Professional Documents
Culture Documents
MA<~ '|S|
Jjigraiiis
MARTIN GARDNER
$2.25
ItuurrBitjj
of $int\bn
(Ui?e
(gift
of
LIBRARY
LOGIC MACHINES
AND DIAGRAMS
Martin Gardner
New
York
Toronto
INC.
London
ARCH &
FINE ARTS
LIBRARY
:*>?*
.^
BBP
Copyright
may
not be reproduced
in
publishers.
any
for C. G.
Who
Lyrasis
http://www.archive.org/details/logicmachinesdia227gard
Preface
A logic machine
is
diagram
is
a geometrical
logic
histories.
Let no reader imagine that logic machines are merely the play-
things of engineers
automata constantly encounter problems that are less mathematical in form than logical. It has been discovered, for example,
that symbolic logic can be applied fruitfully to the design and
simplification of switching circuits.
It
And
tell
them what
in the
new
field
for
The
Although the book consists for the most part of material drawn
relatively inaccessible books and
journals, it also contains much that has not previously been pubfrom widely separated and often
The reader
will find,
PREFACE
viii
for example,
some unfamiliar
known Venn
circles;
so
much
of the
first
chapter
is
that
Ramon's
dull to those
life
life is
its
modern formalization
it
is
book
so
little
notation
is
class
it
as
symbols for
it
seemed
best
it.
all sorts
of help in
all sorts
of ways.
Martin Gardner
Contents
PREFACE
Vll
Ramon
2.
Logic Diagrams
3.
4.
5.
Jevons's Logic
6.
104
7.
Window Cards
117
8.
Electrical Logic
9.
The Future
of
28
Network Diagram
index
Lull
Machine
60
80
91
Machines
125
of Logic Machines
140
153
1:
Ramon
of
Lull
isles off
Mount Randa
rises
abruptly
logic
Logic Machines
and Diagrams
cis
Bacon
And
yet
in
to mention, that
De augmentis (Book
VI,
ostentatious
than learned, have laboured about a kind of method not worthy to be called
method, being rather a method of imposture, which neverthewould no doubt be very acceptable to certain meddling wits. The object
of it is to sprinkle little drops of science about, in such a manner that any
sciolist may make some show and ostentation of learning. Such was the Art
a legitimate
less
Typocosmy
may
all
arts, to
who
are ready
Such
collections are like a fripper's or broker's shop, that has ends of everything,
Swift
is
in
III,
or study."
On
the other
hand we
find
teaching
it
to wealthy
noblemen
we
in
find
it had become a
young Leibnitz fascinated by
Venice where
The Ars
Lull's
Magna
of
Ramon
Lull
his Dissertio
de arte
were
to arise," Leibnitz
countants. For
to
sit
down
to witness,
it
would
to their slates,
if
and
to say to
fantastic
work of
Or was
Lull's
method
little
now
interest to sketch
the extraordinary,
briefly
forgotten Art,
it
and attempting
will
perhaps be of
almost unbelievable
inventor. 2
its
Ramon
teens he
the
career of
more than
became
King James
Aragon
of
It
her, only to
by
this
him
praising in
to her
poems
* Superscript
numbers designate
partially
and Diagrams
Logic Machines
is
eternal!"
Lull retired in great shame and agitation. Shortly after this incident, while he
was
in his
lyrics,
On
four
to
morning confession
as a
win nothing
less than the entire Moslem world for Christianity. It was an obsession that dominated the remainder of his life and eventually brought
about his violent death. As the first necessary step in this ambitious
Lull's conversion
desire to
home
He purchased
him
a Moorish slave
who
lived
hearing
the name of Christ. Soon thereafter the Moor reby attacking Lull with a knife. Lull succeeded in disarming
him and the slave was jailed while Lull pondered the type of punishment he should receive. Expecting to be put to death, the Moor
hanged himself with the rope that bound him.
him blaspheme
taliated
Before
this
tion. It is
to prove
managed
to finish
book, the Book of Contemplaa massive, dull work of several thousand pages that seeks
by "necessary reasons"
first
all
the
major
truths of Christianity.
tween truths of natural theology that he believed could be established by reason, and truths of revelation that could be known only
Lull found this distinction unnecessary.
by
faith.
all
the leading
dogmas
is
believed that
He
by
Book of Contemplation reveals his early preoccupation with a number symbolism that was characteristic of many scholars of his time. The work
Lull had not yet discovered his Great Art, but the
The Ars
is
Magna
of
divided into
Ramon
five
Lull
books
honor of the
in
wounds
five
of Christ.
Forty subdivisions signify the forty days Christ spent in the wilderness.
making a
Of
introduced as metaphors.
special interest to
modern
logicians
is
example,
in
tion
own work
but
asserts,
stronger
it
is
who
in the value
the
one true
follows
moral
its
will
It
he
faith,
teaching a
the
world," and he assures the reader that he has "neither place nor
all
and great."
These immodest sentiments are
who become
characteristic of
and
it
is
book
is
good
most eccentrics
similar sentiments
God
the
Holy
Spirit.
An
men
A dan, Solomon
el
oft-repeated jingle
mundo,
Ray mundo.
in
the world
Logic Machines
many
of
is
Some
much
and Diagrams
numerous although
and
repetition of material
Most
name
are
known
to be spurious).
doctrines by
means
of "necessary reasons," or to
Some
combat Averroism,
Many
some 6,000
Book
of
most of which concern the application of his Great Art, are devoted
to almost every subject matter with which his contemporaries were
concerned
astronomy, chemistry, physics, medicine, law, psychology, mnemonics, military tactics, grammar, rhetoric, mathematics,
The
protagonist,
who
closely resembles
The Book
is
become
of the
office,
a contemplative hermit.
Lull's best
known work,
hermit's pen.
it
was
later to
become common
Theresa and
Magna
The Ars
of
Ramon
symbols
a fertile
field.
will find
Lull
we
Book
of
Marvels,
find
Miller's fiction. It
is
difficult
Mount Randa,
it
became church
his conviction
grew
evangelical character.
by
He
his
from his father, distributing his possessions to the poor. His wife
and child were abandoned, though he provided liberally for their
welfare. He made endless pilgrimages, seeking the aid of popes and
princes in the founding of schools and monasteries where his Great
Logic Machines
is
no
and Diagrams
is
ceeded to stand and deliver a loud and lengthy oration on the perfections of
haved
On
as a
God. The story is believable because Lull always bepossessed by inspired, irrefutable truth.
man
made voyages
to Africa to clash
with his
life.
white and his eyes burning with desire for the crown of martyrdom,
he
set sail
the streets
of
Moslem
faith.
merchant ship
to
mob and
which
man from
his bruised
the merchants
.
Lull in his
own
The Soul
carried.
Columbus)
new world.
Ellis (in a
person so brilliantly
Spain.
He was
,!
legend
their country.
all
of one of
chapter on
summed up
in his
making of
lover, a soldier,
San Francisco,
at
relics
now
saint,
rest in
venerated as those of a saint, in spite of the fact that Lull has never
been canonized.
In turning
now
to
itself,
it
is
One
it
were otherwise.
It
it is
much
it
adequately,
Magna
The Ars
Ramon
of
Lull
in the
exhausting
to explore
minds.
To
all
Figure
and rotating
all
circles.
we
call
are able
upon the
list
1.
sets of categories in
haust
we
two
all
two
vertical
columns (Figure
Or we can arrange
a set
lines as
shown.
we
third
pride,
shown
is
to place
two or more
in Figure 3.
By
in
sets of
we
easily obtain a
combine,
this
mechanical method
Logic Machines
10
short-cut
method
men
of
and Diagrams
little
We
by Count Alfred Korzybski to illustrate principles of general semantics. Perhaps there is even a touch of the same awe in the reverence
with which some philosophers
view symbolic logic as a tool of
philosophical analysis.
Before
going
complicated
method,
aspects
of
concrete illustrations of
circles.
The
is
letter
sixteen
3.
first
is
of his
circle.
now
had no existence
for sixteen
greatness
in
Lull
called
Around
Figure
how
A.
"A," representing God,
The
Lull's
let
used his
more
the
into
through
we
(omitting / which
divine attributes
(magnitudo),
for goodness
for eternity
(bonitas),
(etemitas),
for
and so on.
By drawing
BC BD BE BF BG
CD CE CF CG
DE DF DG
EF EG
FG
BH
CH
BK
CK
DH DI DK
EH EI EK
FH FI FK
GH GI GK
BI
CI
HI
BM
CM
DL DM
EL EM
BN
CN
DN
EN
FL FM FN
GL GM GN
BL
CL
HK HL HM
IK
BO BP BQ
CO CP CQ
DO DP DQ
EO EP EQ
FO FP FQ
GO GP GQ
HN HO HP HQ
BR
CR
DR
ER
FR
GR
HR
IL
IM
KL
KM KN KO KP KQ KR
IN
IO
LM LN LO
IP
IQ
IR
LP LQ LR
MN MO MP MQ MR
NO NP NQ NR
OP OQ OR
PQ PR
QR
Figures 4 to 9,
left
to
right,
celona, 1923.)
II
Logic Machines
12
and Diagrams
many
theological difficulties.
this a
demonstration
we
them back
to a common origin. Free will and predestination sprout from
equally necessary attributes of God, like two twigs growing on
branches attached to the trunk of a single
Lull's
letter S.
tree.
Four
trace
is
designated by the
ignorant (P), and a will that loves and hates (Q). Lull considered
this last state the unhealthiest of the four.
way
We now
superimpose the
The Ars
Magna
of
Ramon
The
136 combinations of
to obtain
It
13
Lull
faculties
faculties.
notion of
imposed
to
form a
one
As
letter at
each vertex
compartments that bear the same color as tHe polygon for which
they mark the vertices. The meanings of the letters are: God, creature, and operation (blue triangle); difference, similarity, contrariety (green); beginning, middle, end (red); majority, equality,
minority (yellow); affirmation, negation, and doubt (black). Roin
same
which he called
and the seven deadly sins. The fourteen categories are arranged alternately around a circle in red (sinful) and blue (virtuous) compartments (Figure 7). Drawing connecting lines, or rovirtues
when
it
our attention
when
lust
figure
letters
combinations of terms.
14
Logic Machines
and Diagrams
this method. He
even produced a book on how preachers could use his Art to discover new topics for sermons, supplying the reader with 1 00 sample
sermons produced by his spinning wheels! In every case the tech-
nique
is
Dozens of
his
Some of these
more comprehensive treatises. Some are
less intellectual readers who find it hard
to
to
comprehend
the
more involved
Figure 9 )
in
figures.
ophy, Vol. 3), and in most modern histories of thought that find
space for Lull's method. Two of Lull's treatises on his Art are
written entirely in Catalan verse.
One
God,
is
angel, heaven,
Where does
the cold go
when
a stone
is
warmed?
In another interesting work Lull uses his Art to explain to a hermit the meaning of some of the Sentences of Peter Lombard. The
slain in the
tism of
a bapangels
Can God
make matter without form? Can He damn Peter and save Judas?
Can a fallen angel repent? In one book, the Tree of Science, over
pass from one place to another in an instant of time?
Magna
The Ars
of
Ramon
15
Lull
four thousand such questions are raised! Sometimes Lull gives the
combination of terms
in
At other times he
un-
answered.
The number
varies
its
two
The mind
figure
common.
same
reels at the
no
less
of topics that
can be explored by
Before passing on to an evaluation of Lull's method, it should be
mentioned that he also frequently employed the diagrammatic device of the tree to indicate subdivisions of genera and species. For
this fantastic instrument.
Lull
it
was both an
illustrative
example, pictures
his subject
matter as a
tree with four roots (the four humors) and two trunks (ancient
and modern medicine). The trunks branch off into various boughs
on which flowers bloom, each flower having a symbolic meaning
(air, exercise, food, sleep, etc.). Colored triangles, squares, and
None
all his
medical works,
argued,
if
There
God had
is
no doubt about
Lull's
how
his circles
By means
of
fire,
Logic Machines
16
and Diagrams
These are then combined in various ways with the signs of the zodiac
answer medical questions concerning diet, evacuation, prepara-
to
and so on.
Doctor Illuminatus
as he was later called, ever seriously doubted that his Art was the
product of divine illumination. But one remarkable poem, the
Desconort ("Disconsolateness"), suggests that at times he may have
been tormented by the thought that possibly his Art was worthless.
tion of medicines, fevers, color of urine,
There
is
The poem
no indication
that
Ramon
Lull, the
is
on
same rhyme.
It
opens with
woods, he comes upon the inevitable hermit and pours out to him
the nature of his sorrows.
laugh
at
him and
He
him a
call
is
fool.
is
men
and
ridiculed
ig-
Ramon
"as fast as a cat that runs through burning coals," perhaps this
is
would there be in believing them? In addition, the hermit argues, if Lull's method is so valureason. If they could be, then what merit
able,
And
how
if it
is
it
truly
to fear
it
will
ever be lost?
Lull replies so eloquently to these objections that
the hermit begging forgiveness for
Ramon
in his labors,
all
he has
we soon
find
life!
method
is
the legend of
of Franciscans.
North Africa, but at the last moment, tormented by doubts and fears of imprisonment and death, he allowed
the boat to sail without him. This precipitated a mental breakdown
that threw him into a state of profound depression. He was carried
into a Dominican church and while praying there he saw a light
like a star and heard a voice speak from above: "Within this order
thou shalt be saved." Lull hesitated to join the order because he
knew the Dominicans had little interest in his Art whereas the
first
missionary
trip to
The Ars
Magna
Ramon
of
from the
light,
17
Lull
of value.
it
time threateningly:
this
"And
did
not
tell
thee
that only in the order of the Preachers thou wouldst find salvation?"
it
would be
better to
clear
possessing
from
many
values.
standing by making
it
method
as
When
of doctrines. It
is
combined in all
mind thinking along novel channels and one is led to
discover fresh truths and arguments, or to make new inventions.
ideas are
method
He was thoroughly
and his writings even include the
popular medieval diagrams of immediate inference and the various
syllogistic figures and moods. He certainly did not think that the
mere juxtaposition of terms provided in themselves a proof by
"necessary reasons." He did think, however, that by the mechanical
formal logic of Aristotle and the schoolmen.
natural for
him
to
all
possible structures of
truth
obviously will do this for us in an irrefutable way. Considered mathematically, the technique
is
in
its
day
it
was
es-
Logic Machines
18
sentially trivial.
and Diagrams
and
if
was
to
is
it
does
We
saw in
For example, in his Dissertio de arte combinatoria Leibnitz constructs an
exhaustive table of all possible combinations of premises and conLull's
method
it
it
first
to formal logic.
somewhat
was used
how
restricted
terms built
dream.
Lull's similar
It
is
is
is
there
is
so remote that
it
it
The Ars
Magna
of
Ramon
19
Lull
ing
Anagram
make
use of Lullian-type
for
letters.
and half on the sheet to which the wheel is fastened. Turning the
produces amusing combinations
a giraffe's head on the
body of a hippopotamus, and so on. One thinks also of Sam Loyd's
famous "Get off the earth" paradox. Renan once described Lull's
circles as "magic," but in turning Loyd's wheel the picture of an
circle
to fiction writers
many
By
titled,
if
remember
cor-
(One
suspects that
He
wheels
Mention
also should be
made
of the
book
wheels,
To
it
dramatize
this dialectical
20
Logic Machines
and Diagrams
letter
might prove
to picture
useful.
all
means
covered. Lull's crude anticipation was a circle bearing the four tra-
a testing of
all
stances or techniques.
of finding
new and
Thomas Edison
What
is
invention, after
all,
number
of sub-
When
1944 a
book called The Trick Brain in which he explains a technique for
combining ideas in Lullian fashion for the purpose of inventing new
magic tricks.
If the reader will take the trouble to construct some Lullian
circles related to a subject matter of special interest to himself, and
play with them for a while, he will find it an effective way of getting
close to Lull's mind. There is an undeniable fascination in twisting
the wheels and letting the mind dwell on the strange combinations
professional magician, Dariel Fitzkee, actually published in
many
centuries.
For persist it did. 10 Fifty years after Lull's death it was strong
enough to provoke a vigorous campaign against Lullism, led by
Dominican inquisitors. They succeeded in having Lull condemned
as a heretic by a papal bull, though later church officials decided
that the bull had been a forgery. Lullist schools, supported chiefly by
Franciscans, flourished throughout the late Middle Ages and Renaissance, mostly in Spain but also in other parts of Europe.
We
The Ars
Magna
of
Ramon
21
Lull
The
great ex-
compass made
in
Rome
Museum
in
1593
in the
Figure 10. Sixteenth-century portable sundial engraved with Lullian figures. (From Archaeologies,
On
Oxford, 1925.)
and back of the two gilt copper "covers" are enLullian circles shown in Figures 11 to 14. For an ex-
the front
graved the
"A
rived from
Ramon
referred to O.
is
Form
M.
Dalton's
89-102.
The seven smaller diagrams in Figure 12 are all from Lull's
writings u and perhaps worth a few comments. The square in the
upper left corner is designed to show how the mind can conceive
make
two large
triangles,
B and
C.
one of which
Each
diagonal diis
subdivided
However, we can
easily
all.
Vol.
Arciiaeologia
LXXIV
Plate
Sapientia ^sdificavtt
SIBI DOMVM.i^Jromrr.ji
&
tacta
A
wonrw
RERVltfvo
BONITATEM
AL.
1UET
I.
Tin
DIScrFLI.
Upper
RI.
mrr
ABAC AT
SABCBD
CACBCD
DDAMDC
fjif.
2.
:nt*i
CONCVPIS-CEN5 5APIET
COKJERVA IVJTITIAM
PRE JEBIT 1RA M TUI C3:.
XXX.
Upper
FlGVRA DEMONSTRATIVE.^*
Efl^PE,
)CTRINIS
Ante obitvmo?
re iv.rnTJAM _
Figures
11
to
Lower
14,
left
to
right,
22
Lower
*S**r\yL
by Renaissance
Lullists.
(From
The Ars
Magna
Ramon
of
23
Lull
way our
The
do not
eyes, but
we can
intellect,
In this
all.
We
see
geometrical truths.
The top
right square
is
The two
only one
is
circles repre-
God
could create at
will, since
to higher
His power
is infinite.
The four
The four
next chapter).
letters
Esse (being),
(the good).
(the true),
and Bonum
The overlapping
of
the circles indicates that the four qualities are inseparable. Nothing
can
exist
The
verse, but I
am
problem
The lower
in navigation.
It
left
The lower
right square
A BCD
taken two
letters at a time.
We
first
the bottom,
is
a primi-
common
fairly
in
24
Logic Machines
we draw
and Diagrams
a third square
midway
be-
asserts,
has
area equal to the circle's area. Lull's discussion of this figure (in his
Ars magna
et
how
behind he was of the geommethod does not provide even a close approxi-
ultima) reveals
far
Books on the Lullian art proliferated throughout the seventeenth century, many of them carrying inserted sheets of circles to
be cut out, or actual rotating circles with centers attached permanently to the page. Wildly exaggerated claims were made for the
method. The German Jesuit Athanasius Kircher (1601-1680),
scientist,
was also a confirmed Lullist. He published in Am1669 a huge tome of nearly 500 pages titled Ars magna
sciendi sive combinatoria. It abounds with Lullian figures and cirhieroglyphics,
sterdam
in
The
symbols of his
devising.
14
and pamphlets
in the
own
second volume of
his
was followed
in
1778 by
defense of Lull.
periodical devoted
many
enthusiastic
from 1901 to
admirers of Lull in
all
his
Art
seems
much
that
it
sistence
that
on the coloration of a suicide. And finally, his inon the divine origin of his Art and his constant emphasis
takes
The Ars
on
its
Magna
of
Ramon
25
Lull
life
to"
was
Lull's personality.
a fantastic one.
He
even wrote a book called The Dispute of a Cleric and Ramon the
Fantastic in which he and a priest each try to prove that the other
who
life.
of
of
men
No
versy over his Great Art has at last been laid to rest and that the
world
is
free to
first
References
1.
In later years Leibnitz was often critical of Lull, but he always regarded as
sound the basic project sketched in his Dissertio de arte combinatoria. In a
letter written in 1714 he makes the following comments:
"When I was young, I found pleasure in the Lullian art, yet I thought also
that I found some defects in it, and I said something about these in a little
schoolboyish essay called On the Art of Combinations, published in 1666,
and later reprinted without my permission. But I do not readily disdain anything
except the arts of divination, which are nothing but pure cheating
and I have found something valuable, too, in the art of Lully and in the
Digestum sapientiae of the Capuchin, Father Ives, which pleased me greatly
because he found a way to apply Lully's generalities to useful particular
problems. But it seems to me that Descartes had a profundity of an entirely
different level." (Gottfried Wilhelm von Leibniz: Philosophical Papers and
Letters, edited and translated by Leroy E. Loemker, University of Chicago
In sketching Lull's
life I
nificent biography,
have relied almost entirely on E. Allison Peers's magLull, London, 1929, the only adequate study of
Ramon
and briefer biography, Raymond Lull, the Illumiin London, 1904, by W. T. A. Barber, who also
contributed an informative article on Lull to the Encyclopedia of Religion and
Ethics. Other English references worth noting are: Otto Zockler's article in
the Religious Encyclopedia; William Turner's article in the Catholic Encyclopedia; George Sarton, Introduction to the History of Science, 1931, Vol. II,
pp. 900 ff.; and Lynn Thorndike, A History of Magic and Experimental
Lull in English.
An
earlier
peared
in Paris in
The most
It is
life
1927.
An
5.
in the
En-
of Historia de
la
cit.,
p. 64.
Lull's death
Lully," in his
7.
on Lull
1
4.
6.
and Diagrams
Logic Machines
26
The only
is
satisfactory description
in
is
in
Vol.
may
be consulted for a
list
Chapter 7 of my Mathematics, Magic, and Mystery, 1956, contains a reproduction and analysis of Loyd's "Get off the earth" puzzle and several related
paradoxes.
10.
11.
1925.
My
12.
plying
13.
It
on
Lull's figure as
my
AB
shown
attention
is
in
drawn
Figure
the circle.
14.
Kircher's
fas-
and
have anticipated motion pictures by constructing a magic lantern that threw
Figure 15.
images on a screen in fairly rapid
succession to illustrate such events
as the ascension of Christ. He invented (as did Leibnitz) an early calculating
machine. On the other hand, he devoted a 250-page treatise to details in the
construction of Noah's Ark!
cinating
nonsense.
mixtures
of
He seems
to
science
The Ars
Magna
Kircher's
of
Ramon
work on
27
Lull
the
thing similar to
analysis of
papers and
it,
human
title
(see reference
1).
but that the author had not even dreamed of the true
thoughts."
(Vol.
1,
1.)
p.
2:
Logic Diagrams
logic
diagram
is
a two-dimensional geo-
metric figure with spatial relations that are isomorphic with the
structure of a logical statement. These spatial relations are usually
is
fact that logic relations are the primitive relations underlying all
spatial structures.
Logic diagrams
same basic
structure.
among
grams
down
their noses
at logic
dia-
it.
when
Many
individuals
Logic Diagrams
29
may be
in the
same
efficient
is
way
that a
graph
True
and more reliable, but this is not always the case, and even
when it is, the diagram affords a convenient technique for checking
results obtained by other means. Finally, the study of logic diagrams is an intensely interesting and relatively unexplored field. It
is closely allied with the rapidly growing subject of topology, and
its kinship with the network theory underlying the construction of
electronic calculators and other automata suggests that it may
have contributions to make in the near future that will be much
faster
more than
trivial
or recreational.
Historically, the
logic
first
state-
ments in what today is called the logic of relations. The tree figure,
for example, was certainly known to Aristotle as a handy way of
picturing successive subdivisions of matter and form, or genera and
species. The so-called tree of Porphyry, so often found in medieval
and Renaissance
logics,
and
its
is
one example of
we spoke
type of diagram. In
figures)
this
(actually an interlocking of
many
way
way
to determine a relationship
of saying that
it
is
separate tree
is
often the
between two
such as
"taller than," "heavier than," "to the left of," and so on, are so
easily diagramed that the technique must have been a familiar one
little
interest to us
Logic Machines
30
and Diagrams
it
became attached
later
ently
first
appeared in a
Tartaretus.
(The
The phrase
to the elaborate
fifteenth century
work on
logic
by Petrus
norum became
common
proposition
fifth
On
had
difficulty in crossing.
resemblance of
grams pictured
Figure
6.
it
signifies)
MAJOR
rela-
CONCLUSION
MINOR
CONCLUSION
CONCLUSION
Figure 16.
tions
between terms
era.
They
fifth
commentary on
The
Figure 17.
first
important
diagrammatic
step
method
toward
sufficiently
Logic Diagrams
31
logic
We
that
to represent a class.
the
The
transcendental
circles to
predicates
of
show
truth,
illus-
Lange
(in
his
mathematician,
it
who was
brilliant Swiss
He
first
them
into the
letters,
first
time
we meet with
in class logic.
sciences,
method
will not
Cambridge
may be
revised
its
his-
may
also be con-
as well as
and other parallelograms. Since class inclusion is not concerned with numerical
quantity, any closed curve topologically equivalent to a circle can
be used. Of special interest is a linear method of diagraming, closely
allied to the Euler circles, that was developed by Euler's contemporary, the German mathematician Johann Heinrich Lambert, and
later variations that
employ
triangles, squares,
Logic Machines
32
and Diagrams
They were
more
discussed
fully in his
book
is
cited above.
moment and
doing so
it
be of
will
be
on much
make
it
S P
S P
Some 5 P
O Some S
All
is
(universal affirmative)
E No
is
(universal negative)
is
is
(particular affirmative)
not
(particular negative)
correctly inferred
other relating
sion.
For example:
All
It
is
All S
is
All S
is
was
this specific
was the
first
to dissect
way
of a mid-
to
Logic Diagrams
whether
33
concerned
or part of
all
its
class)
was divided
into
The
is
this case
inferences.
modern symbolic
logic
we now
see
it
as a restricted
form of
class-
speech.
line of Philosophy,
modern
1927,
Russell's
An
Out-
is
This form of inference does actually occur, though very rarely. The only
instance
produced
much
among
others to a certain
he argued: "Everything in
this book is a joke, therefore the advertisements are jokes." I have never
come across any other case of new knowledge obtained by means of a
last
syllogism.
Bacon spoke
it
to cover
new forms
of inference. Francis
and over
Logic Machines
34
and Diagrams
if
anything
is
it
in a
is
hypothetical form:
M;
therefore
human
is
if
all
is
anything
is
it
is
is
P.
it
is
Franz
not a not-mortal
If
anything
human;
there
is
not a not-
helm Schuppe decided that Aristotle was mistaken when he said that
no conclusion could be derived from two negative premises. For can
we not reason: No
is
is P; no 5 is M; therefore neither S nor
P? And if that didn't prove the point, Schuppe had another example: No
is P; no S is M; therefore S may be P. Of course S may
be P even without the premises, and in either case it may also not
be P. Nevertheless we can say for certain that we cannot say for
certain anything about the relation of S to P. Schuppe felt that this
Schuppe
Thus:
Some
Logic Diagrams
35
before him, that the predicate term in each of Aristotle's four basic
assertions (A,E,I,0)
us whether
not,
we
is
ambiguous
it
does not
tell
Why
ambiguous "All S is P" we substitute the two fully quantified assertions, "All S is all P" and "All S is some P." The old logic would
treat "All men are mortal" and "All men are featherless bipeds"
as identical in form; whereas in the new system we see at once that
the first statement is an example of "All S is some P (all men are
some mortals) and the second is an example of "All 5 is all P" (all
men
we
at
is
The
no reason
trouble
at all
is
mon
why
way
really
the predi-
that in doing so
in
we
which com-
we develop
the
English
mathematician
Augustus
by
his con-
De Morgan
here because
it has the superficial appearance of a diagram. Actuno attempt to find a spatial analogue of classes. The
system consists only of symbols with agreed-upon meanings, and
rather cumbersome symbols at that in spite of Hamilton's own
opinion that they were "easy, simple, compendious, all-sufficient,
consistent, manifest, precise, and complete." The system employed
the English C and the Greek capital gamma (each the third letter
ally,
in
its
there
is
alphabet) for the two terms of the conclusion, and A/ tor the
Logic Machines
36
and Diagrams
c,
comma
term, a
Figure is.
to signify
an undistributed (par-
ticular) term.
how Hamilton
wedge-shaped marks
his
to
form triangular
them
to
Figure 19.
Bruno's similar
even width,
effort.
The
Hamilton's
represents
"unfigured
syllogism."
By
This
transform-
in statements of equality.
Socrates and
Socrates and
Some
it
For example:
because
it
equal.
who were
seeking
Logic Diagrams
37
comprehend even
common
In
it
and
proved
one says "Some of Picasso's paintings are mediocre," one does not
want to quantify the predicate, it being obvious that there are
other mediocre things that are not paintings by Picasso. It is one
if
it is
constructed
De Morgan
By
more
to
same thing. Hamilton accused De Morgan of plagiarand for many years the two men argued with each other in
books and magazine articles perhaps the bitterest and funniest
of saying the
ism,
debate about formal logic since the time of the schoolmen, though
They
fought,
as well as insight
De Morgan once
was on the
side of
De Morgan
De Morgan.
always maintained
health,
though
at
in the
ailing
De Morgan's many
matical
skill,
in the
the
arm.
proved more
fruitful
men
we cannot be sure
who are in-
term "exemplar"
refers.
De Morgan
writes, to a system he
worked
38
Logic Machines
and Diagrams
some X's
are Z's.
The second
ex-
ample presupposes that 100 Y's exist. We can now reason: 70 A"s
are Y's; 40 Z's are Y's; therefore at least 10 X's are Z's. Boole,
Jevons, and many other pioneers of modern logic discussed syllogisms of this type at considerable length. 3 "Onzymatic" refers to
De Morgan's expansion of the Aristotelian system by the use of
negative terms and quantified predicates (see his Syllabus of a
Proposed System of Logic, 1860). As an example of a transposed
syllogism he cites: some A"s are not Y's; for every X there is a Y
which is Z; therefore some Z's are not A"s.
The
initial
syllogism,
letters
of the
De Morgan
names of these
six
new
varieties of
"Rue
He
adds, however, that followers of the old logic can take comfort
from the
No!"
same
letters
can be transposed to
spell
"True?
proposed by various logicians of the last century) are of little interest to a modern logician. They were courageous verbal attempts
to extend the domain of formal logic beyond its traditional boundaries,
making an old
Some
are merely
new
minus
verbal ways
3 equals 5 in-
39
Logic Diagrams
no longer a
Few
so.
duced
to the
figures or four.
is
or
is
not re-
first
or whether
quickly was that, shortly after Boole laid the foundations for an
algebraic notation, John Venn came forth with an ingenious improvement on Euler's circles. The result was a diagrammatic
method so perfectly isomorphic with the Boolean class algebra, and
picturing the structure of class logic with such visual clarity, that
new
logic
was
To understand
first
sect
all
to a syllogism.
like
the
about.
exactly
We
works,
it
us apply
let
(subject),
(Figure 20).
The
(predi-
(In this
tion
book we
shall
by placing a
way
that,
to indicate
if
its
The
region outside
P, or
all
all
and
their nega-
Logic Machines
40
and Diagrams
~S~M~P
Figure 21.
Figure 20.
therefore
we shade
all
compartments
in
which we
find these
two
Our second
premise,
let
us say,
is
"No
is
Figure 23
Figure 24.
Figure 25.
Logic Diagrams
41
doing, in a sense,
is
now
tion
arises,
the other.
We
have
is
in the
Venn
circles
and
in the syntax of a
same
as
structure
Peirce ex-
it, is
showing
all
is
either
or Z, taking "or" in
z
Figure 26.
this
is
done.
require
other
stratagems.
Peirce
way
suggested
(Collected
Papers,
volves the use of X's and O's to stand for presence or absence of
members, then connecting them by a line to indicate disjunction.
For example, Figure 28 shows how Peirce diagramed the statement "Either all S is P or some P is not-5, and either no S is P or
no not-S is not-F."
all
is
then
all
42
and Diagrams
Logic Machines
We
cit., p.
how
shall see
this is
done when we
Venn
Venn diagram
cir-
numerical
Figure 29 shows
AAAAAAAAAA
V_
Figure 29.
A's of which four are Z?s; eight A's are Cs; therefore at least two
#'s are Cs.
method
315), but
his proposal
is
more notational
than diagrammatic.
One
being impossible to
(it
desired manner).
is
make
to use ellipses as
shown
in
Figure 30
method.
Every
Every
All
If
we
is
either
X and
not Z, or
and not X.
is
we diagram
is
As
either
shown
X
in
or
W.
Figure 3
it
im-
to exist.
number of terms increases, the diagram of course becomes more involved. It is possible to draw any number of closed
the
Logic Diagrams
43
volved, the
five terms,
Venn proposed
shown
the diagram
Venn thought
Figure 30.
ing
all
parts of
it
in
best to
Figure 32.
and Y.
abandon hope of keepbut inside
Figure 31.
one
in-
Beyond
it is
number
and simply
to di-
of subcompartments,
The
sort
first
was an
"A
this
at
It
ap-
~ AB ~ C ~DEF. By
problems involving
with
Venn
circles.
the areas
known
to
have members,
six
of a noniconic nature.
Logic Machines
44
/
\
~B
N
~c
and Diagrams
~c
~c
~c
E^
~F
o<
-E<
~F
E^
-D<
'E-<
~F
Figure 33.
row
strip
spectrum." Null classes are indicated by gray shadCompartments excluded by the premises are shaded black.
this a "logical
ing.
~B
AAAAAAAA
~C
~D
~D
~C
~D
M2
~D
~D
ABCD
~D
~C
~D
~D
~A~B~C~0
Figure 34.
Logic Diagrams
45
first
explained in an article
ment
American Association
"The
titled
p.
286.
Advance-
Method of the Logical Spectrum to Boole's Problem," Macshowed how easily his diagram solved an involved problem
posed by Boole on p. 146 of his Laws of Thought.
A method of dividing a square, somewhat different from Marquand's, was proposed by William J. Newlin, of Amherst College,
of the
farlane
in
an
article titled
"A New
539.
Still
3, Sept.
13, 1906,
dis-
moments
"Two
No.
1909, p. 31.
Another interesting Marquand-type graph, using colored counters to indicate presence or absence of class members, was invented
(not-Z). The
left side
of the square
is
is
between
this
inner
Y\
(the
square
M'
is
the area
dia-
gram.
To diagram
we simply mark
the ap-
known
to contain
empty.
If
Logic Machines
46
and Diagrams
has members, but are not sure which one, the red counter is placed
on the border between the two areas. After we have suitably
marked the graph in accord with our premises, inspection of the
diagram
if
any, that
we may reach
THE GAME
"
LOGIC."
Instructions,
this
Game,
for
playing
will be found in
is
to
in
Y.
The
clusion.
Venn and
Logic Diagrams
syllogistic
47
form.
typically
Carrollian problem:
All teetotalers like sugar.
No
If
we
All
is
not-M
we
"No Y
have:
M are X
No Y
By
we
not-A"' or
"No
we may draw
we have
seen,
De Morgan and
others in-
"They have a
sort of
And
thus, having
of,
are of
no use
at all
when you
consider
possible forms."
if
we can but
In a later
all!'
let
of Euler
My Method of Diagrams resembles Mr. Venn's, in having separate Compartments assigned to the various Classes, and in marking these Compartments as occupied or as empty; but it differs from his Method, in assigning
a closed area to the Universe of Discourse, so that the Class which, under
Mr. Venn's liberal sway, has been ranging
suddenly dismayed to find itself "cabin'd,
Cell like any other Class!
at will
through
cribb'd,
Infinite Space,
confined,"
in
is
limited
it
Logic Machines
48
and Diagrams
pictures a
for eight
U
bat
Syllogism
out.
toorfccfc
me
off
toljen
tije
gatonmg
Fm
listening to some-
\Lf\t
premisses, separately.
Or
premisses, rombmrt.
o
o
c
fir (Conclusion.
Oat
tfje
36, shows
how
were developed before the modern truth-value propositional calculus assumed its present form and importance. As we have seen, the
terms of class logic stand for classes. In the propositional calculus,
terms stand for statements (such as "It is raining") which may be
Logic Diagrams
49
regarded as true or
false,
logically related
"if
by such
then."
The
problems
The answer
yes, as
is
Venn
with surpris-
ing efficiency.
The
member
interpreted to mean,
is
"The
is
true then
same thing
doors,"
it
is
ways of
To
stating the
is
is
true
true." "If
if
when
it
say the
stay in-
make
same underlying
Venn
may be
true"
class statement.
is
becomes a
on which
logical structure.
we must first
them in a different way. Each circle now stands for a
proposition which may be either true or false, rather than a class
which may or may not have members. The labels on the various
compartments (Figure 37) indicate possible or impossible combinations of true and false values of the respective terms. Just as we
formerly shaded a compartment to show that it had no members,
use the
interpret
we now shade
it
to indicate that
it
is
an impossible combination
compartment indicates
~A ~B ~C
three. This
is
done
is
shown
when
necessary, of this
area.)
If
we wish
containing
all
to
~A
show
areas containing
A and ~ A
that
(Figure 38).
is
(Figure 39).
Logic Machines
50
and Diagrams
lowing
like positive
this
simple procedure.
Let us see
tion, "If
is
true, then
table for this relation tells us there are four possible true
and
false
combinations (TT, TF, FT, and FF) of which only the combina-
TF
Hence we
compartments containing
A ~ B. The result (shown in Figure 40) is, as we would expect, a
diagram identical with the diagram for the class statement, "All A
tion
is
is
invalid.
eliminate
all
BT
The nature of
we add
figure. If
"material implication"
to
it
is
it
will
tell
easily explained
is
false
appear as
us that
by
(by shading
B may
in
this
all
Figure 39.
be either true
or false.
On
if
we make A
true (Figure
Logic Diagrams
41
),
we
51
B must be
any other true one.
therefore, implies
true.
Any
true proposition,
throughout
tions for
this
book
is
shown on
the
left.
On
the right
is
the "nega-
more than
in the head.
methods,
just as
is,
in
game of
much
of course,
Tautologous or equivalent statements are rendered visually obwe make separate diagrams for
two
assertions:
A v~ B
B
DA
now
to
be identical.
following premises:
A DB
(A implies B)
B^C
(Either
AsC
C DA
(C implies^)
or
shown
in
we
are
left
with
AB ~
C,
NOTATION
Conjunction ("And")
Implication ("If
^
=
Equivalence
then
and only
("if
")
if
or
then
")
and
")
or both")
or
Negation ("Not")
NEGATION
BINARY RELATION
Wo
A:>B
If
is
true, then
true
is
is
true
and B
is
B~A
B=A
o
If
is
false
true, then
true
is
is
true
and A
is
false
Avb
O
Either
or
is
true, or both
A^B
=B
If
and only
o
Either
or
is
if
is
true,
is
true
A|B
o
A and B cannot both be
true
AB
If
and only
if
is
true,
is
Either
true
Both
A and B
or
is
A ond B cannot
are true
Figure 42.
52
both be true
53
Logic Diagrams
We
on the basis of the logical structure asserted by the four premises, A and B must be true and C false.
Four-term problems can be solved in the same manner on Venn's
four-term figure of intersecting ellipses. Problems with larger num-
that,
on
Chapter
5, this
it
not a
is
all
all
eliminations
we
discover that
all
this
How
can
"A
B"
is
also be true.
this
be shown on the
Venn
circles?
We
can of course
54
and Diagrams
Logic Machines
other
Figure 44 shows
of circles ex-
sets. This
shaded black because we
lower
set is
know
is
because
Gray
sets of circles
we have no way
of
knowing
dis-
other areas.
and B, and
is
true or false.
is
If,
in
some
cases
sufficient to tell
we
for example,
learn that both terms are false, this clearly contradicts the relation
this is all
is
true, for
it
will
is
that
true,
is
relation
true or false.
The reader who desires to pursue this further will not find it difficult
work out rules for handling any type of compound statement,
to
including statements that are mixtures of class and propositional assertions. It also
is
and Marquand-type
more than
intricate that
it
if
becomes so
Peirce (1839-1914),
men-
carded;
it
what he called an
"entitative
graph" (later
dis-
Logic Diagrams
finally
55
worked out
a comprehensive system by
which he believed he
it
his
any existing
power of depicting
found
yet
it
necessary to
original scheme.
graph as
tial
by
this
his
For the rest of his life Peirce regarded the existenmost important contribution to logic; his chef-
his
not
it.
Some
how much
idea of
store Peirce
The
".
if
aid that the system of graphs thus affords to the process of logi-
cal analysis,
by virtue of
its
own
analytical purity,
is
surprisingly
tant questions that the analysis of ideas does not help to answer.
The
depends on
{Col-
this."
not space available in this volume to give a coherent account of Peirce's fantastic diagrammatic method even if I understood it fully, and I am far from assured I do. His several papers on
the topic (reprinted in Vol. 4 of his Collected Papers) are written
There
in
is
such an
elliptic,
is
led to
wonder
cloudy
if
Peirce
writing
that
56
and Diagrams
Logic Machines
Add
complain
to
to this
indeed.
create a
method
clear. Peirce
to
one would
devote several hours a day for a week or two to practicing with the
graphs, Peirce wrote, he would soon be able to solve problems with
a facility "about equal" to that of any algebraic
structure of
all
method
yet devised,
in detail the
ment
What
its
In this
way
the
thought."
They render
all
the "pueril-
ities
it
it
could then be
scientist experi-
new
same way
much
the
new
old ones.
Peirce's system
is
size
is
if
the
Logic Diagrams
57
and
inside regions.
The system
heavy unbroken
on
"lines of
no
two signs, one at each end of the line.
These properties of enclosure and connectivity are of course topological, and in one interesting passage (Collected Papers, Vol. 4,
p. 346) Peirce says that he expects his system of graphs to contribute toward an understanding of topological laws. On more than
one occasion he likens his graphs to chemical diagrams that show
how the molecules of a given substance are bonded together in
various complex topological structures. The graphs bear an even
stronger resemblance, in both appearance and purpose, to the
topological figures employed by Kurt Lewin in his Principles of
Topological Psychology* 1936, and by followers of Lewin who are
presently laboring in the field known as "group dynamics."
Peirce sought to make his diagrams as iconic as possible, and in
this he partially succeeded. For example: his use of two closed
curves, one inside the other (he called this a "scroll"), to indicate
what is now called material implication. The outer circle ("outidentity,"
lines
true,
if
necessitates the
truth of all that part of the graph inside the smaller circle
loop"). There
is
called a "ligature."
dotted, wavy,
But
its
in
("in-
an iconic aspect to
his
many
and saw-toothed
other respects
lines, the
(e.g., the
use of
aspect
is
the iconic
is
and necessary
color, fur,
truth. Peirce
Logic Machines
58
How Ramon
and Diagrams
References
1.
tion,
1906.
in
edi-
59
Logic Diagrams
2.
If the
The numerical
Chapter 4.
paper on "Boolean Algebra (the technique for manipulating 'and,' 'or,' 'not,' and conditions) and Applications to Insurance"
(first published in The Record of the American Institute of Actuaries, 1937logic will be given in
4.
Edmund
C. Berkeley,
in his
Edmund
of forming
5.
Venn diagrams
in
In my opinion Lewin's "topological psychology," as well as his "vector psychology" and "field theory," belong among the century's many premature and
his
admirers
obvious.
6.
Thomas A. Goudge,
in his
Thought of C.
S.
is
3:
Network Diagram
for
T enr
enn
and other diagrams of the shadedwe have seen in the previous chapter, be
circles
as
many
is clumsy
what Peirce called "iconicity" formal resemblance
to the logical structure for which they are intended to be visual aids.
This is understandable since these diagrammatic methods were
originally devised for class logic. To use them for truth-value problems we have to think of the problems in terms of class logic before
the diagram takes on an iconic aspect. Is it possible to diagram statements in the calculus of propositions in such a way that the diagrams exhibit more directly the formal structure of truth-value
and lacking
in
relations?
In 1951
I set
upon
this chapter. It
in
the network
obviously
efficiency with
have,
it
is
method
not intended as
method of
in helping novices
and
analysis. In addition,
it
60
it
does
such as myself
improved,
that will
it
it
can be
much
fancy of some
A Network Diagram
who
61
reader
eliminating
chief defects
its
to propositional problems,
is
from each other on the diagram so that they can be analyzed separately or altered as desired. This might be done by using sheets of
transparent paper (shading each premise on a different sheet) but
such a procedure is troublesome, and of course it cannot be applied
to classroom blackboards. The network method to be explained
here requires only paper and pencil, or chalk and blackboard, and
it
diagrams a
series of
makes
it
is
visually separate
from the
others. This
structure in a
way
that
is
difficult
type.
Essentially, the
or matrix
Like
all
the geometric
methods considered so
far, this
one also
is
works
in
we
shall
five
45
is
A
cross
drawn.
simple assertion that a term
mark on
is
true or false
is
indicated by a
shown
in
Figure 46.
Logic Machines
62
A
TF
TF
TF
E
T F
TF
and Diagrams
B
(DA
Figure 45.
These horizontal
give
them some
Figure 46.
lines will
sort of
in solving a problem, as
(2KB
be called "shuttles."
name, and
we
It
is
necessary to
this
shall see,
and forth along these lines in much the same manner that Man42nd Street shuttle train moves back and forth between
the Seventh Avenue and Lexington Avenue subway lines.
hattan's
It is
drawn
combinations.
To
illustrate,
let
us consider
first
a truth table
is
valid
(that
is,
if it is
it
states
un-
we im-
TT
TF
FT
FF
~AB
A-B
A~B
Figure 47.
Figure 48.
Figure 49.
A Network Diagram
63
at
in
Figure 49.
compound statement,
crosses for we have no way of knowing whether
is true or false. This will be made clear later
the diagraming of compound statements.
If,
we cannot add
the
is
part of a
when we consider
The biconditional, or statement of equivalence (symbolized by
=), requires two shuttles. In ordinary speech this is expressed by
and only if A is true, then B is true." Its truth table has
two valid lines, TT and FF; therefore we diagram it as in Figure 50.
The two shuttles show clearly that, if we are "riding" (to labor
the subway metaphor a moment) on A's T line, we have only one
shuttle that will carry us to B, and it will land us on fi's T line.
Similarly, the only available shuttle on A's F line carries us to B's F
line. The same relations hold if we move backward from B to A. In
other words, if either term is true the other must be true; if either
is false, the other must be false.
saying, "If
A=B
A^B
The
Figure 52.
Figure 51.
Figure 50.
wise
AvB
at
if one term
comparison of
a glance that
made up
is
gram
like-
one
#0
is
from the
and white
areas, so
we can
its
is
true the
this pattern
Each diagram
is
we transformed
Venn
dia-
in this
method
effect the
same
trans-
The
by
v,
it is
the negation of
~A
show
that
B.
are true"
(sometimes
re|
B.
64
Logic Machines
It is
the negation of
and Diagrams
ure 53).
The
"If
symmetrical. That
A
3
not
it is
the diagram
is,
A=>B
A|B
The
B^A
Figure 53.
Figure 54.
shuttles reveal
A DB
that
immediately
the negation of
is
~ B,
of ~ A
cellent
material implication.
implies
its
all
(~ A) may
we
if
we
indicated by
combinations of truth
B. Hence,
proposition
is
let
A and B
On
that
any
false proposition
false
(B or
B).
true proposition
vanishes as soon as
we
then" of material
common
speech.
It is
the
are
common
bols in logic.
relation
easily
discussed occur so often, however, that one's use of the graph will
A Network Diagram
65
be greatly facilitated
if
memory
so that
it
will
The order
to be graphed.
are
drawn
is
as patterns,
in
which the
it
But
if
shuttles for
is
in truth-
table lines.
When one or both terms of a binary relation are negative, as, for
~ A v B, how do we go about drawing the required shut-
example,
tles?
The procedure
is
simple.
We
other words,
all shuttles
on A's
v B,
lines.
In
line; all
shuttles on A's F line are shifted to the T line. The terminal points
on B remain unchanged. After we have done this we shall discover
that the resulting diagram is identical with the diagram for A 3 B.
The same pattern also results if we diagram ~B D ~A (in this
case we must of course exchange the terminal points on the truthvalue lines of both terms). Whenever the diagrams for two asser-
be "tautologies," that
We
~A
identity of
is,
called
laws."
One
two
tells
make
other law
tells
is
De Morgan called
known as "De Morgan's
clear.
this
interesting tautologies
in a disjunctive relation.
A \B
SymThe
= ~Av~B.
compound
statements involving
66
Logic Machines
parentheses,
and Diagrams
let
true then
1.
If
2.
Either
is
true or
3.
Either
is
true or
4.
is
is
is
C
C
OB)
(A
true.
is
is
true, or both.
(A
will
appear as shown
The next
C)
(B)
true.
(B #^ C)
this is
done, our
in Figure 55.
is
to
will
premise
2.
we
offer us a choice of
point on C's
line.
Our passage on
its
terminal
mark on
premise
C's
C's
4, forces us to
premise
2, in
false.
conclude that
is
by
The
combination with
We
next inspect
there also.
We
now determined
that
premise
knowledge that B
is
mark
is
false
is
We mark
line,
and mark
its
terminal
no other
single shuttles
we con-
clude that the premises are consistent, and consistent only with
the truth of
and B, and falsity of C. If the premises had conit would have forced us, in our exploration
tained a contradiction,
ABC
A Network Diagram
67
(l)A=>B
f
'\
)<
) \
(4)B
Figure 56.
Figure 56 shows
v
/\
) (
\
Figure 55.
(2)B*C
(3)AvC
and
falsity of at least
one
terms.
how
been solved.
In August
2.
either
tie.
3.
In August
I either
tie
and some-
times both.
To
we
A I
B
I
I
stated sym-
(l)A*C
bolically as:
1.
2.
^C
[2)B C
B C
A wB
|
3.
A
will
(3)AvB
show
Figure 57.
68
and Diagrams
Logic Machines
We
must now
test this
suppose we
We may
terms.
start
structure to see
its
the
this
is
line.
The
tle in
first
We
single shutfalse,
is
but
about B.
The next
A's
step
is
all
We
line.
we
If
a cross
we
on
explore
and
falsity
three terms.
final step
see
to
if
contradictions
arise.
No
such
A and
lem
but
is
August
that in
may
or
may
truth values of
all
Hence
are
the falsity of
enare
not sport a
test
contradictions
test
bow
tie.
of one term
is
of one term will give values for only a portion of the remaining
made
to see
if
the unde-
may
it may be that certain premises are contraany case, the graph gives a clear visual picture of the
structure that is open for inspection and experimentation in a way
terms undecidable. Or
dictory. In
that
is
if
if
we have
one
is
we may wish
A and F
It
answer such
to
to
be true when
if
they lead to
how many
are given, we
how many
binary relations
it
Compound
We
by con-
A Network Diagram
the
sidering
assertion
statements
inside
manner, as
if
69
(A
B)
^(CvD). The
parentheses are
two disjunctive
diagramed in the usual
first
statements.)
lines.
when Venn
circles
compound
symbolize
this is
yet
is
done by supplying
we immediately
tle,
if
place crosses on
its
is
from
a single shut-
now be connected on
the graph by a relation of implication. To do this, we adopt the
following procedure. At the right of the graph we draw two pairs
The two
one of the
state-
a binary relation.
plication to
tles
show
On
this
in this case
each term
is
itself
lines of im-
one relation implies the other. These shutis no uncertainty about their validity.
now appear like Figure 58.
that
The
If
graph
entire
will
a single term
ample,
we show
(B
is
involved in a
compound
is
we can
of course
70
Logic Machines
C
and Diagrams
CvD
n
Ave
Figure 58.
ABC
B"C
A~BD
Figure 59.
Figure 60.
compound
The
relation.
we
if
it
a valid
Cs
is
diagramed
Z?'s
is
treated exactly as
part of
If
two
= B = C,
can be
on
truth-value lines.
is
true
and
all false,
is
tell
us
terms cannot be
permitted except
cases, of course,
we
are
chain
by the
six
When
we would have
to replace the
two
shuttles
missing ones.
there are
of shuttles involved
we may
find
it
A Network Diagram
71
C
/
\
/
\ )
A=B = C
A=>(B"C)
\ )
^*
Figure 61.
This
ment
Figure 62.
a diagrammatic way of expanding the original statewhat logicians call its "normal disjunctive" form. Each
in the above diagram represents a valid line of the eightis
to
shuttle
The
(A*B*C)v(~AmBmC)v(~A'B*~C)v(~A*~B'C)
y(~Am~B*~C).
When
it
often
the
is
in a statement
a simpler form
before diagraming
it.
as in
it is
much
simpler
Figure 63.
is
still
fl's
line
(Figure 64).
A,
a tautology.
Thus we
see
how
(A*~
B)
(~ A
~B)
= B
and Diagrams
Logic Machines
72
most eco-
its
and
is
shown
in Figure 65.
A
>:
Figure 63.
Figure 64.
Figure 65.
many
shown
alone
shown
cannot go into
expressed by
all
compound
is
except for
shuttles)
in the
is
We
and dotted
(half crosses
as tentative
it
to
difficult
make
rules should
who
work out
is
suf-
his
own
If
all
to
one of the
known
is
to be true.
information
there are
there
is
we need
two
to
know
that
shuttles leading
a shuttle for
FT
ADB
from A's
and another
is
is
all
is
In
the
F
for
line.
In other words,
FF; so regardless of
A Network Diagram
whether
is
73
sufficient to tell
is
situations with
us that
we know
When
follows.
and there are two shuttles leading from this truth-value line, we can then affirm the truth
of the entire binary function. If there is only one shuttle, we lack
sufficient information to do this.
2. If the terms are known to have a combination of truth values
not indicated by a shuttle, the entire relation is known to be false.
3. Whenever a parenthetical statement is known to be true,
either because of knowledge of its terms or because it is found to
be true in the process of exploring the entire structure, its shuttles
are changed to solid lines or its half crosses to crosses. The truth of
the entire statement is then indicated by a cross mark on the T line
in the pair of truth-value lines (to the right or below) that correspond to the statement.
4. Whenever a parenthetical statement is known to be false,
in either of the two ways mentioned above, we add the missing
the truth value of one term only,
Let us
then
it
(A D B) D (B ^ A)
To determine this we
We must now test
we wish
to
know whether
a valid theorem.
If
is,
it
and B.
in Figure 66.
this structure for the four possible combinaand B a diagrammatic procedure correspondthe matrix method of testing a theorem. If none of these
ing to
all
first
is
FT we
Our
testing
FF
are
all
we know
procedure
the structure
will
consistent, but
show
that
when we
how this
occurs.
The
first
FT
is
to
make
a cross on A's
line
this
lower statement,
we know
is
true.
Conse-
Logic Machines
74
quently,
we change
and Diagrams
its
We
changing
like
tells
line of the
line,
on
we
B D A,
is
indicate this
shuttles to
its
shuttle to the
is
line
Figure 67.
A
B=A
B=>A
-*
>C
) (
--
A=>B
A=>B
-x-
Figure 66.
We
/4'sF
that
Figure 67.
is
us to
fl's
line.
But
For
we encounter
this contradicts
if
we
our assumption
will carry us to
A's
line,
explore
a single shut-
line
is
false.
we supply
us that
is
true
assumptions about
least
This in turn
counter a contradiction,
is
not a law.
ample, in
If
theorem.
Syllogisms can be tested in this
is
way
to determine
awkward, especially
if
if
Network Diagram
involved.
To
75
we wished
All A
No B
is
is
Therefore, no
is
D B)
C] 3
C.
When we
test
to
be valid in
all cases.
for the
it
it
proves
"Some A
is
is
not suf-
when
(in
that
method
for the
It is
that
also possible,
circles
is
I
beyond
my
think, to
is
a use
Venn
capacities to explore.
combine
this
mixture of class-inclusion
method
is
table matrices. 2
network
based on truth-
For example, a three-value logic could be diagramed by increasing the number of truth-value lines under each
term from two to three. Since shuttles in such a logic represent
two types of relations, true and "indeterminate" (or whatever one
wishes to call the third value), it will be necessary to distinguish two
types of shuttle lines. This could be done by making "true" shuttles
solid and using a saw-toothed line for the new value.
Exactly what pattern of shuttles to use for a given function will
depend on the type of three-value logic we are considering. Here
76
Logic Machines
and Diagrams
words meant
just
AB
Figure 68.
This shuttle
tells
A-B
Figure 69.
logic, a pleasing
meaning
is
permitted
A Network Diagram
77
#"
in a three-value
by that relation. If instead of saying "A implies
system we said "A galumphs fl," considerable clarity might result.
Figure 71 shows how the shuttles for "A implies B" appear when
we diagram
B
T ? F
A=B
AB
Figure 71.
Figure 70.
There
is
as
little
"mean-
The pattern
It is interesting to
the
itself is all
note that
meaning the
relation
need have.
if
is
a sense, therefore, in
Logic Machines
78
make
false.
and Diagrams
is
is
to use only
ABC
oo
o o
o
U.'
$
o
=>
ters
may
missible true
g
#
i<
A
B
=C
false
combinations.
as the "board,"
Figure 72.
on the board
To
and
as
shown
in Figure 72.
theses
we
side,
we wished
if
we would
to
its
counters.
Ob-
viously,
ways
ways
It
less iconic
it
more
it
method of
truth tables.
it
really a primitive
as
as a "diagram."
maLord Stanhope's
it is
is
modern
electrical
Network Diagram
79
ter.
References
1.
two "one-way"
one direction only. Thus
A D B
in
is
some
use, but
it
cases, this
is
an
efficient type
in the
of diagram to
chapter because
duces
the
electrical
(circuits
system's
iconicity.
In
the
it
it
re-
construction
of
made one-way by
4:
The Stanhope
Demonstrator
A.
Llthough
Ramon
Lull
made
use of rotating
machines"
in the
was a
and scientist, Charles
Stanhope, third Earl Stanhope (1753-1816). His curious device,
which he called a "demonstrator," is interesting in more ways than
one. Not only could it be used for solving traditional syllogisms by
a method closely linked to the Venn circles; it also took care of
numerical syllogisms (anticipating De Morgan's analysis of such
forms) as well as elementary problems of probability. In addition,
it was based on a system of logical notation which clearly foreshadowed Hamilton's technique of reducing syllogisms to statements of identity by making use of negative terms and quantified
world's
first
logic
machine, in
predicates.
printing
finished
upon
New
Principles.
beyond
lest
81
came
for his
many
scientific inventions.
His
first
Pitt.
earl
the ministry to
At
Royal Society and for the rest of his life he devoted a large segment of his time and income to scientific pursuits. His best known
inventions were the Stanhope microscopic lens, the Stanhope hand
printing press, a
monochord
necessary
first
can take
this to
mean
It rests
men
is
identical to a
Logic Machines
82
we mean
members of
all
mortal things.
we
If
mortal,"
that this
of the
the class of
all
and Diagrams
say "Socrates
identical to
is
is
one
by changing them to affirmative statements and employing negative terms. For example, "No swans are
green" tells us that the class of all swans is identical with a portion
of the class of
not-green things.
all
The following
table
A
No A
A.
All
E.
is
is
Some A
Some A
I.
O.
A
All A
Some A
Some A
B
B
is
All
is
not
= Some
= Some
= Some
= Some
B
not-5
B
not-fi
Stanhope used the term "holos" (the Greek word for "whole")
term of a syllogism, choosing this word
to underscore the fact that the middle term must be universally
to stand for the middle
distributed
in
at
least
can successfully
it
The order
in the second.
premise
is
"The reader
los
may
will observe,"
first
in the first
if it is
of premises
is
premise, los
The
immaterial.
first
considered).
Stanhope
ho
writes, "that
as well as
be identic with holos, but that neither ho nor los can ever
exceed holos."
In Stanhope's terminology, the demonstrator
for determining
what relation of ho
is
simply a device
can be deduced by relating each term to the holos. The contrivance consists of a block
of
mahogany 4
to los
tall,
and
inch thick,
with a brass plate mounted on the face (Figure 73). In the center
is a square window
n
.,
or depression about an inch wide
of the plate
Holorw
Red
slide
Grayslide^
calls this
gism
panel of gray
premise that
is
the "holon." It
Figure 73.
window
wood
being examined.
syllo-
is
It
first
demonstrator through a
the holon.
The
middle term)
slot
83
on the
is
in the
left until
it
covers part or
is
all
of
not the
ment. As
we
and inserted through another slot above the holon for workwhat Stanhope calls the "logic of probability."
A scale from to 10 appears on the brass frame above and to
the left of the holon. The same scale also is found on the lower
edge of the red slide. Figure 74 shows the face of the demonstrator
its slot
ing problems in
DEMONSTRATOR,
INVENTED BY
The
of
The
The
area of
on
all cases.
same
To
the gray,
add the
dark red,)
The proportion, between the area of the dark red and the area of the holon,
is the probability which results from the gray and the ""\
Figure 74. Reproduction of the face of Lord Stanhope's demonstrator. (From Mind, April, 1879.)
as
and Diagrams
Logic Machines
84
appears
it
when both
slides are
pushed
the red above the gray and both covering the entire holon.
The working
we have
The
We
is
No
All
first
ments of
of the device
is
is
step
is
All
All
is
some not-A
some
is
it
it
The next
tially
all
step
of
is
to
M. Whenever
is
made
M.
all
of
slide (all
is
B)
until
it
only par-
is
visible
clude that an identity has been established between the ho and the
In this case
follows:
1.
When
we
push the panel for that term over the entire holon.
2. When a premise relates a term to "some" of the middle term,
we push the panel for that term over part of the holon.
3.
(e.g.,
When
"No A
is
M," or
is
not-M"), we
withdraw the panel for that term so that no part of the holon is
covered by it.
4. After the two slides are properly adjusted to represent the
two premises, we inspect the holon to see if the slides must of
necessity overlap. If so,
we may
85
no
identity
is
If
the
established and
No
No
M
M
is
is
will
M
M
is:
some not-A
some not-Z?
is
is
make
The two slides, standing for "some not-A" and "some not-fl" are
each pushed over the entire holon. They must overlap; so we conclude that
"Some not-A
is
some
not-Z?." This
we have committed
is
not, of course, a
to B. Nevertheless,
it is
a valid conclusion
(assuming that the middle term has members), as can be seen easily
by making a Venn diagram of the syllogism. We may even go a
step further, Stanhope points out, and conclude that as many
not-/Ts are not-B's as there are M's.
M. The
is
Some B
M
not
is
must be converted
All not-M
Some B
The gray
is
to:
some not-^4
some not-M
is
(some not-A)
slide
commonly
expressed,
"Some B
is
"Some B
not
is
some not-A" or
A"
Logic Machines
86
as
an
means
efficient
which
in
some ways
rectangular form,
mind
its
and Diagrams
it
it is
is
it
markedly
anticipates.
more
is
On
efficient
happening
in syllogistic
inferior to the
Venn
Venn
than the
cir-
owing
circles
to
when
used for syllogisms with terms that have numerically definite quantifiers
half."
Some
Some
M
M
is
is
some A
some B
The gray panel is "some A"; the red is "some B." Each is pushed
way over the holon (some M). The two slides are not
only part
M." In
this case,
We
is,
slide
in
made
is
and
To
more than
it is
De Morgan's
in the
to cover "most," or
each
M"
syllogisms discussed
second chapter.
illustrate
how
the demonstrator
may be
used for
De Mor-
us consider what
let
We
let
pushed
in
from the
pushed
is
in
from the
covers s/i o
red slide (4 pictures by Picasso) is
scale. In other
The
The
words, until
number 4 on
its
it
lower scale
coincides with the right edge of the holon; in other words, until
it
scale,
This
is
what Stanhope
calls,
on the face of
is
%o
of the holon.
los.
87
may
be more (four
is
middle term
is
why two
less
is
then we
simply regard as our holon only that portion of the aperture which
extends from the
is
unaffected by
member
left side to
this,
to bring the
It is
clearly
how
more or
less
than
half, or "all"
and "some."
"Behold, then," the Earl writes in one of his notes, "the luminous
perspicuity
this
new system
of
logic!"
how he used
but from
the rule given at the bottom of the brass face (see Figure 74),
easy to understand
how
it is
the instrument
rule states,
is,
the
scale; that
is,
until
it
To
Logic Machines
88
represent the chance of a head
pushed
The dark
it
on
and Diagrams
is
two slides, is obviously one-fourth of the holon. The chance that the penny will fall
heads on both tosses is therefore one-fourth. The same procedure
would be used, Harley points out, to give exactly the same answer,
if we wanted to know the probability that the coin would fall tails
twice, or heads and then tails, or tails and then heads.
As in dealing with numerical syllogisms, probability problems
holon.
by giving the device whatever type of scale it required. As we shall see in Chapter 9, Stanhope's demonstrator,
suitably handled
used
at a
in the
manner
is
a crude
first
attempt
If for
it.
Draw on
a sheet
23456789
10
4
5
6
7
8
9
10
HOLON
..;
GRAY SLIDE
Figures 75 (top), 76,
MJJmm
RED SLIDE
and
77.
89
slide,
left as
shown
in
Figure 75.
shown
this
in Figure 77.
when
it
on cardboard
to
all
three squares
durable.
They obviously
are ca-
Stanhope device.
Refe rences
Stanhope had three daughters by his first wife. The youngest created a scandal
by eloping with the family druggist. The E arl was never reconciled to the pair,
but Prime Minister Pitt made the druggist controller-general of the customs.
The eldest daughter, Lady Hester Lucy Stanhope (1776-1839), kept house for
Pitt until his death in 1806. In 1814, with a generous pension from the government, she settled in an abandoned convent in Lebanon where until her death
she ruled over her retinue of some thirty servants like an Oriental potentate.
For a time she was a political power in the Syrian area. In later years, her
interest in the occult intensified, and she claimed to be an inspired prophetess.
For fascinating accounts of her beauty, sarcastic wit, arrogance, and legendary
exploits, see Alphonse de Lamartine's Voyages en Orient, Alexander Kinglake's
Eothen, and the six volumes of Lady Hester's memoirs prepared after her death
by her physician, Dr. Charles Lewis Meryon.
The fourth Earl Stanhope, Philip Henry (one of Charles's three sons by
his second wife), is best known for his espousal of the famous German imposter, Kaspar Hauser. Philip paid for Hauser's education and even wrote a
book in German about him.
Lord Stanhope either made or had made for him several models of both his
logic and arithmetical machines. Photographs of one model of each machine
are reproduced opposite page 127 of Early Science in Oxford, by Robert
W. T. Gunther, Vol. 1, Part 1, 1922. These two models are owned by the
Oxford Museum of the History of Science, Old Ashmolean Building, Oxford.
A second model of the logic machine is owned by the present Earl Stanhope.
A quotation from one of Stanhope's letters gives an amusing insight into the
value he placed on recognizing that class propositions could be expressed as
identities:
"When
talk of identity,
dit
rien,
do not
but
say, as
say thus:
my
find
new
star, so distinguished,
constellation,
Logic Machines
90
which 'moves
that
fast,'
Have
is,
which belongs
made an advance
and Diagrams
knowledge by
move
fast,
my
having so perceived, though in point of fact, it is the same comet, the identical comet, originally described by me incompletely, before I perceived, or could predicate,
are identic.
not
in
such identity? Voild tout. Would it not sound to your ears very droll if a
person were to say that that star moving fast means that it is identic with
some star which does not move fast? Now if that would be evidently wrong,
and if I have by my method only two opposite classes, viz., stars moving fast
and
stars not
moving
is
fast,
if
my
two
mean
that
it
is
it
5:
amon
first
to use a
mechanical
first
to use
The next great step in the history of logic machines took place in
1869 when William Stanley Jevons, British economist and logician,
produced the first working model of his famous logic machine. It
was the first such machine with sufficient power to solve a complicated problem faster than the problem could be solved without the
machine's
aid.
He
spend
in
five
master of
now
arts degree,
the double
at
He soon found
Owens
College,
himself carrying
and mental and moral philosophy" and "professor of political economy." In 1876 he became
professor of political economy at his alma mater, the University
College, London, where he remained until he resigned the chair in
1880 because of failing health. Two years later, at the age of only
forty-seven, he was drowned while swimming alone off the beach
at
title
of "professor of logic
As an economist, Jevons
is
economic
issues.
1871,
is
the
91
and Diagrams
Logic Machines
92
most important of
his
nomic theory
are less well
his speculations
on the
rela-
knowledge and
statistical sophistication,
At
a time
when most
damned
many
of
greatest
its
defects.
advance
He
He
virtually
effect
on the specula-
modern
logicians agree
no
On
make
letter,
saw
clearly the
weakness
notation.
"I
am
quite
.
mathematical expressions."
in Boole's
his
con-
have no
He
also
we
shall see,
93
is
as well as a primitive
technique. Since
it
necessary to review
Putting
it
method
in
some
it
will
be
detail.
amount
of justification,
one might say that the method is a linking of Lull's Ars magna
with one of Sherlock Holmes's favorite canons of deduction;
namely, that if you eliminate all possible explanations of a crime
but one, that one explanation is certain to be correct. As the technique of reductio ad absurdum, this procedure is an ancient one,
but the realization that it could be applied to the Boolean logic
awoke
in the morning," he recorded in his journal in 1866, "the sun was
shining brightly into my room. There was a consciousness on my
mind that I was the discoverer of the true logic of the future. For
a few minutes I felt a delight such as one can seldom hope to
came
feel."
to
The
way
easiest
amples of
how
premises, All
it
and
B,
No B
and
all
is
first
to give a
the
two
few ex-
syllogistic
to
make
possible combinations of
ABC
is
C.
Our
first
step
is
De Morgan) we
shall
adopt
The
table appears
as follows:
ABC
ABc
AbC
Abe
aBC
aBc
abC
abc
These eight classes correspond of course to the eight compartments of Venn's three-circle diagram (including the area representing abc, which
lies
such an exhaustive
list
difficult to
first
called
it
this
Logic Machines
94
5
bet."
and Diagrams
compartments on the
The
first
premise, All
is
B,
tells
us that the
classes Abe and AbC are empty; therefore we draw a line through
them. Similarly, the second premise, No B is C, will eliminate
classes ABC and aBC. The logical alphabet will now look like this:
ABC
ABc
-AbG-Abe-
-aBG^
aBc
abC
abc
The
remaining
to C.
We
note at
On
is
we may
tains a c. This
cause
it
is
what
this is
is
none
"Some
conclusion
"No A
We may
C."
Lewis Carroll,
also
"Some
not-/4
all
possible inferences
"somes,"
Jevons's Logic
Machine
95
more
efficiently
we took
liberties
explaining
in
how
the
its
He combined
like
part or
all
all
A and B
is
either
or
D)
making a
faint be-
its
significance.
keeping
meaning
are both
A and
stitute for
is
is
Z?"
is
the equation
= AB,
him
to sub-
Because of
96
Logic Machines
how
and Diagrams
We
are
and only
Either
We
As
if
wish to
is
true.
before, the
The
step
first
first
is
infer
to write
premise
tells
about
down
us that
B and
C.
we must
eliminate
all
Ab
and aB. The second premise elimiBC and be. 9 The third premise
combinations containing A. The alphabet will now
combinations containing
all
is false.
three terms.
nates
true, then
is
combinations containing
excludes
all
look like
this:
B must
be false and
The method
is
true.
essentially the
same
as the
method described
in
The
logical alphabet
is
simply another
nated just as
all
all
If
the
become shaded
if
contra-
truth-value
premises
are
(and, as
logic
we
97
machine),
is
England's
first
first
electric
to
make
11
use of matrix analysis.
To
devised a
number
a rubber stamp made of the logical alphabet for three terms, another
stamp for four. This eliminates the annoyance of having to jot down
all the combinations each time you tackle a new problem. For
a problem of five terms, you have only to make two impressions
with the A BCD stamp, heading one of them E and the other e.
(Venn also suggested having rubber stamps made for his circular
figures. See his Symbolic Logic, revised 1894 edition, p. 135.)
As early as 1863 Jevons was usirfg a "logical slate." This was a
slate on which a logical alphabet was permanently engraved so that
problems could be solved by chalking out the inconsistent lines.
Still another device, suggested to Jevons by a correspondent, is to
pencil the alphabet along the extreme edge of a sheet of paper, then
cut the sheet between each pair of adjacent combinations. When
a combination is to be eliminated, it is simply folded back out of
sight.
was
book The
become a
small rectangular wooden
logic machine.
boards,
all
the
"On
Mechanical Performance of Logical Inference." The paper was printed in full, together with plates showin a
paper
titled
the
and Diagrams
Logic Machines
98
p.
ciety's
Vol.
1,
appearance
In
Jevons's
ma-
about
piano
feet
high
(Figure 78). On
piano are openings through which
the face of the
representing
letters
16 possible combinations of
the
four
(Each combination forms a vertical row of four terms.) The keyboard consists of 21 keys arranged
as
negative
Figure
78.
Jevons's
logic
frontispiece of Principles of
by William
machine.
Science,
(From
1874,
Jevons.)
S.
ation keys."
The "copula"
forms,
are
represented
in the center
is
an equation. The
on the extreme right is pressed after a complete equation has been fed to the machine. When the "finis" key on the exsign of equality connecting left
and
right sides of
"full stop"
<
CO
_|0_
_i
1-
0.
o
o
u.
1-
_io
U_CO
Figure 79.
treme
tion.
left is
pressed,
it
restores the
.|..
machine
to
its
original condi-
They
left
Jevons's Logic
To
Machine
99
it
is
in the
=
is
AB, which
6."
the right),
B (on
let
all
combinations of terms
its
face
is
prem-
all
by no means a com-
dled.
way how
it
the
Jevons did not think that his machine had any practical use,
owing
complex
everyday
life.
But he did
logical questions
feel that
it
was valuable
seldom
arise in
as a classroom
it
113
is
striking because
it
{Prin-
reveals
as well as
many
of
its
how
work
defects:
gations of the late Dr. Boole must be recognized at their true value, and the
in
it
re-
quire the manipulation of mathematical symbols in a very intricate and perplexing manner, but the results when obtained were devoid of demon-
Logic Machines
100
strative force,
and Diagrams
mathematical reasoning, so that Boole inverted the true order of proof when
he proposed to infer logical truths by algebraic processes. It is wonderful
evidence of his mental power that by methods fundamentally false he should
have succeeded in reaching true conclusions and widening the sphere of
reason.
both of the truth of Boole's results and of the mistaken nature of his
mode
of deducing them. Conclusions which he could obtain only by pages of intricate calculation, are exhibited
of manipulation.
And
by the machine
after
are demonstratively true, because every step of the process involves nothing
Laws of Thought.
first
of
would have defects that could be remedied by later machines operating on essentially the same principles. By insisting that
statements be fed to the machine in a clumsy equational form, it is
made unnecessarily complicated. There is no efficient procedure for
feeding "some" propositions to the machine. The mechanism does
not permit of easy extension to a larger number of terms. (Jevons
once contemplated building a machine for ten terms but abandoned
the project when it became clear that the device would occupy the
entire wall space of one side of his study.)
A more serious objection to the machine, and one that may not
permit of remedy within the framework of Jevons's combinatorial
logic, was voiced by the British philosopher Francis H. Bradley in
a section on the machine in his The Principles of Logic, 1883. As a
conclusion, the machine merely exhibits all the consistent lines of
its
kind,
(i.e., all
combined premises).
It
analyzing these lines so that one can see the desired conclusion.
The process
and
In
many
is
false,
false values for individual terms but will give a series of con-
sistent
101
the
power
to
reduce
this
answer
It
may
reveal
one can
this,
involved, but in
is
He
would be
it
p.
139) comparing
it
with the "sieve of Eratosthenes" by which one can search for prime
8 that electric
"minimizing machines"
References
1.
from 1721
mates
in his
of
conform
boom and
to 1878
in
Owing
it
to vagueness in deciding
it
is
as easy to squeeze
sion in 1944.
More
in
turn conditioned
The Letters and Journal of W. Stanley Jevons, 1886, p. 350. The book is
who, it is interesting to note, was the daughter of
the founder and first owner of the Manchester Guardian.
edited by Jevons's wife
102
3.
Logic Machines
and Diagrams
text has
is
still
in
print.
4.
5.
duality
is
7.
is
Stoic-Megaric school
number
it
any modern logician. Sextus Empiricus quotes a third century B.C. remark by the head of the Alexandrian Library, that "even the crows on the rooftops are cawing over which conditional is true." See chap. 15 of Ancient Formal Logic, by I. M. Bochehski,
as disinterested in Aristotle's class logic as
Amsterdam, 1951.
8.
efficient
notation for
and
true
J.
first
as early as
using the values "necessary," "impossible," and "possible but not necessary."
9.
Jevons contrasts
show
p.
350)
the
symbol
of his
pressing each relation both in equational form and in terms of the valid lines
of
10.
its
truth table.
This holds only for the propositional calculus. In class logic a contradiction
when all the combinations specified by a "some"
is revealed in two ways: (
)
1
known
to
have members
is
when
all
a class
combi-
Machine
Jevons's Logic
11.
The answer
103
who was the first to use a matrix method debroadly or narrowly the term "matrix method" is
defined. As Venn points out (Symbolic Logic, revised 1894 edition, p. 415),
C. A. Semler, in an 1811 German work, suggested the procedure of listing
to the question of
pends of course on
all
how
possible combinations of terms, then striking out those that are contra-
meaning of
a conditional
statement as
it
Megara.
The
however, that Jevons was probably the first to make exwhat is substantially a truth-table method for solving problems, even though its first explicit application to truth-value statements came
fact remains,
tensive use of
later.
used the
1
12.
or
now
W. Mays
machine was given by his son to the Science Museum, South Kensington, London, but in 1934 it was transferred to the Oxford Museum of the History of Science, Old Ashmolean Building, Oxford,
In 1914 Jevons's original
where
it
is
now on
display.
6:
Others
"
much
of Jevons'
p.
work
165),
and economical
I cannot but hold that in the domain of logic his inconand contradictions are remarkable." There was a strong
element of rivalry between the two men; hence it is not surprising to
find Venn dismissing Jevons's logic machine as essentially trivial.
science,
sistencies
op.cit., p.
two grounds.
In the
first
place,
complicated examples
in
it
is
upon
this
on
it
rather
is
for
marked
may
second place,
known or
it
me
that
any contrivances
at
present
be discovered really deserve the name of logical machines. It is but a very small part of the entire process, which goes to form
a piece of reasoning, which they are capable of performing. For, if we begin
likely to
from the beginning, that process would involve four tolerably distinct steps.
There is, first, the statement of our data in accurate logical language. This
step deserves to be reckoned, since the variations of popular language are so
may need
104
105
throw these statements into a form fit for the engine to work
to reduce each proposition to its elementary denials. It
would task the energies of any machine to deal at once with the premises
employed even in such simple examples as we have offered, if they were
presented to it in their original form. Thirdly, there is the comhination or
we have
to
with
this case
in
further treatment of our premises after such reduction. Finally, the results
have to be interpreted or read off. This last generally gives rise to much
in most cases there are many ways
opening for skill and sagacity; for
of reading off the answer. It then becomes a question of tact and judgment
which of these is the simplest and best. ... I hardly see how any machine
can hope to assist us except in the third of these steps; so that it seems very
doubtful whether any thing of this sort really deserves the name of a logical
.
engine.
Venn goes on
to say that
"So
little
trouble
is
required to sketch
as
felt
much
really
it is
work
for us.
such attempts,
interest in
it
seemed worth while seeing how the thing could be effected here."
then suggests several mechanical procedures of his own. First,
the use of a rubber stamp to form the intersecting circles. Second,
draw the desired diagrams on thin board, then cut out the compartments so they fit together like a
He
now can
compartments, you
move
nn n n 1
re-
of what he
calls
"logical-diagram machine."
This machine
is
merely a three-
The
device.
inders
uu Liuy
Figure 80.
allowing the
box
is
To
wooden
eliminate a
cell,
you
The
all
The
box
looking
down from
ment
is
above.
You
will
Logic Machines
106
left-hand peg.
succeed
in
Venn was
doing
all
and Diagrams
"would
machine."
As Venn himself recognized, his logic box was simply a mechaniway of handling a Venn diagram. It is surprising that Venn did
not think of placing counters on his diagrams, after the manner of
cal
Lewis Carroll's method, since this is much simpler than the contrivances he proposed and has the additional advantage that
counters of two different colors may be used
one to show that a
cell has no members, the other to indicate cells involved in particular ("some") assertions. Although it is true that Venn's devices can
perform any of the operations possible on Jevons's logic machine,
Venn would not permit himself to appreciate the fact that Jevons's
machine was a pioneer effort to work out a mechanical laborsaving
method in which desired operations could be effected merely by
pressing keys.
The first real advance over Jevons's machine was made by Allan
Marquand ( 1853-1924). The son of Henry G. Marquand, a prominent American philanthropist and art collector, Allan Marquand
began
kins University.
in
He became
and ethics
at
Johns Hop-
and archeology at Princeton. His books include a Textbook of the History of Sculpture, 1896; Greek Architecture, 1909;
and several books on della Robbia, the famous Florentine sculptor,
and his family.
The first logical device built by Marquand is of no special interest. It was merely a more elaborate version of Venn's logicaldiagram machine, making use of 256 separate wooden parts to
accommodate eight terms. In 1881 he turned his attention toward
a machine of the Jevons type, describing the final outcome of his
labors in a brief article titled "A New Logical Machine," in the
Proceedings of the American Academy of Arts and Sciences, Vol.
21, 1885, p. 303. The following account of his machine is taken
from this paper.
A crude first model was built by Marquand some time in 1881;
then during the winter of 1881-1882 an improved and final model
was constructed for Marquand by his friend Prof. Charles G. Rockwood, Jr., of Princeton's department of mathematics. "The machine
fessor of art
107
was made from the wood of a red-cedar post," Marquand tells us,
"which once formed part of the enclosure of Princeton's oldest
homestead."
Photographs of front and back views of this machine, as it appears today, are reproduced in Figures 81 and 82. In external
appearance
it
more than
trifle
inches wide, and six inches from front to back. The inner mechanism (Figure 82) consists of an ingenious arrangement of rods and
view.
by catgut
levers connected
and
spiral springs.
it is
The
sixteen
and
on the face of the machine. These pointers
accordance with Marquand's logic diagram which he
are labeled in
had described
The method
upper
left
the right
in a
of labeling
is
easily understood.
AbCD,
The
ABCD.
is
pointer in the
Next to
reached
it
on
at the
lower right corner. Each pointer can be raised to a horizontal position, pointing leftward, to indicate that the
combination
is
"true"
108
Logic Machines
dropped
and Diagrams
to a vertical position,
(one ex-
The keyboard
shown
in
Figure 83.
A
and
(capital letters)
keys
the
the
for
by eight keys,
being
negatives
The remain-
Figure 83.
ation key")
key"), are called "operation keys." Their use will become clear in
a moment.
Before a problem
is
is
pointers must
all
first
be
first
the
Okey.
Each premise
is
example, suppose
This
for
is
The
a
B
Ab
_J
in
Figure 84.
shown
in exactly the
_Jj
what combina-
<
*
<
<_o
the
machine operates
as does Jevons's
i
Fi 9
<o
with the
ure 84
what
109
duced.
makes
it
third advantage
is
machine
is
enormously
re-
mechanism
more than four
(American
Journal of Psychology, Vol. 1, November, 1887, p. 165), summarizes these advantages in the following interesting and characteristic
manner:
The nature
that of Jevons.
is
a vastly
it.
would have been no inconvenience in embracing six. Incumbrous equations of Jevons, Mr. Marquand uses
Professor Mitchell's method throughout. 1 There are virtually no keys except
the eight for the letters and their negatives, for two keys used in the process
used, though there
stead
of using the
of erasing,
etc.,
Any number
may
of keys
be put
down
to-
dif-
ference that the two dimensions of the plane are taken advantage of to ar-
To work
way
a simple syllogism,
is
in-
necessary, two keys being pressed each time. A cord has also to be pulled
each time so as to actualize the statement which the pressure of the keys
only formulates. This is good logic: philosophers are too apt to forget this
cord to be pulled,
this
illogical.
in existence,
To work
the
clumsy manner
in
and thus
to re-
syllogism with
owing
Mr.
to the relatively
syllo-
ample
All
is
B, All
is
The premises
Logic Machines
110
Be
The
and Diagrams
=0
combinations:
ABC
aBC
abC
abc
we examine
If
A and C we
This
tells
us that "All
is
combination of
Ac
The
first
problem
is
stated
by Marquand
as follows:
Let us suppose that there are four girls at school, Anna, Bertha, Cora, and
Dora, and that someone had observed that:
To
each
solve this
letter
problem we
let
A, B, C,
corresponding to the
home" and
initial
at
of a
girl's
name.
home,
girls,
capital
(AvB)DC
2.
~B
3.
CD A
~A
Converting these to the required negative form and using Marquand's symbolism,
1.
Ac
2.
Be
bA
Ca
3.
When
=
=
=
=
these four statements are fed to the machine, the face will
appear as
mine
we have:
in
Figure 85.
An
D may
ABC
combine with
either
or abc.
In other words,
when Dora
is
III
ABC
either
or abc.
at
home
all at
holds true
Figure 85
While preparing this chapter in 1956 I wrote to Princeton Unian effort to learn the present whereabouts of Marquand's
historic machine, the first device of its kind to be constructed in the
United States. No one in either the philosophy or psychology deversity in
it
might be.
An
exten-
Logic Machines
112
classified scrapbooks,
Marquand.
James Mark Baldwin,
his Dictionary of
In 1882
and Diagrams
in
an
article
on "Logical Machines"
(in
all
from view as the premises are imhad designs made by means of which the
same operations could be accomplished by means of electro-magnets.
inconsistent combinations are concealed
Nothing whatever
He
is
also
known
its
Professor Alonzo
in detail
probably the
first
is
This
is
was ever actually built. Dr. Mays gives 1885 as the probable date
the design was drawn. The wiring diagram is of no special interest,
since it merely provides an electrical method by which the keys of
Marquand's machine can turn the pointers. Had small light bulbs
been available at the time, Marquand would undoubtedly have used
them instead of his electromagnetically operated pointers.
Mention also should be made of another logic device invented
by Marquand and described in his article "A Machine for Producing Syllogistic Variations," Johns Hopkins University Studies in
Logic, edited by Charles Peirce, 1883, p. 12. (A paper by Marquand on "The Logic of the Epicureans" also appears in this volume.) By turning a crank, the face of this machine exhibits in turn
the eight different forms in which a syllogism can be expressed,
assuming that each statement of the syllogism has two "contraposed" forms (e.g., "All A is fi" or "All not-fi is not-A ) The device
would have delighted Ramon Lull, for it is merely an elaboration
.
such a way
that,
113
machine automatically runs through a Lullian table of exhaustive combinations for three statements, each of which is expressible
in two forms. As Marquand points out, the same device can be used
the
it
make
for an extremely
compact and
is
"may be made
To
illustrate,
Ma-
and D,
reader
may
wish to solve
this
problem by Jevons's
logical alphabet,
Although
traditional syllogisms
is,
as
we have
seen, rather
awkward. Since
114
Logic Machines
and Diagrams
tastic
Genoa,
and author of
Italy,
sev-
machine
is
Mechanical Models)
of
an attempt
translate
to
the structure of
movement somewhat
It consists essentially
man-
in the
of a triangular
consideration.
Each group
representing "all" of
is
B)
is
B)
is
class.
under
class,
so that,
When two
a negative relation
is
its
belts that
to the syllogism
a small wheel of
direction.
manner appropriate
is
when A
is
turned,
Thus
opposite directions,
in
A and B
An
(No A
with a belt
small wheels
is
tive
(Some A
B) and
a crossed
belt
to a large
particular
negative
one
is
(Some
the
is
is
syllogism: All
No A
is
not B).
is
B,
Thus
No B
the
is
C,
Figure 86.
to
same
shown
in
Figure 86.
If
cause
all
will
is
not valid.
The device
is
designed primarily
model of
115
than a machine
To make
by the
belts, a
essary.
it
nec-
is
two smaller wheels on the same axle (since if all A is B, then some
be B), but when a small wheel rotates, the large wheel must
remain stationary (since if some A is B, we know nothing about all
of A ). But this is not all. To extend the power of the machine beyond the bounds of the traditional syllogism, Mr. Pastore added a
second small wheel to each axle so that "some A is #" and "some
A is not fi" could be represented simultaneously by permitting the
two small wheels of A to rotate in opposite directions while the
large wheel remained motionless (since we know nothing about all
of A).
A chart at the back of Pastore's book shows the belt connections
for 256 possible combinations of syllogistic premises and conclu-
A must
Of
sions.
Some A
is
Some A
is
is
B,
Some B
is
C,
As clumsy
as Pastore's contrivance
is,
it
can be
no reason
logic
translated into
is
would turn
in the
same
direction.
"A
implies
#"
set
Logic Machines
116
and Diagrams
electric
1.
2.
3.
The Mitchell to whom Peirce refers is one of his students. Prof. O. H. Mitchell,
whose influential paper "On a New Algebra of Logic" appeared in the Johns
Hopkins University Studies in Logic, edited by Peirce in 1883.
book is reviewed at length by Andre Lelande in Vol. 63, 1907, pp.
268 ff., of the Revue philosophique de la France et de I'etranger. Lelande is
sharply critical of Pastore's confused belief that his machine provides a genuine "experimental" approach to formal logic.
There is a sense in which all mechanical phenomena are expressive of logical
relations. A lever with a fulcrum at one end will lift a weight at its center "if
and only if" the other end is raised. But if the fulcrum is at the center, a
weight on either end is raised only when the other end is lowered, a precise
analogue of exclusive disjunction. A typewriter contains hundreds of working
parts that can be considered expressions of "and," "or," "if, then," and other
logical relations. This is what Peirce had in mind when he wrote, ".
every
machine is a reasoning machine, in so much as there are certain relations between its parts, which relations involve other relations that were not expressly
intended. A piece of apparatus for performing a physical or chemical experiment is also a reasoning machine, with this difference, that it does not depend
on the laws of the human mind, but on the objective reason embodied in the
laws of nature. Accordingly, it is no figure of speech to say that the alembics
and cucurbits of the chemist are instruments of thought, or logical machines."
(Charles Peirce, "Logical Machines," American Journal of Psychology, Vol. 1,
November, 1887).
Pastore's
7:
Window
Cards
B.
oth the Stanhope demonstrator and the Pasmachine operate along mechanical lines that are in some ways
analogous to the formal structure of syllogistic inference. If one is
tore
unconcerned with such analogy, desiring only a device that will produce the required conclusion from any pair of premises, then it is
possible to invent a wide variety of simple gadgets for such purpose.
Perhaps the simplest is a set of cards, one for each possible premise,
with openings or "windows" cut on the cards in such a
is
way
that,
my
article
When
on "Logic Machines"
if
designed
(Scientific
middle term
all
If
we
All
No
No
is
is
is
Some
is
All
is
Some 5
P is M
Some
Some ^ is
not
is
Some 5
Some S
is
is
is
not
not Z5
No P
is
No S
is
Some
P is M
Some
P is not M
NoSisP
Some S
Some
is
3" is
P
not
Some S
is
is
P
Some s
is
not
Minor Cards.
All
All
is
is
No
S"
No
VIZ is
is
118
Some 5
Some
in
is
5"
Some 5 is
Some
not
is
not
S.
Window Cards
appear
119
in the
applied the
cylinder, with
windows,
in
chapter just
in the
to the Science
simply a
it
solid
third device
Cunynghame
around an inner
flat
One cardboard
When two
smaller one.
sions,
if
sense,
it is
possible
all
com-
which combinations yield valid conclusions and what those conclusions are.
None
its
construction or
necessary
For
may
in
this reason,
new answers
My
in
set of
is
reproduced
Figure 88. In some respects these cards are simpler than Cunyng-
"No A
is
ZT means
is
the
fi" is the
it
of the other.
on
top,
is
put on top
that
shows
To make
this clearer,
if
the fol-
^
^
ii
M ^
W
^
o
c
co
0)
o
CO
CO
o
c
o
CO
CO
^x^j
^ 1
^
&
~i
L_
in
CO
01
....
ST
11
<b
Ml
o
CO
13
L1
1
,M-,\
ffl
^
^
^
^ m
120
5
z
0-
CO
CO
^
m 1
CO
CL
CO
1
z
CO
Window Cards
121
of Oz,
Somebody
Oz
in
Frank Baum's
valid:
is
stole
We
in L.
Ozma.
steal.
in
Oz
is
wicked.
Oz
people
in
subject (5),
in
of the
mood IAl
(Dimaris)
in
the
figure:
first
All
Some 5
Some S
To
is
is
is
we
two premise cards and place them together, either above the other. When we cover them with the conclusion card we find that the window for "Some S is P" is solid
black. This confirms the syllogism and indicates that the Patchwork
Girl was not so scatterbrained as her behavior and conversation
test this,
find the
often suggested.
A
To
more elaborate
test a
card, put
set of
syllogism cards
is
on top of
it
"T"
will
appear
in
invalid,
If
on top
is
valid, the
the
syllogism
will
appear along
the
ward
to use
problems
more than
in
fairly well
with elementary
number
If
there arc
of cards required
is
so
no useful purpose.
Figure 90 shows a
set of
~.
o O
5 <
CO
li.
vxy
o
c
E
D
>
>
a>
4)
<U
CD
-2
>
"a
-r,
-r,
'Z
a
>
>
9)
10
I
U
Q.
cu
0)
E
o
co
00
10
tc
Q.
< m o Q
~7
o
c
1-
*r
a>
d
v
0)
Z 3
W
m
Q-5
UJ
u.
50.
oo5
a>
o a
i^
e>
005
5o_
o o
Sen
\vs
o
CO
1
i
<3-
<
j*
CO
1-
ELS
EEo
1L
j/5
1) 0)
EE
UJ
CQ
5oo
o o
tow
a.
<
1
Q_
::
</>
00
a o
zz
zz
1-
S\N
i-
P0
P0
aa
ai
UJ
5
a
<
((-
CM
UJ
li.
CL
c/>
V-
hI
CO
<
a
3 5
<
LU <
5 o: o
Cl
122
CO
cj
rr
IT
ffl
,n <T
rr CJ
O 2
o<
o
o
Window Cards
123
BASE CARD
2.
TRUE
3.
FALSE
4.
IMPLICATION
5.
IMPLICATION
6.
DISJUNCTION
7.
EXCLUSIVE
8.
EQUIVALENCE
9.
NON-CONJUNCTION
DISJUNCTION
Figure 90. Triangular grid cards for propositional logic. Shaded areas are cut out.
Premises
for
through 9 are for binary relations. The basic form of the relation
shown on
commonly encountered
is
equivalent
To
solve a problem, pick out the required card for each premise, turn-
Logic Machines
124
of the triangle. After
all
and Diagrams
1.
(it
does
Combina-
this
procedure
if
clear.
Suppose we
B and C from
the
following premises:
is
true
If
is
Either
{A
true,
then
or
is
false
(B D
cards on card
through the
When no
Only a
1.
grid.
We
is
at the
2, 6, 9.
We
single combination,
combinations
~A
(fivC)
is
A ~BC,
false
and
all
is
is
three
visible
true.
it
to the use of
Venn
problems. Discussions of
and 3 may be reviewed for addion how to handle the triangular cards.
Window cards have little value except as novelties, although they
suggest how easily a punch-card machine could be devised that
would take care of more complicated problems of formal logic with
considerable efficiency. The most promising line of development,
however, is offered by the recent electric network machines which
these procedures in Chapters 2
tional details
References
1.
was probably a
in attempts to learn
more about
its
construction.
Electrical Logic
8:
Machines
have developed models of logic machines as compact as the Comptometer and the electrical adding machine. Unfortunately, problems
in formal logic, too difficult to be solved in the head, are seldom
encountered in either everyday
life
As
first
machines of Leibnitz, Pascal, and Babbage. At present, logic machines that operate by electric relays or electronic switching elements offer the most promising future, but even in this domain
current research
Who
is still
in a
was the
first
to build
in
As
p.
machine was
first
built
published descrip-
and demonstrated
as early as 1936.
125
Logic Machines
126
Burack's machine
is
designed to
test
all
and Diagrams
including
syllogisms,
Figure 91
case
).
The lower
contains
part of this
plywood
thin
To
test a
required
three
the
of
tills
the
in
side of
upper half
Metal
suitcase.
the
left
contact
panel so that,
invalid,
if
points
on the
the s\llogism
up on the
I
connections
electrical
between COntad
is
light
of
lyllogitm machine.
Cour-
Benjamin Burack
one
for the
and one
An
Since Burack
first
many
other types of
machines have been built, most of them by unistudents interested in logic and cybernetics, that it would be
electrical syllogism
versity
is
even possible
Geniac
parts of a
to build
kit
an
None
sponds
like the
syllogism machines has a network that correany formal wa> with the structure of class logic. They are
of these
in
window cards
When, however, we
turn to the electrical machines that have been designed for the
propositional calculus
we
enter an
altogether
different
domain.
Electrical Logic
Here we
127
Machines
now
member
a staff
of
M.I.T.,
it
appeared
in
back to
circuit design.
Or
new
circuit
can be devised by
light
on the
logical aspects of
more
structed with
is
attention paid
to
special circuits
work. 3
Since Shannon's article appeared, rapid advances have been
made
important papers on
journals.
academic
Once again
this topic
interest only,
in
engineering
enormous
practical value.
seemingly
its
Logic Machines
128
new
work
It is
every reason
is
and Diagrams
all types,
from
adding machines to the giant brains, are not designed for han-
computer constructed
culus.
The
electrical
taken.
is
is
true.
valid,
is
false.
Thus
Electrical Logic
Machines
129
Figure 92. The Kalin-Burkhart machine, front and rear views. (Courtesy of William Burkhart.)
stop automatically
when
is
again manually.
At present
ber of the
the
staff of
machine is
Arthur D.
is
in the possession of
Little, Inc., a
now
Burkhart, a
mem-
research organization in
chief of the
computer laboratory.
Logic Machines
130
and Diagrams
is
it.
It
can
tunately by the time such a problem has been translated into a form
the machine can handle, and the answer translated back into circuit
theory, the
the Kalin-Burkhart
when
machine
it
It is
ing rapidly back and forth from true to false. In a letter to Burkhart
in
may be
racket."
It
chine
is
in a sense
Its
handle
many more
terms, a
power derived
is its power to
from the fact that
piano.
chiefly
truth
if
table
many
down
it
simultaneously
may be
consid-
ever,
by adding
pleted.
to the
Logic Machines
Electrical
131
announced or described
in journals.
of references in English to
some
The following
is
a partial
list
of these machines.
its
solutions
on an
oscilloscope screen.
133.)
2.
machine
to the
when
all
the
"Think" button is pressed. ("Mechanized Reasoning. Logical Computers and Their Design," by D. M. McCallum and J. B. Smith,
Electronic Engineering, Vol. 23, April, 1951
3.
New
).
first
it
delivered
its
answers
in a
way
is
one "true" line of a combined truth table) without running through an entire truth table
until
ciple
that
it
it
encounters one.
is
is,
If all
which only a
cuts
down
vice
was
single
answer
is
complex problems in
demanded. In such cases the machine
built in
1951
at the
Edinburgh
The
de-
Logic Machines
132
and Diagrams
A ten-term "truth-function
5.
J.
B.
It
proposed by Jan Lukasiewicz in which all parentheses are eliminated by placing the variables and connectives in order from left to
right according to the conventions of the notation. This has certain
mechanical advantages.
It is
In a
letter to
machines,
(i.e.,
I'll
bet this
is
way
to
8,
1954.)
through truth
of
modern
but
if
logic
machine
wishes to
is
is
believe, to run
much
know what
problems
in
which one
determined by a given
set of statements,
it
should be possible to
if
It is
at
modern
an accelerating
number
rate.
of possible
methods would
seems
likely, therefore, that logic machines will be constructed some day
on a basis that will eliminate the scanning procedure. Such macombinations that even the
fastest
electronic
1?
Figure 93. Front and back views of the Burroughs truth-function evaluator. (Courtesy of Bur-
roughs Corporation.)
Figure 94. Control panel of the Burroughs truth-function evaluator. (From the Journal of the
Association for Computing Machinery, April, 1957.)
133
134
and Diagrams
Logic Machines
chines would
show
instantly, after
is
fed to
them, the exact status of every variable. For example, each term
would have,
say, a red
false. If
the
premises are consistent only with the truth of A, the falsity of B, but
Although such a machine would be clumsy to use for testing theoit would solve certain types of logical problems with greater
speed than any of the present machines that operate on a scanning
rems,
basis.
It
was mentioned
computer,
problems
to assign binary
numbers
To do
this
it
is
first
necessary
is
done
as follows:
We begin
A
B
The
now be
tells
us
similar
in the
first
manner we can
arrive at a designation
binary functions:
=
A ^B =
A \B =
A DB =
B DA =
A
vfl
number
Logic Machines
Electrical
now
It is
is
135
when
it
binary numbers,
it
in
will
expresses the
to the truth or falsity of the individual terms. In fact the first Fer-
ranti
in just the
way
this
way,
To make
to be the designation
is
analogous to the
corresponding pairs of
digits.
two premises:
If
and only
is
if
is
true,
is
true.
true.
states the
number
equivalence of
originally assigned to
corresponding pairs of
number
the
last
A and
for the
is
B,
we
first
premise,
A, namely, 0101.
digits
is
for the
If
we now
obtain 0001
multiply the
as the designation
we
is
tells
us that only
true.
Referring to
chine.
If
number
final
designation
compartments
and Diagrams
Logic Machines
136
become shaded
Venn system
in the
or
all
of ones,
indicates that
it
all lines
number con-
are true. In other words, the statements fed to the machine consti-
tute a
how
its
terms vary in
numbers. Thus
terms
we must
if
number
number
of
of digits required
A
B
10 10 10
110 11
1111
will also
"or" of disjunction
is
handled
The
inclusive
when
the
first
implication
is
To
we add
the
numbers according
nation
number
of 1111. This
is
tells
true;
computer so
it
it is
is
a tautology.
in this binary
easily.
However,
to ask
137
Machines
Electrical Logic
an
like asking
is
The
giant brains
it
problems.
we
in the simplest,
tric
it
is
is
truth table
A ~B.
tive statement
is
AB
~AB
machine
v B, but the
valid:
is
powerless to perform
this
reduction.
The
task of "minimizing" a
if
number
is
often
this as well as
of variables,
all
How
there
is
efficient
constructed.
The
inventor,
made
built, for
is still
undecided, but
when he designed
minimizing machine. The
is
now
where
in
its
powers
are,
Logic Machines
138
may
logic computers.
and Diagrams
area in the
field of
References
1.
Inc.,
cybernetics.
2.
of Couturat's Algebra logiki. Details were worked out in 1934-1935 by the Russian physicist V.
I.
of this work, the same views were set forth in 1936 in a Japanese journal by
Akira Nakasima and Masao Hanzawa. See Journal of Symbolic Logic, Vol.
December, 1953,
3.
18,
p. 345.
The author
4.
problems
arise in
not
the
first
to
work out
5.
this
method of manipulating
Electrical Logic
Machines
139
six variables.
most economical
circuit.
is
"most economical"
mizing, all based on essentially the same procedure, but a fully satisfactory
technique has not yet been worked out.
Recent papers on the topic include "Circuit Minimization," by Edward Samson and Burton Mills, AFCRC Technical Report 54-2 (a United States Air
Force
6.
p.
627.
For early attempts to tackle the minimizing problem, see the last chapter
of lohn N. Keynes's Studies and Exercises in Formal Logic, fourth edition,
1906. Keynes gives several methods, including a system of "multiplying out"
devised by W. E. Johnson and published as early as 1892.
A close relation to a logic minimizing machine is the "relay circuit analyzer" built by Claude E. Shannon and Edward F. Moore for the Bell Telephone Laboratories. Murray Hill, N.J. The machine verifies whether a circuit
satisfies given logical specifications and also makes a systematic attempt to
reduce the circuit to simplest form. Since the machine considers each switch in
terms of open, closed, or "don't care" positions, it is capable of applying a
three-valued logic to switching circuits. See "Machine Aid for Switching
Circuit Design." by Shannon and Moore, Proceedings of the Institute of Radio
Engineers, Vol. 41, October, 1953,
p.
1348.
The Future
Machines
9:
of Logic
little
mean
The few
logic
have
been constructed are obviously crude models, and there are probably all kinds of ways in which compact little logic machines, operating along mechanical lines, can be designed. But the power of
such devices
is
that
likely
it is
unlikely
much thought
will
be given
devoid
no formal analogy
not, to
my
be confined
ises
140
and conclusions.
It
M, P
prem-
variables,
141
to the
When
it
in the
be
difficult to
construct electrical
machines designed
(like the
now
We
is
going on.
J.
Than Thought,
edited by B. V.
Bow-
sorts,
and
Edmund
Burkhart machine {Giant Brains, 1949, Chapter 9), gives a complicated problem involving insurance coverage and shows how
it
will
Logic Machines
142
make new
and Diagrams
deductions, translate
ver-
its
of philosophy
by
domain
vast
electrical
of axioms.
of multivalued logics
machines. As yet
machine explore
letting the
new combinations
the implications of
The
and
and
know
lies
open
to exploration
ma-
of no attempt to build
why
this
cannot be done. The fact that electrical switching circuits are "twovalued" (on or off)
is
on matrix
(An
exciting possibility
is
tables
Such
have even
less practical
logic
machines would
but as tools for research in multivalued logics they might save the
professors a vast
to
strict
machines
I.
first,
little
may
machal-
work along
brilliant creative
these lines.
it
is
true that
we encounter
Church
143
it
machine or program a
computer
digital
is
no
is so complex and time-conbeyond the speed and capacity of present machines? The answer is certainly yes. A strenuous effort is now
under way - to develop a complex information-processing system
(called by its inventors the "logic-theory machine") capable of
searching for proofs of logic theorems in a manner closely analogous
to the way a human logician searches for such proofs, namely, by
trial and error, intuition, and sheer luck. The machine starts with a
few basic axioms in its memory, then tries to find a chain of
theorems that will eventually include the theorem it is trying to
prove. Various cues guide the machine along the most promising
lines, and the machine also modifies its search in the light of its trial
suming
that
is
it
experiences.
Although the logic-theory machine has been worked out only for
the propositional calculus (in which proofs can also be found by
the exhaustive truth-table
method)
its
general technique
decision procedure
is
is
appli-
possible, as well as
if
known,
(e.g., solving
is
beyond the
a difficult chess
human
is
it
will
be
and solve
problems within a reasonable computing time in areas where exhaustive mechanical procedures are impossible, not known, or
duplicates the
impractical to adopt.
The system
is
wiring of
way
is
such a
that
in
electronic
devices,
Logic Machines
144
and Diagrams
way
about
its
human
Memex would
quickly
machine. 3
It is
machine
little
such machines on a
much
made
it
possible to construct
dared dream.
Were
"world brain" that would make use of electronic storage units and
logical circuits for
its
Gargantuan task of
collecting, collating,
and
At present (1958)
the Soviet
strides
New
145
Institute of Scien-
tific
year.
Institute of Scientific
in
Informa-
articles
ing 1,400 from the United States. Dr. D. R. Newth, a British zoologist,
trip to Russia.
"No
God
is
doing
this job,"
Hard Facts on
in
he
Robert
By
1957 only thirty of 1,200 Russian scientific journals received by the United States government were being translated
and made available. A group of American industries, after spending
$200,000 for five years of research on an electric circuit problem,
discovered that the answers had been published in Russia before
they had even started their work. A Soviet paper on hydrodynamics
was secretly translated seven times by seven separate government
contrast, in
agencies, each
unaware of the
ner, president of a
New York
other's effort.
When Lloyd
V. Berk-
this
To
logic
logicians
and probability theory, the possible use of calculating mastill further vistas for the imagination. Can a ma-
chines opens up
number
of observational re-
ports
Logic Machines
146
either of these purposes.
The invention
is
and Diagrams
from under-
combine concepts
in
certain combinations.
An
element of luck
is
Certainly there
is
ability,
1950, takes
He
no machine)
is
no procedure (hence
new theorems
in logic or mathematics.
As
in
theorems
rests
147
possible,
He
lated.
gives as a trivial
fore, that the hypothesis that the individual! has black hair has a
underlies inductive reasoning, and to the degree that such reasoning employs this logic, to that degree will
it
be possible to devise
it.
The
point has sometimes been formulated [Carnap writes, p. 193]
by saying that it is not possible to construct an inductive machine. The
latter is presumably meant as a mechanical contrivance which, when fed an
observational report, would furnish a suitable hypothesis, just as a computing machine when supplied with two factors furnishes their product. I am
completely in agreement that an inductive machine of this kind is not possible. However, I think we must be careful not to draw too far reaching
negative consequences from this fact. I do not believe that this fact excludes
.
the possibility of a system of inductive logic with exact rules or the possibility
to
me
in
deductive logic. 4
more
limited, aim. It
The
is
seems
similar to that
From our
all
into
areas of seemingly
edge
rests ultimately
with
all
knowlnumber,
scientific
finite in
situation
Logic Machines
148
tion rests squarely
on the
possibility of obtaining
and Diagrams
an exhaustive
list
its
essential aspects. It
is
in-
it
might), with a
finite
number of
new law
returns of discovering a
much
same way
new gadget
or inventing a
in
than mechanical.
that a
Add
all
Kurt Godel,
one
in
enough
am
(artificial creatures,
modern
mammoth
ness,
brains
become
that
lack
duplicate
psychotic,
themselves,
often
outlast
or
give objective,
moral and
it
became
unemotional answers to
questions,
all
political ones.
it
grew
in
including
power
until
and
short stories.
The
of their
ler's
own
may develop
creative powers
and a
will
who
Bierce's terri-
losing a game. Charles Peirce did not share these fears. In his article
"Logical Machines"
in
Chapter 6) he
149
to
itself
its
...
would
still
original initiative,
calculated to do. 5
it
do
to
stance,
We
its
is
thread
it
if
own
that
it
proofs,
much
has too
The
difficulty
initiative, that
it
is
in-
original
an original professor.
If
The mouse
it
first
blunders his
way
by trial and error; but he "remembers" all that he has learned about the maze so that when given a
second trial he runs the maze without a single false turn. Machines
of this sort can acquire unpredictable behavior patterns,
and
to the
different matter
from
quote with
his permission):
it
and
do next on the
repeats.
what
basis of
For example,
it
it
first
to
in calculating a table
when
answer
is
of sines or
it
computes
same as
the
improving
its
and Diagrams
Logic Machines
150
we must
ing
generalize, and this is a creative process. Before building true learnmachines we must first learn to build creative and inductive ma-
chines.
Machines are things which manipulate symbols exactly as they are wired
to do. If we interpret their inputs and outputs as being numbers, then the
machine is a computer. If we interpret the answers as logical statements, it is
a logic machine. If we connect the outputs to a motor, the machine is a
modern elevator control system. If we connect them to a mechanical mouse,
it is a parlor game.
.
man
as nothing
is
a philosophical mechanist,
regarding
its
it
is
environ-
tool
idealist
who
believes
which learns
learns.
much
arise
from an extremely
intricate, as yet
unknown
may
type of neural
two
attitudes
may
plorers
(first
published in
November, 1949) tells of a group of space exwho come upon a small uninhabited planet completely sur-
Startling Stories,
151
downward
could be
built.
spoon
slides
as
he
structed.
References
1.
p.
"On
the Representa-
by Toshihiko
Kurihara, and "On the Representation of Many-valued Propositional Logics
by Relay Circuits," by Kamenosuke Yasuura, both in the Technology Reports
tion
of
of the
2.
Finitely
Kyushu
Many-valued
Logics
by
Electric
Circuits,"
See "The Logic Theory Machine" by Allen Newell and Herbert A. Simon,
Rand Corporation Report P-868, July 12, 1956, and Institute of Radio Engi-
"Empirical
3.
Shaw, and Simon, Rand Corporation Report P-951, Jan. 11, 1957; and "Programming the Logic Theory Machine" by Newell and Shaw, Rand Corporation Report P-954, Jan. 11, 1957.
The analogy between electrical circuits and nets of nerve cells in the brain is
so striking that the question immediately arises: is "thinking" a process involving actual circuits, switches, and scanning devices? Many important papers
groping in this cloudy area have appeared. The interested reader should consult chap. XLVI on "Boolean Algebra of Neural Nets" in Mathematical
Biophysics by Nicolas Rashevsky, revised 1948 edition (Rashevsky's earlier
articles on the subject are given in a list of references at the end of the
chapter);
"A
by Warren
S.
p.
49); and
Immanent
in
Nervous
Activity,"
as
Pitts,
Bulletin
On p. 297 a pair of diagrams illusbetween deductive logic (topological and nonquantitative) and inductive logic (requiring numerical measurement) could be interpreted as illustrations of the two ways Lord Stanhope used his demonstrator
one for syllogisms and the other for what he called "The logic of probability" (see chap. 4). Actually, Stanhope's device is the first crude attempt at
precisely the kind of "inductive machine" Carnap has in mind.
Cf. pp.
trating
the
difference
Logic Machines
152
5
and Diagrams
In
live falsehood."
Index
Bowden,
Aquinas, Thomas, 4
Archer, A., 131 n
Aristotle,
and formal
logic,
17-18
Buridan, Jean, 30
Lull)
acceptance
Astrology,
100-101
Brentano, Franz, 34
Ramon
B. V., 141
of,
by Lull,
15-16
137
on machines, 149-150
Burks, Arthur W., 132
on syllogisms, 33-34
system of induction,
147-148
Baum,
Bentham, George, 81
Berkeley,
137,
Edmund C,
138,
141
Carroll, Lewis, 51
counter method
Biconditional, 63
Bierce,
of, 78,
106
graphs, 45-48
Ambrose, 148
153
and Diagrams
Logic Machines
154
by Renaissance Lul22-23
by Venn, 32, 49-50, 61, 86,
96-97
Venn
(See also
and propositional
circles)
137
133
logic, 128,
135
Disjunction, 63
139
Conjunction, 62, 76
Conditional, 64
Electrical
puter)
Couturat, Louis,
Cunynghame, Henry,
of,
grid principle
117-119
logic
Havelock, on Lull, 8
Euclid,
Demonstrator, Stanhope,
De
on, 86-87
84-89
De Morgan, Augustus, on Hamilrules for operating,
125-138,
Morgan
machines,
140
26/?, 138/7
Euler,
fifth
138/?
proposition of, 30
Leonhard,
circle
method
of,
23, 31, 39
graphs
(see
in,
20
Graphs,
existential)
37
laws of, 65
Feedback
87
Fitzkee, Dariel,
logic,
lators,
method, 31
network
calculus,
as
60-79
pedagogic device, 29
Peirce's
20
Functional
2,
calculi,
20-21
machines
for,
142
29
computer, 131
views on Lull,
87
Diagrams,
logical
method, 54-58
"Gorgon syllogism," 35
Graphs, existential of Peirce, 5458
Great Art of Ramon Lull, controversy over, 1-2
examination of, 8-9
methods, 9-20, 146
(See also Lull,
Ramon)
155
Index
Grid-cards
Cunynghame
principle,
method, 117-118
123-124
119-121
for syllogisms,
Hamilton,
William,
De Morgan. 37
controversy with
of
quantification
34-35
Hanzawa, Masao,
59
34-37.
30,
the
138/;
Georg Wilhelm
Friedrich, 14
Clark
syllogism
L.,
Wilhelm von, on
38-39
arithmetical machine of. 125
machine
Huxley, Aldous,
19,
I.,
75, 142
26
mechanical
Hypothesis,
syllogisms,
124/7
of,
dif-
ferential" of, 10
Leibnitz, Gottfried
"structural
14,
predicate,
Alfred,
of,
procedure
Logical
108, 113
Inductive
machines,
possibility
of,
145-148, 151
Father
Ives,
Yves, Father, of
(see
Lombard, Peter,
Loyd, Sam, 19
Lull,
Paris)
(See
Jevons,
theories of.
George
logic
100
machines
106,
Boole,
of,
92-93,
99104-
91-101,
135
of, by,
96-97
A.,
128,
129,
141
125-139
124/;
130.
91-101.
104-106.
128-130,
109
machine,
Ramon
132
Kalin-Burkhart
of
Jevons,
Kalin, Theodore
Great Art
economic
Lull, 18
also
Lull)
91-92
and
on
14/7
Marquand, 106-1
Pastore, 114-116
13
Logic Machines
156
Machines, logic, Stanhope, 80-89
Venn, 105-106, 109
Marks, Robert W., 131
Marquand. Allan, graphs, 43
logic machines of, 106-108, 112
and Diagrams
Walter,
Pitts,
15 b/
102
of, 29,
76
Post. Emil,
113, 125
diagram,
Rabelais, Francois, 2
112
Mediate inference,
analysis
by
of,
32-33
Aristotle.
Memory
35, 37
Reichenbach, Hans, 77
Relational
algebras,
machine
101.
137-138
Jr.,
Moore. Edward
Russell, Bertrand, 33
F..
for,
142
Reshevsky. Nicolas, 151//
machine. 143
139
paradox
Morris, Charles. 19
of,
106
130
Schiller. F. C. S.,
151
Multivalued
machines,
logic
142,
Schuppe, Wilhelm, 34
Semantics, 10, 141-142
151
Sestakov, V.
Newth, Dr. D.
J..
R..
Shaw,
45
138
of,
J.
C,
27
151
145
mechanical
and
work
I.,
Shannon, Claude,
mouse, 149
63-64
Nonequivalence. 63
machine
114-116
Patterson, George W., 112
Pastore. Annibale, logic
of,
Smith,
41-
42
on Euler's
circles,
existential
graphs
58
(see
Graphs,
existential)
J.
B.,
Square of opposition, 29
Stanhope, Charles, demonstrator of,
82-89, 141, 151
other methods, as compared to,
86-87
syllogisms, technique of, 80-89
Stanhope, Lady Hester, 89//
Strict implication,
75
157
Inde)
Syllogisms,
Aristotelian,
30-35
18,
De Morgan's
87
Euler's
circles
for
testing,
Hamilton's system
31
17-124
139/;
of Sigwart. 34
machine
for
135-136
testing,
80-90
Venn
36-37
34, 37, 80
of,
Stanhope's
topological
39-40
41,
properties
of.
28-29,
141
Tarski, Alfred, 77
W. Gray, 149
Warren, Don W., 132
Wells, H. G., 144-145
Window cards (see Grid cards)
Three-valued
logic,
75-77
Walter,
Wright, Jesse
B.,
132
eg
Date Due
ie d
Due
to
Returned
ata
160 78G227I
12b2 D3Sb1
ARCH*
FINE ARTS
UBAAJTY
RINE ARTS
160.78
G227^
LIBRARY
AUTHOR
TITLE
DEC
21 mx /t>+
Ml*/
fi^fib
This
is
LOGIC M^CHIZffES
complete survey
and
&
of mechanical
electrical
to solve
machines designed
problems
in
MARTIN GARDNER
geometrical methods
for doing the
same
thing.
Ramon
goes
through the fascinating history of logic diagrams
and machines; and concludes with the
complex, efficient electrical
machines of today.
Spanish mystic,
Lull;
The Stanhope
Demonstrator, Jevons's
Logic Machine, the Marquand
Machine, and others are completely described
and illustrated, and logic window cards easily
constructed cards for solving
elementary logic problems
contribute a thorough
discussion of electrical
are explained.
logic
machines, emphasiz-
such devices in an
automated age, and of
future applications of logic
machines
to
such
fields
as operations research,
information storing and
processing, and efficient
circuit designing.
22841