent.
Shew
Example.
that, if Xq be a characteristic
4>{x)=f{.v)
number, the equation
+ \J'' E{x, ^)^{^)d^

J a
certainly has no sokition unless f{x)

.
orthogonal to
is
all
the characteristic functions
corresponding to \.
The solution of Fredhohns equation by a
11 "71.
Retaining the notation of
^ {x) =f{x) +\f K (x,

-'
where
If
{x, ^) is
series.
11 '7, consider the integral equation
^)<^{^)
dl
symmetric.
^ (^)
we assume that
can be expanded into a uniformly convergent
00
series
an4>n (1),
we have
00
00
an(f)n{x!)=f(x)+
w=l
SO that f{x) can be
expanded
S
if the
bn<}>n(ix),
then
the
an
series
range
-A
-\
(f)n
{x).
b "^ \
"^
n-l ^n
n=l
the
'
function f{x) can be expanded into the convergent series
<
an(f>n(^),
M=l ^n
in the series
00
Hence
{a, b), is the
solution of
\^
<f>n{^),
Fredhohns
if
it
equation.
converges uniformly in
ir71-ir8l]
INTEGRAL EQUATIONS
229
X
To determine the
coefficients
6,^
we observe
that
bn (f)n (^) converges uni-
n=l
formly by
3'35*; then, multiplying by
and integrating, we get
(f)n(x)
J a
Solution of AbeVs integral equation.
11 "8.
This equation
is
of the form
d$
^^^^-llw
where/'
(x) is
Let (f)(x)=
continuous
aud/(a)=0; we proceed
a^^x^b),
{0<fi<l,
to find a continuous solution u (x)
u (I) o?|, and take the formula t
J a
7T
sin/iTT
multiply by u
(^)
and
integrate,
and we
dx
{z-xY~>^{x
^y
on using Dirichlet's formula
get,
( 4*51
f{x)dx^
f'
Since the original expression has a continuous derivate, so has the final one
continuous solution, if it
it
can be verified
11 "SI.
l)y
therefore the
can be none other than
exist,
""^'^^
and
corollary\
'V~dzj{z-xy->^'
substitution
that this function actually
is
a solution.
Schlomilch's^ integral equation.
Let f{x) have a continuous differential

2
f{x)=^
coefficient
when
tt
^.r
^ tt.
Then
the equation
/"i"
(f){xsmd)de
J n
has one solution with a continuoiis differential
when -
coefficient
it
^x ^ it,
namely
/in
<^{x)=f{0) + x
From
4-2 it follows
that
we have
* Since the
(0)=/(0), ^'
numbers
other positive, the
X,^
members
evident that X/(X - X)
is
f'{xsm6)d6.
that
f
(so
'
Jo
(.v)
=^
(0)
sin
Oct)'
Zeitschrift filr
(9)
dd
= |7r/' (0)).
are all real
we may arrange them
in each set being in order of
in two sets, one negative the
magnitude
then,
when
'
>
X, it is
a monotonic sequence in the case of either set.
t This follows from 6"24 example 1, by writing

X For the details we refer to Bocher's tract.
Math, und Phys.
to a case of Volterra's
{x sin
ii.
(1857).
(z
x)l(x
The reader
equation with a discontinuous nucleus.
f ) in place of x.
will easily see that this is reducible
230
Write
.r
sin
for x\
^//
Change the order

in iDlace of
\f/,
and we have on multiplying by x and integrating
/' {x sin
d\lr
\//-)
<
sin
^(^'
(.i-
sin 6 sin
fi"^
.,
fi^
2.V
,,
.s
= sin 6 sin
;^
f^
\//-)
c?^
\ dyjr.
and take a new variable x
of integration in the repeated integral ( 4'3)
defined by the equation sin
rr,.
[CHAP. XI
yj/.
(x sin v) cos Y dv]
(b'
,^
xj^f'ixsn.i.)d^=^j^\j^^^-~J^^-~^-^\de.
Then
Changing the order of integration again

H'^
P"^'/ix sn. ,^^l = xj^f

^) ^^
2jt;
^
But
-77
n/(COs2
Jx
x
and so
Jo
;^
j^
- cos-5^N=
^)
f {x sin
/"i'r
sin^rf<?
/i'^
( 4'51),
y\r)
(/)'(A-sinx)cosxsin^
\/(sin^'^-sin
/cos^Xn^'T
-arc
=^7r,
sin
\COS x/ Jx
|_
d^ = x
</>'
^'/ ^^^
sin x) cos
(.r
;>(
o?;^
= 0Gt-)-(^(O).
Since
(f)
(0)
=/ (0), we must
(j)
have
(.r)
=/ (0)+x
'
and
it
f (x sin
;/.) câ/^
can be verified by substitution that this function actually
is
a solution.
REFERENCES.
H. Bateman, Report
M. BocHER,
to the
British Association*, 1910.
Introditction to Integral Equations
(Cambridge Math. Tracts, No.
10,
1909).
H. B. Heywood
et
M. Fr^chet, Liquation de Fredkolm.
V. Volte^ra, Lego7is sur
les
equations integrales
et les
(Paris, 1912).
equations integro-differentielles
(Paris, 1913).
T. Lalesco, Introduction
I.
a la
the'orie
des equations integrales (Paris, 1912).
Fredholm, Acta Mathematica, xxvii.
(1903), pp. 365-390.
D. HiLBERT, Orundzilge einer allgemeinen
Theorie der linearen Integralgleichungen
(Leipzig, 1912).
E. Schmidt, Math. Ann. lxiii. (1907), pp. 433-476.

E. GouRSAT, Cours d' Analyse,
iii.
(Paris, 1915), Chs. xxx-xxxill.
MiSCELT.ANEOUS EXAMPLES.
if the time of descent of a particle down a smooth curve to its lowest
independent of the starting point (the particle starting from rest) the curve is a
Shew that
1.
point
is
(Abel.)
cycloid.
*
here.
The
reader will find a
more complete bibliugrapby
in this
Report than
it is
possible to give
INTEGRAL EQUATIONS
2.
Shew
231
that, if/(^) is continuous, the sohition of
(f)
(x)
=f {x) + \
COS
+X
/(.r)
(2.rs)
(s)
<f)
ds
f{s)cos{2jcs)ds
I
*(^>=
'"
Hrjxv
assuming the legitimacy of a certain change of order of integration.

3.
Shew
that the Weber-Hermite functions
satisfy
4>{x)
=\
e^ '*-^
(s)
ds
for the characteristic values of X.

4.
Shew
(A. Milne.)
that even periodic solutions (with period Stt) of the diflferential equation
^-^ + (2 +
>l-2
cos2 x)
cf)
(.r)
satisfy the integral equation

<f>{x)
5.
Shew that the
=\
<f}
6.
Shew
is
any
(x)
= cos m.v,
(x)
Shew
sin inx,
integer.
that
<p
has the discontinuous solution

7.
19-21.)
characteristic functions of the equation
are
where \=m'^ and
(Whittaker; see
e^<ôo'=<^^^'(f,{s)ds.
"^
<f)
^-^-f
(x)
= Lr^ ~
(|)
d$
(Bocher.)
that a solution of the integral equation with a symmetric nucleus
f{:v)=rK{x,^)<t>i$)di
J a
0(^)= 2
is
a\^(f)^(x),
n=l
provided that this series converges uniformly, where X, 0
numbers and functions
of
{x, |)
and 2 a0
(x) is the
(x) are
the characteristic
expansion off(x).
n=l
8.
Shew
that, if
< 1,
the characteristic functions of the equation
'f'^''^=LI[^
are
1, cos?n.r,
m takes all
sin
i-2hcL{i-^Hh'' ^
^^^ "^^
mx, the corresponding characteristic numbers being
positive integral values.
1, l/A'", Ijh'",
where
PART
II
CHAPTER
XII
THE GAMMA FUNCTION

12"1.
Gamma-function.
Definitions of the
Historically, the
The Weierstrassian product.
Gamma-function* V{z) was
first
defined by Euler as the
limit of a product ( 12*11) from which can be derived the infinite integral
1
i^~ê~^dt;
but in developing the theory of the function,
it is
more con-
Jo
venient to define
by means of an
it
product of Weierstrass' canonical
infinite
form.
Consider the product
7=
where
[The constant y
exists
we observe
IS
positive
and
is
less
hm
The value
1 -H
known
log
as Euler's or
?/i[
=0-5772157....
Mascheroni's constant
than
[^ dt
/
J-
+ -+...+
,
n^
of y has been calculated
.a<=

log
values of z
to prove
that
it
-,
"^
,; therefore 2 w converges,
'i^
n=l
"'
,.
logwl= hm
1
by
J. C.
<
Adams
?<
+ log
for, if
and
m + l\
\=
*
2 .
to 260 places of decimals.]
The product under consideration represents an

all
tliat, if
Jo
,.
-I
lim ]-+--f-...-l-
u=
fl
ze^^
analytic function of
N be an integer such that \z\^ îV, we havef
if
z,
for
> N,
236
S
of
]log(l
any integer
and
exponential
its
so zey^
an absolutely and uniformly convergent series
is
functions,
analytic
sequently
+-
so
is
$l+-]e~nl
is
5'3)
an analytic function, and
an analytic function when \z\^^N, where
is
con-
is
that
is to say,
,=^-T-
from this equation

1,
it
11
the product
The Gamma-function was
z=0,
[CHAP. XII
2,
it is
where
is
analytic for
all finite
values of
z.
defined by Weierstrass* by the equation
zey^
U \(l+-]e
T (z)
apparent that
is
~n[analytic except at the points
has simple poles.
it
Proofs have been published by Holder t, Moore J, and Bai-nes of a theorem known to
Weierstrass that the Gamma-function does not satisfy any differential equation with
rational coefficients.
Example
1.
Prove that
r(i) =
where y
is
r'(i)=-y,
i,
Euler's constant.
[Justify differentiating logarithmically the equation
~=
Viz)
by
4-7,
and put
Example
2.
=l
zey''n\(l+ -)e~ri\
I
nJ
IV
1
after the differentiations
Shew that
,11
2
and hence that Euler's constant y

lim
Example
3.
have been performed.]
Shew
is
P 1 -(1-""*,
Jo
given by||
[/:{'-04)"}?-/:(-r.)"T']'
that
e>^r(g + l)
n = \\\
*
(
+ n)
T{z-x + $'
Journal filr Math. li. (1856). This formula for F {z) had been obtained from Euler's formula
by F. W. Newman, Cambridge and Dublin Math. Journal, iii. (1848), p. 60.
12-11) in 1848
t Math. Ann. sxvm. (1887), pp. 1-13.

+ Math. Ann. xlviii. (1897), pp. 70-74.
Messenger of Math. xxix. (1900), pp. 122-128.
li
The reader
will see later ( 12*2
example
4)
that this limit
may
be written
THE GAMMA FUNCTION
12-11, 12-12]
12-11.
By
Elder s formula for
the
Gamma-function.
we have
the definition of an infinite product
=z
T{z)
lim e^
lim
;n-*-x
...
logni
lim
=z
-.+
...
237
login]
+ -le
K[[^^>1\
"
238
[CHAP. XII
Prove that
Example.
111
[Consider the expression
+ l)(2 + 2)
2(s
z{z-\-\)
z{z
+ l)...{z+m)
in
It
_(-)/
m-^<X)
when
z is
itt-t
By means
of the Gamma-function,
+n
11
;h-^0 as
2
[r=m-n+irl]
infinite products.
Un,
=l
we can
n.
Avrite
the product in the form
-a^)(n-a2) ...(n- ak)

(n-b,).:{n-bi)
I'
(n
n=i[
supposed that no factor in the denominator vanishes.
In order that
numerator must
denominator, and
would not tend
We
possible to evaluate the general
is
it
any rational function of the index
For, resolving m into its factors,
it is
products of the form
7t
is
not a negative integer.]
class of infinite
where Un
where
prove that 2 ^ .n=o n\
The evaluation of a general class of
12'13.
"'(-)"!(="
{m-n + l)\
r=m-ri+\r\
]_(-)"
\^
"=1
,<-,
2
Noting that
and
d^
can be expressed in partial fractions in the form 2
this product
clearly
A ^1;
also
may
converge, the
for,
of factors in the
otherwise, the general factor of the product
to the value unity as
have therefore
number
be the same as the number of factors in the
k=
I,
n tends
to infinity.
and, denoting the product
by P, we may
\vrite
'
=i
The general term
]{n-b^)
{n-bk)i
...
in this product can be written
(-^(-^)(-r-(-r
=
where ^4^
is
(h~-)
when n
In order that the
+ a^+
a^
+ak-b,-
...
product
infinite
h Jin}
is large.
may be
therefore necessary further ( 2*7) that

tti
...-bk
Ojk
^1
^/fc
= 0.
it
is
THE GAMMA FUNCTION
12 13, 12-14]
We
239
can therefore introduce the factor
exp
{/i-i (!
...
+ak-b,-
...
hk)
into the general factor of the product, without altering its value
and thus
we have
(i_^),^(i_^)e"...(i-^*)."
p= n
(-^)--(-^)-
"=N
But
it
obvious from the Weierstrassian definition of the
is
Gamma-
function that
,"|('")''"]"-^r(-^)e-T''
P=
and so
^r(- 6,)6,r(-6^^^(-/,.) ^ ^ TJ^l-K)

a^T{-CH)...akV{-ak)
= ir(l-a^)
a formula which expresses the general infinite product
in terms of the
Gamma-function.
Example
Prove that
1.
(a-i-6 + )
"
s=i
Example
Shew
2.
(a+)
that, if a
\ /
(6-l-s)
_r(a + l)r (6-1-1)

T{a + b + \)
= cos (27r/n) +
sin {27r jn\ then

^
'
^(i-^j(i-|;)... = {-r(-x)r(-aa-)...r(-a"-i:r)}-'.
1214.
We
Connexion betiveen the Gamma-function and the circular functions.
now proceed
to establish another
most important property of the
Gamma-function, expressed by the equation
r(^)r(i-.^)= .
sm TTZ
We
have, by the definition of Weierstrass
( 12"1),
r(.)r(-.)=-in{(i+i)r^rnj(i-i);"
Z Sin TTZ
by
7-5
example
1.
Since,
by
12-12,
r(i-z) = -zr(-z)
we have the
result stated.
240
Corollary
we
If
1.
assign to
the formula of Weierstrass, r (^)
Corollary
since,
= 7r cot
(1 -z)--/r (0)
i/^
tts.
The multiplication-theorem of Gauss* and Legendre.
12-15.
We
then
formula gives {r(^)}2 = 7r
\, this
we have
is positive,
= r'(s)/r(2),
If r/.(s)
"l.
the value
[CHAP. XII
shall next obtain the result
r(.)r(. + ^)r(. + ^)...r(. + 'i^) = (2.)n-.-^-rM.
For
let
n)
(^(.)
have, by Euler's formnla
Then we
,.
n lim
\im
i(m
1)
...
nz (nz
!j
(m
)
1)
(z) is
(nm)'*^
+ n7n
1)
{nm)'"
'n
wi"^
evident from this last equation that

(j)
1)
(71Z
. . .
(wm-1)!
môo
Thus
l).m^-^/'*
1)!}
{n'ln
is
.2...(m-
lim
^(Z)^
It
_.
12-11 example),
n'"-^
îT^;^;^
equal to the value which
Therefore
{</,
(z)Y
n|
<p
(l
independent of
when
has
it
(^-)
(z) is
sm sm
n
Thus, since
cj)
27r
;:)|
(27r)"-^
.
...
so
tt'^-i
TT
and
sm
(?i l)7r
(n~^) is positive,
<#,W = (2,r)4<"-''-*,
i.e.
Corollary.
(.)
(.
^)
(.
'-iî)
Taking n = 2, we have
(2,r)4 <
'>
r (.).
2^^-ir(.)r(^+i)=,r4r(24
This
is
called the duplication formula.

*
Werke,
iii.
p. 149.
The case
in
which
z.
was given by Legendre.
by
THE GAMMA FUNCTION
12-15-12-2]
Example.
If
^, o)
241
= EMlM
shew that
1216.
We
D(p,,)b{p + \ q)...B(p^'^,q)
no) = a
B (np,
Expansions for
the logarithmic derivates of the
{r (2+l)}-J
have
'>'^
Gamma-function.
+ -") e~]-
/("l
DifiFerentiating logarithmically ( 4"7), this gives
l^gJîiiLi)= _ '''^1
+
Therefore, since log r
(2
(s
--
have
Iogr(2)=-y--.+
2 2
'
n=in(2 + n)'
d-.
_,
(f
,.
^,logr(.+
l)
^{^^ +
(2+l>'^
These expansions are occasionally used
Elder s expression of T
The
infinite integral
(2
+ 2)2
in ai)plications of
{z) as
(2
+ 2) + -}
122.
+ 1 = log 2 + r (2), we
dz
Differentiating again,
^
+
""
+ 1) '*"2(2 + 2) "^3(2 + 3)
!_
rfs
an
""
the theory.
injinite integral.
e~H^~^dt represents an analytic function of z
Jo
the real part of z
Second Kindf.
equal to
(z).
positive ( 5-32);
is
it is
now be shewn
It will
Denoting the
when*
called the Eulerian Integral of the
that,
real part of z
when li{z)>0, the integral is

x, we have x > 0.
Now, if J
by
n(z,n)=j'Ul-yJ\'-^dt,
we have
(z,
n)
n^
(1
- TyW-'hlr,
\)
we write t= nr; it is easily shewn by repeated integrations by parts

when a; > and n is a positive integer,
if
f^
{1
t)"t^-^
dr
Jo
n
11 (2,
J
and
so
Hence, by the example of

*
If the real part of z is
larity of the integrand at

t
n-
- Tf
1^
n ;!
(I
- Tf-'T'dr
\J
Jo
z(5
n(n-l)...l
fi
+ l)...(^ + n-l)Jo
n)
z{z^
12*11, 11
{z,
2
.
. . .
'*^'
?i
\) ...{z^-n)
n)
S'.
^ r(^)
n^
as
/i
-*
not positive the integral does not converge on account of the singu-
= 0.
The name was given by Legendre
t The many-valued function

purely real.
W. M. A.
T- (1
that,
<^~i is
see 12-4 for the Eulerian Integral of the First Kind.
made
precise by the equation *~i
= c(*-i)'s',
log
being
16'
242
Consequently
And
(^)
T, (2)
so, if
(1
lim
I
J
[CHAP. XII
t-~^dt.
e-H'-'dt,
(\
we have
ri(^)-r(^)=
r ie-'-(l-^y\t'-'dt+j
lim
Now
lim
e'U'-'dt
e-H'-'dt
= 0,
since
e~H^~^dt converges.
.'0
To shew
formula
that zero
for Fj {z)
the limit of the
is
{z)
first
[To establish these inequalities, we proceed as follows
from the
and
series for e^
(1
3/)"^
1
Writing
+-
two integrals in the
when
< y < 1,
we have
tjn for y,
ê-'^
^
and
of the
we observe that
HJ-
0ê--fl--
so
71
= e-Ml-eM 1 -1)1
<e-MlNow,
7ia
^1
if
O-Sa^l,
and, writing
(1
a)"^l na by
t^/n^ for a,
is
induction
when ?ia<l and obviously when
get
0ê-*-(l--)
and so*
which
we
%e-H^ln,
the required result.]
From
the inequalities,
r'
it
follows at once that
--
t'-'dt
'
n-'e-ff'-^'dt
<n-'j
e-H'^-^'dtÔ,
Jo
as n -^
*
since the last integral converges.
This analysis
Analysis,
ii.
p. 243.
a modification of that given
is
for the object required)
by Schloinilch, Compendium der hoheren
simple method of obtaining a less precise inequality (which

is
given by Bromwich, Infinite Series, p. 459.
is sufficient
THE GAMMA FUNCTION
12-21]
Consequently Fi {2)= V
converges; that
is
when the integral, by which Tj (z)

when the real part of z is positive,
(z)
to say that,
And
so,
when
the real part of z
this integral or
243
defined,
positive,
is
is
(z)
may be
defined either by
by the Weierstrassian product.
Example
1,
Prove that, when
Example
2.
Prove that,
R (z)
if li {z)
>
is positive,
and
R (s) > 0,
e-"=x-'cla.-=^.
Jo
Example
3.
Prove that,
Example
4.
From
12'1
^:;
if
R{z)>Q and R
example
2,
is)
>
1,
by using the inequality
0%e-^-(\-^X ^fie-^ln,
deduce that
<
12*21.
l_e-_e-i/<
-at.
t
Extension of the infinite integral
Gamma-function
to
the case in
which the argument of the
is negative.
The formula of the

Cauchy* and
last article
negative.
an analogous theorem
is
ni>
longer applicable
when the
real part of z is
Saalschiitzt have shewn, however, that, for negative arguments,
This can be obtained
exists.
in the following
way.
Consider the function
r,{z)=j\'-^(e-'-\ + t-^, + ... + {-)>'-^^^dt,

where k
By
is
k>x>-k-l,
when s < 1,
the integer so chosen that
partial integration
we
have,
x being the
real part of
z.
r,(.)=[f(.-.-n.<-il + ...+(-r.^,)];
+.'j7''(^-'-i+'--+(-''(j^)*
The
positive
integrated part tends to zero at each limit, since

:
so
x+k
is
negative and
r,{z)
The same proof
applies
when x
lies
= lT,{z+i).
between
and 1, and leads to the result
(0>.r>-i).
r{z+i)=zr2{z)
The
last
x+k+l
we have
equation shews that, between the values

To{z)
* Exercices de
and 1 of a*,
= r(z).
Math. u. (1827), pp. 01-92.
t Zeitschrift fur Math, und Phys. xxxii. (1887), xxxiii. (1888).
162
is
244
The preceding equation then shews that

values of Ji{z) less than 1.
Cauchy and
where k
Thus, for
is
{z)
the same as T
negative values of R(z),
all
(2)
for all negative
we have the
result of
Saalschiitz
the integer next less than
is
To
[CHAP. XII
Example.
If a function
(z).
be such that for positive values of
(ft)
/x
we have
1 ^-x
e~'-' dx,
J
and
negative values of
if for
/x
we
define Pj
by the equation
(fi)
Pi(;u)=|\i'--i(e-^-l + .r-... + (-)*- + i^,) dx.

where k
- ju, shew
the integer next less than
is
that
/',W./'W-l + ^^-... + (-)'--.-^,.

Hankers expression of V {z) as a contour
12"22.
The
integrals obtained for
T {z)
(Saalschiitz.)
integral.
in 12-2, 12-21 are
members
of a large
by which the Gamma-function can be defined.

The most general integral of the class in question is due to Hankel*; this
class
of definite integrals
now be
integral will
Let
D be
investigated.
a contour which starts from a point p on the real axis, encircles
the origin once counter-clockwise and returns to
Consider
when the
(t)~~ê~hlt,
p.
real part of z is positive
and
z is not
an integer.
The many-valued function ( ty~^

that ( ty~^
= e'^~^' ^^ '"*' and
log
part of the real axis, so that, on D,
The integrand
of integration
is
h,
ir
to be
made
definite
when
purely real
% arg (
t)
not analytic inside D, but, by
may be deformed
into the path of integration

to
is
t) is
by the convention
is
on the negative
it.
5'2
corollary
1,
the path
(without affecting the value of the integral)
which
starts
from
p,
proceeds along the real axis
describes a circle of radius h counter-clockwise round the origin and
returns to p along the real axis.
On
so that
the real axis in the
part of the
On
f
ty~'^
first
e~^'' ^^~^H^~^
new path (-
the circle
we
{-ty-'e-Ht-=
ty-^
write
\
part of this
(where log
= e^"
i=
'^-^'
8e*^;
new path we have arg ( ^) =

is
t^'^.
then we get
e-'-(--i)^'-ie-'f?^+f"(8e')--ie(cose+;sm<)gg,ej^7^
=*
2i sin
{-jTz)
Zeitschrift filr Math,
t'-'
e-hit
und Phys.
tt,
purely real); and on the last
"
ih'
{''e'^'^^-'^H'-'ê'^dt
e<>e+(cose+tsine)^^_
ix. (1864), p. 7.
THE GAMMA FUNCTION
12 22]
This
is
true for
gize+s
and
JI
(COS
positive values of S
all
0+isin e^
^^0
TT
îzeîQ
j
245
^ p; now make
^[^^^^
^\^q
then
integrand tends to
8'
limit
its
J IT
uniformly.
We
consequently infer that
(-ty-'e-'dt
This
is
= -21
true for
sin {7r2)[''t'-'e-*dt.
Jo
positive values of p
all
make
p^ x
and
let
C be
the
limit of the contour D.
Then
f
J
Now,
2i sin {ttz) f t'-'e-^dt.
Jo
T (^) = - r^4
'2i sm
Therefore
is
=-
(- ty-'e-^dt
c
since the contour
I (- tf-'e-'dt.
TTzJ c
does not pass through the point
no need longer to stipulate that the real part of 2

(ty~ê~^dt
c
Hence, by
is
a one-valued analytic function of z for
is
= 0,
positive
all
there
;
and
values of
2.
just proved when the real part
5"5, the equation,
0/2 is
1, +2
positive,
and make use of
12"14,
persists for all values of z with the exception of the values 0,
Consequently,
for all
except integer values of
r(0)
This
is
we get the
= -_^J
Hankel's formula;
if
{-ty-'e-'dt.
we write
^,
2 for z
further result that
V
'(0
We
+)
meaning thereby that the path of inteC

gration starts at 'infinity' on the real axis, encircles the origin in the positive
shall write
for
re
direction and returns to the starting point.
Example
1.
Shew
that, if the real part
constant, l{ t)~ê~^dt tends to zero as p^-ao
of
,
be positive and
when the path
if
a be any positive
of integration
is
either of
the quadrants of circles of radius p + a with centres at a, the end points of one quadrant
being p and a + 1 (p -|- a), and of the other p and a i{p-â).
24G
"
lim
'\-t)-' e-^dt= lim
p-*xy-a + 'P
p-^y^
Deduce that
{-t)-'e-^dt,
[CHAP. XII
and hence, by writing t= a-iu, shew that

['^
\
\
-= -
"*-^"
(a
_x
ztt y
(s)
+ <)"* f^M-
[This formula was given by Laplace, Theorie Analytique des Prohahilites (1812), p. 134,
it is substantially equivalent to Hankel's formula involving a contour integral.]
and
Example
By taking a = 1, and putting t= \ + i tan
2,
""
-J = r
Example
shew
that, if
(2)
> 0,
cos (tan
(9
- 2(9)
cos^ -
1,
shew that
2 (9c?<9.
Jo
as contour of integration a parabola whose focus
By taking
3.
77
B in example
is
the origin,
then
Y{z)=-.
e-''(l
sin nz J
+ ^-)^-Jcos{2a^ + (22-l)arctani;}(/i!.
(Bourguet, Acta Math,
Example
4.
Investigate the values of x for which the integral

2
TT
converges
that
it is
i.)
for such values of
r*
^^-^sin tdt
x express
it
in
terms of Gamma-functions, and thence shew
equal to
(St John's, 1902.)
Example
of
example
Prove that
5.
4,
evaluate
Jot
when
it
(log
m=l
'
and when
when
converges
dt
t)"^
m > 0,
m = 2.
and, bv
means
(St John's, 1902.)
Gauss' expression for the logarithmic derivate of the Gamma-function
12'3.
as an infinite integral*.
We
integral
shall
now
express the function
when the
frequently written
Jo
real part of z is positive

yjr
Take the formula
where A = l-e-^ since
We
(z).
( 12'2
J\
=1
-e~"
y=
5^0
This
is
(z)
an
as
infinite
the function in question
need a new formula
for 7.
4)
Us
J^
= log
aôlJA^
J&
J'
^.-^Ô as 8-*-0.
l-e"^
in the first of these integrals
lim I
5-*o
first
example
./a
^Yriting
log
-7-
and then replacing u by
e-^dt.
r ^,dt- Jsr '-^dt] = Jor \j~^,-]]
U-e
Us l-e
the formula for y which was required.

*
Wtr'ke, HI, p. 159.
'
t)
we have
is
THE GAMMA FUNCTION
12-3]
To get Gauss'
formula, take the equation ( 12"16)
r'(^)
T{z)
and write
z -^
is
permissible
m = 0,
when
/I
..
z-vm)
âcm=iVm
'
this
247
r e-'(^+'">rf^;
Jo
1, 2, ... if
the real part of z
positive.
is
It follows that
^'=-7- .0
-7+
lim
1-
< ^ 1,
<
and when
(e-""-e-('+^)')f^<
rt->ooJowi = l
- e-'
Now, when
e-^^?^+ lim
r(^)
1-e1 e
is
W-
dt
lim
-e-(''+'(/t
^1
a bounded function of
^-^^ < -^J-^^ < ^^^
1,
whose
limit as ^-^[0 is 6nite
'I
.
Therefore we can find a
number
A'
independent of
1
I
'
r\zl~^,-K^^'^Vdt\<K[
andso
We
jo 1-e-'
such that, on the path of integration,
e-i"*')'
rf
= /r(H + l)-'^0
as /i-^-x
./o
have thus proved the formula
t<^) = s'"'^*'->=.C(T-r^')'"which
is
Gauss'
of -^{z) as an infinite integral.
expression
It
may be
integral which we have encountered connected
remarked that this is the first

with the Gamma-function in which the integrand
Writing <=log
(1
-f-.i')
in Gauss' result,
we
get, if
A=(J
is
a single-valued function.
1,
^-if=limf^t'-,^,U.
<
since
T'(z)
Hence
sothat
^
dt
y =log g- ^0 as
<
j
11
= ûaj^|,-.____|_,
,.
/"
rf.r
''<=)=rM/J{e-'-(^,}',
an equation due to Dirichlet*.

Werke,
i.
p. 275.
8^0.
248
Example
Prove that,
Example
Shew
the real part of
if
y=\
that
z is
positive,
(Dirichlet.)
{{l^-t)-'^ -e-^]t-'^dt.
Binet's first expression for log
12'31.
[CHAP. XII
T {z)
terms of an infinite integral.
in
Binet* has given two expressions for logr(^) which are of great
To
importance as shewing the way in which log V {z) behaves as 2 -* oo
.
obtain the
first
of these expressions,
we observe
that,
when
the real part of
z is positive,
r'
r(^
writing z
+\
Now, by
for
2^
r*
_L 1 \
/'-r
i)
so,
dt,
e^-i
[^
Jo
in 12"3.
6'222 example
6,
we have
r~^^'
log2=)
and
o-tz
(o-t
(22^)"^
since
f" 1
e~^^
^
'Jo
dt,
we have
dz
logr(. +
The integrand
-
+ f
in the
bounded
is
z.
= l + log.-/"g-;+^-^Je-*.
l)
as
integral converges uniformly
consequently integrate from
we
integral
last
-^ 00
it
when the
1 to ^
is
continuous as
-^
and since
follows without difficulty that the

real part of z is posijiive;
under the sign of integration
we may
( 4*44)
and
get-f-
iogr(.+i) = (. +
Since
\-^
-^
|)iog.-.H-i+/;g-^ + ^p--^-
î
log
^^
continuous as
r (z +
1)
= log^ +
^-
by
log
r (z),
7 '2,
dt.
and since
have
/
iogr(.)
("^
1\
CI
e-*^
= (.-2)iog.-. + i.-j^|--- + ^-^_--^|-^.
(2
e'^
l)t
* Journal de I'Ecole Poly technique, xvi. (1839), pp. 123-143.
Logr(2 + l) meaus
the
the Weierstrassian product.
sum
of the principal values of the logarithms in the factors of
THE GAMMA FU^XTION
12 -31]
To
249
evaluate the second of these integrals, let*
so that, taking z = ^ in the last expression for logr(2:),
i log
7r
we
get
= i + t/-/.
dt,
we have
r^'
\ dt
-jo
e'-\)
=/:r-^'-HT
~jo
c/^V
'2tj
'
rf<
= 2 + Uogi
/=1 ^log(27r).
Consequently
We
therefore have Binet's result that,
real part of z is positive,
when the
3tZ
log r(^)
li z
= [z-^^\ogz-z +
= x-\- iy, we see that,
values of
is
log(27r)
+
J^
the upper bound of
if
G
(
^dt.
^"'-t)
"^
when x
is
I.
Prove
terms (z
large, the
approximate expression
Example
that,
for log
when
2.
*^^ '^^^
'
Prove that, when
* This artifice
is
due
to
- ^\og z -
\-
^"'"^^
^\o^
(2'jr)
furnish an
V (z).
A' (a)
logr(.-)=
Example
_i
K, then
logr(^)-(^-i)log^ + ^-|log(27r)'<i(:|^
so that,
> 0,
r {*i4.1-r+(^-l)"j 7-
(Malmsten.)
R (z) > 0,
Priagsheim, Math. Ann. xxxi. (1888),
p. 473.
250
Example
From
3.
the formula of 12" 14, shew that,
<;r
if
[CHAP. XII
< 1,
2logrW-log.+logsin..-=/; |-'Ûi=fl-'-(l-2.).-.}f
(Kummer.)
Example
sinh (| .v)^ and
By expanding
4.
example 3 that,
1 2x'
in Fourier sine series,
shew from
< < 1,
if
.v
00
log
(.p)
= i log TT
/"
jo V^ + 4wV2
"
a,i
e-n
2?i7r
Deduce from example
log sin -nx + 2 2
-^
sin 2?i7r^,
dt
2%7rJ
i!
2 of 12 "3 that
"*" ""
2^
^"^ "*"
^^^ 2"" + ^*^g
''*
(Kummer, Journal fiir Math. xxxv.

12*32.
Binet's second expression
(1847), p.
for log V {z) in terms of an
1.)
infinite
integral.
Consider the application of example 7 of Chapter vii

equation
The
the real part of z be positive
Since
145) to the
conditions there stated as sufficient for the transformation of a
series into integrals are obviously satisfied

if
(p.
( 12-16)
+ /i)
^^
!5'(^,
is
and
easily seen to be less
n, it follows
Hence
Since
V;: log
^
dz^
~ 2^ -
I*
(^)
^
than K^t/n, where Ki
that the limit of the last integral
= 2,
\-
-I
2z-
does not exceed
r^,
and we have
pendent of
by the function ^(^)=
Jo
{z^
(where
+ t'f
is
inde-
is 'zero.
-.
e-^^
K depends only on 8) when the
real
part of z exceeds
8,
the integral converges uniformly and
integrate under the integral sign
We
4'44) from 1 to
we may
z.
get
log
-I
where C
is
log
where 6"
is
(.)
= - ^;+ log
a constant.
r (.) =
(.
^^^.^^JL
/;
Integrating again,
a constant.
1) log .
(c
1) .
c + 2/; '-^^^^dt.
THE GAMMA FUNCTION
12-32, 12-33]
Now,
and
if
^ arc tan
is real,
tjz
251
$ tlz,
so
logrW-(^-i)log^-(C-l).-C"|<?/"^d(.
But
has been shewn in
it
12"31 that
\^^%^{z)-[z--^\ogz-^z-^\og{1'K) -0,
as 2 -*
through
Comparing these
real values.
results
we
see that
C=
0,
6"=ilog(27r).
Hence
for all values of z
whose
real part is positive,
logrW = (.-l)log.-. + llog(2.) + 2|;-I^^^rf*.

where arc tan
by the equation
u is defined
arc tan
xi
.'o
in
which the path of integration

This
l+<-
a straight
is
Binet's second expression for log
is
Example.
line.
V {z).
Justify differentiating with regard to z under the sign of integration, so as
to get the equation
The asymptotic expansion of the logarithm of the Gamma-
12 33.
function (Stirling's series).
We
can now obtain an expansion which represents the function log
asymptotically ( 8-2) for large values of \z\, and which

calculation of the Gamma-function.
Let us assume that,
if
^^
= x -f iy,
then
a-'
>
r (^) = (^ -
where
^(.)
^)
log ^
= 2/;
- log (27r)
</)
(^),
^^^^^c/.
Now
arc
,,,
It'
(_)n-i^2n-i
(_)n-i ^2,1-1
It'
(_)H
(_y^
rt
u^n^^^
u^
Substituting and remembering
( 7-2)
Jo e'^'^^-l
that
~4w'
{z)
and we have, by Binet's
second formula,
log
used in the
is
+ z'
'
252
where Bi, Bo,
^
(-)'"'
we have
are Bernoulli's numbers,
...
.=1
2r
(2?-
Br
2 (-)'*
^^'--l
1)
^271-1
Let the upper bound * of
+ ^^
w^
[CHAP. XII
"
r
j
U'^'
2^2
()
du
dt
g2,r<
^2 j
positive values of
for ^
_I
u be K,
Then
f
u'^'du]
dt
r^
ftl
uow^ +
e-'^'-li"
^'^j
r"
Jo
'
'
(^^
\e'-'-l
[Jo
^n+1
Ji-z
^4(n+l)(2M+l)|^|2"
Hence
2(-y^
z^"-' Jo
_d^
u^^l
f*
Vou' + z^
KA+^
-l^ 2{n + l){2n + l)\z p+i
e^^^
and it is obvious that this tends to zero uniformly as ^ ^where ^tt > A > 0, so that K^ ^ cosec 2 A.
|
Also
clear that if arg
it is
^r
first
71
^ Itt
Kz =
(so that
oo if
arg ^ ^ ^ tt
- A,
l) the error in taking the
terms of the series
(-y-'Br
rti 2r (2r
as
an approximation
Since, if
to
(z) is
<f)
1) z^-'
numerically
than the (n
less
arg z\^^7r A,
-,2111
!</>
(^)
- i
r=i
Z '^\
^
2r(2r-l)l
cosec^ 2 A
<
^,
2(n
l)(2,
l)(2n
l)th term.
l)
-0,
as
^^
^^
00
it is
clear that
B,
B,
1.2.2
is
is
3. 4.
the asymptotic expansion f
We
( 8-2)
2=*
'
B,
5.6.2^
"*
of ^(2).
see therefore that the series
the asymptotic expansion of log
V {z) when
arg ^ ^ ^tt
- A.
-Kj
^ 18
the lower bound of
4^2y2
,
-!
.n o
^^
.j^
or 1 as
and
is
consequentlj' equal to
x2<m2
or
x^>y^.
t The development is asymptotic for if it converged when 2 ^ p, by 2-6 we could find K,

such that B<(2;i-l)2nA>2"; and then the series 2 ^~^"~'^"^^" would define
an integral
;
function
this
is
contrary to
7-2.
THE GAMMA FUNCTION
12 4]
This
known
generally
is
In particular when z
Uo
Hence, when
n terms
and
the
(=
tt
x),
(x)
(f>
Hence
ix)
=e
This
.r- ^
~
<
+ l){2n +
2{n
l)af
of <})(oc) always lies between the sm7i of

terms of the series for all values of n.
so that
<f>
(./)
= y^ where
<
<
1.
/'I
we get
571
139
-,1
= x'-^- e''^ i2'rrf e^^'^^'l
^ ^^^-^ 288^ ~ 51840^ ~ 2488320^
(27r)
be estab-
value
1
T{x)
_x
it will
Bn+.
dt
Also, taking the exponential of Stirling's series,
13'6
we have
<
In particular
A.
+ ^je^'^'-l
^j e^'^'-l
w^
x>0, the
sum ofn +
arg^
u?^du']
r^
Jo
positive
is
In
as Stirling s series.
lished over the extended range
253
"*"
V^
In conjunction
an asymptotic formula for the Gamma-function.

= xr{x), it is very useful for the purpose of com-
is
with the formula T {x + l)
puting the numerical value of the function
for real values of x.
(.^), correct to 12 decimal places, for values of .v between

were constructed in this way by Legendre, and published in his Exercices de
Calcul Integral, ii. p. 85, in 1817, and his Traite des fonctions elliptiques (1826), p. 489.
Tables of the function logioT
and
It
2,
observed that V (x) has one
may be
minimum
for
positive
values of
.>;,
when
.r= 1-4616321..., the value of log,or(.r) then being 1-9472391....
R {z) > 0,
=
\og,T {z) {z-l)\og, z- z+^\og,{27r) + J (z),
Obtain the expansion, convergent when
Example.
where
-^
in
W- 5 |, +
2 (-^+1)
+ 2) ^ 3 (.-+1) + 2) {z + S)^(.-
(.^
which
and generally
c=
r (x+l) {x + 2)
...
{x
+ n-\)
{2x -\) xdx.
(Binet.)
12 "4.
The Eulenan Integral of the First Kind.
The name Eulerian Integral of
the First
Kind was given by Legendre
to
the integral
B (p,
q)
xP-' (1
- x)'i-' dec,
In this integral, the real

first studied by Euler and Legendre*.
xp~^,
to
positive;
and
and
are
supposed
be
parts of p
(1 x)^~^ are to be
q
e('v-i)'e(i-=)
which correspond
understood to mean those values of e(^~^)^* and
which was
to the real determinations of the logarithms.

* Euler,
Nov. Comvi. Petrop.
xvi. (1772);
Legendre, Exercices,
i.
p. 221.
254
With
these stipulations,
( 4*5
improper) integral
We
easily seen that
it is
example
*)
have, on writing (1
B (p, q)
[CHAP. XII
exists, as a (possibly
2).
for w,
B{p,q) = B{q,p).
Also, integrating
by
parts,
Jo
B{p,q +
so that
Example
1.
Shew
PJo
Jo
i^
= ^B{p+l,q).
l)
that
B{p,q)=îp + l,q) + B{p,q + l).

Example
2.
Deduce from example

B{P,<l
Example
3.
Prove that
if
that
+ ^)=^^B{p,q).
a positive integei*,
is
Example
4.
Prove that
Example
5.
Prove that
12'41.
the
,,
B{p,n + \) =
(2)
...
p{p + l)...{p + n)_
lim n^
n).
{z,
Expression of the Eulerian Integral of the First Kind in terms of
Gammafunction.
We
shall
now
establish the important
^x_ r(m)r(n)
R/.,,
First let the real parts of
theorem that
and n exceed
r (m) r (n) =
e-* '"-i
câ;
writing
a;'-
for
.t,
and
V (m) r (n) =
for
g-?' if'-^
dy.
Jo
rR
fR
the values of
then
y^ for y, this gives
4 lim
=4
Now,
Jo
On
lim
e"*' x'-""-'
dx x
g-?/' ^/-"-i
dy
e-(^'+2'')^2m-iy2n-i(^^(^^_
and n under consideration, the integrand
continuous over the range of integration, and so the integral

sidered as a double integral taken over a square Sji.
may be
is
con-
Calling the integrand
THE GAMMA FUNCTION
12-41]
f{x,
we
255
and calling Qg the quadrant with centre at the origin and radius R,
y),
have,
be the part of S^, outside Qb,
if Tji
f(x, y) dxdy
f{oc,y)
dxdy
JJqr
JJsr
f(a:,y)dxdy
^JJ Tr
-^0
since
|/(;,
JJ Sr
y)
JJ Sir
R^^ X
as
dxdy converges
\f{^,y)dxdy\
\f{^>y)ida^dy-\\
JJSr
x""-'
Jo'
e-^'-
namely
to a limit,
idxx2\
y-''-'
e'^
dy.
Jo
Therefore
Imi
Changing
to polar* coordinates {x
f{x, y) dxdy
J J
f(x,y)dxdy=
(
|
Or
Oh
lim
f{x,y)dxdy.
//
= rcos 0,
r sin
6),
we have
e-"" (r cos 0)^'"-' (r sin 0)-''-'
rdrd0.
Hence
r (w) r
()
=4
e-r'^r-('"+")-i
dr
cos="-i
sin^'^-^
^c?^
Jo
Jo
= 2r (i + ?z)
cos^'"-'
sin^''-^ <9c?^.
Jo
Writing
cos-
u we at once get
r (m) r (?i) = r (?u + )
z^
(w,
?i).
This has only been proved when the real parts of

it
can obviously be deduced when these are
less
ni
and n exceed | but

124 example 2.
;
than | by
This result, discovered by Euler, connects the Eulerian Integral of the

First
Kind with the Gamma-function.
Example
Shew
1.
that
['
} -I
* It is easily proved
(l-f-A-)'^-i(l-.r)-i(^^=2P + 9-i^^Mli2).
^
'
îp+q)
'
by the methods of
411
that the areas A^^
angles provided only that their greatest diameters can be
number
by
of areas
radii vectores
sufficiently large
and
circular arcs.
so the areas
may
made
/x
of 4-3 need not be rect-
arbitrarily small
by taking the
be taken to be the regions bounded
256
Example
2.
Shew
j\isj
[CHAP. XII
that, if
.v
x+ 1
2!
.r
+2
3!
+d
.r
then
f{x,y)=f{y + \,x-\\
where ^ and y have such vakies that the series are convergent.
Example
3.
(Jesus, 1901.)
Prove that
j'J'j{xy){l-xT-^y>î\-yr-^dxdy = ^-^>^^^'j
(Math. Trip. 1894.)
Evaluation of trigonometrical integrals in terms of the
1242.
Gamma-
function.
We
now
can
evaluate the integral
cos"^~^
x sm'^~^ xdx, where
and n
Jo
are not restricted to be integers, but have their real parts positive.
For, writing cosô;

I'*"
t,
we
have, as in 12'41,
.
cos'-^ cc sin" -1 xdx
The well-known elementary formulae
IF am) ran)
^^
'
^^f.
for
m and
the cases in which
n are
integers can be at once derived from this result.

Prove that, when \k\<.\,
Example.
_ T (lm+^)T {^/i+^)
~ Tlh-a + hi+l)^7r
fh^ CDS'" e sin'êdd
\l^ksm^6)i
cos'" -'"Odd
fh^
(l
-k
sin2 ^)i+^
(Trinity, 1898.)
Pochhammers*
12'43.
extension of the Ealerian Integral of the First
Kind.
We
have seen in
integral for
F {z) by
12'22 that
it is
possible to replace the second Eulerian
a contour integral which converges for
similar process has been carried out by
Pochhammer
for
all
values of
z.
Eulerian integrals
of the first kind.
Let
be any point on the real axis between
and
consider the
integral
r(i+,o+, 1-, 0-)
t^-^ (1
e-'"'<'^+^)
- 0^-' dt = e {a,
/3).
The notation employed is that introduced at the end of 12"22 and

means that the path of integration starts from P, encircles the point 1 in the
positive (counter-clockwise) direction and returns to P, then encircles the
origin in the positive direction and returns to P, and so on.
,
Math. Ann. xxxv. (1890),
p. 495.
THE GAMMA FUNCTION
12-42, 12-43]
At the
starting-point the arguments of
and
the circuit (1 +) they are
27r; after the circuit (1
l-ir
they are
\t
and
257
are both zero; after
+) they are
after the circuit (0
lir
and
and
after the circuit (0
are both zero, so that the final vahie of the integrand
is
Itt
and
they
the same as the
initial value.
It
is
easily seen that, since the
any way so long as
it
may be taken
integrand, the path
path of integration may be deformed in
does not pass over the branch points
shewn
to be that
0,
in the figure,
of the
wherein
the four parallel lines are supposed to coincide with the real axis.
>^
//
the real parts
of
and
a.
are positive the integrals round the circles
y9
tend to zero as the radii of the circles
paths marked
a, h,
d are
c,
â-Ig-'W(a-l) (1
_ ^)3-lg2.r<(8-l)^
arguments of
respectively, the
tend to zero*; the integrands on the
Hence we may write
and
â-l g2,r.(a-])
(^
_ ^)fl-i
nuiu being zero in each case.
^) as the sum of four (possibly improper)
e (a,
integrals, thus
e (a, /3)
= e '(''+^)
t^-' (1
0^"'^^^
0^~'e-'' ^dt
-' (1
- tf-'e-^^'
t"-' (1
I
/""
t"-' (1
tf-'
e^-'t"*^)
dt
+
.'
dt
Hence
= e '('^-^)
(a, /3)
(1
- e^')
e---) (1
^-' (1
tf-' dt
JO
l^(a)r(/9)
^,
-isin (a7r)sin (ott)
^~
47r*
"r(l-a)r(l-/:^)r(a +
Now
and
e (a, /S)
for all values of a
values of a and
l3.
and of /3
by the theory of analytic continuation, this
parts of a and ^ are positive, holds for all
this last expression are analytic functions of a
and
/S.
when the
equality, proved
^)-
So,
real
Hence for
/
all values of a
and
/3 lue
have proved that
-47r-
J^
r{i-cc)l\l-id)l\a+id)'
AV.
M. A.
The reader ought
to
have uo
difficulty in
proving this.
17
258
Dirichlet's integral*.
12"5.
We
now shew how the repeated
shall
ff
f{t,
t,+ ...+
integral
tn) î"'-^ô"^-^
may be reduced to a simple integral, where/is

and the integration
that
+ 4+
^1
To
...
extended over
is
p'M^
simplify
= T {\
v)/v
r~U'
Jo
f\t
t,
T, a, j3 for
(?*
1, 2, ...
n)
for
^3
+ ^4 +
...
+tn),
0)
V-''-'
r-+3-i dvdT.
the integral becomes
( 4*51),
V(^ + T/v){l- vy-'v''-' T-^^-'dTdv.
T = VTo,
/(^ + "^2)
(if X.
and X
Og
t^, t^, cui,
/(x + Tlv){l - vy-'
dt
J T/(l-\)
>
+ T+\)t'^-'T^-'dtdT
Changing the order of integration
Putting
positive values of the variables such
continuous, a^
Jo
(where we have written
all
tn'^n-idt^dt,
. . .
$1.
tn
Jo
put
[CHAP. XII
(1
- vy-' v^-'
r.^+^-' dr., dv
r (a)r(8)
r( + /3)
r^-^
Jo
Hence
r(,)r(,)
r(e
^^^^J|...
j/(T.
+ ^3+
...
the integration being extended over

that
T2
ts
...
r((x,
is
all
...
tn^'^-'dr.dt,
...
dtn,
positive values of the variables such
+tn-^^.
Continually reducing in this
which
+Qt/.+^-i^3"^-i
way we get
+ a,+
...+an) Jo-^^
'
Dirichlet's result.
Example
1.
Reduce
to a simple integral
the range of integration being extended over
all
positive values
of the variables such that
it
being assumed that
a, 6,
c.
a, ^, y,
p, q, r are positive.
Werkt,
I.
pp. 375, 391.
(Dirichlet.)
.
'
THE GAMMA FUNCTION
12o]
Example
2.
Evaluate
259
x^yi dxdy,
and n being positive and

'
Example
3.
Shew
a, 6, c are
Example
4.
2/^0,
moment
that the
density, taken about the axis of
where
>0,
x>"
+ y" ^
Pembroke,
homogeneous
of inertia of a
1 907
ellipsoid of unit
z, is
the semi-axes.
Shew
that the area of the epicycloid
x-'
=P
+y^
is fn-r-.
REFERENCES.
N. Nielsen, Handhnch der Theorie aer (Jamma-funktion*.
0.
ScHLOMiLCH, Compendium der ho heren Analysis,
E. L. LiNDELOF, Le Calcul des Residus, Ch. iv.
(Leipzig, 1906.)
(Brun.swick, 1874,)
ll.
(Paris, 1905.)
A. Pringsheim, Math. Ann. xxxi. (1888), pp. 455-481.
1.
Shew
that
2.
Shew
that
(Trinity, 1897.)
.,'i"i
3.
vr.
Prove that
r'(i)
f(i)
4.
Shew
Shew
T'
(is)
r(|)
=^
('^^"^' ^^^2-)
^^S ^-
that
{r(i)!^
"16;;^"
5.
''"'^' ''>
rrp rTr.-i7i;""='"<"+"-
52-1
32
= 3^^
92-1
72
5^ r^T
92
IP
^^
(Trmity, 1891.)
.
i^-ZTi
that
f
\
I
(-a)(M+^
+y
n
=o
a \\
) /,
-7-7
r^
H1+
^
l/j
(+/:<) (+y)
\
;
+
,
1
= - - sm (an),.,_,
B (/3,
^"
.
TT
'
y).
'^^
(Trinity, 1905.)
6.
Shew
that
7.
Shew
that, if
^
z
= iC
where ^
^ (^)
is real,
^^ ('Isf
(Peterhouse, 1906.)
then
(Trinity, 1904.)
l^^^^l=\/(fSVc)8.
When x
is
positive,
shew thatt
r (x) r
.
.
(h)
Tx
2n
-^^
^^
(Math. Trip. 1897.)
This work contains a complete bibliography.
t This
and some other examples are most easily proved by the result
of 14-11.
172
260
If
9.
is
positive,
shew that
-)"(- 1) ((f r(^)r(+i)

^ i (
!
r(2+)
=o
10.
If
.V
>
[cHAP. XII
2)
...(g-H)
s+'
and
JO
shew that
and
11.
12.
Shew that
X>
if
Prove that,
,c> 0, - ^tt
0,
when
^-le-^^cosacos(X^;siua)c?^ = X-^^(.^')
V^~^ e-'î'fosasin
if 6
> 0,
cosfto;
^^^
< < 1,
cosa.^,
sin
(Euler.)
a.^".
< s < 2,
(s),
-^
If
(.2;)
-^7" dx = \u\f-^ cosec (J7rs)/r
< s < 1,
(Xisin a) dt==\~^ r
when
then,
^ ^^^^
^^^ (a.^.)/p
(Euler.)
(5),
.^
_/
13.
<\tt, then
and,
<a
prove that
/i
jj(l+.t'-cos^-rf.r=r(,0 {cos
(f -1)
-^^^+^,^
(Peterhouse, 1895.)
14.
By taking as contour
of integration a parabola with its vertex at the origin, derive
from the formula

1
r**^"")
2^ sin
an
the result
f^
r(a)=
e-^''x"--'^{l+:c''-)^^[Ziim{x
-.
2 sui utt J
+ a?(.YCcot(-Jc)}
{)
+ sin {x + {a - 2) arc cot - x)}] dx,

(
the arc cot denoting an obtuse angle.

(Bourguet, Acta Math.
15.
Shew
that,
if
the real part of
a-n is
positive
and 2
n
is
convergent when
16.
m > 2,
where
i.
p. 367.)
l/a^ is convergent, then
=l
d'
\//-()
(2)
= -r^^ log r (z).
(Math. Trip. 1907.)
Prove that
d\ogr{z)_ /"e- g-e-g'^

<^2
"jo
1'
= r{(l+a)-i-(l+a)-}---),
a
Jo
/1
Jo
^.2
_
^
x-l
dx-y.
(Legendre.)
THE GAMMA FUNCTION
261
Prove that, when R{z)>0,
17.
\ogr{z)=r i-^^-xiz-l)]
Prove that,
18.
(Bmet.)
for all values of z except negative real values,
log r (2) = (3 - 1) log 2 - 2 + Uog (27r)
Prove that, when
19.
Prove that, when
20.
=^
{l-a.-+log.r}.
r^j
Jo (l-.r)log.r^
R (2) > 0,
c/2'^
/2+1
If
21.
_3_
J;
^ (2) > 0,
-rlogr(2)^ = log2dz
logr(Oo?^ = ?',
shew that
and deduce from
12-33 that, for all values of 2
except negative real
value.s,
u = z log z z + i log (2n-).
(Raabe, Jovnial fur Math, xxv.)
Prove that, for
22.
all
wilues of
except negative real values,

00
f^
logr(2) = (2-i)log2-2 + Uog(27r)+ 2 /

n=lj
dx'
-
sin 2nTrx
*+'
(Bourguet*.
Prove that
23.
Bip,p)B{p+h,p+h)=^^^y
Prove
24.
that,
when -(</<(,
,1
r,
Prove that, when
25.
B{p,
q)
Prove that, when
26.
+ Bip + \,
28.
Shew
q)
+ B{p + 2,
q-\).
'ît;4-?'*' 1.
2.
r)
(/>
(/)
(io
+ ? + l)
= B{q, r)B{q+r,
,,,
"
.,
''
d.r
(.r+?>) +
''
1
r(a)r(b)
r(a + 6) (l+jo)jo"'
> 0, 6 > 0, p > 0.

*
(Euler.)
p).
that
Jo
a
+ ...=B{p,
a(a + l) g(g + l)
aq
B{p, q)B{p^q,
if
q)
pa>0,
^(i^, ?)
Prove that
cosh (2n*) c?M
q>\,
B{p-ci,q)
27.
/"""
(Binet.)
This result
is
(Trinity,. 1908.)
attributed to Bourguet by Stieltjes, Journal de Math,
(i), v. p.
432.
'
262
29.
Shew
that, if
m > 0, n>0,
n
and deduce
that,
then
i+.^)2>-i (1 -^-y^n-i
(
when
is real
/'_i,r
r(m)r{n)
,^^_^
and not an integer multiple of
/cos 6 + sin ^\ cos 2a
'i^
veos
C^
[chap. XII
ijr,
d6=.
SUl
2 sin
(tt
cos^ a)
(St John's, 1904.)

30.
Shew
> 0, ^ > 0,
that, if a
/:
and
(Kummer.)
31
Shew
that, if a
> 0, + 6 > 0,
fr(a)r(S)
r..r--(i-^)^^_,,^lrw
Deduce
50
I-''*''
Jo
that, if in addition
r(a + b)r{8)
l=^/^(a+6)-^/'(a).
T{a + b + 8)
^(a+ 8)
'
a + c>0,
a-\-b
+ c>0,
/lâ-l (l_^b)(l_.^.c)
32.
Shew
that, if a,
b, c
fUl-x")(l-afi)(l-x')
(i-A-)(-log.r)
JO
By
(^_+_'^)
+ 6)r(a+c)"
be such that the integral converges,
'
33.
(t^).!
*^~ ^*r(a
jo "(i-.r)(-log.r)
T{b + c+l)T{c + a + l)T{a+b + l)

^^r(a + '\)r{b + l)r{c + l)r{a + h + c+l)'
a.v=los
'"''
the substitution cos ^ =
1-2 tan
shew that
^<^,
(St John's, 1896.)

4v/7r
(3-cos^)5
34.
f^
Evaluate in terms of Gamma-functions the integral
sinP.v
fraction greater than unity
[Shew that the integral

'-
'
sin''
x\~+
n=i
Jo
- )"
\x+mr
/) is
'*-"
whose numerator and denominator are both odd

is
when
dx,
integers.
[ d.v.]
x-mrj)
(Clare, 1898.)
35.
Shew
that
93?-
/:
36.
2" +
-a-mi-
Prove that
log
s
37.
2^^r=o2/'!(n-r)
Prove that,
if
log (^+i)
Bip,g)^
v/', -/;
\ pq
p>0, p + s>0,
4-
'
j(,
^\: '"\^l^'"^
dv.
then
s{s~l)
B(p,p)
s(s-l)(s-2)(s-3)
+ 2.4.(2^ + fdî,--}^(P>P + ^)=-^f7^'^l+^(^^)

{'
38.
The curve r"*=2'"~i a'"cos
is
??i^
composed of
the length of the arc of half of one of the loops

i~^
(i cos x)
total perimeter of the curve is
a iV
lm)\
?n
is
Jo
and hence that the
(Euler.)
(1-V)l0g^7
'
dx,
equal closed loops.
<-Shew that
THE GAMMA FUNCTION

Draw
39.
the straight line joiniug the points i, and the semicircle of
on the right of this
lies
0,
?',
i.
By
263
considering
Let
line.
2P--i {z + z-^)p
" cosP*i--6
cos
"-- dz,
6 d6 =
ip-q)
^
^
Jo
Prove that the result
is
true for
\z\
=\
which
be the contour formed by indenting this figure at
shew
p + q>l, q <l,
that, if
"!
r.
,
+ q-l)2Pî-^B(p,q)
,
ip
values of p and q such that
all
p + q>'l.
(Cauchy.)
40.
Mild
s is
cos
series
and of the function
cos*.i'.
.f-2^, r
(*+i)[r(|+ia+i)r(i5-â+i)
find the values of
42.
and - in ^.r ^hn, shew that
positive (not necessarily integral),
draw graphs of the

41.
and
If
Prove that,
for
if />
which
r(h+?,a+i)raj-â+iy-]'
(Cauchy.)
applicable.
it is
> A,
r
22p-i
2/>"-
""Wi'U +
'"'''"=-A
"^
12.32
12
"|2
i'^2(2yT3-) + 2.4.(2p+3)(2f+5)+ }.
'
(Binet.)
43.
Shew
that,
if .c
< 0, + > 0,
r(-.f) [-.f
r(5)
and deduce
that,
./
then
--
,(-..-)(l-.r)
^-
"^^
s(l+2)
when x + r>
44.
^j
1
0,
m^ +*) _ f _ 4
^ Ina
^
r>)
f/-'
(-.r)(l-^)(2-f )
z{l+z){2 + z)
^^^-1)
,
'(^-l)(-^-2)
^ z(z + l)^-^ z{z + l){z + 2)
Using the result of example
logr(2 + a) = logr(0 + rtlogs
43, prove that
2z
fa
a [\{\-t)(2-t) ...(n-t)dt- f" t{l-t){2-t) ...(n-t)'dt
_ 5
.'
.'o
{n
nti
+ l)z{z+l){z + 2)
investigating the region of convergence of the
...{z
+ n)
.series.
(Binet, Journal de V Ecole polytechnique, xvi. (1839), p. 256.)
45.
Prove that,
if />
> 0, ^ > 0,
B ip,
then
=
-.
q)
P^
i (2^)* ^
"' '
264
[CHAP, XII
where
II (p, g)
= 2p
and
p2
and
the function
if
shew
(,r)
^-^
-.
^-^,}-,
=^2 + q^ +pq-
6^=2*-'7r(l-i'r),
If
46.
arc tan
F=2^-^7r(i-iA'),
be defined by the equation
that F{o:) satisfies the equation
(1)
F{x+\)=xF{x) +
(2) that, for all positive integral
^
^^
r(l-A-)'
values of x,
F{x) = rix\
(3)
that F(x)
(4)
that
is
analytic for all finite values of x,
1
"F(x)= ~- ^
(7
"*
-r-
'
log^
2
^
Expand
47.
as a series of ascending powei's of a.
(Various evaluations of the coefficients in this expansion have been given by Bourguet,
Schlomilch,'
Set. Math. v. (1881), p. 43; Bourguet, Acta Math. ll. (1883), p. 261
Bull des
Zeitschrift fiir
48.
Math, und Phys. xxv. (1880), pp. 35, 351.)
Prove that the G^-function, defined by the equation
G(z + \) = (2nf'e-^'^^+^-^^
is
an integral function which
|(i+i)%-'+-'-^/(2")|,
satisfies the relations
0{z + \) = V{z)G{z),
(n
!)'V6-'
(n
+ 1) = 11
22
(?(1)
.
33
...
= 1,
(Alexeiewsky.)
w".
(The most important properties of the G^-function are discussed in Barnes' memoir,
Quarterly Journal, xxxi.)
49.
Shew
that
and deduce that

log
50.
Shew
^Q ^
^s cot Tvzdz-z log
(27r).
that
logrri + l)c^<=i3log(2rr)-|3(2+i;+3logr(s + l)-logG'(s + l).
/:
CHAPTER
XIII
THE ZETA FCNCTIOX OF RIEMANX

13"1.
Let
Definition of the Zeta-function.
= a + it where a and
are real*; then,

=^
if S
>
the series
0,
n=i n^
is
a uniformly convergent series of analytic functions
domain
in
which
was known
<r
+ 8;
domam.
of s in such a
and consequently the
The function
although
series is
called the Zeta-function
is
any
( 2"33, 3'34) in
most remarkable properties were not discovered

before RiemannJ who discussed it in his memoir on prime numbers it has
since proved to be of fundamental importance, not onl}- in the Theory of
Prime Numbers, but also in the higher theory of the Gamma-function and
it
to
Eulerf
its
allied functions.
1311.
Many
The generalised Zeta-function^.

of the properties possessed by the Zeta-function are particular cases
more general function
of properties possessed by a
when
defined,
cr'^1
+ 8,
by the equation
where a is a constant. For simplicity, we shall suppose that

then we take arg(a + n) = 0. It is evident that ^(s, 1) = ^(s).
||
1312.
The expression of
Since (a-F
fortiori
??)-'
when
T (s) =
a^\
-\-
S),
V{s)^{s,a)= \\m
^{s, a)
.v^xjio
*
The
letters
o",
will be
t Commentatioiies Acad.
.-r*-'
The
and
1,
o-
e" <+'*
arg.r
and
o-
>
(and
8,
câ;
1-6-^
.'o
l-e-^"
used in this sense throughout the chapter.

ix. (1737), pp. 160-188.
ScL Imp. Petropolitunae,
+ Berliner MonaUherichte, 1659, pp. 671-680.
when
do-.,
we have, when
:L
as an infinite integral.
af-ê'^''^'''''
<
Ges.
Werke
(1876), pp. 136-141.
definition of this function appears to be due to Hurwitz, Zeitschrift fiir
Math,
Phys. xxvii. (1882), pp. 86-101.

When a has this range of values, the properties of the function are, in general,
ttnd
much
II
simpler than the corresponding properties for other values of

for all values of a (negative integer values excepted);
true
when
R (a) > 0.
and the
a.
The
results of 13*14 are true
results of 13-12, 13-13, 13-2 are
266
Now, when
ic'î),
e^^l +
x,
and
[CHAP. XIII
modulus of the second of these
so the
integrals does not exceed

""
A-'-2e-(-^^+)^rf^
= (N
aV-^r (o- -
1),
-'0
which (when
o-
Hence, when
+ 8)
tends to
a^l +
as JV
and arg x
this formula corresponds in
^x
0,
some respects
to Euler's integral for the
Gamma-
function.
The expression* of^(s, a) as a contour
13"13.
When a ^1 +
8,
integral.
consider
(0+)
^y-i g-a2
'
( 12*22) and not containing

which are poles of the integrand; it is
the contour of integration being of Hankel's t3rpe
the points + 2n7ri(n
supposed (as in
l, 2, 3, ,..)
12"22) that
arg(
z)\ îr.
It is legitimate to modify the contour, precisely as in 12*22,

o-
B;
whenf
and we get
'(0+)
{ A^-ip-az
r^
^8-ip-az
Therefore
27n
Now
of
of
(1
this last integral is a one-valued analytic function of s for all values
Hence the only
s.
.'
s),
i.e.
possible singularities of ^{s, a) are at the singularities
at the points
1, 2, 3, ...,
and, with the exception of these
points, the integral affords a representation of ^{s, a) valid over the whole
The
plane.
function.
result obtained corresponds to Hankel's integral for the
Also,
we have seen that
so the only singularity of ^{s, a)

integral,
^{s, a) is analytic
at the point
is
is
the residue at
Hence
r(o+)
e-^
1
lim
Writing
6-
Gamma-
1 -H 8,
1 in
and
this residue is 1.
(1
Given by Riemann for the ordinary Zeta-function.
If
1,
^^f
\ = -l.
-s)
1.
'
of the integrand,
*îl
(7
o-
we get
ZTTl J ^
which
.s
when
the integral taken along any straight line up to the origin does not converge.
and
the
THE ZETA FUNXTION OF RIEMAXN
13'13, 18-14]
Since
(1
s)
has a single pole at s
the only singularity of ^{s, a)
Example
Shew
1.
when
that,
(1
-Shew that,
Example
3.
Shew that
21 -*) /-(s)
to say,
1.
when R{s)>
when
a one- valued function of
so that ^
is
follows that
1 at 5
2*
3'
/* x^~^
1,
+rr/, 877/, 57ri',
....
Values of ^{s, a) for special values of s.
In the special case

is
1, it
= i: _ 1 + 1 _ i ~
4s
where the contour does not include any of the points
1314.
R (s) > 0,
2.
with residue
a simple pole with residue
is
Exuraple
267
it is
dz
is
the residue of the intesfrand at ^
the coefficient of z'" in
V^
we
this coefficient
.
e-"^
that
(-)n<^(a)2
__
a,
= 0,
differentiate the expansion ( 7"2)
e-^-l
term-by-term with regard to
theorem,
e~^
To obtain
^^
We may consequently apply Cauchy's
z.
^-r:
an integer (positive or negative),
s is
n'.
n=\
where
denotes the Bernoullian poly-
(f)n{iO
nomial.
(This
into a
is
power
obviously legitimate, by
series in
when
<
^tt,
since
imiformly convergent with respect to
-^_, can be expanded
a.)
i^^ll-r^^Ml".
Then
Therefore
4-7,
fs
is
zero or a negative integer (=
^{- m,
a)
=-
In the special case when a

/
of z-~^ in the expansion of
-^
,
'
</)',+o(a)/[(m
=
yji
if s
1,
I
'
Z
.
-f-
= m,
m),
1) (i
we have
-f-
2)|.
then ^(5)
is
the coefficient
268
Hence, by
[CHAP. XIII
7*2,
^(-2m) =
^{l-2m) = (-y-BJ{2m)
0,
(m
1, 2, S, ...),
r(0)=-|.
These equations give the value of ^
of Hurwitz for ^{s, a) when cr<
TJie forniula*
13"15.
Consider
e~^ dz
-^r-^
27n J c
ivhen s is a negative integer or zero.
(s)
0.
taken round a contour
consisting
of
a (large) circle of radius (2iV+l)7r, (iV an integer), starting at the point

(2iV"+ l)7r and encircling the origin in the positive direction, arg ( z) being
zero at z
= -(2N+l)
ir.
C and
In the region between

contour of
13"lo
the contour {2NTr +7r;
the limiting form, (
is
one-valued except at the simple poles +
zy~'ê~'^^ (1
2Tri,
47^^, ...,
+), of
which the
e~^)~^ is analytic
and
2N7ri.
Hence
l-e~'
27riJc
where
At the
i2,
Rn
27ri
l-e'-
^^n+i) n
n=i
are the residues of the integrand at 2n7ri,
point at which
z=
2mre~^'^^, the residue
27^^ respectively.
is
(2n7r)'-le-^'''('-l)e-2^"''',
Rn + Rn =
and hence
(27?7r)^~^ 2 sin
i^^'^
"^
2'n-an j
Hence
1
(_^)s-ig-a^
/(0+)
l-e-^
27riJ(2iv'+i)^
_
"
Now,
< a^
since
2 sin ^STT
"(27rr^,=i
1,
2 cos ^sir
cos {2'rran)
sin (27ra?)
'^
"
/?^-^
n'-'
\2^)^=^n':i
+ H
2in
it
easy to see that we can find a
independent of N' such that
is
|
e~"^^
(1
dz.
-,^
e~^)~^ <
\
K when z
is
on
C.
Hence
( z\^~'^
1
--
e~"'^
.
1
dz\<^K\
f""
{(2.Y
,
27r{
1) -rrYe'"' rf^
I
<ir{(2iV-|-l)7r|'ê-l*l
^^
* Zeitschrift fill-
as
i\"
^X
if
cr
<
Math, und Phys. xsvii. (1882),
0.
p. 95,
number
THE ZETA FUNCTION OF RIEMANN
13- 15-13-2]
Making
N ^ y:
û. ^
we obtain the
^r(l-,v)
/I
result of
Hurwitz
that, if
o-
/i
^ cos(27ra)
^^
269
<
0,
sin(27ra/i))
each of these series being convergent.
13151.
If
Riemairii's relation between ^(s)
and ^(l
s).
we write a = 1 in the formula of Hurwitz given in 13'15, and employ

we get the remarkable result, due bo Riemann, that
12-14,
2^-*
r (S)^(S)
cos (I STTJ
^^^(1
_ s).
Since both sides of this equation are analytic functions of
save for isolated
s,
values of s at which they have poles, this equation, proved
when a <
0,
persists (by 5"5) for all values of s save those isolated values.
Example
If
1.
be a positive integer, shew that

C (2>n) = S-^"*- 1
Example
Shew
2.
that r {is)n~^^ C
7r2"'
BJ{2m)
unaltered by replacing
(s) is
by
- s.
(Riemann.)
Example
Deduce from Riemann's
3.
are zeros of the
13"2.
(p
relation that the zeros of f
at
(5)
2,
4,
6,
...
tirst order.
Hennites* fornuda for
^(s, a).
Let us apply Plana's theorem (example 7, p. 145)

=(a + z)~\ where arg {ii + z) has its principal value.
the
to
function
(z)
Define the function q

<^''
(/
y)
9-^-
{x,
y)
by the equation
K + ^ + wT' -
= Ha + A')- +
if]
A'
iy)~']
~ ^* sin \s arc tan

{
Since +
arc tan
^
X
-^
X
a)
does not exceed the smaller of ^tt

and
'
+a
'^
x+
we
have
+ x)'+f-]^
\q{x,ij)\^ {(a
'
Using the
*
q (X, y)
first
[{a
result
when
^=
jsinh
|^^|
.
|
y > a and the second when

(3),fv. (1901), pp.
:,<.
sinh JItt 5
j
j
+ xf + f]-^-''\
Annali di Matemalica,
t If t>0, arc tau
y-'
*<^
1-rt-
5;
1-i-t-
57-72.
and arc tan ^<
Jo
dt.
y < a
it
is
270
evident that,
if
o-
>
0,
y) (e-"^
(.v,
Jo
tends to
as
ic
-^
(a
also
1)
+ x)~^dx
dy
[CHAP. XIII
convergent when x ^
is
converges
if
a >
and
1.
.0
Hence,
if o-
>
^(s,a)
make
legitimate to
1, it is
the example cited
^'o
^-
oo in
the result contained in
and we have
= la-^+\^ {a+ocy^dx +
'lj^
(a2
+ ^2)-i
jsin fs arc tan
0j-
^^J^
So
^{s, a)
This
la-s
+ f^^+2j\a^ + f)-y
^)|
Hermite's formula*; using the results that,
is
arc tan y/a
we
jsin (. arc tan
^ y/a
y <
^ dT^j
arc tan y/a
<
if
^-^.
y^
0,
(y>:â7r],
2 tt
see that the integral involved in the formula converges for all values of
Further, the integral defines an analytic function of
values of
s for all
s.
s.
To prove this, it is sufficient ( 5*31) to shew that the integral obtained by differentiating
under the sign of integration converges uniformly that is to say we have to prove that
;
- 1 log
{a-
dij
(a'^+y-) - 5* sin i s arc tan
+ _y2)
o^'^y
^*
{((^+y^)
arc tan
cos
arc tan
di/
converges uniformly with respect to
where A
is
any
positive
in
any domain of values of
(a2+y-) ~^* arc tan
'-
^j
,2îA
{a-+f)^
the path of integration of the

and using the results
sin'lsarctau
it
V
in the respective parts,
^ A,
< {a- +
i/'^')i^
^ cosh
(|n-A)
yy_
dividing
)
converges, the second integral converges uniformly
Now when
s.
cos (s arc tan
By
-I
number, we have
since
Una,
~'^y
<sinh
by
first
4"431
(I).
integral into
two parts
(0,
^rra),
sin (s arc tan -) l<sinhJr7rA
we can simihxrly shew that the
first
integral converges uniformly.
Consequently Hermite's formula is valid ( 5"5) for all values of s, and

legitimate to differentiate under the sign of integration, and the
is
differentiated integral
*
is
The corresponding formula when
= 1 had
s.
been previously giveu by Jensen.
13 "21, 13 '3]
271
Deductions from Hermites formula.
13'21.
Writing
Making
-^
1^(,,
î(^^'
we
in Hermite's formula,
see that
from the uniformity of convergence of the integral involved
1,
in Hermite's formula
li,,
we
see that
= lim
+1+2 r
a)--^\
2a
*-lj
s-\
,-*!
Jo
^
Hence, by the example of
,^-^,
^,
+ y-'){e^^y-l)
{a-
we have
12'32,
hm|ri.,)-^-^^| =
-j^.
Further, differentiating* the formula for
l^{s,
a) and then
making
-^
0,
we get
A
\d ^,
a^-*'
log a
o log (a- + 2/-)
2
Jo
a'-'
(a-
+ y-) ~
^*'
sin
s arc
tan -)
ci/
+ ^V-)^ ~ ^^ arc
+ (a^
I
i\
= (â-,jlog-
+
,
tan - cos
a
^
{'^
s arc
tan
a)] e'^y
^
- 1
(Wa)
___A^rfy.
arc tan
2J^
Hence, by
1232,
These results had previously been obtained

Lerch
Corollary.
13"3.
Let
lim
^1 = 7> T (0) = -5 log(27r).
k(s)
Euler's product for ^(s).
(T^l+B; and
let
2, 3, 5,
Then, subtracting the series
This was justified in
.../>,...
for 2~* ^ (s)
^(.).(l-2-) =
be the prime numbers in order.
from the series
~ + ^,+~ +
^,
for ^(s),
we get
...,
13 -2.
The formula for f (s,

published by the Academy
t
manner by
in a different
-j*.
a)
from which Lerch derived these
of Sciences of Prague.
A summary
the Jahrbuch iiber die Fortschritte der Math. 1893-1894, p. 484.
results
of his
is
given in a
memoir
is
memoir
contained in
;.
272
the terms of Sn~^ for which n
all
a multiple of 2 being omitted
then in
manner
like
the terms for which n
all
is
[CHAP. XIII
is
a miiltiple of 2 or 3 being omitted
and
so on
so that
^(s)
the
'
to 2,
- 2-0(1 - 3-0
(1
t'n-' $ I'n-'-^ $
2'-^
n'^-^
as
ja
^ co
Therefore if a
number p assumes
But the product
the product ^(s) Yl (1

p
8,
prime values
the
11 (1
p
2, 3, 5, ...
p~^) converges
to
1,
tuhere
only.
p~^) converges when a ^
for it consists of
8,
factors of the absolutely convergent product 11 (1
some of the
it
-p-') =
(1
denoting that only those values of n (greater than p) which are prime
3, ... j:? occur in the summation.
Now*
the
n~^).
Consequently we infer that ^(s) has no zeros at which a ^1 + 8;

had any such zeros, IT (1 p~^) would not converge at them.
for if
Therefore,
This
is
ii
a^
Euler's result.
13"31.
Riemanns
It has just
From
hypothesis concerning the zeros of ^(s).
been proved that ^(s) has no zeros at which a >1.
the formula
(|
y{s)
is
it
now apparent
of {r(s)}-^ sec (2
the
8,
13*1 51)
2-i {r(.9)|-isec
STT^
Ul-s)
^'''')
the points
i-e-
= 2, -
was conjectured by Riemann, but
the zeros of
^(.s)
<
that the only zeros of ^ (.5) for which a

-4,
in this strip lie
been proved by Hardy
-f-
on the
it
line
are the zeros
....
Hence all the zeros of f (s) except those at 2, ]>,

domain of the complex variable s ivhich is defined by
It
...
lie
in that strip
-â ^
has not yet been proved, that
=2
o"
'
"^'île it
Riemann's conjecture
is
The
first
term
of S' starts with the
correct,
lie
on
cr
see p. 280.
:^
and the proof of
prime next greater than p.
t Comptes Rendiif!, clviii. (1914), p. 1012
all
has quite recently
would have far-reaching consequences in the theory of Prime Numbers.

*
of
1.
that an infinity of zeros of ^(s) actually
It is highly probable that

it
(^
13-31, 13'4]
Riemanns
13'4.
for
integral
It is easy to see that, if
cr
>
273
^(s).
0,
n'
>
Hence, when a
^ {s)
0,
(^ )
TT
- i
lim
1 e- "''^^ x^-'- ^
dx.
Now,
1
if OT
2ct {x)
{x)= S
= x~
'
e"""'^'", since,
by example 17 of Chapter vi
we
{1 4- 2ts (I/*')},
st(x)x^^~^ dx converges when a
>
have
lim x^
-sr
{x)
(p. 124),
and hence
\.
Jo
Consequently, if
^(s)r
Jo
2,
('.7s')7r-'''"=lim
V"
Now,
a >
^ (x) x^-'''^ dx-
Nô^lJt)
modulus of the
as in 13"12, the
\n = N+l
xi'-'^dx']
last integral
does not exceed
l_e-<^V+l)^x
.'o
e'''''''
M = .V+1
.0
.0
{7r(iY+l)|-'|(iV^H2;\^)7r}l-i<^rQ<r-l)
-*-
Hence, when
cr
>
as uV
--
since
a >
2.
2,
= ri-.^ + ^^-i +x-^-x;7(llx)\x^-'-'^dx+
= -^+^7^^ + f
.rizTOr)a;-i'' +
t!r(^)a:i^-l(fa;
I
l(-^^)(;a;+[
nT(x)x^'-Ux.
Consequently
r(5)r(L')7r-i*- -
Now
values of
v^-r,=
(x^-^^-''>
+x^')x-'^^(x)dx.
the integral on the right represents an analytic function of s for all

s,
by
5*32, since
tn-
{x)
<
on the path of integration

e-'^*
e-""^^
e-""^ (1
- e-"")-'.
;i=0
Consequently, by
all
values of
w. M.
A.
o^S,
the above equation, proved
when cr>
2,
persists for
5.
18
274
[CHAP. XIII
now we put
If
= l + it,
(t)
ls(s-
1) ^(s)
= 1(0,
(}^s^ 7r-^*-
we have
ff^
x-i^ (x) cos
+^j
x~^-st{x)^ log x\ cos
Since
the test of
satisfies
=I-
we may
4'44 corollary,
all
values* of
+ mr) dx
log x
-^
any number of times

Hence, by Taylor's
differentiate
under the sign of integration, and then put

theorem, we have for
tlogx^ dx.
(^
0.
^(0= S
a^^t''',
=o
by considering the
This result
agî is
now
shall
When
cr>
obviously
real.
fundamental in Riemann's researches.
is
when o->
Inequalities satisfied by ^(s, a)
13*5.
We
last integral
0.
investigate the behaviour of ^(s, a) as t^-
easy to see that,
1, it is
+ oo
for given values of
cr.
N be any integer,
if
-)=U- + n)-^- ^-,_s)ilâY- ^-LM'^^
as,
where
u-n
/'"+!
Now, when
tr'> 0,
l/ ()
a-n
"+]
fn
/
Jn
Therefore the series
when
cr
>
i" f^ (s) is
so that 2 /
(s)
is
ai
(u + a)
fn+l
di
(n
{n
+ af-^'
{slin + a)-"-'.
series of analytic functions
a uniformly convergent
an analytic function when <t> 0; and consequently,
bj'
;i=.V
5-5, the function (
(s,
Now
let [t]
a)
(.,
may
a)
be defined when (r>0 by the series
(a
+ )-.- (i_,)(^;^).-. -j/. .
be the greatest integer in
|C)I^
2 \{a + nr\
<
|
+ \{{l-sr'i[t] + ay-^}\+
{a + 7i)-^ +
\t\-mt] + ay
n=0
*
\s\{n + ar''
'
[t]
'^
+ \s\
(n
+ a)
'^
i>=[t]
In this particular piece of analysis it

and then |
s = ^ + it
2
n=[t]
71=0
< 2
Then
and take iV=[^].
is
convenieut to regard
(t) is
as a complex variable,
t.
13-5, 13'51]
Using the Maclaurin-Cauchy sum formula
( 4-43),
we
get
r^
r[t]
Jo
Now when
^(s, a)
Hence f (s,
But,
J[t]-l
^ a- ^ 1 - S
where
> 0, we
have
<a-^+{l-a)-^{a + [t]y-'' -a^-^} +
when
a)
1
= 0{\t
275
p"*^), the
\-^ {[t]
+ af-'' +
constant implied in the symbol
s\<r-H[t]-l
+ a)-''.
being independent of
s.
8^cr^l + S, we have
I
C (, a)
+ A-)-'' dx
\'-n
f-")
+ {'-'+( + tf-"]
''
<0{\t
since
o"
^ 1,
(a
+ x)~'^ â^~'^ {a+x)-'>- when
and
o-^l,
(a+x)
dx,
(a+x)'"^ ^{a+[t])^~'^ (a + x)-^
when
anoJ so
Cis, a)
{'Itr'' log \t\}.
When 0-^1+8,
|C(,
13'51.
We
)|â~^+ i
Inequalities satisfied by f
(a
a) wlien
(s,
+ r'-* = 6'(l).
cr
^ 0.
next obtain inequalities of a similar nature
function f
(s)
we use Riemaim's
C(s)
Now, when
o-
< 1 - 8,
when
<j
^h.
In the case of the
relation
= 2''7r-i r
we have, by
(1
-s) f (1 -s) sin
(isTr).
1233,
r(l-s) = 0{e(*"*)'*'(^-*)-(^-*'}
and ao
C
(s)
Since arc tan
[exp {.V
i/(l
+ (^-o--i01og|l-s|+iarctan/(l-o-)}]C(l-s).
cr)= i7r4-0
(^~^),
the results already obtained for f
according as
a),
(s,
i{B)
is
positive or negative,
we
see,
from
that
= 0{\t\^-''\i{\-s).
In the case of the function ^ (s, ), we have to use the formula of

to obtain the generalisation of this result we have, when o0,
Hurwitz
( 13"15)
<
i{s,a):^-i{nY-^V{\-s)\_e^'^' Uiy-^)-^-''''" i-a{\-s)\
where
Ca
(!-')= 2
1
Hence
(1 -e"^''"')
%'-
-^^^
f(l -) = e2'^^+ 2
e2'^'
+ (S-1)
[/i'-i-
(n-
l)-i]
/T'/""
tt-2^^;
on the right is a uniformly convergent series of analytic functions

whenever o-^l-S, this equation gives the continuation of fa(l-*) over the range
Oô-:^1-S; so that, whenever cr ^ 1 - S, we have
since the series
sin7raCa(l-s)
1^1+
2
1=2
{/i'^~'
+ (n-l)'^-iH-|s-l
/"
n"--
dv..
re=iV+l ]
nX
182
276
Taking ^V=[^], we obtain, as
[cHAP. XIII
in 13"5,
{8â^l-8)
Ca{l-s)=0{\tr)
= 0{\tf\og\t\) {-8^(T<8).
And
obviously
{a<-8).
C{\-s)=0{\)
Consequently, whether a
is
unity or not, we have the results
C{s,a) = 0{\t\^-'')
(a
= 0{\t\^)
^8)
(8^0-^1-8)
= Oi\t\^\og\t\) {-8â^8).
We may
combine these results and those
of 13"5, into the single formula
C(s,a) = 0(i<r"^'log|^|),
where*
r(o-)-i-(r, (o-Ô);
and the log
7-(o-)
From
may
12"1
example
Now,
^S
o-
{z
when
or
S^cr^l + S.
a).
follows that
3, it
aJ =i
a+nJ
(.V
(z
+ a)
the principal values of the logarithms being taken,
az
ji{a + n)J
= 2
n=l
<
^^
I
r(a) = l-(r, (*-$<t^1); r(cr) = 0, (r^l);
The asymptotic expansion of log T
13*6.
If
= A, (Oâî);
be suppressed except when - 8
a,
the double series
is
(-)"'"'
^
^2
^
(a
'ni
(_yn-i
^r
a''
+ 2
w)"'J
j,,'^!
absolutely convergent since
'
'
{aA+ n)
= 1 [_/i(a
- log
^^ + ^
n)
a+ n
a+
converges.
Consequently
e-v^r(a)
log
Now
(^
a)
1
consider
;;;
1
;
ziri J
az
^=1
71
(a
n)
^
+ 2
1,
0,
1,-2,
,,,=2
772'
-.
c ssimrs
^ z^ ^{m, a);
and
integral converges if
*
It
contour of integration being

^
tis, a) ds, the
2^
|
<
t {a)
= 2,
3, 4,
since, as
can be proved that
zahlen, % 237.
...;
z'"'^{m,a).
wi
'
similar to that of 12"22 enclosing the points 5
points
yn
5^^^
cr
^x
(where
s= a +
it),
...
but not the
s = m(m^2)
^{s, a)
(1),
is
the
1.
may
be taken to be i
(1
- a)
when
^ ^ 1.
(t
See Landau, Prim-
THE ZETA FUNCTION" OF RIEMANN
13 '6]
277
Consequently
,
e-y'r{a)
+ a)
I {z
az
+ m)
=i n (a
..
..
-rrz^
Ztti }
s sin its
Hence
V{a)
Now
r (^^ +
let
1
-TTZ'
^.
f
27ri j c- 5 sin tts ^ ^
r (a)
be a semicircle of (large) radius
on the right of
lying
semicircle
V'{a)
a)
the
'
N with
On
f the
semicircle
centre at s
<r = |.
line
this
;ê-'î'f-^'^^s3l.
a)=0(l), !ir*| = |^;<'e-^'^'-g', and so the integrand is*
Hence if
< 1 and tt + 8 ^ arg ^ ^ tt 8, where S is positive, the integrand
is
{\zY e'^^*"^), and hence
^(s,
5:
^^
as
iV^
-* 00
T(a)
log
1Y^~-^-.
(^ + a)
(a)
is
consider
arg
^TT
z]
and
^''
^,
'rrz'
~.
<
1,
^(s, a) ds.
^îssimrs
^ for all values of
12^
|
if
8.
when
^^
[
<
1,
persists for all
8.
--.
-n-îii
going to tend to
where
r^
'Itti J
:^7r
-/
arg z
the above equation, proved
Now
5"5,
when arg zI'^tt
^^
j
Hence, by
^0
a)ds
7r5
i
=-z r'(a)
+ ^r-.
^^
1
this integral defines
values of
t,{s,
sm
It follows at once that, if
But
j)S
cr :$ -
where
n is a fixed integer
and
TTS'^
By
infinity.
;
^(s, a) ds,
sin
and hence
if
13"51, the integrand
is
[z'^
the upper signs be taken, or
signs be taken, the integral tends to zero as i2 -^
e'^^
if
R"'^'^^},"'
the lower
Therefore, by Cauchy's theorem,
Via)
log^
\\ = -z
where
R^
is
T'ia)
j^-~-
î
f->'-h +
^.
Now, on the new path
^^
^(s,a)ds+ X
R,n,
= in.
of integration
s sin ITS
I
\
where
is
independent of z and
t,
and
t{(t) is the function defined in
13-51.
*
The constants implied
in the
symbol
are independent of
and
throughout.
278
Consequently, since
when
[CHAP. XIII
e~^-^',t]'^^-^^-i)dt converges,
we have
\2\ is large.
Now, when
a positive integer,
is
Rm = ^^
by 13-14,
r^'-^,,
m(m + l)(m
+ 2)
i?,
^^
^-^^
'^ and
so
where 6,/ (a)^ denotes the derivate of

^ ^
Bernoulli's polynomial.
Also Ro
is
the residue at
and so
i2
=
by
(i
of
a j log ^ +
) log ^
^' (0,
log
a)
r () -
log
(27r),
13-21.
And, using
13-21,
R_,
is
the residue* at >Sf=
of
_ia_^^s=_..,(i,^>...),(i,s,g,+..,(l_i:g,...).
XT
Consequently,
log
= z\oo'z+ z
75
R_^
Hence
r (z +
a)
==
finally, if
(z
r' (a)
^tt^
r(a)
arg z\-^ ir S and
a -~^^\og z
+ z.
|
is large,
- z + l\og{27r)
+ i
^, __^s
(-)"'~^</>w+2(-)
w=i w(77H- l)(m
^*
2)^^
In the special case when a = 1, this reduces to the formula found

previously in 12-33 for a more restricted range of values of arg 2.
The asymptotic expansion just obtained is valid when a is not restricted

by the inequality < a ^ 1 but the investigation of it involves the rather
more elaborate methods which are necessary for obtaining inequalities satisfied
by ^(s, a) when a does not satisfy the inequality 0<a%l. But if, in the
;
formula just obtained, we write

seen that,
log
when arg (^ +
j
r (2 + a +
1)
a)
f^
<
a
tt
+
*
a=l
and then put

have
+a
for z, it is easily
8, we
2)
log (z
+ a)-z-a + l
Writings = 5+1.
^^^ ^^'^^
o{l);
13*6]
subtracting log (z
a) from each side,
arg
+ a)
{z
we
when both
easily see that
\-^7r 8 and
arg zI^^tt
279
8,
we have the asymptotic formula
logr(^
a)=(^^
where the expression which
+ a-^jlog^-^ + .^log(27r) +
is o
o(l),
(1) tends to zero as \Zi>X).
REFERENCES.
G. F. B.
RiEMANN,
E. G. H.
Landau, Handbuch der Primzahlen.

Le Calcid des Residue, Ch.
E. L. LiNDELOF,
E.
Ges. Werke, pp. 145-155.
W. Barnes, Messenger of Mathematics,
G. H.
Hardy and
J.
(Leipzig, 1909.)
(Paris, 1905.)
iv.
xxix. (1899), pp. 64-128.
E. Littlewood, Acta Mathematical xli. (1917), pp. 119-196.
1.
Shew
that
(2
(5)
+
-
(i
+y2)-* sin
(Jensen,
2.
(s
arc tan 2y)
D Intermediaire des Math.
(1895), 'p. 346.)
Shew that
/*
2-i
C(s)=^_j-2j^ (1+/)-**
sin
(5
arc tan y)
dv
^^-^-
(Jensen.)
3.
Discuss the asymptotic expansion of
\ogG
{z
+ a),
(Chapter xii
example 48) by
aid of the generahsed Zeta-function.

4.
Shew
that, if
cr
>
(Barnes.)
log C
(5)
=2
p
2
m=.l
mp"
the summation extending over the prime numbers
jd
(Dirichlet,
5.
Shew
where A (w) =
prime p.
6.
that, if
o->
when n
is
= 2,
3, 5,
Journal de Math.
iv. (1839), p. 407.)
1,
not a power of a prime, and
{n)
= \ogp when
is
a power of a
Prove that
e~-^-dx
**
^^&
r'(is)
(is) Jo
1
(Lerch,
See the Jahrbuch
ilber die Fortschritte
KraMw
der Math. 1893-1894,
p.
Rozprawy*,
482.
ll.
280
If
7.
00
where
and,
[CHAP. XIII
a;
if 5
<
<
1,
and the
real part of s is positive,
1,
lim
- xy-^
(1
(j)
shew that
=r
(s, A-)
(1
- s).
(Appell, Comptes Rendus, lxxxvii.)

8.
If X, a,
and
be
real,
< a < 1,
and
and
> 1,
and
if
<^(--''^).=?,(^:^.'
shew that
and
s) = tttVo
(b (x, a, 1
{
(Lerch, Acta Math, xi.)
9.
By
contour,
evaluating the residues at the poles on the
shew
that, if k
> 0,
and arg
3/
that, if
^-
>
< Att,
rhi)y-''du,
i,
^'^'^
9^- f
k xi
and thence
that, if a is
of the straight line taken as
fk+cci
e-y=^--.l
and deduce
left
(")
('"^)" " ^^^'')
du = w{x\
an acute angle.
r ^^^1^
^ (0 ^ =
+
t^
TT
cos ia -
W"
{1
+ 2 or (e-)}
(Hardy.)
10.
By differentiating 2?i times under
and then making a -^
Jtt,
the integral sign in the last result of example
deduce from example 17 on
p.
9,
124 that
r-*i- <*.(,)*= <^-cos|

By taking n large, deduce that there is no number ^0 such that | {t)
when t > to, and thence that f (s) has an infinity of zeros on the line <r = ^.
is
of fixed sign
(Hardy.)
[Hardy and Littlewood,

of zeros on the hue
number
Kiemann hypothesis
Landau, Pnmzahlen,
is
i.
P?'oc.
o-
true,
p. 370.]
=i
the
London Math.
for
which
number
Soc. xix. (1920),
< <
^
is
7'
T' is
log
at least
have shewn that the

if the
( T) as ^ ^- 00
;
^-
"^
7"+
(log
7^
see
CHAPTER XIV
THE HYPERGEOMETRIC FUNCTION
The hypergeometric
141.
We
1
have already
+ ^-^.
l.c
series.
2*38) considered the hypergeometric series*
( + l)^(^ + l)
1.2.c(c + l)
"*
a(a + l)(a+2)6(6+l)(6 +
1.2.3.c(c + l)(c + 2)
2)
^"^
from the point of view of its convergence. It follows from 2-38 and
that the series defines a function which is analytic when 2^ < 1.
i
It will
and that
appear later
if
a cutf
14-53) that this function has a branch point at 2
an impassable barrier)
(i.e.
the real axis, the function

plane.
The function
Many
5-3
will
is
is
made from +
1 to
along
analytic and one-valued throughout the cut
be denoted by
F {a,
b;c;z).
important functions employed in Analysis can be expressed by
means of hypergeometric
functions.
(1
Thus|
+2)" = F(-n, ^:^;-z),
log{l
z)
= zF(l, l;2;-z),
e'=\im F{l,0;l;zl0).
Example.
Shew
that
1411.
c; z)
= ~F{a + \, 6 + 1; c + 1;
The valuel of F(a, b;
The reader
*
F{a, b;
will easily verify,
The name was given
bj-
c; I)
when
2).
E(c-a-b)>0.
by considering the
coefficients of
Wallis in 1655 to the series whose Kth term
a{ + b}
{a
+ 2b}
...
{a
.2;'*
in the
is
+ {n-l)b}.
Euler used the term hypergeometric in this sense, the modern use of the term being apparently
due to Kummer, Journal fiir Math. xv. (1S36).
t The plane of the variable z is said to be cut along a curve when it is convenient to consider
only such variations in z which do not involve a passage across the curve in question so that
the cut may be regarded as an impassable barrier.
;
X It will be a good exercise for the reader to construct a rigorous proof of the third of these
results.
This analysis
is
due
to justify is given in 14-6
to
Gauss.
example
2.
A method more
easy to i-emember but more
difficult
282
various series, that
if
ir
<
1,
then
c{c-l-{2c-a-b-l)x]F (a,
+ (c - a){c - b) xF (a, b c +
= c(c-l){l-x)F(a,b;c-l;x)
c
x)
= C(C -l)jl+ ^
where Un
the coefficient of ^" in
is
Now make
1+2
(un
By
x>l.
tin-i)
[CHAP. XIV
i''(a, 6
converges to zero,
x)
x).
if ?i,j>0,
i.e.
(Un-Un-i)x'>'l,
the right-hand
3*71,
tends
side
which
to
zero
when
the case
is
if
=i
R{c- a Also,
>
6)
by
0.
2*38
and
the left-hand side tends to
3'71,
c(a+b-c)F{a,b;c;l) + {c-a)(c- b) F(a, b;c+l;l)

under the same condition
and therefore
Repeating this process, we see that

F(a, b: c;l)
^
'
'
Yl
lira
(\-F(a,
^' b:c
-,
+ n){c-a-b + n)\
\n=o{c
=
if
'
{[
7n; 1)
'
^
lim F(a, b;c
+ m;
1),
these two limits exist.
But
(^
^
12'13) the former limit
integer; and,
m>
c
I
if Un(ci, b,
c)
is ^tt
1 (c
T {c a)r-Fr,
rz, if c is
6)
be the coefficient oi x^ in
F (a,
not a negative
b; c; x),
and
we have
00
\F{a, b]
+ m;
l)
l\^ 2
\un(a,
+ m)\
b, c
n=l
00
^ X Un (ja,,
ib\,
m \c\)
n=l
'
<
Now the
Q,b
last series converges,
decreasing function of
m;
when
??i
therefore, since
lim F(a, b:
and
"
^, 2 Mn (I a
m-\c\n=o
-h 1,
6
I
^,
1,
7?l -f-
a -f j^l 1, and is
[m '\c\}~^>0, we have
>
ic| 4-
+ m 1)=
;
therefore, finally,
r(c)r(c-a-6)
C
I
I).
a positive
14*2, 14-3]
The
14"2.
F (a, b;
differential equation satisfied by
The reader
will verify
the hypergeometric series
283
c:
z).
without
is
difficulty, by the methods of lO'S, that

an integral valid near ^ = of the hypergeometric
equation*
^ ^^
from
10"3, it
is
~ ^^
d? +
(c
- (a +
Shew
1) ^} ^-^
apparent that every point
equation, with the exception of 0,
Example.
1,
- &'^ =
an 'ordinary point' of this
is
and that these are regular
points.'
'
that an integral of the equation
z<^F{a-\-a, 6
10"72
z).
Solutions of Riemann's P-equation by hypergeometric functions.
14'3.
In
+ a; a-^ + 1;
was observed that Riemann's
it
1-^-/3'
j]^-a^-a'
dru
dz-
{aa'
_^
l-y-y\
du
dz
zb
(a-b)(a-c) ^ ^jB' {h -
c){b
-a
_^
77 (c-a)(c-6)
z
u
{z
by a suitable change of
equation
equation
variables, could
we
and, carrying out the change,
a) {z
b){z
0.
c)
be reduced to a hypergeometric
see that a solution of Riemann's
is
fz-ay /z-c\y {
provided that a
a'
is
r./
not a negative integer
(^-a)(c-&)]
'
for
simplicity,
we
shall,
throughout thi^ section, suppose that no one of the exponent differences

a a', /3 ^, 7 7' is zero or an integer, as ( 10"32) in this exceptional
case the general solution of the differential equation
terms
Lindelof, to
Now
still
may
involve logarithmic
the formulae in the exceptional case will be found in a
if
satisfy
which the reader
is
memoir by
:J:
referred.
a', or 7 with 7', in this expression, it must

Riemann's equation, since the latter is unaffected by this change.
a be interchanged with
This equation was given by Gauss.
The constants
are subject to the condition a
J Acta Soc. Sclent. Fennicae, xix. (1893).

hypergeometrische Funktion (Leipzig, 1894).
+ a' + ^ + ^' + y + y' = l.
See also Klein's lithographed Lectures, Veher die
284
We
M,
[CHAP. XIV
thus obtain altogether four expressions, namely,
'z-aX /z-c\y p
^f
= (z-ay
^
^,
.
,
^ry,
-,
^+t;
==(.i)
which are
solutions of the differential equation.
J^|
ft)
all
/3
7',
Moreover, the differential equation
is
,_{c-h){z-a)\
{c-b){z-a)]
>
l+-''';
unaltered
such changes in the above solutions, they will
the triads
(a, a', a),
If therefore
we make
if
are interchanged in any manner.
(yS, y8', h), (y, y', c)
(o-a)(.-6) j-
still
be solutions of the
differential equation.
There are
five
{b, c, a\,
in turn in place of
We
make
thus obtain 4 x
write
[b, a, c]
with corresponding changes of
= 20 new
a,
a,
/S, (3',
7, 7'.
which with the original four
expressions,
altogether twenty-four particular solutions of Riemann's equation, in
The twenty new
series.
solutions
may
be written down as follows
(^
= z-by fz-aY iP'j/8
+ 7 + a,
z-b\^' [z-ay t^{q,^

('-^Y
^^s=('^)
\z \z - cj
/3
+ 7'+a;
^,
^
+ yS-/3';
^.
^
o.
Fy+y + oc,^'+y'+a-l
+ /3'-/3-
cj
o'
',
/D'
"^
(z-b^fz a-y _
\z cl \z c
\Z-C' \Z-C'
Ua
[c, b, a],
{a, c, b],
{c, a, b],
[a, b, c],
terms of hypergeometric
u.
may
such changes possible, for we
(^
'
= (z-c\y (z-b\^ ^{
;
)
-.
(a-c)(^-&)]
,.^
(a
c)(z b)
b){z c)
(^^^
_ J^^^ - c)\
+ /3'-/3;
rr'
^,
{a-b){z- c)
(a
^,
i^^/3+7+a,
/3+7+a
rI
y^
{a-c)(z-b)\
)
'-\
{b-a){z- c)]
^,7+a-+^;l + y-7; ^^_^;^^^_^^

"^"=U-^) U-^J ^|7^ +
|
U,,=
l^-^)'(^-^r F\yâ + ^',y^c^^^';

"^
^
(lllfy7l^Yf f/'+o +
\z-aj \z-al
\z-aj
\z-a)
['
{'
ZS'.
"^
7'+a+/3';
^
'
'
+ 7-7';
'
'
'
'
+ 7-7;
'
>
^j^^^H
(l,-c)(z-a)\
P^-^
(i,-c){z- a)
14-4]
285
Cj
FL + y + ^',a +
y'
+ ^';
+a-
'\\'
/^-cy'/2-a\
a,.=
MlQ
'
/o'
^_^;(^_^^
1^
By
\z-a)
aJ
aj
V^-/ \z-aj
\z
writing
AU
0,
14"4.
^,
C,
respectively,
^
equation satisfied hy
The
F {A,
(c-b)(z-
'
\z
'
6J
\z
'^22
{a-b)(z-c))
j
1,
^|t+/3 + ,7+^+<: 1+7-7;
|,
.T^b)
' ''\
^l'
{b-a){z-c)}\
{c
'
B,
0,
^'
6'-^ -5,
we obtain 24
B; G
r^
"^
a;
for
a,
'
a)}
o)(z a))
(c-6)(^-a)J
a',
yg,
^S',
y,
'
y,
solutions of the hypergeoraetric
x).
existence of these 24 solutions was
first
shewn by Kuinmer*.
Relations betiueen particular solutions of the hypergeometric equation.
It has just
been shewn that 24 expressions involving hypergeometric

hypergeometric equation and, from the general
series are solutions of the
theory of linear differential equations of the second order,
it
follows that, if
any three have a common domain of existence, there must be a linear relation
with constant coefficients connecting those three solutions.
If Ave simplify
*
n-^,
u.,,
ih, ih] u^^, u^g-,
uî, W22
Journal fUr Math. xv. (1836), pp. 39-83, 127-172.

in Forsyth's Treatise on Differential Equations, Chap.
in the
manner indicated at
They are obtained

vi.
in a different
manner

[CHAP. XIV
286
the end of
solutions of the hypergeometric
we obtain the following
14-3,
equation with elements A, B, G, x:
=F{A,B;C-x),
= (- xy-^'FiA -C+\,B-G+1;2-C;x),
y,
y^
y,=(l-xr-''-^F{C-B,C-A;G;xl
y^ = (_ ccY-''{l - xf-^-^'Fil -B,l-A;2-G;x),
- G + 1 1 - x),
y,r = F{A,B; A + B
y^^ = {\-xf-^-''F{G-B, G-A; G-A-B-hl; l-x),
y^^ = (_ xY^'FiA, A-G+\;A-B + \; x-%
y^ = (- x)-^ F{B,B-G + l]B-A + l; x-').
<
it is easy to see from 2-53 that, when
arg(l a;) <
;
If
.r
TT,
1,
the
relations connecting y^, y.2, y y^ must be y^^y-i, y2= rj^, by considering the
of the series involved.
form of the expansions near =
c^;
In this manner we can group the functions

viz.
u-i, v-i,
Un, Hi, u^^, iiis;
Mi3, Wis;
Wjo, U12, u-18, ''ho,
such that
ii^,
members
Wj, ...
z/24
into six sets of four*,
u^, M21, u^', Wg, u^, U22, M24
of the
same
'"9?
Wji,
t<i7> '^19
set are constant multiples of
one another throughout a suitably chosen domain.

In particular, we observe that Ui, u^, u^s, u^^ are constant multiples of a
function which (by 5*4, 2"o3) can be expanded in the form
{z-aY\l+ i en(2-a)4
when
(z
a\
is
small
sufficiently
when arg {z
a)* is one-valued, this solution of
And
P'"*.
\z
\z
b\, \z
To obtain the
is
that
usually written
manner when
difficult.
The
of separate sets
been obtained by
integrals which will be obtained
relations have
but a more simple and singularly elegant method has recently
been discovered by Barnes
of his investigation
Barnes contour integrals for
Consider
members
relations which connect three
much more
is
later in 14-61
^r.
273-1
j_xj
p/
r(c
we
shall give a brief account.
the hypergeometric function^.
--
+ s)
,
'(-zyds,
where |arg( -z)] <7r, and the path of integration

ensure that the poles of r(a + 5)r(6 + 6'), viz. s = a
The
so restricted
c\ respectively are sufficiently small.
elaborate transformations of the double circuit
is
P'"'; P'^', P'^'; P*^', P*^* are defined in a similar
of solutions
14"5.
a)
Riemann's equation
is
curved
(if
necessary) to
n, hn {n = 0,
1, 2,
.),
special formula
FiA,l;C;.)^^^F(^C-A,l;C;^^),
which is derivable from the relation counectiug
Methodus Differential is, prop. vii.
t Piuc. London Math. Soc.
topics by Pincherle, Mellin
(2), vi.
Mj
with
(1908), pp. 141-177.
and Barnes are there given.
i3,
was discovered
in
1730 by Stirling,
References to previous work on similar
14-5]
lie
on the
left
r( 5),
of the path and the poles of
287
viz. s
= 0,
1, 2, ..., lie
on
the right of the path *.
From
13'6 it
follows that the integrand

j+''-c-i
Oils
exp {- arg {-
z)
is
(s)
7r\I {s)\\]
sCO on the contour, and hence it is easily seen (| 5"32) that the integrand
an analytic function of z throughout the domain defined by the inequality
arg 2\ ^TT S, where B is any positive number.
as
is
Now, taking note
of the relation
If
27rij(7
IttiJ c
where
is
the semicircle of radius iV
Now, by
13"6,
when
s is
(1
s)
on the semicircle
sir
cosec
consider
s-rr,
on the right of the imaginary axis
is
^)
-n-i-zy
an integer.
^ ^ ^.â+,_,_,) (-^y
and,
if
6^
iV
.j j
exp
sin stt
'
sinsTT
being independent of arg s
the constant implied in the symbol
( zy cosec
= tt
we have
r(a+^)F(6 +
T(c + s)r{l+s)
N> 00
5)
T(a+s)r(b+s) iri-zy
ds,
F(c + s) F(l +s) sinsTT
with centre at the origin, and
as
F (
j\^
+^
and
e''*
cos 6 log ^
|
lY
<
we have
1,
sin ^ arg
z)
-(iV + ^)7r|sm^||]
exp
^ i\^
+ .^)
cos
6'
log ^
I
(iV
+ ^]
exp|2-^îV^+.^)logj^
exp|-2-^8(i\r +
Hence if
log ^
|
is
negative
(i.e.
8 sin
1
i)|
(^
6*
I
k
zii
11
TT,
\z\< 1), the integrand tends to zero

under consideration) to ensure that
sufficiently rapidly (for all values of 6

'0 as
iVoo
/.
Now
J -ooi
by Cauchy's theorem,
is
[J
~xi
equal to minus
of the integrand at the points s

* It is
= 0,
2'Tri
1,2,
...
J (iV+i)
i)
sum of the residues

Make N>cc and the last
times the
N.
assumed that a and b are such that the contour can be drawn, i.e. that a and
which case the hypergeometric series is merely a polynomial).
are not negative integers (in
288
when
three integrals tend to zero
arg ( ^)
tt
S,
[CHAB. XIV
and
\z
\<1, and
so,
in
these circumstances,
the general term in this summation being the residue of the integrand at
S
71.
Thus, an analytic function {namely the integral under consideration) exists
throughout the domain defined by the inequality

this analytic
function
may
arg ^
<
tt,
and, when \z\<
1,
he represented by the sei'ies
V r(a + n)r(b +
r {c + n) .n
n)
The symbol F(a,
h; c; z) will, in future, be used to denote this function
divided by r(.)r(6)/r(c).
The continuation of the hypergeometric
14'51.
series.
To obtain a representation of the function F (a, b; c; z) in the form

when
> 1, we shall employ the integral obtained
series convergent
2^
If -D be the semicircle of radius p
14'5.
with centre at the origin,
it
on the
may be shewn* by
left
of
in
of the imaginary axis
the methods of
14'o
that
V{a + s)V{h + s)T{-s)

r{c + s)
Jd
as
px
provided that
arg { z)
<
tt,
that the lower bound of the distance of

a positive
number
Hence
it
>
and p><xi
in such a
way
from poles of the integiând
is
(not zero).
can be proved (as in the corresponding work of
arg {z)\<7r and
\z',
>
14"5) that,
when
1,
r{a + s)r(b + s)T{-s)
sin ( c- a - n) tt
r(a+7i)r(l -c + a + n)
/_ x-^^-h
_
~ V r (1 + w) r (1 - 6 + + n) cos n-rr sin (b - - n) it
rZo
sin (c - 6 - ?i) tt
^ T(b + n)r{l-c + b + n)
6 w) tt
n=o r (1 + n) r (1 + 6 + n) cos nir sin (ft
ft
ft
f^_^^
ft
the expressions in these summations being the residues of the integrand at
the points
It
*
or
= - a-
n, s
=-b
n respectively.
then follows at once on simplifying these series that the analytic
In considering the asymptotic expansion of the integrand when s is large on the contour
12-14.
it is simplest to transform T (a + s), V (b + s), T{c + s) by the relation of
on D,
14-51, 14*52]
289
continuation of the series, by which the hypergeometric function was originally

defined,
given by the equation
is
r(c)
^""'^'-^'^^^
where arg ( z)\<

It
-(-^r'^-^(>i-^ + ^^;i-^ +
r(a-c)
{0
c)
tt.
readily seen that each of the three terms in this equation
is
^^^^
is
a solution
of the hypergeometric equation (see 14"4).

This result has to be modified when a 6 is an integer or zero, as some of the poles of
poles, and the right-hand side then may involve logarithmic
r{a + s)r{h + s) are double
terms, in accordance with ^
r
(1
43.
Putting h = c, we see that,
Corollary.
where
if
arg {- z)\< n,
"'
{\-z)-" = ^-. f
r {a+s) r ( -s)
'ZniJ _xt
(a)
-^)~<'--l as 2-*.0, and so the value of
arg(l -
z)
by the inequality arg( |
14*52.
{y
Barnes'
s)T{b s)
lie
z)
lemma
on
\<
14'1)
i:;
is
less
from
than
to
tt
always
-f-oo
caused
n.
if the 'path of integration

uf the path and the poles of T
that,
the right
which
has to be taken in this equation, in virtue of the cut (sec
zyds,
is
(a
curved so that the poles of

lie on the left*,
+ s) r (/3 + s)
then
Write / for the expression on the

If
C be
defined to be the semicircle of radius p on the right of the imaginary axis with
centre at the origin, and
left.
from the poles of r
(y
if
p-s-oo in such a
- s) r (5 -
s) is
way that the lower bound
positive (not zero),
it is
of the distance of
readily seen that
T{a + s)T{& + s)T{y-s)T{b-s) = ^-^ ^}'''^^'^^^^^^^^^^
= 0[..'^-^^+>+-êxp{-2^1/()l}J,
as
s |--oo
on the imaginary axis or on
Hence the
Thus, as in
C.
original integral converges
14'5,
the integral involved in 7
integrand at the poles oiT{y s)T{b s)
/=
*
It is
supposed that
/3,
y, 5 are
is
when^(a + /3-|-y-|-S- 1)<0.
- 2ni times the sum of the residues of the
I
,,=o
we gett
V{a + b + n)T{^ + b + n)
T (?i+l) r (1 -f-S-y + w) sin(y-S)7r*
such that no pole of the
first set
coincides with any pole
set.
These two
W. M. A.
a,
-^^-Oasp-a-oc
evaluating these residues
r( a + y+%)rO + y + ?^)
TT
i
a=oV{n + \)T{\+y-8 + 7l) sin(S-y)7r
of the second
t
and
series converge
2-38).
19
290
And
using the result of ^ 12"14 freely, by 14-11
so,
_ 7rr(l-a-/3-y -5)
+ d) T (3 + 5)
(a
r(a + y )
|r(l-a-y)r(l-/3-y)
sin(y-8)7r
[cHAP. XIV
(1
rO + y)
- a -8)~r (1 - ^-8)J
r(a+y)r(^ + y)r(a + g)r(^ + a)

^
ro^ ^
- r(a+/J + y + 8)sm(a + ^ + + 5).sin(y-8). ^^"^ ( +
^) ^ '^"^ ('^^^^
y
-sin (a + S) 77 sin 04-S)7r|.
But
2sin(fi+y)7rsiu(/3 + y)7r-2sin(a + S)7rsin (3 + S)7r
= cos (a /3) TT - cos (a + /3 + 2y) tt - cos (a jS) + cos (a + ^ + 28) tt
= 2sin(y-8)7r sin(a+/3 + y + S)7r.
^Vja + y)n^ym-_+^) V {^ ^ ^)
^
Therefore
^(a + ^ + y + S)
r
tt
which
the required result
is
has, however, only been prt)ved
it
when
^(o + /3 + y + 8-l)<0;
it is true throughout the domain through
which both sides of the equation are analytic functions of, say, a: and hence it is true for
all values of a,
y, 8 for which none of the poles of r (a + s) r (/3 + s), qua function of s,
coincide with any of the poles of r (y - s) r (8 - s).
hut,
})y
the theory of analytic continuation,
/ii,
Writing
Corollary.
the result
is still
true
s+
a k, $-k, y + k, 8 + k in
when the limits of integration
k.,
place of
are
;{,
a,
s,
cx)
i,
y, 8,
/3,
we
where k
is
see that
any
real
constant.
14 "53.
We
The connexion between hypergeometrie functions of z and of
hav seen that,
r(c)
'
'
if
arg
- s)
<
T{c + s)
_x
~kx
l^TTl J
"
'
T{a^t)Y{h-irt)T{s-t)V{c-a-h-t)dt\
\^r-.\
''^TTi
z.
tt,
2ni j _x,-
'
.T{c-a)T{c-b) '
by Barnes' leinma.
If k be so chosen that the lower
t
contour
may
is
positive (not zero),
it
bound
may
of the distance
contour and the
be interchanged.
Carrying out the interchange, we see that
if
arg
(1
T(C'-a)T {c-h)T {a)r{b)F{a,
= s/
1
= ^r.
-z) be given
its
principal value,
b; c; 2)/r (c)
r{a + t)r{b+t)r{c-a-b~t)\-.
^T^lJ-k-Jii
{'ZtvI
r{s-t)ri-s)(~zYds\J -jci
dt
f-k+oci
I
2ni J -A-K/
*
between the
be shewn that the order of the integrations*
Methods similar
{a
+ t)r {b+t)r {c~a-b-t)r {-t) {i-zy dt.
to those of 4-51
may
be used, or
it
may be
proved without
that conditions established by Bromwich, Infinite Series, % 177, are satisfied.
much
difficulty
14'53, 14 6]
Now, when arg (1 -k)

methods of Barnes' lemma
< Stt and
( 14-52)
291
I -z\ < 1, this last integral may be evaluated by the

and so we deduce that
|
r{c-a)r{c-b)T{a)rib)F{a,b;c;z)
= r {c)r {a)r {b)T {c-a-b) F {a,

+ r {c)r {c - a)
r (c-b) r {a + b -
a + b-c + l
c) {I
-zy-'"-"
This result has to be modified
if c
{a--\-t)
when arg {z)\

|
Taking
<
1,
tt,
arg
an integer or
{c
- a -b - t)T {-t)
{b->r t)
may
<
With
appear.
z)
at
zero, as
\-z),
z=\.
then
this exception, the result is
tt.
tend to a real value, and we see that the result
<2<
such that
is
- 2)
(1
we may make
holds for real values of
still
1.
Solution of Riemann's equation hy a contour integral.
14"6.
We
<
a6
has double poles, and logarithmic terms

valid
F{c- a, c-b; c-a-b + l;
F {a,
a result which shews the nature of the singularity of
l-z)
next proceed to establish a result relating to the expression of the
liypergeometric function by means of contour integrals.
Let the dependent variable u in Riemann's equation

by a new dependent variable /, defined by the relation
u=
The
- aY
(z
{z
differential equation satisfied
d^
dz-
l+^-'
+ aâ'
z a
(
+ /3 +
7) {(g
- bf
(z
by /
is
in the
10"7) be replaced
c)y I.
easily found to be
1-f^- 7) dl
z c
dz
]
+ ^ + ry +
1) ^
{z a){z
which can be written
+ Sa
(g
+ /?' + 7' - 1)1
0) {z c)
form
Q(z)^^^-[{X-2)Q'{z)+R{z)]f^
+
where
li (^
2) (X
(/' (z)
-^(X-DR (z)] 1 = 0,
\=l-ot-^-j = a' + ^' +
y',
iQ(z)^{z-a)(z-h){z-c),
(R(z) = X{a' +/3
+ ry){z-b){z-c).
It must be observed that the function / is not analytic at x and consequently the
above differential equation in / is not a case of the generalised hypergeometric equation.
,
We
shall noiu sheiu that this differential equation can be satisfied by
integral of the form
1=1
J
{t- ay'+^+y-^
(t
- 6)-+^'+y-i (t -
c)-+^-y -1 {z
- #)--^-y
dt,
jirovlded that C,
tJie
contour of integration,
is
suitably cJiosen.
192
an
[CHAP. XIV
292
For,
we
if
substitute this value of
in the differential equation, the con-
dition* that the equation should be satisfied becomes
j
J
(t- ay+^+y-'
{t
- by+^'+y-'
{t
cY+p+y'-' {z
- t)-''-^-y--Kdt =
0,
where
ir
= (\ - 2) |q (z) +
{t-
z) Q' (z)
+ l(t- zj Q" {z)
+ (t-z){R(z) + (t-z)R'(z)]
= (\-2) {Q (t) -it- zy] ^-it-z) [R (t) -a- zy s
= - (1 + a + /3 + 7) (^ - a) (^ - 6) (i - c)
+ S (a +
+ 7) (^ -
/9
^) (
c) {t
It follows that the condition to be satisfied reduces to
The
is
integral
ay+^+y
is
/3
7)1
dV =
-7- dt 0, where
c dt
- by+^'+y (t - cy+^+y (t - ^)-(i+-+p+v.
therefore a solution of the differential equation,
V resumes
such that
(t
z).
v=(t-
('
value after
its initial
when
has described C.
Now
V={t-
ay'+^+y-'
- by+^'+y-'
(t
U = (t-
where
Now
Z7 is
a)
- cy+^+y-'
-b){t-
a one-valued function of
must be such that the integrand

after
{t
(t
c) (z
hence,
in the integral
[z
t)-''-^-y U,
- ty\
if
be a closed contour,
/ resumes
its original
it
value
has described the contour.
Hence
finally
any integral of the type
(z-ay(z-by(z - c)y
(t-ay+y+''-\t-b)y+''+^'-'{t-cy+^^y'-'{z-ty''~^-y
dt,
JC
where
resumes
that
either
is
its initial
a closed contour
value after
V lias the same value
at
its
in
the t-plane such
has described
it,
that the integrand
or else is a simple curve such
termini, is a solution of the differential equation
of the general hypergeometric function.
The reader is
and
referred to
the memoirs of Pochhammer, Math. An)i. xxxv. (1890),
Hob.son, Phil. Trans. 187 a (1896), pp. 443-531, for an account of the
methods by which integrals of this type are transformed so as to give rise to the relations
pp. 495-526,
of ^ 14-51
and
Example
1.
14-53.
To deduce a
real
definite
integral
which, in certain circumstances,
represents the hypergeometric series.
The
imder the sign of integration are legitimate ( 4-2) if the path C does
a3d does not pass through the points a, b, c, z if C be an infinite contour or if
iliEferentiations
not depend on
passes through the points a,
b, c
or
z,
further conditions are necessary.
14-61]
The hypergeometric series F{a, b; c;

equation defined by the scheme
z) is,
as already shewn, a solution of the differential
00
zy
\\c
293
c- a-
If in the integral
which
a constant multiple of that just obtained, we
is
we
attention to the validity of this process),
and this tends to zero at t=
We
V in
the limiting form of
accordingly consider
greater tlum
(without paying
t''''^{t-\y-^-^{t-zydt.
Now
make 6^-x
are led to consider
question
and ^=
cc
R (c) > R(b)>0.
provided
Vf-''-'^
t"~'= {t
is
{t-z)~"
dt,
where
is
not* positive and
1.
In this integral, write
= u~^
/

(1-jt)'^-''-' (l-M2)-0??<.
?{''-'
.'
We
are therefore led to expect that this integral
for the hypergeometric
The reader
will easily see that if
branch of l-uz
found
is
specified
be
a solution of
the differential equation
R (c) > R{b)> 0, and if arg u = arg (1 - m) =0, while the
by the
(1-m2)-^1
that
fact
as m-^0, the integral just
is
T{b)r{c-b)
^^
This can be proved by expandingf
using
may
series.
(1
^(, &; c;z}.
s)-a
in ascending
powers of
when
<1
and
12 "41.
Example
We
now shew how an
shall
( 14-3) of
satisfies
from the preceding example.
which represents the particular solution i^'^
integral
the hypergeometric difterential equation can be found.
have seen
l={^z-
result of 14-11
Determination of an integral which represents i^"\
14'61.
We
Deduce the
2.
( 14-6)
df [z~bf
the
{z
that the integral
- cy ( {t-a)P+y+-^'-\t-b)y+<'+P'-^(t-c)'' + ^+y'-Ht-z)-<'~^-ydt
difl:erential
equation of the hypergeometric function, provided
is
a closed
contour such that the integrand resumes its initial value after t has described C. Now the
singularities of this integrand in the ^plane are the points a, b, c, z; and after describing
the double circuit contour
( 12'43)
symbolised by
(6
c-h, 6
the integi-and returns
to its original value.

*
This ensures that the point
The
ljz
justification of this process
by
is
not on the path of integration,
4'7 is left to the reader.
294
Now,
if z lie
whose centre
iu a circle
is ,
the circle not containing either of the points
we can choose the path of integration

\z a\<i\t a\ for all points t on the path.
h and
c,
Now
so that
choose arg(2 a) to be numericall}' less than
they reduce to* arg(a
arg(a-c) when z-â\
6),
fix
outside this circle, and so
is
and arg(i 6), ai*g(2-c)
it
{t
a) when
so that
6), arg
(<
also choose arg
{t
arg(^ a), arg
the point iV at which the path of integration starts and ends
reduce to arg
[CHAP. XIV
(<
c)
- z)
at
to
z-â.
(z-bf = (a-bf
Then
|l
+/3 (j^'^)
...j-
(.-c)V=(-o)'{l+^(i^?) + ...l,
and since we can expand (t-z)~'^~~'^ into an absolutely and uuifornily convergent
we may expand the
integral into a series which converges absolutely.
Multiplying up the absolutely convergent

z
multiplied by
We
(s
can define
series
a).
P^'''\
series,
we get a
series of integer
powers of
Consequently we must have
P^^\ P^'\ F^y\
P^^'^
by double circuit integrals in a similar
manner.
Relations hetiueen contiguous hypergeometric functions.
14'7.
Let P{z) be a solution of Riemann's equation with argument z, singularities

Further let P(z) be a constant
c, and exponents a, a', /3, j3', y, 7'.
multiple of one of the six functions P*, P'"'', P'^', P'^\ P<y, P't*. Let
Pi+i,m-i(z) denote the function which is obtained by replacing two of the
a,
h,
exponents,
and m,
in
functions, since
It
its
was
first
and
P{2) hy
and
be contiguous to
P;+i,i_i (z) are said to
m may
m1
P (z).
be any two
Such functions
respectively.
There are 6 x 5
= -30
shewn by Riemannf that
the function
P(z) and any
contiguous functions are connected hy a linear relation, the
which are polynomials in
There
will clearly
to obtain them,
F{z) = {z- ay
where
*
is
we
(z
be
x 30 x 29
(z
P {z^
of
in
c)y f
J c
435 of these
relations.
To shew how
in the form
{t-
ay+y+'^'-^
(t
(t
- cY+^+y'-'
6)v+'^+^ -^
(z
- ty-^-y dt,
a double circuit contour of the type considered in 14*61.
The values of arg (a-b), arg (a - c) being
t Abh. der
tivo
coeffi,cients
z.
shall take
- hY
contiguous
of the six exponents.
k.
fixed.
Ges. der Wiss. zu Gottingen, 1857;
between contiguous hypergeometric functions.
Gauss had previously obtained 15 rehitions
147]
First, since the integral
resumes
[
J
On
Ic dt
{(t
round
value after
its initial
295
of the differential of
has described
- ay+^+y (t - by^^'^y-Ht -
is zero,
any function which
we have
cY+^+y'-'^ {t
zy-^-^-y]
dt.
performing the differentiation by differentiating each factor in turn,
we get
(a'
+ y)P + (a + ^' + y-l) Pa-+,,^'-, + (a + (3 + y'-l) Pa'+i,y-i

_ (a + /3 + 7) p
symmetry shew that the right-hand
Considerations of
side
of
this
equation can be replaced by
These, together with the analogous formulae obtained by cyclical inter-
change* of
(a,
a) with
a,
connecting the
(6,
/3,
^')
and
(c,
hypergeometric function
7,
are six linear relations
7'),
with
contiguous
twelve
the
functions
Pa+],^'-l,
-I
-I
-I
a.'+l,^' 1)
Next, writing
of writing
a'
p+i^y'i,
-/y+i,a_l,
a'4-],Y'_i,
^'4.),-y'_i,
-ta-],v'-l5
-t
j3'+l,
a = (t- h) + (6 ), and
1 for a' in
a'-l
>
-Lfi+iâ'-l,
''^y'+I, a'-
usingf
Pa'_i to
^y~\,^'~\i
-*7'+l,^' 1-
denote the result
P, we have
P = P.._:,,.+:-^(6-)P.-:.
P = P_i,y'+i + (c - a) Pa'-i-
S-imilarly
Eliminating Pa-i from these equations, we have
{c-h)P + (a - c) Pa-_:,^vi + (b- a) Pa'-.,y-,, =
0.
This and the analogous formulae are three more linear relations conP with the last six of the twelve contiguous functions written above.
necting
Next, writing
P=
(^
P^+:, v'-i
^) = (^
a)
(^ a), we
readily find the relation
- (^ - )"^' (^ - f^y (^ - 0^
Jc
which gives the equations

(^
_ a)-i {P - (^ - 6)- P^+,, Y--:} = (^ - h)-^ [P-{z- c)-^ Py+, a'-i}

= Kz - c)-' [P-{z- a)-' Pa+i,^ _i}.
The interchange
is
to be
made only
in
the integrands
the contour
is
to
remain
unaltered.
t Pa'-l
unity.
is
not a function of Riemann's type since the
sum
of
its
exponents at
a, h, c is
not
296
These are two more linear equations between

contiguous functions.
[CHAP. XIV
and the above twelve
We have therefore now altogether found eleven linear relations between

and these twelve functions, the coefficients in these relations being rational
Hence each of these functions can be expressed linearly in
functions of 2.
terms of P and some selected one of them that is, between P and any two of
a linear relation. The coefficients in this

relation will be rational functions of z, and therefore will become polynomials
in z when the relation is multiplied throughout by the least common multiple
the above functions there exists
of their denominators.
The theorem
therefore proved, so far as the above twelve contiguous
is
It can, without difficulty, be
functions are concerned.
extended so as to be
established for the rest of the thirty contiguous functions.

If functions
Corollary.
by a+p,
a'
+ g,
/3
+ r,
jS'+s,
be derived from P by replacing the exponents a, a, /3, /3', y, y

y + t, y' + x, where p, q, r, s, f, u are integers satisfying the
relation
p-{-q + r + s + t + u=0,
P and any two such functions there

which are polynomials in z.
exists a linear relation, the coefficients
with the two functions by a chain of
then between
in
This result can be obtained by connecting
down the
linear relations which connect them

and the two functions, and from these relations eliminating the intermediate
intermediate contiguous functions, writing
with
contiguous functions.
Many theorems which

for the
will
Legendre functions
be established subsequently,
( 15-21),
e.g.
the recurrence-formulae
are really cases of the theorem of this article.
REFERENCES.
in. pp. 12.3-163, 207-229.
C. F.
Gauss, Ges. Werke,
E. E.
KuMMER, Journal fur Math.
G. F. B.
RiEMANN,
Ges.
Math. Werke, pp. 67-84.
E. Papperitz, Math. Ann. xxv.

S.
xv. (1836), pp. 39-83, 127-172.
(1885), pp. 212-221.
PiNCHERLE, Rend. Accad. Lincei
E. ^Y.
Barxes, Proc. London Math.
(4), iv.
(1888), pp. 694-700, 792-799.
Soc. (2), vi. (1908), pp. 141-177.
Hj. Mellin, Acta Soc. Fennicae, xx. (1895), No. 12.
1.
Shew
that
F{a,
2.
F(a,
Shew
IB;
that
if
+ \;
c;
z)-F{a,
b; c;
z)
= F{a + l, h + l;
- .v)-7-F{a, ^
A') is
+ l;
z).
/3 and y are not integers, then the ratio

independent of x, and find its value.
o is a negative integer while
+ /3 + l -y;

3.
P iz)
If
P and
terms of
d'-P
and -TV
4.
jP(|, I
i.e.
Shew
1
be a hypergeometrie function, express
dP
its
d^P
--^
derivates -,- and
linearly in
contiguous functions, and hence find the linear relation between P,
P satisfies
verify that
that i^{j, j
Shew
z).
functions are equal
equal to F{\, |
297
the hypergeometrie difterential equation.
4^(1 -z)) satisfies the hypergeometrie equation satisfied by
that, in the left-hand half of the lemniscate
and
1
dP
-j-
\z{\-z)
= j,
these two
in the right-hand half of the lemniscate, the former function is
2).
5.
\i Fu,^=F {a + ^,h; c; .r), /^_ =i^(a - 1, h; c; x), determine the 15 linear relations
with polynomial coefficients which connect F{a, h; c; x) with pairs of the six functions
Fa^, Fa-, F,,, F,_, F,^, F,_.

6.
(Gauss.)
Shew that the hypergeometrie equation
x(x-l/J^^-{y-(a+^ + l)x}'^ + aî/ =

by the two integrals (suppo.sed convergent)
is satisfied
[\^-\l-z)y-^-'^{l-xz)-''dz
J
['/-i(l-2)"-i'{l-(l-^)2}~"rf2.
and
.'
7.
Shew
that, for values of
AF{^â,i^; i;
is
where A,
8.
B are
Shew
"
and
x between
{\ --Ixy-l
+ B {l-2x)
the solution of the equation
1,
F {h{a + l),
h{fi
+ l);
|; {l-2xy-},
arbitrary constants and F{a, ^,y;x) represents the hypergeometrie series.

(Math. Trip. 1896.)
that
Jim|F(,/3,y,..)-^J^(-)
,;r(y-a)r(y-^)r()r(^)
_ r(y-a-^)r(y
^'
^^
^(y-a)^(y-^)
where h
is
the integer such that k
(This specifies the

.r--l -0
9.
manner
in
provided that o-f |3-y
Shew
that,
when
i? (y
^R (a + ^ y)<k-\-\.
which the hj'pergeometric function becomes

not an integer.)
infinite
- a - /S)
< 0,
when
(Hardy.)
is
then
-/3-y
T{y)n''^^-y
^"(4-^-y)r(a)r(^)
as '/i^-x
where S^ denotes the sum of the

(M.
first
J.
--l
n terms
M.
of the series for F{a,
Hill, Proc.
London Math.
/3
1).
Soe. (2), v.)
,.
298
Shew
10.
that, ii
be indeijeiident sokitions of
i/i, y-i
2-^1-^^-.
then the
is z
[CHAP. XIV
oreneral sohition of
= Ay^-\-Byiy.i-\-Cyi',
where J, B,
are constants.
(Appell, Comptes Rendus, xci.)
Deduce from example 10
11.
^F(a h^'
'
r-
r)\^=
^'
'
'
that, if
+ | = c,
ft
^_rW_r(2c-l)
T {2a + 7>) F ia + b+n)T {2b+7i)

*'
nl T {c + n)T {2c- 1+n)
^^,,
r{2a)T{2b)T{a + b)n=o
(Clausen, Journal fiir Math,

12.
Shew
that,
if
F{2a,
<i and
13.
Deduce from example 12 that
14.
Shew
that, if
co
= e^'^' and
i^(a,3a-l; 2a; "
i^ (a)
-^)
iii.)
x{\ s) <|,
a+/3 + *; x} = F{a,
2/3;
'
<
/3
a+^ + l
(Kummer.)
4.r(l-.r)}.
1,
= 3^'' '* exp [K^^Sa - 1)} ^|^^^|
i^(,3a-l; 2a; -co) =3^^^- ^ exp {^W
(1
-3a)}
^M^^|
(Watson, Quarterly Journal, xhi.)

15.
Shew
that
V
-rj>
-r_j
..y
(Heymann,
16.
If
il-xY+^-y F{2a,2^;2y;
x)
V9;
r(|)r('/i + f)
Zeitschrift fur
Math, und Phys. XLiv.)
= \+Bx + C:'f- + Dx^+...,
shew that
i^(a, /3;
y+^;
^)
i^(y-a, y-/3
(4),
y+l;
x)
(7+l)(7 + i-)
y+l
(Cayley, PAz7. Mag.
(y+*)(y + f)(7+|)
See also Orr, Camb.
xvi. (1858), pp. 356-357.
Phil.
Trans, xvii. (1899), pp. 1-15.)

17.
If the function
i^(a,^,/3', y;
x,y) =
then shew that between
{a,
ji, /3',
_-^^
r (y -
y) be defined by the equation
.*;,
\\i'^-\\~xc)y-''-\\- ux)-\l-uy)-^' du,
(a)
a) j
F and
any three
i^(al),
of its eight contiguous functions
F{ii\),
F{ii'l),
F{y\\
there exists a homogeneous linear equation, whose coefficients are polynomials in x and y.
(Le Vavasseur.)

If
18.
and
y-a-
that, if
< 0,
iS
shew
^-0-/3 = 0,
that,
299
x -^1-0,
a.s
the corresponding approximate formula
is
(Math. Trip. 1893.)
Shew
19.
/(2
when \x\<
that,
1,
+ ,0 + ,a:-,0-)
fl
= _ ^,r/' sin OTT sin (y - a) TT ^'^"^^-^"V (a,

.
where
of v x
denotes a point on the straight line joining the points
and of
same
are the
as that of
and arg
.>;,
(1
j/)
/3
^),
the initial arguments
0, x,
-*-0 as v-^0.
(Pochhammer.)
20.
when
If,
arg
(1
-x)\< 2n,
K{x)=r
_
LTTl J
and,
when
arg.*;
y,
{T{-s)Ti}; + s)\^'{\-xYds,
,
< 27r,
l-lZl
by changing the variable
(.r)
s in
J _K,-
the integral or otherwise, obtain the following relations
= A" (1
- ^),
K{l-x) = K'(x),
K{x) = {l -
A"
(1
.r)-4 A'
(^-^^
.i?
|
<
tt
'I-
iT (^)
(1
\<n.
.r)
arg.r
if
<7r.
arg.r |'<
if
tt.
A"
(^
if
_") ,
_J_-
arg
(1
.x-)
I
< ^-
(Barnes.)
j'-^- {log^-4log 2 + 4 (1
f^lt
+ --^^}
and
= + i - ^) - 5
(
when arg { x)\<ir, the ambiguous

j
<^
2^K' {x)= - l^
of the preceding example, obtain the following results
11=0
when \x\<l, arg
arg
if
2 /{' (Ijx),
- a;) = (1 - X) - i
With the notation
21.
- :f)
(1
if|argj;|<7r.
K{l-x)^x-h K f"- ^
K' {x) = x
arg
if
a: (l/.r)
+ - ^) - 2 ^' (1/:??),
sign being the
same as the
.sign
of /
(x).
(Barnes.)
300
22.
Hypergeometric
series in
two variables are defined by the equations
F,{a;^,^';y;
.r,
y)
^'
,r,
y)
-ts{a,a,^,^; y;
x,
y)= 2
F,
(a
/3,
y,
[CHAP. XIV
"Xn^f"^
^'"J/'S
^^^'-, x^y^.
^n(^n Pm Pn
--,-_
,(
^^
??l
,
.
yi ^
x^'y"
,1
FUa,^;y,y';.,y)^ ^
.XpV^'^'"'"^''
where
a,
= a(a + l). ..(a + m 1),
and 2 means 2
2
m=0 =0
?n,
Obtain the differential equations

?F^
?Ft
d'F^
c"F.
-(l-)8^+y(l-)5j;5^+{r-(+e+l)-}5;j'-*^-.^/-,=0,
d^F^
d^F^
dF;
.il-o:)-^+y^^^Hy-{a + ^ + l)x]j^-a^F, = 0,
and four similar equations, derived from these by interchanging x with y and
y when a', ^', y occur in the corresponding series.
a,
/3,
y with
a', ^3',
(Appell, Comptes Rendus, xc.)

23.
If
where
v is
is
negative,
and
if
a= V + a,
an integer and a
is positive,
r (a)
D = (-)"(-l)(a-2)
^
'^ ^
^-^^
u
where
"i
24.
(x)
shew that
When
< 1,
(*/
= ^^^
"^
(Hermite,
Joiirnal
\
mr
Math,
xcii.)/
shew that
{a)
^ _l
Rn
n=\X a n''
I
n=\X + n
(-)a(a + l)
where
...
(a
+ .-l)
^
7i
25.
(a-%)' ^,
G(^-n),
x+n
T{x)T{a-x)^
a,
...
When
> 1,
and
and a are respectively the integral and
fractional parts of
shew that
T{x)T{a-x)^ G{x)p.,_
r(a)
=i
.r
?i
- (?( ^')p,^
=i .r-a
f'-'X
-G{x)\-^+^^^
...+
'\_X a
X-a-\ +
.r-a-i^ + lj'
^

(?(^)=(l-?^
where
--^
a+l/)
...
ajjj
_(-)"a(a+l)...
(a
(l
a/ V
It,
(l _
301
.^_^)
+ v lj
+ -l)
(Hermite, Journal fiir Math, xcil.)

If
26.
,\-\1 .)-
(T
./ (.r, y,
/
where n
is
r
C,
-^'(jZ
-
+ ^ + n-l) +C.3
., x{x+l)(7/ + v + n-l)(^ + v + n)
- ...
3,(^ + 1) (^ + ,)(^.^, + i)
-^-^^,)
a positive integer and
...
are binomial coefficients, shew that
r(y)rCy-^-H/^)r(.r-n;)r(y+?0
,^
f (T y
./H^.y.
C2,
C'i,
Y{y-x)T{y + n)V{:6)V{x-irV^n)'
xxxv. a number of similar results are given

by Dougall, Proc. Edinburgh Math. Soc. xxv.)
(Saalschiitz, Zeitschrift fur Math.
shew
that,
liib +
when
\)> 0,
-},a
i^(a,a-5+l,-c
+ l';
^'
^
S, e
'
'
then
l)=2-
^
'
'
..A^^^P^W^"^"^
r(8-â)r(f-ia)r(^ + ia)r(S + e-a-l)

London Math. Soc. xxxv.)
(A. C. Dixon, P/'oc.
Shew
28.
that,
R (a) <
if
.if,
then
a(a+l)... (a-l-n- l)l^
r(l-3a)
,.^,,,
(Morley, Proc. London Math. Soc. xxxiv.)

If
2!).
[' /'....-I
J
i>
(1
_.,.)>-iy-i
_^)A-.
(1
-a,yy--J->'dxdi/ = B(i,J,
(1
X-,
/,
m),
i>
shew, by integrating with respect to
symmetric function of i+J,j + i\
jc,
/,
and also with respect to
+ in,
i/,
that
B {i,j,
/,
/,
m)
is
/n+i.
Deduce that
F{a,ld,y;
is
a symmetric function of
(A.
('.
8, f ,
fi,
l)^r
(8)
S + e a - /i,
()
8+e ^-
Dixon, Proc. London Math. Soc.
(6
+6
y, 8
(2), II.
- a -/3-y)
+ e - y - a.
(1905),
8-16.
jjp.
For a proof of
a special case by Barnes' methods, see Barnes, Quarterly Journal, XLI.

(1910), pp. 136-140.)
If
30.
/ = i^(-, a + ;.; y,
shew
that,
when
^=
where
.r
.r)
x^~^ il-x^f'"-
n is a large positive integer,
,^-/,
d"
-^-^^y^-^^^^
and
^^^
(\-x)
0< < 1,
.r
(sin,^)i-V(cos0)V--icos{(2u + a)(^-i,r(2y-l)} + o(
1^),
= sin2^.
(This result
is
contained in the great memoir by Darboux, "Sur I'approxi-
mation des fonctions de

(1878),
pp.
5-56,
trfes
377-416.
grands nombres," Journal de Math.
(3),
iv.
For a systematic development of hyper-
geometric functions in which one (or more) of the constants
Camh. Phil. Trans, xxii. (1918), pp. 277-308.)
is large,
see
CHAPTER XV
LEGENDRE FUNCTIONS
151.
Definition of Legendre polynomials.
Consider the expression (1 -2^/i -|-/?^)~ S when \'izh-h'^\<l, it can be

expanded in a series of ascending powers of '^.zh - Ii-. If, in addition,
2zh
+ \h\-<l,
these powers can be multiplied out and the resulting series
rearranged in any manner

in powers of \2zh\
+ \h\'^
rearrange in powers of
(1
2zh
h,
2-52) since the expansion of [1
2zh
+ \h\^]]~^
In particular,
then converges absolutely.
we
if
we get
+ /r) -i = P, (z) + hP, (z) + h^-P, (z) + Ps {z)+...,
where
P,{z)^l,
p^ (z)
P,{z)
= z,
= l{Sz^-n
P,{z)
= I {S5z' - Wz' + 3),
P,
(z)
P,{z)
= l(5z^-Szl
= I {QSz' - 70z' +
5z),
and generally
^n
{z)
2,,
_
^^^ ,^,
??i
If a, 6
**
and
^^ 2 ^"
1)'
^2.4. (2n - 1) (2n - 3)
1)
whichever
>
It will
"
being so small that 2ah + h'^^l -8, the expansion
ft
which
The expressions Pq {^), Pi (^).
known as Legendre jwlynomials* Pn
polynomial of degree
'
an integer.
is
converges uniformly with respect to
are
'
-r)!(n-2r)!
be positive constants,
- 2zh + /i^) ~ ^
of (1
2 (2n
2".r!(7?
=
r,to
where
1^
and k when
are clearly
{z)
all
â,
2
j
A ^6.
|
polynomials in
z,
being called the Legendre
n.
appear later (^15"2) that these polynomials are particular cases of a more
known as Legendre functions.
extensive class of functions
Example
1.
By
giving z special values in the expression
other names are Legendre
coefficients
(1
and Zonal Harmonies.
analysis in 1784 by Legendre, Memoires par divers saians,
x.
(1785).
izh
+ K^) "
=-',
shew that
They were iutrodueed
into
LEGENDRE FUNCTIONS
15-1-15'12]
Example
(1
From
2.
the expansion
''
cos ^ + A^')
- 2A
303
= (\+\he^^+^^Ji-e^'^+..\
shew that
= l-3---(^ -l)
(cos 6)
^
k cos
2.4... (2w)
"^
'/i^
+ 2.(2?i
J",!f'"\.
1)
2 cos (
^
-2)5
'
1.3.(2n)(2?i-2)
^
,,
+ 2 .4.(2.-l)(2 -3)^^^^("-^)^
+ -|l
Deduce
that, if 5 be a real angle,
!/>(cos5)|<
so that
I
P,, (cos 5)
Example
^ 1.
Shew
3.
r + 2.(2-l)-^ + 2.^4Tl>3iy(2^rr3)-^ + -j
2. 4. ..2m
(Legendre.)
when z= -\,
that,
Pn
--=
1\ P'ln^ Pi P-zn
Rodrig lies' * formula for
15"11.
It is evident that,
when n
(In
is
an
fin
,.=0^
the general formuhi for P,; (z)
this result
known
is
Shew
Example.
15'12.
1),
(2n-2r)!
r!(n-r)! (n-2r)!
'i!
'
the coefficients of negative powers of z vanishing.

it
as Rodrigues' formula.
that
P^
(2)
has n real roots,
all
lying between
Schldfii'sf integral for Pn(z).
From the
(Ckre, 1905.
Legendre polynomials.
.,
From
or ^ (n
integer,
w=
?i
the
71
where
;j
+ P> P,. -,-... + P...
result of
4,
loll combined with
5*22,' it
where
is
result
is
called Schldfiis integral formula for the
+ 1.
a contour which encircles the point z once counter-clockwise; this
Corresp. sur VEcole poly technique,
t Schlafli,
iii.
(1814-1816), pp. 361-385.
Ueber die zwei Heine'schen Kngelfunctionen (Bern, 1881).
304
Legendres
15"13.
We
shall
[CHAP.
XV
now prove
= P^ (^)
that the function u
is
a solution of the
.
,^
which
is
du
called Legendre's differential equation for functions of degree
substituting
For,
by
d-u
Schlafli's
integral
the
in
left-hand
>l
we
side,
have,
5-22,
^^
" ^"^ ~~dz^ ~
1)
1)"+^
^ (^
this integral is zero, since
C when n
after describing
satisfies
The
(^-
'
z)~"-~-
1)'*+^ {t
resumes
its original
The L^gendre polynomial
an integer.
is
(^)
,,
2'7ri.2^']cdt\{t-zy'+-']
and
(n
^dV ^ " ^" + 1) ^n
^^
value
therefore
the differential equation.

result just obtained can be written in the form
d_ [,,
dz |(1
^,, dPn
- z")
^^^1 +
{z)\
n (n
1)
P. (^) = 0.
It will be observed that Legendre's equation is a particular case of
Riemann's equation,
-1
oc
71
Example
1.
P<^-r
Example
Shew
If
that this
Example
3.
z^
is
t],
+1
z\.
-n
satisfied
by
^"r
+ r+l
-r
z\.
-n+r
2.
7i
shew that Legendre's
is
differential equation takes the
form
a hypergeometric equation.
Deduce
Schlafli's integral for
the Legendre functions, as a limiting case of
the general hypergeometric integral of 14-6.

[Since Legendre's eqiiation
is
given by the scheme
P,
-71
LEGENDRE FUNCTIONS
15 13, 15*14]
the integral suggested
is
=
C such
taken round a contour

it
and
305
,{f^-l){t-z)-''-^dt,
that the integrand resumes its initial value after describing
this gives Schlafli's integral.]
The integral properties of

shall now shew that*
1514.
We
the
=
.'
(m
2
-1
2n+l
Let
and
dJû
denote -r^
[u].r
so, if
Now,
<
n, [{z^
:jfe
n),
{m =
n).
P.n{z)Pn{z)dz^
['
of the two
then,
1)"}^
if
r-^n,
vanishes
numbers m,
[{z-
when
1)"},. is
z=\
n, let
divisible
and when
by
be that one which
{z^
1)""'';
1.
is
equal to or
greater than the other.
Then, integrating by parts continually,
j'
{{z"--i)%,,{{z^-iy%dz
{(^^-ir}^._, {(^^--i)l
-1
(-)'"
r
.'
since
{{z"^
(z-^-i)"'{(z^-ir],,^,,dz,
-1
l)'"}t_i,
Now, when
-f-1 {(z^-lru-^{{^-lru^dz
K^'
m > n,
[(2-
l)'"}7-2.
vanish at both limits.
!)"},-;-= 0,
of order higher than 2 vanish
and
since differential coefficients of (2so,
luhen ni
is
greater than
n, it
follows
from Rodrigues' formula that
'
P^(z)Pn{z)dz =
0.
/:
When
ni
n,
we
have,
by the transformation just obtained,
/:
= (2n) if
y
(1
= 2.(2w)!
- z-Y dz
(l-2-)"fZ^
Jo
2.(2n)!
sin-"+'
dcW
Jo
2 (2w)
.
2.4...(2n)
3.5...(2w+l)'
These two results were given by Legendre in 1784 and 1789.

W. M. A.
1)"
20
'
306
where cos 6 has been written
the integral
for z in
[CHAP.
XV
hence, by Rodrigues'
formula,
r
J _^
We
(P n
I
2.(2n)! (2".7i!)^
M-^^
/
V
/,
^2*^
!)2
(2w
+ 1)
have therefore obtained both the required

language of Chapter
It follows that, in the
orthogonal functions for the interval
Example
1.
Shew
that, if
(?i
+ |)2
P(3) are normal
1, 1).
x> 0,
(cosh2.?^-j)-4p(2)c^s = 25(n + i)-ie-(2"
[^
"
results.
the functions
xi,
2/?
i)=:.
(Clare, 1908.)
Example
2.
If
/=
P,
{z)
P^
(z) dz,
then
.'
/-l/(2/i
(i)
+ l)
(m=w),
{m n
7=0
(ii)
+ "
'-~^,
7= 2'"
,,'
..
^^.^
+ "-i(%-?)0 (" + "i + l) ("
(iii)
\
'*'*
'
?2.
'
(?i
'
.,
even),
= 2i/ + I, m = 2u.).
(m
!)
(Clare, 1902.)
Leg endre functions.
15'2.
P {z^
when n
Hitherto we have supposed that the degree n of

integer
integer.
P (^)
shall now
in fact,
has not been defined except
We
see
how
P,j {z)
is
a positive
is
a positive
can be defined for values of n which
are not necessarily integers.
An
z
integral values of z
to be
The expression
analogy can be drawn from the theory of the Gamma-function.
as ordinarily defined
(2
+ I), and
Referring to
is satisfied
as 2
(viz.
(2
(2
2)
...
1)
has a meaning only for positive
but when the Gamma-function has been introduced,
so a function
1.5"13,
we
will exist for values of z
which are not
can be defined
integers.
see that the differential equation
by the expression
''~27rijc 2-(t-zr+^'^''
even when n is not a positive integer, provided that C is a contour such that
(t2_ i)n4-i^^_^^-n-2 resumes its original value after describing C.
Suppose then that n
is
no longer taken to be a positive integer.
(f^ 1)"+' (^ 2^)~"~^ has three singularities, namely the

= l, t = 1, t = z and it is clear that after describing a circuit round
= 1 counter-clockwise, the function resumes its original value
point
The function
points
the
multiplied by
e^Trtfre-i-i)
counter-clockwise, the
while after describing a/<5ircuit round the point

function
resumes
its
original
t =^
value multiplied by
LEGENDRE FUNCTIONS
15-2, 15 '21]
^>nii-n-:>)
jf therefore
its
t=l and t = z, but

(^^ 1)"+^ (^ 2^)-'*"^ will
be a contour enclosing the points
not enclosing the point
resume
307
= -1,
original value
then the function
after
has described the contour C.
Legendre's differential equation for functions of degree
is satisfied
,x'^^^*
/I
du
Hence,
n,
hy the expression
(1+,^+)
{t^-l)
"=2^-/.^
'2-''
{t
- zf^'^*'
for all values of n the many- valued functions will be specified precisely
by taking A on the real axis on the right of the point t=l (and on the
;
right
of 2
\^Yg{t
z}\<'Tr
if
be
dX
real),
and by taking arg
(^
1)
= arg(^4-
1)
and
A.
''This expression will be denoted by
F^ (2^).
(^nd will be
termed the Legeiidre
function of degree n of the first kind.
We
n
is
have thus defined a function P

an integer or not.
The function Pn{z) thus

two contours as shewn
to
make
in
defined
is
then, as
this involves
the definition being valid whether
not a one-valued function of
for
we might take
the figure, and the integrals along them would not be the same
make
the contour integral unique,
real axis
{z),
a cut in the
making a similar cut
plane from
in the z plane, for
if
to
oo
along the
the cut were not made,
varied continuously across the negative part of the real axis, the contour would
not vary continuously.

It follows,
15'21.
We
by
5'31, that
If
{z) is
analytic throughout the cut plane.
The Recurrence Formulae.
proceed to establish a group of formulae (which are really particular
cases of the relations
shewn
between contiguous Riemann P-functions which were

Legendre functions of different degrees.
to exist in 14'7) connecting
G be
the contour of
JL_[
F
M-- 2^,^^.
^n {z)
j
15*2,
we have*
(^'-^y
dt
(TZTp-i '^^'
P'(z)-''-^^
^" ^^> - 2"+'7ri
( ^-^^^^dt
j c {t - z)--'-'
WewriteP'{z)for-P(2).
202
308
- (t"--iy'^' ^ 2(7i + l)t(t^-l) _ (n + l){t^-iy'+'
Now
and
XV
[CHAP.
integrating,
so,
Therefore
2-+1 iri
Jc{t- zY
J c (t
'2^^+-^7ri
2+^ Tri j
zY^'
^^
{t-zY+'
Consequently
i>,,w-.i'.,{.)=^.jj;:-i^'rf*
Differentiating*,
we get
P'n+, (Z)
'^ndso
(A).
- zP'n (Z) - Pn (Z) = nP, (z),

- zP'n(z) = (n + 1) P,,{2)
P'n^,(z)
This
the
is
first
(I).
of the required formulae.
Next, expanding the equation
we
find that
/.c{t-zY
Jc
Writing (^--1)4-1
for
and
t-
î c{t-zY
Using
J^(^t-ZY^'
for
(t-zY
]c
we get
in this equation,
Jc{t-zY-''
(A), Ave have at once
(n
That
(t-zY
(t z) + z
is
1) {P+, (z)
- zPn
(z)]
7iP_, (z)
nzP
(z)
0.
to say
(n
1) P,î (z)
{2>i
1)
zPn
(z)
nP_, U)
(II),
a relation f connecting three Legendre functions of consecutive degrees.

is
This
the second of the required formulae.
We
can deduce the remaining formulae from (T) and (II) thus
Differentiating (II),
{n
1) [P'+, (z)
Using
(I) to
we have
- zP'n
{z)\
eliminate P'n+i
[zP\
(z),
{z)
P'_,
The process of
If
v.
n,
1)
differentiating under the sign of integration
Pi)[z)
= l,
P_i
(2)
= !, and
the result (III)
is
0.
readily justified by 4-2.
memoir on
is
{z)
(III).
Probability, Misc.
(1770-1773), pp. 167-232.
n = 0, we have
P^
we get
nP,,{z)
t This relation was given in substance by Lagrange in a
Taurinemia,
- {2n +
and then dividing hyl
zP'n{z)-P'n_,{z) =
*
(^)i
true but trivial.
LEGENDRE FUNCTIONS
15-21]
Adding
(I)
309
and (III) we get
P'n+A2)-P'n-^{z)==(2n+l)P,,(z)
Lastly, writing n
for n in (I)
equation so obtained and
(III),
(I)
and eliminating P'_i
(z)
between the
we have
(2-'-l)P'n{z)
The formulae
(IV).
= nzP,{z)-nFn-,(z)
(V).
(V) are called the recurrence formidae.
The above proof holds whether n
is an integer or not, i.e. it is applicable to the general

Another proof which, however, only applies to the case when n is
Legendre functions.
a positive integer
(i.e.
is
only applicable to the Legendre polynomials)
Then, equating coefficients* of powei's of h

equation
in the expansions
{l-2hz + h^)~ = {z-h)
on each side of the
V,
nP{z)-{2n-\)zP^_i{z) + {n-\) P^_o{z) = 0,
we have
is
as follows
V={l-2hz + h^-)-^.
Write
which
is
the formula
(II).
Similarly, equating coeflficients* of powers of h in the expansions
on each side of the
equation
^Th='^'-^^Tz^
we have
which
is
the formula (III).
Example
1.
Shew
dPî^)
" ' ^,-î^"'
' -?-^' - dPr,_y{z)
= nP^
dz
dz
The
(z\
^
others can be deduced from these.
that, for all values of ,
jSz{Pn'+P'n^l)-^PnPn^l] = {^n + -i)P\^,-{2n+\)P^.

(Hargreaves.)
Example
2.
310
The expression of any polynomial as a
15'211.
[CHAP.
series
XV
of Legendre
polynomials.
Let fn
Then
be a polynomial of degree n \n
{z)
it is
always possible to choose
/
for,
{z)
aoPo
on equating coefficients of
which determine
powers of z in fn
a, a_i,
To determine (/q,
by Pr{z), and
= 0,
Example
...
Given
r"
1, 2,
1.
z'\
iPi
z"-~'^,
(5)
...
...
a-^,
so that
+ anPn {z),
on each
side,
we obtain equations
...
uniquely in turn, in terms of the coefficients of
>i
(z).
cti,
the most simple manner, multiply the

Then, by 15"14,
in
integrate.
identity
when
{z)
a,,,
z.
when
>
= aoPo (2) +
<î
n,
the integral on the
(2)
+ -.. + ,ii'n (2),
left vanishes.
to determine ao ii
(Legendre, Exercices de Calc. Int.
[Equate coefficients of
on both sides
s"
"""
Let In,m=
~"'Pm
SO that,
{^) d^,
li.
'^'n-
p.
352
this gives
i^r
by the
result just given,
_2"> + i(m!)^
Now when
vanishes
and
n ru
7
To evaluate
is
odd,
7,
is
also vanishes
/,,
the integral of an odd function with limits
when
when n m
is
71- in is negative
1, and
so
and even.
a positive even integer, we have from Legendre's
equation
m {m + 1)
f ^ z^Pra{z)dz=-(^^^ 2" ^_
{(1
- z^)
P,' {z)}
dz
= J^znii-z^)P^î^,)^+nj'^z-^{\-z^)P,:{z)dz
^
n\^--^{\-z^)P,{z)^
on integrating by parts twice
m {m +
-nj'j,{n-\)z--^-{n + \)z")P,{z)dz,
and so
1 ) 7, ,
=n
(/
+ 1)
/,
-n{7i-l)
/ _ 2. ,
Therefore
T
n{n-l)
n{n-\)
{n
m){n-'2m)
by carrying on the process
...
...
2. ( + ?
of reduction.
(m + l)
+ !)( + ?- 1)
j
...
(27H
+ 3)
LEGENDKE FUNCTIONS
15'211, 15-22]
Consequently
and so
!,,=
,h
= 0, when n m
-,
is
odd or negative,
(2m + l)2"'n\{ln + ^m)l
m=~ri
Example
'
When n m
.i\,
TTT
2.
Express cos
3.
Evaluate the integrals
7id
311
-,
even and positive.!
is
as a series of Legendre polynomials of cos 6
when
n is
an
integer.
Example
j' ^zP,{z)P^,{z)dz,
j[^z'Pn(z)Pn^liz)dz.
(St John's, 1899.)
Example
Shew that
4.
j[^{l- z^)
Example
{P'
(z)}'^
dz=^^^^
(Trinity, 1894.)
Shew that
5.
nP
(cos
6)^ 2 cos reP,,_,.
(cos d).
(St John's, 1898.)
r=l
Example
If
6.
,.
- 22)
/'^^ (s)
(?i-wi)(2 +
Since
where
m<
/<,
shew that
2m + l)M = 2n-'-i-
(Trinity, 1895.)
Murphy's expression* of Pn{z) as a hypergeometric function.
15'22.
15'13) Legendre's equation
equation,
rfs,
it is
is
a particular case of Riemann's
to be expected that a formula can be obtained giving
To determine
terms of hypergeometric functions.

the contour C, let
z is such that
^^
|
Since
convergent series
2-28
-^
\l-t\
1-z
1-t
the
Legendre function and suppose that 1 ^ < 2 to

B be any constant such that
< 5 < 1, and suppose that
^ 2 (1 S) and then take C to be the circle^f*
integral of 15"2 for the

fix
Pn(z) in
this formula, take
^<
1,
= 2-8.
we may expand
(t
z)~'^~^
into the uniformly
:|:
(*-.)--.=o-i) {i+( + i)5j+*'iiiii^>(;^;y+...|.

Substituting this result in Schlafli's integral, and integrating term-by-
term
( 4'7),
we get
.2
(z
- ly (n +
l)(n
2) ...(n
+ r)
r<^+'^+'
(t'
1)"
{t-iy
2-.{r\y
rio
* Electricity (1833).
This
circle
Murphy's
result
contains the points
X The series terminates
if
was obtained only
= l, = z.
t
n be a negative
integer.
[dr
for the
^^
dt
^^
312
by
Since arg (t+l)
5*22.
l'('+^>"
and
so,
when
p
^^ "
^
.
(n
z ^2
(1
+ l)(n +
This
00
When
a polynomial in
Exam/pie
2"-'- n
B) <
2)...{n
is
it is
+ r).(-n)(l-n)...(r-l-n)
that, if
supplies a reason ( 14"53)
it
- 2 ^j
why
the cut
15*2.
obvious that, for
all
values of n.
coeificients.
r(2m + ^+2)
that the Legendre polynomial
(cos 6)
l,
-n;
\; cos^ 16),
hS F { -
n,
-n
next shew that,
the Legendre function
Pn
(-)F(?i +
Laplace's integrals* for
shall
V2
be a positive integer,
cos''
We
V"
(i
a positive integer, the result gives the Legendre polynomial as
- s with simple
and to
15*23.
-r + 1),
^-^^).
1;
this result,
Shew
2.
(n
we have
2,
f--'P.n..{^\
Example
we get
(n-1)...
could not be avoided in
Shew
1.
1,
(H)^
From
Corollary.
XV
t=i
the required expression
is
1 to
Note.
,.=0
= F(^n + l,-n;
from
when
[CHAP.
/Trinitv 1907)
is
equal to
(Murphy.)
tan^ ^^).
P {z).
for all values of
n and
for certain values of z,
Pn
{z)
can be represented by the integral (called
"(^
Laplace's first integral)
(A)
Proof applicable only
When
is
(^'
to the
a positive integer,
we
- 1)*
cos <^}" d^.
have, by 15"12,
G is any contour which encircles the point z counter-clockwise.

Take G to be the circle with centre z and radius 2^ 1 *, so that, on G,
{z- 1 )^ e**, where
t ~ z
may be taken to increase from tt to tt.
where
-\-
* Mecanique Celeste, Livre xi. Ch. 2.

For the contour employed in this section, and
some others introduced later in the chapter, we are indebted to Mr J. Hodgkinaon.
for
LEGENDRE FUNCTIONS
15-23]
Making the
= ir-
{^
f"
_^
Ztt J
1
=-
Pn {z)
(^--l)*cos<f)}c?<f>
iz"- 1)5 COS 4)Y #'
an even function of
is
The
<^.
choice of the branch of
matter of indifference.
1)- is obviously a
Legendre functions, where n
to the
is um-estricted.
the same substitution as in (A) in Schlafli's integral defining
however, necessary in addition to verify that
it is,
contour and ^ =
the branch of
The
{z^
Proof applicable
z,
[z
the two- valued function
Make
have, for nil values of
'
f"
since the integrand
(B)
we
substitution,
313
[z
outside
+ {z"^
and
it,
which
1)^ cos ^}",
conditions that
\,
= \
now
is
inside the
1 is
we should
also necessary that
it is
specify
a many- valued function of
should be inside and outside
<^.
re-
spectively are that the distances of z from these points should be less and
These conditions are both
greater than \z^l\^.
R {z) > 0,
which gives
|arg^|<
and
satisfied if
2:
|
so (giving arg z its principal value)
1 <
1
+1
we must have
^TT.
P {z) = ^
Ztt
Therefore
where the value of arg
{2^ -I-
["
[z
(^^
l)i cos
(j>Y
(^<l>>
J _,r
(^'^ 1)2 cos
arg(^
<f)}
is
specified
by the
as
(f)
arg(i'-
1)
increases from
to
that
fact
numerically less than

on the real axis and on the right of z (see 15 "2).
[being equal to
Now
^^
|
z)] is
tt
when
it
is
+ {z'^-1)^ cos describes a straight line in the

z + {z^ l)~ and back again; and since this line
cos0} does not change by so much as

[2 +
tt, 2
Argand diagram going from z-{z'- 1)2 to

does not pass through the origin*, arg
77
Now
it
l)'-^
(s"-
on the range of integration.
suppose that the branch of
reduces to
z" e'""-'"^^
(where
is
{z
+ {z^-
cos (^|" which has to be taken
l)'-*
an integer) when
(^
is
such that
= ^7r.
Jnkni -
Then
P,. {z)
2.TT
{z
+ (z^ - 1 )*
where now that branch of the many-valued function
Now make
2--l
b}'
And
* It
so
we
get
only does so
<^}" d4>,
is
taken which
a path which avoids the zeros of Pn{z)
integral are analytic functions of z
path.
cos
J -TT
(r'^^'''
if ^ is
',
is
since Pn{z)
when arg 2 <hr, k does not change

|
equal to s"
as
when
and the
describes the
= 1.
a pure imaginary; and such values of z have been excluded.
314
Therefore,
when arg
^n(^)
where arg
[z
+ (z- -
1)- cos
This expression
<
2^
- If
COS <1>Y
d<f>,
{z-
when
(f>
^tt.
may, again, obviously be written
Avhich
[z
{Z'
to be taken equal to arg z
cp] is
XV
unrestricted,
is
[Z
^j[
Pn (z),
for
and n
- tt
[cHAP.
1)* cos
d<p,
</)]"
TT J
is
known
as Laplace's first integral for P(^).
From
Corollary.
15'22 corollary, it is
evident that,
when arg 2 <i7r,

|
dcj)
^cos^}"+i'
a
result,
due to Jacobi, Journal
second integral for P
Example
fiir
Obtain Laplace's
1.
first
2 A"
,1
and
using 6-21
Example
Math. xxvi. (1843), pp. 81-87, known as Laplace's
(z).
example
integral
{z
j
by considering
+ {z--\)^
cos
(^}" c?^,
1)
1.
Shew, by direct differentiation, that Laplace's integral
2.
is
a solution of
Legendre's equation.
Example
If s
3.
< 1,
<1
and
(l-2/icos(9 + A2)-= 2 bncosnd,

11=0
shew that
Example
^=
When 2>1,
4.
-^T- jo
^^"^^*-)
(TT^TO^^P?^-"-
deduce Laplace's second integral from his
first integral
by
the substitution
{S- (22 -1)5 cos ^}
Example
values of
5.
By expanding
in
Example
powers of
6.
{z
l)*COS(^}
cos<^,
= l.
shew that
+ (z'--l)hcoscl)]"d(t) = z^F{-hi,i,-in;
-
for
a certain range of
Shew
1; I-2-2).
that Legendre's equation

r
+ {Z^-
z,
^ 7I"
where
IZ
is
00

1
p} -hi
i + hi
[^+hi
-hi
M,
J
= ^(1^+^"^).
15 -231.
The Mehler-Dirichlet integral* for P (2).
Another expression
for the
Legendre function as a definite integral
may be
obtained in
the following way

* Dirichlet,
Journal fur Math.
xvii. (1837), p.
35; Mehler, Math. Ann.
v. (1872), p. 141.
LEGENDRE FUNCTIONS
15 "231]
For
all
values of
we
/*,
have, by the preceding theorem,
Pn{z) = - 1^
TT
+ COS
{Z
P (3) = -
^ = |7r, and
Now
let z
= cos d
Now
{6 being restricted
through A =
1,
1)^ } dcp.
arg
line,
l)^ sin
)Ti
COS
k,
/i
is
(cos
.so
6)^'
determined by the fact that k = z when
(1
- ihz + /i2)
<6 <\it
that -^tt
when n
is
not a positive integer) the
and joins the points A = e~*^ A = e'*, since the integrand
h=e^ we
its
analytic throughout
Jt+i)*
[e
-d,
-(2cos</)--2cos^)*
.'
Pn (cos 6) = -
'
i
^.'0 {2(cos0-cos^)P
easy to see that the positive value of the square root
known
is
chord t.
get
and so
is
hdh.
be deformed* into that arc of the circle |A| = 1 which passes
may
This
(f).
A"
all
dh
P(cos^)=
it is
(j),
{l-2kz + ffi) '
h"-
- i {z- -
the region between this arc and
Writing
then
Pn
path of integration
a straight
is
{l--2hz + h'^)~^=
(22
by a new variable
A = 2 + (s2 -
(i>
In this integral, replace the variable
and we get
315
^^
is to
>
be taken.
as Mehler's simplified form of Dirichlefs integral.
The
result is valid for
values of n.
Example
1.
Prove that, when a
is
a positive integer,
/""
\ (cos 6) = (Write 7r-6 for 6 and
Exa'mple
2.
-^
tt
for
(^
sin( + ^)0o?(/)
"'-"--!
=
{2(cos^-cos</>)}^
in the result just obtained.)
Prove that
(cos ^)
= -,
27ri./
A"
T ^^5
(A2_2Acos^ + l)4
the integral being taken along a closed path which encircles the two points
a suitable meaning being assigned to the radical.
* If e be
difficulties.
complex and
The reader
+ The integrand
is
h=e^
and
R (cos ^) >
will easily
the deformation of the contour presents slightly greater

modify the analysis given to cover this case.
not analytic at the ends of the arc but behaves
like
(ft-e*'^)~^ near
them but if the region be indented ( 6-'23) at e*'^ and the radii of the indentations be
tend to zero, we see that the deformation is legitimate.
;
made
to
316
Hence
prove that,
(or otherwise)
,,
(cos ^)
if
lie
=-
?r-T
77.
+TT^
3.5...(27i
77
between
2(2?i
+ 3)
(2 sin ^)^
12.32
cos(%^ + 5(^)
2.4.(2n + 3)(2?i+5)
(^smO)^
|-
denotes \d jtt.
<^
this
an approximate value of P (cos
also that the first few terms of the series give
Shew
may
theorem
6)
which are not nearly equal to either or it. And explain

be used to approximate to the roots of the equation P^ (cos ^) = 0.
and
between
for all values of 6
how
cosM + 3(^)\
^2gij^^yj
'
Itt,
1^
r^ H
^^
l)
where
and
Jtt
ico&(n6 + d>)
2.4...2
XV
[CHAP.
n-
(See Heine, Kugelfunktionen,
178
p.
i.
Darboux, Comptes Rendus, Lxxxii. (1876),
pp. 365, 404.)
Legendre functions of
15"3.
We
second kind.
have hitherto considered only one solution of Legendre's equation,
We
namely P(^).
We
tite
have seen
proceed to find a second solution.
( 15"2)
that Legendre's equation
j{f^-ir{t-zy
is satisfied
by
dt,
taken round any contour such that the integrand returns to its initial value
be a figure-of-eight contour formed in the following
Let
it.
way let z be not a real number between + 1 draw an ellipse in the i-plane
after describing
with the points
Let
outside.
of
Let
+ l)
real axis
(cf.
start from
let
= arg(i-l) =
is
and describe the
circuits (1
1 +),
arg (z
t)\> arg ^
as
>
on the contour.
Let
at A.
solution of Legendre's equation valid in the plane (cut along the
from
1 to
00 ) is
is
2^
12-43), and lying wholly inside the ellipse.
arg zlîr and
Then a
1.
returning to
if
be the end of the major axis of the ellipse on the right
Let the contour
arg(i
the ellipse being so small that the point
1 as foci,
it'
1)"
not an integer.
When
i^
(?^
-I-
1)
>
0,
we may deform the path
of integration as in 12-43,
and get
(where arg(l
arg(l
-|-^)
0); this will be taken as the definition of
Qn (z) when n is a positive integer or zero. When n is a negative integer

(= m 1) Legendre's differential equation for functions of degree n is
identical with that for functions of degree m, and accordingly we shall take
the two fundamental solutions to be
Qn (z)
is
P.^ (z),
Q,n
{z).
called the Legendre function of degree n of the second kind.
LEGENDRE FUNCTIONS
15-3, 15-31]
Expansion of Qn (z) as a
15'31.
We now
j)ower-series.
proceed to express the Legendre function of the second kind as
a power-series in
We
317
z~'^.
when
have,
the real part of n
>
z\
\
1 is positive,
Suppose that
Then the integrand can be expanded
1.
uniformly convergent with regard to
t,
in a series
so that
n
dt
dt
where r =
the integrals arising from odd values of
2^,
Writing
t-
= u, we
get without difficulty, from
The proof given above

(see 4'5)
applies only
when the
vanishing.
12'4'1,
real part of (n
1) is positive
but a similar process can be applied to the integral
Qn {Z)
-. .^
4i sin UTT J\
D^l^
(f-
the coefficients being evaluated by writing
- t)^
1) {Z
{t-
1)"
f dt
dt,
in the form
J JD
/(-! + )
/(1-)
Jo
Jo
and then, writing
f-
and
using 12'43, the
same
result
is
reached, so
that the formula

TT^
is
all
r(/i
1^/1
l)
,11
1^
true for unrestricted values of n (negative integer values excepted) and for
values* of
Example
z,
1.
When
such that
Shew
is
that,
2:
>
when n
1,
is
a positive integer
arg z\
<
tt.
a positive integer,
it is
unnecessary to restrict the value of arg
z.
318
[It is easily verified that Legendre's
by
Two
independent solutions of this equation are found to be
and
(32-1)"
It follows that
|(2-^
1)
j^ (v^-
(i;2-l)--irft;.
(s2-l)/'
XV
equation can be derived from the equation
n times and writmg = ^^
diflferentiating
[CHAP.
1)
-1 dv^
As this expression, when expanded in ascending

is a solution of Legendre's equation.
2--i, it must be a constant multiple* of Q (z)
powers of z~i, commences with a term in
^-''-i
in this expression with the coefficient of s~"~' in
and on comparing the coetficient of
obtain the required result.]
above,
we
found
as
of
the expansion
Q^ (2),
Example
2.
Shew
that,
when
n is a positive integer, the
Legendre function of the
second kind can be expressed by the formula
Example
3.
Shew
when %
that,
t=0
is
.[11
a positive integer,
I)
J z
[This result can be obtained by applying the general integration -theorem
to the preceding result.]
15'32.
The recurrence- formulae for Qn{^)-
P
The
same form, namely
functions
(2)
and
have been defined by means of integrals of precisely the
Q,^ {z)
[{t--iy'{t-z)-''-^dt,
taken round different contours.
is
15-21,
It follows that the general proof of the recurrence-formulae for P{z), given in
and hence that the Legendre function of the
(z)
applicable to the function
Qn
equally
second kind
satisfies the
recurrence-formulae
(w -H 1) $ + 1
(2)
- (2 + 1 ) zQn {z) + n Qn-l
(z)
= 0,
zQ'n{z)-Q'n-l{^) = nQ{z),
Q'n^l{z)-Qn~l{^)=={2ri+1)Q,,{Z),
{z'
Example
1.
-l)Q'n (2) = nzQ (2) - n$_i
(2).
Shew that
$o(2)
and deduce that
= 41ogj4|>
Q>
P (2) contains
(2)
i(^) = S^logJ^-l,
= ^^'2 (2) log
positive
powers of
- h,
when n
is
an integer.
LEGENDRE FUNCTIONS
15*32, 15*33]
Example
Shew by the recurrence-formulae
2.
^1/
where /_i
hPn
(s)
log
powers of
in descending
when
a positive integer*,
is
i
(and zero) powers of
consists of the positive
(z)
z+1
21
that,
319
in
the expansion of
z.
[This example shews the nature of the singulai'ities of Q^ (^) at 1, when n is an integer,
which make the cut from 1 to +1 necessary. For the connexion of the result with
the theory of continued fractions, see Gauss,
Journal fur Math. Lxxiii. (1871),
Werke, in.
and Frobenius,
pp. 165-206,
p. 16.]
The Laplacian integral f for Legendre functions of the second kind.
15 33.
now be proved
It will
Qn (^)
R {n+ 1)> 0,
when
that,
"
[2
(2'
1)* cosh 6>;--i de,
Jo
where arg
\z
+ {z-
1)* cosh 6] has its principal value
when 6 =
0, if
n be not
an integer.
First suppose that
z>\.
so that the range (
of real values of
1, 1)
lo*3, viz.
,^^^^+l)J-(-l)J^
write
In the integral of
of real values of
corresponds to the range (
It then follows (as in 15*23
6.
oo
oo )
A) by straightforward
substitution that
Q^{z)
= \\
[z^
{z
{z^-
1 )^
cosh (9}-"-
+ {z"--\)^cos\ie]-''-'de,
Jo
since the integrand
is
an even function of
0.
To prove the result for values of z not comprised in the range of real values greater
1, we observe that the branch points of the integrand, qua function of z, are at the
than
1 and at points where

which z= coth d.
points
Hence Q^
(z)
and
{z+
{z^
+ {z^-l)^
1)-
cosh^ vanishes; the
cosh B}-''-^d$ are botli analytic + at
all
points of the
Jo
plane when cut along the line joining the points

* If
latter are the points at
z= I.
-1<3<1,
it is apparent from these formulae that Q{z + Oi) - Q(z-Oi)= -7riP(z).

convenient to define Qn(z) for such values of z to be J(?(^ + Oi) + JQ(0-Oi). The
reader will observe that this function satisfies Legendre's equation for real values of z.
It
t
:J:
is
This formula was

It is
first given by Heine

see his Kugelfunktionen, p. 147.
easy to shew that the integral has a unique derivate in the cut plane.
;
320
By
XV
[CHAP.
the theory of analytic continuation the equation proved for positive vahies of
values of z in the cut plane, provided that arg {z+(z^- l)^cosh^}
persists for all
a suitable value, namely that one which reduces to zero when
The integrand
integer; but Arg
is
{z
one-valued in the cut plane [and so
is
{z
to
Obtain this result for complex values of
when n
is
a positive
and therefore
z= oc
if
has
n be
and the cut when n
+ (2^ - 1 )^} = 2 arg {{z + 1 )^ + (2 - 1)-)
is positive.
Qn{^)]
2= 1
These cuts give the necessary limitations on the value of

an integer ensures that arg
1.
{z^- l)2cosh^} increases by 27r as arg 2 does so,
not a positive integer, a further cut has to be made from
Example
given
is
is
not
principal value.
its
by taking the path of
making the substitution
integration to be a certain circular arc before
^_/ (2 + 1)^ -(2-1)^

/ (2+1)* + (2-1)^'
where 6
is real.
Example
Shew
2.
that, if 2
Qn
>1
(2)
and coth a = z,
{2
- 1)*
(2-
cosh
u}"-
du,
where arg
{z
-{z^-\ )5 cosh
u]
= 0.
(Trinity, 1893.)
Neumann's* formula for
15*34.
When
function Qn
is
Qn{^)i
a positive integer, and
when n
is
an
integer.
not a real
2 is
number between
expressed in terms of the Legendre function of the
(2) is
first
and 1, the
kind by the
relation
which we
When
shall
2
I
>
now
1
establish.
we can expand the
integi-and in the uniformly conve|'gent series
wj=0 2
Consequently
^-y
^y _i
The
integrals for
which m n
^ J -\
is
J -1
^ TO=o
odd or negative vanish
^ m=o
^1
_.^,_,^_,
I
2,=o'
<0n
(2ri
by
15-31.
The theorem
is
(^
( 15-211)
and so
-I
2-^^n + 27n) (n + m)
ni\{2n + 2m+l)l
'.
!N2
^^2i +
+ l)!
o, .,?i-t-i,
+ .,,*
thus established for the case in which |2|>1.
Since each
side of the equation
represents an analytic function, even
not a real number between

true ( 5*5) for
all
values of
* F.
and
when
4- 1, it
is
not greater than unity, provided that
2.
Neumann, Journal
2 is
follows that, with this exception, the result
fur Mdth. xxxvn. (1818), p. 24.
is-
LEGENDRE FUNCTIONS
15-34, 15 -4]
The reader should
321
Neumann's formula apparently expresses Qn{z)
notice that
as a
whereas it is known to be many- valued ( 15-32 example 2).

The reason for the apparent discrepancy is that Neumann's formula has been established
when the z plane is cut from - 1 to +\, and Q^ {z) is one-valued in the cut plane.
one- valued function of
Example
values of
2,
Shew
1.
Q^
Example
(2)
[For,
(1
when A
|
when -
that,
that,
2A2
-|-
when
fi')
~ ^
is sufficiently
i2 (2)
^ 1,
does not exceed the larger of
Shew
2.
the expansion of
z,
ai-c
is
2-
(z)
j
1 |~i,
7(2) |-'
2-I-
a positive integer, Qn{z)
cosh
is
We
the coefficient of h" in
small,
This result has been investigated by Heine, Kugelfunktionen,

Journal de Math. (3), I. p. 373.]
Heine s* development of
nomials in
for other
= (1 - 2A2+A2) - i arc cosh 1-^--,
15*4.
and that
1 |~'.
{t
p.
I.
134,
and Laurent,
as a series of Legendre poly-
z)~^
z.
now
shall
obtain an expansion which will
serve as the basis of
a general class of expansions involving Legendre polynomials.
The reader
by induction from the recurrence-formulae
will readily prove
(2m +
{2m +
1) tq,
(0
[) zP, {z)
- ("i +
- {m +
- mQ,,_, (t) = 0,
ml\_, {z) = 0,
1) P,+i {z)
1) Q,+i {t)
that
-- =
I (2m +
Using Laplace's
P,, {Z) Qrn
1 )
integrals,
(t) 4-
"
^
I
[Pn^, {z)
Q (t) ~ P (z) Q^.
(t)}.
we have
PnÂz)Qn{t)-Pn{z)Qn+x{t)
_\ p
TT j
[z-îz-^- l)^C0S</)j
1^ -I- (t-
-if- COsh
X [2
+ (^-
I
r*row consider
Let cosh
z
a,
and
+ (2^
1)'^
COS
This
is
a
I
(^
\t-\- (t-
)-
cosh |~^] d(f>du.
+ {z^-lfcos(p
j-
+ {t^-l)^coshu
foci
+ 1 which
pass
cosh
(a
+ 10),
^1 = cosh (a
id) sinh {a + id) cos

= {cosh- a sin^ 9 + (cosh^ a cos^ B) cos'-^ ^ 4 sinh
1
-f-
(f)
for real values of
cos
|2
<^
when
cos
0= + 1
a cosh a cos
<^}2
and hence
^ 2 cosh2 a -I + 2 cosh a (cosh^ a - l)^=exp (2a).

\t + (t^- 1)2 cosh u ^ exp a.
\
* Journal
W. M. A.
COS
cosh a be the semi-major axes of the ellipses with

respectively.
Let
be the eccentric angle of z then
maximum
2 (2^ - 1)2
Similarly
)-
2
2
through
?<}"+'
fur Math.
XLir. (1851), p. 72.
21
322
[cHAP.
XV
Therefore
I
P+i
Qn (t) - Pa
(z)
{2) Qn+i (t)
TT-i
exp
{n (a
T f*
a)}
(2-
Ip
cos<f>
+ {^ + (^--l)^cosh u]
+ {f-l)^coshu
Pn+i (z) Qn (0 ~ Pn (^) Qn+i (0 > 0, as ?i -> 00 provided
\V\
rhere
Vd<f>du,
Therefore
And
further, if
the upper bound of
remaining constant,
varies, a
Vd<j)du
independent of
is
t,
is
it
a <
a.
easy to see that
and so
Jo Jo
Pn+^{z)Qn(t)-Pn{^)Qn+^(t)
tends to zero uniformly with regard to
Hence
the point
if the point z
and has
is in the interior
the points
for
r^^=l
z
valid;
and
{2n
ellipse
+ \)Pn{z)Qn{t)
Neiiinanns* exj)ansion of an arbitrary
proceed
now
such that z
the expansion converges uniformly with regard to
it,
fmotion
We
which passes through
then the expansion
if the a variable point on an ellipse with foci
a fixed point inside

15"41.
of the
its foci,
=o
is
t.
in
The expansion
is
of special interest, as
among expansions
it
stands next
in series of poljnaomials.
Let f{2) be any function which is analj^tic inside and on an

whose foci are the points z^ \. We shall shew that
where
a,,,
= a,Po (z) +
Oj, a^, ...
ttiA
{z)
are independent of
points z in the interior of
Let
aoPo {2)
2,
this
ttaPa
ellipse
C,
{2)+...,
expansion being valid
for all
the ellipse C.
be any point on the circumference of the
Then, since
series of
to discuss the expansion of a function in a series of
in simplicity to Taylor's series,
f{2)
is
t.
S (2n+ l)Pn(z)
ellipse.
Qn(t) converges uniformly with regard to
t,
w=0
/(^>
= 2^-
l/t- f - 2^-
+
!.
.L<2"
1>
^"
<^>
" (')'">
"
= %
anPn(z),
w=0
where
* K.
1862).
2n
^
1
.
f
I
f(t) Qn
(t) dt.
Neumann, Ueher die Entwickeluny einer Funktion nach den Kugelfunktionen (Halle,
Thomd, Journal fur Math. lxvi. (1866), pp. 337-343. Neumann also gives an ex-
See also
pansion, in Legendre functions of both kinds, valid in the annulus bounded by two ellipses.
LEGENDRE FUNCTIONS
15*41, 15-5]
323
00
This
the required expansion
is
since
to converge uniformly with regard to z
wholly inside
C, the
Another form
when
Pn{z) Qn(t) may be proved*
any domain C" lying
z lies in
expansion converges uniformly throughout C.
for a
can therefore be obtained by integrating, as in
15-211, so that
an
+ 1)
{2n
('^
form of this equation which
is
/(^^ ^ (^^
2) j
frequently useful
^^
is
"'^Îrli/V*'''^"^^-^^""^'^"^'^'
which
Pn
obtained by substituting for
is
from Rodrigues' formula and
(^)
integrating by parts.
The theorem
theorem
bears the
Avhich
Let fit) he defined
'ivhen
ahsohitely convergent ; also
1 ^^^
Then 2aPn{a;)
1< ar <
For a
is
convergent
if
1,
1,
1.
and has
any condition of
Shey
as follows
and
= {n-\-h)
integral of
let the
that, if p
(^
1)
2.
z=(^^\\
If
the su?n
t-y^f{t)
i{f ic+0) +
'
3.
Shew
he
f{jc
- 0)} at any point

9 "43 is satisfied.
x,
for
end of
memoirs by Hobsont and Burkhardt J.
/[2
= 5''^~ JJ
^
fj^l)
= {(a: + 1
(y
(p+p~'), \
where
- 1 ))i
(p
p"^).
y>x>\,
i P
(.*:)
Q,, (y).
[Substitute Laplace's integrals on the right and integrate with regard to
Example
and
exist
be the radius of convergence of the series 2cs", then
/"
prove that
{\
f{t)P{t)dt.
2cP(2) converges inside an ellipse whose semi-axes are
Example
Fourier's
the type stated at the
proof, the reader is referred to
Example
Neumann's expansion as
relation to
is
let
which
same
the expansion of 9'11
beai's to
(^.]
that
2V-) '"^ lEI)[vi-!l = Jo'^^ '> *"
<^'-
(Frobenius, Journal fur Math. LXXiil. (1871),
p. 1.)
Ferrers associated Leg endre functions Pn^{z) and Qnî^)-
15'5.
We shall now introduce a more extended class of Legendre functions.

If m be a positive integer and 1 < ^ < 1, n being unrestricted ,
the
functions
P- (Z) = (1 - Zf^- ^"^"1

*
The proof is similar

London Math.
t Proc.
to
Q^rn
,
(,)
(1
,.)i^
the proof in 15-4 that that convergence
Soc.
(2), vi.
(1908), pp. 388-395;
X Miinchener Sitzungsberichte, xxxix, (1909), No. 10.

See p. 317, footnote. Ferrers writes r' {z) for P^"'
(2), vii.
is
d^"Qn(^)
uniform with regard
(1909), pp. 24-39.
(z).
21
to
t.
[CHAP.
324
be called Ferrers' associated Legendre functions of degi-ee

of the first and second kinds respectively.
71
will
XV
and order
on
It may be shewn that these functions satisfy a differential equation

analogous to Legendre's equation.
For, differentiate Legendre's equation
(1
^
111
Write
- ^^) ^, -
= (1 2'-)^'"
lu
(1
This
From
fiinctious
ox
-^-^^
2^
2^
dz-
-^
times and write v for
(1
^_
^2)
We
(^^^
1)
+
^
dz
1)
obtain the equation
^+
('i
'>0 ('^
1 X
m^
) d.--
dw
2^
s + r ( + 1) - 1^}
/
P,,"* (z)
the definitions given above, several expressions for
may
Schlafli's
î
A
1.
when n
is
in (2)
Shew that Legendre's

/
and
tlie
Qj'' (z).
associated
Legendre
'
associated equation
oc
im
n+1
Ui
- im
- hn
The integral pro'perties of
is
defined
l-u\-
(Olbricht.)
Legendre functions.
the associated
n, r, m, are positive integers
by the scheme
f)
15-14
generalisation of the theorem of
When
= 0-
a positive integer, we have, by Rodrigues' formula,
15"51.
= 0.
formula we have
where the contour does not enclose the point ^=
Example.
1) ^
be obtained.
J^n-ri
Further,
and we get
^'.
d"W
the differential equation satisfied by
is
Thus, from
The
(n
,,
the following
is
and n > m,
> m, then
(ri^n),
.1
I
Pn'''(z)Pr"H2)dz\
J
2?i
.,
(n+ m)\
1 (n
m)l
To obtain the first result, multiply the differential equations

Pr"^ {z) by P/" (z), Pn"^ (z) respectively and subtract this gives
for P^"(z),.
dz
+ (n-
r)
(n
+r+
1)
P^ (^) Pn'" {2) = 0.
325
LEGENDRE FUNCTIONS
15-51, 15-6]
On
integrating between the limits
+1, the
1,
result follows
when n
in square brackets vanishes at each
and r are unequal, since the expression

limit.
To obtain the second

P"+^ {z)
we observe
result,
= {l-
-z"-)-^ P, - {z)
+ mz (1
-^^
z-f
that
squaring and integrating, we get
^^'"(^)r\ o-..z.,./..^^"(^)
(l-^^)|'^'^^^HV27n^P-(^)
j'j^P,r^^(z)Ydz=j'^
dz
.'
on integrating the
now
first
two terms
in the first line
12^-
on the right by
parts.
If
use the differential equation for Pn^" (z) to simplify the
integral in the second line, we at once get
Ave
{P'"+^ (z)\'
dz
(n
- m)
(71
m + 1)
first
[Pn'" i^)]' dz.
By
repeated applications of this result we get

{P"* {z)]-
1
.'
dz
= (n - m +l)(n- m +
2)
??
-1
X {n
+ m) (n + vi-l)...
(n
and
f-1 |P.-(.);.&= 2?l +
so
15"6.
So
Hohsons
far it
{P (z)Y dz,
I)
-1
<i!!i^|.
1
definition of the associated
{n
7/t):
Legendre functions.
has been taken for granted that the function
(1 2^)^'" which
and
is purely real
more elementary physical applications of Legendre functions, it
usually happens that 1<2^<1, no complications arise.
But as we wish
to consider the associated functions as functions of a complex variable, it is
undesirable to introduce an additional cut in the 2^-plane by giving arg(l z)
occurs in Ferrers' definition of the associated functions
since, in the
its principal value.
we
when z
Hobson in defining
a real number such that
\<z<
Accoi'dingly, in futmê,
is not
shall follow
the associated functions by the equations
p- (.) =
(.'
i)i' ''"'j;;^^)
where ni is a positive integer, n

have their principal values.
is
Q.r
(.)
iz'-
unrestricted and arg
i)i*
z,
1,
^^>
arg (^
+ 1),
arg (^
1)
326
When
unrestricted,
is
111
may
definition of '"
{z)
2 + ir(7i
sinWTT
to be
from which the formula
sm( + m) 7f T(n+m + \)T{\)
m
O
^" ^>
XV
^(-...+i;i-.;i-J^);
f(T3;^)(^)
and Barnes has given a
Hobson
defined by
P;i'" (s) is
[cHAP.
(^^-1)^"'
^n + m + l
+ f)
be obtained.
Throughout
work we
shall take
to be a positive integer.
Expression of Pn^^{z) as an integral of Laplace
15"61.
If
this
we make the necessary
modification in the Schlafli integral of 15"5,
in accordance with the definition of 15'6,
p, .
(,)
we have
= (n + !)(' +^2)^;.. ( + -0 (,, _ 1 )J
Write
where a
=^+
(2-
1)2
the value of
is
e'*,
s type.
as in 15-23
^ when Hs
arg
{z-
at
--(,,
_ 1).. (, _ ,)__-. rf.
then
so that
1)-
|-
,,.
<
TT.
Now,
as in 15'23, the integrand is a one- valued periodic function of the
real variable
Since
^ with
{2^
(^^^
period
and
27r,
1)^ cos^]"
so
an even function of
is
the range of integration into the parts (PrJ^ {z)
'
[z
I
Jo
TT
The ranges
n
is
or
is
of validity of this formula,
Shew
Pn"'{^)
that, if
= (-)"
arg z
+ {z^ yf
which
is
we
get,
n{n
on dividing
(0, tt),
cos ^j'* cos m<\>
d<f).
due to Heine, (according as
15 "23.
< ^tt,
I) ...
{n-m-+ !) f^
cosm(f)d(f)
'
Jo
where the many-valued functions are
{3
+ (32_i)icos<^} + i'
specified as in 15 23.
The addition theorem for the Legendre polynomials*.
Let z=xaf -{x^ - 1)^ (x'^*
</>,
not an integer) are precisely those of the formula of
Example.
15 "7.
0) and
tt,
Legendre, Calc. Int.
11.
Vp cosco, where
pp. 262-269.
reasoning will be given subsequently
An
18 "4).
x, x',
a are unrestricted complex numbers.
investigation of the theorem based on physical
15*61,
LEGENDRE FUNCTIONS
157]
Then we
First let
shew that
shall
Pn
= Pn {X) Pn {x') + 2
{Z)
R (of) > 0,
< <^ < 27r.
If
M be
2
m=i
- )' ^^""^^ PrT (^) ?/" (^') COS r.m.

{n + ni)
\
^+(-^ "
so that
af
range
327
<-OM<-9J
j^^
+ {x''^-l)^cos(f>
upper bound and
its
if
^ bounded function of
in the
< i/""',
then
^{.r + ( .r2-l)^cos( a,-0)}
converges uniformly with regard to 0, and so ( 4"7)

{^ +
''
A'^
=0
f
j-r
^'-1)^ cos (0,-0 )}"^^^
{y + (y2_l)^COS0} +
/""^
-'^
-^
Now, by a
p-n
example
of 6-21,
/""
(;ar:'2
1)^ COS 0}" +
-n A
+ B COS + C sin (^
follows that
(A-- B^ C'^)^'
({}
radic;il is
it
2v
d(f)
where that value of the
{.r'
^^^
y + (j7'2- l)i COH(f>-h {x+{x--'l)^COS{<0-(f>)}
slight modification of
A"{:c + (^2-l)^cos(co-(^)}"
n=i)
'
taken which makes
Therefore
+ (.r'2 - l)i cos (f)-h {x + {x2 - 1)^ cos ( - (j))}

2n
{{a/
- hxf - {{of' _
1 )i
-A
1 )i
cos
Po
(^')
by
(a;2
a)}2
- [h (^2 _
)i în a,}2ji
27r
'
(l-2>^0+A2)i
and when h~^0,
tliis
expression has to tend to
of h and equating coefficients,
Now
we
a polynomial of degree
/'(2) is
27r
j5
Expanding
15-23.
in
powers
get
ti
in
cos, and can C(jnsequently be expressed in
the form i.4o+ 2 J,cosjrtco, where the coefficients

to
determine them, we use Fourier's rule
A ,n =
'"'
/"
-n
I"
2-'''
/"'^
f""
{.V
are independent of w
and we get
+ {x^-l P cos (<a - </>)}" cos Wlm
r /"
{a;' +(a;'2- 1)4 cos (/>} +
{.r
"
on changing the order of

from +7r ^ to ir.
["^
{y +
(.r2
(a/2_l)4cos<^} +
- 1 )4 cos
\|/-}"
cos
7W<0
"]
>//'
-]
m (0 + >//),. 1
+ (a;'2-l)4cos0}+i
= + and
integration, writing
{^'
ij
+ (.^2 - 1)4 cos (o) - 0)} cos
i -T L j -T
{^ +
r
f
27r2J_;,Li-T
- J^
...
Pn (s) cos ma d(o
2r2J_Li-T
__1_
(^ 9"12),
Jj,
J^,,
,,
J
changing the limits for
>//
328
/"
Now
and
so,
i
{x + {x^ l)^
by
cos-yj/}"^ sinm-\j/d-\lr
= 0,
since the integrand
15"61,
[^
111
7r{n + m)l J
cos mcf)
-^
{,r'
(,f'2
P^ (x)
-1
cos
)3
is
XV
an odd function;
(^}
when arg / <^7r,
Therefore,
2 (-)'nL
P(^) = P(^0i*nG^') + 2
(% + m;
t=i
p'(.^')P(.^'')cosma>.
!
But this is a mere algebraical identity in x, x and cos <o

and so is true independently of the sign of R {x').
The
[CHAP.
(since n is a positive integer)
result stated has therefore been proved.
11
The corresponding theorem with

Pr,{xx'
+ {1
Ferrers' definition is
^
-X')h (1-^2)2 cos co}^P,,{x) P
ix')
+2
771^
( Jl
2
Pn'"
{li-i-jn)
m=i )-^~J.-\
{^) P.î-V') COS mco.
15 "71.
Let
^77
The addition theorem for
(^'1)^, (*''1)^ be given their princiiJal values
let
z=xx'
Then
Legendre functions.
be two constants, real or complex, whose arguments are numerically less than
x'
.r,
and
the
shall sheiv that, if
loe
arg
-{x'^
s
< ^tt
let
co
be real and
let
1)2 {x''^~ 1)2 cos W.
for all valves of the real variable
co,
and n
he
not a positive integer,
P
Let cosh
= P {X) P {X') +-2^^2^ - )- jT^^:;^) ^n'" (^') /^n'" (^') COS mo,.
{z)
a,
cosh
Let a + i^ = ^,
Now
as
G)
a'
be the semi-major axes of the ellipses with foci +1 passing through

^, jS' be the eccentric angles of x, x' on these ellipses so that
Let
X, x' respectively.
= cosh^, .r'=:Cosh ^'.

R (z) oscillates between
- 1 2 = cosh (a a') cos (/3 ^'),
R (xx') R{x^-l)i
a'
i/3'
= ^',
passes through
so that
.*'
all real
values,
(a''2
so that
it is
Now
necessary that ^/3' he acute angles positive or negative.
take Schlafli's integral
and write
_e^ {e~"" sinh ^ cosh |g' - cosh | sinh \^'] + cosh g cosh l^ -e"" sinh ^ sinh ^g'
'
sinh ^^'
cosh ig'+e
The path
of
?:,
increases from
as
tt
to
tt,
may
be shewn to be a circle
and the
reader will verify that
^~
_ 2 {e' ^*
_2
{i
^"^
~"^
cosh Jg
+ s inh^g} {sinh igcoshjf
"
"
^^
cosh ig' + e
" "^
id,
siuh ig + cos h ^g} {cosh
cosh Jg' + e
_ {e^ cosh ^g'
-|-
sinh ^g'}
{e'"
if}
'
sinh ig'
igco^|g' - / "
sinh ^g sinh i|'}
sinh |g'
sinh g sinh^ ig'
cosh
sinh
e'" cosh |g
'
ig'
+e
*"
sinh g cosh^ ig'
+ e'* sinh ig'
- cosh
g sinlij'}
LEGENDRE FUNCTIONS
1571, 15 -8]
Since*
cosh ^^'
by
increases
>
i|'
sinh
,
|
329
the argument of the denominators does not change
when
for similar reasons, the
arguments of the first and third numerators

increase by 27r, and the argument of the second does not change; therefore the circle
contains the points i = l, t = z, and not t= - 1, so it is a possible contour.
<^
27r
Making these substitutions
it is
readily found that
J_
{x
+ (x^-l)^ cos{a>-(t>)}''
2n J -^ {y (^'2-l)icos 0}" +
+
and the
work
rest of the
#,
follows the course of 15-7 except that the general form of
Fourier's theorem has to be employed.
Shew
Example.
{^^'
Qn
when
n be a positive integer,
if
+ (x^-l )^ (.r'2 - l)i cos
is real,
(o
that,
(.*')
^ 0,
and
a,J
{x
= $ (x) /> {x') +
1)
{x+ l)\<\ (x-
(Heiiie, Kiigelfunktionen
15-8.
Tlie furictioni
'/
-2
1)
$'" (x)
(^ + 1)
P- ' {x)
cos m<o,
|.
K. Neumann, Leipziger Abh. 1886.)
{z).
function connected with the associated Legeudre function Pn'" (z) is the function
^n" (2), which for integral values of n is defined to be the coefficient of /< in the expansion
of (1 2kz + h^)~'' in ascending powers of /t.
It is easily seen that C,," (2) satisfies the dift'erential
c?2y
(2i/
For
is
the contour of
The reader
When
(I)
n(H + 2v)
dz
z^-\
^>
will easily
P (s) = C -
(II)
is
v;
1 ")"2
{t-zY^^
this corresponds to Schlafli's integral.
prove the following results
When
(s),
r is
Rodrigues' formula
C*"^^
,z\=
'
(z)
n-r^
last
is
a_,^^h--dl.^^_î^n+.-h.
+ 2vy
'
dz'^
'
^'
a particular case of this result.
an integer,
n-,'^
whence
an integer
c*"^-'
The
'
"'''
jc
{-2Yv{v + \)...{v+ n-\)

C''(z)=
" ^'
n\{2n + 2v-\){2n + 2v-2)...{n
since
_
y
i/, it may l)e shewn

that we can define a function, satisfying
by a contour integral of the form
^^
o?y
values of n and
all
this equation,
where
+ l) 2
z^-l
rf^"*"
equation
'
L,
(2r-l)(2r-3)...3.1
-^
~^^
(2r-l)(2/--3)...3.
^p
dz^
(,\
"^
''
Pj
(2).
"
^
equation gives the connexion between the functions C^
(2)
and
P/ (2).
This follows from the fact that cos/3'>0.
t This function has been studied by Gegenbauer, Wiener Sitzungsberichte, lxx. (1874), pp. 434443; Lxxv. (1877), pp. 891-896; xcvii. (1888), pp. 259-316; en. (1893), p. 942.
330
Modifications of the recurrence-formulae for /*
(III)
^^^ = 2^0" +
(4
>i67(2)
(2)
= (-H-2,/)2C"'
[CHAP.
are the following
XV
(2)-2,.(l-22)(7''-l(2).
REFERENCES.
A. M. Legendre, Calcul Integral,
11.
(Paris, 1817).
H. E. Heine* Handbuch der Kugelfimktionen (Berlin, 1878).

N. M. Ferrers, Spherical Harmonics (1877).
I.
L.
Todhunter, Functions of Laplace, Lame and
Bessel (1875).
ScHLAFLi, Ueber die zwei Heine^schen Kugelfunhtionen (Bern, 1881).
E.
W. HoBSON,
Phil. Trans, of the
E.
W. Barnes,
Quarterly Journal, xxxix. (1908), pp. 97-204.
Royal
R. Olbricht, Stiidien ueber die Kugel-
Acta Acad. Leop.
Society, 187
(1896), pp. 443-531.
und Cylinder funhtionen
(Halle, 1887).
[Nova
pp. 1-48.]
lii. (1888),
N. Nielsen, Theorie des fonctions metaspheriques
(Paris, 1911).
Miscellaneous Examples f.
1.
Prove that when n
is
a positive integer,
(Math. Trip. 1898.)

yi
2.
Prove that
I
is
zero unless rn
n= + 1, and
dP
z{\-z^)^
as
_i
determine
its
dz
value in these cases.
(Math. Trip. 1896.)

3.
(27t
Shew
(by induction or otherwise) that
when n
is
a positive integer,
+ l) ^P2(^)^, = l_^p^^2_2,(P,2 + p,2+... + p2^_,) + 2(PlPo + P2A + +^-l^)

(Math. Trip. 1899.)
4.
Shew
that
zP,:
(z)
= nP^{z) +
2n - 3) P_ 2
(z)
+ {2n - 7)
P _ 4
(2)
(Clare, 1906.)
5.
Shew
that
z'P"
where p = hi or
*
(z)
= nin-l) Pn{z)+
2 (2n-4'/-+l) {r{2n-2r + l)-2} P_2,
(Math. Trip. 1904.)
(n-l).
Before studying the Legendre function P
{z),
(z)
in this treatise, the reader should consult
Hobson's memoir, as some of Heine's work is incorrect.

t The functions involved in examples 1-30 are Legendre polynomials.
LEGENDRE FUNCTIONS
6.
Shew
331
that the Legendre polynomial satisfies the relation
{z^-\f^ = n{n-\){n + \){n + 2)
j^^dz
^Â^dz.
(Trin. Coll. Dublin.)
7.
Shew
that
n{n + \)
7 Pap
\ n
j^z
^ + i^^^^-i(^^^^-(2_l)(2ri
f
+ l)(2 + 3)(Peterhouse, 1905.)
Shew
8.
that the values of
a - z^f P^"
(i)
8'rt
when
m-n
is
for other values of
(iii)
Shew
{z)
dz are as follows
and even,
positive
-2(2_i)(7i-2)/(2n + l)
(ii)
9.
(n + 1)
P'
(z)
-1
.'
when
and
?/i=n,
(Peterhouse, 1907.)
n.
that
sin^/'(sin^;= I
/^
)*"
'
cos'-
gP^ (cos
^).
(Math. Trip. 1907.)
when n
is
/' (/x)
P (cos
6)
dd
( 15-1
example
'"
/I
parts, that
11.
"
Shew, by evaluating
10.
arc sni
/x
2),
and then integrating by
:,.
c/^x is
when n
zero
even and
is
a.3...(?i-2)]
is
equal to
tt
<^
odd.
If wi
and be positive
^,^
^,^
m^n,
and
integer.s,
/2?t
V ^lm-r^^r^-r
shew by induction that
+ 2 l - 4/-+
^1,,,=
r!
(Adams, Proc. Royal
By expanding
2(^
is
to
Soc. xxvil.)
powers of u shew that
in ascending
where
1.3.5...(2m-l)
where
12.
(nA-'W
(Clare, 1903.)
\n
ffn
be replaced by (1-2^) after the
diflFerentiation
has been performed.
13.
Shew that P (z) can be expressed as a constant multiple of a determinant in
which all elements parallel to the auxiliary diagonal are equal (i.e. all elements are equal
the determinant
for which the sum of the row-index and column-index is the same)
;
containing n rows, and
its
elements being
_1
3'
11
3^'
'5'
5^'""2-l'*
(Heun,
14.
Shew
that, if the
p (,.1 [-{zJl-J)^2Hl-z^)^}- dt.
By
writing cot
^'
Nach. 1881.)
path of integration passes above i=l,
^"^^"rrt jo
15.
Gott.
"
(Silva.)
(l-f2)"î
= cot d k cosec d
and expanding
sin
ff
in
powers of A by Taylor's
theorem, shew that
P (cos
6)
'
= ^-^
ft
cosec" "^
'
6 ^V^^î
d (cot 6)^
(Math. Trip.
1 893.
332
By
16.
The equation
small
is
shew that
1
London Math.
Soc. vi.)
of a nearly spherical surfixce of revolution is
r^l + a{Pi
where a
XV
considering S /i"P(0), shew that
(Glaisher, Proc.
17.
[CHAP.
+a
(cos 6)
+ Ps (cos d) + ... + P-2n-i (cos 0)},
be neglected the radius of curvature of the meridian
if a^
is
2 {n(4j+3)-(m + l)(8/H + 3)}P2m + i(cos<9).

(Math. Trip. 1894.)
The equation
18.
of a nearly spherical surface of revolution

r
where
e is
Shew
small.
that
if e^
be neglected,
its
area
is
+ 1,2 ^^^^ ^l
47ra2 ji
19.
is
= a {l+ePn i^ôs 6)},
Shew
that, if
is
(Trinity, 1894.)
an integer and
(l-2As + /i2)-^^= 2 anPni^),

then
where x and
?/
are to be replaced by unity after the differentiations have been performed.
(Routh, Proc. London Math. Soc. xxvi.)

20.
Shew that
^^ '^ w ^"-1
'
(^)
'^"-1
(^)
('y>
^-^'^
-^'
= 1.
21.
Let x^+9/'^+z^=r-,
= fir,
(Catalan.)
the numbers involved being real, so that
-l</x<l.
Shew that
/
A(m)
where r
is
\n
j.n
+1
gn.
/J~>
C2"
to be treated as a function of the independent variables
a;,
^, z in
performing
the differentiations.
22.
With the notation of the preceding example
(+i)p(m)+/-A/(/x)=
23.
Shew
that,
if
and
(of. p.
319, footnote *),
,^,
a7>(,-3J-
are sufficiently small,
1-^2
(1-2^2 + ^2)^
2 (2 + l)/i"P(2).
=0
shew that
LEGENDRE FUNCTIONS
333
Prove that
24.
(Math. Trip. 1894.)

If the arbitrary function /(.r)
25.
can be expanded in the series

X
f{x)= 2 anPn{x),
n=0
converging uniformly in a domain which inchides the point
.y
= l, sliew
that the expansion
of the integral of this function is
fmd.= -a,-\a,^V {^^ - ;^--- )

Determine the
26.
coefficients in
Neumann's expansion of
(Bauer.)
/..(...),
a series of Legendre
e"^ in
(Bauer, Journal fur Math, lvi.)
polynomials.
27.
TT
'-
(1
.3. 5... (2n-l))
{^....g.^.,, ]
arcsn.3-=-^ 2
.,
,,,
{/^.^. (^")-
A-
(.)}
(Catalan.)
Shew
28.
that
Qn{z) =
\\0g(^^^. P{z)-\Pa-x{z) P,(Z) + \P...,{Z)
P,{Z)
+ \Pn-z{^P->{z)+--^\P.{^Pn-l{z)^
(Schlafli; Herraite.)
Shew
29.
that
Pn.ve also that
=
f^..(^-.)(';r>
^*//. " _1
.
where
30.
^-
(i)
=i
/>.
W log ^^
Ci-X'-.-'-o
_1 _ 1\ ( ^-1)(^^-2)(
2
-/, _
3/
/^
''<"--'A<s-'->>-t^ (^-i-7]
+ l)(/i + 2)(n+ 3) /^ÎV "^"

,
122232
= 1 +^ + ^ + ... + 2
3
Shew
(2),
(Math. Trip. 1898.)
that the complete solution of Legendre's differential equation
the path of integration being the straight line which
is
when produced backwards passes
through the point ^=0.

*
The
(1858).
first
of these expressions for /_i(2)
was given by
Christoffel, Journal fiir
Math. lv.
334
Shew
31.
that
TO!r(m-a + l)
27r
Shew
that,
XV
+ (s2- 1)^=2 A$2m-a-i (2),
{s
where
32.
[chap.
(SchlaBi.
'
when R{)i+1)>0,
Qn {Z)=
dh,
-,
{\-%hz + h?f
kn
2r-(z--'-l)5
dh
and
(l-2As + ^2)5
Shew
33.
that
where the
is
(/i
+ 1)
is
+1
cosh mtt
/"
du,
(Hobson.)
greater than m.
Obtain the expansion of
34.
real part of
(?
when arg
/* {z)
2
|
<
as a series of powers of
tt
I/2,
when
not an integer, nameh'
IT
T{n + \)T{\y
[This
35.
is
is
most
Shew
easily obtained
2'
'
zy
'''
by the method
of
i^
14'51.]
that the differential equation for the associated Legendre function
/V"
defined by the schemes*

00
hn
00
-\n
^^^
1^- I
P\ -\n
\m
T^y
(Olbricht.
36.
Shew
that the difi'erential equation for
-1
\
n + 2v
37.
Prove that,
,,
^'
then
scheme
,-
if
(2?^ + l)(2n + 3)... (27^ + 2.-1) ^

n{n^-l){n^-4) ...{?i^-{s-iy}{n + s)^
.
2(2 + l)
2_iv^
'
dz
'
+3
^^î*,...^-^P + 2;^P-.
y3-^ + 3
and
i-
-n
defined by the
C,," {) is
271
3(2n + 3)
3J27i_+5)
2^^_^ ^+i+ 2^-_3 fu-x
.
(2v^
+ 3) (271 + 5)
(271- 1) (271-3)
"-"'
(Math. Trip. 1896.)
find the general formula.
See also
15*5
example.
(2)
LEGENDRE FUNCTIONS
Shew that
38.
335
/cos ^)
r(?t+wi +
f cos
l)
V - Am^
+ i) - ^TT + |??t7r }
<9
r(n + |)
Vtt
{{ii
(2 sin 0)5
2 (2%
+ |)<9-|7r+^7ft7r}
cos{(?t
+ 3)
(2sin0)^
-4m2) (32-4^2 ) cos{(n + f)0^f7r + î7r}

'
'^2.4.(2 + 3)(2n + 5)
(12
"1
J'
(2sin0)^
obtaining the ranges of values of m, n and 6 for which
it is
valid.
(Math. Trip. 1901.)
Shew that the values of

TT when m is positive
39.
from
values of d between
and
which Pn~"' (cos 6) vanishes, decrease as 6 increases

and that the number of real zeros of /*""' (cos 6) for
the greatest integer less than n m+l.
(Macdonald, Proc. London Math. Soc. xxxi, xxxiv.)
n, for
to
tt is
Obtain the formula
40.
rn-
- 2A{cosa)
[1
cos
+ sin
o)
cos(0'-
sin
0)! +/i-]
^
,
^dd=
2 /i"P (cos
&>)/' (cos 0).
n=o
2ir J -TT
(Legendre.)
H f{x) = x^{xÔ)
41.
Aud f(x)=x^{x<0), shew
f(x) can be expanded
that, if
into a uniformly convergent series of Legendre polâiomials in the range (-1,
expansion
1),
the
is
(Trinity, 1893.)
42.
7^
^T-TTTTT,
(1-2A^+ AT
If
=
(Z)
2 h"
":.. 0/
shew that
C,,"
{xxi
- (a-2 _
i)h
(a-j2
1)^
^ r(2v-i)
cos 0}
4^r(w-X4-i){r(i/+x)}2(2i/+2X-i)
r(n + 2,. + X)
{r(wp ;,io^~)""
X {x^-
1)*^ (.r,2- i)i^ c;+;; (^) (7;+;;
(Gegenbauer,
43.
If
where
cj is
o-,.
the least root of
fi
(2)
P'
('
(a;,)
c[-'^ (cos<^).
IFiewe?* Sitsimg/sberichte, cil. (1893), p. 942.)
-3tz + l)-h
Ztz + 1 = 0, shew
t'^dt,
that
(2 + l)o- + ,-3(2?i-l).'o-_,
+ 2(-l)o-_.^ = 0,
and
4 (4;3 _
1)
an" + 1442V" -
(1
2?i-'
where
(r
- 24? - 291
"
) a-,,'
- (n - 3)
(2 - 7) {2n + 5)
- 0,
etc.
(Pincherle, Rendiconti Lincei

44.
o-
(4), vii.
(1891), p. 74.)
(P-3/i2-fl)-^= 2 Rn{z)h'\
If
71=0
shew that
2 ( + l) /2 + i ~
3(2n+l)2 + (2?i-
1)
i?_2-0,
/?+/?'_., -2i?' = 0,
and
4 (4^3-1) R"'
,
where
+ 9Gz'iR,:'-z
{\2n^ + 24n-9l)
, d^R^
Rj," = -j-^
(Pincherle,
R,,'
-}i{2n
+ 3)
{2n + 9)R^=0,
etc.
Mem.
1st.
Bologna
(5),
i.
(1889), p. 337.)
336
45.
If
[CHAP.
XV
^(-)=2^r;nWr)^^'"'-')"^"-')>'
obtain the recurrence-formula

(
+ l)(2H-l).4(.r)-{(47i2_l)^+l},^,^_l(.r)-|-(-l)(2?^ + l)^_2(^') = 0.
(Schendel, Journal fur Math, lxxx.)
46.
If
11
not negative and
is
(2
-1
a positive integer, shew that the equation
is
(2'/i
^=m
+ 2) ^
((
+ 2/i +
1 ) 3/
has the two solutions
when X
47.
not a real
is
number such that
.?;
1.
Prove that
fClare, 1901.)
48.
F,,^{x)= I
If
m=0
shew that
where P
^"'"^'"-'^^
(*',
a)
is
C-^,
=e^P
(e+*')l
a polynomial of degree
Pn +
^~;^x-\
7i
in
.r
and deduce that
= (^' + <*) ^n (*'5
(^, a),
+*-''
d^
TT,
dx
(='^)
(Trinity, 1905.)
49.
If
Fn
(x)
be the coefficient of
s" iu
the expansion of
2kz
gflZ
in ascending powers of
Fo
g~
''3
so that
z,
{x)
=1
Fi{x) = X,
F2
(x)
3^.2
_ ^2
-g
etc.,
shew that
F,i {x) is
(2)
'î^^=Fn-r{^-)
(3)
(4)
If
i^(.r)<;.r
y = a(^F(^
then the mean value

50.
a homogeneous polynomial of degree n in x and
(1)
If
Fn
(*)
of
{n^n
{x)
Cî^l),
+ aiFi
-j~,
h,
(x)
+ a2F2 {x) + ..., where
in the interval from
x= h
to
Uo, !, aa?
x= +A
...
^*'i'e
(cos
-^-2,-
cos -^-
(Leaute.)
is a^.
be defined as in the preceding example, shew that, when
^2,W = (-0-2-^.
real constants,
<x <h,
+^^cos-^ + ...j,
(Appell.)
CHAPTER XVI
THE CONFLUENT HYPERGEOMETRIC FUNCTION
The confluence of two singularities of Riemanns equation.
16"1.
We
have seen
10'8) that the linear differential equation with
two
regular singularities only can be integrated in terms of elementary functions
while the solution of the linear differential equation with three regular
singularities
is
substantially the topic of Chapter xiv.
in order of complexity,
equation which
two of the
we
shall consider a modified
As the next type
form of the differential
obtained from Riemann's equation by the confluence of
is
This confluence gives an equation with an irregular
singularities.
singularity (corresponding to the confluent singularities of Riemann's equation)
and a regular singularity corresponding
to the third singularity of
Riemann's
equation.
The
confluent equation
is
obtained by making
-* oo in the equation
-.-\-m
2
1;
.7
The equation
We
in question
ni
is
readily found to be
modify this equation by writing
equation*
for
= e~ ^^ Wk, m (2)
and obtain as the
Wk^mi^)
^^-1-^-^^^-^]'^The reader
6 and
not
an
is
verify
that the singularities of this equation are at
and when 2m
two integrals of equation (B) which are regular near and
finite values of z are given by the series
the former being regular and the latter irregular
M^,ra{z)-z
ji
i,(2/H
+ l)^^ 2!(2m+l)(2//.+
This equation was given by Whittaker, Bulletin American Math. Soc.

W. M.
integer,
valid for all
will
.-<-)
A.
2)
'^"
x. (1904), pp.
125-134.
22
338
1
J/,._,W
These
.iI -
".
[CHAP. XVI
Ck m k) (| m k)
^ mk
^'|l+iV(i^:2^^+ 2!(l-2m)(2-20 ^+-j-
- xz
series obviously
-,
form a fundamental system of solutions.
[Note. Series of the type in { } have been considered by Kummer* and more recently
by Jticobsthalt and Barnes J; the special series in which k = had been investigated by
Lagrange in 1762-1765 (Oeuvres, I. p. 480).
In the notation of Kummer, modified by
Barnes, they would be written iF-i {hni k; 2m + 1 2}
the reason for discussing
;
solutions of equation (B) rather than those of the equation 2 -j^ (z p) ~~ ai/=0, of
(a
p; z) is a solution, is the greater appearance of symmetry in the formulae,
together with a simplicity in the equations giving various functions of Applied Mathe-
which iFi
matics (see
16"11.
(I)
16"2) in
terms "of solutions of equation (B).]
Kummer s formulae.
We
shall
now shew
that, if
^-^--il/,,(^)
that
is
'2'm is
not a negative integer, then
= (-^)-i~''W_,,,(-^),
to say,
^ ""
h
(I
.,
-^"^
+ m - k) (f + m - k)
(2w + 1) (27^1 + 2)
!
16-11,16"12] THE CONFLUENT HYPERGEOMETRIC FUNCTION
339
of which the second and third factors possess absolutely convergent expansions,
,*:
-A^,-^
^^
.^
"^-^(-11,
77i
2m-n; n + i-m:
is ( 3'73)
i)
(i
m)(#+m)...(/t-m-f-A) ^..
= --T^+
-n+^-m;
-^Fi hi, -m-in;
w^ + 2) ...{27n+7i)
nl{2m+l){2m
;
.
,,
1),
7::
by Kummer's relation*
F{2a,2^; a+/3 + l;
valid
when
(|
â'^ i
and so the
+ w)( + m)
...
A-)
2p\
odd this vanishes
solutions
it is
. . .
(^
- 1?0
'
2*P.p\ {m + 1) {m
+ 2) ... {m+p)'
of equation (B) of 16"1 are not, however, the
-il/t.imC^')
to
take as the standard solutions, on account of the
disappearance of one of them
when 2m
e^^,
valued by taking
tends to zero as
=z
arg (
i -*-
an integer.
^)
that of
14*6,
when
multiplied
is;}:
supposed that arg^ has
chosen that the point
is
by confluence from
integral obtained
by a constant multiple of
It is
w - hi)
even values of n{ = 2p)
{2m -\- 2)
Definition f of the function Wic,m (2)
most convenient
The
{^ )
(^ -
is
{\-m-hi)T{\-\n)
l.S...{2p-l)
2pl 23p(m + l)(m + 2)...(m+p)
The
(by 14-11)
{2m + n) r
for
is{l-x)},
r(i-/)(-^)(-|)...(i-p)
{m + f) ...{m+p-^) {m + \) {m + 2) ... {7n+p) F {^-m-p)
2?P {in-ir\)
16'12.
a+^ + h;
T {-n + h-m)r
{n-m-\-^ )
n\ (2m +
is
^;
coefficient of 2"+'""'"^
n\ (2i+l)(2m + 2)... (2m + 7i) T
and when n
= F{a,
is
its
principal value
outside
it.
and that the contour
The integrand
is
and taking that value of arg
tt
is
so
rendered one-
(1
+ t/z)
which
by a path lying inside the contour.
Under these circumstances it follows from 5"32 that the integral

To shew that it satisfies equation (B), write
z.
is
an
v=r^\* See Chapter xiv,
t)-^-i + "'(l
examples 12 and
The function Wt^^(z) was
+ tjz)^-i+'ê-^dt;
13, p. 298.
defined by
means
of
an integral in
this
manner by Whittaker,
ioc. cit. p. 125.
suitable contour has been chosen
and the variable
of 14*6 replaced
by -
1.
222
[cHAP. XVI
340
and we have without
dH
difficulty*
(^ _-,\dv
i-
7n'
+ k{k-l)
0,
since the expression in
tends to zero as
-*
oo
and
this is the condition
that e'^^z^v should satisfy (B).
Accordingly the function Wk, m
is
a solution of the differential equation (B).
The formula
and k
for TfA;,m (^)
To overcome
integer.
an
defined by the integral
{z)
^ m
is
becomes nugatory when
this difficulty,
not an integer,
infinite integral, after the
manner of 12-22
k and
a positive integer,
-i-:^ 7n\
and
Example.
all
and
is
a negative
we observe that whenever
and
integral into
when
so,
^0,
Wi^miz) in the
This formula suffices to define

is
-2-w
we may transform the contour
Rik
m + 2->t
^'
critical
when
cases
so Wk,m{z) is defined for all values of
values of z except negative real values f.
Solve the equation
dû
in terms of functions of the type Tft_,
(2),
c\
where
a, 6, c are
any constants.
16"2.
Expression of various functions by functions of the type Wk^,n{z).
It has
been shewn:!: that various functions employed in Applied Mathe-
matics are expressible by means of the function Wk,m
(2)
the following are a
few examples
differentiations under the sign of integration are legitimate by 4-44 corollary.
The
When
Wji^jn (^
""
is
real
00) whichever
and negative, T^\^(z) may be defined

is more convenient.
+ Whittaker, Bulletin American Math. Soc.
than
is
given here.
x.
to
be either
this paper contains a
Jrji._^(^
+ Oi)
or
more complete account
16*2]
341
The Error function* which occurs in connexion with the theories of

Probability, Errors of Observation, Refraction and Conduction of Heat is
(I)
Erfc(^)=[
where
a;
is real.
Writing
we
e-''dt,
= x^{w^ l)
and then
iu
= s/x
in the integral for TT.j. .^(a^),
get
Ji
e'^'ds,
2xê^^^
J
and
by the formula
so the error function is given
Erfc
(a;)
Other integrals which occur
- i e - i^' Tf
_
a;
in
j^
{x').
connexion with the theory of Conduction
rb
of Heat, e.g.
e'^^~^^^^^ dt,
can be expressed in terms of error functions, and
J a
SO in terms of
functions.
the formula for the
eri'or
Gamma function,
The Incomplete
(II)
is
W^^m
Shew that
Exaviple.
function
is
true for complex values of x.
studied by Legendre and othersf,
x)=
y{n,
f'-'e-^dt.
Jo
By
writing
=sx
in the integral for Wi.,,^_i\ in(x), the reader will
verify that
(n,
7
(III)
This name
is
:|:,
(x).
also applied to the function
Erf(x)=
{
l), ._,
The Logarithmic-integral function, which has been discussed by

arg { log ^j < tt, by the equation
is defined, when
Euler and others
x)=r (n) - xi ( " l) e " ^^ If^ ( _
Legendre, Exercices,
i.
p.
339
''e-'d(
= 7r^-Erfc(j-).
Hocevar, Zeitschrift fiir Math, und Phys. xxi. (1876), p. 449;

Prym, Journal filr Math, lxxxii.
Schlomilch; Zeitschrift fur Math, und Phys. xvi. (1871), p. 261
(1877), p. 165.
Soldner, Monatliche- Correspondenz, von Zach (1811), p. 182;

X Euler, Inst. Calc. Int. i.
Briefwechsel zwischen Gauss und Bessel (1880), pp. 114-120; Bessel, Konigsherger Archiv, i. (1812),
;
pp. 369-405; Laguerre, Bulletin de la
VEcole norm. sup.
in the higher parts of
Soc.Math.de France,
vii.
(1879), p. 72; Stieltjes,
Ann. de
The logarithmic-integral function is of considerable importance

the Theory of Prime Numbers. See Landau, Primzahlen, p. 11.
(3), iii.
(1886).
342
On
it
writing
may be
log z =
= log
u and then u
[cHAP. XVI
in the integral for
verified that
\i{z)
= -{-\ogzy'^zhv_._^Q{-\ogz).
Weber's Parabolic Cylinder functions
It will appear later that
16*5)
and
Bessel's Circular Cylinder functions (Chapter xvii) are particular cases of the
Wk^rn function.
Examples
at the
The
[Note.
Other functions of like nature are given

end of this chapter.
error function has been tabulated
and Burgess, Tram. Roy.
pp. 248-304,
in the Miscellaneous
by Encke, Berliner
tabulated by Bessel and by Soldner.
integral function has been
Fiinktionentafeln (Leipzig, 1909),
ast.
Jahrbuch, 1834,
The
Soc. Edin. xxxix. (1900), p. 257.
logarithmic-
Jahnke und Emde,
and Glaisher, Factor Tables (London,
1883), should also
be consulted.]
The asymptotic expansion of
16*3.
From
{z),
TT'â;,w
when \z\is
the contour integral by which Wk,rn,{z) was defined,
to obtain an asymptotic expansion for W]c^m{z) valid
For
large.
we employ the
this purpose,
1+-)
z)
=1+^+
\ z
n\\
possible
tt.
example
result given in Chap, v,
...+
it is
when |arg^| <
- + Rn
(t,
that
6,
Z),
z^
where
Substituting this in the formula of

it
and integrating term-by-term,
16'12,
follows from the result of 12-22 that

yyjc,m{2)
= ^ -z
ji
YY^
[m^
- (k -
^y-]
{m'
-(k- ^f]
n
provided that n be taken so large that
Now,
arg 2
if
tt
and
\z\
>
(1
I
-{k-n-if ^f]
(m^
. . .
k-^-\-
m > 0.
j
then
1^1(1 + ^/0)1^1+^
tjz)
{n
21^2
R{z)^Q\
sin a
R (z) ^ O]
^)iM (cosec a)l^l f
'
and so*
1
R^(t, z) ^
^(^-l)---(^-^')
It is
supposed that \
nl
is real
^'^"
1(1
I
'
'
w"(l
+ w)'"
du.
Jo
the inequality has to be slightly modified for complex values of
\.
16*3-1 6 '4]
343
Therefore
\Rn{t,z)\
^(^-^^-^^-'^'>
<
(1
^yK\ (cosec a)'î
(V^)
l+ (1
+ ty^Hn +
l)-\
since
when
Therefore,
>
<
;<
1,
= o|l
r^' + ^ + '" + "(l + 0'^'^'|^r"~ê"'^4
= O(^--0,
The constant implied
since the integral converges.
independent of
That
arg2^,
but depends on
kU
-^z
and
asymptotic expansion of W]c^^n{z)
is to say, the
a,
Wi'
(^
-^H
Im-'
\m'
tlie
(k
formula
-n +
hy] {
(B) of
tt.
The second solution of the equation for
16*31.
given by
is
is
a0.
nlz"
for large values of\z\ when arg 2; ^ tt - a <
The
- (k -
=i
the symbol
in
tends to infinity as
16"1 satisfied
W^^^ni^)-
by Wk,m{z)
unaltered
is
if
the signs of z and k are changed throughout.
Hence,
Since,
if
arg { z)\<
ir,
W^k^m (
when arg z\<Tr,

|
ir,,,(^)
whereas,
z) is a solution of the equation.
when arg ( |
2)
= e-^^'/{l +
0(^-1)1,
< tt,
W.k,.n{-z)
e^'H-z)-^[l
+ 0{z-%
the ratio W]c^m{z)jW-]c^jn{ z) cannot be a constant, and so Wk^m{z) and

^-k,m{ z) form a fundamental system of solutions of the differential
equation.
16'4.
Contour integrals of Barnes type for
W]c^,n{z)-
Consider now
g-JV^-rri
where arg^; <

|
Y{s)T{-s-k-m+l)V{-s-k + m-\-\)
^
^---lW>
T{-k-m-^\)V{-k + m + \)
J-oc.j-xi
tt,
and neither of the numbers k +
m + 5 is a positive
integer
344
or zero*; the contour has loops
those of
It
is
r (s
km
and so the
in the
by
-w
a<
1"^
(e-^<'\\s\-^^-^),
which
analytic at
is
all
points
^tt.
N so that the poles oi T ( s k m + ^j T ( s k+ m + ^j
choose
are on the right of the line R(s)
= N-^; and
round the rectangle whose corners are
and
m+
r (- s - k +
integi-al represents a function of 2
it.
s>cc on the contour,
13'6, that, as
+'^^
T (s) and
^j are on opposite sides of
domain arg z\r^^7r-
Now
necessary so that the poles of
+ 5 F ( s k + m +
easily verified,
T {s)T (- s -
if
[CHAP. XVI
consider the integral taken
N^ ^, where
î,
is
positive J
large.
The reader
when arg z\^-7r
will verify that,
tend to zero as
e-h^^k
27ri
^> 00
fi
so,
and
the integrals
r-N-i + î
pA'-J-|J
'
<x,
î
m + i)T (- s - k + m +1)^S(^^
T{-k-vi+^)l{-k + m + ^)
T (s)T (- s - k -
|-ocr
j-ooi
= e-^-'zH 1 R
=0
^TriJ
where Rn
Write
is
T{-k-m + ^)T(-k + m + ^)
_^^_^_î
NI +
it,
e-*!'l
n.
and the modulus of the
where the constant implied

Since
in the
symbol
\^~'^^dt converges,
is
we
J'
last
integrand
independent of
is
z.
find that
I=e-^'z^\ I Rn + 0{\z\-^-^)[.
*
In these cases the series of
16-3 terminates
and W^
is
i (z)
a combination of elementary'
functions.
t
The
The
integral is rendered one- valued

line joining ^j
may have
when
J? (z)
<0
by specifying arg
z.
loops to avoid poles of the integrand as explained above.
16 4]
we get
But, on calculating the residue Rn,

7?
""'
_
and
/ has
so
Further
substituting
T(n-k-m+^)r(n-k + m + ^)
nir(-k-m + ^)ri-k + m+^)^
-{k-
{m^
[in'
^Y]
-(k- f
... {rri'
)^}
the
same asymptotic expansion as
satisfies
the expression (given in
,dH
Wk^mX'^).
for
F (- s - k +
^j
^f]
Wk, m
(^)
for,
on
for v
in
m + ^j z^ds
16*12)
dv
^,
m+
__
^
- (k-n+
equation
differential
F (- s - k -
r{s)
the
345
l\
i\ /,
/,
dv
we get
p' r(s)T(-s-k-m. + l)r(-s-k + m + |)
z'ds
-j"\r(s + i)r(-s-k-m + l]r(-s-k + m + iy'+'ds

= (r' - r^"'^
r(s)
rf-s-k-m + f] rf-s-k + m + l^z'ds.
Since there are no poles of the
integrand between the contours, and

> oo s being between the contours,
last
since the integrand tends to zero as
the expression under consideration vanishes, by Cauchy's theorem
the equation for
satisfies
and
and
so
{2).
I=AWk,m(z) + B W^^, m (-2),
Therefore
where
TTjt,
are constants.
Making
when R(z)>0 we
Wk,m{ ^). that
^ -> 00
|
the asymptotic expansions obtained for / and
see,
from
^ = 1, 5 = 0.
Accordingly, by the theory of analytic continuation, the equality
I=W,,,n(z)
persists for all values of z such that
such that
TT
Example
< arg ^ <

I
1.
TT,
Wk,m
(z)
arg^r|<7r; and, for values* of arg^'
may be
defined to be the expression /.
Shew that
taken along a suitable contour.

* It
would have been
by modifying the path of integration in 16-3, to have shewn

But the
to define an analytic function when \a.rg z\<^w.
unnecessary to do so, as Barnes' integral affords a simpler definition
possible,
that that integral could be
reader will see that

of the function.
it is
made
'
346
Example
Obtain Barnes' integral for W^^^(z) by writing
2.
2Trl
for (l
+ tjz)
'^'^"'
If
J _
in the integral of
i^
16"12 and changing the order of integration.
Relations betiveen Wk,m{z)
16"41.
[CHAP. XVI
'-^nd
Mk,m{z)-
we take the expression

F{s)
= V (s) vi-s-k-m + '^^Tl-s-k-^- m + 1'
which occurs in Barnes' integral
and write
for W]c, m (^).
it in
the form
^rj)
'
+ k + m + ^)r{s + k m + ^)cos{s + k + m) ir cos {s-\-k - m) it

by 13'6, that, when B (s) ^ 0, we have, as 5 > x
r(s
we
see,
F{s)
exp|C-s-^-2^'jlog5 +
Hence,
if
arg
2^
|
<
tt,
sec {s
-\-
^
'
m)
tt
sec {s
m)
ir.
taken round a semicircle on the
jF{s)z--ds,
right of the imaginary axis, tends to zero as the radius of the semicircle
tends to infinity, provided the lower bound of the distance of the seraicircle
from the poles of the integrand
positive (not zero).
= - r(_^_,,, +^)r(-yfc + m
+ f)
Tf,,..(.)
Therefore
is
where SR' denotes the sum of the residues of F(s) at its poles on the
right of the contour (cf. 14"5) which occurs in equation (C) of 16*4.
Evaluating these residues we find without difficulty that, when
arg 2
<
and
2m
TT,
not an integer*,
is
Example
1.
T(-2m)
Shew
that,
,.
r(2m)
when arg ( -s) <|7r and 2m

|
is
,^
not an integer,
(Earnest.)
Example
AVhen - -stt
2.
When 1m
is
'^)t,m (2)
may
tt
and - f tt
< arg
- ^)
< ^tt,
an integer some of the poles are generally double
shew that
poles,
and their residues
has not been proved when fe- |i 7h. is a positive integer or

be obtained for such values of k and m by comparing the terminating series for
involve logarithms of
zero, but
< arg s < f
z.
with the series
The
result
for Mj.^, (2).
16-1.
t Barnes' results are given in the notation explained in
16*41-16-51] THE CONFLUENT HYPERGEOMETRIC FUNCTION

Example
Obtain Kummer's
3.
first
formula
gng-z^
T{n-s)z'ds.
The parabolic cylinder functions.
Consider
the
equation
differential
16'11) from the result
r^^'
Inl JI -xi
16"5.
347
(Barnes.)
Weber's equation.
by
satisfied
= z'^Wj^
iv
_i[-^Z'];
it is
d ( d{wz^)
zdz \ zdz )
2k
z-
\2k
I z'A
this reduces to
-.
\
\-
dz-
ix^î^^.
'
z*\
w = 0.
Therefore the function
satisfies
the differential equation
Accordingly Dn(z)
is
one of the functions associated with the parabolic
cylinder in harmonic analysis*; the equation satisfied by
it
will
be called
Weber's equation.
From
16'41, it follows
when arg z\ <

I
But
that
Q
-tt.
z -i
and these are one-valued analytic functions of z throughout the ^^-plane.

Accordingly Dn (z) is a one-valued function of z throughout the ^-plane and,
;
by
16"4, its
asymptotic expansion when
n(n-l)
-iz2nL
^
16'51.
r"
< - tt
is
n{n-l)(n-2){n-S)
"^
2z^
arg ^
2.4^
-y
The second solution of Weber's equation.
Since Weber's equation is unaltered if we simultaneously replace n

and z by n 1 and + iz respectively, it follows that D_n-i (iz) and
D-n-i ( iz) are solutions of Weber's equation, as is also Dn ( z).
*
Weber, Math. Ann.
pp. 417-427.
i.
(1869), pp. 1-36; Whittaker, Proc,
London Math. Soc. xxxv.
(1903),
348
It is obvious from the asymptotic expansions of
^ tt <
valid in the range
<
arg z
-ir,
[CHAP. XVI
Dn{z) and
Z)__i(2e^'^*),
that the ratio of these two solutions
is
not a constant.
The relation between
16'511.
From
the functions Dn{z),
iz).
the theory of linear diiSerential equations, a relation of the form
Dn {z) = aD_n-i
must hold when the
To obtain
we observe
z,
h D__i
iz)
on the right
that
not a constant.
is
the functions involved be
if
i=nr^
...
two terms we get
first
(27r) - *
(-
the expansions are
Comparing the
{iz)
ratio of the functions
this relation,
expanded in ascending powers of
and
D^n_^ (+
r (w + 1) e^''''\
(27r) " ^
T (w +
"
1) e
so
I>n{z)
= r(n + l)
V(27r)
The general asymptotic expansion of Dn {z).
16'52.
So
+ e-'^''^^ D_n-,{-iz)
D_n-^{iz)
L
far the
been given
D^ (z)
asymptotic expansion of
( ] 6*5)
in the sector
< jTt.
arg z
of arg z not comprised in this range
for large values of z
we
write
To obtain
iz
for z
its
and
has only
form for values
n 1
for
7i
in
the formula of the preceding section, and get
DAz) = e''^'D,(-z) + ^^^êî^'^'^-'D_r.-^{-iz^

Now,
+
TT
if - TT
>
>
arg z
and arg ( z)
jTt,
arg z
we can
ir,
the asymptotic expansion of

II
TT
>
/.\
arg ^r
assign to .
arg (
iz)
^^
and
1
= arg z ^tt;
Dn{ z) and
16"5 to
iz arguments between
and then, applying
D^n-i{iz), we see that,
> - TT,
-i^-,nJi
^^
(^t
- 1)
V(27r) ^n^,-^,-2^_-i
r(-w)
L
[
{n
_^
(w_
1) (n
1)
(71
2)
(/t
3)
+ 2)
2z'
(n
+ l)(+2)(n + 3)(n +
4)
16"511-16"6] THE CONFLUENT HYPERGEOMETRIC FUNCTION

This formula
validity, viz. Jtt
not inconsistent with that of
is
<arg2 <|7r,
To obtain a formula
16-5 since in their
common
e^^" 2-2.1-1 jg o(z-"') for all positive values of
valid in the range
349
range of
w.
- tt > arg z > jir, we
use the
formula
and we get an asymptotic expansion which

been obtained only in containing
Since Dn(z)
is
e""'^' in place of
asymptotic expansion of
tt ^ arg z
i) (z)
^tt, the complete
has been obtained.
contour integral for Dn{z).
no+)
Consider
e'"^'.
one-valued and one or other of the expansions obtained
is
valid for all values of arg z in the range
16*6.
from that which has just
differs
_ ,1,2
^^
2^
(-t)'"'-'^ dt,
where arg ( |
<
it
represents a one-valued
00
throughout the ^-plane
( 5-32)
and further
the differentiations under the sign of integration being easily justified

integral satisfies the differential equation satisfied by e ~ i^ Z)
(2)
accordingly the
and therefore
e-'f-î\-t)-'^-'dt = aB{z) + bD__i{iz),
e-^'where a and b are constants.
Now,
if
the expression on the right be called
e-^{-t)-^-^dt,
En{0)=\
To evaluate
R {n) <0
e-\-t)-^dt.
En'{0)=
these integrals, which are analytic functions of n,
then, deforming the paths of integration,
i:(0)=-2isin( +
=2"^" i'sin
= 2 - 4"
l)7r
7r
e'
a=
T - hi).
(
sm{nn)r{^- ^71).
^^^^ 2 r
(i)
Therefore
the equations are true for
D,,
(z)
= - ^'^^^
^'' i
sin {nn)
T ( - ^^/i)
- ^< -
22"
= 2zr n)
{
that
u~ ^^~^ dti
&=0,
first
^(- t'^'-'^dt
Both sides of these equations being analytic functions of

n and therefore
values of
we suppose
we get
e~'^
sin (nn)
' (0) = - 2^ " 4"
Similarly
all
E^ (2)5 we have
sin mr.
' i^' f*"^^
*'
- O"""
dt.
350
From
Dn {z).
Recurrence formulae for
16*61.
[CHAP. XVI
the equation
after using 16*6,
we
see that
Dn+,
(z)
- z Dn (z) + n Dn-, {z) = 0.
Further, by differentiating the integral of 16"6,
it
follows that
D^ {z) + \zDn {z) - nDn-, {z) = 0.

Example.
Obtain these results from the ascendiug power series of
16"7.
Properties of
When
Dn
{z) ivhen n is
n 'e"i^^
If
an integer.
an integer, we may write the integral of
is
now we
write
= v z, we
16"5.
/-(0+)
g - 2 i
-
16"6 in the form
get
dv
zf
= (-rei''^Je-n
dz'
a result due to Hermite*.

Also, if
and n be unequal
integers,
we
see from
the differential
equations that
Dn {z) Dm" (z) - D, {z) Dn" (z) + (m - n) Dm (z) Dn (z) = 0,

and
so
{m - 71)
Di (z)
Dn (z) dz = Dn{z)DJ{z)-Dm(z)Dn'{z)
J ex
0,
by the expansion of 16"5 in descending powers of z (which terminates

and is valid for all values of arg z when n is a positive integer).
Therefore
if
m and
n are unequal positive integers
D,iz)Dn{z)dz=0.
>
Comptes Rendus,
lviii. (1864), pp.
266-273.
167] THE CONFLUENT HYPERGEOMETRIC FUNCTION
16'6l,
On
when
the other hand,
7n
351
we have
n,
cc
= D, (Z) Dn+, (Z)]
zDn
\l
iz) D,,^, {Z)
- Dn+^
(z)
D^' (z)\ dz
=rX [D,^,{z)Ydz,
J
on using the recurrence formula, integrating by parts and then using the
recurrence formula again.
It follows
by induction that
{Dr,(z)Y-dz
= nir
[D,{z)Ydz
J CO
= (27r)^n!,
by
12-14 corollary 1 and
12-2.
It follows at once that
/(z)
oc
for
a function /(^), an expansion of the form
= aoDo (z) + a, D,{z)+... + a^Dn {z)+
and if it is legitimate
and oo then
exists,
if,
to integrate
...
term-by-term between the limits
^"
= (270*^f_^^^^-^(*>^^REFERENCES.
W. Jacobsthal,
E.
W. Barnes,
E. T.
Math. Ann. LVi. (1903), pp. 129-154.
Trans. Camb. Phil. Soe. xx. (1908), pp. 253-279.
Whittaker,
Bulletin American Math. Soc. x. (1904), pp. 125-134.
H. Weber, Math. Ann.
i.
(1869), pp. 1-36.
A. Adamoff, Ann. de VInstitut Polytechnique de St Petershourg, v. (1906), pp. 127-143.

E. T.
G. N.
Whittaker,
Watson,
Proc.
Proc.
London Math.
London Math.
Soc.
Soc.
xxxv. (1903), pp. 417-427.
(2),
viii.
(1910), pp.
393-421; xvn. (1919),
pp. 116-148.
H. E.
J.
CuRZON, Proc. London Math.
Soc. (2), xii. (1913), pp. 236-259.
A. Milne, Proc. Edinburgh Math. Soc. xxxii. (1914), pp. 2-14; xxxill. (1915), pp. 48-64.
N. Nielsen, Meddelelser K. Danske Videnskabernes
Selskab,
i.
(1918), no.
6.
the transcendental functions
352
[chap. xvi
1.
Shew
that, if the integral is convergent, then
^^'^^')-
r{l+m-^k)T{h + m-k)
that Mk,,n{z) = z^
+ "^ e'
2.
Shew
3.
Obtain the recurrence formulae
4.
Prove that Ht
numbers k-h + m
5.
Shew
is
that,
.,^^^''^
w)
(1
lim i^(^ + wi-X-, ^ + to-X- + p
2^
du.
e-
2m + 1;
the integral of an elementary function
is
,,1(2)
when
z/p).
either of the
a negative integer,
by a suitable change of
{a.2
variables, the equation
+ b2.v)^ + {ai + biX)- + {ao + box)i/ =
can be brought to the form
derive this equation from the equation for F{a, b; c; x) by writing x = ^lb and
making
6-*-
6.
VI.),
Shew
that the cosine integral of Schlomilch and Besso {Oiornale di Matematiche,
ci(.)=[;^^.^.,
is
equal to
Jz
" ^ e^'^ + *'^''
F_ 1
iz)
+ 1.- " 2 g - i'- " i'^'
Shew
Tf _
iz).
also that Schlomilch's function, defined {Zeitschrift filr Math,
und
Physik,
(1859), p. 390) by the equations
S{v,z)=j
is
î"
equal to
7.
Express in terms of
W,cî
Si
8.
Shew
equal to
1 î^
W^ _
^du,
1 ^ (2).
functions the two functions

^"'
'
f
Jot
^dt,
-Ei{z)=['^~dt.
J
that Sonine's polynomial, defined {Math. Ann. xvi. p. 41) by the equation
-'m"(2)=;^
is
(?)
+ t)-''e-'^dt = z''-ê'
{l
/_,
(i
^\,
+ ?i)!
.
(x\
0!
rz.
1 \
/ iTm +:(n-2)I (m + -2)! 2!
(??i + /i-l)! 1!
(w-1)!
I
1
-
-.,
-.-,
M' +
1) ^,^~
{z).
"'
iv.
THE CXDNFLUENT HYPERGEOMETRIC FUNCTION

9.
p.
(1
by Lagrange
defined
<})mi^)
1762-1765
in
520) and by Abel (Oeuvres, 1881, p. 284) as the coefficient of

- A)-i e-''Mi-ft) is eqnal to
Shew
10*.
<>>n,m ('))
353
/('*'
{Oeuv)-es,
i.
in the expansion of
that the Pearson-Cunningham function {Proc. Royal Soc. Lxxxi.
p. 310),
defined as
T{n
n- hm
is
^ "^,
equal to
Shew
11.
that, if
arg z
"
,,
< ^n,
"^
("^
and arg (1
|
1) e
"
^MF ^ ,
+ ^) <
{z).
tt,
(Whittaker.)
Shew
12.
that, if
n be not a positive integer and
^"^'^
and that
arg
< frr,
^^^^
T{-n)
j .î
then
'^^'
'
this result holds for all values of args if the integral be
enclosing the poles of
Shew
13.
27r/
if
that, if
r(j
but not those of r
arg a
{^t
the contours
-\n).
< \it,
tc
^-^
n^(-*". hn-i<;hn-hn + \r,
r(-w)r(^m-î + l)aH'+i)
Deduce from example 13
14.
|J
- i''
z""'
that, if the integral is convergent,
A+i
(2) c/2
l-ia-).
then
= (v/2)-i-'" r (i + 1) sin (^ - {m)
n.
(Watson.)
Shew
15.
that, if
n be a positive integer, and
E,
{x)
TJ-
^''
B {.v)= le'fnni ^K^^^) r
then
{n
{z
if
- .v)
+ 1)
n,, (z) dz,
"
^"''"^
i)__i
+ ix),
the upper or lower signs being taken according as the imaginary part of
Shew
16.
fi
is
^n or J(n-
^{-f 2" + 2 (27r) - ^ ei-^' Jo

i'
1),
whichever
is
u'' e
2"' ''^^
sm
{2xu)
W. M.
The
A.
positive
dii,
an integer, and the cosine or sine
even or odd.
*
is
that, if n be a positive integer,

i> {X)
where
(Watson.)
or negative.
is
taken as n
(AdamoflF.)
results of
examples
8, 9,
10 were communicated to us by
Mr Bateman.
23
is
354
Shew
17.
[CHAP. XVI
that, if n be a positive integer,
Ji= (^ e'
where
^^
(^ "
"""^^
sin
J _oo
Jo=
"
(t(v)
sni
jo
{xv ^hi) di;

^ '
'^
Jn)
(xv
^
'
dv,
J, = j[j-ni^-^f^2^xv,,n)d.,
and
o-(^)
With
18.
= e^^"(l-^'')i;"-e-"(^-l)'.
(Adamoff)
the notation of the preceding examples, shew that,

T
J,
when x
is real,
-\ ~ 5*
J-x^COS
\
= -,T^n
ê
(xJn);
,
sni
while Ja satisfies both the inequalities
\J,\<(^^^~ e-'\
|./3i<2e--{|.rUM},
Shew
Jijn) <
and
oI
(1
-Ai)
< J
'
By employing
19.
-,
co,j(.r)
satisfies
''^
to
at i'=l
decreases, its
mean
= V2.(v'0
decreases from
o- (/')
minimum
sj{^ln)
.-1
cos(.r2
<
''-^
<
{In)
'
=0-0742....
when
:^-)iTr)
necessary,
(Adamoff.)
shew that
'
s'n J'
both the inequalities
ia;2
,^,
.1
_i
^'TT
I
is real
and n
is
an integer greater than
Shew
that,
if
n be positive
(AdamofF.)
2.
but otherwise unrestricted, and
if
be a positive
integer (or zero), then the equation in z
has
m positive
roots
at-
oo, o-(v) increases
where
to a
from 1 to
limit being zero where
v increases
value theorem
3-35...
20.
'
< An ~ i,
(1 +/^2)
when X
1,
and as
< x/(l)
the second
A.(.^')
where
from
also that as v increases
v=\hi and then increases to

to a maximum at l + /?2 and then
when 2m
-\<n< 2/h +
1.
(Milne.)
CHAPTER XYII
BESSEL FUNCTIONS
The Bessel
171.
coefficients.
In this chapter we shall consider a class of functions known as Bessel

functions or cylindrical fimctions which have many analogies with the Legendre
Just as the Legendre functions proved to be parti-
functions of Chapter xv.
cular forms of the hypergeometric function with three regular singularities, so
the Bessel functions are particular forms of the confluent hypergeometric

function with one regular and one irregular singularity.
the Legendre functions,
we
first
As
in the case of
introduce* a certain set of the Bessel functions
as coefficients in an expansion.
For
all
values of z and
{t
excepted), the function
can be expanded by Laurent's theorem in a series of positive and negative

powers of t. If the coefficient of V\ where n is any integer positive or
negative: be denoted
To express Jn {z)
since the contour
we may take
is
by J^
as a
power
series in
any one which
to be the circle
it
from
{z), it follows,
z,
5 "6, that
write u
"^
= ^. 2
(V
as the integrand can be
/i
\n~-2r
on
expanded
this contour, it follows
a positive integer or zero
is
,'(0+)
M- [\
Now
residue
t'-ê'dt.
^
when n +
is
[(n
is
+ r)l}~^
by
61, when
a negative integer the
is zero.
Therefore,
if
is
ll
a positive integer or zero,
2"n!(
*
then
4*7 that
Jn {z)
encircles the origin once counter-clockwise,

\t
in a series of powers of z uniformly convergent
from
= 'Itjz
This procedure
is
due
^t
">
'"]'
2M(?z + l)'^2M.2(7H-l)(w + 2)
to Schlomilch, Zeitschrift filr
Math. undPhys.
ir.
(1857), pp. 137-165.
232
356
when n
whereas,
r=m r.{r-m)l
and so Jn
The
of
(z)
now been
function Jn{z), which has
and negative,
is
defined for
all
integral values
^.
shall see later ( 17'2) that Bessel coefficients are a particular case of a class of
known
functions
The
s)lsl
called the Bessel coeffi^cient of order n] the
series defining it converges for all values of
We
(m +
^=0
= {-y^Jm {z).
positive
7},
vi,
a negative integer equal to
is
[CHAP. XVII
as Bessel functions.
by which Jn (z) is defined occurs in a memoir by Euler, on the vibrations

membrane, Novi Comm. Acad. Petrop. x.- (1764) [Published 1766],
it also occurs in a memoir
pp. 243-260, an investigation dealt with below in 18'51
by Lagrange on elliptic motion, Hist, de I'Acad. R. des Sci. de Berlin, xxv. (1769) [PubUshed
series
of a stretched circular
1771], p. 223.
The earliest systematic study of the functions was made in 1824 by Bessel in his
Vntersuchung des Theils der planetarischen Storungen welcher aus der Bewegung der Sonne
entsteht {Berliner
Ahh. 1824)
researches published before
special cases of Bessel coefficients had, however, appeared in
769
to Leibniz*, in
function of order
^-;
memoir on the
the earliest of these
Jakob Bernoulli
which occurs a
in a letter, dated Oct. 3, 1 703, from

which is now described as a Bessel
is
series
the Bessel coefficient of order zero occurs in 1732 in Daniel Bernoulli's
oscillations of
heavy chains, Comm. Acad.
Sci.
Imp. Petrop.
vi.
(1732-1733)
[Published 1738], pp. 108-122.
In reading some of the earlier papers on the subject, it should be remembered that the
{z) being now written / {2z).
notation has changed, what was formerly written J^
Example
1.
Prove that
then
if
sin
e2
hz= AiJi
(z)
+ A2J2 (2) + ^3 J3 (2) +
. . .
(Math. Trip. 1896.)

[For, if the contour
enough
in the ?;-plane be a circle with centre
n)-
e-
u-
if^J
replacing the integrals by Bessel coefficients,
^^
i^fn-l)
(-'"''
2a
??j
4b~
and radius large
u) A,,u-''-\
the series on the right converging uniformly on the contour
TTi
we have
to include the zeros of the denominator,
and
so,
using 4*7 and
we have
Izin-'^-)
27ri J
fA^
A2
A3
\ii^
u^
u*
= A-^Ji{z) + A2J2{z) + A3Js{z) + ....

*
Published in Leibnizens Ges. Werke, Dritte Folge,
iii.
(Halle, 1855), p. 75.
du
17* 11]
BESSEL FUNCTIONS
357
In the integral on the

radius
left write ^{u-u-'^)-a=-t, so that as it describes a circle of

describes an ellipse with semiaxes cosh/3 and sinh/3 with foci at -ai; then
eP, t
we have
the contour being the ellipse just specified, which contains the zeros of
the integral by 6*1, we have the required result.]
Example
2.
Shew
that,
when n
4
(.y
is
an
+ 2) =
t--\-lfi.
Evaluating
integer,
2 J,
Vl=-x
(7/)
/_ , (z).
Xeumann and
(K.
Schlafli.)"
[Consider the expansion of each side of the equation
exp {i (y+.) (<-!)} =exp
Example
ex,,
{k
4.
^ j^
Shew that
Jo
{r)
+ 2i cos 4> Ji (s) + 2i'^ cos 2(^^ (2) +

= ^+y2
(^2)
if r2
= Jo ix) Jo (y) -
2.7:3 (.;)
J2 (y )
+ 2-/4
(0,-)
We
is
.]
])}
. . .
My)-....
(K.
17"11.
(<
Shew that
3.
e" <=os,|,
Example
[y (,-])]
Neumann and Lommel.)
Bessel's differential equation.
have seen that, when n
is
an integer, the Bessel
coefficient of order n
given by the formula
From
this formula
we
shall
now shew
that Jn{z)
is
a solution of the
linear differential equation
which
is
dhj
dy
n'^\
dz~
z dz
z-j
_
'
"^
called Bessel's equation for functions of order
For we
find
dz-
since ^~"~^
on performing the differentiations
dz
= 0,
exp {t z^l^t)
dz-
The reader
will
is
observe
( 4'2)
that
z'-J
one-valued.
n.
dz
that ^
Thus
have proved that
z^l
lue
is
a regular point and ^
=x
irregular point, all other points being ordinary points of this equation.
an
358
Example
By
1.
with regard to
difi'erentiatiiig
and with regard
[CHAP. XVII
the expansion
to
t,
shew that the Bessel
coefficients satisfy Bessel's-
(St John's, 1899.)
equation.
Example
The function
2.
P,,'"
(\- ^\
X
An-
the equation defined by the scheme
satisfies
shew that
J,,, (^)
satisfies
The
17-2.
+1
-n
il
^/
i'/Jl
hn
the confluent form of this equation obtained 'by making
solution
of Bessel's equation when n
is
/^
^x
not necessarily an
integer.
We
J a {z)
now
to the case
methods similar
when n
is
to those of
dzis satisfied
manner of
proceed, after the
15-2, to
any number,
ITII
extend the definition of
real or complex.
It
appears by
that, for all values of w, the equation
z dz
z-J
by an integral of the form

y
= z'"
r"-^ exp
(t
- ^\
dt
(t - Z'/U) resumes its initial value after describing

and that differentiations under the sign of integration are justified.
provided that i~"~^ exp
Accordingly,
we
define / (z)
2n
by the equation
'-(0
+)
^2v
the expression being rendered precise by giving arg z

taking arg t\^7r on the contour.
its
principal value and
To
express this integral as a power series, we observe that it is an

and we^ may obtain the coefficients in the Taylor's
analytic function of z;
series in
and
powers of z by differentiating under the sign of integration

Hence we deduce that
4-44).
dt
V
by
12-22.
This
is
{-Yzr~
the expansion in question.
5"32
'
17"2, 17'21]
BESSEL FUNCTIONS
Accordingly, for general values of
n,
359
we define the Bessel function Jn(2)
by the equations
1
~
,.
y- r(o+)
/I
2"+2'-r;r(n
2-\
+r+
l)
This function reduces to a Bessel coefficient when n
is
an integer;
it is
sometimes called a Bessel function of the first kind.
The reader
will
n,
is
observe that since Bessel's equation
is
unaltered
by-
fundamental solutions are Jn{z), J_n{z), except when

an integer, in which case the solutions are not independent. With this
writing
n for
exception the general solution of Bessel's equation
where a and
is
second solution of Bessel's equation when n
is
an integer
will
be given
later ( 17-6).
17*21.
The recurrence formulae for
As the
Bessel function satisfies a confluent form of the hypergeometric
equation,
it is
to
to the relations
the Bessel functions.
be expected that recurrence formulae
will exist,
corresponding
between contiguous hypergeometric functions indicated
in
14-7.
To
establish these relations for general values of
have recourse to the result of
at length,
=
=
and
On
17"2.
n, real
or complex,
we
writing the equation
we have
U-''
27ri
+ \ zH-''-- -
|(2^-if -^/_,
so
Next we
- )
(20"+^/+,
(z) 4- 1 z'
J^^_^(z)
have, by
nt-''-A exp (t
+ Jn+i{z}=
dt
(z)
('Iz-^Jn (z)\
2n
Jn(z)
z
(A).
4*44,
=-
z
^ ^.
=Z
r(0"^)
J n+\
\2)y
^-\
r-" exp
/
-U
dt
360
and consequentl)^
if
[cHAP. XVII
primes denote differentiations with regard to
z,
(B).
J.,:{2)J^J,,{z)-J,^,{z)
z
From (A) and
(B)
easy to derive the other recurrence formulae
it is
(C),
Jn{z)==\[Jn-,{z)-Jn^,{z)]
and
Example
1.
Obtain these results from the power series for
Example
2.
She w t hat
Example
3.
Shew
that
Example
4.
Shew
that
j_
{z^J,,
J^
16^- {z)=Jn-i
Example
5.
Shew
Example
6.
Shew that
(-')}
(2)
=-Ji (2).
- 4/ _ 2 (2) + ^n (2) - 4/ ^ 2
(2)
(2)
+ Jnî {z).
that
J-2(2)
17"211.
t/ {z).
= 2" Jn -lis)-
^2(2)-./o(2)
an
W-
= Jn-,{z)--Jn{z)
Jn'{z)
between
Relation
= 2./o"(2)-
= Jo"(2)-2-lJo'(2)-
tiuo
Bessel functions luhose orders differ
by
integer.
-From the
can be deduced an equation connecting any two
last article
Bessel functions whose
ordei-s differ
.-"-/^. (z)
by an integer, namely
= {-y
^^ [z-j,
(z)},
where n is unrestricted and r is any positive integer. This result follows at

once by induction from formula (B), when it is written in the form
17'212.
The connexion
The reader
write y
which
will verify
= z~^v and
is
bettceen
Jn(z) and Wk^mfi-nctions.
without difficulty that,
then write z
= xjli, we
d-v
dx-
the equation satisfied by Wo^n{x);

./ {z)
Comparing the
= Az
- ijlfo.n (2t>)
coefficients of
Jn
{z)
get
i-w^N
^^
if in Bessel's
_^
)
it
+ Bz
follows that
-
*#,_ {2iz).
^*" on each side we see that
= -TI
z-^
ri
M^,n
iîz).
equation
we
17"211-17-22]
BESSEL FUNCTIONS
except in the
when
critical cases
2/; is
361
when n is half of
Kummer's second formula
a negative integer
a negative odd integer, the result follows from
( 16-11).
The zeros of Bessel functions
17'22.
The
ivhose order n is real.
deduce the interesting theorem that

lies one and only one
relations of 17"21 enable us to
hehveen any two consecutive real zeros of z~^^Jn{z), there

zero* of z-'^Jn+îz).
For, from relation (B)
when written
Z-J,,^,{z)
it
follows from Rolle's
of ^~"J(^') there
is
follows that
theorem + that between each consecutive pair of zeros
when written
former function
satisfies
easily verified
-j-
is
zero
by
same
there
is
zeros;
for
the
the equation
by induction on
differentiating this equation that
z, all
differential coeflficients of
if
y vanish,
5'4.
The theorem required

zeros
common
have no
[z~^^Jn{z)]
both y and -^ vanish for any value of
and y
z''^'^'^Jn+i{z)
z^^'^''^Jn{z).
Further z~'^Jn{z) and
it is
in the form
between each consecutive pair of zeros of
at least one zero oi
and
= -^^\Z-Jn{z)],
at least one zero of ^~"'J,n.i(2').
Similarly, from relation (D)
it
in the form
is
now obvious except
for the numerically smallest

^~" J (z) and 2^"+ V (z) have the
f of z~'^Jn {z), since (except for z = 0),

zeros.
But z = is a zero of z~'"'Jn+i(z), and
if
there were any other
|, then z^^'^^Jn(z)
would have a zero between
and ^1, which contradicts the hypothesis that
and ^.
there were no zeros o{ z''^'^^Jn(z) between
positive zero oi z~'"'Jn+i(z), say
The theorem
is
less
than
4,
and example 19
at the
end of the chapter.]
Proofs of this theorem have been given by Bocher, Bull. American Math. Soc.
206; Gegenbauer, Monatshefte

Math. Soc. IV. (1898), p. 274.
p.
t This
is
fiir
Math.
viii.
(1897), p.
iv. (1897),
383; and Porter, Bull. American
proved in Burnside and Panton's Theory of Equations (i. p. 157) for polynomials.
with continuous differential coefiBcients by using the First
may be deduced for any functions

Mean Value Theorem ( 4-14).
It
which was
therefore proved.
[See also 17'3 examples 3 and

*
^1,
362
17'23.
We
BesseVs integral for the Bessel
shall
next obtain an integral
coefficients.
given by Bessel in the particular
first
case of the Bessel functions for which n
[CHAP. XVII
a positive integer; in
is
the result resembles Laplace's integrals given in
1523 and
some
respects
15*33 for the
Legendre functions.
In the integral of
17"1, viz.
Ziri J
take the contour to be the circle
and write u =
e'*",
so that
Ztt J _^
Bisect the range of integration and in the former part write
- 6
for
we get
27r j
and so
(^)
TT
which
is
27r J
f"
cos {n6
z sin B) dd,
Jo
the formula in question.
Example
1.
Shew
that,
when
Example
2.
Shew
that, for all values of n (real or complex), the integral
z is real
/'
v=-
and n
COH ( nd -
satisfies
1
dy
2v
which reduces to Bessel's equation when n

on the
left is
sin d)
dd
is
sin7r /I
d^^Jdz'^y'V-}^
[It is easy to
integer,
jo
n-
d^'i/
an
is
V~
n\
\^~7^)'
an integer.
shew, by differentiating under the integral sign, that the expression
equal to
17*231.
The modification of BesseVs integral ivhen n
We
now shew
shall
J^ {z)
when R{z) >
0.
"^-
that*, for general values of
r cos {nd - z sin 6) dd - ^^"
is
not an integer.
n,
'''^
e-^^'-'^'^He
This obviously reduces to the result of
17'23
an integer.
Taking the integral of
This result
is
17*2, viz.
due to
Schliifli,
Math. Ann. in. (1871),
p. 148.
.(A),
when
is
BESSEL FUNCTIONS
17-231]
17-2.^,
and supposing that
But,
if
r^+>
1 to
taken twice and the
when
when arg z\<^7r; and
e.
i.
^ uz,
i\)
fi
presents an analytic function of z

path,
have, on writing
the contour be taken to be that of the figure consisting of the real
axis from
we
z is positive,
363
so,
circle
R (zu)
=1,
this integral re-
negative as
is
m -^
|
on the
oc
by the theory of analytic continuation,
the formula (which has been proved by a direct transformation for positive
values of z)
R {z) >
true whenever
is
0.
Hence
where
= l, and arg
on the third path.
denotes the circle |w]
integration while arg u
=+
tt
;/
tt
on the
path of
first
CO
Writing u
each case arg
J^
iz)
'
= te'^'"^ in
= 0), and
the
u
and third integrals respectively
first
= e'^
in the second,
^\ _ e-"'+'>-n6 ae+W27n
2'7r J
- ^^^
'Itti
which
(2)
=-
"cos {nd
I
- sm
the required result,
is
cW +
6)
arg
lies
between
^tt
'
n, since ./
(s)
= e*'*"'' J,^
\I^COS{7ie + ZSm djde-
"(..;
the upper or lower sign being taken as arg
"When n
make use
is
^n
e^ for t
e-^'-''''''^'
sin UTT
or
<-
z),
we have
^ 6-"^+^^*'"^ V^
/
5> irr
cW,
tt.
Jn{s) =
t)
17 "23 and writing
when arg ^ <

and
e"
j^
^'"^""^^^^
When
r"->
I
Modifying the former of these integrals as in

in the latter, we have at once
J^
(so that in
we have
(B),
Itt.
an integer (A) reduces at once to BessePs integral, and (B) does so when we
{?) = { )'^J-n {z), which is true for integer values of n.
of the equation J^^
"
THE TRANSCENDEXTAL FUNCTIONS
364
Equation (A), as already stated, is due to Schlafli, Math. Ann.

equation (B) was given bv Sonine, Math. Ann. xvi. (1880), p. 14.
These trigonometric integrals
the confluent form (obtained by making ?(-*-x
But Laplace's
integral for this function
=
Tlie limit of the integrand as
of Laplace's integral for
Example
integration,
From
1.
shew
that,
-âc
'A
is
example
iz
the formula
n
is
J^^{.r)
{n~^)>
(f)
e'-cos(<)
g-i^cos*
^^^
Shew
2.
change of order of
^^
e-^'=Vo(.i-sin^).rrf^.
(2),
xv. (1891), p. 121.)
that, with Ferrers' definition of P^"* (cos 6),
P,r (cos 0) =
when n and
dxfi.
> 0,
(Callandreau, Bull, des Set. Math.
Example
cos m(f)
J,,, (z).
^^^
a i)0sitive integer and cos ^
JJ
z^/2n^).
^^^ ^^^^ ^^^ ^^j^ exhibits the similarity
"
= --
P (cos^)=-^^_^-^^
2) /,(2) satisfies
i-*/' (1
+ COS +
to the Bessel-Schlafli integral for
P,,"' (z)
when
( 17-11
of the equation for
^^^J^^^^^^ (J
are positive integers
e-^'^^'
J, {X sin 0) .v-
dx
and cos ^ > 0.
(Hobson, Proc. London Math. Soc. xxv. (1894),
17"24.
We
is
^^
and
a multiple of
is
For
(1871), p. 148,
iii.
may be regarded as corresponding
for the Bessel functions
to Laplace's integrals for the Legendro functions.
[CHAP.'XVII
Bessel functions whose order is half
have seen
( 17-2)
that
when
an odd
p. 49.)
integer.
the order n of a Bessel function Jn {z)
half an odd integer, the difference of the roots of the indicial equation at
is
2w,
which
is
an integer.
We now
shew
that, in such eases,
Jn
{2) is
expressible in terms of elementary functions.
^
For
r
J,
and therefore
(.)
_h
2^îf.
17*211)
r
if
A-
is
^'^'-^'^=
On
z'
z'
^^ + ^^^^
...|
i-fi^zf^'^
Jk + J (^)
Pj^,
Example
Shew
that
t/_.j
/2\*
sm
.,
d^
/sin^N
d^'\^}-
.i
= Pk sin
Q^ are polynomials in ^"

1.
a positive integer
differentiating out the expression on the right,
where
-.
2\i
{z)=i\
cos
-.
z-iirQk cos
we obtain the
z,
result that
17 '24, 17"3]
Example
2.
BESSEL FUNCTIONS
Prove by induction that
if
365
k be an integer and n = k + \, then
(2r-l)! 2'-3 22r-i
%=i
J'
the summations being continued as far as the terms with the vanishing factors
the numerators.
Example
3.
Shew
that s^+^^-r^fe
4_i
Example
4.
Shew
that the sokition of s'"^
^ = S^'-^^ 2
where
(^^)
Cq, Ci, ... C2j
d^m + l y
^.im +
Cp {/_,_, (2a,.^)
are arbitrary and oq,
oi,
...
sohition
a-^m
+3/ =
of
Bessel's
in
equation for
is
+ ^,,^^(2ap2%
are the roots of
(Lommel.)
HankeVs contour integral* for
17'3.
Jni^z).
Consider the integral

r(i+,-i-)
where
is
a point on the right of the point t=l, and
= arg( +
arg(^-l)
at
the contour
may
l)
conveniently be regarded as being in the shape of
a figure of eight.
We
shall
shew that
this integral
is
a constant multiple of Jn{z).
easily seen that the integrand returns to its initial value after
for (^
1)" ~ ^ is multiplied
the circuit (1+) has been described, and

factor e-f^'i-Dft after
the circuit (
(^
,.=0
To evaluate these
functions of n for
all
factor
e'^n-i)
m after
multiplied by the
(2r)!
integrals,
( 4"7)
we observe
firstly
values of n, and secondly that,
the contour into the circles |^ 1|
axis joining the points
is
is
has been described.
converges uniformly on the contour, we have
+ 1)""^
It
has described
l<-lît^-if-l
Since
may deform
by the
t= {1 h) taken
twice,
that they are analytic
when
i^
(w
2)
> 0, we
|i+l| = S and the real

and then we may make
S,
the integrals round the circles tend to zero and, assigning to

* Math. Ann.
i.
(1869), pp. 467-501.
366
and
we
[CHAP. XVII
arguments on the moditied path of integration,

and t- = u,
their appropriate
get, if arg (1 t-)
/(i+.-i-)
^ g( - i) î
'
( 1
f'Y - ^
f/^
- < - ^)
2isin
P (1 -
"/
= 2{sin fn + ^)
f-)"
'""^1
n -r^] TT
t-'il- r-f - i dt
J -1
= - -ii sin in-^^TT

=-
""^
.1
" ^ dt
- v)^~^du
Jo
TT
.^j
T ( +
^V r {n + r + 1).
Since the initial and final expressions are analytic functions of n for
values of n, it follows from 5o that this equation, proved when
all
RCn+f\>0,
is
true for
all
values of
n.
Accordingly
-
^"-0
=
2"+i
(-)' z''+"- 2i sin (
-t-
h)
(2r)\r{7i
+ ^)
sin (^n
Trr
(^
-jrV
(r
+ h)T ( +
h)
+ r + l)'
+ ^j r
(^|)
(0),
on reduction.
Accordingly, when \T (^
Corollary.
n ]>
=^0, lue have
'^n iz)
\;
When
ii(>j
+ ^)>0, we may
-p^.y^
(r-
1)" -
COS {Zt)
dt.
deform the path of integration, and obtain
the result
'-r7^fr(J)/',('-''>""''=''(^'>*
=
Example
Example
1.
2.
Shew
'-,
-,.
that,
when
/?(?j
Jn
= ^,
^%l"
(2)
sin-" d) cos (2 cos d>) d<b.
+ i)>0,
.1,
" e^'^ ""'^ sin2"
<^ c^(^.
Obtain the result
when R{n)>0, by expanding
in
powers of
and integrating
(^ 4-7)
term-by- term.
BESSEL FUNCTIONS
17 '41
Example
[Let
Shew
3.
= {m+^)
IT
when - i <
that
where
is
< ^,
Jn
(s)
367
has an infinite number of real zeros.
zero or a positive integer
then by the corollary above
where
J 2r-l
h-^<0,
so, since
Mto> wi_i> t?,_2>

= is due to
This method of proof for n
and hence
...,
J',i(w7r+^7r) has the sign of (-)'".
Bessel.]
4.
Shew that if n be real, Ju (z) has an infinite number of real zeros
an upjicr limit to the numerically smallest of them.
Example
find
17 '22.]
[Use example 3 combined with

Connexion
17 "4.
We
now
shall
example
17'11
bettoeen Bessel coefficients
establish a resvilt due to

2,
and
and Legendre functions.
Heine* which renders precise the statement of
concerning the expression of Bessel coefficients as limiting forms of
hypergeometric functions.
When
I
arg (1+0)
<
TT,
is
unrestricted and
is
a positive integer,
follows by
it
differentiating the formula of 15-22 that, with Ferrers' definition of Pn^{z),
'"'^' ^"*'^'
^'"^"^=2^^rî^S^Ti)^^"'^''"^^+'^^^^^
and
so, if
arg
z
\
< ^n,
arg
^z^ln^) < w, we
(1
2'"'.m! r(?i
2n''J
have
m+1)
'
in-
Now make ^- + qo {n being positive, but not

8-^0 continuously through positive values.
Then -}
(-
a-l,
by
S 13-6,
and
necessarily integral), so that,
- v^,
and the
series of
if S
...
{\
-{m + rfb'^]
3-32,
which this
is
is
the (r+l)th term

;
is
so,
2'.m
,.=0
(m + 1)
(i
+ 2)
...
{m + r)rl
the relation required.
Example
1.
Shew thatt
lim
%^)]
= '^,n(4
* The apparently different result given in Heine's Kugelfunktionen

between Heine's associated Legeudre function and Ferrers' function.
p.
'
we have
JimJ.-"^/>,n(l-ii)]:
which
= ;i~i,
^-"^
easily seen to converge uniformly in a range of values of 8 including the point 6 =:
by
--l.
Further, the (r + 1 )th term of the hypergeometric series is

{-y {I -mb) {\+^ + mb + rb) {\ -{m + l)n-^\ {\-{m+2f h^]
{on + l) {m-\-'2) ...{m + r).r\
this is a continuous function of S
is
due
to the difference
t The special case of this when m = was given by Mehler, Journal fur Math, lxviii. (1868),
140; see also Math. Ann. v. (1872), pp. 141-144.
368
Example 2. Shew that

by the schemes
defined
\ + ic
ic
ic
\-
We
ic
Asymptotic
\Yhere
have seen
it is
But
by
series for
J^
---I,
\z\ is large.
tt,
.^
(z)
c-
i(c-)
[-hi ^{c + n) n + 1
hi
when
(z)
p'
-i
17-212) that
< arg {'2iz) <
tt
.^
tt.
range of values of z
16*41 example 2,
c^x
making
supposed that arg z\ <
for this
'\
e^p} n
^
[-n f-2ic 2u'-l
the confluence being obtained by

17-5.
the confluent foi-m of the equations
is
IrOx
oc
(0n
Bessel's equation
[CHAP. XVII
if
-J_
3
2 tt
{ei (
< arg (
+ i) -'
TFo,
iiz)
(2i>)
<
''";
and
+ e " H + i)
so,
'^^
when arg z\<7r,

|
Wo,n (-
2i>)}.
(27r^)*
But, for the values of z under consideration, the asymptotic expansion of
Wo,ni2iz)is
,.
.-.
1 1
(4n^-P)(4>i'-3^)
+ (^^1:13
^ ( ly I4n^
r-] i4n^
3-1
{4..
- (2r -
1)^|
^^
(8i2')''
and therefore, combining the series, the asymptotic expansion of J

2;
< tt, is
is large and arg
(z),
when
2;
^(^)~Q'
cos [Z
11
r
''^'^
(-X
'^
{4n^
- P]
{4>i^
- 3^}
. . .
{4h^
- (4r - 1)^}
]
(2r)!2'-2*'-
-,.=1
+ sinf.-^n.-J.)
i (-)M4^^--i;M4^'--3^..-{4n--(4r-3)-i
4
2
(2r - 1) 2'-='^=^'-i
;^.^j
V
!
cos
where Un {z),
Vn {z)
The reader
will
2 ??7r
- ^ TT j
[/ {z)
sin ("z
2 "tt
have been written in place of the
observe that
terminate and give the result of
if
is
- J ttJ
T,, (z)
series.
half an odd integer these series
17-24 example
2.
17 "5, 17*6]
BESSEL FUNCTIONS
369
Even when z is not very large, the value of Jn iz) can be computed with great accuracy
from this formula. Thus, for all positive values of z greater than 8, the first three terms
of the asymptotic expansion give the value of e/o (2) find J^ (z) to six places of decimals.
(for n = 0) and by Jacobi t (for
Complex values of z were considered by
The method of obtaining the expansion here
This asymptotic expansion was given by Poisson*

general integral values of n) for real values of
and several subsequent

due to Barnes .
Hankel;]:
given
is
Asymptotic expansions
Jn
for
writers.
z.
when the order n
(z)
is
by Debye
large have been given
{Math. Ann. Lxvii. (1909), pp. 535-558, Miinchener Sitzangsberichte, XL. (1910), no. 5) and
Nicholson {Phil. Mag. 1907).
An
apj)roximate formula for Jn
a;"
{2nnf
was obtained by Carlini
Celeste v.
supplement
is
< 1,
and 0<.i"
large
(1
- .r2)i {1 + v'(l
'
x'^)Y
also investigated
by Laplace
in 1827 in his
on the hypothesis that
[Oeicvres, v. (1882)]
.v
more extended account of researches on Bessel functions of

Proc. London Math. Soc. (2), xvi. (1917), pp. 150-174.
Example
arg
I
2
I
< ^TT
By suitably modifying
1.
and
(?i
namely
exp {n^{\ x-)}
1817 in a memoir reprinted in Jacobi's Ges. Werke,
in
The formula was
pp. 189-245.
when n
{n.v)
Hankel's contour integral
vil.
Mecanique
purely imaginary.
is
large order is given in
( 17"3),
shew
that,
when
+ 1) > 0,
and deduce the asymptotic expansion of J^
(2)
when
is
large
and arg z
|
< ^n.
[Take the contour to be the rectangle whose corners are 1, \+iN, the rectangle
being indented at 1, and make X-âz
the integrand being (1 i^)**"'^'^.]
;
Example
Shew
2.
= -^^.
Jn{^)
[Write M = 2scot
Example
that,
3.
Ae'~z''
Shew
<^n
and
[*" e-2^cot'<>^^g-i
R {n + h) > 0,
cosec'""'
d)
sin
{z-(n-h)
cb] deb.
in the preceding example.]
that, if
vr^-^-
{l
Jo
is
when args
arg
z\<\n and R
{n
+ 1) > 0,
then
+ iv)''-^ e-^''^ dv + Be-'^ z" f" v"-Hl-ivf~ê--'''dv

Jo
a solution of Bessel's equation.

Further, determine
A and B
so that this
may
represent J^
{z).
(Schafheitlin, Journal fur Math, cxiv.)
The second solution of Bessel's equation ivhen
17'6.
We
have seen in
equation
is
172
that,
when the
the order is
not an integer, the general solution of the equation
aJn
where a and
/3
{z)
an
integer.
order n of Bessel's differential

is
+ ^J-n {Z),
* Journal de VEcole Polytechnique (1), cah. 19 (1823), p. 350.
t Astr. Nach. xxviii. p. 94.

+ Math. Ann. 1. (1869), pp. 467-501.
Trans.
W. M. A.
Camb. Phil. Soc.
xx. (1908), p. 274.
24
370
When, however,
is
an integer, we have seen that
Jn{z)
{-YJ~n{z),
and consequently the two solutions
We
[CHAP. XVII
and
J,, {z)
J"_,i {z)
are not really distinct.
therefore require in this case to find another particular solution of the
equation,
differential
distinct
from
order to have the general
in
{z),
solution.
We
shall
now
consider the function
Jn{z)cosn;^J^^^
Y. (.) = 2.."-sin znir
which
a solution of Bessel's equation
is
introduction of this function Y^ {z)
When
11
is
an integer,
due
is
Y,j {z)
is
when 2n
to
is
not an integer.
The
Hankel *.
defined by the limiting form of this
equation, namely
Imi 27rel'-' ^
^(.) = ,-*o
^. ..
(^) cos (
g-.o
= lim
sm
- J,,^. (.)
+ e)7r
-f e7r)
sin2(?i
Zeir
-' !/+, {Z)
- (-)" J_n-. {Z)\.
(2-) in terms of Wj-^n functions, we have recourse to the

which gives
To express Y^
result of 17'5,
Y(^)=lim _il^r{e^( + ^ + J)-nf,,^,(2t>) + e-^(^^ + ^)-''Tro,n+e(-2i>)}

-*o (27r2)- L
remembering that W^^ m
^^k,
-m
Hence, since f lim W^^n^^ (2iz)
This function
equation
it is
(?(
"H\(2t2^),
being an integer)
is
we have
obviously a solution of Bessel's
called a Bessel function of the second kind.
Another function (also called a function of the second kind) was first used
by Weber, il/a^/i. Ann. vi. (1873), p. 148 and by Schlafli, Ann. di Mat. (2), vi.
it is defined by the equation
(1875), p. 17
;
Yn{z)
= Jn {z) COS nir J^n {z)

sm nir
Math. Ann. i. (1869), p. 472.

t This is most easily seen from the uniformity
Barnes' contour integral
16'4) for
W^
^j^^{2iz).
of
Yn {z) cos mr
ire''
the convergence with regard to
of
BESSEL FUNCTIONS
17'Gl]
371
by the limits of these expressions when n is an integer. This function

which exists for all values of n is taken as the canonical function of the
second kind by Nielsen, Sandbuch der Cylinderfunktionen (Leipzig, 1904),
and formulae involving it are generally (but not always) simpler than the
or
corresponding formulae involving Hankel's function.
The asymptotic expansion for Yn{z), corresponding

that, when arg^' < tt and n is an integer,
Jn
{z), is
(z-^mr-^TTJ
where Un{z) and Vn
leading terms being
Example
where n
1.
2.
(z) are
1
The
To
+ cos
(z
-2 mr-^
(^)
17 '5, their
respectively.
integer after differentiation.
Shew
that
Y (s) be
if
when n
The ascending
an
is
(Hankel.)
defined by the equation of example
1,
it
series for Yniz)-
series of 17*6 is convenient for calculating Y(^)

2^
|
j
the ascending series for
functions* of
may
e,
/^-(n+e) (z)
each term
may be expanded
we have
to
powers of
in
e (
and
this double
53, 5 4).
sum
the coefficients of the
if s
'^
^to r
T (- w- e + r +
be a positive integer or zero and
following expansions in powers of
are valid
'
1)
a negative integer, the
G^r"=(|^)"">.elog(l.) +
.
T'is + l)
r(s + i)^--'[
,
r(s + e +
i)
r(s + i)l
__^^_sin(^^
is
first
in the terms of the series
^=or!r(n + e + r+l)
Now,
when \z\ is large.

we observe that, since
are uniformly convergent series of analytic
then be arranged in powers of
Accordingly, to obtain Yn(z),
power of
where 7
is
integer.
obtain a convenient series for small values of
series
tt
the asymptotic expansions defined in

(4w- l)l8z
and
solution of Bessel's equation
17'61.
.Un{2)
Prove that
made an
is
Example
to that of 17*5 for
Euler's constant ( 12*1).

*
The proof
of this
is left to
the reader.
24
372
Accordingly, picking out the coefficient of
- log 1^, z) ^X^ -^^,_-___ +
(.)
(-XCi^)""*""*
'^
>Zorir{n +
(_)
y^fy^î
5:
we
(-)
see that
^,p^_^_^
,rr!r(-n
+ r+l)J
"i''
7i
=1
(_)r-+ir(_r),
>*=0
and
+ l)^^
e,
[cHAP. XVll
so
Yn(^)= 2 tnif}!!!:'
When
?i
is
= 0,
Karl
(.^.~)
-=^-
2<
r=o
near z
Jo-log
+ 27-
Tm- - S m-
an integer, fundamental solutions* of Bessel's equations, regular

are Jn{z) and Yn{z) or ^n{z).
Neumannf
took as the second solution the function F'"* {z) defined
by the equation
F'-' (z)
.^
Y {z)
+ Jn {z)
but Yn{z) and Y(^) are more useful

Example
1.
Shew
that the function F
F'(.-)
Shew
(log 2
- 7)
for physical applications.

{z)
satisfies
the recurrence formulae
= i{r_i(2)-r,i(^)}.
also that Hankel's function
Y^ (s) and Neumann's
function
}'() {z)
satisfy the
same recurrence formulae.

[These are the same as the recurrence formulae satisfied by J^
Example
2.
Shew
that,
when arg
|
z
\
f"sin(2sin^-^)(^^-
7rF(.-)=
(2).]
< \n,
e"- *'b^{e + (-
|"^
)" e"''^} rf^.
(Schljifli,
Example
3.
Shew
r(o)(0)
17'7.
The
= Jo(5)logs + 2
{Jo.{z)-^J,{z)
+ lMz)-...}.
Bessel functions with purely imaginary argument.
function j
l,{z)
= i-'\h{iz)= 2
^=0
Math. Ann. in.)
that
^^
+ r)l
rl{n
Euler gave a second solution (involving a logarithm) of the equation in the special cases
n = 0, n =
l, Inst.
Calc. Int.
11.
(Petersburg, 1769), pp. 187, 233.
t Theorie der BesseV schen Funktionen (Leipzig, 1867), p. 41.
X This notation was introduced by Basset, fiydrodyuamicfi

defined 7(2) as iJ(i2); see Proc. Camb. Phil. Soc.
vi.
11.
(1888), p. 17; in 1886 he
(1889), p. 11.
had
is
373
BESSEL FUNCTIO^"S
177, 17'71]
of frequent occurrence in various branches of applied mathematics;
these apphcations z
is
The reader should have no
difficulty in obtaining the following
/_i(^)-/,,^,(^)
(ii)
'^^{z'Î,,{z)]=zH^_M-
(iii)
^Jz-'Î,,(z)]=z-^Î,,^,(z).
When
/ (z)
cosh (z cos 6)
^-^=-7
In(z)-
^
^ tt
<
arg z
+ s {-y
<
_ pi
(4^2
the asymptotic expansion of / (z)

{4n2
'-
- 3^
when
arg ^r
seen to be valid over the extended range
arg z
is
{4n2
>-=i
the second series being negligible
qj.
. . .
<
3
.^
tt
2""
^'^
The
"".
<
arg z
<
result
3
v,7r
if
is
easily
we
write
g-(n + j)Tri^ ^^^ Upper or lower sign being taken according as
Modified Bessel functions of the second kind.
When n
is
a positive integer or zero, I_a (^) =In (^)
solution of the modified Bessel equation (iv) of 17'7,

for all values of n
Kn {z) =
so that
The notation
A'
(z)
to obtain a
second
we define* the function
by the equation
Kn (z) =
- (2r - 1)^
positive or negative.
17'71.
Kn (z)
is
r!2'-2''
(27r>)'
gîn + jjiri
sin-'* (bd4),
^-rr,
,,=1
{l-Kzf
rL + 1'j>0,
When
(vi)
formulae
= "^/n(^).
(i)
(v)
in
usually positive.
cos UTT Wo, n (^z),

J
îr [I^n (z)
was used by Basset
In {z)] cot nir.
in 1886, Proc.
denote a function which differed from the function
now
"
Camb. Phil.
Soc. vi. (1889), p. 11, to
defined by the omission of the factor
and Basset's notation has since been used by various writers, notably Macdonald. The
is to make Z,j(~) and A',; (2) satisfy the same recurrence
formulae. Subsequently Basset, Hydrodynamics ii. (1888), p. 19, used the notation K^(z) to
denote a slightly different function, but the latter usage has not been followed by other writers.
The definition of K,^ (z) for integral values of n which is given here is due to Gray and Mathews,
Bessel Functions, p. 68, and is now common (see example 40, p. 384), but the corresponding deficos/iTT,
object of the insertion of the factor
nition for non-integral values has the serious disadvantage that the function vanishes identically
when
2rt is
pp. 130-139
an odd integer. The function was considered by Eiemann, Ann, der Phys. xcv. (1855),
and Hankel, Math. Ann. i. (1869), p. 498.
374
Whether n be an integer or
Bessel equation, and
-^n (^)
~ (|;)
for large vakies of
When
^^
j
a solution of the modified
expansi
possesses the asymptotic expansion
TT it
tt
{4n-
+ S
Kn (2)
Kn (2) = li^n
which gives
;^
is
l'']
{4??''
3^1
. . .
{4n'^-(2r-l)'}'
|.
an integer,
is
when arg ^ <
cos {mr)
'
function
not, this
[CHAP, XVII
is
defined
by the equation
-^n+e (z)]
2 TT {/_n-. (s)
cot
7re,
17'61)
(cf.
^^
r=0^!0^
+ ÔU
^m =
m=l
I'^^Vi N~"+^''(-V'"^(?i-'''-l)!
{o2
+5 2
r=
as
an ascending
Shew
Example.
that
Neumanns
17'8.
'
'
series.
Kn (2)
satisfies the
same recurrence formulae
as /
(2).
expansion * of an analytic function in a series of Bessel
coeffi-cients.
We
shall
analytic in a
now
consider
the
expansion of an arbitrary function f{z),
domain including the
origin, in a sei-ies of Bessel coefficients, in
the form
= a Jo (z) +
f{z)
where
Oo,
otj,
Oo.
Assuming the
o{l/{t-z);
let it
ccjJi {z)
are independent of
J2 iz)
...,
possibility of expansions of this type, let us first consider the expansion
= ^
(^) -^0 iz)
+ 20,(t)J, {z) + 2 O2 (0 Mz)-\-...,
where the functions 0 {t) are independent of
We shall
a.,
be
fTz
is
z.
z.
now determine conditions which On
(0 niust satisfy
to be a uniformly convergent series of analytic functions
if
the series on the right
by these conditions On (t)
and it will then be shewn that, if On (t) is so determined, then the

on the right actually converges to the sum \j{t z) when 2 <| ^ I.
will be determined,
series
/9
8\
we have
0^{t)J^{z) + 2 2 0'(t)Jn(z)+Oo{t)jQ'{z) + 2 2 On{t)Jn'{z) = 0,
n=l
=1
SO that, on replacing
2.7' (2)
by J_i
Ji
K.
Neumann, Journal fur Math,

(3), x. (1893), p.
106.
(*),
we
find
2 {20n'it) + 0^,{t)-0.i{t)}Jn{z) = 0.
W(0 + <^l(0}'/o(2)+
norm. sup.
(2) - 1/ +
lxvii. (1867), p. 310; see also
Kapteyn, Ann. de VKcole
BESSEL FUNCTIONS
17"8, 17'81]
Accordingly the
siuscessioe
fwactions 0\
(t), O-, (i),
O3
375
are deterrained hy the recurrence
(t), ...
forimdae
On^,{t) = 0,,.,{t)-20,:{t),
Odt)=-0o{t),
and, putting
in
the original expansion,
we
see
that Oo{t)
is
be defined hy-the
to
equation
0,{t)
These formulae shew without
We
by induction that 0
shall next prove
when
or
E (t) > 0.
f'
is
ii
so defined,
v/( n^
+1
l/t.
equal to
is
(t)
"]
du
or Oy
d /"
e-^"{ic^(u^+l)]^-'^dtt--j-
(t)
when n
is
equal to
e-'" {?<^/(2
(it
+ l)}"f??
e-'{v/(2<2 + l)}-i{l
whence the induction

Writing
a polynomial of degree n in
obviously equal to Oq
"jo
=h
(t),
(t) is
h
r
On
e- '" [{u + Jiu^ + 1 )}" + {u -
For the expression

and
respectively
= llt.
difficulty that
obvious.
is
= sinh^, we
h
+ 222Kv'(wHl)}t^
see that, according as n is even or odd*,
[{<
+ v/(m2 + !)} + {u - Jiu^ + 1 )}"] =
nd
Î'^Jj
2{2n-2)
4(2n-2) (2-4)
J'
and hence, when R{t)>0, we have on integration,
^W--
^n +
i- |^
+ 2(2,i-2)'^2.4(2?i-2)(2n-4)"'""-J
P
the series terminating with the term in

On{t)
is
defined as a polynomial in
value of On
for all values of
(t)
Shew
Example.
or ^"~' now, whether R{t) be positive or not,

and so the expansion obtained for On{t) is the
;
t.
that, for all values of
On (0 = 2^1
and
l/t
'
f^
t,
e- [{.r + ^/(.rH ^2)|.+ {.^-- V(.;2 + ^2)}n]rf,,,
verify that the expression on the right satisfies the recurrence formulae for On
17'81.
if).
Proof of Neumanns expansion.
The method
of
178 merely
deterrained the coefficients in
Neumann's
^),
on the hypothesis that the expansion existed and that

the rearrangements were legitimate.
expansion of
]/(^
To obtain a proof
Jn
(^)
* Cf.
of the validity of the expansion,
=-^[^ + ^4>
On
it)
^-
Hobson, Plane Trigonometry (1918),
we observe
{1
+ 4>n],
79, 2G4.
that
376
> 0,
where ^
>
(f>,i
as
>*
-^
oo
when
and
[CHAP. XVII
Hence the
are fixed.
series
C\(t)Jo(z)+2 S 0,{t)Jn{z) = F(z,t)

w
is
=l
comparable with the geometrical progression whose general term
and
progression
this
expansion for F{z,
is
is
t)
when
is 2"/^"^+^
and so the
the same circum-
2:|<|^|,
2'34) in
stances.
where r < R, the series is comparable with the

geometrical progression whose general term is r'^jR'^^'^, and so the expansion
for F {z, t) converges uniformly throughout the domains 2^ $r and \t\'^R
by 3'34. Hence, by 53, term-by-term differentiations are permissible,
Again
^r,
if \z\
It]"^ R,
and so
(I
I) ^^^'
"^
^^
^"' ^^^
^^^ "^ ^
'^'
-f-Oo(0/o'(^)+2
^"'
^^^
^" ^^-^
i On{t)Jn{z)
n=l
{Oo'{t)
+ 0,{t)]Jo(z)+ i
{20,/(0 + 0+i(0-0_i(01/(^)
=1
= 0,
by the recurrence formulae.
d
'd
(| + ^-^)^(M)=0,
Since
it
follows that F(z,
t) is
expressible as a function of
.
it is
F (z,
clear that
It
is
t)
F{0,t)=0,{t) =
provided that
<
since
l/t,
= o^ZTTl
i On
(0 ^n
{Z),
=i
Hence, if/(^) be analytic when
\f(t) \0.
J
(t)
\z\
Jo (^)
^r,
we
have,
=1
(On
dt
it) fit) dt,
the paths of integration being the circle
the validity of Neumann's expansion
when \z\<r,
+ 2 i On (t) Jn {Z)\
= Jo (z)f(O) + i '^'^
\z\<r.
therefore proved that
4'7,
^ and
- z).
l/(^
-i = Oo (0 J, (^) + 2
by
\t\
r;
and
when \z\<r and/* (2')
this establishes
is
analytic
when
377
BESSEL FUNCTIONS
17*82]
Example
Shew
1.
that
= Jo (2) - 2./2 (2) + 2-/4 (2) - ...

sin z = 2./i (j) - 2 J3 , + 2/5 (s) -
cos
. . .
Example
Shew
2.
Shew
that,
when
<
Oo(0ô(2) + 2 2 O(0^n(2)=
n=l
=
17*82.
J(s)
2
(t=-x
m=-x
'
<-"-'e-^{^ + v/(^H.2)}"c;'*'
(Kapteyn.)
an arbitrary function in a
SchliJmilch's expansion of
Neumann.)\
series
of Bessel coejicients
zero.
Schlomilch* has given an expansion of quite a
Neumann.
.
xr
|,
Jo
of order
/i-
i^-
>-=o
3.
Neumann.
that
^ (n + 2r).in + r-l)\ ^
,,
(i0)=
2 ^--^ ^ ^
-J7i+2ri^)-
Example
(K.
Any
His result may be stated thus
character from that of
diflFerent
function f{x), which has a continuous differential
fluctuation for all values of x in the closed range
(0, tt),
may
he
coefficient
with limited
expanded in
f {x) = ao + i Ja {x) + aîJ^ {2x) + a^J^^ (3x) +
. .
total
the series
valid in this range; where
ao=f{0)+TT
an=^
Schlomilch's proof
is
u
J
t)
u cos nu
jo"
de du,
{u în B)
f {u sin 6) d6 du
{n
> 0).
J Q
substantially as follows
Let F{x) be the continuous solution of the integral equation
f{x) = -
("^ F{xHm(j))d(t>.
Then (11-81)
F{x)=f(0) + x
irTT
"
I
/'(.rsin^)c^^.
In order to obtain Schlomilch's expansion,
We
theorem to the function i^(xsin<^).

f(x)
=-
-''
F(u) du + - 2
d(j)\I
=TT
/"tt
I
it
merely necessary to apply Fourier's
cos
nu cos {nx sin
(p)
F{u) du
F(u) du +
TT
is
thus have
^
'2
[tt
/
cos
nu F{u)
J(,
(nx) du,
n=l J
the interchange of summation and integration being permissible by 4*7 and 9'44.
* Zeitschrift filr
XLiii. (1912), pp.
Math, und
34-37.
Phijs.
11,
(1857), pp. 137-165.
See Chapman, Quarterly Journal,
378
In this equation, replace F{u) by
f{x)=TT
its
[CHAP. XVII
Thus we have
value in terms of f{u).
'"^
/"l/W + Jo'f (u 8m 0) dd\ du

Jt
+IT
2 /o
"
(?>')
cos
nu
Jo
n=i
/(0)
'"
?<
f ( sin 6)de\ du,
Jo'
['
which gives Schlomilch's expansion.
Shew
Example.
^ .v < tt,
that, if
^' - 2 |jo
is
equal to
.r
but that,
if tt
^
.r
where arc cos
(n-.r-i) is
(.f )
the expression
+ ^ Jo
+ ^ ^0
(3.^-)
(5:i-)
. .
j-
.r ^ 2ir, its value is

+ 277 arc cos (n-.r"^) 2x^(.?.-^ - it-),
and -
taken between
Find the value of the expression when
.v
lies
between 2n and
Stt.
(Math. Trip. 1895.)

17 'S.
Tabulation of Bessel functions.
Hansen used the asymptotic expansion

given in Lommel's Stiidien
ilbcr die-
(^ 17'5) to calculate tables of
Jnix) which are
BesseV schen Fu7iktionen.
Meissel tabulated Jo(x) and Ji {x) to 12 places of decimals from x = Q to x=\b-b {Abk.
der Akad. zu Berlin, 1888), while the British Assoc. Report (1909), p. 33 gives tables by
which Jn
(x)
and F (x) may be calculated when x>\0.
Tables of J^
{x),
Phys. XVIII. (1911),
Jîx),
J_
{x),
J _^
{x) are
tmd
given by Dinnik, Archiv der Math,
p. 337.
Tables of the second solution of Bessel's equation have been given by the following
Mag.
Mag. (6),
writers: B. A. Smith, Messenger, xxvi. (1897), p. 98; Phil.

Aldis, Proc.
Royal
The functions
Soc. Lxvi. (1900), p. 32; Airey, Phil.
/ {x)
have been tabulated
(1893) p. 223, (1896) p. 98, (1907) p. 94
also
(5),
XLV. (1898),
p.
106
xxil. (1911), p. 658.
in the British Assoc. Reports, (1889) p. 28,
by
Aldis, Proc.
Royal
Soc. Lxiv. (1899);
by
Isherwood, Proc. Manchester Lit. and Phil. Soc. xlviii. (1904); and by E. Anding, Sechsstellige
Tafeln der BesseVschen Fxmhtionen imagindren Argumentes (Leipzig, 1911).
Tables of Jn{x\li), a function employed in the theory of alternating currents in wires,
have been given in the British Assoc. Reports, 1889, 1893, 1896 and 1912 by Kelvin, Math,
and Phys. Papers, iii. p. 493; by Aldis, Proc. Royal Soc. Lxvi. (1900), p. 32; and by
;
Savidge, Phil. Mag.
(6),
xix. (1910), p. 49.
Formulae for computing the zeros of J'o {z) were given by Stokes, Camb. Phil. Trans, ix.
and the 40 smallest zeros were tabulated by Willson and Peirce, Bull. American Math.
Soc. III. (1897), p. 153. The roots of an equation involving Bessel functions were computed
by Kalahne, Zeitschrift fur Math, und Phys. Liv. (1907), p. 55.
A number
of tables connected with Bessel functions are given in British Assoc. Reports,
1910-1914, and also by Jahnke imd Emde, Funktionentafeln (Leipzig, 1909).
REFERENCES.
R. LiPSCHiTZ,
Journal fiir Math. LVI. (1859), pp. 189-196.
H. Hankel, Math. Ann.

K.
Neumann,
i.
(1869), pp. 467-501.
Theorie der BesseVschen Funktioneji.
(Leipzig, 1867.)
17 '9]
E,
BESSEL FUNCTIONS
LoMMEL, Studien
379
BesseVschen Funktionen.
iiber die
Math. Ann.
(Leipzig, 1868.)
III. IV.
H. E. Heine, Handhuch der Kugelfunktionen.

R. Olbricht, Studien
(Berlin, 1878.)
Kugel- und Cylinder-funktionen.
iiber die
(Halle, 1887.)
A. SoMMERFELD, Math. Ann. XLVii. (1896), pp. 317-374.

N. Nielsen, Handhuch der Theorie der Cylinderfunktionen.
A.
J.
(Leipzig, 1904.)
Gray and G. B. Mathews, A Treatise on Bessel Functions.

W. Nicholson, Quarterly Journal, xlii. (1911), pp. 216-224.
(Loudon, 1895.)
1.
Shew that
cos
{z
sin 6)
sin
(2
sin d)
= Jq (^) +
i^) cos 28 + 2Ji (z) cos 4^ +
= 2Jj (s) sin ^ + 2 J3 (z) sin 30 + 2J (2) sin 5^ +
'^'^2
. .
...
(K.
2.
By expanding each
example
side of the equations of
Neumann.)
powers of sin
in
6,
express
2**
as a series of Bessel coefficients.
3.
By multiplying
the expansions for exp-^-
considering the terms independent of

{Jo(2)}2 + 2 {Ji
Deduce
4.
)y
and
+2
{J2(2)}2
+2
U3(2)}2 +
...
= 1.
that, for the Bessel coefficients,
when
-]- and exp]--2U
shew that
t,
(2)}2
2 (?
Jo
(2)
(n^l)
lJn(2)i<2-i,
1^1,
2 is real.
J^ (2) = n1
If
i"
2^'
cos* u cos {rnu
when k
(this function reduces to a Bessel coefficient
JJ{z)= 2
where ^-m,k,p
is
du
?<)
zero
is
and
m an
integer),
shew that
^,{iz)PiV_,,,,^^,
the 'Cauchy's number' defined by the equation

/"'T
^-m,k,p = ir
Shew
2 sin
e''"(e^ + e-'')*(e"'-e-')PrfM.
further that
J''{z)=j''~\{z) + j''-\iz),
2j'=+^(2)
and
'
= 2>H./''+V2)-2(/l-+l){./*^m A^-J^m+l
Â^))^
m,
'
'
'
1 ^
'>
(Bourget, Journal de Math.

5.
If V
and
M=E e sin E,
cos
v=-
cos
v=M+ 2 (1 - e2)i
shew that
where Jrn{z)
is
(2), vi.)
M are connected by the equations
defined as in example
E e
e cos E^
where
JJ' {me)
2 2 {hef
"
m=l k=ti
4.
i
I
< 1,
'
m sin mM,
(Bourget.)
380
6.
where
Prove that,
if
= rcosd,
i/-
x^
and
[cHAP. XVII
are integers,
= r-iim- 0,
and
c,,"'
is
independent of
2.
(Math. Trip. 1893.)

7.
where
Shew that the
(f)
and
\//-
solution of
are arbitrary functions of
8.
Shew
that
9.
Shew
that
equation
tlie differential
z, is
(Trinity, 1908.)
M J H=
J
for all values of
(Schlafli,
10.
(m - 1)
Shew
a;'
fi
and
/.
Math. Ami.
that, if
/ + 1
(x)
(-)"r(^+..+2 + i)(|g)'^+''+^-"
is
J_i
iii.
and Schonholzer, Bern dissertation,
(1871), p. 142;
a positive integer and m4-2M +

{x)
dr = .r'^ 1 {./+i
(.r)
/_i
(.*)
l is
1877.)
positive,
A^ (x)} + {m + 1)
P .r'V2
(.r)
rf^
(Math. Trip. 1899.)

11.
Shew that
^3(e).3ô(--)+4^=0.
12.
Shew that
n
J{z)
13.
+ \ -n-h2-
+ 3-...'
Shew that
J-n{z)Jn-x{z)+J-n^v{z)Jn{t) =
2 sin niT
(Lommel.)
ttz
'
14.
If ^":V^^^ be
denoted by Q
(2),
shew that
'^Jniz)
dQni^ ^
dz
1 5.
Shew
that, if
^^ = ,.2 ^ ^.î _
Jo{R)= Jo
2(;^
{>)
+ l)
Qn{z)+z{Qn{z)Y.
2rr^ cos 6
Jo
(r-i)
+2
and
?-,
2 Jn
> r > 0,
(r)
Jn
{ri)
cos
710,
11=1
r,{R)=J,{r) To (7-0 + 2 2 Jn{r) r (î) cos
^.
n=l
16.
Shew
that, if
(K.
Neumann.)
(K.
Neumann.)
R (n -h i) > 0,
p" ./.
(22 cos ^) (/^
=W
{Jn{z)V.
BESSEL FUNCTIONS
Shew how
17.
nomials in
to express z^'^J^ni^)
381
the form AJ2{z) + BJq{z), where J,
iii
B are
poly-
and prove that
Jt (6*) +3,/o (6^) = 0,
(30i)
3.76
+ 5 J, (30*) = 0.
(Math. Trip. 1896.)
Shew
18.
(a2
that,
- ^2)
if
'
a 4=
and n
i3
>
1,
^ J
xJn {ax) Jn {&x) dx = x J {ax)

|
^ {Jn {ax)f dx=(a^x^ - n^)
2a2
(^.^)
{J (ax)}^ +
- J^ (^)
^_ J (a^)l
|r
I'''
Prove that,
1 9.
xJn
if
>-
and
./ (a)
= J
while a
fi
Jn i^x) dx = 0, and
{oix)
(/3)
[J^ {ax))-
/:
Hence prove
that,
n> -1,
when
the roots of
./(.r)
= 0,
^ J
{ax)\'
=t= /3,
dx= i {^ +
(a)}2.
other than zero, are
all real
and
unequal.
be complex, take
[If a could
to be the conjugate complex.]
/3
(Lommel, Studien
iiber
die BesseV schen Funktionen^ p. 69.)
20.
Let x^ f{x) have an absolutely convergent integral in the range
:$.r:$ 1
\ei
be a real constant and let n ^ 0. Then, if /-j, ko, ... denote the positive roots of the equation
;
k--{U^{k) + HJr,{]c)} = 0,
shew that, at any point x for which
<
9*43, /(.r) can be expanded in the form
.r
<
and f{x)
satisfies
one of the conditions of
f{x)= 2 ArJni^r^),
where
-4,.
{Jn{krX)}'^d.i'
In the special case when 11= n,
mining
^"i,
^2)
being J + i
(X')
= 0, and
be taken to be zero, the equation deter-
the
term of the expansion
first
{./
when If
+
is AqX"'
where
.v''*'^f{x)dx.
^r = 2
(k^x) dx.
ki is to
Ao = {2n + 2)
Discuss, in particular, the case
xf (x) J^
'>
is infinite,
so that J^ {k) = 0, shewing that
av)}-- [\rf(x)J{k,x)dx.
J
due to Hobson, Proc. London Math. Soc. (2), vii. (1909), p. 349 see
The formal
also W. H. Young, Proc. London Math. Soc. (2), xviii. (1920), pp. 163-200.
expansion was given with ZT infinite (when w = 0) by Fovu-ier and (for general values of n)
by Lommel proofs were given by Hankel and Schlafli. The formula when IL= n was
given incorrectly by Dini, Serie di Fourier (Pisa, 1880), the term A^x'^ being printed as Jq,
and this error was not corrected by Nielsen. See Bridgeman, Phil. Mag. (6), xvi. (1908),
The expansion is usually called the
p. 947 and Chree, Phil. Mag. (6), xvii. (1909), p. 330.
[This result
is
Fo urier- Bessel expansio n

21.
Prove that,
if
a^
x^ = AiJq (Xj x)
exists as a vuiiformly convergent series
roots of /o(Xa)
= 0,
'
the expansion
+ A2JQ 0^2'V) + ...
when a
^x -â,
where
Xj, X2,
...
are the positive
then
J = 8 {aXVi
(X)}-i.
(Clare, 1900.)
382
If
22.
i'l,
t2,
...
are the positive roots of J (/a)
= 0,
and
[CHAP. XVII
if
r=l
when Oâ; â, then
this series converging uniformly
2a'-'-
dJ,,
(Ka)
da
(Math. Trip. 1906.)
Shew that
23.
when 7i>7n> -
1.
(Sonine, Math. Ann. xvi.)
> 0,
24.
Shew
25.
If i be a positive integer
that, if
o-
(Nicholson, PAj7. Mag.
and w >
g-isinh
0,
(6), xviii. (1909;, p. 6.)
deduce from Bessel's integral formula that
J^ ( ^.) (/^. _ g
- .
sech
u.
(Math. Trip. 1904.)

Pi'ove that,
26.
when x > 0,
J- {x) = IT
sin
(.1-
cosh
dt,
J'o (.r)
[Take the contour of
"^
(.r
cosh
dt.
to be the imaginary axis indented at the origin
17"1
cos
and a
semicircle on the left of this line.]
Shew that
27.
< <
x~^ Jq ixt) sin xdx = \n
=arocosecf
^>1
= ^-^ {1 - (1 - 1-)^
< < ll
= <-'
f>l
and that
.r-i
./i
{xt) sin .r(i.r
/:
J
(Weber, Journal fur Math. LXXV.)

28.
Shew that
w
"
e'"-co8 e
[A
+ 5 log
(?
sin'- 6)}
dO
Jo
is
the solution of
du
dû
ar''
r ar
(Poisson, Journal de VEcole Poly technique, xil. (1823), p. 476
Camh. Phil. Trans,

29.
see also Stokes,
ix. (1856), p. [38].)
Prove that no relation of the form
2 xV,/^,(.r) =
s=o
can exist for rational values of Ng, n and x except relations which are satisfied when the
Bessel functions are replaced by arbitrary solutions of the recurrence formula of 1721 (A),
(Math. Trip. 1901.)
[Express the left-hand side in terms of Jn{^) and Jn + i
+ i {x)lJn{x) is irrational when n and x are rational.]
that J
(x),
and shew by example 12
BESSEL FUNCTIONS
383
Prove that, when R{7i)> -h,
30.
^^^~2"-ir(?i4-i)r(-i-)V
(Hargreave, Phil. Trans. 1848
Shew
31.
C"
Shew
32.
P"/ (z sin
+ 1 > ; >
that, if 271
Macdonald, Proc. London Math. Soc. xxix.)
^) sin"'
+ 6 dd = z-i Jr+i
i
r(7i-im + A
{Weher, Journal fur Math. lxix.
Shew
33.
Math. Trip. 1898.)
that
-=l '^^
(Lommel.
{./,,+i(s)F.
In the equation
34.
d^y
?i
(Hobson.)
(z).
1,
jo
\^
R (in-\-h) > 0,
when
that,
c^^^y
is real
shew that a solution

cos (n log
0)
is
1 rfy
n'^\
/,
given by
(
2
,=1 22m
- )'" z^^ cos (u,n - n log z)

(1
^âî
'
(4
+ 7i2)4
(^m2
+ n^)h
TO
where u^ denotes 2 arc tan
(Math. Trip. 1894.)
(n/r).
r=l
35.
Shew
that,
when n
is
large
and
positive,
J(?i) = 2-t3-^7r-ir(i)-i + o(/t-i).
(Cauchy, Comptes Rendus, xxxvili. (1854),

(6),
36.
Shew
p.
993
Nicholson, Phil. Mag.
XVI. (1908), p. 276.)
that
tJ^{tx)
(Mehler, Journal fur Math, txviii.

37.
Shew
that
e\cose=,2-ir() 2 ( +
/[)
(?" (cos
^)X-"/+i(X).
(Math. Trip. 1900.)
38.
Shew
that, if
^=
a, b, c
being positive, and
W=0
W=
is
Jm
Jm (bx) Ji (ex)
A-l-"' dx,
a positive integer or zero, then
{a-b)^>c^,
ft
m hm
f,m
23m-i^^r(m + |)
WÔ
(*')
(a + 6)2>c2.
{226V -2a*r-^
(a
+ 6)2>c2>(a-&)2,

384
Shew
39.
that, if
n> \,
tn> i and
W=j'j{a.ii.v) t7
a, b, c
[CHAP. XVII
(bx) J,n
{c.v)
.I''
~"' dx,
being positive, then
ir=(27rr*a"'-i6'"-ic-"'(l-;t2)J(2m-l) p^-'j'
where
= {a^ + f^-c^)/2ab,
fi
{a +
^^)
iiy= -
bf>c^>{a-bf,
î.
(Macdonald, Proc. London Math. Soc.
Shew
40.
that,
if i? (wi
/, (z)
and,
if
arg
(2), vii.)
+ ^) > 0,
TT
pr^
cosh
COS
(z
sin-'" (b d(b,
(f>)
< W,
2
I
Prove also that
{z)
= n~-2"'z"'T (m + h) cos
mir
{u'^
+ z'^)~^'^~^
cos udzt.
(Math. Trip. 1898.
[The
by-term.
first
integral
To
may
Camb.
Of. Basset, Proc.
be obtained by expanding in powers of
Phil. Soc. Vi.)
and integrating term-
obtain the second, consider

/(!
s'
>
-1 +
e-''{t^-\r-^dt,
arg (^-1) = arg (4-l) = 0. Take \t\>l on the contour, expand

descending powers of t, and integrate term-by-term. The result is
where
initially
îe^"""' sin
Also, deforming the contour

2ie2'""*'2"'sin2m77
{2mn) T {2m) 2 " ' r
by flattening
it,
_!)-*
j^^
- wi) ^- m (z)-
(1
-\r~^ dt + 2ie-"''''
e-^' {f
(^^
z'^^cosmw
e-'îl -t^)'"-^ dt
and consequently
/-
2i-"'sin(m7r)2"'
41
Shew that 0
(2)
satisfies
the
5'=z~V^
where
diflferential
^ven),
g,i
iy-*^M
equation
= 'i^
'^
{71
(K.Neumann.)
odd).
be analytic throughout the ring-shaped region bounded by the

whose centres are at the origin, establish the expansion
42.
If f{z)
f{z) = haoJo{z) + a,J^{z)
+ iô ^0
0,,= -. I f (t) 0
J c
where
43.
Shew
that,
if
x and y
(--)
/3i
(t) dt,
0,
circles
c,
+ a2Jo{s)+...
(z)
(i
+^. 0,(z) + ...
= ^.
î
(fit)
.1
J{t)dt.
(K.Neumann.)
are positive,
'---jQ{ky}kdk=/:
where
/= +\'(x2-l-?/2)
and /3=
-|-^/(/---l) or
tV(l -^") according as ^> 1 or /< 1.

(Math. Trip. 1905.)
'
BESSEL FUNCTIONS
Shew
44.
on n and on the form of the function /(a:),
that, with suitable restrictions
/ (.r) =[
J,, (tx) t
385
fix-')
theorem
historical account of this important
[A proof with an
It is due to Hankel, but

Cylinderfxmktionen, pp. 360-363.
it is often called the Fourier-Bessel integral.']
If
45.
/"^
C contains
unless
and n are
0,
(Z)
On
7i
in
which case the
be any closed contour, and
but the third
J,
{Z)
Jn
dz=f
{Z)
the origin and
rji
equal to ni (or 2iTi
is
m = 0)
if
if
J,
(z)
is
given by Nielsen'
view of the result of
(in
On
(z)
dz = 0,
two integrals are
first
Shew
still
zero,
the origin once counter-
encircles
clockwise.
46.
9"7)
shew that
integers,
dz=f
(z)
dx\ dL
(taf) x'
(K.
Neumann.)
(K.
Neumann.)
that, if
i-y = a,
~
p' *
/y
P- 9-
and
n be a positive
if
integer, then
Z
while
Z^-^''
47.
It
'
2 Qn-m,
m=l
n+
m-l ^2m-l
\Z)i
= an-i,n-lO(>{z) + 2
2 a _ TO _ i, + , _ Oim
m=l
"W-J^
2-,
(z).
y;;^
>
shew that
(J/2_.^2)-l
when
the. series
Shew
48.
= Oo(y){Jo(.^)F+2
2 Qn(y){Jn{x)r-
on the right converges.
that,
\ic>0,
J (a) J
{b)
(/i)
>-1
(K.
and
= ^. r^"' <-
R{a bf > 0,
exp
{(^2
_ 2 _
Neumann, Math. Ann.
ill.)
then
52)/(20}
/. {abjt)
dt.
^TTl J c i
(Macdonald, Proc. London Math. Soc. xxxil.)
Deduce from example
49.
(2
48, or otherwise prove, that
+ b^-2ab cos d) ~ *" J {(a2 + 52 _ 2ab cos d)^}

=-2''T{n) 2
(m + ?0 ""&"" '^m + n()/.tt + n(&) Cm" (cos ^).
(Gegenbauer, Wiener Sitzungsberichte, LXix. lxxiv.)

50.
Shew
that
y=j^JnAt)Jn{tz^)t'-'dt
satisfies
the equation
di/
dz^
kt" J,
if
resumes
its initial
Deduce
that,
/:
'
\z
'
(t) J,, (tz^)
z-lj
dz
-t"^^
JJ
(t) J,, {tz^}
+ zi
t^
+
1
J,, (0 J,: (tzi)
value after describing the contour.
when 0<2<1,
J..,l,)Jr^, lt^),-*'-r
rf'
= a>-'-"
(1-m
r(y)
-^<'
(Schafheithn, Math. Ann. xxx.
w. M. A.
^'
y-'
Math. Trip. 1903.)
25
CHAPTER
XYIII
THE EQUATIONS OF MATHEMATICAL PHYSICS

The differential equations of mathematical
18"1.
The
pliysics.
functions which have been introduced in the preceding chapters are
of importance in the applications of mathematics to physical investigations.
Such applications are outside the province of this book but most of them
depend essentially on the fact that, by means of these functions, it is possible
;
to construct solutions of certain partial differential equations, of
following are
among
Laplace
(I)
s equation
dx-
in a
memoir* on Saturn's
rings.
be the rectangular coordinates of any point in space, this equation is

in various branches of mathematical
by the following functions which occur
satisfied
physics
z)
dz^
dy'^
which was originally introduced

If {x, y,
which the
the most important
(i)
The
gravitational potential in regions not occupied
(ii)
The
electrostatic potential in a
uniform
by
attractii:ig
dielectric, in the
matter.
theory of electro-
statics.
(iii)
The magnetic
(iv)
The
potential in free aether, in the theory of magnetostatics.
electric potential, in the theory of the steady flow of electric currents in
solid conductoi's.
in
(v)
The temperature,
(vi)
The
in the theory of
thermal equilibrium in
velocity potential at points of a
homogeneous
liquid
solids.
moving
irrotationally,
hydrodynamical problems,
mathematical investiproblem of thermal equilibrium in a

solid when the points of its surface are maintained at given temperatures is mathematically iden.tical with the problem of determining the electric intensity in a region
when the points of its boundary are maintained at given potentials.
Notwithstanding the physical
gations are
(II)
much
the
same
diflferences of these theories, the
for all of
them
thus, the
The equation of ivave motions

dx^
dy'^
dz^
C'
dt-
This equation is of general occurrence in investigations of undidatory disturbances

propagated with velocity c independent of the wave length for example, in the theory of
electric waves and the electro-magnetic theory of light, it is the equation satisfied by each
component of the electric or magnetic vector; in the theory of elastic vibrations, it
and in the theory
is the equati<Mi satisfied by each component of the displacement
of sound, it is the equation satisfied by the velocity potential in a perfect gas.
;
Mem.
de FAcad. des Sciences, 1787 (published 1789), p. 252.
18*1, 18 "2]
The equation of conduction of heat
(III)
This
body
387
is
dfV
dec-
dy^
dfV_ldV
k
dz-
dt
the equation satisfied by the temperature at a point of a homogeneous isotropic
the constant k
proportional to
(IV)
dfV
proportional to the heat conductivity of the body and inversely
is
heat and density.
its specific
particular case of the preceding equation (II),
when the
variable
z is absent, is
dx-
This
of a
dy-
c-
dt'^
the equation satisfied by the displacement in the theory of transverse vibrations
is
membrane
the equation also occurs in the theory of wave motion in two dimensions.
The equation of telegraphy
(V)
This is the equation satisfied by the potential in a telegraph cable when the inductance
X, the capacity K, and the resistance It per unit length are taken into account.
would not be
attempt
an exhaustive account of the theories of these and the other differential
equations of mathematical physics; but, by considering selected 'typical
cases, we shall expound some of the principal methods employed, with
It
possible, within the limits of this chapter, to
special reference to the uses of the transcendental functions.
Boundary
18'2.
A problem
equations of
ditions
which arises very frequently
18*1, of
thus we
conditions.
may
a solution which
is
is
the determination, for one of the
subject to certain boundary con-
desire to find the temperature at
any point inside a
when
homogeneous
isotropic conducting solid in thermal equilibrium
points of
outer surface are maintained at given temperatures.
amounts
surface,
its
the
This
to finding a solution of Laplace's equation at points inside a given
when
the value of the solution at points on the surface
is
given.
more complicated problem of a similar nature occurs in discussing
small oscillations of a liquid in a basin, the liquid being exposed to the
atmosphere
in this
problem we are given,
effectively, the velocity potential
at points of the free surface and the normal derivate of the velocity potential
where the liquid
The nature
is
in contact with the basin.
of the
uniquely, varies very
boundary conditions, necessary
much with
even in the case of equations which, at
Thus a
to
determine a solution
the form of differential equation considered,

first
sight,
seem very much
alike.
solution of the equation
dx-
dy'^
252
388
(which occurs in the problem of thermal equilibrium

cylinder)
in
a conducting
uniquely determined at points inside a closed curve in the
is
by a knowledge of the value of
x7/-ip\a,ne
[CHAP. XVIII
at points on the curve
but
in the case of the equation
da;^
c-
dt-
(which effectively only differs from the former in a change of sign), occurring
in
connexion with transverse vibrations of a stretched string, where
denotes the displacement at time
string, it is physically evident that a solution is
both
dV
-^
and
(where
are given for
all
x from the end of the
at distance
determined uniquely only
values of x such that
if
^x^l, when ^ =
denotes the length of the string).
intuitions
Physical
conditions which
are
will
usually indicate the nature of the boundary
necessary to determine a
solution of a differential
but the existence theorems which are necessary from

the point of view of the pure mathematician are usually very tedious and
equation uniquely
difficult*.
18"3.
general solution of Laplace
equation^.
It is possible to construct a general solution of Laplace's
equation in the
This solution can be employed to solve various
form of a definite integral.
problems involving boundary conditions.
Let
into a
V {x, y, z) be
power
a solution of Laplace's equation which can be expanded
series in three variables valid for points of {x, y, z) sufficiently
near a given point
{Xf^,
yo, z^).
Accordingly we write
= Xq + X,
y=zy^
+ Y,
= Zo+ Z;
and we assume the expansion
V=
c(o
+ a^X +
biY + CiZ +
a..
X- +
b.
Y-
c^ Z^
+ 2cLYZ+2e,ZX + 2f,XY+...,
it
being supposed that this series
is
absolutely convergent whenever
\X'f+\Yr +
\Z'^^â,
If this expansion exists, V is said to

where a is some positive constant
It
can
be
proved
by the methods of 3*7, 4'7
be analytic at (xo, yo,
:]:.
S'o).
* See e.g. Forsyth,
problem
is
Theory of Functions (1918),
216-220, where an apparently simple
discussed.
t Whittaker, Math. Ann. lvii. (1902), p. 333.

* The functions of applied mathematics satisfy this condition.
18-3]
389
that the series converges uniformly throughout the domain indicated and
may
be differentiated term-by-term with regard to X,
Y or Z any
number
of
times at points inside the domain.

If
we
substitute the expansion in Laplace's equation, which
may be
written
d'V
a^F
d^V
dX-'
dY-'
dZ-'
^
'
and equate to zero ( 373) the coefficients of the various powers of X, Y

and Z, we get an infinite set of linear relations between the coefficients,
of which
a.2
may be taken
There are ^n(n
c.
these relations* between the ^{n
l) of
^(n+2)(n +
only
-I-
-^n
1)
2n+l
a linear combination of
(n
1)
2n
-f 1
and cosines of multiples of
u,
may
thus
must be
4-
Y and
iFsin
m)"
Z.
it is
be expanded in a series of sines
Y, Z) COS
mu+ 2
m=
111
gîX,
highest power of
h,,^(X, Y,
Z) sin. mu,
=1
Z) and h^{X, Y, Z) being independent of u. The

gm {X, Y, Z) and A,, (X, F, Z) is Z"'"^ and the former
Y,
in
an even function of Y, the latter an odd function; hence

are linearly independent.
They therefore form a set of
functions
the
2?i
l)
2 gm{X,
is
of degree n in
(Z + iX cos u
find a set of such solutions, consider
a solution of Laplace's equation which
function
linearly independent particular solutions of
Laplace's equation, these solutions being each of degree n in X,
the functions
2){n
independent coefficients in
Hence the terms
the terms of degree n in V.
To
terms of degree n in the expansion of V, so that there
coefficients of the
are
4- 62
as typical.
-f-
functions of the type sought.
Now
by Fourier's rulef
Trgm {X, Y,
Z)=
{X, Y,
Z)=
irhyn
( 9'1'2)
{Z + iX cos
{Z + iX cos
u -H
+ i Fsin
iFsin
u)" cos
mudu,
uY sin mudu,
J 77
* If a^,s,j
"~
d-V
WV
"*"
the coefficient
(where r + s +
d^V
Wr-
dW
"*"
fl^.s.t
dZ^
<loes
hence the relations are

t 27r
= n)
be the coefficient of A''TZ' in V, and

,
.,
arranged primarily in powers of
not occur in any term after
all linearly
must be written
for
tt
X and
Z'-rZ<
independent.
in the coefficient of
g^^
{X, Y, Z).
(or
if
the terms of degree
secondarily in powers of F,
ZT*-^^^
if r
or
1),
and
390
and so any linear combination of the

form
{Z + iX COS
/:
where /(w)
Now
assumed
it is
may
{u) du,
e*".
\X\^
if
+ \Y\-Jr\Z\-
be sufficiently small, and so
write
^ {Z+iX cos
V =\
+ iY sin ?()"/ (w) du.
-TT n-i)
But any expression
F=
of this form
may be
written
F{Z -\-iX cos u + i Y sin u, u) du,
where
i()'^fn
written in this form, the series of terms under the integral
sign converges uniformly

( 4*7) wê
solutions can be written in the
readily verified that, if the terms of degree n in the expression
V be
for
+ iY sin
i(
a rational function of
is
2/;
[CHAP. XVIII
a function such that differentiations with regard to X,
i^ is
under the sign of integration are permissible.
V is
function of this type,
This result
may be
And, conversely,
if
Y or Z
F be
any
a solution of Laplace's equation.
written
J TT
on absorbing the terms
2^,1
*ô cos m I'yo sin
m into the second variable;
and, if differentiations under the sign of integi-ation are permissible, this

gives a general solution of Laplace's equation
of Laplace's equation which
sphere
is
is
that
is
to say, every solution
analytic throughout the interior of
some
expressible by an integral of the form given.
This result
is
the three-dimensional analogue of the theorem that
V=f{x-\-iy)+g{x-ii/)
is
the general solution of

ox^
[NoTE.
distinction has to be
equation and general integrals of a
oy-
drawn between the primitive

i)artial differential
of an ordinary diflFerential
equation of order higher than the
first*.
Two
apparently distinct primitives are always directly transformable into one another
by means of suitable relations between the constants; thus
in the case of ;T"f^+y = 0,
we
can obtain the primitive Csin {x + t) from A cos .r + 5 sin .r by defining C and e by the
equations Csin e = A, C'cos f = B. On the other hand, every solution of Laplace's equation
is
expressible in each of the forms
/:
*
f{x cos < +//

'
For a discussion
Equations,
vi. (1906),
sin
+ iz,
t) dt,
I
'
TT
(y cos
?<
+ sin u + ix,
of general integrals of such equations, see Forsyth,
Ch.
xii.
du
Theory of Differential

18 "31]
but
if
these are
known
relation, connecting
to be the
same
the functions
no general analytical
solution, there appears to be
/ and
g,
which
will directly
391
transform one form of
the solution into the other.]
Example
Shew
1.
that the potential of a particle of unit mass at
["
2tt
at all points for
Example
which
z>
Shew
2.
du
-TT {z
(a, b, c) is
c) + i
{x
a) cos u
+ i{y b) sin u
c.
that a general solution of Laplace's equation of zero degree in
X, y, z is
/
log (.rcos <+j/sin
Express the solutions
Example
Shew
3.
-^
and
+ jj)^
il
log-;
_^
(<)(:/i,
if
g{t)dt = Q.
where
in this form,
r-
= x'^+y'^-\-z^.
p^ + g;^ = x+y
(
where jo = ^, q = ^), integrals of Charpit's subsidiary equations
(see Forsyth, Differential
Equations, Chap, ix.) are

(i)
p^--x=y-q^- = a,
(ii)
p = q + a?.
Deduce that the corresponding general integrals are derived from
= ^,{:c + af+l{y-af+F{a)\
(i)
Q = {x + af-{y-af + F'
(ii)
Az
j'
{a)
= \ (.^+y)3 + 2a2 {x-y)-a^ {x^-y)-^ + G{a)\

j'
= Aa{x-y)-Aa^{x->ry)-'^ + G' [a)
and thence obtain a diflferential equation determining the function

function E{a) when the two general integrals are the same.
18'31.
Xq
we have seen
(z
that
assumed
= y^ ^ Zq ^
we can express F
in 18'3, exists
+ ix cos u + iy sin u)"' cos mudu,
\),
(z
ix cos u
+ iy sin uy^ sin inudu,
J -TT
where n and
If
shall
are integers such that
now examine
we take
^m^n.
these expressions more closely.
polar coordinates, defined by the equations
x= r sin 6 cos
(f),
when
as a series of expressions of the type
J TT
We
in terms of the
Solutions of Laplace s equation involving Leg endive functions.
If an expansion for V, of the form
O (a)
r sin 6 sin
(/>,
= r cos
6,
392
[CHAP. XVIII
we have
{z
I
ix cos u
+ iy sin w)" cos mu du
J ]T
{cos ^
?"
sin 6 cos (
^)j"
cos
mudu
rn-<t>
= ?'"
=
+ i sin
(cos
TT
?'"
^ cos
cos
yjr]"
7?l (</)
+ ylr)
rf-v/r
I^
+ I sin ^ cos -v^}" cos m {(p + ^fr) dy\r
{cos ^
J It
= ?" COS
{cos
m(f>
d + i sin
cos
i^}"^
cos
m'yjrd'yjr,
J -TT
since the integrand
is
a periodic function of -^
(cos ^
is
an odd function of
sin 6 cos
Therefore
yjr.
i/r)"
and
sin
^îi/r
with Ferrers' definition of the
( 15"61),
associated Legendre function,

rir
27ri"'
(z
-^
cos ??m c?ii = 7

+ ix cos u + iyc sin )"
/
(-.^j
^-;'
+ m)
r^Pn'^ (cos ^) cos md).
Similarly
(^
+ iv cos + 1?/ sin k)"

i/
/:
sin
'^
?/it<c?w
=7
(71
Theiêfore
7''^P,i"^
(cos 6) cos 7?i^
and are particular
in X, y, z
18"3, every solution
aiic? r^''Pn'^
^-! r^^Fn^'^ (cos ^) sin md>.
+ m)
(cos ^) sin
7/i</>
are polynomials
solutions of Laplace's equation.
of Laplace's equation, which
Further, by
analytic near the origin,
is
can be exprêssed in the form
F= i
Any
r'^
\AnFn (cos (9) + I
(yl"' cos m<\>
+ Bn^'''^
sin
m^) P"^ (cos 6)1
expression of the form
AnPn (cos ^) + i
(^n<"" cos
?</)
+ 5<'
sin m(f>)
P,r
(cos ^),
m=\
where w
is
a positive integer,
is
harmonic of degree 7i
called a solid harmonic
called a surface
a surface harmonic of degi-ee n multiplied by
?'"
is
(or a spherical Jiarmonic) of degree n.
The curves on a unit sphere (with centre
at the origin)
on which P (cos 6) vanishes

and so P (cos d)
are n parallels of latitude which divide the surface of the sphere into zones,
is called (see S 15*1)
are
n7n
a zonal harmonic
parallels of latitude
and the curves on which
mtb Pn"^ (cos 6) vanishes

.
and 2m meridians, which divide the surface of the sphere

and so these functions are called tesseral
into quadrangles wh(>e angles are right angles,
harmonics.
18 '4]
A
X, y, z

harmonic of degree n
solid
and
is
.evidently a
homogeneous polynomial
393
of degree
in
Laplace's equation.
it satisfies
It is evident that, if a change of rectangular coordinates* is made l;)y rotating the axes
about the origin, a solid harmonic (or a surface harmonic) of degree n transforms into
a solid harmonic (or a suz'face harmonic) of degree n in the new coordinates.
Spherical harmonics were investigated with the aid of Cartesian coordinates by
W. Thomson
Phil. Trans.
in 1862, see
on Natural Philosophy
Treatise
independently in
and Thomson and Tait,
(1863), pp. 573-582,
171-218
they were also investigated

the same manner at about the same time by Clebsch, Journal fur Math.
i.
(1879),
jip.
LXi. (1863), pp. 195-262.
Example.
If coordinates
shew that
r, ^,
are defined
by the equations
1
a; = rcos^,
?/
= (/' 1)- sin ^ cos 0,
= (/'2- 1)- sin ^sin(^,
P,/" (r) P'" (cos 6) cos wi0 satisfies Laplace's equation.
The solution of Laplace's equation which
18'4.
satisfies
assigned boundary
conditions at the surface of a sphere.
We
is
have seen
18"31) that any solution of Laplace's equation which
analytic near the origin can be expanded in the form
w=
+ i
+ 5"'
(^<"' cos m(f)
m=l
and, from
say
3*7,
a, for all
it
evident that
is
values of 6 and
absolutely and uniformly
sin m(f>) P,,"* (cos 6)
(f)
when
if
it
converges for a given value of
such that O^^^tt,

r
<
tt^^^tt,
it
r,
converges
a.
To determine the constants, we must know the boundary conditions

V must satisfy. A boundary condition of frequent occurrence is
that F is a given bounded integrable function of 6 and
say f(0, (f>), on
the surface of a given sphere, which we take to have radius a, and V is
which
(f),
analytic at points inside this sphere.
We
then have to determine the coefficients An,
-4<'"',
5,i""*
from the
equation
f{d,(^)= S a'M^^Pn
rt
(cos
^)+ 2
Assuming that
(^'^' cos
?n</)
+ "'
sin m</))P,;'^ (cos
m=l
this series converges uniformly f
throughout the domain
multiplying by
P' (cos ^) ^%i<i,
sm
Laplace's operator
This
^ +
,
ox^
is
;r
dij-
;^-5 is
6')
invariant for changes of rectangular axes.
oz-
usually the case in physical problems.
394
integrating term-by-term
( 4'7)
and using the results of
the integral properties of Legendre functions,
fid',
Cn
we
m<f>'
sin e'dd'dcf^'
= 7ra -
6') P' (cos d') sin
wt<i)'
sin O'dO'd^'
= -jra"
P,r (cos
6')
6*',
f V(^''
when
Therefore,
F(r,
<
Pn (cos
^') sin e'dB'dcf)'
f7(^',
f-)" f^
i (!i:i^; p^ m (cos
(yx
=-
^ ^"',
-1- 171, "^
5<'',
lira" ^5-?. ^.
a,
?^
e,4>)=l
<l>')
^
9
fir
1551 on
15'14,
find that
cos
4>')
[CHAP. XVIII
f )]p,(cos^)P(cos^')
6) Pn'^ (cos ^') COS
m (6 - d)')} sin O'dO'd^)'.
The series which is here integrated term-by-term converges uniformly

when r <a, since the expression under the integral sign is a bounded
function of
47rF(r,
6, 0',
0)
6*,
Now
(^1'. ^1')
1^/(0',
so ( 4*7
cf>')
(2n
^Pn'"" (cos
+ m)l
^y'^
m=i(n
-h
1)
f-VlPn (cos ^) P (cos ^)
d)Pn''' (cos d') cos 7n(4>-(f>')\ sin e'de'd(f>'.

J
new
polar axis and let
suppose that we
(f>)
be the new coordinates of the line whose old coordinates were
as a
take the line (6,
we consequently have
so we get
47rF(r,
e,
(f>)=r
J -TT
=
If,
and
(f>',
(f),
in this formula,
(p. 332),
to replace
t^ f(0',
Pn (cos
</)')
6)
(2n
by
and
+ l)(-Y
P,{!^ (cos 6)
(6', <^')
by zero
and
Pn(cos0,')sme,'de,'d<f>,'
\^V
W=0
rf(0''<f>')
i (2n+l)(-Y Pn(cose,')sme'dO'd(f>'.
we make use
of the result of
example 23 of Chapter xv
we get
^'
'^^
J-nJo^
^\r'-2arcos6,'+a-)^
and so
47r F(r, e,
</))
a(a-
r')\
J
f"
I
-nJ
f(0',
')
sin e'dd'dd)'
^-
n-
[r- 2ar {cos 6 cos ^' -f sin 6 sin ^' cos ((^ </>)} + a"]^
In this compact formula the Legendre functions have ceased to appear

explicitly.
18 '5]
The
395
For properties
Natural Philosophy.,
formula can be obtained by the theory of Green's functions.
last
of such functions the reader
Thomson and
to
refei'red
is
Tait,
499-519.
From
[Note.
V (r,
the integrals for
6,
we can obtain
of cos^, cos^' respectively,
involving Legendre functions of cos
cf))
new proof
theorem
of the addition
^j'
and
for the
Legendre polynomial.
For
let
Xn
{&',
=Pn
(t>')
- |P
(COS d{)
(cos 6)
+2
and we
get,
(cOS 6')^
^-f; P- (COS 6) P,;" (cos
on comparing the two formulae for V
0=
f'/iff,
J -TT J
<^')
2 (2+i)
{r, 6,
6')
cos ra (0
<^')
(f)),
(-Xxn (d\
4>')
siD e'd&d^,'.
\/
n=0
If we take/(^', 0') to be a surface harmonic of degree n, the term involving r" is the only
one which occurs in the integrated series and in particular, if we take/(^', 4>') = Xn i^'i 0')>
;
we
get
= 0.
'{Xni6',fp')}^^^^'id'de'd(t)'
-n
Since the integrand
Xn{ff,
(f)')
Pn
= 0;
(cos î')
that
continuous and
is
is
not negative
it
must be
zero
and so
we have proved the formula
to say
is
= Pn (cos 6) Pn (COS 6') + 2
2 )
Pn'" (COS
m-1 {n + M)
'-^.
6) Pn"" (COS 6') COS
711
{(j>-
(f)'),
wherein
obvious that
it is
cos
^]'
= cos 6 cos 6' + sin 6 sin 0' cos (0 - <^'),
from geometrical considerations.
We
have thus obtained a physical proof of a theorem proved elsewhere*
( I5"7)
by
purely analytical reasoning.]
Example 1. Find the solution of Laplace's equation analytic inside the sphere /=1
which has the value sin 36 cos <^ at the surface of the sphere.
[^s^r^Pgi (cos 6) cos
Example
in ^, y,
Let
2.
fii{r,
6,
(f))
- irPji
(f)
(cos 0) cos 0.]
be equal to a homogeneous polynomial of degree n
Shew that
z.
l"
J
fnla, G,
(f))
Pn
{cos 6 con
d'
+ siudamd' cos
{(f)
-4>')}
a'^
sin 6 d0dcf>
J o'
-^;^/.(.^'<^)[Take the direction

18*5.
(6', <^')
as a
new
polar axis.]
Solutions of Laplace's equation luhich involve Bessel
particular case of the result of
18'3
Qk(z+ixcosu+iysmu,
is
is
coeffi,cients.
that
gQg f)iuclu
a solution of Laplace's equation, k being any constant and
being any
integer.
*
The absence
now employed
of the factor
)'"
which occurs in
are Ferrers' associated functions.
1-5-7 is
due to the
fact that the functions
396
Taking cylindrical-polar coordinates
=p
cos ^,
{p,
(f),
z) defined
p sin
[CHAP. XVIII
by the equations
4>,
the above solution becomes

^kz
gikpcos iu-4>)
cos
= gkz
( dii
J IT
gikpcos
g^g
.^
(v
(f))
clv
JT
2e*^
6**'"^^ "
cos ?HV cos m<pdv
Jo
2e*^ cos (mcj))
e'*'"^^ "
cos mvdv,
Jo
and
using
so,
17*1
example
we
3,
see that
'Itti^'^ e''^
cos (m(f))
Jm(f^'p) is
solution of Laplace's equation analytic near the origin.
Similarly, from the expression
where
is
an integer,
ive
deduce that
27rz'"
sin
e*^^
(??i</))
/, (kp) is a solution
of Laplace's equation.
18 "SI.
The periods of vibration of a uniform membrane*.
The equation satisfied by the displacement V

plane membrane vibrating harmonically is
cy-
'dx^
where
at time
c- Ct-
of a point {x, y) of a uniform
'
It follows,
The
a constant depending on the tension and density of the membrane.
is
equation can be reduced to Laplace's equation by the change of variable given by
from
18'5, that expressions of
satisfy the equation of
sin
possible type of vibration
F=0 when p = R
a membrane with a fixed
that is to say
is
so that
circular
given by the equation
r=t/, {kp) cos
provided that
motion of the membrane.
Take as a particular case a drum,

boundary of radius R.
Then one
= cti.
the form
sm
'^
m<\i
we have
cos
ckt,
to choose k to satisfy the equation
J,{kR)=0.
This equation to determine / has an
^1, ^2j ^3j
S3,y.
r=/,
This
is
infinite
possible type of vibration

(f-'rp)
a periodic motion with period
number
is
of real roots ( 17'3
cosmcf) cos ckrt
1iTJ{ckj.)
example
3),
then given by
{r=
1, 2, 3, ,..).
and so the calculation of the periods
depends essentially on calculating the zeros of Bessel
coefficients (see ^ 17'9).
Novi Covim. Acad. Petrop. x. (1764) [published 1766], pp. 243-260; Poisson, Mem.
viii. (1829), pp. 357-570; Bourget, Ann. de I'Ecole nvrm.sup. iii. (1866), pp. 55-95.
For a detailed discussion of vibrations of membranes, see also Rayleigh, Theory of Sound,
Euler,
de I'Academie,
Chapter
ix.
18 "5 1-1 8 '61]
The equation
Example.
OZ
pendicularly to the axis
of
motion of
dV
is
finding the zeros of
18'6.
may
It
boundary condition
that the determination of the free periods depends on
J,,/ (C) == 0-
general solution of the equation of wave motions.
183
that a general solution of
of wave motions
dx^
F=
'
dt-
be shewn* by the methods of
the equation
IS
c'
dy'^
If the cylinder have radius R, the
the velocity potential.
=0 when p = R. Shew
that n
a circular cylinder vibrating per-
in
of the cylinder is
dx^
V denoting
air
397
f(x sin u cos
where /is a function
dz-
c-
+ y sin u
sin v
dy''
di?
+ z cos
+ ct,
u, v)
dudv,
(of three variables) of the type considered in 18"3.
Regarding an integral as a limit of a sum, we see that a physical

is that the velocity potential V is produced
by a number of plane waves, the disturbance represented by the element
interpretation of this equation
f(x sin u
cos v
+ y sin
u sin v
+ z cos u +
being propagated in the direction (sin u cos
v,
ct, u,
sin u sin
v)
8u Bv
cos u) with velocity
v,
The solution therefore represents an aggregate of plane waves

all directions with velocity
c.
travelling in
c.
Solutions of the equation of wave motions which involve Bessel
18'61.
functions.
We
shall
now
obtain a class of particular solutions of the equation of
wave motions, useful
for the solution of certain special problems.
In physical investigations,
means
we should
desirable to have the time occurring
it is
of a factor sin ckt or cos ckt,
where k
is
by
This suggests that
constant.
consider solutions of the type

/tt
y =,
Physically this
rn
I
-TT
giA;(a;sinttcost;+2/sinMsiuv+2cosM+cti
f (u
i)\
dudv
means that we consider motions
in
which
all
the elementary waves
have the same period.
Now
let
the polar coordinates of
(x, y, z)
be
(r, 6,
(f>)
polar coordinates of the direction (u, v) referred
the polar axis
OZ;
is
the direction
(f)),
and the plane
let (o),
yjr)
be the
new axes such

-v/r
that
passes through
so that
cos
CO
= cos
sin
(ff,
and
to
II
+ sin d sin u cos

v) = sin w sin
6 cos u
sin
((f)
See the paper previously cited, Math. Ann.
matics, XXXVI. (1907), pp. 98-106.
((f)
v),
yjr.
lvii. (1902), pp.
342-345, or Messenger of Mathe-
398
Also, take the arbitrary function /(w, y) to be
denotes a surface harmonic in
Sn(u,
where
18*31) Sn
We
of degree
u, v
= Sn
V)
{0, 4>]
a surface harmonic in
is
?i
CO,
co,
-sjr
(w, y) sin
;S,j
so that
[CHAP. XVIII
we may
i/,
where Sn
write
yjr),
of degree
n.
thus get
V=
e^'^'^f
' e''""""
Sn
w,
{6, <^\
sin
y\r)
a>
dw df.
Now we may write ( 18 "SI)

0},y}r) = An{e,(f>). Pn (COS Oj)
Sn (d,
(i>;
+ S
{^"'" (0,
<f>)
cos myfr
+ Bn'"'^
{0,
m=l
where
An {6,
</>),
and 5^""'
^n'"'* {6, <p)
{6,
27re''^'^
An
27re'^'-^
A n {0,(ii)j
e''"-'^
27r.-^'*
An
e'*^'^
{0, 4>)
J'^
by Rodrigues' formula
^= o;r^ ^'""^n {0,
'It
-yjr,
P,,
4>)
to)
so
F is
sin codco
-^^ (f.^ -
l)n d^,
a constant multiple of
ti
times and using
e'>"-^^
(1
- /x^)" c^yu
= (27r)^^ iV'*^'^ {kr) -^-J + ^ (kr) An (6,

and
co.
the equation
{ikrY I
J
and
(/i) djM
we obtain
yjr
co),
we get
loll); on integrating by parts
Hankel's integral ( 17 '3 corollary),
Zi
to
f"e^'*"^"P (cos
Jo
^, (^)
sin myfr] P"' (cos
0) are independent of
Performing the integration with respect
F=
(f))
c}>),
e'^'^h-' ^ Jj^_^_i(kr)
An(0,
(f>).
Now
the equation of wave motions is unaffected if we multiply x, y, z

by the same constant factor, i.e. if we multiply r and t by the same
constant factor leaving Q and ^ unaltered so that An{B, <f)) may be taken
to be independent of the arbitrary constant k which multiplies r and t.
and
Hence lim
e**-'''
r~
H' ~ " ~ ^ /
A:-*0
of
i
^
(kr)
An
{0,
9)
is
wave motions; and therefore r'Ân(0, <f>) is a solution (independent of t)

wave motions, and is consequently a solution of Laplace's
of the equation of
equation
it is,
accordingly, permissible to take
harmonic of degree n
ris
and
so
Jn+r
^2
we
ih-) Pn''' (cos 0)

^
An
(0,
<f>)
obtain the result that
^
sm
md>
^
^ ckt
sm
a particular solution of the equation of wave motions.
to be
any surface
I8"61lj
Application of 18-61
18*611.
The
a physical problem.
to
wave motions may be used
solution just obtained for the equation of
manner
following
399
in the
to determine the periods of free vibration of air contained in a rigid
sphere.
The
velocity potential
condition
is
that
dV
-^=-
the equation of wave motions and the boundary
T' satisfies
=0 when
= ,
where a
,i.r,
,.(/-/) ^,^cos<9)
Y-r ~\*J"+"
-,
gives a possible motion
if
Z*
/iv
Hence
the radius of the sphere.

^
is
or
COS
.
sm
COS
sin
wd)
ckt
so chosen that
is
on using
This equation determines k;
17'24,
we
see that
it
may
be written in
the form
tan ka =/, {ka)^
where f^ (ka) is a rational function of ka.

In particular the radial vibrations, in which V is independent of 6 and
taking n = 0; then the equation to determine k becomes simply
tan ka = ka
and the pitches of the fundamental
<^,
are given by
radial vibrations correspond to the roots of this
equation.
REFERENCES.
Fourier, La
J.
W. Thomson and P. G.
Lord Rayleigh, Theory
F. PoCKELS,
(Translated by A. Freeman.)
theorie analytique de la Chaleur.
Tait, Natural Philosophy.
of Sound.
(1879.)
(London, 1894-1896.)
Uber die partielle Diferentialgleichung /\u + khi. = 0.
H. BuRKHARDT, Eiitivickelungen nach oscillirenden Funktionen.

H. Bateman, Electrical and Optical Wave-motion.
E. T.
Whittaker, History of
the Theories of Aether
(Leipzig, 1891.)
(Leipzig, 1908.)
(1915.)
and
Electricity.
(Dublin, 1910.)
A. E. H. Love, Proc. London Math. Soc. xxx. (1899), pp. 308-321.
H. Bateman, Proc. London Math.
Soc. (2),
Magazine
L. N. G. FiLON, Philosophical
H. Bateman, Proc. London Math. Soc.
(6),
(2),
(1904), pp. 451-458.
i.
vi.
vii.
(1903), pp. 193-213.
(1909), pp. 70-89.
1.
and
if
If
be a solution of Laplace's equation which is symmetrical with respect to OZ,

if f{^\ be a function which is analytic in a domain of
y=f{z) on OZ, shew that
values (which contains the origin) of the complex variable
'=i
^
f,
then
r^-{^ + i(.r2+j/2)5cos0}rf(^
/
at any point of a certain three-dimensional region.
Deduce that the potential of a uniform

the plane
XO T
with
its
circular ring of radius c
centre at the origin
^^
7!"
r [^^ + +
{^
(^'
is
+ '/)^ cos (/)}2] - I d<t>.
and of mass
M lying in
400
2.
(p,
(f>,
If
r be
z)
are cylindrical coordinates, and
a solution of Laplace's equation, which
p"'e""*/(i) near the axis of
where /(f)
z,
of the form e""*jP(p,
is
2),
where
solution is approximately equal to
this
if
[CHAP. XVIII
of the character described in example
is
1,
shew that
^= r (i+
If
3.
1 )
^^' "^ ^^
(*)
''''^
'^
^'"""
.r, y and z by means

A.v+By + Cz=\,
be determined as a function of
where A, B,
are functions of
shew that (subject
to
(DougaU.)
^'^^-
of the equation
such that
?<
certain general
conditions) any function of
is
a solution of
Laplace's equation.
(Forsyth, Messenger, xxvii. (1898), pp. 99-118.)
A,
4.
are two points outside a sphere whose centre
matter on the surface of the sphere

the formula
is
such that
C.
is
layer of attracting
surface density (Tp at
its
P is
given by
apCc{AP.BP)-\
Shew
that the total quantity of matter
CA CB and ACB
unaffected by varying
is
A and B
so long as
and prove that this result is equivalent to the theorem

that the surface integral of two harmonics of different degrees taken over the sphere
are unaltered
IS zero.
(Sylvester, Phil.
V (.r,
Let
5.
of masses
X + iy.,
Shew
respectively.
if
the point
and
(.r,
Mag.
y,
final values of
V {x,
that
z)
y, z) is infinite at all points of
6.
a certain real
circle,
and
describes a circuit intertwined once with this circle the initial
V{x,y,
z)
are numerically equal, but opposite in sign.

(Appell, Math.
it
(1876), pp. 291-307.)
(5), il.
be the potential function defined analytically as due to particles

X - z/x at the points (a + ia', b + ib', c + ic') and (a ia', b - ib', c ic')
y, z)
Find the solution of Laplace's equation analytic
Ann. xxx.
(1887), pp. 155-156.)
in the region for
which a<r<iA,
being given that on the spheres /=a and r = A the solution reduces to
2 cP(cos(9),
2 C;P(cos<9),
respectively.
Let 0' have coordinates
7.
(0, 0, c),
Pdz=6,
and
let
PO =
P0'Z=6',
r,
PO' = r'.
Shew that
P{cos6')
_ P{cosd
Y"
according as
.s
~^
"^
r>c
At a
point
whose density
(r, 6, (p)
is p,
shew that the potential

;\^_
3/-
ni^
P^ + ^
{cos d)
C""^^
a--b- and
(Trinity, 1893.)
^).
outside a uniform oblate spheroid whose semi-axes are a, b and
)a-b
where
c''
or r<c.
Obtain a similar expansion for /"P' (cos

8.
+ 2)
+ 2) r^'P,{oo,d) T...|,
1)^-^1 (^"^)'^ {n + l){n

pn + 2
2!
(n
f
+ l^V"^^/
Un
I
jn + 1) Qi
cPnî(cos^)
?>?..
is
m^ Pj (cos 0)
3.5
r^
Obtain the
m^ P^ (cos 6)
b.l
-...].
iôtential at points for
which r<m.
(St John's, 1899.)
401
Shew that
9.
eirco,e=(^^)i
in [271
n=
+ 1) r-lP {cos 6) J+i(r).

^
(Bauer, Journal fiir Math, lvi.)
Shew that
10*.
if
x iy = h cosh (| + irj), the equation
and
in the coordinates ^
wave motions
of two-dimensional
is
+ 5^ = ^ (h- 1 - cos2 -^
X {c + r cos 6) cos (p, y = {c + r cos 6) sin cp, 2 = /-sin^;
(Lame.)
rj)
a|7
IL
Let
respectively are constant form an orthogonal

shew that the surfaces for which
d,
.system; and shew that Laplace's equation in the coordinates /, &,
is
/,
-^r(c
vr
^
\
(f)
+ ?-cos(9)-;^[+
+ '/-cos^)^^^+-;^{(c
^^7., =0.
cd \
r 06
ac + rcosdd(f>j
'
\j-
(W. D. Xiven, J/essenger,
P have
plane POZ meet
Let
12.
the
Cartesian coordinates
the circle
x'^
t/'
{x,
= k-, s =
aPy = (o,
and polar coordinates

the points a, y; and let
z)
?/,
in
log (Pa/Py)
Shew that Laplace's equation in the coordinates

8
da-
[cosh
and shew
bV]
sinho-
o-
cos a da
that a solution
sinho-
d^ j
8w
o)
6,
x.)
Let
(f)).
= a.
is
a>, (p
o-,
_ 871
ca (cosh a cos
(;,
sinh
o-
(cosh
cr
cos w)
^-08(^2
'
is
V= (cosh cos 0))^ cos

o"
cos ??i0
/iQ)
(cosh
o-).
(Hicks, Phil. Trans. CLXXii. p. 617 et seq.)
Shew that
13.
^-hnî
00
{R^ + p^-2Rpcoscl)
+ c-)-h=
and deduce an expression

14.
Shew
that
2
m=0
dk
e-"" J,
if a, b, c
{I- p)e'^'^''"'
cos rauclu,
J -T
for the potential of a particle in
are constants and
the roots of the equation in
X,
/x,
terms of Bessel functions.
are confocal coordinates, defined as
a2
then Laplace's equation
r-jI
may
+ e'^6-^ + e'^
6-2
+6
'
be written
Ax(M-)^xf^^}+^M(-X)|{A/J}+A.(X-M)^{A.^]=0,
where
A;,
= V{(aHX)
(i'^
+ X)
(c-^
+ X)}.
(Lame.)
Examples
VEcole Polyt.
10, 11, 12
and 14 are most
easily proved
by using Lame's
result (Journal de
23 (1834), pp. 191-288) that if (X, /ul, v) be orthogonal coordinates for

which the line-element is given by the formula {dx)'' + (8yf + {5zy^=(H^d\y^ + (H2diM)'- + {H^5v)'^,
xiv. cahier
Laplace's equation in these coordinates

d
fH.H^cV\
d\ V Hi
simple method (due to
a,\
is
/ H.^H,
dfjL\
H.,
dV\
/ H,H.,
'Hi-
d/jiJ^di'\Hs
W. Thomson, Camh. Math.
Journal,
dV\_
dv
iv. (1845),
pp. 83-42) of proving
by means of arguments of a physical character, is reproduced by Lamb, HydroAnalytical proofs, based on Lame's proof, are given by Bertrand,
dynamics (1916), 111.
this result,
Traite de Calcul Differentielle (1864), pp. 181-187, and Goursat, Coiirs d'Analyse,
pp. 155-159, the last proof being appreciably the simplest.
Theorie der Kugelfunctionen,
W. M. A.
i.
Another proof
is
i.
(1910),
given by Heine,
(1878), pp. 303-306.
26
402
Shew
15.
wave motions
that a general solution of the equation of
r=
{.v
J IT
coH 6 + ^
iiin
$+
iz,
>/
[CHAP. XVIII
+ iz siu 6 + ct cos B,
B)dd.
(Bateman, Proc. London Math. Soc.

r=/(.r, y,
If
16.
cx'
prove that another solution of the equation
Shew
17.
cos
(f>,
ct
e,
+p
is
/a
-{ z + ct cos d\ ,
^
F (a,
r
arc suih
(1
.'
^,
6)
-TT
p sin ^
'
18.
,.
dBda,
are cylindrical coordinates and a, h are arbitrary constants.
If r=/(.r, y,
z) is
(2)
i.
(1904), p. 458.)
a solution of Laplace's equation, shew that
1
/ r^-a^
~(^_,^)i-' V2(.'--i'
r^+gg
Y^
is
az
î{x-iyy x-iy)
^
another solution.
(Bateman, Proc. London Math.

19.
If
U=f{x,
another solution
y,
z,
t)
is
Soc. (2) vii. (1909), p. 77.)
a solution of the equation of wave motions, shew that
is
rj^
1
.( X
z-ct-'Xz-ct''
y
z-ct"
_r2-l
rHl \
^{z-cty 2c{z-ct))'

20.
l=x-iy,
If
\ = j:+iy,
m = z + hv,
= z-uv,
fji
so that
n=
+
p= \,
.^:^-
y'^
(2) vii. (1909), p. 77.)
+ + iv%
z^-
+ 7nfi + nv=0,
l\
shew that any homogeneous
solution, of degree zero, of
cHT
cH[
dH/
dx'^
dy^
cz^
cÛ^Q
dw^
^+Z+^^=0dndv
dmdfi
satisfies
dldX
and obtain a solution
where
is
'

"^
when the motion
dO
sin 6)
/""
f'
J
p,
(1904), p. 457.)
<^, is
+
where
l.
'
dz-
cy'^
that a general solution of the equation of wave motions,
/ (2 + ip
(2)
be a solution of
z, t)
d^ ct
independent of
is
^^
of this equation in the
= (^-c)
(f-a),.
?/Li
'
form
= (c-a)
j',
b,
U',
^',
y'
(f-6),
np = (a-
b)
((-c).
(2) vii. (1909),
pp. 78-82.)

21*.
If
(r,
jr
where
6,
(f))
=c
x, y, z are
are spheroidal coordinates, defined
(r^
by the equations
+ iy^sin ^ cos<^, y=c (r2-|-l)^sin^ sin(^,
rectangular coordinates and
c is
403
z=crcosd,
when n and
a constant, shew that,
m are
integers,
('"
/.^cos^
Pn\
j -n-
i2\
+ y sin^-f
^^]
cos
/Sin
(n
^P,r{tr)Pn'"
in)\^
mtdt = 27ry^
(Blades, Proc.
22.
With the notation
/^
-r
Qn
/cPcos^
\
of example 21,
sin^ +i'2\ cos

+
^-^y
^
c
)
J Sin
shew
,
intdt
that, if 2
,.
"
are
cos
^
^
md).
sin
Soc. xxxiii.)
0,
(n m)l^
' {ir)
(n+m) ^"
=,^277 )---fj
The functions introduced in examples 21 and 22
{cos 6)
^
Edinburgh Math.
=t=
(Jeffery, Proc.
*
{n+m)l
-'
-r,
"
Edinburgh Math.
known
^^
P' cos e)
'
cos
m4>.
sin
Soc. xxxiil.)
as internal
and external
spheroidal harmonics respectively.
262
CHAPTER XIX
MATHIEU FUNCTIONS
differential equation of Mathieu.
The
19"1.
The preceding
chapters have been occupied with the discussion of
five
functions which belong to what
may be
generally described as the hyper-
now
well
In the present chapter we enter upon a region of Analysis which

beyond this, and which is, as yet, only very imperfectly explored.
lies
geometric type, and
many
simple properties of these functions are
known.
The
Mathematical Physics and which come
functions which occur in
next in order of complication to

called
Mathieu functions
functions
of hypergeometric
known
these functions are also
They
associated ivith the elliptic cylinder.
arise
type
are
as the functions
from the equation of two-
dimensional wave motion, namely
dy^
dx^
This partial
magnetic waves
c-
dt'
difterential equation occurs in the theory of the propagation of electro;
if
the electric vector in the wave-front
is parallel
to
OZ and
if
E denotes
the electric force, while {Hx, ffy, 0) are the components of magnetic force, Maxwell's
fundamental equations are
lo^_8^_a^
~
<? dt
denoting the velocity of light
dx
;
ly
dHy_dE
c_H^__dE
'
at
dij
ct
dx
'
and these equations give at once
'
c^ ct^
cx'^
cy'^
In the case of the scattering of waves, propagated parallel to OX, incident on an

elliptic cylinder
condition
is
that
The same
for
which
OX
and
OY
are axes of a pi-incipal section, the boundary
should vanish at the surface of the cylinder.
partial differential equation occurs in connexion with the vibrations of
a uniform plane membrane, the dependent variable being the displacement perpendiculaito the
membrane
if
the
the boundary condition
The
is
membrane be
in the
shape of an
ellipse
was discussed by Mathieu* in 1868
with the problem of vibrations of an
elliptic
membrane
manner
*
with a rigid boundary,
the same as in the electromagnetic problem just discussed.
Journal de Math.
(2), xiii.
(1868), p. 137.
in
connexion
in the following
MA.THIEU FUXCTIOXS
19*1]
Suppose that the membrane, which

in equilibrium, is vibrating
V= u
(cc,
Then,
p.
y) cos {pt
XOY
plane
in the
is
with frequency
405
if
when
it
is
we write
e),
the equation becomes

d'û
Let the
foci of
real variables*
^,
i]
+ iy =
curves, on
which ^ or
The
differential equation for
af-2
if
sinh | sin
r].
and tt < ?? $ tt, to each

and only onef value of (^, ??).
we take ^ ^
u transforms
into;]:
+ a^ + -^ (cosh- f - cos- 7?) u = 0.
solution of this equation of the form
where the
and introduce new
+ irj),
=h
of the plane corresponds one
we assume a
h, 0, 0),
constant, are evidently ellipses or hyper-
is
r]
bolas confocal with the boundary

(w, y, 0)
li
c-
h cosh (^
x = h cosh ^ cos t},
The
p-
^
+ ^ = 0.
defined by the complex equation
so that
If
dy^
membrane be (+
the elliptic
cc
point
dû
^-,
^-,
cw
= F{^)G{v),
factors are functions of ^ only
and of
t)
only respectively, we see
that
d'J(g)
Ay
)_
d'GM
i^jf
Since the left-hand side contains ^ but not ?/, while the right-hand side
ri but not ^,
{^) and G (tj) must be such that each side is a constant,
contains
say, since |
We
and
tj
are independent variables.
thus arrive at the equations
^) +
By
('i!%osl.f-^)l-(f) =
0,
a slight change of independent variable in the former equation,
we
see
that both of these equations are linear differential equations, of the second
order, of the
form
-T-; -f
They
(a
-1-
65 cos 2z) u
= 0,
The iutroduction of these variables is due to Lame, who called ^ the thermometric parameter.
See Lame, Sur les fonctions inverses des
are more usually known as confocal coordinates.
transcendantes,
1'"''^
Le(?on.
be seen most easily by considering the ellipses obtained by giving f various

If the ellipse be drawn through a definite point (t, v) ot the plane, r? is the
positive values.
+ This
may
eccentric angle of that point on the ellipse.
proof of this result, due to Lame,
is
given in
numerous text-books
see p. 401, footnote.
[CHAP. XIX
406
where a and q are constants*.

cepted)
This
It
is
obvious that every point (infinity ex-
a regular point of this equation.
is
the equation which
is
circumstances
( 19'2),
known
is
as Jfathieu's equation and, in certain
particular solutions of
are called Mathieu functions.
it
The form of the solution of Mathieu^ s equation.
1911.
In the physical problems which suggested Mathieu's equation, the constant
a
It
not given a priori, and we have to consider
is
that u
a one-valued function of
(x, y) is
by increasing
by
77
27r
a,
terms of
in
when a has not one
19*41) that,
and
jôsition,
and the condition f
determine a set of values of
it is
to be determined.
q.
G (r) +
And
is
membrane
consequently unaltered
lif)
= G {rj)
it will
is sufficient
appear
to
later ( 19'4,
of these values, the equation
G{'n
is
how
obvious from physical considerations in the problem of the
is
2'rr)=G (v)
no longer true.
When
F{^)
that
is
thus determined, q (and thence p) is determined by the fact

on the boundary; and so the periods of the free vibrations of
membrane
the
are obtained.
Other problems of Mathematical Phy.sics which involve Mathieu functions in their

(i) Tidal waves in a cylindrical vessel with an elliptic boundary, (ii) Certain
solution are
forms of steady vortex motion in an
The equation
in a metal cylinder^.
is
The decay of magnetic force

problem of Rigid Dynamics which
elliptic cylinder, (iii)
also occurs in a
of general interest g.
19'12.
Hill's equation.
but of a more general nature,

method of determining the motion of the Lunar
Adams'^ determination of the motion of the Lunar Node.
differential equation, similar to Mathieu's
arises in G.
Perigee,
and
W.
in
HiU's]]
Hill's equation is
^' +
The theory
k+
of Hill's equation
i ^cos2^) = 0.
is
very similar to that of Mathieu's (in spite
of the increase in generality due to the presence of the infinite series), so the
two equations
*
will, to
some
extent, be considered together.
Their actual values are a = A - h-p-l{2c-), q = h-2)-l(S2c-); the factor 16
is
inserted to avoid
powers of 2 in the solution.

t
An
and only
elementary analogue of this result

if,
is
* K. C. Maclaurin, Trans.
A.
II
W. Young,
Acta Math.
is
that a solution of -j-^ +aii =
has period 2v
if,
the square of an integer.
Camb. Phil.
Soc. xvii. p. 41.
Proc. Edinburgh Math. Soc. xxxii. p. 81.
viii.
(1S86).
Hill's
U.S.A.
H Monthly Notices R.A.S. xxxviii.
memoir was
p. 43.
originally published in 1877 at Cambridge,
MATHIEU FUNCTIONS
19"l]-19"2l]
In the astronomical applications
problem of choosing them

does not
fact,
The
arise.
ô> î>
in such a
are
way that the
407
known
constants, so the
solution
solution of Hill's equation in the
may be
periodic
Lunar Theory
is,
in
not periodic.
Periodic solutions of Mathieus equation.
19'2.
We have seen that in physical (as distinguished from astronomical)

problems the constant a in Mathieu's equation has to be chosen to be such
a function of q that the equation possesses a periodic solution.
Let this solution be

integral function of
G (z)
then
Three
z.
G {z),
in addition to being periodic, is
possibilities arise as to the nature of
G (z) may be an even function of z, (ii) G {z) may

G (z) may be neither even nor odd.
(i)
an
G {z)
be an odd function of
z,
(iii)
In case
is
^ {G
(iii),
(z)
+ G {- z)]
an even periodic solution and
^{G{z)-G{-z)]
is
an odd periodic solution of Mathieu's equation, these two solutions forming

It is therefore sufficient to confine our attention to
a fundamental system.
periodic solutions of Mathieu's equation which are either even or odd.
and
solutions,
these only, will
These
be called 31 athieu functions.
z=
are
and 1,
two odd) periodic solutions of Mathieu's equation cannot form a fundamental
system. But, so far, there seems to be no reason why Mathieu's equation, for special
values of a and q, should not have one even and one odd periodic solution
for comparatively small values of 5' it can be seen [ 19*3 example 2, (ii) and (iii)] that Mathieu's
equation has two periodic solutions only in the trivial case in which q = 0; but for larger
values o{ \q\ there may be pairs of periodic solutions, though no such pairs have, as
It will
two even
be observed that, suice the roots of the indicial equation at
(or
'
yet,
been discovered.
19"21.
An
It will
now be shewn
integral equation satisfied hy even Mathieu functions*.

that, if
G {r})
is
any even Mathieu function, then
G(r)) satisfies the homogeneous integral equation
G(v)=-\
where k
\/(32^).
equation given in
*
This result
[" e''''^''''^G(e)dO,
is
suggested by the solution of Laplace's
1 83.
This integral equation and the expansions of
Int. Congress of
Math. 1912.
The
Trans. Camb. Phil. Soc. xxi. (1912),
integral equation
p. 193.
19"3
were published by Wbittaker, Proc.
was known
to
him
as early as 1904;
see
[CHAP. XIX
408
For,
if
A'
+ {i/ = h
cosh,
i^+iri) and
if
G (v)
F{^) and
are solutions of the
differential equations
^d^ - (^ +
(^'_^M
then, by 191,
F {^) G
If this solution
is
+ (A +
nvJf- cos-
7;)
F(^) =
{v)
0,
= 0,
a particular solution of Laplace's equation.
is
{rj) e"^^^
a special case of the general solution
/(// cosh
given in
^^^f^ cosh^ I)
^ cos
cos ^
77
+h
sinh ^ sin
sin 6
7;
+ iz,
6) dO,
natural to expect that*
18"3, it is
f(v,e)^F{0)e'^'<f>(e),
where
<f>
(6) is a function of 6 to be determined.
F{^)G (v) e""' =1

.
Thus
exp {mh cosh ^ cos v cos 6
F(0)(f) (6)
IT
+ mh sinh ^ sin rjsmO + miz] dd.

independent, we may put ^ = 0; and we are thus led to
Since ^ and tj are

consider the possibility of Mathieu's equation possessing a solution of the
form
J 77
Proof that
19-22.
It
and
is
the even
Mathieu functions satisfy
readily verified ( 5-31) that, if
(f>
(6)
"
G (77) =
e'"''
cosncose
G (77)
j-^ +
be analytic in the range (-
tt, tt)
be defined by the equation
if 6^(7;)
then
the integral equation.
is
(A
^ (^) ^0^
an even periodic integral function of
7;
and
+ m-h- cos- 7;) G (7;)
drf-
["
J
=-
{?u-/(2 (sin-
77
cos-
+ cos^ 77) - mh cos
77
cos ^
+ ^}
g'"'' cos
TV
n^H/i sin e cos
{<^" (6')
77</)
(^)
+ (^ +
</)'
{6)]
ni'h'' cos'^
^)
e'''coscos6
(/)
(^)}
e"'''cos.,cosfl
^/^^
on integrating by parts.
*
The constant
F (0)
is
inserted to simplify the algebra.
>,
cose
^^
(^^
^q
MATHIEU FUNCTIONS
19 "22, 19 '3]
But
if(f){6) be
])eriodic function {ivith
(6)
(f)"
both the integral
by the
and
<f)
is
stants
</>
part vanish
{6)
= 0,
that
is
to say,
G (rj),
defined
integral, is a periodic sohition of Mathieu's equation.
Consequently
(0)
period 27r) such that
+ {A+ m^h- cos^ 6)
the integrated
409
and therefore
[In the case
we have
G (tj)
is
an even periodic solution of Mathieu's equation
if
a periodic solution of Mathieu's equation formed with the same con-
(p {d)
when
(ft
(6) is a constant multiple of
G (6)
let it
the Mathieu equation has two periodic sokitions,
= XG (d) + Gi {6)
where
6-'i
{6) is
an odd periodic function
be \G(6).
this case exist,
if
but
^mh cosv cos eg^^g^^^

r
vanishes, so the subsequent
If
we take a and
A and
mh,
We
it is
work
is
unaffected.]
q as the parameters of the
mh =
obvious that
have thus proved
\/{S2q)
that, if
Mathieu equation instead of
= k.
G{'r])
be an even periodic solution of
Mathieu's equation, then

G{r])
which
is
\r
e''-'^^^'^'^G(0)de,
19'21.
the result stated in
From 11 "23, it is known that this integral equation has a solution

when X has one of the characteristic values.' It will be shewn in 19-3
'
for
such values of
\,
only
that
the integral equation affords a simple means of con-
structing the even Mathieu functions.
Example
1.
Shew
that the odd Mathieu functions satisfy the integral equation

G{i])
Example
Shew
2.
=\
sin
(/
sin
r]
sin 6)
G (d)
dd.
that both the even and the odd Mathieu functions satisfy the
integral equation
G{,j)
Example
3.
Shew
that
= x["
when the
e^'''''''^^^^G{d)de.
eccentricity of the fundamental ellipse tends to zero,
the confluent form of the integral equation for the even Mathieu functions
J,,(.r)
is
/" e'-*' cose cos ^ d6.

-
ZTTi" j
19"3.
We
The construction of Mathieu functions.

shall
now make
Mathieu functions
use of the integral equation of
19*21 to construct
the canonical form of Mathieu's equation will be taken as

-T^
4-
(o
6g cos ^z) u
= 0.
410
[CHAP. XIX
In the special case when q is zero, the periodic sohitions are obtained by
a= n-, where n is any integer; the solutions are then
taking
cos^',
cos
sin
sin ^z,
1,
The Mathieu
z,
which reduce
functions,
precise,
functions cen{z,
q), sen i^,
Let us now construct

Since
q)
and
see that A,-*.(27r)~^ as
Oo,
oixler n.
Accordingly
--* 0.
(y
the characteristic value of
q,
= 1 + ttig + Oioq- +
=
l+
q)
qî {z) + q-ô (^) +
(27r\)-i
ce^ (z,
Oj,
The
be unity.
sen {z, q) to
Mathie it functions of
X which
we
gives
and that
where
ce^ (z, q).
suppose that, for general values of

rise to ce^ {z, q)
the coefficients of cos nz and sin nz
{z,
Q) will be called
0)=1, we
ceo(3',
be called
will
seo (z, q), ....
we take
in the respective Fourier series for ce
when q-^0,
ce. {z, q),
{z, q),
se?!
...,
to these
cei (z, q),
ceo {z, q),
To make the functions
22^,
. .
are numerical constants and ^x{z),
...
(z), ...
13.2
are periodic
functions of z which are independent of q and which contain no constant
term.
On
substituting in the integral equation,

(1
T"
(1
Equating
oc,q+a,q-
+ \/(32g)
{l+q/3,
...)
cos ^ cos ^
coefficients of successive
use of the fact that î(z),
/SjC^),
,.
..
we
find that
(z) -i-q'^,{z)
16r/cos-^'cos-^
...}
...}
powers of q in this result and making

contain no constant term, we find in
succession
Oj
= 4,
/3i
(z)
oTo
/Sa
(z)
14,
= 4 cos 2z,
= 2 cos 4^,
and we thus obtain the following expansion
29
-77
cco (z, q)
+ Uq - 28q^ + "-^
+ U'f-'^
the terms not written
q'
cos 6z
down being
(q^)
5'
of a
is
-S2q- +
224>q'
(^
r/
as ^ -*
210
The value
+ i2q--
...jcos2z
.
.
160
q^+
N
...j
cos
4iz
cos 8^
0.
99
l^f/+0(r/);
that the coefficient of cos 2z in the series for
C6?o(2,
q)
is
it will
ai(8q).
be observed
19 'SI]
MATHIEU FUNCTIONS
411
The Mathieu functions of higher order may be obtained in a similar

manner from the same integral equation and from the integral equation of
The consideration of the convergence of the series thus
19"22 example 1.
obtained
postponed
is
Example
Obtain the following expansions*:
1.
(i)
ce,
(n)
cei(2, o)
(.,
to 19"61.
s,
_1+
^,
J_^
(ni)
(iv)
se, {z, q)
= sm .+
q)
C<?2 (^,
= cos5+
2(2
+
^^^\y^^î^~
I"
r
\
r=i i(?'+l)!?*!
2 {(^j-yy^
+ g3 + ()
(r
m
+ 1)!
,
.)}
CO, 2..,
^.
"ttv.
(r+1)!
/
+ (.+ 1)
^5-)
(,/
'^r+l^qr + l
gV
oc
-.
(.+ 1)
COS 22
where, in each case, the constant implied in the symbol
depends on r but not on
z.
(Whittaker.)
Example
2.
Shew
(iii) sfij (2, 5'), (iv) ce2 {z,
that the values of a associated with

210
(i)
Example
19*31.
Since
3.
Shew
-32(?2
+ 224(?4
2Q
__
+
,^6
ceo(5,
</),
(ii)
cei{z, q),
((?),
(ii)
l-8q-^^^ + ^^-\qi + 0{q%
(iii)
(iv)
4+
that, if
+ 8j-8j2_823-^<?* + 0(j5
^j2-^g* + 0(?).
(Mathieu.)
n be an integer,
Tê integral formulae for

all
(i)
q) are respectively
the
Mathieu functions.
the Mathieu functions satisfy a homogeneous integral equation
with a symmetrical nucleus
.'
19*22 example
3), it
follows ( 11'61) that
cem
{z,
q) cen (z,
q)dz
(m
n),
sem
(z,
q) sen
(z,
q)dz
{m
i=
n)
cem
{z,
q) sen
(z,
q)dz =
- jr
0.
The leading terms
of these series, as given in
were obtained by Mathieu.
example 4
at the
end of the chapter
(p.
427),
[CHAP. XIX
THE TRAXSCENDENTAL FUNCTIONS
412
Example
Obtain expansions of the form
1.
cos
(ii)
(/
sin z sin ^)
2 Bce (z,
q) ce^ {6, q),
?l=0
sin (/ sin
(iii)
sin
6)= 2 Cse
q) se {6, q),
(2,
where i=,^{32q).
Example
2.
)!
as a confluent form of expansions
(ii)
and
= 3C
(iii)
of
example
The nature of tJie solution of Mathieu
19"4.
1.
general equation
Floquet's
theory.
We
now
shall
discuss the nature of the solution of Mathieu's equation
when the parameter a

solutions; this
is
no longer restricted so as to give
is
the case which
rise to periodic
of importance in astronomical problems, as
is
distinguished from other ph^-sical applications of the theory.
The method is applicable to any linear equation with periodic coefficients

which are one-valued functions of the independent variable the nature of
the general solution of particular equations of this type has long been perceived by astronomers, by inference from the circumstances in which the
;
These inferences have been confirmed by the following

analytical investigation which was published in 1883 by Floquet*.
equations
arise.
Let g{z), h
(or,
(z)
be a fundamental system of solutions of Mathieu's equation
indeed, of any linear equation in which the coefficients have period 27r)
then,
if
F{z) be any other integral of such an equation, we must have
F{z)=Ag(z)+Bh(z),
where
and
are definite constants.
Since g {z+ 27r), h (z -f 27r) are obviously solutions of the equationf, they
can be expressed in terms of the continuations of g (z) and Ji (z) by equations
of the type
g(z +
where
27r)
ttj, a.>, /S^,
/?._>
a,h
(z),
{z
Itt)
= ^,g {z) + 0,h (z),
are definite constants; and then
F{z +
*
a^g (z)
27r)
Ann. de VEcole norm.
sitj).
(Aa^
+ B^,) g {z) +
(2), xii.
B/3,)
(z).
Floquet's analysis
is
a natural sequel
with doubly-periodic coefficients
20-1),
(1883), p. 47.
to Picard's theory of differential equations
{Aoi.,
and
to the
theory of the fundamental equation due to Fuchs and Hamburger.

t
These solutions
may
not be identical with
equation with periodic coefficients

is
a solution of
-r
dz
(1
^
+ cot ^)
'
1/
is
= 0.
(j(z),
h(z) respectively, as the solution of an
not necessarily periodic.
To
take a simple case, H
= e^sin
MATHIEU FUNCTIONS
19-4, 19'41]
Consequently
F{z + 27r)= kF(z), where k
418
a constant*, if A and
is
are
chosen so that
A a, + BI3, = hA, A a, + B/3o = kB.

These equations
only
have a solution, other than
will
oc,-k,
ofo
and
A = B = 0,
and
if,
if,
i{
/5o
=0;
A.-
k be taken to be either root of this equation, the function F(2) can be
constructed so as to be a solution of the differential equation such that
F(z+27r) = kF(z).
Defining
fi
by the equation
k=
e-'"^
and writing
(^{z) for e~'^^F{z),
we
see
that
(f>(z
Hence
e'^^ (f){z),
We
the
where
+ 27r) = -''''+-''>
F{z+-2tt)=(^
(z).
equation has a particular solution of

a periodic function with period 27r.
differential
(f)(z) is
the
form
have seen that in physical problems, the jjarameters involved in the

have to be so chosen that k = l is a root of the quadratic,
and a solution is periodic. In general, however,

which the parameters are given, A- ^^ 1 and there
in astronomical problems, in
is
no periodic solution.
In the particular case of Mathieu's general equation or Hill's equation, a
fundamental system of solutions f

equation is unaltered by writing
Mathieu's general equation is then
u
where
Ci, c,
Example.
Cie'*-(f)
then
is
^r
{z)
for ^
ao
Now
q.
=0
,3i
)io
are independent of the particular pair of solutions,
Hill's
a definite function of a and
/i is
roots of the equation
a,-k,
19"41.
e~>^^^{z), since the
+ Coe~i^^(^ ( z),
are arbitrary constants, and
Shew that the
e'^-(f)(z),
so that the complete solution of
{z)
and h
(z),
chosen.
method of solution.
that the general functional character of the solution of equations
with periodic coefficients has been found by Floquet's theory,

expected that the determination of an explicit expression
for
it
might be
the solutions of
Mathieu's and Hill's equations would be a comparatively easy matter
this
For example, in the particular case of Mathieu's

has to be obtained in the form
a
solution
equation,
general
however
is
not the case.
y
*
The symbol
with the constant

t
The
k
A;
is
= e^^(f)
(z),
used in this particular sense only in this section.
of 19-21,
which was associated with the parameter
ratio of these solutions is not even periodic
still less is it
It
must not be confused
q of Mathieu's equation.
a constant.
414
where
(z) is periodic
<f)
crux of the problem

of
(f)
and /i is a function of the parameters a and q. The

when this is done, the determination
determine
to
is
/j. ;
(2) presents comparatively
The
successful
first
Hill in the
no more
memoir
difficult
[CHAP. XIX
little difficulty.
method of attacking the problem was published by

method for Hill's equation is
cited in 19'12; since the
than
Mathieu's general equation,
for the special case of
we
shall discuss the case of Hill's equation, viz.
where
J (z)
is
an even function of z with period
Two
tt.
cases are of interest,
the analysis being the same in each
The astronomical
(I)
when
case
real and, for real values of z,
is
J (z)
J(z)
00
the coefficients 6n are
+ 2^1 cos
known
2z
202 cos
4>z
constants and
26^ cos 6^
. .
6n converges absolutely.
n=0
The
(II)
case
when
^ is a
complex variable and J (2) is analytic in a

whose sides are parallel to the
strip of the plane (containing the real axis),
The expansion
real axis.
of J(z) in the Fourier series 6^
+2
6n cos 2nz
"Z
H=l
is
then valid
9"11) throughout the interior of the strip, and, as before,
00
On converges absolutely.
=o
Defining ^_ to be equal to 6n, we assume

00
71= -OC
as a solution of Hill's equation.

[In case (II) this
is
the solution analytic in the strip
have to be shewn ultimately
which
will
(see the note at the
be determined are such as to make
(>^
end of
10-2, 19'4)
19*42)
in case (I) it will
that the values of 6
n-bn absolutely convergent, in order to
n=
justify the processes which
On
we
shall
now
carry out.]
substitution in the equation,
M= 00
we
find
\n=-x
\?j= 00
Multiplying out the absolutely convergent series and equating coefficients

of powers of e^' to zero ( 9"6-9"632),
(,i
+ 2niyb+
eX_,,
we obtain the system

(n
...,
of equations
-2,-1,0,1,2,
...).
MATHIEU FUNCTIONS
19-42]
If
we eliminate the
typical equation
by
coefficients bn determinantally (after dividing the
6o 4^n"
minantal equation
(^>+4)^-
415
to secure convergence)
we obtain*
Hill's deter-
416
the definition of an infinite determinant
From
Now,
(i)
sin ^TT (?>
A (i^) = - A, (i^)
and so
the determinant A,
if
that Ai
(ifi) is
follows that
( 2'8) it
Vô) sin ^tt (i>
+ \/Oo)
'
smM*7rV^)
be written out in
{i/j,)
an even periodic function of
an analytic function (cf. 2'81, 8"34, 5'3) of

poles), which tends to unity as the real part of
yu,
now we choose the constant
full, it is
easy to see
with period 2tV(ii) that Aj
/x
is
If
[CHAP. XIX
(except at
/j,
its
tends to
so that the function
oo
{{//,)
obvious simple
.
(/a),
defined by
the equation
(ijl)
= Ai
(i"/u.)
-K
has no pole at the point

function of
yu.,
it
[cot îr
= i ^6^,
ij,
D (/x)
follows that
{ifi,
function
(/j.)
which has no poles,
then, since
(ifj,
{/j,)
is
- Vô)},
an even periodic
Vô)
is
therefore a periodic function of
and which
obviously bounded as
is
conditions postulated in Liouville's theorem

is
- cot i TT
any integer.
is
The
\/0(,)
has no pole at any of the points
2ni
where n
a constant
making
//,-
go
we
/n
(with period 2^)
(//.)
^+
5'63) are satisfied,
see that this constant
is
and so
The
D (fi)
unity.
Therefore
Ai
and
(ifi)
1+K {cot Itt
{ifi
\/d,)
- cot
TT {ifi
-h
- V(9o)},
so
{rfi)
sinl7r(t>Vô)sin^7r(i>+\/ô)
^
sinM^-Vô)
To determine K, put
/a
A(0)
wi ia\
-^' '"' ^^" ^^"^" o7,
then
= l + 2/i:cot(i7rv/ô)-
Hence, on subtraction,
A(, =
which
the result stated.
is
The
MO)-|^li^
roots of Hill's determinantal equation are therefore the roots of the
equation
sin^ {^-nifi)
When
in
fj,
terms of
equation
is
=A
(0)
sin^ (^tt Vô)-
has thus been determined, the coefficients bn can be determined

b^,
and cofactors of
complete.
A {ifi);
and the solution of
Hill's differential
19 '5, 19 '51]
MATHIEU FUNCTIOXS
[In case (I) of 19*41, the convergence of 2 6

|
of 2-82
for 22 6
is
equal to
6o
ni= x>
Cj, o
follows from the rearrangement theorem
-^
I
417
ô, o
where C^, n
is
the cofactor of B^,
in Ai (ifi.)', and 2 Ci,o is the determinant obtained by replacing the elements of the row
through the origin by numbers whose moduli are bounded.]
|
It was shewn by Hill that, for the purposes of his astronomical problem, a remarkably
good approximation to the value of
could be obtained by considering only the three
central rows and columns of his determinant.
fj.
The Lindemann-Stieltjes' theory of Mathieus general equation.
19'5.
Up
the present, Mathieu's equation has been treated as a linear
to
Some extremely
differential equation with periodic coefficients.
properties of the equation have been obtained by
interesting
Lindemann* by the sub-
^=cos^^, Avhich transforms the equation into an equation with

namely
stitution
rational coefficients,
4^(1
- O ^, +
- 20
2 (1
^+
(a
-I6q + 32^0
^^
= 0.
This equation, though it somewhat resembles the hypergeometric equation,

type than the equations dealt with in Chapters xiv and xvi, inasmuch as
and
regular singularities at
and an irregular singularity
at
is
of higher
it
has two
whereas the three
singularities of the hypergeometric equation are all regular, while tlie equation for TFj._^(3)
has one irregular singularity and only one regular singularity.
We
now
shall
give a short account of Linderaann's analysis, with some
modifications due to Stieltjesf.
Lindemann' s form of Floquet's theorem.
19"51.
Since Mathieu's equation (in Lindemann's form) has singularities at
^=
and
the exponents at each being
1,
0,
M=0
W=0
2/,o
= i
>i
the
first
two
series
- ^y\
an (1
û
When
00
- 0* i
^n (1
- KT;
M=
converge when
the ^-plane
is
and
<
^j
1,
the last two
when
cut along the real axis from
to
^|
<
1.
+ x and from
so relations of the form
Vw
will exist
(1
the four functions defined by these series are one-valued in the
cut plane
=
]
to
^=
\, there exist solutions of the form
ay 00
+ ^Voi
Vn
73/00
î/oi
throughout the cut plane.
Now suppose
that ^ describes a closed circuit round the origin, so that the

circuit crosses the cut from oo to
the analytic continuation of 3/10 is
;
Math. Ann.
xxii. (1883), p. 117.
t Astr. Nach. cix. (1884), cols. 145-152, 261-266. The analysis is very similar to that
employed by Hermite in his lectures at the Ecole Polytechnique in 1872-1873 [Oeuvres, iii.
(Paris, 1912), pp. 118-122] in
W. M. A.
connexion with Lame's equation.
See
23"7.
27
THE TRAXSCENDEXTAL FLECTIONS
418
oj/oo
/3yoi
(since
unaffected by the description of the circuit, but
is
l/^^
[CHAP. XIX
changes sign) and the continuation of j/u
is 7^00
"c? so Ay^^-
^j/oi
+ By^{-
?/oi
will
he unaffected by the description of the circuit if
A
i.e.
{ay,o
+ ySyoi)' + B {yy^ + SyoO' s A
{ay^
- /3?/oi)' + B (73/00 - Sôi)-,
Aa^+ ByS = 0.
if
Also Ay^f-h Byii- obviously has not a branch-point at f=l, and so, if
or 1, and, as it has no
Aal3 + By8 0, this function has no branch-points at
other possible singularities in the finite part of the plane,
it
must
be
an
is
an
^.
The two expressions

A^y,o
are consequently two
iB^-yn,
z) is
iB^'l/u
whose product
^.
( 19"4)
the product of
that
a periodic integral function of
The determination of
1952.
solutions of Mathieu's equation
[This amounts to the fact

e~'^~^
^-j/io
ef"^
(}>
(z)
and
z.']
the integral function associated with Mathieu's
general equation.
The
integral function F(z)
mined without
difficulty
Ayô"
for, if ^jo
+ By^^^,
just introduced, can be deter-
and y^ are any solutions of
^:+p(n|+Q(r)=o,
their squares (and consequently
satisfy the
any linear combination of their squares)
equation*
^! + 3P (D ^^ + [P' (0 + 4Q (0 + 2
in the case
under consideration,
[P (or^]^
this result reduces to
+ (a-l-l6q + S2q^) ^^J^
-\-
16qF (f) =
0.
Let the Maclaurin series
for F{!^)
be
c,i^";
obtain the recurrence formula for the coefficients
on substitution, we easily
c,
namely
where
"""
*
(n
-\- ly - a
ieq(2n + l)
I) {(n
__
16q]
^''"
'
Appell, Comptes Rendus, xci. (1880), pp. 211-214
cf.
n (n
-h
l)(2n
S2q{2n-1)
example
10, p.
l)
"
298 supra.
MATHIEU FUNCTIONS
19-52, 19-53]
At
first sight, it
419
appears from the recurrence formula that
be chosen arbitrarily, and the remaining coefficients
C2,
C3,
and
Co
terms of them; but the third order equation has a singularity at t=

the series thus obtained would have only unit radius of convergence.
necessary to choose the value of the ratio
verge for
all
values of
The recurrence
so that the series
Ci/Cq
when written
W2
V,
Uu+'^
W^jô
1"
lim
(n,
1*/,,
+ ^^
nn+i
on the right can be
fraction
+ m)IK (n +
w,i/r (n, n
where
It is
may
con-
in the form
suggests the consideration of the infinite continued
The continued
^nd
1)
^.
formula,
n+i
can
Ci
calculated in
...
+ m) =
1,
firaction
...+
-sNTitten*
+ m),
u -1
The
limit of this, as
type (by the example of
??i
-^
2'82)
Uni->
is
a convergent determinant of von Koch's
and since
Vr+i
as n ^- 00
llrUr+l
it is
easily seen that
Therefore,
(n,
x) ^1
Cn
if
Cn+1
then Cn
satisfies
is
obvious that
7i
-*
the recurrence formula and, since Cnî/Cn ^-
resulting series for

it is
x
Un K {n, X )
K{n + 1, co)'
as
F (^) is an integral function.
all
the coefficients c are
sufficiently large.
The
From
finite, since
as
?i
^-
the
the recurrence formula
they are
finite
construction of the integral function
when n
F (^)
has
therefore been effected.

19"53.
If Wi
The solution of Mathieus equation in terms of F(^).
and Wo be two particular solutions of
g+P(r)|+(3(f)"=o,
thenf
* Sylvester, Phil.
W^Wi
Mag.
(4), v.
t Abel, Journal fiir Math.
ii.
W1IV2r=cexp|-j^^P(r)fzr}>
(1S53), p.
446 [Math. Papers,
(1827), p. 22.
i.
p. 609].
Primes denote differentiations with regard
97
to
f.
[CHAP. XIX
THE TRANSCENDENTAL FDNCTIONS
420
Taking iv^ and w^ to be those two solutions of

Mathieu's general equation whose product is -P(^), we have
where
a definite constant.
is
w^
w.^
W,
Wo
w/
f^(l- t)*i^(0'
'"''
^r(^)
îO'
w,'
^^'2
the latter following at once from the equation tvîu.^Fi^).
Solving these equations for
iv^lic,
and
tuJ/wo,
and then integrating, we at
once get
where
y..
71,
by taking
From
Cq
are constants of integration; obviously no real generality
71
72
is lost
1-
the former result we have, for small values of
we have aJao = â+
while, in the notation of 1^"51,
Hence
C^
This equation determines
I69
^|,
Sq.
c^.
in terms of a, q
and
the value of
Cj,
being
Ci
K(l, cc)^{uoK(0, x)}.

Example
periodic,
If
1.
the solutions of Mathieu's equation be e'^'^^(p{z), where
</>
(s)
is
shew that
Example
Shew
2.
that the zeros oi F{C) are
all
simple, unless (7=0.

(Stieltjes.)
[If
F{() could have a re^jeated
19"6.
So
zero, v^
and
ivo
would then have an essential singularity.]
second method of constructing the Mathieu function.
far, it
has been assumed that
in the expressions for cey(2, q)

sheton that ce^
{z,
q)
and scy
all
the various series of
and sey(z,
{z,
q) are convergent.
19'3 involved
It will noiv be
q) are integral functions of z
and
that the
coefficients in their expansions as Fourier series are power series in q which

converge absolutely when \q\ is sufficiently small*.
To obtain
this result for the functions ce2f{z, q),
we
shall
shew how
to
determine a particular integral of the equation
-^ + (a +
*
The
essential part of this
in the coefficients;
integral
it is
functions of
expansion follows from
already
z,
and
9*11.
\Qq cos 2z) u
theorem
known
is
y\r
{a, q)
cos
Nz
the proof of the convergence of the series which occur
{ 10'2, 10-21) that solutions of
(in the case of periodic solutions)
Mathieu's equation are
the existence of the Fourier
MATHIEU FCNCTIOXS
19-6]
form of a Fourier series converging over the whole
in the
yjr
421
(a, q) is
a function of the parameters a and
2^-plane,
The equation
q.
where
(a, q)
-v/r
then determines a relation between a and q which gives rise to a Mathieu

function.
The reader who is acquainted with the method of Frobenius* as
applied to the solution of linear differential equations in power series will
recognise the resemblance of the following analysis to his work.
Write a
= iY^ +
where
8p,
JSf is
zero or a positive or negative integer.
Mathieu's equation becomes
^_+:N'Hi =
-S (p +
and q are neglected, a solution of
If jj
2q cos 2z)
11.
this equation is u
cos Nz=
Uo{^)>
say.
To obtain
of cosines,
a closer approximation, write
i.e.
{^ cos
as a
ITq (z)
{X-2)z+p cos Nz + q cos (N + 2) z] = Fj {z),
Then, instead of solving
which involve cos Nz;
sum
-r--
+ X'-u = V^ {z),
say.
suppress the terms f in V^ {z)
consider the function W^ {z) wherej
i.e.
If,
8(p + 2q cos 2z)
in the form
(^)=F,(^)
8^;
cos iY^.
particular integral of
^^,+NHl=W,{z)
18
^^' +
'^' (^ - 2) ^
^^ 1 = ^'
+ roTiT) ''
iroîT)
Now express S(p + 2q cos 2z) L\ (z) as a sum of cosines
11
sum
Vo
(z),
choose
a. to
process.
are thus obtained, such
Three
sets of functions
that U,n {z)
and W,n
F, (^)
a function of p and q hut not of
Journal fiir Math, lxxvi. (1873), pp. 214-224.
The reason
{z)
=-
^^J^ + NL\(z)
, is
this
8 {p
N = \,
cos
Nz
Um{z), Vm{z), Wm{z)
2q cos 2z) U,n-i
suppression
is
that the
in which case \\\{z)-1\{z)
Nz
(z),
W,{z),
z.
particular
contains non-periodic terms.

+ Unless
a^.
contain no term in cos
for
+ N- u = Wo {z),
m ^ 0, and
W^ {z) = F, {z) + a, cos Nz,
where
calling this
cP u
-r-j
Solve the equation
and continue the
be such a function of p and q that V^ (z)

and let V., {z) + a., cos Nz = W.^ {z).
Nz
contains no term in cos
when
^'^' '^^
9.
+ 9{i) + q)
co%z.
integral
of
^+
d-u
-,
A'- = cosA^
422
[CHAP. XIX
It follows that
\az^
/=i
j=o
ni
=-
(;9
25 COS
2ir)
[7',rt_i
(^)
a,
cos Nz.
U{z)= S
f/',^ (^'^ be a uniformly convergent series of analytic

m=0
functions throughout a two-dimensional region in the ^-plane, we have
Therefore,
if
( 5-3)
d?U(z^
+ ('^ +
7-2
^^9. cos 'Iz)
U (z) =
ylr
(a, q) cos
Nz,
oc
Avhere
It is obvious that, if a
to cey
^fr{a,q)=
0^.
be so chosen that
yjf
(a, q)
0,
then
U {z)
reduces
{z).
similar process can obviously be carried out for the functions 5e y
by making use of sines of multiples of

19*61.
The convergence of the
We shall
series defining
now examine the expansion
{z,
q)
z.
Mathieu functions.
of 19"6
more
closely,
with a view to investigating
the convergence of the series involved.
When
n"^
we may obviously write
1,
/!
U-a\Z)=
2 */3,,.cos(iy-2r)-|-
2 a^rCOfi{N->r'2.r)z,
r=l
r=l
the asterisk denoting that the
first
summation ceases
at the greatest value of r for which
r^^N.
1
{(12
-jpi+^^\ ^n+i
it
(-)
= an +
follows on equating coefficients of cos

0.1
/(>-
(iV^
cos
Nz -8{p + 2q
2r)
cos 22)
= ^!? ("h,1+3,i),
+ .y)a+,,, = 2{/?a,, + j(a,^r-i + a,r + i)}
Vo = 3.o =
(ii)
^^j
(i")
^.^j
= ^^ iv-i
^li6i'
-^^
When A'=0
The conventions
(''=1,2,
i*^
is
^^'en
(;>),
and r=|i\',
odd and r=h{N-\).
these equations must be modified by the suppression of
(ii)
and
(iii)
...),
a,. = ^,^^^ =
= 1,2,...);
^n.H-V+D^'^n.H^V-i) ^ên .V
or
(2),
on each side of the equation t that
These formulae hold universally with the following conventions %

(i)
/"
are due to the fact that cos2;
= cos (-2),
all
cos 22
the coefficients
= cos (- 22).
MATHIEU FUNCTIONS
19-61]
The reader
(I)
will easily obtain the following special

a^
(III)
a,y
formulae
= 8p,
(iV=^l);
and
homogeneous polynomials
^n,r ^^^
we have
>/^(a, j)
= 8ip + q),
ai
8jt?
423
(iV=l),
of degree n in
+ 8y (Ji+5,)
and
q.
(iV-^D,
rir + ]V)A,=2ipAr+q(Ar-, + A,î)]
(A),
r(r-.V) B, = 2{pB,+q {B,_, + Brî)}

where
Ao=B^ = 1 and
Now
The
where
write
w,.=
B,. is
-q
subject to conventions due to
{r {r
+ ]V)-2p}-\
result of eliminating Ai, Ao,
A,, is
...
-q
'/=
Aj.-^,
A^ + i,
...
{r
(B),
and
(ii)
(iii)
above.
(r-y)-2p}-\
from the
set of equations (A) is
the infinite determinant of von Koch's type ( 2'82)

A,.=
The determinant converges

and Ar-*-l as r--cc (cf.
2p^r{r-\-N), where r = l,
il'r+u
0,0,....
Wr+3,
W^+3,
...
absolutely ( 2"82 example) if no denominator vanishes

If p and q be given such values that Ao^fcO,
19"52).
3,
2,
the series
...,
2 { yiVx%o.2...Wj.Ar^(r^ cos(iV+2r)
r=\
represents an integral function of
In like manner B,.D(,=
{-Y
Wi
z.
tc^ ...w^' J),.,
w'r +
w'r + 2,
the last row being
is
0, 0, ... 0,
2w\^,
or
0,
where
D,. is
,
...
2,
i,
the
finite
determinant
J,
,
w'r +
0,
...
0, ^<''i(jvr-i))
+ ^'''i(,v_i)
according as
even or odd.
The
series
Un (z)
is
therefore
?t=0
COS^V^ + Ao"^
(-)'"?<?i?i'2...?<'rArCOS(iV+2r)2
r=l
+ D-
these series converging uniformly in any bounded
)'
domain
iv;
yj/
(,
g')
is
pAoô-q
If
we multiply by
equivalent to
{w-^
. .
w,' I), cos
of values of
term differentiations are permissible.

Further, the condition
10,'
Ai Do + Wi'Di Aq) = 0.
z,
(iY-
2;-) z,
so that term-by-
[CHAP. XIX
424
the expression on the
terms of
ôw
^ (a,
becomes an integral function of both p and
left
q\ which are of lowest degrees.in p and
expand
in ascending powers of q
|
Then
contour.
^-^-^
^ (iV^ + Sjo,
in the p-plane, with
cii'cle
q) has only
one zero inside the

\q\,
as a
substituting for
series involved in cex
(s,
terms of q throughout the series for U (z), we see that the

convergent when ^ is sufficiently small.
in
q) are absolutely
series involved in se^
(2,
q)
may
obviously be investigated in a similar manner.
The method of change of parameter \.

of Hill and of Lindemann-Stieltjes are
19'7
the
ap
the contour being a small
being so small that
(a, q), h-a\
p and
power series in q commencing* with a term in q^; and if

small D^ and Aq will not vanish, since both are equal to 1 when ^=0.
be sufficiently
The
follows, just as in 7"31, that, for sufficiently small values of
it
we may expand p
On
..o
7"31),
(cf.
centre at the origin, and
q,
are respectively
-.
q,
The methods
effective in
determining
^,
but
Such analysis is inevitable, as ^ is by no means a simple

function of q this may be seen by giving q an assigned real value and making a vary
from C30 to + 00 then /x alternates between real and complex values, the changes taking
only after elaborate analysis.
;
place when, with the Hill-]\Iathieu notation,
and
elaborate expression involving both a and
passes through the values
(0) sin- [hir s,â)
the complicated nature of this condition
is
due to the
fact that
(0) is
an
q.
It is, however, possible to express

and a in terms of q and of a new parameter tr, and
the results are very well adapted for purposes of numerical computation when 5' is small J.
fj.
The
introduction of the parameter
a-
is
suggested by the series for
cei {z, q)
and
sey {z, q)
a consideration of the.se series leads us to investigate the

given in 19"3 example 1
potentialities of a solution of Mathieu's general equation in the form y=e'^'^0(s), where
;
^ if) = sin {z-(r) + a-i cos (82 - o-) + 63 sin

the parameter
in
o-
(j){z);
or
is
On
o-
(82
- cr) + 05 cos
{hz
- cr) + 65 sin
being rendered definite by the fact that no term in cos
the special functions 5^1(2,
q), cei{z, q)
tr)
a)
is
to appear
(52
{z
are the cases of this solution in
cr
.
.
which
\ir.
substituting this expression in Mathieu's equation, the reader will have no difficulty
in obtaining the following approximations, valid for small viilues of q
of
and
real values
^ =4^' sin 2o-- 12^-" sin 2o-- 12j*sin 4o- + (2"^),

a = 1+ 8y cos 2(7 + ( - 16 + 8 cos 4o-) ^2 _ î cos 2o- + (^f^
a3=Sq^ sin 2a- + '3q^ sin 4:a + {-^^ sin 2a- + 9 sin 6a) q^ +
--
88 cos
h^=q + q^ cos 2o- + - J^f + 5 cos 4(r) j^ + _ IJ. cos 2o- + 7 cos 60-)
a- = Y?^ sin 2o-4-|f ?^ sin 4o- +
{q%
+ q^ cos 2(7 + - Vr + f f cos Aa) q^+0 {q->),
67 = ^^q^ + ^^ (/* cos 2<t+0 iq%
7 = f'ifs 9* sin 2a +
(q^),
h, = ^l,,q^ + 0{q%
a, = 0{q%
(
h=W
=l
this result has to be modified, since there
the term q^jiN -
1)
+ O {q%
</*
{q%
^j
the constants involved in the various functions 0{q^^) depending on

* If A^
4o-) q^
{q%
does not appear.
is
a-.
an additional term
on the right and
t Wbittaker, Proc. Edinburgh Math. Soc. xxxn. (1914), pp. 75-80.

X They have been applied to Hill's problem by luce, Monthly Notices of the R. A. S. lxxv.
(1915), pp. 436-448.
The parameters
a and q as hitherto.
and a are
to be regarded as
fundamental
in this analysis, instead of
197, 19 '8]
MATHIEU FUNCTIONS
The domains
and
of values of q
which these
for
o-
425
series converge
have not yet been
determined*.
If the sokition thus obtained be called A {z, a, q), then A (z, cr, q) and A
a fundamental system of solutions of Mathieu's general equation if /x=#0.
Example
shew
Shew
1.
also that,
if
(r
that,
= / x 0'5
if o-
/x
= ?x 0-046,993,5 ...
= i and = 0'01, then

= l-,321, 169,3...,
q)
form
/^
= ix 0-145,027,6....
2'
Obtain the equations
2.
/x
rt
= 4^ sin 2a- 4 ja3

1 + 83' cos 2(r 8363,
/it-
expressing n and a in finite terms as functions of
Example
o",
and ^ = 0-01, then
a = M24,841,4...,
rt
Example
{z,
q, a, a^
and
63.
Obtain the recurrence formulae
3.
{-4:n{n+l) + 8qcos2a--8qb38qi{2n+l){as-sin2(r)}z.2n + i + 8q{z2_i+Z2^3) = 0,

where
22jh-i
denotes bon+i
ict-in
oi" ^-in
i2)i
i>
according as the upper or lower sign
is
taken.
The asymptotic solution of MathiexCs equation.
19"8.
If in
Mathieu's equation
d'^v.
-5
dz-
we write k
sin 2
= ^, we
a+-
,.,
A,-
^ \
cos 22
M=
get
where i/ ^ = + ^P.
This equation has an irregular singularity at
equation,
we are
e'^
led to write u
|~-
v,
infinity.
From
its
resemblance to Bessel's
and substitute
V=l+{a,/^) + {a,!e) + ...

in the resulting equation for v
we then
(i - 3P + F),
ai = - î
a.,
find that
Hi - ^^^ +
^')
(f
- -^^H F) + IF,
the general coefficient being given by the recurrence formula
2i(r+l)a, + i = {J-J/2 + F + /-(r+l)} + (2?--l)zFa^_l-(/-2-2/- + |)Fa,_2.
The two
series
e'U~-(l+^
+ jl+-],
|,p,...),
.-.,-.,.
e-''r'(l-^^.^_
+
are formal solutions of Mathieu's equation,

solutions of Bessel's equation ( 17-5)
them with the
when
reducing to the well-known asymptotic
^--^0.
The complete formulae which connect
solutions e^'^^ (f)(z) have not yet been published, though
towards obtaining them have been made by Dougall,
F7'oc.
Edinburgh Math.
some steps
Soc. xxxiv.
(1916), pp. 176-196.

* It
seems highly probable
that, if
|
is
sufficiently small, the series converge for all real
and also for complex values of cr for which |I((r) is

noticed that, when q is real, real and purely imaginary values
to real and purely imaginary values of fi.
values of
a,
sufficiently small.
of
cr
It
may
be
correspond respectively
THE TRANSCEXDEXTAL FUNXTIONS
426
[CHAP. XIX
KEFEREXCES.
E. L.
Mathiec, Journal de Math.
W. Hill, Acta Mathematica,
G.
viii. (1886), pp. 1-36.
G. Floquet, Ann. de VEcole norm. sup.

C. L. F.
LiNDEMANN,
Mcith.
pp. 137-203.
(2), xiii. (1868),
(2), xii.
(1883), pp. 47-88.
Ann. xxii. (1883), pp. 117-123.
T. J. Stieltjes, Astr. Naeh. cix. (1884), cols. 145-152, 261-266.
LiXDSTEDT, Astr. Nach. cm. (1882),
A.
145-150; cv. (1883),
H. Bruns, Astr. Xach.

R. C.
cvi. (1883), cols.
Maclaurin, Trans. Camb.
193-204;
I.
cvii. (1884), cols. 129-132.
(2), iv.
(1908), pp. 266-278.
Proc. International Congress of Mathematicians, Cambridge, 1912,
Whittaker,
pp. 366-371.
E. T.
Whittaker,
G. N,
Watson,
W. Young,
A.
(1883), cols.
Phil. Soc. xvii. (1899), pp. 41-108.
K. AiCHi, Proc. Tokyo Math, and Phys. Soc.

E. T.
211-220, 257-268; Civ.
cols.
97-112.
cols.
Proc. Edinburgh Math. Soc. xxxii. (1914), pp. 75-80.
Proc. Edinburgh Math. Soc. xxxiii. (1915), pp. 25-30.

Proc.
Edinburgh Math.
Soc. xxxii. (1914), pp. 81-90.
E. Lindsay Inge, Proc. Edinburgh Math. Soc. xxxiii. (1915), pp. 2-15.
J.
Soc. xxxiv. (1916), pp. 176-196.
Dougall, Proc. Edinburgh Math.
1.
Shew
that, if
k=^l{Z^\
2wce(, {s, q)
= cco {0,q)
cos {k sin
sin 6) ccq (0, q) d6.
J It
2.
Shew that the even Mathieu functions
G (2)=X
3.
Shew
Jo [ik (cos
+ cos
6)}
{6) d6.
that the equation

(a22
(where
satisfy the integral equation
o, c, X,
+ c)^+2a5^ + (X%2+0 =
are constants)
is satisfied
by
u = \^^v{s)ds
tAken round an appropriate contour, provided that
{as^
which
is
v (s) satisfies
+ c)^^ + 2as^-\-{X^cs^- + m)v
the same as the equation for
{3)
= 0,
u.
Derive the integral equations satisfied by the Mathieu functions as particular cases of
this result.
MATHIEU FUNCTIONS
Shew
4.
that, if
ce?! (s,
427
powers of q above the fourth are neglected, then
q) = cos 2 + J cos 32 + q^ (J cos hz cos 3s)
+ (^ (i^8 cos 7s f cos 5s + J cos 3s)
+ ?* (rl^ <^^^ ^^ ~
sei
(s,
j) = sin s + g sin 3s +
'2
+ <f
+
C(?2 (s, g-)
= cos 2s +
g-
(cos 4s
(i sin 5s
jij
(
I's
^0*^
'''^
+H
cos 5s + ^^ cos
3s),
+ sin 3s)
sin 7s
+ f sin 5s + A
+ Jj sin
q^ (yIcj sin 9s
7s
sin 3s)
+ ^ sin 5s - ^- sin 3s),
2) + ^ g^ cog g^
+ ?^ (-/s cos 8s + If cos 4s + -*^)

+ ?* (sTO cos 10s + f| cos 6s).
(Mathieu.)
Shew that
5.
663(0,
2')
= cos 32 + 2'( coss+l cos5s)

+ j2 (cos s + J(5 COS 7s) + j3
and
_ I COS
+ -^^ COS 5s + Jq
cos 9s)
{q*),
that, in the case of this function
a = 9 + 4q^-8q^+0{q^).
(Mathieu.)
Shew that,
6.
differential
if
equation
1/
(s)
be a Mathieu function, then a second sokition of the corresponding
is
that a second solution * of the equation for
Shew
zceQ
If
7.
?/
(2)
(z,
q) 4:qsiB 2s-
ce^
Sq- sin
(s,
q) is
4s-
....
be a solution of Mathieu's general equation, shew that

{y(s + 2^)+3/(s-2:r)}/j/(2)
is
constant.
Express the Mathieu functions as series of Bessel functions in which the coefficients
8.
are multiples of the coefficients in the Fourier series for the Mathieu functions.
[Substitute the Fourier series under the integral sign
in the
integral equations of
19-22.]
Shew
9.
that the confluent form of the equations for ce^
eccentricity of the fundamental ellipse tends to zero,
by
J,, {ii-
eoH
is,
(s,
q)
and
se^
(2,
q),
when the
in each case, the equation satisfied
z).
10.
Obtain the parabolic cylinder functions of Chapter xvi as confluent forms of the
Mathieu functions, by making the eccentricity of the fundamental ellipse tend to unity.
11.
Shew that
ce
(s,
q)
can be expanded in series of the form

2 J,cos2'"3 or 2
m=0
5,,,cos2' + i2,
n=0
according as %
is
even or odd; and that these series converge when
This solution
is
called in^
(z,
q)
coss
<
1.
the second sohitions of the equations satisfied by Mathieu
functions have been investigated by Ince, Proc. Edinburgh Math. Soc. xxxiii. (1915), pp. 2-15.
See also
19-2.
428
With the notation
12.
of
ce
example
{z,
q)
11,
shew
= X
[cHAP. XIX
that, if
e*cos3oo60
ê,, ((9,
q) dd,
then A
is
given by one or other of the series
provided that these series converge.
Shew
13.
that the differential equation satisfied by the product of any two solutions
of Bessel's equation for functions of order
is
S{S-2n) {S+21l)u + -I.^^S + l) u = 0,

where 3 denotes
Shew
z -j-
dz
is an integral function of 2; and thence, by the

methods of 5;^$ 19'5-19'53, obtain the Bessel functions, discussing particularly the case in
which a is an integer.
14.
that one solution of this equation
Shew
that an approximate solution of the equation
-^-ir(A+k-sm\i-z)u=Q
dz^
is
where
\i
C and
very large and
= C (cosech s)
are constants of integration

z is
not small.
sin {k cosh z + e),

it is
to be
assumed that k
is large,
is
not
CHAPTER XX
GENERAL THEOREMS AND THE
ELLIPTIC FUNCTIONS.
WEIERSTRASSIAN FUNCTIONS
Doubly -periodic functions.
20"1.
is
most important property of the circular functions sin^, cos^, tan^,

that, if/(^) denote any one of them,
f{z
and hfmce f{z-\-'2mr)=f{z),
of this property that
the
which
will
+ 2^)=f{z),
for all integer values of n.
functions are
circular
periodic functions with period
...
on account
It is
frequently described as
To distinguish them from the functions
27r.
be discussed in this and the two following chapters, they are
called singly-periodic functions.
Let
Q)i, 0)2
real.
be any two numbers (real or complex) whose ratio*
f(z
for all values of z for
of
z,
is
not purely
function which satisfies the equations
with periods
2a,0 =f{z),
which f(z)
2&)i, 2co2.
f(z
2a,,)
=f(z),
exists, is called a doubly-periodic function
doubly-periodic function which
is
analytic
(except at poles), and which has no singularities other than poles in the
finite part of
the plane,
What
[Note.
is
called
now known
is
integrals in connexion with the
an
elliptic
elliptic function.
as an elliptic integral^ occurs in the researches of Jakob
Maclaurin, Fagnano, Legendre, and others considered such
Bernoulli on the Elastica.
'inverting'
an
problem of rectifying an arc of an ellipse the idea of

21 '7) to obtain an elliptic function is due to Abel,
;
integral (
Jacobi and Gauss.]
The
periods
functions as
is
2(i,
2&,o
play
much
the same part in the theory of elliptic
played by the single period in the case of the circular
functions.
Before actually constructing any elliptic functions, and, indeed, before

establishing the existence of such functions,
general
theorems
elliptic functions
(|
20"11-20'14)
it
concerning
is
convenient to prove some
properties
* If w.,/wj is real, the parallelograms defined in 20-11 collapse,
a singly-periodic function
by
when
Jacobi, Journal fiir Math. xiii. (183.5),
is
to
all
convenient
and the function reduces to

and when w^/wj is irrational, it has been shewn
pp. 55-56 [Ges. Werke, 11. (1882), pp. 25-26] that the
Wg/wj is rational
function reduces to a constant.

t
common
this procedure, though not strictly logical,
brief discussion of elliptic integrals will be
found in
22-7-22*741.
[cHAP.
XX
elliptic functions
can
430
because a large number of the properties of particular

be obtained at once by an appeal to these theorems.
The
Example.
diflferential
coefficient
an
of
elliptic
function
is
itself
an
elliptic
function.
2011.
Pei-iod-parallelograms.
The study
of elliptic functions
representation afforded by the
is
Argaud
much
facilitated
by the geometrical
diagi-am.
Suppose that in the plane of the variable z we mark the points 0, 2(Wi,
complex coordinates are
2(02, 2a)i + 2&)2, and, generally, all the points whose
of the form 2mo)i
2h&j.,,
where
and n are
integers.
Join in succession consecutive points of the set 0, 2(Oi, 2(Oi + 2o}.2, 2ei).,, 0,
and we obtain a parallelogram. If there is no point a inside or on the
boundary of
this parallelogram (the vertices excepted)
such that
f{z + <o)=f{z)
for all values of
gram
for
an
z,
this parallelogram is called & fundamental period-parallelo-
elliptic function
with periods
It is clear that the ^-plane
may be
2a)i, 2(0^.
covered with a network of parallelo-
grams equal to the fundamental period-parallelogram and similarly situated,

each of the points 2?/ia)i + 2na), being a vertex of four parallelograms.
These parallelograms are called peHod-parallelograms, or meshes
for all
values of z, the points z, z-\-2w^, ... z +2m(o^+2nw. ...

corresponding positions in the meshes any pair of such points are said to
be congruent to one another. The congruence of two points z, z is expressed
manifestly occupy
by the notation / =
From
elliptic
points
^ (mod. 2&)i,
2w.2).
the fundamental property of elliptic functions,
it
follows that
an
function assumes the same value at every one of a set of congruent

and so its values in any mesh are a mere repetition of its values in
any other mesh.

For purposes of integration it is not convenient to deal with the actual
meshes if they have singularities of the integrand on their boundaries on
account of the periodic properties of elliptic functions nothing is lost by
;
taking as a contour, not an actual mesh, but a parallelogram obtained

by translating a mesh (without rotation) in such a way that none of the poles
of the integrands considered are on the sides of the parallelogi-am.
Such a
Obviously the values assumed by an

parallelogram is called a cell.
function in a cell are a mere repetition of its values in any mesh.
elliptic
set of poles (or zeros) of
an
an irreducible set all other poles

one or other of them.
;
elliptic function in
any given
(or zeros) of the function are
cell is called
congruent to
ELLIPTIC FUNCTIONS
20"11, 20'12]
Simple properties of
20'12.
elliptic functions.
(I)
The number of poles of an
For,
if not,
analogue of
function
This point
by
so,
is
For,
cell is finite.
5'61) an essential singularity of the
definition, the function is not
The number of zeros of an
(II)
any
the poles would have a limit point, by the two-dimensional
2*21.
and
431
an
elliptic function.
any
cell is finite.
the reciprocal of the function would have an infinite
if not,
number
and would therefore have an essential singularity and

this point would be an essential singularity of the original function, which
would therefore not be an elliptic function, [This argument presupposes
of poles in the
cell,
that the function
not identically zero.]
The sum of the residues of an
(III)
any
is
poles in
cell is zero.
Let
be the contour formed by the edges of the
of the cell be
+ 2&ji, t +
^,
2(yi
but that one

imaginary part and then,
will
if
cell,
elliptic
function will not be called
2ci)i,
2(B2
is counter-clockioise.
the contour formed by
of the residues oif{z) at its poles inside
is
shall
^.\ f{z)dz=-~\\
In the second and third integrals write z+2a)i, z

and the right-hand side becomes
^.
{/(^)
-/{^ +
2a>.))
dz
and each of these integrals vanishes

f{z); and so
f{z) dz
= 0,
{f{z)
and the theorem
Liouvilles theorem*.
An
+ 2(Oo
\f{z)dz.
respectively for
-f{z +
20,01 dz,
in virtue of the periodic properties of

is
established.
elliptic function, f{z), luith
no poles in a
merely a constant.
For if f(z) has no poles inside the cell,

bounded) inside and on the boundary of the
to say, there
boundary of
on
it is
analytic (and consequently
cell ( 3-61 corollary ii)
that
is
number K such that f(z) < K when z is inside or on the

From the periodic properties oi f{z) it follows that
the cell.
is
* This modification of the
lectures
the corners
cell.]
The sum
cell is
let
denote by the symbol
Throughout the chapter, we
(IV)
and
2&)2.
be called 2wi which makes the ratio o),/! have a positive

be described in the sense indicated by the order of the
corners given above, the description of
the edges of a
2&).,, t
In future, the periods of an
[Note.
indifferently
2,
elliptic function, f{z), at its
elliptic functions.
theorem of
5-63 is the result
on which Liouville based his
432
f{z)
analytic and \f{z)
is
K for
<
values of z; and
all
[CHAP.
by
so,
5-63,
XX
f{z)
is
a constant.
be seen later that a very large number of theorems concerning
It will
can be proved by the aid of this
elliptic functions
2013.
The order of an
It will
now be shewn
constant, the
elliptic function.
that, \^ f{z) be an
number of roots of the equation
lie
in
any
elliptic function
and
be any
=c
f(^)
which
result.
depends only on f{z), and not on c; this number is

and is equal to the number of poles
cell
called the order of the elliptic function,
off(z) in the
By
cell.
which
Since /' {z
parts, precisely as in 20'12(III),
/'<^>
number
conversely; hence the
which
find that this integral
of zeros of /'(^) c
number
is
d..
by dividing the contour into four
{z),
but any pole oif{z)
c\
of poles of f(z),
we
c is
of zeros of
independent of
number
of zeros and the
is
f{z) c
+ 2&)i) =/' {z + 2(Wo) = /"
Therefore the
in the cell
lie
'
1
27ri J r
poles oi f{z)
number
the difference between the
6*31,
of poles o{f(z)
equal to the
is
number
of
obviously a pole of f(z) and
f(z)
is zero.
c is
number
equal to the
the required result
is
therefore
established.
[Note.
In determining the order of an
its irreducible poles, it is
elliptic
function by counting the
number
of
obvious, from 6'31, that each pole has to be reckoned according
to its multiplicity.]
The order
of an elliptic function
never less than 2
is
function of order 1 would have a single irreducible pole
for
and
an
elliptic
if this
point
actually were a pole (and not an ordinary point) the residue there would
not be zero, which
So
is
contrary to the result of
201 2
(III).
far as singularities are concerned, the simplest elliptic functions are
Such functions may be divided into two classes, (i) those

which have a single irreducible double pole, at which the residue is zero in
accordance with 20"12 (III) (ii) those which have two simple poles at which,
those of order
2.
by
20"!
2 (III), the residues are numerically equal but opposite in sign.
Functions belonging to these respective classes
and
chapter
Jacobian
elliptic
types.
in
elliptic
function
Chapter xxii under

functions
is
respectively
expressible
in
will
be discussed in this
names of Weierstrassian and

and it will be shewn that any
the
;
terms of functions of either of these
ELLIPTIC FUNCTIONS
20-13-20'2]
Relation hetiueen the zeros and poles of an elliptic function.
20"14.
We
shall
zeros of
433
an
now shew
elliptic
sum of the affixes of a set of irreducible

congruent to the sum of the affixes of a set of
that the
function
is
irreducible poles.
For, with the notation previously employed,
the difference between the sums in question
27ri J C
^TTl [J
f {2)
^ j^
rt+2.^
~27ri]t
f{z +
\f{^)
2ay,)
6"31,
that
{2:)
'
f{z+2c.,)
\f{z)
f (z)
Jt
from
t+2ojJ
j^^) _ {z+2oy,)f'(z + 2co.^
27rij,
27ri
J f+2a,,+2a,2
J ^+2<o,
follows,
it
is
f{z)
t+iu
J.j-2.,[log/(.)
2(0,
log/(^)
t
on making use of the substitutions used in
2012
(III)
and of the periodic
properties off{z) and f'{z).
Now f{z) has the same values at the points t + 2(Oi, t 4-20), as at t, so
the values of \ogf(z) at these points can only differ from the value of f{z)
at
by integer multiples of
27ri,
say
2n7ri, 2///'7rt;
then we have
2'mJc fiz)
and
sum
so the
the poles
is
20'2.
of the affixes of the zeros
a period
and
this is the result
The construction of an
minus the suin of the affixes of

which had to be established.
elliptic function.
Definition of^{z).
It was seen in | 20'1 that elliptic functions may be expected to have

some properties analogous to those of the circular functions. It is therefore
natural to introduce elliptic functions into analysis by some definition
analogous to one of the definitions which may be made the foundation
of the theory of circular functions.
One mode
of developing the theory of the circular functions
S {z-m7r)~-;
from the series
JW= -
to
deduce
all
the
calling this series (sin
^')~-,
it
is
is
to start
possible
CO
known
properties of sin z
the method of doing so
is
briefly
indicated in 20-222.
W. M.
A.
28

434
The analogous method

define the function
(,.>
of founding- the theory of elliptic functions
XX
is
to
by the equation*
(^)
^"
[CHAP.
m.n
\{2;
2vio)i '2n(02y
'
(2?&)i+2?ieu2)-j
Avhere (o^, Wo satisfy the conditions laid down in ^0*1, 20"12(III); the
summation extends over all integer values (positive, negative and zero) of
m and n, simultaneous zero values of ni and n excepted.
For brevity, we write Qm,n in place of
'Iinw^
'Inwo, so
that
m, n
When m and
series defining
ip
n are such that \^m,n\

(z)
0(\flm,n\~% and
is
large, the general terra of the
is
the series converges
so (3-4)
absolutely and uniformly (with regard to z) except near its poles, namely
the points Clm,n-
Therefore
( 5"3),
^J
analytic throughout the whole 2^-plane except
(z) is
at the points n,_, where
it
has double poles.
The introduction of this function p{z) is due to Weierstrassf we now

proceed to discuss properties of ^ (z), and in the course of the investigation
it will appear that ^j{z) is an elliptic function with penods 2&)i, 2a)2.
;
For purposes of numerical computation the series for

slowness of
its
convergence.
Elliptic functions free
(z) is
useless on account of the
from this defect
will
be obtained in
Chapter xxi.
Example.
Pi'ove that
P(2)=U
20*21.
-^+
Periodicity
and
Since the series for
cosec2
=7r
cosec-
"tt
other properties of ^{z).
a uniformly convergent series of analytic
{z) is
and so
functions, term-by-terra differentiation is legitimate ( 5'3),

ij'{z)
The function ^' {z)

^y(^),
we
= ^,j{z)=-2 t
is
^^
m, n \^
an odd functio7i of z\
- '>,
for,
n)
from the definition of
at once get
^'(-^)
(^4-n,,)-^.
m, n
*
of
Throughout the chapter 2
m and
n, a
will be written to
prime being inserted
(2')
when
denote a summation over
the term for which j=:n =
all
integer values
has to be omitted
m, n
also customary
from the summation. It is

the prime in two senses will not cause confusion.
to write ^'
{z)
for the derivate of
^4>
{z).
The
use of
The subject-matter of the greater part of this chapter is

t Werke, ii. (1895), pp. 245-2.55.
to Weierstrass, and is contained in his lectures, of which an account has been published by
due
Schwarz, Formeln und Lehrsatze zum Gebrauche der elUptischen Fiinktionen, Xach Vorle.tnngen
und Aufzeichnungen des Herrn Prof. K. Weierstrass (Berlin, 1893). See also Cayley, Journal de
Math.
X. (1845),
pp. 385-420 [Math. Papers,
XXXV. (1847), pp. 137-184, 18-5-274.
i.
pp. 156-182],
and Eisenstein, Journal fUr Math.
20 '2 1]
ELLIPTIC FUNCTIONS
But the
terms of
D,(_
set of points
^'
the same as the set Qm, n and so the

^' (z), but in a different
is
the same as those of
z) are just
But, the series for
order.
435
being absolutely convergent
(z)
^j'
derangement of the terms does not
the
sum, and therefore
affect its
^y {-z) = -
( 3"4),
^y (z).
In like manner, the terms of the absolutely convergent series
ni,n
are the terms of the series
y
m,
in a different order,
Hz -
is to
say, ^ (z)
Further,
\-^
n~^
and hence
i^
that
(-
z)
i^
U)
is
an even functipn_oj[j.
^j'
{z
2co,)
=-
(^
- D^,^ + 2a},r^
m, n
but the set of points
for ^y (z
is
'Zcoi)
n,^
is
^y (z) is
If
^y'
2ft)i)
an
except at
(z) is analytic
= ^y (z)
in like
where A is constant.
even function, we get
manner
^j
{z
Putting z
^=
0,
being
it
has the period
2ft)o.
follows from this result that
'Icoo)
2oy,)
'2(o^)
= <^' {z), we
get
= <î{z) + A,
w^
and using the
fact that
<^
{z) is
an
so that
{z
iO
(jp
2ft)i)
= ^J (z)
(z).
Since p (z) has no singularities but poles,

that ^J(2) is an elliptic function.
it
follows from these
There are other methods of introducing both the circular and

for the circular functions the following may be noticed
(1)
the series
series
manner
its poles, it
integrate the equation ^' {z

<p{z
analysis
The
elliptic function.
now we
in like
to say, '/ (z) has the period 2fOi
Since
Q,, , so
(z).
we have
^y (z
that
the same as the set
2&ji is
a derangement of the series for ^'
two results
elliptic functions into
The geometrical
definition in
which sin z
is
the ratio of the side opposite the angle
hypotenuse in a right-angled triangle of which one angle is z. This is the definition

given in elementary text-books on Trigonometry from our point of view it has various
disadvantages, some of which are stated in the Appendix.
2
to the
(2)
The
definition
by the power
series
Z^
sm2 = 2--, +
282
'
436
(3)
The
definition
by the product
(4)
The
definition
by 'inversion' of an integral
[CHAP.
XX
fBim
J
The
may be
periodicity properties
obtained easily from
(4)
by taking suitable paths of

it is extremely difficult to
Forsyth, Theory of Functions, (1918), ^ 104), but

prove that sin z defined in this way is an analytic function.
integration
(of.
The reader
may be
Deduce the periodicity of
^.>
[It is not difficult to justify the necessary
2022.
We
now
first
(z)
and
fourth.
directly
from
its definition
as a double series.
derangement.]
The differential equation
shall
20-53 example 4) that
defined by definitions analogous to each of these, with corre-
sponding disadvantages in the cases of the
Example.
and
will 'see later ( 22-82, 22-1, 20-42, 20-22
elliptic functions
hy
satisfied
{z).
^j
obtain an equation satisfied by
<^
{z),
which
will
prove to
be of great importance in the theory of the function.
The
function
^j
(2)
,2"^
which
equal to S'
is
{(2
n,,^,i)~2
H^^},
is
m, n
analytic in a region of which the origin
even function of
of the form
s.
an internal point, and
is
valid for sufficiently small values of
^^
|
|.
It
is
m, n
Thus
(z)
differentiating this result,

^y (z)
f
Hence
That
Z-'
+ 20
{z)
^'- {z)
to
say,
is
the
also analytic at all
it is
z-'
respectively,
(z^).
we get
+ .l~^g,z-^ + l^g..+
0(z^),
- 4^^=* (z) = - g.z" -gs + {z'),

- 4f (z) + g^^ (z) + gs=0 (z').
function ^^^ {z) 4^^^ (z) + g^^ (z) + g-s,
is
analytic at the origin,
which is
and consequently
But such points
congruent points.
therefore a constant
On making z^0, we
+ 28 ^3^ + ^ (^')
possible singularities of the function, and, so
no singularities;
an
we have
obviously an elliptic function,

it
g^z'
= - 2z-' + ^^ g,z ^ 1 g,z^ +
^'^ (z)
is
is
easy to see that

ni,
Cubing and squaring these
and so
it
Consequently, by Taylor's theorem, we have an expansion
it
is
an
20'12, lY).
see that this constant
is zero.
elliptic
are the only
function luith
"
elliptic functions
20'22, 20-221]
437
Thus, finally, the function (^{z) satisfies the differential equation
where
^2
and
given by the equations
g^ (called the invariants) are

g,
mT
),
n^f,
5r3
i4o 1' n-\.

m, n
Conversely, given the equation
if numbers
can he determined* such that
&)i, (o^
m, n
i,
then the general solution of the differential equation
iO{z
is
a),
where a is the constant of integration. This may be seen by taking a new

dependent variable u defined by the equationf y = <^ (u), when the differential
equation reduces to
Since
g?
{z) is
[^\ =1.
an even function of
we have y =
z,
{0{z
and so the
a),
solution of the equation can be written in the form
y
without
= ^j{z +
a)
loss of generality.
Example.
Deduce from the
differential equation that,
)i
then
C2=5r2/22
Ci=g^l^^
5,
^~2*.5.7.11'
20"221.
if
=l
CQ=gîj2i.'^
7,
52,
"~25..3.5M.3"^2*.72.13'
The integral formula for
<^
''
2^
52
11
{z).
Consider the equation
z=\ (U'-g,t-g,)-^dt,
i
determining z in terms of
if
does not pass through a zero of M^
On
differentiation,
may
be any curve which
g.^t g^.
we get
(i/ = 4?'-</,f-...
and so
where a
*
The
^=Sf> (^
is
oc),
a constant.
difficult
problem of establishing the existence
g^ are given is solved in 21 73.
t This equation in
it
always has solutions, by
20-13.
of such
numbers
w^
and Wj when g^ and
438
Make ^> x
of the function ^;
The
^>0,
then
i.e.,
z=l
result that the equation
the equation
^=^{z)
is
An
20*222.
illustration
results in the
^=^j(2
gd
(/i)~-dt
the theory
in
a pole
equivalent to
is
of the circular functions.
20'2-20'221
>;
from the
is
in the form
may
be illustrated by the corresponding

Thus we may deduce the properties
theory of the circular functions.
of the function cosec- z
so a
+ n,^n) = ^(z).
f,
from
The theorems obtained
so
{4:t'^
and
,i,
sometimes written
and
since the integral converges,
of the form n,^
is
XX
[CHAP.
{z mTT)~^ in the following manner
series
m=-a)
Denote the
to
z)
series
by/ (2);
except near the points
the series converges absolutely and uniformly* (with regard
mn
which
at
it
Except at these
obviously has double poles.
The effect of adding any multiple of tt to ^ is to give a

the same as tho.se occurring in the original series; since the
points, /(j) is analytic.
series
whose terms are
series
converges absolutely, the
of
z icith
sum
of the series is unaffected,
and 80/(2)
a periodic function
is
period n.
Now consider the behaviour oi f{z) in the .strip for which -\7r ^R{z)^\n. From
the periodicity of f{z), the value off{z) at any point in the plane is equal to its value at
the corresponding point of the strip. In the strip/ (2) has one .singularity, namely 2 = ;
and /(2)
bounded as
is
2-*-oc in the strip, becau.se the
terms of the series for/ (2
are
ac
small compared with the corresponding terms of the comparison series
2'
m~^.
m= - 00
In a domain including the point z=0, f{z)-z~^
and consequently there is a Maclaurin expansion
/(2)-2-2= 2
valid
when
'
z'
<n.
analytic,
is
and
is
an even function;
a,z^^
It is easily seen that
= 27r--(2 + l)
a2
and so
o=^
Hence, for small values of
2 jn-'"~-,
m=l
a2 = 67r~* 2
in'*^^.
m=l
f{z) = z-^ + ^ + ^\z^ + 0(z*).

Differentiating this result twice,
is
to
say, the
obviously periodic.
also squaring
it,
we have
f"(z)=ez-^+^+0{z%
^Hz)=z-*+p-^+n+o{z^).
/" (2) -e/'-i (2) + 4/(2)= (z^).
It follows that
That
and
function
Since
its
/"
(2)
6/^ (2) + 4/" (2)
from the periodic property of the function that

*
is
analytic at the
origin
only possible singularities are at the points mn,
By comparison with
it is
the series
2'
m=-
m~'^.
30
and
it
it
is
follows
ELLIPTIC FUNCTIONS
20*222]
Furthei*,
bounded as
it is
and so is* f"
îr^R{z)^^7r,
z-s-qc in the strip
since f{z)
Hence/" (2) 6/^ (3) + 4/(2) is bounded in the strip, and

is bounded everywhere.
By Liouville's theorem ( 5"63)
(z).
its periodicity it
By making
a constant.
439
z^~0, we see that the constant
zero.
is
bounded
is
therefore from
it is
therefore
Hence the function
cosec^ z satisfies the equation
f"(z)
Multiplying by 2/'
= 6fHz)-4f{z).
and integrating, we get
(s)
f^^z) = 4P{z){f(z)-l}+c,
where
for/'
a constant, which
c is
(2)
We
is
on making use of the power series
easily seen to be zero
and/ (2).
thence deduce that
2s
f'^ {t-l)~i
dt,
J fiz)
when an appropriate path

Example
If
1.
j/
=^
of integration is chosen.
(2)
and primes denote difierentiations with regard to
2,
shew that
4^-|j3=tV{(y-î)-H0/-.'.2)-H(3/-.3)-^}-|y(3/-î)-'(y-^2)-V.y-^3)--\
where
ej, e^, e^
are the roots of the equation Afi
[AVe have
y"^
got gz = ^-
= 4f-go,y-g^
Differentiating logarithmically
and dividing by y\ we have
r=\
Differentiating again,
Adding
this
multiplied by ^^,
It
we have
2y"'
4?/"2
y^
y^
r=i
'
equation multiplied by j to the square of the preceding equation,

we readily obtain the desired result.
should be noted that the left-hand side of the equation
derivative t of
with respect to y
and so
is
is
half the Schwarziaii
the quotient of two solutions of the
equation
c^+
Example
Ire
(//-O^-gy^n
(.-..)
,.!,
.=0.]
Obtain the 'properties of homogeneity of the function ^(2); namely that
2.
'
^^H>''')=^''^f^'l'"')'
A2/
CO2/
\
\
^^^^; ^~'92, X-V3) = X-'&?(--;
<72,5'3),
where ^(s
denotes the function formed with periods
2a)2
2q)i,
and ^(2;
g^, g^)
denotes the function formed with invariants g-n g^-
[The former
latter
is
may then be
*
The
a direct consequence of the definition of
{z)
by a double
derived from the double series defining the g invariants.]
series for
/"
(s)
may
be compared with
2'
t~'*.
m= -
t Cayley,
Gamh. Phil. Tram.
xiii. (1883), p. 5
[Math. Papers,
xi. p. 148].
series
the
440
The addition-theorem for
20"3.
[chap.
XX
the function ^{z).
The function ^ {z) possesses what is known as an addition-theorem that

to say, there exists a formula expressing ^J {z + y) as an algebraic function
;
is
oi^{z) and ^(y)
for general values*
and
<){
?/.
Consider the equations

i^'{z)
which determine
z
y (mod. 2wi,
Now consider
and
= Af{z) +
B,
<,j'{y)Âf{^j)
+ B,
and y unless ^{z) = <^{y),
in terms of ^
i.e.
unlessf
2&)2),
&>'
(0 - îP (t) -
^>
and consequently it has

It has a triple pole at ^ =
f.
and only three, irreducible zeros, by 20"13; the sum of these is a
period, by 20'14, and as ^=z, ^=y are two zeros, the third irreducible zero
must be congruent to - ^ y. Hence - z y h a. zero of ^' (^) - Af (^) B,
and so
î/a function of
three,
ô'{-z-y)=A<^{-z-y) +
Eliminating
and
and
B.
from this equation and the equations by which
were defined, we have

^{z)
-^'{z + y)
iîz^y)
=0.
f'{z)
Since the derived functions occurring in this result can be expressed

algebraically in terms of
result really expresses
It
is
^.>
{z),
^
{z
+ y)
î>
{y),
<fP
(z
+ y)
respectively
algebraically in
terms of
i^
( 20"22),
{z)
and
<^
(y).
therefore an addition-theorem.
Other methods of obtaining the addition-theorem are indicated in

examples
and
2,
and
+V
-I-
w=
20-311
may
be noticed, namely
then
0,
20*31.
20"312.
symmetrical form of the addition-theorem
that, if u
ij(u)
iîu)
&(v)
^'(v)
^ (lu)
^' (lu)
=0.
Another form of the addition-theorem.
Retaining the notation of

^' (^)
this
- A^ (^) - B
20-3,
we
vanish, are congruent
see that the values of

to
one of the points
z,
^,
which make
y,z y.
* It is, of course, unnecessary to consider the special cases when y, or z, or ij +z is n period.

t The function ^{z)-^(y), qua function of z, has double poles at points congruent to 2 = 0,
and no other singularities it therefore ( 20-13) has only two irreducible zeros and the points
;
congruent to
z=^y
therefore give all the zeros of
^ (z) -
^J (y).
20-3-20 '3 11]
ELLIPTIC FUNCTIONS
Hence ^'^(0
points
(O + B]And so
l-^fr^
z y.
z, y,
4j.nr)
vanishes
when
^j
(^)
when t
vanishes
441
is
- AY{K) - {ÂB + g.^ f {0 equal to any one of ^
is
(z),
congruent to any of the
+ 9s)
(B-'
{y),
(z
^j
(y)
and ^{z
+ y)
- {-lAB + g^)Z-
(B- +g,)
For general values of z and
y,
^j
(z), ô
y).
are unequal
and
so they are all the roots of the equation

4>Z^
A-Z--
Consequently, by the ordinary formula for the
sum
0.
of the roots of a cubic
equation,
^(^)
and
so
î.
+ iHy) +
y)
= lA%
+ y)=li^-^^^-^^,).ô(y),
on solving the equations by which

This result expresses
of
i^(2
^j
(z
+ y)
and
were defined.
explicitly in terms of functions of z
and
2/.
T}ê duplication
20'311.
formula for
<^^{z).
of the addition-theorem which have been obtained are both
The forms
nugatory when y = z. But the result of

given value of z, for general values of y.
result
when y approaches
From
this equation,
/ox
g (2z)
z,
we
see that, if
22- is
(z)
-J
-.
20*31 is true, in the case of any

Taking the limiting form of the
we have
W
= Irhm V-^-x
=
not a period,
{^
i^'
^-7
we have
h)\+ ~-y
o (z)
- 2p
/
11m
on applying Taylor's theorem to
^J
{z
+ h),
^j'
{z
+ h)\ and
so
^<^^>=i{|<f-^^(^>'
unless 2z
is
Example
qua function
a period.
1.
of
This result
is
called the duplicatioih formula.
Prove that
j,
has no singularities at points congruent with
use of Liouville's theorem, deduce the addition-theorem.
= 0,
i/; and,
by making
442
Example
Apply the process indicated
2.
example
iy)
^(2/)
in
[CHAP.
XX
to the function
I,
and deduce the addition-theorem.
Example
Shew
3.
that
^{z^.y) + ^{z-y) = W{z)-<p{:y)}-'-[{2^{z)ip{y)-\g^]{<>{z)-^^{y)}-g,\
[By the addition-theorem we have
(y) by 4^^ (^)_^^^ (2)_^3 and ^<&^y) - g^^

and reducing, we obtain the required result.]
Replacing ^'^{z) and

tively,
Example
{y)
- 9^
respec-
Shew, by Liouville's theorem, that
4.
j^W{z-a)^{z-h)] = <p{a-h){<p'{z-a) + '^'{z-h)}-iy{a-b){^{z-a)-^{z-h)].

(Trinity, 1905.)
Abel's* method of proving the addition-theorem for
20"312.
The
{z).
method of establishing the addition-theorem
following outline of a
instructive,
for
p (z)
is
though a completely rigorous proof would be long and tedious.
Let the invariants of p{z) be go,
take rectangular axes OX,
g^',
OY in
a plane, and
consider the intersections of the cubic curve

y^-
If
any point
(^î,
= Ax^-g.2X-g^
y = mx-\-n.
with a variable line

y^)
be taken on the cubic, the equation in
P{z)-x, = Q
has two solutions +2i,
Since P'^
which
{z)
î
( 20"13)
and
= ip^{z)g2P{z)gz, we
other solutions are congruent to these two.
all
have P"-{z)=yi^
choose
z-^
to
be the solution for
p' (2i)= -Hyi, not -y^.
number
Now
Zi
thus chosen
the abscissae î,
will
x.2,
x.^
be called the parameter of
(î, j/i)
on the cubic.
of the intersections of the cubic with the variable line
are the roots of

<^
ix)
~ 4x^ ^gi^gz- ("^-^ +nY= 0,
(^{x)=.A{x- x^) (x x,^ (x x^).
and so
The variation bx^ in one of these abscissae due to the variation in position of the line
consequent on small changes hn, 8n in the coefficients m, n is given by the equation
^
(b'
and so
(Xr)
S*v+ ^- 8m + ?^
*
(f)'
2
,
whence
(Xr) 8^r
8Xr
=1 mjCr
+n
Cm
Journal fiir Math.
ii.
= 0,
= 2 (mx^ + n) {x^tm
=2
0'
r=l
(1827), pp. 101-181;
-\-
8n),
Xr8m-\- 871
provided that Xi, x^, x^ are unequal, so that
1839), pp. 141-2.52].
8?t
en
<^'
(-^r)
'
{Xr)+0.
iii.
(1828), pp. 160-190 [Oeuvres,
i.
(Christiania,
elliptic functions
20-312, 20'32]
Now,
we put X (x 8)n + 8n)/4> (s), qua function
if
of
443
into partial fractions, the result
.r,
is
r=\
where
^,.= lim
.r
-p^
(^Sm + Sw)
= x^ {x^. 8m + 8n)
lim (x av)/^
+ 8n)l(f)'
:=.c,{x,8m
(^)
{x^\
by Taylor's theoi-em.
3
Putting x=0, we get
S.r,./j/,.
= 0,
i.e.
r=l
62,.
= 0.
r=l
That is to say, the sum of the parameters of the points of intersection

independent of the position of the line.
Vary the
parameters when the line
parameter
move
line so that all the points of intersection
coinciding during this process), and
a period of
is
Hence the sum
is
(2)
it is
evident that
21
and therefore
21
+ 22 + 23
is
equal to the
of the parameters of three collinear points
sum
of the
line is at infinity,
a period of
is
two points
off to infinity (no
+ 22 + 23
when the
the line at infinity; but
a constant
is
each
^ (2).
on the cubic
congruent to
is
This result having been obtained, the determinantal form of the addition-theorem
zero.
follows as in 20"3.
all
2032.
The constants
It will
now be shewn
unequal
equation
and,
62, 63.
that
^J (coi),
their values be
&>'
<^'
(&)i).
iî)
=~
i^'
(- î) = -
Similarly
Since
e^, e^, e^,
Since ^' {z)
so
<^'
^j((Oo), iîfOs),
then
(where 0)3=
e-^,
e.2,
e-,
co^ coo),
are
are the roots of the
g^t g^Ô.
4^*
First consider
and
if
e-^,
<^'
{z) is
an
i^'
an odd periodic function,

(2<yi
- i)
=-
wê
have
<^' (ft)i),
= 0.
fr''(&>i)
{w.^ =
(0)3) = 0.
(^'
elliptic function
at points congruent to the origin,
irreducible zeros.
is
whose only singularities are triple poles

{z) has three, and only three ( 20'13),
<^'
Therefore the only zeros of
f' {z) are points
congruent to
COj, &),, 6)3.
Next consider
a double zero at
i^{z)
ei.
This vanishes at
from 20*13 that the only zeros of
(^{z)
manner, the only. zeros of ^{z)-
^J(z)
gruent to
Hence
(1)2,
î
ois
(o^
and, since <^'{wy)
= 0,
Since i^{z) has only two irreducible poles,
(o^.
e-.,
ei
es
are congruent to
it
it
has
follows
In like
coj.
are double zeros at points con-
respectively.
^ 62 =^ ^s-
For
if gj
= e^,
then ^
(z)
e^
has a zero at
w.^,
which
is
a point not congruent to Wj.

Also, since ^'^ (z)
6)3, it
follows that
That
= 4^^^ (z) gz^ (z) g^ and since

(z) g2^ (z) g^ vanishes when
^''
4^^=*
is to say, e^, 62, e^
(z)
are the roots of the equation
4>f-g2t-gs =
0.
(z)
vanishes at
e^, 62
or
coj, to,,
e^.
444
From
[CHAP.
XX
the well-known formulae connecting roots of equations with their
coefficients, it follows that
e,
Example
that
When
1.
ex, e2 ^s
are
and
0)3
0)1 is
Example
real
and
2.
Shew
0,
{^t^-g^t-g-^~^dt,
=-
'
{g:i+g.it
- 4fi) ~ ^
dt,
a pure imaginary.
003
that, in the circumstances of
example
+ cos,
0, 0)3,
wi
of a half-period
to the
ai-gument of
From
shew
00
meter of the rectangle whose corners are

20*33.
is positive,
choosing them so that e-^>eo>
0,1=1
so that
^3
and the discriminant g.? %1g-^

63, shew that
g^ and^rg are real
all real
e.
T/ie addition
1,
{z) is
real
on the peri-
coi.
the form of the addition-theorem given in 20'31,
(z).
we have
,(,).,+,K)=H|-l;!-^r.
3
and
so, since
^J'-(z)
=4 n
{P{z)
er},
r=l
we have
^(-'+o,.)
= '^<";!l'î''"'"
-^ W-.
^J {Z)
Bi
on using the result
2 6^=0
r=l
this formula expresses
Example
1.'
Shew
2.
From
^ (s + coi)
in
terms of ^^(2).
that
^ (|6,i)=ei {(61-62) (61-63)}Example

shew that
the formula for ^{z + at.^ combined with the result of example
^ (^CO, +
Q).,)
= 61 + {(61 - 62) (61 - 63)}^
1,
(Math. Trip. 1913.)
Shew that the value of ^'(2)

the discriminant of the equation 4t^ got g3=0.
Example
3.
[Differentiating the result of 20%33,
(2)
results,
^' (2 + Wi)
(61
- 63)
^'
(2)
{if>
(2)
- 6,}-2
g)' (2
+ CO2) p' (2
-I-
+ W3)
CO3)
r=l
= 16(6l-62)M^2-3)M3-e,)^
is
(2
we have
= (61 -62)2 (62 -63)-^ (63-61)2^(2) n
which
we have
r (2 -f l) = - (61 - 62)
from this and analogous
^'(z-t-wi) ^' (s-l-wo)
the discriminant g^^ Tig-} in question.]
{p(2)-6,}-2
is
equal to
ELLIPTIC FUNCTIONS
20*33, 20*4]
Example
Shew
4.
445
that, with appropriate interpretations of the radicals,
^' (icoi)=
-2 {{e,-e.,) {e,-e^)]i {lêi-e^)^ + {e^-e^)h
(Math. Trip. 1913.)
Example
{p
Shew
5.
- e^}^ {p
{2z)
that, with appropriate interpretations of the radicals,
{2z)
- 63}* +
(22)
- ^3}* {P
We
shall
iP
- e^}^
m~e,}i
ip {2z)-e.2}^ = p{z)-p
The function*
Quasi-periodic functions.
20'4.
(22)
{2z).-
^(z).
next introduce the function ^(z) defined by the equation
dz
coupled with the condition lim [^{z)
^'^^^'
z~^]=0.
z~'^ converges
uniformly throughout any
domain from which the neighbourhoods of the points f ^'m,n are excluded, we
may integrate term-by-term ( 4'7) and get
Since the series for
l;iz)-z-^
{z)
= -n^{z)-z-^,dz
J
= -S' r{{z-n,,nr'-^7n%]dz,
711,
andso
+N
^(^)=
The reader
!__+
will easily see that the general
0(
and hence
n J
(cf 20*2), ^{z)
as
ni,;-=^)
is
in,^,,,
term of
this series is
-^co;
an analytic function of z over the Avhole
except at simple poles (the residue at each pole being
-H
sr-plane
1) at all the points
of the set i^m, tilt
is
evident that
m.n \Z
-T î-m,n
^îii,n
I2,j^,j)
and, since this series consists of the terms of the series for ^(z), deranged in
the same
way
as in the corresponding series of 20'21,
we
have,
by
2'52,
^{-z) = -^(z),
that
*
is
to say, ^{z) is
an odd function of z.
This function should not, of course, be confused with the Zeta-funetion of Eieraann,
discussed in Chapter xni.

t
(cf.
The symbol
20-2).
il'^
,^
is
used to denote
all
the points
fl,,,i
with the exception of the origin
446
Following up the analogy of
2~'+
defined by the series
^f
cori-eponding to
2041
The heading
proved, that
of
i^{z) is
^{z) of increasing z
by
2(i or
by
2w.. will
is
taking account of the fact that ^ {z)
manner,
^{z
(j)
C {y\ C
Example
2
is
is
is
1
1
evident from
view of the fact that
+ 277i,
z= ai^,
u.
and
2??2,
K (^2)if A--i-?/-|-2
= 0,
then
Stiekelberger, Journal fur Math, lxxxviii.)

It is not
not algebraic functions of (
(x),
(.?/),
a true addition-theorem since

C
(2)-]
Prove by Liouville's theorem that
2.
on
an odd function, we have
a pseudo-addition theorem.
(2) ai'e
p'{x)\^=âx+y + z)-i{x)-({jj)-C{.z).
(x)
p{x)
PLy)
PHy)
^(y)
F(i^)
^(2)
^(2)
^(2)
F'(2)
p^
{X)
'
It
now be
effect
{z\
Prove by Liouville's theorem that,
1.
[This result
i-^),
= - cosec^
1.
1-1
(Frobenius
bo considered.
2&)o)= ^{z)-\-
where
Example
and the
the constant introduced by integration; putting
is
like
.of z;
elliptic function, in
^(^+2a,,)=r(~0
In
cot
weget
27^1
-7;
i^{z).
not a doubly-periodic function
ip{z-^ 2&)i)
where
the equation
20'4 was an anticipation of the result, which will
the residue of ^{z) at every pole
now we
^{z) with the function cot 2
= ^ (2).
20*12 (III) that ^{z) cannot be an
If
+ (m7r)~i},
{(2-i7r)~'
quasi-periodicity of the function
77/ e
(2)
2'
we may compare
20-222,
J^
XX
[cHAP.
Obtain a generalisation of this theorem involving n variables.

(Math. Trip. 1894.)
20'411.
Tlie relation hetiueen
We
now shew
shall
ij^
and
rj.,.
that
1
To obtain
There
cell.
Hence
1:^
this result consider
one pole of
is
(z)
dz
27ri.
i^{z)dz taken round the
i^{z) inside
the
cell,
boundary of a
the residue there being
-I-
1.
20 '4 1-20 -421]
ELLIPTIC FUNCTIONS
447
Modifying the contour integral in the manner of

^-Ki
{K{z)
=and
r(^ +
rf^
27?2
which
is
{^ (^)
^(^
+ 2o}
(^^
iVi^î
'^Vi^-z,
The function a{z).
20-42.
We
dz
fZt
27/1
27rt
so
2a),)i
we get
20"12,
shall next introduce the function
{z),
j-^\og(7{z)=l;{z)
coupled with the condition lim [a{z)!z]
On
1.
account of the uniformity of convergence of the series for
near the poles of ^{z), we
may
t,{z),
except
Doing
integrate the series term-by-term.
so,
and taking the exponential of each side of the resulting equation, we get
'
l-^^-)exp
a(z)^zn'\[l(^) = ^n'
^'
z^
--m.n
2S2T.
the constant of integration has been adjusted in accordance with the condition
stated.
By
the methods employed in
obtain the following results
will easily
202, 20-21, 20-4, the reader
The product for a (z) converges absolutely and uniformly

bounded domain of values of z.
(I)
The function a(z)
(II)
at
all
is
in
any
an odd integral function of ^ with simple zeros
the points flm,n'
The function a
(z)
may be compared
with the function sin z defined by
the product
z
= - CO
the relation
20-421.
If
we
-j-
log sin z
= cot z
corresponding to
-i-
The quasi-periodicity of the function
log
o-
(z)
^(z).
cr{z).
^{z
we get
where
(_
a
c is
(z
+ 2(o,) =
+ 2(Wi) =
!:{z}+2v
ce-^'^-a (z),
the constant of integration; to determine
then
(coi)
= ce~-''''">o- (coi).
c,
we put
= coi,
and
448
Consequently
In like manner
e-'''"'
gî^'z-i-'xj)
These results exhibit the behaviour of
If,
XX
o- (z + 2&)j) = - e2''i(^+'-i) a {z).

(t(z + 2<i)o) =
a (z).
and
period of
[CHAP.
(t{z)
when
is
increased by a
(z).
as in 20"32,
we
wTite
ais
= co^
then three other Sigma-functions
ay.,,
are defined by the equations

a, {z)
The
= e-'i'-'o- (z +
V (wr)
(?
!, 2, 3).
four Siguia-functions are analogous to the four Theta-functions dis-
cussed in Chapter XXI (see
Example
Shew
that, if
21'9).
m and
<r{z + 277i(Oi + 2na>2) = )"' + " a- (z)

(
and deduce that
r]xi>ii-
r]-i<Ji\
Example
Shew
that,
then
&),.)
2.
an
/^(2) is
integi-al
is
n are any integers,

exp
{{27nrii
if
5'
= exp {iriu)-!! o>i)i
function with the
doubly-periodic function of
+ 2nr).>) z + 2m-r)^<3>i + 4??i;j,a)2 + 2n'^r\<ia>^
an integer multiple of
with periods
so that
same
\ni.
< 1,
and
zeros as (t{z)
if
and also F{z)la-{z)
is
2a)i, 2a).2-
Example
3.
Deduce from example
Example
4.
Obtain the result of example 3 by expressing each factor on the right as
2,
by using
Liouville's theorem, that
a singty infinite product.

20"5.
functions
Formulae expressing any

same periods.
elliptic
function in terms of Weie7'strassian
ivith the
There are various formulae analogous to the expression of any rational

two sets of products of linear factors, (II) a sum
fraction as (I) a quotient of
of partial fractions
of the first type there are
functions and Weierstrassian
type there
will
is
now be
20*51.
two formulae involving Sigma-
elliptic functions respectively;
a formula involving derivates of Zeta-functions.
of the second
These formulae
obtained.
The expression of any
terms of ^{z) and
\(>'
{z).
Let f{z) be any elliptic function, and let i^ {z) be the Weierstrassian
formed with the same periods 2&)i, 2&)o.
elliptic function
We
first ^^îte
fiz)
= I Uiz) +/(- z)] +
[{f{z)
-f{- z)\ Wizfr] ^'{z).
20*5-20 'SS]
'
ELLIPTIC FUNCTIONS
44!^
The functions
f{z)
+/(- z\
-/(-
[f{z)
^)]
Wi^)]-'
are both even functions, and they are obviously elliptic functions
an
when/(^)
is
elliptic function.
The
solution of the problem before us
any even
expiêss
elliptic
Let a be a zero of
- a
to
be a
will also
arranged in two
^2
>
In
hi,
Cln
(f>
any
(z) in
The
zero.
say
a^, a^,
cell
i^)
irreducible zeros of
(z)
cell
may
congruent
therefore be
and certain points congruent
...an
we can
(z).
then the point in the
to
a-^,
and certain points congruent
bn,
therefore effected if
is
say, in terms of
<^ (2),
may be arranged in two

60, ... 6.
manner, the irreducible poles
like
...
b..,
sets,
function
sets,
say
to 61,
Consider now the function*

1
H')(^)-iP("r)
fi
d>{z)-, = i
It is
(b
an
elliptic function of
and
^)
clearly
{br)}
it
has no poles
are zeros f of the numerator of the product, and
(z)
denominator
theorem
oi"
the product are polesf of
<^
to say,
is
for the zeros of
the zeros of the
Consequently by Liouville's
4>(z).
a constant, A^, say.
it is
Therefore
that
z,
{Z)
iiO
<f)
(z)
(.)
AH
|^^^^?^^1
has been expressed as a rational function of
^J
(z).
Carrying out this process with each of the functions
f{z) +fi-z),
we obtain the theorem that any

of
-f{-z)}
{f(z)
{ip\z)}-\
function f (z) can be expressed in terms,
elliptic
^ (z) and p' {z) luith

and linear in <^' {z).
the Weierstrassian elliptic functions
the expression being rational in
{z)
20"52.
The expression of any
Zeta functions and their derivates.
Let f{z) be any
elliptic
elliptic function
irreducible poles of f(z) be i,
c/o,
...
same
periods.,
function as a linear combination of
with periods
a,
the
and
let
2coi,
2(iJo:
Let a set of
the principal part (5*61)
off{z) near the pole a^ be

^k,i
* If
factor
of
(j)
îi>
[z)
any one of the points

(rt,.) or
^J (^) - ^
(2) ^.'
(z- a^f
number
a,,
Ck,
'"
{z-
rje
ttkY^
we omit the corresponding^

product and the zero (or pole)
or h^ is congruent to the origin,
(&,.).
at the origin are then of the
20-5o, factors
Ck,2
z-ajc
The zero
same order
(or pole) of the
of multiplicit.y.
In this product, and in that of
corresponding to multiple zeros and poles have to be repeated the appropriate
of times.
f Of the same order of multiplifity.
W. M. A,
29
450
Then we can shew
f{z)
.4,
+i
k:
c l
where
A., is
[CHAP.
XX
that
\c,,,^{2
- (/,) -Ct,,r (^ -
"X")
+ .
denotes -r-
^{z).
(.
a constant, and
^^^^
(z)
Denoting the summation on the right by F{z), we see that
F{z + -2co,)-F{z)= I
27J,Ck,^,
k=l
by
20"41, since all
the derivates of the Zeta-functions are periodic.
But S
is
C/cî
the
sum
of the residues of
of
poles in a
y'(2^)
at
all
and similarly
it
has period
its
cell,
k=l
and
consequently
is
20'12) zero.
Therefore F(z) has period
f{z)
F (z)
is
an
2&)i,
2coo;
and so
elliptic function.
Moreover F{z) has been so constructed that f{z) F{z) has no poles at
It is
nj, o, ... a; and hence it has no poles in a certain cell.
the points
consequently a constant,
Thus
the function
A.,,
f {z)
by
Liouville's theorem.
A=is=i(s
This result
is
known
i;
of importance in the
when the principal

we obviously have
function f(z)
for
f(z)dz = A.z+ 2
Ck,i\og
part of
problem of integrating an
its
expansion at each of
elliptic
its
poles
is
a (z - ak)
k=i
where
C is
Example.
a constant of integration.
Shew by
the method of this article that
and deduce that
where
C is
2053.
a constant of integration.
TJie
expression of
any
elliptic
function as a quotient of Sigma-
fiinctions.
Let f{z) be any
elliptic function,
of irreducible zeros of /(^) be
with periods
a,, a^, ... an.
2&)i
and
2a)2,
and
let
a set
Then (20'14) we can choose a
20 '5 3]
ELLIPTIC FUNCTIONS
set of poles bi,bo,
other of
0^/(2) are congruent to one or
hn such that all poles
...
451
them andf
+ a2+
a^
Consider
now
...
= bi +
an
b.2
...
+bn.
the function
(r(z-
ar)
This product obviously has the same poles and zeros as f{z)
effect of increasing z by 2(0j is to multiply the function by
^
,.=1
The function
and
2&)2),
is
an
exp
exp
[2771
{277i
therefore has period
{z
g,.)}
(^
br)]
also the
^^
(and in like manner
2&)i
it
has period
so the quotient
elliptic function
be a constant,
By
with no zeros or poles.
Liouville's theorem, it
must
A-^ say.
Thus the function/' (2^) can be expressed
in the form
r=l(T{Z-b,)
An
elliptic function is
constant)
when
Example
1.
Shew
Example
2.
Deduce by
and by further
consequently determinate (save for a multiplicative
periods and a set of irreducible zeros and poles are known.
its
that
from example
difFerentiation,
1,
that
differentiation obtain the addition-theorem for
(z).
II
Example
If 2 a,.=
3.
r\.
22
=1
r=]
b,.,
6,.,
shew that
(r{ar-bi)(r(ar-b.^...(T{ar-bJ _^
- ai) o- (a,. - a J
o- (a^ - a)
^
,.=1 o- (a,.
the * denoting that the vanishing factor
Example
Shew
4.
o-
(a^
. .
. .
a^)
is
that
^{z)-e,. = a/{z)la'^{z)
[It is
customary to define
Example
d
{z
+ a)
Establish,
5.
a-
{z
- a)
a-
'
to be omitted.
{b
{g? {z)
e^}^
by example
+ c)
a-
{b
to
I,
(r=l,2,
mean
the
'
o-,.
(2)/(r (2),
3).
not
o-,.
(s)/'o- (2).]
three-term equation,' namely,
- c) + <T {z + b) a-
{z
0-
b)
{z
a-
{c
+ a) a
{c
- a)
+ c)(t{z-c) (r{a + b)a {a-b) = 0.
t Multiple zeros or poles are, of course, to be reckoned according to their degree of multiplicity ; to determine b^, h-i, ...b^, we choose 6i bo,, ... b,^_i 6,/ to be the set of poles in the cell in
which ai, a-i, ...a lie, and then choose &, congruent to 6,/, in such a way that the required
,
equation
is satisfied.
292
452
[This result
due to Weierstrass
is
The equation
Fonctions ElUptiques,
Sigma-function
see p. 47 of the edition of his lectures
characteristic of the Sigma-function
is
I.
(Paris, 1886), p. 187, that
an equation of this type.
satisfies
The connexion between
20"54.
by Schwarz.]
has been proved by Halphen,
no function essentially diflferent from the

See p. 461, example 38.
hvo
ani/
it
XX
[CHAP.
the
same
an algebraic relation
exists
functions with
elliptic
periods.
We
now prove the important
shall
any
bettveen
by
For,
functions, f{z)
ttuo elliptic
we can
20*51,
where Ri and
jB,
= R, [p (z),
io'
(z)
and
with the same periods.

the
(z) as rational functions of
with the same periods, so that
^' (z)
(z)],
</,
(z)
^' (z) algebraically
= R, {^ (z),
^' (z)],
from these two equations and
= 4>f'{z)-go_&{z)-g
^'Uz)
is
<f)(z),
denote rational functions of two variables.
Eliminating ^
we
and
express f(z) and
Weierstrassian functions ^ (z) and
f(z)
result that
obtain an algebraic relation connecting f{z) and
(f>
(z)
and the theorem
proved.
particular case of the proposition
nected with
If
now
its
Ave
is
that every elliptic function
con-
is
derivate by an algebraic relation.
take the orders of the elliptic functions f(z) and 4>{z) to be

to any given value of/(^) there
and n respectively, then, corresponding

(
2018) a
set of
iiTeducible values of
values (in general distinct) of

are ni values of
The
in
(f)
relation
and
The
cf)
n in
(2) = ^2 ^^^^
(j>
(z)
As an
(z).
and consequently there are
So, corresponding to each value off, there
and, similarly, to each value of
between f(z) and
cf)
(f)
correspond n values of /.
(z) is therefore (in
may be
of lower degree.
of order 4, the relation is/-
general) of degree
of order
degree 3 in f; this
is,
The
8.
in
f;ict,
= p (z),
Thus, iff(z)
of order
2,
and
2,
and
cf).
illustration of the general result take
= ^y (z),
is
relation
(j)
cf)
z,
f(z)
= ô{z),
of order
relation should be of degree 2 in
the case, for the relation
is <^-
(/>
and of
= 4/'-' ^.2/ ^3.
Example. If u, v, ic are three elliptic functions of their argument of the second order
with the same periods, shew that, in general, there exist two distinct relations which are
linear in each of ^l, v, w, namely
A uvtv+Bvw + Cicu + Btiv + E u + F r + G iv+ H =0,

A'uvw + B'vw + C'lvti + D'u v + E'u + F'v + G'w + H' = 0,
where
^1
B,
On
20"6.
It will
insoluble
IT' are constants.
the iy\tegration
now be shewn
by means
of
[aoX*'
+ ââf' +
Qa^x^
+ 4ea^x + a^ ~^.
that certain problems of integration, which are
of elementary functions only, can be solved
duction of the function
<^ {z).
by the
intro-
20'54, 20*6]
ELLIPTIC FUNCTIONS
Let ttoX^ + ââf + Qa.^x- + â^x

which has no repeated factors; and
g
gs
if
to
aâoCti
a^
=f(x) be any
let its
= ciocti 4aia3 +
polynomial
quartic
invariants* be
3a./,
2aia2a3
ai
aâ^^
a-ca^.
-1
/'*
Let
453
{/(t)]
'dt,
where
a-'o
is
any root of the equation/(a;)
the function ^ {z) be construe tedf with the invariants g^ and

express x as a rational function of <,p{z go, g^).
g^, it is
then,
possible
The reason
[Note.
assuming tbat/(.r) has no repeated factors
for
is that,
when/(.r)
has a repeated factor, the integration can be eftected with the aid of circular or logarithmic
functions only. For the same reason, the case in which aQ = ai = need not be considered.]
By
Taylor's theorem,
f{t)
(since
(xq)
we have
= 4^3 {t - X,) + QA. (t - x,y + 4>A, (t -
x,y
^'oY,
= 0), where
A = UoXq^ + 2aiXQ +
As = o^V'' + 'ÛiXq- + â.2Xo +
a.,,
On
+ ^0 (^ -
writing
(t
Xa)~^
t,
Xq)~'- =
(x
{4.43T=^
1
'
we have
+ QA,T^ + 4A,T + Ao] ~ irfr.
To remove the second term
in the cubic involved, write^
= Ar(cr-h^.^
^,
ttj.
^=A,-^(s-iA,),
and we get
r
z^
['ia'-{SA.^-Â,A,)a-{2A,A.,A,-A,^-AoAs')]~^-d(T.
.
The reader
will verify,
3i4.;-
difficulty,
that
- 4^1^ 3 and iA-^Â.Â^-
are respectively equal to
and
without
g.^
and
g-^,
A^ - AÂ^^
the invariants of the original quartic,
so
s=io{z;go,g.;).
Now
and hence
so that
*
= Xq + Az
= Xo + \f' (ô)
{{^
{z
\s
',
Â.^~'^,
92, Qi)
- -hf" (^'o)}~S
X has been expressed as a rational function of (^
Burnside and Pauton, Tlieonj of Equations,
ii.
(z
g.,,
g^.
p. 113.
+ See 21-73.
J This substitution
is
legitimate
having x Xq as a repeated root.
since
A3 + O;
for
the
equation
^-13
involves /(x)
454
This formula
for
cc
XX
be regarded as the integral equivalent of the
to
is
[CHAP.
relation
Example
1.
With the notation
Example
2.
Shew
of this article,
shew that
that, if
where a is ani/ constant, not necessarily a zero of f{x), and / (x)

with no repeated factors, then
,
the function
p (z)
a quartic polynomial
{/()}^F(^)+i/'W{^(^-)-^V/"(^)}+A/()/'"()
is
being formed with the invariants of the quartic /(:c).

(Weierstrass.)
[This result was
first
published in 1865, in an
_
where /(:f, y)
is
C'^v y dx
by
due to Mordell, Messengery
Inaugural-dissertation at Berlin
An alternative result,
Biermann, who ascribed it to Weierstrass.

XLIV. (1915), pp. 138-141 is that, if
X dy
a homogeneous quartic whose Hessian
is
{x, y),
then we
may
take
x=ap'{z)^ff+ip{z)f,+^h
y-hp-{z)^lf-\p{z)f,-Ua,
where /and h stand
Example
3.
(.
^
for /(a, 6)
Shew
M
^'
that,
and A
and
(a, 6),
denote partial dMferentiations.]
suffixes
with the notation of example
,rw
/'()
4(.r-a)
{/w/()}^+/()
I
^{x-af
2,
24
'
- 1,^3 - rr^l{/()}^ - j/-^3

F(^)=
*
{{x-af
{{x-af A{x-a)-y-'
and
'
'
20"7.
The uniformisation* of curves of genus
The theorem
If
of 20*6
the variables
may be
stated
+/Û
A{x-ay]
{/(-)}*
^'
^'
^
unity.
somewhat
differently thus
x and y are connected by an equation of

y"- = aîd^ + â-iâ? + â^x^ + â^x + a^,
tlie
form
then they can be expressed as one-valued functions of a variable z by the
equations
^
.^
x=x,+\r{x,)\p {z) - j-j" {x,yr

,
y = -U'{ô)p'{z){p{z)-i-J"{xo)]
where f{x) = a^x^ + 4aia^ + Qa.,x^ + 'ia^x
a^, Xq is any zero of f{x), and the
function <^{z) is formed with the invariants of the quartic ; and z is such that
-t-
z=r [f{t)]-^dt.
*
This term employs the word uniform
in the sense one-valued.
To prevent coufusion with
the idea of uniformity as explained in Chapter in, tliioughout the present work we have used the
phrase 'one-valued function' as being preferable to 'uniform function.'
ELLIPTIC FUNCTIONS
!07]
It
is
obvious that y
fact,
x and
a two-valued function of
is
and the
function of y
functions of the variable
455
z,
a;
is
a four-valued
that x and y can be expi'essed as one-valued
makes
importance
this variable z of considerable
in the theory of algebraic equations of the type considered
called the
is
uniformising vmîahle of the equation

y"
The reader who
= a^x*
-f-
4ai.'C^ -f â.^x"
} â^x + a^
acquainted with the theory of algebraic plane curves will be aware

number whose geometrical
is
that they are classified according to their deficiency/ or genus*, a

significance
is
that
it
by the curve and the
is
the difterence between the
maximum number
number
of double points
of double points possessed
which can be possessed by a
curve of the same degree as the given curve.
Curves whose deficiency
is
of a unicursal curve, it is well
zero are called tmicursal curves.
known t
that x and
?/
If/ (x, y) =
Since rational functions are one-valued, this parameter
of a 'parameter.
is
the equation
can be expressed as rational functions
a uniformising
is
variable for the curve in question.
Next consider curves of genus unity;
let /(.r,
?/)
be such a curve; then
it
has
been shewn by CIebsch| that x and y can be expressed as rational functions of ^ and
where ry' is a polynomial in | of degree three or four. Hence, by 20-6, ^ and r\ can be
expressed as rational functions of ^ (2) and ^' {z\ (these functions being formed with
?;
suitable invariants),
which
is
and so x and y can be expressed as one-valued
functions of
(elliptic)
z^
therefore a uniformising variable for the equation under consideration.
When
the genus of the algebraic curve /(.r,
sation can be effected by
means
of
3/)
greater than unity, the uniformi-
is
what are known as automorphic
Two
functions.
of such functions of genus greater than unity have been constructed, the first
classes
by Weber,
Oottinger Nach. (1886), pp. 359-370, the other by Whittaker, Phil. Trans, cxcii. (1898),
pp. 1-32.
The analogue
of the period-parallelogram
In the case of Weber's functions this polygon
is
known
as the 'fundamental polygon.'
mviltiply-connected,'
i.e.
it
consists of
region containing islands which have to be regarded as not belonging to
it
is
'
the case of the second class of functions, the polygon
no such
islands.
The
latter class of functions
is
may
'
simply-connected,'
i.e. it
contains
therefore be regarded as a
immediate generalisation of elliptic functions. Cf. Ford, Introduction

morphic Functions., Edinburgh Math. Tracts, No. 6 (1915).
to
a.
whereas in
more
theory of Auto-
REFERENCES.
K. Weierstrass, Werke,
i.
(1894), pp. 1-49,
11.
(1895), pp. 245-255, 257-309.
Briot et J. C. Bouquet, Theorie des fonctions elliptiques. (Paris, 1875.)

H. A. ScHWARZ, Formeln und Lehrsdtze zuni Gehrauche der elliptischen Funktionen. Nach
Vorlesungen und Aufzeichnungen des Herrn Prof. K. Weierstrass. (Berlin, 1893.)
A. L. Daniels, 'Notes on Weierstrass' methods,' American Journal of Math. vi. (1884),
vn. (1885), pp. 82-99.
pp. 177-182, 253-269
J. LiouviLLE (Lectures published by C. W. Borchardt), Journal fiir Math. Lxxxviii.
C.
(1880), pp. 277-310.
A. Enneper, ElliptiS'-he Funktionen.

J.
Tannery
et J.
Molk, Fonctions
(Zweite Auflage, von F. Miiller, Halle, 1890.)
Elliptiques.
(Paris, 1893-1902.)
French genre, German Geschlecht.

t See Salmon, Higher Plane Curves (Dublin, 1873), Ch.
A proof
X Journal ffir Math. lxiv. (1865), pp. 210-270.
*
11.
of the result of Clebsch is given
Forsyth, Theory of Functions (1918), 248. See also Cayley, Proc. London Math. Sac.
pp. 347-352 [Math. Papers, vni. pp. 181-187].
by
iv. (1873),
TH E TRANSCENDENTAL FUNCTIONS
456
[cHAP.
XX
Shew that
1.
9{^'ry)-^{z-y)=-&'{z)^'{y){ip{z)-^{y)]-\
Prove that
2.
where, on the right-hand side, the subject of diflferentiation
symmetrical in
is
y,
2,
and w.
(Math. Trip. 1897.)
Shew
ti.
that
n^-y)
_ 55^2
1
r'o^-^)
r\y-^'^
r"(2-y)
(- y)
r"(3/-"')
r'('-2)
(y-Ô
(w'-^)
(Trinity, 1898.)
dy
.y'=^
,
y=^(2)-ei,
If
4.
shew that y
one of the vahies of
is
if
\i
\ d'^
|/(y-4^2logyj
+(ei-e2)(ei-e3)|
(Math. Trip. 1897.)
Prove that
5.
(2)
- } (P
(y)
where the sign of summation

roots
-P
(')}'
IP (y +)
refers to the three
- 4* (P (y - '^) - )* = O'
arguments
2,
y,
v.;
and
e is
any one of the
Ci, Co, So.
(Math. Trip. 1896.)

6.
Shew
that
P'{z + <oi)_
(P (ii)-(i)r
........
P'(^)
7.
(Math. Trip. 1894.)
Prove that
P (22) - p (a,,) = IP' (2)}-2
[^
(2)
- ^ (|a)}2 {^ (2) - ^ (a>2 +*o,i)}^.

(Math. Trip. 1894.)
8.
Shew
that
p{u + v)iO(u-v) = iP
(u)
p (v) + ig^l+giilMllMl
,,.V2
{p{u)-p{v)
(Trinity, 1908.
9.
And/i
If
p(u) have primitive periods
(m) are similarly
2(2
2a)i,
constructed with periods
Pi{ii)
= P{ti)+"'2
and
2a)i/yi
f {u) = {p {u) - p
and
{p{u + '2ma>iln)
2a)2,
{co-^}'^
while
g:)i
(?0
prove that
-p {2mcoi/7i)},
m=l
n-1
and
/i (w)
n f{u + 2ma)iin)
= "^^î
n f{2mailn)
m=r
(Math. Trip. 1914 the first of the formulae

Journal fur Math. Lxxvi. (1873), p. 39.)
;
is
due to Kiepert,
ELLIPTIC FUNCTIONS
If
10.
where a
where
is
.r
= p{u + a),
457
y = <p{u-a),
constant, shew that the curve on which
(.r,
y) lies is
{xy + ex + cy + ^g.^f = 4 (^ +y + c) [cxy - lg^\
= p(2a).
(Burnside, Messenger^ xxi.)
Shew that
11.
2^"3
(.0
- Zg."^ {u) +gi = 21
{^'^
{u)+g,Y.
(Trinity, 1909.)
z=r
If
12.
{xi
+ 6cx^ + e^-)-idx,
x=^^^,
verify that
the elliptic function being formed with the roots c, ^
{c
+ e),
(c
e).
(Trinity, 1905.)
If
13.
be any constant, prove that
/e'{^(2)-S'(2')fp(3)(;2
F(i/)i
êMS'(2)-S'(y)}^y
^,^^^
"
^{z)-<P{y)
''
^(2)-^(y)
-I?
where the summation
refers to the values
3 of
1, 2,
IT
{P(^)-r}{P(y)-er}
and the
'
integrals are indefinite.
(Math. Trip. 1897.)
and
R {x) = Ax^ + Bx"^ + Cx"- + Dx + ^',
Let
1 4.
let
|=^ (.r)
be the function defined by the equation
.^'=|^^(^)}-^^/^,
where the lower limit of the integral
20' ()
<j>'
a-\-y)
is
^' ia )
<^(o+^)-0(a)
(^(A'+2/)-0(a)
Shew
arbitrary.
0'
that
(a-y)+<^'(a) _ <^'(a+y)-<^'(^)
(]i{a+y)-(j) {x)
(i>{a-y)-<f>{a)
0'(-y)-0'(-^')
'
c}){a-y)-cl){x)
[Hermite, Proc. J/^/^ Congress (Chicago, 1896), p. 105. This formula is an

addition-formula which is satisfied by every elliptic function of order 2.]
Shew
15.
is
that,
when the change
of variables
applied to the equations
r + .(l+i>^) + l^ = 0,
a.-^-^^ =
0,
they transform into the similar equations
Shew
is
that the result of performing this change of variables three times in succession
a retiirn to the original variables
functions of
where
is
^^
by
-Eiû)
and F{u)
r^
and hence prove
that, if ^
and
r)
be denoted as
one-third of a period of the functions E{u) and F{u).
E (u) = ^-P{it;
Shew that
where
^,
respectively, then
92
= '2p + p\ 93=
g2, g^),
-'i--
qP^-^P^-
(De Brun, Ofversigt af K.
Vet.
Akad., Stockholm, Liv.)
458
[chap.
XX
Shew that
16.
&'{^
= 2o- (2 + Wj)
F (^)=
aud
60- (z
(T
(2
C (ft)i)
O-^ (s)
(Z
<t>2) <T
(T (0)2) O"
0)9)
i
(i +0)2)
+ a) a{z-a)(r{z + c) a{z c)
(r*(2)<r2(a)o-2(c)
where
(Math. Trip. 1913.)
Prove that
17.
+ &'(a-h){C{z-a)-a'-b) + C{<')-ah)]
(Math. Trip. 1895.)
Shew
IS.
that
2l^(M)-^(t^)
^^^^
'^^^
^^
gJ(i;)-^(M')J
''^
(Math. Trip. 1910.)
Shew that
19.
C(i)
+ C,W2) + C("3)-r(W] +
2 {^
(?î)
*2
+ 3)
g> (M2)}
F (l) W (2) - ^ (Ws)} + ^'

Shew
20.
{^
(-2)
(^^2)
{P
- 9 {n^)]
-
(3)
<r {1/
- z) a {z -
x)
^1
a3(.r)cr3(y)cr3(2)
(%) {^ (l) - ^ (2)}

(Math. Trip. 1912.)
i;-''
{z)
^ ^^^
^, ^^^
p{z)
p'{z)
from this
^'
(-0
i-^)
result.
induction, or otherwise, that
_/_xiM(n-l)j
'
2^
the addition-theorem for the function
Shew by
21.
- P (^<l)}
(3)
that
a{x+y+z) o- {x-y)
Obtain
{P
^^ (l)}
O-
,
I
+2 +
(20
. . .
+ 2n)
no-
(2;^
- Z^)
... ?i
'<T^^^z^)...a"*'{z)
^(î)
PizJ P'(2)...p-1)(2)
^'(î)...^<"-^)(2i)
where the product

that X
<
is
taken for pairs of
(Frobenius
22.
all
integral values of X
and
/i
from
to
71,
such
/I.
u. Stickelberger*,
Journal fiir Math, lxxxiii. (1877),
p. 179.)
Express
1
p{x)
P(^)
p'{x)
Piy)
F(y)
r(3')
^(2)
P(2)
F(^)
p{u)
p()
f (w)
as a fraction
*
whose numerator and denominator are products of Sigma-functions.
See also Kiepert, Journal
Lxxxn. (1877),
p.
346.
filr
Math.
L.xs.yi.
(1873), pp.
21-33
Hermite, Journal fiir Math.
'
ELLIPTIC FUNCTIONS
Deducethat
if
= p(.r),
8 = p{y), y=p{z),
- 63)
{(a
- ei)
(/3
- ei)
(y
+ (es - ei)
{(a
- e^)
(/3
- e,)
(7 -
+ (^1 - ea)
{(a
- 63)
(/3
(y
(^2
e^)
- e{)
=p
{8
- ej)}^
e^) (S
- 63)}*
- 63)
(ii),
459
where x + 9/+z+u = 0, then
- 63)}* = {e^ - 63)
(8
(^3
- ei)
(ej
- eg)-
(Math. Trip. 191L)

23.
Shew
that
2C(2)-4aiO = |J'^''^
3t(32i)-9C(20 =
(Math. Trip. 1905.)
24.
Shew that
and prove that
a-
{nu)j{cr ()}"' is
a doubly-periodic function of
?6.
(Math. Trip. 1912.)

25.
Prove that
a-
^{z-a)-({z-b)-C(a-b) + C{2a-2h)-
a-
(z-2a + b)
(26 2a)
o-
(z -2b + a)
a) o- (^ 6)
a-
(2
(Math. Trip. 1895.)

26.
Shew
that, if
+ S2 + ^3 + î = 0, then
{2C (2,.)}^ = 3 {2t
taken for r = l, 2, 3, 4.
Sj
(2,.)}
the summations being

27.
two
Shew
28.
+ 2^'
(.-.),
(Math. Trip. 1897.)
form
aiPiz +
b, rtj,
(^,.)}
that every elliptic function of order n can be expressed as the quotient of
ex2)ressions of the
where
{SP
02,
...
Taking
+ a,p'{z + b) + ...+a,,p(^^-')(z + b),
b)
(Painleve, Bulletin de la Soc. Math, xxvii.)
a^ are constants.
e^>e2>e;,
p{a>)
= ei,
p(co')
= es,
consider the values assumed by
C()-Mf (')/'
as
u passes along the perimeter
of the rectangle
whose corners are - co,
+ w',
a> + w
(Math. Trip. 1914.)

29.
Obtain an integral of the equation

1 d^w
,^
,
- -Ty = 6^(2)
+ 36
,
in the
form
where
c is
dzla{z)a{c)^^'^\b-2p{c)
^^^'^fj'
(62- 39^2)^(0) = 3
(63
+^3).
Also, obtain another integral in the form
^f
j^
^(i) +
where
and neither
ai
+ a2
nor ai-a-i
is
exp{-sf(ai)-2aa2)},
P (2) = &,
F (i) + P (2) = 0,
congruent to a period.
(Math. Trip. 1912.;
460
Prove
30.
XX
[CHAP.
tliat
a- {z + Z3) a- (z + Zj)
+ i(2i + 22 + 3 + 24)}
._a{z + Zi) (r{z + Z2)

^^^'~ '
,
r{22
is
(2)
+5r
z,
such that
+ coj) + (2+
+ (2+ 0)1 + C02)
= - 20- {| (20 + 23 - 2i - 24)}
(2
o).,)
(/
<7
or
{\ (23
+ 2i - Zo - 24)}
(T
(21
{J
+ 22 - 23 - 24)}.
(Math. Trip. 1893.)

If f{z) be a doubly-periodic function of the third order, with poles at 2
31.
2=C3, and
if
and poles at
(2).
= 0,
= Cj, 2 = 02,
be a doubly-periodic function of the second order with the same periods

being
: = 3, its value in the neighbourhood of 2 =
(^{z)
=
z
+ \^{z-a) + }<o{z-af -{...,
prove that
{/"
iX-'
(a)
-/" (3)} - \ {/'
()
+/'
2</) (^1)
+ {/ (a) -fm
{sXXi
+ 2ct>
{c,)
cj>
(C3)|
= 0.
(Math. Trip. 1894.)

If X
32.
be an
(2)
function with two poles
elliptic
aj, a^,
and
if z^,
z-j,,
...
22n
be 2n
constants subject only to the condition

Zl
shew that the determinant whose

1,
[where
X2(2,.),
\{Zi),
Xj
(zi)
+ Z2 + ...+Z.^ = 7l{ai + Cto),

ith
X,(2,),
\^{Zi),
...
row
is
X2
X(2i)X,(2i),
denotes the result of writing
Xi
(2,)
(2,),
Deduce from example 21 by a limiting
\F{z)
34.
(2)],
(2f)
vanishes
(Math. Trip. 1893.)
identically.
33.
X^-M^t) ^l
..
the derivate of X
for 2 in
2^
=(-)"-Ml!
...p-')(2)
P"{z)
p"{z)
r'(2) -P'^H^)
P-1)(2)
p)(2)...F""'K2)
Shew
process, or otherwise prove, that
2!... (-l)!}2cr(;m)/{a-(?0}'^.
(Kiepert, Journal fur Math, lxxvi.)
that, provided certain conditions of inequality are satisfied,
{z)(T{y)^
2coi
'
where the summation applies to
all
^"'1/
2(01
<"!
positive integer values of
m and n, and j = exp (7ria)2/<i)

(Math. Trip. 1895.)
35.
Assuming the formula

>)i
a-(2)
=e
s?
'
^sm^r-nz
2<x)i
2a),
.
- 2o- cos
+ q*"
tO]
:r-r,
7^
prove that
(P
when
2 satisfies
(2)=
- +
cosec2
cos
the inequalities
-2li(^)<R(^)<2R(?A,
\la>l/
\i(Oi/
V(Oi/
(Math. Trip. 1896.)
ELLIPTIC FUNCTIONS
Shew
36.
then X
that
if
2 is- be
any expression
461
of the form 2ma)i
+ 2n(02 and
if
a root of the sextic
is
,r- 55'2a;*-405'3A-3
and obtain
all
- 5g.^x^ - Sg.2gzX-bg^^ = 0,
the roots of the sextic.
(Trinity, 1898.)
Shew that
37.
/.-)(.-.)ri..=-iiog:-|-;-f;î.og:jj^>,
where
(Dolbnia, Darhoux' Bulletin
Prove that every
38.
analjiiic
function/ (3) which
satisfies
(2),
xix.)
the three-term equation
2 /(2 + a)/(2-a)/(6+c)/(6-c) = 0,
for general values of ,
6,
and
^,
is
expressible as a finite combination of elementary
functions, together with a Sigma-function (including a circular function or an algebraic
function as degenerate cases).
(Hermite, Fonctions elliptiqnes,
[Put
0=a = 6 = c=O,
that/ (2)
If
and then/(0) = 0;
put
= c,
i.
p. 187.)
and then /( 6)+/(6-(x) = 0,
so
an odd function.
is
is
[z)
the logarithmic derivate of
with respect to
6,
and then putting
Differentiate with respect to
6,
= c,
the result of differentiating the relation
/'(2),
is
and put
/(. + a)/(^-a) {/'
= 0;
then
(0)F _
"
{/(^)/(a)F
^
^
^-
If/' (0) "were zei'o, /^' {z) would be a constant and, by integration,/ (2) would be of the
form A exp (Bz+Cz^), and this is an odd function only in the trivial case when it is zero.
If /' (0)
^ 0,
and we write F' (s)= 4>
(z),
it is
found that the
coefficient of a* in the
expansion of
l2f{z+a)f{z-a)/{fiz)Y
is
of
6 {$ (z)}^-^"
(2).
function by $'
coefficient of a* in 12 {/()}^ {* (a) * (2)} is a linear function

a quadratic function of $(2); and when we multiply this
and integrate we find that
(j),
Hence
(2)
and the
4>" (2)
is
{<!>'
(2)}2
=4
{* (2)}3+ 12J {*
C are constants. If the

{z) = ^ {z + a) + A, where a is
where A, B,
20"6,

K and L are constants
since/ (2)
/(2)
12^$ (2)4-46',
constant, and on integration
f{z) = (r{z + a)
where
(2)j-'+
cubic on the right has no re2:)eated factors, then, by
exp{-Âz'-Kz-L),
is
an odd function a = K=0, and
= (r(2)exp{-iJ22-Z}.
If the cubic has a repeated factor, the Sigma-function
and
the sine of a multiple of
z,
replaced by a multiple of
2.]
if
the cubic
is
is
to be replaced
(cf.
20-222)
a perfect cube the Sigma-function
is
by
to be
CHAPTER XXI
THE THETA FUNCTIONS
211.
The definition of a
When
it
Tli eta-function.
desired to obtain definite
is
numerical results in problems
involving Elliptic functions, the calculations are most simply performed

with the aid of certain auxiliary functions known as Thetafunctions. These
functions are of considerable intrinsic interest, apart from their connexion
with Elliptic functions, and we shall now give an account of their funda-
mental properties.
The Theta-functions were first systematically studied by Jacobi*, who

and his analysis
obtained their properties by purely algebraical methods
in this chapter
contained
results
all
the
practically
that
complete
was so
;
(with the exception of the discussion of the problem of inversion in 21-7

In accordance with the general scheme
et seq.) are to be found in his works.
employ the methods of Jacobi, but the more

powerful methods based on the use of Cauchy's theorem. These methods
Avere first employed in the theory of Elliptic and allied functions by Liouville
in his lectures and have since been given in several treatises on Elliptic
functions, the earliest of these works being that by Briot and Bouquet.
of this book,
The
[Note.
we
first
function^
Partition
shall not
function of the Theta-function type to appear in Analysis was the
(l-.*'"^)"^
Euler,
of
Introductio
in
Anaiysin Infinitorum,
i.
(Lausanne, 1748), 304; by means of the results given in 5^ 21-3, it is easy to express
Theta-functions in terms of Partition functions. Euler also obtained properties of products
of the type
(iA-"),
(ia;2),
n=l
n=l
(i^-2-i).
n=\
associated series 2 ?n^"('*+^\ 2 m-"^""*^^^ and 2 m' had previously occurred in the
M=0
n=0
>i=0
posthumous work of Jakob Bei'nouUi, Ars Conjectandi (1713), p. 55.
The
*
I.
Fundamenta Nova Theoriae Fimctionum Ellipticarum (Konigsberg, 1829), and
Ges. Werke,
pp. 497-538.
t
The
Partition function
and
i.
Cauchy, Coviptes liendus,
x. (1840),
involving what are
known
as-sociated functions
(1811), pp. 7-l'2 [Werke,
reg. sci. Gottinnensis rec.
ii.
have been studied by Gauss, Comm. Soc.

and Werke, iii. pp. 433-480 and
pp. 16-21]
pp. 178-181. For a discussion of properties of various functions

as Basic ^lumbers (which are closely connected with Partition functions)
London Math. Soc. (1) xxviii. (1897),

and Watson, Camb. Fhil. Trails.
the tlieory of Basic numbers was given by
see Jackson, Froc. Roijal Soc. Lxxiv. (1905), pp. 64-72, Froc.
pp. 475-486 and
(2)
i.
(1904), pp. 63-88,
XXI. (1912), pp. 281-299.
ii.
(1904), pp. 192-220;
fundamental formula in
Heine, Kugelfunktionen (Berlin, 1878),
i.
p. 107.
2ri, 2ril]
THE THETA FUNCTIONS
463
La Theorie Analytique de la Chaleur (Paris,

265 of Freeman's translation (Cambridge, 1878).
Theta-functions also occur in Fourier's

1822),
cf. p.
The theory
by Jacobi
was developed from the theory of

Fundameata Nova Theoriae Functionum Ellipticarum
of Theta-functions
in his
elliptic
functions
(1829), reprinted
Werke, i. pp. 49-239; the notation there employed is explained in 21-62.

In his subsequent lectures, he introduced the functions discussed in this chapter an
account of these lectures (1838) is given by Borchardt in Jacobi's Ges. Werl-e, i. pp. 497-538.
in his Ges.
The most important

cf.
results contained in them seem to have been discovered in 1835,

Kronecker, Sitzungsherichte der ALad. zu Berlin (1891), pp. 653-659.]
Let T be a (constant) complex number whose imaginary part

and write q = e'^^'", so that q\<l.
Consider the function ^{z,
q),
is
positive
defined by the series

00
qua function of the variable

If
z.
be any positive constant, then, when \2\Â, we have

qII' Q-k2.niz
i
I
<
'*'"
I
(7
QinA
n being a positive integer.

00
Now
d'Alembert's ratio
which tends
to zero as
?i
^x
vS",
and
so
it is
= 00
The
series for
analytic functions, uniformly convergent (
values of
2'36) for the series
an integral function
^ {z,
334)
in
q)
( 5"3, 5*64).
70
5)
is
["'e^ji^ is
j
|2n+ig2^^
therefore a series of
any bounded domain of
It is evident that
^ (^,
{-y'q"^' COS 2nz,
464
The
function
^3(2',
^3
Next, %(z, q)
is

/
g)
(^,
is
^)
^4
^^
(
2 TT,
q)
7"- COS 2)12.
by the equation
^4(2^, q)
(-)"(/(" +
sin (2 n
*)'
l)z.
by the equation
q) is defined
'^.,{z,q}
= 00
2
71
(^.
defined in terms of
%{2,
Lastly, ^2
"^
and hence*
[CHAP. XXI
= %(z + l7r,q) =
^
q'<"
+ ^-^' cos{2n+ \)z.
H=0
Writing down the series at length, we have
^2
{z,
^3
(z,
bz ...,
2q'^''
=
2q* cos z + 2g' cos Sz +
cos 5^ +
q)
fy) = 1 + 2^' cos 2z + 2q* cos 4^ + 29^ cos 62 + ...
^4
(^>
^)
^1
(^)
q)
= ^q*
sin z - 2q* sin 3^
+ 25"*"" sin
~ 2g cos
It is obvious that ^j
(2,
22
is
q)
2q* cos 4^
cos Qz
S^*"
an odd function of 2 and that the other
Theta-functions are even functions of
z.
The notation which has now been introduced is a modified form of

Molk the only difference
that employed in the treatise of Tannery and
between it and Jacobi's notation is that ^4 (2, q) is written where Jacobi

would have written ^ (2, q). There are, unfortunately, several notations in
use a scheme, giving the connexions between them, will be found in 2r9.
;
For brevity, the parameter q
will usually
not be specified, so that ^1
(2), ...
be written for ^1 (2, q), .... When it is desired to exhibit the dependence
Also
of a Theta-function on the parameter r, it will be written ^ (2 t).
will
^2(0), ^3(0), '^4(0) will be replaced by ^2> ^3,
'^4
respectively;
and ^/
will
denote the result of making 2 equal to zero in the derivate of ^1(2).

Example
1.
Shew that
Example
2.
01)tain the results
3i{z, q)
-B.2{z + i^)
$2(2)=
MS3{z + Ut)=
^3(2)=
3i{z+U) =
where
*
exp
(2)
= - iJfSi
(2
q^)-h{2z,
q*).
=-iMB-i{z + \-n + UT)=-iMSi{z + \'!TT\
Bi{z)=
3i
= 9s{2z,
+ ^777") =
mi{z + ^_n + ^^nr)=

+ W+i7rr)=
im.2 {z+U+ i^rr) =
S^Z + hTr),
m,{z + UT),
^3 (2 + U),
.l/^,(3
M=q* e".
Throughout the chapter, the many-valued function q^
(Xttit).
is
to
be interpreted to
mean
THE THETA FUNCTIONS
21-12]
Example
periods
tt,
ttt
3.
Shew that the
465
multipliers of the Theta-functions associated with the
are given by the scheme
[cHAP. XXI
466
Since one zero of î
obviously z
{z) is
= 0,
it
follows that the zeros of
%(z), %,{z), ^4(^) are the points congruent respectively to
î(^),
Itt-I-Ittt, Ittt.
The reader
will
0,
^,
observe that these four points form the
corners of a parallelogram described counter-clockwise.
The relations between
21-2.
the squares
uf the Theta-functions.
the Thcsta-functions be regarded as functions of a

single variable z, this variable can be eliminated from the equations defining
any pair of Theta-functions, the result being a relation* between the functions
It
evident that,
is
if
which might be expected, on general grounds, to be non-algebraic; there

however, extremely simple relations connecting any three of the Theta-
are,
functions
now be
these relations will
obtained.
{z) is analytic for all

Each of the four functions î" {z), "^^ {z), ^3- {z),
q-"
with the periods
associated
e-*^^
factors
periodicity
1,
values of z and has
cell.
in
any
other
zeros)
and each has a double zero (and no
IT, TTT
;
From
these considerations
obvious that,
it is
and
if a, b, a'
are suitably
b'
chosen constants, each of the functions
a'% '(z)+b'X'(z)
aV(g)-h6V (^)
V(^)
is
'
V(^)
a doubly-periodic function (with periods
simple pole in each
cell.
By
and obviously we can adjust
tt,
ttt)
having at most only a

is merely a constant;
20-13, such a function
a, b,
a,
b'
so as to
make the
constants, in each
of the cases under consideration, equal to unity.
There
exist, therefore, relations of
^^ {z)
To determine
we have
'^3-
= a%' (z) + 6V (z),

a, b,
a,
= ((^4-,
b' ,
the form
%'
(z)
= a%' (z) +
give z the special values
^2"
= ^^4";
%- = a"^^,
^^
ttt
'^3-
b'X'
(z).
and
since
= 6"^4-.
Consequently, we have obtained the relations
X' {z) ^4' = ^4^ {z) X' - X' (z) %',

If
we
write z
%' (z)
+ -îr
V = %'
By means
(z)
for
z,
we get the
%' - ^.^ (z) ^3%
of these results
it is
%^ (z) %' = X' {z) %' - %' {z) ^,1

additional relations
V V = %' V - X'
(z)
(z)
possible to express
(z)
V.
any Theta-function in
terms of any other pair of Theta-functions.

*
The
analoo;ou.s relation for the fuuctions sinz
and cos
2 is, of course, (sin2)"''+(cos,:)2= 1.
21-2
21-22]
Writing
Corollary.
that
z=0
these formulae are not addition-theorems in the strict sense, as
functions of z and
of
two
results just obtained are particular cases of formulae containing
variables
Theta functions.
the
they do not express Theta-functions of 2
-{-y,
we have
in the last relation,
The addition-formulae for
21*21.
467
say
is to
The
THE THETA FUNCTIONS
and
y,
but
all
+y
algebraically in terms of Theta-
involve Theta-functions of z
as well as of
y.
To obtain one of these formulae, consider ^3 {z -h y) '^3 {z y) qua function

z.
The periodicity factors of this function associated with the periods tt
and
TTT are 1
and
(f^
e-2''2+!/)
= q-2 ^-Hz^
q-^ Q-iiKz-y)
But the function a^j- {z) + 6^1- {z) has the same periodicity factors, and
we can obviously choose the ratio a:b so that the doubly -periodic function
a%'{z) + }fh{Hz)
%{z + y)%{z-y)
^3 {z y) it then has,
has no poles at the zeros of

at most, a single simple
pole in any cell, namely the zero of ^3(2:4- y) in that cell, and consequently
and, as only the ratio a 6 is
( 20'13) it is a constant, i.e. independent of z
;
so far fixed,
We
we may choose a and
then have to determine a and
a%' (z) +
To do
aX' =
and
^Hy\
h^{'(^^7r
(z) =%(z-h y) X (z -
+ \'rTT^ =
and
%{^îr
= ^3- {y)l^.^,
is
unity.
from the identity in
put z in turn equal to
this,
so
We
b^-'
the constant
b so that
-îr
-^
ttt,
+ \7rr +
y).
and we get
y)%{^^'rr
The
+ y) % {z -
set of formulae, of
and 2 at the end of

21 '22.
= "î^ {y)l^3\
Jacobi's
y)
^/ =
V V
{y)
which this
is
{z)
+ î^ iy) X' {z).
typical, will
be found in examples
fundameMal formulae *.
as four independent variables, w, x, y,

iv',
x\ y\
z'
this chapter.
The addition-formulae just obtained are particular cases of a set of

by Jacobi, who obtained them by purely algebraical methods each
Let
+ \'rrT-y
have therefore obtained an addition-formula, namely

^3 (^
many
z,
be defined in terms of
2w'
2x'
2?/'
22'
lu,
identities first given
identity involves as
z.
x, y, z
by the
set of equations
= w-\-x-\-y-\-z^
= iv x+y + z,
= w+x y + z,
= tv-\-x-\-y z.
* Ges.
Werke,
i.
p. 505.
302
463
The
connexion between w,
re<\der will easily verify that the
[chap. XXI
x, y, z
and
iv\
x\ y\
z'
is
reciprocal one*.
For brevity t, write
[?] for
5, (w) 5^
(.r)
5^ {y) 3^
{z)
and
[r]'
for S,
(to')
$,
(x')
$,
(if)
S,
(z').
The effect of increasing by tt or nr

[4J qua functions of z.
is to transform the functions in the first row of the following table into those in the second
or third row respectively.
Consider
[3], [1]', [2]', [3]',
THE THETA FUNCTIONS
21-3]
Example 2. By writing
shew that
for y\ /),
+ ^n, x + \tv
tv
for
where [3344] means ^3 {w) S3 (x) S^
Example
3.
Shew that
Example
4.
Shew
x (and consequently
iv,
[3344] + [2211] = [4433]'
469
}/-\-\i^,
s'
+ ^tt
+ [1122j,
{y) ^4 (z), etc.
2[1234] = [3412]'+[2143]'-[1234]' + [4321]'.
21"3.
We
that
Jacobi's expressions for the Theta-functions as infinite products*.
shall
now
establish the result
M=l
(where
independent of
is
Let
fi2)=
and three similar formulae.
z),
(1
^^n-i
e--^'>)
{l-q->'-ê--'');
each of the two products converges absolutely and uniformly in any bounded
domain
of values of
z,
by
3'341, on account of the absolute convergence of
00
^2n-i.
and
so
hence
it is
The
is
(2:)
analytic throughout the finite part of the 2^-plane,
= g(2H+l)T^
g2iz
i.e.
where
zeros of/(2') are simple zeros at the points
where
2iz
(2?i
1) ttit
(w=..., -2,
f(z) and ^4
2ni7ri; so that
(z)
1,0, 1,2,
...)
have the same
consequently the quotient 'î{z)/f{z) has neither zeros nor poles in

the finite part of the plane.
zeros;
Now, obviously / (2^ +
tt)
=f{z)
GO
and
f{z+'77r)=
QO
IT
(1
5-+ie-''^)
IT (1
n=\
=1
52^-3 g-2fe)
=f{z){l-q-U-^^)!{\-qe^^^)
=
That
is
example
to
3).
-q-ê'-^f{z).
say f{z) arid î{z) have the same periodicity factors ( 2 I'll
Therefore î{z)/f(z) is a doubly-periodic function with no
zeros or poles,
and
so (| 20*12) it is a constant G, say; consequently
^4
(z)=G
{1-
2q"'"-'
cos 22
5^"--).
11=1
00
[It will
appear in
2142
that
G= U
(1
q-'').]
n= l
Write z +
^TT for
2^
in this result,
%,(z)=G
(1
and we get
25^'^-!
cos 2z
+ q'^'-').
n=l
* Cf.
Fundamenta Nova,
p. 145.
[CHAP. XXI
470
Also
^1 (z)
= - iq^ e" ^,
^ ttt)
00
=_
igi
e'z
3C
IJ
- 9=" e-'')
(1
n=l
=
and so
26^5* sin ^ 11 (1
w=1
^, (z)
= 2Gq^
TI (1
f"e-")
- q'''~"- e
-'''')
{I
^ê'-'O,
=1
sin ^ ft
- 25-" cos 2z + q')
n=\
+ l7r]
'^,{z)='^Jz
while
= 2Gq^ cosz U
{1
2q-''
cos 2z
+ ^").
n=l
Shew that*
Example.
(i-?2-i)i +16?-
bi = l
ISfoc
= n
(-00
18
cc
(i+?2)l
'n=l
18
(i+92-i)^
'
ln = l
(Jacobi.)
TZ/e differential equation satisfied hy the Theta-functions.
21"4.
We may
and t; and
number
regard ^s(z t) as a function of two independent variables z

is permissible to differentiate the series for ^3(2^|t) any
it
of times with regard to z or
convergence of the resulting series
'
ft
Consequently, the function
'^3
(2
1
The reader
OZ
will readily
r,
on account of the uniformity of
( 4*7 corollary)
n-
exip{n-7nT
in particular
+ 2mz)
= 00
t) satisfies the partial differential equation
.d^y
dy
prove that the other three Theta-functions also
satisfy this equation.
21*41.
relation between Theta-functions of zero argument.
The remarkable
result that
V(0) = ^2 (0)^3 (0)^4(0)

will
now be
established f.
It
is first
necessary to obtain some formulae for
differential coefficients of all the Theta-functions.

* Jacobi describes this result (Fund. Nova, p. 90) as 'aequatio identica satis abstrusa.'
t Several proofs of this important proposition have been given, but none are simple.
Jacobi's original proof (Gfis. Werke, i. pp. 515-517), though somewhat more difficult than the
proof given here,
is
well worth study.
THE THETA FUNCTIONS
21-4, 21-41]
471
Since the resulting series converge uniformly, except near the zeros of
we may
the respective Theta-functions,
differentiate the formulae for the
logarithms of Theta-functions, obtainable from
21"3,
many
as
times as
we
please.
Denoting differentiations with regard
to z
by primes, we thus get

11 a'iiZ
%'{z) = X{z)
a%^ (>2n
L=i (1
Making
2r
^ 0,
(0)
=1 (1
n=\
g-2i3
g^"-i e--^^)-
we get
^3" (0)
0,
=-
8^3 (0)
J^ (3^-^.^
In like manner.
V(0) = 8^4(0) 2
V(0)-0,
..=i(]
^2' (0)
^2" (0)
0,
-cf-^r
-1-8
^2 (0)
S
=i(l
and,
if
we
write ^1 {z)
</)'(0)
If,
however,
we
sin z
(z),
= 0,
<?^'?J
we get
^'"
= 8(/>(0) i
(0)
differentiate the equation ^1 (z)
sin z
cj)
(z)
three times,
we get
^/ (0) =
(0),
(/>
^/"
(0)
= S<p" (0) -
V"(0) = 24
2
Therefore
V(0)
=1 (1
</>
(0).
-1;
q'^^y
and
V(0)
v:iO)
"^^,(0) ^ Sf3(0)
V:(0)
^4(0)
8-2
=i (1
^2?l
00
^,2/11
q'^'f
q^'T-
_ V
.=1 (1
.=1 (1
+ 2
.=1 (1
^2W
CO
+ 2
q'^'-'r
- 2
q^^y
.=1 (1
n=i (1
?''"-'/
q^^y
on combining the first two series and writing the third as the difference of
two series. If we add corresponding terms of the first two series in the last
line,
we get
at once
^^V(0)^V10)_^Vi0)^2^^
^,(0)
%(0)
%iO)
n=i{l~q'''y
= 1 + V(0)
V(0)
[cHAP. XXI
472
may be
Utilising the differential equations of 21-4, this
d%' (0
written
t)
I
dr
V(0|t)
d%(0\T)
~^2(0|t)
dr
Integrating with regard to
V
where
(0, q)
= 1;
see that
which
d%(0\r)
dr
%(0\t)
dr
q)
(0, q)
(0, q),
To determine
q).
lim^3
qÔ
C,
make q-^0;
lim^4
l,
q^-O
since
l,
(/^-O
and so
the result stated.
is
The value of
21-42.
From
the constant G.
the result just obtained,
constant
^/ = 0(0)=
we can
at once deduce the value of the
in 21-3.
which was introduced
by the formulae
For,
= C% (0,
\imq--^% =
qÔ
d%{0\r)
we get
r,
a constant (independent of
is
\imq-^X = %
we
"^^3(0|t)
of that section,
2q^
GU{1- q'^% % = 2qiGU{l + f)^

n=l
M=l
^3
and
so,
by
G^
(1
Now
all
- (f^y =
G^
(1
q;"')'
(i
+ q^^'-^y n
(i
(1
- g^"-') n
- ?^'ol
(1
so
the
I fi (1
(i
(t-')
(1
+ ^^'ol
g*^)
W=l
M=l
q^^^-y.
= n (i-9'O n
=
\q\<l, and
the products converge absolutely, since
following rearrangements are permissible
CO
OO
O)
(1
q^'^-'f,
we have
| 21-41,
00
x = GU{i-
r''-')\
(1
g^*^),
M=l
the
first
step following from the consideration that
comprised under the forms 2n
and
Hence the equation determining
n
n=\
andso
G=+ n
(1
- r/^
(1
2n.
is
- (f'J = G\
all
positive integers are
21*42
21-5]
THE THETA FUNCTIONS
To determine the ambiguity
in sign,
473
we observe that G
is
an analytic
function of q (and consequently one-valued) throughout the domain \q\<\;

and from the product for ^3(2^), we see that G-^1 as q>0. Hence the
plus sign
must always be taken
and so we have established the result
G=Yl
Example
1.
Shew
that
Example
2.
Shew
that
Example
3.
Shew
that
^1=22*0^.
+2
)!
21 ^S.
Connexion of
has been seen
It
the periods
2a)i, 2co2, is
where (/=exp
If
{^
(l-^^'O-
the
i y-= n {(i-(72)(i+g2'i-i)2).
=1
n =l
Sigma-function with
the Theta-functions.
20-421 example 3) that the function
a-
[z
wx, M2),
formed with
expressible in the form
(ttiojo/wi).
we compare
this result with the product of 21
.(.) =
jî
in
t),
we
see at once that
êxp(^Y^,-in(l-./)-3i(^|-^).
^
2
COj/
\2coi/
TT
To express
for ^1 [z
'4
\2coi
=i
terms of Theta-functious, take logarithms and differentiate twice,
so that
^<^)=:i-(0=(e,)-(^.T
where
= ^7rzja)^
Expanding
we get
and the function
in ascending
cf)
is
powers of
.4>'
that defined in 21"41.

z
and equating the terms independent of
z in this
result,
a>i
3 \2(oiJ
\2(Ui/
(f)
(0)
,;=_ -
and SO
Lzooi
Consequently
a-
{z
wi,
wo)
î
can be expressed in terms of Theta-functions by the
formula
'031
I'-.^l
o
a(^lo,co,)=-^,exp(^--g^J5,(^v
"2^
I
mi
where v^hivzlwi.
Example.
Prove that
/n'-a-iBi"
21"5.
The expression of elliptic functions hy means of Theta-functions.
been seen that Theta-functions are substantially equivalent

and so, corresponding to the formulae of 20'5-20"58,
exist expressions for elliptic functions in terms of Theta-functions.
It has just
to Sigma-functions,
there will
7n'\
474
From
the theoretical point of view, the formulae of
more important on account
20"5-20"53 are the
of their S3'mmetry in the periods,
the Theta-function formulae have two advantages,

are
[CHAP. XXI
(i)
more readily computed than Sigma-functions,
but in practice
that Theta-functions
that
(ii)
the Theta-
functions have a specially simple behaviour with respect to the real period,
which
generally the significant period in applications of elliptic functions
is
in Applied Mathematics.
Let f{z) be an
set of zeros (aj,
elliptic function
Oa, ...
an)
and poles
with periods 2wi,

(/3j,
/3...,
...
/3)
2&J2; let
a fundamental
be chosen, so that
as in 20o3.
Then, by the methods of
20"53,
the reader will at once verify that
TTZ TTCUr W.jX

TTZ TTySr
= A,\\\X[^-^-'~'^\^%^
1
f{z)
r=\
where A^
a constant
is
Iw,
V
^
'
(.
and
(jdJ
.^
2&),
^Wg
CO
if
1)1).
m=l
be the principal part oi f{z) at
r=i
where
pole ^^, then,
(m-1)!
by the methods of 20-')2,
"
dz""
2&)i
This formula
is
important in connexion with the integration of
An
functions.
will
Example.
&)i/J
a constant.
J. 2 is
problem
(,=.!
its
example of an application of the formula

be found in 22741.
Shew
elliptic
dynamical
to a
that
^3^
(2)_ _
5i2
(2)
\^
'^{^
dz
-^l'
(^)
-^3 -^3"
,
.9i (2)
î'3
'
and deduce that
21'51.
Jacohis imaginary transformation.
If an elliptic function be constructed with periods
((2/a)i)
>
2&)i,
2w2, such that
0,
might be convenient to regard the periods as being 'Iw^, 2&)i for these
numbers are periods and, if I (co-i/coi) >0, then also /( oji/wo)> 0. In the
case of the elliptic functions which have been considered up to this point,
the periods have appeared in a symmetrical manner and nothing is gained
by this point of view. But in the case of the Theta-functions, which are
it
only quasi-periodic, the behaviour of the function with respect to the real
period
tt
period
ttt.
is
quite different from
its
behaviour with respect to the complex
Consequently, in view of the result of 21"43, we
may
expect to
475
THE THETA FUNCTIONS
21-51]
obtain transformations of Theta-functions in which the period-ratios of the
l/r.
two Theta-functions involved are respectively t and
The transformations of the four Theta-functions were first obtained by

who obtained them from the theory of elliptic functions but Poissonf
Jacobi*,
had previously obtained a formida identical with one of the transformations

and the other three transformations can be obtained from this one by elementary algebra. A direct proof of the transformations is due to Landsberg,
who used the methods of contour integration

formulae, which we shall now give, is based on
formula which we shall establish is
where (-
ir)' ^
For brevity, we
The only
The
shall write
zeros of
'^3
{z
/t
and
t)
respectively,
mir
-\-
UTTT
{t z
'^3
+2
where m,
m, n
n,
= m, we
is
an integral function with no

1
^{z +
while
yjr
(z
Consequently
<
^'^^
at the points
'""'''
TZ = mir +
take
all
,,,1,1/
tiTTT 4- ^ vr 4-
.3
ttt
integer values; taking
m =- \,
)i
-v/r
Thus
and writing z
zeros.
= exp
ttt) ^ylf{z)
'tt)
-^ yjr
= exp
z)
(z
(z
-.
constant,
t)
-t-
- tt
(z\t)=
-i-
^q
x
_ -'^
.
ê
q~ê~-'^''
(independent of
1,
= 1.
ttt in
2).
r')
turn for
% (zr
exp (itZ^/tt) % (ZT
exp {Wz-j-n) % (zt
exp
A% (z\t)=
J.'î
+ 7r-T-\
t) = exp (tW/7r) ^3 (^r
+ ^ ttt,
^^,
(IZTTT
doubly-periodic function with no zeros or poles
must be a
^^3
+ ^-tt,
(z) is a
so ( 20'12) i/^(^)
the precise
see that the quotient
A
Also
iV)
t) are simple zeros
+o
'"'
and
theorem
exp {ttW).
r',
at which
z
investigation of Jacobi's
Liouville's
be interpreted by the convention arg(-
to
is
ij:.
{irz-'/Tr)
z,
we
easily get
r),
t),
I
t').
Journal fur Math. in. (1828), pp. 403-404 [Ges. Werke, i. (1881), pp. 264-265].
Mem. dc VAcad. des Sci. vi. (1827), p. 592; the special case of the formula in which
had been given
earlier
by Poisson, Journal de VEcole polyteclmique,
xii.
z-0
(cahier xix), (1823),
p. 420.
+ This method
Math.
is
CXI. (1893), pp.
indicated in example 17 of Chapter
234-253.
vi, p. 124.
See Landsberg, Journal
fiir
We
[CHAP. XXI
47 G
still
have to prove that
equation and then put 2
A ={ it)-
do
to
^^/(OJT) = -iVV(0
%'
But
^/
and
(0
(0
differentiate
so,
the last
0; we get
= ^, (0
t') = ^, (0
t)
t) ^3 (0
r')
t').
t) ^4 (0
t)
^3 (0 r) ^, (0
t')
on dividing these results and substituting, we at once get A~'-
A=
To determine the ambiguity
= W, and
so
{-iT)^-.
in sign,
we observe
that
^^3(0 t)=^3(0|t'),
both the Theta-functions being analytic functions of t when 7 (t) >
thus A is analytic and one-valued in the upper half r-plane. Since the
Theta-functions are both positive when t is a pure imaginary, the plus sign
;
Hence, by the theory of analytic continuation, we
must then be taken.

always have
A = + {- ir)^
this gives the transformation stated.
been shewn that
It has thus
X
5"
on-Trir+tniz
As -mr]-
'V
{1:17)
7j= -oe
Example
when
TT
Shew
1.
that
B,{Q\t)
_ %AO\t')
53(0|r)
53(0|r')
-\.
Example
Shew
2.
that
MOPr + 1)^
53(0 r + 1)
i^2iOJjr)
54(0|r)'
Example
Shew
3.
that
and shew that the plus sign should be taken.
Landens type of transformation.
21"52.
transformation
interest, is
due
integrals
of elliptic
Landen
to
( 22-42);
this
227), which
is
of historical
transformation follows at once
from a transformation connecting Theta-functions with parameters t and 2t,
namely
X{ z\t)X {z
t)
I
It) ^4(0
^4(0|2t)
^ ^3(0
^4(22|2t)
which we
The
z
shall
IT
prove.
of ^^{z
zeros
= [m + -\
now
-\-
\n
JT)
+ -A
r)î{z\r) are simple zeros at the points where

TTT
and where
imr
+in +
-Airr,
where
and n
THE THETA FUNCTIONS
21-52, 21-6]
take
all
integral values
477
these are the points where 1z
which are the zeros of ^4 (2^
mir
Ui
Ar
-îr
.'Ir,
Hence the quotient
2t).
^4 (2^ 2t)
I
Moreover, associated with the periods
has no zeros or poles.

has multipliers
and
{cf-^
e"-''^)
{^-
a doubly-periodic function, and
value of this constant
may be
is
q-^ e''^^^)
-^
{-
q~~"'
consequen-tly
e-^^"^)
obtained by putting
it
it
and
is
ttt, it
therefore
The
we then have
20"12) a constant.
z
^ and
the result stated.

If
we
write \^ +;- TTT for
we get a corresponding
z^
result for the other
Theta-functions, namely
The
21-6.
From
^,(^|t)î(^|t)
^ ^3 (0 1^)>4 (OK)
î(2^|2r)
'""^4(0i2T)
of Theta-functions.
differential equations satisfied hy quotients
21-11 example
has periodicity factors
and consequently
its
3, it is
1,
obvious that the function
associated with the periods
ttt respectively;
tt,
derivative
[X {z) ^4 {z) - %' (^) ^1 (^)l -
(^)
has the same periodicity factors.
But
+1
it is
easy to verify that ^.(z)^,,{z)/^J^ (z) has periodicity
and consequently,
if <^ (z)
{X (z) % (z) then
(f)
poles of
(f)
Now
we
tt
and
ttt
simple poles at points congruent to
(z) are
consider
cf)
iz
+ I ttt]
1,
- [% (z) % (z)},
(^) ^1 i^)]
doubly-periodic with periods
(z) is
f;xctors
be defined as the quotient
from the relations of
and the only possible
^ tt
and
^ vr -h
ttt.
.^
namely
21-11,
= iq-ie-''%(z),
%^Z + l7rT^=iq-h-''X{z\
'^,(^z
%\z+l7rT^=q-ie'''X(z),
x{z-^l'rrT^=q~ê-''%{z),
l7rT'j
easily see that
(^ (^^
-i-
Hence ^ (z)
these periods,
is
the
congruent
^ ttt)
to ^ vr
2
{-
(z) ^1 (z)
(^)
^4
doubly-periodic with periods

only possible
(^)l
tt
- [% (z) % (z)}.
and
-^
ttt
and, relative
poles of ^{z) are simple poles
to
at points
[CHAP. XXI
478
Therefore (^20-12),
</)(^)
the value of this constant
We
a constant; and making z-^0, we see that
is
is [î' '^4} -^ 1^2'â}
= ^Z-
have therefore established the important result that
writing ^
= *î
(^)/^4 {2)
and making use of the results of
()' = ^^-^ ~
^'^^''^ ^^^'
"
21
2,
we
see that
^'^''^'
It
This differential equation possesses the solution î(^)/^4(2).

difficult
is
see that the general solution
to
the constant of integration
increased by
the negative sign
tt,
%{z + a)/%(z + a)
is
since this quotient changes sign
may
is
not
where a
when
is
be suppressed without affecting the
generality of the solution.
Example
Shew that
1.
dz [Si
Example
Shew
2.
thcat
{z;j
S4
(2)
^4
(z)
lPiii)l=_q2îi?) Mi)
The genesis of the Jacohian
21*61.
The
which was obtained in

change of variable.
21-6,
if A:-
^^''
"
^'^^'^ ^^''
may be brought
^'^''^'
"
to a canonical form
by a
slight
^%/% = y, ^V =
Writet
of M
function* snu.
(S)' ^
then,
Ellip.tic
be written in place of ^2/^3, the equation determining y in terms
is
(|y = a-/)(i-A.y).
This differential equation has the particular solution
The function
the periods
7r%",
periods 27r^3^
of u on the right has multipliers

irT^-,;-;
ttt^^-.
congruent to hirrâ-
it
is
In any
and
tt^-J^
therefore
cell,
-1,4-1
associated with
doubly-periodic function with
has two simple poles at the points
it
iTTT^j-
and, on account of the nature of the
quasi-periodicity of y, the residues at these points are equal and opj3osite in

and tt^sI
sign the zeros of the function are the points congruent to
;
* Jacobi
and other early writers used the notation
sin
am
in phice of sn.
when \q\<l, except when q

t Notice, from the formulae of 21-3, that ^2 + 0, ^3 +
which case the Theta-functions degenerate; the substitutions are therefore legitimate.
= 0,
in
THE THETA FUNCTIONS
21-61, 21*62]
It
is
customary
to regard
y as depending on k rather than on q

k,
or simply
y = sn
2/
is
now
evident that sn
of the types described in
that sn(w, A")>sin
to
u.
an
is
elliptic
when g>0
and
function of the second
(so that ^'>0), it is easy to see
v.
The constant k
k' is called
k)
(u,
20'13
we write
= sn {u, k),
exhibit y as a function of u and
It
479
is
modulus;
called the
if ^''^
The
the complementary modulus.
= ^4/^3,
so that k^
+ k''^=l,
quasi-periods ir^^, ttt^.^ are
usually written 2K, 2iK', so that sn (u, k) has periods 4iK, 2{K'.
From
we
21-51,
function of t as
Example
is
Shew
1.
see that
of
t,
when
2K' =
tt'
that, if
?y
= -^
^J
Example
"
~.
-"^
is
the same
and u = z$^-, theu
Shew that
2.
that, if
j/
3^'
3.4(2)
-r^ 6"7 \
^3 ^4 [z)
>
^^'^ u^zS^^, then
Obtain the following results
3.
S,{z)3,{z)'
(z)
[These results are convenient for calculating
21'62.
K'
S,(z) S,{z)'
'''
Example
that
-^4 (^)
rf2
and deduce
so
1.
that
dzS,{z)
and deduce
7r'^.^^(0 \r'),
Jacohi's
Eta-fanction
earlier
notation*.
/,
k\
A", A''
when
is
given.]
The Theta-function (?<) and
the
H ().
The presence
of the factors ^3"- in the expression for sn {u, k) renders
it
sometimes desirable to use the notation which Jacobi employed in the

Fvndamenta Nova, and subsequently discarded. The function which is of
primary importance with this notation is (u), defined by the equation
Ch)
(u)
= ^4 {u%-'
so that the periods associated with

* This
is
(u)
are
t),
i
2K
and 2iK'.
the notation employed throughout the
Fundamenta Nova.
[CHAP. XXI
480
The
function
&{u + K)
have the function
then replaces ^3 ('^)

(u) defined by the equation
e (u +
H () = - iq - ie'""^"
replaced by H (?/ + A").
<-'^'^
and ^. {z)
is
i7v")
and
in place of î(-?)
= ^^ {u%-'-
we
t),
The reader will have no difficulty in translating the analysis of this

chapter into Jacobi's earlier notation.
Example
{u)=-
If e'
1.
at the points congruent to
Example
Shew
2.
(mod
I'K'
(0)
W A'-i H
and the residue at each singularity
A')
(0)
is 1.
e (A').
The problem of Inversion.
21"7.
Up
2A', -liK')
that
H'
singularities of -t-t-t are simple poles
shew that the
the Jacobian elliptic function sn (i/, k) has been

depending on the parameter q rather than on the
has been shewn that it satisfies the differential equation
to the present,
implicitly regarded as
modulus k
and
it
-^
where
= (1 -
k- sn- u),
sn- u) (1
A:^
(0, ?)/
(0, 5).
But, in those problems of Applied Mathematics in which elliptic functions

we have to deal with the solution of the differential equation
occur,
{I)-('-^'>^'-'y->
in
which the modulus k
is
given, and
we have no a
priori knowledge of the
value of q\ and, to prove the existence of an analytic function sn(M, k)

satisfies this equation,
which
we have
to
shew that a number t
exists* such
that
When
this
number
t has
been shewn
to exist, the function sn(w, k) can
be constructed as a quotient of Theta-functions, satisfying the differential

equation and possessing the properties of being doubly-periodic and analytic
except at simple poles
and
also
lim sn{u,
That
is
to say,
we can
k)/i(
1.
invert the integral
[y
dt
^^~Jo {l-(')^{l-k'i^)^'
so as to obtain the equation
The
that
I
after the
<
existence of a
1.
An
manner
number
r,
y= sn
for
(u, k).
which /
(t)
>
0,
involves
alternative procedure would be to discuss

of Chapter x.
t)ie
tlie
existence of a
number
q such
differential equation directly,
THE THETA FUNCTIONS
217, 2r7l]
The
481
shewing that the equation
difficulty, of course, arises in
c=V(0|tW(0|t),
(where
has been written for
When*
<
<
1,
it
is
has a solution.
L-).
From
easy to shew that a solution exists.
identity given in 21*2 corollary,
evident that
it is
prove
sufficient to
it is
the
the existence of a solution of the equation
1-c = V(0|t)/V(0!t),
/^ _ Q2n-1\
1 - c = 11
^CO
which may be written
Now,
as q increases from
to
the product on the right
1,
is
continuous
passes through the
and steadily decreases from 1 to

and
value 1 c once and only once.
Consequently a solution of the equation
in T exists and the problem of inversion may be regarded as solved.
so ( 3'63)
The problem of inversion
21 "71.
complex values of
for
it
c.
The modular functions
f{r),g{T),h{T).
The
may
i^roblem of inversion
may
be regarded as a problem of Integral Calcuhis, and
it
be j^roved, by somewhat lengthy algebraical investigations involving a discussion of
the behaviour of
(I
;!-)
~2
(1
k^fi) ~
when y
2 dt,
on a 'Eiemann
lies
surface,' that the
problem of inversion possesses a solution. For an exhaustive discussion of this aspect of

the problem, the reader is referred to Hancock, Elliptic Functions, i. (New York, 1910).
It
however, more in accordance with the
is,
method
domain
( 6-.31)
of the T-j)lane
function of
c,
= ^2* (^
that the equation
when
c is
and that (subject
work
of this
sjjirit
'^V-^s* (^
'')
to prove
to certain limitations) this root is
regarded as variable.
It
by Cauchy's
^^^ one root lying in a certain
an analytic
has been seen that the existence of this
root yields the solution of the inversion problem, so that the existence of the Jacobian
elliptic function
with given modulus k will have been demonstrated.
The method just indicated has the advantage of exhibiting the potentialities of what
The general theory of these functions (which are of
are known as modular /mictions.
great importance in connexion with the Theories of Transformation of Elliptic Functions)
has been considered in a treatise by Klein and Fricket.
r^^
:Mnir
/ (r) = Weîr n
Let
..
^,
IgV""^^^
_g(2n-l)T^8
"
,s
S./{0\t)
^3 (0
'"'
54*(0ir)
=53-'(0
h{r)=-f{r)lg{r).
Then,
by
if
tt
= 1,
This
is
+ 2)=5r(r),
/(r + 2)=/(r),
5r(r
f{r + \)^h
/ (r') =g (r),
21 '2 corollary,
*
the functions just introduced possess the following properties
(r),
2r51 example
the case which
is
f{r)+g{r) =
g
(r')
\,
=/(r),
1.
of practical importance.
t F. Klein, Vorlesungen uber die Theorie der clliptischen Modnlfunktionen (ausgearbeitet
vervoUstandigt von E. Fricke).
W. M. A.
(Leipzig, 1890.)
31
und
482
+ QO the functions iV<*~""/('") = /i W and g (t) tend to

R (r), when - 1 ^ ^ (r) ^ 1 and the derivates of these two
It is easy|_to see that as /(r)--
unity, uniformly with resi)ect to
functions (with regard to
tend uniformly to zero* in the same circumstances.
The principal solution off (t) r = 0.
21 'Til.
It
r)
[CHAP. XXI
has been seen in 6*31 that, if /(t) is analytic inside and on any contour,
of roots of the equation /(t) c = inside the contour is equal to
iiri
times
number
the
df{r)
Ifir^cîr^^'
taken ]round the contour in question.
Take the contour

temporarily! that /(t)
ABCDEFE' D'C B' A

c
shewn
the figure,
in
E'
FE is drawn
is
By
elementary geometry,
it
D\
circle is the reflexion of the arc
The
parallel to the real axis, at a large distance
inverse, the circle on
supposed
constructed in the following manner
AB la the inverse of FE with respect to

BC is the inverse of ED with respect to
own
being
-1
The contour
it
has no zero actually on the contour.
the circle
[
= 1,
follows that, since
Z)
in the line
left-hand half of the figure
is
it.
= 1.
being chosen so that
C and
D are
as diameter passes through
AB
from
inverse points and
C and
;
D\=AO.
so the arc
CD
is its
of this
R (r) = ^.
the reflexion of the right-hand half in the line
R{t) = 0.
*
This follows from the expressions for the Tlieta-functions as power series in
as I (t) ^- -|- oo
9 -
observed that
The
values of/{T) at points on the contour are discussed in 21'712.
q, it
being
THE THETA FUNCTIONS
2r71l]
now be shewn
It will
c^
that, unless*
483
or c^Ô, the equation /(r)
c=0 has
one,
and
FE
only one, root inside the contour, provided that

is sufficiently distant from the real
axis.
This root will be called the principal root of the equation.
To
establish the existence of this root, consider
-rr\
--y^
dr taken along; the
various portions of the contour.
Since/(r + 2)=/(r), we have
BC
Also, as T describes
BE
ED-
and
r'(=
B'C",
-C
(t)
dr
l/r) describes
E'D' and
ED
respectively;
and so
{}
BC
dr
f{T)-C
j C-B']
dr
Jc-B']g{T)-C
\j
BC
{]
ED'
DE) g{r)-C
dr
= 0,
because g
(r'
+ 2)=^(r'), and consequently

1 = A (r), we have
Since / (r
corresponding elements of the integrals cancel.

^
[j
jCD]f{r)-C
D'C
but, as T describes
B'AB,
jB:ABh{T)-c
dr
EE\
r describes
and
so the integral
dr
round the complete contoui*
reduces to
f_]
/"
df{r)
dh{r')
dr
h(T')
{^
î
Now
as
EE' moves
- lim
f
J EE'
f{'r')
dr
dfjr)
dh{r)
dr
h{r)\\-c.h{r)]
dr
]EE'\f{T)-C
limit of the integral
dr
dfiyiXdr
jEE'\f{-r)-c
/(t) c-*- -c=t=0,
off to infinityt,
'^
{r)
-C
dlMidr
dr
'
^(t)-c--1 c4=0, and so the
is
^^
{log
-^
l-C.A(r) dr^
(r)}
'^
dr
^X^m
.'
EE \-C.h{r)\
But 1- c.A(t)^1, fi(r)~^\,g,{r)-^l, ^^^^-^0,

dr
.^
^logACr) _ rflog^(r)
^^
dr
^^-^0,
clr
dr
'
and so the limit of the
integral is
nidr = 2-i
Now,
g
{r)
of /(r)
from the real axis that / (r) c, \c.h (r),

above EE', then the contour will pass over no zeros
to infinity and the radii of the arcs CD, D'C, B'AB diminish
if we choose EE' to be
have no zeros when r
to zero
E'E
as EE' moves off

and then the integral
initially so far
is
will not
original contour integral will be Sttz,
change as the contour
and the number
is
of zeros oif{r)
modified,
and so the
inside the original
contour will be precisely one.

* It is
shewn in
21'71'2 that, if
c^l
or
cÔ, then/(T) -c has a

and 4=1.
zero on the contour.
t It has been supposed temporarily that c=|=0
312
484
21 '712.
The values of
modular function f {t) on
the
[CHAP. XXI
the contoitr considered.
We now
have to discuss the point mentioned at the beginning of

21-711, concerning
c on the lines* joining +1 to l + x t and on the semicircles of
05C1, {-\)C'B'0.
^5
the zeros of f{r)
As
T goes from
to 1
real negative values.
+x
or from
E'
sponding indentation in D'
1+x
to
i,
/(t) goes from -
we make an indentation
So, if c is negative,
in
DE
through
to
and a
and the integrals along the indentations cancel
corre-
in virtue of
the relation /(t + 2) +/(r).
As r describes the semicircle 0^C1,t' goes from - 1 + x I'to 1, and/(r) = 5r(T') = l /(r),
and goes from 1 to +x through real values it would be possible to make indentations in
BC and B'C to avoid this difficulty, but we do not do so for the following reason the
ettect of changing the sign of the imaginary part of the number is to change the sign of the
real part of r.
Now, if < 7^ (c) < 1 and I (c) be small, this merely make-s t cross OF by a
short i>ath if R (c) < 0, t goes from DE to D' E' (or vice versa) and the value of q alters
only slightly but if R{c) > 1, r goes from BC to B'C, and so q is not a one-valued function
of c so far as circuits round c = +1 are concerned to make q a one-valued function of c,
we cut the c-plane from -l-l to -fx
and then for values of c in the cut plane, q is
;
'
determined as a one-valued analytic function of
c,
say q
(c),
by the formula q
(c)
e'^'"'^^'''
where
.
2771
as
may be
seen from
j/(r)-C
by using the method of
6'3,
df{r),
dr
5-22.
If c describes a circuit not siu-rounding the point
one-valued only
21 '72.
if,
in addition, the circuit does not
c=l, q{c)
is
one-valued, but t{c)
surround the point
is
= 0.
The periods, regarded as functions of the modidus.
21-712 that K is a one-valued analytic function of

Since K=\7rB:^ (0, q) we see from
c{=k^) when a cut from 1 to -l-x is made in the c-plane; but since K'= -irK, we see
that K' is not a one-valued function of c unless an additional cut is made from
to x
it will appear later ( 22-32) that the cut fi'om 1 to -fx which was necessary so far as
5^
K is
concerned
not necessary
c\s
regards K'.
The inversion-problem associated
2173.
It will
is
is
now
when
be shewn that,
iv it h
invariants
Weierstrassiari elliptic functions.
g.^
and g^ are
given, such that
g^^%lg^,
possible to construct the Weierstrassian elliptic function with these invariants
we
to say,
fp {z\(oi,
shall
shew that
it is
do) has invariants g^
The problem
is
solved
if
possible to construct \ periods
of the
we can obtain a
Let
*
We
= Az,
effect
such that the function
solution of the differential equation
if -9-^^-93
^ = ^ (2
form
it
is
and g^.
m
"We proceed to
2(i)i, 2u>.2
that
wi,
<t>2)-
the solution of the equation with the aid of Theta-functions.
where
.4 is
have seen that
a constant to be determined presently.
EE
'
can be so chosen that
f (t) -c
lias
uo zeros either on
EE
'
or
on
the small circular arcs.

t
On
the actual calculation of the periods, see E. T. A. Innes, Proc. Edinburgh Royal Sue.
XXVII. (1907), pp. 357-368.
THE THETA FUNCTIONS
21*7 12-21-8]
By
the methods of 21 -e,
and hence, using the
Now
that
let ei,
(^1
e.^,
485
seeu that
it is easily
we have
results of 21 "2,
be the roots of the equation '^y'^g-iygz = ^j chosen in such an order

is not* a real number greater than unity or negative.
e^
e'2)l{ei 63)
In these circumstances the equation
^3'(U|r)
61-63
possesses a solution ( 21 "712) such that
which has, up
T of the Theta-functions,
Choosing t in this manner,
let
/(r)>0;
this eqvxation determines the
parameter
now, been at our disposal.
till
be next chosen so thatt
Then the function

.V
= ^' |!^j 53MO

(
The
r)
I
+ ei
r)
(0
I
the equation
satisfies
= 4 (y - ei) {y - 62) {y - 63).
periods of y, qua function of
are ttJ, tttJA
z,
calling these 2(bi, 2W2
we have
/(co2/<ai)>0.
The function ^{z

seen that
^(3)-^%\f-f-J
the origin |
If 6*2
6*3
it is
may
oa.^
wi,
^,2(0
be constructed with these periods, and
t)
V(0
r)-ei
is
an
elliptic
it
is
easily
function with no pole at
therefore a constant, C, say.
be the invariants of
^ (2
wi
a>.^,
we have
Aip^(^z)-G.^{z)-G, = f^{z) = ^{ip{z)-C-e,]{^{z)-C-e.^l{iJ{z)-C-e,},
and
so,
comparing
coefiicients of
= 12C,
Hence
and
^{z
co^,
<^ (z),
we have
G3=g3-g,C+iC^
G.2=g2,
(7=0,
so the function
Gz^g^;
wo) with the required invariants
proceeding in ascending powers of q are convenient for
series
calculating Theta-functions generally, even
when
|
it
usually happens in practice that the modulus k

* If
^_iZ:^>i thenO<^^^^^<:l; andif ^^^^<0, then

e^-e^
ei-i'j
ei-e^
^j
The
t
values
The
has been constructed.
The numerical Computation of elliptic functions.
21'8.
The
powers of
G2=g-2-l2C%
0, 1, qo
of
(ej
<'2)/(î
sign attached to
functions of
and
t The terms in
is
^3)
-ej
is
is
|
as large as 0"9.
given and the calculation
1-^^^^>1,
and
Ci-e^
^ ri_^lZ^1"^<l.
a-i'e
excluded since
(12^
4= 27(/3^.
a matter of indifference, since
we
deal exclusively with even
-.
z'"^
cancel,
and there
But
is
no term
in z~i because the function
is
even.
'
[CHAP. XXI
486
be seen later
22'801, 22'32) that
of K,
K' and q
K, K'
are expressible in terms of hypergeometric functions, by the equations
is
It
necessary.
^v^ll
when k and
series converge slowly except
but these
quite small
k'
respectively are
so that the series are never simultaneously suitable for numerical
calculations.
To obtain more convenient

as a root of the equation k
formula
K=
rîr'î (0,
numerical work, we
series for
= ô"
first
and then obtain
(0, q)l^'i' (0, q),
calculate q
K from the
q) and K' from the formula
K'=7r-^K\og,{l!q).
k=
The equation
is
equivalent to*
(0, ?)/
= ^4 (0,
^Jk'
(0, q)
5)/^3 (0,
q).
111
-i
Writing 2e
j,
have seen
solution which
is
<
(so that
<
;j
when
we get
q)^ %(0,q')
%{0,q) + %{0,q) %{0,q^)-
21*71-21-712) that this equation in
e*
expansible in a Maclaurin series in powers of

It
0<k< 1),
_%i O,q)-'ÂO,
^'
We
when
e
<
g^
possesses a
and so q
will
be
in this domainf.
remains to determine the coefficients in this expansion from the
equation
^
+ r^+ ...
+ 2^* + 25" +
g
1
(/
.
.
which may be written
q=6 +
2qU-q''+2q'^-q-'+
...;
the reader will easily verify by continually substituting
on the right that the
for q w^herever q occurs
q
It has just
[Note.
The
= e + 2e' +
been seen that

first
Example.
= |,
loOe'^
= 0-0000609,
Given k = k' = \IJ2, calculate
q,
2q*e
q^
...
U'').
when
this expansion usually suffice
2-^
two termsj are given by
this series converges
two terms of
large as ^(0-8704) = 0-933...,
15e
first
<
thus, even if k be as
15f = 0-0000002.]
K, A" by means of the expansion just
obtained, and also by observing that t=i, so that q = e~^

[y
^= A" = 1-854075.]
numerical work
< < 1, and so q is positive and < ^k' < 1.
The Theta-functions do not vanish when |5|<1 except at = 0,
* In
= 0-04.32 139,
A;
'^
possible branch point.
X This expansion was given by Weierstrass, Werke,
ii.
(1895), p. '276.
so this gives the only
THE THETA FUNCTIONS
21-9]
21 'Q.
The
487
The notations employed for the Theta-functions.

following scheme indicates the principal systems of notation which have been
employed by various writers

function.
î(tj)
the symbols in any one column
all
denote the same
[chap. XXI
488
Obtain the addition-formulae
2.
94(y+2)54Cy-^)V=V(y)^2M2)+V(y)V(2)=V(3/)V(2)+V(y)^3'(^).
and,
by increasing
}/
by half
periods, obtain the corresponding formulae for
5r (i/ + z) Sr Ly -
where r=l,
3.
--)
^2^
and Br{y-{-z)Br{2f-z)
V.
(Jacobi.)
2, 3.
Obtain the formulae

5l 0/
--)
^2 (y
+ ^) h -94 = -9, Q/) $2 (i/) ^3 (2) ^4 (2) ^3 (y) ^4 C^) ^1 C^) ^2 {z),
Sl(]/z)Ss(l/ + z) ^2^4 = ! Cy) ^3 (y) ^2
(2)
^4
C^)
^2 i^) ^4 (y) ^1 (z) ^3 i^l
= 5, Cy) ^4 (y) ^2 (2) ^3 C^) ô (y) ^3 (y) Si {z) Si (z),

h{l/z) -34 (y +
So{i/z)SsQ/ + z) 3. S^ = So (y) S3 (v/) S., (2) ^3 (z) + S^ {y) S^ {1/) S^ (z) 3^ (z),
') 5.2-93
S,{yz)
Si{i/
+ z) S.,Si=S, (j/) Si (y) ^2 {z) Si {z) + Si (j/) ^3 (y) ^1 (2) ^3 (2),

+ S,iy)S.{i/)S,{z)S2{z).
S3{i/z)S,{y + z)SsSi = S,{y)Si{9/)S3{z)Siiz)
(Jacobi.)
4.
Obtain the duplication-fomiulae

5, {2y) S,Si^ = 5,2 (y)
53 (2y) 53
V=
Si {2y) Si'
(3/)
^3^ (y) 54^ 0/)
= S3*
(1/)
- 3i2 (3,) S^2 (_y)^
- -î' (y)
-92'^
(.^),
- S,* (y ) = Si*{y)- S,*
(y).
(Jacobi.)
Obtain the dupHcation-formula

S, (2y) 52^3^4
= 2^1 (y) -9, (y) ^3 (y) ^4 (y).

(Jacobi.)
Obtain dupUcation-fomuilae from the results indicated in example
Shew
2.
that, with the notation of 21 '22,
[l]-[2] = [4]'-[3]',
[l]-[3] = [l]'-[3]',
[l]-[4] = [2]'-[3]',
[2]-[3] = [l]'-[4]',
[2]-[4] = [2]-[4]',
[3]-[4] = [2]'-[l]'.
Shew that
= [11 22]' + [2211]' -[4433]' + [3344]',

=
2[1133] [1133]' + [3311]'-[4422]' + [2244]',
2[1144] = [1144]'+[4411]'-[3322]' + [2233]',
2 [2233] = [2233]' + [3322]' - [441 1]' + [1 1 44]',
2[2244] = [2244]' + [4422]'-[3311]' + [1133]',
2 [3344] = [3344]' + [4433]' - [221 1 + [1 1 22]'.
2 [11 22]
]'
(Jacobi.)
Obtain the formulae
2n-^fa^=-2qi
k^k'
10.
Deduce the following
= 2fyi n
i-esults
{(1
-?2")2
{(l+?2H)2(l-f/-i)-
from example 9
- j2n- 1)6 = 2î >{'/(->,
(1
{l-f'^f
=27v-'q-^M'K\
(i-j)
= An-^q-^k^k'-'K\
n=l
-y2n-l)-2|^.
(1
(l+?2-i)=22^(M')-i,
1=1
(i+92)
(1+9"
=lq^^-kk'-^,
Iq-^k^k'-K
n=l
(Jacobi.)
489
THE THETA FUNCTIONS

11.
By considering
whose corners are ^tt,
and deduce
12.
that,
when
.*
/*
Itt, ^Tr
I{z)
e^"" dz taken along the
contour formed by the parallelogram
i (2)
+ Trr, ^tt + ttt,
shew that
<Î{ttt),
Obtain the following expansions
S3'{z)_
)'^
^ sin 23
each expansion being valid in the strip of the s-plane in which the series involved
is
(Jacobi.)
13.
Shew
that, if
\I{y)\<l
(ttt-)
and / (2)
1
cot y + cot
^^^^=
^1
^1
<7
+4 i
(ttt),
then
q'"'^ sin
{2my + 2nz).
m=l =i
(2)
(y)
(Math. Trip. 1908.)

14.
Shew
that, if
/(s)
< \I (ttt),
then
^--rT=io+
^,(S)"2'
TT
a = 2 2
where
2 acos2?i2,
j('
+ ^)(2'+'+^).
(Math. Trip. 1903.)

[Obtain a reduction formula for by considering
^ {Si(z)}-U-''^'dz
taken round the
contour of example 11.]

15.
Shew that
is
=il-2j2ncos:
cos 22 + g*" J
^1(2)
a doublv-periodic function of
with no singularities, and deduce that
it is zero.
Prove similarly that
^2(^)_
^3
"~
(2)
16.
Obtain the values of
lil
+ 2?^" cos 22 + 2^"
+ 2g'''-icos22 + j4-2'
g2n-l
^4'.(g)^.
^4(2)
g-" sin 22
A%
!=i 1
.92 (z)
sii,
2g
$^"-2'
=i l-2g^''-icos22 +
k, k',
[/{.=
K, K' correct to six places of decimals when q=^^.

0-895769,
^=2-262700,
F = 0-444518,
Zr' = 1-658414.]
,
.
[CHAP. XXI
490
Shew
17.
that, if
w+x + + 2=0,
i/
then, with the notation of 21-22,
[3]
+ [l] = [2] + [4],
[1234] + [3412] + [2143] + [4321] =0.
Shew
18.
that
^4(y)
By putting x=y = z, w=3a;
19.
results
^4(2)
hilZ + z)
in Jacobi's
'
'î(y)Si{^)h(2/ + ^y
fundamental formulae, obtain the following
= ^^3
5^
^33 {x) S3 (ar) - 3i^ (x) $i (ar) = B^^ {2x) 5-2,
= ^33 (2^) ^3
So3 (.f) 3. (Sx) + 3^3 (^) 5^
Si3 (^) ^j (3^.) ^. ^^3
(^.)
^^ (ar)
(2.1,.)
(3.1;)
20.
{Si^ (x) Si {3x) Si^
(^) ^4 (3^) ^2^}^

+ Si^ {x) Si {Sx) Si^} ^ + {Ss^ (x) S3 {3x) S^^ ^
= {52=*
^2 (3.f) S3- + ^4^ (x) Si {3x) .932}
('*-)
(Trinity, 1882.)
CHAPTER XXII
THE JACOBIAN ELLIPTIC FUNCTIONS
Elliptic functions with two simple poles.
22"1.
In the course of pr6ving general theorems concerning

at the
beginning of Chapter xx,
it
was shewn that two
elliptic functions
classes of elliptic
functions were simpler than any others so far as their singularities were
concerned, namely the elliptic functions of order
2.
The
first class
of those with a single double pole (with zero residue) in each

consists of those with
two simple poles in each
cell,
the
sum
cell,
consists
the second
of the residues
at these poles being zero.
An example of the first class, namely ^ {z), was discussed at length in

Chapter xx in the present chapter we shall discuss various examples of
;
the second
It will
class,
known
be seen
as Jacobian elliptic functions*.
22"122, note) that, in certain circumstances, the Jacobian
ordinary circular functions
functions degenerate into the
notation (invented by Jacobi and modified by
accordingly, a
Gudermann and
Glaisher) will
be employed which emphasizes an analogy between the Jacobian functions
and the circular functions.
From
the theoretical aspect,
it
is
most simple
functions as quotients of Theta-functions ( 21"61).
fundamental
properties
to regard the
But
as
many
Jacobian
of their
can be obtained by quite elementary methods,
without appealing to the theory of Theta-functions, we shall discuss the
making use
functions without
do so
for the
22'11.
It
of Chapter
xxi except when
The Jacobian
was shewn in
elliptic functions,
21 "61 that
desirable to
snu, cnu, dnu.
if
'$,,%{ul%'y
the Theta-functions being formed with parameter
where k^
it is
sake of brevity or simplicity.
= ô (0
t)/^3 (0 r).
|
Conversely,
modulus-f) be given, then, unless
k-^1
if
t,
then
the constant h (called the
or /c-Ô, a value of r can be found
These functions were introduced by Jacobi, but many of their properties were obtained
who used a different notation. See the note on p. 512.
is the acute angle such that sin d = k, 6 is called the modular angle.
t If 0</c<l, and
*
independently by Abel,
[CHAP. XXII
492
21-7-21-712) for which
t)
(0
(0 t)
A--,
so
that
the
sokition
of the differential equation
;|y=(i-yn(i-%')
subject to the condition
-y^
1 is
the Theta-functions being formed with the parameter t which has been
determined.
The
may be
written
to=i\l-t-)-^-{l-k-H')-^-dt,
Jo
and, by the methods of
may be shewn that, if y and a are conmay be expressed in terms of u as the
21-73, it
nected by this integral formula, y
quotient of two Theta-functions, in the form already given.
Thus,
if
Jo
y may
be regarded as the function of u defined by the quotient of the Thetaan analytic function of u except at its singularities,
simple poles to denote this functional dependence, we write
functions, so that y
which are
all
is
y=
or simply
The
y= sn
u,
when
function sn u
is
sn(?/, k),
unnecessary to emphasize the modulus*.
it is
known
as a Jacobian elliptic function of
[Unless the theory of the Theta-functions
poles.
Cf.
ôw
(New York,
i.
cnOa-) =
''"(" ^>
Then, from the relation of
^ sn
The modulus
21"6,
du
*
assumed,
it is
exceedingly difficult to shew
as a function of u which is analytic except at simple
Hancock, Elliptic Functions,
write
is
and
^^^-
^"=a;:w5?)
that the integral formula defines
ii,
i(
1910).]
(B),
^^^;(^;y^
a,
*^>-
aTWV)
we have
= en M dn
will alwaj's be iuserted
(I),
(/
when
it is
not
k.
22-121]
2'2'12,
and from the relations of
21
k'^
cn^ u
en
-=
dn
when
poles
make sn
necessary, that sn
;
it
From
on
u,
en
u,
in the plane of the
(Ill),
(I^)u,
en
dn u as defined
u,
(IV)
(I), (II), (III),
the integral
the
dn u determinate functions of u. It will be assumed,

dn u are one-valued functions of ti, analytic except at their
complex variable
u=
bj'
these four relations
u,
m,
Simple pi'operties of sn
22'12.
be assumed that they are one-valued analytic functions of
will also
made
are
(II),
discuss the properties of the functions sn
equations (A), (B), (C) by using the four relations

are sufficient to
=1
u + dn- u
sn-
and, obviously,
now
493
we have
'2,
sn- u
"We shall
(1
en
u,
u,
^ (1
t-)
from
k^
dn
to -f oo
and from
k'^
to
when
cuts
xi
u.
k'-t-)
evident, on writing
df^ it is
Jo
for
that, if the sign of
t,
Hence sn
is
Since sn (
y be changed, the sign of u
an odd function of
ii)
sn
u.
account of the one-valuedness of en

it
follows that either the
upper
In the special case u
taken.
has to be taken always; hence
of
u.
In
like
manner, dn u
( u) = + en u; on
by the theory of analytic continuation
u,
sign, or else the lower sign,

0,
must always be
the upper sign has to be taken, and so
cn{-u) = cnu, and cnu
an even function of
is
also changed.
follows from (II) that en
it
u,
is
it
cm even function
is
u.
These results are also obvious from the definitions (A), (B) and (C) of
2211.
Next,
let
us differentiate the equation sn- u
+ en- u =
on using equation
we get
(I),
du = sn
d en a
z
in like
manner, from equations (III) and
d dn u
=
du
-,
22"121.
Tlie
A;-
dn u
(I)
we have
sn u en
valued function of
^'
-f 1
as k
0, k' is
as the
-f
complementary
k' is
a one-
k.
that, if
then
known
of the cut in the ^--plane from 1 to
[With the aid of the Theta-functions, we can make
Shew
u.
compleme)dary modulus.
+ k'- = 1 and
On account
modulus.
If k-
Example.
?/
= en {u,
k).
k'^ one-valued,
by defining
it
to
be
[CHAP. XXII
494
shew
Also,
= [ \l
that, if
then
= dn
3/
(,
^-')
J (^2
_ /.'2) - I dt^
/().
[These results are sometimes written in the form
J en
22122.
J An
ti
It
Glaiskers notation* for quotients.
A short and
convenient notation has been invented by Glaisher to express

the reciprocals
elliptic functions
and quotients of the Jacobian
reciprocals
are denoted by reversing the order of the letters which express the function,
thus
ns u
1/sn
nc u
u,
= l /en u,
nd
/dn u
while quotients are denoted by writing in order the
= sn u/cn
u,
w = en w/sn
w,
sc u
[Note.
Aamw,
letters of the
sn ?^/dn
ds u = dn ujsn
sd u
= en w/dn w,
u = dn u/cn u.
u,
cd u
u,
dc
rt,
en u, dn u was sinam u, cosam m,
due to Gudermannt, who also wrote tuu,
Jacobi's notation for the functions sn
now
tanam u,
the abbreviations
as an abbreviation for
The
first
denominator functions, thus
numerator and
cs
in use being
in place of
what
is
now
written sc
u.
reason for Jacobi's notation was that he regarded the inverse of the integral
u= |*(l-Fsin2^)-^rf^
and wrote J
as fundamental,
(^
= am w;
./
r'^'\i-k'-'fi)-îi+kH^)--^
dt=
Jo
.]
{\-t^)-^{l-kH^)-^dt
r
J dc
{t-^-l)-^{t^-k^)~^dt
Ud M
{f'-f^)-^^t''+k^r^dt
H
=['
i(
{t''-l)-^{t^-k^)-^dt
u
["'"
{t''-\)-^-{FH'-+k-^)~^-dt
{fi-i)-i{i-r-f^)-^dt.
The addition-theorem for

shall
cs
7 ds
/'
J cd
We
sin^c^)- for -j-
Obtain the following results
Example.
22'2.
A^ = (l-/?--
he also wrote
now shew how
the function Qnu.
to express sn (u
+ v)
in terms of the Jacobian
the result will be the addition-theorem for the

elliptic functions of u and v
function sn u it will be an addition-theorem in the strict sense, as it can
be written in the form of an algebraic relation connecting sn u, sn v, sn {u + v).
;
Messenger of Mathematics,
t Journal fiir Math,

J Fundamenta Nova,
xi.
(1882), p. 86.
xviii. (1838), pp. 12, 20.
p. 30.
As k*-Q,
am )(--(/.
the jacobian elliptic functions
22'122, 22*2]
495
[There are numerous methods of establishing the result the one given
due to Euler*, who was the first to obtain (in 1756, 1757) the integral of
;
in the form of
X and
of
an algebraic relation between x and
Y is the same quartic
function of
^,
when
and
xi
essentially
denotes a quartic function
y.
Three t other methods are given as examples, at the end of this
Suppose that
is
section.]
vary while u-\-v remains constant and equal to
a,
say, so that
dv
die
Now
new
introduce, as
variables,
Si
sn
and
Sj
Sn
%i,
= sn v,
so thatj
.V
= (1 - s^) (1 -]es^\
and
4-
= (1 s.f) (1
I:?
defined by the equations
Sg
since xr
si),
Differentiating with regard to u and dividing by
we
find that, for general values of u

Si
=-
and
+ A;-) Si + IhH^,
if.
\.
2si
and 24 respectively,
v,
= - (1 + A;^ 52 + Ih^s-f.
Hence, by some easy algebra,

S\Sn^
S^Sx
Zk
~
Si's.:r-S./S,'
and
^2
{si-Sx^){l-k'Sx^si)
'
so
(s,s,
on integrating
- s,s,)-' ~ (s,s, - s,s;) = (1 - k's.W)-' ^ (1 - k's,%')

this equation
S1S2
where C
is
s^Si
k^ 81^82'
the constant of integration.
Replacing the expressions on the
we get
we have
left
by their values
in
terms of u and
en
S1S2 \Si
Acta Petropolitana,
?i
dn M sn y + en w dn
1 k^ sn^ u sn^ v
vi. (1761),
pp. 35-57.
w sn u
Euler had obtaiued some special cases of this
result a few years earlier.
t Another method
is
given by Legendre, Fonctions Elliptiques,
interesting geometrical proof was given
X For brevity,
we
bj-
i.
Jacobi, Journal fur Math.
(Paris, 1825), p. 20,

iii.
shall denote differential coefficients with regard to
dv
du
8 I.e. those values for
..
d^v
dii^
which en u dn u and cnv ^nv do not vanish.
(1828), p. 376.
u by
dots, thus
and an
496
That
{{)
to say,
is
u + v=
a and
we have two
integrals of the equation
[CHAP. XXII
du + dv =
0,
namely
(ii)
+ sn en
k^ sn"^ u sn^ v
sn u en V dn V
1
M dn u
?;
G,
By
each integral involving an arbitrary constant.
the general theory of
cannot be functionally
differential equations of the first order, these integrals
independent, and so
sn u en v dn v
1
is
expressible as a function oi u
On
putting
i;
= 0, we
see that
sn
sn^
A;-
en u dn u
v
ii sn"-*
function
call this
/(u)=snw; and
f(u +
v).
so the function
is
the
sn function.
We
have thus demonstrated the result that

sn {u
v)
= sn tt
which
is
en V dn V
j-^
sn v en u
4-
dn u
,
sn- u sn- V
^'-
the addition-theorem.
Using an obvious notation*, wc may write

s.cdo
Example
1.
Obtain the addition-theorem
Example
2.
Prove from
first
SoCidi
for sin
ic
by using the
results
principles that
d\
SiC2d2
cuj
cuJ
\cv
ycv
and deduce the addition-theorem
+ s.2Cidi_
k'-s^-s-r
for sn u.
(Abel, Journal fUr Math. n. (1827), p. 105.)
Example
3.
Shew
that
s^8^
s^c^d^- s^Cidi
_ SiCid^+ s^Cidi ^ SidiC^+SjdiCi

did2+khiS2CiC2'
CiC2 + Sidj^Sod2
(Cayley, Elliptic Functions (1876), p. 63.)
Example
4.
Oljtain the addition-theorem for sn u from the results
î(y-H^)^4(y-^)^253=-9,(y)54(y)^2(^)^3(^)+-^2(y)^3(3/)5i(^)54(4
^4 {y+z) Si(2/-2)
V= V
(y) î' (z)
- ^1- ii/)
given in Chapter xxi, Miscellaneous Examples
Example
5.
Assuming that the coordinates

?/-
can be expressed in the form (sn

Abel's
method
*
u,
^l'
and 3
of
(z),
(Jacobi.)
(pp. 487, 488).
any point on the curve
= (l-;f2)(l-y5;2a;2)
en u dn
u),
obtain the addition-theorem for sn u by
( 20-312).
This uotation
is
due
to Glaisher, Messenger, x. (1881), pj).
9'2,
124.
497
22*21]
[Consider the intersections of the given curve with the variable curve y =
one
is (0,
1)
let
the others have parameters u^, Uz, M3, of which u^.

choice of
arbitrarily Vjy suitable
method
and deduce that
of 20-312,
and
Shew
n.
that U1
+ U2 + U3
xii
+ mx + n.v'^
may
be chosen
is
constant, by the
(^1
+ ^2+^3)
by taking
this constant is zero
= 0, = -|(1+F).
Observe also that, by reason of the relations
we have
^3(1 -k'^Xi^x.^) = X3-{\+ ,2_
2)
^mx^x^^x^-^mxix.,- nxxX2
= {xi+x-2-{-x^ nxi x-2 X3) {Xi + X2) 2mxi Xo - nx^ Xo {xi + ^2)
The addition-theorems for en u and dn
22-21.
We
shall
now
establish the results
en
-
dn
(z<
(u
v)
en M en w
k's^'s.-y en- (u
dn dn V k- sn w sn v en
^
+ v) =
rw-^^
v)
them
at the
is
en (u
from the formula
end of
22*2,
+ v) = + -^ r^~
1
k-sisi
it is
li
0,
for other values, so the
we
see
ambiguous sign
is
analytic
u,
1 for
some values of
really definite
that the plus sign has to be taken.
The
first
formula
for dn {u + v) follows in like manner from the identity

- k'^sisiy - k^ (s^CoJz + s.Cxdn^y
= (1 - k'-si) (1 - fi'si) + k*sisi (1 - si) (1 - si) - 2k's,SoC,C2d,d2,
w.
ii.
A.
is left
u,
putting
The formula
the proof of which
v).
we have
consequently proved.
(1
ineonsistent with the theory
of analytic continuation that their ratio should be

1
sn (u
for
of these expressions are one- valued funetions of
exeept at isolated poles and zeros, and
= (l - k^s^^^s^^Y {1 - sn- (u + v)]

= (1 - k'^S^'^S^y - (SjCaCô + SoCidiY
= 1 - 'Ik'-si's^ + k*s,*s.2' - sf (1 - si) (1 - k^si)
si (1 si) (1 k-si) 2sxS.2CiC.2did2
= (1 si) (1 si) + sisi (1 - k'si) (1 - k'si)
2siS2CiC.,dxd.2
= (ciCo - SySod^d^y
and
u en v
and so
But both
Y,
Using the notation introdueed
sn w sn V dn w dn w
the most simple method of obtaining
(1
u.
to the reader.
32
is
498
Example
[CHAP. XXII
Shew that
1.
du
{u
+ v) dn {u-v) = -j^j^î^'
(Jacobi.)
[A
in the
set of
33 formulae of this nature connecting functions of u
Fundamenta Nova,
Example
Shew
2.
cnu + cnv
dn
so that (en r(+cn v)/(snMdn?' + sn vdnii)
+cn
v is
given
pp. 32-34.]
cu sn M dn V + sn
{1
and of ^^-
that
equal to
+v
is
cnw+cnv
ov sn
?<
dn v + sn dn u
'
i;
a function of
î
+v
only; and deduce that
it is
+ v)}lsvi{u + v).
{u
Obtain a corresponding result for the function
(iC9
+ S2Ci)/(c?j+(a?2).
(Cayley, Messenger, xiv. (1885), pp. 56-61.)
Example
Shew
3.
that
- Fsn2 {u + v) sn2 {u-v) = {\- Fsn* u){l- F sn* v) (1 - Fsn2 sn2 y) -2,
^'- + >fc2cn2(?< + v) (in^{u~v) = {k"^ + k-cn*u) (/{'2+Fcn' r) (1 -Fsn2%sn2w)-2.
1
?<
(Jacobi and Glaisher.)
Example
22-2
Example
Obtain the addition-theorems for cn(-w
4.
example
5.
+ r),
by the method
dn(w-|-'i?)
of
4.
Using Glaisher's abridged notation {Messenger,

s, c,
d=s\\
u,
en
u,
dn
and
u,
C,
>S',
Z) = sn 2i,
x. (1881), p. 105),
en
2i(,
dn
namely
2,
prove that
2gCt^
^-
l->{;254'
l-2g2^^.2g4
l_y[,.254
'
^-
- 2/t2s2 + y[;y
1
L y[^s4-
'
(i+^)4_(i_î
Example
6.
With the notation
(l
of
example
-?:,S')^
+ (l-Z-,S')^'
5,
shew that
\-D
D-BC-k-^
D-G
= 1-C
k^{D-G)
l+Z>~F(l + (7)
k'^'+D-BC
^_ D + C _ D + k-^C-k _ /;'2(i-Z)) _ /j:'2(i + (7)
~ k-^D -C)~ k"- + D- PC
^ ~l-|-i>~
k:^{l + G)'
k'^{\ + D
k'^+D + k'^G _ D+C _k'^{\-C)
\ + D
"l + C" D~C ~k'^ + D-k-^G'
^
'^
(Glaisher.)
22'3.
We
The constant K.
have seen that,
if
u=\\l-^)-^~{l-kH'')-^~dt,
J
y=
then
If
we take
sn
{u, k).
the upper limit to be unity (the path of integration being
a straight line), it is customary

symbol K, so that sn {K, k)= 1.
[It will
to denote the value of the integral
be seen in 22*302 that this detinition of A'
^TTâ^ in 21-61.]
is
by the
equivalent to the definition as
22 3-22 '302]
and dn
k' to fix the ambiguity in sign,
and trace the change in (1 k'H-)^ as t increases from to 1
expression is initially unity and as neither of its branch points (at
is encountered, the final value of the expression is positive, and so
is a continuous function of k, its value is
and therefore, since dn
It is obvious that
<
suppose
since this
t
= k~^)
+
it is
A^'
The
A'
<
en
1,
always
K=
^=
499
k'.
K are
elliptic functions of
K in terms of
The expression of
22'301.
thus given by the formulae
cn^ = 0,
sniT-l,
In the integral defining K, write
K^\
dn
k'.
k.
= sin <^,
K=
and we have at once
(l-A-2sin-(/))-^cZ(/>.
When \k\<\,
may be expanded in a series of powers of k,

^ (by 3'34, since sin-" (^ ^ 1
the integi-and
the series converging uniformly with regard to

integrating term-by-term
( 4*7),
K=\^F{\,
where
k"^.
values of c
all
\;
we
at once get
1;
k^=\^Fl\,
\-
1;
By the theory of analytic continuation, this result holds for

oo in the c-plane, since
when a cut is made from 1 to
-l-
both the integrand and the hypergeometric function are one-valued and
analytic in the cut plane.
Shew
Example.
that
(Legendre, Fonctions Elliptiques,
The eqvAvalence of the
22'302.
Taking ^ = ^^3Consequently,
of sn w in the
sn
cell
with corners
0,
and n are
{t-\-\)
and so
Therefore, since the
\7rî^.
^nS^, n
sum
(1825), p. 62.)
K-
see at once that &n{hTrd^) =
m has a double zero at
20'13 that the only zeros of 1
definitions of
we
in 21-61,
i.
B^,
tt
- 1)
= Jn-
(r
are at the points
.93'-^
is
cn(|7r53-)
number
two,
= 0.
of poles
follows from
it
(4?n-l-l-|-2?ir) ^3^,
where
Therefore, with the definition of 22'3,
integers.
A'=|7r(4m-l-l+2r)532.
Now
?i
= 0,
take t to be a pure imaginary, so that
< < 1,
/(
and
is
real
and we have
so that
\TT{Am
where
If
is
+ l)B^^=
a positive integer or zero
m is a
positive integer, since
it is
(1
''(l->t2sin2 0)~2(/^,
obviously not a negative integer.
k'^
sin^ 4>)~^
^^ i^
* continuous function of a and
J
so passes through
a value of a
less
all
than
values between
^tt,
A7(4?ra -I- 1 )
and
= hrrS^^ =
sn
so
which
is
untrue, since
and
as a increases from
to ^tt,
we can
such that
f
(^TrSs^)
sn
"
(
-P
sin2 cj))-^ dc^
= sin a < 1
= 1.
(-177^32)
322
find
500
Therefore
must he
that
zero,
we have
to say
is
[CHAP. XXIT
But both
and ^n^^- are analytic functions of / when tlie o-plane is cut from 1 to
and so, by the theory of analytic continuation, this result, proved when 0<^-<l,
00,
persists throughout the cut plane.
The equivalence
Example
By
1.
Example
therefore been established.
considering the integral
2K= 0.
shew that sn
K has
of the definitions of
Prove that
2.
snÂ'=(l
+/{;')"
dn|A' = /CA
cnhK=M^{\ + k')'^,
[Notice that when u = \K, cn2M = 0. The simplest way of determining the signs to
be attached to the various radicals is ^to make ^--a-O, X'-a-l, and then sn ?<, en w, dn u
degenerate into sin
Example
cos
u,
u, 1.]
Prove, by means of the theory of Theta-functions, that
3.
iA'=dn \K=k'^.
cs
The periodic
22*31.
K)
with
(associated
proper^ties
of the Jacobian
elliptic functions.
The intimate connexion

snu, en
of
with periodic properties of the functions
dnu, which may be anticipated from the periodic properties of
11,
Theta-functions associated with -
will
tt,
now be demonstrated
directly from
the addition-theorem.
By
22"2, Ave
have
sn(u
^
+ K)
^
In like manner, from

en
TT
Hence
sn
(it
/
(i
(
^
snucnK dnK + iiKcn.udnu =
Tj^x
2it
^'
sd
u,
cn((t
/^)
= en
ÎK)
sn {u
-h
^K) =
is
dn (a +
-^
7^;
dn(u-l-ir)
-f
Thus 4iK
cd
u.
22*2],
and, similarly, en {u
Finally,
-77
.,
k' sn- a sn^
+ K) =
4-
dn
u,
sn {u
-f
=
{it
2K) =
ii )
^'sdw
^^~
.->
A;ndM
'IK)
sn
u,
^'
nd
u.
= sn u,
= dn
en
u.
(?(
-I-
a period of each of the functions sn u, en
4 A")
(/,
= en u.
while dn u has
the smaller period 2K.
Example
1.
Obtain the results

sn ( u + A') = cd
directly
u,
en (m + K) = -
from the definitions of sn
Example
2.
Shew that
?t,
en
cs
?<,
cs
k' sd u,
dn ( u + K) = k' nd
dn u as quotients
K u) = k'.
(
m,
of Theta-functions.
22*31,
the jagobiax elliptic functions
2'2-3'2]
The constant K'.
22-32.
We
501
shall
denote the
integi'al
Jo
by the symbol K',

and so
k^ (= c)
K'
so that
the same function of
is
k'-
(= c) as
is
of
K'=\'rrFi^-,
when
to
the c'-plane
-X
1 to
oo
1; k"^
when
i.e.
the c-plane
is
cut from
To shew that
that
cut from
is
\;
1,
if T7-'=
K'
this definition of
is
equivalent to the definition of 21 "61, we observe
is
the one-valued function of P, in the cut plane, defined by the equations
K= M^- (0
F = 5,^ (0
r),
r)
^ ^3* (0
r),
I
while, with the definition of 21'51,
^' = 1^^32(010,
so that K'
;:'2
must be the same function of

K' as
kf^
= 52*(01r')^^3'*(0|r'),
as
K is of
k"^
and
this is consistent with the
integral definition of
Jo
It will
1 to
+ X
now be shewn
that, if the c-plane be cut from
then, in the cut plane,
K' may be
K' =r\s'-l)--^{lJ
First suppose that
concerned are
real.
to
00
and from

k's')
-^
<
1,
ds.
0<^'<1,
so that
<
^'
and then the
integrals
In the integral
f\l -t^)-iO--k'H')-^dt
.'0
make
the substitution
s
{l-k'H-)-'^,
which gives
(s'
1)4
= k't (1 -
I^-f) -i,
(1
- k's'-)i = k' (1 - i?f (1 - k'H-) ' *,
k'H
_
dt^Xl-k'H-'f
ds
it
being understood that the positive value of each radical
On
substitution,
we
at once get the result stated,
K' =
provided that
of^.
< k <\; the
'
(6-^
1)
- i (1
is
to be taken.
namely that
- k's') - i ds,
result has next to be
extended to complex values
502
Consider
T'^'
{l-t^)~^{l -kH"-) "
the path of integration passing above the point
The path may be taken
dt.
and not crossing the imaginary axis*.

to 1 - 8 and 1 + 8 to k~^ together
1,
to be the straight lines joining
with a semicircle of (small) radius S above the real
+1
reduce to
each radical
at
is
the
positive
8-*-0,
and the
Now
which t
integral
\^'^
+/;'.
and as
/l/A;
is
cut from
to
and so
K is analytic throughout the cut plane.
it is
equal to the analytic function K'
qo
that
is
when
to say
and from
K + iK'=\
Since
like 8-
(F-?('-i)"2(i_,,2)-2o;?(,
the equality persists throughout the cut plane
c-plane
(\-k'^t-)^
{i:^-\y^{\-hH^y^ dt
analytic thi'oughout the cut plane,
< X-'< 1,
when the
always
analytic throughout the cut plane, while
Hence
is
it is
round the semicircle tends to zero
{\-f-)-^-{\-k'H^y^-clt=\
is
(lt-y and
If
axis.
value of the former at 1 + S is e"^"" S^ (2 + 8)^ = - i (t^ - Vf, where

while the value of the latter at <=1 is +/' when k is real, and
hence by the theory of analytic continuation
Make
[CHAP. XXII
to
+ oo
{l-t-)--îl-hH')-^~dt,
Jo
we have
sn
(K + iK') =
while the value of en
(K +
iK')
the prescribed path to the point
Example
1.
i/k,
is
the value of (1
1/A-,
and so
{-t{l-t){\- kH)} -^dt = ^
Example
its
P)^ when
value
is
has followed
ik'/k, not +ik'lk.
Shew that
=
I f^ \t{l-t){l-Pt)}~-^dt l
-
dn (K + iK') =
2.
Shew
{t{t-'^){kH-\)}~^^dt=K,
{^ (^
1) (1
- FOl- ~^'dt = A".
that K' satisfies the same linear differential equation as
K ( 22'301
example).
The periodic propertiesX {associated with
22*33.
K + iK') of the Jacohian
elliptic functions.
If
we make use
sn
* II
(A:)
{K +
>
of the three equations
iK')
because
= k-\
arg c
en
(K + iK') = -
ik'jk,
dn {K + iK')
0,
< tt.
t The path of integration passes above the point u = k.

J The double periodicity of snw may be inferred from dynamical considerations.
Whittaker, Analytical Dynamics (1917), 44.
See
at once, from the addition-theorems for sn u, en u,
we get
results
sn (
503
22 33-22-341]
dn
u,
the following
,^
4- ii
^,
+ *A
)
^
= sn u en {K +
and similarly
By
'^
= k~^ do
(K + iK') + sn (Z H- iK') cnudnw

Tz
2/ ir
ir'\
k^
sn^ u sn^ {K +%K )
iK') dn
(t,
en
'{u
dn
(zt
K + iK') = ik'k~^ nc
+
+
if
iK')
= ik'
sc w.
we have
repeated applications of these formulae

f
sn (u
+ 2K+
2iK')
=-
^cn(u+2K-\-2iK')^
[dn (m
Hence
+ 2Z +
period
the smaller
2K +
u,
en
t<,
= - dn w,
2iK')
the functions sn
sn
u,
sn (a
4>K
+ UK') =
en (u
4^"
[dn {u
4<K
-.
u and dn u have period
4tX')
sn
u,
=cn
m,
+ UK') = dn u.
4iK + UK', ivhile en u
2iK'.
The periodic proper-ties (associated with iK') of
22-34.
has
the
Jacobian
elliptic functions.
By
the addition-theorem
sn (u
we have
= Sn(u-K + K+ iK')
= k-' dc (u - K)
iK')
=
Similarly
we
find the equations
ik~^ ds u,
+ iK') = ics u.
en (u
dn {u
By
k~^ ns u.
iK')
repeated applications of these formulae

'
Hence
Bn{u-\- 2iK')
sn
u,
en (u
2iK')
= - en u,
,dn (u
2iK')
= dn
the functions
u,
we have
+ UK') =
en {u + UK) =
sn
sn
u,
en
u,
+ UK') = dn
u.
(ii
[dn {u
en u and dn u have period UK',
luhile
sn u has the
smaller period 2iK'

Example.
Obtain the formulae

sn iu + 2mK+ 2mK') = (
en
dn
22'341.
)'"
sn
ti,
(u + 27nK+ 2niK') = - ) + " en

( + 2mK+2niK') = - )" dn u.
(
i^,
The behaviour of
the
Jacobian
elliptic
functions near the origin
and near iK'.
We
d
-^ sn u
du
have
= en M dn u,
d^
-i sn
du^
4^^ gn^ u en
u dn m - en m dn u (dn^ u
k^-
cn^ u).
504
Hence, by Maclaurin's theorem, we have,

sn u
on using the fact that sn u
=u-~(l + k-)
[CHAP. XXII
for small values of
u'
w|,
(u'),
an odd function.
is
In like manner
dn
= l-l khr +
(u').
It follows that
sn (u
+ iK') = k~^ ns u
ku
,-
ku
^'-^
-7rr-
bk
^ 3,
u+0(u');
,
2k- 1
iu
^,
en (u
i
+ iK') = tt; +
dn (u
+ iK') = - - + ^/^' ill +

u
b
and similarly
It follows that at the point
poles with residues k~^,
iK'
ik~^, i
(u'),
(u').
^
the functions sn
r,
dnv have simple
cnv,
ixspectively.
Obtain the residues of snw, cnw,
Example.
dnw
at iK'
hy the theory of Theta-
fiinctions.
General description of the functiotis sn
2235.
The foregoing
summarised
(I)
(mod.
4jfir,
en
investigations of the functions sn
u,
u,
dn
u.
en u and dn u
may be
in the following terms
The function sn m
^K, 2iK'.
u,
is
a doubly-periodic function of u with period:?
It is analytic except at the points
congruent to iK' or to
2iK'); these points are simple poles, the residues at the
being k~^ and the residues at the second set

has a simple zero at all points congruent to
all
being - k~^
2K + iK'
first set all
and the function
(mod. 2K, 2iK').
may
be observed that sn u is the only function of u satisfying this description for

were another such function, sn m (m) would have no singularities and would be
a doubly-periodic function; hence ( 20-12) it would be a constant, and this constant
vanishes, as may be seen by putting u = 0; so that ^ (tt) = sn u.
It
if
(m)
When 0< A;^<

real values of u
(II)
^K
1, it is
and
is
obvious that
The function cm*
and 2K 2iK'.
2ir-f- iK' (mod. 4/1",
-+
K and K' are real, and sn u
a pure imaginary
It
2K
is
-\-
is
when
a is a
is
real for
pure imaginary.
a doubly-periodic function of u with periods
analytic except at points congruent to iK' or to
2iK'); these points are simple poles, the residues
22*35-22"4]
505
at the first set being ik~\ and the residues at the second set being ik~^
and the function has a simple zero at all points congruent to
(mod. 2K, 2iK').
The function dn u
(Ill)
2,K and
(mod.
being
''2K,
4tiK')
i, and
a doubly-periodic function of u with periods
is
congruent to iK' or to ZiK'
It is analytic except at points
4iiK'.
these points are simple poles, the residues at the
a simple zero at
points congruent to
all
first set
and the function has

+ iK' (mod. 2K, 2iK').
the residues at the second set being
[To see that the fvinctions have no zeros or poles other than those just
specified,
recourse must be had to their definitions in terms of Theta-functions.]

22*351.
If ej,
The connexion between Weierstrassian and Jacobian
62) ^3
be any three distinct numbers whose
sum
elliptic functions.
and
is zero,
if
we write
61-63
^ = ^3 + sn2
we have
^^^^
(
;7
~"
('^'i
=4
^z)'^
(ns^Xw
= 4X2 (ci - 63) ~ V^/ - ^3)

Hence,
if X2
= 6i 63 and
and so
where a
e3
is
+ (ei-63)
>?'2
(e-.
63)/(6i
(u; g2, 93)
the Jacobian elhptic function having
(/
63),
^^
- ei)
then y
(ns^Xw-^^)
- 63) - 63}.
{e^
the equation*
satisfies
ns2 {u
(ci
Qi, 9z\
and so
see that a is a period,
= 63 + {ei - 63)
its
1)
{y -
A/^^zyr = ^ (' + ;
ns^ Jm (61-63)2,
Making u -^ 0, we
a constant.
'
^" "'^^ ^** ^^^ ^^' ^^^
~e'i)-\^ ns^Xzi
(ei
(Xm, k)
- 63)%
modulus given by the
eqviation
1-63
61-63
22'4.
The
Jacobi's
imaginary transformation'f
which gave a transformation from Theta-functions
result of | 21-51,
with parameter t to Theta-functions with parameter
produces a transformation of Jacobian

is
t'
elliptic functions
l/r, naturally
this transformation
expressed by the equations

sn
(iu, k)
= i sc (u,
en
k'),
{iu, k)
Suppose, for simplicity, that
<
nc
<
dn
(u, k'),
and y >
{in, k)
= dc (a,
).
let
''
f
Jo
{l-t')'^{l-
so that
iy
kH-) ' ^dt
= sn (iu,
k)
take the path of integration to be a straight
en
(iu,
k)
= (1 + y'^)^,
dn
{iu,
iu,
line,
A;)
and we have
= (1
-1-
k-y-)^.
The values of ^2 ^''^^ 93 ^^^, ^^ usual, - J2e23 and êieê-s.

f Fundamenta Nova, pp. 34, 35. Abel {Journal fUr Math. 11. (1827), p. 104) derives the
double periodicity of elliptic functions from this result. Cf. a letter of Jan. 12, 1828, from Jacobi
*
to Legendre [Jacobi, Ges. Werke,
i.
(1881), p. 402].
506
Now
of
put V
= 77/(1
from
is
values of
to
Then
to
<
77
and hence,
177/(177-)-,
from
is
t^
<
where
?;-)-,
[CHAP. XXII
1,
so that the range of values
if
iUl{\
t^)^,
the range of
77.
= i{\-t^)-^-idU, (1 - r-)^ = (1 - ^,-)-*,

1 - kH' = (1 - kH^) ~^-{l- /r) - i
dt
in=\
and we have
fj-) " ^ (1
(1
îdti,
h'-ti-) '
Jo
so that
77
sn
and therefore
tj
sc {u,
We
(u, k')
k').
have thus obtained the result that
Also
and
Now
sn
{hi,
en
{in, k)
dn
{iu,
sn
k)
k)
= i sc {u,
=
(1
-f-
y'-)- --
77-) ~ ^ = nc
(1
= {l- if)^ =
and isc
{iu, k)
k').
(w, k')
u, k'),
- k'-'n'')^ ( 1 -
77-)
= dc
{a, k').
are one-valued functions of u and
Hence by the theory
cut c-plane) with isolated poles.
~ *
A;
(in the
of analytic continuation
the results proved for real values of u and k hold for general complex values
of u and
k.
Proof of
22*41.
Jacohi's imagimiry transformation by the aid of Theta-
functions.
The
may be proved
results just obtained
Thus, from
Theta-functions.
sndu
very simply by the aid of
21 61,
M_ ^3(0|T) ^,(iîT)
^''^''''^^-%{Oit)%{J^t)'
where
and
so,
by
21-51,
sn
{tu, k)
= m/^3- (0
= |4^|4> "f^^^^^'"' ^
.
= isc (v,
where
= izr"^^ (0
so that, finally,
Example
1.
sn
Prove that en
= izr'
t)
{iu, k)
{iu,
t),
l-)
k'),
(-
ir) ^3- (0
t)
= - w,
sc {u, k').
= nc(u,
k'},
dn
(in,
i-)
= do{ri,
k')
by the
aid of Theta-
functions.
Example^.
Shew that
sn {hiK', k) = iiic {^K',
en iîK', k) = {l+k)^ k-^,
[There
is
k')
= ik~-,
dn {iiK',
k)
= {l+k)K
great difficulty in determining the signs of sn îK',
method other than Jaeobi's transformation
is
used.]
ci\ jîE',
dnit'A'',
if
any
22-42]
2i^-41,
Example
3.
Shew
Example
4.
If
507
that
< <
?:/:')
and
Z-
{K +
sn \
if
^ be the
*'^
e''^*' "
V(cosec
modular
angle,
shew that
en h {K + iK') = e " i'^' ^/(cot
^),
6),
(Glaisher.)
Landens transformation*
22"42.
We
now
shall
'
(1
obtain the formula
^^l-
sin- d,)
~^dd,
= {l-^ k')
Jo
=
k, =
where
sin ^j
and
(1
cos
k') sin (^
{l-k')l{l
<^
f (1
Jo
(1
sin- 6) " *c^^,
^-'^
kr sin-
^)
k').
This formula, of which Landen was the discoverer,
means
of Jacobian
elliptic functions in
sn
{(1
+ k') u,
on writing
(f)
To obtain
this result,
k^\
{\
k')
(f)^
we make use
sn
{u, k)
am
be expressed
by-
cd
{u, k),
u^
of the equations of | 21-52,
%{z\r)^,{z\r ) ^ %{z\t)^;{z\t ) ^
-^^
may
the form
->r
= am u,
^4(2^12t)
~^
^3 (0
t) ^4 (0
namely
t)
I
^4(0|2t)
(22I2t)
and let k-^, A, A' be the modulus and quarter-periods

formed with parameter Tj then the equation
Writef
Ti
2t,
may
%{z\r)X{z\r) _
'^^{'lz\n;)
%{z\r)^,{z\r)
^,(20iTO
obviously be written
k sn {2Kz/7r, k) cd (2 A^^/tt, k)
To determine
k^ in
terms of
k,
put z
^/(l
which
gives,
on squaring,
To determine A,
k^
(1
k')
^'')/(l
= k,^ sn (4
jTt,
=
+
(A).
^tt, k,)
and we immediately get
k^,
k'),
divide equation (A) by
as stated above.
z,
and then make
^>0;
and
we get
2Kk =
so that
4^'l*
A,
A=~{[+k')K.
* Phil. Trans, of the Royal Sac. lxv. (1775), p. 285.

t It will be supposed that \R(t)\<:^, to avoid difficulties of sign which arise if R (tj) does
not lie between 1. This condition is satisfied when 0<k<l, for r is then a pure imaginary.
508
Hence, writing
(1
ii
in place of ^Kzjir,
A;')
sn
since
we
cd {u, k) = sn {(1 + k') u,

= 2Au/K =(1 + k')ii;
Example
1.
Shew
that \'JA = 2K'jK, and thence that A'
Example
2.
Shew
that
dn{u,
+1-') K'.
/i}
dn{(l+^')w,
Shew
= (l
= {l -(!+>(') sn2(, H')} nd {n, k),

î] = W + (l - k') cn^ (u, X-)}nd(M, k).
en {(!+/')",
3.
ki],
been completely proved.
so that Landen's result has
Example
[CHAP. XXII
at once get from (A)
{u, k)
4<Az/7r
that
k)
= {l-k')cn{{l+k')u,
where
ki]
+ {I +
dn
i-')
{{I
+ k') u,
l\},
X-=2^-ii/(l+^'i).
22*421.
Transformations of
The formula
Landen
of
of elliptic functions
is
elliptic functions.
what
a particular case of
is
known
as a transformation
a transformation consists in the expression of elliptic functions with
parameter t in terms of those with parameter {a + bT)j{c + dT^, where a, b, c, d are integers.
"We have had another transformation in which = - 1, 6 = 0, c = 0, c?=l, namely Jacobi's
imaginary transformation.
For the general theory of transformations, which is out-
Fundamenta
side the range of this book, the reader is referred to Jacobi,
JYova, to Klein,
Vorlesungen iiher die Theorie der elliptischen Modulfunktionen (edited by Fricke), and to
Cayley, Elliptic Functions (London, 1895).
Example.
By
considering the transformation
ro
= r+l,
shew, by the method of
22-42, that
sn
k2)=k' sd
{k'u,
{u, k),
where ^-0= ik/k', and the upper or lower sign is taken according as
and obtain formulae for en {k'u, ^2) and dn {k'^l, k'^.
I nfhdte products for
22'5.
The products
products
for
for
the
Jacohian
elliptic
or
R (r) >
functions*
the Theta-functions, obtained in
the Jacobian elliptic functions
R{t)<0
writing
once yield
21-3, at
= ^KxJtt, we
obviously
have, from 22-11, formulae (A), (B) and (C),

^ i
It
en
II
-4
i;47
g
2q*k
-k - cos
^
- lA
From
=i
u
''1
(l
[1
n
H
- 2m^ cos
+ o^"
cos 1x + (7^"-+ 29^" COS 2^ + ^^"
2a;
25-'*-!
{ _
,jî (1
?r-^r~~,
2^2"-> cos
^^
2x +
^t;;
i
5'*"-2)
+ 2g^-^ cos2a; + g^^-'^

2cf'^'''^ COS 2x + q
these results the products for the nine reciprocals and quotients can
be written down.
There are twenty-four other formulae which may be obtained
From
the duplication-formulae ( 22-21 example 5)
1-cntt
Huu
,1= sn -1 udc
u,
2
l-HduM
sn
2
*
,1
=as -u
2
dn?i-|-cnM
uc -
in the following
manner
we have
Fundamenta Nova, pp. 84-115.
sn m
= en
,1
1
?^
as 2
u.
first
509
22-421, 22*5]
Take the
and use the products
of these,
cnu
en
dnû
|i(,
fl
-2 -o)" cos j;-
sin^
=i
(1
+2
snu
for sn û,
l-cos.^'
we
get
'
7+q^^j
(-g^)"cosa;+g'2
on combining the various products.
Write u + K for
u,
x+h-rr for
and ve have
x,
dnw + ^snw l + sin.-?;

"^
cos j;
cnu
fl
+2
=i
[1
- 2 ( - g)"
(-g)"
sin.-g
+ g^"
sin
+ (^-j
a;
!
'
Writing u + iK' for in these formulae we have

k sn
and the expression
if
for ^ cd
dn
?i
?<
ri
=I n
.
+ ik' nd u
+ 2^
- =
^rr-.
) o" -^ sin
;
.v
"
<f''
^-
-.
^]
^ t
obtained by writing cos x for sin
is
in this product.
.t?
From the identities {I - cmi) (I + cmi) = an^ ti, (ksmi + idnu) {l-sm(-idmi)^l, etc.,
we at once get four other formuhxe, making eight in all the other sixteen follow in the
same way from the expressions for ds-|Mnc^M and cn^tidsîc. The reader may obtain
;
these as an example, noting specially the following
Example
1.
Shew that
^^
Example
2.
Deduce
>-i
example
fron\
((i-ij^'-'Xi+t}'""')!
and from
22 '41 example
that, if 6 be the
4,
modular angle, then
"tl+(-)" .}+*/'
and thence, by taking logarithms, obtain Jacobi's
J^= 2
H=
'
)" arc
q"'~^
tan
= arc tan
quae inter formulas elegantissimas censeri
Example
3.
By expanding each term
powers of
e^'^^^,
s,'q^
{Fund. Nova,
2 {log
and rearranging the resulting double

,
e'^^^)
series,
**
K
logsnw=log(2g')-|log/{- + logsm.r+ 2
1
Jq^ - ,
p. 108.)
- j^" e^*^)
- j2ne-2ia;)_log(l-j2n-l
+ arc tan
in the equation
/t
in
- arc tan
debet.'
log sn M = log (2 J') - i log + log sin x +
+ log (1
s,U]
result
-\og
(1
- J^n-l
g -2ia:-)j
shew that
2o'" cos 2mA'
when l/(2)l<-|7r/(r).
Obtain similar series for log en
u, log
dn î.
(Jacobi,
Example
4.
Fundamenta Nova,
j).
99.)
K
log sn
udu=
îrK'
\K log k.
/:
(Glaisher, Proc.
Royal
Soc. xxix.)
510
Fourier series for the Jacobian
226.
If u
^Rxjir, sn u
therefore
we may expand sn w
9*22)
multiples of
/mictions*
elliptic
an odd periodic function of x (with period
is
Dirichlet's conditions ( 9"2)
obviously satisfies
[cHAP. XXII
real
for
27r),
values of x
which
;
and
as a Fourier sine-series in sines of
thus
x,
sn a
hn sin nx,
n=\
the expansion being valid for
values of
all real
It is easily seen that the
x.
given by the formula
coefficients 6 are
TTibn
sn u exp {nix) dx.

.
J 77
To evaluate
this integi'al, consider
tt,
parallelogram whose corners are
tt,
snu. exp {nix) dx taken round the
ttt,
27r + ttt.
Tttt
IT
the periodic properties of sn w and exp {nix),
From
r -It
'
and
since
so,
tt + ^ttt and
we
see that
cancels
only poles of the integrand
^ ttt are the
{qua function of x) inside the contour, with residuesf
^~^
and
tt/K ] exp
niir
+ - niriT
j
we have
\\
tt +
a;
[1
sn
li
exp {nix)dx
= -^
ttt for
(-)"?"}
Hence, when
/i
g-"
()"}.
{1
-"-^"
J -2;7 + 7rTj
(J -TT
Writing
Q Tr/iTJ exp Q nirir)
k-^
respectively,
x in the second integral, we get

sn
I*
exp (mic)
is
even, hn
27r
rfa;
= ^-| 3^"
but when n
is
- (-)~}.
odd
Consequently
sn
when X
q^"^ exTp
hand
is
real
{nix)
side
*
t
is
and
ti
J5'-
sin
a;
g^ sin Sx
q^ sin
oa;
but the right-hand side of this equation

q^^
exp {nix) both tend to zero as
is
w>x,
analytic
and the
analytic except at the poles of sn u.
These results are substantially due to Jacobi, Fundamenta Nova, p. 101.

The factor ^îrlK has to be inserted because we are dealing with sn (2KxJTr).
when
left-
THE JACOBIAX ELLIPTIC FUNCTIONS
22*6, 22'6l]
Hence both
so,
defined by the inequality
is
by the theory of
= ^KxJtt),
Example
1.
Shew
CU M =
?i
A^
= 2A'.iY7r,
n=o
"
A-
?^-i
I
7.
Example
2.
By
when
/(^)
{r).
we have the
/ (a;) <
Stt
tt
'''''*"2Z +
'
result
tt/ (r)
cos
'"
!i
2??.r
l+j-^
'
^ 29" sin 2?i^

%(l+j2)
,1=1
./o
these results being valid
irl
then
cos(2?;.+ l)
^ 5
27r
<
complex
I g"^^sin(2;? + \)x
27r
valid throughout the strip

that, if
I(oc)
analj^tic continuation,
sn
(where u
511
sides are analytic in the strip (in the plane of the
variable x) which
And
<c\tt I {r).
writing x-k-\iT for
in results ah'eady obtained,
shew
that, if
u=^Kxl-K and \I{x)\<\ivI{t\

1
cdw-^,^^2^
cd-^
then
(-rg--+^cos(2^ + l )^
l_j2 +
i
TT
nd =-.>,, +
2AX-'
'"^^'-A'M'io
""
l+j2n
)" o" COS 2a;

^-^
2
Kk' =i
^^
Fourier series for reciprocals of Jacobian
22"61.
In the result of
22"6,
write u
then we see that, ifO>/(a7)>
and
27r
-FT,
+ iK'
tt/
(-^ ?"+^sin (2n + l).:

T+^î
sd..--^
'
for u
'
elliptic functions.
and consequentl}^
a;
+ ^ ttt
for
a;
(t),
so ( 22-:34)
ns
li
= (- iirjK) S
^" + * [5" + ie(2+i)
^_
- - ie-(2n+i)
_ ^2+i^
ix|/(î
w=
(- iV/iT)
[2t5-"+i sin (2n
+ 1) + (1 - q--'^-^) e-(-+i 'a'J/(i _ ^2n+i^

a;
=o
^
That
27r -
is to
r+' sin (27^ + 1) a _

:
iV ^
^,,^^,
.^
say
TT
ns=jêosee^ +
^
yJ_^L__
27r
o-'*+i
sin (2/i 4- 1)
But, apart from isolated poles at the points x
equation
is
an analytic function of x
IT
is
= wrr,
each side of this
the strip in which
I (r) >I{x)>-'TrI (r)
width of that in which the equation has been proved to

and so, by the theory of analytic continuation, this expansion for
valid throughout the wider strip, except at the points x = iiir.
strip double the
be true
ns u
\r\
a;
[cHAP. XXII
512
Obtain the following expansions, valid throughout the strip I {x)\< itI{t)
except at the poles of the first term on the right-hand sides of the respective expansions
Example.
'
ds u =
27r
TT
cosec..-
TT
-'
2K
cot
27r
y;r
TT
^.
==
"
o-^"
=i
27r
"^
TT
27r
2 A A:
227.
An
'
(-)g2n + lcos(2?l + l).r

l-g-^-i
'
^-^,,
(
)" o^"
=i
sin 2?U7
l+y^n
Elliptic integrals.
integi-al of the
w and
of
Kk
>
2nx
sin
1+?=^"
nc = 2ZX^,sec.r -^,^2^
"^
+ l)^
i:^^.T-^
+^^2^
secA-
a2n + isin(2?i
^2^
x,
form IR(w, x)dx, where
denotes a rational function
and w^ is a QUARTIG, or CUBIC function of x (without repeated

an elliptic integral*.
factors), is called
Elliptic integrals are of considerable historical importance,
[Note.
that a very large
number
owing to the fact
of important properties of such integrals were discovered
by
Euler and Legendre before it was realised that the inverses of certain standard types of
such integrals, rather than the integrals themselves, should be regarded as fundamental
functions of analjsis.
The
first
was Gauss
mathematician to deal with

22 "S), but the
first results
elliptic functions as
opposed to
elliptic integrals
published were by Abelt and Jacobil.
The results obtained by Abel were brought to the notice of Legendre by Jacobi
immediately after the publication by Legendre of the Traite des fonciions elliptiques. In
the supplement (tome in. (1828), p. 1), Legendre comments on their discoveries in the
following terms "A peine mon ouvrage avait-il vu le jour, a peine son titre pouvait-il tre
connu des .savans etrangers, que j'appris, avec autant d'etonnement que de satisfaction,
:
que deux jeunes geometres, ÎM. Jacobi
(C.-G.-J.) de
Koeuigsberg
et
Abel de Christiania,
avaient reussi, par leurs travaux particuliers, a perfectionner considerablement la theorie
des fonctions elliptiques dans ses points les plus eleves."
An interesting
correspondence between Legendre and Jacobi was printed in Journal fur
Math. Lxxx. (1875), pp. 20.5-279; in one of the letters Legendre refers to the claim of
Gauss to have made in 1809 many of the discoveries published by Jacobi and Abel. The
validity of this claim was established by Schering (see Gauss, Werke, in. (1876), pp. 493,
494),
though the researches of Gauss
Werke, in. pp. 404-460) remained unpubli-shed until
after his death.]
We
shall
now
elliptic integral
give a brief outline of the important theorem that every
can be evaluated by the aid of Theta-functions, combined
* Strictly speaking,
means
it
is
only called an
elliptic integral
when
it
cannot be integrated by
of the elementary functions, and consequently involves one of the three kinds of elliptic
integrals introduced in 22 "72.
t Journal fur Math.
ii.
(1827), pp. 101-196.
i Jacobi announced his discovery in two letters (dated June 13, 1827 and August 2, 1827) to
Schumacher, who published extracts from them in Astr. Nach. vi. (No. 123) in September 1827
the
month
in
which Abel's memoir appeared.
22 '7, 2271]
with the elementary functions of analysis
has already been seen
it
513
(
20'6)
that this process can be carried out in the special case of jiv~^dx, since
the Weierstrassian elliptic functions can easily be expressed in terms of

Theta-functions and their derivates
( 21'73).
[The most important case practically is that in which R is a real function of x and w,
which are themselves real on the path of integration it will be shewn how, in such
circumstances, the integral may be expressed in a real form.]
;
Since
(lu,
x)
is
R (w,
where
x)
in
R(w x)=~ ^^ ^^^'

wQ (w,
x)
Q ( w,
unaffected by changing
rational function of
If
x)
it
and x
^^
x)
x),
then we have
^ ^~ ^' ^^
Q ( w, x)
'
a rational function of w- and x, since it is

therefore expressible as a
it is
of lu
sign
x.
now we multiply out
of X wherever
is
the
and x we may write
P (w, x)IQ {w,
and Q denote polynomials
Now Q (w,
a rational function of
wP {w,
x)
Q ( w;,
occurs in the expression,
x)
and substitute
for w- in
we ultimately reduce
nomial in X and w, the polynomial being linear in
iv.
it
terms
to a poly-
We
thus have an
jRj
and R2 denote
identity of the form
R (w,
by reason of the expression
x)
{Ri (x)
tvRz {x)]lw,
for w- as a quartic in
where
rational functions of x.
Now
so the
\Ro {x) dx can be evaluated by means of elementary functions only*;
problem
this process
it is
now proceed
To
reduced to that of evaluating jw~^Ri (x) dx.

necessary to obtain a canonical expression for
w'-,
carry out
which we
to do.
The expression of a quartic as
22'71.
It will
is
now be shewn
the
product of sums of squares.
that any quartic (or cubicf) in x (with no repeated
factors) can be expressed in the form
{A,(x- a)^ +
where,
if
BJx- ^Y]
{A, (x
the coefficients in the quartic are
- af + B, (x - ^f]
real,
A^, B^,
A..,
B^,
a,
/3
are
all
real.
*
The
integratiou of rational functions of one variable is discussed in text-books on Integral
Calculus.
t In the following analysis, a cubic
may
be regarded as a quartic in which the eoefiScient of
X* vanishes.
W. M. A.
33
514
To obtain
the form
S^S..
we
this result,
where
Now, X being a
observe that any quartic can be expressed yi

quadratic in
Si, S2 are
<S'i
a^x^
+ 2biX +
Ci,
S.2
constant, ^1
\S.
will
(tti
Xftj) (ci
Xco)
Let the roots of this equation be

a, yQ
[CHAP. XXII
say*
x,
aox-
2h..x
c.,.
be a perfect square in x
(61
\h^then,
X^, X^',
if
0.
by hypothesis, numbers
exist such that
\1S.2 =
Si
Xitto) (x
(o-i
on solving these as equations in
= Ai{x-
Si
a)-
^j,
Si
a)-.
S.,,
we
+ Bi{x- /3)-,
X^S..^ (a I
[Note.
s A^
S.^
(x
then
and
Xi
when X =
When
and
Si
S-i
effected.
factors, let Si
have com-
Xa2)
(c'l
XC2) (61 X62)^
and negative! when X = ai/2.

have real
that Xi and X2 should be real
is
(^1
a condition which
Xi are real
(!
is positive
^f
- a)- + Bo {x - /S)^
and has two or four complex

and distinct since
If the quartic is real

;
Xstu) {x
obviously get results of the form
and the required reduction of the quartic has been
plex factors
factors, say {x
- ^2)
(6'
- ^2)
when the
is satisfied
- î)
{-v
î),
{^
- ^2)
{^ ^2), the condition
found to be
easily
(^1
- ^2')
(^1'
&
zeros of Si
was, of course, the reason for choosing the factors
- f/) > 0,
and those of
do not interlace this
and So of the quartic in such a way
;
^S"!
that their zeros do not interlace.]
The three kinds of elliptic
22'72.
Let
a,
/3
integrals.
be determined by the rule just obtained in 22'71, and, in the
integral \'w~^Ri {x) dx, take a
new
variable
defined by the equation;!:
t=(x-a)l{x-^);
^,
= +
dx
we then have
(aB)~^dt
^^
{{Aif'
+ Bi)(Ad' + B.2)\i
* If the coefficients in the quartic are real, the factorisation
coefficients in
.Sj
and So are
factors, these factors
real.
can be carried out so that the
In the special case of the quartic having four real linear
should be associated in pairs
(to give Sj
roots of one pair do not interlace the roots of the other pair
and
tlie
S2) in
such a waj* that the
reason for this will be seen in
the note at the end of the section.
t Unless
ttj
02
= ^1
&2i i"
which case
Siâi {x-a)^ + Bi,

t
It is
S.,
= ao{x - a)^+ Bo.
rather remarkable that Jacobi did not realise the existence of this homographic
substitution
in his reduction he
employed a quadratic substitution, equivalent to the

elliptic functions which we shall introduce.
applying a Landen transformation to the
result of
22 72]
If
we
write R^ {x) in the form (a
fRi (x) dx
Now
R,
(t)
R, (-
t)
R^
/3)
{(A,t'
~ 2R, {tr),
But
l{(A,t;'
(t)
is
rational,
dt
R,
{t)
t-,
R, (-
and
R, (f ) + tR,
t)
2tR, {t%
so
(t^).
+ B,)(A.J' + Bo)\-^tR,{t")dt
can be evaluated in terms of elementary functions by taking

variable*; so that,
integrating
(tv,
x)
the following types
we put
if
we get
B,)(A,t'+B,)]i'
where R^ and R^ are rational functions of

Ro
where R^
(t),
R., (t)
~f
-'
515
t"
new
as a
Ri{t') into partial fractions, the problem of
dx has been reduced
to the integration of integrals of
^f^ {{A,P +
[(1
+ m^y
in the former of these
B,) {A,t'
{{A,t'
B,)] - ^dt,
B,) {A,t'
an integer,
is
B,)]
in the latter
- ^dt
is
a positive integer
andi\^^0.
By
differentiating expressions of the form
t?^-' [{A,t^
it
is
easy to
B,)
{A.J-^
obtain
i?,)]*,
(1
+ m^y-^ [{A,i:' +
B,) {A,t'
formulae by means of which
reduction
5,);i,
the
above
integrals can be expressed in terms of one of the three canonical forms
+ B,){A,P + B.;)]-idt,
(i)
^[{A,t^
(ii)
1^^ [(.4,^^
(iii)
[(1
+ B,) {A,t- + B,)] -Ut,
+ m^)-'
{(A,t'
+ B,) {A,r- +
These integrals were called by Legendref

second and
The
5,)} -idt.
elliptic integrals
of the
first,
thii'd kinds, respectively.
kind presents no difficulty, as it can be

once by a substitution based on the integral formulae of
22121, 22-122; thus, if A B A., B., are all positive and A^B,>A,B.,
elliptic integral of the first
integrated at
we
write
A,^t
* See, e.g.,
= Bi cs (u,
k).
[k'-'
= (A,B,)/(A,B,).]
Hardy, Integration of Functions of a single Variable (Camb. Math. Tracts, No.
t Exercices de Calcul Integral,
i.
(Paris, 1811), p. 19.

.
332
2).
516
Example
all
1.
Verify that, in the case of real
the following scheme gives

and determine the appropriate
integi'als,
possible essentially diflerent arrangements of sign,
substitutions necessary to evaluate the corresponding integrals.
^1
[cHAP. XXII
2273]
Example
that, if
where
Deduce from example
= (V3-f)
The
22"73.
(2^3)*,
/32
= (v/3 + f) (2^3)*,
for en m,
We
B,) (A,t^
The function* E(u).
B,)l
- idt,
the
kind which has the same expression under the
first
are thus led to one of the twelve integrals
jsnûdu,
a(aH^^)"^-
the same elliptic function substitution as in the case of that
elliptic integral of
radical.
is

jf' {(A,t'
we employ
and the modulus
of the second kind.
elliptic integral
To reduce an
By
combined with the integral formula
5,
is positive,
g-i
a2
6.
517
22'72 example
functions of u and
icn^
these are
3,
Jdn-udu
it is
udu,
...
jnd'ûdu.
expressible in terms of u, elliptic
all
convenient to regard
E(ii)
fu
dn- udu
Jo
as the fundamental elliptic integral of the second kind, in terms of which all
others can be expressed
E {u, k)
We
in place oi
when
du
Further, since dnû
2mK + (2n + l)iK,

is
of dn
u.
is
E{0) =
0.
an even function with double poles at the points
now be shewn
that E{^()
may
it is
easy to see that
be expressed in terms of Theta-
the most convenient type to employ
m {%U\
is
the function
(u).
'
a doubly-periodic function of u with double poles at the zeros of
at the poles of
*
I.
= dn-u,
the residue at each pole being zero,
^''''''^''
it is
write
an odd one-valuedf function of u with simple poles at the poles
It will
functions
i.e.
we
observe that
dE(it)
E(u)
the modulus has to be emphasized,
E{u).
dn u, and
so, if
dn-u
- A^-
(u),
be a suitably chosen constant,
du
-^
'
((z<.)
This notation was introduced by Jacobi, Journal fiir Math. iv. (1829), p. 373 [Ges. Werke,
In the Fundamenta Nova, he wrote E (am u) where we write E (u).
(1881), p. 299J.
+ Since the residues of dn^ m are zero, the integral defining
chosen
( 6-1).
{u) is
independent
of the
path
518
is
simple pole in any
u,
with periods 2K, 2{K', with only a single
It is therefore
cell.
the principal part of dn- u at iK'
d 10'(u))
'
-y-
<
^,
{n iK
IS
is
&' (u)IS (u) at this pole
of
du [yd{u))
_.,
)
a constant
this constant is usually
To determine the constant A, we
written in the form JE/K.
residue
[CHAP. XXII
-.
Hence
yi
(u iK')~^, by
unity,
is
= 1,
and the
the principal
part
of
so
= 0, we
Integrating and observing that H' (0)
so
observe that
22"341
get
E (w) = ' )/0 (w) + tiEjK.

we have E (K) = E hence
(;f
Since ' {K)
E=(
It
first
is
= 0,
dn-
usual
ucZw=r (l-k'sm"-0dcf> = l7rF(-l,
(cf.
22"3) to call
and second kinds.
are given
1.
Shew
that
1;
kA
K and E the complete elliptic integrals of the
Tables of them qua functions of the modular angle
by Legendre, Fonctions
Example
~;
Elliptiques,
ii.
E {u + 2nK)=Ê{ti) + 2nE,
where n
is
any
integer.
Example
2.
By expressing e{u) in terms of the function 9^(îrulK), and expanding
about the point u = iK', shew that
^=^(2-P-V7W5i')}ir.
The Zeta-function
22-731.
The
function
Z (w).
E (u) is not periodic
with these periods,
in either
2K
or in 2iK', but, associated
has additive constants 2E, {2iK'E
7ri]/K
it
is
convenient to have a function of the same general type as E{u) which
is
singly-periodic,
it
and such a function
Z (u) =
is
' (m)/ (u)
from this definition, we have*
Z
22*732.
(u)
= E (ti) - uE/K,
(m) = (0) exp I j" Z
The addition-formulae for
E {u)
{t)
dt\
and Z{u).
Consider the expression
W(v)
e'(u + v) '(m)
-
- /^ +
/ S{v)
(m)
@(u + v) T^TT^x
'
/-v
,
f^'
,
sn u sn V sn {u
v)
* The integral in the expression for

(u) is not one-valued as Z (f) has residue 1 at its poles;
bat the difference of the integrals taken along any two paths with the same end points is 2/i7r/
where n is the number of poles enclosed, and the exponential of the integral is therefore one-
valued, as
it
should be, since 0(m)
is
one-valued.
22 731-22-734]
519
qua function of m. It is doubly-periodic* (periods 2if and 2iK') with simple

poles congruent to iK' and to iK' v the residue of the first two terms at
iK' is 1, and the residue of sn u sn v sn (u + v) is k~^ sn v sn (iK' + v) = k~^.
;
Hence the function is doubly-periodic and has no poles at points

By
to iK' or (similarly) at points congruent to iK' v.
Liouville's theorem, it is therefore a constant, and, putting u = 0, we see
congruent
that the constant
is zero.
Hence we have the addition-formulae
Z (w)
-f-
Since Z{u) and
[Note.
(v)
Z(u+v) = k^ sn M sn V sn (?< +
v),
E (v) Eiu 4- v) = k^ sn u sn v sn{u + v).
E(u) +
E (u)
are not doubly- periodic,
algebraic relation can exist connecting
them with sn
u,
is
it
possible to prove that no
en u and dn
u, so
these are not
addition-theoreras in the strict sense t.]
22'733.
From
imaginary transformation\ of7i{u).
Jacobi's
21'51
evident that there must be a transformation of
it is fairly
Jacobi's type for the function
^2 {ix t)
To obtain
{ii).
= ( ir)^ exp ( iTX^/ir)
we
it,
translate the formula
^4 (ixr
r)
into Jacobi's earlier notation,
H (iu + K,
k)
when
it
becomes
= (- ir)i exp (_^^
(^, k'),
and hence
en
k)
(in,
= (- ^V)* exp
Taking the logarithmic
Tru-"
^4 (0
t)
(^^^ j ^^^
I
(u,
^^^
we
ditferential of each side,
k')
get,
on making use of
22-4,
Z {iu,
22 734.
It is
k)
= i dn
Jacobi's
(u, k') sc {u, k')
iTj {u,
7riu/(2KK').
k')
imaginary transformation of E(u).
convenient to obtain the transformation of
integral definition
E (u)
directly from the
we have
E (in,
k)
"dn-
{t,
k)
dt=\
Jo
=i
dn-
(it',
k) idt'
Jo
dc-
(t',
k') dt',
on writing
2iK'
A theorem
theorem
is
it'
and making use of
22 "4.
first two terms cancel.

due to Weierstrass states that an analytic function,/ (2), possessing an addition-
a jêriod since the additive constants for the
in the strict sense
must be
either
an algebraic function of z,
(ii)
an algebraic function of exp (Trizjw),
or
(iiij
an algebraic function of ^ (2 wi W2)
or
where w, wj, W2 are suitably chosen constants. See Forsyth, Theory of Functions
J Fundamenta Nova, p. 161.
(i)
(1918), Ch. xiii.
520
Hence, from
22*72 example
E {hi,
k)
E {iu,
and so
This
=i
\u
+ dn {u,
E'
k') sc (, k')
= E(K',
dn=
{u, k') sc {u, k')
{t',
iE {u,
k') dt'
k').
E'
to denote the
same function of
k'
as
E is
of
k,
k'X so that
= 2i(K'
E{2{K',k)
-E').
Legendres relation*.
22'735.
From
the transformation stated.
is
It is convenient to write
i.e.
we have
3,
= iu + i dn
k)
[CHAP. XXII
the transformations of
E{u) and Z{u)
to derive a remarkable relation connecting the
just obtained,
it is
possible
two kinds of complete
elliptic
namely
integrals,
EK' + E'K-EK' = \ir.

For we have, by the transformations of 22"733, 22*734,
E {iu,
k)
- Z {iu,
and on making use
Z {u,
k), this
k)
= iu - i
of the connexion
equation
we may take m
;
it is
1.
k')\
+ '7riu/{2KK'),
between the functions
=|^
0,
[uE'jK']
E {u,
k)
and
+ 'Triu/{2KK').
the result stated follows at once ft-om this
the analogue of the relation
the Weierstrassian theory
Example
- Z {u,
k')
gives
iuE/K = iu Since
[E{u,
Shew
rj^coo
Vzfî = 9
tJ"*
which arose in
2041 1).
that
E{u + K)-F{ii) = E-k-mucdu.

Example
2.
Shew
that
E(2u + 2iK')^ E (2u) + 2i {K' Example
3.
E
Example
4.
E').
{ti
sn^ u + iK') sn (2m + 2iK')

+ iK') = Ê{2u + 2iK') +
= E (u) -fen M ds M + i (A" - ").
Shew
that
E {u + K+ iK') = E (u) - sn u dc u + E+i {E' - E').

Example
5.
Obtain the expansions, vaHd when
/7 r'NO
tt;
,'T-,
'^
n=i
iw" COS
2ii.r
\-q^"
I{x) l<^7r/(T),
.^
-,
n=\
Q'^s\Vi2nx
1-?''"
(Jacobi.)
* Exercices de
Messenger,
Calcul Integral,
iv. (1874),
pp. 95-96,
i.
(1811),
p.
01.
For a geometrical proof see Glaisher,
22-735-22737]
521
Properties of the complete elliptic integrals, regarded as functions
22'736.
of the modulus.
If,
B=\
in the formulae
(1
k^ sin^ (f)y
Jo
sign of integration
L
dE
dk
fh^
I
dk
.
.
k sm-
,
k"
sm^
..
_i
.._!
(1
(f)
..
-d(J3
(f))
^'
under the
EE-K
j-
Drmula for
Treating the formula
dJ^
differentiate
we have
4"2),
),
( 4*2
we
d<f>,
sin-
K in the same manner, we have

ha'
<^ ( 1
^'
sm"^
=k
^d(f)
(j>)
sd^ u di
Jo
J a
jLj/^dn'u*,.by
22-72 example 3
so that
dk
If
we
write k-
= c,
k'-=
E-K
dc
Shew
1.
^dK
E-Kc'
dc
cc
that
^dE^^ K'~E'
dc
Example
kk'-
these results assume the forms
c',
^dE
Example
k!''-u
^=
cK'-E'
dc
'
c'
cc'
Shew, by difterentiation with regard to
2.
that
c,
EK' + E'K KK'
is
constant.
Example
Shew
3.
that
K and K' are solutions of

d^
|m'^|^|=^..,
dk
and that
E and
E' K' are sohitions of

^'"
~M-
The values of
22'737.
From
the complete elliptic integrals for
the integral definitions of
in powers of
k,
K = lim E^l-TT,
lim E'
In like manner,
it is
easy to see, by expanding
not
possible
to
cos
(K - E)lk' = ^ tt.
lim
(f)d(j)
1.
k-^-o
is
E and
small values of k.
that
lim
It
(Legendre.
\^ 'M^'^^^~ ^'
determine lim/i' in the same way because

fcÔ
(1
^'-sin^(^)"
is
discontinuous
example 21 of Chapter xiv
(p.
at
</>
0,
299) that, when
k
|
argA-
lim{i^'-log(4/^)l=0.
but
< tt,
it
follows
from
522
This result
q-*-0; or
By
1
is
may
it
also deducible from the formulae 2iK' = -!TTî^, k^^.^jBi^, by making

be proved for real values of k by the following elementary method
:
A" =
22-32.
[CHAP. XXII
- k-)
{f-
' ^
{l-t')'^dt; now, when
and 1-/-; and, when s,'k<t<\^
{fi-k^)/t^ lies
between
< < Jk,
{l-t-)
and
~k.
lies
between
Therefore
A'' lies
between
Jk
and
i^/A-
(^2_^.2)-ij;+
(i-/.)-*]/"^
t--^{l-f-^)-hdt\;
and therefore
=(1 -
ijlog
6k)
= (1 - Ok) - h [2 log {1 + v'(l - k)} - log

where 0^^^^
lira [2 (1
- Ok) ' i log {1 + v/(l - k)} - log 4] = 0,

\\m
Deduce Legendre's
The
22"74.
f(l
known
+ Nt')-'
functions,
sponding integrals of the

integral
is
or,
- du
yS,
v are constants
^
n (u, a) =
,
To express
k-
by making k-^0.
we make the substitution made

and second kinds ( 22*72,
in the corre22"73).
first
+ (p
ait
if
i^
0,
first
av)
The
or
may be
[^
Jo
the integral can be
for other values
= k^ sn- a,
and then
it
fundamental integral of the third kind
k-snacn a dn a sn'^ u
du.
j~.
1 k^ sn^ a sn^ u
,
this in terms of Theta-functions,
we observe
that the inte-
written in the form
sn u sn a {sn (u
a)
-1-
sn (u
)}
Legendre, Exercices de Calcul Integral,
pp. 14-18, 74, 75; Jacobi,
not Legendre's.
^^
du,
and second kinds
we determine the parameter a by the equation
grand
2,
+ B,) {A,t'+B.^] - ^ dt
{{A,P
evidentl}' permissible to take as the
22-736 example
of the third kind*.
expressed in terais of integrals of the
thereby reduced to
I
is
= 0,
/<
in terms of
from
relation
elliptic integral
To evaluate an
of V
(4/){,-)}
Example.
where
- {I - 6k) -^ log k = 0,
{I
lim {A' - log
and therefore
is
/],
1.
Xow
which
1-- /(i^l.^}
log
Fundamenta Nova
i.
(^ ("
-^
("
a)
+ 2Z (a)},
(1811), p. 17; Fonctions EUiptiques,
(1829), pp. 137-172;
we employ
i.
(1825),
Jacobi's notation,
by the addition-theorem
(i()
523
2274, 22741]
= B'(m)/^ (u), we
the Zeta-fimction
for
making use
of the formula
at once get
with
a result which shews that U(u, a) is a many- valued function of u
logarithmic singularities at the zeros of @ (u a).
Example
Obtain the addition-formula*
1.
n{u, a)+n{v, a)-niu+v,
e (u+v + a) e tt-g) e (v -a)

= iloge(+.,-a)e(-ha)e(^^+^)
(
j^
Psnasn -^sn-?;sn (tc + v a)

(w + v + a)'
~2 "l + Psn asnwsnvsn
(Legendre.)
(Take
a-
tf = m
-h r
i(
a in Jacobi's fundamental formula
[4]
-Example
Shew
2.
+ [l] = [4]'-r[l]'.)
that
n {u,
a)
-n
m)
(a,
= uZ (a) - aZ (tt).
(Legendre and Jacobi.)
[This
is
known
Example
n (iL
a)
Shew
3.
argument and parameter.]
as the formula for interchange of
+ n hi,
h)
that
n (u,
a + b) = i log
lPsnaHnbsnuHn{a + b-u)
j^
_,
-.
i
,
,,\
+ uk^ sn a sn 6 sn (a 4- 6).
(Jacobi.)
[This
is
known
Example
as the formula for addition of parameters.]
Shew that
4.
IT
Example
(m
+ v,
22 741.
ia
+ /i,
k)
= Il{u,a + K\
(Jacobi.)
k').
Shew that
5.
a-irb)
+n
= - F sn a sn 6
and obtain
{iio,
(u-v, a-b)-2n{u, a)-2U{v,

.
{(u
+ v) sn (a -f b) - (w -
^y)
sn (
b)
,
- 6)}
,
-H
i log
l-Fsn2(2<-a)sn2(i;-6)
i
+ ^2 .^^2
( ;,
^ ^,) s^^ (^ + 6)
special forms of this result by putting v or h equal to zero.
from the expression for
(Jacobi.)
the elliptic integral of the third kind.
dynamical application of
It is evident
'
n (m,
are real, the average rate of increase of
a) in
(m, a)
terms of Theta- functions that if u, a, k

is Z (a), since 9 {iia) is
as u increases
periodic with resjject to the real period 2K.
This result determines the mean precession about the invariable line in the motion of
its centre of gravity under forces whose resultant passes through
gravity.
centre
of
It
is evident that, for purposes of computation, a result of this nature
its
is preferable to the corresponding result in terms of Sigma-functions and Weierstrassian
a rigid body relative to
Zeta-functions, for the reasons that the Theta-functious have a specially simple behaviour
with respect to their real period

matics
and that
by which the Sigma-function

*
No
the period
the ^-series are

is
much
which is of importance in Applied Mathebetter adapted for computation than the product
most simply
defined.
fewer than 96 forms have been obtained for the expression on the right.
Messenger,
x. (1881), p.
124.
See Glaisher,
524
The lemniscate functions.
22'8.
The
[CHAP. XXII
integral
(1
^)
problem of rectifying the arc of
dt occurs in the
Jo
if the integral be denoted by ^, we
between ^ and x by writingi* x sin lemn <^.
the lemniscate*;
relation
In like manner,
express the
shall
if
-^
.'
we write
and we have the
<^
= cos lemn
cos
lemn
^j,
relation
lemn
sin
^^
tn-
These lemniscate functions, which were the
(^
first functions:]:
defined by the
inversion of an integral, can easily be expressed in terms of elliptic functions
with modulus l/VS;
from the formula
for,
22-122 example)
sd u
u=\
[{l-k'^f){l+l:^f)]-Uy,
easy to see (on writing y
it is
\/2) that
= 2 " ^ sd (<^ \/2, l/V^)

cos lemn ^ = en
V2, 1/^2).
sin
similarly,
lemn ^
(</>
^^
Further,
is
the smallest positive value of
cn(<^V2, 1/V2)
so that
the
for
which
= 0,
OT=V2ô,
suffix
attached to the complete elliptic integral denoting that
formed with the particular modulus

This result renders
it
it
is
l/\/2.
possible to express
Kq
in terms of Gamma-functions,
thus
K,=
2^[^ {\-P)-^-dt='2---\^ u -^{l-u)-^-di

Jo
a result
obtained by Legendre.
first
Since k
*
//
The equation
{%)' = ar~*
Jo
^''^'"
when k =
of the
l/\/2, it follows that
lemniscate being
'^' ^^'^"^"^^
r-
= a'^ cos
{tj= 1 + (1-)'
7^
26,
it
=
is
7iV.
and so
easj'
to
ô
e^".
derive the equation
t Gauss wrote si and cl for siu lemn and cos lemn, Werke, iii. (1876), p. 493.
t Gauss, Werke, iii. (187(5), p. 40i. The idea of investigating the functions occurred
on January 8, 1797.
Exercices de Calcul Integral,
while 07 = 2-62205756....
i.
(Paris, 1811), p. 209.
The value
of
Kq
is
to
Gauss
1-85407468...
22-8, 22*81]
525
Example
\.
Express A'o in terms of Gamma-fuuctions by usiug Kummer's formula
Chapter xiv, example 12, p. 298).
(see
Example
2.
By
writing
shew that
2^ ô
= {\ -m^)^
=/
and deduce that
Example
22-736
Jô
= Stt^ {r
j)}
ic*)
i du,
~ 2,
from example
relation ( 22-735)
combined with
2.
Shew
that
.
sm
The values of
22-81.
formula
'^du+T u^{l-
?**)
Deduce Legendre's
3.
4.
2ô -
example
Example
j^ the
Iemn2d) = ,
cos lemn^ (b
--t-
H-coslemn''9
K and K' for special values of

K
k.
been seen that, when k=ljJ2,

can be evaluated in terms of Gamma-functions,
and K=E' this is a special case of a general theorem* that, whenever
It has
_ a + hjn
c + djn''
is
a root of an algebraic equation with integral
E'
where
b,
c,
d,
are integers, k
coefficients.
This theorem is based on the theory of the transformation of elliptic functions and is
beyond the scope of this book but there are three distinct cases in which k, 7i, E' all
have fairly simj^le values, namely
;
= V2-l,
E' = EJ2,
(II)
k = Hm^7r,
E' = E^%
(III)
k=tsin^7r,
E'^2E.
(I)
Of these we
(I)
>(-
shall give a brief investigation
The quarter-periods with the modulus
t.
^/2
1.
Landen's transformation gives a relation between elliptic functions with any modulus k
and those with modulus ki = {l k')/{l+k'); and the quarter-periods A, A' associated with
the modulus k^ satisfy the relation A'/a = 2E'/E.
If we choose k so that ky = k', then A = /i'' and k{ = k
A7A=2A'7A'gives A'2=2a2.
A' associated with the
Therefore the quarter-periods A,
equation
î
= (l î)/(l-|-X-i)
so that
are such that
A'=Av/2;
i.e.
i\.'
= E;
modulus
\i ki
and the relation
k-^
= ^l2-\,
given by the
then
A'
= Av/2
(since A, A' obviously are both positive).

(II)
The
The quarter-periods associated

case of k = s,\n^-rrr
ivith the
modidus sin
y^tt.
was discussed by Legendre|; he obtained the remarkable
result that, with this value of
k,
E' = Ey/3.
* Abel, Journal fiir
Math. in. p. 184 [Oeuvres, i. (1881), p. 377].

For some similar formulae of a less simple nature, see Kronecker, Berliner Sitzungsherichte,
1857, 1862.
X Exercices de Calcul Integral,

pp. 59, 60.
i.
(1811),
pp. 59, 210;
Fonctions Elliptiques,
i.
(1825),
[CHAP. XXII
526
This result follows from the relation between detinite
To obtain
this relation, consider
part of the real axis (indented at
integi-als
l{l-z^)~^dz taken round the contour formed by the

arc of radius R'^) joining the points
s=l by an
and
and the arc of radius B joining the points R and Re^""^ as

the line joining ê*"' to
indentation,
the integral round the arc tends to zero, as does the integral round the
Râ:
and
so,
on writing
.v
and
respectively for z on the two straight lines.
ê*'"*
Writing
we have
so,
j\l+.f^)--^ds=j\l-x^)-^dxÎ
j\x-^-l)-^dx = L,
[\l-x^)-^-dx = Ii,
/j
+ iV. = 4 (1 + is/S) h
equating real and imaginary parts,
A = 5-^3
is
the relation stated*.
Now, by
example
22-72
where the modulus
is
6,
= 4(a2 + ^2)-*',
/2
a {d'
+ ^^) ~ ^
/^
We
>{:2
,
+ /3 = 4(a2 + /32)-4^',
and
a2=2v3-3,
so that
= 5-^3X^3,
h-V I^- Lis''^ = \h + h-i=^^
and therefore
which
-^2
= 2V3 + 3,
/32
= i(2-V3) = sin2Js7r.
therefore have
3-i.
2/1
= 3-2. 2^' = /2 = 3^/i

= 3-^|V"s(i-^)-ic^^=i,r^r(J)/r(|),
the modulus k
when
(III)
If,
now
in
is
sin jV^-
The quarter-periods
iiith the
Landen's transformation
modtdus
( 22-42),
tan^-i-Tr.
we take
l::=i;^'2,
we have A'/A=2K'/K=2;
this value of k gives

7
and the corresponding quarter-periods
Example
1.
V^-1
\, A' are
Discuss the quarter-i)eriods
"1
|(1
+2
and
^) A'o
when k has the
(1
+2
^)
Kq.
values (2;,y2-2)^, sin f^7r,
and 2^(^2-1).
*
Another method of obtaining the relation
functions by writing
defined.
t^
is
to express Ij
t~^, (-' - 1)* respectively for
I^,
in terms of
x in the integrals by which
Gamma-
Ii, I2, I3 are
22'82]
Example
Shew
2.
527
that
?l
)i
=l
(Glaisher, Messenger, v.)
Example
Express the coordinates of any point on the curve y^=ofi
3.
^~
l+cn?<
considering
Example
4.
y~^dx = 3~*
Shew
that,
_2.3^snMdnM
'^~
(l + cnM)2
'
where the moduhis of the eUiptic functions
By
is
form
=^~ iy.
-=-
K in terms of Gamma-functions when
when y'^=x^l,
r y-Hx-lYdx=[-^y-^ {l-x-'^flJi
in the
'
and shew that
sin jV"",
evahiate
die,
3^ (1 en u)
+'^
T {x~^y-^-x-h/-^)dx;
'^y
'^
and thence, by using example 3 and expressing the
terms of
last integral in
Gamma-
functions by the substitution x = t~^, obtain the formula of Legendre {Calcul Integral,
p. 60)
connecting the
Example
By
5.
first
and second complete
elliptic integrals
with modulus sin
expressing the coordinates of any point on the curve
jJjtt:
Y'^=\-X^
in
the form
32(l-cnv)
1+cni?
in which the
modulus of the
_ 2. 3*sn vdn??
(1-l-cn^')^
'
elliptic functions is sin y^tt,
'
and evaluating
{/>/} Y-^\-XfdX
in
terms of Gamma-functions, obtain Legendre's result that*, when
22*82.
geometrical illustration of
tlie
functions sn u, en
ii,
dn
k^sm^w,
u.
geometrical representation of Jacobian elliptic functions with
k=ljJ2
is
afforded
by
the arc of the lemniscate, as has been seen in 22"8 to represent the Jacobian functions
with any modulus k {0
<\), we may make use of a curve described on a sphere, known
;
<k
as Seifferfs spherical spiral^.
Take a sphere of radius unity with centre

coordinates of any point on
is
it lê
(p,
^,
z),
at the origin,
and
let
the cylindrical polar
so that the arc of a curve traced on the sphere
given by the formula \
{dsr^=pHdcf>y+{i-p^)-^{dpy.
* It is
interesting
EK' + E'K- KK'
is
to
observe that,
when Legendre had proved by
differentiation
that
constant, he used the results of examples 4 and 5 to determine the constant,
before using the methods of 22-8 example 3 and of 22-737.

t Seiffert, Ueher eine neue geometrische Eiiifilhrung in die Theorie der elliptisclien Funktiunen
(Charlottenburg, 1896).
X This
is
an obvious transformation of the formula
are connected by the relation
p'^
z'-^
l.
{ds)"={dp)'^
+ p'^
(d(p)~
+ {dz)^ when
and
528
[CHAP. XXII
by the equation
Seiffert's spiral is defined
(f)
= ks,
where s is the arc measured from the pole of the sphere (i.e. the point where the axis of
meets the sphere) and k is a positive constant, less than unity*.
For this curve we have
and
so, since s
and p vanish together,

p = sn
The
arc
(s,
k).
any point on the curve expressed
cylindrical polar coordinates of
in
terms of the
measured from the pole are therefore

{p,
z)
cf),
= (sn
s, /ts,
en
s)
be the cosine of the angle at which the curve cuts the meridian.
be the arc of the curve from the pole to the equator, then
Hence it may be seen that, if
sn s and cu s have period AK, while dn has period 2K.
and dn
s is
easily seen to
6*
REFERENCES.
A. M. Legendre, Traite des Fonctions Elliptiques (Paris, 1825-1828).
C. G. J. Jacobi, F\hiidamenta
Nova Theoriae Functionv.m Ellipticanim (Konigsberg,
1829).
J.
Tannery
Molk, Fonctions
et J.
Elliptiques (Paris, 1893-1902).
A. Cayley, Elliptic Functions (London, 1895).

P. F.
Verhulst, Traite eUmentaire des
fonctions elliptiques (Brussels, 1841).
A. Enneper, Elliptische Funktionen, Zweite Auflage von F. Miiller (Halle, 1890).
11
1.
Shew that one of the values of
dnM + cnwVl+cn?t /
is
/dn
\\dinu k'su.u)
w-cn wV'j
\cnn J )
l-sn%
+ sni* \^'\
\6.nu + k' snu)
\-
(Math. Trip. 1904.)
2 (1 +k').
2.
li
x + iy = sn"^ ( u + iv) and x iy sn^

{(^-l)2+2/-}-
(% iv),
shew that
= (^-+y^)^dn22( + cn2w.
(Math. Trip. 1911.)
3.
Shew
that
{lcn( + ^.)}{lcn(-i;)} = /'-'?/=^^^^.

1 '= sn^ w sn^
4.
Shew
that
1
+cn
(w
^
+ v)
en {u-v) = ,
cn^
-p,
cn^ V
A:^sn-M sn^ V
(Jacobi.)
* If fc>l, the
curve
is
imaginary.
THE JACOB IAN ELLIPTIC FUNCTIONS

5.
p
Express
^
6.
Shew that
+ cn(ti + v)cn(u-v)
\
v)
\+an{u
+ v)axi(u
as a function orc sn-'w + snî

o
529
v.
(Math. Trip. 1909.

sn {u v) dn {u + v) =
sn M dn
-M
en V sn v dn
^2
sn"^
su"^
en u
i?
(Jacobi.)
Shew
7.
that
{\-{\+k')s,r).us,n{u-\-E)){\-{l~ k') sn m sn {u + K)] = {su
(.
+ ^) - sn uf.
(Math. Trip. 1914.)
Shew that
8.
(1 -/(:')
sn
(i<
+ |iA''') = ^
(1
+ ^) sn M +
1
en w dni<
+ ^ sn^ M
Shew that
9.
cos
sm {am
i^,,
{am
{u
,,
2sn?icnMdnv
.,
cn^ V sn^ V dn^ m
+ v) + -^m{u-v)}^j-j^^^^^^,~,
+ v)-^m (-^)H ^ ^ ^2 32,,s^2
(Jacobi.)
Shew that
10.
dn {u+v) dn
{u
ds2?ids- v + F^'2
v) =
ns^
?i
ns^ V
^-2
'
and hence express

(u
+ v) 2
iP
{u vj
e-{
as a rational function of ip{u) and ^{v).
(Trinity, 1903.)
From the formulae for cn(2^ ) and

+cn 2u and l + dnSw, shew that
11.
for
diW
{2K-u) combined with
(l-cn|ir)(H-dn^) = l.
12.
With
notation similar to that of 22*2,

Ci d-i
c<idx
en
(i
and deduce
that, if Mi
(Trinity, 1906.)
shew that
+ u-i) dn
Sj-sa
the formulae
Mj
+ ^2)
sn(i4-2)
f-% + ^'4 = 2A', then
(/o
(Ci 0?9
- C2 0?i)
(C3 c/4
- C4 (^3) = ^2
(sj
_ s.,)
(S3
- S4).
(Trinity, 1906.)
13.
Shew
that, if u
1
+ + ?r = 0, then
i^
- dn^ XL dn^ v - dn- + 2 dn
dn
dn u'=k* sn^
?<
sn^ v sn^
m;.
(Math. Trip. 1907.)

14.
By
Liouville's
dn?< dn {u
theorem or otherwise, shew that
+ w)-dn
<;dn {v
+ w) = k^ {sn
yen u sn
(v
+ w) en
(u
+ w)
snMcn vsn{u + xo) en
(y
+ ;)}.
(Math. Trip. 1910.)

15.
Shew
that
2 en Uo en u^ sn
(?<2
- %) dn
Mj
+ sn (^2 - M3) sn (W3 - ?<i)
the summation applying to the suffices
W. M. A.
1, 2, 3.
sn
(i<i
- Mo) dn
?ii
dn
Mo
dn
% = 0,
(Math. Trip. 1894.)
34
'
530
16.
[CHAP. XXII
Obtain the formulae

cii 3?< = BjD,
dn 'Mi = CLD,
^ = 3s - 4 1 + s\+ Qk'^ s^ - k* s%
B = c{l-4:S^ + 6k^s*-4kU'^ + k^s},
^4 _ 4X.2 ^c + ia gsx^
C=d{l-4k^s^ +
D = l- 6^-2 + 4:k- 1 + 0 - Zk^ s%
= cni(, c/=dn;<.
s = sn ,
sn 3 = AID,
where
/f-^)
6/5-2
S-*
and
/?;2)
c'
17.
where
Shew
that
- dn 3u _
/1 -
+ dn
\l
'iu
dn
?t\
/I
+ dn tcj
\l
?<
?<
Ui, ao, 3, a^ are constants to be
18.
+ a^ dn u + 02 dn^ u + c'3 dn^ + a^ dn

- aj dn + ^2 dn^ 14 - ag dn^ % + a^ dn* uj
determined.
,
If
^
P{u) + F{tt+2iK')
F{u)-F{u + 2iK')
shew that
P (u)
Determine the poles and zeros of
(Trinity, 1912.)
+ dn 3i(\+ dn ?<
sn 2u en u
en 2m sn u
and the
term in the expansion of the
fii'st
function about each pole and zero.
(Math. Trip. 1908.)

19.
Shew
sn
that
{ui
+ U-2 + U3) = A/D,
en
+ U2 + ii^) = BjD,
{ui
dn
{ui
+ M2 + %) = C'/i),
where
A=SiSoSi {-
1 --?-2
+ 2F2V-('<'-'^ +
^-'')
252^532+ 2ytisi2s22 532)
+ 2 {s^c.c^dodî {l + 2k'^s.isi-B^s.^si)],
B =CiC2Cs {1
- k^2s.2^S3^+2k*S^^S2^S-/}
+ 2 {CiS2.?3O?20?3(- 1 +2Ps2^S32 + 2Fsi2-F25225g2))^

C = d,d2d3{l-P2s./ss'^ + 2F-s,^s.2H:i^}
+ F2 {diS^SsCoC-i - 1 +2/{-2s22s32 + 2Si2-F2S22532)},
D=l - 2P2S2'-^S32 + 4 (F + F) Si22^3- " 2k*Si^S2-S3~2Si- + B2s.*S3*,
(
and the summations

20.
refer to the suffices
(Glaisher, Messenger, xi.)
1, 2, 3.
Shew that
sn (ui
+ W2 +
^'-s)
= -^'/D',
en
{ui
+ U2 + its) = ^'/-^'j
dn
(u^
+ U2 + u.^) = G'jD',
J' = 2siC2C3C^2^3--5iS2S3(l+^^ ^^2si2+/{;4si2s22s32),
where
- ^* Si2s2232) di d^d^'Ss^s^Ci di
5' = Ci C2C3 (1
C" = 0?i C?2 0?3 1 F Si2 22 S32) - F Cj C2 C3 2^2 S3

Z)' = l - F 25-22 S32 + (F + ^*) Si2s22s32-FsjS2S32SiC2C3C^2'^3<?!
C^l
((/'ayley,
21.
By
applying Abel's method
;j72_|.y2_i^ 22
+ F_^.2_.i
Journal fur Matli.
plane Ix + my + nzl,
+ M2 + % + 4 = 0j
îtii the variable

'Mi
then
XLi.)
( 20*3 12) to the intersections of the twisted curve
Si
Ci
d^
52
C2
^2
53
C3
0?3
54
C4
0^4
shew
that,
if
=0.
OVjtain this result also from the equation

(2
- i)
(^3^4
- ^40^3) + (4 - s
which may be proved by the method of example
{cid2~C2di)
= 0,
12.
(Cay ley, Messenger, xiv.)
531
Shew that
22.
by expressing each
side in terms of
Si, 2, s^, s^
and deduce from example 21
that, if
+ U2 + U3 + Ui = 0,
S4 Ci o?2 + S3 C2C/1 + 52 63(^4 + 51 040^3 = 0,
S4 C2 cî + S3 Ci o?2 +
^3 + ^1 C3 c^4 = 0.
iii
then
''2 <?4
(Forsyth, Messenger, xiv.)
Deduce from
23.
Jacobi's fundamental Theta-function formulae that,

Ui
if
+ U.2 + M3 + M4 = 0,
k'- k^ k"'si So S3 S4 4- k^ Ci c, c^ c^ di d.2 d^ d^ = 0.
then
(Gudermann, Journal fur Math,
Deduce from
24.
Jacobi's fundamental Theta-function formulae that,
xviii.)
if
+ U.2 + lt3 + Ui = 0,
S3 S4) cîa?2 + <^3 0^4=^0,
k'^ {S1S2- 838^) +did2C3Ci CiC2d3di = 0,
.SiS2C^3(^4 did2S^Si + C3C4 Ci C2 = 0.
Ul
F (si S2 C3 C4 - Ci
then
6'2
(H. J.
Smith, Proc. London Math. Soc.
S.
+ Uo + n3-)rUi = 0, shew that the cross-ratio of sn Ui, sn Uo,

cross-ratio of sn (u^ + K)., sn {u-i+K), sn
+ K\ sn {Ui + K).
25.
to the
If Ui
(1), x.)
sn M3, sn
?<4
is
equal
{11,3
(Math. Trip. 1905.)
Shew
26.
that
sn (w-
iin^{ii,+v)
sn
<;)
fiTi?{u-v)
cn2 (w -I- v)
en (m -f v) en ( v)
en- (w - v)
dn
dn^{u v)^
dn'^(-|-i^)
(ji-f-y)
(m -t-
2^)
dn
(?(
y)
Sk"' SiS2^ CiC^did^
(,
^'Si S2
(Math. Trip. 1913.)
Find
27.
are real and

If
28.
all
systems of values of u and
k'
= J {a ~
^
a)'-^,
< a < 1,
where
'
^" ^
and that
f for
which
iivi^{u
+ iv)
0<k-<\.
sn'-^fiT is
shew that
4a3
r-
(l-f-a2)(l
obtained by writing
is real when u and

(Math. Trip. 1901.)
a~^
+ 2a-a-^)'
for
a in this expression.
(Math. Trip. 1902.)
29.
If the values of en
z,
which are such that en 3z = a, are

9
3k*
Cj, C2, ... Cg,
shew that
Cr
+ k'*
c,.
= 0.
(Math. Trip. 1899.)
y,.
g-l-sn (M-l-y) _ 6-|-cn(^f-f y)
a + su{u
and
if
none of snv, cnu, dnu,
v)
1
+ cn{u v)
^-'^sn^wsn^
_c + dn{u + v)
c + dii{it v)'
vanishes,
shew that
ti
is
given by the
equation
X-2
(t'^d'
+ b''- c^) sn- u = k'-+k'^b'' - c\

(King's, 1900.)
34
:i
532
31.
Shew that
32.
Shew
[CHAP. XXII
(Math. Trip. 1912.)

that
{dn {2KxlTr)
Jl-2j2n-lgin-j; + j4n-2^
"^
l-sn(2^.r/7r)
sn (2A'j
/;'
(Math. Trip. 1904.)
Shew that
if
for small values of
?.
33.
k be so small that
k^
may be
sn M = sin u ^k-cosic.
Shew
34.
neglected, then
(t<
sin iccos
u),
(Trinity, 1904.)
that, if
/ {x)
< ^nl
then
(r),
= log cos .r
log en (2 A'.r/Tr)
4j" sin^
=iw{l
nx
+ (-?)"}
(Math. Trip. 1907.)
[Integrate the Fourier series for sn (2A''^/7r)dc (2A'r/7r).]
Shew
35.
that
cn^
71
dn'=
(Math. Trip. 1906.)

[Express the integrand in terms of functions of 2u.]
Shew
36.
that
/cnvdu
_.
snv-snT<~
where
^^
+ -|^ i7r)î(|^.r + 5y-|7r -iTrr)

Ml-^ - y) ^{hîl/ - i ^r)
5i (|a'
2Kx = Tru, 2Ky=irv.

Shew
37.
{iZ + hirr)
Ij^y+i"^)
î
'*
'
(Math. Trip. 1912.)
that
^
jo {l+cuu)dnû
(Math. Trip. 1903.)
Shew
38.
that
,
k/
,'+^
I
sn
v'r
/-/)/
?(a;
-i/
l+X-snasn/3
= io2;
,
Irvnr
/;sn asn
j3'
(St John's, 1914.)
By
39.
integrating
je^'^^dmicsudz
round a
Jo'
\
cos (iru/K) logsn
t(,dii
whose corners are i7r,

shew that, if </:- < 1, then
rectangle
îr + cci (where 2Kz = itii) and then integrating by
parts,
= ^ K timh. (îrir).
(Math. Trip. 1902.)
Shew that
40.
K and K' satisfy the equation

c(l-)^ + (l-2c)^-i = 0,
where
= ^2.
j^j-kJ
^ with argument
deduce that they satisfy Legendre's equation for functions of degree

1
2k'^.
533
Express the coordinates of any point on the curve x^+y^=l in the form
41.
^^ 2.3^snMdnM-(l-cn u)^
2.3isn%dntt + (l-cn w)2
_ 2^ cos ^3^ tt
2. 3^
the modulus of the elliptic functions being sin yV 't
further that the
- on w) {1 + tan jLtt
snwdn m + (1
en u)
-cnw)'^
and shew that
Jo
J X
Shew
(1
sum
of the parameters of three coUinear points on the cubic
is
period.
[See Richelot, Journal fiir Math. ix. (1832), pp. 407-408 and Cayley, Proc. Camb. Phil.
uniformising variable for the general cubic in the canonical
A
X^+ F^ + Z^ + 6rnXYZ=0
Soc. IV. (1883), pp. 106-109.
form
has been obtained by Bobek, Einleitung in die Theorie dex

Funktionen (Leipzig, 1884), p. 251. Dixon {Quarterly Journal, xxiv. (1890),
pp. 167-233) has developed the theory of elliptic functions by taking the equivalent curve
a^+y'^ ^cuicy=\ as fundamental, instead of the curve
elliptischen
y2=(l-.^2)(1_^.2^.2).]
42.
first
Express
{{2x-x-)
(4^-2
+ 9)} ~ "^dx
in terms of a complete elliptic integral of the
kind with a real modulus.
43.
(Math. Trip. 1911.)
u=l
If
{{t
+ \){t:^ + + \)]-^dt,
t
express x in terms of Jacobian elliptic functions of u with a real modulus.
(Math. Trip. 1899.)

44.
î=
If
P (1+^2 _ 2^4)- i^;^
express x in terms of by means of either Jacobian or Weierstrassian elliptic functions.
(Math. Trip. 1914.)

45.
Shew
that
2'^7r^
(Trinity, 1881.)
46.
When
a>x> ^>y,
reduce the integrals

j-"
\\{a-t){t-^){t-y)]-^^dt,
{(a-t){t-^)(t-y)}-idt
by the substitutions
x-y = {a- y) dn^ w,

respectively,
Deduce
where
that, ii
k'^
??
= {a ^)/{a y).
u + v = K, then
1-sn^ u sn^
By
x y = (p-y) nd-
+ k'snûiin^v = 0.
the substitution y = {a t) {t j3)l{t - y) applied to the above integral taken between

j3 and a, obtain the Gaussian form of Landen's transformation,
the limits
where
ai, bi are
{a{^ cos^
d + bi'^Hm-d)~^de=
{a- eos^ 6
sin^
(9)
the arithmetic and geometric means between a and
b.
(Gauss, Werke,
ill.
p.
b'^
dd,
352; Math. Trip. 1895.)
534
47.
Shew
that
sc
= - k'
?f
{C {u
A')
- C (u - K - 2iK') - C {2iK%
where the Zeta-functions are formed with periods

48.
where
49.
Shew that
c=lfi,
= 2K,
AiE'.
(Math. Trip. 1903.)
E k"^K sati.sfies the
equation
(^
ni^ k^ K' dk={n-\)
that
(71
50.
2a)i, 2co2
and obtain the primitive of this equation.
Shew
[cHAP. XXII
+ 2)
t'E'dk={n + l)
/(''
(Math. Trip. 1911.)

-
E' dk,
k'^K'dk.
(Trinity, 1906.)
{t{\-t){\-ct)}~^dt,
u^W""
^
If
o(c-l)^+(2c-l)^ + 4. =
shew that
^|^3^}
(Trinity, 1896.)
51.
SHew that the
primitive of
du
k _
u^
dk'^J'^T^'^
A{E-K) + A'E'
'^
""'AE+A'iE'-K'y
where J, A' are constants.

52.
(Math. Trip. 1906.)
Deduce from the addition-formula

Ui
then
is
(sn
?<!
sn
for
E {u)
that, if
+ U2 + U^ + Ui = 0,
^l<l
sn u^ sn Ui) sn (uy + %i^
unaltered by any permutation of suffices.

53.
Shew
54.
Shew that
(Math. Trip. 1910.)
that
(Math. Trip. 1913.)
i'^ u cd^ udu = 2K
U-^
[Write = A''+r.]
55.
By
+ P)
A'-
2 (1 -l-F) E).
(Math. Trip. 1904.)
considering the curves
+ Uo + Us + Ui = 0,
{(2
'
?/2 = ^ (i _^.)
(1 _^.2^,)^
y = l-\-mx +
n.v'^,
shew
that, if
then
E{%lx)+E{U2)
+ E{ll3) + E{Ui) = k\
.V^
+ 2CiC2C3C4-2SiS2S3S4-2r.
(Math. Trip. 1908.)
56.
By the method
of
example
E{ui) + E{u2) + E(u3) + E{tii) when Ui

l
+ k^s^s^Si
r=i
'
'^
"
k^SiS2S3Sidid2d3di
k'-'
Pkf^SiS2S3Si-did2d3di r=l
21,
obtain
seven expressions for
+ U2 + U3 + Ui = 0:
+ d^ d^dsdi
Ks d
'''''
r=i
k^cc^c^c,
- k^CiC2 C3Cidi d2d3di
- X-'2
'''''
,=1
u^
d)
d^d2d3di-lrk'^CiC2C3Ci r=l
+ CiC2C3Ci
+ A:2siS253S4
k'SiS2S3Si
ClC2C3C4
the following
,.=i
4
-^2;(gj52S3S4)-l
+ (CiC2C3C4)-l + X'^(a?lC/2C^3<:^4)~'}~^
ll{s,Crdr).
r=l
(Forsyth, Messenger, xv.)
'
535

Shew that
57.
when I{x) <7r/(r)
and, by differentiation, deduce that
f^^y ns*
")
= 6 cosec* x + i
(1
(^y -
+ F)
TT
-,
Shew
also that,
when /(^)
|
2Kx\ _
^j -
f/,
/2/r\2
TT
o2 COS 2A'
< |7r/(r),
fl-hf-2
,!o 1
79S
"]
cosec^
{2n
+ lf fjn^yi
-
2F
2P
\2k)
n.M
27rg"-^-2 sin (2ot
^ (1 - ?^-
+ 1) ^
1)
(Jacobi.)
58.
Shew
that, if
a be the semi-major axis of an
the perimeter of the ellipse
ellipse
whose eccentricity
is
sin
j-Vtt,
is
(Ramanujan, Quarterly Journal, XLV.)

59.
Deduce from examjjle 19 of Chapter xxi that
dnûdn3u
+
r
FTO cn3 2u = -k"^
l+k^snûsn3u
o
k'^ + k^ cn^ u en Su
,_
5
dn^ 2u = ^r71
î~
T-,
+ Fsn^wsnSM
(Trinity, 1882.)
60.
From
the formula sd
deduce that
k')
rq^'+i
^ f{n + i)7ru\_
(^rv^î.br^'î^^V1
where
and u
{i.u,
k) = i sd (n,
q
lies inside
= exp ttK'/K),
(
q^
- ^-^^g^^""'
(-)'*?i'^"'^
;.^
f{7i
+ i)nu^
K'
= exp irKjK'),
(
the parallelogram whose vertices are
iKK'.
By
integrating from u to
K\
from
to u
and again from
ti
to K', prove that
are
[A formula which may be derived from this by writing 2i = $ + iri, where | and
and equating imaginary parts on either side of the equation was obtained by Thomson
and Tait, Natural Philosopki/, ii. (1883), p. 249, but they failed to observe that their formula
was nothing but a consequence of Jacobi's imaginary transformation. The formula was
suggested to Thomson and Tait by the solution of a problem in the theory of Elasticity.]
rj
real,
CHAPTER
XXIII
ELLIPSOIDAL HARMONICS AND LAME'S EQUATION

The definition of ellipsoidal harmonics.
231.
been seen earlier in this work ( 18-4) that solutions of Laplace's

equation, which are analytic near the origin and which are appropriate for
the discussion of physical problems connected with a sphere, may be conIt has
veniently expressed as linear combinations of functions of the type

cos
r'^P (cos
where n and
md),
are positive integers (zero included).
When Pn (cos 6)
is
resolved into a product of factors which are linear in
when n
6 (multiplied by cos 6
cos^
r'^Pn'" (cos 6)
^),
we
odd),
is
see that,
if
cos 9 is replaced
by
then the zonal harmonic r"P (cos 6) is expressible as a product of factors

which are linear in a^, y^ and z^, the whole being multiplied by z when n is
zjr,
odd.
The
linear in
tesseral
a^, y^
and
harmonics are similarly resoluble into factors which are

z-
multiplied by one of the eight products
1,
cc,
y, z, yz, zx,
xy, xyz.
The
surfaces on which
any given zonal or tesseral harmonic vanishes are

has some constant value, so that they are
surfaces on which either 6 or
(f)
circular cones or planes, the coordinate planes being included in certain cases.
When we
deal with physical problems connected with ellipsoids, the
structure of spheres, cones and planes associated with polar coordinates
is
The property of spherical

replaced by a structure of confocal quadrics.
the construction of a set
explained
suggests
been
harmonics which has just
of harmonics which shall vanish on certain members of the confocal system.
Such harmonics are known
Lame*
as ellipsoidal harmonics; they were studied by
in the early part of the nineteenth century
coordinates.
The
expressions for ellipsoidal
coordinates were obtained
many
years later by
following account of their construction
The fundamental
ellipsoid is
and any confocal quadric
is
by means of confocal
harmonics in terms of Cartesian
W. D.
Niven-f-,
based on his researches.
taken to be
x^
ir
z^
a^
0^
is
* Journal de Math.
iv. (1839),
pp. 100-125, 126-163.
t Phil. Trans. 182 a (1892), pp. 231-278.
and the
ELLIPSOIDAL HARMONICS
23'1, 23-2]
where ^ is a constant. It
and it conduces to brevity
-^
The equation
will
be necessary to consider sets of such quadrics,
to write
of any
member
(h)
of the set
Qp =
The
analysis
is
537
made more
definite
is
then
0.
by taking the
a;-axis as
the longest axis
of the fundamental ellipsoid and the ^-axis as the shortest, so that

23'2.
The
foiir species
a>b>
c.
of ellipsoidal hat^monics.
consideration of the expressions for
spherical harmonics in factors
indicates that there are four possible species of ellipsoidal harmonics to be

investigated.
These are included in the scheme

X,
1,
y,
zx,
xyz r%^%.^...%,
538
23'21.
of ellipsoidal harmonics of the
Tlie construction
As a simple
case let us
9-
d^
x^
so @i
isa harmonic if
(0
e,
6^ is
+ ) {6 + cO +
y-
((9
b'-
2
0,
c'
'
e,
a root of the quadratic equation
d")
(6
a')
c^
This quadratic has one root between
and
z"
'+T,^Ver
â^d,'^
2
ci"
6^
of the
the effect of applying Laplace's operator, namely

9-
species
first
Such a hal-monic must be simply
9^^'^9p'^9?
and
first species.
consider the harmonics of the
first
which are of the second degree.

form @i.
Now
[CHAP. XXTII
(f)
a') (6
+ b') =
0.
and b~ and another between

by giving 6^ the value of
Its roots are therefore unequal, and,
a-.
each root in turn,
we obtain two*
ellipsoidal
harmonics of the
first
species
of the second degree.
Next consider the general product @i02

by
...
@^
this product will
n (0)
and it will be supposed that it has no repeated

tion which will be justified later ( 23-43).
If
we temporarily regard
... (S)j
Bj, @,
factors
be denoted
a supposi-
as a set of auxiliary variables, the
ordinary formula of partial differentiation gives
9n
(0)
dx
and,
if
we
dU ((H)) 90^ _ ^ dUi}
'^
d^p
pî
2cc
'
+ dp'
a-
differentiate again,
9^n(@)_ ^ 9n(0)
dx'
where the
1, 2, ...
dx
9@p
pZi
m.
9@
p=i
dm()
j,^,
dBpdB,
2
'
a'+0p'
Sx^
"
(a'
+ dp) {a" +
'
6^)
summation extends over all unequal pairs of the integers

The terms for which p = q may be omitted because none of the
last
expressions 0,, @o,
...
0^
enters into
fl
(0) to a degree higher than the
It follows that the result of applying Laplace's operator to 11 (0)
i5II()
V S'n (0 ) (_
j\ow
(a-^
8y'
+ e~,)
is
Sx'
jZ^ Wpm, t(a^ + 0p)
first.
"*"
(6^
0p)
^
(.ry,A{ci'+0p)(O''+f^g)
( +
-
dg)
8f
(c^
+ Op) (c^ + d,)\
'
0g-0p'
\a,b,c )
*
The complete
set of 5 ellipsoidal
harmonics
of the second degree is
composed of these two
together with the three harmonics yz, zx, xy, which are of the third species.
23 21]
539
9n (S)/dSp consists of the product 11 (0) with the factor p omitted,

while dÛ (S)/dpdSq consists of the product IT (0) with the factors 0^ and
and
That
0^ omitted.
to say
is
ân(0)
3^n(0)
If
we make
we
these substitutions,
9^
may be
a^n(0)^8n()
see that
ô n
9-
3'
n(0)
written in the form
g 9n
(0)
V'
the prime indicating that the term for which q
=p
__?_
has to be omitted from
the summation.
If
n (0)
is
to be a
it will certainly
harmonic
annihilated by Laplace's operator
it is
be so annihilated
if it
possible to choose
is
0^,
O.^,
...
and
0,n so
that each of the equations
a'
Now
e^'^
b'
^ c'+dp^
ep
where p takes the values
is satisfied,
let
be a variable and
i9
let
1, 2, ...
Op-e^
m.
Aj (6) denote the polynomial of degree
in 6
n
q
If
A/ (6)
A/(^) denotes
is
m2
cZAj (6)/cW, then,
sum
equal to the
a time, and A/' (6)
{d-e,).
=\
twice the
is
by
direct differentiation,
of all products of ^
sum
î,
...
6.,,
of all products of the
it is
seen that
6^,ni l
at
same expressions,
at a time.
Hence,
if
6 be given the special value 6p, the quotient A^" (6p)IA^ (6p)
becomes equal
to twice the
(the expression 6p
6p
sum
of the reciprocals of 6p6y, 6p
6.2, ...
6p
6^,
being omitted).
Consequently the set of equations derived from the hypothesis that
(0J,)
is
a harmonic shews that the expression
''^'
1
a'
6'^
+6'^ c'+d^
2A/'(^J
A,' (6)
vanishes whenever 6 has any of the special values
6^, 60, ...
6j,.
Hence the expression

(a'
6) (b'
6) (c^
+ 6) A," {6) + l\ 1
"
a, h, c
( +
6) (c'
6)
,'
A/
{6)
540
is
when 6 has any
a polynomial in 6 which vanishes
has 66^,
and so
it
degree
m+
6.2,
and the
1 in ^
d,n as factors.
coefficient of ^'"+^ is
[CHAP. XXIII
of the values
Now this
m (m + |),
factors are
known, the remaining factor must be of the form
where
a constant which
0^, d^,
polynomial
Since
^m,
of
is
of the
m (m + ^) 6 + IG,
is
We
will
be determined subsequently.
have therefore shewn that
(.
0) ^f2
d) (c^
6) A/' {6)
+ \\ 1
{h^-
+ 6) {& + ^)l A/ {6)
= |m(m +
That
is
degree n
is
where
0^,
any
to say,
harmonic of the
ellipsoidal
*)6'
first
i-C'}Ai(6').
species of (even)
6.,,
î are the zeros of a pol}Tiomial Ai(^) of degree
...,
this polynomial
^n; and
must be a solution
of a differential equation of the type
+ ^)(c^ + ^)}|^
+ e){ + e){c^-ê)]^^'>
4V{(a^+^)(6^
^\{a^
=
This equation
is
known
as
Lame's
[n{n
l)e
C].\,{e).
It will
be
in-
vestigated in considerable detail in 23"4-23*81, and in the course of the

investigation it will be shewn that (I) there are precisely ^n + 1 different
real values of
for
The
analysis of this section
the existence of
degree
which the equation has a solution which is a polynomial

(II) these polynomials have no repeated factors.
and
in d of degree \n,
n,
-J
may then be
1 ellipsoidal
reversed step by step to establish
harmonics of the
first
species of (even)
and the elementary theory of the harmonics of the
first
species will
then be complete.
The corresponding results

now be indicated
species will
for
harmonics of the second, third and fourth

notation already introduced being
briefly, the
adhered to so far as possible.

23'22.
Ellipsoidal liarmonics of the second species.

in
We
2m +
1.
take x II (0^) as a typical harmonic of the second species of degree
The
result of applying Laplace's operator to
r: 311(0)
Îpî
dSp
6
I
\a'
X dSpde^
dp'^
\(a'-
b'
it is
Op'^
c'
+ dp]
+ dp) (a^ + 6,^) ^
(6'^
+ dp) (b' + 6,^) "^
(c=
+ 0p) (c^ + 6,)\]
'
23-22, 23'23]
ellipsoidal harmonics
and
Consequently,
this has to vanish.
541
if
3=1
we
find,
by
the reasoning of 23"21, that A^ {0)
is
a solution of the differential
equation
(cc-
d){b'-
+
where Cn
is
^ {3 (6^
(c^~
d) (a'
+ (^ + 0) ib' + 0)} A/ (0)

= {m(m + f)^ + ia}Ao(6'),
0)
a constant to be determined.
now we
If
+ e)A./'(e)
+ 0) ic' + e) +
e){c'
write Ag (0) =
A (^)/\/(a- + 0), we
A (0)
find that
a solution
is
of the differential equation

4>^{(a^
0)(h^-
+ 0){c^+0)}^
V((a^
^)(6-^
^)(c^+^)].^'^^^^
d0
2)
It will be observed that the last differential equation
is
C=
where
6*2
{(2m
6^
1) {2ru
Hence the discussion

first
multiplied by \/{a^
now under
2m +
r
The
31]
\/(b-
-+-
0)
and ^(c"
m+1
all, 8???,
0) respectively.
values of
be
It will
can be associated with

3 harmonics of the second species
are obtained.
Ip'î
(@^) as a typical harmonic of the third species of degree
(Ch))
[
^%
^^
('
dm (@)
+ j^q d%d%
and
reduced to
In the case of
m
take yz
2.
'
1.
Ellipsoidal harmonics of the third species.
23*23.
We
2m +
each of the three types, so that, in

1
is
is
type the solutions are required to be polynomials in

the corresponding factors for harmonics of the
0)
shewn subsequently that precisely
2m 4-
equal to the
consideration,
of harmonics of the second species
second and third types are
of degree
{0),
same type
of the
still
the discussion of solutions of Lame's differential equation.
harmonics of the
C}
c-.
as the equation derived in 23"21, the constant n being
degree of the harmonic, which, in the case
+ 0p^h^ + 0p^
f
c'-{-0,
8^^
8^^
%if
((a^ + 0p) {a^ + 0q) ^
Consequently,
if
this has to vanish.
it is
+ dp) (6^ +
A3(^)= n{0-0,;),
3=1
"^
^,)
(c^
+ 0^) (c' + 0q)
n
'
542
we
by the reasoning of
find,
that A3 {6)
23'21,
[cHAP. XXIII
a solution of the differential
is
equation
(a'
6) (b'
If
we
is
now we
Vl(a-^
6) A," {&)
d)
(c-
6')
+ 3 (c^ +
6) {tr
+ ^) + 3 {te + 6){b' + 6)] A,' {0}

= [m(vi + ^)e + iC,]As(e),
find that
{c'
K^'-' +
+h
where Oj
6)
A {6)
A {0)l^[{b' + 6)
A3 (6) =
write
is
+ 0) { +
(c-'
+ 6)],
a solution of the differential equation
6) (c^
^)|
s/{{ce
+ 6) { +
6)
(c'
dA(e)'
6)}
cie
=
C=
where
C3
-F 4^(1^
{(2m
b-
-f
2)
c-.
It will be observed that the last equation is of the
equation derived in
23'21, the constant n being
of the harmonic, which, in the case
Hence the
now under
A (0),
(2m + 3) ^ + C]
same type
as the
equal to the degree
still
consideration,
is
discussion of harmonics of the third species
2m +
is
2.
reduced to
In the case of
the discussion of solutions of Lame's differential equation.
harmonics of the first type, the solutions are required to be jôlynomials in
^^ corresponding factors for harmonics of
multiplied by VK^" + ^) (^^ + ^)j
are
third
types
^J{(c" + 6) (a- + 6)] and \J[{a- + 6) (b- + 6)]
the second and
+ 1 values of C
It will be shewn subsequently that precisely
respectively.
'
can be associated with each of the three types, so that, in

of the third species of degree 2m + 2 are obtained.
Sni
all,
3 harmonics
Ellipsoidal harmonics of the fourth species.
23'24.
The harmonic
^m + 8
of the fourth species of degree
is
expressible in the
form xyz
The
11 (@p).
^ an^0)
l3i ^%
r
"^^^
and
i>__
t'
^ d'U (B)
"^
,i aeâB,
+^P
^''
'^P
__L
c'
^p
this has to vanish.
e^) {a?
_%'^
8a,-^
[(a-^
it is
+ e,) "^
(ft^
Consequently,
+ e^j +
e^)
"^
(c^
e,;) (c^
+ e,,)\\
if
Hi
A,(^)=ll(^-^,),
7=
we
find
(^
by the reasoning of 23'21 that A4(^)
^)(6'^+^)(c-^
+ ^)a;'(^)+
:s
I
is
(6^
<9)(c^
6^)|a;(^)
= {m(m + |)^ +
where C4
is
iC,}A,(^),
'
ellipsoidal harmonics
23*24, 23-25]
If
now we
write
A, {6)
we
find that
^'^{{ce~
A (0)
is
+ e){h^- +
= A {0)/^{(a^ + 6) ( +
e)(c'
C^
be observed that the
4<
(a-
last
{(2m
h-
3)
0)},
{2m +
4) ^
equation
of the
is
as the
2m + 3.
discussion of harmonics of the fourth species
is
reduced to the
be shewn subsequently that precisely
with solutions of this type, so that
23*25.
same type
is
now under
required to be polynomials in 6 multij)lied by
2rii -f
A (6),
equal to the degree
still
discussion of solutions of Lame's differential equation.
degree
C}
consideration,
of the harmonic which, in the case
will
c'-).
equation derived in 23'21, the constant n being
Hence the
+ 0)}~
C=
where
d) (c'
a solution of the differential equation
=
It will
543
m+1
\/{{a"
m+1
6)
The
{b"-
values of
6)
solutions are
(c^
+ n)].
It
can be associated
harmonics of the fourth species of
3 are obtained.
Nivens expressions fur
ellipsoidal
harmonics in terms of homo-
geneous harmonics.
If Gn (x, y, z) denotes any of the harmonics of degree n which have just
been tentatively constructed, then Gn (^", y. z) consists of a finite number of
terms of degrees n, n 2, w - 4, ... in x, y, z. If H^ {x, y, z) denotes the
aggregate of terms of degree n, it follows from the homogeneity of Laplace's
operator that
Hn (^,
y, z) is itself
obviously be obtained from Gn
occur in the expression of Gn
a solution of Laplace's equation, and
{x, y, z)
y, z)
{oc,
as
it
may
by replacing the factors (h)^, which

a product, by the factors Kp.
It has been shewn by Niven {loc. cit., pp. 243-245) that Gn (x, y, z) may
be derived from Hn {oc, y, z) by applying to the latter function the differential
operator
2(2?i-l)"*'2.4.(2w-l)(2w-3)
where
D'-
2.4.6(2^1
- l)(27i-3)(27i-5)
"^
"
'
stands for
92
B^
92
da^
dy^
dz^
a^ + b^- + c^~
and terms containing powers of
'
higher than the nth
may be omitted from
the operator.
We shall now give a proof of this result for any harmonic of the first species*.
*
The
proofs for harmonies of the other three species are
proof applicable to fuiictious of
all
left to
the reader as examples.
four species has been given by Hobson, Proc. London
Math. Soc. XXIV. (1893), pp. (30-64. In constructing the proof given
cations have been made in Niven's proof.
in the text, several modifi-
544
For such harmonics the degree
even and we write
is
p=i
where Sn Sn-2 Sn-i

and
,
are
...
[CHAP. XXIII
p=\
homogeneous functions of degrees
n,
w 2, w 4,
. .
respectively,
in
Sn
= Hn (x,
y,
Z)=U
Kp.
p=l
The
taken
|-
function Sn-2r
evidently the
is
sum
of the products of K^, K2,
...
Ki,^
at a time.
If K^, K2,
... /iTi^n
be regarded as an auxiliary system of variables, then, by
the ordinary formula of partial differentiation

a>s,
doc
dKp
p=-i
p=i
and,
if
we
9r
dKp
a~
+ Bp
differentiate again,
= t
dx^
The terms
Kp
dx
dS
in
pLx
-^-^^+
t
dKp a-'+dp p^^dKpdKq{c(r +
dp){a^
+ 6^)'
in d-Sn-^rl^Kp^ can be omitted because each of the functions
does not occur in Sn_2r to a degree higher than the
first.
It follows that
p=i
^;,
It will
dKp
BKpdK,
\a^
\(a'
now be shewn
6^
^p
&+
<9p
dp\
+ Op) (a^ + 6,) ^ { +
Op) {b'
+ 6,)
that the expression on the right
is
OI >Oji_2r2-
We
first
observe that
OpKp dqKq
a?x^
\a,
and
that,
by the
b,
c)
differential equation of 23-21,
a,T,
"
+ 0p
-\-
Up
a, b, c
2,
a-
^
q = \ tTp
Op
^
Vq
(c^
+ dp) (c^ +
6,)]
a constant multiple
23-25]
545
so that
D'^nor
^^
oKp
p=i
Now
dSn-or/dKp
is
the
dKpdKq
sum
e,
p=i
d"Sn2r
pjpq
dSn-.
in
bK
6p
0q
dpKp UqKq
% (^q
of the products of the expressions K^, Ko,
(Kp being omitted) taken ^n ?' 1 at a time and Kqd^Sn-2rldKpdKq

consists of those terms of this sum which contain Kg as a factor.
.
iTijj
~ - K,
C'0,i_2'
Hence
dK.,
is
equal to the
sum
of the products of the expressions K^, Ko,
Kq both being omitted) taken hn
r1
and
at a time;
... -K"i;
therefore,
(Kp and
by sym-
metry, we have
K,
''dKpdKq
dKr.
is
-KP dKpdKq'
dSn-
so that
On
dK,
dKpdkq "
\Kq-Kp).
dK.
~dKr
,p
w-i^q
substituting by this formula for the second differential coefficients,
it
found that
hi
^
hi
pî
{4:11
6-8
dKp
5io,
"I"
denotes the
\Kp-Kq).
"^-Q-p
p^
of the products of the expressions K^, K.^,

in at
dSn-
"
Now
it is
D'-8n-,r
dK
{Kp-Kq)Sn-,r-2.
p-l cIVp
K.2,
p^q
Kx,^ of degree
...
It is therefore a multiple of Sn-^r-zA.
... /fj^,,
and we then see that
= (4n -2) t
power of any of the expressions K^, K^,
w. M.
a time
clear that the expression on the right
metric function of K^,
first.
Kq)_
K
Kp Kq_
(Kp and Kq both being omitted) taken
Hence
= \{Gp^q){Kp
write Sn2r in the form
^nir
where
I)
p=\
Now we may
i'
2=1
UpKp UqKq
ii
Gq
q = \(^p
^n
is
r\,
homogeneous symand it contains no
Ki^^ to a degree higher
To
than the
determine the multiple

35
we
546
when Sn-^-^
observe that
while the
number
hn (4n -
and
this
is
is
contains
jtC;.+i
terms
2)
^,^_,C,
- 8 â
.
i_,C,-i-
consequently
equal to (2r
It has consequently
+ 2) (2?i - 2r -
It follows at once
1).
been proved that
D-^Sn-.r
^n-2r
it
of terms in
is
The multiple
written out at length
is
[CHAP. XXIII
(2r
2) (2n
2r
- 1) ^_.^_,.
by induction that
2.4...
2/-
{2n
- 1) (271 -
3)
...
(2n
2r
1)'
and the formula

Gn (x,
is
y, z)
(-YD"-'-
now obvious when Gn {x,

Example
y, z) is
an
ellipsoidal
Prove Niven's fonmila when G
1.
Hn {x,
,.=o2.4...2r.(2n-l)(2n-:3)...(2n-2r + l)
(.r,
harmonic of the
y, z) is
an
ellipsoidal
y, z)
first species.
harmonic of the
second, third or fourth species.
Example
Obtain the symbolic formula
2.
G^{x,
2326.
y, 2)
= r(i-).
(iZ))"+^/__^(Z)). ^(.r, y,
z).
Ellipsoidal harmonics of degree n.

in 23"21-23"24
shew that when n is

^n + 1 harmonics of the first species and |/i harmonics of the
third species; when 7i is odd there are |(?i + l) harmonics of the second
species and ^ {n 1) harmonics of the fourth species, so that, in either case,
The
results obtained
and stated
even, there are
there are 2n
harmonics in
all.
It follows
from
18*3 that,
if
the terms of
degree n in these harmonics are linearly independent, they form a funda-
mental system of harmonics of degree n and any homogeneous harmonic of

degree n is expressible as a linear combination of the homogeneous harmonics
which are obtained by selecting the terms of degree n from the 2)i + 1 ellip;
soidal harmonics.
In order to prove the results concerning the number of harmonics of

degree n and to establish their linear independence, it is necessary to make
an intensive study of Lame's equation but before we pursue this investigation
;
we
shall
study the construction of ellipsoidal harmonics in terms of confocal
coordinates.
'
23-26, 23-3]
547
These expressions for ellipsoidal harmonics are of historical importance in view of

Lame's investigations, but the expressions which have just been obtained by Niven's
method
are, in
some
respects,
suitable for physical applications.
harmonics to the investigation of the Figure of the Earth,

harmonics to forms adapted for numerical computation, the
referred to the memoir by G. H. Darwin, Phil. Trans. 197 a (1901), pp. 461-537.
For applications of
and
more
ellipsoidal
for the reduction of the
reader
is
23"3.
Confocal coordinates.
If {X, Y,
a, b, c
Z) denote current coordinates
are positive (a
>
> c),
in three-dimensional space,
X'
Z^
Y'
+7^+^
=
0-
a-
represents an ellipsoid
is
,+
e
b'
is
^2
7-2
X2
and 6
if
c-
the equation of any confocal quadric
a'
and
the equation
^'
+ 0'^c^ + e
called the pcuximeter of this quadric.
The quadric
passes through a particular point
{x, y, z)
if
is
chosen
so that
ar
Whether 6
+d
y-
+
z'
yl_
a--ve
+
c-
equation or not,
satisfies this
1-
62 +
(9
e
c''
= ].
e
it is
convenient to write
/(^)
+ e~
{a'
+ d) (6^ +
0)
(c-
+ 6>)
and, since f{d) is a cubic function of 6, it is clear that, in general, three

quadrics of the confocal system pass through any particular point (x, y, z).
To determine the
species of these three quadrics,
Table
e
we
construct the following
548
Now
take the identity in
6,
e-\){e-ii ){e-v)
+ e){b"- + d){c- + e)'
+ 0, c- + 6 and after so doing,
a?
_ j/^ _
2'
ar-ê
b-'+tl
C--+6
and multiply it, in turn, by a- +

h-, c- respectively.
6 by
6, h-
It
(('-,
^ _ (6^ +
^'
(a-
{a'
\) (a^
fj.)
(a-
+ v)
(6^+^6- +
X)
(c-^ X )
(c-
fj,)(c-
(a-c2)(6--c=)
From
/Lt,
these equations
V denote the
it is
replace
thus found that
is
lO
'(a--6-)(^'-c-)
_
^'~
if X,
[CHAP. XXIII
clear that, if
(a-,
y, z)
'
v)
*
be any point of space and
parameters of the quadrics confocal with
=
+ 1T + Z-
X'-
Y-'
a-
0^
which pass through the point, then

terms of (X,
v) and vice versa.
c-
(x-, y", z-)
are uniquely determinate in
fj,,
The parameters
{x, y, z)
(X,
relative to the
It is easy to
the confocal coordinates of the point
v) are called
fi,
fundamental ellipsoid
X'
Y-
Z'-
a-
b-
c-
shew that confocal coordinates form an orthogonal system
for consider the direction cosines of the
of the surfaces
and smce
and
(/a)
dx dx
--
(v)
/dx
dy
dz
\d\'
dX'
dx
+ ^^4._-- _= ^^
dXdJi^ ^Xd^l'^
dy dy
tangent to the curve of intersection
these direction cosines are proportional to
dz dz
^
%
_a^+
^-_v
aXc (a^-6')(a2- c^)
dXdJjL^
it is
evident that the directions
are perpendicular
fdx
dy
dz\
/dx
dy
dz\
Vax'
dx'
dx)'
\d^'
Yfi'
dfx)
and, similarly, each of these directions
dy
/dx
is
perpendicular to
dz\
'
\dv
It has therefore
X,
jji,
'
dv
'
been shewn that the three systems of surfaces, on which
V respectively are constant, form a triply orthogonal system.
Hence the square
of the line-element, namely
{hxy-^{hjr
is
dv)
+ {hz)\
23-31]
with similar expressions in
To evaluate Hi-
"
4^^ va^y
^x
if
and
v for H.^^
in terras of (X,
'
But,
and
/j,
û,
^ 4^^
Ux-y
4^/^
(g^
H3'.
observe that
y),
"^
s;
549
+ fx) {a? +
\dx
i^)
we express
(A,
(X, v)
+ X) (6^ + X) (c" +
/Lt)
(a^
qua function of
sum
X, as
X)
'
we
of partial fractions,
see that
it is
precisely
equal to
(g^ + fi) (g" + v)
+ X)(g2-6^)(a^-cO'.
^
.5,c(g'
and consequently
^
^
The values
H^-
of H.^
interchanges of
(X,
/i,
Formulae equivalent
II.
-r
'
and
Hf
(a2
z-A
c
X) (62 + X) (c^ + X)
+ ^.îl
are obtained from this expression
by
cyclical
i').
were obtained by Lame, Journal de Math.
to those of this section
(1837), pp. 147-183.
Example
1.
With the notation
shew that
of this section,
.r2+/ + 5- = a2 + 62 + c2 + X +
Example
2.
Shew
It has
i'.
that
y"^
-^^
ATTI-
23 '31.
jtx
^^
{a;'
+ \f^{h^ + \f^{c'' + \f'
Uniformising variables associated
been seen
in 23'3 that
ivith
confocal coordinates.
when the Cartesian
are expressed in terms of the confocal coordinates (X,
obtained are not one-valued functions of
thereby produced, we express
(X,
/i,
v) in
(X,
/x,
v).
fx,
coordinates
v),
(x, y, z)
the expressions so
To avoid the inconvenience

{u, v, w)
terms of three new variables
respectively by writing
HP
(u)
^j(v)
'^
the invariants
g.2
and
i'w)
= X + |(a- + 6^ + C-),
=/i +!('
+ ^' + c'),
=v
\-\{d^
h'"'
+ c^),
g^ of the Weierstrassian elliptic functions
by the identity
4>(a-+X){b'
+ \){c'+X)^^f{u)-g^^{u)-g,.
being defined
The discriminant
example 3)
so
it is
with
associated
the
positive
{a-
b-Y
{b"-
&).,+ 0)3
>e.2>e:;,
e,
= a- + b-- '2c-,
Next we express
and
(cf.
20-33,
a'Y,
2a)2
the imaginary part of Wo
In these circumstances
3e,
c'Y- (c-
and, therefore*, of the periods
positive while 2a)3 is a pure imaginary
since Wj
functions
L-lHptic
is
16
and
[CHAP. XXIII
550
Se^
and
= c" +
terms of
(x, y, z) in
its real
and
2&)3, 2g)i is
part negative,
positive since
((o.J(o^)
> 0.
we have
so
a-
is
2a),, 2&)o
has
w)
{u, v,
= b-
8e,
2b-,
c-
- 2a-.
we have
{a^-b^){a^-c-')
{f
(m)
- e,]
_
~
by
20-53,
example
- 63}
\^ (v)
(gj
a^- (u) (T3- (V)
0-3-
q--
(tv)
(w )
- eal
(6)1) 0-- {(O.)

'
'
cr {v) a- {w)
o-^ (it)
0-3- (a)i) 0-3- (td.)
we have
Therefore, by 20'421,
4.
x.-^-.
0-3
"/
^
and similarly
{g>
- es) {bo - e^)
3/
(w)
o- (t;) o-
a-2
(w)
(7o
e-^-a-(a,,)
^ (,,)
_ 4.-X
').'"i
g -T,co
"/
cr-
(w)
0-3 (t') o-s
(")
o-
^-
{w)
(w)
(v)
cTa
(^)
^ ^^)
ôl(")o-l(y)ol(w)
--
a{u)
a-
(v)
a (w)
each of u, v, w by 2(0^ is to change the sign of the

the expressions for y and z remain unaltered
while
expression given for x
and simila'r statements hold for increases by 2&)._. and 2&)i and again each of
The
effect of increasing
the three expressions
changed in sign by changing the signs of
is
u, v, iv.
the upper signs be taken in the ambiguities, there is a unique

correspondence between all sets of values of {x, y, z), real or complex, and all
Hence,
if
the sets of values of
given
(it,
v,
w) whose three representative points
lie
in
any
cell.
The uniformisation
is
consequently effected by taking

.,
-.u.
^
'^
0/
0-3
{v)
0-3
{v)
a (w)
o-2
(w)
{u)
cr {ii)
o-:,
a-
(t/-)o-.,(t;)o-,(w)
{a)
{v)
cr
{w)
(t{u) (T{v)cr{w)
Formulae which differ from these only by the interchange of the suffixesand 3 were given by Halphen, Fonctions Elliptiques, li. (1888), p. 459.
Cf. 20-32,
example
1.
23-32]
23'32.
Laplace's equation referred
551
confocal coordinates.
to
been shewn by Lame and by W. Thomson* that Laplace's equation

when referred to any system of orthogonal coordinates (X,
v) assumes the
form
It has
/j.,
0,
H,
where {H^,
is
H.2,
H-^ are to be determined from the consideration that
Although W. Thomson's proof of
to be the square of the line-element.
based on arguments of a physical character,
result,
dv
this
extremely simple,
is
all
the analytical proofs are extremely long and cumbrous.

It has, however,
(X,
/JL,
Lamef
been shewn by
that, in the special -case in
which
v) represent confocal coordinates, Laplace's equation assumes a simple
form obtainable without elaborate analysis
when
the uniformising variables
w) of 23"31 are adopted as coordinates, the form of Laplace's equation

becomes still simpler.
(u, V,
By
straightforward differentiation
independent functions
(X,
may be proved
it
v) of {x, y, z) are
/n,
that,
when any
three
taken as independent variables,
then
a^F
a^F
dx-
dy^
a^F
dz-"-
transforms into
yax
t
+
\dx
dv
dfx
J<1 +
A.M. V
dv
d/ji
dx dx
dz.
d/Ji
dy dy
dv
dz dzj
d^\
d^\
d'\
d_V
\_^x-
dy-
dz^\
dx'
d/jidv
In order to reduce this expression, we 'observe that X

y^
X-
oMO.
and
so,
by
a-
z-
PTX cM^ =
"^
(a^
(a^
dx
X)-
y^
,+
+X
+
x^
14a;
the equation
1,
differentiation with x, y, z as independent variables,
2x
a^
"*"
satisfies
dx
+ xy
2
{{a^
(6-
+ X)
* Cf. the footnote
on
t Journal de Math.
(c^
y.
+
(6^
X(a2
1 ^'^
^^
^
+
xy
X)-
+ X)=
+ x)-^j
+
(c^
+
{h-
+ \y
X)^
d'X
p. 401.
iv. (1839),
pp. 133-136.
_A
dx
(c^
X)*-^]
dx
= 0.
'
552
^^~^=4<Hi^^,
da;
a.' + \
Hence
+X
a'
[CHAP. XXIII
(a-'
+ \y //j-
2^1^
{a'
xy^ /^,7, ^v (a-
+ Xf
'
d.v'
\a, b, c)
with similar equations in
and
fx,
d-V
From
equations of the
type
first
d-V
and the
V5r-
y, z.
second type obtained by interchanging
with similar equations in

If,
for brevity,
we
[x
with similar meanings

"^
and
for
0Z-]
a^-
so Laplace's equation
that
is
>'
o,
i.c"
of the
found that
+^
v.
+ X){h- + \){c'^-X)] = ^K,
A^ and
"^
a^^
x.T",
we add up equations
x, y, z cyclically, it is
we
A^,
('^
'
see that
(x
- /i) (x - v)
(X
- /x)
4Ax
and
if
+x
a-
6-
+'x
"^
c^
+X
cÂx
(X
'
rfX
I/)
assumes the form
A*) (^ ~ ^)
dx ax
^^'
= 0,
to say
('^-^)^â-xWI^]+<^-^)^'IKa + ^^-'')^'^.K'5 =
The equivalent equation with
[i^j
or,
is
write
^/[{a^
a^;-
dy-
and
coefficient of tt-t^ is zero
[dx-
seen that the coefficient of -^
it is
more
{v)
- i^ {w)] 1^ +
{^ {w)
{u, v, lu) as
- ^ in)]
independent variables
^+
[iO
{u)
- ^ {v)]
simply
0,
briefly,
^^-^>a^ + <^-^>-a^ + ^^-^>a^^ =

The
^=
is
last three
'-
equations will be regarded as canonical forms of Laplace's
equation in the subsequent analysis.

23*33.
When
Ellipsoidal harmonics referred to confocal coordinates.
Niven's function 0p, defined as
a-
6,pJ
b''
+
^
6,,
^p
c- ^
^
^p
23*33]
is
it
expressed in terms of the confocal coordinates {\

assumes the form
{\-e^){t,-ej){v-dp
and consequently, when constant
(^
553
[x,
v) of
the point
{x, y, z),
factors of the form
( +
e,,)
dp) (c^
Op)
are omitted, ellij)soidal harmonics assume the form

X,
yz
y,
zx,
1,
xyz
xy
z,
If
any
now we
replace
(/i-6'p)
=1
73
{v-B^).
p=\
x, y,
harmonic
ellipsoidal
- ^^) H
p=l
(X
is
z b}' their values in terms of X,
form of a
expressible in the
/*,
v,
wê see that
constant multiple of
where A is a function of \ only, and

and N are the same functions
and V respectively as A is of X. Further A is a polynomial of degree nn
in \ multiplied, in the case of harmonics of the second, third or fourth
species, by one, two or three of the expressions \/{a' + X), \/{b- + X), \/(c^ + ^)-
AMN,
of
/Li,
Since the polynomial involved in
is
i3
A is
sideration of 23'21-23-24 that
si [{a-
X) {
(c^
+ X)}
(X
6p),
is
V((a-^
the degree of the harmonic in
may
This result
also
follows from a con-
a solution of Lame's differential equation
X)(6^
+ XX)(c^ + X)}^^
=
where n
it
=l
[n{n 4-l)X+C'} A,
{x, y, z).
be attained from a consideration of solutions of
Laplace's equation which are of the type*
V= AMN,
where A, M,
For
if
we
are functions only of X,
in 23'32, on division
^{v)-io{w)
A
The
1
so
-r-
last
dÂ
-y
by V, we
d}K^^
v respectively.
find that
w) -
^J
{u)
d'M
dii^
must be a
^0
(u)
u,
- ipO^) d'N
dv^
two terms, qua functions of
coordinates v and w,
where
/x,
substitute this expression in Laplace's equation, as transformed
" div-
^^
are linear functions of
linear function of ô(u); since
it
is
ip
(u),
and
independent of the
we have
K and B are constants.
harmonic which is the product of three functions, each of which depends on one coordinate only, is sometimes called a normal solution of Laplace's equation. Thus normal solutions
with polar coordinates are
18-31|
r"
P " '
(cos d)
'
^
m(p.
sin
554
If
we make
function of
this substitution in the differential equation,
(k) equated (identically) to zero,
(,)
must vanish
linear function
that
m,.
is
X.
is
d-M
rf'N
taken as independent variable, the

4 A. ^^|a.
and
Î
we get a
linear
so the coefficients in this
to say
s)
and
and on solving these with the observation that ^ (v)

zero, we obtain the three equations
When
[CHAP. XXIII
= [K\ -^ i? +
first
i/i (a^
j^J
equation becomes
+ + c^)]
this is the equation already obtained for A, the
not identically
(lu) is
A,
degree n of the harmonic
being given by the formula
n (n
We
is
have now progressed so
far
l)
= K.
with the study of ellipsoidal harmonics as
convenient without making use of properties of Lame's equation.
We
now proceed
234.
We
and
to the detailed consideration of this equation.
Various forms of Lame
s differential equation.
have already encountered two forms of Lame's equation, namely
this
may
d\'
also be written
\a"
+\
h-
+X
c-
+X} d\ ~ 4 (a- + X) {b- + X) (c- + X)
which may be termed the algebraic form
'
and
^^ = {n{n + l)^j(ii) + B]A,

which, since
it
contains the Weierstrassian elliptic function ^(u),
termed the Weierstrassian form
the constants
and C are connected by the
c-)
relation
B+
in
1) (a-
b-
may be
C.
23 "4]
If
lame's equation
we take
^j
new
(u) as a
which
variable,
will
555
be called
^,
we obtain the
slightly modified algebraic form (c 10"6)
This differential equation has singularities at
exponents are
0,
exponents are
^n,\{n + \).
The
eo,
gj,
and a singularity at
ir.
e^
at
infinity, at
"Weierstrassian form of the equation has been studied
Elliptiques,
The
^ in each case
which the
which the
by Halphen, Fonctions
(Paris, 1888), pp. 457-531.
algebraic forms have been studied
by
Stieltjes,
Acta Math.
vi. (1885), y>V.
321-326,
Klein, Vorlesimgen iiber lineare Diferentielgleichunge/i {lithogr&iAiGd, Gottingen, 1894),
and
Bocher, l/ber die Reihenentioickelungen der Potentialtheorie (Leipzig, 1894).
The more general
exponents are arbitrary (save that the
sum
with
arbitrary singularities at which the
foiu*
of all the exponents at all the singularities is 2)
has been discussed by Heun, Math. Ann. xsxiii. (1889), pp. 161-179 the gain in generality
by taking the singularities arbitrary is only apparent, because by a homographic change
;
of the independent variable one of
then a change of origin
is
them can be transferred to the point at infinity, and

make the sum of the complex coordinates of the
sufficient to
three finite singularities equal to zero.
Another important form of Lame's equation

notation of Jacobian elliptic functions
Zl
if
we
U V(ei
is
obtained by using the
write
63),
the Weierstrassian form becomes

n {n
dzi"
and putting
2^
=a
+ 1) I -^'
- > 4-ns2îV + e,e
1^1
iK' w^here 2iK'
is
A,
the imaginary period of sn z-^
we
obtain the simple form
i- = \n (n + 1) k- sn-a + A] A,
doP
where
is
a constant connected with
B+
e-iii
(n
>
1)
by the relation
= A{e,-
e-,).
The Jacobian form has been studied by Hermits, Sur quelques

elliptiques,
applications des fonctions
Comptes Rendns, Lxxxv. (1877), published separately, Paris, 1885.
In studying the properties of Lame's equation, it is best not to use one

form only, but to take the form best fitted for the purpose in hand. For
practical applications the Jacobian form, leading to the Theta functions, is
For obtaining the properties of the solutions of the

equation, the best form to use is, in general, the second algebraic form,
though in some problems analysis is simpler with the Weierstrassian form.
the most suitable.
'
556
23'41.
Solittioiis in series
Lame
of
[cHAP. XXIII
s equation.
Let us now assume a solution of Lamp's equation, which may be written
in the form
A=
6.(1-^,)^"-''.
)-=0
The
on the right,
series
if it
sufficiently small values of \^
e.,\',
is
a solution, will converge
but our object
of the convergence but the choice of
will
10"31) for
be not the discussion
in such a wây that the series
may
terminate, so that considerations of convergence will be superfluous.
The
result of substituting this series for
on the left-hand side of the
result in powers of ^
and arranging the
go
is
minus the
series
(^-e.,)^'^-r + i[r(n-r-[-hJbr-{Se,(^n-r
+ ir--in{n+l)e,-lB]br-^
r=--0
+
in
which the
Hence,
coefficients
if
the series
hj.
(e,
- e,) (^ w -
e,) (e,
2)
(i n
with negative suffixes are to be taken
+ f ) 6,-2],
to
be zero.
to be a solution, the relation connecting successive
is
coefficients is
r (n
-r +
^)br
and
(n
If
bj.
we take
&
1,
(^n
-r+
- (ei
- e^) (^2 - 63) i^n - r +
[Se^
bj.
as
?i
we may do without
a polynomial in
The sign
(ii)
?' :$
is
(n
+l)eo-lB\
2)
br-i
{^n - r + f) br-.,
- i) 61 = {^n% -ln{n + l)e,- ^B}b,.
are seen to be functions of

(i)
1)-
loss of generality,
the coefficients
with the following properties

of degree
of the coefficient of
the actual coefficient of B^
B''
r.
in br
is
that of
()'",
provided that
is
i-Y
2.4>...2r(2n-l){2n-3)... ( 2n - 2r
(iii)
of
6,.
and
Now
If
^1, e.,
6,._o
e-i
and
suppose that n
If this choice
are real and
î
> 62 >
^3,
are opposite in sign, provided that r
is
is
+ 1)
= 0, the values
+ 'A) and r < n.
then, if b^-i
<
even and that we choose
^ {n
in such a
made, the recurrence formula shews that
hn ^2 = 0,
way
that
23-41]
lame's equation
by putting r =
^1)1
^7i
+2
557
in the formula in question;
and
+ 2 ^^6 ^^^^ ^^^ subsequent recurrence formulae are

^1/1
+3
î + 4
...
if
both
satisfied
^n +
of degree
When
is
that
is
when n
1,
is
=0.
even.
coefficients
so that the condition,
also vanishes,
when
is
odd,
is
1).
shew that, when e^ > e., > e-^, these algebraic equations have
For the properties (ii) and (iii) shew that, qua functions
their roots real.
all
of B, the expressions
< ^ (n + 3), and
has
all its
h^, b^, b, ...
roots realf
for
form a set of Sturm's functions* when
and unequal.
Hence, when the constants
b,.
so the equation
importance, as was seen in

of
and then 6i/^3x
should be a root of a certain algebraic equation of degree \ (n
It is easy to
solution which
should be a root of a certain algebraic equation
odd, Ave take 6x/,jî\ to vanish
and so do the subsequent

that
and
hy taking
Hence the condition that Lame's equation should have a

a polynomial in ^
is
6i,j_î
gj, e,, es
are real (which
23'31), there are ^?i
is
the case of practical
1 real
and distinct values
which Lame's equation has a solution of the type
tbr(^-e^^-'
Avhen n
is
even; and there are
^(n+1)
real
and
distinct values of
for
which Lame's equation has a solution of the type
when
When
is
odd.
the constants
ej, eo, e^
are not
all real, it is
possible for the equation satisfied
have equal roots the solutions of Lame's equation in such cases have been
discussed by Cohn in a Konigsberg dissertation (1888).
by
to
Example
1.
Discuss solutions of Lame's equation of the types
_^^
(^-îF i h;{^-e.^^^-'--^,
(i)
(ii)
(iii)
'
r=0
(^-e,)^ 2
(^-e,)^{^-e,)i I b;"{^-e,)^"-'-\
Mem. pr^senUs par
t This procedure
is
*?l-r-J
V'd-eo)^
les
Savans Etrangers,
vi. (1835),
due to Liouville, Journal de Math.
pp. 271-318.
xi. (1846), p.
221.
558
[CHAP. XXIII
ohtainiug the recurrence relations

r {n
(i)
-r + h) W
{hi - r + hf + ie^-e^)
=-{Se.-,
{h
- (ei - 62)
u - r + |)
(^2
- 63)
-^n (n + 1) 63 - i^} 6',.-i

-r + %) {hi -r + l) 6V-2,
{hi
r{n-r + h)b," = {3e.,{hn-r+hf-{ei-e2){in-r + ^)-^7i{7i+l)e.,-iB\b"r_,

- {ei- 62) {62-63) {^n - r+^) {hi-r + l) b",._.i,
(ii)
(iii)
r{n-r + h)b;" = {3e2{^n-r + h)--ke2{n' + n + l)-^B}b"'r_i

- (^1 - eo) (62 - e-i) ih)i-r+l){l n -r + h)
Example
values of
2.
for
b"\. _ 2
With the notation of example 1 shew that the numbers of real distinct
which Lamp's equation is satisfied by terminating series of the several
species are
\{n-\)
(i)
or
The
2342.
When we
\{n-2);
(ii)
definition of
\{n-\)
Lame
collect the results
ov h{n-9.);
(iii)
^{n-2)
or h{n-Z).
functions.
which have been obtained
in 23*41,
it
is
clear that, given the equation
n being a positive integer, there are 2n + 1 values of B for which the equation
has a solution of one or other of the four species described in 23"21-23'24.
If,
when such a
solution
expanded
is
term |^"
coefficient of the leading
the function so obtained
is
called a
is
in descending powers of
taken to be unity, as was done in
Lame
function of degree
the
2341,
of the
n,
The 2n +
kind, of the first (second, third or fourth) species.
^,
first
functions so
obtained are denoted by the symbol
(m =
E,^-{^);
and,
when we have
to deal
l, 2,
with only one such function,
it
...2
+ l).
may be denoted by
the symbol
Tables of the expressions representing Lame functions for ?i = l,
compiled by Guerritore, Giornale di Mat. (2) xvi. (1909), pp. 164-172.
Example
1.
Obtain the
V(X +
Example
2.
i-)v/(X
Lame
five
+ c2),
functions of degree
v/(X
Obtain the seven
+ c2)^/(X + a2),
Lame
and six functions obtained by interchanges of
23'43.
It will
unequal.
+ a-')
[X
v'(X
+ -)V(X +
3,
namely
V{(X + a'^)(X + &'')(X4-c2)},
v'(X
...10 have been
namely
2,
functions of degree
2,
a, b, c in
the expressions
+ 1 (a2 + 262 + .2c') I J{a* + ib* + 4c* - 76V-' - f'â^ _ a'b^}].
27te non-repetition
now be shewn
of factors in
that
all
Lame functions.
the rational linear factors of ^"' (|) are
This result follows most simply from the differential equation which
-^n (?) satisfies
for, if
?i
be any factor of ^'"(|), where
|i is
not one of
lame's equation
23-42-23-44]
the numbers
gj, eo
or
63,
then
fj
559
a regular point of the equation ( 10"H),
is
and any solution of the equation which, when expanded in powers of f ^1,
does not begin with a term in (^ ^1) or (^ ^^Y must be identically zero.
Again,
if |i
appropriate to
were one of the numbers

would have the roots
^1
En"' (^) in ascending powers of
^1
gj, 63
or
and
J,
^s.
the indicial equation
and so the expansion of

would begin with a term in (f ^j)" or
Hence, in no circumstances has
^,i'"(^), g'wa
function of
^,
a repeated
factor.
The determination of the numbers 0^, 6.2, ... 0^ introduced in

may now be regarded as complete for it has been seen that
23*24
23-21-
solutions
of Lame's equation can be constructed with non-repeated factors, and the

values of
0^, 62, ...
which correspond to the roots of En^(^)
satisfy the
equations which are requisite to ensure that Niven's products are solutions of
Laplace's equation.
It still
remains to be shewn that the
structed in this
'2n
ellipsoidal
way form a fundamental system
harmonics con-
of solutions of degree
71
of
Laplace's equation.
23'44.
The linear independence of Lame functions.
now be shewn that the 2n -i- 1 Lame functions E)/^ {^) which are
n are linearly independent, that is to say that no linear relation can
which connects them identically for general values of ^.
It will
of degree
exist
In the
first j)lace, if
different species
such a linear relation existed in which functions of

it is obvious that by suitable changes of signs
were involved,
of the radicals \/(^
î),
\/(^
^2),
\/(f
^s) we
could obtain other relations
which, on being combined by addition or subtraction with the original relation,
would give
rise to
two
(or
more) linear relations each of which involved
functions restricted not merely to be of the
same
species but also of the
same
type.
Let one of these latter relations,
and
let this relation
be
involve r of the functions.
Operate on this identity
The
if it exists,
7^
times with the operator
results of the successive operations are
ta,, {Bn'-y Err{^)=0
where Bn'^
is
the particular value of
jB
which
is
(5
1, 2, ...
associated with E^^"-
1),
{^).
'
560
Eliminate
a-^,
iu, ...
a,,
[CHAP. XXIII
from the r equations now obtained; and
it is
found
that
1
R
Dii
J->n
{B,^y-\
Now
...
>
-'-'71
=0.
1
Tl
-'-'(!
(B/y-'
{B,;r-\
the only factors of the determinant on the
left
are differences of the
and these differences cannot vanish, by 23"41. Hence the

numbers
determinant cannot vanish and so the postulated relation does not exist.
5'",
The
linear independence of the
2m +1 Lame functions of degree n
is
The linear independence of ellipsoidal harmonics.
2345.
Let Gn'^
{^, y, z)
E,^ (^), and
let ^T,/"
It is
easy to shew that not only are the
now
linearly independent, but also the
Gn"^ (^, y, z)
Hn''
In the
first
place, if a linear relation existed
y, z).
{oc,
then,
focal coordinates {X,
/x,
seen that no such relation
type
+ 1 harmonics of the
+ 1 harmonics of the
type
type
between harmonics of the
when we expressed these harmonics in terms of conv), we should obtain a linear relation between Lame
functions of the type En'î^) where
Again,
2??
2?;
are linearly independent.
{x, y, z)
type Gn'"
be the ellipsoidal harmonic of degree n associated with

(cc, y, z) be the corresponding homogeneous harmonic.
^=X+
^{a-
+ b-+
c"),
and
it
has been
exists.
a linear relation existed between homogeneous harmonics of the
if
Hn^'^ {x,
y,
z),
by operating on the relation with Niven's operator
( 23-25),
^- T) "^2.4
2 (2n
D\
(2
- 1 {2n- 3)
"
'
we should obtain a linear relation connecting functions of the type (z,/'' {x, y, z),
and since it has just been seen that no such relation exists, it follows that the
homogeneous harmonics of degree n are linearly independent.
2346.
It has
Stieltjes'
theorem on the zeros of Lame functions.
been seen that any Lam6 function of degree n
is
expressible in the
form
m
(6
+ a'Y' (d + h-^r (0 + cy^ .u{d- dp),

;>
=1
or ^ and the numbers 6^, 6., ...

where i, k^, k-^ are equal to
unequal both to each other and to a^, b^, c- and ^n =
;
When
type
/cj,
is ru
/Co,
Ks are
1.
given the
number
of
Lame
6,^
ni
are real and
Ki
+ Kn + k^.
functions of this degi-ee and
lame's equation
23*45, 23*46]
The remai'kable
between
of its êros
functions,
of its zeros f
6-^,
6.2,
To prove
0^
...
6-
lie
w+1
been proved by Stieltjes* that these
result has
way
functions can be arranged in order in such a

set has r
561
that the rth function of the
between a^ and h- and the remaining m r +

and c^ and, incidentally, that, for all the 7?i +
between a- and c-.
this result, let 0j, ^2;
be any real variables such that
4^m
a- ^(jip^
h^,
[-h"^(})p^-
1,2,...') 1)
(p = r,r+l,... m)
(p
c^
and consider the product

'
11
J-
= n
[| ((/>p
4-1
+aO T'+i
(</),.
r^-'i
h"-)
(<^^
a.1
This product
when
(c/>^
]
- <^,)
|.
pdpq
p=l
also
c^) i;3+4]
zero
is
when
the variables
all
have their greatest values
all
both to each other and to
a-,
when
c", then
If,
have their
0^^
and
least values
the variables ^p are unequal

and it is obviously
IT is positive
a continuous bounded function of the variables.
Hence there
a set of values of the variables for which 11 attains
is
upper bound, which
is
positive
and not zero (cf
For this set of values of the variables the conditions
eiogn _
that
is
<f)p
where p assumes
Now
this
+4
+
_^
+ 4T
<f>p
Ko
-r
a^
6^
cf)p
in turn the values 1, 2,
system of equations
dm has a
c...
^',
fjli
and
Hence,
zeros
<j>p
1 of its zeros
between
h-
and
Since there are

obtained
when
theorem due to
(j),i
so the system
by which
is
between
a^
1, 2, ...
and
h"
(p
(jo
1,
= l,2, ...r-1)
= ?',?+ 1, ... to)
a
Lame
function exists
and the remaining
vi
c'-.
m + 1 Lame
functions of the specified type, they are
given in turn the values
1, 2, ... to
and
all
this is the
Stieltjes.
Acta Mathematica,
The zeros
-a'^,
\i. (1885),
-b^, -
c^
pp. 321-326.
are to be omitted from this enumeration,
6^, d.^,
...
^,
only being
taken into account.
W. M. A.
6^
6^, 6^, ...
of equations determining
solution for which
r has any of the values
if
give
m.
j-a'<dp<-b%
\ b"< dp < C-.
with r
maximum
precisely the system
is
are determined (cf | 23*21-23'24)

Oi, 02, ...
+T
4
'^3
for a
n_
111
to say
fCi
a log
its
3"62).
36
562
An
that
was given by
interesting statical interpretation of the theorem
if wi
+3 particles which attract one another according
move on the
namely
Stieltjcs,
to the law of the inverse distance
are placed on a line, and three of these particles, whose masses are
fixed at points with coordinates a-, b-, c^, the
[CHAP. XXIII
Ki
-r
f2+
>
''s
+ ti
remainder being of unit mass and
^^'^
free
and the
positions of equilibrium of the system are those in which the coordinates of the moveable
particles are 61,6-2, ... 6^, i-e. the values of 6 for which a certain one of the Lame functions
of degree 2 {m + ki + K2 + <z) vanishes.
to
Example.
then log
line,
the gravitational potential of the system
is
Discuss the positions of the zeros of polynomials which satisfy an equation
of the type
"*
d6'^
rti
e-a, d6
*
'
^~
,^
'
{6-a,)
s=l
where
(f)r-2{6) is
a polynomial of degree r 2 in 6 in which the coefficient of 6^~-
is
- m {m + r - 1 2
a^}
.-=1
m being a positive
integer,
and the remaining
coefficients in <f>r-2{6) are
determined from
the consideration that the equation has a polynomial solution.

(Stieltjes.)
Lame functions of the second kind.

The functions En^ (^), hitherto discussed,
23'47.
of the first kind.
It
is
are
known
as
Lame
functions
easy to verify that an independent solution of Lame's
equation
duis
the function
i^,i'"
(^) defined
i^,i'" (1^) is
From
by the equation*
= (2H-l)^'"(f)/;jj;^,
/..(f)
and
'
termed a Lam6 function of the second kind.
this formula it is clear that, near u
F^^ (^) =
(271
1)
u-^
{1
"
(lOl
u-''
I
= 0,
{1+0 {u)]
du
= u''+'
{1
(u)],
Jo
and we obviously have
E,r{^) = u-[l + 0{u)].

It
is
the
clear
first
satisfied by
It
is
from these results that Fn"^ (|) can never be a Lame function of
and so there is no value of Bn"^ for luhich Lame's equation is
kind,
two Lame functions of the first kind of
possible to obtain an expression for
different species or types.
Fn'\^) which
is
free
from
quadratures, analogous to Christoffel's formula for Qni^), given on p. 333,

example 29.
shall give the analysis in the case when En"^ (|) is of the
We
first species.
The only
at a set of points
u^,
u.,,
irreducible poles of l/{^,j"'(^)j^ qua function of u, are

...
Un which are none of
* This definitiou of the function

p. 194.
i<''" (4)
is
them periods
or half periods.
due to Heine, Journal fUr Math. xxix. (1845),
23-47, 23-5]
lame's equation
Near any one
of these points
En'" (I)
/.-i
{U
Ur)
563
we have an expansion
h (W - Urf +
^'3
{u
of the form
- U,)^ +
...
and, by substitution of this series in the differential equation,
it is
found that
^2 is zero.
Hence the
principal part of l/{'n"* {^)\' near
iir
is
1
k]-
and the residue
(u
'
Uff
is zero.
Hence we can
Ar such
find constants
that
{E,r(^yr- I Ario(u-u,.)
r=l
has no poles at any points congruent to any of the points
a constant A, by Liouville's theorem, since

of
it
is
iif
it is
therefore
a doubly periodic function
M.
Now
Ey^ (I)
If
dn
fit
the points Uf can be grouped in pairs whose
is
an even function of
we take
iin-,-
'';+i,
--~:^ =
Jo i^n
sum
is
zero, since
u.
we have
Au-
2 Ar{^{u-Ur) +
K{u
+ U>)]
r=l
{g)i
= AlC -
2^{u)
t
r=l
-ff'^'']
Ar-t= lip{y-)-^{Ur)
and therefore
where
'?<î,j_i(|^) is
Example.
=!
a polj^nomial in | of degree ^n
1.
Obtain formulae analogous to this expression for
i^,
{^) when
E,^^^ (^) is
of
the second, third or fourth species.

23'5.
Lame's equation in
association tvith Jacobian elliptic functions.
All the results \Yhich have so far been obtained in connexion with
Lame
functions of course have their analogues in the notation of Jacobian elliptic

functions, and, in the hands of
Hermite
(cf.
23*7 1),
elliptic functions in the discussion of generalisations of
produced extremely interesting

Unfortunately
all
it is
the use of Jacobian
Lame's equation has
results.
not possible to use Jacobian elliptic functions in which
the variables involved are
real,
without a
loss of
symmetry.
362
564
[CHAP. XXIII
The symmetrical formulae may be obtained by taking new variables

7 defined by the equations
fa = iK' + u V(ei - e-^),
= iK' + w V(ei -
1^7
and then the formulae of
a, /3,
e-,),
23"31 are equivalent to
A;-
(f^-/k') V(ft"
{ijk')
\/{a~
c")
sn a sn
/3
sn
7,
C-) en a en ^ en 7,
\/(a- C-) dn a dn /8 dn 7,
.
the modulus of the elliptic functions being
V W-cVThe equation
stant
of the quadric of the confocal system on which a
is
con-
is
X'
Y^
Z-
- b') cn^ a ~ (a^ - c-) dn" a ~

is an ellipsoid if a lies between iK' and K + iK'
the quadric on which
constant is an hyperboloid of one sheet if ^ lies between K + iK' and
"
(a2-62)sn2a
This
yS is
-fir
(a^
and the quadric on which 7 is constant is an hyperboloid of two sheets if

between
and
and with this determination of (a, ^, 7) the point
lies
{x, y, z) lies
in the positive octant.
It has already
been seen
23*4) that, with this notation, Lame's equation
assumes the form

--T-v
{^^
{n
1)
A,'-
sn- a
+ A]
A,
and the solutions expressible as periodic functions of a will be called* -'" (a).
The first species of Lame' function is then a polynomial in sn^ a, and generally
the species
may be
1,
defined by a scheme analogous to that of
23*2,
sn
a,
en a dn
en
or,
dnasna,
a,
sn a en a dn a
(sn-
oc
sn-Qp).
J}
dn
23'6.
a,
sn a en
a,
The integral equation
satisfied by
Lame
functions of the first and
second species'^.
We
shall
now shew
that, if En'^(a)
is
any Lamd function of the
first
species (n being even) or of the second species (n being odd) with sn a as a

* There is no risk of confusing these with the corresponding functions i-'/'
(t).
t This integral equation and the corresponding formulae of 28 -62 associated with ellipsoidal
harmonics were given by Whittaker, Froc. London Math.- Soc. (2) xiv. (1915), pp. 260-268.
Proofs of the formulae involving functions of the third and fourth species have not been
previously published.
23*6]
lame's equation
then E'"'(a)
factor,
is
a solution of the integral equation
E,r () = X
Pn
J
where \
To
(k sn a sn 6)
-2K
one of the 'characteristic numbers'
is
establish
565
this
result
E,r (0) dd
( 11-23).
we need the lemma
that
P,i
(A:
sn a sn ^)
is
annihilated by the partial differential operator
3^2
- 9^ - "
(^
To prove the lemma, observe

0, we have
1)
that,
^''
(sn' a
when
- sn^ d).
/x is
written for brevity in place
of k sn a sn
{a|-î^'^^^'-'""''^^^>
en- 6 dn- 6 sn- a] Pn" (/i)

+ ^k" sn a sn ^ (sn^ a - sn^ 0) P,,' (/x)
= k' (sn^ a - sn^ 6)
- 1 ) P,/' C/.) + 2/iP,/ (^)]
^2
= (sn- a sn^ 6) n (n + l) P
k- (en- a
dn- a sn- 6
[(fj,^-
(jj,),
when we
use Legendre's differential equation
( 16-13).
And
the
lemma
is
established.
The
result of applying the operator
^2 - n
+ l)k"sn-oc-A '
(n
to the integi'al
r2K
Pn
J
is
now seen
\
to
be
-n{n + l)
k' sn^ a
- .4
,,'
-2K (ca
P (k sn
a sn 6) E^"^ (6)
dO
^,
\^.-^^{n + \)k-s,nê-AyvPn{knasne) E,r{e)dd,
and when we integrate twice by parts

-B
E^T (0) dO
(k sn a sn 6)
~2K
PAfcsnasnB)
^^^ ^^^ _
this
^^^ ^^ ^^ ^ ^^ ^^
becomes
dEê)
2K
_-2K
+j
Hence
it
Pn (^n asn d) Uj,-
n (n
1) k' sn' d
- A .^ [
En'"" {d)
follows that the integral
riK
Pn
J
is
(k sn a sn 6) En"" (0)
d9
-2K
annihilated by the operator

d^
T2
{n
l)k^ sn- a
J.,l"^
.dd
= 0.
566
and
it is
evidently a polynomial of degree n in sn-
has
onlj'^
one integral of this type*,
of
not zero
En^'^ (a) if it is
and the result
Since Lame's equation
a.
follows that the integral is a multiple
it
established.
is
does not appear to have been proved that the only vahies of
It
f{a)
=\
[CHAP. XXIII
A, for
which the equation
J\{kHnaKn6)f{6)dd
["^
has a solution, are those which make/(o) a solution of Lame's equation.
Example
Shew
1.
that the nucleus of an integral equation satisfied by
Lame
functions
of the
first
species {n being even) or of the second species {n being odd) with en a as a
factor,
may
be taken to be
/
Example
Shew
2.
of the first species
that the nucleus of an integral equation satisfied by
Lame
functions
being even) or of the second species {n being odd) with dn a as a
(?i
may be taken
factor,
(-p en a en 6
to be
PYpdnadn^y
Tlie integral equation satisfied
23'61.
fourth
Lame functions
by
of the third and
species.
The theorem analogous
to that of 23'6, in the case of
the third and fourth species,
is
that
any Lame
Lame
(n being odd) or of the third species (n being even) with en a

satisfies
functions of
function of the fourth species
dn a
as a factor,,
the integral equation

r2K
E,r (a) = X
en a dn a en 6 dn dPn' {k sn a sn 6) E/'' (0) dO.

J -2 A'
The preliminary lemma
that the nucleus
is
en a dn a en d dn
dPn'
To
lemma
verify the
^-^,
ctf-
- n {n +
6),
by the operator
like the nucleus of 23"6, is annihilated

TT-,
dot-
{k sn a sn
l)
A'-
(sn- a
- sn- 6).
observe that
|^JcnadnaP/(/^^snasn^)}
k- cn^'
and
.
dn''
a sn- ^PJ''
en a dn a (dn- a
(/u.)
- 3^- sn a en a dn
k- en- a
a sn
(dn- a
k' en- a)
Pn" (fi)
Pn" (fi),
sn" a)
so
I
.
(en a
dn a en ^ dn 6Pn {k sn a sn
= kcnadnacne dn 6 (sn^ a - sn'^ 0)

= k- en a dn
= k- n {n +
*
4k'^
The
(sn a)--.
{(,x''
0)\
PJ^"
d^
a en ^
1)
dn ^
en a dn
other solution
(sn- a
- sn- ^) ^
en 6 dn 6 (sn-
when expanded
a.
{(m-'
sn'-
6)
in descending
(/i) -f
GfxPn" (^) + GPn" (/*)]
1) Pv/ (a^)}
Pn"
(/a),
powers of sn a begins with a term in
. .
23 '61, 23 "62]
ellipsoidal hahmonics
and the lemma
now
equation
The proof
established.
is
567
En^ (a)
that
Example 1. Shew that the nucleus of an integral equation which

functions of the fourth species ( being odd) or of the third species
sn a dn a as a factor, may be taken to be
sn a dn a sn 6 dn 6 P"
Example
Shew
2.
sn a en a as a factor,
( -77
may be
{ii
is satisfied
being odd) or of the third species
is satisfied
(/i
by Lame
being even) with
taken ta be
sn a en a sn 6 en 6P,l'
{ j,
dn
dn ^
Obtain the following three integral equations satisfied by

and of the third species (?i being even)
3.
by Lame
being even) with
{ii
en a en ^ )
that the nucleus of an integral equation which
functions of the fourth species
Example
the integral
satisfies
follows precisely as in the case of the integral equation of 23"6.
Lame
functions
of the fourth species {n being odd)
k"
(i)
(ii)
..'-
E.r () -X c dn / ^^^ P. (i. ., sn )

^^
,.
{jjJj^ '-^] M,
-ic=,.i.V'W = ^-.,nd/^^P,.(fcnc)^{^-^/^>}o(,
(iii)
I.-'
in' a
Er M = Xr- s a c /^^î>.
in the case of functions of
even
Q. dn dn )
^ {^J^^ ^} M
the functions of the different types each satisfy one
ordei-,
of these equations only.
Integral formulae fur ellipsoidal harmonics.
23'62.
The
integral equations just considered
make it possible to obtain elegant

Gn^ {x, y, z) and of the corre-
representations of the ellipsoidal harmonic
sponding homogeneous harmonic
From
is
H^^ {x,
the general equation formula of
expressible in the
y, z) in
terms of definite integrals.
18"3, it is
evident that
Hn^ {x, y, z)
form
Hn''' (x,
y,z)=
(x cost
+y
sin
+ izTf(t) dt,
J TV
where y(^)
Now
is
a periodic function to be determined.
the result of applying Niven's operator D- to {x cos
n (n
and
so,
1 )
(a- cos"-
by Niven's formula
+
(
b'-
sin^
23*25)
c'-)
we
(x cos
+ y sin t +
+ y sin t +
t^)" is
izy^~^,
find that Gn"^ {x, y, s)
is
expressible
in the form
n{n-l)(n-2)(n-S)
+ 2.4(2n-l)(2u-3)
^'
^/{t)at,
[CHAP. XXIII
568
where
*^l
= xcost + y
sin
is,
^ = V{(' - C-) cos- + {b-- C-) sin^

t
t],
so that
Now
write sin
usual, given
= cd
the modulus of the elliptic functions being, as
^,
by the equation
,
b'
a- C-
tt"
The new limits of integration are - ^K and K, but they may be replaced
by - 'IK and 'IK on account of the periodicity of the integrand.
It is thus
found that
{-^
e"
where
to
(^
(^. y, ^)
T^
= j_^^.
/^'a;sn^
i^ (
+ ycn^+>dn^\
:j^_-^^
(^) is a periodic function of 6,
j #> (^)
independent of
x, y, z,
''<'.
which
is,
as yet,
be determined.
If
we express the
ellipsoidal
harmonic as the product of three Lame
functions, with the aid of the formulae of
235 we
find that
rlK
E^^ (a) E,r
(/3)
Pa
E,r {l)=G\
(/x) <^
(0) dO,
J -2A"
known constant and
where
is
/x
k-
sn a sn
/3
sn 7 sn ^
- {k-/k'-) en a en /? en 7 en
- (l/'k'-) dn a dn y8 dn 7 dn 6.
harmonic is of the first species or of the second species

type, we now give /3 and 7 the special values
If the ellipsoidal
and
first
^ = K,
K + iK',
and we see that
Pn(ksnasnd)(f3{e)(W
['
J
-2 A'
(^) is a solution of Lame's

a solution of Lame's equation, and so, by 23-6,
multiple
a
E,i'"
than
(6).
of
equation which can be no other*
is
Hence
it
follows that
l"-^
where X
is
T.
/'k'x
sn0+
en $
iz
dn ^\
,^
,^,
a constant.
* If (p{e) involved the
second solution, the integral would not converge.
7^
23-63]
If
569
be of the second species and of the second or third type
Gn!"^ {x, y, z)
we put
or
/3
0,
/3
0,
= iiT +
7 = /i
iK',
and we obtain anew the same formula.
respectively,
It thus follows that if
be any ellipsoidal harmonic of the
z)
(th'" {x, y,
first
or second species, then
GrT {oo, y,z) = \
""
f-2
^'
(x, y, z)
E,i- {6)
^^'
= X.
where
[x
23'63.
Pn (/^)
cie,
A'
(k'x
sn6 + y end +
I nteg7'al formulae for
+ 3/ en
(^'^ sn ^
iz
ellipsoidal
dn
6?
d)/\/{b-
iz
dn df En''' (0) dd,
c").
harmonics of the third and fom^th
species.
In order to obtain integral expressions for harmonics of the third and

fourth species,
we turn
namely
to the equation of 23'62,
,' (a) En'-
(/8)
=C
En^- (7)
r
J
Pn (yu)
cf>
(0) d0,
-iK
where
fi
= A;^sn a sn /3 sn 7 sn ^ {k-/k'-) en
this equation
is
satisfied
Suppose now that
We
it
(l/k'-) dn a dn /S dn 7 dn ^
en 7 en ^
/3
by harmonics of any
'/'^ (a) is
the third species so that
a en
species.
of the fourth species or of the first type of
has en a dn a as a factor.
next differentiate the equation with respect to
/3
and
7,
and then put
^ = K,y = K + iK'.
It is thus
found that
d
En'"(^)
En'îct)
d/3
V = A'+iA"
i.-=A-L7
2A
K
dPn(f^)
Now
_
87
(0) d0.
C'
?/3S7
{ilk')
dn a dn
(?
/S
= K, y = K+iK'}
dn 0Pn
(/x),
_ y = A'+?A''
SO that
d^dy J{^=K,y=K+iK:')
Hence
r
J
^^
en a dn a en ^ dn
en a dn a en ^ dn 0Pn" (k sn a sn ^)
0Pn" (k
(^
(0)
sn a sn
0).
d0
-2A
is
a solution of Lame's equation with en a
(ji
(0)
dn a
as a factor
can he none other than a constant multiple of.E^^-
(a).
and
so,
by
23'61,
570
We
have thus found that the equation
G,r {oc, y,z) = \\

J
by any
is satisfied
Pn
(fi)
E,r (0)
harmonic which has en a dn a as a factor
ellipsoidal
^^^
=\
(x, XL z)
(le
-IK
corresponding formula for the homogeneous harmonic

Hn'"
[CHAP. XXIII
/.'
J-
X sn
e-^xi en 6
Example.
Shew that the equation of this section
harmonics which have sn a dn a or sn a en a as a factor.
is
the
is
+ t>dn ^) E.îO) dd.
satisfied
by the
ellipsoidal
Generalisations of Lame s equation.
23"7.
Two
obvious generalisations of Lame's equation at once suggest themIn the first, the constant B has net one of the characteristic values
selves.
Bn^, for which a solution
is
in the second, the degree
expressible as an algebraic function of

is
generalisation has been fully
{u)
and
no longer supposed to be an integer. The first

dealt with by Hermite* and Halphenf, but the
only case of the second which has received any attention

half of an odd integer; this has been discussed
is that in which n is
by Brioschij, Halphen and
Crawford
II
We
shall
now examine the
solution of the equation
^_ =
where
is
arbitrary and n
is
+ l)
[n{n
io{u)
-^
B]
i\,
a positive integer, by the method of Lindemannconnexion with Mathieu's equation ( 19'5-
Stieltjes already explained in

19-52).
The product
of
'^ ~ ^
by
19"52.
4 (^
The
e,)
(I
any pair of solutions of
^'^ ^'^
^^
^"^
^"^
+ ^^
this equation is a solution of
S~
-"
algebraic form of this equation
e,)
(^
.3)
-4
{(2
+ 3 (6^ -
^'^
i^,)
:J:
11.
^-
ti89-(;95,
go
-In (;i
1)
728-732, 821-826.
Comptes Remlus, i.xxxvi. (1878), pp. 313-315.

ii. (Paris, 1888), pp. 471-473.
Quarterly Journal, xxvii. (1895), pp. 93-98.
X = 0.
be taken to be
(Paris, 1888), pp. 494-502.
Fonctions Elliptiques,
II
^ = ^'
'^
+ - 3) ^ + i^l
Comptes liendus, lxxxv. (1877), pp.
^''^
is
If a solution of this in descending powers of ^
t Fonctions Elliptiques,
^''
^^
'
237]
lame's equation
the recurrence formula for the coefficients
c,.
571
is
r + I) (2/1 - ? + 1) Cr
= ( - r + 1) {12e (n - r) (n - r + 2) - 4eo (n- + n - 3) - 4J5} c,_i
- 2 (n - r + ] (n - r + 2) (ei - e^) (e. - 63) (2h - 2r + 3) c,_2.
Write r = 4- 1, and it is seen that Cn+i =
then write r = n + 2 and Cn+2 =
and the recurrence formulae with r > n + 2 are all satisfied by taking
4r
(n,
?i
71
Hence Lame's generalised equation always has two

is
solutions tuhose product
of the form
2c,(^-e,)-'-.
r=0
This polynomial
may be
written in the form

n
n
where
i, a^, ...
{^j(w)-&>(r)l>
a^ are, as yet, undetermined as to their signs
and the two
solutions of Lame's equation will be called Aj, A,.
Two
cases
when A1/A2
when Aj/A^
is
not
multiplied possibly by expressions of the type ^
e^,
arise, (I)
is
constant, (II)
constant.
The
(I)
first
case
is
easily disposed of; for unless the polynomial
r=l
is
a perfect square in
^ ^2,
63,
^,
then the algebraic form of Lame's equation has an indicial
equation, one of whose roots
is
at one or
more of the points
|^
î>
(a,-)
and
this is not the case ( 23'43).
Hence the polynomial must be a square multiplied possibly by one or

more of ^ e^, ^ 62, ^ e-s, and then Aj is a Lame function, so that B has
one of the characteristic values 5,/"; and this is the case which has been
discussed at length in 23-1-23-47.
(II)
In the second case we have
du
where
is
a constant which
rf
log A,
so that
^^^^
is
19*53)
du
'
Then
not zero.
d log A2
du
d log Ai
d log A2
du
d log Ai
du
_ 1 dX
- 2^ ^^^
_ 26
du
dX
du
^logAa^
du
2X
dX
du ^
572
On
integration,
A,
Again,
we
if
may
see that Ave
"sve
take
= VXexp|-(>[~|.
differentiate the equation
dA,_
~
Ai du
1
we
[chap. XXIII
dX
(i
'2X du
'
find that
{1 dA,\^
d'A,
A,
dir-
Ai du
2X
d'X
fdXy
g dX
dti?
2X'
du
X^ du
'
and hence, with the aid of Lame's equation, we obtain the interesting
formula
If
now
that, \i
^,.
{a,),
jf)
we
from this formula (when multiplied by X-),

cir, then
find
u be given the special value
m-
'dXV
We
now
fix
the signs of
aj, an, ... cin
by taking
'dX\
fdX\
2g
2(
(.)
And
then,
we put 26/X,
if
g-ita
function of
|,
into partial fractions,
it is
seen
that
2(J
(a'
(n
'^
and therefore
Ai
n[i^(H)-li>(a.)i
X exp
whence
it
{log
follows that ( 20'53,
Ai= n
r=l
and
^2=
n
,.=1
The complete
constant B.
a-
(ctr
it)
example
(ar
lO" (z<) (7
log
(a,
- n) -
2m ^ (a,)|
1)
-m)
;x[exp|-?t 2^^(a,)|,
(a,
I|o-(iOo-(a,)J-exp|M
^
-
^=1
?(a,.)h
'^
solution has therefore been obtained for arbitrary values of the
lame's EQUATION
2371]
23'71.
We
The Jacobian form of tJie generalised Lame equation.
shall
now
construct the solution of the equation
- =
for general values of
The
where
[n {n
+^
1) k' sn^ a
A,
A, in a form resembling that of
solution which corresponds to that of
23"6
23*6.
is
seen to be*
are constants to be determined.
p, a^, a^, ...cun
On
573
differentiating this equation
Ada
,-=i|H(a
it is
seen that
B\a)\^^
a,)
= X
{Z (a
+ iK') - Z
a,
+p+
(a)}
\mrilK,
r=l
,
^^ ^^^'
1 C^Â
(1 dA\-
JA
^-
^1
.w
^^"" ^"
.-X
-rr/,
^^
+ ^^^ ^ -
i
'^^"
""
and therefore, since A is a solution of Lame's equation, the constants

Oo, ... a,i are to be determined from the consideration that the equation
n {n
1) k- sn- a
-\-
A=
{dn- (a
a,-
+ iK')
p, a^,
dn- a
r=l
2
r=
is to
be an identity
that
is
{Z(a
+ ar+
iK')
- Z {a}]
-f
^ niri/K
to say
n'k" sn- a
71
+A + 2
cs-(a
+ a,-)
2 {Z
(a
+ a, + t7f - Z (a)] + p +^mrilK

')
r=l
Now
both sides of the proposed identity are doubly periodic functions of
a with periods 2K, 2iK', and their singularities are double poles at points
congruent to iK', Qi, a.,, ... a^^; the dominant terms near iK' and
;.
are respectively
n-
(a+TKy
in the case of each of the expressions
The
1
'
~ (a +
arf
residues of the expression on the left are
p, !, Ko, ...
all
zero and so, if
we choose
On SO that the residues of the expression on the right are zero,
* This solution
was published
at the Ecole polytecbnique.
in 1872 in Hermite's lithographed notes of his lectures delivered
574
it will
made
We
n+
thus obtain
these equations are not

It is easy to
theorem that the two expressions

by proper choice of A.
follow from Liouville's
constant which can be
{Z{a
[CHAP. XXIII
by a
differ
to vanish
2 equations connecting p,
a^,
a..,
...
a,,,
with A, but
independent.
all
o^,
prove that, near
+ a, + iK') - Z {a)\ +
îîri/K
r=l
-^
I'
(Op
- a, +
Hi')
+ nZ
where the prime denotes that the term
I
r=
+ a, + iK') - Z (a)] + p +
[Z (a
(a,)
which
for
i (n
jj
1) -nilK
is
?'
omitted
(a
a,-),
and, near
i n-ni\K
=
ft
Hence the
^+ i
+ til
>.
+ p+0(a+27r).
Z(,)
=!
residues of
{Z (a
will all vanish if p, a,,

'
Z(a,)
(ft,,
a,
+ i7i
... ft
a.^,
i'
ft,
')
- Z (a)} +
| ??.7rt7A"
are chosen so that the equations
+ iK') + '/iZ (ft,) + p + ^ ( -
1)
7r?:/ii
= 0,
+P=
V}-=1
are
all satisfied.
The
last
equation merely gives the value of
p,
namely
- 2 Z (ft,),
and,
when we
substitute this value in the

2' [Z
where r
= 1,
when
7t-A''-
sn- a
i,
?i
a.,,
...
ft, -f-
By
2, ... n.
of these equations
and,
(ftp
is zero,
51
system,
example
2,
the
sum
so they are equivalent to n
any values which
cs- (
we
find that
+ Z (ft,) - Z (a,,) + t^TTzyZ] = 0,
iK')
22*735,
a^ have
+ -4 +
first
satisfy
of the left-hand sides
equations at most
them, the difference
ft;-)
|Z
(ft
ft,
IK')
-Z
(ft)
-Z
(ft,)
+ \irijK\
lame's equation
2371]
is
By
constant.
n
taking a
+A+ S
cs-
Of,.
i.e.
2
)
if
S en
seen that the constant
0, it is
r=l
Op,
a,,
[Z (a,
Of,,
ds
cif,.[-
reduce the system of n equations
{up
-ar + iK') + Z
and
-Z
(a,)
(a,)
zero if
^Tri/K]
ot,.
^.
with the notation of
22 '2,
if
functions of
easy to see that
2, it is
(op)
ns-
r=l
be denoted by the suffixes
-Z
iK')
is
=!
ij-=l
We now
575
+ ^ Tri'/A'
= Z (a - Or + iK') + Z(ar) -Z{ap + iK') + Cj 0?i /Si

= sn (op + iK') sn a^ sn (ap + iX' o^) + Ci /si
k'^
ci?i
2
Sj
sn (op
Cidi
n,.)
Si
_ SiCiO?i+g2C2C^2
~
V-2^
Consequently a solution of Lame's equation
--=
{71.
I
act-
A= n
IS
provided that
!, a2>
(n
+ l)^-sn-a+ A]
'
@(^)
exp-aZ(a,)}
n be chosen to satisfy the
n independent equations
comprised in the system

/
^, sn
ttp
en
ttp
dn
ctp
sn^ Op
Jî
en
ttr
ds
+ sn en o^ dn
sn^ ar
of^
a,.
ns^
a,.
=A
;.
r=l
and
if this
solution
solution of Lame's equation
n
The
not doubly periodic, a second
is
is
"H(a-a,)
@ (a)
exp
[otZ
(,.)]
existence of a solution of the system of n
=
-\-l
0.
equations follows from
23-7.
REFERENCES.
G. Lam^, Journal de Math.
385
les
ii.
(1837), pp. 147-188; iv. (1839), pp. 100-125, 126-163,
vin. (1843), pp. 397-434.
Lecons sur
surfaces isothermes (Paris, 1857).
les
Lemons sur
351-
fonctions inverses des transcendantes

les
et
coordonnees curvilignes (Paris, 1859).
E. Heine, Journal fUr Math. xxix. (1845), pp. 185-208.
Theorie der Kugelfunctionen,
ii.
(Berlin, 1880).
C. Herhite, Comptes Rendus, lxxxv. (1877), pp. 689-695, 728-732, 821-826

(2) IX. (1878), pp. 21-24.
Oeuvres Mathematiques (Paris, 1905-1917).
Ann. di Mat.
576
G. H.
F.
Halphen, Fonctions
Elliptiques,
LixuEMANN, Math. Ann.
(Paris,
il.
[CHAP. XXIII
li
xix. (1882), pp. 323-386.
K. Hecn, JIath. Ann. xxxili. (1889), pp. 161-179, 180-196.

L.
Crawford, Quarterly Journal,
W. D. NiVEN,
Phil. Trans, of the
xxvii. (1895), pp. 93-98; xxix. (1898), pp. 196-201.
Royal
Society, 182
(1891), pp. 231-278.
A. Cayley, Phil. Trans, of the Royal Society, 165 (1875), pp. 675-774.
G. H. DAR^VIN, Phil. Trans, of the Royal Society, 197 A (1901), pp. 461-557
pp. 301-331.
198 a (1901),
1.
Obtain the formula
Gn
{X, y,
^=^^,
/J^^"
Pn
(J)
e-'^dn.Hn {x,
y,
z).
(Niven, Phil. Travis. 182 a (1891), p. 245.)

2.
Shew that
rr
fl
l\
^(-r.{2n)l
Hn{x,y,z)
(Hobson, Proc. London Math. Soc. xxiv.)

3.
Shew
that the 'external ellipsoidal harmonic' F^'"
($) En'"
(rj) E,,''"
(C) is
a constant
multiple of
"Xcx' dy'
82;
(Niven
4.
(2yi + 3) (2?i + 5)
--J ^{x^+y^ + z^)'
and Hobson, Proc. London Math. Soc. xxiv.)
2.(271 + 3)^ 2. 4
Discuss the confluent form of Lame's equation when the invariants ^2 and g^ of the
made to tend to zero; express the solution in terms of
Weierstrassian elliptic function are

Bessel functions.
(Haentzschel, Zeitschrift fiir Math,

5
If
denotes
t!
G(a)^
exp
[{X
-Z
Lamd's equation has a solution which
dn-l^
where
X-
and
sn^
p.
is
(/x)} a],
where X and
/x
und
Phys. xxxi.)
are constants, shew that
expressible as a linear combination of
cln-3^
dn-b^
are algebraic functions of the constant A.
(Hermite.)
6.
Obtain solutions of
-^=12Psn22-4(l + F)5V(I-F +
Z-').
dz-
(Stenberg, Acta Math, x.)

7.
Discuss the solution of the equation
2(2-l)(2-a)^ + [(a + ^ + l)s'^-{n + /3-g + l+(y + S)a}i+ay]^^ + ^(5-j)3/ =

in the
form of the series
Gn{q){^hY
i=i!y(y+l)...(y+)'
l+a,3 2
Gi{q) = q,
where
6-'
+ 8)a} j-ay,
(?2(?) = a/3?2 + {(a + /3-8+ l) +
=
-l)o}
+ a/3j]6'(g)
[ {(a + /3-8 + n) + (y + S +
(?)
-(a + -l)(/3 + ?i-l)(y + n-l)?i.6'_i(j).
(-y
(Heun, Math. Ann. xxxiii.)
LAME S EQUATION
8.
Shew
that the exponents at the singularities

(0,
1-y),
1-S),
(0,
577
0, \, a, cc
(0,1-0,
of
Heun's equation are
(,^),
y + 8 + e = a + l3 + l.
where
(Heun, Math. Ann. xxxiii.)

9.
Obtain the following group of variables for Heun's equation, corresponding to the
group
\
for the hypergeometric equation
2-1
[CHAP. XXIII
578
pô^^ded that
ir{u-r + ^)l>,. + ['^-2e.>{-2n-i- + \){r-l)-4ein{'2n-l) + -iB]b,_i
and
is
+ 4(ei-e2)('2-t'3)(2-r + 2)(-/-+S)6r-2 = 0,
= 0^,j^.j^
so determined that
(Brioschi
Shew
14.
that,
if
an odd integer, a sohition of
n is half of
xpressible in finite form
tlie
and Halphen.)
equation of example 11
is
Z'="i'V(C-eo)"-^-^
provided that
4/) (n+jt?
+ A) bp - [Ue., {n -p+^^ [ +p - *) - 1<'2 (2 - 1) + 45] //p_i

+ 4 (ei - f 2) - ^3) in-p + '^){p-l)
(-.'
and
1=0
?>'
is
?>',,
-2 =
the equation which determines B.

(Crawford.)
"With the notation of examples 13 and 14 shew that,
15.
V=
the equations which determine
/)o,
61, ...^>
_i
and deduce
half of an odd integer)
is
- )" (^1 -
Cq, Cj,
that, if
^'2)''
{e.
.<^_j!.
if
- es)" c_^_.j.
'Î'e
identical with those
which determine
one of the solutions of Lame's equation
expressible as an algebraic function of
{v\ so also
(in
is
which n
is
the other.
(Crawford.)
Prove that the values of
16.
are
determined
in
example 13 are
re;\l
when
e^, e^
and
real.
Shew
17.
that the complete solution of
A = {iO'{U)r^-{Ap{iu) + B],
is
where
A and B
are arbiti-aiy constants.
(Halphen,
Shew
18.
Jle'm.
par
divers savants, xxviii.
(i),
(1880), p. 105.)
that the complete solution of
ig = |Fsnâ-i(l4-F)
\ = {sn^{C-a)cQ^{C-a)dn^{C-a)r^{A + Bsnn.iC-a)},
is
where
.-1
and
are arbitrary constants
and C=2K+iK'.
(Jamet, Comptes Rendus, cxi.)
6-3
APPENDIX
THE ELEMENTARY TRAXSCEXDEXTAL FTXCTIOXS
A*l.
It
On
certain results
assumed in Chapters I-IV.
was conveuient, in the
fii-st
four chapters of this work, to assume some of the
properties of the elementary transcendental functions, namely the exponential, logarithmic
and
circular functions
it
was
make
also convenient to
use of a
number
of results which
the reader would be prepared to accept intuitively by reason of his familiarity with the
means
geometrical repi-esentation of complex nambere by
of points in a plane.
was assumed
lim (exp i) = exp (lim *}, and
Argand diagram made it appear plausible
that the argument of a complex number was a many-valued function, possessing the
property that any two of its values diffei-ed by an integer multiple of îr.
To take two
instances,
(i)
( 2"7) that
it
the geometrical concept of an angle in the
(ii)
The assumption
of results of the first type
was
base ai'ithmetical results on geometrical reasoning.

of geometry on a satisfactory basis,
arithmetic, but
it is
it
is
clearly illogical
it
wjis also illogical to
For, in order to put the foundations
not only desirj^ble to employ the axioms of
also necessary to utilise a further set of
axioms of a more definitely
geometrical character, concerning properties of points, straight lines and planes*.

fm-ther, the arithmetical theory of the logarithm of a
And,
complex nimiber appears to be
a necessary preliminary to the development of a logical theory of angles.
Apart from this, it seems unsatisfcictory to the cêsthetic taste of the mathematician to
employ one branch of mathematics as an essential constituent in the structui-e of another
particularly when the former has, to some extent, a material basis whereas the latter is of
a purely abstract nature f.
The reasons
for pursuing the
somewhat
the earlier part of this work, were,
firstly,
illogical
and unaesthetic procedure, adopted
in
that the properties of the elementary transcen-
dental functions were required gradually in the com"se of Chapter n, and
seemed
it
imdesirable that the coiuê of a general development of the various infinite processes
should be frequently interrupted in oi"der to prove theorems (with which the reader was,
in all probabihty, already famihar} concerning a single particuliU* function
and, secondly,
that (in counexioji with the assumption of results based on geometrical considei-ations)
a pui'ely arithmetical
mode
of development of Chapters l-iv, deriving no help or illus-
trations from geometrical processes, would have very greatly inci-eased the difliculties of
the reader unacquainted with the methods and the spirit of the analyst,
* It is not our object to give any account of the foundations of geometry in this work.
They
are investigated by various writei-s. such as ^Tiitehead. Axioms of Projective Geometry (Cambridge
1906) and Mathews, Projectile Geometry ^London, 1914). A perusal of

and xxv of the latter work will convince the reader that it is even more
develop geometiy in a logical manner, from the minimum number of axioms, than
Math. Tracts, no.

Chapters
i,
laborious to
it
is
i.
xx, xxii
to evolve the theory of the circular functions
by purely analytical methods.
complete
account of the elements both of arithmetic and of geometry has been given by Whitehead and
Bussell, Principia
Cf.
(note 2)
Mathematica (1910-1913).
Merz, History of European Thought in the Nineteenth Century, n. (London, 1903), pp. 631
and 707 (note 1), where a letter from Weierstrass to Schwarz is quoted. See also
Sylvester, P/ii7.
Mag.
(5), ii.
(1876), p. 307 [Math. Papers, lu. (1909). p.
-50].
APPENDIX
580
Summary
A'll.
In
of the Appendix.
general course of the Appendix
The
is
as follows
A*2-A-22, the exponential function
v5
From
defined by a power series.
is
this
combined with results contained in Chapter ll, are derived the elementary
It is then easy to deduce
properties (apart from the periodic properties) of this function.
corresponding properties of logarithms of positive numbers ( A'3-A'33).
definition,
Next, the sine and cosine are defined by power series from which follows the connexion
A brief
of these functions with the exponential function.
the formulae of elementary trigonometry
The
may be
derived
sketch of the manner in which
is
then given (^ A'4-A-42).
results thus obtained render it possible to discuss the periodicity of the exponential
and circular functions by purely arithmetical methods ( A"5, A*51).

In A"52-A'522, we consider, substantially, the continuity of the inverse circular
When these functions have been investigated, the theory of logarithms of
functions.
complex numbers
A'6) presents no further difficulty.

it is shewn that an angle, defined in a purely analytical manner,
which are consistent with the ordinary concept of an angle, based on
Finally, in A"7,
possesises properties
our experience of the material world.

obvious to the reader that we do not profess to give a complete account of
It will be
we have confined ourselves to a brief sketch

The developments have been given by writers-
the elementary transcendental functions, but

of the logical foundations of the theory*.
of various treatises, such as Hobson, Plane
Mathematics
A'12.
-1
will find it instructive to read Chax)ters i-iv
the following order
in
Chapter
Chapter
ii
Chapter
iii
Chapter
to the
ii,
Chapter
of
5^
1 "5
except the
first
of 2*61 (omitting the
end of
3-34
and
v^
two
paragi-aphs).
examples
in
,^
2-31-2-61).
3-5-3-73.
A-2-A-6 (omitting
A-32, A-33).-
2-31-2'61.
3-341-3-4.
iv, inserting
ii,
all
the examples of
Chapter in,
Chapter
end
to the
and the Appendix a second
(omitting +
The Appendix,
He
Hardy, A course of Pure
Series.
logical order of development of the elements of Analysis.
The reader
time
Trigonometry
and Bromwich, Theory of Infinite
A"32, A-33,
Aw
after 4'13.
2-7-2-82.
should try thus to convince himself that
(in
that order)
it is
possible to elaborate
a purely arithmetical development of the subject, in which the graphic and familiar
language of geometry J
* In writing the
is
to be regarded as merely conventional.
Appendix, frequent reference has been made to the
article
on Algebraic
Analysis in the Encijklopiidie der Math. Wissenschaften by Pringsheim and Faber, to the same
article translated
and
revised by
Molk
for the Encyclopedic des Sciences Math.,
and
to
Tannery,
Introduction h la Theorie dis Fonctions d'uue Variable (Paris, 1904).

t
The
properties of the argument (or phase) of a complex
number
are not required in the
text before Chapter v.
i E.p. 'a point
'
for
'an ordered number-pair,'
origin' for 'the set of ordered number-pairs
points of a straight line
the type
'
for
Ax + By + C =0,' and
'
(.r,
y)
'
the circle of unit radius with centre at the
which
satisfy the condition
the set of ordered number-pairs
so on.
(x,
y)
which
.x'-
+ y-=l,' 'the
satisfy a relation of
A-ll-A-21] THE ELEMENTARY TRANSCENDENTAL FUNCTIONS
The exponential function exp
"2.
The exponential
z.
function, of a complex variable
exp.= l + + |^ + |^ +
f;
This series converges absolutely for
lim
ratio test ( 2'36) since
{z\n)
defined bj the series*
2, is
= l+J^|;.
...
values of
all
= 0<1
581
and complex) by D'Alembert's
z (real
so the definition
is
valid for all values of
z.
Further, the series converges uniformly throughout any bounded domain of values of z
for, if
the domain be such that
when
/iJ
z is in
the domain, then
j(sVm!)I ^R"/nl,
and the uniformity
of the convergence
is
a consequence of the test of Weierstrass ( 3"34),

X
by reason of the convergence of the
1+2
series
{R"i'n
which the terms are indepen-
in
!),
n=l
dent of
z.
Moreover, since, for any fixed value of
from
(cf.
n, s'V
is
3'32 tha^t the exponential function is continuous for all values of z
3 '2), if z
be a variable which tends to the limit

lim exp
The addition-theorem
A*21.
From Cauchy's theorem on
2= exp
z, it
follows
and hence
we have
(,
f.
and
for the exponential function,
its
consequences.
multiplication of absolutely convergent series ( 2-53),
it
follows thatf
(exp 2i)(exp2,) =
(l+-^V|', + -)(l + ^ +
= 1 + 2i + 22
Zi^
+ 2z^Z2 +
-Yj-+
= exp (21 + 22),
so that exp
(21
+ 22)
!;'
+ ..)
Z-2^
+...
,
2",
can be expressed in terms of exponential functions of
21
and of
22
by
the formula
exp
This result
we
see
is
known
(21
+ 22) = (exp 2i)
(exp
z.^.
as the addition-theorem for the exponential function.
From
it,
by induction that
(exp2i) (exp
22) ... (exp2,i)
= exp(2i + 22+...+2),
and, in particular,
{exp 2} {exp
From
the last equation,
it is
there were such a value of
for, if
- 2)} =exp 0=1.
apparent that there

2,
is
no value of
2 for
which exp
since exp (2) would exist for this value of
2,
we
should have 0=1.
when x is real, exp^>0 for, from the

when x ^ 0, exp x= 1/exp ( - .r)>0.
It also follows that,
when
xÔ
* It
p. 167.
and,
was formerly customary
Cauchy
to define
{ibid. pp. 168, 309) also
but his investigation when z
is
The reader
will at
many
2 as
lim
+-
cf.
Cauchy, Coins
exp.r^l
cV Analyse,
i.
derived the properties of the function from the series,
not rational
alg. Analysis (1889), pp. 29, 178, 246.
the limit definition has
exp
series definition,
is
incomplete.
See also Sclilomilch, Handbuch der
Hardy has pointed out (Math.
Gazette, in. p. 284) that
disadvantages.
once verify that the general term in the product
(2i + Ci2i"-^22 + C22i"-22._,2 +
. . .
series is
+ z^'^^n = (î + z^)''ln

!
APPENDIX
582
Further, exp
.>
is
an increasing function of the real variable

exp {x + k) exp x = exp x {exp
.
,v
for, if X.'>0,
1 > 0,
}
>l-
because exp .r>0 and exp />!.
{expA-
Also, since
and the
series
values of
on the right
= 1 +(A/2 + (/i2/3 + ...,
1}/A
!)
:)
seen (by the methods of
is
we have
h,
lim {exp/i-
A"2) to be continuous for
all
= 1,
1}//;
exp(2+A)-exp2
rfexpa
^ = exp
= hm ^-^
y
,.
and so
^^
Variom properties of
A'22.
the exponential function.
Returning to the formula (exp
when n
is
^j)
(exp
(exp
(exp
2.>) ...
a positive integer,
2)"
2-)
= exp (21+22 + +2n) we
In particular, taking
when
is
an
2= 1
if
and writing
- ?iz).
place of exp
e in
= 2-71828..., we
see that,
integer, positive or negative,

e'
Also,
see that,
= exp (ns),
(exp z)-" = 1 /(exp 2)" = 1 /exp {nz) = exp
and
z.
= exp Hi = l + (i/i:) + (?-/2 +

!)
....
be any rational number {=Pi'q, where// and q are integers, q being positive)
/Li
(exp
so that the ^th power of
exp^
is
fi)'i
et'
= exp fiq = exp p = e'\
that
is to say,
expM
a value of
is
and
ef'ii=ei^,
it is
obviously ( A-21) the real positive value.

If X be an irrational-real number (defined by a section in which Oj and ao are typical
members of the X-class and the ^ff-class respectively), the irrational power e^ is most
simply defined as exp x we thus have, for all real values of x, rational and irrational,
;
e^=\
It
write
exp 2 when
z is
regarded as being a power of
From
complex.
e),
thus defined,
Tannery, Lecons d'Algehre

irrational, as the only
the definition
For e\])x( = X)
satisfies
it
*
it
De
(i,
a-2)
such that
c'"-
rî
X'
{^X)
for every !
and
number
ao.
exists.
also did so, then
is
continuous,
a-i
a^ cannot be chosen
arbitrarily
does not define a section.]
Aitdlysi per aequat. nu)n. term. inf. (written before 16G9, but not published
in letters to
published by Wallis in 1685 in his Treatise on Algebra,

explicitly stated
viz.
p. 45, practically defines e^,
I.
^Xê"-,
/'' ^ X' - -V
was also given both by Newtou and by Leibniz
was
e'*^
easily seen that such a unique

i(
so that, since the exponential function
and so
is
the inequality, and
- exp
subject to the ordinary laws of indices,,
is
et d' Analyse (1906),
number
we have given
exj) 2
small,
exp x when x is real, and it is customary to

The function e^ (which, of course, must not be
therefore, legitimate tg write e^ for

for
is
'
given by Newton*.
first
e^
[Note.
2!
is,
when X
1-
an equation
x'^
-I
by Schlomilch, Handbuch der
alg.
p. 343.
Oldenburg
in 1676
The equation when x
Analysis (1889), p. 182.
till
it
is
1711)
was
first
irrational
A-22-A-32] THE ELiÎENTARY TRANSCENDENTAL FUNCTIONS

A'3.
Logarithms of positive numbers*.
when x
It has been seen (^ A"2, A"21) that,
increasing function of x, and obviously exp
any positive number,
then, a be
-^ +
a-
follows
it
expA*
real,
is
( ^')-^>-0
exp.'r=l/exp
If,
583
a positive continuous
is
x -- + oo while
as
x-^ cc.
as
from
3-63 that the equation in x,
exp X = a,
has one real root and only one.
written t LoggW or simply
Log a
This root (which

it is
is,
of course, a function of a) will be
Logarithm of the positive number
called the
Since a one-one correspondence has been established between
an increasing function of
Logarithm
is
It will
Deduce from
The continuity of
now be shewn
the
that,
a.
and since a
that
is
is
to say, the
Logarithm.
when a
is
e^=a,
First suppose that
A>0,
{hla)
is
a.
Log (a + A) = .r + k^
e''^=a-Vh,
so that ^>0,
l
Log a
positive.
Log a = jp,
so that
+ (hja) =
e''.
and then
= l+k+U' + ...>l + k,
0<k<hja,
0<Log( + A) Loga<A/a.
and so
is
a,
A-21 that Log a + Log 6 = Log a6.
Let
that
of a
and
an increasing function.
Example.
A "31.
x must be an increasing function
x,
i-
to say
Hence, h being positive, Log (a + A) Log can be mjide arbitrarily small by taking k
sufficiently small.
Next, suppose that A<0, so that Z-<0, and then a/{a + h) = e~''.
Hence (taking
and
0< -h<â,
as
is
obviousl}' permissible)
^-
Therefore, whether h be positive or negative,

I
be taken
less
if e
be an arbitrary positive number and
than both \a and |ae, we have
and
Log {a + h) Log a\ <e,
so the condition for continuity ( 3-2) is satisfied,
A'32.
{a
get
al(a + h) = l + {-k) + U'^+...>l-k,

- < - 1+ a/ (a + A) = - h/{a + h)<- ^hja.
so
if
we
Diferentiation of the Logarithm.
Retaining the notation of A-31, we see, from results there proved, that,
fixed), then also k^~Q.
Therefore, when a>0,
if
h-Ô
being
(ILosa
Since Log 1=0, we have, by 4"13 example
Loga=
*
Many mathematicians
reader should consult a
define the
11
,.
3,
^~i dt.
Logarithm by the integral formula given in
memoir by Hurwitz {Math. Ann.
lxx. (1911), pp. 33-47)
A.-32.
The
on the founda-
tions of the theory of the logarithm.

t This is in agreement with the notation of most text-books, in which
principal value (see A-6) of the logarithm of a complex number.
Log denotes the
584
APPENDIX
The expansion of Log
A'33.
From
A "32
Log
where
/^
if
powers of a.
in
+ a) =i" {I + t)~idt
(l
Now,
+)
(1
we have
>
= r{\-t+f^ -...
+ {-y>-H'-^ + {-)"(" {i+t)-^}dt
= a - ia- +
+ (-)"-!- a" + /?
= (_)"
-...
itf''
/"%
(i
+<)-i
c^^.
l<a<l, we have
t"{l-\a\)-^dt
i^nlî
= |a|" + i{(n+l)a-|i)}-'
--^
feence,
as
?t
-^ ac
when !<<!, Log(l + a) can be expanded
into the convergent series*
Log(l+) = o--ia2 + ia3-...= 2
(-)"-!"/?;.
n=l
If
a=+l,
|/?,J=/ i(H-;)-it^(;<
so the expansion
Example.
is
[We have
a= + 1
when
valid
Shew
i;"rf^
I
+ l)-i-Ôas
(>i
n-9-x,
./
./
it is
lim
that
log (l
Jm.^ n
1)
not valid when
(l+-j
1.
=e.
Jiu.^ (^
^,
+ sTT^ " -)
= 1,
and the
result required follows
The
A*4.
from the result of
definition of the sine
The functions t
sin z
and cos
and
these series converge absolutely for
25
new
functions by the equations

2i sin z
* This
Trans.
11.
method
CO
by means of power
/_.)22 +
thus
series,
z.
these series with the exponential scries,
cosine are not essentially
= e^]
values of z (real and complex) by 2"36, and so the
all
definitions are valid for all Vcilues of
e-
cosine.
are defined analytically

23
On comparing
A-2 that lim
it is
ap})arcnt that the sine
and
functions, but they can be expressed in terms of exponential
= exp {iz) exp iz),
of obtaining the
2 cos 2 = exp
Logarithmic expansion
(iz)
is,
+ exp iz).
(
in effect,
due to Walhs,
P]iil.
(1668), p. 754.
t These series were given by Newtou,
Dc
trigouometrical functions are defined in the
Anahjsi... (1711), see A'22 footnote.
manner with which
the reader
is
The other
famiHar, as
quotients and reciprocals of sines and cosines.

X These equations were derived by Euler [they were given in a letter to Jobann liernoulH in
1740 and pubhshed in the Hist. Acad. Berlin, v. (1749), p. 279] from the geometrical definitions
of the sine
based.
and cosine, npon which the theory of the circular functions was
tlien universally
THE ELEMENTARY TRANSCENDENTAL FUNCTIONS
A'33-A"5]
It
obvious that sin
is
and cos z are odd and even functions of
respectively
585
;
that
is
to say
sin
2)= sin 2,
The fundaraental properties of sin
A*41.
cos
and
2) = cos 2.
cos
z.
may be
proved, just as in the case of the exponential function ( A"2), that the series
for sin 2 and cos 2 conyerge uniformly in any bounded domain of values of z, and consequently that sin 2 and cos 2 are continuous functions of 2 for all values of 2.
It
Further,
may
it
manner that the
be proved in a similar
22
1-3-1
+ 5---
defines a continuous function of 2 for all values of

is
continuous at 2=0, and so
The additio7i-theorems for sin 2 and cos 2.
"42.
By
and, in particular, this function
2,
follows that
it
lim (2~^sin2) =
series
z*
using Euler's equations
A*4),
(^
it is
easy to prove from properties of the exponential
function that
and
these results are
may
It
known
sin
(2^
cos
{z^
+ 22) = sin z^ cos 23 + cos 2^ sin z^

+ 22) cos 21 cos 22 sin z^ sin 22
as the additiorit-theorems for sin 2
also be proved,
by using Euler's equations, that

sin2 2
By means
and cos 2.
+ cos-2 = l.
+ 22) can be expressed as an algebraic function of sin2i

can similarly be expressed as an algebraic function of cos 21
of this result, sin(2j
and sin 22, while cos (21 + 22)

and cos 22 so the addition-formulae may be regarded as addition-theorems
;
sense
(cf.
By
in the strict
20-3, 22-732 note).
differentiating Euler's equations,
dz =
it is
Shew
Example.
cos
= sm
5
2,
2.
dz
that
sin 2i
these results are
obvious that
dcoaz
(/sin 2
known
= 2 sin 2 cos
cos 22 = 2 cos-
2,
as the duplication-formulae.
The periodicity of the exponential function.
A'5.
22 are such that exp2j = exp22, then, multiplying both sides of the equation by
exp( 22), we get exp (21 22) = ! and writing y for z^ Z2, we see that, for all values of 2
If
2i
and
and
all integral
values of n,
exp
(2
-f-
ny) = exp
(exp
y)''
= exp 2.
The exponential function is then said to have period y, since the effect of increasing
by y, or by an integral multiple thereof, does not affect the value of the function.
It will
now be shewn
the numbers
y,
that such numbers y (other than zero) actually exist, and that all
possessing the property just described, are comprised in the expression

2mvi,
where it
than 4.
*
The
is
a certain jjositive
number* which happens
(^=1, 2, 3,
...)
than 2v'2 and
to be greater
less
an irrational number, whose value is 3'141-59... is irrelevant to the

For an account of attempts at determining the value of tt, concluding
with a proof of the theorem that it satisfies no algebraic equation with rational coefficients, see
Hobson's monograph Squaring the Circle (1913).
fact that
tt
present investigation.
is
APPENDIX
586
The sold tion of
A*51.
Let y = a +
expy=l
is
the equation
where a and
ifi,
/3
e\\>y=\.
are real
then the problem of solving the equation
identical with that of solving the equation
exp expi/3=l.
.
Comparing the real and imaginary parts of each side of this equation, we have
expa.cos^=l,
expo,
sin j3=0.
Squaring and adding these equations, and using the identity cos-/3 + sin2/3=
1,
we
get
exp2a=l.
Xow
if
would be
a were positive, exp 2a would be greater than
less
than
and
cos /3 =
It follows that
Now
the equation
sin/ii
if a
were negative, exp 2a
sin
1 ,
3 = 0-
a necessary consequence of the equation cos^ =
is
account of the identity cos2/3 + sin-î3=l.

(if
and
1,
so the only possible value for a is zero.
l,
on
It is therefore sufficient to consider solutions
such solutions exist) of the equation cos/3 = l.

Instead, however, of considering the equation
cos/3
= l,
is
it
more convenient
to
consider the equation* cos 07=0.

It will
now be shewn
and
between
2,
that the equation cos.r=0 has one root, and only one, lying
and that
this root exceeds ^^2
of the following considerations
The function
(I)
(II)
When
so,
>
(Ill)
The value of cos 2
is
^.i-
^ y'2,
cos
2<'
we make use
Oâ'^2.
have +
-__>0
6!^'
4!
'
when
to prove these statements,
cos.r is certainly continuous in the range
^ A- ^ ^'2, we
1- >0
2!^
and
___>o
10!=^ '"'
8!
0.
4 \
210 /
-^ + S-72o('-7:Ti)-IO!('-l-Tn-2)--=-i--<WhenO<.r^2,
(IV)
sin.v
and
so,
when
^ x 4. 2,
/,
sin
.v"\
A''*
x^ \
/,
x'^
0-6) + 12oO-6r7) + ->l-6^^'

,
x ^ \x.
3-63 that the

from (II) and (III) combined with the results of (I) and of
equation cosa'=0 has at least one root in the range v^2 <.i'< 2, and it has no root in the
range $ x ^ J
It follows
^^
Further, there
is
oiot
more
than,
there were two, Xi and X2{x.2>Xi)
sin (a% Xij
and this
is
one root in the range
J2<x<2;
for,
suppose that
then 0<.r2-A'j<2-y'2<l, and
sin .Cg cos x^ sin Xi cos x^ = 0,
incompatible with (IV) which shews that sin
(0:^2
-^'i)
^M'^2~'^'i)-
The equation cos:r=0 therefore has one and onli/ one root lying betiveen and 2. This
root lies between ^^2 and 2, and it is called \n and, as stated in the footnote to A"5, its
actual value happens to be 1 "57079....
;
* If cos.r
and
= 0,
an immediate consecjuence of the duplication-formulae that cos2.r= -1

so, if x is a solution of cos.c = 0, 4.t is a solutiou of cos/3 = l.
may be replaced by
except when x = ,^2 in the first place where it occurs,
it is
tht'uce that co8 4.t
t The symbol ^
and except when x =
= 1,
>
in the other places.
A-52] THE ELEMENTARY TRANSCENDENTAL FUNCTIONS
A.-51,
From
the addition-formulae,
may
it
cos
where
is
?i
any
In particular, cos2)i7r = l, where n
choose the integer
no value of
is
which cosj3=l
for
ra
integer.
other than those values which are of the form
/3,
must be
it
real*,
2?i7r,
and so we can
so that
- ^/3
WITT
^27mr I3<7r.
TT
= sin(m^-l/3)=sini^=2~2(l_cos/3)i = 0,
this is inconsistent t with sin
wtt ^/3
Consequently the numbers 2n7r,
equal
any
is
then have
sin
and
= 0,
there were such a value,
if
We
sin?i7r
integer.
Moreover, there
for
be proved at once by induction that
= 1)",
?;7r
587
(/i
= 0,
^J
j/itt
i/3
unless
1, 2,...), a/id no
/3
= 2mTr.
others,
hare their cosines
to unity.
number n
It follows that a positive
exp2 has no period fundamentally
exists such that
from
distinct
exps has period
The formulae of elementary trigonometry concerning the

functions, with which the reader
methods without any
is
îri
and
that
liri.
already acquainted, can
periodicity of the circular
now be proved by
analytical
difficulty.
Example
\.
Shew that
Example
2.
Shew
sini^r
is
equal to 1,-not to
when 0<x<\Tr.
that tan.i->.z;
[For cos.i'>0 and
sm.r orcosA':
and every term
Example
l)!i
= 1 (4
4>i
in the series is positive.]
Shew
3.
x^
that
-V+
2
vanishes when
7-^1,
+ lj'
--A^n-'i71
.r
25
v^
77^
~ ;^
20
ir7
24
i^
positive
when x = ,r^, and that
= (6 2 ^/3)2 = 1-5924...
lb
x^
x^
Jrr
-V+
24
and deduce that J
3-125 <7r<3-185.
The solution of a pair of trigonometncal equations.
A'52.
Let
X,
Then,
/x
be a pair of real numbers such that X^ + jit^^l.
if X=|=
the equations
1,
cos
have an
infinity of solutions of
X and
First, let
be not negative
fi
X = X,
sin
X = î
which one and only one

;
lies
between
tt
and
tt.
then (3-63) the equation cos.r = X has at least one

= 1, cos^7r=0. The equation has not two
solution Xi such that O^.x-jîTr, since cos
and x.2, were distinct solutions we could prove

would contradict A-51 (IV), since
solutions in this range, for if Xi
that sin
(.t'l
.i"2)
= 0,
and
this
(cf.
A-51)
0<|^2--î <i7r<2.
I
Further, sin.ri= +\'(1 coh'^ X])=

*
The equation cos/3=l implies that exp
complex
See
i(i
= M5
so x^ is a solution of both equations.
l, and we
have seen that this equation has no
roots.
+ The inequality
+
-1-^/(1 --^^)
is
true by (IV) since
De Morgan, A Budget
irnr
|/3
:^
i7r<2.
of Purado.ves (London, 1872), pp. .S16
et.
sfq., for
proving that 7r>3^.

If
X=: -
1,
7r are solutions and there are no others in the range
tt,
tt).
reasons for
APPENDIX
588
The equations have no
x or cos x
one, in the range

If
X or
rr,
since, in these
soUition,
and only
we may
investigate the equations in a similar
manner;
the rea<ier.
left to
a solution of the equations, so also
is
x-^
and therefore the equations have an
integer,
tt)
tt).
It is obvious that, if
any
negative.
is
(or both) is negative,
fi
the details are
( - tt, 0) and (^tt,

Thus the equations have one
solutions iu the ranges
ranges, either sin
is Xy
+ 'iniY,
where n
is
infinity of real solutions.
The principal solution of the trigonometrical equations.
A'521.
The unique solution of the equations cos.t' = X, sin,t-=/i. (where X^ + fx'^=\) which lies
between - it and n is called the principal solution*, and any other solution differs from it
by an integer multiple of
Stt.
principal valuei of the argument of a complex
The
number
can now be defined
^(=1=0)
analytically as the principal solution of the equations

I
and then,
d=
It will
+ 2mr,
(P
through the value

Let
z,
that
it is
Zq, it is
j,
2i
lê
(I)
I
(III)
(I)
and
argument 6
provided that
of
.Ti - ^0
- 2/0
?/i
'-"o
t).
+ ^Vo
>
^1
.^'i
+ >i
<2
<2
-^0
J/o
provided that Xq =t= 0,
provided that
|Xi-.%|<Je|2(,|,
3/0
=t=
0,
i?/i-3/o|<ie|2ol.
(II) it follows that x^Vi
and
^ â'o^
xo .vi
are not negative, and
y^f/i
/oî
+ ?/o2/i ^ J
> i.yo^
20
2.
j
that value of 61 be taken which differs from ^0 by less tlian

have not opposite signs and yo f^'id 1/1 have not opposite signs J^,
the solution of the equations of ^ A"52 that ^^ and ^0 differ by less than ^n.
let
Xi
it
it
then, since
follows from
tan(î-^,.) = ^':i^^",
Now
\= - 1, we take +7r as the principal solution of. p. 9.

The term principal value was introduced in 1845 by Bjorling
* If
und
(z),
does not pass
.'ToA-i
written
chosen to be so small that the following inequalities are satisfied J
so that
and
is
its argument
then, to prove that
shew that a number 61 exists such that î=arg2i
than an arbitrary positive number e by giving 21 20
less
- So
z,
be any value of
let 6^
than some positive number
less
(II)
x^,
and
variable.
possible to choose such a value of the
and
fo
Now
argument of
the
sufficient to
^0 can be made
^1
let
From
a value of
called
is
it
Let
Also
is
zero.
continuous at
and that
any value
that
^ given ^'alue of
^^6
2,1
and 6
The continuity of the argument of a complex
now be shewn
a complex variable
(z) is
= /^ (2), 2 sin ^ = / {z),

= 2 (cos ^ + / sin 6),
(^
1-5).
i^
A "522.
cos
we must have
arg2 (cf
if
see the Archiv der
Math,
Fhijs. ix. (1847), p. 408.
X
(ii) 2/0
(I)
or
(II)
respectively
= 0.
The gtometrical
is
simply to be suppressed in the case when
interpretation of these conditions
different quadrants of the plane.
is
merely that
(i)
.TqÔ, or
and
z^
when
are not in
A"521-A7] THE ELEMENTARY TRANSCENDENTAL FUNCTIONS

aud so
A"51 example
2),
ô(yi-yo)-y o('î--ô)
^2
But
A'o
I
589
ko and also yo
|
(9i
- 6>o
2i - zq
2o
|~^
^:o
=$
yi-ô
{I
^0
therefore
2o
1~^
{|
+ l^'o
yi- ?/o
^1
-^'i
- .^0
-^0
!}.
|}<e-
(if ô + 0) and | e ^g
J^'o
ô + 0)
so that, if ri be the smallest of the
the inequalities (I), (II), (III) above are satisfied
<
and this
three numbers * i a-o -I yo i ko j by taking Sj - ^q
'?, we have
î - ô
is the condition that 6 {z) should be a continuous function of the complex âriable z.
Further,
if
we take
l^ss
than
^-
and i yo
(if
>
<
<
Logarithms of comflex numhers.
'6.
f is said to be a logarithm of z if 2 = e^
The number
To
solve this equation in
(,
write
z
Taking the modulus
where
^=^ +
= e^ (cos + i sin
of each side,
Tj
we
see that
and
^'7?
rj
are real
and then we have
?;).
3|
=e
so that ( A-3), ^ =
Log
and
\z\;
then
z=\z\ (cos
so that
T]
must be a value of
The logarithm
can be expressed
argz.
complex number is consequently a many-valued function, and

terms of more elementary functions by the equation
log z
of logs
The
= Log
2 4-
arg z.
(when s4=0) follows from
A'31 and
If a^
A'522, since \z\ is
z.
any particular branch of logs (

and the expansion of A"33 may be established for
Corollary.
it
differential coefficient of
as in A*32
when
+ 1 sin/;),
of a
in
The continuity
7/
be defined to mean
e'iog, a^ is
5'7)
may
log (1
be determined
+) when
aud
< 1.
of
a=t=0.
The analytical definition of an angle.
A'7.
^1, 22, 23 be three complex numbers represented by the points Pj, P.^^, F3 in the
Argand diagram. Then the angle between the lines ( A"12, footnote) PiPo and P1P3 is
defined to be any value of arg (23 Zj) - arg (22 Zi).
Let
It will
now be shewn f that the area
radii of a given circle
and the arc of the
(defined as
circle
an
integi'al),
which
terminated by the
is
bounded by two
radii, is proportional to
one of the values of the angle between the radii, so that an angle (in the analytical sense)
which is given at the beginning of all text-books on Trigonometry |.
jjossesses the propei'ty
any of these numbers is zero, it is to be omitted.

The proof here given applies only to acute angles
* If
extending the result to angles greater than
bounding
lir,
and
the reader should have no difficulty in
to the case
when OX
is
not one of the
radii.
X Euclid's definition of an angle does not, in
itself, afford a measure of an angle

it is shewn
Hobson, Plane Trigonometry (1918), Ch. i) that an angle is
measured by twice the area of the sector which the angle cuts off from a unit circle whose centre
in treatises on Trigonometry
is
at the vertex of the angle.
(cf.
APPENDIX
590
Let
x'^-\-y-
yi)
(o-'i,
be any point (both of whose coordinates are positive) of the circle
= a?{a>0).
Let 6 be the principal value of arg(î +
Then the area bounded by

circle joining
{.Vi,
OX
y{) to (o, 0) is
and the
^ .r ^ a cos ^),
/(^)=a;tan^
(0
/ {x) = {a^ x^)-
(a cos d
remains to be proved that
f{x) dx
J
/"a
fa
Now
Z f{x) dx =
I)
^xâ),
is
(a^
(I
^ha-^smecose + hr
{a^-x-)
- x-)i dx
cose
Lâ^--x^')-^ + ^^x{a^-x^)^\ dx
1
-
dx
(1
-)
'id(- p'^
(1
on writing x = at and using the example worked out on
That
is
6.
o COS Q
= |a2 1 r
this extent,
61).
dx +
J
proportional to
fa
X tan
Jo
= â-jf"
(cf. p.
"
cose
0<6 <Itt.
and the arc of the
'
an area be defined as meaning a suitably chosen integral

It
so that
i/y)
f{x)dx, where*
/
./
if
iyi),
line joining (0, 0) to {x\,
to say, the area of the sector
is
- f') -
-^
dt\
p. 64.
proportional to the angle of the sector.
we have shewn that the popular conception
of
an angle
is
To
consistent with
the analytical definition.

*
The reader
figure.
will easily see
the geometrical interpretation of the integral by drawing a
LIST OF
\^The
numbers
AUTHOES QUOTED
refer to the pages.
Initials which are rarely used are
given in
Abel, N. H., 16, 17, 50, 57, 58, 211, 229, 230,
353, 419, 429, 442, 491, 496, 505, 512, 525
Adamoff, A., 351, 353, 354
Adams, J. C, 125, 235, 331, 406
Aichi, K., 426
Airev, J. R., 378
Aldis, W. S., 378
Alexeiewsky,
W.
P., 264
Amigues,E. P. ill., 122

AndinK, E., 378
J. R., 9
Bachmann,
P., 11
Baker, H. F., 53
Barues, E. W., 127, 159, 236, 264, 279, 286,
289, 296, 299, 301, 326, 330, 338, 346,
351, 369, 370
Basset, A. B., 372, 373, 384
Bateman, H., 230, 353, 399, 402
Bauer, G., 333, 401
Beau, 0., 193
Berger, A., 191
Bernoulli, Daniel (1700-1782), 160, 356
Bernoulli, Jakob (1654-1705), 125, 127,
429, 402
Bernoulli, Johann (1667-1748), 584
Bertraud, J. L. F., 71. 401
Bessel, F. W., 204, 205, 341, 342, 356,
362, 367
Besso, D., 352
Biermann, W. G. A., 454
Binet, J. P. M., 248, 249, 250, 251, 253,
262, 263, 314
Bjorling, E. G., 588
Blades, E., 403
Biirmann, Heinricb, 128, 129

Burnside, William, 457
Burnside, W. S., 212, 361, 453
Cajori, F., 59, 60
Callandreau, 0., 364
Cantor, Georg F. L. P., 4, 161, 182, 183, 186
Carda, K., 126
Carliui, F., 369
Garse, G. A., 189
Catalan, E. C, 332, 333

Cauchy, (Baron) A. L., 12, 13, 21, 25, 27, 29,
Appell, P. E., 280, 298, 300, 336, 400, 418
Argand,
italics"^
347,
40, 42, 59, 71, 77, 83, 85, 86, 91, 93, 96,
99, 105, 119, 122, 123, 160, 243, 263, 379,
383, 462, 581
Cavlev, A., 147, 198, 298, 434, 439, 455, 487,
496, 498, 508, 528, 530, 533, 576
Cellerier,
C, 110
Cesaro, E., 38, 39, 58, 155, 156
Chapman,
356,
S., 159,
377
Charpit, P., 891

Chartier, J., 72, 77
Chree, C, 381
Christoffel, E. B., 333, 562
Chrystal, G., 23
Clairaut, A. C, 166
Clausen, T., 298
Clebsch, B. F. A. 393, 455
,
357,
Cohn,
F.,
557
Corey, S. A., 108

Craig, T., 208
Crawford, L. 570, 576, 578
Cuuninghani, E.,,210, 353
Curzou, H. E. J., 210, 351
,
261,
Bobek, K., 533

Bocher, M., 81, 203, 212, 221, 229, 230, 231,
361, 555
Bolzano, B.', 12, 13
Bonnet, P. Ossian, 66, 161, 176
Borchardt, C. W., 105, 455, 463
D'Alembert,
J. le
Bond,
22, 23, 160
Daniels, A. L., 455

Dantscher, V. von, 4, 10
Darboux, J. G., 43, 61, 96, 125, 301, 316
Darwin, Sir George H., 547, 576
De Brun, F. D., 457
Debye, P. 369
Dedekind. J. W. Eicbard, 4, 10
De la Vailee Poussin, Ch. J., 39, 59, 72, 73,
80, 81, 108, 161, 173, 179, 181, 189
De Morgan, A., 23, 587
,
Borel, E., 53, 83, 99, 108, 140, 141, 144, 154,
155, 159
Bouquet, J. C, 455, 462, 487
Bourget, J., 379, 396
Bourguet, L., 246, 260, 261, 264
Bridgeman, P. W., 381
Brioschi, F., 570, 577, 578
Briot, J. A. A., 455, 462, 487
Bromwich, T. J. I'a., 10, 25, 26, 33, 38, 45,
50, 59, 75, 80, 144, 156, 159, 242, 290,
Descartes, E. du P., 3
Dini, U., 179, 381
580
Brouucker, William (Viscount), 16
Bruns, H., 426
Burgess, J., 342
Burkhardt, H. F. K. L., 189, 323, 399
Dirkseu, E.
173
Dixon, A. C, 128, 301, 533
Dolbnia, J. P. (Dolbnja, Iwan), 461
Dougall, J., 301, 400, 426
Du Bois Eeymond, P. D. G., 66, 81, 110
Dinnik, A., 378

Dirichlet, P. G. Lejeune, 17, 25, 50, 71, 76, 77,
80, 81, 161. 163,
164, 176, 179, 182, 247,

248, 258, 279, 314, 315
LIST OF
592
AUTHORS QUOTED
Hardy, G. H.,
Eisenstein. F. G. M., 52, 434
Emde,
159,
F., 342, 378

Encke, J. F., 342
Enneper, A., 455, 528
Euclid, 589
Euler, L., 16, 09, 119, 127, 151, 155, 159, 100,
163, 235. 236, 237. 241, 253, 255, 260, 201,
262, 265, 266, 271, 272, 281, 341, 356, 372,
896, 462, 487, 495, 512, 584
Faber, G.. 580

(II Marchese) Giulio Carlo de Toschi
429
Fatou, P., 163
F6aux, B., 249
Fejer, L.. 161, 109
Ferrers, N. M., 323, 324, 330
Filon, L. N. G., 399
Floquet, A. M. G., 412, 413, 426
Ford, L. R., 455
Forsvth, A. R., 194, 201, 203, 204, 208, 285,
Faguano,
di,
388,
390,
391,
400, 436, 455,
519,
531,
534
Fourier, (Baron) J. B. Joseph, 160, 163, 164,
175, 188, 211, 381, 399, 463
Frechet, M., 230
Fredholm, E. I., 212, 213, 215, 217, 221,
230
A., 399, 463
Freud, ^Y., 3
Fricke, K. E. R., 481, 487, 508
FiobeuiuP. F. G., 197, 201, 208, 319, 323, 421,
446, 458
Fuchs, I. L., 197, 208, 412
Fiirstenau, E., 36
Fuss, P. H., 237
Freeman,
Gambioli, D., 148

Gauss, Carl F. (Johann Friedrich Karl),
7, 9,
240, 246, 247, 248, 281, 283, 294, 296,

297, 319, 429, 462, 512, 524, 533
Gegenbauer, L., 329, 335, 361, 385
Glaisher, James, 342
Glaisher, J. W. L., 332, 491, 494, 496, 498,
507, 509, 516, 520, 523, 527, 530
Gmeiner, J. A., 14
Goldbach, C, 237
Goursat, Edouard J. B., 53, 59, 61, 80, 85,
1U8, 121, 144, 208, 230, 401
Grace, J. H., 123
Gray, A. 373, 379
Green, George, 395
Gregory, James, 16
Gudermann, C, 491, 494, 531
Guerritore, G., 558
Guichard, C, 148
Gutzmer, C. F. A., 109
,
Hadamard,
J., 108, 212

Haentzschel, E., 576
Halm. J. K. E., 210
Halphcn, G. H., 452, 487, 550, 555, 570, 576,
577, 578
Hamburger, M., 412
Hamilton, Sir William Rowan, 8, 173
Hancock, H., 481, 487, 492
Hankel, H., 10, 244, 245, 246, 266, 365, 369,

370, 371, 373, 378, 381, 385
Hansen, P. A., 378
10, 17, 38, 50, 59, 69, 81, 156,
272,
173,
279, 280, 297, 515, 580,
581
Hargreave, C. J., 383
Hargreaves, R., 309
Harkucss, J. 487
Harnack, A., 160
,
W. J., 41
Heine, H. E., 53, 54, 316, 319, 321, 326, 329,
330, 367, 379, 401, 462, 562, 575
Hermite, C, 204, 231, 269, 270, 271, 300, 301,
Hcffter, L.
333, 350, 417, 457, 458, 4G1, 487, 555, 563,

570, 573, 575, 576
Heun, K., 331, 555, 576, 577
Heymann, K. W., 298

Heywood, H.
Hicks,
W.
B., 230
M., 401
Hilbert, D., 213, 227, 230

Hill, G. W., 36, 40, 406, 413, 414, 415, 416,
417, 424, 426
M. J. M., 97, 297

Hobson, E. W., 3, 10, 53,
Hill,
56, 67, 75, 80, 160,

161, 189, 292, 323, 325, 326, 330, 334, 364,
375, 381, 383, 543, 576, 580, 585, 589
Hocevar, F., 341
J., 312
Holder. 0., 66, 236
Hurwitz, Adolf, 161, 180, 265, 268, 269, 583
Hurwitz, Wallie Abraham, 76
Hodgkinson,
Ince, E. Lindsay, 424, 426, 427

Innes, R. T. A.. 484
Isherwood, J. G., 378
Jackson, F. H., 462

Jacobi, Karl (Carl) G. J., 39, 108,
314, 369, 429, 402, 463, 464, 467,
470, 474, 475, 478, 479, 480, 487,
491, 494, 495, 496, 498, 505, 508,
512, 514, 516, 517, 519, 520, 522,
529, 535
Jacobsthal, W., 338, 351
Jahnke, P. R. E., 342, 378
Jamet, E. V., 578
Jeffery, G. B., 403
Jensen, J. L. W. V., 270, 279
Je^.ek, 0.,
Jordan,
31.
147
E.
C,
109,
468,
488,
509,
523,
17, 115, 121, 178, 179,
124,
469,
489,
510,
528,
487
Kalahne, A., 378

Kapteyn, W., 146, 374, 377
Kelvin (Sir William Thomson), Lord, 378, 393,
395, 399, 401, 535, 551
Kiepert, L., 456, 458, 460
Klein, C. Felix, 203, 204, 208, 209, 283, 481,
487, 508, 555
Kluyver, J. C.
142
Kiieser, J. C. C. A., 223

Koch, N. F. H. von, 36, 37, 419, 423
Kronecker, L., 123, 146, 463, 468, 525
Kummer, E. E., 250, 262, 281, 285, 296, 298,
338
Lagrange, C, 147
Lagrange, J. L., 96, 132, 133, 149, 166, 308,
338, 353, 356
Laguerre, E. N., 341
Lalesco (Lalescu), T., 230
Lamb, H., 62, 401
LIST OF
AUTHORS QUOTED
Lame,
G., 204, 205, 206, 401, 405, 417, 536,

540, 546, 549, 551, 554, 575
Landau, E. G. H., 11, 276, 279, 280, 341
Laudeu, J., 476, 507, 508
Landsberg, G., 124, 475
Laplace, P. S. (Le marquis de), 211, 246, 312,
314, 369, 386
Laureut, Paul Mathieu Hermann, 123, 321
Laurent, Pierre Alphonse, 100
Leathem, J. G. , 11
Leaute, H., 336
Lebesgue, H., 53, 63, 163, 172, 173, 178, 179,
189
Legendre, A. M., 122, 159, 204, 205, 235, 240,
241, 253, 260, 302, 303, 304, 305, 310, 316,
320, 330, 335, 341,429, 495, 499, 505,512,
515, 518, 520, 521, 522, 523, 524, 525, 527,
528
Leibniz (Leibnitz), G. W., 16, 67, 356, 582
Lerch, M., 81, 108, 110, 149, 271, 279, 280
Le Vavasseur, E., 298
Levi-Civita, T., 144
Liapounoff, A., 180
Lie,'il/. Sophus, 41
Lindelof, Ernst L., 108, 121, 146, 259, 279,
283
Lindemann, C. L. F., 210, 417, 424. 426, 570,
576
Lindstedt, A., 426
Liouville, J., 105, 211, 221, 431, 455, 462, 557
Lipsehitz, E. 0. S., 179, 378
Littlewood, J. E., 156, 159, 279, 280
Lommel, E. C. J. von, 357, 365, 378, 379,380,
381, 383
London, F., 26
Love, A. E. H., 399
McClintock, E., 133
Macdonald, H. M., 121, 335, 373, 383, 384, 385
Maclaurin, Colin, 71, 77, 94, 127, 151, 429
Maclaurin, E. C, 406, 426
Malmsteu, C. J., 249
Mangeot, S., 147
Manning, H. P., 25
Mansion, P., 43
Mascheroni, L. 235
Maseres, Francis (Baron), 3
Mathews, G. B., 373, 379, 579
Mathieu, E. L., 204, 205, 404, 406, 407, 411,
426, 427
Maxwell, J. Clerk, 404
Mehler, F. G., 314, 315, 367, 383
Meissel, D. F.E., 378
Mellin, R. Hj., 286, 296
Merz, J. T., 579
Meyer, F. G., 80
Mildner, E., 148
Milne, A., 231, 351, 354
Minding, E. F. A., 119
Mittag-Leffler, M. G., 134
Molk, C. F. J., 16, 455, 464, 487, 528, 580
Moore, E. H., 236
Mordell, L. J., 454
Morera, G., 87, 110
Morley, F.. 301, 487
Miiller, H. F., 455, 528
Murphy, E., 224, 311, 312
,
Netto, E., 64
W. M.
A.
593
Neumann, Franz Ernst, 320, 321

Neumann, Karl (Carl) Gottfried, 221,
322, 329,
357, 372, 374, 376, 377, 378, 379, 380, 384,
385
Newman,
F. W., 236
Newton, Sir Isaac, 10, 582, 584
Nicholson, J. W., 369, 379, 382, 883
Nicole, F., 142
Nielsen, N., 142, 259, 330, 351, 371, 379, 381,
385
Niven, Sir William D.,
547, 576
Norlund, N. E., 142
401,
536,
543,
546,
Olbricht, E., 324, 330, 334, 379
Oldenburg, H., 582

Orr, W. McF., 298
Osgood, W. F., 45, 49, 59, 72, 87
Painleve, P. 459
Panton, A. W., 212, 361, 453
Papperitz, J. E., 206, 207, 296
Parseval, M. A., 182
,
Peano, G., vii

Pearson, Karl, 353
Peirce, B. 0., 378
Pieard, C. E., 412
Pierpont, J., 75
Pincherle, S., 109, 142, 149, 286, 296, 335
Plana, G. A. A., 146
Pochhammer, L., 256, 292, 299
Pockels, F. C. A., 399
Poincare, J. Henri, 36, 151, 159
Poisson, S. D., 124, 160, 369, 382, 396, 475
Porter, M. B., 361
Pringsheim, A., 16, 17, 26, 27, 28, 33, 38, 110,
249, 259, 580
Prym, F. E., 341
Eaabe,
261
535
J. L., 126,
Eamanujan,
S.
Eavut, L., 87
Eayleigh (J. W. Strutt), Lord, 189, 396, 399
Eeiff, E. A., 16
Eichelot, F. J. 533
Eiemaun, G. F. Bernhard, 26, 38, 63, 80, 84,
,
85, 161, 172, 173, 178, 182, 183, 184, 185,
186, 187, 189, 206,207, 208, 209, 265,266,

269, 272, 273, 274, 279, 280, 283, 294, 296,
373
Eiesz, M., 156
Eitt, J. F., 153
Eodrigues, 0., 303

Eouth, E. J., 332
Eussell, Hon. Bertrand A. W.,
5, 10,
579
Saalschiitz, L., 243, 244, 301
Salmon, G. 455
Savidge, H. G., 378
,
Schafkeitliu, P., 369, 335

Scheibner, W., 108, 109
Schendel, L., 336

Schering, E. C. J., 512
Schlatii, L., 303, 330, 333,
334, 357, 362, 364,

370, 372, 380, 381
Schlesiuger, L., 208
Schlomilch, 0. X., 142, 144, 159, 229, 242,
259, 264, 341, 352. 355, 377, 378, 581, 582
Schmidt, O. J. E., 223, 227, 230
38
LIST OF
594
AUTHORS QUOTED
J. J., 380
Schumacher, H. C, 512
Schwarz, A'. H. A,, 186, 434, 452, 455, 579
Seidel, P. L.,44
Seiffert, L. G. A., 527
Schonholzer,
Shearer, G., 189

Silva, J. A. Martins da, 331
Simou, H., 38
Smith, B. A., 378
Smith, H. J. S., 468, 487, 531
Solduer, J. von, 341, 342
Sommerfeld, A. J. W., 379
Souine (Sonin, Ss6niu), N. J., 352, 364, 382,
383
Stenberg, 0. A., 576
Stickelberger, L. 446, 458
T. J., 261, 341, 417, 420, 424, 426,
555, 560, 561, 562, 570
Stirling, James, 94, 127, 142, 151, 251, 253, 286
Stokes, Sir George G., 44, 73, 77, 81, 152, 167,
173, 204, 378, 382
Stolz, 0., 10, 14, 27
Stormer, F. C. M., 122
,
Stieltjes,
Thome, L. W., 197, 208, 322

Thomson, Sir William (see Kelvin)
Todhunter, I., 204, 330
Transon, A. E. L., 147
Tweedie, C, 127, 142
Verhulst, P. F., 528
Volterra, V., 213, 218, 221, 230
Wallis, John, 11, 32, 281, 582, 584
43, 53, 59, 77, 108, 159, 298,
351, 353, 426, 462
Weber, H., 204, 205, 231, 342, 347, 351, 370,
382, 383, 455, 487
Weierstrass, Karl (Carl) T. H'., 4, 12, 34, 41.
44, 49, 99, 108, 110, 137, 235, 236, 434,
452, 454, 455, 486, 487, 519, 579
Wessel, Caspar, 9
Whitehead, A. N., 10, 579
Whittaker, E. T., 211, 231. 337, 339, 340, 347,
351, 353, 388, 399. 407, 411, 424, 426,-455,
502, 564
Willson, R. W., 378
Watson, G. N..
Sturm,
J. C. F., 557
Sylvester, J. J., 36, 400, 419, 579
Wolstenhohne, J., 123

Wronski, J. Hoen6, 147
Tait, P. G., 393, 395, 399, 535

Tannerv, J., 455, 464, 487, 528, 580, 582
Young, A. W., 406, 426

Young, W. H., 56, 381
Taylor, Brook, 93
Teixeira, F. G., 131, 132, 146
Zach, (Freiherr) F.
A',
von, 341
GENERAL INDEX
[TVie
numbers refer to the pages. References to theorems contained in a few of

the more important examples are given by numbers in italics]
Abel's discovery of elliptic functions, 429, 512

inequality, 16
integral equation, 211, 229, 230;
method of establishing addition theorems, 442, 496, 497, 530, 534; special form, ^,(2), of
the confluent hypergeometric function, 353
test for convergence, 17
theorem on continuity
theorem on multiplication of convergent series, 58, 59
of power series, 57
;
Abridged
notation for products of Theta-f unctions, 468,
469
for quotients
and reciprocals
of
elHptic functions, 494, 498
Absolute convergence,
28
18,
Cauchy's
test
for,
21
D'Alembert's ratio
test
22
for,
De
^Morgan's test for, 23
Absolute value, see Modulus

Absolutely convergent double
property
28
series,
infinite
products,
32
series,
(fundamental
18,
(multiplication of) 29
of) 25,
Addition formula for Bessel functions, 357, 380 for Gegenbauer's function, 335 for Legendre
polynomials, 326, 395
for Legendre functions, 328
for the Sigma-function, 451
for
Theta-functions, 467; for the Jacobian Zeta-function and for E[u), 518, 534; for the
third kind of elliptic integral, 523
for the Weierstrassian Zeta-function, 446
;
Addition formulae, distinguished from addition theorems, 519

Addition theorem for circular functions, 535 for the exponential function, 531 for Jacobian
elliptic functions, 494, 497, 530; for the Weierstrassian elliptic function, 440, 457; proofs
of, by Abel's method, 442, 496, 497, 530, 534
;
Affix, 9
Air in a sphere, vibrations
of,
399
Amplitude, 9
Analytic continuation, 96, (not always possible) 98; and Borel's
geometric function, 288. See also Asymptotic expansions
Analytic functions, 82-110 (Chapter v)
defined, 83
integi-al,
141; of the hyper-"
derivates of, 89, (inequality satisfied by) 91

represented by integi-als, 92; Eiemann's
equations connected with, 84; values of, at points inside a contour, 88; uniformly convergent
;
from monogenic functions, 99
clistinguislied
series of, 91
Angle, analytical definition
of,
589
and popular conception of an angle, 589, 590
Angle, modular, 492
Area represented by an integi-al, 61, 589

Argand diagram, 9
Argument, 9, 588 principal value of, 9, 588 continuity of, 588
Associated function of Borel, 141 of Eiemann, 183 of Legendre
Asymptotic expansions, 150-159 (Chapter viii) differentiation
;
[P"' (z)
and Q'"
(z)],
323-326
153
integration of, 153
multiplication of, 152; of Bessel functions, 368, 369, 371, 373, 374; of confluent hypergeometric functions, 342, 343; of Gamma-functions, 251, 276; of parabolic cylinder functions,
uniqueness of, 153, 154
347, 348
;
of,
Asymptotic inequality for parabolic cylinder functions

Asymptotic solutions of Mathieu's equation, 425
*
Auto -functions, 226
Automorphic functions, 455
Axioms of arithmetic and geometry, 579
of large order, 354
Barnes' contour integrals for the hypergeometric function, 286, 289

geometric function, 343-345
for the confluent hyper-
Barnes' G-function, 264, 278

Barnes'
Lemma, 289
Basic numbers, 462
Bernoullian numbers, 125; polynomials, 126, 127

Bertrand's test for convergence of infinite integi-als, 71
Bessel coefficients [/(^)], 101, 355; addition fonnulae for, 357; Bessel's integral for, 362;
differential equation satisfied by, 357
expansion of, as power series, 355 expansion of
;
?..S
GENERAL INDEX
596
functions in series of (by Neumann), 374, 375, 3S4, (by Schlomilcb), 377
expansion of
{t-z)~^ in series of, 374. 375, 376; expressible as a confluent fomi of Legendre functions,
expressible
as
confluent
hypergeometric
functions,
inequality
satisfied by, 879;
358;
367;
Neumann's function 0,^(z) connected with, sec Neumann's function; order of, 356; recurrence formulae for, 359 special case of confluent hypergeometric functions, 358.
See also
Bessel functions
;
Bessel functions, 355-385 (Chapter xvn), Jn{z) defined, 358-360; addition formulae for, 380;
asymptotic expansion of, 368, 369, 371, 373, 374 expansion of, as an ascending series, 358,
expansion of functions in series of, 374, 375, 377, 5.S'i first kind of, 359; Hankel's
371
integi'al connecting Legendre functions with, 364, 401
integi'al for, 365
integral properties
of, 380, 381, 384, 385; integi-als involving products of, 380, 383, 385; notations for, 356,
order
of,
products
of,
recurrence
formulae for, 359,
372, 373;
356;
379, 360, 383, 385, 428;
373, 374; relations between, 360, 371, 372; relation between Gegenbauer's function and,
form
of
Schliifli's
Bessel's
integral
for,
kind
second
of,
(Hankel), 370;
362,
372;
Y(2)
378;
yC' (-) (Neumann), 372; 1',, (z) (Weber-Schliifli), 370; second kind of modified, K{z), 373;
solution of Laplace's equation by, 395
solution of the wave-motion equation by, 397
whose order is large, 368, 383; whose order is half an odd integer, 364 ;
tabulation of, 378
with imaginary argument. I{z), K{z), 372, 373, 384; zeros of, 361, 367, 378,381.
See
aho Bessel coefficients (ind Bessel's equation
;
fundamental system of solutions of (when )i is not an integer),

Bessel's equation, 204, 357, 373
second solution when n is an integer, 370, 373. See also Bessel functions
359. 372
;
Binefs integrals for log V (z), 248-251

Binomial theorem, 95
B6clier's theorem on linear differential equations with five singularities, 203
Bolzano's theorem on limit points, 12
Bonnet's foi-m of the second mean value theorem. 66
integral, 140; integral and analytic continuation, 141
Borel's associated function, 141
theorem (the modified Heine-Borel theorem), 58
of summing series, 154
Boundary, 44
Boundary conditions, 387 and Laplace's equation, 393
Bounds of continuous functions, 55
Branch of a function, 106
Branch -point. 106
Biirmanns theorem, 128 extended by Teixeira, 131
;
'
'
^'
method
Cantor's Lemma, 183

Cauchy's condition for the existence of a limit, 13 discontinuous factor, 123 formula for the
remainder in Taylor's series, 96 inequality for derivatives of an analytic function, 91
integi'al, 119; integi'al representing V [z), 243
numbers, 372 tests for convergence of series
;
and
integrals, 21, 71
Cauchy's theorem, 85
CeU, 430
extension to curves on a cone, 87
Morera's converse
of, 87,
110
method of summing series, 155 generalised, 156

Change of order of terms in a series, 25 in an infinite deteiTninant, 37 in an infinite product, 33
Change of parameter (method of solution of Mathieu's equation), 424
Characteristic functions, 226
numbers, 219 numbers associated with symmetric nuclei are
Cesaro's
'
'
real.
226
Chartier's test for convergence of infinite integrals, 72

Circle, iirea of sector of, 589
limiting, 98 ; of convergence, 30
Circular functions, 435, 584
addition theorems for, 585 continuity of, 585
differentiation
of, 585
duplication fonnulae, 585 periodicity of, 587 ; relation with Gamma-functions,
;
239
Circular
membrane,
Class, left (L), 4
vibrations of, 356, 396
right
{li),
Closed. 44
Cluster point, 13
CoefiBcients, equating, 59
ot, 163, 165
in Fourier series, nature of, 167,
174
in trigonometrical series, values
Coefficients of Bessel. sec Bessel coefficients
Comparison theorem for convergence of integrals, 71 for convergence of series, 20

Complementary moduli, 479, 493 elliptic integrals with, 479, 501, 520
Complete elliptic integrals [K, K, K', 7v"] (first and second kinds), 498, 499, 518 Legendre's
;
lation lietween,
520
re-
properties of {qua functions of the modulus), 484, 498, 499, 501, 521
GENERAL INDEX
597
299 tables of, 518 the Gaussian transformation, 533 values for small values
values (as Gamma-functions) for special values of k, 524-527; with compleof \k\, 521
mentary moduli, 479, 501, 520
Complex integrals, 77 upper limit to value of, 78
Complex integration, fundamental theorem of, 78
Complex numbers, 3-10 (Chapter i), defined, 6; amplitude of, 9 argument of, 9, 588 dependence
imaginary part of (I), 9 logarithm of, 589 modulus of, 8 real part
of one on another, 41
series for,
oi\R), 9
representative point of, 9
Complex variable, continuous function

Computation of elliptic functions, 485
of a, 44
of solutions of integi'al equations, 211
See also Convergence
Conditional convergence of series, 18 of infinite determinants, 415.

Absolute convergence
Condition of integrability (Eiemann's), 63
;
and
Conditions, Dirichlet's, 161, 163, 164, 176
Conduction of Heat, equation of, 387

Confluence, 202, 337
Confluent form, 203, 337
equation for, 337 general
Confluent hypergeometric function {Wj^^ to (^)]' 337-354 (Chapter xv)
asymptotic expansion of, 342, 345 integral defining, 339 integrals of Barnes' type for,
343-345 Kummer's formulae for, 338 recurrence formulae for, 352 relations with Bessel
functions, 360 the functions W^^^^^ (z) and xl/j._, (z), 337-339 the relations between functions
of these types, 346 various functions expressed in terms of JFj,,,; (z), 340, 352, 353, 360. See
also Bessel functions mid Parabolic cylinder functions
Confocal coordinates, 405, 547; form a triply orthogonal system, 548; in association with
uniformising variables
ellipsoidal harmonics, 552
Laplace's equation referred to, 551
associated with, 549
Congruence of points in the Argand diagram, 430
Constant, Euler's or Mascheroni's, [7], 235, 246, 248
Constants ci e.^, e^, 443; E, E' 518, 520; of Fourier, 164; 771, r/.," 446, (relation between -qi
and 7?,,) 446"; G, 469, 472 A', 484, 498, 499 K', 484, .501, 503
of Mathieu functions, 409, (second method)
Construction of elliptic functions, 438, 478, 492
420
'Contiguous hypergeometric functions, 294
Continua, 43
;
Continuants, 36
Continuation, analytic, 96, (not always possible) 98 and Borel's integral, 141
geometric function, 288. See also Asymptotic expansions
;
of the hyper-
of the argument of a complex variable,

of power series, 57, (Abel's theorem) 57
Continuity, 41
588 of the circular functions, 585 of the exponential function, 581 of the logarithmic
function, 583, 589
uniformity of, 54
Continuous functions, 41-60 (Chapter in), defined, 41 bounds of, 55 integrability of, 63 of a
complex variable, 44 of two variables, 67
;
Contour, 85
roots of
an equation
in the interior of a, 119,
123
evaluation of definite integrals by, 112-124 the Mellin-Barnes type of,

286, 343
see also under the special function represented by the integral
Convergence, 11-40 (Chapter 11), defined, 13, 15; circle of, 30; conditional, 18; of a double
of an infinite integral,
of an infinite product, 32
series, 27 ; of an infinite determinant, 36
of Fourier
of a series 15, (Abel's test for) 17, (Dirichlet's test for) 17
70, (tests for) 71, 72
of the series
of the hypergeometric series, 24
series, 174-179
of the geometric series, 19
of trigonometrical series, 161
2n~*', 19
of the series occurring in Mathieu functions, 422
principle of, 13
theorem on (Hardy's), 156. See also Absolute convergence.
radius of, 30
Non-uniform convergence and Uniformity of convergence
Contour integrals, 85
Coordinates, confocal, 405, 547

Cosecant, series for, 135
orthogonal, 401, 548
Cosine, see Circular functions

(z)], 352
-series (Fourier series), 165
Cotangents, expansion of a function in series of, 139
Cubic function, integration problem connected with, 452, 512
Cunningham's function [w, j (z)], 353
Cosine- integral [Ci
Curve, simple, 43
spiral),
on a cone, extension of Cauchy's theorem
527
Cut, 281
Cylindrical functions, 355.
See Bessel functions
to,
87
on a sphere
(Seiffert's
GENERAL INDEX
598
D'Alemberfs ratio test
Darboux" formula. 125
for
convergence of
series,
22
Decreasing- sequence, 12
Dedekind's theory of irrational numbers, 4

Deficiency of a plane curve, 455
Definite integrals, evaluation of, 111-124 (Chapter
Degree of Legendre functions, 302, 307, 324
vi)
De la Vallee Poussins test for uniformity of convergence of an

De Morgan's test for convergence of series, 23
Dependence of one complex number on another, 41
Derangement of convergent series, 25 of double series, 28
;
intinite products, 33,
infinite integral, 72
of infinite determinants, 37
of
34
Derivates of an analytic function,

Derivates of elliptic functions, 4^0
fi9
Cauchy's inequality
Determinant. Hadamard's, 212

Determinants, infinite, 36 convergence
91
for,
integrals for, 89
415 discussed by Hill, 36, 415 ;

rearrangement of, 37
Difiference equation satisfied by the Gamma-function, 237
Diflferential equations satisfied by elliptic functions and quotients of Theta-f unctions, 436, 477,
492; (partial) satisfied by Theta-functions, 470; Weierstrass' theorem on Gamma-functions
and, 236.
See also Linear diflferential equations and Partial diflferential equations
DiflFerentiation of an asymptotic expansion, 153
of a Fourier series, 168
of an infinite
integral, 74
of an integral, 67
of a series, 79, 91
of elliptic functions, 480, 493
of the
circular functions, 585
of the exponential function, 582
of the logarithmic function, 583,
589
Dirichlefs conditions, 161, 163, 164, 176; form of Fourier's theorem, 161, 163, 176; formula
connecting repeated integrals, 75, 76, 77
integral, 252
integral for f (z), 247
integral for
Legendre functions, 314 test for convergence, 17
Discontinuities, 42; and non-uniform convergence, 47 of Fourier series, 167, 169; ordinary, 42;
regular distribution of, 212; removable, 42
Discontinuous factor, Cauchy's, 123
Discriminant associated with Weierstrassian elliptic functions, 444, 550
Divergence of a series, 15 of infinite products, 33
Domain, 44
Double circuit integrals, 256, 293
Double integrals, 68, 254
Double series, 26; absolute convergence of, 28; convergence of (Stolz' condition), 27; methods
of summing, 27
a particular form of, 51
rearrangement of, 28
Doubly periodic functions, 429-535. See alt> Jacobian elliptic functions, Theta-functions and
Weierstrassian elliptic functions
Duplication formula for the circular functions, 585; for the Gamma-function, 240; for the
Jacobian ellij)tic functions, 498; for the Sigma-function, 459, 460; for the Theta-functions,
48S; for the Weierstrassian elliptic function, 441
for the Weierstrassian Zeta-function, 459
;
of, 36, (conditional)
evaluated by Hill in a particular case, 415
Electromagnetic waves, equations

Elementary functions, S2
for,
404
Elementary transcendental functions, 579-590 (Appendix).

Sec (dm Circular functions.
Exponential function and Logarithm
Ellipsoidal harmonics, 536-578 (Chapter xxiii)
associated with confocal coordinates, 552
derived from Lame's equation, 538-543, 552-554; external, 576; integi'al equations con
;
nected with, 567; linear independence

physical applications of, 547; .species
and Lame functions
Elliptic cylinder functions, see
560; number of, when the degi-ee is given, 546.

537; types of, 537. See also Lamp's equation
of,
of,
Mathieu functions
429-535 (Chapters xx-xxii) computation of 485 construction of, 433, 478

derivate of, 480; discovery of, by Abel, Gauss and Jacobi, 429, 512, 524; expressed by
means of Theta-functions, 473 expressed by means of Weierstrassian functions, 448-451
general addition formula, 457; number of zeros (or poles) in a cell, 431, 432; order of
432; periodicity of, 429, 479, 500, 502, 503; period parallelogram of, 430; relation be
tween zeros and poles of, 433; residues of, 431, 504; transformations of, 508; with no
poles (are constant), 431; with one double pole, 432, 434; with the same periods (relations
between), 452; with two simple poles, 432, 491.
See also Jacobian elliptic functions,
Theta-functions and Weierstrassian elliptic functions
Elliptic functions,
GENERAL INDEX
599
first kind of, 515; function E (it) and, 517; function Z () and,
inversion of, 429, 452, 454, 480, 484, 512, 524 second kind of, 517, (addition formulae
for) 518, 519, 534, (imaginary transformation of) 519
third kind of, 522, 523, (dynamical
application of) 523, (parameter of) 522
three kinds of, 514.
See also Complete elliptic
integrals
Elliptic integrals, 429, 512;
518
membrane, vibrations of, 404

Equating coefiQcients, 59, 186
Equation of degree m has m roots, 120
Elliptic
Equations, indicial, 198; number of roots inside a contour, 119, 123; of Mathematical Physics,
with periodic coefficients, 412.
203, 386-403
See also Difference equation, Integral
equations, Linear differential equations, and under the names of special equations
;
Equivalence of curvilinear integrals, 83

Error- function [Erf (,() and Erfc (,r)], 341
Essential singularity, 102 at infinity, 104
Eta-function [H ()], 479, 480
Eulerian integrals, first kind of [B (m, n)], 253; expressed by Gamma-functions, 254; extended
by Pochhammer, 256
Eulerian integrals, second kind of, 241 .see Gamma-function
expansion (Maclaurin's), 127; method of 'summing' series,
Euler's constant [7], 235, 246, 24S
155 product for the Gamma-function, 237 product for the Zeta-function of Riemann, 271
Evaluation of definite integrals and of infinite integrals, 111-124 (Chapter vi)
;
Evaluation of
Hill's infinite
Even functions,
determinant, 415
115, 165; of
Mathieu
[cc,^{z, q)],
Existence of derivatives of analytic function, 89
407
-theorems, 888
Expansions of functions, 125-149 (Chapter vii)

by Biirmann, 128, 131 by Darboux, 125 by
Euler and Maclaurin, 127; by Fourier, see Fourier Series; by Fourier (the Fourier-Bessel
expansion), 381
by Lagrange, 132, 149 by Laurent, 100 by Maclaurin, 94 by Pincherle,
149 by Plana, 145 by Taylor, 93 by Wronski, 147 in infinite products, 136 in series of
Bessel coefficients or Bessel functions, 374, 375, 381, 384 in series of cotangents, 139 in
series of inverse factorials, 142
in series of Legendre polynomials or Legendre functions,
;
310, 322, 330, 331, 335; in series of Neumann functions, 374, 375, 384; in series of parabolic
cylinder functions, 351 ; in series of rational functions, 134. See also Asymptotic expansions.
Series, and under the itaines of special functions
Exponential function, 581

addition theorem
582 periodicity of, 585
;
for,
581
continuity of, 581
differentiation of,
Exponential-integrral [Ei
Exponents
(z)],
352
at a regular point of a linear differential equation, 198
Exterior, 44
External harmonics,
(ellipsoidal) 576, (spheroidal)
403
Factor, Cauchy's discontinuous, 123; periodicity-, 463

Factorials, expansion in a series of inverse, 142
Factor-theorem of Weierstrass, 137

Fej^r's theorem on the summability of Fourier series, 169, 178
Ferrers' associated Legendre functions [P^'" (z) and Q^"^ (z)], 323
First kind, Bessel functions of, 359; elliptic integi-als of, 515, (complete) 518, (integration of)
515 Eulerian integral of, 253, (expressed by Gamma-functions) 254 integral equation of,
;
221
Legendre functions of, 307

First mean-value theorem, 65, 96
First species of ellipsoidal harmonic, 537, (construction of) 538
Floquet's solution of differential equations with periodic coefficients, 412
Fluctuation, 56 total, 57
Foundations of arithmetic and geometry, 579
Fourier-Bessel expansion, 381
integral, 385
Fourier constants, 164
Fourier series, 160-193 (Chapter ix) coefficients in, 167, 174; convergence of, 174-179; differentiation of, 168
discontinuities of, 167, 169
distinction between any trigonometrical
series and, 160, 163
expansions of a function in, 163, 165, 175, 176; expansions of Jacobian
elliptic functions in, 510, 511
expansion of Mathieu functions in, 409, 411, 414, 420; Fejer's
theorem on, 169; Hurwitz-Liapounotf theorem on, 180; Parseval's theorem on, 182; series
of sines and series of cosines, 165
summability of, 169, 178 uniformity of convergence of,
168, 179.
See also Trigonometrical series
Fourier's theorem, Dirichlet's statement of, 161, 163, 176
;
GENERAL INDEX
600
Fourier's theorem on integrals. 188, 211
Fourtb species of
harmonic, 537, (construction
ellipsoidal
of)
542
Fredliolm's integral equation, 213-217, 228
Functionality, concept
Functions, branches
of,
41
identity of two, 98
limits of, 42
without
principal parts of, 102
essential singularities, 105
which cannot be continued, 98. See also inidcr the mnnes of
special functions! or special types offitncti(yns, e.g. Legendre functions. Analytic functions
106
of,
Fundamental formulae of Jacobi connecting Theta-functions, 467, 4S8

Fundamental period parallelogram, 430; polygon (of automorphic functions), 455
Fundamental system of solutions of a linear differential equation, 197, 200, 389, 559.
iDulcr the
.S^(r
also
names of special equations
Gamma-function
asymptotic expansion of, 251, 276; circular

[r(^)], 235-264 (Chapter xii)
functions and, 239; complete elliptic integrals and, 524-527, 535 contour integi-al (Hankel's)
for, 244; difference equation satisfied by. 237; differential equations and, 236; duplication
formula, 240
Euler's integi-al of the second
Euler's integral of the first kind and, 254
kind, 241. (modified by Cauchy and Saalschiitz) 243, (modified by Hankel) 244; Euler's
product, 237; incomplete form of, 341; integrals for, (Binet's) 248-251, (Euler's) 241;
minimum value of, 253; multiplication formula, 240; series, (Rummer's) :^50, (Stirling's)
Weierstrassian product, 235,
251
tabulation of. 253
trigonometrical integrals and, 256
236.
See also Eulerian integrals and Logaritlimic derivate of the Gamma-function
;
elliptic functions, 429, 512, 524; integral for r'(~)/r(r), 246;

functions, see Lemniscate functions
transformation of elliptic integrals, 533
Gauss' discovery of
lemniscate
addition formula, 335

differential equation for, 329
Gegenbauer's function [C,/ (z)], 329
recuiTence formulae, 330 relation with Legendre functions, 329 relation involving Bessel
functions and, 56-5 Rodrigues' formula (analogue), 329
Schliifii's integral (analogue), 329
Genus of a plane curve, 455
Geometric series, 19
Glaisher's notation for quotients and reciprocals of elliptic functions, 494, 498
;
Greatest of the limits, 13

Green's functions, 395
Hadamard's lemma, 212

Half-periods of Weierstrassian elliptic functions, 444
Hankel's Bessel function of the second kind, Y,j(^), 370 contour integral for T (z), 244 integral
for J(^), 365
Hardy's convergence theorem, 156 test for uniform convergence, 50
tesseral, 392, 536
zonal, 302, 392, 536
spheroidal, 403
Harmonics, solid and surface, 392
See also Ellipsoidal harmonics
Sylvester's theorem concerning integi-als of, 400.
;
of, 387
Heine-Borel theorem (modified), 53
Heat, equation of conduction

Heine's expansion of
(t
- z)-^ in series of Legendre polynomials, 321
Hermite's equation, 204, 209, 342, 347. See also Parabolic cylinder functions
Hermite's formula for the generalised Zeta-function f (s, ), 269
Hermite's solution of Lame's equation, 573-575
Heun's equation, 576, 577

Hill's equation, 406. 413-417
Hill's method of solution, 413

determinant, 36, 40, 415 evaluation of, 415
Hobson's associated Legendre functions, 325
Holomorphic, 83
Homogeneity of Weierstrassian elliptic functions, 439
Homogeneous harmonics (associated with ellipsoid), 543, 57G\
from (Nivcn's formula), 543; linear independence of, 560
;
Hill's infinite
ellipsoidal
harmonics derived
Homogeneous integral equations,
217, 219
Hurwitz' definition of the generalised Zeta-function
theorem concerning Foiu'ier constants, 180
j'(x,
Hypergeometric equation, see Hypergeometric functions

Hypergeometric functions, 281-301 (Chapter xiv) Barnes'
),
265; formula for
(.<,(/),
268;
contiguous, 294
integrals, 286, 289
continuation of, 2H,S
contour integrals for, 291; differential equation for, 202, 207, 283;
functions expressed in terms of, 281, 311 of two variables (Appell's), 300; relations between
twenty-four expressions involving, 284, 285, 290; liiemann's P-equation and, 208, 283;
series for (convergence of), 24, 281
squares and products of, ^96'; value of F(a, l>; c; 1),
;
GENERAL INDEX
601
values of special forms of bypergeometric functions, 298, 301.

281, 393
functions, Confluent hypergeometric functions and Legendre functions
\
See also Bessel
Hypergeometric series, >iee Hypergeometric functions

Hypothesis of Riemann on zeros of f(.s), 272, 280
Identically
vanisMng power
series,
58
Identity of two functions, 98

Imaginary argument, Bessel functions with [I,j(2) and K^{z)'], 372, 373, 384
Imaginary part (/) of a complex number, 9

Imaginary transformation (Jacobi's) of elliptic functions, 505, 506,
474; of E[u) and Z(h), 519
Improper integrals, 75
Incomplete Gamma-functions \_y{n, c)], 341
5-3.5
of Theta-functions, 124,
Increasing sequence, 12
Indicial equation, 198
Inequality (Abel's), 16 (Hadamard's), 212 satisfied by Bessel coefficients, 379

Legendre polynomials, 303; satisfied by parabolic cylinder functions, 354;
j-(.s', a), 274, 275
;
Infinite determinants, see
satisfied
satisfied
by
by
Determinants
evaluation of, 111-124

of, 70, 71, 72 ; differentiation of, 74
representing analytic
functions represented by, sec under the names of special functions
uniform convergence of, 70, 72, 73. See also
functions, 92
theorems concerning, 73
Integrals and Integration
Infinite integrals, 69
convergence
Infinite products, 32
absolute convergence of, 32 convergence of, 32 divergence to zero, 33
expansions of functions as, 136, 137 {see also under the names of special functions) expressed
by means of Theta-functions, 473, 488; uniform convergence of, 49
;
Infinite series, see Series
104; point
Infinity, 11, 103; essential singularity at,
Integers, positive, 3
103; pole
Eiemann's condition
and analytic continuation, 141
Integrability of continuous functions, 63

Integral, Borel's, 140
at,
at,
104; zero
104
at,
signless, 3
63
of,
Integral, Cauchy's, 119

Integral, Dirichlet's, 258
Abel's, 211, 229, 330; Fredholm's, 213-217, 228;
Integral equations, 211-231 (Chapter xi)
Liouville-Neumann method of solution of, 221
homogeneous, 217, 219
kernel of, 213
nucleus of, 213 numbers (characteristic) associated with, 219 numerical solutions of, 211
satisfied by
of the first and second kinds, 213, 221
satisfied by Lame functions, 564-567
Mathieu functions, 407 satisfied by parabolic cylinder functions, 331 Schlomilch's, 229
solutions in series, 228; Volterra's, 221
with variable upper limit, 213, 221
Integral formulae for ellipsoidal harmonics, 567 for the Jacobian elliptic functions, 492, 494
for the Weierstrassian elliptic function, 437
;
Integral functions, 106; and Lame's equation, 571; and Mathieu's equation, 418
Integral properties of Bessel functions, 380, 381, 385; of Legendre functions, 325, 305, 324; of
Mathieu functions, 411 of Neumann's function, 385; of parabolic cylinder functions, 350
;
Integrals, 61-81 (Chapter iv)

along curves (equivalence of), 87; complex, 77, 78; differentiation
of, 67; double, 68, 255; double-circuit, 256, 293; evaluation of, 111-124; for derivates of an
analytic function, 89
functions represented by, see under the juinies of the special functions ;
imjjroper, 75; lower, 61; of harmonics (Sylvester's theorem), 400; of irrational functions,
regular, 201
repeated,
principal values of, 75, 117
452, 512
of periodic functions, 112
round a contour,
representing areas, 61, 589
68, 75
representing analytic functions, 92
85 upper, 61.
See also Elliptic integrals. Infinite integrals, and Integration
;
of Fourier-Bessel, 385
complex, 77; contour-, 77; general theorem on, 63; general theorem on
proof series, 78
of asymptotic expansions, 153
of integi'als, 68, 74, 75
blem connected with cubics or quartics and elliptic functions, 452, 512. See also Infinite
Integral theorem, Fourier's, 188, 211

Integration, 61;
complex, 78
integrals and Integrals

Interior, 44
Internal spheroidal harmonics, 403
Invariants of Weierstrassian elliptic functions, 437

Inverse factorials, expansions in series of, 142
Inversion of
elliptic integrals, 429, 452, 454, 480, 484, 512,
Irrational functions, integration of, 452, 512

Irrational-real numbers, 5
524
GENERAL INDEX
602
Irreducible set of zeros or poles, 430
Irregular points singularities) of differential equations, 197, 202

Iterated functions. 222
i
elliptic functions [sn //, en ii, dn ;(], 432, 478, 491-535 (Chapter xxii)
addition theorems
494, 497, 530, 535; connexion with Weierstrassian functions, 505; definitions of am ,
differential equations satisfied by, 477, 492 ;
A<p, sn M (sin am ), en u, dn u, 478, 492, 494
differentiation of, 493; duplication formulae for, 498; Fourier series for, 510, 511, 55.5;
general description of, 504
geometrical illustration of, 524, 527
Glaisher's notation for
infinite products for, 508, 53:^
integral formulae for, 492,
quotients and reciprocals of, 494
494 Jacobi's imaginary transformation of, 505, 506 Lame functions expressed in terms of,
Landen's transfonnation of, 507 modular angle of, 492 modulus of, 479, 492,
564, 573
(complementary) 479, 493 parametric representation of points on curves by, 524, 537, 527,
periodicity of, 479, 500, 502, 503
poles of, 432, 503, 504
quarter periods. A', iK', of,
555
relations between, 492
residues of, 504
Seiffert's spherical spiral and,
479, 498, 499, 501
527; triplication formulae, 530. 534, 535; values of, when ii is hK, kiK' or h {K riK'), 500,
See ahit Elliptic functions. Elliptic
506. 5(17; values of. when the modulus is small, 533.
integrals. Lemniscate functions, Tbeta- functions, and Weierstrassian elliptic functions
Jacobian
for,
earlier notation for Theta-functions, 479

Jacobi"s discovery of elliptic functions, 429, 512
fundamental Theta-function fomiulae, 467, 4H8; imaginary transformations, 124, 474, 505,
Zeta-function, têe under Zeta-function of Jacobi
506, 519, 535
;
Jordan's lemma, 115

Kernel. 213
Klein's
theorem on
linear differential equations with five singularities, 203
Kummer's formulae
for confluent hypergeometric functions, 338
series for
logF
(z),
250
Lacunary function. 98
Lagrange's expansion, 132, 149; form for the remainder in Taylor's series, 96
Lame functions, defined, 558; expressed as algebraic functions, 556; 577; expressed by Jacobian
elliptic functions, 573-575; expressed by Weierstrassian elliptic functions, 570-572; integi'al
equations satisfied by, 564-567
linear independence of, 559
reality and distinctness of
zeros of, 557, 558, 578
second kind of, 562 values of, 558 zeros of (Stieltjes' theorem),
See alsi) Lames equation and Ellipsoidal harmonics
560.
derived from theory of ellipsoidal harmonics,
Lame's equation, 204, 536-578 (Chapter xxiii)
538-543, 552-554; different forms of. 554, 573; generalised,' 204, 570, 573, 576, 577;
series solutions of, 556, 577, 578; solutions expressed in finite fomi, 459, 556, 576, 577, 578;
;
solutions of a generalised equation in finite foim, 570, 573.

Ellipsoidal harmonics
Lam6
See alao
functions and
Landens transformation
of Jacobian elliptic functions, 476, 507, 533

Laplace's equation, 386 its general solution, 388 ; normal solutions of, 553 solutions involving
functions of Legendre and Bessel, 391, 395; solution with given boundary conditions, 393;
synnnetrical solution of, 399
transformations of, 401, 407, 551, 553
;
Laplace's integrals for Legendre polynomials and functions, 312, 313, 314, 319, 326, 337
Laurent's expansion, 100
Least of limits. 13
Lebesgue's lemma, 172
Left (L-) class. 4
for associated functions, 324
Legendre's equation, 204, 304
Legendre functions and Legendre polynomials
;
second solution
See
of, 316.
aho
Legendre functions, 302-336 (Chapter xv) P{z), Q{z), P^"Uz), Q'"{z) defined, 306, 316, 323,
325
addition formulae for, 328, 395
Bessel functions and, 364, 367, 401
degree of, 307,
differential equation for, 204, 306, 324
distinguished from Legendre polynomials,
324
306; expansions in a.scending series, 311, 326; expansions in descending series, 302, 317,
326, 334 expansion of a function as a series of, 334
expressed by Murphy as hypergeometric
functions, 311, 312; expression of Qn{z) in terms of Legendre polynomials. 319, 320, 333;
Ferrers' functions associated with, 323. 324; first kind of, 307; Gegenbauer's function,
Cn" (z), associated with, xee Gegenbauer's function Heine's expansion of {t - e)~i as a series
of, 321
Hobson's functions associated with, 325 integral connecting Bessel functions with,
364 integi-al properties of, 324 Laplace's integials for, 312, 313, 319, 326, 334 MehlerDirichlet integral for, 314
Schlafli's
order of, 326
recuiTence fomiulae for, 307, 318
integral for, 304, 306; second kind of, 316-320, 325, 320; summation of ::ii"P(:) and
Z/f" Q (z), 302, 321
zeros of, 303, 316, 335. See also Legendre polynomials and Legendre's
equation
Legendre polynomials [P {z)], 95, 302 addition foraiula for, 326, 387 degi-ee of, 302 differential equation for, 204, 304
expansion in ascending series, 311
expansion in descending
;
GENERAL INDEX
603
334; expansion of a function as a series of, 310, 322, 330, 331, 332, 335;
expressed by Mui-phy as a hypergeometric function, 311, 312; Heine's expansion of (t - z)~'^
as a series of, 321; integi-al connecting Bessel functions with, 364; integi'al properties of,
Laplace's integrals for, 312, 314 Mehler-Dirichlet
Laplace's equation and, 391
225, 305
Neumann's expansion in series of, 322 numerical inequality satisfied by,
integi-al for, 314
303 recurrence fonnulae for, 307, 309 Eodrigues' fonnula for, 225, 303 Schlafli's integral
summation of 2/;" P (z), 302 zeros of, 303, 316. See aim Legendre functions
for, 303, 304
series, 302,
Legendre's relation between complete
elliptic integi-als,
520
and cos lemn 0], 524

Lemniscate functions [sin lemn
LiapounofF's theorem concerning Fourier constants, 180
Limit, condition for existence of, 13
Limit of a function, 42 of a sequence, 11, 12 -point (the Bolzano -Weierstrass theorem), 12

Limiting circle, 98
Limits, greatest of and least of, 13
Limit to the value of a complex integi'al, 78
Lindemann's theory of Mathieu's equation, 417 the similar theory of Lame's equation, 570
Linear differential equations, 194-210 (Chapter x), 386-403 (Chapter xviii) exponents of, 198;
fundamental system of solutions of, 197, 200 iixegular singularities of, 197, 202 ordinary
point of, 194; regular integi-al of, 201; regular point of, 197; singular points of, 194, 197,
special types of equations
(confluence of) 202
solution of, 194, 197, (uniqueness of) 196
Bessel's for circular cylinder functions, 204. 342, 357, 358, 373; Gauss' for hypergeometric functions, 202, 207, 283; Gegenbauer's, 329; Hermite's, 204, 209, 342, 3^7; Hill's,
406, 413; .Tacobi's for Theta-functions, 463; Lame's, 204, 540-543, .554-558, 570-575;
Laplace's, 386, 388, 536, 551; Legendre's for zonal and surface harmonics, 204, 304, 324;
Riemann's for
Neumann's, 3&5
Mathieu's for elliptic cylinder functions, 204, 406
P-functions, 206, 283, 291, 294; Stokes', 204; Weber's for parabolic cylinder functions,
Whittaker's for confluent hypergeometric functions, 337 equation for
204, 209, 342, 347
conduction of Heat, 387 equation of Telegraphy, 387 equation of wave motions, 386, 397,
402; equations with five singularities (the Klein-Bocher theorem), 203; equations with three
equations with r singularities,
singularities, 206
equations with two singularities, 208
209 equation of the third order with regular integrals, 210
Liouville's method of solving integral equations, 221
;
Liouville's theorem, 105, 431
Logarithm, 583 continuity of, 583, 589

of complex numbers, 589
;
Logarithmic derivate
of the
differentiation of, 586, 589
expansion
Gamma-function
circular functions and, 240
589
of, 584,
[i^(z)], 240, 241; Binet's integrals for,

Dirichlet's integi-al for, 247 ; Gauss' integi-al for, 246
248-251;
Logarithmic derivate of the Riemann Zeta-function, 279

Logarithmic-integral function [Liz], 341
Lower integral, 61
Lunar perigee and node, motions
of,
406
Maclaurin's (and Euler's) expansion, 127
test for
convergence of infinite integrals, 71
series,
110
94, (failure of) 104,
Many-valued functions, 106

[7], 235, 246, 248
Mathematical Physics, equations of, 203, 386-403 (Chapter x\an). See aho under Linear differential equations and the names of special equations
Mathieu functions [oc(, q), .sp(z, q), in,^{z, q)], 404-428 (Chapter xix) construction of, 409,
420; convergence of series in, 422; even and odd, 407; expansions as Fourier series, 409,
411, 420; integral equations satisfied by, 407, 409; integral formulae, 411; order of 410;
second kind of, 427
general form, solutions by Floquet, 412, by
Mathieu's equation, 204, 404-428 (Chapter xix)
Lindemann and Stieltjes, 417, by the method of change of parameter, 424 second solution
of, 413, 420, 427; solutions in" asymptotic series, 425; solutions which are periodic, .tee
Mathieu functions the integial function associated with, 418. See also Hill's equation
Mascheroni's constant
Mean-value theorems, 65, 66, 96

Mehler's integral for Legendre functions, 314
Mellin's (and Barnes') type of contour integi-al, 286, 343
Membranes, vibrations
of,
356, 396, 404, 405
Mesh, 430
Methods of summing
'
'
series,
Minding's formula, 119

Minimum value of T (.r) 253
,
154-156
;;
GENERAL INDEX
604
Modified Heine-Borel theorem,
Modular
aiiplo, 49'i
-53
function, 481, (equation connected with) 482; -surf.ice, 41
Modulus, 430 of a complex number, 8 of Jacobian elliptic functions, 479, 492, (complementary)
479. 493
periods of elliptic functions regarded as functions of the, 484, 498, 499, 501, 521
Monogenic, 83 distinguished from analytic, 99
Monotonic, 57
Morera's theorem (converse of Cauchy's theorem), 87, 110
Motions (if lunar perigee and node, 406
M-test for uniformity of convergence, 49
Multiplication formula for T [z), 240 for the Sigma-function, 460
Multiplication of absolutely convergent series, 29 of asymptotic expansions, 152 of convergent
series (Abel's theorem), 58, 59
Multipliers of Thcta-functions, 463
Murphy's formulae for Legendre functions and polynomials, 311, 312
;
Neumann's
definition of Bessel functions of the second kind, 372

expansions in series of
Legendre and Bessel functions, 322, 374
(F. E. Neumann's) integral for the Legendre
function of the second kind, 320 method of solving integral equations, 221
Neumann's function [0,j (z)], 374 differential equation satisfied by, 385 expansion of, 374
expansion of functions in series of, 376, 384
integral for, 375
integral properties of,
56.5
recurrence formulae for, 375
Non-uniform convergence. 44 and discontinuity, 47
Normal functions, 224
Normal solutions of Laplace's equation, 553
Notations, for Bessel functions, 356, 372, 373
for Legendre functions, 325, 326
for quotients
and reciprocals of elliptic functions, 494, 498 for Theta-functions, 464, 479, 487
Nucleus of an integi-al equation, 213 symmetric, 223, 228
Numbers, 3-10 (Chapter i) basic, 462; Bernoulli's, 125; Cauchy's, 379; characteristic, 219,
(reality of) 226; complex, 6; irrational, 6; irrational-real, 5; pairs of, 6; rational, 3, 4;
;
rational-real, 5
Odd
real, 5
functions, 115, 166; of Mathieu,
Open. 44
Order [O and
11
o),
functions, 432
function, 102
function, 94
[.s('(^, 7)],
407
of Bernoullian polynomials, 126 ; of Bessel functions, 356 ; of elliptic

of poles of a
of Legendre functions, 324 ; of Mathieu functions, 410
of zeros of a
of terms in a series, 25 ; of the factors of a product, 33
;
Ordinary discontinuity, 42
Ordinary point of a linear differential equation, 194
Orthogonal coordinates, 394 functions, 224
;
Oscillation, 11
Parabolic cylinder functions [/> (2:)], 347

contour integi-al for, 349 differential equation for,
expansion in a power series, 347 expansion of a function as a series of, 351
204, ;i()9, 347
inequalities satisfied by, 354
general asymptotic expansion of, 348
integral equation
satisfied by, 231
integral properties, 350
integrals involving, 353
integrals representing,
353 properties when n is an integer, 350, 353, 354 recurrence formulae, 350 relations
between different kinds of [D,^{z) and D-n-ii i^)]^ 348; zeros of, 354. See also Weber's
;
equation
Parallelogram of periods, 430
Parameter, change of (method of solving Mathieu's equation), 424; connected with Thetafunctions, 463, 464; of a point on a curve, 442, 496, 497, 527, 530, 533; of members of
confocal systems of quadrics, 547
of third kind of elliptic integral, 522
thermometric, 405
Parse val's theorem, 182
Partial differential equations, property of, 390, 391.
See (iIko Linear differential equations
Partition function. 462
Parts, real an<l imaginary, 9
;
Pearson's function [w,
, (z)],
353
connexion with the hypergeometric equation, 208, 283 solutransformations of, 207
Periodic coefficients, equations with (Floquet's theory of), 412
Periodic functions, integrals involving, 112, 256.
See aUo Fourier series mnl Doubly periodic
P-equation, Riemann's, 206, 337
functions
tions of, 2S3, 291, (relations between) 294
GENERAL INDEX
605
Periodicity factors, 463

Periodicity of circular and exponential functions, 585-587
of Theta-funetions, 463
500, 502, 503
of elliptic functions, 429, 434, 479,
Periodic solutions of Mathieu's equation, 407
Period-parallelogram, 430
fundamental, 430
Periods of elliptic functions, 429
qua functions
of the
modulus, 484, 498, 499, 501, 521
Phase, 9
Pincherle's functions (modified Legendre functions), 335
Plana's expansion, 145
Pochhammer's extension
Point, at infinity, 103
of Eulerian integrals,
limit-, 12
256
representative, 9
singular, 194, 202
Poles of a function, 102 at infinity, 104 irreducible set of, 430 number in a cell, 431
between zeros of elliptic functions and, 433 residues at, 432, 504 simple, 102
;
relations
Polygon, (fundamental) of automoi-phic functions, 455

Polynomials, expi'essed as series of Legendre polynomials, 310 of Abel, 333
127 of Legendre, xce Legendre polynomials of Sonine, 352
;
of Bernoulli, 126,
Popular conception
an angle, 589
of
of continuity, 41
Positive integers, 3
Power
circle of convergence of, 30

continuity of, 57, (Abel's theorem) 57
series, 29
expansions of functions in, xee under the name of special functiomi
identically vanishing, 58;
Maclaurin's expansion in, 94; radius of convergence of, 30, 32; series derived from, 31;
;
Taylor's expansion in, 93
uniformity of convergence
57
of,
solution of a certain equation, 482

Principal part of a function, 102
117 value of the argument of a complex number, 9, 588
;
value of an integral, 75,
Principle of convergence, 13
Pringsheim's theorem on summation of double series, 28

Products of Bessel functions, 379, 380, 383, 385, 428; of hypergeometric functions, 298.
also Infinite products
Quarter periods K, iK', 479, 498, 499, 501.

Quartic, canonical form of, 513
integi'ation
See
See also Elliptic integrals
problem connected with, 452, 512
Quasi-periodicity, 445, 447, 463
Quotients of
elliptic
functions (Glai^her's notation), 494, 511
Radius of convergence
Rational functions, 105
Rational numbers, 3, 4
of
;
power series, 30, 32

expansions in series
-real
of Theta-f unctions, 477
134
of,
numbers, 5
Real functions of real variables, 56

Reality of characteristic numbers, 226
Real numbers, rational and irrational, 5
Real part (li) of a complex number, 9
Rearrangement
of convergent series, 25
infinite products,
of double series, 28
of infinite determinants, 37
of
33
Reciprocal functions, Volterra's, 218

elliptic functions (Glaisher's notation), 494, 511
Recurrence formulae, for Bessel functions, 359, 373, 374; for confluent hypergeometric functions,
352 for Gegenbauer's function, 330 for Legendre functions, 307, 309, 318 for Neumann's
function, 375
for parabolic cylinder functions, 350.
See also Contiguous hypergeometric
Reciprocals of
functions
Region, 44
Regvilar, 83
(of
distribution of discontinuities, 212; integrals of linear differential equations, 201,

the third order) 210 points (singularities) of linear differential equations, 197
;
Relations between Bessel functions, 360, 371

between confluent hypergeometric functions
Tr_;^.
(^) and M/^. ^^^^{z}, 346; between contiguous hypergeometric functions, 294; bejjj
tween elliptic functions, 452 between parabolic cylinder functions D,j ( z) and D__i ( iz),
348 between poles and zeros of elliptic functions, 433 between Riemann Zeta-f unctions
;
and f (1 s), 269. See also Recurrence formulae

Remainder after terms of a series, 15 in Taylor's series, 95
Removable discontinuity, 42
Repeated integrals, 68, 75
f
(s)
*;
Representative point, 9
Residues, 111-124 (Chapter
vi),
defined. 111
of elliptic functions, 425, 497
GENERAL INDEX
(506
equations satisfied
condition of integrability, 63
f(j,), 272, 280; lemmas, 172, 184, 185;
/'-equation, 206, 283, 291, 294, (transformation of) 207, (and the hypergeometric equation)
Zetatheory of trigonometiical series, 182-188
208, nee (tho Hypergeometric functions
Riemann's associated function, 183, 184, 185
by analytic functions, S4; hypothesis concerning
function, xer Zeta-function (of

Riesz'
method
Right {U-)
of
summing
'
Riemann)
series,
'
156
class, 4
Rodrigues' formula for Legendre polynomials, 303 modified, for Gegenbauer's function, 329
Roots of an equation, number of, 120, (inside a contour) 119, 123; of Weierstrassian elliptic
;
functions
(<'i
eo, e^),
443
Saalschiitz' integral for the

Schlafli'-s
Gamma-function, 243
Bessel function of the second kind, [r
(2)],
870
for Legendre polynomials and functions, 303,

integral for Bessel functions, 362, 372
modified, for Gegenbauer's function, 329
304, 306
function, 352; integi-al equation, 229
Schlomilch's expansion in series of Bessel coefficients, 377
Sclilafli's
Schmidt's theorem, 223

Schwarz" lemma, 186
Second kind. Bessel function of, (Hankel's) 370, (Neumann's) 372, (Weber-Schliifli), 370,
(modified) 373; elliptic integral of [-E (m), Z ((/)], 517, (complete) 518; Eulerian integral of,
Lame functions of, 562 Legendre
integi-al equation of, 213, 221
241, (extended) 244
functions of, 316-320, 325, 326
;
Second mean-value theorem, 66

Second solution of Bessel's equation, 370, 372, (modified) 373 of Legendre's equation, 316
Mathieu's equation, 413, 427; of the hypergeometric equation, 286, (confluent form) 343
Weber's equation, 347
Second species of ellipsoidal harmonics, 537, (construction of) 540
;
of
of
Section, 4
527
Seififerfs spherical spiral,
Sequences, 11
decreasing, 12
increasing, 12
Series (infinite series), 15: absolutely convergent, 18; change of order of terms in, 25; conditionally convergent, 18; convergence of, 15; differentiation of, 31, 79, 92; divergence of,
methods of summing, 154-156 multiplication
integration of, 32, 78
15 geometric, 19
of inverse
of cotangents, 139
of cosines, 185
of analytic functions, 91
of, 29, 58, 59
of sines, 166 ; of
of rational functions, 134
of powers, see Power series
factorials, 142
remainder of,
order of terms in, 25
variable terms, 44 (see also Uniformity of convergence)
solutions of
15 representing particular functions, see ii)}iler tlie name of the fitiietioit
Sec also Asymptotic
differential and integi'al equations in, 194-202, 228; Taylor's, 93.
expansions. Convergence, Expansions, Foiirier series. Trigonometrical series and Uniformity

of convergence
430
Sigma-functions of Weierstrass [(t(z), <ri{z), 0-2(2), 0-3(2)], 447, 448; addition formula for, 451,
duplication formulae, 459, 460
four
analogy -with circular functions, 447
458, 460
quasi-periodic properties, 447
expression of elliptic functions by, 450
types of, 448
singly infinite product for, 448; three-term equation involving, 451, 461; Theta-functions
connected with, 448, 473, 487 triplication formula, 459
Set, Irreducible (of zeros or poles),
Signless integers, 3
Simple curve, 43 pole, 102 zero, 94

Simply-connected region, 455
;
Sine, product for, 137.

Sine-integi-al [Si (2)],
352
See also Circular functions

;
166
-series (Fourier scries),
See also Circular functions

Singly-periodic functions, 429.
at infinity, 104; confluence of, 203, 337; equations
Singularities, 83, 84, 102, 194, 197, 202
with five, 203; equations with three, 206, 210; equations with two, 208 equations with r,
;
209; essential, 102, 104 irregular, 197, 202 regular, 197

Singular points (singularities) of linear differential equations, 194, 202
Solid harmonics. 392
Solution of Riemann's P-equation by hypergeometric functions, 283, 288
Solutions of differential equations, see Chapters x, xviii, xxiii, and under the names of special
;
iiiuiitions
Solutions of integral equations, see Chapter xi

Sonine s polynomial ['/'," (2)], 352
Species (various) of ellipsoidal harmonics, 537
GENERAL INDEX
Spherical harmonics,
see
607
Harmonics
Spherical spiral, Seiffert's, 527

Spheroidal harmonics, 403
Squares of Bessel functions, 379, 380; of liypergeometric functions, 298; of Jacobian

functions (relations between), 492 of Theta-functions (relations between), 466
Statement of Fourier's theorem, Dirichlet's, 161, 163, 164, 176
elliptic
Steadily tending to zero, 17

Stieltjes' theorem on zeros of
equation, 417
Stirling's series for the
Lame
functions, 560,
(generalised) 562
theory of Mathieu's
Gamma-function, 251
Stokes' equation, 204
convergence of double
Stolz' condition for
series,
27
Strings, vibrations of, 160
Successive substitutions, method of, 221

Sum-formula of Euler and Maclaurin, 127
Summability, methods of, 154-156 of Fourier series, 169 uniform, 156

Surface harmonic, 392
Surface, modular, 41
Surfaces, nearly spherical, 332
Sylvesters theorem concerning integi-als of harmonics, 400
Symmetric nucleus, 223, 228
;
Tabulation of Bessel functions, 378 of complete elliptic integi-als, 518 of Gamma-functions, 253
failure of, 100, 104, 110
Taylor's series, 93 remainder in, 95
Teixeira's extension of Biirmann's theorem, 131
Telegraphy, equation of, 387
factorisation of, 536
Tesseral harmonics, 392
Tests for convergence, see Infinite integrals, Infinite products and Series
Thermometric parameter, 405
Theta-functions [î (;), S-o (z), &3 {z), ^4 {z) or ^ (,-), 9 ()], 462-490 (Chapter xxi) abridged notation for products, 468, 469
addition formulae, 467 connexion with Sigma- functions, 448,
duplication formulae, 488 expression of elliptic functions by, 473 four types
473, 487
of, 463; fundamental formulae (Jacobi's), 467, 488; infinite products for, 469, 473, 488;
Jacobi's first notation, G () and H (;/), 479; multipliers, 463; notations, 464, 479, 487;
parameters q, t, 463; jmrtial differential equation satisfied by, 470; periodicity factors,
463; periods, 463; quotients of, 477; quotients yielding Jacobian elliptic functions, 478;
relation S-i' = a-jS^3 ^4, 470; squares of (relations between), 466; transformation of, (Jacobi's
imaginary) 124, 474, (Landen's) 476
triplication formulae for, 490
with zero argument
zeros of, 465
(^9 ^:j ^4 ^1'), 464
Third kind of elliptic integral, IT (u, a), 522 a dynamical application of, 523
Third order, linear differential equations of, 210, 298, 418, 428
Third species of ellipsoidal harmonics, 537, (construction of) 541
Three kinds of elliptic integi-als, 514
Three-term equation involving Sigma-f unctions, 451, 461
;
Total fluctuation, 57
Transcendental functions, see under the names of special functions

Transformations of elliptic functions and Theta-functions, 508; Jacobi's imaginary, 474, 505,
Landen's, 476, 507 of Eiemann's P-equation, 207
506, 519
Trigonometrical equations, 587, 588
Trigonometrical integrals, 112, 263; and Gamma-functions, 256
Trigonometrical series, 160-193 (Chapter ix); convergence of, 161 values of coefficients in, 163
Eiemann's theory of, 182-188; which are not Fourier series, 160, 163. See also Fourier series
Triplication formulae for Jacobian elliptic functions and E {u), 530,534; for Sigma-functions,
459 for Theta-functions, 490 for Zeta-functions, 459
Twenty-four solutions of the hypergeometric equation, 284; relations between, 285, 288, 290
Two-dimensional continuum, 43
Two variables, continuous functions of, 67 hypergeometric functions (Appell's) of, 300
Types of ellipsoidal hannonics, 587
;
Unicursal, 455
Uniformisation, 454
GENERAL INDEX
608
Uniformising variables, 455 associated with confocal coordinates, 549

of, 52
Uniformity of continuity, 54; of sumniability, 156
Uniformity of convergence. 41-60 (Chapter iii), defined, 44; of Fourier series, 172, 179, 180; of
intiuite intei;nils. 70. 72, 73
of infinite products, 49
of power series, 57
of series, 44,
(condition for) 45, (Hardy's test for) 50, (Weierstrass'ili-test for) 49
Uniformly convergent infinite integrals, properties of, 73
series of analytic functions, 91,
(differentiation of) 92
Uniqueness of an asymptotic expansion, 153 of solutions of linear differential equations, 196
Upper bound, 55 integi-al, 61
Upper limit, integral equation with variable, 213, 221 to the value of a complex integi-al, 78, 91
;
Uniformity, concept
of the coefficients
Value, absolute, iêe Modulus of the argument of a complex number, 9, 588
in Fourier series and trigonometrical series, 163, 165, 167, 174
of particular hypergeometric
functions, 281, :i93, 298, 301; of Jacobian elliptic functions of Â', îK', h(K+iK'), 500,
506, 507; of K, K' for special values of A", 521, 524, 525; of ^(a) for special values of .*,
;
267, 269
Vanishing of power
series,
58
Variable, uniformising, 455; terms (series

integi-al equation with, 213, 221
Vibrations of air in a sphere, 399

of strings, 160
Volterra's integi'al equation, 221
Wave
motions, equation
of,
of circular
of),
see
Uniformity of convergence;
membranes, 396
of elliptic
upper limit,
membranes, 404, 405
reciprocal functions, 218
386; general solution, 397, 402; solution involving Bessel functions,
3S)7
Weber's Bessel function of the second kind [r,^(^)], 370

Weber's equation, 204. 209, 342, 347. See also Parabolic cylinder functions
Weierstrass' factor theorem, 137 il/-test for uniform convergence, 49 product for the Gammatheorem on limit points, 12
function, 235
Weierstrassian elliptic function [^>(^)], 429-461 (Chapter xx), defined and constructed, 432,
analogy with circular functions,
addition theorem for, 440, (Abel's method) 442
433
438; definition of \^ [z) - e^]^, 451; differential equation for, 436; discriminant of, 444;
;
expression of elliptic functions by, 448; expression of ip (z) (y)

homogeneity properties, 439 integral formula
half-periods, 444
invariants of, 437
inversion problem
integi-ation of in-ational functions by, 452
for, 437
for, 484; Jacobian elliptic functions and, 505; periodicity, 434; roots fj, e.,, e.^, 443.
See
also Sigma-fimctions and Zeta-function (of Weierstrass)
duplication formula, 441

by Sigma-functions, 451
Whittaker's function TFj. ,(),

Wronskis expansion, 147
see
Confluent hypergeometric functions
Zero argument, Theta-functions with, 464 relation between, 470

Zero of a function, 94 at infinity, 104 simple, 94
Zeros of a function and poles (relation between), 438 connected with zeros of its derivate, 121,
123 in-educible set of, 430 number of, in a cell, 431 order of, 94
Zeros of fimctions, (Bessel's) 361, 367, 378, 381, (Lame's) 557, 558, 560, 578, (Legendre's) 303,
316, 335, (parabolic cylinder) 354, (Eiemann's Zeta-) 268, 269, 272, 280, (Theta-) 465
Zeta-function, Z("). (of Jacobi), 518; addition formula for, 518; connexion with E{u), 518;
See also Jacobian
Jacobi's imaginary transformation of, 519.
Fourier series for, 520
elliptic functions
Zeta-fimction, j'(*), i'(s,a), (of Riemann) 265-280 (Chapter xiii), (generalised by Hurwitz) 265;
Hurwitz' integral for, 268
inHermite's integi-al for, 269
Euler's product for, 271
equalities satisfied by, 274, 275; logarithmic derivate of, 279; Eiemann's hypothesis
concerning, 272, 280; Eiemann's integrals for, 266, 273; Eiemann's relation connecting f (s)
and f (1 - *), 269 values of, for special values of .s, 267, 269 zeros of, 268, 269, 272, 280
Zeta-function, f(2), (of Weierstrass), 445; addition formula, 446; analogy with circular
connected with, 446; duplication formulae for, 459; exfunctions, 446; constants t/j
triplication formulae, 459.
quasi-periodicity, 445
pression of elliptic functions by, 449
See also Weierstrassian elliptic functions
Zonal harmonics, 302, 392 factorisation of, 536
;
t).,
CAMBRIDGE: PRINTED BY
J.
B.
PEACE, M.A., AT THE UNIVERSITY PRESS.
University of California
SOUTHERN REGIONAL LIBRARY

305 De Neve Drive
LOS
FACILITY
Parking Lot 17 Box 951388

ANGELES, CALIFORNIA 90095-1388
-
Return this material to the library from which
it
was borrowed.
UNIVERSITY
of
CALIFORNIA
-AT
LOS Ai-^GELES
LIBRARY

Course of Modern An 00 Whit

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Course of Modern An 00 Whit

Uploaded by

Copyright:

Available Formats

^

CAMBRIDGE UNIVERSITY PRESS

NEW YORK THE MACMILLAN

PROFESSOR OF MATHEMATICS IN THE UNIVERSITY OF EDINBURGH

PROFESSOR OF MATHEMATICS IN THE UNIVERSITY OF BIRMINGHAM

First Edition 1902

Second Edition 1915

Third Edition 1920

has been taken of the preparation of the third edition of

add a chapter on Ellipsoidal Harmonics and Lame's Equation,

chapter on Trigonometrical Series so that the parts

which are used in Applied Mathematics come at the beginning of the

are due to Miss

reading the greater part of the

to the staff of the University Press for

courtesy and con-

sideration during the progress of the printing.

THE PROCESSES OF ANALYSIS

Continuous Functions and Uniform Convergence

properties of Analytic Functions

The Theoryof Residues; application

to the evaluation of Definite Integrals

Asymptotic Expansions and Summable Series

of functions in Infinite Series

Fourier Series and Trigonometrical Series

Linear DiflFerential Equations

THE TRANSCENDENTAL FUNCTIONS

The Zeta Function

The Hypergeometric Function

The Confluent Hypergeometric Function

The Theta Functions

Hamionics and Lame's Equation

The decimal system of paragraphing, introduced by Peano, is adopted in this

fractional parts are arranged in each chapter in order of magnitude.

pp. 187, 188, 9-632 precedes

THE PROCESSES OF ANALYSIS

instance from the

consideration of the set of positive* integral numbers, or positive integers;

Positive integers have

in elementary Arithmetic, classes of

numbers are constructed such that the

operations of subtraction and division can always bB performed

numbers among which the operation of subtraction

have thus introduced three classes of numbers,

the signless integers,

the rational numbers.

not part of the scheme of this work to discuss the construction of

the class of integers or the logical foundations of the theory of I'ational

number, which has just been described, was not

without some opposition from the more conservative mathematicians.

In the latter half

and Frend (1757-1841) published works

of the eighteenth century, Maseras (1731-1824)

epithet would be, not positive, but signless.

t In the strict sense.

Such a discussion, defining a rational number as an ordered number-pair of iute^ers in a

THE PROCESSES OF ANALYSIS

number x may be represented

(Pi being on the right of 0),

P^ or the displacement OP^ (which

be written OPx) as repre-

All the rational

but the converse