^ 'V

r \

» 4 *'
" frâ5S^E> . “■
-Vi" Ф ';^ ’..--г, ( b ,i‘‘ ¿Й - •

.... ■Щ : ^ J - i
*-,, ¿¡) ■!:р1^"1к rJ,l\ ilii/ •• - ^ - - ■-|Д
: r-г- > - Л .:
|1

■f = J
Ö ‘iM* Щ
_.1 ^i iaWi ■—J
 LOGIC
Techniques of Formal Reasoning
 LOGIC
Techniques of Formal
Reasoning

Donald Kalish
Richard Montague
U N I V E R S I T Y OF C A L I F O R N I A , LOS A N G E L E S

7^

NEW YORK CHICAGO SAN FRANCISCO ATLANTA

Harcourt, Brace & World, Inc.

 © 1964 BY H a r c o u r t , B race & W o r l d , I n c .

All rights reserved. No part of this publication may be repro­

duced or transmitted in any form or by any means, electronic
or mechanical, including photocopy, recording, or any informa­
tion storage and retrieval system, without permission in writing
from the publisher.
ISBN: 0-15-551180-7

Library of Congress Catalog Card Number 64-10046

PRINTED IN THE UNITED STATES OF AMERICA
 CO N TEN TS

Preface IX

Chapter I. ‘N O T ’ and T F 3
1. Symbols and sentences • 4 / e x e r c i s e s • 8
2. From symbols to English and back • 8 / exer cises • I2
3. Derivability and validity • 13 / e x e r c i s e s , group i • 26 / exer ­
cises, GROUP II • 29
4. * Fallacies • 30
5. Theorems • 34 / exer cises • 37
6. * Historical remarks • 37

Chapter I L ‘A N D ’, ‘ O R ’, ‘I F A N D O N L Y I F ’ 39
1. Symbols and sentences • 39 / e x e r c i s e s • 41
2. Translation and symbolization • 42 / exer cises • 45
3. Inference rules; theorems with unabbreviated proofs • 47 /
EXERCISES •55
4. Abbreviated derivations • 57 / exer cises • 61
5. Theorems with abbreviated proofs *62 / e x e r c i s e s , group
1*68 EXERCISES, GROUP II • 7I *EXERCISES, GROUP III
•72
6. *Truth-value analysis of sentences • 73 / e x e r c i s e s *75
7. * Truth-value analysis of arguments *75 / »e x e r c i s e s • 77
8. * Historical remarks • 79
9. Appendix: list of theorems of chapters I and II • 80

Chapter II I . ‘A L L ’ and ‘ S O M E ’ 85
1. Variables, quantifiers, formulas • 85 / e x e r c i s e s • 88
2. Bondage and freedom • 89 / e x e r c i s e s • 90
3. Informal notational conventions • 90
vi . Contents

4. Translation and symbolization *91 / e x e r c i s e s -95

5. Inference rules and forms of derivation; theorems with unabbre­

viated proofs • 99 / EXERCISES • IO5
6. Abbreviated derivations; theorems with abbreviated proofs • io6
/ EXERCISES • II4
7. Arguments * 117 / e x e rci s e s • 117
8. *Fallacies • 118 / exercises • 122
9. * Invalidity • 124 / EXERCISES • 126
10. * Historical remarks • 128
11. Appendix: list of theorems of chapter III • 128

Chapter IV . 'A L L ’ and 'S O M E ’, continued 13 1

1. Terms and formulas • 131 / e x e rci s e s • 136
2. Bondage and freedom • 136 / exe rci s e s • 137
3. Informal notational conventions • 137 / e xerci ses • 138
4. Translation and symbolization • 138 / e xerci ses • 143
5. Revised inference rules • 148 / e xerci ses • 150
6. Theorems • 151 / e x e rci s e s * 155
7. Alphabetic variance • 155 / e xerci ses • 156
8. Substitution • 157 / EXERCISES • l6l
9. Abbreviated derivations reconsidered • 164 / e xe rci s e s ,
GROUP I • 167 EXERCISES, GROUP II • 168 *EXERCISES,
GROUP III • 169
10. *Invalidity • 170 / EXERCISES • 174
11. * Paradoxical inferences • 177
12. * Historical remarks • 178
13. Appendix: summary of the system of logic developed in chapters
I - I V * 179
14. Appendix: list of theorems of chapter IV • 182

* Chapter V. Automatic procedures 185

1. Introduction • 185
2. Tautologies reconsidered • 185 / e xerci ses • 191
3. Tautological implication; generalized indirect derivation • 192
4. A proof procedure for prenex formulas • 193 / exerci ses * 198
5. A derivation procedure for prenex arguments • 199 / exerci ses
• 200
 Contents . vii

6. Conversion to prenex form * 201 / exerci ses • 204

7. A derivation procedure for arbitrary symbolic arguments • 204 /
EXERCISES • 207
8. A decision procedure for certain prenex formulas • 208 /
EXERCISES • 210
9. A decision procedure for certain prenex arguments *210 /
EXERCISES• 2II
10. Historical remarks *211

Chapter V I. ‘I S ’ (in one sense) 2 13

1. Terms and formulas *213
2. Translation and symbolization *215 / e x e r c i s e s * 218

3. Inference rules; theorems -219 / e x e r c i s e s * 226

4. * Invalidity * 226 / exerci ses * 227

5. * Paradoxical inferences * 229
6. * Historical remarks * 230
7. Appendix: list of theorems of chapter VI * 230

Chapter V II. ‘T H E ’ 233

1. Descriptive phrases, terms, formulas * 233 / exerci ses * 236
2. Bondage and freedom • 237 / exerci ses * 238
3. Informal notational conventions • 238 / exerci ses * 239
4. Translation and symbolization • 239 / exerci ses * 241
5. Inference rules • 242 / exerci ses • 244
6. Theorems with unabbreviated proofs • 244 / exerci ses • 248
7. Alphabetic variance; proper substitution; abbreviated deriva­
tions • 248 / exerci ses • 255
8. Theorems with abbreviated proofs • 256 / exerci ses * 258
9. Arguments * 258 / exerci ses • 259
10. * Invalidity * 260 / exerci ses • 262
11. * Historical remarks • 264
12. Appendix: summary of the system of logic developed in chapters
I - V I I *, 264
13. Appendix: list of theorems of chapter VII • 270

Chapter V III. Definitions; formal theories 271

I. The vocabulary of formal languages • 271
viii . Contents

2. Bondage and freedom; proper substitution; alphabetic vari­

ance • 274
3. Informal notational conventions • 277
4. Derivability • 278
5. Formal theories; the theory of commutative ordered fields • 279
6. Extensions of theories; the theory of real numbers • 285
7. Definitions * 294

Chapter IX . Variable-binding operators 308

1. Definitions reconsidered • 308
2. The theory of convergence * 310
3. A sketch of further developments • 324
4. * Historical remarks • 330
5. Appendix: list of constants used in chapters V III and IX • 331
6. Appendix: list of axioms, definitions, and theorems of chapters
V III and IX • 332

B iblio g ra ph y 341
I ndexes 347
 PREFACE

T he expressions ‘logic’, ‘formal logic’, ‘symbolic logic’, and ‘mathematical

logic’ are in the just acceptation synonyms. They refer to a discipline
created by Aristotle, extended by the Stoics, studied by the Scholastics,
developed in the esoteric writings of Leibniz, and sent on its modern
career in the late nineteenth century. The discoveries of Leibniz and the
Stoics, though extensive, were until recently either ignored or misinter­
preted, and thus had to be duplicated in the early twentieth century.
Presently logic is cultivated as a branch of both philosophy and
mathematics.
This book is an introduction to logic, requiring no prior knowledge of
philosophy or mathematics. It does not aim at communicating results
about logical systems (as do several excellent but more advanced texts),
but instead at imparting a skill—the ability to recognize and construct
correct deductions. Metamathematical results are indeed sometimes
mentioned, but only incidentally and as an aid to understanding.
The logical apparatus introduced in the first seven chapters is essentially
that of the first-order predicate calculus with identity and descriptive
phrases. The theorems of those chapters (with a few exceptions connected
with descriptive phrases) are thus standard and well known. (Indeed, the
completeness and universal validity of the system has been proved in the
publication listed in the bibliography as Montague and Kalish [i].)
Some originality, however, may be claimed for the rules of the system,
which constitute, we think, a closer approximation to everyday and mathe­
matical reasoning than has previously been achieved by any formal system,
and which suggest a number of simple and practical strategies for the dis­
covery of proofs. In these chapters heavy emphasis is placed on applications
to ordinary language.
To achieve the objective of mirroring everyday reasoning, we have
sometimes had to sacrifice economy, though never precision. We have
constructed a graduated but unified development of logic from the sen­
tential calculus through the theory of descriptions, steadfastly resisting the
temptation to introduce fragmentary though perhaps attractive procedures
appropriate to isolated branches of logic. (For instance, sentential logic
X . Preface

is not developed by means of truth tables, though they are introduced in

an optional section.)
In the last two chapters first-order logic is extended (for the first time,
it seems) so as to comprehend arbitrary variable-binding operators and
is employed in a fairly detailed development of a familiar mathematical
theory. The general theory of variable-binding operators and their defini­
tions was developed in collaboration with Professor Dana Scott and will
appear also in the monograph Montague, Scott, Tarski [i].
Although the book is written so as to be comprehensible without a
teacher, it is primarily designed for use in a first course in symbolic logic,
taught in a department of either philosophy or mathematics. Such a course
may occupy either one or two semesters, and be set at any undergraduate
level, from freshman to senior. A one-semester course should treat at least
the first three, and at most the first seven, chapters, with chapter V
probably omitted. A year course could be restricted to the first seven
chapters (again with the possible omission of chapter V), in which case all
the standard material of elementary logic would be treated, or preferably
could cover all nine chapters. In a mathematics department it might be
found desirable to begin the second semester with chapter V III and to
extend the mathematical development beyond the limits of chapter IX
but along the lines suggested there. Another alternative is to cover the book
in two one-semester courses set at different levels, the second semester
beginning with chapter V, VI, or V III. To facilitate this alternative,
summaries suitable for reviewing the preceding material are given as
appendices to chapters IV and V II. As an additional pedagogical aid a list
of special symbols is given at the end of chapter IX (pp. 331 -3 2 ).
The sections and chapter marked with an asterisk, together with the
exposition in fine print, may be regarded as optional and can be omitted,
in whole or in part, without loss of continuity to the main development.
For historical and bibliographical information we have relied heavily
on two works of Alonzo Church, A bibliography of symbolic logic and
Introduction to mathematical logic. For helpful suggestions and discussions
we are indebted to Dr. J. D. Halpern and to Professors David Kaplan,
Benson Mates, Ruth Anna Mathers, and Dana Scott.
 LOGIC
Techniques of Formal Reasoning
 Chapter I
‘ N O T ’ and ‘ I F ’

L ogi c is concerned with arguments, good and bad. With the docile and
the reasonable, arguments are sometimes useful in settling disputes.
With the reasonable, this utility attaches only to good arguments. It is the
logician’s business to serve the reasonable. Therefore, in the realm of
arguments, it is he who distinguishes good from bad.
Virtue among arguments is known as validity. An argument is valid if
it is impossible for its premises to be true and its conclusion false. But this
is to speak loosely, and the reasonable do not countenance looseness.
That intuition is not a reliable judge of validity will appear from a few
examples.
(1) Suppose that a student, Alfred, satisfies the following conditions.
If he studies, then he receives good grades; if he does not study, then he
enjoys college; if he does not receive good grades, then he does not enjoy
college. Is it correct to draw any conclusion concerning Alfred’s academic
performance?
It is correct to conclude that Alfred receives good grades. This example
is simpler than most of those we shall encounter, yet it is sufficiently
complex to puzzle an unschooled intuition. The situation is worse when
intuitively plausible premises lead by way of an intuitively valid argument
to an obviously false conclusion, as in the next two examples.
(2) Suppose that Alfred, an inhabitant of Berkeley, shaves all and only
those inhabitants of Berkeley who do not shave themselves. Does he shave
himself or not? The answer is neither. For suppose he shaves himself.
Then, since he shaves only those inhabitants of Berkeley who do not
shave themselves, he clearly does not shave himself. Suppose, on the other
hand, that he does not shave himself. Then, since he is himself an in­
habitant of Berkeley and he shaves all inhabitants of Berkeley who fail
to shave themselves, he must shave himself. Thus both possibilities lead
to absurdity.
(3) Is the following boxed sentence true or false?

The boxed sentence is false.

Again, the answer is neither. For suppose the boxed sentence is true.
4 . ‘Not’ and ‘If’ Chapter I

Then it is true that the boxed sentence is false. Hence the boxed sentence
is false. On the other hand, suppose the boxed sentence is false. Then it
is false that the boxed sentence is false. Hence the boxed sentence is true.
Thus, if the sentence is true, it is false; and if it is false, it is true.
In the next example, which we owe to St. Anselm, the conclusion is
not obviously false but only controversial. To those who believe it, how­
ever, the argument should appear blasphemously short.
(4) Even the atheist, who denies that God exists, must be able to con­
ceive of Him in order to know what he denies. Suppose that God does not
exist. Then the atheist can nevertheless conceive of Him as existing and
thereby can conceive of something greater than God. But, by definition,
God is that than which nothing greater can be conceived. Hence the
atheist’s supposition leads to contradiction. Therefore God exists.

I. Symbols and sentences. We shall analyze validity in steps.

At first we shall restrict ourselves to valid arguments of a very special
kind—roughly speaking, those arguments whose validity depends only
on the phrases 'it is not the case that’ and ' i f . . . , then’.
Let us adopt the symbol ' ^ ’ as an abbreviation for 'it is not the case
that’. Thus

(1) Socrates is bald

is read

(2) It is not the case that Socrates is bald.

The expression (i) is called the negation of 'Socrates is bald’.
The phrase 'if . . . , then’ is used to combine two sentences, say,
(3) Rex is a dog
and
(4) Rex is carnivorous,
into a new sentence,
(5) If Rex is a dog, then Rex is carnivorous.
For the phrase ‘i f . . . , then’ let us use the symbol accompanied
by a pair of parentheses. Thus (5) becomes
(6) (Rex is a dog -> Rex is carnivorous) .
(6) is called the conditional formed from (3) and (4); (3) is known as the
antecedent^ and (4) as the consequent^ of (6).
In addition to and which abbreviate certain phrases, we shall
use the capital letters ‘P ’ through 'Z ’ as abbreviations for English sentences;
'A ’ through ‘O’ will be reserved for later purposes. For example, we may
Section I Symbols and Sentences . 5

let ‘P ’ stand for ‘Rex is a dog’ and ‘Q’ for ‘Rex is carnivorous’ . The
sentence (6) will then acquire the concise form

(7) ( P - .Q )

The relation between a capital letter and the sentence that it abbreviates,
unlike that between the symbols ‘ ^ ’ and and their English counter­
parts, is subject to change. By allowing ‘P’ and ‘Q’ to represent other
sentences, we may, for instance, consider (7) an abbreviation for
If snow is white, then grass is green
or
If Empedocles is hoary, then Dalmatia is icebound.

Thus, in different contexts, we may ascribe different significance to the

capital letters ‘P ’ through ‘Z ’. The ascription will take the form of a
scheme of abbreviation; such schemes will be discussed more fully in a
moment.
Despite the latitude achieved by permitting shifts of abbreviation,
it is conceivable that occasions will arise when we shall require more than
eleven sentential abbreviations. To prevent embarrassment, therefore,
we admit as possible abbreviations for English sentences not only the
letters ‘P ’ through ‘Z ’ but also any variant of these letters obtained by
adding a subscript, for example.

Q23 •

Thus the language we consider initially is English, supplemented by

and parentheses, and sentence letters (that is, capital letters ‘P ’
through ‘Z ’, with or without numerical subscripts). To avoid accusations
of provincialism, we should mention that the preferred status of English
is a matter only of the authors’ convenience; the subsequent treatment
would apply as well to French, German, or Coptic.
Our interest will not extend to all sentences of English. We shall
arbitrarily avoid questions, commands, and exclamations, leaving their
treatment to logicians of the future. We shall be concerned exclusively
with declarative sentences, that is, those sentences which are capable of
truth or falsehood. The sentences of our language fall, then, into the
following categories:
(i) declarative sentences of English, for example:
The text is readable,
Alfred will pass or the text is not readable;
6 . ‘Not’ and ‘If’ Chapter I

(2) sentence letters, for example:

T ,
P3 ;
(3) sentences correctly constructed from sentences of categories (i)
and (2) by means of and parentheses, for example:
(Alfred will pass or the text is not readable the text is readable) ,
(T -> Alfred will pass or the text is not readable) ,
-(T->-P) .

To be more explicit, the class of sentences can be exhaustively character­

ized as follows:

(j) A ll declarative sentences of English are sentences.

(2) Sentence letters are sentences.
{3) The negation of a sentence is a sentence.
{4) The conditional formed from two sentences is a sentence.

Clauses (3) and (4) admit of the following alternative formulations:

i f ) I f <t> is a sentence^ then so is the result of writing ' followed by <f>.

{4') I f <f) and iff are sentences^ then so is the result of writing followed by
(f) followed by ‘ -5^’ followed by i/j followed by *)’.

We shall use typographical displays in such a way that (3') and (4')
will be respectively synonymous with the following more succinct formula­
tions :

(3") I f <f> is a sentence^ then so is

^ <j> .
(4") I f (f) and iff are sentences^ then so is

It is evident from the preceding discussion that no sentence of our

language will contain Greek letters; although is a sentence, is
not. We reserve Greek letters for the purpose of making general statements
about sentences and, later, other expressions. Thus (3'), for example, is
to be regarded as a generalization having the following assertion among
its special cases: if ‘(P Q)’ is a sentence, then the result of writing
followed by ‘(P -> Q)* (that is, ‘ ^ (P Q)’) is again a sentence.
The symbolic part of our language comprises those sentences which
are constructed exclusively from sentence letters, parentheses,
Section I Symbols and Sentences . 7

and Accordingly, the class of symbolic sentences can be exhaustively

characterized as follows:

(j) Sentence letters are symbolic sentences.

{2) I f (f) and i/j are symbolic sentences^ then so are

and

In (6) ahd in the characterizations of a sentence and a symbolic sentence,

it may seem curious that parentheses accompany Their function is
like that of punctuation in written English and becomes conspicuous
in the case of complex sentences. For example,
( 8) 0)
is a conditional whose antecedent is

on the other hand.

'{<!> - > 4>)
is the negation of a conditional. Similarly, it is important to distinguish
between

(9)
and
(10) •
To drop the inner parentheses of (9) and (10) would obliterate the dis­
tinction.
Although parentheses are generally required to prevent ambiguity,
no confusion will arise if we omit the outermost parentheses of a sentence,
and this we shall do frequently. Thus, for example, (8) and (9) may
alternatively be written

'<!>
and
(4> ^>P) -^ x
respectively. Further, when parentheses He within parentheses, some
pairs may for perspicuity be replaced by pairs of brackets. For example,

i{4> -><P) <f>

8 . ‘Not’ and ‘If’ Chapter I

may also be written

([<^ ->0] •

EX ER C ISES
State whether each of the following expressions is a nonsymbolic
sentence, a symbolic sentence, or neither. Exercises i and 2 are
solved for illustration.

1. (^ p _^ (Q _> p ))
According to clauses (2) and (3) of the characterization of the class
of sentences (p. 6), is a sentence; according to clauses (2) and
(4), ‘(Q -> P)^ is a sentence; thus, according to clause (4) again. No. i
is a sentence. Further, it is a symbolic sentence, for it is constructed
exclusively from sentence letters, parentheses, and

2. (P -> Q -> R)
Because the characterization of the class of sentences on page 6
is stipulated to be exhaustive. No. 2, if a sentence, must result from
one of the four clauses of that characterization. But clause (i) is
clearly inapplicable, and neither of the clauses (2) or (3) can generate
an expression beginning with a parenthesis. Thus clause (4) must
yield No. 2; and for this to happen, either ‘P* and *Q -> R*, or ‘P -> Q’
and ‘R’, must be sentences. But in view of considerations like the
foregoing, neither ‘Q -> R’ nor ‘P Q* is, strictly speaking, a sentence;
each fails to be a conditional by a pair of peripheral parentheses. Hence
No. 2 is not a sentence.

3. If if if Alfred concentrates, then he will pass, then he will secure

employment, then he will marry.
4. (^ Alfred will pass -> (the text is not readable -> P))
5. ( ( ( p ^ p ) ^ p ) ^ p )
6. If the text is readable Alfred concentrates, then he passes.

2. From symbols to English and back* Frequently it will be de­

sirable to translate a symbolic sentence into English and perform the re­
verse process of symbolizing an English sentence. We consider first the
passage from symbols to English.
As was mentioned earlier, the correlation between sentence letters
and the English sentences which they abbreviate is quite arbitrary. Thus
the passage from a symbolic sentence to a sentence of English must
proceed on the basis of a scheme of abbreviation^ which will establish such
a correlation.
More explicitly, let us understand by an abbreviation an ordered pair
of sentences, the first of which is a sentence letter and the second an
Section 2 From Symbols to English and Back . 9

English sentence. A scheme of abbreviation is a collection of abbreviations

such that no two abbreviations in the collection have the same first mem­
ber.
For example, the collections
(I) Q : the lectures are dull
T : the text is readable
P : Alfred will pass
and
S : the lectures are dull
Q : the lectures are dull
T : the text is readable
P : Alfred will pass
are schemes of abbreviation; but the collection
Q : the lectures are dull
Q : the text is readable
P : Alfred will pass
is not. (The two abbreviations whose first member is ‘Q’ are at fault.)
The process of literal translation into English on the basis of a given
scheme of abbreviation begins with a symbolic sentence and if successful
ends with a sentence of English. The process consists of the following
steps:

\i) Restore any parentheses that may have disappeared as a result of

applying the informal conventions of the last section,
{it) Replace sentence letters by English sentences in accordance with the
given scheme of abbreviation; that is, each sentence letter is to be replaced
by the English sentence with which it is paired in the scheme,
{in) Replace all occurrences of
^(¡> ,
where <f> is a sentencey by
it is not the case that (f> ,
{iv) Replace all occurrences of

where <f> and ip are sentences, by

{if (f>y then ijj) .

For example, under the scheme of abbreviation (i) the sentence

(2) Q -> ( ^ T -P )
10 . ‘Not* and ‘If* Chapter I

becomes in step (i)

(Q - > ( - T -> ^ P ))

in step (ii)
(The lectures are dull the text is readable -> Alfred
will pass)) ,
in step (iii)
(The lectures are dull -> (it is not the case that the text is
readable -> it is not the case that Alfred will pass)) ,
and in step (iv)
(3) (If the lectures are dull, then (if it is not the case that
the text is readable, then it is not the case that
Alfred will pass)) .
(The parentheses in sentences such as (3), as in symbolic sentences, serve
as marks of punctuation.)
We shall generally wish to be more liberal in translating from symbols
to English than the notion of a literal translation will permit. For example,
we should like to consider the sentence
(4) Assuming that the lectures are dull, if the text is not
readable, then Alfred will not pass
as a translation of (2) on the basis of the scheme (i). Accordingly, we say
that an English sentence is a translation (or, when a distinction is to be
drawn, a free translation) of a symbolic sentence <f> on the basis of a given
scheme of abbreviation if it is a stylistic variant of the literal translation
of <f> into English on the basis of that scheme.
Because (4) differs only in style from (3), the former as well as the latter
qualifies as an English translation of (2) on the basis of scheme (i).
In the realm of free translations, we countenance looseness. Specifically,
we attempt no precise description of stylistic variance; in this connection,
intuition (here identified with linguistic insight) rather than exact rules
must guide the reader. To remove this source of looseness would require
systematic exploration of the English language, indeed of what might be
called the ‘logic of ordinary English’, and would be either extremely
laborious or impossible. In any case, the authors of the present book
would not find it rewarding.
Although no exact definition of stylistic variance will be offered, we
shall not leave the reader entirely to his own devices. Two stylistic variants
of
It is not the case that Alfred concentrated
Section 2 From Symbols to English and Back . i i

are
Alfred did not concentrate
and
Alfred failed to concentrate.
English idiom provides a number of stylistic variants for
(5) (If Rex is a dog, then Rex is carnivorous) ,
for example,
(6) Rex is carnivorous if Rex is a dog,
(7) Rex is carnivorous provided that Rex is a dog,
(8) Rex is a dog only if Rex is carnivorous,
(9) Only if Rex is carnivorous is Rex a dog.
Between (5) and (6) the only difference (apart from the use of parentheses)
is in word order. (7) comes from (6) upon replacement of ‘if’ by its intuitive
equivalent ‘provided that’. To see the intuitive equivalence between (5)
and (8), the following consideration should be of assistance: to assert
that Rex is a dog only if carnivorous is to deny that Rex is a dog and not
carnivorous, which is to assert that if Rex is a dog, he is also carnivorous;
but this amounts to (5). (9) comes from (8) by inversion of word order.
If <f) and ifj are any two sentences, a partial list of stylistic variants of
(if then ip)
is the following:
if (py then ip y
Ip if <P y
tp provided that <p ,
(p only if tp y
only if tp <p y
given that <jt, tp ,
tp in case <p y
tp assuming that (p ,
tp on the condition that <p .
Further instances of stylistic variance may be obtained by introducing
pronouns in place of nouns and by altering word order. Examples of these
and other sorts of stylistic variance will be found among the exercises of
this and later chapters.
We shall also be interested in the passage from English to symbols.
Accordingly, we say that is a symbolization of an English sentence tp
on the basis of a given scheme of abbreviation if and only if ^ is a symbolic
sentence that has ^ as a translation on the basis of that scheme, in other
12 . ‘Not* and ‘If* Chapter I

words, if and only if 0 is a stylistic variant of the literal translation of (f>

on the basis of the scheme.
To find a symbolization of a given English sentence on the basis of a
given scheme of abbreviation, the reader will find it useful to proceed
roughly as follows:

(j) Introduce 'it is not the case that' and \ i f , , . , then)' in place of their
respective stylistic variants,
(2) Reverse the steps leading from a symbolic sentence to a literal English
translation; that isy
{2a) replace all parts of the form
{if <f>, then ip) ,
where <f>attd ip are sefitences^ by

(0->0) ;
(26) replace all parts of the form
it is 7Wt the case that <p ,
where <p is a sentence^ by
^<P ;
{2c) replace English components by sentence letters in accordance with
the scheme of abbreviation; that is, replace each English component by a
sentence letter with which it is paired in the scheme of abbreviation;
{2d) omit peripheral parentheses and replace parentheses by brackets in
accordance with the informal conventions of the preceding section,

EX ER C ISES
On the basis of the scheme of abbreviation
P logic is difficult
Q Alfred will pass
R Alfred concentrates
S the text is readable
T Alfred will secure enjplo^Tnent
u Alfred will marry
V the lectures are dull ,
translate the following symbolic sentences into idiomatic English.
1. P - > ( Q ^ R )
2. S - > ( R - > [ - P - > Q ] )
3. ( R ^ p ) _ ^ ^ Q

On the basis of the scheme of abbreviation above, symbolize the

following English sentences. Exercise 4 is solved for illustration.
Section 3 Derivability and Validity . 13

4. Only if Alfred concentrates will he pass.

Sentence No. 4 becomes in step (i) of the procedure given on page
12
(If Alfred will pass, then Alfred concentrates) ;
in taking this step we first made minor stylistic changes and replaced
‘will he pass* by ‘Alfred will pass*, and then, in conformity with the
discussion on page ii, we inverted the order of the component
sentences. In step (2a) the sentence becomes
(Alfred will pass ^ Alfred concentrates) ,
step (2b) is inapplicable, step (2c) leads to
(Q ^ R ) ,
and step (2d) to
Q->R .
5. If logic is difficult, Alfred will pass only if he concentrates.
6. Alfred will pass on the condition that if he will pass only if he
concentrates then he will pass.
7. If if if Alfred concentrates, then he will pass, then he will secure
employment, then he will marry.
8. It is not the case that if Alfred will secure employment provided
that the text is readable, then he will marry only if he concentrates.

The following sentences are ambiguous, in the sense that the

placement of parentheses in their symbolizations is not uniquely deter­
mined. Give all plausible symbolizations of these sentences on the basis
of the scheme of abbreviation that appears above.
9. Alfred will pass only if he concentrates provided that the text
is not readable.
10. It is not the case that Alfred concentrates if the lectures are
dull.
11. Alfred will not secure employment if he fails to concentrate on
the condition that the lectures are dull.

3, Derivability and validity. An argument^ as we shall understand

it, consists of two parts—first, a sequence of sentences called its premises^
and secondly, an additional sentence called its conclusion. In the case of a
valid argument the premises constitute conclusive evidence for the con­
clusion. Ordinarily we shall present an argument by listing first the premises,
and then the conclusion, set off by the sign or the word ‘therefore'.
Three examples follow.
(1) P->Q . Q ^ R .*. P - > R
(2) .*. Alfred will pass or he will not pass.
 14 . ‘Not* and ‘If* Chapter I

(3) If Socrates did not die of old age, then the Athenians
condemned him to death. The Athenians did not con­
demn Socrates to death. Socrates died of old age.
It is possible for an argument to have an empty sequence of premises
(that is, to have no premises at all), as in example (2). It is also possible
for a valid argument to have a false conclusion, as in example (3). (When,
at the end of this section, we define validity, it will be seen that argument
(3) is indeed valid.) In the case of a valid argument we may be sure only
that if all the premises are true, then the conclusion will be true.
Let us for a moment confine our attention to symbolic arguments^ that is,
arguments whose premises and conclusions are symbolic sentences.
To establish the validity of a symbolic argument with the sentence <f>
as its conclusion, we construct a derivation of (f> from the premises of the
argument—that is, a sequence of steps, each justified in some way, which
lead from the premises to the establishing of
Before presenting the rules for constructing a derivation, let us consider
an example. Suppose that from the premise

(4) p
we wish to derive the conclusion

(5) (P ^ Q) -> Q
An appropriate derivation may be constructed as follows. We write first
the conclusion, (5), together with an indication that it is to be established:
Show (P Q) -> Q
Now (5) is a conditional, and we may establish a conditional by assuming
its antecedent and deriving its consequent. Accordingly we add, as an
assumption, the antecedent of (5):
Show (P -> Q) Q
P ->Q
Now we may add the premise, (4):
Show (P Q) -> Q
P ->Q
P
From the second and third lines of the derivation we may infer ‘Q’ by
means of an inference rule, (A list of inference rules will be given shortly.)
The rule used here is modus ponens; by modus ponens a symbolic sentence ifs
may be inferred from symbolic sentences

(<!>-> >P)
Section 3 Derivability and Validity . 15

and (f>. Thus we obtain:

Show (P Q) ^ Q
P
P
Q
We have now succeeded in deriving the consequent of (5). Hence (5)
is established. To indicate this fact we cancel the occurrence of *Show'
that precedes ‘(P Q) -> Q\ Further, we box the second, third, and
fourth lines of the derivation to indicate that they have served their
purpose. Thus we finally obtain:
Show- (P -> Q) Q

The derivation of (5) from (4) is now complete.

The inference rules we shall employ are the following:

Modus ponens (M P): {<j> ^ifj)
i _____

Modus tollens (M T ): ^

(j)
Double negation (DN), in two forms: ^ 0 <f>

Repetition (R): <f>

J
That is, a symbolic sentence ifs is said to follow by modus ponens from two
other symbolic sentences if and only if these sentences have the forms

{4>
and (f); a symbolic sentence follows by modus tollens from two other symbolic
sentences if and only if it has the form
(j)
and the other sentences have the forms
i6 . ‘N ot’ and ‘I f ’ Chapter I

and
.if, ;
a symbolic sentence follows from another by double negation if and only
if the two symbolic sentences have the forms (f> and
r>^(j) *
and a sentence <f> follows by repetition from a symbolic sentence iff if and
only if (f) and if, are the same sentence. M P and M T correspond to familiar
forms of reasoning; DN is the principle that a double negative amounts
to an affirmative; and the function of the trivial rule R will become clear
later.
For example, in the arguments

(P ^Q ) . P -Q ,
(P -> ^Q ) . /^P ,
^^(P->Q) .\(P->Q) ,
( P - > Q ) /. ^ ^ ( P - > Q ) ,
(P ^ Q) .-. (P -> Q) ,

the conclusion follows from the premises by the respective rules MP,
M T , DN, DN, and R.
Suppose now that we have certain symbolic premises and that we wish
to derive as a conclusion the symbolic sentence <f>. We begin by writing
Show <f> .
We may continue in one of three ways, each of which has numerous in­
tuitive counterparts in the derivations of mathematics and in the reasonings
of law courts and everyday life:
(i) By direct derivation. We write next a line that can be established
independently (for instance, a premise or a sentence accompanied by a
subsidiary derivation) and proceed by inference rules, subsidiary deriva­
tions, and citing of premises until we secure (f>. A direct derivation of
then, will have the form
( 6) Show <f)

(ii) By conditional derivation^ in case (f> is of the form

where if, and x ^^e symbolic sentences. In this case we write next, as an
Section 3 Derivability and Validity . 17

assumption, the sentence 0, and proceed by inference rules, subsidiary

derivations, and citing of premises until we secure x- ^ conditional
derivation of

then, will have the form

(7) Show {iff x)

0 (Assumption)

(iii) By indirect derivation. In this case we write next, as an assumption,

the sentence
(f)

and proceed by inference rules, subsidiary derivations, and citing of

premises until we secure a symbolic sentence x and its negation,

If (f> is itself a negation, say

we may assume ip instead of

and proceed as above. Thus an indirect derivation of cp will have the

form

(8) Show <p

(p (Assumption)

or, in case <p is the negation of a sentence ip.

i8 . ‘Not* and ‘If* Chapter 1

(9) Show iff

0 (Assumption)

Indirect derivation is also known as derivation by reductio ad absurdum,

and depends for its cogency on the following consideration. To show
that an assertion <f> holds, it is sufficient to assume that it does not, and
from this assumption to derive a contradiction (that is, a pair of sentences,
one of which is the negation of the other); for then our assumption must
be mistaken, and <f) must hold.
When the derivation of <f>, accomplished by one of these methods, is
complete, we indicate its completion by cancelling the occurrence of
^Show^ in the first line and boxing the remaining lines. (6) will then
become
Show (f) (Direct derivation)

<!>

(7) will become

Show {iff -> x) (Conditional derivation)

(8) will become

Shew(f> (Indirect derivation)
^ (f)
Section 3 Derivability and Validity . 19

and (9) will become

Show- (Indirect derivation)

An example will clarify the use of subsidiary derivations. Suppose

that we wish to derive T -> ([Q -> R] -> R)* from T -> Q\ The sentence
we wish to derive is a conditional; we therefore start a conditional deriva­
tion:
1. Show P ([Q -> R] -> R)
2. P (Assumption for
conditional derivation)

To complete the conditional derivation, we must obtain the consequent

of line I, ‘ [Q -> R] R \ One way of proceeding is to begin with a sub­
sidiary derivation; and since the sentence we wish to establish is a con­
ditional, it is natural to start a conditional derivation:
1. Show P -> ([Q R] ^ R)
2. P
3. Show [Q R] R
4. Q R (Assumption for
conditional derivation)

We continue the subsidiary derivation, employing inference rules and the

premise (in a way that will be explained more fully later), until we reach
the consequent of line 3.
1. Show P ([Q -> R] -> R)
2. P
3. Show [Q -> R] -> R
20 . ‘Not* and ‘If* Chapter I

4. Q ->■ R
5. P->Q (Premise)
6. Q (2, 5. MP)
7. R (4. 6 , MP)
The subsidiary derivation establishing the assertion in line 3 is complete.
Thus we may box and cancel to obtain:

1. Show P -> ([Q -> R] ^ R)

2. P
3. Show [Q -> R] -> R (Conditional derivation)

Now the main derivation is complete. Hence we obtain:

Show P ([Q -> R] R) (Conditional derivation)

The foregoing remarks on derivations constitute only an informal

introduction. The following is an explicit set of directions for constructing
a derivation from given symbolic premises: _________ ____
--------------- ;------------- ^
j (i) I f (f> IS any symbolic sentence^ then
Show (f)
may occur as a line, {Such lines may be accompanied by the annotation
^Assertion',)
(2) Any one of the premises may occur as a line, {Annotation: ^Premise',)
{3) U ^ symbolic sentences such that
Show {<f> 0)
occurs as a line^ then (f> may occur as the next line, {Annotation: ^Assumption
for conditional derivation^ or simply *Assumption*,)
{4) I f (f> is a symbolic sentence such that
Show <f>
 Derivaijility aiid'Valtillty ~

occurs as a line^ then

may occur as the next line; if (f) is a symbolic sentence such that
Show ^(f>
occurs as a liney then <f> may occur as the next line. {Annotation: 'Assumptioîi
for indirect derivation^ or simply 'Assumption'.)
(5) A symbolic sentence may occur as a line if it follows by an inference
rule from antecedent lines, that iSy preceding lines which neither are boxed
nor contain uncancelled 'Show'. {The annotation should refer to the inference
rule employed and the numbers of the preceding lines involved.)
{6) When the following arrangement of lines has appeared:
Show (f>
‘ XI

Xm y
where none of yi through xm contains uncancelled 'Show' and either
{i) (f) occurs unboxed among through xmy
{it) <f>is of the form
(«Ai ^ h )
and 02 occurs unboxed among xi through Xmy or
{Hi) for some sentence Xj both x negation occur unboxed among
XI through Xmy
then one may simultaneously cancel the displayed occurrence of 'Show'
and box all subsequent lines. {When we say that a sentence 0 occurs among
certain lines of a derivation y we mean that either (f> or <f>preceded by 'S how'
is one of those lines.)

derivation of ‘(-^Q -> Q' from the premise T ’ will illustrate

applications of clauses (5) and (6).
Show ( ' ^ Q - > ^ P ) - > Q Assertion
'^Q Assumption for conditional
derivation
Show Q Assertion
Q Assumption for indirect
derivation
P Premise
^P 2, 4, MP
22 . ‘Not’ and ‘If* Chapter I

The subjunction of line 6 constitutes an application of clause (5), for line

6 follows from the antecedent lines 2 and 4. The derivation can be con­
tinued by an application of clause (6), part (iii); for lines 3 - 6 have the
pattern
SJww <f>
XI

Xrn J

where none of through xm contains uncancelled *Show' and both T ’

and its negation occur unboxed among xi through xm- Thus we may simul­
taneously cancel the last unboxed occurrence of ^Show* and box all sub­
sequent lines, to obtain:

■'P) -> Q

Now an application of clause (6), part (ii), is possible, for lines 1 - 6 have
the pattern
Show
XI

Xm
where none of xi through xm contains uncancelled 'Show' and ‘Q* occurs
unboxed among through xm- Thus we may again simultaneously cancel
the last unboxed occurrence of 'Show' and box all subsequent lines, to
obtain:
Section 3 Derivability and Validity • 23

To ensure correct application of clauses ( i) - ( 6 ) , any parentheses

that may have been dropped must be mentally restored. For example,
we might be tempted to begin an indirect derivation of ‘P Q’ as follows:
Show P -> Q
P -> Q (Assumption for
indirect derivation)
But when missing parentheses are restored, this becomes:
Show (P -> Q)
(-P -> Q ) ,
in which the second line is clearly not a negation, and hence cannot be
obtained correctly by clause (4). A correct indirect derivation would begin
as follows :
Show P -> Q
- ( P ->Q) (Clause (4))
Similarly, given
(10) P ^ Q
as a line, we might be tempted to infer by DN the sentence
P -> Q
But this inference is not permitted by our present rules. Instead, re­
calling the missing parentheses of (10), we obtain

(P -> Q)
A derivation is said to be complete if each of its lines either is boxed or
contains cancelled ^Show\
A symbolic sentence (f) is said to be derivable from given symbolic
premises if, by using only clauses (i)-(6), a complete derivation from those
premises can be constructed in which
Show <f>
occurs as an unboxed line.
For example, the derivation
1. Show P ([Q R] -> R)
2. P
3. Show [Q -> R] -> R
4. Q -> R
5. P -> Q (Premise)
6. Q
7. R
24 . ‘Not* and ‘If* Chapter I

is not complete, but the derivation

I . Show P -> ([Q R] ^ R)
2
3
4
5
6

is complete; and in view of the latter derivation, T -> ([Q R] ^ R)’

is derivable from T -> Q’.
A very important feature of the notion given above of a derivation,
and one that will be shared by all analogous notions to be introduced
subsequently, is that there is a purely automatic procedure for checking
the correctness of an alleged derivation from a finite class of premises;
more generally, whenever a class of premises is such that one can auto­
matically determine of any sentence whether it belongs to that class, there
will be an automatic procedure for determining whether any alleged
derivation is indeed a correct derivation from that class of premises.
Thus the correctness of derivations, unlike the cogency of everyday
reasonings, is removed from the realm of controversy.
A symbolic argument is said to be valid if its conclusion is derivable from
its premises. Argument (i) of page 13 happens to be valid in view of the
following derivation.
I. Show- P -> R Assertion
Assumption
ist premise
2, 3, M P
2nd premise
4, 5, MP
By an English argument is understood an argument whose premises
and conclusion are sentences of English. A symbolization of an English
argument on the basis of a given scheme of abbreviation is a symbolic argument
whose premises and conclusion are respectively symbolizations, on the
basis of that scheme, of the premises and conclusion of the English
argument. A symbolic argument is called simply a symbolization of an
English argument if there is some scheme of abbreviation on the basis
of which it is a symbolization of the English argument. Consider, as an
example, the argument:
( ii) Free love is justified only if the sex drive is primary. Freud's
theory is correct provided that the sex drive is primary.
Section 3 Derivability and Validity . 25

It is not the case that if psychoanalysis is therapeutic,

Freud’s theory is correct. Free love is not justified.
On the basis of a natural scheme of abbreviation, the argument may be
symbolized as follows:
(12) Q -> S . S ^ T . ^ (P ^ T ) -Q
We confine the application of deductive procedures (that is, the pro­
cedures whereby derivations are constructed) to symbolic sentences, for
otherwise difficulties of analysis would arise which we prefer either not to
treat at all or to treat only in connection with stylistic variance. Thus,
in order to define validity for English arguments, we must proceed in­
directly, as follows.
An English argument is said to be valid (in the branch of logic presently
under consideration) if and only if it has a symbolization whose conclusion
is derivable from its premises. (Other branches of logic, and wider notions
of validity, will be considered later.) Argument ( ii) is valid, in view of the
symbolization (12) and the derivation
Show Assertion
Assumption
Premise
2, 3, M P
Premise
Assertion
Assumption
4, 5, MP
Premise

Annotations do not, strictly speaking, form part of a derivation. Their

inclusion, however, is often convenient for permitting a quick decision
as to the correctness of a proposed derivation. For example, that ‘ ^ P ’
is derivable from the premises ‘ (Q -> R)’ and T R ’ is settled
by a brief inspection of the following lines.
Show Assertion
P Assumption
P R Premise
/-wR 2, 3, MP
Show Q -> R Assertion
Q Assumption
R 4, DN
(Q -^ R ) Premise
26 . ‘Not* and ‘If* Chapter I

E X E R C IS E S , G R O U P I
Show by constructing derivations that the following arguments are
valid.
12. (P->Q )-^Q . Q -> P P
13. P -> (Q -> R ) . P - > ( R -> S ) P -> (Q -> S )
14. ([P -> Q ]-> P )-> P
15. ~P^Q . P -^ Q Q
16. Q —>^^R . '^ P —^R '^ P —
17. -(R -> Q ) . Q P
18. P —> Q P —> Q P
In solving exercises 1 2 - 1 8 (as well as later exercises) the reader will
find the following suggestions helpful but not infallible. They are intended
merely as informal advice and have not the same status as the official
directions for constructing a derivation.

(1) To derive a conditional, use conditional derivation,

(2) To derive anything else, use indirect derivation unless another pro­
cedure is immediately obvious.
(j) Enter all premises as lines, but not until one or another form of deriva­
tion has been commenced.
{4) Whenever a sentence follows from antecedent lines by M P or M T,
enter that sentence as a line.
(5) When using indirect derivation, determine whether any of the ante­
cedent lines is the negation of a conditional; if so, attempt to derive that
conditional.

We solve exercise 12 for illustration. According to suggestion (2), we

begin an indirect derivation:
1. Show P Assertion
2. P Assumption
Next, in accordance with suggestion (3), we enter the premises
1. Show P Assertion
2. '^P Assumption
3. (P -> Q) Q Premise
4. Q -> P Premise
We then notice that 2 and 4 permit an application of M T , and we follow
suggestion (4):
1. Show P Assertion
2. -~P Assumption
3. (P -> Q) ^ Q Premise
4. Q -> P Premise
5. ~Q 2, 4, M T
Section 3 Derivability and Validity . 27

Again we apply M T :
1. Show P Assertion
2. Assumption
3. (P -vQ )-. Premise
4. Q -> P Premise
5. ~Q 2, 4, M T
6. ~ (P ^ Q ) 3> 5, M T
We have now obtained, in our indirect derivation, the negation of a con­
ditional. Thus, following suggestion (5), we should attempt to derive
T Q’, and and this we do, following suggestion (i), by conditional
derivation:
1 . Show P Assertion
2. '^P Assumption
3. (P->Q )->Q Premise
4. Q -> P Premise
5. '-Q 2, 4, M T
6. - ( P ->Q) 3, 5, M T
7. Show P Q Assertion
8. P Assumption

To complete the subsidiary conditional derivation, we must derive ‘Q’ ;

thus, by suggestion (2), we try indirect derivation :
I . Shozv P Assertion
2. ~ P Assumption
3. (P Q) -> Q Premise
4- Q Premise
5. ~ Q 2, 4, M T
6. ~ ( P ^ Q ) 3, 5. M T
7. Show P -> Q Assertion
8. P Assumption
9. Show Q Assertion
10. ~ Q Assumption
It is now obvious how to complete the subsidiary indirect derivation:
1. Show P Assertion
2. -P Assumption
3. (P -> Q )-^ Q Premise
4. Q ^p Premise
5. 2, 4, M T
6. ^ (P -> Q ) 3, 5, M T
7. Show P -> Q Assertion
8. P Assumption
28 . ‘Not’ and ‘If’ Chapter I

9. Show Q Assertion
10. Assumption
11. 8, R
12. 2, R
But this completes the subsidiary conditional derivation, whose completion
in turn completes the main derivation:
Show P Assertion
2 Assumption
3 Premise
4 Premise
5 2, 4, M T
6 3, 5, M T
7 Assertion
8 Assumption
9 Assertion
10 Assumption
11 8, R
12 2, R

Actually, the derivation to which our suggestions led us is unnecessarily

complicated. An insight after we reached line 8 would have allowed us to
complete the derivation in one additional line, as follows:
Show-V Assertion
-P Assumption
(P Q) -> Q Premise
Q -> P Premise
-Q 2, 4, M T
- ( P ->Q) 3, 5, M T
Show P -> Q Assertion
Assumption
2, R

Here the subsidiary derivation of line 7 is accomplished in an unusual­

way. The assumption, line 8, is made in accordance with conditional
derivation, but the boxing and cancelling are done in accordance with
indirect derivation. This procedure is legitimate in view of the directions
for constructing a derivation; see clauses (3) and (6, iii), pp. 2 0 - 2 1 .
Other mixed forms of derivation are possible and indeed convenient, but
in each case the mixture could be avoided at the expense of a few extra
lines.
Section 3 D erivability and Validity . 29

E X E R C IS E S , G R O U P II

Show the following arguments valid by constructing symbolizations

and deriving the conclusions of the symbolizations from their premises.
Indicate the scheme of abbreviation used. Exercise 19 is solved for
illustration.

19. Argument (3) of page 14.

On the basis of the scheme
P : Socrates died of old age
Q : the Athenians condemned Socrates to death ,
we obtain as a symbolization of the argument (3) the symbolic argument
-P -> Q . -Q .\ P ;
its validity, and hence that of (3), is established by the following
derivation:
T ^ h/ivt} P

Premise
Premise
2, 3, MT
4, DN

The reader should note that (3) has other symbolizations; for example,
on the basis of the scheme
P : if Socrates did not die of old age, then the
Athenians condemned him to death
Q : the Athenians did not condemn Socrates to death
R : Socrates died of old age
the argument
P . Q /. R
is a symbolization of (3); and on the basis of the scheme
P Socrates did not die of old age
Q Socrates died of old age
R the Athenians condemned Socrates to death
so is the argument
R R /. Q
But neither of the last two symbolizations is valid (as we shall be able
to show in the next chapter). In general, the longer the symbolization,
the more likely it is to be valid; and among symbolizations of equal
length, the likelihood of validity increases as the number of distinct
sentence letters decreases.
30 . ‘Not’ and ‘If’ Chapter I

20. If Alfred studies, then he receives good grades. If he does not

study, then he enjoys college. If he does not receive good grades,
then he does not enjoy college. .’. Alfred receives good grades.
(Compare (i), p. 3.)
21. If Herbert can take the apartment only if he divorces his wife,
then he should think twice. If Herbert keeps Fido, then he cannot
take the apartment. Herbert's wife insists on keeping Fido. If Herbert
does not keep Fido, then he will divorce his wife provided that she
insists on keeping Fido. Herbert should think twice.
22. If Herbert grows rich, then he can take the apartment. If he
divorces his wife, then he will not receive his inheritance. Herbert
will grow rich if he receives his inheritance. Herbert can take the
apartment only if he divorces his wife. If Herbert receives his
inheritance, then Fido does not matter.
23. If God exists, then He is omnipotent. If God exists, then He
is omniscient. If God exists, then He is benevolent. If God can
prevent evil, then if He knows that evil exists, then He is not benevolent
if He does not prevent it. If God is omnipotent, then He can prevent
evil. If God is omniscient, then He knows that evil exists if it does in­
deed exist. Evil does not exist if God prevents it. Evil exists. God
does not exist.
24. If business will flourish on the condition that taxes will not
increase, then the standard of living will improve. It is not the case that
the standard of living will improve. If business does not flourish, then
taxes will increase. .*. Unemployment will be a problem.
25. The standard of living will improve provided that taxes will
increase. Business will flourish only if unemployment is not a problem.
Business will flourish if the standard of living will improve. .*. On
the condition that taxes will increase, unemployment is not a problem.

4. * F allacie s. In the directions for constructing a derivation, a number

of restrictions appear whose significance is perhaps not immediately
obvious. However, if we define a fallacy as a procedure that permits the
validation of a false English argument, that is, an argument whose premises
are true sentences of English and whose conclusion is a false sentence of
English, then the neglect of any but one of the restrictions would lead to
fallacies.
The fallacies involved here are of two kinds—those of begging the
question and those of unwarranted assumptions. These kinds of fallacies
can be characterized in a rough, informal way, as follows.
When, in deriving something, one uses the assertion that is to be derived,
one is said to be guilty of begging the question.
Assumptions are useful tools in logical derivations, but they must be
employed with caution. Only on special occasions may they be made or
inferences drawn from them. The use of assumptions at other junctures is
the fallacy of unwarranted assumptions, (The reader should observe the
Section 4 Fallacies • 31

difference between the use of premises^ which is justified by clause (2), and
the use of assumptions. A premise may be stated at any point in the course
of a derivation; there are no restrictions here whose neglect would lead
to fallacies.)
In clauses (3) and (4) of the directions for constructing a derivation,
it is specified that an assumption may be made only in connection with a
line containing uncancelled 'Show'. Let us examine the consequences
of ignoring this restriction, in connection first with clause (3). Consider
the argument
If Schopenhauer was married, then he had a wife.
Schopenhauer had a wife.
Here the premise is obviously true, and the conclusion is, as a matter of
history, false; hence the argument is false. But by an incorrect use of
clause (3), the validity of this argument could be established by means of
the following symbolization and accompanying derivation.
(I) P ^Q Q
I. Show Q Assertion
Premise
Unwarranted assumption
2, 3, M P
Now let us employ clause (4) incorrectly, again making an assumption
in connection with a line not beginning with uncancelled 'Show'. Consider
the false argument
Snow is white. Grass is red.
and the following symbolization and derivation:
(2) P Q
I. Show C Assertion
2. Premise
3- Unwarranted assumption
In both (i) and (2), then, we have validated false arguments by committing
one form of the fallacy of unwarranted assumptions; in neither case was
the assumption made in connection with a line containing uncancelled
'Show'.
The other restriction involved in clauses (3) and (4) is that the assump­
tion be made immediately after the line with which it is connected. It
is not yet possible to illustrate the necessity of this restriction; in fact,
its neglect will produce no fallacies until our logical apparatus is enlarged
by the introduction, in chapter III, of a new form of derivation.
Clause (5) permits us to draw inferences only from antecedent lilies.
This restriction can be violated in two ways—by applying an inference
32 . ‘Not’ and ‘If’ Chapter I

rule to a line containing uncancelled 'Show' or by applying a rule to a

boxed line. Let us illustrate the first violation.
Argument:
Snow is white. If snow is white, then grass is red.
Symbolization:
P P ^ Q
Derivation:
I. Show P -> Q Assertion
2. P Premise
3- Q I, 2, incorrect application
of M P
The application of MP is incorrect because at the time it occurred the
'Show' in line I had not been cancelled. Intuitively, we used the conclusion
that was to be derived in order to obtain one of the lines in its derivation.
Thus we begged the question.
Let us now illustrate the consequences of applying an inference rule to
a boxed line.
Argument:
Snow is white. Grass is red.
Symbolization:
Q
We consider several stages in constructing a derivation corresponding
to this symbolization.

(3) Show Q Assertion

Show Q -> P Assertion
3- Q Assumption
4- P Premise

(4) Show Q
Show Q ^ P
3- Q
4- P

(5) kJriZJZV

3, incorrect application of R
Section 4 Fallacies . 33

The partial derivations (3) and (4) are correctly constructed, but (5)
involves an unwarranted use of assumptions. Although the assumption
made in line 3 is legitimate, an illegitimate consequence has been drawn
from it; for line 3 is boxed in stage (4) and hence no longer available for
further inferences.
The derivation (5) can be obtained by another fallacious procedure.
From (3) we can proceed legitimately not only to (4), but also to the follow­
ing partial derivation:

(4') 1. Show Q
2. Show Q
3* Q
4- P
5- Q 3.R
Then we might apply clause (6) of the directions for constructing deriva­
tions, ignoring the injunction that when an occurrence of 'Show' is can­
celled, all subsequent lines must be boxed:

(4") 1 . Show Q
2. Show Q
3- Q
4- P
5- Q 3>R
From this (incorrect) partial derivation, we might then proceed legitimately
to (5). Since line 5 depends on an assumption, it is legitimately available
only as long as the assumption is; thus the failure to box line 5 along with
lines 3 and 4 in passing from (4') to (4") can be construed, like the previous
fallacy, as an unwarranted use of an assumption.
In boxing and cancelling by clause (6), the lines to be boxed must be
free of uncancelled 'Show', Let us consider two examples in which this
restriction is violated.
Argument:
Snow is white. Grass is red.
Symbolization :
Q
Derivation:
(6) I. Show- Q Assertion
2. P Premise
3- Show Q Assertion
34 . ‘Not* and ‘If* Chapter I

Derivation:

(7) I. Show Q Assertion

2. Show Q- Assertion
3- Q Assumption

In the derivation (6) we have begged the question, and in (7) we have
made unwarranted use of an assumption; in both cases the lines boxed
contain uncancelled 'Show\
There is another restriction imposed in clause (6). In each of our three
forms of derivation certain lines are crucial, and these must occur unboxed.
In direct derivation, for instance, the assertion that is to be established
must occur in some unboxed line following its initial statement. Let us
consider a case in which the crucial line is boxed.
Argument :
Snow is white. Grass is red.
Symbolization:
P .\ Q
We consider two stages in constructing a corresponding derivation.

(8 ) I . Show Q Assertion
2 KWJftUtv I I Assertion
3- Q Assumption
4- p Premise

(9) I. Skew Q
2. Shew Q P
3- Q
4- P

The incomplete derivation (8) is correctly constructed. The passage from

(8) to (9), however, fails to satisfy the requirements of clause (6), for the
occurrence of ‘Q’ in line 3 of (8) is boxed. This passage is, moreover,
intuitively unsatisfactory; it involves an unwarranted use of an assumption.
We leave to the reader illustrations of the same sort in connection with
conditional and indirect derivation.

5, Theorems. As we stated earlier, it is possible for an argument

to have an empty sequence of premises; an example is (2) on page 13.
If such an argument is valid and is furthermore symbolic, its conclusion
is called a theorem; for example, the sentence in exercise 14 (p. 26) is a
theorem. A derivation corresponding to such an argument will contain
Section 5 Theorems , 35

no line justified by clause (2) (the clause that admits premises), and will
be called a proof. The following theorems, some of which are accompanied
by proofs, will be useful in subsequent chapters.
Ti 1. Shofo P -> P Assertion
2. Assumption

T2 1 S h e w Q (P ^ Q) Assertion
2 Assumption
3 Assertion
4 2, R

T3 1 Assertion
2 Assumption
3 Assertion
4 Assumption
5 2, 4, MP

T4 and T5 are known as principles of syllogism. Their derivations are

left to the reader.
T4 (P ->Q) ^ ( [Q _>R] ->[P ->R])
T5 (Q -> R ) ^ ( [ P ^ Q ] _>[P -^R])

T6 and T7 are called principles of distribution of over

T6 I. ,SAow(P ->[Q ->R])
([P -> Q ]^ [P ^ R ]) Assertion
Assumption
Assertion
Assumption
Assertion
Assumption
4, 6, MP
2, 6, MP
7> 8> MP

T7 ([P ^ Q] -> [P ^ R]) -> (P -> [Q R])

The principle of commutation:
T8 I. Show (P “> [Q -> R]) ^
(Q ^ [P ^ R ]) Assertion
36 . ’Not’ and ‘If’ Chapter 1

Assumption
Assertion
Assumption
Assertion
Assumption
7- 2, 6, M P
8. 4, 7, M P

T9 (P ^ [P -> Q]) ^ (P ^ Q)

T io I. 5Ae«>-([P - > Q ] - ^ Q ) - >

([Q ^P]->P) Assertion

2 [P ^ Q] Q Assumption
3 Shot» [Q -> P] -> P Assertion
4 Q -> P Assumption
5 Sho^jo P Assertion
6 -P Assumption
7 -Q 4. 6, M T
8 -[P ^Q ] 2, 7, M T
9 Show P -> Q Assertion
10 Assumption
II 6, R

T i i and T i 2 are the two laws of double negation,

T ii I . Show ^ ^ P -> P Assertion
2. Assumption
3- 2, DN

T l2 P ^

T 1 3 - T 1 6 are known as principles of transposition. We shall henceforth

omit the annotations ‘Assertion* and ‘Assumption*.

T 13 1 . Show (P -> Q) (~ Q ~ P)
2.

2, 4, M T

T 14 (P ^~Q )->(Q P)
Section 6 Historical Remarks . 37

T 15 (~ P -> Q )-> (-Q ^P)

T 16 ( ~ P -> ~Q) ^ ( Q ->P)
T 17 1. Show P -> ( ~ P Q)
2.

2 ,R

T 18 P -^(P ->Q )
T 19 and T20 are known as laws of reductio ad absurdum.
T 19 Show ( ^ P -> P) ^ P

2, 4, MP

T20 (P - P ) -> - P
According to T 2 1 and T22, the denial of a conditional leads to the affir­
mation of its antecedent and the denial of its consequent.
T21 ^ (P Q) P
T22 ^ ( P - > Q ) - > '^ Q
T23 is known as Peirce's law (after the nineteenth-century American
philosopher C. S. Peirce) and is already familiar to the reader who has
solved exercise 14 (p. 26).
T23 ([P ->Q] ->P) -> P

EXERCISES
26. Prove T4, T7, T15.
27. Prove T18, T20, T 21, T22.
In solving exercises 26 and 27, the reader will again find useful the
suggestions made on page 26.

6. * Historical remarks. The logic of ‘ ^ ’ and ‘ w^as first studied

by the Stoics in the fourth and third centuries B.c. (see Lukasiewicz [2]
and Mates [i]). Its first complete formalization (in a technical sense)
occurs in Frege [1]. An extensive investigation into this branch of logic
is reported in Lukasiewicz and Tarski [1].
38 . ‘Not’ and ‘If’ Chapter I
The treatment set forth in the present chapter differs in a marked way
from that of Frege and Lukasiewicz. The earlier systems relied on only
one form of derivation—direct derivation. The simplicity thus obtained
was secured at the expense of postulating certain theorems as axioms.
The idea of dispensing with axioms in favor of conditional derivation
stems from Gentzen [i] and Jaskowski [i] and depends on a result obtained
independently by Herbrand and Tarski—the so-called dediictioji theorem
(see Herbrand [i], [2], and Tarski [i]). Systems of the later variety are
said to employ natural deduction and, as this designation indicates, are
intended to reflect intuitive forms of reasoning.
Other systems of natural deduction may be found in Quine [3], Copi
[i], Suppes [i], and Hilbert and Ackermann [4].
Our system could be simplified at the expense of making certain deriva­
tions unintuitive. In particular, indirect derivation is theoretically super­
fluous; further, we could replace our four rules of inference by the follow­
ing two:
Modus ponens:

A variant of modus toUens:

4^ ___
4>
 Chapter II
‘A N D ’, ‘ O R ’, ‘ I F AND ONLY I F ’

!• Symbols and sentences. Let us add to our language three new

symbols, ‘ a’, ‘ v’, and symbolizing respectively ‘and*, ‘or*, and ‘if
and only if*. Each of them is used, like ‘ ^*, to combine two sentences, say,
(1) Socrates is snub-nosed
and

(2) Socrates is bald,

into a new sentence. Thus we obtain

(3) (Socrates is snub-nosed a Socrates is bald)

(4) (Socrates is snub-nosed v Socrates is bald)

and

(5) (Socrates is snub-nosed ^ Socrates is bald)

which are read

Socrates is snub-nosed and Socrates is bald,
Socrates is snub-nosed or Socrates is bald.
and
Socrates is snub-nosed if and only if Socrates is bald.
(3) is called the conjunctioti of (i) and (2), which are its conjuncts, (4) is
the disjunction of (i) and (2), which are its disjuncts, (5) is the biconditional
formed from (i) and (2), which are its constituents.
The symbols ‘ ^^*, ‘ ->*, ‘ a*,‘ v*, and ‘<->* are called sentential connectives.
The corresponding phrases ‘it is not the case that’, ‘i f . . . , then*, ‘and*,
‘or*, and ‘if and only if* are czWtA phrases o j connection.
The language we now consider is that of chapter I, supplemented by
‘ a*, ‘ v*, and Thus the sentences of our language fall into the following
categories:
40 . ‘And*, ‘Or*, ‘If and Only If* Chapter II

(1) declarative sentences of English,

(2) sentence letters,
(3) sentences correctly constructed from sentences of categories (i)
and (2) by means of sentential connectives and parentheses.
To be more explicit, the class of sentences can be exhaustively charac­
terized as follows:

(j) A ll declarative sentences of English are sentences.

{2) Sentence letters are sentences.
(3) U ^ sentencesy then so are

0) ,
A i/i) ,
V i/j) ,

and

The symbolic sentences of our language are those sentences which are con­
structed exclusively from sentence letters, parentheses, and sentential
connectives. More precisely, the class of symbolic sentences is exhaustively
characterized by clauses (2) and (3) above, reading everywhere ‘symbolic
sentences’ for ‘sentences’ .
Notice that ‘ a’, ‘ v’, and like bring with them a pair of paren­
theses. The ambiguity which would otherwise result is illustrated by the
invitation
Bring your wife or come stag and have a good time.
As in chapter I, we shall usually omit the outermost parentheses of a
sentence and sometimes replace parentheses by brackets. In addition,
and will be regarded as marking a greater break than the other
connectives. Consequently, the parentheses accompanying ‘ a* and ‘v’
may be omitted in certain contexts, as in the following examples. The
sentence
(P A Q) -> (R V S)
becomes
P AQ R VS
and
( P v Q )<->(R a S)
becomes
P y Q ^ R a S
Section I Symbols and Sentences . 41

Thus we arrive informally at a larger class of sentences than that pro­

vided by clauses ( i) - ( 3 ) . Sentences of the smaller class may, in those
few cases when it is necessary to draw a distinction, be called setttences
in the official sense.

EXERCISES
Which of the following are sentences in the official sense? Exercises
I and 2 are solved for illustration.

1. - ( - P - ^ ( Q v R ) )
According to clauses (2) and (3) of the characterization of sentences
(p. 40), ‘(Q V R)’ is a sentence, and is a sentence. By clause (3),
then, ‘('^P -> (Q V R))’ is a sentence; thus, by clause (3) again, No. i
is a sentence in the official sense.

2. (P v (Q -^ R ) a - P )
Because the characterization of sentences given by clauses (i) -(3)
is stipulated to be exhaustive, and clauses (i) and (2), together with
the first and fifth parts of clause (3), are inapplicable, No. 2, if a
sentence, must be either a disjunction, a conditional, or a conjunction
of two sentences. Thus either

and
(1) (Q ->R) A ~ P
are sentences, or
(2) P v(Q
and
(3) R) A ~P

are sentences, or

(4) P V (Q -> R)
and

are sentences. But (i), (2), and (4) (and, incidentally, (3)) are not senten­
ces in the official sense, as the reader can easily verify. Thus No. 2
is not a sentence.

3. ((P <-> Q ) <-> ((P A Q ) V ( ~ P A ~ Q )))

4. ~ ( ( P v Q ^ R ) - > ( ( P - ^ R ) a (Q->R)))
3, .^ ( '^ ( '^ '^ P v Q ) V R ) — (P <—►(Q > R ))
42 . *And\ ‘Or*, ‘If and Only If* Chapter II

From each of the following sentences obtain a sentence in the official

sense by restoring parentheses omitted by informal conventions.
Exercise 6 is solved for illustration.

6. (P A Q P VQ) ^ P V(Q A R)
This sentence becomes
(((P a Q )-> (P v Q ) ) ^ ( P v (Q a R))) .

7 . Q v R - > ( P a R->(Q<->R v P))

8. ( P a Q) v ( - P A -Q )->((P<->Q)4->(Q^P))
9. (P->Q vR)v(Q vR- >P) ^( QvR<- >P)
10. - ( P v ( Q a R)->( ( P v Q) a ( P v R ) ^ P a Q))

2. Translation and sym bolization. The process of literal translation

into English on the basis of a given scheme of abbreviation begins as before
with a symbolic sentence and if successful ends with a sentence of English.
The process consists of the following steps:

(z*) Restore any parentheses that may have disappeared as a result of

the informal conventions of the last section,
{ii) Replace sentence letters by English sentefices in accordance with the
scheme of abbreviation,
{Hi) Eliminate sentential connectives in favor of the corresponding phrases
of connection^ preserving all parentheses.

As before, we call an English sentence a free translation (or simply a

translation) of a symbolic sentence (f> on the basis of a given scheme of
abbreviation if it is a stylistic variant of the literal English translation of <f>
based on that scheme.
Let us consider, for example, the scheme of abbreviation
P Alfred dances with Alice
Q Alfred dances with Mary
R Alfred will enjoy the party
and the symbolic sentence

(I) P a Q -> R
In step (i) of the process of literal translation into English, this sentence
becomes
((P a Q )-> R )
in step (ii)
((Alfred dances with Alice a Alfred dances with Mary)
Alfred will enjoy the party) ,
Section 2 Translation and Symbolization . 43

and in step (iii)

(2) (If (Alfred dances with Alice and Alfred dances with Mary),
then Alfred will enjoy the party) .

Thus (2) is the literal translation of (i) into English on the basis of the
scheme above, and the more idiomatic sentence,
(3) Alfred will enjoy the party if he dances with both Alice and
Mary,
being a stylistic variant of (2), qualifies as a free translation of (i) on the
basis of the same scheme.
As regards stylistic variance, we persist in our loose practices, giving
no exact definition but only a few examples, ‘But’, ‘although’, ‘even
though’, ‘both . . . and’, and sometimes the relative pronouns ‘who’,
‘which’, ‘that’, serve as stylistic variants of ‘and’ ; ‘unless’, ‘either . . .
or’, as stylistic variants of ‘or’ ; and ‘exactly on condition that’, and some­
times ‘just in case’, as stylistic variants of ‘if and only if’. Thus each of
Alfred slept through class, but he passed.
Even though Alfred slept through class, he passed,
Alfred, who slept through class, passed

is regarded as a stylistic variant of

Alfred slept through class and Alfred passed;

The needle on the ammeter will deflect unless the battery

is dead

as a stylistic variant of
The needle on the ammeter will deflect or the battery is
dead;
and

Alfred will be elected exactly on condition that he stand

for office.

Just in case Alfred stands for office will he be elected

as stylistic variants of
Alfred will be elected if and only if Alfred stands for office.
As a further illustration of stylistic variance, observe that the phrase
‘neither . . . nor’ is expressible by means of phrases of connection.
Neither (j> nor iJj ,
44 • ‘And’, ‘Or’, ‘If and Only If’ Chapter II

where <f> and i/j are sentences, has the stylistic variants
(It is not the case that <f> and it is not the case that 0 )
and
It is not the case that (<^ or ifj) .
It should be emphasized that the binary sentential connectives, that
is, ‘ a’, ‘ v*, may stand only between sentences (or, in later chapters,
between formulas). English usage, however, provides other contexts
for the corresponding phrases of connection, as in the sentence (3) above,
(4) Socrates is either snub-nosed or bald,
and
(5) Arcadia lies between Laconia and Achaea.
(3) and (4) have canonical stylistic variants, that is, stylistic variants
in which phrases of connection operate only on sentences. We have
seen this already in the case of (3), and (4) can be expanded into
(Socrates is snub-nosed or Socrates is bald) .
But (5) cannot be similarly treated; it is clearly not synonymous with
Arcadia lies between Laconia and Arcadia lies between
Achaea.
As in chapter I, we say that ^ is a symbolization of an English sentence
iff on the basis of a given scheme of abbreviation just in case is a symbolic
sentence which has 0 as a translation on the basis of that scheme.
To find a symbolization of a given English sentence on the basis of a
given scheme of abbreviation, the reader will find it useful to proceed
roughly as follows:

(j) Introduce phrases of connection^ accompanied by parentheses and

occurring canonically {that w, standing only before or between sentences)^
in place of their stylistic variants.
(2) Reverse the steps leading from a symbolic sentence to a literal English
translation; that iSy
{2a) replace all parts having one of the forms
it is not the case that <f> ,
{if <f>y then iji) ,
{<f> and iff) ,
l<f> or ifl) ,
{(f) if and only if iff) ,
where <f> and iff are sentences^ by
Section 2 Translation and Symbolization . 45

'^<f> I

A ,

{<{>*-*>!>)
respectively;
(26) replace English components by sentence letters in accordance with the
scheme of abbreviation;
(2c) omit parentheses and insert brackets in accordance with the informal
conventions of the preceding section.

EXERCISES
11. On the basis of the scheme of abbreviation
P : Alice will dance with Alfred
Q : Mary will dance with Alfred
R : Alfred will improve his deportment
translate the following symbolic sentence into idiomatic English:

( [ - P a - Q ] v R ) a ( R 4- > P a Q) .

Symbolize each of the following sentences on the basis of the scheme

of abbreviation that accompanies it. Exercise 12 is solved for illustra­
tion.

12. Errors will decrease in the subject’s performance just in case

neither motivation is absent nor learning has not occurred. (P : errors
will decrease in the subject’s performance; Q : motivation is absent;
R : learning has occurred)
In step (i) of the informal procedure for symbolizing English
sentences (pp. 44 - 45), we may transform No. 12 into
(Errors will decrease in the subject’s performance if and
only if it is not the case that (motivation is absent or it is
not the case that learning has occurred)) .
In step (2a) the sentence becomes
(Errors will decrease in the subject’s performance
^ (motivation is absent v learning has occurred)) ;
in step (2b),
(P<-> -(Q V -R ))
and in step (2c),
P ^ (Q V R)
46 . ‘And*, ‘Or’, ‘If and Only If’ Chapter II

13. Assuming either that logic is difficult or that the text is not
readable, Alfred will pass only if he concentrates. (P : logic is dif­
ficult; Q : the text is readable; R : Alfred will pass; S : Alfred
concentrates)
14. Unless logic is difficult, Alfred will pass if he concentrates.
(P : logic is difficult; R : Alfred will pass; S : Alfred concentrates)
15. Mary will arrive at 10:30 A.M. unless the plane is late. (P :
Mary will arrive at 10:30 a.m .; Q : the plane will be late)
16. Assuming that the professor is a Communist, he will sign
the loyalty oath; but if he is an idealist, he will neither sign the loyalty
oath nor speak to those who do. (P : the professor is a Communist;
Q : the professor will sign the loyalty oath; R : the professor is an
idealist; S : the professor will speak to those who sign the loyalty
oath)
17. If Alfred and Mary are playing dice together, it is the first
throw of the game, and Mary is throwing the dice, then she wins
the game on the first throw if and only if she throws 7 or i i. (P :
Alfred is playing dice; Q : Mary is playing dice; R : Alfred and
Mary are playing dice together; S : it is the first throw of the game;
T : Mary is throwing the dice; U : Mary wins on the first throw;
V : Mary throws 7 or i i ; W : Mary throws 7; X : Mary throws ii)
18. If the world is a progressively realized community of inter­
pretation, then either quadruplicity will drink procrastination or,
provided that the Nothing negates, boredom will ensue seldom more
often than frequently. (P : the world is a progressively realized
community of interpretation; Q: quadruplicity will drink procrastina­
tion; R : the Nothing negates; S : boredom will ensue seldom more
often than frequently)

The following sentences are ambiguous in the sense that the place­
ment of parentheses in their symbolizations is not uniquely deter­
mined. Give all plausible symbolizations of each on the basis of the
given scheme of abbreviation. Exercise 19 is solved for illustration.

19. Errors will occur in the subject’s performance if and only if

motivation is absent or learning has not taken place.
Given the scheme of abbreviation
P : errors will occur in the subject’s performance
Q : motivation is absent
R : learning has taken place ,
No. 19 becomes in step (i) either
(Errors will occur in the subject’s performance if and only
if (motivation is absent or it is not the case that learning
has taken place))
or
Section 3 Inference Rules; Theorems with Unabbreviated Proofs . 47

(Errors will occur in the subject’s performance if and

only if motivation is absent) or it is not the case that learn­
ing has taken place) ;
and in step (2) it correspondingly becomes either
P<->Q V - R
or
(P Q) V ^ R
20. If either a war or a depression occurs then neither science
nor music and literature will flourish unless the government supports
research and provides patronage for artists. (P : a war occurs; Q : a
depression occurs; R : science will flourish; S : music will flourish;
T : literature will flourish ; U : the government will support research ;
V : the government will provide patronage for artists)
21. If Mary belongs to a sorority then she will be graduated from
college only if she resists temptation, provided that she is not attractive
or intelligent. (P : Mary belongs to a sorority; Q : Mary will be
graduated from college; R : Mary resists temptation; S : Mary
is attractive; T : Mary is intelligent)

3, Inference ru les; theorem s with unabbreviated proofs« The

sentential calculus is that branch of logic which essentially involves the
sentential connectives. For a complete formulation of the sentential
calculus, we must add to the apparatus of chapter I the following inference
rules for the new sentential connectives.
Simplification (S), in two forms: A A

4>
Adjunction (Adj): ^

(^A ^)
Addition (Add), in two forms:

M o d u s t o l l e n d o p o n e n s (M TP), in two forms; (^ v ^ )

tjf

Biconditional-conditional (BC), in two forms: ■<p)

Conditional-biconditional (CB) : i< f>

48 . ‘And*, ‘Or*, ‘If and Only If* Chapter II

Thus one symbolic sentence is said to follow from another by simplification

just in case the former is a conjunct of the latter; a symbolic sentence follows
by adjunction from two others just in case it is their conjunction; and so on.
When

or

is inferred from <f> by Add, the symbolic sentence is called the added
disjunct.
The new rules are exemplified by the following arguments.
Simplification:
( - P a Q)
(P A (Q R)) /. (Q R)
Adjunction:
Q (-P a Q)
Addition:
P (P V ~Q ) (Here is the added disjunct.)
~Q (P V ^Q ) (Here T ' is the added disjunct.)
Modus toUendo ponens:
(P V ~ Q) . ~P
(P V Q) . Q P
Biconditional-conditional:
(P<->(Q vR)) ( P ( Q V R))
(P<->(Q vR )) ((QV R ) -> P)
Conditional-biconditional:
(P (Q V R)) . ((Q V R) -> P) /. (P<-> (Q V R))
We have at our disposal several conventions for dropping parentheses,
but they must be used with caution. In particular, when applying inference
rules, we must mentally restore omitted parentheses. For example,

(1) P aQ -> R
Q -^ R
might seem to be a case of simplification, and
(2) P
PvQ
Section 3 Inference Rules; Theorems with Unabbreviated Proofs . 49

a case of addition. But when parentheses are restored according to our

conventions, (i) and (2) become respectively
((P a Q )-> R )
(Q ^ R )
and

((P v Q )^ R )

which clearly do not constitute applications of our rules. Indeed, the

inferences (i) and (2) should not fall under our rules. For (i) is comparable
to the inference from
If San Francisco is larger than New York and New York is
larger than Los Angeles, then San Francisco is larger than
Los Angeles
to
If New York is larger than Los Angeles then San Francisco is
larger than Los Angeles,
and (2) to the inference from
Socrates is bald
to
If either Socrates is bald or Socrates is snub-nosed, then
Socrates is unmarried.
‘Or’ has two senses.
<f> or ijj
may mean either
either <f> or ijj, but not both
or
either </>, or i/r, or both
We select the second sense. This choice is reflected in our adoption of
the rule Add; for if the first sense were selected, the inference
Socrates is bald, *. Socrates is bald or Socrates is snub-
nosed
would lead from truth to falsehood. Though we shall use ‘v’ in the second
sense of ‘or’, it is clear that the first sense can still be expressed by a
combination of ‘v’, and ‘ a\
50 . ‘And,’ ‘O r’, ‘I f and Only I f ’ Chapter II

‘ If and only if’ is a composite of ‘if’ and ‘only if’.

4> if and only if ifi
asserts

and
^ only if Ip .
This is reflected in our two rules BC and CB, by which the biconditional
(f) <
—>tjj
is related to the conditional
Ip -><i>
and its converse.
<p ->ip
The definitions of an argument, a symbolic argument, and an English
argument are carried over intact from chapter I (pp. 13 and 24). The
directions for constructing a derivation from given symbolic premises
(pp. 20 - 21), and the definitions of a complete derivation (p. 23),
of derivability (p. 23), of a valid symbolic argument (p. 24), of a valid
English argumejit (p. 25), of a theorem (p. 34), and of proof (p. 35) also remain
unchanged. The interpretation, however, of the phrase ‘an inference
rule* (which occurs in clause (5) of the directions for constructing a deriva­
tion) is extended so as to include our new inference rules (S, Adj, Add,
M TP, BC, and CB) as well as MP, M T , DN, and R.
It is convenient, in developing the full sentential calculus, to depart
from the order of presentation of chapter I. We begin, not with derivations
involving premises, but with proofs of theorems. The first group of
theorems primarily concern the connectives ‘ and ‘ a *. T24 is
the commutative law for ‘ a *.
T24 1. Show P A Q<-> Q A P
2. Show P A Q Q A P
3* P aQ
4- Q 3» S
5* P 3. S
6. Q aP 4. 5. Adj
Show Q A P -> P A Q
Similar
8. P a Q<->Q a P 2, 7, CB
Section 3 In feren ce R u les; Theorem s with U nabbreviated Proofs . 5 1

T25 is the associative law for

T2S P a ( Q a R)<->(P a Q ) a R

Here, as in the case of T24, we prove explicitly only one conditional.

The converse can be proved in a similar way, and the biconditional will
then follow by CB.
Shew P A (Q A R) -> (P A Q) A R
P Л (Q A R)
P 2, s
QaR 2,s
Q 4, s
P aQ 3. 5. Adj
R 4, S
(P A Q) A R 6, 7, Adj

A new informal convention for the omission of parentheses will prove

convenient. In a repeated conjunction in which all terms are associated
to the left, the internal parentheses may be omitted. For instance,

(P A Q) A R
becomes
P A QAR
and
([P A Q] A R) A S
becomes
P A QAR A S

However, the parentheses of

P A (Q A R)

may not be omitted. T25 may now be formulated as

P a ( Q a R)<-^P a Q a R
T26, like T4 and T 5, is a principle of syllogism.
T26 (P Q) A (Q R) (P R)
T27 is known as the law of exportation.
T27 I. S h e w {? A Q -> R) (P -> [Q R])
2. I .S A e w (P A Q ^ R )-> (P -^ [Q ^ R ]) |
52 . ‘And’, ‘Or’, ‘If and Only If’ Chapter II

P a Q ->R
Shew- P [Q -> R]

Show Q R
Q
P aQ 5. 7. Adj
R 3, 8, MP

10 Shew {P ( P aQ ^ R )
11
12
13
14 13. s
15 I I , 14, MP
16 13. S
17 15, 16, MP

18 (Pa Q [Q ^ R ]) 2, 10, CB

T28 (P A Q R ) < - ^ (P A ~ R ->■ Q)

T29 is the principle of distribution of over ‘ a’ .
T29 (P->Q a R)<->(P ^ Q ) a ( P - > R )

T30 and T 3 1 are factor principles.

T30 (P ^ Q) ^ (R A P R A Q)
T 31 (P->Q )->(Pa R - > Q a R)
T 32 is Leibniz’ praeclarum theorema.
T 32
Section 3 Inference Rules; Theorems with Unabbreviated Proofs . 53

T33 is a principle of dilemma.

T 33 Shem {^ - > Q ) a ( ~ P -> Q) ^ Q

2, S
4. S. M T
2, S
4. 7. M T

T34, like T 19 and T20, is a law of reductio ad absurdum,

T34 (P->Q )a (P -> ^ Q )-> ^ P
T35 I. Shew ( ^ P -> R) A (Q -> R)<-> ([P Q R)
2. Show ( P R) A (Q ^ R) ->
([P -> Q ]^ R )
3-
4-
5-
6.
7-
8. 3. 5
9- 7, 8, M T
10. 9. 5. M T
11. 3. 5
12. 10 , II, MP

13- Shem ([P Q] -> R) ->

__________ ( ~ P - > R ) a ( Q - > R )
14. [P -> Q] ^ R
15- Show ~ P R
16.
17-
18.
19. 16, R
20. 14, 17, MP
21. Show Q R
22. Q
23- S how P -> Q
54 . ‘And*, ‘Or*, ‘If and Only If* Chapter II

24. Q 22, R
25- R 14, 23, MP
26. (~ P ^ R ) a (Q ^R) 15, 21, Adj
27- ( ~ P ^ R) A (Q -> R)<->([P ^ Q ]^ R) 2, 13, CB

T36 is the law of contradiction.

T36 I. Show ~ (P A ~ P )
2.
2, S
2, s

T 37 shows how to express by means of and ‘ a’, andT38 how to

express V by means of and T39 and T40 provide alternative
expressions for the negation of a conjunction and of a conditional.
T 37 I . iS A e w -(P Q ) ~ (P A ~ Q)
2.

5. 5
3. 6, MP
5. 5

10
11
12
13
H
15 12, 14, Adj
16 10, R

17- 2, 9, CB

T38 P a Q<->

T39 -(P a Q)<-^(P -Q )

T40 ^ (P “> Q) <—> P A Q
T 41 is the law of indempotence for ‘ a*, and T 4 2 - T 4 4 will be cited in
chapter V.
Section 3 In feren ce R u les; T h eo rem s with U nabbreviated Proofs . 55

T41 P P AP
T42 P a ^Q —
> ^(P -> Q)
T43 -> -^(P A Q)
T44 0 ^0)

EXERCISES
22. Prove T28, T29, T34.
23. Prove T38, T44.

In solving these exercises, which essentially involve only

and ‘ a^ the reader will find the following informal suggestions, some
of which appeared in chapter I, helpful (but not infallible).

(j) To derive a conditional^ use conditional derivation,

{2) To derive a conjunction^ derive first both conjuncts and then use
Adj,
(j) To derive a biconditional, derive first the two corresponding con­
ditionals and then use CB.
(4) To derive anything else^ use indirect derivation unless another
procedure is immediately obvious.
(5) Whenever a sentence follows from antecedent lines by MP,
MTy Sy MTPy or BCy enter that sentence as a line.
(6) When using indirect derivation^ determine whether any of the
antecedent lines is the negation of a conditional; if so, attempt to derive
that conditional,

T40 is proved for illustration. We begin the proof by writing

'Show' followed by the sentence to be proved. This sentence is a
biconditional; thus, following (3) and (i) of the foregoing suggestions,
we begin a conditional derivation of one of the corresponding con­
ditionals.
1. Show (P -> Q )< ^ P a Q
2, Show '(P ->Q )^P A 'Q
3- ~ ( P - Q)
In order to complete the subsidiary conditional derivation we must
derive a conjunction; thus, following suggestions (2) and (4), we begin
an indirect derivation of one of its conjuncts.
1. Show (P->Q)- P a 'Q
2. Show />/(P->Q ) P a
3. ~ (P ^ Q )
4. Show P
5- ~P
56 . ‘And’, ‘Or’, ‘If and Only If’ Chapter II

Consideration of the lines now before us and suggestion (6) lead us to

begin next a conditional derivation of the sentence whose negation
occurs in line 3, and this can be completed after an application of rule
R.
Show ^ (P ■ Q)' P a Q
Show ^ (P - Q) P a
~ (P ^ Q )
Show P
~P
6. Show-P
7-
8. s .R
We now complete the subsidiary derivation of T* and, returning to
suggestions (2) and (4), begin an indirect derivation of which
we can complete by employing again suggestion (6).
1. Show (p- Q)^ P a Q
2. Show (p- Q) P a Q
3 ~ (P -> Q )
p
4

S .R
3. R

II, R
3. R
Now, following through suggestion (2), we employ adjunction in order
to complete the subsidiary derivation started in line 2.
1 . Show ^(P->Q )<->'P a
2. Show- '^ ( P - > Q ) - > P a
^ (P -> Q )
iShov^ P

5. R
3. R
Section 4 Abbreviated Derivations . 57

10.
11.
12.
13- II, R
H- 3. R
IS- 4. 10. Adj

To complete the main derivation we derive next the converse of the

conditional that occurs in line 2 and then, following through suggestion
(3), employ CB.
I• /(P —> Q) ► P A ~ Q
2. 5 A<9w ~ ( P ^ Q ) - ^ P a ~Q
3-
4-
5-
6.
7-
8. s .R
9- 3.R
10.
11.
12.
13- II, R
14- 3.R
15- 4,10, Adj
16.
17-
18.
19.
20. 17, S
21. 19, 20, MP
22. 17, S

23- ' (P —>■ Q) P A~ Q 2, 16, CB

4. Abbreviated derivations. It will be useful, from time to time,

to append to the directions for constructing a derivation what we shall
call abbreviatory clauses; we shall thus arrive at a more comprehensive
class of derivations, called abbreviated derivations, which will be con­
structed on the basis of clauses (i) - (6) along with the new clauses.
58 . ‘And’, ‘Or’, ‘If and Only If’ Chapter II

Abbreviatory clauses must satisfy two conditions. In the first place, they
must be theoretically dispensable; that is to say, whenever a conclusion
can be derived from a class of premises by means of an abbreviated deriva­
tion, it must also be derivable from the same premises by means of some
unabbreviated derivation (a derivation constructed on the basis of clauses
( i ) - ( 6 ) alone). In the second place, abbreviated derivations must share
the important characteristic of unabbreviated derivations pointed out
earlier: there must be an automatic procedure for checking the correctness
of an abbreviated derivation (at least when membership in the class of
premises is automatically decidable).
For the sentential calculus we shall adopt only two abbreviatory clauses,
(7) and (8), which, when adjoined to clauses ( i) - ( 6 ) , will constitute the
directions for constructing an abbreviated derivation^ or, as we shall
henceforth say, a derivation. The fact that clauses (7) and (8) satisfy the
two conditions mentioned above will be relatively obvious but will not
be established in a formal way.
For clause (7) we must introduce a new notion. By an instance of a
symbolic sentence <f> we shall understand any symbolic sentence obtained
from (f) by replacing sentence letters uniformly by symbolic sentences.
(The replacement is uniform just in case all occurrences of a sentence letter
are replaced by the same sentence.)
For example, T2,
Q -> (P ^ Q )
has as an instance
Q - > ( - R ->Q)
In this case, we have replaced T ’ by and ‘Q’ by ‘Q’ itself. We may
indicate this replacement diagrammatically as follows:
Q P
R
Other instances of T2 are
(1) ^R (Q R) ,
(2) P ^ ([R -> T ]-> P ) ,
(3) Q VP ^ ([P -> Q] -> Q VP)
(4) Q ^ (P -> Q ) ,
(5) P ^ (Q ^ P ) ,
obtained by the respective replacements
(!') _Q ____P
~R Q ’
Section 4 Abbreviated Derivations . 59

(2') Q
R -> T
(3') P
Q vP P
(4') Q P
Q P
(S') Q P
______
P Q
As a further example, T26,
(P->Q )a(Q ->R)->(P->R) ,

has as an instance
([S -> T ] - > T ) a (T -> S ) ->([S -> T ] -> S)
obtained by the replacement
P Q R

The relation is an instance of ifi is intended to apply only to symbolic

sentences. (This accords with our previous decision to apply deductive
procedures only to symbolic sentences.)
It is because of clause (7) that theorems are important. Having proved
a theorem, we may then use it in a later derivation without repeating its
proof. Indeed, we may use as well any of its instances, for their proofs
could be obtained by a simple procedure of replacement from the original
proof.

(7) I f (f> is an instance of a theorem already proved^ then <l> may occur as
a line. {Annotation: the number of the theorem of which (f) is an instance^
sometimes together with a diagrammatic indication of the replacement in-
volved.

For example, clause (7) provides the justification for lines 3 and 5
in the following abbreviated derivation.
I. Skew-{[? -> Q] -> Q) A (Q -> P) ^ P
2. ([P ^ Q ] ->Q) a (Q >P)
3- ([P ^ Q ] ^ Q ) a (Q -■>P)-^ Q R'
T 26|i ^
([P ->Q]->P) Ip ->Q Q p-
4- [P ->Q] ->p 2, 3. MP
5- ([P -^Q] -^P) ->P T 23
6. P 4. 5. MP
6o . ‘And’, ‘Or’, ‘If and Only If’ Chapter II

(The reader should compare this brief proof with the unabbreviated
derivation, given on page 26 ff., corresponding to exercise 12 of chapter I.)
It is often convenient to compress several steps into one, omitting some
lines that an unabbreviated derivation would require. Such compression
will be allowed only when no clauses other than (2) (premises), (5) (in­
ference rules), and (7) (instances of previously proved theorems) are
involved, and is legitimized by clause (8).

(5) A sentence may occur as a line if it follows from antecedent lines by

a succession of stepSy and each intermediate step can be justified by one of
clauses (2), (5), or (7). (The annotation should determine the succession of
steps leading to the line in question. This can be done by indicatingy in order
of applicationy the antecedent linesy the premisesy the inference rulesy and the
previously proved theorems employed. In additiony when the rule Add is
usedy the added disjunct should be indicated; and when an instance of a
previously proved theorem is involvedy the relevant replacement should be
indicated.)

For example, clause (8) provides the justification for line 5 in the
following abbreviated derivation.
1. Shew-(? ->Q ) -> (R a P -> R a Q)
2.

4, S, 2, MP, 4,
S, Adj
The omitted lines are, in order:
i. P (4, S)
ii. Q (i, 2, MP)
iii. R (4, S)
Line 5 follows from (ii) and (iii) by Adj. Clause (8) also provides the
justification for lines 3, 4, and 5 in the derivation accompanying the
following argument:

(Q ->P) -> P P^Q Q

I. Show Q
2. ~Q
3- ~P 2, 2nd premise, M T

4- Q -> P
?)• “ P
5- P 4, ist premise, MP
Section 4 Abbreviated Derivations . 6i

An unabbreviated derivation of the conclusion from the given premises

can be obtained by inserting in the abbreviated derivation the omitted
lines, along v^ith a subsidiary derivation of the instance of the theorem
employed. The following is an example of such a derivation.
SLlii^aix
JCTTTvTtr

3 Premise
4 2, 3, M T
5
6
7
8
9 6, R

10 2, 5, MP
11 Premise
12 10, II, M P

Abbreviated derivations, unlike unabbreviated derivations, require

as an essential part their annotations—at least those annotations given
in connection with clause (8). For without an indication of omitted
steps, it would be impossible, even for the simple logical system now
under consideration, to give an automatic procedure for checking the
correctness of an abbreviated derivation.

EX ER C ISES
Corresponding to the following argument, two abbreviated deriva-
tions are given
(6) ~ P -^ R P -> Q . Q -^ R R
(7) I. ^hewR
2. ~R
3- ~Q
4- ~P
5- R
(8) I. Shew- R
2. P -^ R 2nd premise, 3rd premise, Adj, T26, MP
3- R 2, ist premise, Adj, "^33

24. Annotate the derivation (7).

25. Construct an unabbreviated derivation corresponding to (6),
using indirect derivation. (Consider (7) and its annotations.)
62 • ‘And', ‘Or*, ‘If and Only If' Chapter II

26. Construct an unabbreviated derivation corresponding to (6), using

direct derivation. (Consider (8).)

Show, by constructing derivations in which at least one step is

justified either by clause (7) or by clause (8), that the following argu­
ments are valid.
27. (p Q) R R
28. (p Q) R Q R
(The reader should attempt two-line derivations in each case.)

5. T heorem s with abbreviated proofs. Many of the remaining

derivations of this chapter are considerably simplified by the addition
of clauses (7) and (8) to the directions for constructing a derivation.
Now we consider theorems primarily concerned with ‘v’ as well as
and ‘ a ’. T45 shows how to express ‘v’ in terms of and
and T46 shows how to express in terms of and ‘v’ . Clause
(8) is used in the proof of T45.

T45 1 . Show-P v Q < - > ( ~ P ^ Q )

2.
3
4
5-
6. 5. 3. M T P

7
8
9
10
11
12
13 12, Add
14 10, R

15 8, I I, MP, Add

16. 2, 7, CB

T46 (P -> Q )^ ~ P v Q
T47 is the law of idempotence for ‘v\
T47 I. P V P
2. I S h ew V P VP
Section 5 Theorems with Abbreviated Proofs • 63

3,A d d

6, 8, M T P

2, 5. CB

T48 and T49, like T 33, are principles of dilemma.

T48 I. 5 A e w ( P v Q ) A ( P - > R ) a (Q -^ S ) - > R v S
2. (P V Q) A (P -> R) A (Q S)
3- Shov> ~ R S
4- ~R
5- (P V Q) A (P R) 2, S
6. Q 2, s
7- S 5» S, 4, M T , 5, S,
M TP, 6, M P
Rv S

3, M P

T49 I. S h 9 w { V V Q) A (P -> R) A (Q ^ R) -> R

2.1 ( P v Q ) a (P ^ R ) a ( Q -> R ) I ^ /P Q R S \
\P Q R R /
2, MP
3- L R --------- ---------------------------1 T
T 44 77(Q
i ) .B
. B Cc ., M P
T50 is a principle of composition.

T50 I. Show (P -> R) A (Q -> 1^))4-^(P vQ ->R )

2 Shaw- (P R) A (Q ^ R) ->
____________ (P v Q ->R )
3- (P R) A (Q ^ R)
4- Show P V Q -> R
5- P vQ
6. R 5, 3, Adj, T49, MP

Show (P V Q ^ R)
(P -> R) A (Q -> R)
 64 . ‘And*, ‘Or*, ‘If and Only If* Chapter II

8.
9-
10.
11. lo, Add, 8, MP
12 .

13-
H- 13, Add, 8, MP
15* 9, 12, Adj
1 6. (P - > R ) a (Q - > R ) ^
______ (P v Q -> R ) 2, 7, CB

T51 (P V Q) A (P ^ R) A ( ~ P A Q ^ R) -> R
T52 (P ^ R ) a ( ~ P a Q ^ R ) < - > ( P v Q - > R )

The following patterns of inference, each corresponding to a theorem

proved by now, will appear frequently in what follows:

/(0 4>
( 2) ^ <f> —
> ijj
/
(i), (2), Adj, T 33 MP
(3) <f> V if /

(4) <f> -^X

(5) ^X
(3), (4), Adj. (5), Adj, T49 MP
(6 ) <f> ~^X
(7) -^ x
^^— / /P Q R\
(9/(7). Adj, T 5 0 - — ------ ,B C ,M P
\9 'f* X '
We shall refer jointly to these three patterns arinferences by the^im ved.
rule o j separation of cases^ and we shall employf^QC Ito indicate their use.
More explicitly, we^^hill abbreviate the lengthy annotations above by ‘SC ’,
together with the numbers of the antecedent lines involved. Thus, for
example, the second pattern becomes:

(3) <f> y ifj

(4) 4 -^ x
(5 ) 'I' -^X
(3), (4), (5), SC

Illustrations of the first and third forms of SC are provided by the following
derivations.
Section 5 Theorems with Abbreviated Proofs • 65

Argument:
P -> R R -> Q Q
Derivation:

2nd premise, 3rd premise,

/ P Q R\

I St premise, 2, SC

Argument:
~ P v Q ->(P ->Q)
Derivation:
I . Shot» ~ P V Q ->■ (P -> Q)
2. ~ P ^ (P -> Q ) T 18
3- Q -> (P -> Q ) T2
4- ~ P v Q ^ ( P ->Q) 2, 3, SC
Another useful pattern of inference is called the derived rule of conditional-
disjunction, and is justified by T 4 5 :

(8) ^__________
] (8), T4S, BC, MP
We shall henceforth replace the annotation ‘T45, BC, M P’ by the abbrevia­
tion ‘CD ’ . An illustration is provided by the following proof.

I. Shot» (P ->■ Q
2. P -^Q
3- Show ^ P
4- P
5- Q 4, DN, 2, MP
6. ~ P VQ 3, CD

T53 is the commutative law for ‘v’.

T53 P V Q <-> Q V P
The associative law for ‘v’ :
T54 P V (Q V R) (P V Q) V R
As in a conjunction, internal parentheses may be omitted from a
repeated disjunction in which all terms are associated to the left. Thus
66 , ‘And’, ‘Or’, ‘If and Only If’ Chapter U

T 54 becomes
P v ( Q v R ) .< - > P v Q v R .
T55 is the distributive law for over ‘v’.
T55 (P ^ Q v R ) < - > ( P - > Q ) v ( P ^ R )
T56 (P ^ Q) ^ (R V P R V Q)
TS7 (P Q) ^ (P V R Q V R)
T58 (P ^ Q )v (Q ^ R )
The law of the excluded middle:
T59 P V ~P
T60, like T50, is a principle of composition.
T60 ( P - ^ R ) v(Q ^ R ) < - > ( P a Q - > R )
T 61 and T62 are laws of distribution.
T 61 P a (Q v R)<-».(P a Q ) v ( P a R)
T62 P v (Q a R )<-> (P v Q ) a ( P v R)
T63 - T66 are known as De Morgan's laws, after the nineteenth-century
logician, Augustus De Morgan; T67, a corollary of T66, will play an
important role in chapter V.
T63 P a Q ^ ~Q )
T64 P V Q<-> ^ ( ^ P A
T65 (P A Q) <-> ~ P v - Q
T66 (P V Q) 4-)^ -~ P A ~ Q
T67 ^P A Q -> - ( P vQ )
T68 and T69 provide redundant but useful forms of expression.
T68 P < -> (P a Q ) v ( P a ~Q )
T69 P t-» -(P v Q )A (P v ~Q )
We come now to theorems which primarily concern All but T78,
T80, T94, and T95 (which are included because of their unintuitive
character) will be found extremely useful in what follows.
T70 Q ^ (P a Q ^P)
Tyi ~ Q ^ (P v Q < -)-P )
T 72 (P ->■ Q) •<-+ (P A Q <-> P)
Section 5 T h eo rem s with Abbrevis

T73 (P - > Q ) ^ ( P v Q < - ^ Q )

T74 (P ^ Q) A P ^ Q

T75 (P 4-^ Q ) A Q ^ P

T76 (P<->Q ) a ~ P - ^ '~ Q

T77 (P<->Q ) a ~ Q - > ~ P
T78 (P [Q R ] ) ^ ([P Q] [P ^ R ])

T79 (P -> [Q 4-^ R]) (P A Q 4-> P A R)

T80 (P 4-)^ Q ) V (P 4-> ~ Q )

T 81 (P 4-y Q ) 4-> (P ^ Q ) A (Q ^ P)

T82 ( P ^ Q ) ^ ~ ( [ P ^ 'Q ] ^ ~ [Q ^ P ])

T83 (p 4_y Q ) 4-> (P A Q ) V ( ~ P A ~ Q )

T84 P A Q -> (P 4-> Q )

T85 ~P a ~ Q - > ( P 4->Q)

T 86 ([P ^ Q] R) (P A Q R) A ( ~ P A .~ Q

T87 ~ (P 4-> Q ) 4-> (P A ~ Q) V ( ~ P A Q)

T88 P a ~Q ~ ( P 4->Q)
T89 ~P a Q -> ~ ( P 4-^Q)
T90 ~ (P 4-^ Q ) 4-> (P 4-^ ~ Q )

T91 P 4 -^ P

T92 (P 4-^. Q ) 4-^. ( Q 4-^ P )

T93 (P 4-» Q ) A ( Q 4-> R) -> (P 4-^ R)

T94 (P 4-y [Q 4-> R]) 4-). ([P 4-> Q ] 4-> R)

T 9S (P 4-> Q ) 4-> ([P 4-^ R] 4-> [Q 4-> R])

T96 ( P 4 - > Q ) 4 - ^ .( ~ P 4 - ^ ~ Q )

T97 (P 4-> R) A ( Q 4-^ S) -> ([P Q ] 4-> [R -> S])

T98 (P 4-^ R) A (Q 4-^ S) -> (P A Q 4-> R A S)
T99 (P 4-> R) A (Q 4-> S) -> (P V Q 4-> R V S)
T ioo (P 4-> R) A (Q 4-> S) -> ([P 4-> Q] 4-> [R 4H.S ])

T io i (Q 4-> S) -> ([P ^ Q] 4-> [P -> S]) A ([Q ^ P]

68 . ‘And*, ‘Or’, ‘If and Only If’ Chapter 11

T 102 (Q S) (P A Q P A S)
T 10 3 (Q S) -> (P V Q P V S)
T 10 4 (Q <-> S) -> ([P <-> Q] [P <- S])
T 10 5 P A (Q •«-> R ) ->■ (P AQ ^ R)
Each of the following theorems is an occasionally useful biconditional
corresponding to a conditional listed in chapter I. The number of the
corresponding conditional (and in some cases that of its converse) is indi­
cated in parentheses.
T I0 6 (P [Q -> R]) <-> ([P Q] ^ [P ^ R]) (T6, T7)
T I0 7 (P -^ [Q ^ R ])< -» (Q -> [P ^ R ]) (T8)
T I08 (P -» [P ^ Q ])< -> (P ^ Q ) (T9)
TI09 ([P -> Q ] -> Q )< -^ ([Q -> P ] ->P) (T io)
T iio P ~ ~P ( T lI,T l2 )
T in (P - ^ Q ) 4-> (~ Q ~P) (T 1 3 )
T112 (P ^ ~ Q ) (Q -> ~ P ) (T 1 4 )
T113 ( - P -^ Q )< -> (-Q ->P ) (T 15 )
T114 (~ P (T 1 9 )
T 115 (P -> ~P)<-^ - P (T20)

EXERCISES, GROUP I

29. In the following derivation, which is a proof of T83, insert

with annotation all lines omitted by the use of clause (8).
1. Shev)-(? ^ Q) ^ (P A Q) V ( ~ P ^ ~Q )
2. Show (P<->Q)->
(P a Q) v ( ~ P A ~Q)
3
4

S-
6.
7'
8 3, BC, 7, MP, 7, Adj
9 s .R
10 3, BC, 6, MT, 6, Adj
11 4. CD
Section s Theorems with Abbreviated Proofs

12. Show P A Q (P Q)
13- PaQ
14. P<->Q 13, S, T2, MP, 13, s,
T 2 ( | - ^ ) ,M P ,C B

15- Show ~Q
A

16. ~ P A '■ ^Q
17- P<-^Q 16, S, T18, MP, i6, s,
T i8 ( ^ ) ,M P ,C B
18. (P^Q)-
( P a Q ) v (~ P A -.Q) 12, 15, SC, 2, CB
Of the preceding theorems, prove
30. T55, T58, T 6 0 -T 6 2 ;
31. T63, T64, T68, T69;
32. T 7 0 -T 7 3 ; ^
33. T 78 -T 8 0 ;
34. T86, T87, T90, T9s;
35* T 9 7 -T 10 0 . ^ ^

In solving exercises 3 0 -3 5 (as well as later exercises), the reader will

find the following informal suggestions helpful (though again not
infallible).

(j) To derive a sentence

derive first
<I>^X
and

and then use SC.

(2) To derive
<t>^^ ,
where <f>is not a disjunction^ use conditional derivation.
(5) To derive a conjunction, derive first both conjuncts, and then use
Adj.
(4) To derive

derive first
<j>—>■ \ji
and then use CD.
70 . ‘And*, ‘Or*, ‘If and Only If* Chapter II

(5) To derive a biconditionaly derive first the two corresponding con-

ditionalsy and then use CB.
(6) To derive anything elsey use either indirect derivation or separation
of cases.

A partial derivation of T94 is given for illustration. After writing

the initial assertion line, we begin, according to suggestions (5) and (2),
a conditional derivation of one of the relevant conditionals.
1 . Show (P ^ •R]). ,( [ P ^ Q ] ^ R )
2. Show (P < >[Q< ■ R]) ([P^Q ]<->R )
3. P<-> [Q^ R]
To complete the conditional derivation started in line 2 we must
derive a biconditional neither of whose constituents is a disjunction;
thus we again follow suggestions (5) and (2). (Note that it would be
superfluous here to enter a line with ^Shozo' follow’ed by the bi­
conditional in question.)
1. Show (P [Q R]) ^ ([p. Q]< R)
2. Show (P <-> [Q <-> R]) ([p. Q]< ■ R)
3. P<-> [Q<-> R]
4. Show [P Q] R
5- P<-^Q
Since ‘R’, the sentence to be derived to complete the subsidiary
derivation for line 4, is not an obvious consequence of the antecedent
lines available (lines 3 and 5), we turn to suggestion (6). To begin an
indirect derivation of ‘R ’ would not improve our position, for it is not
obvious that a contradiction can be derived from ‘ /^R* along with
lines 3 and 5. The most useful strategy is to separate cases. We see
clearly that ‘R^ follows once we assume ‘P’ :
1. Show (P ^ ■ »[Q- ■ R]). ■ ([P. Q]<->R)
2. Show (P < R]) ([P. Q]^>R)
3* P<->[Q<- ■ R]
4- Show [P< ■ >Q]- R
5- P<->Q
6. R
7- P
8. Q 5, BC, 7, MP
9- Q<->R 3, BC, 7, MP
10. R 9, BC, 8, MP
Now we derive ‘R ’ with as an assumption:
1. Show (P ^ [Q ^ R])<->([P. Q]<->R)
2. Show (P [Q ^ R])->([P< Q]<->R)
3- P<->[Q<->R]
4- Show [P <-> Q] R
Section 5 Theorems with Abbreviated Proofs . 71

5- P<-^Q
6 , ,Show P —^ R
7- P
8. Q S, BC, 7, MP

9* Q ^R 3. BC, 7, MP
10. R 9, BC, 8, MP
II. Show- -> R
12. -P
13- [Q <—> R] 3, BC, 12, MT

14. Q<-+ ~ R T9 o( q j ^), BC, 13, MP

IS- ~Q 5, BC, 12, MT
16. R 14, BC, 15, MT, DN

Then, in employing SC, we are able to complete the subsidiary deriva­

tion started in line 4.
1. Show (P <-> [Q <-> R]) <-► ([P <-> Q] <->• R)
2. Show (P <-> [Q ^ R]) ([P ^ Q] R)
3. P<-> [Q w R]
4. ¿:Aew[P<->Q]->R

5, BC, 7, MP
3, BC, 7, MP
9, BC, 8, MP

3, BC, 12, MT
/P Q\
MP
' 5'
5, BC, 12, MT
14, BC, 15, MT, DN
6, II, SC
To complete the subsidiary derivation started in line 2, the con­
verse of the conditional in line 4 must be derived; and then to com­
plete the main derivation, the converse of the conditional in line 2
must be derived. We leave these details to the reader.

E X E R C I S E S , G R O UP II

Show the following arguments valid by constructing symbolizations

and deriving the conclusions of the symbolizations from their premises.
72 . ‘And’, ‘Or’, ‘If and Only If’ Chapter II

Indicate in each case the scheme of abbreviation used. (Before sym­

bolizing Nos. 36 - 39, the reader should review the remarks made in
connection with the solution of exercise 19 of chapter I, p. 29.)

36. If Mary joins a sorority and gives in to her inclinations, then

her social life will flourish. If her social life flourishes, her academic
life will suffer. Fortunately, Mary will give in to her inclinations, but
her academic life will not suffer. Therefore she will not join a sorority.
37. Either love is blind and men are not aware of the fact that love
is blind, or love is blind and women take advantage of the fact that
love is blind. If men are not aware of the fact that love is blind, then
love is not blind. Therefore women take advantage of the fact that
love is blind.
38. If Alfred is a lover of logic who organizes his time, then he
enjoys Mozart in the morning or whiskey at night, but not both. If
he enjoys whiskey at night, then either he enjoys Mozart in the morn­
ing and organizes his time, or he does not enjoy Mozart in the morning
and does not organize his time, or else he is not a lover of logic. Alfred
enjoys whiskey at night provided that he both enjoys Mozart in the
morning and organizes his time. Therefore Alfred is a lover of logic
only if he does not organize his time.
39. Either it is not the case that Alfred pays attention and does not
lose track of the argument, or it is not the case that he does not take
notes and does not do well in the course. Alfred neither does well in
the course nor loses track of the argument. If Alfred studies logic,
then he does not do well in the course only if he does not take notes and
pays attention. Therefore Alfred does not study logic.

♦EX ERC ISES, GROUP III

We can generalize the definitions of conjunction and disjunction: by

the conjunction or disjunction of the sentences ^ 1 ,.. . , we shall under­
stand the sentence
<f>i A A <f>n

or
V ... V

respectively. (In case « is i, both these sentences reduce to (f>i,)

By a basic sentence we understand either a sentence letter or the
negation of a sentence letter. We say that a sentence is in conjunctive
normal form if it is a conjunction of disjunctions of basic sentences,
and in disjunctive normal form if it is a disjunction of conjunctions of
basic sentences. Two symbolic sentences <f> and i/j are said to be
equivalent if the biconditional
(j>i—y ijj
is a theorem. It happens that every symbolic sentence is equivalent
Section 6 Truth-Value Analysis of Sentences . 73

to some sentence in conjunctive normal form, and also to one in dis­

junctive normal form. The theorems in the next exercise illustrate this
point.
40. Prove:
(P-^[Q<->R])<->(^P V - Q vR) a (-Pv - R v Q)

( P [ Q <-> R ] ) ^ - P v(Q a R) v (-Q A -R )

- [ ( - P a - Q) v ( - R a - S ) ] < - > ( P v Q) a ( R v S)

—[(—P A —Q) V ( —R A —S)] <— >

(P A R) V (P A S) V (Q A R) V (Q A S)
41. For each of the following sentences, find an equivalent sen­
tence in conjunctive normal form and one in disjunctive normal form.
P -> Q a R
- ( P v Q -> R )
p<_>(Q<_>R)
([P -> Q ]-> P )-> P
(P->Q ) a ( P - > - Q ) ^ P
42. Formulate a general procedure for transforming any symbolic
sentence (of the sentential calculus) into an equivalent sentence in
conjunctive normal form.
43. Formulate a general procedure for transforming any sentence
in conjunctive normal form into an equivalent sentence in disjunc­
tive normal form.
6. * Truth-value analysis of sentences. Inability to construct a
derivation corresponding to a given argument is not conclusive evidence
that no such derivation exists. The question naturally arises whether
there is an automatic method of testing an argument for validity. For
symbolic arguments of the sentential calculus there is such a method.
Its presentation requires some new notions.
A sentence may have one of two truth values^ truth (T) or falsehood (F).
We may consider arbitrary assignments of truth values to sentence letters;
such an assignment correlates with each sentence letter either T or F ,
Given an assignment we may determine the truth value of an arbitrary
symbolic sentence (f> with respect to A by the following rules:
(1) If is a sentence letter, then the truth value of <f> is the truth value
correlated with ^ by
(2) The truth value of
—(j)
is T if that of (f> is Fy and F if that of <f> is T.
(3) The truth value of
74 . ‘And’, ‘Or*, ‘If and Only If* Chapter II

is F if that of (/> is T and that of 0 is F ; otherwise the truth value of

(¡> ->tfi
is T.
(4) The truth value of
(f> A ifj

is T if the truth values of (f> and i/j are both T ; otherwise the truth value
of
(f) A ijj
is F .
(5) The truth value of

is F if the truth values of (f) and ifj are both

1 F ; otherwise the truth value
of
V</r

is T,
(6) The truth value of
(j) i-► Ip
is T if (p and ip have the same truth value; otherwise the truth value of
(p <—> p
is F.

A tautology is a symbolic sentence whose truth value is T with respect

to every possible assignment. For instance,
(i) P -^ P
is a tautology. For consider any assignment A, It will correlate with ‘P*
either T or F, In neither case can the antecedent of (i) have the value T
and the consequent the value F, Hence, by rule 3, the value of (i) is
always T.
It is convenient, in deciding whether a symbolic sentence is a tautology,
to construct a truth table. The following is a truth table for the sentence
‘(P ^(p A

p Q 1P ~Q P A Q (P A Q) (P -> Q) <—
> (P A Q)

T T T F F T T
T F F T T F T
F T T F F T T
F F T T F T T
Section 6 Truth-Value Analysis of Sentences . 75

The table can be regarded as an abbreviation of the following considera­

tions. Let A be an arbitrary assignment. If A correlates T with both T*
and ‘Q’ (first row), then (by rule 3, p. 73) the value of T -> Q’ with
respect to A is T*, that of is (by rule 2) F, hence (by rule 4)
the value of T a is F, hence (by rule 2) that of ‘ ^ (P a ^ Q )’ is
F, and hence (by rule 6) the value of ‘(P -> Q)<-> ^~(P a is T, If
A correlates T with T ’ and F with ‘Q* (second row), then the value of
T ->Q* isF , that o f i s F, that of T a is F, that o f ( P a ^ Q )’
is F, and hence that of ‘(P Q)<-> ^ (P a ^Q )' is F. If A correlates F
with T ’ and F with ‘Q’ (third row), then, similarly, the value of ‘(P -> Q)
^ (P A ^Q )' is F. Finally, if A correlates Fw ith both ‘P’ and ‘Q’, the
value of ‘(P Q)<-> ^ (P A ^ Q )’ is F. Since in all cases the value of
‘(P -> Q)<-> ~ (P A ^ Q )’ is F, this sentence is a tautology.
In general, to construct a truth table for a given symbolic sentence
proceed as follows. Provide a column for each component of the sentence,
including the sentence itself. Arrange the components in order of in­
creasing complexity, placing all sentence letters to the left of a double
line. To the left of the double line, construct a row for each possible
combination of truth values. (If there are 71 sentence letters, there will
be 2” rows.) Proceed to the right in each row, filling in truth values for the
sentences heading the columns, on the basis of previous entries and the
rules on pages 73 - 74 . The sentence is then a tautology just in case its
table contains nothing but ‘ F ’ in the final column.
It happens that a sentence is a theorem of the sentential calculus if
and only if it is a tautology. (This fact will not be proved here; part of it
will be proved in chapter V.) Thus truth tables provide an automatic
test for validity in the case of symbolic arguments without premises.

EXERC ISES
Determine by means of truth tables which of the following sentences
are tautologies. (In order to show that a sentence is not a tautology,
it is sufficient to give, rather than a complete truth table, one row
whose last entry is ‘F ’.) Exercise 44 is solved for illustration.

44. (P A Q -> R) -> (P R)

The following single row of a truth table is sufficient to show that
No. 44 is not a tautology:
p Q R 1 P aQ P -^ R P a Q- ^R (P a Q -> R )-> (P -^ R )

T F F 1 F F T F

45. ( P ^ R ) - > ( P v Q ^ R )
46. T58, T61, T66, T73, T78, T94.
47. The converse of T32 and the converse of T93.
76 . ‘And’, ‘Or’, ‘If and Only If’ Chapter II

7. * Truth-value an alysis o f argum ents. A symbolic sentence <f>

is tautologically implied by other symbolic sentences just in case there is
no assignment of truth values with respect to which each of the latter
sentences has the truth value T while (f> has the truth value F. It happens
(but will not be proved) that a symbolic argument is valid in the sentential
calculus if and only if its conclusion is tautologically implied by its
premises.
In the case of symbolic arguments with finitely many premises, truth
tables are convenient for determining tautological implication. Consider,
for example, the argument

(I) P ->Q P -> ~ Q

The following is a truth table for this argument:
P Q ~P ~Q P^Q P-

T T F F T F
T F F T F T
F T T F T T
F F T T T T
The table can be regarded as an abbreviation of the following considera­
tions. Let A be an arbitrary assignment. I f A correlates T with both ‘P’
and ‘Q’ (first row), then the value of ‘ ~ P ’ with respect to A is F , and that
of ‘ ~ Q’ is F, that of ‘P -> Q’ is T, and that of ‘P ^ ~ Q’ is F. The other
rows have a similar meaning, and the four rows exhaust all possible cases.
We observe that in all cases in which the premises of (i) all receive the
value T (rows 3 and 4), the conclusion also receives the value T. Thus the
conclusion of (i) is tautologically implied by its premises, and hence
(i) is valid.
In general, to construct a truth table for a given symbolic argument
proceed as follows. Provide a column for each component of the premises
and the conclusion, including these sentences themselves. Arrange the
components in order of increasing complexity, placing all sentence letters
to the left of a double line. To the left of the double line, construct a row
for each possible combination of truth values. Proceed to the right in
each row, filling in truth values for the sentences heading the columns,
on the basis of previous entries and the rules on pages 73 - 74. The
conclusion of the argument is tautologically implied by its premises just
in case there is no row in which the premises all have the value T and the
conclusion has the value F .
Since a symbolic argument is valid in the sentential calculus just in
case its conclusion is tautologically implied by its premises, truth tables
provide both an automatic test for validity and a method of showing
invalidity in the case of symbolic arguments with finitely many premises.
Section 7 Truth-Value Analysis of Arguments . 77

The truth-table method of showing invalidity applies directly only to

symbolic arguments. The reader will recall that an English argument is
valid in the sentential calculus just in case it has a symbolization whose
conclusion is derivable from its premises (within the sentential calculus).
Hence, to show the invalidity of an English argument within this branch
of logic, it is necessary to show that no symbolization of it is valid. Using
truth tables, we may be able to show that various particular symbolizations
are invalid, but without more precise information about the notion of a
symbolization (in particular, about the notion of stylistic variance, on which
it depends) we shall be unable to establish general assertions about all
possible symbolizations of a given argument. Thus, although it would be
desirable to develop a test of validity for arguments expressed in idiomatic
English, this does not seem possible without a more detailed analysis of the
notion of stylistic variance than we care to undertake.

EXERC ISES
Test each of the following arguments for validity, using the method
of truth tables. It is sufficient in showing invalidity to exhibit a single
row of a truth table. (The invalidity of Nos. 48 and 49 was forecast
on page 29.)

48. P . Q .*. R
49. P -> R . ^R .*. Q
50. P - > Q v R . R -> Q aP . Q -> R P<->Q
51. ^ S v P v Q . . P~>Ra S Q
52. Q -> (Q V [R S] ^ [R V S])

Although we cannot establish in a clear-cut way the invalidity (with­

in the sentential calculus) of arguments expressed in idiomatic English,
we can at least, if they are invalid, establish the invalidity of any one of
their symbolizations. Together with each of the arguments 5 3 - 5 7
below we give an interesting symbolization, that is, one which appears
to reflect to a maximal degree the logical structure of the argument
that it symbolizes. Such a symbolization will be one that is as long as
possible and, among those of maximal length, one which uses the small­
est possible number of different sentence letters. For each of the
symbolizations below, either derive its conclusion from its premises or
show by the method of truth tables that it is invalid. In finding deriva­
tions, the reader will again find helpful the suggestions made on pages
69-70. (Nos. 53 and 54 correspond to an example on page 3.)
53. If Alfred studies, then he receives good grades; if he does not
study, then he enjoys college; if he does not receive good grades, then
he does not enjoy college. Therefore Alfred studies.
P -> Q . T -> R . ' Q —> R P
78 . ‘And^ *Or\ ‘If and Only If’ Chapter II

54. If Alfred studies, then he receives good grades; if he does not

study, then he enjoys college; if he does not receive good grades, then
he does not enjoy college. Therefore Alfred enjoys college.
P Q . ^P—
>R . Q R R
55. Alfred will marry either Alice or Mary, but not both. If he
marries Alice but not Mary, then he is fortunate. If he does not marry
Alice but marries Mary, he is also fortunate. Therefore Alfred is
fortunate.
( P v Q )a - ( P a Q) P A P a Q -> R R
56. Alfred passes just in case he is both intelligent and industrious,
and he is intelligent. Alfred is industrious if and only if he both is
intelligent and does not pass, provided that he does not succumb to
temptation. Therefore Alfred succumbs to temptation.
(R ^ P a Q )a P ' S —> (Q <—
►PA R)
57. If men prefer chaste women just in case chastity is a virtue,
then men prefer virtuous women; but men do not prefer virtuous
women. Therefore it is not the case that if men prefer chaste women,
then'chastity is a virtue.
( [ P ^ Q] R) A -R -(P Q)

Give an interesting symbolization of each of the following arguments,

and for each symbolization either derive its conclusion from its premises
or show by the method of truth tables that it is invalid.
58. If God is willing to prevent evil but unable to do so, he is im­
potent. If God is able to prevent evil but unwilling to do so, he is
malevolent. Evil exists if and only if God is either unwilling or unable
to prevent it. God exists only if he is neither impotent nor malevolent.
Therefore if God exists, evil does not exist.
59. Caesar went to England, and either Pompey went to Spain or
Crassus marched against the Parthians. It is not the case that
both Caesar went to England and Crassus was not killed by the
Parthians. If Caesar went to England and Cicero denounced Catiline,
then Pompey did not go to Spain. If Cicero did not denounce Catiline,
then Crassus marched against the Parthians. Therefore Caesar was
murdered on the Ides of March, 44 B.c.
60. If neither an adequate social life nor a comfortable income can
replace Alfred’s love of logic, then, provided that he has normal
intelligence, he can expect nothing better than an instructorship. An
adequate social life can replace Alfred’s love of logic just in case a
comfortable income can replace his love of logic. Therefore either an
adequate social life can replace Alfred’s love of logic or he can expect
nothing better than an instructorship.
Section 8 Historical Remarks • 79

8. * H istorical rem arks. What is now called the sentential calculus

has been designated variously the ‘calculus of equivalent statements'
(MacColl [i]), the ‘Aussagenkalkiil* (Schröder [i]), the ‘propositional
calculus' (Russell [2]), and the ‘theory of deduction' (Russell [3]).
The sentential calculus was first investigated by the Stoics (see
Lukasiewicz [i] and Mates [i]). Its modern development was adumbrated
by Leibniz (see Couturat [i] and [2]), in Boole [i] and [2], De Morgan
[i], Peirce [i], and MacColl [i], but properly begins with Frege [1]. Its
fullest development, based on the ideas of Frege, occurs in Whitehead
and Russell [1].
Besides the system of symbols used here, which is due to Tarski, there
are two other principal systems. The notation that is' most frequent in
the literature is that of Whitehead and Russell [i]; its symbols are the
following:
' fo r ‘^
3 ' fo r
•' fo r ‘ a
V ' fo r ‘ V
=' fo r

(To indicate grouping, Whitehead and Russell use, along with parentheses,
a complicated system of dots.)
A much more economical notation is that of Lukasiewicz [1]. The
essential feature is the placement of sentential connectives. A binary
connective is placed beforcy rather than betweeUy the two sentences which
it connects. Thus, using our symbols ‘ ^ ' and we would write
P ->Q R
for
(~ P ^ (Q -> R )) .
and
->-> ~ P Q R
for
((-P -> Q )^ R ) .
The advantage of Lukasiewicz' notation is that it makes unnecessary
special symbols for grouping, such as parentheses.
The notion of an instance of a theorem and the principle that an in­
stance of a theorem is again a theorem occur more or less explicitly in Frege
[3], Couturat [3], and Russell [2].
The method of truth tables occurs informally in Frege [i], and explicitly
in Peirce [2]. The assertion that a symbolic argument is valid in the sen­
tential calculus if and only if its premises tautologically imply its conclusion
was first established in Post [1].
8o . ‘And*, ‘Or*, ‘If and Only If* Chapter II

Our formulation of the sentential calculus admits of a certain simplifica­

tion, at the expense of detracting from its intuitive character. We could,
as in chapter I, dispense with indirect proof and replace the four inference
rules of that chapter by two, modus ponens and a variant of modus tollens;
further, we could replace the primitive inference rules introduced in the
present chapter by the following rules, which embody in some sense
‘definitions’ of ‘a ’, ‘v ’, in terms of and

^{<f> ^tj,)
~ ('A
(f}yi/j
(«^ v«A)

{{<f> <A) A (l/r <f,)) {4 ,^ 4 .)

The observation made at the end of section 4, that there is no automatic

procedure which, in the absence of annotations, will permit a decision as
to the correctness of an abbreviated derivation, holds in the sentential
calculus only when infinitely many premises are permitted. It can be
shown, however, that even when the class of premises, though allowed to
be infinite, is required to be decidable (that is, loosely speaking, such that
it can automatically be decided of any given sentence whether it is a member
of the class), it is not generally possible to formulate an automatic test
for a correct abbreviated derivation from that class of premises without
recourse to annotations. In abbreviated derivations of the quantifier
calculus (that is, the system of chapters III and IV), annotations are
indispensable even when only finitely many premises are involved.

9. A p p en d ix : list o f th eo re m s o f chapters I and IL

Ti P -> P
T2 Q -> (P -> Q)
T3 P -> ([P -> Q ]^ Q )
T4 (P ^ Q) -> ([Q -> R] [P ^ R])
T5 (Q -> R )^ ([P -> Q ]-> [P ^ R ])
T6 (p -> [Q R]) -> ([P ^ Q] -> [P ^ R])

T? ([P ->Q ] ->[P -^R ]) ->(P ^ [ Q ->R])

T8 (P -> [Q ^ R]) (Q -> [P R])
T9 (P ^ [P -^ Q ])^ (P -> Q )
Section 9 Appendix: List of Theorems of Chapters !
T ío ([P ^ Q ]^ Q )-> ([Q -> P ]-> P )
Tu ~ ~ P -> P
T l2 P ~ ~P

T I3 (P ^ Q )-> (~ Q ^ ~ P )
T I4 (P ^ ~ Q )-> (Q ^ ~ P )
T I5 (~ P ^ Q ) ^ ( ~ Q ^ P )
T I6 (~ P ~Q ) ^ (Q ^ P )
T I7 P ^ (~ P ^Q )
T I8 ~ P -^(P ^ Q )
T I9 (~ P ^ P ) ^ P
T20 (P ^ ~ P ) -> ~ P
T2I ~ (P ^ Q ) ^ P
T22 ^ (P Q) ^Q
T23 ([P Q] ^ P) ^ P
T24 P A Q <-> Q A P
T25 P A (Q A R) (P A Q) A R
T26 (P ^ Q) A (Q ^ R) (P R)
T27 (P A Q ^ R) ^ (P ^ [Q ^ R])
T28 (P a Q ^ R ) ^ ( P a ~ R ^ ~Q )
T29 (P ^ Q A R) ^ (P ^ Q) A (P ^ R)
T30 (P ^ Q) -> (R A P -> R A Q)

T 3I (P ^ Q) -> (P A R ^ Q A R)
T32 (P ^ R) A (Q ^ S) ^ (P A Q R A S)

T33 (P ^ Q ) a ( ~ P ^ Q )

T34 (P ->Q) A (P ^ ~Q ) ^ ~ P

T35 ( ~ P ^ R ) a (Q -> R) ([P ^ Q] ^ R)

T36 ~ (P A ~ P )

T37 (P ^ Q ) ^ > ~ ( P a ~Q )
T38 P a Q < -> ~ (P -^ ~ Q )
 Chapter II

T39 ~ (P A Q) <-> (P ~ Q)
T40 ~ (P Q) >P A ~Q
T 4I P P A P
T42 P a ~Q —
> ~ ( P -> Q)

T43 ^P — (P A Q)
T44 ^ Q “ >■ (P A Q)

T45 P v Q < - > ( ~ P ->Q)

T46 (P Q) 4 -)^ ~P V Q

T47 P 4-> P V P
T48 (P V Q) A (P ^ R) A (Q ^ S) ^ R V S
T49 (P V Q) A (P ^ R) A (Q -> R) -> R
T so (P ^ R) A (Q -> R) 4-4 (P V Q -> R)

T 51 (P V Q) A (P ^ R) A (~ P A Q -> R) -> R
TS2 (P - > R ) a ( ~ P a Q ^ R ) 4->(P v Q ->R )

T 53 P V Q 4-> Q V P

T54 P V (Q V R) 4 -^ (P V Q) V R

T55 (P Q V R) 4-> (P Q) V (P ^ R)
T56 (P ^ Q) ^ (R V P R V Q)

T 57 (P ^ Q) ^ (P V R ^ Q V R)
T58 (P ^ Q) V (Q -> R)

T59 P V ~P
T60 (P ^ R) V (Q ^ R) 4-> (P A Q ^ R)
T61 P A (Q V R) 4-> (P A Q) V (P A R)
T62 P V (Q A R) 4-> (P V Q) A (P V R)

T63 P a Q 4-4 ~ ( ~ P v ~Q )

T64 P V Q 4 -4 ~ (~ P A ~ Q)

T65 ~ (P A Q) 4— > ~ P V ~Q

T66 ~ (P V Q) 4— > ~ P A~ Q

T67 ~P a ~ Q -> ~ ( P v Q)
Section 9 Appendix: List o f T h eo rem s o f Cl

T68 P < -> (P a Q) v (P a ~Q )

T69 P (P V Q) A (P V ~ Q)
T70 Q -> (P A Q <-> P)
T 71 ~ Q ^ ( P vQ <-^P)
T72 (P ->Q )<->(P a Q<->P)

T73 (P -> Q) <-> (P V Q Q)

T74 (P Q) A P -> Q

T75 (P ^ Q) A Q -> P
T76 (P 4-> Q) A ~P ~Q
T77 (P<->Q ) a ~ Q - > ~ P
T78 (P ^ [Q<-> R]) ([P ^ Q] -(-> [P -> R])
T79 (P ^ [Q ^ R]) ^ (P A Q P A R)
T80 (P<->Q) v ( P ^ ~Q )
T 81 (P Q) (P Q) A (Q ^ P)
T82 (P ^ Q )< -> ~ ([P ^ Q ] ^ P ])
T83 (P<-> Q) <-> (P A Q) V ( ~ P A ,~Q)

T84 P aQ ^(P < -^ Q )

T85 ~P A ~Q (P Q)
T86 ([P ^ Q] ^ R) (P A Q ^ R) A (~ P A Q ->R )
T87 ~ (P > Q) <—> (P A ~ Q) V (~ P A Q)
T88 P A ~Q ~ (P > Q)
T89 ~P A Q ~ (P <—> Q)
T90 ~(P<-^Q )<-).(P 4-> ~Q )

T 91 P<->P

T92 (P<->Q)<^(Q<-^P)

T93 (P ^ Q) A (Q R) -> (P<-^ R)

T94 (P [Q<-^ R]) <-> ([P<-> Q] <-> R)

T95 (P Q) <-> ([P ^ R] <-> [Q R])

T96 (P Q) ~ P ~ Q)
84 • ‘And’, ‘Or’, ‘If and Only If’ Chapter 11

T97 (P R) A (Q . S) ^ ([P ^ Q] ^ [R ^ S])

T98 (P ^ R) A (Q S) -> (P A Q R A S)
T99 (P R) A (Q S) ^ (P V Q R V S)
T io o (P <-> R) A (Q ♦ -> S) ^ ([P Q] <-». [R+-> S])
T io i ( Q ^ S ) ^ ( [ P ^ Q ] ^ [ P ^ S ] ) a ([Q ^ P ] [S ^ P ])
T 10 2 (Q<->S) ^ ( P a Q <->.P a S)
T 10 3 (Q S) -> (P V Q ^ P V S)
T 10 4 ( Q ^ S ) - > ( [P 4- ^ Q ] ^ [ P ^ S ] )
T 10 5 P A (Q R) ^ (P A Q <-> R)
T 10 6 (p ^ [Q ^ R]) ^ ([P ^ Q] ^ [P ^ R])

T 107 (P ^ [Q ^ R ])< ^ (Q ^ [P -> R ])

T 108 (P ^ [P _ > Q ])^ (P ^ Q )

T 109 ([P ^ Q ] ^ Q ) ^ ( [ Q ^ P ] ->P)

T iio p >~ ~ p
T in (P ^ Q )^ (^ Q ^ ^ P )

T 112 (P -> ~ Q ) ^ ( Q ^ ~ P )
T 113 ( ~ P ^ Q ) ^ ( ~ Q ->P )
T 114 ( ~ p ^ P )< -).P

T 115 (P ->■ ~ P ) 4-> ~ P

 Chapter I II
‘ A L L ’ and ‘ S O M E ’

I. Variables, quantifiers, formulas. Not all intuitively valid argu­

ments can be reached by the procedures of chapters I and II. For example,
the argument
(i) All Communists are Marxists. Some Communists are American.
Some Marxists are American,
unlike the arguments in chapters I and II, depends for its validity on
the meaning of the words ‘all’ and ‘some’.
Before embarking on the logic of such words, we must engage
in grammatical considerations. Of sentences we have already spoken.
A sentence of English is an expression of English that is either true or
false. An English name is a word or group of words that designates (at
least within a given context) a single object. For example, ‘Socrates’,
‘David Hume’, ‘7 ’, ‘7 + 5’, and ‘the author of Waverley^ are names re­
spectively designating Socrates, David Hume, 7, 12, and Sir Walter
Scott. A variable is a lower-case Latin letter, with or without a numerical
subscript. For example, ‘2 ’, ‘jci’, ‘^3’, ‘2:2’ are variables.
Using variables we can construct expressions that closely resemble
English sentences but fail to be sentences. For instance, the expression

(2) X is bald.
unlike

(3) Socrates is bald

and
(4) Samson is bald,
is neither true nor false, and hence is not a sentence.
Although we cannot ascribe truth or falsehood to (2), we can make
other assertions about it. For instance, we can assert that a particular
object, say, Socrates or Samson, satisfies (2); this amounts to asserting
(3) or (4). Or we can say that every object, or at least one object, satisfies
(2); these assertions amount respectively to
86 . ‘A ir and ‘Some* Chapter III

is) For each Xy x is bald

and
(6) There is an object x such that x is bald.
(5) and (6), although not completely idiomatic, can nevertheless, like
their more idiomatic counterparts ‘Everything is bald’ and ‘Something
is bald’, be construed as sentences of English; in fact, (5) is false and
(6) is true. We have thus discovered three ways of converting (2) into a
sentence—by replacing its variable ‘^’ by a name or by prefixing one
of the phrases ‘for each x ’ or ‘there is an object x such that’.
We shall abbreviate ‘for each’ by ‘A’ and ‘there is an object. . . such
that’ by ‘V’. (5) and (6) then become
K x X is bald
and
Vx X is bald
The symbols ‘A’ and ‘V’ are known respectively as the universal quantifier
and the existential quantifier, A quantifier may be written before any
variable to form a quantifier phrasey which is universal or existential accord­
ing to the quantifier used. The English counterparts of quantifier phrases,
that is, expressions of the form
for each a
or
there is an object a such that ,
where a is a variable, are czWtd phrases of quantity.
_ By a formula of English we shall understand either a sentence of English
or an expressfon containing occurrences oLvariables which becomes a
sentence of English when some or all of these occurrences are replaced
by English names. (2) above, as well as each of the following, is a formula
of English:
David Hume is human.
(7) X loves y y
(8) there is an object y such that x loves y.
X -{■ y = Z y
(9)
if X loves yy then y loves x.
To see, for example, that (8) is a formula, we need only replace the variable
^x^ by ‘Romeo’, and to see that (9) is a formula, we may replace the vari­
ables ‘x ’, ‘y , 'z' by the names ‘2’, ‘3 ’, ‘5’.
In symbolic abbreviations, we shall henceforth employ not only sentence
letters but also predicate letters. These are to be capitals from ‘F ’ through
Section I Variables, Quantifiers, Formulas . 87

‘O’, with or without numerical subscripts; T ’ through ‘Z ’ continue in

their role as sentence letters, and ‘A ’ through ‘E ’ are reserved for later
uses. A predicate letter accompanied by a variable will serve as an abbre­
viation of a formula of English containing that variable. For example,
the formula (2) may be abbreviated by
F.V .
The language with which we are now concerned is obtained by adding
to English the following symbols:

(1) the sentential connectives;

(2) parentheses;
(3) the quantifiers, that is, ‘A’ and ‘V’ ;
(4) variables, that is, lower-case letters with or without subscripts;
(5) sentence letters, that is, capital letters T ’ through ‘Z ’ with or
without subscripts;
(6) predicate letters, that is, capital letters ‘F ’ through ‘O’, with or
without subscripts.

As formulas we count formulas of English, their symbolic counter­

parts, and mixed combinations. To be more explicit, the class of for­
mulas can be exhaustively characterized as follows:

(1) A ll formulas of English {that iSy sentences of English or expressions

like sentences of English except for the occurrence of variables) are formulas,
(2) Sentence letters are formulas.
(j) The result of writing a predicate letter followed by a variable is a
formula.
{4) I f (f> and ip a re fo rm u la S y then so a r e
^ (f> ,

{<f> A ifl) ,
(<f> y ,

(5) I f <f) is a formula and a a variable^ then

A0L<f)
Va^
are formulas.

Clause (5) leads not only to

Ax¥x
88 . ‘A ir and ‘Some’ Chapter III

but also to seemingly meaningless combinations such as

AjcP
It would be artificial to exclude these possibilities; their meaning will
be explained in due course.
A symbolic formula is a formula constructed exclusively from variables,
sentences letters, predicate letters, parentheses, sentential connectives,
and quantifiers. More precisely, the class of symbolic formulas is ex­
haustively characterized by clauses (2 )-(5 ) above, reading everywhere
‘symbolic formula* for ‘formula*.
A few more terms will be useful. The result of prefixing one or more
universal quantifier phrases to a formula is called a universal generalization
of that formula. Similarly, one forms existential generalizations of a formula.
For example, the formulas
Kx X loves y ,
hxhy X loves y

are universal generalizations of (7), and the formulas

Vjc X loves y ,
yy^x loves y

are existential generalizations of (7).

EXERC ISES

Which of the following are formulas? For each expression that is a

formula, state whether it is a symbolic formula. Exercises 1 - 3 are
solved for illustration.

I. ((A:v(:v is a man is mortal) a Socrates is a man) -> Socrates

is mortal)
According to clause (i) of the characterization of the class of for­
mulas, 'x is a man’, *x is mortal’, ‘Socrates is a man’, and ‘Socrates
is mortal’ are formulas; according to clause (4), then, \x is a man
X is mortal)’ is a formula, and according to clause (5), 'Ax{x is a man
X is mortal)’ is a formula. According to clause (4) again, ‘(Ajc(^ is a man
X i s mortal) a Socrates is a man)’ and hence No. i are formulas.
Expression No. i is not a symbolic formula, however, for it contains
words of English.

2. A2(AWjv(Fjc V Gy) ^ H2)

According to clause (3), 'Fx\ ‘Gy*, and ‘H2* are formulas; thus
according to clause (4), ‘(Fa; v Gy)* is a formula, and according to
clause (5), first ‘Vy(FA; v Gy)* and then *AxWy(Fx v Gy)* are formulas.
Section 2 Bondage and Freedom . 89

By clause (4) again, ‘(AjcYy(F^ v Gy) -> is a formula, and ac­

cording to clause (5) again. No. 2 is a formula. It is a symbolic
formula, for it contains only variables, sentence letters, predicate
letters, parentheses, sentential connectives, and quantifiers.

2>Ax x is even v A yyisodd

This could be a formula only of the kind introduced by clause (5),
for it is not a formula of English (clause (i)), it is not a sentence letter
(clause (2)), it is not a predicate letter followed by a variable (clause
(3)), and its first symbol is neither a negation sign nor a left parenthesis
(clause (4)). But clause (5) is applicable only if what follows the quan­
tifier phrase *Ax* is a formula; that is, only if *x is even v Ajy y is odd^
is a formula, and, strictly speaking, this is not the case, as another
review of clauses (i) - (5) would reveal. Thus No. 3 is not a formula.

4. /^( /^ Ax Fx V Gy)
5- (Ficy ->■ Gy»)
6. Aa(H» <-> Gy)
7- VA»(H» G»)
8. Ay ■ ^V x(H;c A Gy)
9- Ay(Hy -> WxUx)
10. Ax{Ax{Hx y ^G y) V is between Berkeley and Los Angeles)

2. Bondage and freedom. A variable may occur more than once in

a formula. For example, in each of the formulas

(ï) Vjc(F;c A G jc)

(2) (VjcF jc a G:r)
there are three occurrences of the variable 'x\ All three occurrences
of V are bound in (i). Only the first and second occurrences of ^x' are
bound in (2); the third occurrence of ^x' (that following ‘G') is not bound in
(2) but instead is free in (2).
In general, an occurrence of a variable a is bound in a symbolic formula <l>
just in case it stands within an occurrence in (/► of a formula
Aoii/f
or
Va0
where i/r is a formula. An occurrence of a variable is free in a symbolic
formula <f>just in case it stands within (f) but is not bound in <f>.
The claim that all three occurrences of ‘a:’ are bound in (i) is verified
by noticing that each stands within an occurrence in ( i) of ‘Vx (F a: a G jc)*.
Likewise, the first and second occurrences of *x' in (2) stand within an
occurrence in (2) of y x ¥ x \ However, the third occurrence of 'x' in (2)
does not stand within an occurrence of any formula beginning with a
90 . *A\V and *Some’ Chapter III

quantifier phrase; thus the third occurrence of in (2) is free in (2). In

the formula

(3) ((F;c A Gy) V hx{Fx a Gy))

the second and third occurrences of the variable ^x^ are bound, because
they stand within an occurrence in (3) of

(4) h x { ^ x A Gy) .
The first occurrence of 'x\ however, as well as both occurrences of ^y\
is free in (3). Although the second occurrence of ‘y* in (3) stands within
an occurrence of (4), it is still free in (3), because the quantifier phrase
with which (4) begins does not contain ‘y*.
The previous considerations apply to occurrences of a variable. A
variable itself is bound in a symbolic formula (f>just in case some occurrence
of it is bound in (f>. Similarly, a variable is free in ^ just in case some
occurrence of it is free in <f>. For example, the variable ^x* is both bound
and free in (3), while ‘y* is only free in (3).
By a symbolic sentence is understood a symbolic formula in which no
variable is free. For example, (i) but not (2) is a symbolic sentence;
this fact is important in connection with exercises 15 and 16 of section 4.
Notice that the definitions above apply only to symbolic formulas;
it would be both unnecessary and extremely difficult to extend the
notions of bondage and freedom to formulas containing English com­
ponents.

EXERCISES

Consider the formula

((VxFx V hz{{Gz A Hx) -> (V^F^ v Hz))) <-> V2(Fy v ¥z)) .
11. In this formula identify each occurrence of a variable as bound
or free.
12. Which variables are bound in the formula?
13. Which variables are free in the formula?

3. Inform al notational conventions. We shall continue to use

the conventions of chapter II (pp. 40, 51, 65) for omitting parentheses
and replacing them by brackets. For practical purposes, then, we shall
deal with a larger class of formulas than that given in section i. It
should be emphasized, however, that in theoretical discussions—for
instance, in the criteria of bondage and freedom given in the last section
and in the rules of derivation that will appear in section 5—the word
‘formula’ is always to be understood in the official sense characterized in
section i.
Section 4 Translation and Symbolization . 91

4. T ra n s la tio n and s y m b o lizatio n . The English sentence

For each a:, x is bald
may be paraphrased in a variety of ways, for instance, as
Everything is bald.
Each thing is bald.
All things are bald.
For all Xy X is bald;
and the English sentence
There is an object x such that x is bald
may be paraphrased as
For some x, x is bald,
Something is bald.
At least one thing is bald.
There is a bald thing.
Using phrases of quantity together with the phrases of connection of
chapters I and II, we can develop stylistic variants of a good number of
English expressions in addition to those just mentioned. For instance,
Nothing is bald
may be paraphrased as either
For each x, it is not the case that x is bald
or
It is not the case that there is an object x such that x is bald .
Other examples are afforded by sentences of the familiar Aristotelian
forms.

(1) All men are mortal.

(2) Some cats are dogs.
(3) No men are mortal.
(4) Some cats are not dogs
are stylistic variants of
For each x (if x is a man, then x is mortal) ,
There is an object x such that (x is a cat and x is a dog) ,
For each x (if x is a man, then it is not the case that x is
mortal) ,
There is an object x such that (x is a cat and it is not the
case that x is a dog)
92 . ‘All’ and ‘Some’ Chapter III

respectively. The supposition that

(5) For each (a: is a man and x is mortal)
is a stylistic variant of the true sentence (i) can be rejected by observing
that (5) is false, for it logically implies the falsehood ‘for each Xy x is
man\ (See exercise 54.) Similarly, the supposition that
(6) There is an object jc such that (if is a cat, then .v is a dog)
is a stylistic variant of the false sentence (2) can be rejected by observing
that (6) is true, for it is implied by the truth ‘there is an object x such
that is a dog’ . (See Exercise 55.)
Somewhat less obvious examples of stylistic variance are provided by
the sentences
Only citizens are voters
and
None but citizens are voters.
both of which may be paraphrased as
For each x (if ;c is a voter, then x is 3. citizen) .
It is impractical to list all the combinations that are expressible in
terms of phrases of quantity. For example, context must be consulted
to determine whether ‘any’ should pass into a universal or an existential
phrase of quantity. Thus here, as in the sentential calculus, the notion
of stylistic variance is employed in an intuitive rather than an exact manner.
The process of translation is somewhat more involved here than in the
sentential calculus. We must first introduce an auxiliary notion: if <f> is
a formula of English, then a is said to be an apparent variable of <f> if
either <f) is 3 sentence of English containing a or there is an occurrence
of a in ^ such that, upon replacement of some other occurrences of
variables by English names, <f>becomes a sentence of English. For example,
is an apparent variable of each of the formulas
‘^’ is the first letter of the alphabet
For each a, a is bald
For some ay a is father of b .
The reason is that the first two formulas above are already sentences
containing ^a*y and the third becomes a sentence upon replacement of
‘A’ by any name.
We must also extend the notion of an abbreviation, for now formulas
as well as sentences of English will require symbolic representation.
Accordingly, we presently understand by an abbreviation either one in
the earlier sense (that is, an ordered pair consisting of a sentence letter
Section 4 Translation and Symbolization . 93

and a sentence of English) or an ordered pair of which the first member

is a predicate letter and the second member a formula of English whose
only variable is 'a' and which contains no apparent variables. Again, a
scheme of abbreviation is to be a collection of abbreviations such that no
two abbreviations in the collection have the same first member.
For example, the following is a scheme of abbreviation:

(7) P Moby Dick is a fish

F a is 2L whale
G ^ is a mammal .
The process of literal translation into English on the basis of a given
scheme of abbreviation now begins with a symbolic formula and if suc­
cessful ends with a formula of English. The process consists of the follow­
ing steps:

(f) Restore any parentheses that may have disappeared as a result of

informal conventions,
{it) Replace sentence letters by English sentences in accordance with the
scheme of abbreviation,
{in) Replace each predicate letter by the formula of English with which
it is paired in the scheme of abbreviation, flanking the latter with a pair of
braces, {The result of this step will not in general be a formulay for it will
contain meaningless parts of the form

y >}« .
with <f>a formtda and ol a variable,)
{iv) Replace all parts of the form

y>] a .
where <f> is a formula of English and a a variablcy by the result of re­
placing in <f) all occurrences of the variable 'a by a.
{v) Eliminate sentential connectives and quantifier phrases in favor
of the corresponding phrases of connection and quantityy preserving all
parentheses.

As before, we say that an English formula is a free translation (or simply

a translation) of a symbolic formula <j) on the basis of a given scheme of
abbreviation if it is a stylistic variant of the literal English translation
of (f> based on that scheme.
Consider, for example, the sentence

(8 ) Ax{Fx G:c)
Let us translate it into English on the basis of the scheme (7). In step
(i) of the process of literal translation into English, (8) becomes
94 • ‘A ir and ‘Some* Chapter III

(Ajc(F^ Gx) ^ P) ,
in step (ii)
(Ajc(Fx -> G:v) -> ^ Moby Dick is a fish) ,
in step (iii)
(A;c({iz is a whale} a; -> (tz is a mammal) x) Moby Dick is a fish) ,
in step (iv)
{Ax{x is a whale is a mammal) ^ Moby Dick is a fish) ,
and in step (v)
(9) (If for each x (if x is a whale, then x is a mammal), then it
is not the case that Moby Dick is a fish) .
Thus (9) is the literal translation of (8) into English on the basis of the
scheme (7), and the more idiomatic sentence.
If all whales are mammals, then Moby Dick is not a fish,
being a stylistic variant of (9), qualifies as a free translation of (8) on the
basis of the same scheme.
We say that ^ is a symbolization of a formula \fj of English on the basis
of a given scheme of abbreviation just in case ^ is a symbolic formula
which has i/r as a translation on the basis of that scheme.
To find a symbolization of a given formula of English on the basis of
a given scheme of abbreviation the reader will find it useful to proceed
roughly as follows:

(j) Introduce phrases of quantity and connection^ the latter accompanied

by parentheses and occurring canonically (that is, standing only before or
between formulas), in place of their stylistic variants.
(2) Reverse the steps leading from a symbolic formida to a literal English
translation.

For example, consider the sentence

(10) If a professor is a Communist and all Communists are
subversive, then he is subversive,
together with the scheme of abbreviation
F : a is a professor
G : is a Communist
H : a is subversive .
Some points should be noted before a symbolization is attempted. First,
the indefinite article is often used as a stylistic variant of a phrase of
Section 4 Translation and Symbolization . 95

quantity, and this is true of the first occurrence of ‘a’ in (10). Secondly,
despite the fact that (10) begins with ‘if’, its symbolization should clearly
be not a conditional but a universal generalization of a conditional.
Thirdly, English pronouns often play a role like that of variables in our
symbolic language; this is the case with ‘he’ in (10). These points suggest
that (10) should become in step (i)
For each x (if ((:c is a professor and a: is a Communist)
and for each y (if j is a Communist, then y is subversive)),
then X is subversive) .
Let us now perform step (2) of the process of symbolization. Reversing
step (v) of the process of literal translation into English, we obtain
Kx (((^ is a professor a is a Communist) a hy{y is a Com­
munist j is subversive)) -> jc is subversive) .
Reversing step (iv), this becomes
hx ((({a is a professor} a (a is a Communist} x) a ^y{{ci is a
Communist} y -> {a is subversive} j) ) {a is subver­
sive} x) .
Reversing step (iii), we obtain
f\x (((F:x: a G ^ ) a hy{G y -> Hy)) ^ H;c) .
Finally, reversing step (i) (for step (ii) is irrelevant), we obtain
( ii) A;c(F:c A G;c A hy{Gy -> Hy) -> H;c) .
(The use of two variables, ‘a:’ and ‘jy’, is not necessary for a symbolization
of (10); it could equally well become ‘A;c(F a: a G x a A;c(G a: H jc) ->
H jc)’, which is, however, somewhat less perspicuous than (ii).)
By an argument we now understand a sequence of formulas^ called
\ts premiseSy together with another/omw/«, called its conclusion. A symbolic
argument or English argument is one whose premises and conclusion
are respectively symbolic formulas or formulas of English. A symboliza­
tion of an English argument on the basis of a given scheme of abbreviation
is a symbolic argument whose premises and conclusion are respective sym­
bolizations, on the basis of that scheme, of the premises and conclusion
of the English argument. A symbolic argument is called simply a sytnboliza-
tion of an English argument if there is some scheme of abbreviation on
the basis of which it is a symbolization of that argument.

EXERCISES
On the basis of the scheme of abbreviation,
F : Û is an even number
G : a is a prime number
96 . ‘A ir and ‘Some’ Chapter III

H a is honest
J a is a person
P 2 is a prime number
Q 4 is a prime number
R the son of Lysimachus is honest
translate the following symbolic formulas into idiomatic English.
Exercises 1 4 - 16 are solved for illustration.

14. A;v(J;c A H:c ^ R )

In steps (i) and (ii) of the process of literal translation into English,
No. 14 becomes
A H x) the son of Lysimachus is honest) ,
in steps (iii) and (iv)
A^((.x is a person Axis honest) the son of Lysimachus
is honest) ,
and in step (v)
(12) For each x (if (;c is a person and x is honest), then the son
of Lysimachus is honest) .
Thus (12) is the literal English translation of No. 14, and the following
stylistic variant of (12) is a translation of that symbolic formula into
idiomatic English:
If anyone is honest, then the son of Lysimachus is honest.

15. p ■ Vjc(F:v a G jc)

16. p- ' WxFx A G:v
On the basis of the given scheme, the literal translations into English
of Nos. 15 and 16 are, respectively.
(If 2 is a prime number, then there is an object x such that
(a: is an even number and jc is a prime number))
and
(If 2 is a prime number, then (there is an object x such
that X is an even number and is a prime number)) ;
and the following stylistic variants are respective translations into
idiomatic English of the symbolic formulas:
If 2 is a prime number, then there is an even prime num­
ber.
If 2 is a prime number, then there is an even number and
X is ^ prime number.
Section 4 Translation and Symbolization . 97

To account for the fact that No. 15 has a different translation from
No. 16, it is sufficient to note that the latter, in contrast with the former,
is not a symbolic sentence,

17. Q <-> Ax(Fx -> G:v)

18. V:v(Fa: a G x ) -> a VxG x
19. Q A V;cF;c -> Wx(Fx a G jc)
20. A^(Fa: a G jc -> P)
21. V x {¥ x a G x ) - > F

Symbolize each of the following sentences on the basis of the scheme

of abbreviation that accompanies it. Exercises 22 - 24 are solved
for illustration.

22. Among snakes, only copperheads and rattlers are poisonous.

(F : is a snake; G : a is poisonous; H : a is a copperhead; J : a is
a rattler)
To symbolize No. 22 we must supplement our preceding comment
on ‘only' (p. 92) with an intuitive comprehension of English; this
intuition suggests that the sentence pass in step (i) of the process of
symbolization into either
For each x (if is a snake, then (if x is poisonous, then
{x is a copperhead or ic is a rattler)))
or
For each x (if {x is a snake and x is poisonous), then
(x is a copperhead or ac is a rattler)) .
Let us now perform step (2) of the process of symbolization, using
the first alternative above. Reversing step (v) of the process of literal
translation into English, we obtain
Ax(x is a snake -> (x is poisonous (jc is a copperhead
V jc is a rattler))) ;

reversing steps (iv) and (iii), we obtain

A^(F;r -> {Gx -> {Hx v Joi))) ;
and finally, reversing step (i) (for step (ii) is irrelevant), we obtain
Ajt(F:v -> [Gx -> Hjt v J^]) .

23. Nothing is a dog unless it is an animal. (F : a is a dog; G : a

is an animal)
Intuition, aided by the preceding comments on ‘nothing' (p. 91)
and ‘unless' (p. 43), suggests that the sentence should become in
step (i) of the process of symbolization
98 . ‘A ll’ and ‘Some* Chapter III

For each x (it is not the case that x is 3. dog or x is an

animal) ,
and in step (2)
A:v( ^ F jc V G ac) .
24. No one but a criminal is in a penitentiary. (F : a is in a peni­
tentiary; G : is a criminal)
It will sometimes be our practice (admittedly questionable) to treat
‘no one’, ‘someone’, ‘everyone’ as stylistic variants of ‘nothing’, ‘some­
thing’, ‘everything’. Thus No. 24 may become in step (i) of the
process of symbolization
For each x (if is in a penitentiary, then is a criminal) ,
rather than
For each x (if (;c is a person and is in a penitentiary),
then jc is a criminal) ;
in step (2) it becomes
Ax{Fx -> Gjf) .
25. Drunkards are not admitted. (F : is 3 drunkard; G : a is ad­
a

mitted)
26. None but the brave deserve the fair. (F : a is brave; G : a

deserves the fair)

27. There is a round square. (F : is round; G : is square)
a

28. Something is round and something is square. (F : a is round;

G : a is square)
29. Single women are decorous only if they are chaperoned. (F : a

is a single woman; G : c is decorous; : is chaperoned)

H a

30. Single women are decorous if they are chaperoned. (F : is 3 a

single woman; G : is decorous; H : is chaperoned)

a a

31. Some soldiers love war, but not all who love war are soldiers.
(F : is 3 soldier; G : loves war)
a a

32. If all men are mortal, then Christ is not a man. (F : a is a man;
G : is mortal; P : Christ is a man)
33. Men and women who are over twenty-one are permitted to
vote. (F : a is a man; G : a is a woman; : is over twenty-one;
H a

I : is permitted to vote)
34. Women without husbands are unhappy unless they have
paramours. (F : a is a woman; G : a is without a husband; H : a is
unhappy; I : a has a paramour)
35. If only Republicans support the incumbent and no Democrat
supports the candidate, then if anyone is a Democrat, someone supports
neither the incumbent nor the candidate. (F: a is a Republican; G : a

supports the incumbent; H : is 3 Democrat; I : supports the

a a

candidate)
Section 5 Inference Rules and Forms of Derivation . 99

36. If those who believe in God have immortal souls, then, giv^en
that God exists, they will have eternal bliss. (F : a believes in God;
G : a has an immortal soul; H : a will have eternal bliss; P : God
exists)

5* Inference rules and forms of derivation; theorems with

unabbreviated proofs. We say that a symbolic formula ip comes from a
symbolic formula <f> by proper substitution of a variable for a variable
a if i/f is like (p except for having free occurrences of jS wherever <p has
free occurrences of a. Consider, for example, the formula

(I) Fx A G y -> V;cH;c

The formula
A Gy VjcH jc

comes from (i) by proper substitution of V for ‘x:*, and

Fy A Gy ^ WxFlx

comes from (i) by proper substitution of 'y^ for V .

For the logic of quantifiers we must add three inference rules and a
form of derivation to our original stock.
The rules of universal instantiation and existential generalization lead
respectively from
Aoi(p
to
<!>' .
and from

to
Mx4> ,

where a is a variable, ^ is a symbolic formula, and <p' comes from <p by

proper substitution of some variable for a. In both cases a is called the
variable of generalization and the variable that replaces a the variable
of instantiation. Universal instantiation corresponds to the intuitive
principle that what is true of everything is true of any given thing, and
existential generalization to the principle that what is true of a given
thing is true of something.
For example, in the arguments
AxFx Fy ,
(Fx: A Gx) Vy (Fy a Gy) ,
the conclusion follows from the premise by the respective rules of universal
loo . ‘A ir and ‘Some* Chapter III

instantiation and existential generalization. In the first case the variable

of generalization is and the variable of instantiation in the second
case these roles are reversed.

Now consider the following informal derivation:

(1) Show that if m is even, then w • w is even.
(2) Assume that m is even.
(3) For some ky m = 2 • k,
(4) Let ko be such a k; thus w = 2 • ^o-
(5) Hence m ’ n = {2 ' ko) ' n = 2 • {ko - n),
(6) Therefore m • n is even.

The third new inference rule, called the rule of existential instantiationy
accounts for the intuitive transition from step (3) to step (4), and corre­
sponds to the following principle; what is true of something may be
asserted to hold for some particular object. In general, existential instan­
tiation leads from
Va<^
to
4>' .
where again a is a variable, ^ is a symbolic formula, and <f>' comes from <f>
by proper substitution of some variable for a. To avoid fallacies, a re­
striction on the use of existential instantiation is incorporated into clause
(5) of the directions given below for constructing a derivation: the variable
of instantiation (that is, the variable which replaces a) must be new.
(Without such a restriction we should risk an unjustifiable identification
of the object of which (f> is asserted to hold with objects already men­
tioned in the derivation.)

Diagrammatically the three new rules appear as follows.

Universal instantiation ( U I) : Aa<^

~ T

Existential generalization (E G ): <f>'

Existential instantiation ( E l) : Vac^

In all three cases a is to be a variable, (f> is to be a symbolic formula, and

<f>' is to come from <f) by proper substitution of some variable for a.
Section 5 Inference Rules and Forms of Derivation . loi

The new form of derivation is known as universal derivation and in

its simplest form appears as follows:
Show A<x(f>
XI

Xm >
where <f> occurs unboxed among xi through xm* In a universal derivation
one shows that everything has a certain property by showing that an
arbitrary thing has that property. To ensure arbitrariness a restriction
will be imposed: the variable a, called the variable of generalization, must
not be free in any antecedent line.
This form of derivation is familiar to every student of plane geometry,
wherein one shows, for instance, that every triangle has a certain property
by considering an arbitrary triangle and showing that it has the property
in question.
It is natural to permit a more inclusive form of universal derivation,
in which several variables of generalization occur. Thus universal deriva­
tion appears in general as follows:

Show A a i. . . hcLjc (f>

XI

Xm )

where (f> occurs unboxed among xi through xm- Here one shows that all
objects stand in a certain relation to one another by showing that arbitrary
objects do so. To insure arbitrariness, the following restriction is im­
posed in clause (6) below: the variables of generalization, ai through oljc,
may not be free in antecedent lines.
We shall continue to employ the inference rules of the sentential calculus,
but they must be reconstrued in such a way as to admit as premises and
conclusions any symbolic formulas of appropriate sentential structure
(not merely symbolic sentences). For example, in the argument

(Fx -> Gx) Fx Gx

the conclusion will now be considered to follow from the premises by MP.
We may call that branch of logic which essentially involves quantifiers
as well as sentential connectives the quantifier calculus. That part of the
quantifier calculus which concerns the rather restricted symbolic language
102 . *A11* and *Some* Chapter III

of the present chapter is called the monadic quantifier calculus. The direc­
tions for constructing a derivation within this discipline (which in the
next chapter will be extended only slightly in order to arrive at the full
quantifier calculus) transcend in two ways those given for the sentential
calculus. On the one hand, the earlier procedures are extended so as to
apply not only to symbolic sentences but to arbitrary symbolic formulas;
on the other hand, additional provisions are made for the accommodation
of quantifiers. The directions for constructing a derivation from given
symbolic premises become, then, the following.

(j) I f (f> is any symbolic formula^ then

Show <f>
may occur as a line, {Annotation: 'Assertion',)
{2) Any one of the premises may occur as a line, {Annotation: 'Premise',)
{3) U ^ symbolic formulas such that
Show {(f) -> ip)
occurs as a linCy then <f>may occur as the next line. {Annotation: 'Assumptioti'.)
{4) I f (f) is a symbolic formula such that
Show <f>
occurs as a line, then
^ (j}
may occur as the next line; if <f>is a symbolic formula such that
Show ^ (p

occurs as a line, then <p may occur as the next line, {Annotation: 'Assump­
tion',)
{^a) A symbolic formula may occur as a line if it follows from antecedent
lines by an inference rule of the sentential calculus {that is, M P, M T , DN,
R, S, Adj, Add, M T P , B C , or CB), by UI, or by EG ,
(56) A symbolic formula may occur as a line if it follows from an antecedent
line by E l, provided that the variable of instantiation does not occur in any
preceding line, {The annotation for {5a) and (56) should refer to the inference
rule employed and the numbers of the antecedent lines involved,)
(6) When the following arrangement of lines has appeared:

Show <p
XI

Xm
Section 5 Inference Rules and Forms of Derivation . 103

where none of xi through xm contains uncancelled 'Show' and either

(i) (j> occurs unboxed among xi through xmy
(ii) (f> is of the form
(t/ri ^ ^2)
and ifj2 occurs unboxed among xi through xmy
{iii) for some formula Xy ^oth x cind its negation occur unboxed among
XI through Xmy or
{iv) <f) is of the form
Aai . . . hcLjc ifj ,
ip occurs unboxed among the lines xi through xmy cind the variables
ai through oLfcare not free in lines antecedent to the displayed occur­
rence of
Show (f> ,
then one may simultaneously cancel the displayed occurrence of 'Show' and
box all subsequent lines.

As before, a derivation is complete ]ust in case every line either is boxed

or contains cancelled 'Show', A symbolic formula (p is derivable (in the
quantifier calculus) from given symbolic premises just in case, by using
only clauses (i) through (6), a complete derivation from those premises
can be constructed in v^hich
OL__I
kjtiuw
occurs as an unboxed line. A proof is a derivation from an empty class of
formulas (that is, a derivation employing no premises), and a theorem
is a symbolic formula derivable from an empty class of premises.
We turn next to some theorems and their proofs. For clerical con­
venience, we now assign numbers beginning with 201.
T201 is established by a conditional derivation with subsidiary con­
ditional and universal derivations.
T 201 Show f\x{¥x -> Gx) {hxYx -> A;cG:v)

4, UI
2 ,U I
6, 7, MP

If, in this derivation, line 6 had preceded line 5, we should not have been
104 • ‘AH' and ‘Some' Chapter III

able to complete the universal derivation of line 5; for then the variable
of generalization, ^x\ would have been free in an antecedent line. As a
general practice, if a universal derivation is to be employed, it is advisable
to begin it before using U I.
In the proof of T 201, only one of our new inference rules, UI, was
employed; T202 illustrates the application of E l and EG .
T202 yxYx -^ yxG x)

4, E l
2, U I
5, 6, MP
7, e g

In this derivation the order of lines is again important. If line 6 had

preceded line 5, we should have been unable to choose ‘j ’ as the variable
of instantiation in line 5. As a general practice, it is advisable to use
E l before U I.
Two symbolic formulas (f> and ifj are said to be equivalent if the formula
(f><—> ip
is a theorem. T 2 0 3 -T 2 0 6 provide equivalent expressions for formulas
beginning with various combinations of quantifier phrases and negation
signs.
T203 I, Skew ^ A x F x <r->Wx'^Fx
2. Show ^ AxFx Wx'^ Fx
3- ^AxFx
4- Show Wx- Fx
5-
6. Show AxFx
7- Show Fx
8. ^Fx
9* Vx'^ Fx 8, EG
10. Vx Fx 5. R
1 1. ^AxFx 3. R
12. SkowWx'^Fx -> '^AxFx
13- Vx'^ Fx
14« Show
Section 5 Inference Rules and Forms of Derivation . 105

15-
16. 13. El
17- 15. u i
18. ^AjcF x ^ Vx ^ F jc 2, 12, CB

T204 I. S k e m ^ y x F x < r-> K x ^ Y x

6, EG
3. R

12, E l
10, U I

2, 9, CB

T205 AxFx<r-> '^V x'^Fx

T206 yxFx<r-> Ax'^ F jc

EXERCISES

37. Annotate each line of the following derivation.

I. Show AxFx A yxGx V.v(F:c a G:v)
2. AxFx A
3- NxGx
4- Gy
5- hxFx
6. Fy
7- Fy A Gy
8. y x ( F x A G jc)

38. Prove T205 and T206.

The reader will find careful study of the proofs of T203 and T204
useful in solving exercise 38.
io6 • ‘A ir and ‘Some* Chapter III

6. Abbreviated derivations; theorems with abbreviated proofs.

It is useful at this point to introduce counterparts to the methods for
abbreviating derivations that appeared in chapter II and, in addition,
one new method. As before, the methods of abbreviation are obtained
by supplementing the directions for constructing derivations. The new
clauses, (7)-(9), will satisfy the requirements given on page 58; that is,
they will be theoretically dispensable, and there will be an automatic
procedure for checking the correctness of an abbreviated derivation
(that is, a derivation constructed on the basis of clauses (i)-(9 )), at least
when membership in the class of premises is automatically decidable
and annotations are present.
We must modify slightly the notion of an instance of a sentential
theorem, now that formulas as well as sentences are at hand. By an instance
of a sentential theorem <f>we now understand any symbolic formula obtained
from <f> by replacing sentence letters uniformly by symbolic formulas.
Thus
Q -> (P ^ Q)
has as an instance
F:x: -> {G y -> Fx)
even though the latter is not a sentence.
We introduce now the counterpart to clause (7) of chapter II.

(7) I f (f> is an instance of a theorem of the sentential calculus that has

already been proved^ then <f>may occur as a line. {Annotation as in Chapter II.)

We do not admit at this point instances of theorems of the quantifier

calculus. The relevant notion of an instance is rather complex and will
be reserved for chapter IV.
It is convenient, however, to add to our resources the following derived
rule involving quantifiers, whose various forms correspond to JIa e j'a n d
T204.
Quantifier negation (QN), in four forms:
'•^ A o i(f> Va '^<l> ^Voc<^ Aa '^ (f>

Va 'Aoi(f> Aol
Here a is to be a variable and <f> a symbolic formula. QN corresponds
to the following intuitive principles: to deny that every object satisfies
a given condition is to assert that some object satisfies its negation, and
to deny that there is an object satisfying a given condition is to assert
that every object satisfies its negation.
For example, in each of the arguments
Section 6 Abbreviated Derivations; Theorems with Abbreviated Proofs . 107

Ajc(F jc -> G;c) /. V;c^(F;c -> G jc)

Vx ^ (F jc Gx) /. ^A x(Fx Gx)
Vx(Fx A G^) A x '^ (F a; a G^)
Ax (Fx A G jc) Vjc(F;x: a G^)
the conclusion follows from the premise by QN.
We incorporate QN into our system by adding the following clause
to the directions for constructing a derivation.

(8) A symbolic formula may occur as a line if it follows from an antecedent

line by QN. {Annotation: 'O N \ together with the number of the antecedent
line involved.)

In the quantifier calculus, as in the sentential calculus, it is again

convenient, when constructing a derivation, to compress several steps
into one, omitting some lines which an unabbreviated derivation would
require. As in chapter II, such compression will be allowed only in
certain cases, specifically, only when the omitted steps can be justified on
the basis of clause (2) (premises), (5a) (inference rules other than El),
(7) (instances of previously proved theorems of the sentential calculus),
or (8) (quantifier negation). We legitimize such compressions by clause
(9)—the counterpart to clause (8) of chapter II.

(9) A symbolic fo rm u la m ay occur as a line i f it follow s fro m antecedent

lines by a succession o f stepsy a n d each interm ediate step can be ju stified by
one o f clauses (2), (5a), (7), or {8 ). {T h e annotation should determ ine u n ­
ambiguously the succession o f steps leading to the line in question. T his can
be done by indicating^ in order o f applicationy the antecedent lineSy the p r e -
miseSy the inference ruleSy a n d the p reviously p ro v e d theorem s em ployed.
AlsOy in connection with the rules Addy C/7, a n d EGy the a d d ed disjuncty
the variable o f instantiationy a n d the variable o f gen eraliza tion should r e ­
spectively be indicated w henever there is a chance o f am bigu ity ; a n d when
an instance o f a previously p ro v ed theorem is involvedy the releva nt replace-
menty i f not obviouSy should be indicated. T h e special annotations
and 'C D \ which were introduced on pages 64 and 65 to indicate a certain
kind of compressiony may again be used.)

For example, clause (9) provides the justification for line 4 in the follow­
ing abbreviated derivation:

2 ,Q N ,U I ,T i 8, MP
io8 . ‘A ir and ‘Some* Chapter III

The omitted lines are, in order:

(i) A x'^ F x 2, QN
(ii) ^ F x (i), u i
(iii) Fx (F.v - G a’) T i8
Line 4 follows from (ii) and (iii) by MP.
Another illustration of the application of clause (9) is provided by
the derivation
Show- AxFx -> Va (Fa -> G a)

2, QN, E l
3 , T i8 ,M P , EG

(The reader should note that the omission of line 3 would not be justified
by clause (9); clause (5b), pertaining to E l, is not listed as one of the
clauses by which an omitted step may be justified.)
The reader should recall that abbreviated derivations, unlike unabbre­
viated derivations, require as an essential part their annotations—at least
those given in connection with clause (9). Without an indication of omit­
ted steps, it would be impossible to give an automatic procedure for
checking the correctness of an abbreviated derivation. (See the remarks
on page 80.)
T 201 and T202 are called distributiofi laws for quantifiers; we state
now some additional distribution laws.

T207 Show Va(Fa V G a)<->Va F a v Va G a

Show Va(Fa V G a) ->Va F a v Va G a
3 Va(Fa V G a)
4 Show Va F a Va G a

5 Va F a
6 Aa '^Fa 5, QN
7 Fy V Gy 3> E l
8 Va G a e[ U I, 7, M TP,
EG
9 Va F a v Va G a 4, CD
10 S how Va F a v Va G a -> Va(Fa v G a )
11 ShewWxFx -> Va (Fa v G a)
12 Va F a
13 Fz 12, E l
14 Va (Fa V G a) 13, Add, E G

15 S k o w W x G x - ^ y x ( F x V G a)
Section 6 Abbreviated Derivations; Theorems with Abbreviated Proofs . 109

16.
17- 16, E l
18. 17, Add, EG
19. VxFx V V.vG.v^V.v(F a- v G.v) II, 15, SC
20. V.r(pA* V G.v) VxF.v V V.yG a' 2, 10, CB

It is often convenient to move in an initial negation sign by means of QN,

as in the transition from line 5 to line 6 in the proof of T207.

T208 Ал(Гл' Л G.v) <-> AvF.v л AvG.v

T209 V.v(F.v Л G.v) -> VxFx A WxGx
T 210 AxFx V A.vG.v -> Av(Fл' v G.v)
T 2 11 (V.vFvV V.vG.y) -> V.v(F.v Gx)
T 2 12 {AxFx -> A.vG.v) -> V.y(F.v -> G.v)
T 2 13 A.v(F.v G.v) -> (A.vF.v <-> A.vG.v)
T 2 14 Av(F.v <-> G.v) (V.vF.v <-> V.vG.v)

(None of the biconditionals corresponding to T 201, T202, T 2 0 9 -T 2 14

can be proved; see section 9.)
The intuitive plausibility of logical laws can often be seen by translation
into English. For example, on the basis of the scheme,

F ¿7 is a man
G a is a. mortal

the distribution laws T201 and T202 can be translated, respectively, into

If all men are mortal, then if everything is a man, then

everything is mortal
and
If all men are mortal, then if something is a man, then
something is mortal.

T 2 15 - T 2 2 2 are called confinement laws. According to these theorems,

a generalization of a conjunction, disjunction, or conditional in which only
one component contains the relevant variable is equivalent to a formula
in which the quantifier phrase is confined to that component. Note that
when the relevant component is the antecedent of a conditional (T221
and T222), the quantifier phrase must be changed from universal to
existential, or conversely.
no . ‘All* and ‘Some’ Chapter III

T 2 15 1 . Shew- hx{? A F jc) P A AA;Fa;

2.
3
4
5 3> U I, s
6 3, U I, S, 4, Adj
7
8
9
10 8, S, S, U I, Adj

11 2, 7, CB
T216 V^(P A F;c) P A Va;F^
T 2 17 Aji(P V F;k)<-)• P V AjcFjc
T 2 18 Va:(P V F;v) ^ P V Va;Fii
T 2 19 Ajf(P Fa:) (P -> Aa:Fa:)
T220 Va:(P ->Fa:)<->(P ->Va:Fa:)
T 2 21 Aa(Fa: P)<-> (Va:Fjc ->■ P)
T222 1. Show Va:(Fa: P) (AaFa: P)
2. Shew Vx(Fx ^ P) -> (Aa:Fa: -> P)
3 Va:(Fjc -> P)
4 Show AaFa: -> P
5 Aa:Fa:
6 Fy P 3. El
7 P 5, U I, 6, M P

8 Shew- ~ AaFa: -> Va:(Fa: ^ P)

9 ~ Aa;Fa:
10 ~ F2: 9. QN, E l
11 Va:(Fa: -> P) 10, T 18 , MP,
EG
12 Show P -> Va:(Fa: ^ P)
13 P
14 Va:(Fa: P) 13, T 2, MP, E G
15 (Aa:Fa: P) ^ Va:(Fa: -> P) 8, 12, Adj, T 35,
BC, M P
16 Va:(Fa: ^ P) (Aa:Fa: ^ P) 2, 15, CB
Section 6 Abbreviated Derivations; Theorems with Abbreviated Proofs . ii t

In idiomatic English, counterparts to unconfined quantifier phrases

seldom occur. On the basis, however, of a suitable scheme of abbreviation,
the two constituents of T 221 can be translated into
If anyone is honest, then the son of Lysimachus is honest
and
If someone is honest, then the son of Lysimachus is honest.
There are no simple confinement laws for biconditionals. We can say
no more than is expressed in the following four theorems. In no case
is the converse also a theorem; this is substantiated in section 9.

T223 Ax(Fx P) ^ (AxF.v<-> P)

T224 Ax(Fx P) ^ {WxFx <-> P)

T225 (VxFx ^ P) -> Wx{Fx <-> P)

T226 Show (AxFx P) -> Vx(F.v P)

Show- AxFx Л P -> V.r(F.v P)

3 AxFx Л P
4 Fx 3. S, UI
5 Fx a F 3, S, 4, Adj
6 Fx ^ P 5, T84, MP
7 Vx(Fx P) 6, EG
8 Show ~ AxFx Л ~ P -> Vx(Fx >P)
9 Ax F x a ~ P
10 ,F y 9, S, QN, E l
11 'FjyA ~ P 9> -^4}
12 F j< -> P i i ,T 8 5 , MP

13 Vx(Fx P) 12, E G
14 (A.vF.v P) Vx(Fx <-> P) 2, 8, Adj, T86,
BC, MP
T227 and T228 are the laws of vacuous quantificationÿ by T227 the
seemingly meaningless formula ‘A.vP’ is equivalent to ‘P’.
T227 1 . Show AxP •
2. Show A.vP
3- AxP
P 3 .U I
4-
5- Show P -> AxP
6. I P I
112 . ‘A ll’ and ‘Some’ Chapter IH

7- Skew hxŸ
8. 6, R

AxP <-> P 2, 5, CB

T228 I. Shot» VarP P

o \/vP

3- V^P
4- P 3. E l
5- Shoto P -> V^P
6. P
7- V^P 6, E G
8. V;cP 2, 5, CB

The application of U I, E l, and E G in the proofs of T227 and T228 may

seem suspect. For instance, may ‘Va;P’ and ‘P ’ be taken as
Va<^
and <f>' in the formulation on page 99? Yes; for examination of the quan­
tifier rules will reveal that a need not occur in (f>, and that if it does not,
<f>' is (f>.
T229 and T230 have a slightly paradoxical character. Their validity
depends on the fact that ‘x ’ is not free in ‘VxFx’ or ‘AxFx’, just as the
validity of the confinement laws depends on the fact that ‘x’ is not free
in ‘P ’ .
T229 I. Show Vjc(VjcF;c -> Fx)
2. ^ VxQ/xFx -> Fx)
3- {WxFx -> F^) 2, QN, UI
4- ¥y 3, T 2 1, MP, E l
5- ^ {VxFx -> F^) 2, QN, UI
6. ^ Fy 5, T22, MP

T230 I. Shew-Wx{Fx ->AxFx)

2. ^ V;c(F;c AxFx)
3- (F a: ^ Ax F a:) 2, QN, UI
4- Yy 3, T22, MP, QN, E l
5- (Fy AxFx) 2, QN, UI
6. Fy 5. T 2 1, MP
T 2 31 and T 232 are called lazes of alphabetic variance (for bound vari­
ables). They reflect the fact that a generalization such as ‘Everything
is material’ may pass, on the basis of a suitable scheme of abbreviation.
Section 6 Abbreviated Derivations; Theorems with Abbreviated Proofs . 1 1 3

into either or 'A yFy' ; the choice of variable is unimportant. Detailed

consideration will be given to these laws in the next chapter.

T 231 I . Show AxFx <-> AyFy

2.
3-
4-
5- 3, UI
6.
7*
8.
9- 7, U I
10. 2, 6, CB

T232 1 . S how WxFx <-> yyF y

2. Skew-WxFx yyFy
3- yxFx
4- Fz 3, E l
5- yyF y 4, E G
6. Skow^yyFy yxFx
7- y y F y

8. Fw 7, El
9* yxF x 8, E G
10. yxF x y yF y 2, 6, CB

Any instance, as well as any universal generalization of an instance,

of a theorem of the sentential calculus will clearly be a theorem of the
quantifier calculus. Further, certain ways of distributing the universal
quantifier through such theorems will again lead to theorems. T 233 - T237
are examples. (T235 corresponds to the Aristotelian syllogism in Barbara,)
Further examples are left to the imagination of the reader.
T233 (F jc ^ G;c) A (G jc -> H^) -> (Fx -> H jc)

T234 Ax[{Fx -> G^) A {G x -> Hx) -> (Fx H^)]

T235 Ax{Fx -> G a:) a Ajc(G a: -> H.x) Ax(Fx -> H jc)

T236 A:v(F jc <r-^ G jc) a Aa:(G^ H jc) -> Aa:(F;v <-> Hx)
T237 I. ShewAx(Fx -> G jc) a Ajc(F jc H^) -> Ax‘( F jc -> G x a H jc)
1 14 . ‘A ll’ and ‘Some’ Chapter III

5
6 2, S
7 6, UI
8 5, 7. MP
9 2, S
10 9. UI
11 5, 10, MP
12 8, II, Adj

We list eleven more theorems involving quantifiers.

T238 Ax F a' -> VvFv

T239 AvFv A VvGv Vjc(F.v a G.y)

T240 A.v(F.v -> Gv) A V.v (F.v a H v ) Vv(Gx a H x)

T 24I Ax(Fx G x VHx) ^ Ax(Fx -> Gx) v Vx(Fx a H x)

T242 ~ Ax(Fx ^ Gx) Vx(F.v a ~ Gx)

T243 ~ Vx(Fx A Gx) Ax(Fx -> ~ Gx)

T244 ~ VxFx -> Ax(Fx G.v)

T245 VxFx Ax(Fx -> Gx) a A x (F x -> ~ Gx)

T246 AxFx AxAy(Fx A F j)

T247 VxFx A VxGx (Ax[Fx ^ H.x] a A x [G x -> Jx] <->

Ax Ay[Fx A Gy Hx a Jjy])

T248 (V:vF:v <-> WxGx) a AxAy{Fx a G y -> [H.v Jjy]) ->

(Ajc[F x ^ Hx] A^[G^ -> J a:])

EXERCISES
39. Fill in the lines that have been omitted from the following
abbreviated derivation.
I. Show-Ax(Fx -> Gx v Hx) a ^ A.v(F.v -> G^) ->
V;c(Fjc a H;c)
Section 6 Abbreviated Derivations; Theorems with Abbreviated Proofs . 115

A.v(Fjc- ^Gx V Нл:) Л Ax{Fx Gvv)

^Gy) 2, S, QX,
El
4. Wx{¥x Л Ux) 3, T21, MP,
3, T22, MP,
2, S, UI,
MP, MTP,
Adj, EG

(The reader should note that clause (9) would not justify the omission of
line 3. This derivation constitutes, essentially, a proof of T241.)

40. Annotate the following abbreviated derivation.

1 Show ^ Ул’(Ел: Л Gx) Лл(Ел: -> Gx)
2
3
4
5
6
7
8
9
10
11
12
The reader should note the order of steps in this derivation, specif­
ically, that a^ universal derivation is started before UI is employed,
and that application of El precedes that of UI. (This derivation is
a proof of T243.)

41. Prove T 20 8 -T 214.

42. Prove T218, T219, T221.
43. Prove T 223-T 225.
44. Prove T240, T242, T245.
45. Prove T246 - T248.

In solving exercises 4 1 - 4 5 (as well as later exercises) the reader will

find the following informal suggestions helpful (though again not
infallible).

(i) To derive a formula

ф уф
ii6 . ‘A ir and ‘Some’ Chapter III

derive first

and

and then use SC.

(2) To derive

derive first

and
>P^X .
and then use Adj, T35, BC, MP. {See lines 8 - 1 5 0 / the proof of T222,
p. n o , for an application of this suggestion.)
(j) To derive
(4> ^ 'f> )-*x
derive first

and
A ~ ^->X .

and then use Adj, T86, BC, MP. {See the proof of T226, p. i l l , for
an application of this suggestion.)
{¿¡) To derive

where is neither a disjunction ^ a conditional, nor a biconditional, use

conditional derivation.
(5) To derive a conjunction, derive first both conjuncts, and then use
Adj.
(6) To derive
(j> w ip
derive first
^ <j) —>• ijj ,
and then use CD.
(7) To derive a biconditional, derive first the two corresponding con-
ditionals, and then use CB.
{8) To derive a formula
Aoi<l> ,
use universal derivation.
Section 7 Arguments . 117

(9) To derive
Va^ ,
either derive first (f>\ where <f>' comes from <f>by proper substitution of some
variable for <Xy and then use EG, or else use indirect derivation, {See lines
S - 13 of the proof of T226y p. i l l yfor an application of thefirst alterna­
tive and the proof of T'229, p, ii2y for an application of the second,)
(10) To derive anything elsey either use indirect derivation or separate
cases.
The reader should also review the remarks accompanying T201
(p. 103), T202 (p. 104), and T207 (p. 108).

7. Argum ents. Let us pass now to derivations involving premises.

As before, we call a symbolic argument valid if its conclusion is derivable
from its premises, and an English argument valid (in the monadic quanti­
fier calculus) if it has a valid symbolization.
Consider, for example, the argument (i) of page 85:
All Communists are Marxists. Some Communists are
American. .*. Some Marxists are American.
Its validity is established by the following symbolization and derivation
hx{¥x -> G ^ ) . Vjc(F x a H x) /. Vjc(G jc a H^)
I. Shew-yx{Gx K H:v)
2nd premise, E l
I S t premise, U I

2, S, 3, M P
2, S, 4, Adj, E G

EXERCISES
For each of the following arguments, derive its conclusion from its
premises. (The reader will again find useful the suggestions made on
pages 1 1 5 - 1 7 . )
46. t\y{¥x A Gjv) F^ A Gx
47. F^ A G^ .*. Ay(Fx A Gy)
48. Aa;(F:v AxGx) . Aa:(G^ V Hx) -> Axjx /. Ax(Fx J^)
49. V;v(F a; a G^) . A;c(F;c H.v)
Ax{]x A lx ^ F jc) Wx{Hx A G^) ^ A;c(Ijc 'Hx)
Aa;(J.\: I:c)
50. Ax{Fx -> Gx y H:v) . Ax{Gx y Hx lx) .
^Wx{lx A Gx) . r>.\/xFx-^WxGx . \ W x (F x a H x )
ii8 . ‘A ir and ‘Some* Chapter 111

51. h x ( ^ x G a;) V Va:(Fjc a H x) . -> v ^ H x)

A x { F x -> I jc A ] x ) . VxFx Wx{Gx a F a;)

52. A a:(Fa: A jcG a:) . A a; (G a: v H a:) -> V a;(Ga; a I:v) .

V a:Ia: A j^Qa: -> K a:) A:v(F a; a J a: -> K a;)
53. V;c(F^ P) Va:(P F a:) /. Va:(Fa; P)

Show the following arguments valid by constructing symbolizations

and deriving the conclusions of the symbolizations from their premises.
Indicate in each case the scheme of abbreviation used. (Before sym­
bolizing, the reader should review the remarks made in connection
with the solution of exercise 19 of chapter I, p. 29. Exercises
54 and 55 substantiate claims made on page 92.)
54. For each a: (a; is a man and x is mortal). For each a;, a; is a
man.
55. There is an object x such that a; is a dog. .*. There is an object
X such that (if a; is a cat, then a; is a dog).
56. A gentleman does not prefer blondes only if he is blond.
Every gentleman either is blond or prefers blondes.
57. There is not a single Communist who either likes logic or is
able to construct derivations correctly. Some impartial seekers of
truth are Communists. Anyone who is not able to construct derivations
correctly eschews Philosophy. Some impartial seekers of truth
eschew Philosophy.
58. Everyone who signed the loyalty oath is an honest citizen. If
someone signed the loyalty oath and was convicted of perjury, or
signed the loyalty oath and is a Communist, then not all who signed
the loyalty oath are honest citizens. No one who signed the
loyalty oath is a Communist.
59. All men who have either a sense of humor or the spirit of
adventure seek the company of women. Anyone who seeks the com­
pany of women and has the spirit of adventure finds life exciting.
Whoever gives in to temptation has the spirit of adventure. .*. Every
man who gives in to temptation finds life exciting.
60. No egghead is a good security risk. Every professor lives
in an ivory tower. If there is someone who lives in an ivory tower
and is not a good security risk, then no one who is either a professor
or an egghead should be trusted with confidential information. If
some professor is an egghead, then no professor should be trusted
with confidential information.
61. No one who is either a skeptic or an atheist hates God. (For
who can hate that which he doubts or believes nonexistent?) Everyone
is such that if he is an atheist only if he will not go to heaven, then he
is a skeptic and hates God. Everyone will go to heaven.

8. * F allacie s. In stating rules of inference and directions for con­

structing derivations, we have again imposed a number of restrictions whose
significance is perhaps not immediately obvious. The restrictions are all
Section 8 Fallacies . 119

introduced in order to prevent the validation of false English arguments

(arguments whose premises are true sentences of English and whose
conclusions are false sentences of English), that is to say, in order to
prevent fallacies.
Of the restrictions not pertaining to variables only one needs to be
mentioned here. (For the others, see section 4 of chapter I.) According
to clauses (3) and (4) of the directions for constructing a derivation, an
assumption may be introduced only in the line immediately following the
inception of the derivation to which it is relevant. The following false
argument, symbolization, and derivation establish the need for this
restriction.
Argument:
Something is wise. Everything is wise.
Symbolization :
V^:F.r *. hx¥x
Derivation :
1. Shew-hxVx
2. Shew Y x
3* S kew Ax ^ ¥x
4* ' F.V Fallacious assumption
for indirect derivation
5* Wx¥x Premise
6. ¥ y 5, El
^¥y 3, U I

In this derivation line 4 is incorrect; if line 2 is to begin an indirect deri­

vation, its assumption must occur as the next line. The reader can easily
construct for himself a similar example involving a fallacious assumption
for conditional derivation.
The restriction imposed on the variable of instantiation of E l, that it
be entirely new to the derivation, is more stringent than necessary. It
would be sufficient, but tediously complex, to impose the following
three prohibitions: (i) the variable of instantiation may not occur free in an
antecedent line; (ii) it may not occur bound in preceding lines containing
uncancelled ^Show'\ and (iii) it may not occur free in preceding lines
containing uncancelled *Show\ The following argument, symbolization,
and derivation show the need for (i), and exercises 62 and 63 the need
for (ii) and (iii).
Argument :
120 . ‘A ir and ‘Some* Chapter III

There is an even number. There is an odd number.

Something is both an even number and an odd number.
Symbolization:
WxYx . yxG x / . yx{¥x a G :v)

Derivation:
Wx(Fx A G jc)
Premise
2, E l
Premise
4, E l
3> 5» Adj, E G
In this derivation line 5 is fallaciously inferred from the second premise.
The variable of instantiation, 'y\ is free in line 3, which is antecedent
to line 5.
The restriction imposed on the variable of generalization in a universal
derivation prohibits a free occurrence of that variable in lines antecedent
to that derivation. The need for this restriction is shown by the following
argument, symbolization, and derivation.

Argument:
There is an odd number. /. Everything is an odd number.
Symbolization:
WxFx AxFx
Derivation:
I. S hew A xFx
2. WxYx Premise
3- Fy 2, E l
4- /^yFy
5- 3. R
6. Far 4, U I
In this derivation the cancellation of 'Show' in line 4, along with the
boxing of line 5, is incorrect. The variable of generalization in the sub­
sidiary universal derivation, 'y\ is free in line 3, which is antecedent to
the subsidiary derivation.
The notion of proper substitution is involved in the formulation of each
of the rules U I, EG , and E l. In order for ifj to come from a formula <f> by
proper substitution of a variable j3 for a variable a, two conditions must
Section 8 Fallacies . 121

be fulfilled: (i) every free occurrence of a must be replaced by an occurrence

of j3, and (ii) each of the occurrences of /3 that replaces an occurrence of a
must be free. The false argument (i) below shows the need for (i)
with respect to U I; exercises 64 and 65, the need for (i) with respect to
EG and E l ; the false argument (2), the need for (ii) with respect to E G ;
and exercise 66, the need for (ii) with respect to U I. The requirement
for E l that the variable of instantiation be new already makes impossible
a legitimate application of that rule in which a free occurrence of a is
replaced by a bound occurrence of jS.

Argument:
(i) All dogs are animals. There is a dog. Everything is an
animal.

Symbolization:
A;c(F jc Gx) VxFx /. AxGx
Derivation:

I. Skew- AxGx
1

2. VxFx Premise
3- Fy 2, E l
4- Ax(Fx G jc) Premise
5- ( F j -> G jc) 4, incorrect application of U I
6. G jc 3, 5, MP

In this derivation line 5 comes from line 4 incorrectly; not all free occur­
rences of 'x' in \ F x -> G:c)* have been replaced by 'y\

Argument:
(2) There is a human. /. Everything is human.
Symbolization:
WxFx /. AxFx
Derivation:

1. Skew-AxFx
2. Shew-Ax{Fx Fx)
Fx<-^ Fx T91
4- WyAx(Fx Fy) 2, incorrect application of EG
5- Ax(Fx<-> Fz) 4, El
6. WxFx Premise
122 . ‘Alt’ and ‘Some’ Chapter III

7- Fît) 6, E l
8. 5, U I, BC, 7, MP
F;c 5, U I, BC, 8, MP

In this derivation line 4 comes from line 2 incorrectly; the occurrence

of that replaces the free occurrence of y in ‘A:v(F:)c -> FyY is not free.
It might appear that fallacies can occur even if all the restrictions im­
posed in this chapter on the process of derivation are observed, for consider
the argument
'a* is the first letter of the alphabet, is the first letter
of the alphabet .
This argument is certainly false, and can be considered (making allowance
for vacuous quantification) a stylistic variant of
is the first letter of the alphabet. There is an object
.V such that 'x' is the first letter of the alphabet .
Further, on the basis of the scheme
(3) F : 'a' is the first letter of the alphabet ,
the latter argument seems to be a translation of the valid symbolic argu­
ment
Fa WxFx .
The fault in this line of reasoning is that (3) is not a scheme of abbreviation :
the formula
is the first letter of the alphabet
contains an apparent variable, as was observed on page 92.

EX ER CISES
For each of the following false arguments annotate the derivation
accompanying its symbolization, indicating the fallacious step and
the error that led to the fallacy.

62. Something is wise. Everything is wise.

Symbolization:
WyFy .*. A^F.v
Derivation:
I. Skew-AxFx
WyFy
Fx
Section 8 Fallacies . 123

63. There is a human. Everything is human.

Symbolization:
WxFx AxFx
Derivation:
Skow-AxFx

64. There is an even number. There is an odd number, Some-

thing is both an even number and an odd number.
Symbolization:
WxFx V^G.v V;c(Fa: a G:v)
Derivation:
I. ^A^V^(F^ A (
2. WxFx
3- Fy
4- V^G.v
5- Fz
6. F j A Far
7- VyiFj A Gy)
8. Fv) A Gzo
9- Va:(F« a G;*)

65. There is a prime number. /, Everything is a number.

Symbolization:
V:v(F;c a G.v) /. A:vG a:
Derivation:
I. Skew-AxGx
2. Wx{Fx A G:v)
3- Fy A Gx
4- Gx
124 • ‘All* and ‘Some* Chapter III

66. Something is human. Something is not human, Something

is such that it is human just in case it is not human.
Symbolization:
WxFx V.v(F.v^ ^F.v)
Derivation:
1 Ske%&Wx(Fx <-> Fx)
2 Skme-AyWx(Fx<r-> ^ F j)
3 ^ AjyV.v(F.v <-► Fy)
4 WxFx
5 Fz
6 V x^Fx
7 'x^Fw
8 '^Vx(Fx <r^ ^ F m)
9 Ax ^ (F jc F m)
10
11 F«;
12 Wx{Fx<-^ ~ F jc)

67. It is not the case that everything is male, /. Every male is

female.
Symbolization:
^AaF jc Ajc(Fa: Gx)
Derivation:
1. Skew-Ax{Fx -> G.v)
2. S how F.V -> G.V
F.V
A.tF vV
A^Fa: -> G.v
F. v -> G.v
G. v

(In connection with this exercise the reader should review the com­
ments on parentheses in section 3, p. 90.)

9. * Invalidity. In chapter II truth tables were used to show arguments

invalid. The technique employed there can be extended to a method
for showing invalidity within the present symbolic language. We shall
call the new method that of truth-functional expansions.
Suppose we are presented with a symbolic argument whose premises
and conclusion are sentences. To obtain a truth-functional expansion of
such an argument, proceed as follows:
Section 9 Invalidity . 125

1. Choose a sequence of distinct variables, say ai, . . . , a^, which do

not occur in the given argument.
2. Throughout the argument replace each occurrence of a formula
A0L<f>
or
v«^ ,
where a is a variable and ^ a formula, by
^1 A . . . A <f>n

or
V . . . V

respectively, where comes from (f) by proper substitution of ai for a,

(f>2 comes from <f) by proper substitution of 0L2 for a, and so on.
3. Throughout the resulting argument replace formulas of the form
77 a ,
where tt is a predicate letter and a is a variable, by sentence letters not
occurring in the argument. The replacement must be biiinique; that
is, distinct formulas are to pass into distinct sentence letters.
For example, consider the argument

(I) Va;F^ ^ VjcGx /. Ajc(F jc -> G jc)

In the first step toward obtaining a truth-functional expansion, we may
choose the variables 'a\ 'b\ In the second step, the argument (i) becomes
F^ V F6 Ga V Gb (Fa -> Ga) a {Fb -> Gb)
and in the third step,
(2) P V Q -> R V S (P -> R) A (Q S) .
The argument (i) has, of course, truth-functional expansions other
than (2). The significant differences from (2) would result from a choice
of more or fewer variables in step i.
It happens that a symbolic argument whose premises and conclusion
are sentences is valid if and only if, in each of its truth-functional expan­
sions, the premises tautologically imply the conclusion. Thus to show
such an argument invalid it is sufficient to exhibit a truth-functional
expansion of it, together with a truth table showing that the premises
of that expansion do not tautologically imply its conclusion.
For instance, the argument (i) is seen to be invalid in view of the
truth-functional expansion (2) and the line of a truth table for (2) that
follows:
126 . ‘A ir and ‘Some’ Chapter III

Q R P vQ Rv S R P V Q-> (P -> R) A
RvS ( Q -> S )

With different truth-functional expansions of a given argument, the

test for tautological implication will, of course, in general lead to different
results. It happens, however, that if we obtain tautological implication
for a truth-functional expansion reached by choosing 2^ variables in step
I , where n is the number of different predicates and sentence letters

appearing in the given argument, then that argument is valid. Thus the
method of truth-functional expansions provides not only a means of
showing invalidity but also an automatic test of validity for arguments
of the kind under consideration.
The method given above is directly applicable only to those symbolic
arguments whose premises and conclusions are sentences. We can,
however, extend the method to arbitrary symbolic arguments.
We say that a formula ^ is a closure of a formula <f>just in case ^ is a
sentence and either ^ is or ^ is a universal generalization of <f>; and we
understand by a closure of an argument any argument obtained from it
by replacing each formula comprised in it by a closure of that formula.
It happens that a symbolic argument is valid just in case each of its
closures is valid, and furthermore that one closure of it is valid if and only
if any other is. Thus to test an arbitrary symbolic argument for validity
it is sufficient to construct a closure of that argument and to test the closure
for validity.
Another method of showing invalidity is suggested by the discussion
of fallacies in the preceding section. It happens that if a symbolic argument
has a false English translation it cannot be valid. The method of showing
invalidity based on this principle, unlike the method of truth-functional
expansions, does not directly give rise to an automatic test for validity
and, in addition, suffers from several drawbacks that will be discussed in
chapter IV.

EXERC ISES
Using the method of truth-functional expansions show the following
arguments invalid. (From the invalidity of (i) on page 125 and exercises
68 -78 it follows that the converses of T201, T202, T 209 -T 214,
T223 - T226 are not theorems.)
68. (AxFx -> hxGx) -> f\x{^x -> Gx)
69. /, Wx¥x A VjcGiV -> Vjî(Fa: a G ac)
70. Ax(Fx V Gx) -> (AxFx v AxGx)
Section 9 Invalidity . 127

71 /. V:v(F^ G;c) (Va:Fjc VjcG at)

72 Vx{Fx -> G a;) {AxFx Aa:G;c)

73 {AxFx <-> A:x;G:v) -> Ax(Fjc <-> G^)

74 (V.yF;v ^ V;vG:v) ^ A:v(F:v <-> G.v)

75 /. {AxFx P) -> Aa:(Fx ^ P)

76 (VxF jc <-> P) -> Ax(Fx <-> P)

77 /. Vx{Fx <-> P) {WxFx <-> P)

78 V^(F^ <-> P) -> {AxFx <-> P)

For each of the following arguments either derive the conclusion

from the premises or show it invalid by the method of truth-func­
tional expansions.
79. A;c(F:r -> P) AxFx P
80. A:vF:v P /. Ax(Fx P)
81. Wx{Fx->F) WxFx->F
82. WxFx -> P Wx{Fx -> P)
83. A;cF:v Fx
84. Fx AxFx
85. ,\Fx<r-^AxFx
86. Ax{Fx Fx y K;c) . Va;(G:v a a .
Ax{ L x l x ) . Vjc(K;c a J jc) . A;c(K;c -> I jc) .
Ax{ r>^Fx y ^ Gx ^ L x A K;c) /. Wx{Lx a Fix)

Give an interesting symbolization of each of the following argu­

ments, and for each symbolization either derive the conclusion from
the premises or show it invalid by the method of truth-functional ex­
pansions. (In this connection the reader should review the remarks on
page 77, preceding exercise 53 of chapter II.)
87. All men are mortal. .*. Some men are mortal.
88. All dogs are animals. All mammals are animals. All dogs are
mammals.
89. Dogs are expensive if and only if they are pedigreed. If
there is a dog, then something is expensive if and only if something
is pedigreed.
90. If none but the gracious deserve the respect of their countrymen,
then Coriolanus deserved his fate. Only the magnanimous deserve
the respect of their countrymen. All who are magnanimous are gracious.
Coriolanus deserved his fate.
128 . ‘All* and ‘Some* Chapter III

91. If none but the gracious deserve the respect of their countrymen,
then Coriolanus deserved his fate. All 'who are magnanimous deserve
the respect of their countrymen. Only the gracious are magnanimous.
/. Coriolanus deserved his fate.

10. * Historical remarks. What we call the quantifier calculus has

also been called the functional calculus, the predicate calculus, the first-
order functional calculus, the lower predicate calculus, and the restricted
predicate calculus.
The essential ideas of the quantifier calculus occur first in Frege [1].
The quantifier calculus was extensively developed, though in a rather
informal way, in Whitehead and Russell [i] and received its first com­
pletely explicit formulation, containing one error corrected in a later
edition and another corrected in Pager [i], in Hilbert and Ackermann [1].
The symbols used here are due to Tarski. The quantificational symbols
most frequent in the literature are those of Whitehead and Russell [i]:
\ x y for ‘Ajc’
\ 3 xy for ‘Vjc’ .
The first automatic test of validity for the monadic quantifier calculus
was given in Behmann [1]. The present test is based on a result in Bernays
and Schonfinkel [1].
Our formulation of the quantifier calculus admits of one simplification
in addition to those indicated in connection with the sentential calculus.
The general form of universal derivation may be replaced by the simple
form (see p. loi).

11. Appendix: list of theorems of chapter III.

L AWS OF distribution :
T 20I Ajc(F jc -> G jc) -> (A:vF a: -> Ax G jc)
T202 Ajc(F jc -> G jc) (V.vF a: V^G a:)
T207 Wx(Fx V G jc) <-> WxFx V WxGx
T 20 8 Ax(Fx A G jc) <-> AxFx a A;cG;c
T 20 9 V;c(F.r A G^) -> WxFx A WxGx
T2IO AxFx V AxGx Ax(Fx v G jc)
T2II (V^F:r -> Va:G^) V:c(F jc G:c)
T 2I2 (AxFx -> Aa:G a:) -> Va:(F:v G;c)
T2I3 A^(Fa: Gx) {AxFx <-> A:rG:v)
T2I4 Ax(F;c <-> G^) (VjcF.r WxGx)
Section II Appendix: List of Theorems of Chapter I I I . 129

L A W S OF Q U A N T I F I E R N E G A T I O N :

T203 AxFx ^-^Vx'^Fx

T204 ^ WxFx Ax <^Fx
T205 AxFx<r^ ^Wx^^Fx
T206 WxFx<-^ '^A x'^Fx

L A W S OF c o n f i n e m e n t :

T 2 15 A:v(P A Fx) <-> P A AxFx

T 216 V.y (P a F a:) <-> P a Va:F a:

T 2 17 A a’(P V F a:) <-> P v A a:Fa:

T 218 V a’(P V F a:) <-> P v Va:F a:

T 219 A a’(P -> F a:) -<-> (P -> A x F a:)

T22O V a^(P F a:) ^ (P -> V a:F a:)

T 221 A a:(Fa: -^ F ) <r-> {VxFx -> P)

T222 V a:(Fa: -> P) <-> (A a:Fa: -> P)

T223 A.t (F a: P) (A a:Fa: <-> P)

T224 A a:(Fa: P) -> (Va^Fa: <-> P)

T225 (Va:F a: <-> P) -> V a (F a: P)

T226 (A a:Fa ^ P) ^ V a (F a: P)

L A W S OF V A C U O U S Q U A N T I F I C A '

T227 A a P <-> P

T228 Va P < - > P

T229 Va:(Va F a: -> F a:)

T23O Va:(Fa: -> A a F a:)

L A W S OF A L P H A B E T I C V A R I A N C l

T 231 A jcF a AyFy

T232 WxFx VyFy

T233 (F a -> G a ) a (G a H a ) -> 1

130 . ‘A ir and ‘Some’ Chapter III

T234 Лл:[(Рл: -> Ол:) л (Ол; -> Нл:) -> (Рл; ^ Нл:)]
T235 К х {Р х Сл:) л K x { G x -> Нл:) -> Лл:(Рл: ^ Нл:)
T236 Л^(Рл: Ол:) А Лл:(Сл: <-> Нл:) Лл:(Рл: <-> Нл:)
T237 Лл:(Рл: ^ Сл:) л Лл:(Рл: Нл:) Л^(Рл: -> Ол: а Н л:)
T238 Лл:Рл: -> Ул:Рл:
T239 Л^Рл: А Ул:Ол: -> Ул:(Рл: а О л:)

T240 Лл:(Рл: -> Ол:) а Ул:(Рл: а Н л:) -> Ул:(Ол: а Н л:)

T241 Лл:(Рл: Ол: v Нл:) -> Ал:(Рл: -> Ол:) v Ул:(Рл: а Н л:)
T242 ^ Лл:(Рл: -> O jc) <->Ул:(Рл: а ^ Ол:)

T243 У:^(Рл: А Ол:) Л^с(Рл: ^ ^ Ол:)

T244 ^ У^сРл" -> Ал:(Рл: -> Ол:)
T245 ^ Ул:Рл: <-> Ал:(Рл: Ол:) А Ал:(Рл: -> Ол:)
T246 Ал:Рл: <-> Ал:А^у(Рл: а Pj )
T247 Ул:Рл: А Ул:Ол: -> (Ал:[Рл: Нл:] аАл:[Ол: -> }л:] <->
Ал:А^у[Рл: а G y Нл: а Jjy])
T248 (Ул:Рл: Ул:Ол:) А АлА^(Рл: а G y -> [Нл:<-> Jy]) ->
(Ал:[Рл: -> Нл:] 4-> Ад^[Ол: -> }л:])
 Chapter IV
‘A L L ’ and ‘ S O M E ’ , continued

I. T erm s and form ulas. There are arguments whose validity depends
essentially on phrases of quantity and their stylistic variants, but which
cannot be reached by the theory of chapter III. For example, the arguments

(1) All men are mortal. Socrates is a man. Therefore Socrates

is mortal,

(2) All horses are animals. Therefore every head of a horse

is a head of an animal,

though intuitively valid, cannot be validated by the methods of the pre­

vious chapter. Before extending the apparatus of chapter III to cover
such cases, we must supplement the grammatical considerations given
there.
Of sentences of Englishy formulas of English^ and names of English we
have already spoken (pp. 85 and 86). There are expressions that resemble
names in the manner in which formulas resemble sentences. For example.

(3) T + X

is related to the names ‘7 + 5' and ‘7 + 6’ but is not itself a name of

English; it designates neither 12, 13, nor any other object. Expressions
such as (3) are called terms. As a limiting case, a name itself will also
be regarded as a term. Accordingly, a term of English is either a name of
English or an expression containing occurrences of variables that becomes
a name of English when some or all of these occurrences are replaced by
names of English. Each of the following, then, is a term of English:

(4) Sir Walter Scott,

(5 ) the native state of z,
( 6) the square of
(7) the product of x and y ;

for (4) is a name, and by replacing 'z' by ‘Lincoln’, ‘.v’ by ‘2’, and ‘y ’
(5) ” (7)'become, respectively, the names
132 . ‘A ir and ‘Some’, continued Chapter IV

the native state of Lincoln,

the square of 2,
the product of 2 and 3,

which designate Kentucky, 4, and 6, respectively.

In addition to terms of English, we admit abbreviations of such terms,
constructed with the help of operation letters. The symbols
AO ,
BO ,
CO ,
DO ,
EO
are o-place operation letters;
Ai

El
are i~place operation letters; and
A2 ,

E2
are 2 -place operation letters. Additional o-place^ i-placey or 2-place operation
letters may be obtained by adding numerical subscripts to the symbols
above. For any nonnegative integer ky we may characterize k-place
operation letters in a similar way. Thus an operation letter consists of one
of the letters ‘A* through ‘E ’ accompanied by either a numerical super­
script and subscript or only a superscript. The superscript indicates the
number of places of the operation letter.
Names of English will be abbreviated by o-place operation letters.
Other terms of English will be symbolized by operation letters together
with variables. For instance, the expressions

(8 ) AO ,
(9) Bi^
(10) Ci;c
(11)
may be construed as symbolizing the respective terms (4) - (7).
As in chapter II, we shall admit abbreviations of formulas of English,
but here we shall provide means of abbreviating formulas with arbitrarily
Section X Terms and Formulas . 133

many variables. To this end, we introduce predicate letters of arbitrarily

many places in analogy with the operation letters introduced above. The
symbols
FO ,

Zo ,
with or without subscripts, are o-place predicate letters;

FI ,

again with or without subscripts, are i-place predicate letters; and so on.
Thus 2i predicate letter consists of one of the letters through ‘Z ’ accom­
panied by either a superscript and subscript or only a superscript; the
number of places of the letter in question is denoted by its superscript.
o-place predicate letters will assume the role played by sentence letters
in earlier chapters, and i-place predicate letters that of the predicate
letters of chapter III. Formulas of English containing k variables may
be symbolized by /j-place predicate letters accompanied by those variables.
For instance, the formulas
X loves y
and
X lies between y and z

may be symbolized by
F2jcy
and
G^xyz
respectively.

The extended language with which we are now concerned is obtained

by adding to English the following symbols:

(1) the sentential connectives;

(2) parentheses;
(3) the quantifiers, that is, ‘A’ and ‘V’ ;
134 • ‘Some’, continued Chapter IV

(4) variables, that is, lower-case letters with or without subscripts;

(5) predicate letters; and
(6) operation letters.
As terms we reckon terms of English, their symbolic counterparts,
and mixed combinations. To be more explicit, the class of terms can be
exhaustively characterized as follows:

(f) Terms of English {that isy names of English or expressions that are
like names of English except for the occurrence of variables) are terms. {In
particulary variables are terms.)
{it) The result of writing a k-place operation letter followed by k terms
is a term. {In particulary each o-place operation letter is itself a term.)

formulas we understand formulas of English, their symbolic counter­

parts, and m*ixed combinations. Thus the class of formulas can be ex­
haustively characterized as follows:

(j) A ll formulas of English {that iSy sentences of English or expressions like

sentences of English except for the occurrence of variables) are formulas.
(2) The result of writing a k-place predicate letter followed by k terms is a
formula. {In particular a o~place predicate letter is itself a formula.)
y

{3) I f ^ fo rm u la S y so a r e

^ (f) y

{<f> A l p ) ,

{4) I f ^ ^ formula and a a variabhy then

A(X(f> y

yoi<l>
are formulas.
The symbolic terms and formulas of our present language are those terms
and formulas that can be constructed exclusively from variables, predicate
letters, operation letters, parentheses, sentential connectives, and quanti­
fiers. More precisely, the class of symbolic terms can be exhaustively
characterized as follows:

{a) A ll variables are symbolic terms.

{b) The result of writing a k-place operation letter followed by k symbolic
terms is a symbolic term. {In particulary a o-place operation letter is itself
a symbolic term.)
Section I Terms and Formulas . 135

The class of symbolic formulas can be exhaustively characterized as

follows:

(A) The result of writing a k-place predicate letter followed by k symbolic

terms is a symbolic formula, {In particular^ a o-place predicate letter is itself
a symbolic formula.)
{B) I f (f> and ip are symbolic formulas, so are
^<l> ,
{<f> ,
{^ Kip) ,
V ,
(<p<r-^ip) .

(C) I f <p is a symbolic formula and a a variable, then

Aol<P ,
yoi<P
are symbolic formulas.

The only symbolic terms occurring in the language of chapter III were
variables. We now countenance symbolic terms such as ( 8 ) - ( i i ) above,
as well as more complex terms like

(12) C^D^xy
and
(13) D2 C^xy ,

which may respectively be regarded as symbolizing

the square of the product of x and y

and
the product of x"^ and y.

To see that (12) is a term, we observe first that 'x' and ^y' are terms
(clause (a)), and hence (clause (b)) that is a term. Applying clause
(b) again we see that D^xy' is a term. The analysis of (13) is slightly
different; this time we consider the following sequence of terms: *x\
‘ C V , y , ‘D2 C ^ x y .
The class of symbolic formulas also expands. Clause (A) leads not
only to formulas like
po
and
pijc
136 . *Air and *Som e’, continu ed Chapter IV

but also to formulas like

(14) G^xy ,
(15) piAo ,
(16) G2A«x ,
(17) G2 a; B^y ;
and clauses (B) and (C) provide us with compounds of these.

EXERC ISES

For each of the following expressions, state whether it is (a) a term,

(b) a formula, (c) a symbolic term, (d) a symbolic formula, or (e) no
one of (a) - (d). Exercise i is solved for illustration.
1. (AxF2:vA0
This expression is a formula and also a symbolic formula. We refer
to the clauses characterizing terms and formulas (p. 134). V is
a term (clause (i)) and ‘A®* is a term (clause (ii)); hence ‘F^jcA^’ is a
formula (clause (2)) and is a term (clause (ii)). Since is a
term, is a term (clause (ii)), and hence ‘G^A^B V is a formula
(clause (2)). Since ‘F^xA^’ is a formula, so is ‘A^cF^jcA®’ (clause (4)).
Finally, since ‘Aa:F2:vA®’ and ‘G^A^B^^v’ are formulas. No. i is a formula
(clause (3)). In addition, since it contains only predicate letters,
operation letters, parentheses, sentential connectives, and quantifiers,
it is a symbolic formula.

2. The individual identical with the author of Waverley

3. A^ the father of x
4. the son of X loves the daughter of y
5. AA0(FiA0-> GiAO)
6. A^xiW-yz
7. the square of the product of x
8. G3C13;D1a:A1A0
9. EiFi;c
10. - (Vx(Aj(F23;A0 -> G2;cB0) -> H1A1B3A0a;B0) <-> Fi^)

2. Bondage and freedom . With the addition of terms other than

variables to our language, we must reconsider the notions of bondage
and freedom; they are now to apply not only to variables but to arbitrary
symbolic terms.
An occurrence of a symbolic term ^ is said to be bound in a symbolic
formula 0 if it stands within an occurrence in 0 of a formula

Aa0
or
Va0 ,
Section 3 Informal Notational Conventions . 137

where 0 is a formula and a is a variable occurring in an occurrence

of a symbolic term is free in <f) if and only if it stands within <f> but is not
bound in (f), A symbolic term is itself bound or free in 0 according as there
is a bound or free occurrence of it in (f>.
For example, consider the formula
{Vx piAijc V GiAijc) .
In this formula the first occurrence of the term is bound, the second
occurrence is free, and the term itself is both bound and free.
The reader should note that bondage and freedom of occurrences of
variables and of variables, as characterized in the preceding chapter,
are simply special cases of the present notions, which apply to arbitrary
symbolic terms and their occurrences.
As in chapter III, a symbolic sentence may be characterized as a sym­
bolic formula in which no variable is free. We can also characterize a
symbolic name as a symbolic term in which no variable occurs.
The reader should observe that the notions of bondage and freedom
apply only to symbolic terms and formulas; as before, applications of
these notions to nonsymbolic expressions will not be required for the
formulation of our deductive system.

EX ER CISES
Consider the following formula:
(Ay (Vjc a ® A^x V G 2 B2y;sr Aijc) -> \ z (G^ <2: v G^ B^y^:))
11. In this formula identify each occurrence of a term as bound
or free.
12. Which terms are bound in the formula? Which terms are free
in the formula?

3. Informal notational conventions. We shall again employ the

conventions of chapter II (pp. 40, 51, and 65) for omitting parentheses
and replacing them by brackets.
In addition, it will be our usual practice to omit superscripts from
operation letters and predicate letters, inserting parentheses and brackets
to avoid ambiguity. In the case of o-place operation letters, and o-place
and I-place predicate letters, no parentheses or brackets will be employed.
In the case of operation letters of one or more places, and predicate letters
of two or more places, we shall enclose the terms accompanying the letter
in question in a pair of parentheses or brackets. For instance, the terms
(12) and (13) of section i (p. 135) may become
C[D(^>-)] ,
T>{C{x)y]
138 . *Air and 'Some*, continued Chapter IV

respectively, and the formulas (14) - (17) of page 136 may become
G(^j) ,
FA ,
G(Aa;) ,
G [^ B (j)]
respectively. Restoration of official notation is completely automatic;
the superscript of a letter is determined by the sequence of terms following
it.
Again, we must emphasize that in theoretical discussions—for instance,
the definitions of bondage and freedom and the ensuing characterization
of derivability—the words ‘term’ and ‘formula’ are always to be under­
stood in the official sense.

EX ER C ISES
13. For each of the following formulas in unofficial notation, delete
the parentheses and brackets inserted, and restore the superscripts
and parentheses omitted, in accordance with the informal notational
conventions introduced in this section.
Y A{x)
G(B x)
H[C(B B) A(B) B]
F jc V P G{xy) A \{{xyz)

4. Translation and sym bolization. Many symbolic sentences of

our present language cannot be translated into English on the basis of
the schemes of abbreviation used in chapter I I I ; an example is
(i) ^yxY {xK ) .
For present purposes we must admit, in addition to the abbreviations of
the last chapter, abbreviations involving operation letters as well as
predicate letters of more than one place.
It is convenient to establish a standard order of variables, in fact, the
order represented by the following list :
a , . . . , 5T, ao, . . . , ^0, • aiy
Thus the first variable will be ‘a’, the second variable ‘6’, the twenty-
sixth variable 'z\ and so on.
If ^ is a formula of English, then a is said to be an apparent variable of
(f> if either ^ is a sentence of English containing a or there is an occurrence
of a in ^ such that, upon replacement of some other occurrences of vari­
ables by English names, <f> becomes a sentence of English. Similarly, if i
is a term of English, then a is an apparent variable of ^ if either t, is an
English name containing a or there is an occurrence of a in ^ such that.
Section 4 Translation and Symbolization . 139

upon replacement of some occurrences of variables by English names, I

becomes an English name. For example, 'a' is an apparent variable of the
name
the letter following ‘a ’
and of the term
the object a such that a wrote b
We shall admit abbreviations only of those formulas and terms of
English whose variables are the first k variables, for some number ky
and which contain no apparent variables. (The reason for this limitation
is discussed in section i i below.) We shall abbreviate such a formula
or term by a ^-place predicate letter or operation letter.
Thus an abbreviation will now be either (i) an ordered pair whose first
member is a/f-place predicate letter, for some /5> 0 , and whose second
member is a formula of English containing exactly the first k variables and
having no apparent variables, or (2) an ordered pair whose first member is
a A-place operation letter, for some number /f^o, and whose second
member is a term of English containing exactly the first k variables and
having no apparent variables.
As in earlier chapters, a scheme of abbreviation is to be a collection of
abbreviations such that no two abbreviations in the collection have the
same first member. The following is an example:

(2) AO Adam
F2 a is father of b
The process of literal translation into English on the basis of a given
scheme of abbreviation begins as before with a symbolic formula and if
successful ends with a formula of English. The process consists of the
following steps:

(/) Restore official notation by reversing the conventions of the preceding

section,
{it) Replace o-place operation letters and o-place predicate letters by
English names and sentences in accordance with the scheme of abbreviation,
{Hi) Replace each operation letter and predicate letter of one or more
places by the term or formula of English with which it is paired in the scheme
of abbreviation, flanking the latter with a pair of braces, (As before, the result
of this step will not in general be a formula,)
(iv) Successively replace all parts of the form

where <f) is a term or formula of English containing exactly the first k

variables and t,\, , , , , are terms, by the ter?n or formula obtained from
140 . *A11’ and *Some’, continued Chapter IV

<f>by replacing all occurrences of 'a' by ^i, 'b' by ^2, etc., up to the kth variable,
which is to be replaced by
(v) Eliminate sentential connectives and quantifier phrases in favor of
the corresponding phrases of connection and quantity, preserving all paren­
theses.

As before, we say that an English formula is a free translation (or

simply a translation) of a symbolic formula ^ on the basis of a given
scheme of abbreviation if it is a stylistic variant of the literal English
translation of (f> based on that scheme.
Let us, for example, translate the sentence (i) into English on the basis
of the scheme (2). In step (i) of the process of literal translation into
English, (i) becomes

in step (ii)
^Wx X Adam
in step (iii)
' Wx [a is father of b] x Adam
in step (iv)
Va: jc is father of Adam
and in step (v)
(3) It is not the case that there is an object x such that x is father
of Adam.
Thus (3) is the literal translation of (i) into English on the basis of the
scheme (2); and the more idiomatic stylistic variant of (3),
Adam has no father,
qualifies as a free translation of (i) on the basis of the same scheme.
As a slightly more involved illustration, consider the symbolic sentence
(4) hxi\y{Fx A Fy A Gy -> G A(jcy))
and the scheme of abbreviation

(5) pi is a number
Gi a is even
A2 the product of a and b .
Let us find the literal English translation of (4) based on the scheme (5).
Applying step (i), we obtain
hx/\y{{{F^x A ply) A G^y) G^ A^xy) ,
Section 4 Translation and Symbolization . 141

Step (ii) is irrelevant. In step (iii) we obtain

hxf\y{{{{a is a number} a; a {¿z is a number) j ) a
[a is even) y) -> [a is even)(the product of a and ¿} x y ) .

Applying step (iv), we obtain first

t\xhy{{{{a is a number) a is a number) j') a
{a is even) y) -> {a is even) the product of x and y) ,
and then
t\xt\y{{{x is a number a jy is a number) a j is even) ^
the product of x and y is even) .
Finally, by step (v), we obtain as the literal translation of (4) the English
sentence
(6) For each Xy for each y (if {{x is a number and j is a number)
and y is even), then the product of x and y is even) .
Moreover,
The product of any number and any even number is even,
being a stylistic variant of (6), is a free English translation of (4) on the
basis of (5).
As before, cf) is said to be a symbolization of an English formula i/j on
the basis of a given scheme of abbreviation just in case is a symbolic
formula that has 0 as a translation on the basis of that scheme.
To find a symbolization of a given formula of English on the basis
of a given scheme of abbreviation the reader will again find it useful to
proceed roughly as follows:

(j) Introduce phrases of quantity and connectiony the latter accompanied

by parentheses and occurring canonicallyy in place of their stylistic variants,
(2) Reverse the steps leading from a symbolic formula to a literal English
translation.

For an example of the process of symbolization, consider the sentence

(7) There is no composer whose contrapuntal ingenuity sur­
passes that of Bach,
together with the scheme of abbreviation
FI a \s z composer
G2 a surpasses b
Ai the contrapuntal ingenuity of a
BO Bach .
142 . ‘All* and ‘Some*, continued Chapter IV

In S te p (i), w it h in tu itio n a s g u id e , w e tr a n s fo r m (7) in to

It is not the case that there is an object x such that {x is a

composer and the contrapuntal ingenuity of surpasses the
contrapuntal ingenuity of Bach) .
Let us now perform step (2) of the process of symbolization. Reversing
step (v) of the process of literal translation into English, we obtain
is a composer a the contrapuntal ingenuity of x
surpasses the contrapuntal ingenuity of Bach) .
Reversing step (iv), this becomes first
'^yx{{a is a composer) x k {a surpasses b] the contrapuntal
ingenuity of X the contrapuntal ingenuity of Bach) ,
and th e n

^ yx {{a is a composer) x k {a surpasses b] (the contrapuntal

ingenuity of a] x (the contrapuntal ingenuity of a) Bach) .
Reversing step (iii), we obtain
- yx{Y^x A G2 K^x AiBach) ;
reversing step (ii),
-Vjt'(Fi;cA G 2 A 1 a;A1B0) ;
and reversing step (i),
^Wx{Fx A G[A{x)A{B)]) .
An example of stylistic variance of rather frequent incidence is the
mutual conversion of active and passive voice. Consider, for example, the
sentence
If someone is loved by one whom he does not love and by
nobody else, then he does not love his lover,
to g e th e r w ith th e fo llo w in g s c h e m e o f a b b r e v ia t io n :

L2 : a loves b
T2 . a differs from b
Ai ; the lover of a
In step (i), construing 'x is loved by y ' as a stylistic variant of *y loves x\
the sentence above becomes
For each jc (if there is an object y such that {{y loves a: and
it is not the case that x loves y) and it is not the case that
there is an object 2r such that {z differs fromjy and 2: loves x)),
then it is not the case that loves the lover of x) ;
Section 4 Translation and Symbolitcation * 143

and carrying through the successive parts of step (2), we obtain the
symbolization
Ax[Vy(L{yx) A r^L{xy) A ^Wz[T{^y) a U zx )]) -> ^ L { x A{x))] .
The order of quantifiers in a symbolization is now of particular impor­
tance. For example, if we adopt the scheme of abbreviation
T2 : a differs from b ,
then
AxWy T(icy)
is a symbolization of the true sentence
. Each thing differs from something,
whereas
VyAjc T(A[y)
is a symbolization of the false sentence
Something is such that everything differs from it.

EXERCISES
14. On the basis of the scheme of abbreviation

is a person
: a loves b ,

find for each sentence of group A a sentence of group B that is a

symbolization of it.

GROUP A
(1) Everyone loves someone.
(2) Someone loves someone.
(3) If anything is a person, then someone loves hitnself.
(4) Someone loves everyone.
(5) Everyone loves everyone.

GROUP B
(a) Vx(Fjc a Ay[Fy G(xy)])
(b) Ax(Fx AAy[Fy G{xy)])
(c) Vy(Fy A Vx[Fx A G(yx)])
(d) Ax(Fx -> Vx [Fjc a G{xx)])
(e) Ay(Fy Ax[Fx -> G(>'^)])
(f) Vx(Fx -> Wy[Fy A G(xy)])
(g) Ax(Fx Wy[Fy A G(xy)])
144 • ‘All’ and ‘Some*, continued Chapter IV

15. On the basis of the scheme of abbreviation

pi : a is a student
is a teacher
is a subject
: a studies h with c ,
find for each sentence of group B a sentence of group A that is a
translation of it.

GROUP A

(1) Every student studies every subject with every teacher.

(2) Every student studies some subject with some teacher.
(3) No student studies every subject with a teacher.
(4) Only students study every subject with a teacher.
(5) No subject is such that every student studies it with every teacher.
(6) Every teacher has some subject that some student studies with
him.
(7) Each subject has some student who studies it with some teacher.
(8) None but teachers are such that all students study all subjects
with them.
(9) Anyone who studies any subject with any teacher is a student.
(10) There is a teacher such that every student studies some subject
with him.
(11) There is a teacher such that some student studies every subject
with him.
(12) Teachers who study some subject with a teacher are students.
(13) No student who does not study every subject with a teacher is a
teacher.
(14) Any student who studies any subject with himself is a teacher.
(15) Some teacher who studies every subject with himself is a student.
(16) There is no teacher with whom any student studies all subjects.

GROUP B

(a) A^:(AyA2[Fy a S(jy2.v)] G.v)

(b) V;c(G jc a Yy[Fy a Az {Hz S(y2:v))])
(c) Ax{Fx -> AyA2[Hjy a G^r -> S(xv2r)])
(d) Ax{Ux -> yyWz[Fy a G z a S(yxz)])
(e) A jc(F^ -> VyVz[Hy a G z a S(xy2)])
(f) Ax(Fx A Vy[Hy A S(ryA:)] -> Gx)
(g) V a:(Gjc a Ay[Fy ->Wz{Hz a S(j 2:v))])
(h) AxAyAz(Fx a Hy a G 2 -> S(xv2))
(i) ^ V;c(Ha: a AyAz[Fy A G^r ^ S(yA:2)])
(j) A^(F:c - Ay[Hy -> Vz{Gz a S{xyz))])
(k) A^(F;c A - Ay[Hy -> Wz{Gz a S{xyz))] -> - Gx)
(l) Ax{Fx -> Ay[Hy -> A2(G;^ -> S{xyz))])
Section 4 Translation and Symbolization • 145

(m) A^(Vy A A S(:vjv;2r)] -> Fx)

(n) Vx{Gx A Ay[Hy S(ry:v)] a F a:)
(o) AA:(FA:->y);(H3^AV;^(G;^AS(A:3;;^))])
(p) Ajc(Ajv[Hjv -> V;s:(G;3: A S(xv;s:))] ^ Fx)
(q) Va:(Fa: a Yy[Hy A ^ V;s:(G^ A S(a:jv;2:))] a G a:)
(r) VA:yy(GA: A FjV A A;s:[H;s: -> S(jV2:a:)])
(s) Aa:(Ga: -> W y W z lF y a H2: a S(>'5:a:)])
(t) - Wx(Fx A Ay[Hy -> yz{Gz a S{xyz))])
(u) Aa:(Ga: a WyWz[Hy a G z A S(a:jv2:)] - > F a:)
Translate each of the following symbolic formulas into idiomatic
English on the basis of the scheme of abbreviation that accompanies
it. Exercise 16 is solved for illustration.

16. AxAyAzAzvlFx a G ( va:) a G(2:a:) a G(«;a:) A Fiiyz) a Fiiyw) a

H(^«;)->I(A(Bb] B[z]) B H )]
(FI : ¿2 is a triangle; G^ : a is 3. side of 6; is different from
b; 12 : is greater than b; A2 : the sum of a and : the length
of a)
In step (i) of the process of literal translation into English, No. 16
becomes
AxAyAzAw{{{{{{{F^x a G 2ja :) a G ^ z x ) a G^^w x ) a Fi^yz) a

H ^yw ) A H ^zw ) 12 A ^ F ^ y F ^ z B ^ w ) ;

in step (iii), for no change occurs in step (ii),

AxAyAzAw{{{{{{{{a is a triangle} a: a {¿1 is a side of b) y
x) A {a is a side of b) z x) A {a is a side of b)w x) a {a is
different from b}y z) a {a is different from b}y w) a {a is
different from b) z w) {a is greater than b) {the sum
of a and b) (the length of a) y (the length of a) z (the
length of a) w) ;
in step (iv), first
AxAyAzAw{{{{{{{x is a triangle a y is 3 side of a:) a ;s: is a
side of a:) a w is a side of a:) A jy is different from 2:) a jy is
different from w) A zis different from w) {a is greater
than b) (the sum of a and b] the length of y the length of
z the length of w) ,
next
AxAyAzAw{{{{{[{x is a triangle a jy is a side of a:) a 2: is
a side of a:) a w is a side of a:) a 3^ is different from z) A y
is different from w) a ;s: is different from w ) { a is greater
than b) the sum of the length of y and the length of z the
length of zv) ,
146 , *Air mt^d *Som9% continued Chapter IV
and finally
A.vAyA^Aw(((((((a: is a triangle a jy is a side of jc) a ^ is a
side of at) A nr is a side o i x ) h y is different from 5 :) A y is
different from w) a ar is different frojm w ) the sum of
the length of y and the length of z is greater than the
length of nr) ;
and in step (v)
For each Xy for each y, for each ar, for each w (if {x is a
triangle and y is a side of x) and is a side of x) and zv is
a side of x) and y is different from z) and y is different
from to) and z is different from zv), then the sum of the
length of y and the length of z is greater than the length
of tv) ,
which is the literal translation into English of No. 16; and the following
stylistic variant is a free translation:
The sum of the lengths of any two sides of a triangle
is greater than the length of the third side.

17. yx{Fx A Ay[Fy -> GA(yAc)]) Wx(Fx a G x)

(F^ : fl is a number; : a is even; : the product of a and b)
18. A x:(Fa; VyG(yAf) a VyH(yAf)) a V:v(F jc a /^[VyG(jcy) v
VyH(^)])
(F^ : a is a person; G^ : a is father of b\ : a is mother of b)
19. Ax{F{x E) -> Wy[G(y E) a H(xy) a l(y B(C(E)))])
(F2 : a is a student of b; E® : the course; G^ : a is a quiz section of b;
H* : a attends b; : a is taught by ft; : the teaching assistant of a ;
: the instructor of a)
Symbolize each of the following formulas on the basis of the scheme
of abbreviation that accompanies it. Exercise 20 is solved for illustra­
tion.

20. A person pays a dollar for the banquet of the club only if
he belongs to at least one of the club’s committees and attends every
meeting of that committee. (F^ : a is a person; G^ : a pays a dollar
for ft; № ; a is a committee of ft; : a belongs to ft; : a is z
meeting of b ; : a attends ft; A ^: the banquet of a\ : the club)
Linguistic insight would suggest that No. 20 become in step (i)
of the process of symbolization
For each x (if (jt is a person and x pays a dollar for the
banquet of the club), then there is an object y such that
((y is a committee of the club and x belongs toy) and for
each z (if is a meeting of y, then x attends 2r))) .
Section 4 Ti^anslation and Sytnbolixation • 147

By performing the successive stages of step (2), Wc obtain first

Aa:((jc is a person a x pays a dollar for the banquet of the
club) -> ^ y {{y is a committee of the club a x belongs to>^)
A is a meeting of y x attends z))) ,
next
is a person} x A {a pays a dollar for 6} x {the ban-^
quet of a) the club) Yy(({^ is a committee of 6} y the
club A {a belongs to 6} xy) a Az((a is a meeting of t j z y A
{a attends 6} x z))) ,
next
a G ^ x A^ the club) -> Vy((H2 y the club a I^^y) a
Az(Pzy K^xz))) ,
next
Ax((F^x A G 2 .V A^BO) -> Vy((W y B<> a l^xy) a Az(Pzy ->
K^xz))) ,
and finally
Ax[Fx A G(^ A(B)) yy(H(j^ B) a I(A;y) a Az[J(zy)
K(^^)])] .

21. Every husband and wife has a spouse* (F^ : a is a husband;

G^ : a is a wife; : a is a spouse of b)
22. No one lacks a father, but not everyone is a father. (F^ : n is a
person; G^ : a is father of b)
23. Every number has some number as its successor. (F^ : a is a
number; G^ : a is a successor of 6)
24. Some number is a successor of every number. (F^ : a is a
number; G^ : a is a successor of b)
25. A teacher has no scruples if he assigns a problem that has no
solution. (F^ : ¿2 is a teacher; G^ : a has scruples; : a assigns b;
: a is a problem; : a is a solution of A)
26. The net force acting on a particle is equal to the product of its
mass and its acceleration. (F^ : a is a particle; A^ : the net force act­
ing on a ; : a is equal to b; B^ : the product of a and b ; :
the mass of a ; D^: the acceleration of a)
27. No principle is innate unless everyone who hears it gives his
assent. (FI : a is a principle; G ^ : a is innate; : a is a person; I*:
a hears b; P : a gives his assent to b)
28. The cube of the square of 2 is evert. (F^ : a is even; A^ : the
cube of a\ B^ : the square of a; C® : 2)
29. No student likes every course he takes unless the only courses
he takes are philosophical studies. (F^ : a is a student; G^ : u is a
course; : a is a philosophical study; : a takes b\ P : a
likes b)
148 . *A11’ and *Some\ continued Chapter IV

30. A person who loves himself more than he loves anyone else
is not loved by anyone other than himself. (F^ : a is a person; :a
loves b\ H2 : is different from b\ : a loves b more than c loves d)
31. If a father has only male children, then he does not have to
provide a dowry for any one of them. (F^ : a is 3, father; : a is
male; : a is a child of b; : a has to provide a dowry for b)
32. The wife of anyone who marries the daughter of the brother
of his father marries the son of the brother of the husband of her
mother. (F^ : a marries b; : the wife of a; : the daughter of
a; : the brother of 0; : the father of a; : the son of 0;
A^i : the husband of a\ : the mother of a)
33. If X is an integer greater than or equal to zero and every integer
is divisible by x, then x is equal to i. (F^ : a is an integer; G^ : a is
greater than b; ; a is equal to 6; : a is divisible by 6; A° :
zero; B^ : i)
34. The square of the hypotenuse of a right triangle is equal to the
sum of the squares of the other two sides. (F^ : 0 is a right triangle;
G2 : is a side of : 0 is equal to 6; A^ : the square of a; B^ :
the hypotenuse of a; : the sum of a and b)
35. The gravitational force exerted by one particle on another
is directly proportional to the mass of each particle and inversely
proportional to the square of the distance between the two particles.
(FI : a is a particle; G^ : a is directly proportional to b; : a is
inversely proportional to 6; A^ : the mass of a; B^ : the square of a;
C2 : the distance between a and b; : the gravitational force exerted
by a on b)
36. Someone is hit by a car every day. (F^ : a is a day; : a is a
car; P : a is hit by b on c)

5. R e v ise d in feren ce ru les. We must extend the notion of proper

substitution for a variable. In chapter III only variables could be substi­
tuted for variables; now we shall admit the substitution of arbitrary sym­
bolic terms for variables. Thus we say that a symbolic formula ip comes
from a symbolic formula <f> by proper substitution of a symbolic term ^ for
a variable a if i/r is like <f) except for having free occurrences of ^ wherever (p
has free occurrence of a.
In chapter III the only terms to which U I and E G could be applied
were variables. We now reformulate these two rules so as to admit an
arbitrary symbolic term in place of the variable of instantiation.

Universal instantiation (U I): Aa^

Existential generalization (E G ): p
Section 5 Revised Inference Rules . 149

In both cases a is to be a variable, (f> is to be a symbolic formula, and tp

is to come from (f> by proper substitution of some symbolic term for a.
All previous applications of U I and E G are comprehended under the
present formulation; there are additional applications of the following
sort. From the sentence
A.vF;c
we can now infer such formulas as

(1) FA ,
(2) F B (A ) ,
(3) FB (^) ,
(4) F C(A x)
Further, from any of (i) - (4) the sentence
V^F;c
follows by EG . From (4) we can also infer by E G the formulas
V jF C ( jx ) ,
F C(A x) ,
yyYC{Xy) ;
but we cannot infer
Vjc F C(.v x) ,
because (4) does not come from T C(;c x)' by proper substitution for 'x\
It should be emphasized that the present extensions concern only U I
and EG . The formulation of E l remains exactly as before; that is, from
one formula another follows by E l just in case the two formulas have the
respective forms
Va<^
and

<f>’ ,
where a is a variable, is a symbolic formula, and comes from (p by
proper substitution of some variable for a.
The directions for constructing an unabbreviated and (for the moment)
an abbreviated derivation remain as in chapter III. For an unabbreviated
derivation we use clauses ( i) - ( 6 ) of pages 1 0 2 - 0 3 ; for an abbreviated
derivation we supplement these with clauses (7) - (9) of pages 106 - 07.
Throughout these clauses, however, we must now understand the terms
‘formula*, ‘U r , and ‘EG* in the broader sense of the present chapter.
We thus arrive at a characterization of a derivation in the full (rather than
the monadic) quantifier calculus.
x $9 • 'All* sad 'Some'» continued Chapter IV

The characterizations of a complete derivation, derivability, a proof, a

theorem^ an argument, a symbolic argument, a valid symbolic argument,
an English argument, a symbolization of an English argument, and a valid
English argument remain exactly as before. (See pp. 95, 103, and 117.)

EXERCISES
37. Which of the following formulas follow from
bxyzF{\{xyx)z)
byUÎ?
(a) V«F(«)
(b) V«F(A(xyx)^)
(c) YarF(A(393^)a)
(d) VirF(A(xAi:x)a)
(e) VaF(A(CyB)*)
(f) V«F(A(CyC)2)
(g) V^F(A(ByB);a)
(h) VarF(A(A(xyx)yA(xyx))2r)
(i) VtrF(A(B(y)yB(y))z)
(j) VarF(A(B(C(a))yB(C(a)));r)
(k) VaF(A(C(B(w))yC(B(«)));s)

38. Which of the following formulas follow from

AxFA(xB(y)) V G(Cy)
by EG?
(a) V«)[AxFA(xB(y)) v G(«5')]
(b) Ww[bxYv) V G(C>-)]
(c) Vtt[AxFA(xB(a>)) v G(Cw)]
(d) Vw(AxFA(Ate) v G(Cy)]
(e) Vx(AxFA(xB(y)) v G(xy)]
(f) Vx(AxFA(xB(x)) VG(Cie)]
(g) Vy[AxFA(A:BCv))vG(Cy)]
(h) yy(AxFA(xBCv)) VG(xy)]
(i) VAr[AxFA(xB(y)) VG(Cy)]
(j) Vw[AxFA(xB(y)) VG»]

For each of the following arguments give an unabbreviated derivation

of its conclusion from its premise.
39. Ax F A(x B) .*. Vx F A(xx)
40. Vx F A(xx) VxFx
41. AxAy- F A[x B(y) C] Vx F A[B(x) B(C) x]
4Z. AxA>- F A[x B(y) C] .'. Vx F A[B(x) x C]
Section 6 Theorems . 151

Show the following arguments valid by constructing symbolizations

and giving derivations (either abbreviated or unabbreviated) of the
conclusions of the symbolizations from their premises. Exercise 43
is discussed for illustration.

43. Argument (2) of page 13 1.

On the basis of the scheme of abbreviation
is a horse
: a is an animal
H2 : <2 is a head of b ,
one obtains the symbolization
hx{Yx Gx) Ax{yy[H{xy) a Fy] -> Vy[H(^j) a Gy]) ,
whose validity the reader can easily establish.

44. Argument (i) of page 13 1.

45. The square of 2 is even. /. The square of something is even.
46. The square of something is even. Something is even.
47. The product of numbers is a number. There is a number such
that the product of it and any number is even. /. There is an even
number.

6. T h eo rem s. Symbolic formulas corresponding to those of chapter

III are known as monadic formulas. More exactly, a monadic formula
is a symbolic formula that contains no predicate letters of more than one
place and no operation letters at all.
For each of the theorems of chapter III a nonmonadic analogue may
be obtained by multiplying variables. For example,
T249 Ax Ay F(xy) -> F(.vj)
is a two-variable analogue of T238, and
T250 AxAyAz F(xyz) <<-> WxWyWz ^ F{xyz)
is a three-variable analogue of T205. Proofs of these two theorems can
be obtained without essential change from the proofs of their earlier
counterparts, as the reader can easily confirm; and this applies to all
multiple-variable analogues of theorems of chapter III.
There are nonmonadic theorems that cannot be obtained in this way,
for example T 251 - T253, the laws of commutation of quantifiers. Accord­
ing to T 251 and T252, a pair of like quantifiers may be commuted. If,
however, the quantifiers are unlike, only the conditional, T253, holds.
T 251 I . Show AxAyF{xy) <-> AjyA;cF(.xrv)
2. I Show-AxAyF{xy) AjyAxF(rv) I
152 . ‘A ir and ‘Some*, continued Chapter IV

3 AjcAjyF(xy)
4 Show-AyAxF (xy)
5- AyF{xy) 3, UI
6. F{xy) 5 .U I

7 ShewAyAxF{xy) -> AxAyF{xy)

8 AyAxF{xy)
9 Show AxAyF{xy)
10 A^F(xy) 8, UI
11 10, U I

12 A;cA_vF(xj) AjyA:vF(xy) 2, 7, CB

T252 yx^yF{xy) <-> YyVxF(xv)

T253 ■ ShowyxAyF{xy) AyyxF{xy)
yxAyF{xy)
AyF{wy) 2, E l
Show AyyxF{xy)
F{wy) 3. U I
VxF{xy) 5. e g

The reader will find an effort to prove the converse of T253 (which is
not a theorem; see exercise 73) an instructive review of the restrictions
on variables that must be observed in constructing a derivation.
We list a few more nonmonadic theorems. The reader who is familiar
with RusselFs paradox or with axiomatic set theory will perhaps find
diversion in proving T255 and T256.
T254 VjcyyF(.ry) <<-> VxYy[F(xy) v F(yx)]
T255 YyA,v[F(:rv) <-> ^F(jcjc)]
T256 AzWyAx[F{xy) <-> F(;c^) a ^ F ( a:^)] -> '>^WzAxF{xz)
T257 AxAyF(xy) -» AyAxF(yx)
T 2 s8 F{ xA(x )) <-> yy[Az[F{zy) -> F(^A(jc))] a F{xy)]
We consider now some monadic theorems whose intuitive content is
less clear than that of the theorems of chapter III. By a generalization
we understand a formula that is either a universal generalization or an
existential generalization, that is, which has the form
AoLcf)
or
WoLcf>
Section 6 Theorems . 153

where a is a variable and (f> a formula. A symbolic formula is said to be

without overlay if it contains no generalization that in turn contains
another generalization. T 2 5 9 - T 2 6 1 illustrate the following fact: every
monadic formula is equivalent to a monadic formula without overlay.
T259 I. S k e w ^ h y y X ' Fy) <-> yxYx a Wx '^ F x
2 S k o ftrV x F x a Wx '^ F x
AjyVjc(F^<-> '^Fy)
3- V;cF.x: a Wx ^ F x
4- F w 3 . S, E l
5- ^ F z 3. S, E l
6. S h o w t \ y y x { F x i 'Fy]
7- Ayyx(Fx
8. ^ yx{Fx <-> F m) 7, QN, E l
Fz:; <r-^ ^ ^^Fu 8, QN, U I, T90,
BC, M P
10. ^ Fw 9, BC, 4, MP
11. Fz Fu 8,Q N , U I,T 9 o ,
BC, M P
12. ^ ^ ^Fu I I , BC, 5, M T

13- S h o w h y y x { F x ^ F^) ->

yxF x A yx Vx
H- Ajy Vjc(F;x; <-> F^)
15- F v < - ^ F y 14, U I, E l
16. Skew-yxFx
17* W x F x

18. F y QN, U I
1 7 ,

19. Fz; 15, BC, 18, M P

20. 'F v 17, QN, U I
21. Skmt^Wx^Fx
22. ^W x'^Fx
23- F v 22, QN, U I,
DN
24. ■ F y 15, BC, 23, MP
25- ■F y 22, QN, U I
26. W x F x A W x F;c 16, 2 1, Adj
27. h y y x {F x * -* ~ Fjy)<->V;cFA; a Vji:~ F:x: 2, 13, CB

T260 V jcA j (F a :<-> F j ^) ~ V^tFjc V A a :Fx :

T261 Ax(Fx Vy[G>- A (H>; V H;c)]) <-> yx{G x a H;c) v ~ V a:F.v v

(V;cG;)C A A.v[Fa; -> H.v])
154 • *Some*, continued Chapter IV

I 1 1 T 2 5 9 - T 2 6 1 , we were interested in ‘disentangling* quantifiers.

The following theorems, T 2 6 2 -T 2 6 6 , illustrate the reverse process,
that of ‘consolidating’ quantifiers. More precisely, let us say that a formula
is in prenex vormal form just in case it is a symbolic formula consisting of
a string of quantifier phrases followed by a formula without quantifiers.
T262 - T266 illustrate the fact that every symbolic formula is equivalent
to one in prenex normal form. (In connection with T265 the reader should
reconsider exercise 43.)

T262 1. Shew-{hixYx -> AjcG:c) <-> VxAj (F a: -> Gy)

2. Skew- AjcF.v-> yxf\y{¥x-> Gy)
3 '^hxYx
4 3, QN, E l
5 Shew hy{Yz -> Gy)
6 Fz Gy 4, T i 8,M P

7 V;vAy(Fjc Gy) 5, e g
8 Skew AxGx WxAy{Fx -> Gy)
9 AxGx
10 Skew A y{Fx --> G j)
11 F jc ^ G j 9, U I, T 2, MP
12 WxAy{Fx Gy) 10, E G
13 (AxFx ->AxGx) ->
V.x:Ay(Fx -> Gy) 2, 8, AdJ, T35,
BC, MP
H- SkewWxAy{Fx -> Gy)
(Aa:F^ AxGx)
IS- V;cAy(F:v Gy)
16. Ay (Fu Gy) IS, E l
17- Skew AxFx -> AxGx
18. AxFx
19. Show AxGx
20. G.V 16, U I, 18,
U I, MP
2 1. (AxFx AxGx)<-^
WxAy(Fx Gy) 13, 14, CB

T263 {AxFx<r^WxGx)<r^ WxWyAz Az^;([F.v Gy] a [G^r F«;])

T264 Q/xFx [WxGx -> AxH:v]) <-> AxAyAz{Fx a Gy Hz)
Section 7 Alphabetic Variance . 155

T265 Лл:(Уу[Н(лгу) Л Fy] -> Уу[Н(лгу) л Gy]) ^

Лл:ЛуУ^(Н(л:у) a Fy Н(д:^) a G^)
T266 Лл:(Рл: A VjyG(xy) -> Уу[Н(д:у) a hz]{xyz)\) <->
^x^yУw^z[Fx A G ( ^ e y ) U{xw) a ]{ x w z )]

A universal and an existential quantifier can in some cases, though

not of course in general, be commuted; T267 - T 271 are illustrations.
T267 I. Shew-hxyy{Fx -> Gy) yy^x(Fx Gy)
f\x4y (Fx Gy)
■ Shew-Уу hx{Fx -> Gy)

4, QN, U I, QN, E l
2, U I, E l
5, T 2 1, MP, 6, MP
4, QN, U I, QN, E l
8, T 22, MP

T268 VyAx(Fjc Gy) • hxyy{Fx Gy)

T269 yyKx{Fx V Gy) i Лл:Уу(Рл: v Gy)
T270 yyhx{Fx A Gy) i Лл:Уу(Рл: a Gy)
T 271 f\xhyyz{Fx A Gy -> H2) hyyzhx{Fx A Gy Hz)

E X E R C ISE S

In solving these exercises, the reader will still find useful the informal
suggestions made on pages 1 15 - 17 of chapter III.
48. Prove T 254-T258.
49. Prove T260 and T261.
50. Prove T264.
51. Prove T269.
7. Alphabetic varian ce. In chapter III the equivalence of two
generalizations differing only in their initial bound variables was observed.
We may refer to symbolic formulas related in this way as immediate
alphabetic variants of one another. Symbolic formulas differing only in
immediate alphabetic variants are also equivalent and may be called
alphabetic variants.
More precisely, we call two symbolic formulas immediate alphabetic
variants if they have the forms
Ау.ф
and
156 . ‘All* and ‘Some*, continued Chapter IV

or else
Va<^
and
V a 'f ,

where a and a' are variables, (f>and (f>' are symbolic formulas, <f>' comes from
(f> by proper substitution of a' for a, and <f> comes from by proper
substitution of a for a'. We say that a symbolic formula ifj is an alphabetic
variant of a symbolic formula if ip is like 0' except for having one or more
occurrences of a formula (f> where 0' has some immediate alphabetic
variant of <f>.
For example,
\'x{Gx A H^)

is an immediate alphabetic variant of

Vj (Gj a Hj ) ,
and therefore
Az[Fz -> Va:(G;c a H x )]

is an alphabetic variant of

Az[Fz My {Gy AUy)] .

Alphabetic variance will play an important role in section 9, in connec­

tion with abbreviatory clauses.

E X E R C ISE S

52. For each of the following pairs of formulas, state whether the
second formula is an alphabetic variant of the first, (a) is answered for
illustration.

(a) AxVy G(xy)

AxWx G(:vjc)
These formulas are not alphabetic variants; although ‘G(:c;v)’ comes
from ‘G(xy)’ by proper substitution of V for y \ ‘G(xy)* does not
come from ‘G(jc:v)’ by proper substitution of ‘y* for *x\

(b) Ax{Fx -> WxGx A Hy)

A ^ ( F jc -> VyGy A Hy)

(c) Ax{Fx -> V;c[G:c a Hy])

Ax{Fx -> Vy[Gy A Hy])
Section 8 Substitution . 157

(d ) y z fy x H x A G z)
V zC /zH z A G z)

8. Substitution. Present abbreviations make available, as lines of a

derivation, instances of s e n te n tia l theorems that have previously been
proved. The reader may already have considered the possibility of extend­
ing this practice to a l l theorems previously proved, that is, the possibility
of using instances of theorems of the quantifier as well as the sentential
calculus in constructing derivations. Such an extension is indeed possible,
but it has been deferred to the present point because the relevant notion
of an in sta n c e exhibits complexities that did not arise in connection with
sentential theorems.
Let us take as an example the theorem

(0 KxYx FA(<2:)

(For perspicuity we employ here and in other illustrations the informal

notation permitted by our conventions on superscripts and parentheses;
in general formulations, however, the words ‘term’ and ‘formula’ are
always to be understood in the official sense.) In forming instances of (i),
we must make replacements that do not change its logical structure, that
is, replacements of components other than sentential connectives, quan­
tifiers, and parentheses. The replaceable components fall then into three
categories: predicate letters, operation letters, and variables.
Let us first consider the replacement of predicate letters. We wish to
say that the formula

(2) t\X ^ G(.v^) ^ G( A(<s:) y )

comes from (i) by su b stitu tion for the p r e d ic a t e letter ‘F ’ . We may think
of (2) as obtained from (i) in two steps.
We first consider the formula

(3) ' GCaj)

(which may be called the su b stitu en d , and in which, the reader will note,
the variable 'a' is free), and in (i) replace all occurrences of ‘F ’ by (3)
enclosed in braces. Thus we obtain
(4) { ~ G (aj)} X ^ ( ~ G (a j)} A ( z )
(which is not a formula).
Next we replace each part

of (4), where ^ is a term, by

GU y)
158 • ‘All* and ‘Some*, continued Chapter IV

that is, the result of replacing in (3) all free occurrence of ‘a’ by Thus we
obtain the formula (2).
We consider now the general situation. Let S be a /i-place predicate
letter occurring in a symbolic formula <f>. Let be a symbolic formula
that is to be substituted for S. (In the example above, S is <f>is (i), and
X is (3).) The substitution takes place in two steps, as follows:
I. Throughout (f)y replace S by x enclosed in braces.
II. In the expression resulting from I (which will in general not be
a formula), successively consider each part of the form

{x)^l • ik ,
where ^1, . . . , are terms. Replace each such part by a certain formula—
in fact, by the formula obtained from x by replacing all free occurrences of
'a' by ^1, by ^2, etc., up to the ^th variable (in the standard ordering
of variables), whose free occurrences are to be replaced by

Certain restrictions are necessary in connection with substitution.

Otherwise we could obtain nontheorems by substitution on theorems.
For example, let the formula in which the substitution is to take
place, be
(5) AxFx AjcFB .
((5) is obviously a theorem.) By substituting ‘G(:va)’ for ‘F*, we secure
first
Ax{G{xa)} X A:v(G( a:«)} B ,
and then
(6) AxG(xx) A:)cG( a:B) .
But this formula is not a theorem (according to exercise 74 of this
chapter).
As a second example, consider the theorem
Ax F jc ^ Fy ,
and substitute for ^F* the formula
WyG{ay) .
We obtain first
Ax{YyG(aj)} X -> ^ y G {a y )]y ,
and then
(7) AxVjG(xy) ^yG{yy) .
But this formula too is not a theorem (according to exercise 75).
 Section 8 Substitution . 159

The difficulties uncovered by these two examples result from a clash

of variables and can be avoided by requiring that the substituend and the
formula in which the substitution is to take place contain no common
variables. Accordingly, we say that a symbolic formula xfs comes from a
symbolic formula <j> by proper substitution of a symbolic formula x for a
^-place predicate letter 8 just in case p can be obtained from (f> by steps I
and II above, and in addition there is no variable occurring in both <j>
and X-
We turn now to substitution on operation letters. We wish to say that
the formula
(8) AjcF jc -> FB{zy)
comes from (i) by substitution for the operation letter ‘A ’. As in the case
of (2), we may think of (8) as obtained from (i) in two steps.
We consider a certain term, the substituend^

(9) .
and in (i) replace all occurrences of ‘A ’ by (9) enclosed in braces, obtaining
(10) S x F x F{B(izjy)}2:

(which is not a formula).

Next we replace each part

of (10), where ^ is a term, by

that is, the result of replacing in (9) all occurrences of 'a^ by Thus we
obtain the formula (8).
Again we may ascend to the general situation. Let S be a k-plzct opera­
tion letter occurring in a symbolic formula (^, and let >7 be a symbolic
term that is to be substituted for S. (In the example, S is 'A \ (f> is again (i),
and 7] is (9).) The substitution takes place in two steps, exactly analogous
to those involved in substitution on predicate letters.

I. Throughout (f>y replace S by 7^ enclosed in braces.

II. In the expression resulting from I, successively consider each part
of the form
{n} ii - * • ik ,
where ..., are terms. Replace each such part by a certain term—
in fact, by the term obtained from 7} by replacing all occurrences of 'a'
by ii, ‘A’ by ^2, etc., up to the Ath variable, whose occurrences are to be
replaced by Ck-
i6o . ‘A ir and ‘Some*, continued Chapter IV

If substitution is to lead from theorems only to theorems, a restric­

tion is again necessary, as the following example indicates.
Let <f)y the formula in which the substitution is to take place, be

(II) Ay[Fy -> G A (j)] ^ AyAs[Fz GA(2)]

This is a theorem (according to exercise 58). By substituting the term
'B(yay for the operation letter ‘A ’, we secure first

Ay[Fy -> G{B(ji7)}(y)] ^ AyAz[Fz -> G(B(ja)}(2:)]

and then
(12) Ay[Fy GB(jyy)] AyAz[Fz -> GB(j'^)]

But (12) is not a theorem (according to exercise 76).

The difficulty here can be avoided by exactly that restriction which
was introduced in connection with substitution on predicate letters.
Accordingly, we say that a symbolic formula ifj comes from a symbolic
formula (f) by proper substitution of a symbolic term rjfor a ^-place operation
letter S just in case 0 can be obtained from (f> by the steps I and II above,
and in addition there is no variable occurring in both (j) and rj.
Let us now consider substitution on variables. For free occurrences
of variables the relevant notion has already been introduced on page 148;
that is, the notion oi proper substitution of a term for a variable. For bound
occurrences of variables no notion of substitution is required, for the role
it would play can be handled by alphabetic variance.
Thus we have three kinds of proper substitution—for predicate letters,
for operation letters, and for (free) variables. An instance of a symbolic
formula is obtained by iterated proper substitution; also, we shall con­
sider each symbolic formula as an instance of itself. Thus a symbolic
formula p is said to be an instance of a symbolic formula <f>just in case
ifj is <f> or obtainable from (f) by one or more operations of proper substi­
tution.
For example, the theorem

(13) AxFx -> Fy

has as an instance
Ax{Fx V Gx) -> (FA v GA)
In this case we have made two proper substitutions; for T ’ we have
substituted '(F^7 v Ga)’ and for y we have substituted *A\ We may
indicate this sequence of substitutions diagrammatically as follows:

(Fa V G«)
Section 8 Substitution . i6i

Here the order in which the substitutions are made is unimportant, but
this is not always true. For example,
Ajc(VyFj V G.v) (VyFjv v GA)
is also an instance of (13), in view of the sequence of substitutions
J F
A {WyFy V Ga)
The reverse sequence,

{WyFy V Ga) A
would lead, however, to an improper substitution on 'F \
Our new notion of an instance subsumes the earlier notion of an instance
of a sentential theorem. To see this, it is sufficient to consider substitution
on o-place predicate letters.
As before, the relation of instance is intended to hold only between
symbolic formulas. This restriction is now to some extent forced upon us;
to introduce a satisfactory notion of instance for formulas involving
phrases of quantity and their stylistic variants would be extremely labo­
rious.

EXERCISES
53. Substitute (according to steps I and II) each of the formulas
(i) - (x) for T ' in the theorem
F(xy) -> yzF{zy) .
In which cases is the substitution proper? Parts (i) and (ii) are solved
for illustration.
(i) G{hha)
(Steps I and II lead to
G{yyx) -> yzG{yyz) ,
and the substitution is proper.)
(ii) G{bxa)
(In this case substitution leads to
G(yjc.v) -> yzG{yxz) .
The substitution is improper, for ^x' is common to (ii) and the theorem
above.)
(iii) G(baa)
(iv) ywG{awb)
i62 . *Air and ‘Some’, continued Chapter IV

(v) VyGy
(vi) WwGw
(vii) G{abc)
(viii) G(bzv)
(ix) G(azb)
(x) Ga

54. For each of the formulas (i) - (x) below, state whether it comes
from the theorem
(14) Fx -> WyFy
by proper substitution of a formula for ‘F*; if so, indicate the relevant
substituend. Parts (i) and (ii) are solved for illustration.

(i) G(:c;2r) ->Vy ^ G{yz)

(The formula (i) comes from (14) by proper substitution of
G{az)
for T \)
(ii) G(ry) -> Vy G(>7)
(The formula (ii) does not come from (14) by proper substitution on
‘F ’. As the reader can verify, ‘ G{ayY is the only substituend that will
permit the passage from (14) to (ii); but a clash of variables makes
this substitution improper.)

(iii) G(;c:v) -> VyG(yjv)

(iv) G(:v:3c) YyG(xy)
(v) G{xx) \yG{yx)
(vi) /\wG{wx) \yNwG{wy)
(vii) /\wG(ww) \yf\wG{wy)
(viii) NwG{ww) -> \yNwG{ww)
(ix) G(jcjc) yyG{yz)
(x) P ->V yP

55. Substitute each of the terms (i) - (x) below for ‘A* in the
theorem
A;cFA(jC5r) YyFA(y;2r) .
In which cases is the substitution proper?
(i) B{C{b)a)
(ii)
(iii) B(a)
(iv) B(6)
(v) B
(vi) B{acb)
Section 8 Substitution • Z63

(vii) B(C(a))
(viii) w
(ix) a
(x) B (A (H C{a))

56. For each of the formulas (i) - (v) below, state whether it comes
from the theorem
F(A(«j) j ) ^ V* FiAixjv)^)
by proper substitution of a term for ‘A’ ; if so, indicate the relevant
substituend.
(i) F(B[C(^)]>-)-^VxF(B[C(^)]3')
(ii) F B{zyy) VxFB{xyy)
(iii) F{A{yz)y)->\/xF{A{yx)y)
(iv) F(j?3/) -> V;cF(^jv)
(v) F(>’>-) ^ VxF(>:y)

57. For each of the following pairs of formulas state whether the
second is an instance of the first obtainable by the indicated sequence
of substitutions. Cases (i) and (ii) are solved for illustration.

(i) Fx-> V yF y G(:va:) WyG{xy)

F z
G{za)
The second formula is an instance of the first obtainable by the indi­
cated sequence of substitutions, (The reader should compare this
case with part (iv) of exercise 54.)
(ii) F{xy) ->WyF{xy) [V;cH:v V G(y^)] Vy[VA;Hic v G(y;2r)]
F
[VxUx\fG{bz)]
Here the second formula is not an instance of the first obtained by the
indicated substitution. (The second can, however, be obtained as
an instance of the first by the following sequence of substitutions:
jc F
•)
r [WxUxwG{ba)]
(iii) Aa;(Fjc -> P) {WxFx P) ; Ajc[F(:vy) A;sfG(y;sr)]
[V;cF(jcy)->/uG(y;^)]
w
AzG{yz) F(azo) y
(iv) ^ WxFx Ajc(F^ Gx) ^WxVyF{xy)->
Ax\WyF{xy) -> A^G(A(jc) z B(.v))]
164 • ‘A ir and ‘Some’, continued Chapter IV

F
yyF(fljy) A^G(A(a) z B(a))

(v) KxYK{x B) yx¥k(xx) ; AxFD{B{xz)) VxFD{B{xx))

B A
D(B(^z6))
(vi) Fy WxFx ; F(ry) V«x;F(:v:v)
F
¥{za) X
58. Show that formula (ii) of page 160 is a theorem.

9. A bbreviated derivations reconsidered. Using the notions of

the last two sections, we can now introduce more sweeping methods of
abbreviating derivations than those carried over from chapter III. To
begin with, we can drop from clause (7) the restriction to sentential theo­
rems, since the general notion of an instance is now available.
Thus we replace (7) as formulated in chapter III by the following:

(7) I f (f) is an instance of an alphabetic variant of a theorem that has

already been proved, then cf> may occur as a line, {Annotation: the number
of the theorem in question, sometimes together with a diagrammatic indication
of the sequence of substitutions involved,)

Clause (8) of chapter III is no longer needed; any application of QN

can be obtained by the revised clause (7) (applied to T203 or T204),
together with one or two sentential steps. We replace the old clause (8)
by a new clause involving the notion of alphabetic variance:

(<§) A symbolic formula may occur as a line if it is an alphabetic variant

of an antecedent line, {Annotation: 'AV* and the number of the antecedent
line,)

Symbolic formulas differing only in equivalent parts are themselves

equivalent, and this fact provides us with another useful device for
abbreviating derivations. To incorporate this device into our system we
replace the former clause (9) (which will reappear as clause (10)) by the
following:

(9) I f <f), <f>' are symbolic formulas such that

Section 9 Abbreviated Derivations Reconsidered . 165

is an instance o f an alphabetic variant o f a previously p ro ved theoremy and py

p ' are symbolic form ulas such that p ' is like p except fo r having one or more
occurrences o f p ' where p has occurrences o f py then

P^P'
may occur as a line. {In such a case we may call

p <—> i/f'

an interchange of equivalents on the basis o f the previously pro ved theorem;

as annotation we use ' I E ' with a parenthetical reference to this theorem.)

As before, we permit the compression of several steps into one, and the
following clause makes provision for this.

( j o ) a symbolic form ula may occur as a line i f it follow s fro m antecedent

lines by a succession o f stepSy and each intermediate step can be justified
by one o f clauses (2), {5a)y (7), (<S), or (9). {T h e annotation should determine
the omitted succession o f steps by indicatingy in order o f applicatiorty the
antecedent lineSy the premiseSy the inference ruleSy and the previously p roved
theorems employed. W e should also make such indication o f terms and form ulas
involved in the intermediate inferences as may be required to dispel remaining
ambiguity.)

The special annotations 'S C ’ and 'C D ’, which were introduced earlier
to indicate special kinds of compression, will continue to be used. We
shall now use the annotation 'Q N ’ as an abbreviation for 'T203, BC, M P’
or 'T204, BC, M P’ ; it will therefore comprehend all inferences that its
previous use would have justified.
Clauses (7) - (10), like the earlier (7) - (9), satisfy the requirements given
on page 58; that is, they are theoretically dispensable, and derivations
constructed on their basis can be automatically checked for correctness
(at least when membership in the class of premises is automatically
decidable and annotations are present). It should be mentioned that an
exact demonstration of the theoretical dispensability of clauses (7 ) -( 10 )
would be more involved than in the case of previous abbreviations.
To obtain a proof of an instance of a theorem from the proof of the
theorem itself it is not sufficient to make simple replacements on the
original proof: the restrictions imposed on variables in our basic clauses
will sometimes necessitate a systematic change of variables in the original
proof before the replacements can be made. The rule IE (and also AV
in some of its applications) raises a special problem. IE cannot be regarded
as an abbreviation of the same kind as QN, SC, and CD, for there is no
one theorem to whose instances all applications of IE correspond. Theo­
rems particularly useful in eliminating applications of IE are:
i66 . *A11* and ‘Some*, continued Chapter IV

T96 (P Q) <-> (/^P ^Q )

T97 (P <-> R) Л (Q <-> S) -> [(P -> Q) (R -> S )]

T98 (P R) Л (Q <-> S) -> (P A Q <-> R Л S)

T99 (P <-► R) A (Q <-> S) ^ (P V Q <-> R V S)
T 100 (P R) A (Q <-> S) -> [(P <f-> Q) (R <-> S)]

T 2 13 Лл:(Рл: <-> Ол:) ^ {hxVx <-> ЛлсОл:)

T 2 14 Лл:(Рл: Gx) Q/xFx ч-> WxGx)

Elimination of an application of A V would require (in addition to the
procedure necessary with IE) proving a suitable instance of T 2 31 or
T232.
The economy achieved by clauses (7) - (10) is illustrated by the following
proofs:
1. Shew^WxQ/xFx -> Рл:) (T229)
2. WxQ/yFy Рл:) <->
(Vypjy ^ Ул:Рл:) T220, ,
(vvFv)
\Wy¥y)
У^(Ул:Рл: -> Рл:) <r^
(WxFx VxFx) 2, AV
yx{WxFx -> Рл:) 3, BC, T i , M P

(Compare with the proof of T229 on page 112 .)

I. Sket» AxAy¥{xy) AyAxF(yx) (T 25 7 )
AxAyF{xy)
AxAzF{xz) 2, AV
AyAzF(yz) 3, AV
AyAxF{yx) 4, AV

(Compare with the proof of T 257 constructed in solving exercise 48 of

chapter IV, p. 155.)
I. Shot» ~ V x(F x A G x )•<-> Ax(Fii ~ Gx) (T243)
2* ■/Ул:(Рл: A Ол:) I / P \
Лл: (Рл: A Ол:)
' Ул:(Рл: A Ол:) <->
Лл:(Рл: -> ^ Ол:) 2, 1Е (Т з 9),ВС ,
МР
The passage from line 2 to line 3 in this proof would look, if expanded,
as follows:
Section 9 Abbreviated Derivations Reconsidered . 167

2. Vjc(F jc a G:v) <-> Ajc^ (Fx a G^)

2a. [ V:v(F jc a G:v) Ax (F^ a G jc)] <->
[ V;v(F;c A G ^ ) A:v(F^ -> G:v)]
2b. ['^yx(Fx A G jc)<-> A;V'^(F:v a G x)] ->
[ ^ VAi(Fjc A Gx) <-> Ax(Fx -> ^ G:\;)] 2a, BC
3. V jc(F jc a G jc) <-> A:v(F:v G;c) 2, 2b, MP
This sort of inference will occur frequently. We shall therefore often omit
‘B C and ‘M P’ from an annotation like ‘2, IE(T39), BC, M P’.
(Compare the proof given here of T243 with the one given on page 115.)
Another example:
I. Sktm '^WxyyF{xy) AxAy^F{xy)

2, QN
3, IE(T204)

In this proof the reader should observe that he could not pass directly
from line 3 to line 4, as he passed from line 2 to line 3, by means of QN.
The passage from line 3 to line 4 would look, if expanded, as follows:
3. Ax YyF(Arv)
3a. A»<^Vy F(jcy)<-> AjcAj ’ r>^F(xy) Ie ( T 2 0 4 _ f l ]
\ ^Fixa)}
3b. AjC'^YyF(ry) AxAy^ F(ry) 3a, BC
4. Ax Ay F(xy) 3, 3b, M P
(The interchange of equivalents that occurs in line 3 a results from the
substitution of ‘F(^a)’ for ‘F ’ in *^WyFy<-> A y '^ F y \ which is an alpha­
betic variant of T204.)
Variants of the confinement laws, useful for the exercises to follow,
can now be obtained without the tedious proofs they would have required
in chapter III.
T272 I. Show Ax(Fx A P ) AxFx a P
2. A^(P A F jc) <-> P a AxF x T 2 15
3- Ax{Fx A P) <-> P A AxFx 2, IE(T24)
A jc( F : v a P) <-> AxFx a P 3, IE(T24)
4-
T273 V;c(F a: a P) <-> WxFx a P
T274 Ax(Fx V P) <-> AxFx V P
T275 V:v(F:)c v P) ^ WxFx v P

E X E R C ISE S, GROUP I
Using the abbreviations of the present section give brief proofs
of the following theorems. Exercise 59 is solved for illustration.
i68 . *Air and ‘Some*, continued Chapter IV

59. (AjcFa:<-> V:vGjc)<->VjcyyA^Aw([FAi:->Gy] a [G-2:->Fz(;]) (T263)

1. S k i m [t\xVx <r-> yxGx) <-> yx\lyf\zKw{\Fx - > Gjv] a [G^r -> Fzi?])
2. {hxYx <-> yxGx)
<-> ([A;cF;c VjcG^c] a \Wx G x A jcF a:]) T 81
<-> (Vjc[F;v -> WxGx] A [Vx G jc A jcF jc]) 2, AV, IE(T222),
AV
-i-> Vjc([Fje -> WxGx] A [V;cG;v -> A jcF;v]) 3, AV, IE(T273),
AV
5* <-> Vjc(Vy[FA: -> G y] a [Wx G x -> A^vF^i]) 4, AV, IE (T 22 o)
6. V;cVjy([FA: G j] A [V:x:G^->A: vF jc]) 5, IE(T273)
7* <<-> VjcVy([Fjc -> Gy] A A2t[G2 -> A jcF a;]) 6, AV, IE (T 22 i )
8. <-> WxWyAz{[Fx -> Gj^] A [G^ -> AxFx]) 7, IE(T2I5)
9- ^ Vx\fyAz{[Fx Gy] A A« j[G2 -> Fzv]) 8, AV, IE(T2I9)
10. <->VjcYyA2Aw([F;c Gjy] a [G^r -> Fw]) 9, IE (T 2 I s )

(It is understood, in the proof above, that the left side of line 2 is to be
carried over to the blanks of lines 3 - 10.) The passage from line 2 to
line 3 would look, if partially expanded, as follows:

2. (AjcF a: VjcG jc)

^ ([A:cF:v VjcG:x:] a [VjcG a; A^cF^v])
2a. <-> {[AxFx -> VyGy] A [V:vG a; -> A:vF jc]) 2, AV
2b. <-> (V:v[F;c VyGy] a [V;cG^ -> Aj^Fjc]) 2a, IE(T222)
3. <-> (Va:[F:x: -> WxGx] A [VjcG a: -> AajF jc]) 2b, AV
60. V;c(F:)c -> AxF:v) (T230)
61. AJc(Vy[H(A:y) A Fy] -> Vy[H(:vy) a Gy]) <->
AxAyyz{H{xy) A Fy -> H(jc2) a G2) (T26S)
62. A^(F;c AVyG(:xy)->Vy[H(^y) a A2 J(^y2)])<->
AMyVwA2r[F;v A G(:xy) H{xzv) a ] { x w z ) ] (T266)
63. VyA^(Fjc -> Gy) AjtVy(F;c -> Gy) (T268)
64. AxAyWz{Fx A Gy -> Hz) <->
AyyzAx{Fx A Gy -> H2) (T271)

E X E R C IS E S , G R O U P II

Symbolize each of the following arguments, indicating the scheme

of abbreviation used, and derive the conclusion of the symbolization
from its premises. Exercise 65 is solved for illustration.

65. Alfred shaves all and only those inhabitants of Berkeley who
do not shave themselves. Alfred is an inhabitant of Berkeley. There­
fore Alfred does not shave himself.
Section 9 Abbreviated Derivations Reconsidered . 169

On the basis of the scheme of abbreviation

: a is an inhabitant of b
S 2 : a shaves h
A® : Berkeley
BO : Alfred ,
the argument becomes
Ajc(F(jcA) -> [S(Bjc) - S{xx)]) . F(BA) /. ^S(BB)
Its validity is established by the following derivation:
I. Show S(BB)

ist premise, UI
2nd premise, 3
MP, BC, 2, MP

66. Alfred shaves all and only those inhabitants of Berkeley who
do not shave themselves. Alfred is an inhabitant of Berkeley. Therefore
Alfred shaves himself.
(Exercises 65 and 66 justify the assertions under (2) of page 3.)
67. Every student is able to solve some problems and not able
to solve some problems. Some teacher is able to solve all problems.
Some teacher is not a student.
68. All members of the Board of Regents distrust every member
of the Communist party. Every proponent of Marxism is a member
of the Communist party. Some member of the Board of Regents does
not distrust some proponent of Marxism. There is a Communist
on the faculty.
69. No student likes every course he takes unless he takes no
courses but philosophical studies. No mathematical study is a philo­
sophical study. Some student takes nothing but mathematical studies
and likes every course he takes. Some student does not take any
courses.
70. No member of the club owes a debt to the treasurer of the
club. A member of the club has not paid the entrance fee only if he
owes a debt to the treasurer of the club. /. If the treasurer of the
club is a member of the club, then he has paid the entrance fee.

♦E X ER C ISE S, GROUP III

71. For each of the following monadic formulas, find an equiv­

alent monadic formula without overlay (see p. 153).
YyAx(Fjc V2:[G2r - ■ H^])
SxyyiYx '^Fjv)
AjcAy (F^ <-> Fjy)
lyo . 'A ir and 'Some*, continued Chapter IV

72. Formulate a general procedure for transforming any monadic

formula into an equivalent monadic formula without overlay.

10. * Invalidity. As Church has shown in [2], there is no automatic

test for the validity of arguments formulated within the present symbolic
language. There are, however, several ways of showing arguments
invalid.
In the first place, the method of chapter III, that of truth-functional
expansions^ remains available. It is, however, applicable only to a restricted
class of arguments. It must be used only in connection with symbolic
arguments whose premises and conclusion are sentences containing no
terms other than variables ; incorrect results might otherwise be obtained.
Further, even among arguments of this sort, cases will now arise of
invalid arguments whose invalidity cannot be established by this method;
an example is argument (i) below. In the present section, we therefore
introduce methods of wider applicability.
By an English translation of a symbolic argument on the basis of a given
scheme of abbreviation we understand another argument whose premises
and conclusion are English translations, on the basis of that scheme, of
the respective premises and conclusion of the symbolic argument—in
other words, an English argument having the symbolic argument as a
symbolization on the basis of the given scheme. As before, we say that an
English argument is false if its premises are true sentences of English
and its conclusion is a false sentence of English.
It happens that if a symbolic argument whose premises and conclusion
are sentences has, on the basis of some scheme of abbreviation, an English
translation that is false, then the symbolic argument cannot be valid.
(This fact will not be proved here; it follows from Montague and Kalish
[I]-) Thus, to establish the invalidity (in the quantifier calculus) of a
symbolic argument, it is sufficient to show that it has a false English
translation.
For example, the argument
\x¥x . SxG x .\V^(Fjc A G jc)

is invalid, because it has the following, clearly false, English translation:

Something is an odd number. Something is an even number.
Something is both an odd number and an even number.
In showing arguments invalid it is often necessary to determine the
truth or falsehood of English sentences of a rather complex structure.
Intuition is not in all cases an unequivocal guide. For instance, the sen­
tence
I f snow is red, then blood is green
 Section lo Invalidity . 171

may seem true to some, false to others. Our deductive system, however,
provides assistance in the following way.
We say that an English sentence is derivable from other English sentences
just in case there is a valid argument with the latter as premises and the
former as conclusion. If an English sentence is derivable from true
English sentences, then it will be regarded as true, and if a false English
sentence is derivable from it, then it will be regarded as false. Thus, for
example, the sentence above is to be considered true because it is derivable
from the indisputably true sentence ‘Snow is not red’. Also, if we were to
doubt the falsehood of
For each jc, x is red,
it would be sufficient to consider the obviously false sentence
The White House is red.
As a further example, consider the sentence
There is an object x such that if x is identical with 2, then x
is not identical with x.
To show its truth it is sufficient to consider the true sentence
If I is identical with 2, then i is not identical with i.
The method of showing invalidity given above is directly applicable only
to those symbolic arguments whose premises and conclusion are sentences.
We can, however, extend the method to arbitrary symbolic arguments.
As before, it happens that a symbolic argument is valid just in case
each of its closures is valid. (For a definition of closure^ see p. 126.) Thus
to establish the invalidity of a symbolic argument it is sufficient to find a
false English translation of one of its closures.
As a matter of fact, we may make a stronger statement: a symbolic
argument is invalid (in the quantifier calculus) just in case a closure of it
has a false English translation. Thus, for any symbolic argument, either
a derivation may be supplied or else a false English translation of one of
its closures will exist. (Yet there is no automatic way of deciding which
alternative holds.)
The task of establishing invalidity can in some cases be simplified.
The argument
(I) AAiAjyA2 [F(xy) A F{yz) - F( x2)]
AxWy F{xy) .*.Va:F(jca;) ,

has no obvious false translations. At first inspection, the argument

(2) For each Xy for each jy, for each Z y i f x < y and y < Zy then
X < z. For each Xy there is an object y such that x < y.
Therefore there is an object x such that x < x
172 . *A11* and ^Some’, continued Chapter IV

might present the appearance of a false English translation of (i). But

this is not clear, for the second premise of (2) might reasonably be re­
garded as false. It does not assert that
for every number there is a greater,
which is true, but that
for every thing there is a greater,
which is doubtful, (Though such exceptions as the universe are amusing
the doubt attaches to any object that is not a number; the meaning of
* < * is not clearly determined for such objects.) The difficulty would
seem to be resolved if we could in some way restrict the phrases of quan­
tity in (2) so as to apply only to numbers. This is made possible by con­
sidering, instead of (i), another argument, called its relativization.
If ^ is a symbolic sentence and tt a i-place predicate letter not occurring
in 0, then by the relativization o f (j> to tt we understand that sentence
which is obtained from <f> by replacing each part of the form
Aa0
or
Va0
where a is a variable and i p a formula, by
Aa(7ra
or
Va { ttol a i p )

respectively. For example, the relativization of the second premise of

(i) to ‘G i’ is
A:v(G;x: -> Vy[Gjy a F(rv)]) .
If S is a ^-place operation letter and tt a i-place predicate letter, then
by the closure axiom for S with respect to t t we understand the sentence
TT 8

if k = o, or the sentence
A^l . . . h X ]c [7 T X \ A ... A TTX jc “ > 7t8(;vi . . . X k ) ]
if k ^ 1. For example, the closure axiom for ‘A ®’ with respect to ‘G^' is
GA ,
and for with respect to ‘G^' is
AjciAx:2[G^i a G.V2 “> G A(;vi;v2)] .
Section 10 Invalidity . 173

Let A he a symbolic argument whose premises and conclusion are

sentences, let tt be a i-place predicate occurring in none of these, and
let Si, . . . , 8n be all the operation letters occurring in A. Then a rela-
tivization of A to TT is an argument obtained from A by replacing its
premises and conclusion by their relativizations to tt and adjoining, as
additional premises, the closure axioms for 81, . . . , S;i with respect to tt,
together with the sentence
WxTT X .

For example, a relativization of (i) to is the argument

(3) Ax[Gx -> Ay{Gy Az[Gz (F(ry) a F(jy<s:) -> F( jc2:))])] .

A:jc(G^ Y y[G j A F(ry)]) . V;cG:v .’.V^(G:v a F( x’:v)) .

(Other relativizations to ‘G^^ could be obtained by changing the order of

the premises.)
If Ay B are symbolic arguments whose premises and conclusions are
sentences, then A is said to be a relativization of B if there is a i-place
predicate letter t t not occurring in B such that ^ is a relativization of B
to TT,
Now it happens that if an argument whose premises and conclusion are
sentences is valid, then so is its relativization to any i-place predicate
letter not occurring in it. Thus to show such an argument invalid it is
sufficient to find a false English translation of one of its relativizations.
For example, the invalidity of (i) is established by the following false
translation of (3):
For all numbers Xy yy Zy \i x < y and y < Zy then x < z.
For each number Xy there is a number y such that x < y.
There is a number. Therefore there is a number x such that

As another example consider the argument

AxF[A{x) x] ,\yxF[A{x) A{x)] .

To establish the invalidity of this argument we consider the relativization

Ax{Gx F[A(:v) x]) , Axi{Gxi GA(xi)) . WxGx
.*. yx{G x A F[A{x) A{x)]) y
together with the following false translation of the latter:
The successor of a number is greater than that number.
The successor of a number is a number. There is a number.
Therefore the successor of some number is greater than
itself.
174 • *Air and ^Some*, continued Chapter IV

The methods given above for showing invalidity suffer from two
drawbacks. First, two of the notions involved, that of an English translation
of a symbolic sentence, and that of the truth of an English sentence, have
not been precisely characterized. This difficulty could be removed as
follows. We could restrict ourselves to literal translations into English,
and, further, limit consideration to that part of English which might be
called elementary arithmetic (that is, the arithmetic of nonnegative
integers). The first restriction would avoid the looseness involved in the
notion of stylistic variance, and the second would provide the possibility
of a precise characterization of truth; for the latter see Tarski [2]. (In
practice we shall not make these restrictions but proceed informally in
showing invalidity.)
The second drawback concerns the methods of showing sentences
true or false. If we deal with sentences of ordinary English, empirical
knowledge will in general be needed. This need can be obviated by the
device, mentioned above, of replacing English by the language of ele­
mentary arithmetic. But even then the mathematical methods required
for showing truth and falsehood will in some cases be very deep. Indeed,
there are invalid arguments whose invalidity cannot be established
within the mathematical system presently used. (This is one version
of the incompleteness theorem of Godel [2].) For such arguments there
will be translations into elementary arithmetic that are in fact false but
which we shall be unable to prove false.

E X E R C ISE S

The reader will recall that a symbolic formula <f> is 3. theorem if

and only if the argument with no premises and with ^ as its conclusion
is valid. One can thus obtain from the considerations of the present
section methods of showing formulas not to be theorems.
Show that each of the following is not a theorem.
73. The converse of T253 (p. 152).
74. The formula (6) of page 158.
75. The formula (7) of page 158.
76. The formula (12) of page 160.

For each of the following arguments show that it is valid or show that
it is invalid. Exercise 77 is solved for illustration.

77. V x ( F a ; a Ay [Gy a Wz ( F z a H(y2r)) H(y:v)]) .

Vy(Gy A V2r[F2T A H(y2)]) Ay[Gy
.*. V a :( F ; c a H(yx)])
A false English translation of this argument is the argument
There is an even prime such that every prime less than
or equal to some even prime is less than or equal to it.
Section 10 Invalidity . 175

Some prime is less than or equal to an even prime.

There is an even prime such that every prime is less
than or equal to it.
(By a prime is understood any integer x other than i or —i that is
divisible (without remainder) only by i, —i, x, and —x,) The false­
hood of this argument becomes apparent once it is recognized that
there is no largest prime but that there is a largest even prime, in
particular, 2.

78. F(*y) ¥{yx) Y{xx)

79. F(xy) A F(y2r) F(;c2r) . F{xy) -> F(jyje) F(^jc)
80. F{xy) A F(y2r) - > F ( : v 2 ) . F(jcy) - > F(y^)
Ax[Wy¥(xy)->F{xx)]
81. AxAy{Fx A Fy -> FA(jcy)) . yx(Fx a Ay[Fj -> GA(3^a;)])
V^(Fjc a G x)

82. Aje(A3^[Fy-> Fi{xy)] a H(xy)]) .

.*. A;c(Ay[H(jcy) Gy] Ay\Fy '^H(A;y)])
83. Ajv[Fy A V jc G(^yA)-> H(yB)] A y ( V ;e [I jc a H ( 3^^)] - > ]y) .
A x A y \ y z G { x y z ) A ] y - > K(y;e)] . A jr A jv [ G ( jc y A ) A F y - > K (jy jc )]
/. Wx
84. FA .*. Ax(Fx Gx) <-> Aa:([F:v a Gx] y [^ F x a GA])
85. A;c(yy[F(j^A) A G{xy)] V>;[F(3;A) ^ U{xB)]
.-. Ay[F(>^A) A G(By)->H(BB)]

Give an interesting symbolization (see p. 77) of each of the follow­

ing arguments, and for each symbolization either show that it is
valid or show that it is invalid. Exercise 86 is solved for illustration.

86. No member of the club owes a debt to the treasurer of the

club. A member of the club has not paid the entrance fee only if he
owes a debt to the treasurer of the club. The treasurer of the
club has paid the entrance fee.
On the basis of the scheme of abbreviation
F2 : a is a member of b
G2 : a owes a debt to b
R2 : a has paid 6
A® : the club
B^ : the treasurer of a
C® : the entrance fee ,
No. 86 has the symbolization
^ W x [F ( x A ) a G { x B {A ) )] .
Ax[F{x A) A - U{x C) G{x B(A))] /. H(B(A) C) .
176 . ‘A ir and ‘Some*, continued Chapter IV

A relativization of this symbolization to

- yx{]x A [F{x A) AG{x B(A))]) .
AxQx -> [F(jc A) A - H(jc C) -> G(^ B(A))]) .
JA . AxiUxi -> J(B(^i))] .
JC . Wx]x /. H(B(A) C) ,
has the following translation:
No integer equals both 2 and 2^. For each integer Xy if x
both equals and does not equal 2, then x equals 2^. 2 is
an integer. The square of an integer is an integer. 2 is
an integer. There is an integer. Therefore 2^ equals 2.
On the basis of elementary arithmetical facts it is clear that the pre­
mises of the last argument are true and the conclusion false; hence
the symbolization above is invalid.

87. There is a number such that the product of it and any number
is even. There is an even number.
88. The distance between A and B is less than the distance between
B and C. The distance between B and C is less than the distance between
A and C. The distance between A and B is less than the distance
between A and C.
89. Students who solve every problem on the final examination
also solve every problem on the midterm examination. Some student
does not solve the hardest problem on the midterm examination.
There is a problem on the final examination that some student
does not solve.
90. No student who fails some course that Rudolf teaches fails
all the courses that Alfred teaches. Some student fails all the courses
Alfred teaches and also fails all the courses that Rudolf teaches.
/. If Rudolf teaches any courses, then there is a course that every
student fails.
91. No member of the club owes a debt to the treasurer of the club.
A member of the club has not paid the entrance fee only if he owes a
debt to some member of the club. /. If the treasurer of the club is a
member of the club, then he has paid the entrance fee.

For each of the following formulas, either show that it is a theorem

or show that it is not a theorem.
92. WzWx F(:v2r) AzAzvAy(F{yzv) <-> V^[F(^2t) a ^ F(jyy)])
93. AzWyAx(F{xy) F(jc2t) a G jc) VjV;c(F(xy)<-> V<2tF(a;2:) a G^)
94. WyAx(F(xy) V2rF(;c2r) a G;c) -> AWjVjr(F(xjv)<-> F(.v2r) a G x)

95* '^y>'AM2:[F(xV2:)<->
Section n Paradoxical Inferences . 177

I I. * P aradoxical inferences. We have sho\vn earlier that if the

restrictions on variables are ignored, it is possible to validate false English
arguments. Certain examples suggest that this can be done even when
all rules are strictly observed. One such case was discussed on page 122
but was ruled out by the requirement concerning apparent variables im­
posed on schemes of abbreviation.
Another example will indicate the necessity of excluding from schemes
of abbreviation those apparent variables which might be called ‘intuitively
bound\ Consider the theorem

U) A^AjyF(xy) -> V;cF(^;c)

and the scheme

(2) F2 : there is an object a such that a differs from b .

In the process of obtaining an English translation on the basis of (2), (i)

becomes first

AxAy {there is an object a such that a differs from b] {xy) ->

yx {there is an object a such that a differs from b] {xx) ,
next

Ax Ay there is an object such that x differs from y ->

yx there is an object x such that x differs from x ,
and finally

If, for each jc, for each jy, there is an object x such that x
differs from j , then there is an object x such that there is an
object X such that x differs from x ,

which has as a stylistic variant

If each thing differs from something, then something differs

from itself.

We should thus have found a false English translation of a theorem, which

contradicts the claim on page 1 7 1. The trouble is that (2) is not a scheme
of abbreviation: its right-hand member has the apparent variable 'a\
Now consider the argument

(3) Phidias’ chryselephantine statue of Athene had a removable

golden garment. There is an object such that Phidias’
chryselephantine statue of x had a removable golden gar­
ment.

On the basis of the scheme

178 • *A11* and *Some\ continued Chapter IV

(4) : Phidias’ chryselephantine statue of a had a removable

golden garment
A® : Athene ,
(3) seems to have the symbolization
FA :y x Y x ,
which is clearly valid. But the premise of (3) is, as a matter of history,
true; and the conclusion of (3) is false. If there is something whose statue
had a removable golden garment, what is it? Certainly not Athene, for
there is no such thing as Athene.
The paradox is solved by observing that if (4) is to be a scheme of
abbreviation, the word ‘Athene’ must be a name. But is ‘Athene’ a name?
In its ordinary sense, no; for in this sense it designates nothing, and a
name according to our characterization must always have a designation.
Another treatment is possible. We may distort the meaning of English
so as to make a name of every expression having the structural charac­
teristics of a name. To do this, we must assign a designation, which may
be quite arbitrary, to each expression that purports to designate but in
its accepted meaning does not. Let us accordingly choose the number o
as the common designation of all such expressions. The argument (3),
then, is valid. But its premise, which asserts that Phidias’ chryselephantine
statue of o had a removable golden garment, is false. To express the in­
tended meaning of this premise, we must now resort to circumlocution
such as.
The chryselephantine statue executed by Phidias and called
‘Phidias’ statue of Athene’ had a removable golden garment.
12. * H istorical rem arks. The observation that every monadic for­
mula is equivalent to one in which no quantifier occurs within the scope
of another quantifier is found in Behmann [1]. The idea of prenex normal
form is due to Peirce (see Peirce [2] and [3]). The process of reduction
to prenex normal form is given in Whitehead and Russell [1].
The notion of an instance of a theorem of the quantifier calculus has
appeared in many forms. Inadequate versions occur in Hilbert and
Ackermann [i], Carnap [i], and Quine [1]. Correct, though highly
complex, forms are to be found in Hilbert and Bernays [i], Hilbert and
Ackermann [2] and [3], and Church [3]. Our formulation is the result of
simplifying and extending the version of Quine [3] and profits from a
suggestion in Pager [1]. ^
The rule IE was demonstrated for the sentential calculus in Post
[i] and for the quantifier calculus in Hilbert and Ackermann [1].
That there is no automatic test for validity in the quantifier calculus
was derived in Church [2] from the results of Church [1].
Section 13 Appendix: Sum m ary . 179

The assertion that a closure of an invalid argument of the quantifier

calculus always has a false translation can be understood as expressing
the completenessy in a certain sense, of this branch of logic. Completeness,
in a somewhat different sense, was demonstrated in Gödel [i] for an axio­
matic quantification theory, and the present assertion can be established
by GodeFs methods.

13. Appendix: summary of the system of logic developed in

chapters I - I V .

INFERENCE RULES
(Here <f>, X to be symbolic formulas and a a variable.)

PRIMITIVE SENTENTIAL RULES:

<f>
Modus ponens (MP)

4>
Modus tollem (M T)

<f> Double negation (DN)

4

<!> Repetition (R)

<!>

(f>Atp <
pAtp Simplification (S)
<!>

4>
Adjunction (Adj)
<
f}Aip
Addition (Add)
(pWtp <pyip
fpynp <pyip
Modus tollendo ponem (MTP)
i8 o . ‘All* a n d ‘S o m e *, c o n tin u e d Chapter IV

(f) ip Biconditional-conditional (BC)

(j) — ijj ijj —
^ (j)
4>
<P Conditional-biconditional (CB)
(f><-> tp
DERIVED SENTENTIAL RULES:
(f> W iff

^ -^ x -^ x
^(f) -> iff iff X ^ ’^ X Separation of cases (SC)
4> X 4>^>l>^x
'^(f) iff Conditional-disjunction (CD)
<f) y iff
PRIMITIVE QUANTIFICATIONAL RULES:

hoL(f> Universal instantiation (UI)

Existential generalization (EG)

j L
Va^
where iff comes from (f> by proper substitution of a term for a;
W(X(f> Existential instantiation (El)

where iff comes from (f> by proper substitution of a variable for a. (See
p. 148 for a definition of ‘proper substitution’ .)

DERIVED QUANTIFICATIONAL RULES:

Aa^ Va

\/(X ^(f> '^AoL<f>

Quantifier negation (QN)
'^W0L<f> Aa

'^(f> ^ W(x,<f>

D IR E C T IO N S FOR C O N S T R U C T IN G A
D E R I V A T I O N F R O M A C L A S S K OF
SYM BO LIC FORM ULAS
(1) I f ^ is any symbolic formula, then
Show <f>
may occur as a line. (Annotation: ‘Assertion’.)
(2) Any member of K may occur as a line. (Annotation: ‘Premise’.)
Section 13 Ap p e n d i x : S u m m a r y . 181

(3) If <f>, i/j are symbolic formulas such that

Show {(f> -> tp)

occurs as a line, then <f>may occur as the next line. (Annotation: ‘Assump­
tion’.)
(4) If is a symbolic formula such that

Show <{)
occurs as a line, then

may occur as the next line; if is a symbolic formula such that

Show '^(f>

occurs as a line, then <f>may occur as the next line. (Annotation: ‘Assump­
tion’.)
(5a) A symbolic formula may occur as a line if it follows from antecedent
lines (see p. 21) by a primitive inference rule other than E L
(5b) A symbolic formula may occur as a line if it follows from an
antecedent line by the inference rule E l, provided that the variable of
instantiation (see p. 100) does not occur in any preceding line. (The
annotation for (5a) and (5b) should refer to the inference rule employed
and the numbers of the antecedent lines involved.)
(6) When the following arrangement of lines has appeared:

Show <f>
XI

Xm >

where none of xi through xm contain uncancelled 'Show^ and either

(i) (p occurs unboxed among xi through xm;
(ii) <p is of the form

i<Pi h)
and ip2 occurs unboxed among xi through xm;
(iii) for some formula x> both x its negation occur unboxed
among XI through xml or
(iv) <p is of the form
A a i . . . A qljc^ ,
i 82 . *Air and *Some’, continued Chapter IV

ip occurs unboxed among the lines xi through xm, and the vari­
ables ai through cnjc are not free in lines antecedent to the displayed
occurrence of
Show <f> ,
then one may simultaneously cancel the displayed occurrence of *Show'
and box all subsequent lines.

The remaining clauses are abbreviatory (in the sense of page 58).
(7) If <l> is an instance of an alphabetic variant of a theorem that has
already been proved, then (f> may occur as a line. (Annotation: the number
of the theorem of which (p is an instance, sometimes together with a
diagrammatic indication of the sequence of substitutions involved.) (For
the notion of instance^ see chapter IV, section 8.)
(8) A symbolic formula may occur as a line if it is an alphabetic variant
of an antecedent line. (Annotation: ‘AV* and the number of the antecedent
line.) (For the notion of alphabetic variance see chapter IV, section 7.)
(9) If (py <p' are symbolic formulas such that
<P^<P'
is an instance of an alphabetic variant of a previously proved theorem,
and ipy ip' are symbolic formulas such that ip' is like ip except for having
one or more occurrences of <p' where ip has occurrences of <py then
ip<-> ip'
may occur as a line. (Annotation: ‘IE ’ (‘interchange of equivalents’),
together with a parenthetical reference to the theorem involved.)
(10) A symbolic formula may occur as a line if it follows from antecedent
lines by a succession of steps, and each intermediate step can be justified
by one of clauses (2), (5a), (7), (8), or (9). (The annotation should determine
the omitted succession of steps by indicating, in order of application, the
antecedent lines, the premises, the inference rules, and the previously
proved theorems employed.) (The use of derived inference rules is com­
prehended under this clause.)

D E R IV A B IL IT Y
A derivation is complete just in case every line either is boxed or contains
cancelled 'Show\ A symbolic formula <p is derivable from a class K of
symbolic formula just in case a complete derivation from K can be con­
structed in which
Show (p
occurs as an unboxed line.
Section 14 Appendix: L ist o f T h e o re m s o f Chapter IV . 183

14. Appendix: list of theorems of chapter IV.

T 24 9 AjcAjyF(i(y) -> Va;yyF(ji3/)

T 250 A;cAjyA2:F(jc^2) -^VarVyVa: ~F(jC3>2r)

T 251 A jcAj F {x y ) <-> AjyAjcF

T 252 VAcyyF(Aí3/) <-> yyVAcF(A^)

T 253 VAcA3;F(xy) -> Aj VacF( a:j )

T254^ y x S y V {x y ) VA:VjJ^F(A;y) v F( jac)]

T 255 ~VyAAc[F(A^) ~ F ( * ac)]

T 256 A2V_yA*[F(Aíy) <-> F(AC3r) A ~ F ( acaí;)] -> ~ V2A a:F( ac2)

T 257 A ajAj F ( xx ) ^ /\ y S x V {y x )

T 258 F ( acA ( ac)) Vy[A2r[F(23/) -> F (2A ( a;))] a F(Ay)]

T 259 Aj Va¡;(Faí: <-> ~ Fy) «-> VacF ac a V ac ~ F ac

T 260 V acAj (F ac <-> F j ) ~ V acF ac v A acF ac

T 261 A x (F x V j [ G j A (H j V H ac)])<->
V ac( G ac a H ac) V ~ V acF ac v (V acG ac a A ac[ F ac -> H ac])

T262 (A acF ac -> A acG ac) •f-» V acAj ( F ac G y)

T 26 3 ( A acF ac <-> V acG ac) <-> VAcV_yA2Att;([FAc G_y)a (G 2 -> Fn>])

T264 (V acF ac ^ [V acG ac ^ A acH ac] ) A acAj A 2 (F ac a Gy H 2)

T 26S Ace(Vy[H(Ay) A F y l ^ Vy[H(c«y) a G y]) <->

AAcAyV2(H(Ay) A F y H(ccs) A G 2)

T 26 6 A ac( F ac a VyG(ciy) V y [H (A y ) a A2j(Acy2)]) <->

AAcAyV«;A2[FAC a G(ciy') -»■ H(xw) a J( acw2)]

T 267 AAcVy(FAc -» Gy) ^ VyAAc(FAc ^ Gy)

T 26 8 VyAcic(FAc -> G y) -> AAcVy(FAc Gy)

T 26 9 VyAx(Fx V G y) *-* AxVy(Fx v G y)

T 270 VyAAc(FAc A G y) *-> AxVy(Fx a G y)

T 271 AAcAyV2(FAc A G y -> Hir) <-> AyVzAx(Fx a G y H2)

184 . *Air and *Some\ continu ed Chapter IV

T272 A:v(F^ A P) ^ AxFx a P

T273 Wx(Fx A P) <-> VxFx A P

T274 A^(F;x: V P) AxFx v P

T275 yx(Fx V P) <-> WxFx V P

 *Chapter V
Automatic procedures

1. Introduction. In earlier chapters we have given informal suggestions

concerning the construction of derivations. For instance, the reader was
advised on page 69 to derive each conjunct of a conjunction before
attempting to derive the conjunction, and to derive a disjunction by first
deriving a corresponding conditional. It was emphasized on several occa­
sions that these suggestions are not infallible—that they will not lead to
a derivation in every case in which one is possible.
The main purpose of this chapter is to present some infallible procedures
of derivation. The new procedures, like the old suggestions, will consist
of directions that can be followed automatically, without the exercise
of ingenuity. They will differ from the earlier suggestions in two respects.
The new procedures will lead to a derivation whenever one is possible;
they are, however, less intuitive than the suggestions previously offered,
and in some cases lead to much lengthier derivations.
Before presenting these automatic procedures, we must add three
abbreviatory clauses to the directions for constructing a derivation.
The new clauses, which will occupy the two following sections, are in­
tended for use only in the present chapter. (In fact, none of the present
chapter will be presupposed later except in optional sections indicated
by asterisks.)
2. Tautologies reconsidered. The tautologies of chapter II are
always symbolic sentences, but the notion of a tautology can easily be
extended so as to apply to arbitrary symbolic formulas. Accordingly,
we define first a molecular formula as one that has the form
^ ^ ,
i'l' -^x) >

{'l>'fx) »
or
{'!> ^ x ) .
for some symbolic formulas ^ and x- An assignment now correlates with
i86 . Automatic Procedures Chapter V

each nonmolecular symbolic formula one of the truth values T or F, The

truth value of an arbitrary symbolic formula, on the basis of a given
assignment, can be computed just as in chapter II (pp. 73 - 74). A
tautology is a symbolic formula whose truth value is T with respect to
every possible assignment. A symbolic argument is tautologically valid
if the truth value of its conclusion is T under every assignment for which
all its premises have the value T ; under the same conditions, the premises
of the argument are said to imply tautologically the conclusion. Just as
before, we may use truth tables to check whether a symbolic formula is a
tautology or whether a symbolic argument with finitely many premises
is tautologically valid.
A subformula of a formula ^ is any formula that occurs within <f>.
In the present section we shall add the following abbreviatory clause
to the directions for constructing a derivation:

( jj) Any tautology may occur as a line.

The theoretical superfluity of clause ( ii) depends on the fact that

every tautology is a theorem; this was asserted in chapter II and will be
substantiated here. We shall, in fact, give directions on the basis of which
we can construct a proof of any given tautology.
Suppose that we are confronted with a tautology, for example,
(i) ( P - ^ Q ) - > ( P a Q<->P) .
We must first construct a truth table whose last column is headed by
the tautology in question. The last column, then, will not contain
In connection with (i) we have
p Q P ->Q P a Q P a Q<-^P ( P - > Q ) - > ( P a Q<-».P)

T T T T T T
T F F F F T
F T T F T T
F F T F T T
On the basis of the truth table, now, we construct a proof of the tautology;
intuitively, the proof can be regarded as a separation of cases, which
correspond to the rows of the truth table. For simplicity, we give the
construction only in connection with (i); the reader will observe, however,
that the procedure is quite general.
If X is formula (i), then the proof of x will have the following
over-all structure:
1. Shew^x
2. I Show P I
Section 2 Tautologies Reconsidered . 187

3-
4- Shew-Q -> X

10. Shet» ~Q ^ X

16. 4, 10, SC
17- Shew ~ P -> X
18. T p

□
19. S k -> X
e w Q

□
25 - Show -> X

SI- X 19, 25, sc

32 . 2, 17, sc
The argument by which line 4 will be established (in which T ’ and ‘Q’
serve as assumptions) will correspond to the first row of the truth table.
The argument for line 10 (under the assumptions T* and will
correspond to the second row. Similarly, the arguments for lines 19 and
25 will correspond to the third and fourth rows respectively.
In greater detail, the general directions for proving (i) are as follows.
We begin the proof by indicating that the tautology is to be shown:
I . Show (P ^ Q) (P A Q <-> P)
We indicate next that the tautology holds when its first nonmolecular
part is true:
1 . Show (P Q) ^(P a Q<->P)
2. Show P ^ [(P - Q) - > ( P a Q< p)]
We assume the antecedent of line 2:
1. Show (P ^Q) ->(P a Q<-^P)
2. Show P - [ ( P - ^ Q ) - > ( P a Q ^ P ) ]
P
We indicate now that the tautology holds when its second nonmolecular
part is true:
1 . Show (P -> Q) ■ ( P a Q<-^P)
2. Show P ^ [(P ■ Q ) ^ ( P a Q < - > P ) ]
i88 • Automatic Procedures Chapter V

3- P
4. Show Q [(P -> Q) -> (P A Q P)]
Again we assume the antecedent:
1 . Show (P -> Q) ->■ (P A Q <-> P)
2. Show P ^ [(P ^ Q) ^ (P A Q<-> P)]
3- P
4. Show Q ^ [(P -> Q) ^ (P A Q<-> P)]
5- Q
We have now, in lines 3 and 5, the assumptions corresponding to the first
row of the truth table. On the basis of these assumptions, we proceed to
‘compute’ the values of the molecular subformulas of (i). That is, we
treat the molecular subformulas of (i) in the order in which they appear
at the head of our truth table, and for each subformula (f>we derive either (j>
or its negation, according as ‘ T ’ or ‘F ' appears below <f> in the first row
of the truth table. (This can always be done because we have at our
disposal the theorems

T 18 ~ P ^ (P -> Q ) ,
T2 Q -> (P ^ Q ) ,
T42 P a ~Q ^~(P->Q )

T43 ~P (P A Q) ,
T44 ~ Q — ~ (P A Q) ,
T67 ~P a ~Q -^~(P v Q)

T84 P a Q ^(P^Q ) ,
T85 ~PA~Q (P Q)
T88 P a ~Q ~(P<->Q )
T89 ~ P AQ ~ (P <—> Q)
which, together with the inference rules DN, Adj, and Add, correspond
completely to the rules for assigning truth values.) In the truth table
for (i), ‘F ’ does not appear in the first row; thus we obtain
1. Show (P ^ Q) (P A Q^-i. P)
2. Show P -> [(P Q) -> (P A Q* P)]
3- P
4. Show Q ^ [(P Q) ^ (P A Q < .P )]
5. Q
6. P —>• Q 5 > T 2, MP
7- P A Q 3> 5. Adj
Section 2 Tautologies Reconsidered . 189

8. P A Q P 3, 7, Adj, T84, MP
9. (P ^ Q ) - ^ ( P a Q ^ P ) 8, T2, MP
The conditional proof of line 4 is complete. We box and cancel, and pass
to the case in which ‘Q’ is false.
1 . Show (P -> Q) (P A Q ^ P)
2. Show P -> [(P ^ Q) ^ (P A Q ^ P)]
3. P
4. Show Q -> [(P Q) (P A Q <-> P)]

10. Show 'Q - ^ [(P (P a Q ^ P )]

11.
Again we derive ‘computed* values, this time in accordance with the
second row of the truth table:
1. Show (P >■ Q) (P A Q <-> P)
2. Show P - [(P ^ Q) -> (P A Q ^ P)]
3- P
4. Sfme-Q ■ -[(P ^ Q ) ^ ( P a Q ^ P ) ]

10. Show Q )^ (P a Q ^ P )]
11. ~Q
12. ~ (P ->Q) 3, II, Adj, T42, MP
13. ~ ( P a Q) 1 1 , T44, M P
14. (P A Q i ^ P) 3. 13. Adj, T89, MP
15. (P ^ Q) -> (P A Q <-). P) 12, T 18 , MP
The conditional proof of line lo is complete. Again, we box and cancel:
1. Show (P -> Q) ^ (P A Q<-^ P)
2. Show P -> [(P ^ Q) -> (P A Q . P)]
3- P
4. Show Q [(P —
> Q) (P A Q <—> P)]
5- Q
6. P -»Q
190 . Automatic Procedures Chapter V

7- P a Q
8. PAQ P
9- (P -> Q) -> (P A Q <-> P)
10. 1S k ew '-'Q [ Q) -> (P A Q <-> P)]
II. ~Q
12. ~ (P ^ Q )
13- ~ ( P a Q)
14. ~ (P A Q <-> P)
15- (P ^ Q) ^ (1P a Q^ P)
But the tautology in question now follows from lines 4 and 10 by separation
of cases; thus we can complete the conditional proof of line 2:

1 Show (P Q) -> (P A Q <-». P)

2 Skew -? ^ [(P ^ Q) ^ (P A Q <-> P)]
3
4
5
6
7
8
9
10
11
12
13
H
15
16 4, 10, SC

We continue in the same way, constructing ‘cases’ for the remaining rows
of the truth table and ‘computing’ corresponding values, until finally all
possibilities are exhausted:

1 . Show (P -> Q) (P A Q < P)

2. Shew ? [(P -> Q) -> (P A Q t-» P)]
3-
4- -> [(P ^ Q) ^ (P A Q<-c P)]

5- Q
6. P 5, T 2, M P
Section 2 Tautologies Reconsidered . 191

7- P aQ 3. 5. Adj
8. P a Q<-^P 3. 7. Adj,
T84, MP
9- (P ^ Q) (P A Q P) 8, T 2, MP
10 . Q ^ [(P-> Q)-> (P A Q ^ P)]
11.
12. 3. II. Adj,
T42, M P
13- 1 1 , T44,M P
H- 3. 13. Adj,
T89, M P
15- 1 2, T i 8,MP
16. Q) ^ (P A Q 4, 10, SC
17- Skew- ~ P ^ [ ( P ^ Q ) - > ( P a Q<-^P)]
18.
19.
20.
21. 20, T 2, MP
22. i 8,T 43,MP
23 - 18, 22, Adj,
T85, M P
24. 23, T 2, MP
25 -
26.
27. i8,T i8 ,M P
28. 26,T44,MP
29. 18, 28, Adj,
T85, MP
30- 29, T 2, MP
31- 19, 25, SC
32. (P ->Q) ^ ( P a Q<->P) 2, 17, s c

Thus we secure a complete proof of (i)—overly long, to be sure, but

obtained by a simple general method.

EXERCISES

Prove the following tautologies by the procedure of this section.

1. ( P ^ [ P ^ Q ] ) ^ Q
2. (Q <-> P A ~P)<-> ~Q
192 . Automatic Procedures Chapter V

([p Q] - > P ) - P
([p Q] -^Q ). >([Q P]->P)

3* Tautological implication; generalized indirect derivation.

The preceding discussion suggests two additional methods of abbreviating
derivations, which, like clause ( ii) , will be used only in the present
chapter. The first of these consists in employing the following clause
in the construction of derivations:

{12) A symbolic formula may occur as a line if it is tautologically implied

by antecedent lines. {Annotation: ^TI\ together with the numbers of the
antecedent lines involved.)

The second of the new abbreviations concerns indirect derivation.

In order to complete such a derivation, we must, according to earlier
instructions, derive an explicit contradiction^ that is, a pair of formulas
of which one is the negation of the other. Our next abbreviation liberalizes
these instructions: to complete an indirect derivation, we shall henceforth
need only to derive a truth-functional contradictiony that is, a combination
of formulas that cannot all simultaneously receive the truth value T.
To be explicit, we say that the symbolic formulas <¡>1 • • . y <f>n are truth-
functionally incompatible just in case there is no assignment of truth
values to nonmolecular symbolic formulas under which all of ...,
(f>n receive the value T. Then the following clause introduces a new form
of derivation, which we shall call generalized indirect derivation.

{ if) When the following arrangement of lines has appeared:

Show (f>
Xi

Xfc y

where none o f x i through Xk contains uncancelled 'S h o w \ and among x i

through Xk there occur unboxed truth-functionally incompatible lineSy then
one m ay simultaneously cancel the displayed occurrence o f *Sh ow \ and
box a ll subsequent lines.

Thus, for example, in the derivation

I. Shew-Wx{¥x K^yGy)
2. Vjc(F:v A yyGy)
3- ' h x {¥ x H x) Premise
Section 4 A Proof Procedure for Prenex Formulas . 193

4* {Fz -> H^) 3, QN, E l

5- AxQ/yGy Fx) Premise
6. WyGy Fz 5 ,U I
7- ^ (F^ A WyGy) 2, QN, U I

the cancellation of ^Show* and boxing of lines 2 through 7 are legitimized

by clause (13); for, as the reader can verify by means of a truth table,
lines 4, 6, and 7 are truth-functionally incompatible.
Clauses (12) and (13) are reducible to (ii), and hence theoretically
superfluous.

4. A proof procedure for prenex formulas. By a prenex formula

we shall understand a formula in prenex normal form or, equivalently,
a symbolic formula in which no quantifier occurs to the right of a paren­
thesis or negation sign. (This criterion is equivalent to the one on page 154
for formulas in official notation.) We are now in a position to describe an
automatic procedure, applicable to any prenex formula, which will lead
to a proof of that formula, if indeed the formula is a theorem. We shall
content ourselves with presenting the procedure, without attempting to
justify the claim made for it. (See, however, the historical remarks accom­
panying this chapter.) The reader will notice that the proofs generated
by the procedure have a particularly simple structure. Only one form of
derivation is employed—generalized indirect derivation—and no sub­
sidiary derivations appear; further, only three inference rules are used—
IE (used in connection with the laws of quantifier negation), U I, and
E l.
Suppose that x is ^ prenex formula. We begin an indirect derivation,
as follows:
Show X
^ X (Assumption)
We now convert

into prenex normal form by successive applications of IE, used in con­

nection with the theorems
' AxFxir-^ Wx ' F a: (T203)
' WxFx <-> Ax ^F a: (T204)
Thus we obtain
Show X
X
194 • Automatic Procedures Chapter V

where \jj is obtained from

by ‘moving in* the negation sign.

The remainder of the derivation, starting with 0 , constitutes an elimina­
tion columny constructed roughly as follows. We drop initial quantifiers
one at a time by means of U I and E L In the process we make sure that
if the column contains a universal generalization

Aoi<f>
and a free occurrence, in one of its lines, of a term then it will also
contain the formula that comes from (f> by proper substitution of ^ for a.
To be more explicit, we begin by examining 0 . If ip has the form

Aa<^
and ii, . . . , in are all the terms that are free in ipy then we add to the
column all the formulas that come from <f> by proper substitution of one
of ii, . . . , in for a; if there are no terms free in we add a formula that
comes from (f> by proper substitution for a of some variable not bound in
any line of the derivation. If ip has the form

Va<^ ,
we add to the column a formula that comes from <p by proper substitution
for a of some variable new to the derivation. These additions are legit­
imate; they follow from ip by either U I or E L If ip is quantifier-free
(that is, contains no quantifiers), we make no additions; the elimination
column then consists of ip alone.
We now examine the second line of the elimination column (if a second
line has been obtained) and add lines to the column as follows.
(1) If the second line has the form
Aoi<p
and ii, . . . , in are all the terms that are free in the first two lines of the
elimination column, then we extend the column by subjoining all for­
mulas that come from <p by proper substitution of one of ii, . . . , in for a.
(The addition is justified by UI.)
(2) I f the second line has the form
Va<^ ,
Section 4 A Proof Procedure for Prenex Formulas . 195

then we subjoin to the column a formula that comes from (f> by proper
substitution for a of some variable new to the derivation. (The addition
is justified by EL)
(3) Whatever the form of the second line, we determine whether any
terms are free in it that were not already free in the first line. If so, and
if the first line is a universal generalization
Aoi(f) ,
then we subjoin to the column all formulas that come from by proper
substitution of one of these terms for a. (The addition is justified by U I.)
We pass now to the third line (if indeed a third line has been obtained)
and subject it to the same sort of treatment as the second line; and so on,
for all following lines. In general, when we examine any line of the elim­
ination column other than the first, we add lines at the bottom of the
column in accordance with the following instructions. (It is clear that
each addition is justified by either U I or E l.)
(j) I f the line in question has the form
Ky,(f>
and ^1, . . . , cire all the terms that are free in lines of the elimination
column up to and including this line^ then we add as lines all formulas that
come from </>by proper substitution of one 0/ ^i, . . . , for a.
(2) I f the line in question has the form
Va<^ ,
then we add as a line a formula that comes from cj) by proper substitution for a
of some variable new to the derivation,
(j) Whatever the form of the line in question^ we determine whether any
terms are free in it that were not already free in the preceding lines. I f sOy
let these terms be , , , y in- For each universal generalization
Acccf)
that precedes the line in question in the elimination columuy we add as lines
all formulas that comefrom (j) by proper substitution of one 0/^1, . . . , in for a.
Now if the original formula x is a theorem, this procedure will eventually
lead to a truth-functional contradiction. It may be convenient, at various
stages of the derivation, to test by a truth table the truth-functional
compatibility of the formulas so far obtained; in fact, it will be sufficient
to test the quantifier-free formulas of the elimination column, for among
these the truth-functional contradiction, if there is one, must always
appear. When we obtain a truth-functional contradiction, we may box
and cancel in accordance with generalized indirect derivation. We shall
then have a complete proof of x*
196 . Automatic Procedures Chapter V

Let us prove, by means of the general procedure, the formula ‘AjyVjc(Fv

-> Fy)\ We begin an indirect derivation:
1 . Show AyWx(Fx -> Fjy)
2, AjyVjc(F^ -> Fjy)
Next we ‘move in' the negation sign:
y y ^ y x {F x Fy) 2, T203
WyKx^{¥x -> Fy) 3, IE(T204)
Now we are to construct an elimination column, beginning with line 4.
We notice that line 4 is an existential generalization and that the variable
has not yet occurred. Thus we add a fifth line, as follows:
5. A x ^ (F x -> F a) 4, E l
We now examine the fifth line of the derivation (the second of the elim­
ination column), and observe that it is a universal generalization and that
is the only term free in this line or the preceding. Thus instruction (i)
for adding lines leads us to
6. ^ ( F a ^ F a ) 5, U I
Instruction (2) is not applicable to line 5, because line 5 is not an existential
generalization; nor is instruction (3), because the preceding line is not a
universal generalization. Thus we pass to line 6. But none of the instruc­
tions is applicable here. Hence the procedure for extending the column
has come to an end. But we easily see that line 6, the only quantifier-free
line, forms by itself a truth-functional contradiction; that is, its truth
value is F under any assignment of truth values to nonmolecular formulas.
Thus, by generalized indirect derivation, we box and cancel, obtaining
I. Show Ay^xiJ^x Fjy)
2. ^ Ay^x{Fx -> Fjy)

3- Vy ^ Vx (F jc - > Fy) 2, T203

4- WyAx'^(Fx Fjy) 3, IE(T204)
5- A x ^ {F x - > Fa) 4, El
6. ^ {F a - > Fa) 5. UI
In the example just considered, the general procedure for extending
the elimination column comes to an end. This is not always the case*
For example, let us apply the procedure to the formula
V:x:yyA2r([Fjc Gx] -> [Fy -> G^r]) .
We begin as before:
1. Show VxWyAz{[Fx -> G a:] -> [Fy G^r])
2. V;cyyA2:([F^ -> G a:] -> [Fy Gz\)
Section 4 A Proof Procedure for Prenex Formulas . 197

3. A^^yyA2r([F;c -> G^] -> [Fy -> G^r]) 2, T204

4. AxAjy/^ A2t([F:v -> G jc] -> [Fy -> G^r]) 3,IE(T204)
5. A:vAjvV2:'^([F^ ^ Gx] -> [Fy -> G^r]) 4,IE(T203)
Since no terms are free in the first line of the elimination column (line 5),
we choose a variable so far not bound, say *a\ and add a corresponding
sixth line:
6. AJyVsr^([Fíг Ga] [Fy -> G^r]) 5 ,U I
We turn to line 6, find it to be a universal generalization, and observe
that is the only term that is free so far. Thus, following instruction (i)
of page 195, we add a seventh line:
7. yz'^ {F a Ga] -> [Fa -> G2:]) 6, U I
Instruction (3) is also applicable to line 6, for that line contains freely
for the first time the variable ^a\ and the preceding line is a universal
generalization. Thus instruction (3) would lead us to add
AJyV2r'^([F¿г -> Ga] [Fy -> Gz]) .
We do not, however, make this addition, for it would amount only to
repeating line 6. (In general, we shall not add lines that repeat earlier
lines.) We now consider line 7. (2) is the only applicable instruction.
We obtain
8. - {[Fa -> Ga] -> [Fa Gb]) 7, E l
Instruction (3) is applicable to line 8; it contains freely for the first time
the variable and there are two earlier universal generalizations (lines
5 and 6). Thus we obtain
9. AyWz^ ([F6 -> Gb] -> [Fy Gz]) 5, U I
10. Wz^{[Fa Ga] -> [Fb -> G^]) 6, U I
To line 9, only instruction (i) is applicable. There are two terms free up
to this point, 'a' and *b' ; thus we obtain
1 1 . Wz^{[Fb -> Gb] [Fa -> G^]) 9, UI
12. Wz^ {[Fb -> Gb] [F6 G^]) 9, U I
Only instruction (2) is applicable to line 10. We obtain
13. - {[Fa -> Ga] [Fb -> Gc]) 10, E l
Again, in connection with line i i , only instruction (2) is applicable.
We obtain
14. -([F A -> GA] -> [F^ Gd)] II, E l

But now we observe that lines 8, 13, and 14 form a truth-functional

19S • Automatic Procedures Chapter V

contradiction; the reader may check this fact by constructing a truth

table. Thus we may box and cancel, to obtain the complete (abbreviated)
proof
I. ■ §hm»yxyyhz{\^x G a;] ^ [Fy ^
2 ~ VA:YyAs([Fx ^ Gx] [F j Gs:])
3 Ax~YyA2r([Fx G a;] [Fy Gsr]) 2, T204
4 A x A y A 2 t([F a: G a:] ^ [Fy -> Gs-]) 3, IE(T204)
5 .AA:AyVa:~([FA: -> Gx] -> [Fy -» Gar]) 4, IE(T203)
6 AyVar~([Fa -> Ga] -> [Fy -> Gar]) 5, U I
7 V » ~ ([F a ^ Ga] ^ [Fa -> Ga]) 6, U I
8 ~ ([Fa ^ Ga] [Fa Gè]) 7, E l
9 AyVa~([F6 -> G6] -> [Fy -> Ga]) 5. U I
10 V a~ ([F a ^ Ga] -> [Fè ^ Ga]) 6, U I
11 V a~([F6 -> G6] ^ [Fa Ga]) 9, U I
12 V a~([F6 ^ Gb] -> [Fb -> Ga]) 9, U I
13 ~ ([F a ^ G a ] [F6 -> Gc]) 10, E l
14 ~ ([F6 -> Gè] ^ [Fa ^ Gt/]) 11, E l

Instead of boxing and cancelling, w'e might well have added lines to
the column. Consideration of lines 12, 13, and 14 would have led to
new lines, and these to further lines. Indeed, the general procedure
would never, in this case, terminate naturally. It is of course pointless
to continue adding lines after a contradiction has been obtained; there
is, however, no general method for computing when, if evpr, this stage
will be reached.

EXERC ISES

Apply the general procedure just described to obtain proofs of

the following theorems. Exercise 5 is solved for illustration.

5. V x V y [F A (x B )F y ]
I. VxVy[FA(xB) ■ Fy]
~VxVy[FA(xB)->Fy]
A.v~ Vy[FA(xB) Fy] 2, T204
AxAy ~ [FA(xB) -> Fy] 3, IE(T204)
A y~[FA (BB )-^Fy] 4, UI
~ [F A (B B )^ F B ] 5, UI
~ [FA(BB) ^ FA(BB)] 5, UI

6. VxVy[FA(xB) FA(yy)]
7. VxAy(Fx v [Fy Gy])
8. AyVxAaVw[F(xy) F(arw)]
9. AxAyAzVzv{Fx \ ^Fz<-> [Fzü<-> Fy])
Section 5 A Derivation Procedure for Prenex Arguments . X99

5. A derivation procedure for prenex arguments. The procedure

just described in connection with individual prenex formulas can be
extended quite naturally so as to apply to arguments. In particular,
we must consider symbolic arguments with finitely many premises, all
of whose formulas are prenex, and whose premises are sentences.
Suppose that such an argument has ^1, . . . , as its premises and ^
as its conclusion. Then the following general procedure will lead, if
the argument is valid, to a complete derivation of tf/ from <^i, . , . , <f>n-
We begin an indirect proof of tp:
Show i/f
tp
As before, we convert

into prenex normal form, obtaining

Show ip
ip

where x is inferred from

/V/ ip
by ‘moving in’ the negation sign. Now we add the premises of the argu­
ment;
Show ip
^ ip

<Pn

The elimination column is again regarded as beginning with x- Lines

are added in accordance with the instructions given on pages 19 4 - 9 5 .
If the argument is valid, a truth-functional contradiction will be obtained,
and the generalized indirect proof may be completed.
200 . Automatic Procedures Chapter V

To illustrate the present procedure, let us construct a derivation

corresponding to the argument
A x A y [F » A G>< - > . '^ x h y { ¥ x a [Jy ^ ~

/. A;c(G:r -> ~J:v)

Derivation:
Shm»^hx{Gx -> ~ J^ )
2 h x ( G x -> ^ j^)
3 Wx'-^iGx -> ]x)
4 hxhy\Fx A Gy -> H(rv)] Premise
5 yxhyi^x A [Jy ^ ^ H(:ry)]) Premise
6 '^{G a -> 3, EI
7 A y {F b A []y ^ ^ U {b y ) ]) S, E l
8 Ay[F¿г A Gy -> H(^y)] 4, U I
9 F b A []a -> ^U{ba)] 7, U I
10 F b A []b -> -H (W )] 7, U I
11 Ay[F6 A Gy -> H(Ay)] 4> U I
12 F¿г A G a -> H {a a ) 8, U I
13 Fa AG b-> U {a b ) 8, U I
H Fb AG a -> U {b a ) I I , UI
15 Fb AG b ^ H {bb) I I , UI

The reader can easily verify that lines 6, 9, and 14 are truth-functionally
incompatible.
Here and in subsequent sections we consider only arguments with
finitely many premises. The restriction is not essential. A slight modi­
fication of the procedures of this chapter would provide derivation pro­
cedures for arguments with infinitely many premises, provided only
that the set of premises be recursively enumerable. (Loosely speaking,
a set of expressions is recursively enumerable just in case there is an enu­
meration of the members of the set according to which one can find
automatically, for each number w, the nth. expression in the enumeration.
For an exact definition, see, for instance, Kleene [i].)

EXERC ISES
Establish the validity of the following arguments by the derivation
procedure just described.
10. AyV;v([Fy -> Gy] V [F:v a Fix]) , Ax{Fx .
V jcF jc V;v(F;ii; A G^)

11. A:vAyA2r[F(^y) A F(y^)-> F(;v;2r)] . A.rAy[F(Ay) F(y^)]

AxAy\F{xy) F(:!c;c)]
1 2 . A^Ay(F:v A Fy -> FA(^y)) . VA:Ay(F;c a [Fy GA(^y)])
/ . V a:(F jc a G : c)
Section 6 Conversion to Prenex Form . 201

6. Conversion to prenex form. In the foregoing sections we con­

sidered only those arguments whose formulas are prenex. Before extending
our considerations further, we must attend to the task of converting a
formula into prenex form. It was mentioned in chapter IV that every
symbolic formula is equivalent to a formula in prenex normal form.
This assertion can be strengthened. Every symbolic formula (f) is equiva­
lent to a prenex formula with the same free variables as <f>; such a formula
is called a prenex form of (f>. We must now give directions for finding
a prenex form of an arbitrary formula <f> and for proving its equivalence
with (f>.
Suppose we are given a symbolic formula (f> that is not yet in prenex
normal form. First we eliminate that is, we replace, one at a time,
each subformula

by an equivalent formula, on the basis of T 8 1 :

(P<->Q)<->(P->Q ) a (Q ^P) .

This transformation requires several applications of IE, corresponding

to the occurrences of Let the formulas produced by these steps,
which lead from to a formula without *<->’, be, in order, ^ 1,. . . , <f>n*
The formula <f>ny like the original formula 0 , is not yet in prenex normal
form. Thus <f>n contains at least one noninitial occurrence of a quantifier
phrase, that is, an occurrence standing to the right of a parenthesis or
negation sign. (We have in mind our official notation, so that in
WxFx -> yxG x
both occurrences of 'Wx' are regarded as noninitial.)
We fix our attention on the first such occurrence, and convert it to an
initial occurrence. The occurrence will begin a subformula of <f>n of the
form

(I) Aonfj
or
(2) Va0 ,
where a is a variable and p a formula. We first, on the basis of AV, change
a to a variable a' that is new to <f>n\ that is, we replace (i) or (2) by
A aV '
or
V aV '
respectively, where comes from iff by proper substitution of a' for a.
Thus we obtain from <f>n a formula
202 . Autom atic Procedures Chapter V

Now we move the occurrence of

Aa'
or
Va'
to the left of all parentheses and negation signs, proceeding step by step
in accordance with the laws of quantifier negation and confinement;
T203 ~ AjcF jc V;c~ F jc
T204 VarFjc F a;
T 2I9 Aac(P ^ F ac) (P AacF ac)
T220 Vac(P ^ F ac) (P -> VajP ac)
T 22I Aac(F a; -> P) (VacF ac ->■ P)
T222 Wx(Fx P) {AxFx -> P)
T 2 I5 Aac(P a FAf) <-> P A AacF ac
T272 AAc(FAi A P) <-> AacF ac a P
T 2I6 Va?(P a F ac) P a VajF a;
T273 VA*(FAi A P) ^ VacF a(; a P
T 2 I7 Aac(P V F ait) P v AacF ac
T274 Aat(F a; V P) •(-> AacF a; v P
T 2I8 Va;(P V F ac) P V VAcFAf
T275 Va.”(F a; V P) VxFx v P
Let the formulas corresponding to the various steps in this transfor­
mation be ^n+2, ‘ In the last of these formulas the occurrence of
Aa'
or
Va'
has become an initial occurrence. We now consider the first noninitial
occurrence of a quantifier phrase in <f>n+ic and transform this into an
initial occurrence just as before. We repeat the process until all noninitial
occurrences of quantifier phrases have been removed. Thus, when we
combine all stages of the process, we obtain a sequence of formulas
<f>y <f)iy , . . y <f>jy iffy in which each formula is equivalent to the preceding

by IE (applied in connection with the theorems just listed), and such

that the final formula if; has no noninitial occurrences of quantifier phrases.
Section 6 Conversion to Prenex Form .

The formula ift is the prenex form that was sought: it is equivalent to <f>f
it is in prenex normal form, and it contains the same free variables as <f>-
Further, a proof of the biconditional
<j)i-► ijf
can easily be constructed, in terms of the intermediate formulas <^i, . . • »

I. Shew- i f f
2.
3- (j>\ i—► <j>9

j + I.
j + 2, (¡>j <—>\fj
j + 3- >tjj

Each of the lines (2) - (j + 2) is justified by AV (assisted by the theorem

T 4 -> P ’), or by IE, used in connection with an appropriate theorem;
and the line (j + 3) is obtained from (2) - {j 4- 2) by T L
For example, let us convert the formula

(3) AxFx -> WyFy V G jc

into prenex form. We fix our attention on the first noninitial occurrence
of a quantifier phrase, change the variable occurring in it to one new
to (3), and, on the basis of T222, move it to the beginning of (3), obtaining
first
AzFz WyFy V G.v
and then
(4) Wz{Fz WyFy V Gx) .
We treat similarly the first noninitial quantifier phrase in (4), obtaining
successively
V2t(F2: -> WwFw V Gx) ,
Wz(Fz -> Ww{Fw V Gx)) ,
(5) WzWzv{Fz Fw V Gx) .
The formula (5) is a prenex form of (3), and the following is a proof of its
equivalence to (3):
1. Shew-{AxFx -> WyFy v G x)<-> V2Vw(F2t - ^ F ww G x)
2. (AxFx -> WyFy V Gx)
{AzFz -> WyFy V Gx) T 9 1, AV
204 . Automatic Procedures Chapter V

3 {hzFz -> VyFy V G*) ^

Sz{^z VyFjv V Gx) EI(T222)
yz{Ÿz V_yFy V G;c)
V^(F3r VzüFw V Gx) T 9 1, AV
5- WzÇFz WwTw V G jc) <r^
Wz{¥z -> Vw(Fw V G jc)) IE(T275)
6. Vsr(F2 Ww(Tw V G.jc)) *->■
WzWw{Fz ->• F«? V Gx) IE(T220)
{Ax¥x -> Wy¥y V Gx) <->
VWa;(F2 -> Fw v G,*) 2 - 6, T I

EXERCISES

Convert each of the following formulas into prenex form and prove
the equivalence of the formula and its prenex form.
13. VMj'F(ry) “ > AyWxF{xy)
14. Vjc[F jc a A>'G(j([rv)]
15. A^(F;v G;v) ^ A^:(yy[H(:rv) a Fy] Yy[H(jcjv) a Gjy])
16. Ax(Fjc<-> Gx) (VjcF:c<-> V^Ga:)

7. A derivation procedure for arbitrary symbolic arguments. We

consider now an arbitrary symbolic argument with finitely many premises.
Let these premises be <^i, . . . , <f>n, and let the conclusion be The
derivation procedure is roughly this. We first derive from the premises
their closures, then convert both the premises and the conclusion into
prenex form, and finally apply the derivation procedure of section 5.
In greater detail, the derivation procedure is the following. Let

(I) Aai . . . Aajt <j>1

be a closure of ^1, the first premise. We begin the derivation with a deri­
vation of (i):
Show A a i . . . KoLjc <f>\

4>i

By the procedure described in section 6, we find a prenex sentence

that is equivalent to (i), and further a proof of the equivalence. We
add this proof to the derivation to obtain
SkewAoii . . . A oljc <f>i

Skew-Aoci . . . Aoijc
Section 7 A D erivation P roced u re for A rbitrary Sy m b olic A rgum ents . 205

We repeat the procedure for each of the other premises, obtaining

Skew-hcLi. . . KcL]c (f>i

4>,
Skew-Noli . . . hoik

Skew-h^i . . . AjS; (f>n

<{>n

Skew-A^i . . . AjS; <l)n<-> if/fi

where jSi, . . . , jSy are the free variables of (f>n* Again by the procedure of
section 6, we find a prenex form x of the conclusion i/r, together with a
proof of the equivalence. This proof we add to the derivation:

S k m e - h o i i . . . hoLk (f>i

Show A a i . . . hoLk (f>i

Shew-h^i . . . AjSy (f>n

At this point we begin an indirect proof of ijjy and from the assumption
^if) ,
together with an equivalence occurring earlier in the derivation, we infer
2o6 . A u tom atic P ro ced u res Chapter V

Next we ‘move in* the negation sign, obtaining a prenex counterpart x' of
'^X •
Thus the derivation becomes:

’SkeW‘hcf.1. . . AaA: ff>i

Show A a i . . . AaA; (f>i <->

Skew -A Pi. . . A]8; (f>n

<f>n

Show AjSi. . . AjS; <l>n<->

Showifjir^X

Shewifj

'^X

Next we infer the formulas t/ri through obtaining

Show A a i . . . Aait

A a i . • . Aa^t

SAow- AjSi. . . Aj8^ <f>n

Section 7 A Derivation Procedure for Arbitrary Symbolic Arguments • 207

Show Aj8i. •• (f)fi ^—y ip ji

Show X

Shewijs
tji

X
01

0n
We now proceed as in section 5, regarding the elimination column as
beginning with the line x - A truth-functional contradiction will ensue
just in case the conclusion ip is derivable from the premises <f>i through <f>n-
The present method, besides serving as a derivation procedure for
symbolic arguments, provides also a proof procedure for arbitrary sym­
bolic formulas. To apply the method to a formula <f>y we consider the
argument without premises whose conclusion is <f),

EXERC ISES

Establish the validity of each of the following arguments by the

derivation procedure described in this section. Exercise 17 is solved for
illustration.

17. ¥x AxGx . Vjc(F;c A ^ AjcHjc) /. />^A.v(G.v H;v)

1 . Show Ax(Fx -> A;cG:r)
2. F jc->A xG jc Premise
3. ShowAx{Fx -> AxGx) AxAy{Fx -> Gy)
4* Ax(Fx -^AxGx) <->Ax{Fx -^AyGy) T91, AV
5- Ax{Fx -^AyGy) <r->AxAy{Fx Gy) IE(T2I9)
6. Ax{Fx -> AjcG^c) <->AxAy{Fx Gy) 4.5. T i
7. Shew-Wx{Fx A <->VWy(F:v a ^ Hy)
8. yx(FxA >Wx { F x a '^ A y F iy ) T91, AV
9- V*(F;ca A jH j) ► V:c(F jc a Vy ^ Hy) IE(T203)
10. ► Vjcyy(F.vA ^Hy)
V ^(F jc A V j ~ Hjv) I E ( T 2 i 6)
11. Wx{FxK AxHx) < ► VWy(Fjc A ^ Hy) 8, 9, 10, T I
го8 . Automatic Procedures Chapter V

12. ¿Лее»-~Лл:(Сл: ^ Нл;)<->Ул:~(Сл: ^ Нл)

13- T203
ч-
15-
i6. 12, 15, T I
17- 16, T204, T I
i8. 1. 3. T I
19- Premise, 7, T I
го. 17, UI
21 . 19. E l
22. 18, UI
23- 21, E l
24. 17. UI
25- 18, UI
26. 22, UI
27- 22, UI
28. 17. UI
29. 18, UI
30- 22, UI
31- 25. UI
32. 25. UI
33- 25. UI

In line I of the derivation, a closure of the first premise is established.

In each of lines 3, 7, and 12, a closure of a formula of the argument is
shown to be equivalent to a prenex form of that formula. Line 14
begins the derivation of the conclusion of the argument, and the
elimination column of this derivation begins with line 17. The reader
can verify that lines 23, 28, and 33 are truth-functionally incompatible.

18. Fx Vy[Gy Л H(xy)] Л Yy[G j a ~ H(xy)] .

Vx(Tx A Ay[Gy-> Н(лу)]) Wx{Tx a ~ F x)
19. Vy(A2[F(2j) -> F(2 A(x))] a F(a^)) .•. AxF(x A{x))

8. A decision procedure for certain prenex formulas. A subsidiary

purpose of this chapter is to present decision procedures for certain kinds
of formulas and arguments. A decision procedure for a class of form ulas
is an automatic method for determining of each formula in the class
whether or not it is a theorem. A decision procedure for a class of arguments
is an automatic method for determining of each argument in the class
whether or not it is valid.
We have observed earlier (p. 170) that for the class of all symbolic
formulas (of the full quantifier calculus) and for the class of all symbolic
arguments (of this calculus) no decision procedures are available. For
some special classes, however, decision procedures may be found. Indeed,
Section 8 A Decision Procedure for Certain Prenex Formulas • 209

in some cases the automatic derivation procedures described earlier in

this chapter lead quite simply to decision procedures.
The cases we wish to consider depend upon the following fact (which
will not be proved here): for formulas and arguments of certain structures,
the elimination column of our automatic derivation procedure will come to
an end, and in such cases we may test for provability or validity by
generating a complete elimination column and performing a truth-
table check of the truth-functional compatibility of the quantifier-free
formulas in it. It should be mentioned that we exclude from our con­
siderations formulas containing operation letters.
The first case is that of prenex formulas without operation letters in
which no universal quantifier follows an existential quantifier. To deter­
mine whether such a formula is a theorem we need only attempt to prove
it by means of the derivation procedure described in section 4. The
elimination column so obtained will come to an end. Thus it is possible to
test the quantifier-free formulas of the elimination column for truth-
functional compatibility by constructing a truth table. We shall obtain
truth-functional incompatibility just in case the original formula is a
theorem.
For instance, the formula

(1) AyWx{Fx -> Fjy)

has the required form. Its elimination column, constructed on page 196,
was seen to terminate, and a truth-table check would have disclosed that
(i) is a theorem.
As another example, consider the formula

(2) AxAyWz[F{xz) V G( j 2:) ^ F(xy)]

The automatic derivation procedure leads us to the following:
Show AxAyVz[F{xz) v G{yz) F(xy)]
A;cAjV^[F(jc^) V G(y<sr) -> F{xy)]
Aj^V2:[F(;c5:) V G(y2:) ->F(xy)]
V;sr[F(jc^) V G{yz) -> F(rv)]
W x V y A z ' ^ [F(a:<2:) v G(jy5r) F(;cy)]
WyAz'^ [F(^^) ^ G{y^)
A^^ V G{bz) -> F(fl6)]
- [F{aa) V G{ba) -> F{ab)]
- [F{ab) V G{bb) -> F{ab)]
The quantifier-free formulas to be considered are
- [F{aa) V G{ba) ^F{ab)] ,
- [F(ai) V G{bb) F(flA)] .
A truth-table check reveals that these formulas are truth-functionally
210 . Automatic Procedures Chapter V

compatible. Thus, according to our decision procedure, the formula (2)

is not a theorem.

EXERCISES
Using the decision procedure just described, determine which of Nos.
2 0 -2 8 are theorems.
20. AjcAjyV2Vw[F(xv) V F(jyjc) F(5tw)]
21. VWj V2([F^ -> Gx] [Fy -> G;^])
22. AjcAjvA2Vw([Fa: ^ Fw] —>■ Fy V F^)
23. AzAyAxAuAv([F{xy)<r-> F(;ca:) a T(xx)]-> F(mz;))
24. AxAyWzWzodFw -> Gw] -> [F j a H(ry) -> Gz a H(:v2r)])
25. AzAxAwAuWvyy{F{xz) [F(jyw) ^ (F(wz;) F(3>jy))])
26. AjcAjvAjs:Aw([F(a:^) -> G{yw)] -> G(>'w))
27. AzAyyzoVuWx{[F(xu) -> G(:vw) v H(a;w)]
[F(yz) -> G{yz)] A [F(w^) A U{wz)])
28. YyVjcV2[F(xv’2r)<-> F(at2a:)]
29. * Give a decision procedure for symbolic formulas without
operation letters and without overlay (see p. 1 53), that is, an automatic
procedure whereby one can determine whether such a formula is a
theorem. In view of the decision procedure just described, it is sufficient
to give an automatic procedure for transforming any formula of the
kind in question into an equivalent prenex formula, also without
operation letters, in which no universal quantifier follows an existential
quantifier. (One such procedure can be obtained with the aid of the
normal forms mentioned on page 72, in exercises 40 - 43 of chapter II.)
Notice that the solution of this problem, combined with that of
exercise 72 of chapter IV (p. 170), will give a decision procedure for
all monadic formulas. (An alternative decision procedure was given
in section 9 of chapter III.)

9. A decision procedure for certain prenex arguments. The second

decision procedure we consider is a natural extension of the first. In
particular, we treat here arguments (again without operation“ letters)
whose premises, finite in number, are prenex sentences in which no exis­
tential quantifier follows a universal, and whose conclusion is a prenex
formula in which no universal quantifier follows an existential. To deter­
mine whether such an argument is valid we apply to the argument the
derivation procedure of section 5. Again the elimination column will
come to an end, and we may test its quantifier-free formulas for truth-
functional compatibility. We shall obtain truth-functional incompatibility
just in case the argument is valid.
For example, the argument
AxAy[Fx A G y -> FL(xy)] VxAy(Fx A []y H ( a tv ) ] )

Aa;(Ga: j^)
Section lo Historical Remarks . a n

has the required form. Its elimination column, constructed on page

200, was seen to terminate and to contain a truth-functional contradiction.
Consider also the argument
(i) YyAjc(Fjc -> Gy) . VyV2r(Gy -> Kxyz{¥x H2:)
Our derivation procedure leads us to the following:
Show hxyz{¥x -> Hz)
^ Axyz(Fx -> H0)
yx'^Wz(Fx -> H^r)
WxAz'^ (F a? -> H^r)
VyAjc(FA? Gy)
VyV;^(Gy -> H^)
As:'^(Fa -> H0)
A a?(F a? -> Gb)
\/z{Gc -> Hz)
^ {Fa -> Ha)
Fa ^ Gb
Fb->Gb
^{Fa-^Hb)
Gc->H d
^{Fa->Hc)
Fc^Gb
^{Fa-^Hd)
Fd->Fb
A truth-table check of the quantifier-free formulas in this derivation
would reveal that these formulas are truth-functionally compatible. Thus,
according to our decision procedure, the argument (i) is invalid.

EXERCISES
Using the decision procedure just described, determine which of the
following arguments are valid.
30. AxAy[Fx A G y - > H(A?y)]) . VxWy[Fx a J y a ^ H(Acy)]
V a?(Ja? a Gx)
3 1 . W yAxAz(Fx A [G y -> H(Ay)] [Gsr - > H(?C5r)]) .
V yV a?[F a? a H (^y)] .*. VA?Vy[FA? A G y A H(A?y)]

32. WxVyAz[Fz - > G a? a H(2rjc) a G y a ^ H ( 2 y ) ] .

VA?Ay(jA? A [G y - > H(Ay)]) V a?(Ja? a '^ F x )

3 3 . VyV5rAA?(F(A£y) <-> ¥{xz) A G a?) /. AMA?Vy(F(A?y) <-> F(A?;2r) a G a?)

10. H istorical rem arks. The task of giving a derivation procedure,

if no special conditions are imposed, is trivial. We need only settle upon
an ordering of our inference rules and forms of derivation and give
212 . Automatic Procedures Chapter V

directions, based on that ordering, which will ensure that all possible
inferences are drawn. If we are presented with a theorem, we may auto­
matically obtain a proof for it by carrying out these directions until they
lead us to one. The derivation procedures given in this chapter, however,
lead to derivations of a particularly simple structure. For example, we
obtain by the procedure of section 4 (our basic procedure for the quantifier
calculus) derivations whose lines are always subformulas of the formula
to be proved or else come from such subformulas by proper substitution
on variables.
The procedure of section 2 (used to show the superfluity of clause (ii))
is essentially Kalmar’s proof of the completeness of the sentential cal­
culus (see Kalmar [i]). The proof procedure of section 4 is closely related
to a procedure given in Herbrand [2]. The claim made on page 193, that
our procedure will always lead to a derivation if one is possible, follows
from a theorem (known as the Herbrand Theorem) proved in Herbrand [2]
or, alternatively, from Gentzen’s Extended HauptsatZy which is proved
in Gentzen [1]. Here we must use the fact that our development of the
quantifier calculus is equivalent to those of Herbrand and Gentzen;
this follows from Montague and Kalish [1].
A decision procedure (differing considerably from ours) for prenex
formulas in which no operation letters occur and no universal quantifier
follows an existential quantifier was first given in Bernays and Schonfinkel
[ i ] ; a procedure quite similar to ours was given in Quine [5]. For a survey
of other cases for which decision procedures have been found, see Church
[3], pp. 245 - 94, and the comprehensive treatment in Ackermann [1].
 Chapter V I
‘ I S ’ (in one sense)

I. T erm s and form ulas. Not all valid arguments can be reached
by the procedures of chapters I - IV. For example, the argument
(i) Mark Twain is identical with Samuel Clemens. Samuel
Clemens wrote Tom Sawyer. Mark Twain wrote Tom
Sawyer ,
unlike those discussed in the preceding chapters, depends for its validity
on the meaning of the phrase ‘is identical with*.
We shall abbreviate ‘is identical with* by the familiar sign ‘ = *. The
first premise of (i) then becomes

(2) Mark Twain = Samuel Clemens

which is regarded as true just in case the two names ‘Mark Twain* and
‘Samuel Clemens* designate the same object. ‘ = * is known as the identity
sigUy and the result, like (2), of flanking it with terms is the identity formed
from those terms.
The language with which we shall deal is that of chapter IV (see p. 133),
with ‘ = * added to its symbols. We obtain no new terms. To the charac­
terization of the class of formulas given in chapter IV a fifth clause must
be added, leading to the following exhaustive characterization:

(j) A ll formulas of English {that iSy sentences of English or expressions

like sentences of English except for the occurrence of variables) are formulas.
(2) The result of writing a k-place predicate letter followed by k terms
is a formula. {In particulary a o-place predicate letter is itself a formula.)
(3) I f <f> and iff are formulasy so are
r^<f> ,
{(f> iff) y
{(f>Mff) ,
{<f>yiff) y
{<t>^lp) .
{4) I f ^ is a formula and a a variabhy then
 214 • (in one sense) Chapter VI

Aa0 ,
Va^
a r e f o r m u l a s ,

(5) I f ^ a n d 7] a r e t e r m s , t h e n

i s a f o r m u l a .

Thus, for example, each of the following is a formula of our present

language:

(3) Scott = the author of W a v e r l e y

(4) 7 + 5 = 12
(5) X = y
(6) AO = Bijc .

The and s y m of our present language are those

b o l i c t e r m s f o r m u l a s

which can be constructed exclusively from variables, predicate letters,

operation letters, parentheses, sentential connectives, quantifiers, and
the identity sign. More precisely, the are those of chapter IV s y m b o l i c t e r m s

(see clauses (a) and (b) of page 134), and the class of s y m b o l i c f o r m u l a s

is exhaustively characterized as follows:

{ A ) T h e r e s u l t o f w r i t i n g a k - p l a c e p r e d i c a t e l e t t e r f o l l o w e d b y k s y m b o l i c

t e r m s i s a s y m b o l i c f o r m u l a , { I n p a r t i c u l a r , a o - p l a c e p r e d i c a t e l e t t e r i s i t s e l f

a s y m b o l i c f o r m u l a , )

( B ) I f (f) a n d i/ s a r e s y m b o l i c f o r m u l a s , s o a r e

<l> y
>
((^ A ,
V t/l) ,

(C) I f <f> i s a s y m b o l i c f o r m u l a a n d c l a v a r i a b l e , t h e n

S cL (f> ,

W 0L<f>

a r e s y m b o l i c f o r m u l a s ,

{ D ) I f ^ a n d 7] a r e s y m b o l i c t e r m s , t h e n

i = V
i s a s y m b o l i c f o r m u l a .

Of the preceding formulas, then, (3) and (4) are not symbolic, and (5)
and (6) are symbolic.
Section 2 Translation and Symbolization . 215

We shall continue to use the informal notational conventions introduced

in section 3 of chapter IV. Again we emphasize that official notation must
be mentally restored before decisions concerning freedom and bondage
and applications of inference rules can be made.

2. Translation and symbolization. At this point we must relinquish

consideration of a large class of English formulas—those we shall call
nonextensionah An English sentence <f> is said to be extetisional if, whenever
a name occurring in it is replaced in one or more of its occurrences by
another name designating the same object, the resulting sentence has the
same truth value as (f>\ that is, is true or false according as (f) is true or
false. Most sentences which come naturally to mind are extensional, for
instance.

(I) Socrates is fat.

Replacement of ‘Socrates’ by another name designating Socrates will
leave the truth value of (i) unchanged. For example.
The snub-nosed teacher of Plato is fat
is true if and only if (i) is true. However, the sentence
(2) Necessarily Socrates is Socrates
is not extensional. ‘Socrates’ and ‘the snub-nosed teacher of Plato’ desig­
nate the same object, and (2) is true; but
Necessarily the snub-nosed teacher of Plato is Socrates
is false. As an another example consider
Most students believe that Cicero denounced Catiline,
which is seen to be nonextensional by comparison with

Most students believe that Tully denounced Catiline.

The use of so-called modal terms, such as ‘necessarily’ and ‘possibly’,
and the subordinate conjunction ‘that’ is in general likely to produce
nonextensional sentences; and it is perhaps in this way that most non­
extensional sentences originate. There are, however, other cases not so
easy to detect, for example, the truth
Hamlet intended to kill the man behind the arras,
which, upon exchanging two names of the same person, becomes the
falsehood
Hamlet intended to kill Polonius.
2i6 . *Is’ (in one sense) Chapter VI

By an we understand either an extensional

e x t e n s i o n a l f o r m u l a o f E n g l i s h

sentence of English or a formula containing variables which, upon re­

placement of all its variables by English names, becomes an extensional
sentence of English. Thus ‘jc is fat’ is extensional, but ‘necessarily is x

Socrates’, ‘most students believe that denounced and intended x y \ ^ x

to kill jy’ are nonextensional.

An English ^ is said to be
n a m e if, whenever ^ is a name e x t e n s i o n a l

obtained from ^ by replacing one or more occurrences of an English

name within ^ by another English name designating the same object, then
i and designate the same object. By an we e x t e n s i o n a l t e r m o f E n g l i s h

understand either an extensional English name or a term containing

variables which, upon replacement of all its variables by English names,
becomes an extensional English name. For example.
the father of a

and
7 + a
are extensional terms of English, but
the person whom a believed to be b

is not.
We shall now admit abbreviations only of those formulas and terms of
English whose variables are the first variables, for some number which k

contain no apparent variables, and which are extensional. (The reason

for the additional requirement of extensionality is discussed in section 5
below.)
Thus an will now be either (i) an ordered pair whose
a b b r e v i a t i o n

first member is a A-place predicate, for some number A ^ o, and whose

second member is an extensional formula of English containing exactly
the first variables and having no apparent variables, or (2) an ordered
k

pair whose first member is a /f-place operation letter, for some number
/ 5 ^ 0 , and whose second member is an extensional term of English
containing exactly the first variables and having no apparent variables. k

L i t e r a l t r a n s l a t i o n i n t o E n g l i s h o n t h e b a s i s o f a g i v e n s c h e m e o f a b b r e v i a t i o n

is handled as in chapter IV, section 4, with one modification. In step

(v) (p. 140), we replace ‘ = ’ by ‘is identical with’, in addition to eliminat­
ing sentential connectives and quantifier phrases in favor of the corre­
sponding phrases of connection and quantity.
As before, we say that an English formula ^ is a (or f r e e t r a n s l a t i o n

simply a of a symbolic formula <f> (or that ^ is a

t r a n s l a t i o n ) s y m b o l i z a t i o n

of tfi) on the basis of a given scheme of abbreviation if ^ is a stylistic

variant of the literal English translation of <f> based on that scheme.
For example, consider the scheme of abbreviation
Section 2 Translation and Symbolization . 217

(3) : Mark Twain

BO : Samuel Clemens
and the symbolic sentence
(4) A = B .
The literal English translation of (4) on the basis of the scheme (3) is
(5) Mark Twain is identical with Samuel Clemens,
and a free translation of (4) would be
(6) Mark Twain is Samuel Clemens.
Thus we regard (6) as a stylistic variant of (5). Ts' does not, however,
always have the sense of identity. In some contexts, for example,
(7) Socrates is wise,
‘is’ has the sense of predication, (7) is true just in case Socrates has the
characteristic, possibly common to many individuals, of wisdom; it is
not asserted that the words ‘Socrates’ and ‘wise’ designate the same object
(or even that ‘wise’ designates any object), as the sense of identity would
require.
The logic of predication can best be regarded as part of the theory
developed in chapters III and IV. The present chapter concerns the logic
of identity.
(5) has other stylistic variants than (6); for example,
Mark Twain is the same as Samuel Clemens,
Mark Twain and Samuel Clemens are the same,
Mark Twain and Samuel Clemens are identical,
Mark Twain equals Samuel Clemens.
The negation of (4) has free translations other than those suggested by
the free translations of (4). For example,
Mark Twain differs from Samuel Clemens,
Mark Twain is distinct from Samuel Clemens,
and
Mark Twain is other than Samuel Clemens
are all free translations of
- A = B
on the basis of the scheme (3).
To find a symbolization of a given formula of English on the basis of
a given scheme of abbreviation, the reader will find it useful to proceed
roughly as follows:
2x8 • *Is* (in one sense) Chapter VI

( j) I n t r o d u c e * is i d e n t i c a l w i t h \ p h r a s e s o f q u a n t i t y y a n d p h r a s e s o f c o n -

n e c t i o n y t h e l a t t e r a c c o m p a n i e d b y p a r e n t h e s e s a n d o c c u r r i n g c a n o n i c a l l y y i n

p l a c e o f t h e i r s t y l i s t i c v a r i a n t s .

(2) R e v e r s e t h e s t e p s l e a d i n g f r o m a s y m b o l i c f o r m u l a t o a l i t e r a l E n g l i s h

t r a n s l a t i o n .

For example, consider the scheme of abbreviation

: a is a citizen
: the wife of a

and the sentence

Anyone who is the wife of a citizen is also a citizen.
In step (i) we obtain
For each x (if there is an object y such that (y is a citizen
and X is identical with the wife of jy), then x is z citizen) ,

and carrying through the successive parts of step (2), we obtain the
symbolization
Ajc(Yy[Fy AX = A(y)] F:v) .
As another example, consider the scheme of abbreviation
F2 : a is a member of b

Q2 : has defeated
a b

AO : the team
: the captain of , a

and the sentence

The captain of the team has defeated every other member

of the team.

The symbolic translation in this case is

Ax[F(xA) A - jc = B(A) ^ G(B(A) x)] .

EXERCISES

I . For each formula of group A, find the formula of group B that

is a symbolization of it.

GRO UP A

(1) Something differs from everything except itself.

(2) Everything is identical with something.
(3) Everything is distinct from something.
Section 3 Inference Rules; Theorems • 219

(4) Everything equal to a thing differs from everything not equal to

that thing.
(5) Nothing differs from everything.
(6) Two things equal to the same thing are equal to each other.
(7) Nothing differs from itself.

GROUP B

(a) X= y
(b) hxVy X = y
(c) VxAy{^>^y = x -> '> -'X = y )
(d) AxAy{Vz[x = z A y = z ]~ ^ x = :y )
(e) AxWy y = X
(f) r^Vxr^ X X
(g) AxAz[x = z Ay{ r^y = z ^ ' ^ x —y)]
Symbolize each of the following formulas on the basis of the scheme
of abbreviation that accompanies it.
2. None but Alfred and his teacher are able to solve the problem.
(F2 : a is able to solve b; A° : Alfred; B® : the problem; :
the teacher of a)
3. Alfred attended the conference and arrived at it before everyone
else who attended it except Mary. (F^ : a attended b\ : a arrived
at b before c; A® : Alfred; : the conference; C® : Mary)
4. Alfred is the only member of the class who can read Greek.
(F2 : a is z member oí b; : a can read b; A® ; Alfred; B^ :
the class; C® : Greek)
5. Greensleeves can jump farther than any other frog in Calaveras
County. (F2 : a can jump farther than b\ G^ : a is a frog; № : ¿2 is
in i; A® : Greensleeves; B® : Calaveras County)
6. Anyone whose mother is the wife of a citizen and whose father
has no other children is the son of a citizen. (F^ : a is z citizen; G^ : a
is a child of i ; A^ : the mother oí a; B^ : the wife of a; C^ :
the father of a; : the son of a)
7. If Alfred’s sister is the wife of Alfred’s brother-in-law, then
Alfred and his sister have the same father-in-law. (A® : Alfred; B^ :
the wife oí a; C^ : the sister of a ; : the brother-in-law of a ;
: the father-in-law of a)
8. None but the lonely heart can know my sadness. (F^ : a is a
lonely heart; G^ : a can know i ; A® : my sadness)
9. Bach is more ingenious than any other composer. (F^ : a is more
ingenious than b; G^ : a is z composer; B® : Bach)
10. Pergolesi was more promising than any other composer of his
time. (F2 : a was more promising than 6; G^ : a is a composer; :
a was contemporary with b; A® : Pergolesi)

3. Inference rules; theorems. Notions of bondage and freedom,

as well as the notion of proper substitution of a term for a variable, are
 220 . ‘Is’ (in one sense) Chapter VI

to be understood here as in chapter IV (see pp. 136 - 37, 148). For the logic
of identity, two inference rules must be added to those already at our
disposal. The new rules lead, respectively, from a symbolic formula

to the formula
Aa(a = i ,
and from a symbolic formula
Aa(a — ^
to the formula

where ip comes from cp by proper substitution of the symbolic term ^ for

the variable a, and a does not occur in Diagrammatically the two
rules present the following appearance:
Identity-i (Id -i): ip
Aa(a = ^

Identity-2 (Id-2): Aa(a = ^ -x p )

^ ’
where a, ip satisfy the conditions above.
The rule Id-i permits the inference from
FA
to
Ax{x = A Fx) ;
and Id-2 permits the converse inference. For a somewhat more complex
illustration, consider the formulas
(1) GB(Ax)
(2) Ay{y = B{Ax) -> Gy) ;
from (i) we can infer (2) by Id -i, and from (2) we can infer (i) by Id-2.
However, from (i) we cannot infer
Ax{x = B(Ax) -> Gx) ,
for ‘x ’ occurs in 'B{Ax)\
We may call that branch of logic which essentially involves the identity
sign, as well as quantifiers and sentential connectives, the identity calculus.
Section 3 Inference Rules; Theorems . 221

The directions for constructing, within this calculus, an unabbreviated

derivation from given symbolic premises are those of chapter IV (see
section 13, clauses (i)-(6 )), with the understanding that Id-i and Id-2
are now to be included among the primitive inference rules, and the word
‘formula’ is to assume the broader sense of the present chapter.
The characterizations of a complete derivation, derivabilityy a. proofs
a theorem^ an argumenty a symbolic argumenty a valid symbolic argumenty
an English argumenty a symbolization of an English argument, and a
valid English argument remain as before (see pp. 95, 103, and 117).
In the identity calculus it is convenient to have at our disposal the
methods for abbreviating derivations that appear in chapter IV (see
section 13, clauses (7 )-(10 )). We may adopt these clauses without
modification, along with the characterizations of the pertinent notions
{alphabetic varianty instance \ see sections 7 and 8 of chapter IV). Here
too we must of course understand the word ‘formula’ in the sense of
the present chapter. In the context of the identity calculus clauses (7) -
(10) continue to satisfy the requirements given on page 58.
We assign to the theorems of thé identity calculus numbers beginning
with 301. Theorems 301, 302, and 303 are the respective laws of reflexivityy
symmetry y and transitivity of identity.
T 301 I . Show X = X

2. S h ew h y{y = X -> y = x)
3- y = X -> y = X Ti
4- X = X 2, Id-2
The formula ^x = comes from the formula ‘j = x^ by proper substi­
tution of ^x' for *y^\ further, the variable ‘j ’ does not occur in the term ^x\
Thus the transition from line 2 to line 4 constitutes a correct application
of Id-2.
T302 ^
T 301
2, Id-i
3 , UI
In checking the transition from line 2 to line 3, we must observe that
‘j = y^ comes from ‘j = x^ by proper substitution of ‘jy’ for ‘ a:’, and
that ‘ a:’ does not occur in ‘y ’ .
T303

2, s
3, Id-I
4, U I, 2, S, M P
222 . *Is’ (in one sense) Chapter VI

In connection with line 4, we must observe that = z" comes from

= z* by proper substitution of ‘jy’ for ^x\ and that ^x^ does not occur
in 'y \
T304 can be derived easily from T302, and is often useful in connection
with IE. T305 - T 3 0 7 are formulations, alternative to T303, of the law
of transitivity of identity.
T304 X= y y = X
T305 X= yhZ=y-^X = Z
T306 y = Xhy = Z-^X = Z
T307 y = XKZ=y-^X=Z
The pattern of inference corresponding to T302 and those corre­
sponding to T303, T305 - T307 will appear frequently in the sequel; thus,
as with SC, CD, and QN, we shall employ special annotations to indicate
their use. The derived rule corresponding to T302—the rule of symmetry
(annotation: ‘ Sm’)— and those corresponding to T303, T305 - T307—the
rules of transitivity (annotation for all forms: ‘T ’)—appear as follows:
Symmetry (Sm ): i = rj
»? = i
Transitivity (T ): i = ■»? ^ = V V= C V= C
}= e
7 6 = 7] 7) = 6 6 = 7]
t= 6 c= e 1 = 6 1 = 6
Here C, 7], 6 may be any symbolic terms.
Inferences justified by Sm and T are familiar to every student of
mathematics; for example, T corresponds to the principle that things
equal to the same thing are equal to each other.
T308 I. ¿¡Aewx: = j -> (Fx«->
2. X = y
3- Show Fjf -> Fjy
4- F;c
5- Ay{y = X Fy) 4, Id -i
6. y = X -> F y 5, UI
7- y = X 2, Sm
8. Fy 6, 7. M P
9- Skew- F y -> F;c
10. Fy
II. Ax{x = jy F:v) 10, Id-1
12. Fx 1 1 , U I, 2, M P
13- Fx<-> Fjy 3. 9. CB
Section 3 Inference Rules; Theorems . 223

The pattern of inference corresponding to T308 is essentially Leibniz*

principle of the indiscernibility of identicals (the converse of his better-
known principle of the identity of indiscernibles), which asserts that if
two things are identical, then anything true of one is also true of the other.
We introduce the abbreviated annotation ‘LL* for this pattern of inference,
which has the following form:

Í = r

Here Í and are to be symbolic terms, <f> is to be a symbolic formula,

and <^' is to be like ^ except for having one or more free occurrence of
where <f> has free occurrences of (To show that all cases of L L corre­
spond to alphabetic variants of instances of T308 would require a brief
argument, which we do not intend to present here.)
In the following arguments the conclusion follows from the premises
by L L :
X= y /. F(:v;ií: ) ^ F(:rv) .
X= y bizG{xz) Az G{yz) .
T 3 0 9 --T 3 11 will prove useful in connection with the rule IE. The
proof of T309 is a trivial application of Id-i and Id-2; those of T 310
and T 3 1 1 are elementary, now that L L is available.

T309 F jc<-> Ay(y = X F j)

T 310 I. Shew-Fx<-*Vy(y = j c a Fj )

3
4 T 30 1
5 3, 4, Adj, E G
6
7
8 7. El
9 8, S
10 8, S
11 9, 10, L L , BC, M P
12 2, 6, CB

The passage from lines 9 and 10 to line 11 in the last proof would look,
if expanded, as follows:
9. Far
10. ar = a:
224 • ‘Is’ (in one sense) Chapter VI

loa. Fz<-^Fx 10, L L

lob. Fz Fx loa, BC
I I . Fx 9, lob, MP
This sort of inference will occur very frequently. We shall therefore
often replace the annotation ‘L L , BC, M P’ by ‘L L ’.
T 3 11 Fx ^ X = y<r^Fy h X = y
T 3 12 and T 3 13 are instances, and T 3 14 and T 3 15 are two-variable
analogues, of T309 and T 3 10 respectively.
T3I2 X -= jy Az{z = X

T3I3 X == j <-> Wz{z = X

T3I4 F(xy) <-> AzAw{z =

T3I5 F(jcj) VzWw{z =

T3I6 I. 1Shew^x = y ^ K{x)
I = K{y)
2. X = y

3- A(^) = A( a;) T 30 1
4- k{x) = K{y) 2, 3> L L
The pattern of inference corresponding to T 3 16 is Euclid’s postulate,
‘When equals are substituted for equals, the results are equal’. We add
this pattern to our stock of derived rules; it presents the following appear­
ance:
Euclid’s law (EL): ^ = I'

Here i', Tjy 7}' are to be symbolic terms, and tj' is to be like r] except for
having one or more occurrences of where t] has occurrences of
(As with L L , a small argument would be required to show that all cases
of E L can be obtained as instances of T 316.) For example, from
X = y
we may pass to
A(.x:;c) = A(ry)
by E L .
The addition of the identity sign to our symbolism enables us to express
numerical conditions symbolically. For example, if we let ‘F^’ abbreviate
*a is a god’, then
(3) WxFx
asserts that there is at least one god, and
Section 3 In fe re n c e R u le s; T h e o re m s . 225

(4 ) AxAy(Fx A Fy X = y)
that there is at most one god; hence the conjunction of (3) and (4) expresses
the doctrine of monotheism. The position that there are at least two gods
is expressed by
(5) yxWy(Fx A Fy A r^x = y) ,
and that there are no more than two by
(6) AxAyAz{Fx A F y A F z - > x = y v x = z w y = z) .
The conjunction of (5) and (6) asserts that there are exactly two gods.
In this way we can assert the existence of exactly three gods, exactly
four gods, exactly as many gods as we may wish to countenance. And what
has been illustrated by reference to gods holds equally well for apostles,
apples, Africans—any finite collection the number of whose members we
wish to specify.
The fact that exactly one thing satisfies a given condition can be
expressed in a variety of ways, some of them, like the right-hand constit­
uents of T 3 17 and T 318, shorter than the conjunction of (3) and (4).
T 319 and T320 provide similar alternatives to the conjunction of (5) and
( 6).

T 3 17 WxFx A AxAy{Fx A Fy -> X = y)<r-> Wx{Fx a Ay[Fy -^ x ^ y])

T 318 WxFx A AxAy{Fx A Fy x = y) <r-^ yyAx{Fx<-^ x = y)
T 319 yxWy(Fx A F y A X = y) A AxAyAz(Fx a Fy a F z
X = y y x = z y y = z)<-> Wxyy(Fx AFy A ^ X y ^
Az[Fz -> x = z ^ y = z])
T320 V:x:yy(F^ A F y A r ^ x = y ) ^ A x A y A z {F x A F y a F z -> x - y y
x = zw y = z)<^V xyy{^ X = y a A z [ F z <-^ z == x ^ z = y])

T 321 expresses the fact that each thing is identical with at least one thing
(itself); T322, that each thing is identical with exactly one thing.

T 321 Wx X = y
T322 WyAx{x = z <r-^ X = y)
T323 - T329 will be particularly useful in the next chapter.
T323 WyAx{Az[x = z <r-^z = u]<r^ X = y)
T324 Vy(A.v[Fjc<-^ A G j) yyA x{Fx<r->x = y) a
Ax{Fx G y)

^325 yyA x{Fx X = y ) -> (A y [ F y -> G y ] <-> V y [ F y a G y ])

T 326 ^Vy A y ('^y = y <->y = j )

226 . ‘Is* (in one sense) Chapter VI

T 327 K{x) = B{x) [Wx FA{x) <-> Wx FB(;c)]

T328 AaAbAc[{WzAx[Fx <r-> X = z] Fa) a V2rA:v[F;c<-^jc = z]
-> a = c) A {WzAy[Fy <-> y = z] Fb) a {'■ ^WzAylFy <->
y = z] b = c) a = b]
T329 A^j:AAAi:[(y);Ajc[F:)c <r->x = y] F^i) a (^VyAjc[F^ <-> jc = j ]
-> a = c) A {WyAx[Gx <r^ x = y] -> Gb) a ( ^ VyAx[Gx <->
X = y ]-^ b c) A Ax{Fx <-> G jc) ->¿1 = 6]

EXERC ISES

11. Prove T 3 11, T314, T315.

12. Prove T 3 17 -T 3 2 0 .
13. Prove T323 - T327.
14. Prove T328 and T329.
Show that each of the following arguments is valid.
15. Some student is feminine. Every feminine student is liked by
the teacher. The teacher likes at most one student. Every student
whom the teacher likes is feminine.
16. The wife of a citizen is also a citizen. Anyone whose mother
is a citizen is also a citizen. /. Anyone whose mother is the wife of a
citizen is a citizen.
17. No men except Alfred, Rudolf, and Alonzo attended the
meeting. Of those who attended the meeting only men smoked.
Alfred did not smoke. Of those who attended the meeting at most
two smoked.
18. ‘The boxed sentence is not true’ is true if and only if the boxed
sentence is not true. The boxed sentence is ‘the boxed sentence is not
true’. Therefore the boxed sentence is neither true nor not true.
(The reference is to the boxed sentence on page 3; this is the paradox
of the liar as formulated in Tarski [2].)
19. AyV^Ax[F(iC2)<-> = y^ .’. ^Va;Ax(F(:x:«?) <-> A«[F(icw)
yy{F{yu) A ^yz[F{zu) A F(2ry)])])
(Exercise 19 corresponds to t h e of grounded classes as described
in Montague [i].)

4* * Invalidity. We employ again the notions of an English translation

of a symbolic argument, a relativization of a symbolic sentence^ a closure
aodonty and a relativization of a symbolic argument; see section 10 of
chapter IV.
Here as in the quantifier calculus it is impossible to devise an automatic
test for validity. We can, however, establish invalidity by use of the various
methods introduced in chapter IV. (The method of truth-functional
expansions, introduced in chapter III, will no longer yield correct resultsj
even in connection with symbolic arguments containing no operation
Section 4 Invalidity. 227

letters; the reader will perhaps be interested in showing this, and in

adapting the method to the identity calculus.) For instance, if a closure
of a symbolic argument has a false English translation, then the argument
will not be valid. We can, however, no longer claim the converse.
For example, consider the following symbolic argument:
(i) f\xf\yhz{x = y v y = <sr V 5T = ^) VjcVy ^ X = y .
This argument is invalid, yet none of its closures has a false translation.
Clearly, all translations of its closures are stylistic variants of the argu­
ment
There are at most two things. Therefore there are at least two
things;
and none of these is false, for there are more than two things.
We can assert, however, that a symbolic argument is invalid in the
identity calculus just in case a relativization of one of its closures has a
false English translation. Consider, for example, the relativization of (i)
to ‘G i’ :
f\x{Gx -> Ay[Gy ^ /\z{Gz - > x = y w y = z y z = : x)]) .
WxGx Va:(Gjc a Vy[Gy a ^ = y]) .
A false translation of this argument can be obtained on the basis of the
scheme
G i : a is Socrates .

EXERC ISES

For each of the following arguments, either show that it is valid

or show that it is invalid.
20. VWy ^ X = y VjcF;» a Vx ^ T x
2 1 . A^(F:v -> Gx V Hx) . VyA;v(Gji <-> x = y) .
VyAA:(H:v ^ x = y) VxVy(Fx AFy a x = y)
2 2 . A^(F:v G jc V Ha:) . VyAx(GA: <->x = y) .
V y A ;v ( H A : <-> a : = y ) / . AxAyAz{¥x A¥y a F z a ^ x = y a

X = z -^ y = z)
VyAA:(A: = y x = z A F a:)
A ¥y< -> W z

24. AzAyAx{x = y AFy<-^ X = z A F a:)

25. A(a:) = B(a:) \Vx FA ( a:) ^ V a: FB(a:)]
Show that each of the following inference rules, if incorporated
into our system of logic, would permit the validation of a symbolic
argument that, by methods now available, can be shown invalid. (The
restrictions on variables involved in Id-i and Id-2 are thus seen to be
necessary.) In all cases a is to be variable, i a symbolic term, and <f>y
i/j symbolic formulas. (Exercise 26 is solved for illustration.)
228 . *Is* (in one sense) Chapter VI
26. Aa (a = i - ■<!>)

where ip comes from <f>by proper substitution of f for a (but a is per­

mitted to occur in ^).
Argument:
hx[x = A(;cB) (Fx = B)] FA(BB) A(BB) = B
Derivation:
1. ^SA^A(BB) = B
2, ^;keW‘Ax(FA{xB)->A{xB) = B)
3 Ax[x = A(:vB) -> (Fx -> x = B)] I St premise
3, the variant given
4 FA(;cB) -> A{xB) = B
above of Id-2
5 A(BB) = B 2, UI, 2nd premise,
MP

Translation:
For each x, if x equals + i, then if a: is a positive inte­
ger, then X equals i. i + i is a positive integer.
Therefore i 4- i equals i.
It becomes clear that the translation is a false argument (and hence
that the symbolic argument is invalid) once it is recognized that
there is no positive integer equal to itself plus i.
27. <i>
Aa. {(X. = ^ <f>)
where ip comes from <p by proper substitution of f for a (but a is
permitted to occur in ^).
28. Aoc {o(. = i -> (p)
^ ’
where i/r is like (p except for having free occurrences of ^ at some (but
not all) places where <p has free occurrences of a.
29- ^
Aa (a = S -> ^)
where ip is like (p except for having free occurrences of ^ at some (but
not all) places where <p has free occurrences of a.
30. Aa (a = C -> ^)

where ip is like (p except for having occurrences of $ (free or bound)

wherever <p has free occurrences of a.
Section 5 Paradoxical Inferences . 229

31-
Aa (a = ^ -» <^)
where tp is like <f> except for having occurrences of ^ (free or bound)
wherever <f>has free occurrences of a,

5. * Paradoxical inferences. We now consider the consequences of

admitting nonextensional formulas into schemes of abbreviation. Consider
the valid symbolic argument

(I) A = B FA FB

and the scheme

(2) F^ : Kepler was unaware that a is greater than 6
AO : the number of planets
BO : 9 .
(This scheme violates the requirement of section 2 in that ‘Kepler was
unaware that a is greater than 6’ is nonextensional.) On the basis of
(2), the argument
(3) The number of planets is 9. Kepler was unaware that the
number of planets is greater than 6. Therefore Kepler was
unaware that 9 is greater than 6
would be a translation of (i), and hence, by the definition of validity
for English arguments, would be valid. But the premises of (3) are true,
and the conclusion of (3) false. Thus if (2) were admitted as a scheme
of abbreviation, a claim on page 227 would be falsified: we should be
able* to find a valid symbolic argument with a false English translation.
It is not to be expected that all unsatisfactory examples are precluded
by the requirement of extensionality. Indeed, there are English arguments
that completely fulfill the requirements for validity given in sections 2 and
3 but that on intuitive grounds would tend to be regarded as invalid.
Consider, for example, the argument
(4) Alcibiades is the most notorious Athenian traitor. Everyone
believes that Alcibiades is honest. Therefore everyone
believes that the most notorious Athenian traitor is honest,
which, under the scheme

(5) F^ : everyone believes that a is honest

A^ : Alcibiades
B^ : the most notorious Athenian traitor ,
is a translation of the valid symbolic argument (i). ^
230 . *Is’ (in one sense) Chapter VI

Now (5) is indeed a scheme of abbreviation; ‘everyone believes that a is

honest’ is extensional because every sentence obtained from it by re­
placing ‘¿z’ by an English name is false. We have not found a fallacy, for
the second premise of (4) is false ; yet (4) would seem invalid on intuitive
grounds because it is possible^ given a sufficiently gullible and misinformed
world, for its premises to be true and its conclusion false. As far as we
can determine, no such examples can be found which are actually false
English arguments; thus the claim on page 227 can still be maintained.
The difficulties encountered here would arise in connection with any
known development of the quantifier and identity calculi, and are by no
means peculiar to our formulation. (See, for example, Frege [2], Quine [4],
and Carnap [2].) Indeed, it is not to be expected that a completely precise
analysis of validity for a natural language, faithfully reflecting all intuitions
that have accrued during the historical development of the language,
could be either simple, elegant, or of major philosophical interest. Two
alternative courses are appropriate: one might apply formal procedures
quite generally, with measured indifference to ordinary use; or one might
restrict consideration to a definite part of language that is naturally
accessible to these procedures. Our logical system can be enlisted to the
service of either alternative, and the second alternative is explicitly pursued
in chapters V III and IX .
It is interesting that within the framework of chapters III and IV no
clear-cut counterintuitive cases seem possible as long as apparent variables
are excluded from schemes of abbreviation. (The reader may wish to
assay the accuracy of this observation, perhaps by attempting to develop
paradoxes involving nonextensional contexts within the logic of quanti­
fication without identity.)

6. * H istorical rem arks. The identity calculus is often called the

first-order predicate calculus with identity. It has been discussed exten­
sively in Whitehead and Russell [i], Hilbert and Bernays [i], and Scholz

The claim that if no closure of a symbolic argument has a relativization

with a false translation, then the argument is valid in the identity calculus,
is essentially Godel’s completeness theorem for this calculus (see Gödel

The principle of the indiscernibility of identicals is first to be found

in Leibniz [1].

7 . Appendix: list of theorems of chapter VI.

T 30 1 X= X
T302 X = y -*-y = X
Section 7 Appeadix: List of Theorems of Chapter V I . 231

T 303 x = y ^ y = lS-> X = Z

T304 X = y * -> y = X

T305 x = y/^Z=y-^X = Z
T306 y = X K y = Z - ^ X = Z

T307 y = X A Z = y - ^ X = Z

T308 je = j (Fx <->

T309 F;» Sy{y = X ^ F^»)

T 3 10 F jc <->'^yiy = XK F>^)
T 3 11 Vx KX = y tr-^Vy K X = y

T 3 12 X = y*r^ As(sr = ->■ 2: = >»)

T 3 13 X = y<r^'^z{z = x s z — y)

T 3 14 F(«y) <-> A2Aw[2 = X kw = y ^ F(sw)]

T 3 15 F(ii^) yz'^u\z = X hw = y K F(2 w)]

T 3 16 X= y A(*) = A (j)
T 3 17 VjcF * a A*Aj(Fic A F^^ -> X = jy) Vx(Fx a Ajy[F>'-» x = j] )

T 318 VxFx A AxAy{Fx ^ ¥ y x = y) WyAx(Fx x = y)

T 3 19 WxWy(Fx A F^ A ~ X = j ) A AxA>’A3r(Fx a F j a F^
x = y v x = z'^ y = z)<-> VxVy(Fx a Fjva x= y a

Aar[Fsr ~ ^ x = z s y = z '\ )
T320 VxVy(Fx A Fj» A ~ X = j ) A AxAjyA2(Fx h Y y a ¥ z -> x = y \ i
X = z v y = 2r)<->VxVy(~X = jy A A2[F2r<-> ar = X V ¡S = y])

T 32 1 Vx X = jy
T 322 VyAx(x = z <-> X = y)
T323 YyAx(A2[x = z<-^z = u ] ^ x = y )
T324 Vy(Ax[Fx<-> X = >»] A Gy) <-> VyAx(Fx<-^ x = y) a

A x ( F x -> G x )

T325 VyAx(Fx X = y) -> (Ax[Fx -> Gx] <-> Vx[Fx a G x])

T326 ~V yA x(~ X = x i- i’ X = y)
T 327 Ax A(x) = B(x) -> [Vx FA(x) <-> Vx FB(x)]
232 . 'Is* (in one sense) Chapter VI

T328 f\aM)Kc[(yzKx\¥x<r^ X = z\ ->Y a)h{'^yzK x\Fx <-> jc = 2]

-> a = c) h (V2rAjy[Fj <r^ y = z\ -> F6) a V2tAj [Fj <->
y = z^ -^b = c) ^ a = h\
T329 AaAftAc[(YyA:x:[F^<-> x = y'] ^ Fa) a {'>^WyKx\¥xi-^ x = y ]
-> a = c) K {yyKx\Gx <-> jc = jy] -> G6) a ( ^ VyA;c[Gjc <-->
X = y'\ -> b = c) h A;c(F jc <-> G:v) -> a = 6]
 Chapter V II
‘T H E ’

I. Descriptive phrases, term s, form ulas. There are arguments

whose validity depends on the meaning of the singular definite article.
For example, the argument
(1) Exactly one student failed. Alfred is the student who failed.
Anyone who failed is illogical. .’. Alfred is illogical
is intuitively valid, yet the procedures of the identity calculus are in­
sufficient to establish its validity.
The word ‘the*, in many of its occurrences, may be replaced by ‘the
object X such that*. For instance, the second premise of (i) may be para­
phrased as
(2) Alfred is identical with the object jc such that x is z student
and jc failed ,
and
the composer of Don Giovanni
as

(3) the object such that composed Don Giovanni

We shall abbreviate ‘the object . . . such that* by ‘T. (2) and (3) then
become
Alfred = is a student and failed)
and
lx X composed Don Giovanni
respectively. ‘1* is known as the descriptive operator and may be written
before any variable to form a descriptive phrase.
The reader will recall that an Ejiglish name is an expression of English
that designates (at least within a given context) a single object, an English
sentence is an expression of English that is either true or false, a term of
English is either an English name or an expression containing occurrences
234 . ‘The* Chapter VII

of variables that becomes an English name when some or all of these

occurrences are replaced by English names, and a formula of English is
either an English sentence or an expression containing occurrences of
variables that becomes an English sentence when some or all of these
occurrences are replaced by English names.
We shall regard expressions of the form
the object a such that ,
where a is a variable, as belonging to English; such expressions will
be called phrases of description. Accordingly, (2) will be construed as an
English sentence, and (3) as an English name.
An expression of the form
. the object a such that (f> ,
where a is a variable and ^ a formula of English whose only variable is a,
will be called a definite description. Examples are (3) above,
(4) the object x such that x is presently king of France ,
and
(5) the object x such that x wrote lolanthe .
English usage is not always clear as to the designation of definite
descriptions. Some clarification is therefore in order. A definite description

(6) the object a such that <f>

is said to be proper if exactly one object satisfies <f>; in this case, following
ordinary usage, we understand (6) as designating that object. In the
case of improper definite descriptions, that is, expressions of the form (6)
for which either no object or more than one object satisfy ordinary
usage provides no guidance; it therefore falls to us to specify their meaning.
It is convenient to select a common designation for all improper definite
descriptions. What object we choose for this purpose is unimportant,
but for the sake of definiteness let us choose the number o. Thus expres­
sions of the form (6), if improper, will henceforth be understood as desig­
nating o.
It follows that (4) and (5), as well as (3), are English names; ( 3 ) - ( s )
respectively designate Mozart, o, and o. (4) and (5) are improper because
nothing is presently king of France and there are two things (Gilbert
and Sullivan) that wrote lolanthe.
More generally, it is now clear that every expression of the form
the object a such that <f> ,
where a is a variable and <f> an arbitrary formula of English, is a term of
English.
Section I Descriptive Phrases, Terms, Formulas • 235

It also follows from our conventions regarding the designation of

definite descriptions that the English sentences
The object x such that x is presently king of France is
married
and
The object x such that x is presently king of France is less
than I
are respectively false and true.
For the treatment of definite descriptions we now consider a language
obtained from that of chapter V I (see p. 213) by adding the descriptive
operator.
The characterization of terms and formulas must be somewhat more
involved than before, for the following reason. The grammatical function
of descriptive phrases (as of their English counterparts) is to produce
terms from formulas. Consequently, not only may terms occur within
formulas, but also formulas within terms. The best way of introducing
the present notions of term and formula is, it seems, to characterize them
simultaneously. Like earlier characterizations the one below is to be
regarded as exhaustive; that is, nothing is to be regarded as a term or
formula unless it can be reached by successive applications of the following
clauses.

(j) Terms of English are terms. {In particular у variables are terms.)
(2) Formulas of English are formulas.
(5) The result of writing a k-place operation letter followed by к terms is a
term. {In particular у a o-place operation letter is itself a term.)
{4) The result of writing a k-place predicate letter followed by к terms is
a formula. {In particular у a o-place predicate letter is itself a formula.)
(5) I f iy ri are termSy then
t = V
is a formula.
(6) I f Ф, Ф are formulas, then so are
^ф ,
(Ф ->Ф ) ,
(ф^ф) ,
{ф A ф) ,
1 ф *~*ф ) •
(7) I f Ф is а formula and <х а variably then
Ааф у
W0L<f>
are formulas.
236 . ‘The’ Chapter VII

(S) I f <f>is a formula and ol a variable y then

1a<^
is a term.

Clause (8) leads not only to

lx X composed Don Giovanni ,
but also to seemingly meaningless combinations such as
lx Socrates is bald .
As with vacuous quantification, it would be artificial to exclude these
possibilities; their interpretation will be explained in due course.
The symbolic terms and formulas of our language are those terms and
formulas which can be constructed exclusively from variables, predicate
letters, operation letters, parentheses, sentential connectives, quantifiers,
the identity sign, and the descriptive operator. More precisely, our
symbolic terms and formulas can be exhaustively characterized by clauses
(3) ” (^) i^bove, reading everywhere ‘symbolic term’ for ‘term’ and
‘symbolic formula’ for ‘formula’, together with the clause:
(i') All variables are symbolic terms.
Examples of symbolic terms are
IxF^x ,
1^G2^A0 ,
lyAxH^xy ,
Iz z = Iwl^wF^ ,
lxAy{ y = X Pxy) .
Symbolic terms produced by clause (8), that is, terms of the form

l^(f> y
where a is a variable and (f> a symbolic formula, are called descriptive
terms.
EXERC ISES

I. State whether each of the following expressions is a term, a formula,

or neither:
(a) I jcFO
(b) IxF^x
(c)
(d) IxÇx = y)
(e) A Gijc)
Section 2 Bondage and Freedom . 237

(f) IjcAijc
(g)
(h) F^lxG^x
(i) F^lxG^xy
(j) = X
(k) Ijc = Ijca: = ;c
(l) F^A^lxF^xA^
2. Bondage and freedom . The descriptive operator, like the quanti­
fiers, binds variables. For example, in the formula
Gn^cFijc
we shall say that both occurrences of 'x' are bound. Further, we shall
speak of bondage in a term as well as bondage in a formula. For example,
we shall say that ‘jc’ is bound in the term
I jcF^jc .
Revisions are thus necessitated in our characterizations of bondage and
freedom. It is convenient to treat bondage and freedom for variables
separately before considering the case of arbitrary symbolic terms.
An occurrence of a variable a is now said to be bound in a symbolic term or
formula (f>just in case it stands within an occurrence in of an expression
Aoii/j ,
Va^ , /
or

where 0 is a symbolic formula. An occurrence of a variable is free in a. sym­

bolic term or formula <f>just in case it stands within <f> but is not bound
in <^. A variable itself is bound or free in a symbolic term or formula (f>
according as there is a bound or free occurrence of it in <f>.
For example, in the formula
G^'lxF^x X ,
the first and second occurrences of ‘.x’ are bound, but the third occurrence
is free, and 'x^ itself is both bound and free.
An occurrence of an arbitrary symbolic term ^ is said to be bound in
a symbolic term or formula (f) if it stands within an occurrence in <f> of
some expression
Aoup y
Va^ ,
or
lai/f ,
238 . ‘The’ Chapter VII

where ^ is a symbolic formula and a a variable which is free in An

occurrence of a symbolic term is free in <f>if and only if it stands within <f) but
is not bound in <f>. A symbolic term is itself bound ov free in (f> according as
there is a bound or free occurrence of it in
For example, in the formula
(i) Q /yFnxW xy V C n x W x y ) ,
the first occurrence of the term

(2) IxH^xy
is bound, and the second free. Thus (2) is both bound and free in (i).
Observe that bondage and freedom as applied to variables (and as
characterized on p. 237) are special cases of bondage and freedom as
applied to arbitrary symbolic terms.
A symbolic sentence can be characterized as a symbolic formula in which
no variable is free, and a symbolic name as a symbolic term in which no
variable is free.
The characterization of proper substitution of a term for a variable
remains as before, even though our language now includes descriptive
terms. Thus we again say that a symbolic formula ip comes from a symbolic
formula <f>by proper substitution of a symbolic term ^for a variable a if 0 is
like <l> except for having free occurrences of i wherever <f> has free occur­
rences of a.

EXERC ISES

Consider the formula

V:ii I jc jc = B^y = l5r(Ay A^x = B^y v .
2. In this formula identify each occurrence of a term that is not a
variable as bound or free.
3. Which terms (including variables) are bound in the formula?
4. Which terms (including variables) are free in the formula?

3« Inform al notational conventions. Once again we shall employ

the conventions of chapter II (pp. 40, 51, and 65) for omitting
parentheses and replacing some of them by brackets.
As in chapter IV, we shall usually omit superscripts from operation
letters and predicate letters, inserting parentheses and brackets to avoid
ambiguity. In the present context, i-place predicate letters, as well as
predicate letters of two or more places and operation letters of one or
more places, will require parentheses or brackets. Thus we may write

(I) A iljfF i*
informally as
Section 4 Translation and Symbolization • 239

A(1;cF[jc]) ,
but not as
(2) A(1jcF;c) .
The official counterpart of (2) is not (i) but
A^lxYOx .
Thus, in unofficial notation, all predicate and operation letters of one
place or more will bring with them a pair of parentheses or brackets.
(In formulating inference rules and criteria of bondage and freedom,
we always have in mind official notation.)

EXERC ISES

5. For each of the following formulas (in unofficial notation) delete

inserted parentheses and brackets and restore omitted superscripts
in accordance with the conventions of this section.
(a) I jcF(jc)
(b) A(\x¥{x))
(c) ¥(\xG(Ax))
(d) Afl^A(:c) = A:]
(e) F[Al:cF(jcA)]

4. Translation and sym bolization. An abbreviation is to be under­

stood in the same way as in the last chapter (p. 216). The process
of literal translation into English on the basis of a given scheme of abbreviation
may now begin with either a symbolic term or a symbolic formula, and
if successful will end with a term or formula of English. The process
consists of the steps given in chapter IV (pp. 13 9 -4 0 ), with one modi­
fication. Step (v) should now read as follows:

{v) Eliminate sentential connectives^ quantifier phrases^ ‘ and de­

scriptive phrases in favor of the corresponding phrases of English^ preserving
all parentheses.

An English term or formula 0 is said to be a free translation (or simply

a translation) of a symbolic term or formula <f> (and (f> is said to be a sym­
bolization of i/j) on the basis of a given scheme of abbreviation if 0 is a
stylistic variant of the literal English translation of (f>based on that scheme.
For example, on the basis of the scheme
F^ : a is z doctor
G2 : is father of b
H2 : a loves b
P : a wrote b
240 . ‘The* Chapter VII

J2 : a is larger than b
A® : Socrates
BO : Waverley ,
the symbolic terms
(1)
(2) I jcG( jcA) ,
(3) TyA^H(xy) ,
(4) 2T = ’\wl{wB) ,
(5) '^xAy{'^y = X ->
have the following respective literal English translations:
the object x such that is a doctor,
the object x such that x is father of Socrates,
the object y such that, for each Xy x loves y,
the object z such that z is identical with the object w such
that w wrote Waverleyy
the object x such that for each y (if it is not the case that y
is identical with Xy then x is larger than y) .
(Observe that in step (i) of the process of translation, that is, the restora-
ration of official notation, ( i) - ( 5 ) become the examples on page 236.)
The following terms of English are free translations of ( i) - ( 5 ) :
(6) the doctor,
(7) the father of Socrates,
what everyone loves,
he who is identical with the one who wrote Waverleyy
(8) the largest thing.
Thus we regard ‘the’, ‘what’, ‘he who’, and ‘the one who’, in some of
their uses, as stylistic variants of ‘the object . . . such that’.
It should not be supposed that every occurrence of ‘the’ can be sup­
planted by a phrase of description. Consider, for example, the sentences
(9) The whale is a mammal,
(10) The members of the club are Republicans,
(11) The trespassers were prosecuted.
(9) asserts that every whale is a mammal, not that the object x such that
is a whale is a mammal; and attempts to express (10) and ( ii) with the
help of phrases of description would lead to ungrammatical results. In
(6)-(8 ), ‘the’ generates a name, but this is not the case in ( 9 ) - ( i i ) ;
even ‘the whale’ cannot reasonably be construed as designating in (9)
a single object.
Section 4 Translation and Symbolization . 241

To find a symbolization of a given term or formula of English on the

basis of a given scheme of abbreviation the reader will find it useful to
proceed roughly as follows:

(j) Introduce Hs identical with\ phrases of description^ phrases of quantityy

and phrases of connectiony the latter accompanied by parentheses and occurring
canonicallyy in place of their stylistic variants,
(2) Reverse the steps leading from a symbolic term or formula to a literal
English translation.

consider the scheme of abbreviation

F2 : a is wife of b
G2 : a is more salacious than b
HI : a is a woman
P : a is mentioned by b
AO : Justinian
BO : Gibbon
and the sentence
The wife of Justinian is the most salacious woman men­
tioned by Gibbon.
In step (i) we might obtain
The object x such that x is wife of Justinian is identical with
the object x such that ((jc is a woman and x is mentioned
by Gibbon) and for each y (if {{y is a woman and y is men­
tioned by Gibbon) and it is not the case that y is identical
with x)y then X is more salacious than y)) ,
and carrying through the successive parts of step (2) we obtain the sym­
bolization
l;vF(;x:A) = l:v(H(^) a ]{xB) a Ay[H{y) a J(jyB) ^ y = x G(xy)]) .

EXERC ISES

Symbolize each of the following formulas on the basis of the scheme

of abbreviation which accompanies it. Exercise 6 is solved for illustra­
tion.

6. The man who lives at the North Pole does not live there.
On the basis of the scheme of abbreviation
pi is a man
G2 a lives at b
AO the North Pole
242 . ‘The’ Chapter VII

No. 6 becomes
^G(\x [F{x) a G { xA ) ] A ) .

7. The positive square root of 2 is a positive even prime. (F^ : a

is positive; : ¿z is a square root of 6; : zz is even; P : a is
prime; : 2)
8. Mary loves the one who loves Mary. (F^ : a loves b; A® : Mary)
9. If he who murdered Desdemona was murdered by Desdemona’s
murderer, then he committed suicide. (F^ : a murdered b; G^ : a
committed suicide; A® : Desdemona)
10. The hardest problem on the examination was solved by no one.
(F^ : ¿zis a problem on the examination; : zzis harder than b; № :
a was solved by b)
11. If the author of Waverley is the author of Ivanhoey then the
author of Ivanhoe is an author of Waverley, (F^ : zz is an author of b\
A® ; Waverley \ : Ivanhoe)
12. God is that than which nothing greater can be conceived.
(A® : God; G^ : zz is greater than b\ F^ : zz can be conceived)
13. The positive square root of the square of the even prime is
irrational. (F^ : zz is positive; G^ : zz is a square root of 6; № : a
is even; D : zz is prime; : a is irrational; A^ : the square of zz)
14. The smallest positive integer is that positive integer which
when multiplied by itself equals itself. (F^ : zz is less than b\ G^ : zzis
positive; : zz is an integer; A? : the product of zz and b)
15. There is no greatest positive integer. (F^ : a is greater than b\
G^ : zz is positive; Hf : zz is an integer)

5. In feren ce ru les. For the logic of the descriptive operator we

must adopt two new inference rules, the first of which is the following.
Proper descriptions (PD): Vj8Aa(<^ <-> a = jS)

Here a is to be a variable and ф a symbolic formula, jS is to be a variable

not free in фу and ф is to come from ф by proper substitution of
M
for a. PD corresponds to the principle that if a given condition is satisfied
by one and only one object, then the object satisfying that condition
satisfies it. PD leads, for example, from the sentence
WyAx{F{x) 4-> л: = jy)
(which could be translated as the assertion that exactly one thing wrote
Waverley), to the sentence
F[l;cF(^)]
Section 5 Inference Rules . 243

(which would then assert that the author of Waverley wrote Waverley), A
somewhat more complex application of PD is the inference from the
formula
WyAx(F{x) A G{xzw)<r-> X = y)
(which could be taken as asserting that there is exactly one integer between
z and w) to the formula
F(1jc[F(;c) a G(jc-sr«;)]) a G (1 jc[F(jc) a G { x z w )] z w )

(which would then assert that the integer between z and w is an integer
between z and w).
Upon first consideration of such examples, one might believe that
any English sentence that is a translation of a conclusion of PD is true.
But consideration of further examples would quickly dispel this belief.
The sentence
The man who lives at the center of the earth is a man living
at the center of the earth
is false, for it implies the falsehood that some man lives at the center of
the earth. The assumption of propriety that constitutes the premise of
PD is, however, sufficient to exclude such examples.
Because we seldom have occasion to use improper descriptive terms
(that is, symbolic terms corresponding to improper definite descriptions),
we could develop a significant part of the logic of the descriptive operator
on the basis of the single rule PD. There is, however, a strong reason for
introducing along with PD a rule pertaining to improper descriptive terms.
Without such a rule, we could not extend A V and IE to the context
la ^
(where ^ is a symbolic formula and a a variable); for example, we could
not justify the inference from
F(\xG{x))
to
F(]yG{y))
or
F ( lx ~ ~ G ( * ) ) .
There are other reasons for adding a rule concerned with improper
descriptions, and one of these reasons will appear in section i i . The
relevant new rule has the following form.
Improper descriptions (ID): Vj3Aa(^ <-> a = jS)
la<^ = ly /X/ y = y
244 . ‘The’ Chapter VII

Here a, j8, and у are to be variables, and ф a symbolic formula in which

is not free.
The rule ID has no intuitive counterpart, simply because ordinary
language shuns improper definite descriptions. The rule corresponds,
however, to the resolution that every improper definite description is to
designate the same object as the particular improper definite description
the object x such that x differs from x ,
and this resolution accords with the interpretation given on page 234.
The adoption of ID will enable us later to extend A V and IE to con­
texts involving descriptive phrases.

EXERC ISES
16. For each of the following pairs of formulas, state whether the
second follows from the first by PD, by ID, or by neither.
(a) WyAx(F{x) Ax = z<r^x = y)
F( 1 jc[F(jc) л л: = 2t]) л 1 ^[F(^) a x = z ] = z

(b) VyAx{H{z) <r-> X = y)

l2H(2) = Iw W= W
(c) VyAx{F{xy) X = y)
F(\xF{xy)y)
(d) WyAx{¥{z) V A^G(jc^) <r->x = y)
¥{z) V A^G(1^[F(^) V A^G(^;sr)]j2r)
(e) ^УyAл:(УyF(л;y) <r-> x = y)
“\xVyF{xy) = lw w= w
(f) VyAx{VxF{x) ^ X = y)
WxF{x)
(g) УуАх{х = z <r-^x = у)
Лxx = z = Л x ^ x = ^ x
(h) VyAx(V^F(^) V G(л:) <r^x = y)
WxF(lx[VxF(x) V G(^)]) V G(^л:[Vл:F(л:) v G(л;)])

6. T heorem s w ith im abbreviated proofs. We may call that branch

of logic which essentially involves the descriptive operator, as well as
the identity sign, quantifiers, and sentential connectives, the description
calculus. The directions for constructing an unabbreviated derivation
within this calculus remain as before (see chapter IV, section 13, clauses
(i) - (6)), with the understanding that PD and ID , as well as Id-i and Id-2,
are now to be included among the primitive inference rules, and the
words ‘term* and ‘formula*, throughout clauses (i) - (6) as well as in the
formulation of the various inference rules, are to assume the broader
senses of the present chapter.
Section 6 Theorems with Unabbreviated Proofs . 245

The characterizations of a complete derivation, derivability^ a proofy

a theorem, an argument, a symbolic argument, a valid symbolic argument,
an English argument, a symbolization of an English argument, and a valid
English argument remain as before. (See pp. 95, 103, and 117.)
Theorems of the description calculus will receive numbers beginning
with 401.
T401 I. S h e w ^ h x { ¥ { x ) < r -^ X = y) ^ = y
Ax{F{x) X = y)
WyAx(F{x) X = y) 2, EG
F(]xF{x)) 3>PD
F(1:cF(:v)) ^ l^F(;x:) = y 2, UI
F(lxF(;v)) I^F(^) = y 5, BC
1;cF(x) = y 4, 6, MP
T402 WyAx(F{x) <r^x = y) F(1^F(^))
T403 '^WyAx{F{x) X = y) I jcF( jc) = "Iw w —w
The next two theorems have an unusual form. They are conditionals
with antecedents that are theorems of the identity calculus. We shall
be interested only in their consequents, which obviously are also theorems
and appear later as T406 and T407. At present, however, with only
unabbreviated derivations at hand, proofs of the antecedents of these two
theorems would run to exorbitant length; it is thus expedient to postpone
the statement of the consequents until such time as we shall allow pre­
viously proved theorems to figure in a derivation.
T404 I. 1 ^F( a:) = ly F (y )
2. T328
3- AbAc[[yzAx[F(x) ^ X = z ] -> ¥0x¥{x)))
V^A;c[F(:v) X = z] TvF(:x:) = c)
(V2:Ay[F(y) <-^y = z]-^ F(^)) ^
V2:A3;[F(y) = z] b = c) ^
lxF{x) = b] 2, UI
Ac[(V2^A;c[F(:!c) x = z] F(l:vF(^))) a
V 2 :A ^ [ F ( : x:) <-> X = z ] ^ I jcF ( jc) = c) a
(V2'Ay[F(y) y = z] -> F(1yF(j))) A
('^V2Ay[F(y) y = 2 :]^ ly F (y ) = c) ^
^xFix) = 1yF{y)] 3. U I
(V^A:v[F(.r) <r-> X = z] F(1a:F(jc))) a
V2:Ajc[F(.v) ^ X = z]
IxF(.v) = I w w = w) A
(V^Aj [F(j ) ■ (->>’ = ^] -^ F(>F(>'))) A
(~V irA j[F(j) y = s] ^
1j F ( j ) = "\w ~ w = w ) I.vF( a') = I j F ( j ) 4. UI
246 . ‘T h e ’ Chapter VII

6. ^ ac = ^ F( 1 xF( a;))
7- VsiA*[F(A;) a; = «]
8. F( 1 a:F(a;)) 7. P D
Show ~V2AAc[F(Af) <-> a; = 2]
1a;F(ac) = Iw w = tv

10. ~ V3rAA:[F(A;) <->a; = a]

11. 1 a;F(a;) = Iw ~ to = w 10, ID
12 . Skew-VzAy[F(y) <-*y = 2] -» F ( 1j;F(y))
13- V3rA;y[F(jy) ^ y = z]
H- F (V F (j) ) _______ ___ 13. P D
15- Show ~ V2Aj[F(y) ^ y = z] ->
1jyF(j) = Iw ~ w = w
i6. '>^Wz^y[F{y)<-^y = z]
17- I j F ( j ) = Iw w = w 16, ID
i8. (Va:AA:[F(A;)<->Af = sr]->■ F ( 1a:F ( a:))) a
( ~ V 2 A a:[F ( a:) <-> a; = .s ] ->
1a:F ( a:) = I zo ~ w = w ) 6,9, Adj
19. (V2:A a;[F ( a;)'<-> a; = z ] - ^ F ( 1a:F ( a;))) a
( ~ V ^ A a;[F ( a:) <-> a: = s] ^
1a:F(ac) = iw ~ w = w) A.
(VsAj[F(>^)<-^>' = z] -> F(1jF ( j) ) ) 18, 12,
Adj
20. (V2:A a;[F ( a:) <-> a: = sr] -> F ( 1a:F ( a:))) a
( ~ VaiAAr[F(A;) x == z] ->
1a:F ( a:) = Iw ~ w = w) a
(VsA>^[F(y) = sr] -> F (V F (j))) A
( ~ VsrAy[F(y)<->^ = z] ^
1jF ( y ) = 1«) ~ w = w) 19. 15.
Adj
2 1. 1a:F ( a:) = l;v F ( j)
5, 20, MP
T405 I. Show'T-^if) -> fAA:fF(A;) <-> G ( a;)] -> lAgF(A;) — 1a;G ( a:)]
2. T329
3- Show AA:[F(;e) G iA ;!! -> 1j;F(Ag) = ’ixGjx)
4- AA;[F(Ar) *-* G ( a:)]
AAAc[(VyAA:[F(A:)<-> a; = y] F ( 1acF ( a:))) a
5-
( ~ V yAA;[F(A:)<-i- a = y ; ] ^
(yyAAc[G(A;) ♦ -> a: = >'] "^ G (6 )) ^
( ~ VyAA:[G(A:) < - y x ~ y ] ~ ^ ^ ~ ^ ^ ^
2, U I
AA:[F(A:)<->G(Ar)]
1a:F ( a;) = 6]
Section 6 Theorems with Unabbreviated Proofs . 247

6. Ac[(yyA;«[F(jc) jc = j ] -> F (l*F (ir))) a

( ~ VjAA;[F(j!r) •<->*= 3^] i*F(2:) = c)^
(VyA2:[G(ai:) <-> ic = j ] ->• G ( 1jcG ( jc))) a
(~ VyAa;[G(A-) <-> * = 3 /] 1* G (* ) = c) A
A ji[F(^) G (*)]
lA:F(ji;) = liiG(ic)] 5. U I
(VyA2:[F(2:)<-» jc = y] -*■ F(licF(a:))) a
( ~ yyA»:[F(*) * - > ->
x = y ]

ljcF(a:) = Iw ~ w = to) A
(VyAic[G(jc) ■ «-> 2! = >»] -> G ( 1jcG ( jc))) a
( ~ YyAac[G(ic) •<-> 2: = j'] ^
la:G(A;) = la ; ~ to = to) a
Ajc[F(ar)<-> G (x)] -*■
1icF(*) = 1*G(jc) 6, U I
5A«w-VjAA:[F(ic) = j ] ->■ F(ljiF(*))
9- YyAjc[F(a:) <-> a: = j»]
10. F ( 1a:F(ic)) 9. PD
11. Skote ~ Vj'AA;[F(a;) a; = j ] ->
la:F(a:) = Iw ^ w = to
12.
13- 12, ID
14.
IS-
16. 15. PD
17 -

18.
19. 18, ID
20. (yyAjc[F(a:) .(-> a; = j ] ^ F(1*F(*))) a
(~VyAa:[F(a:) <-> a: = jy] ->
la:F(a:) = lto ~ a ; = to) 8 , II,
Adj
21. (VyAa;[F(a:) •<-> a: = j»] F(1a;F(a:))) a
( ~ VyAa;[F(a:) = y]-*-
ia:F(a:) = Ito ~ to = to) a
(VyAa;[G(a:) •<-> a; = >>) -> G(1a:G(a:))) 20, 14,
Adj
22. (YyAa;[F(a:) a: = j'] F(1a;F(*))) a
( ~ VyAa:[F(a) ■«->a: = j»] ^
la:F(a:) = Ito ~ to = to) a
Vj'Ai[G(a) <-> * = jy] ->■ G(la:G(a;))) a
*4 8 . ‘The’ Chapter VII

( ~ VyAjc[G(A;) •(-» Of = j ] ^
l;cG(*:) = la> ~ n> = w) 21, 17,
Adj
23- (VyA*:[F(jc)<->X = y] ^ F(l^FW)) ^
(~yyAjc[F(ic) •<-> Af = >»] ->
1 acF ( a;) = lit) ~ w = w) a

(VjyAA(;[G(Ac) <-> AC = >)] -» G ( 1 acG ( ac) ) ) a

( ~ V j A ac[ G ( ac) <-> AC = j ] ->

I acG ( ac) = la ) ~ o ) = a;) a
A ac[ F ( ac) •«-> G ( ac)] 22, 4, Adj
24. I acF ( ac) = I acG ( ac) 7. 23, MP

EXERCISES

17. Prove T402 and T403.

7. Alphabetic variance; proper substitution; abbreviated deri­

vations. In chapter I II we introduced the relation of equivalence between
Two symbolic form ulas
form ulas. (f> and i/j are said to be equivalent just in
case the biconditional
tfj
is a theorem. This notion can be extended naturally so as to apply as well
to terms. The symbolic terms ^ and rj are equivalent ]ust in case the identity

i = V
is a theorem.
T404, together with T328, will enable us to establish the equivalence
of two descriptive terms differing only in their initial bound variables.
This suggests that we extend to terms related in this way the notion of
alphabetic variance introduced in chapter IV, section 7. Thus we call two
symbolic form ulas immediate alphabetic variants if (as before) they have
the forms
Acf,(f>

and

or else
W0Lcf>
and
Va'c^' ,
Section 7 Abbreviated Derivations • 349

and we call two symbolic t e r m s i m m e d i a t e a l p h a b e t i c v a r i a n t s if they have

the forms

and
l a 'f
in each case a and a' are to be variables, (¡> and are to be symbolic
formulas, <j> is to come from <f> by proper substitution of a' for a, and <f>
is to come from <f>' by proper substitution of a for a'. Symbolic terms
or formulas that differ only in parts that are immediate alphabetic variants
of one another are also equivalent, and may be called alphabetic variants.
Thus we replace the previous notion of alphabetic variance by the follow­
ing: a symbolic term or formula iff is an alphabetic variant of a symbolic
term or formula «/r' just in case ifs is like i/r' except for having one or more
occurrences of a term or formula <f>where i/f' has some immediate alphabetic
variant of <f>.
For example,
lx[¥{x) A G{x)]
is an immediate alphabetic variant of
W M a GCv)] ,
and therefore
= lx[F(^) A G{x)]
is an alphabetic variant of
z = ^ y [F (y ) A G {y )] ;
further,
WxF{xz)
is an immediate alphabetic variant of
WyF{yz) ,
and therefore
lW:vF(x5r)
is an alphabetic variant of
lzWy¥{yz) .
To define the notion of an of a formula of the description
i n s t a n c e

calculus we can employ the notions of p r o p e r s u b s t i t u t i o n o f a f o r m u l a f o r

a and
p r e d i c a t e l e t t e r p r o p e r as characterized in
s u b s t i t u t i o n o n v a r i a b l e s
*50 . ‘The* Chapter VII

section 8 of chapter IV. (In the characterization of these notions we

must now, of course, understand ‘term’, and ‘formula’, in the sense of
the present chapter.) For proper substitution of a term for an operation
letter^ however, we need a slightly modified notion now that variables
may be bound in terms. We say that a symbolic formula ф comes
from a symbolic formula ф by proper substitytion of a symbolic term r)
for a A-place operation letter S just in case there is no variable occurring
in both Ф and 77, and ф can be obtained from ф in the following two steps:
I. Throughout фу replace S by enclosed in braces.
II. In the expression resulting from I, successively consider each
part of the form

where ii, . . . , are terms. Replace each such part by a certain term—
in fact, by the term obtained from r] by replacing all free occurrences
of ‘a’ by ii, ‘6’ by ^2, etc., up to the ^th variable, all of whose free occur­
rences are to be replaced by i*.
(The only alteration is that here we speak in step II only of free occur­
rences of variables; in chapter IV this qualification was unnecessary, for
before the descriptive operator was introduced there was no way for
a variable to be bound in a term.)
For example, from the formula
Wxx = A{y)
the formula
Wxx = lz^ aY {az) v G(y)]
comes by proper substitution of the term
lz \ 4 aY{az) V G(^)]
for the operation letter ‘A ’ .
We say as before that a symbolic formula ф is an instance of a symbolic
formula 0 just in case ф is фот obtainable from ф by one or more operations
of proper substitution—whether on predicate letters, on operation letters,
or on variables.
We once again remind the reader that he should mentally restore official
notation when checking applications of the preceding notions.
Now that the notions of instance and alphabetic variance have been
extended, we may reintroduce three of the abbreviatory clauses of chapter
IV :

(7) I f Ф is an instance of an alphabetic variant of a theorem which has

already been provedy then ф may occur as a line, {Annotation as prescribed in
chapter / F , section 9.)
Section 7 Abbreviated D erivations •251

(5) A symbolic formula may occur as a line if it is an alphabetic variant

of an antecedent line, {Annotation as in chapter IV ,)
(9) I f are symbolic formulas such that

is an instance of an alphabetic variant of a previously proved theorem^ and

iff, iff' are symbolic formulas such that ip' is like ip except for having one or
more occurrences of <f>' where \p has occurrences of then

may occur as a line, (As beforey we call

an interchange of equivalents on the basis of the previously proved theorem

and use the annotation with a parenthetical reference to this theorem.)
We shall now introduce an abbreviation analogous to the derived rule
known as Leibniz’ law (p. 223). The reader will recall that in an application

i = r

of L L , the occurrences of ^ and which are interchanged in passing from

<f> to (f>' must all be free. By strengthening the premise, we can dispense
with this limitation. For example, the inference from
Ax A{x) = B(jc)
to
WxF{A{x))<r^WxF{B{x))
though not justified by L L , is correct in view of T327. The premise of
this inference is no longer an identity, but a universal generalization
of an identity; the important feature is that all variables bound in the
conclusion (in this case, only *x') are universally quantified in the premise.
More generally, the pattern of inference presently contemplated has
the form

(0 A a i . . . A a n C = C'

where ^ and are symbolic terms, <f> and <f>' are symbolic formulas, <f>' is
like <f> except for having one or more occurrences of t , ' where <f> has occur­
rences of i, and a i , . . . , a« are all the variables bound in <f>. (If no variables
are bound in <f>, the premise reduces of course to
c = r .)
252 . ‘The’ Chapter VII

We may consider a similar analogue to Euclid’s law. Here the inference

has the form
A a i...A a n g = ^
(2)
1= 1
where rj, tj' are symbolic terms, rj’ is like tj except for having one or
more occurrences of where tj has occurrences of and a i......... an are
all the variables bound in -q.
In the inferences (i) and (2) the premises contain terms. Comparable
forms of inference are available whose premises instead contain formulas.
They are
A a i. . . Aan(^
(3)

and
A a i. . . Ахп{ф <-> Ф')
(4)

Here i/jy ifj'y (f)y <f>' are to be symbolic formulas, and rjy rj' are to be symbolic
terms. We assume in addition that, in the first case, <^' is like <f> except for
having one or more occurrences of where (f> has occurrences of ^ and
ai, . . . , an are all the variables bound in <f>y and, in the second case, r)'
is like 7] except for having one or more occurrences of ifj' where t] has
occurrences of 0 and ai, . . . , are all the variables bound in rj.
For example, the inference from
Ax[F(x)<-> G{x)]
to
WxF{x)<-^WxG{x)
correct in view of T 214, is comprehended under (3).
In each of the inference patterns (i) - (4), we shall say that the conclusion
follows from the premise by interchangey and we incorporate these patterns
into our system by the following abbreviatory clause:

( j o ) a symbolic formula may occur as a line if it follows from an antecedent

line by interchange. {Annotation: ^InV and the number of the antecedent line.)

As before, we add a clause permitting the compression of several steps

into one:

( jj) A symbolic formula may occur as a line if it follows from antecedent

lines by a succession of stepSy and each intermediate step can be justified by
one of clauses (2), (5a), (7), (5), (9), or (10). {Annotation as in chapter IV.)
Section 7 Abbreviated Derivations . 253

The special annotations ‘S C , ‘C D ’, ‘QN’, ‘Sm’, ‘T ’, and ‘L L ’ will be

used as before. We shall also use ‘E L ’, as in chapter VI, to stand for the
pattern of inference corresponding to T 316. Now that variables can have
bound occurrences in terms, a slight reformulation is required:
Euclid’s law (E L ): i i'

Here i, i', 7], 7]' are to be symbolic terms, and is to be like t] except for
having one or more free occurrences of where tj has free occurrences
ofi.
There are close connections among Leibniz’ law, Euclid’s law, inter­
change of equivalents, and interchange. Indeed, the first three can rather
easily be subsumed under interchange, given the principle of alphabetic
variance (clause (8)).

Many inferences justified by LL, such as that from

(5) ^
to
(6) [Уд;Р(ж) V С(л;)] •<-+ [Уд;Р(л:) v G(^)]
are not immediately comprehended under Int; but this rule, assisted by familiar
laws, leads from (5) to (6) by the following sequence of steps:
(i) л = у
(ii) Az X = у i, T227, BC, MP
(iii) [Wz¥{z)yG{x)] ^\yzF(z)wG{y)] ii, Int (form (i))
(iv) [WxF{x)vG{x)]< [VxF{x)yG{y)] iii, AV
The second form of Int bears a similar relationship to EL. The fact that a
formula is a theorem just in case all its closures are theorems clarifies the relation
between IE and the third form of Int.

Clauses (7) - (i i), like their earlier counterparts, satisfy the requirements
for abbreviatory clauses that are given on page 58. A proof of the theoreti­
cal dispensability of these clauses would be rather involved and is not
attempted here; a few relevant remarks, however, will perhaps be appro­
priate.

To show clause (8) to be dispensable, we should have to show the existence

of unabbreviated proofs, not only of T231 and T232, but also of the formula
Ъ Ц х) = lyP(y)
and its instances. But this would be relatively easy, in view of the unabbreviated
proof of T404 given in the previous section, and the fact that the antecedent
254 • ‘The* Chapter VII

of T404 is a theorem of the identity calculus (for whose theorems we already

suppose unabbreviated proofs to exist).
Similar remarks apply to the relation between clause (10) and the formula

A x[F{x)^ G(x)] ^ lx F ( x ) = 1.vG(x) ,

which has an unabbreviated proof in view of the unabbreviated proof of T405

given in the previous section. (Of course, several other theorems would also be in­
volved in eliminating applications of clause (10), but they are all theorems of pre­
ceding chapters.)
The easiest way to show the eliminability of clause (9) is first to treat clause
(10) and then show that clause (9) is reducible to it.

We mentioned earlier that without the rule ID we should be unable

to justify the use of AV and IE for the description calculus. The reason
is now relatively clear. We could not claim the dispensability of clauses (8)
and (9) without having T404 and T405 at our disposal, and the proofs of
these theorems essentially involve ID.
The four forms of Int are respectively illustrated by the following
arguments and accompanying derivations.

(7) A(rv) = A( jva:) VjcAjF[A(xy)] VjcAjF[A(jy^)]

V;cAjvF[A(jcj)] V;cAjvF[A(jy^] ]
2. SkeW‘ AxAyA{xy) == ^iyx)
3* A(jcy) = A(j :k) Premise
4- VxAy F[A(xy)] VAfAjyF[A(jjc)] 2, Int (form (i))

A(xy) = I jc A(ry) = y = ~\x A(yx) = y

1. Show I jc A{xy') = y = I jc A{yx) = y
Skew AxAy A{xy) = A(jac)
3- A{xy) = A{yx) Premise
4- 1 ;« A(xy) = j = lx A(yx) = y 2, Int (form (2))

~ F(xy) . ~ G(j^) VyAjcF(jcj) <-> VyAjcG(j^)

1 . S how WyAx¥(xy) WyAxG(xy)
2. iSA««>-AjcA;v[F (^ ) •«-> G(jcy)]
F(iiy) G(a:y) 1st premise, 2nd
premise, Adj,
T85, MP
V);AjcF(j(y) ^ yyAjcG(xy) 2, Int (form (3))
Section 7 Abbreviated Derivations . 255

F(jçy) G{xy) /. ’]xAyF{xy) = '\xAyG{xy)

1. Skew-lxAyF(xy) = ’\xAyG{xy)
2.
F(ry^) G(jcjy) I Stpremise, 2nd
premise, Adj,
T85, MP
^xAyF{xy) = l^A jG (rv) 2, Int (form (4))

To appreciate the economy gained by Int the reader should, for example,
construct a derivation corresponding to (7) employing L L rather than
Int.

EXERCISES
18. Which of the following pairs of expressions are alphabetic
variants?
(a) y X lxF(:c) = x
Wx lyF(y) = X
(b) Wx IjcF(xy) = X
y y IxF(ry) = y
(c) lxF{x) = lyG(y)
IxF(x) = lxG(:r)
(d) G(lxF(;c)l:vF(;c)y)
Gn«x^F(:r) lyF(y)y)

19. For each of the terms (i) - (v) below indicate the instance, if
any, that can be obtained from the theorem
Ax y = A(;c) -^ y x y = A{x)
by proper substitution of that term for ‘A*.
(i) "\z z = a
(ii) "^aa = a
(iii) ^z X = a
(iv) ’\z\yaF{az) v G(û)]
(v) a= a
20. The second of each of the following pairs of formulas is an
instance of the first; indicate in each case a sequence of substitutions
by which the instance can be obtained.
(i) S y h {x )= y yy^x Z = X = y
(ii) y y A(x) = y yy"\x X = X = y
(iii) y y A(:c) = y yy"\z X — Z == y
256 . ‘The* Chapter VII

21. Give short proofs of the following theorems.

(i) hx{¥{x) <-> P) {hxF(x) <-> P)
(ii) lxAy[¥{y) -> G{x)] = ly[yx¥(x) -> G(y)]
(iii) Ax A{x) = B(jc) -> ly A(y) = y = ^]z B(2) = z

8. T heorem s w ith abbreviated proofs. The consequents of T404

and T405 can now be proved without pain. It is only necessary to employ
clause (7).

T406 I jcF( jc) = TyF(jv)

T407 A;c[F(:v)<-> G(x)] I jcF( jc) = l.rG(jc)

T408 gives a necessary and sufficient condition for a descriptive term

to be proper.

T408 I. A;c[F(jc) X = 1.rF(jc)] <-> VyA.r(F(;c) x = y)

2. A:v[F(^)<-> a: = 1jcF(:v)] ^
VyA^(F(^)<-^ = y)
3* A;c[F(^)<-> jc = 1 jcF(;c)]
4- VyA;c(F(;c) <-^x = y) 3, eg

5- Show WyAx(¥{x) X = y)
Ajc[F(jc) ^ = IjcF(jc)]
6. WyAx{¥{x) X = y)
7- Ax(¥{x)<r^x = z) 6, E l
8. IjcF(jc) = z 7, T 401, M P
9* A jc[F(:r)^;v = IjcF(jc)] 8, 7, L L
10. A :r[F (x )^ ;r = 1jcF( jc)]<->
YyAjc(F(jc)<->^ = y) 2, 5» CB

T409 and T 410 are useful corollaries of T408.

T409 VyAjc(F(;c)<-> X = y) A F(;c) -> X = 1^F( a:)

T 410 WyAx{¥{x)<-^ X = y) A X = 1jcF(;c) -> ¥{x)

To obtain the converse of T407, we must assume that the descriptive

terms involved are proper:

T 4 11 VyAx(¥(x) X = y) A yyAx{G{x) <r-> x = y)

[Ax[¥{x)<->G(x)] <-> lx¥{x) = 1^G(a:)]

As we have seen (p. 243), an assertion that the object satisfying

has the property expressed by <f> will not always be true. T 412, together
Section 8 Theorems with Abbreviated Proofs . 257

with its instances, gives a necessary and sufficient condition for the truth
of such a sentence.
T 412 I. F(1.x:F(^)) yyKx{F{x) ^ = y) V w = w)
2. iS / i^ € i^ F ( l:v F ( :v ) ) a ' ^ W y A x ( F { x ) < ->

X = y ) -> F(1ztJ ^ w = zv)

3- F(l:vF(;c)) a ^WyAx(F{x) <r^ x = y)

4* F(1tt; = zv) 3, S, ID , 3, S, L L
5- Shew-WyAx{F{x) ^ = y)-> F(1:vF(:v))
6. VyA;c(F(;c) <-> x = jy)
7- F(l;cF(:c)) 6, PD
8. Show ^ YyA:c(F(:v) <-> x = j) a
F(lz(; zv = zv) -> F(1^F(:v))
/^VyAjc(F(A:) <-> a: = jv) A
F(1zi; ^ z v = zv)

10. F(\xF{x)) __________________ 9, S, ID, 9, S, L L

11. F(1xF(^)) <r-^ WyAx(F{x) <r^x = y )^
F(1«; zv = zv) 2, T27, BC, MP,
IE(T4 s ), 5, 8,
Adj, T52, BC,
MP, CB
T 4 13 is a corollary of T 4 12 ; it gives alternative conditions under which
the inference from
X = 1jcF(;c)
to
F(^)
is valid.
T 413 [ '^ x = lzv'>^zv = zvy F(1zc; ^ zo = zv)] a
X = 1jcF(^)->F(;c)

The converse of T401, like that of T407, requires a hypothesis of pro­

priety.
T 414 VyAjc(F(;c) ^ X = y) [A;c(F(jc) ^ a: = y) <-> 1:vF(.v) = y]
On the basis of ID we can show that any two improper definite descrip­
tions designate the same thing.
T415 />
./VyA;c(F(.x:) ^ a '^WyAx{G{x) <r-> x = y)
lxF{x) = lxG {x)
T 416 gives a necessary and sufficient condition for the truth of any
sentence containing a descriptive term; T324 and T325 are useful in
its derivation. T 417, which can be derived easily from an instance of
2S8 . ‘The’ Chapter VII

T 416 , along with an instance of T 310 , gives a necessary and sufficient

condition for the truth of an identity sentence containing a descriptive
term.
T 4 16 G {lx F {x )) W y [A x (F {x ) < r-,x = y ) A G {y )] v
[^yyA:v(F(:v) x = y ) a G ( 1 zü w = w )\

T 4 17 z = 1aî:F(;v) <-> Ajc(F(;c) x = z) v

[<-^YyA;x:(F(;x;) <-> x = y) a z Iw w = w]

Each thing is the thing identical with it.

T 4 18 Ixx = y = y

The self-identical thing and the non-self-identical things are the same.
T 4 19 lw w = w = l w ^ w = ^ w
T420 indicates the meaning which is to be assigned to vacuous de­
scriptive terms.
T420 IxF = Iw w = w

Thus we conclude that

lx Socrates is bald
designates o.

EXERC ISES

22. Prove T 4 11, T 4 13, and T 414.

23. Prove T 416 and T 417.
24. Prove T 4 18 - T420.
25. Show that the following two formulas are equivalent:
(a) ^ (Vy[A^(F(;c) <-> ^ = y ) A G ( y ) ] v [^YyA:»(F(:v) x — y) a
G (lw '^ to = w)])
(b) Vy[A^(F(A:) <-> jc = y) A ^G (y)] V ['^VyAA:(F(:v) x = y) a
G(lw^zü = w)]
26. Show that the following two formulas are equivalent:
(a) A^(Yy[A^(F(jc) X = y ) A G(y2r)] v [/^VyA^(F(^) <-> x = y) a
G(\w'>'ZO = w z)])
(b) Vy[Aic(F(:v) x = y ) A A^G(y2r)] v [-^VyA:v(F(:v) x = y) A
h z G Ç iw ^ w = V) 2r)]

(Exercises 25 and 26 will be of interest to those readers who are

familiar with RusselPs theory of descriptions, as elaborated in White-
head and Russell [i].)

9. A rgum ents. Consider again argument (i) of page 233. On the

basis of the scheme of abbreviation
Section 9 Arguments . 259

is a student
: a failed
is illogical
A® : Alfred ,
(i) has the symbolization
VvAx [F(a:) a G(;c) X = y\ . A = l;x:[F(.x:) a G( x)]
A x [G {x )^ U {x )] /. H(A) ,
which, as the reader who solves exercise 28 will verify, is valid.

EXERCISES

Show that the following arguments are valid. Exercise 27 is dis­

cussed for illustration.

27. Some problem on Examination I is harder than any other

problem on that examination. Students who solve every problem
on Examination II also solve every problem on Examination I. Alfred,
a student, does not solve the hardest problem on Examination I.
For each x and y, if x is harder than y, then y is not harder than x.
There is a problem on Examination II that not every student
solves.

The last premise of this argument is essential to its validity. It

asserts explicitly the asymmetry of the relation harder thariy implicit
in the ordinary use of a comparative such as ‘harder’. Many intuitively
plausible arguments containing relation words are invalid; to make
them valid one must add premises expressing obvious properties of
the relations involved. For example, compare the symbolization of
exercise 27 with exercise 38 (section 10), and the symbolization of
exercise 32 below with exercise 39.
To derive the conclusion from the premises of No. 27 (or more
precisely a symbolization of No. 27), derive first, using the fourth
premise,
(1) At most one problem on Examination I is harder than
any other problem on that examination.
The first premise and (i), together with T318 and PD, lead to the
assertion
(2) The hardest problem on Examination I is a problem on
Examination I;
and with (2) at one’s disposal the derivation of the conclusion of the
argument is straightforward.
28. Argument (i) of page 233.
29. Tully is the man who denounced Catiline. Cicero is a man.
26o . ‘The’ Chapter VII

Cicero preferred Pompey to Caesar. Everyone who preferred Pompey

to Caesar denounced Catiline. There is exactly one man who denounced
Catiline. Tully and Cicero are identical.
30. There is exactly one man whom Mary loves. There is exactly
one man who loves Alice. The man whom Mary loves is the man whom
Alice loves. Alfred is the man whom Alice loves. Mary loves the man
who loves Alice. .*. The man who loves Alice is Alfred.
31. 2 is the even prime. The positive square root of 4 is an even
prime. There is exactly one even prime. 2 equals the positive
square root of 4.
32. There is exactly one sophist who teaches without remunera­
tion, and he is Socrates. Socrates argues better than any other sophist.
Plato argues better than some sophist who teaches without remunera­
tion. For each x and jy, if x argues better than y, then y does not argue
better than x , .*. Plato is not a sophist.
33. Any fish is faster than any smaller one. For each x and y,
if X is faster than y , then y is not faster than x . For each x and y, if
either x is smaller than 3; or jc is the same size as jy, then x is not larger
than y . For each x and jy, if x is neither smaller than y nor the same
size as jy, then y is smaller than x , .*. If there is a largest fish, then
the largest fish is the fastest fish.

10. * Invalidity. The notions of an English translation of a symbolic

argument and a closure axiom are to be characterized as before (chapter IV,
section 10), but the notions of a relativization of a sentence and an argu­
ment must be modified as follows.
If is a symbolic sentence and tt a i-place predicate letter not occurring
in <f)y then by the relativization of <f> to tt we now understand that sentence
which is obtained from <f> by replacing each part of the form
Aa0 ,
Va^ ,
or
la 0 ,
where a is a variable and i/j a symbolic formula, by
Aa(Tra 0) ,
Va(7ra A ifj) ,
or
1 a (Tra A 0 )
respectively. For example, the relativization of the formula
Ax:F[lyG(:vy)]
to the predicate is
Section lo Invalidity . 261

Ax[H{x) -> F(1j[H(3;) a G(^j )])] .

Let A be a symbolic argument whose premises and conclusion are

sentences, let tt be a i-place predicate occurring in none of these, and
let Si, . . . , 87^ be all the operation letters occurring in A. Then a relativi-
zation of A to 7Tis an argument obtained from A by replacing its premises
and conclusion by their relativizations to tt and adjoining, as additional
premises, the closure axioms for Si, . . . , S77 with respect to tt, together
with the sentence
lx X = X

Consider, for example, the argument

(1) Ax[F{x) G(^)] G{lx[F{x) a H(A[;v])]) .

A relativization of (i) to is the argument

(2) AxQix) ^ [F(^) -> G(^)]) . AxU(x) -> ]{A[x])] .

]{lx ^ x = x) G{lx[]{x) A F{x) A H(A[:c])]) .

If Ay B are symbolic arguments whose premises and conclusion are

sentences, then A is said to be a relativization of B if there is a i-place
predicate letter tt not occurring in B such that A is 3. relativization of
B t o TT.
A false translation of a closure of a symbolic argument is still sufficient
to establish invalidity; but as in the case of the identity calculus there
are many invalid symbolic arguments to which this approach will not
apply.
It happens, however, that a symbolic argument is invalid in the de­
scription calculus just in case a relativization of one of its closures has a
false English translation. (This assertion would not hold in the absence
of the rule ID.) In particular, it follows that a symbolic formula <f> is not
a theorem just in case a relativization of a closure of the argument having
<f> as conclusion and no premises has a false translation.
The invalidity of (i) is established by the following false translation
of (2):
(3) Every integer greater than 2 is greater than i. The successor
of an integer is an integer. The object x such that x differs
from X is an integer. The object x such that x is an
integer greater than 2 whose successor is an integer is
greater than i.

To verify that (3) is a false English argument it is sufficient to recognize

that both definite descriptions occurring in it are improper and to recall
that improper definite descriptions designate the number o.
26a . ‘The’ Chapter VII

EXERC ISES
Show of each of the following arguments that it is valid or that it is
invalid.
34. VyA:v[F(jc) <-> jc = jv] . l;cF(jc) = 1:x;G(;x:) .
Ax[G{x) -> H{x)] HflA:F(jc)]
35. 1;cF(:x;) = lzo zv = zo 1;cF(;c) = l;cG(ic)
yy[Ax(F{x) <^x = y) A G(j)]
36. F(lzv '^zv = z v ) w ^ A = lzO'^zo = zv A = 1 a;F(jc)
AAi(F(;c) <r->X = A)
37. /. A(a?) = B(oi) lx A(;c) = y = lx B(jc) = y
38. Vx(F{x) A Ajv[F(j) A y = X G(jc3;)]) .
A^(H(^) A Ay[J{y) -> K{xy)] -> Ay[F{y) -> K(^;;)]) .
H(A) A K[A l:v(F(:v) a Ajy[F(3/) a ^ j = ^->G(^3^)])]
Vjc(J(^) A A3^[H(jc)-> K(j;c)]) (Compare exercise 27, p. 259.)
39. Vy[A:v(F(j»i) A G{x) <r-^ x = y) a y = A] .
Ajc[F (; c) a ^ x = A - > H ( A jc)] . y x [¥ {x ) a G (: v) a H (B a;)]
^F(B) (Compare exercise 32, p. 260.)
40. WxF{xA) . G(B) . -H [B IxF(jcA)]
V:v(F(:vA) a V2[G(2) a /^H(2r;c)])
41. H(A) A A = lx '^Vjv[H(j) A G(yx)] .
A:v(^F(:v) a H(x) ->Vy[G{yx) a U{y)]) F(A)

On the basis of the scheme of abbreviation

AO : God
F^ : a exists
G2 : a is greater than b
: a can be conceived
exercise 41 has the following translation:
God can be conceived, and is that than which nothing
greater can be conceived. If a nonexistent object can be
conceived, then something greater can also be conceived.
Therefore God exists.
(This English argument is, essentially, St. Anselm’s ontological
argument; compare page 4.)

Show that each of the following inference rules, if incorporated into

our system of logic, would permit the validation of a symbolic argu­
ment that, by methods now available, can be shown invalid. (The
restrictions on variables involved in PD and ID are thus seen to be
necessary.) In all cases, a, j8, y are to be variables, and <f>, tfs symbolic
formulas. (Exercise 42 is solved for illustration.)
Section 10 Invalidity . 263

42. Vj8Aa(^ i-^oL == P)

where ip comes from (p by proper substitution of

for a (but j3 is permitted to be free in <p).

Argument:
Ax[F{x) A A(jc) = B ^ jc = B] /. Wx[¥{x) a A { x ) = C]
Derivation:
I. a A(a;) = C]
2. VyA^[F(jc) A A(^) = y<r^ X = y] Premise, EG
3- ¿^A ^A 3'[F(1^[F(a;) a A(:v) = y]) a
A(l^[F(ic) A A(;v) = y]) = y]
F( 1 ^[F(jc) A A(^) = y]) A
A(1:v[F(^) A A(:v)= y]) = y 2, the variant
of PD above
s- F(1jc[F(;c) a A { x ) = C]) A
A(1^[F(^) A A(^) = C]) = C 3, UI
6. Wx[F(x) A A(^) = C] s, eg
Translation:
Zero is the only integer whose square is zero. Therefore
3 is the square of some integer.
Clearly the translation is a false argument, and hence the symbolic
argument is invalid.

43. ^ypAoi{(p ^ a = j8)

la<^ = ly/x/ y = y
(Here j3 is permitted to be free in <p,)
44. VjSAa(^ ^ a =

where jS is not free in <^, and ip is like <pexcept for having free occurrences
of
loLip
at some (but not all) places where <p has free occurrences of a.
45. V^Aa(^ a = j8)
264 . ‘The* Chapter VII

where is not free in <f), and i/j is like (f>except for having occurrences of

(free or bound) wherever <f> has free occurrences of a.

46. Find a symbolic argument that, by methods now available, can
be shown invalid but that could be validated if we were to incorporate
into our present system of logic Euclid’s law as formulated on page
224.

1 1 . * H istorical rem arks. The earliest theories of descriptions are

those of Frege (summarized in Carnap [2]) and Russell [1]. In both
treatments, descriptive terms were regarded as defined expressions. Frege’s
theory has the disadvantage of requiring, in any language to which it is
applied, the presence of at least one name that is not a descriptive term.
Russell’s theory has more serious drawbacks. In the first place, the rules
of the quantifier calculus must be significantly curtailed when applied
to formulas containing descriptive terms. In addition, a descriptive term
must always be accompanied by a scope indicator, and formulas differing
only in the scope of their descriptive terms will not always be equivalent.
(See in this context exercises 25 and 26, p. 258.) In Rosser [i] the
descriptive operator, ‘1 ’, is treated as a primitive symbol; Rosser’s
system, however, is not complete in the sense that a symbolic argument
is valid if none of its closures has a relativization with a false English
translation. The present treatment is like Rosser’s in that ‘1’ is taken
as primitive, but in other respects resembles Frege’s theory. (T416
corresponds to Frege’s contextual definition for descriptive terms.)
The completeness of the calculus of this chapter is proved in Montague
and Kalish [1].
The principle of interchange, in its full generality, is justified in Quine
[2].

12. Appendix: summary of the system of logic developed in

chapters I-VII.

IN FEREN CE RULES
(Here a, jS, y are to be variables, C, V> V > ^ symbolic
terms, and <f>', >J), x symbolic formulas.)

PRIMITIVE SENTENTIAL RULES:

<f>
<!> Modus ponens (MP)
Section 12 Appendix: Summary • 265

^ ijj Modus tollens (M T)

(j)
^ (j) <!> Double negation (DN)
~T~ .<P

<!> Repetition (R)

<f>
(f>Alp <p Alp Simplification (S)
<f> >P

Adjunction (Adj)
(f> Al/j

4> <P Addition (Add)

<j>y {jj <p Wip
(py ip <P y ip
(p ip Modus tollendo ponens (MTP)
•A <t>
<P<—> tp <P<—> ip Biconditional-conditional (BC)
(p -> ip ip -> (p
(j, ->ip
<P ->«A Conditional-biconditional (CB)
<P<-> ip
D E R I V E D S E N T E N T I A L R U L E S :

<p y ip
(f> ^tfi <f> -^ x ~^x
^(p Ip ^ -> x ^ ~^X Separation of cases (SC)
0
’'(p ^ ip Conditional-disjunction (CD)
(p y ip
P R I M I T I V E Q U A N T I F I C A T I O N A L R U L E S ;

Aci<f> Universal instantiation (UI)

Existential generalization (EG)

Va<^ ’
266 . ‘The* Chapter VII

where 0 comes from (f> by proper substitution of a term for a;

Va^ Existential instantiation (El)
T ’
where ^ comes from ij> by proper substitution of a variable for a. (See
page 238 for a definition of ‘proper substitution’ .)

DERIVED QUAN TIFICATIO NA L R U LE S:

V a ^<f>
V a ^(f> Aoi(f>
Quantifier negation (QN)
^ Va<^ Aa
hoi ^<f> Va<^

PRIM ITIVE R U LE S OF I D E N T I T Y :

Identity-i (Id-i)
A a(a — Ç

A a(a = C ^<f>) Identity-2 (Id-2)

where iff comes from <f> by proper substitution of C for a, and a does not
occur in C-

DERIVED R U L E S OF I D E N T I T Y :

Symmetry (Sm)
V = C

7] — 6 0 = 7]
T^e 1^ 0
Transitivity (T)
V= t V= i
rj = 6 0 = 7)
T^0
Leibniz’ law (L L )

where is like <f>except for having one or more free occurrences of C where
<f> has free occurrences of
C = i' Euclid’s law (EL)
7) = 7)
Section 12 Appendix: Summary • 267

where r;' is like t] except for having one or more free occurrences of
where rj has free occurrences of C-

P R I M I T I V E R U L E S OF D E S C R I P T I O N :

ViSAa(<^^a = j8)
Proper descriptions (PD)

where 0 comes from <f> by proper substitution of

la ^
for a, and j8 is not free in <f>,
-V j3A a ((^ ^ a = j8)
Improper descriptions (ID)
= ly ^ 7 = y
where jS is not free in <f>.

INTERCH ANGE
One formula follows from another by interchange (Int) if and only if
the two formulas are respectively the conclusion and the premise of one of
the following four patterns of inference (in all cases ly, 7^' are to be
symbolic terms, and <f>y < f)\ 0' symbolic formulas):
A a i . , . Aan ^ = V

where <f>' is like (f) except for having one or more occurrences of where <f)

has occurrences of and ai, . . . , are all the variables bound in <f>;
A a i . . . A(Xn i = V

V= V
where t]' is like t] except for having one or more occurrences of where rj
has occurrences of and ai, . . . , are all the variables bound in t/;
A a i . . . AoLni^ ^')

where <j> is like <f) except for having one or more occurrences of i f j ' where <f)

has occurrences of 0, and ai, . . . , are all the variables bound in

A a i . . . 0Ln{^ 0')
V= V
where rj' is like r] except for having one or more occurrences of 0' where rj
has occurrences of 0, and ai, . . . , a^ are all the variables bound in 77.
268 . ‘The’ Chapter VII

D IR E C T IO N S FOR C O N S T R U C T IN G A
D E R I V A T I O N F R O M A C L A S S K OF
SYM BO LIC FO RM ULAS
(1) If is any symbolic formula, then
Show <f>
may occur as a line. (Annotation: ‘Assertion’.)
(2) Any member of K may occur as a line. (Annotation: ‘Premise’.)
(3) If (f>y i/j are symbolic formulas such that
Show {(f) -> ip)
occurs as a line, then <j>may occur as the next line. (Annotation: ‘Assump­
tion’.)
(4) If ^ is a symbolic formula such that
Show (p
occurs as a line, then
^ (P
may occur as the next line; if is a symbolic formula such that
Show ^ (p
occurs as a line, then (p may occur as the next line. (Annotation: ‘Assump­
tion’.)
(5a) A symbolic formula may occur as a line if it follows from anteced­
ent lines (see p. 21) by a primitive inference rule other than E l.
(5b) A symbolic formula may occur as a line if it follows from an
antecedent line by the inference rule E l, provided that the variable of
instantiation (see p. 100) does not occur in any preceding line. (The
annotation for (5a) and (5b) should refer to the inference rule employed
and the numbers of the antecedent lines involved.)
(6) When the following arrangement of lines has appeared:
Show <p
XI

Xm y

where none of xi through xm contains uncancelled ^Show' and either

(i) (p occurs unboxed among xi through xmy
(ii) (p is of the form
ih h)
and ip2 occurs unboxed among xi through xm]
Section 12 Appendix: Summary . 269

(iii) for some formula x> both x its negation occur unboxed
among XI through xm\ or
(iv) <f> is of the form

A a i . . . Aajfc i/s ,

ijj occurs unboxed among the lines xi through xm, and the
variables a i through a * are not free in lines antecedent to the
displayed occurrence of

Show (f> ,

then one may simultaneously cancel the displayed occurrence of ^Show^

and box all subsequent lines.

The remaining clauses are abbreviatory (in the sense of page 58):
(7) If (f) is an instance of an alphabetic variant of a theorem that has
already been proved, then <f>may occur as a line. (Annotation: the number
of the theorem in question, sometimes together with a diagrammatic
indication of the sequence of substitutions involved.) (For the notion of
instance see chapter V II, pp. 249 - 50.)
(8) A symbolic formula may occur as a line if it is an alphabetic variant
of an antecedent line. (Annotation: ‘AV* and the number of the antecedent
line.) (For the notion of alphabetic variance see chapter V II, pp. 248 —49.)
(9) If <f>y <f>' are symbolic formulas such that

is an instance of an alphabetic variant of a previously proved theorem,

and 0, 0' are symbolic formulas such that is like i/j except for having
one or more occurrences of (f>' where ^ has occurrences of then

may occur as a line. (Annotation: ‘IE ' (‘interchiange of equivalents')

together with a parenthetical reference to the theorem involved.)
(10) A symbolic formula may occur as a line if it follows from an ante­
cedent line by interchange. (Annotation: ‘Int' and the number of the
antecedent line.)
(11) A symbolic formula may occur as a line if it follows from ante­
cedent lines by a succession of steps, and each intermediate step can be
justified by one of clauses (2), (5a), (7), (8), (9), or (10). (The annotation
should determine the omitted succession of steps by indicating, in order
of application, the antecedent lines, the premises, the inference rules,
and the previously proved theorems employed.) (The use of derived
inference rules is comprehended under this clause.)
*70 . ‘The* Chapter VII

D ERIVABILITY
A derivation is complete just in case every line either is boxed or con­
tains cancelled *Show\ A symbolic formula <f> is derivable from a class K
of symbolic formulas just in case one can construct a complete derivation
from K in which
Shemcf)
occurs as an unboxed line.

13. Appendix: list of theorems of chapter VII.

T401 Ax(F{x) X = y) lA;F(:r) = y
T402 WyAx{F{x) ^ x = y) -> F(l:vF(:r))
T 40 3 VyAx{F{x) X = y) IjcF(^) = '\w w = w
T404 T328 -> 1jcF( a;) = VF(j^)
T405 T329 [Ajc[F(jc) G{x)] -> l^F(;c) = l;rG(jc)]
T406 1:rF(^) = '\yF{y)
T407 Ajc[F(;c) G{x)]-> 1jcF(:r) = lx:G(:r)
T408 A;c[F(:r) 4-> = I jcF( jc)] <-> WyAx{F{x) ^ x = y)
T409 WyAx(F{x) <-^x = y) A F(:r) x = l;cF(^)
T 4 10 VjA;c(F(;c) <r-> X = y) A X = l:rF(:)ii:) ^ F{x)
T411 VjyA:)c(F(:c) < -> X = y) A WyAx{G{x)<-> x = y)
[A^[F(jc) ^ G(^)] lxF{x) = lxG{x)]
T 4 12 F(1jcF( jc)) <-> VjA:r(F(jc) <r-> x = y ) y F(la? w == zv)
T 4 13 [n ^ x = l w '^ w = w y F (1«; w = w )] a
X = l;cF(:v) ^ F(:r)
T 4 14 yyAx(F{x) <-> jc = j) [A:v(F(:r) <-> x = y) <-> l^F(^) = y]
T 4 15 ^YyA;c(F(jc) <-> = 3;)A '^WyAx{G{x)<-^ x = J') ->
l^F(;c) = lxG {x)
T 416 G(1jcF( jc)) <-> yj;[A:r(F(^) x = y) a G(y)] v
[^WyAx{F{x) <-> X = y) A G(\w w = «;)]
T 4 17 z = 1jcF( jc)<(-> A;c(F(:r) <-> :v = 2) v
[^WyAx(F{x) < r ^ x = y ) A z = ']zv'^zv = zv]
T 418 lx X = y = y
T 419 Iw w = w = Iw w = w
T420 Ix F = Iw w = w
 Chapter V III
Definitions; formal theories

I . The vocabulary o f form al languages. In previous chapters we

have considered ordinary English supplemented by various symbols.
To avoid excessive complication, we must at this point renounce English
and limit ourselves to purely symbolic languages. It is in connection with
such languages that the notion of a definition can most conveniently be
treated.
In symbolic languages, as in earlier developments, we distinguish
two kinds of meaningful expressions, terms and formulas. Reverting to
our informal characterization of these expressions, we may say that a
term is an expression that becomes a name once its free variables are
replaced by names, and a formula is an expression that becomes a sentence
once its free variables are replaced by names. Before giving a precise
characterization of terms and formulas, we must consider the basic
symbols from which symbolic languages will be constructed. One category
of such symbols is that of variables^ which are as before lower-case Latin
letters with or without numerical subscripts. It is clear that by the allow­
ance for subscripts there are infinitely many variables. We retain the
standard order established earlier, in fact, the order
a , 6, . . . , 5T, ao, ¿0, • • • > ^0, « i , . . .

We may thus speak of the first variable (which is the second variable
(which is ‘i ’), and so on.
Variables are the simplest terms. Formulas, together with more com­
plicated terms, are formed with the aid of constants^ which fall into two
classes, formula-makers and term-makers^ according to the kind of expres­
sion that they generate. Each constant may be used in combination with
a certain number of variables and previously generated terms and formulas
to construct a new term or formula. To make this procedure precise, we
shall associate with every constant a fixed degree^ which will be a quadruple
<f, m, w, /)> of nonnegative integers, in which i is either o or i. Here i
is o or I according as the constant in question is a term-maker or a formula-
maker, and m, w, and p are respectively the number of variables, the
number of terms, and the number of formulas that the constant demands.
272 . Definitions; Formal Theories Chapter VIII

Our terms and formulas can then be exhaustively characterized as

follows.
(j) Every variable is a term.
(2) I f h is a constant of degree <0, my w, /) >, ai, . . . , cum distinct
variableSy .. ., are terms, and <f>iy • . . , <l>p are formulas, then the
expression
S a i . . . cLffi . . . in • • • <t>p
is a term.
(j) I f h is a constant of degree <1, m, n, p y , ai, . . . , oLm are distinct
variables, , in are terms, and <t>i, . * *, <f>p are formulas, then the
expression
S a i . . . a/n i i • • • iw • • • ^p
is a formula.

(Each of m, n, p may take on the value zero, in which case one of the
strings a i . . . oLrrii i i . . . in, or . . . (f>p will disappear. Thus, for example,
\{ m = p = o, the expression

Sai. . . i l . . . in
will become simply
Sii...in .)
The preceding characterization employs the notion of a constant of degree
<i, m, n, p y and, before becoming completely intelligible, would have to
be supplemented by a characterization of this notion. The most appro­
priate way of providing the required characterization is by giving a list
of symbols that are to be regarded as constants, together with the speci­
fication of a degree for each. Such a list, sufficient for the purposes of this
and the following chapter, is to be found in an appendix to chapter IX .
The list has an arbitrary character, stemming from the accidental features
of chapter IX and the later sections of the present chapter. If the develop­
ments there had been more extensive, the list would have been longer.
Indeed, for some purposes it would be convenient to have an infinite list
of constants. (Such a list could be given, here as in the case of variables,
by a general characterization of the structure of the symbols comprised
in it.)
At this point, however, we shall not give a complete list of the constants
that will be used in the formal languages of later sections; for many
constants will be introduced by definition, and it is in connection with
their definitions that their meaning can most conveniently be elucidated.
We wish tcMnclude among our constants all the special symbols introduced
in the preceding chapters: ‘ a ', ‘V, ‘A ’, ‘V ’, ‘ = ^, 'X. These
Section I The Vocabulary of Formal Languages . 273

symbols are called logical constants and are all formula-makers except for
The identity sign *= ' is of degree <1, o, 2, o ) ; that is, it is a formula-
maker that demands two terms to produce a formula. Thus, for example,
x y is a formula. (The reader will note that the order of symbols,
both here and in the case of the sentential connectives other than
differs from that with which he has become familiar.) The negation
sign is of degree <1, o, o, i >, and ‘ a’, ‘ v*, and are of degree
<1, o, o, 2>; that is, the negation sign is a formula-maker requiring one
formula to produce a new formula, and the other sentential connectives
are formula-makers demanding two formulas to produce a new formula.
Thus, for example, = x y ' is 2l formula (the negation of * = xy^) and
‘a = x y = y is a formula (the conjunction oi ' = x y ' and ‘ = y z').
The quantifiers are symbols of degree <1, i, o, i> ; that is, they are
formula-makers that demand one variable and one formula to produce
a new formula. Thus, for example, 'Ax = x x* is a formula. The de­
scriptive operator is a symbol of degree <0, 1 , 0 , i > ; that is, it is a
term-maker that demands one variable and one formula to produce a new
term. Thus, for example, '~\x — x x ' is a term. Constants other than
‘ a’, W , 'A\ ' = \ and ‘1’ are called nonlogical
We wish also to include among our constants the operation letters
and predicate letters of previous chapters, that is, the symbols
AO, . . . , EO, Ai, . . . , E l , . . . , FO, . . . , Zo, Fi, . . . , Z i , . . . ,
together with their subscripted variants. Each w-place operation letter
is of degree <0, o, «, o ), and each w-place predicate letter of degree
< I, o, w, o>.

As additional examples of nonlogical constants we list the following:

-f- (degree: <0, o, 2, 0 »
^ (degree: <0, o, i, 0 »
e (degree: <1, o, 2, 0 »
I (degree: <1, o, i, 0 »
V (degree: <1, i, o, 1 »
E (degree: <0, i, o, i> )
lim (degree: <0, i, i, 0 »
‘ + * and are to be understood in their familiar mathematical senses,
as signs for addition and cube root, and
€Xy ,
IX ,
^x Fi^ ,
Ex E^x ,
lim n A^n
are to be read respectively
274 • Definitions; Formal Theories Chapter VIII

jc is a member of the set y ,

X is an integer ,
there is exactly one x such that F^x ,
the set of objects x such that F^x ,
the limit of as n approaches infinity .
It will be convenient to provide a terminology for certain kinds of
constants. Those constants which must be followed immediately by one or
more variables (that is, which have degree , w, w, />> with m > o) will be
called operators; these variables will turn out, in the light of section 2, to
be bound. Among the logical constants, ‘A’, ‘V’, and ‘T are the only opera­
tors; and *lim* are also operators. The identity sign is not an
operator; for even though it may be followed immediately by variables,
as in ‘ = X y \ these variables will be free. Further, it is not required that
the identity sign be followed immediately by variables, as the formula
‘ = l x = x x y ' indicates. Operators will sometimes for emphasis be
called variable-binding operators.
A constant of degree <1, o, w, o> is called an n-place predicate; that is,
an w-place predicate is a formula-maker requiring exactly n terms to
produce a formula. The identity symbol, then, is a 2-place predicate;
predicate letters are predicates; 'e and T* are predicates. A constant of
degree <0, o, w, o > is called an n-place operation symbol; that is, an n-place
operation symbol is a term-maker requiring exactly n terms to produce
a new term. Operation letters are operation symbols, and so are ‘ ^ and
Finally, we shall call a o-place operation symbol an individual constant^
and a constant of degree <1, o, o, />> with^ > o a sentential connective.
z. Bondage and freedom; proper substitution; alphabetic vari­
ance. Variables may now be bound not only by quantifiers and ‘T but
by arbitrary operators. For example, if S is an operator of degree <0, i, o,
I >, then each occurrence of the variable a in the term

S a = a j8
is bound. Thus we must give a more general characterization of bondage
and freedom than the earlier treatment provides.
An occurrence of a variable a is said to be bound in a term or formula <f>
just in case it stands within an occurrence in (f) of some expression
Sai. . , . . . l^n <I>1 ••• <l>p ,
where 8 is a constant of degree <0, w, n, /)> or <1, w, w, />>, a i , . . . ,
are distinct variables, ^i, . . . , are terms, <f>h . . . y <f>p are formulas,
and a is one of a i , . . . , ctim» An occurrence of a variable is free in <f> just
in case it stands wdthin ^ but is not bound in <f>. A variable is bound or
free in <f> according as it has a bound or free occurrence in (f>. Thus, in
^x F^xy ,
Section 2 Bondage and Freedom; Proper Substitution; Alphabetic Variance . 275

both occurrences of 'x' are bound, the only occurrence of *y' is free, the
variable 'x* is bound but not free, and the variable y is free but not
bound.
We now consider freedom and bondage of arbitrary terms. If is a
formula or term, then an occurrence of a term ^ is bound in (f>just in case
it stands within an occurrence in (f> of some expression

8(x.i, . . (x,m i i • • * in <ki • • • y

where 8 is a constant of degree <0, my n^ p } or <1, m, w, /)>, ai, . . . ,
are distinct variables, ii, . . ., Cn are terms, (f>iy •. . y <f>p are formulas, and
at least one of the variables ai, . . . y oLm is free in An occurrence of a
term is said to be free in (f> if it stands within <j> but is not bound in <f>,
A term is bound or free in (f> according as it has a bound or free occurrence
in (f>. For example, in
(i) Ax EyG^xy y
the occurrence of *FyG^xy' is bound, but in

(2) A xF^EyG ^zy y

the occurrence of 'EyG^zy* is free. Observe that the characterization of

bondage and freedom for the special case of variables and their occur­
rences is subsumed under the more general characterization just given.
Thus in (i) and (2) both V and are bound, and 'z' is free in (2), no
matter whether we apply the definitions pertaining only to variables or
the more general definitions pertaining to arbitrary terms.
A sentence is as before a formula in which no variable is free, and a
name is a term in which no variable is free.
Proper substitution will be useful in this chapter, and in chapter IX will
acquire additional importance in connection with definitions. The follow­
ing characterization of this notion is essentially identical with the earlier
treatment.
If a is a variable, i a term, and i/j formulas, then ip is said to come from
(f>by proper substitution of i for a just in case ip is like <p except for containing
free occurrences of i wherever <p contains free occurrences of a. For
example, if (p is the formula
€xy
then
Ax € X EzG^^wz
comes from <p by proper substitution of 'EzG^wz^ for
If 8 is an «-place operation symbol or predicate, x is correspondingly
a term or a formula, and <py tp are formulas, then ip is said to come from
(p hy proper substitution of x for 8 just in case \p can be obtained from <p by
276 . Definitions; Formal Theories Chapter VIII

(I) replacing, throughout the constant S by

{x} .
and,
(II) in the expression resulting from (I), successively replacing each
part of the form

where ^i, . . . , In are terms, by the expression obtained from x

by replacing all free occurrences of 'a' by ^1, 'V by ^2, etc., up to
the wth variable, whose free occurrences are to be replaced by ^n\
we require in addition that x (f) have no variables in common. For
example, consider the formula
(3) Ax G^xy .
The formula
Ax = E z = y z x
comes from (3) by proper substitution of
(4) = Ez=bza
for the 2-place predicate In performing the substitution we obtain in
the first step
Ajc { = E z = b z a } xy ,
and in the second step
Ax = E z = y z X
we observe, moreover, that (3) and (4) have no variables in common.
If is a class consisting of variables, operation symbols, and predicates,
then a formula ifj is said to come from a formula <f>by iterated proper substi­
tution on members of K just in case iJj can be obtained from (f> by zero or
more applications of proper substitution, each of which is either the
substitution of (i) a term for a variable belonging to Ky (2) a term for
an operation symbol belonging to Ky or (3) a formula for a predicate
belonging to K . A formula ifj is an instance of a formula (f> if and only if tp
comes from (f> by iterated proper substitution on members of some class
consisting of variables, operation symbols, and predicates.
The notion of alphabetic variance must also be extended. Let 8 be an
operator of degree <0, my ny p y or <1, w, w, /)>, let i ^ m, and let ip
be expressions of the respective forms
S a i . . . ctiffi <pi • • • <pn i>n+i • • • ^n+p y
8 ^ 1. . . ^ni ^»+1 • • • ^n+p y
Section 3 Informal Notational Conventions • 277

where ai, . . . , oLm are distinct variables, Pm are also distinct

variables which, except for pu are identical with ai, . . . , respectively,
<f>h • • • i i>ny ^y^n are terms, and <f>n+h • • • , ^n+py • • • >i^n+p are
formulas. (Thus <f>, ip are both either terms or formulas.) We say that
ijj are immediate alphabetic variants if, for each j ^ n +/>, ipj comes from <f>j
by proper substitution of Pi for and, conversely, (f>j comes from ipj by
proper substitution of for Pi.
This definition is the natural extension to arbitrary operators of the
notion of immediate alphabetic variance considered in chapters IV, VI,
and V II.
More generally, a term or formula tp is said to be an alphabetic variant
of a term or formula <p if there is a finite sequence of expressions beginning
with <p and ending with 0, and such that each expression of the sequence
(except the first) is obtained from its predecessor by replacing an occur­
rence of a term or formula by an immediate alphabetic variant of that term
or formula.
For example,
= ExF^xlyAxG^xy
is an alphabetic variant of
= EyY^y'^xhzG^zx
by virtue of the sequence
= EyE^y’\xhzG'^zx ,
= ,
= ExE^x'\yAzG^zy ,
= ExE^x^yAxG^xy .

3. Informal notational conventions. The official characterization in

section I of term ^nd formula introduces a notation that, though highly
convenient for the formulation of the general definitions of the last two
sections, leads to virtual unreadability when we consider specific formulas
of any length. This remark applies even to the formula (4) of the previous
section, which according to a more customary notational style would
assume the following perspicuous form:
Ez[b = z] = a .
A more compelling example can be provided by considering the associative
law for the addition of integers. According to the style of section i this
law has the following awkward appearance:
AxAyAz -> A A Ix ly lz = ’i-~\‘ X y z + x + y z ,
rather than the more customary:
AxAyAz [I:v A Ijy A I^r (x + y ) + ^ = + (y + ^)] .
278 . Definitions; Formal Theories Chapter VIII

To simplify the appearance of formulas and to bring them into closer

accord with customary usage, we shall adopt some informal notational
conventions.
Although official notation requires that a constant always stand before
the variables, terms, and formulas to which it applies, we shall often depart
from this order. For example, sentential connectives of degree <1, o, o, 2>
will be written between their attendant formulas, as in chapters I - V II,
and operation symbols and predicates of some mathematical currency will
be transposed to the position that mathematical usage prescribes. Paren­
theses and brackets, unnecessary in official notation, must now be in­
troduced to avoid ambiguity and, along with other symbols of punctua­
tion, will sometimes also be used purely for perspicuity. The number of
parentheses required to avoid ambiguity will be somewhat reduced by
adoption of the conventions given in earlier chapters for their omission.
For instance, and will be regarded as marking a greater break
than ‘ a* and and in an iterated conjunction or disjunction without
parentheses the components are to be understood as associated to the
left.
As in earlier chapters, operation letters and predicate letters will be
relieved of their superscripts provided no ambiguity ensues.

4. Derivability. We must now specify the circumstances under which a

formula will be considered derivable from a class of formulas. The charac­
terization will follow the lines of the earlier development. Indeed, clauses
( i ) - ( i i ) of chapter V II (see pp. 2 6 8 -6 9 ) are taken as the directions
for constructing a derivation from a class K of formulas. Of course, we must
now understand ‘symbolic term’ and ‘symbolic formula’ in the sense of
‘term’ and ‘formula’ given in the present chapter. The primitive inference
rules, to which reference is made in clause (5a), are to be just those of chap­
ter V II (see pp. 264 - 67), and interchange (clause (10)) is to be defined
as before (see p. 267); throughout the characterizations, however, ‘sym­
bolic formula’ and ‘symbolic term’ are to be replaced by ‘formula’ and
‘term’ respectively. Alphabetic variance (clauses (7), (8), and (9)) is to be
understood in the sense of the present chapter.
It should be pointed out that clauses (8) and (10) can no longer be
regarded as theoretically dispensable (in the sense of page 58). The
passage from
'^xYx
to
Yy
and the passage from
G^)
Section 5 Formal Theories; the Theory of Commutative Ordered Fields . 279

to
E a!:F;c = ExG x
are justified by AV and Int respectively; but neither of these inferences
can be reduced to applications of clauses (i) - (6). Thus we must regard
the directions for constructing an unabbreviated derivation as consisting
of clauses ( i) - ( 6 ) , (8), and (10). The remaining clauses will then be
theoretically dispensable. (Clause (9), though not reducible to ( i) - ( 6 ) ,
can be eliminated once (10) is available.) In practice, however, we shall
draw no distinction between an abbreviated and an unabbreviated deri­
vation.
As before, a derivation is said to be complete if every line either is
boxed or contains cancelled 'Show\ and a formula (f> is derivable from a
class K of formulas if one can construct a complete derivation from K
in which
Shew-<f>
occurs as an unboxed line.

5. Formal theories; the theory of commutative ordered fields. A

theory consists of two things: (i) a class L of constants, which may be
any class of nonlogical constants, and (2) a class of axioms, which may be
any class of formulas containing no nonlogical constants beyond those in L .
As an example, we consider the theory of commutative ordered fields,
which has the following constants:
+ (2-place operation symbol)
— (I-place operation symbol)
0 (individual constant)
(2-place operation symbol)
(I-place operation symbol)
1 (individual constant)
^ (2-place predicate)
(In characterizing a theory, it is unnecessary to specify any interpre­
tation. To facilitate comprehension, however, a few remarks concerning
what may be regarded as the standard interpretation of this theory will
be useful. The domain of discourse, or the class of objects to which
the variables of the theory are construed as referring, is to consist either of
the rational numbers (that is, numbers expressible as the quotient of two
integers) or of the real numbers (that is, the numbers ordinarily considered
in elementary algebra; in particular, beyond the rationals, such numbers
as 7T and ^/г are included, but complex numbers, built up with the aid
of are not); and the constants listed above are to have their usual
mathematical meaning. The symbol ‘ \ which is often used in two ways,
28o . Definitions; Formal Theories Chapter VIII

as both a i-place operation symbol (as in ‘ —2’) and a 2-place operation

symbol (as in ‘3 — 2’), is here restricted to the first usage; thus we read
' —x' as ‘the negative of x\ The symbol occurs as a superscript, as
in which is read ‘the reciprocal of x\ and designates the number^.
The term to which no meaning is ordinarily assigned, may be
interpreted as designating any fixed number; the axioms below place
no restriction on the choice of that number.)
The axioms of the theory of commutative ordered fields are the following
formulas, A i - A 1 5 :
Ai X + {y + z) = (x + y) z
Az X + y = y + X

A3 X + 0 = X

A4 X + — X = 0

As X •( y •z) = (x •y ) •Z
A6 X •y = y •X

A? X •1 = X

A8 X = 0 -> X •xr^ = I

Ag X •{y + z) = {x •y) + [x •z)

A io /^0=1

A ll 0 ^ X y0 ^ — X

A iz
A 13 O ^ X A O ^ y ~ > O ^ X + y

A14. 0 < X A0 ^ y ^ X •y
A 15 X ^ y <r->0 ^ y + — X

Now if T is an arbitrary theory, then a term^ formula^ or sentence of T

is respectively a term, formula, or sentence that contains no nonlogical
constants beyond the constants of T, and a theorem of T \s 2. formula of T
that is derivable from the axioms of T, (In clause (7) of the directions for
constructing a derivation, which permits the use of previously proved
theorems, a theorem is, as always, to be understood as a theorem of logic—
that is, a formula derivable from the empty class of premises—and not as
a theorem of a theory. A limited use of previously proved theorems of a
theory will be introduced below, after T i.)
For example, let us return to the theory of commutative ordered fields
and derive a few theorems. Theorems i and 2 are cancellation laws for ad­
dition, and 3 and 4 cancellation laws for multiplication, (Theorems of
Section 5 Formal Theories; the Theory of Commutative Ordered Fields . 281

the present theory and its extensions will be numbered T i, Tz, and so on;
we italicize ‘1 ” here to prevent confusion with the numbering system of
the theorems of chapters I - VII.)

Ti 1 Skew-x + z = y + z ^ x = y
2

3 Ai
4
5 ^4
6
7 ^3
8
9
10 9, E L
11 2, U I, UI, UI,
2, U I, UI, UI,
10, T , T
12. 4, U I, I I , L L
13- 6, UI, 6, UI,
12, T , T
The proof of T i is made unduly long by the need for subsidiary deri­
vations of closures of axioms. For example, line 2 is required to obtain
(i) (:v + ^) + - z = X {z -]---- z) ,
which is in turn required to infer line 11 from line 10. Now (i) comes from
-^i by iterated substitution on variables but is not itself an axiom. We
shall in future dispense with such subsidiary derivations as those beginning
with lines 2, 4, 6, and, when showing a formula to be a theorem of a theory
T, permit as a line of a derivation any formula that comes by iterated
substitution on variables from an alphabetic variant of an axiom of T or
a previously proved theorem of T. For annotation we shall simply refer
to the axiom or theorem involved.
Using this informal abbreviation, we may simplify as follows the
derivation of T i :

I. ^Skew-x -\-z = y - { - z ^ x = : y
2. X z = y + z
3- {x z) -\---- = (j; + H-------- z 2, EL
4- X + {z -----z) = 3; + (;sr H----- z) A i , A i ,2 , T , t
X + 0 = y + 0 ^ 4 .4 .1 'L
6. X = y ^ 3 .^ 3 , 5, T , T
282 . Definitions; Formal Theories Chapter VIII

Tz

T3 '^Z = O A X * Z = y ^ Z - > X = y
T4 ^z = o z • X = z •y X = y

TS 1. Show —o = o
O + —0 = 0 ^4
0 + 0 = 0 ^3
— 0 = 0 2 , 3 . T , T2, M P
T6 Shew-x *0 = 0
(a: • l) + (ji • o) = ;!ii: • (l + o) Ag, Sym
( ii • l ) + (jc • o) = :» • I ^ 3 , 2, L L
(a; • i) + o = • I A3
X •o = 0 3 > 4 > T , Tz, M P
Tl X + —y = O ir^ X = y
T8 X = —y y = —X
T9 X = O <r-> —X = o

The next group of theorems, T io - J'14 , are laws of signs.

T io ------- X = X

T il I. Shot»- —(ji + y) = —X + —y
2. {x + y) + ( + -y) =
(a: 4- - ^) + O' + -y) A i, Az, L L
3- (x + y) + ( -» + —y) = 0 A4, 2 , L L , A^, T
4- (x + 3') + -{x + 3^) = 0 ^4
5- —(* + :v) = (-.V + -y) 4. 3. T , Tz, M P
(To obtain line 2 above, several applications of ^ i , Az, and L L are
required. But here and henceforth we shall for the most part omit repeti­
tions in annotations.)
T iz ( - x)-y = - 0 -y)
T 13 X • i-y ) = - { x -y)
T i4 T
X• fDTlXJXXr (1 ■ x ) - ( - y ) = X *y
2. ■ (-y ) = - O ' - O ') ) T iz
3- = - ~ ( x 'y ) 2, T i3 > L L
4- = X 3 . T io , T

(In the derivation above, the left side of line 2 is imagined to be repeated
in the blanks of lines 3 and 4.)
Section 5 Formal Theories; the Theory of Commutative Ordered Fields . 283

T15 1. Show ^ X = o — )-i = - ( ^ 1 )

2. ^ X= o
X • = I 2, ^8, M P
(-* ) • -(^ ^ ) = I 3, T 14, T
—X = o 2, Tg, BC, M T
{-x)-{-x)~ ^ = I 5, A8, M P
i-x )-i = -(^ 1) 4, 6, T , 5, Adj, T4,
MP
We state next two basic principles of multiplication.
T16 X *y = o ir^ x = ovy = o
T iy X - y = o -^ {x *y)^^ = x-^ • y-^
We turn now to theorems on order. T 18 - 7"2i assert that ^ is respectively
reflexive^ antisymmetric^ transitive^ and connected^ and hence is a simple
ordering,
T18 Shew-x ^ X

Show o < o
^ o < o
o < —o 3>
0^0 4> ^ 5>
o < ^ + —X 2, LL
X ^ X 6, A i$
In the annotation for lines 4 and 7 above, reference to theorems and
inference rules of the sentential calculus has been omitted; this practice
will be adopted henceforth.
T19 Shew-x ^ y ^ y ^ x ^ x = y
X ^ y t<y ^ X
0 ^ 3 ^ + —x ^ o ^ x + —y 2, A i s
o < —(jt + —y) A O < ; 3, Tio, T i l , A2,
LL
X + —y = o 4, ^ 1 2
X = y 5. T7
T20 Shew-x 4: y t ^ y ^ z - ^ x < i Z
X ^ y /^y 4: z
O ^ y + —X A O < 2T + —y 2, A i s
o < (y H---- x) + (2' H-----y) 3. ^ 1 3
O ^ Z + —X 4, A i, AZy A3, A4,
LL
X ^ z 5. ^ 1 5
284 • Definitions; Formal Theories Chapter VIII

T 21 I. Shew X < j vy ^ X

2. о ^ jy + —л: V о < —{у Н----д;) A ll

3- 0 ^ 3 ^ + —ХУ о ^ л; + —у 2, T io , T i i , A2,
LL
4. X ^ у у у ^ X 3. ^ 1 5
T22 asserts that squares are always nonnegative; hence (T23) i is non­
negative.
T22 Show-о ^ X • X
2. Show 0 ^ X 0 ^ X •X
3- 0 ^ д:
4- о ^ л; • л; 3 , A 14.

5- Show 0 ^ — X ->o ^ X •X
6. 0 < —X
7- о < (-л ) • {-х ) 6, ^ 1 4
8. о ^ д; • д: 7, LL
9- (0 ^ X •X 2 ,S ,A i i

Г23 0 ^ 1
The remaining theorems of this section are familiar laws of inequality.

Г24 I. S k ew x ^ у -> x -h z ^ у + z
2. X ^ у
3- 0 < + —X 2, A is
4- ( j + г) + -(д; + 2г)
= (у + 2г) 4- ( - л Н---- z) T 11, E L
5- = у -i---- д: 4, А2, А з , А4,
LL
6. о < (у + 2:) + -(д: + z) 3> 5> L L
7- X + Z ^ у + Z 6, A is

Т25 x + z ^ y + z - ^ x ^ y

Т26 x ^ y ^ z ^ w - > д: + ^ ^

Г27 у + 1 ^ у
Г28 у ^ у + - I

Т29 X ^ у —у ^ — д:
Т30 —у ^ — д: -> д: ^

Т 31 X ^ о<->о < — д:
Section 6 Extensions of Theories; the Theory of Real Numbers . 285

TS2

^33

2, A i $
3. A 1 4
4, A6, Ag, T13,
A6, L L
5. A 1 5

T34

2, T31
2, 3 . ^33
4> ^*3> L L
5. ^30
T 3S

4, A n
2, 5, A 1 4
6, T 13, L L
2f LL
T23, A io , A j 2

T36

2, T35, T33
3, A s, 2, A8, A7,
LL

T37 y * Z ^ X ^ Z A z ^ O A ^ Z = 0 -> X ^ y

T38 X = y ^ X ^ y

6. Extensions o f theories; the theory o f re a l num bers. If T is a

theory and an arbitrary formula (whether of T or not), then by an
instance of <f> within T is understood any formula of T that comes from <f>
by iterated proper substitution on some class of symbols that contains
no constant of T.
If T and U are theories, then U is said to be an extension of T if all con­
stants of T are constants of U and all axioms of T are axioms of U. For
286 . Definitions; Formal Theories Chapter VIII

example, the theory of commutative ordered fields has as an extension

the theory of real numbers, which can be characterized as follows: the
constants of the theory of real numbers are those of the theory of commutative
ordered fields together with the i-place predicate ‘F, which is read ‘is
an integer’. The axioms of the theory of real numbers are A i - A i^ above,
together with
A i6 I(o) ,

^ 17 I(:v) -> l{x + i) A I(:v H------ 1) ,

^ 18 I ( :r) A 1(3;) A X ^ y A y ^ X + l -^y = xy y = X + I ,

and all instances within the present theory of the formula

A S ig WxF(x) A WyAx(F{x) -^ x ^ y)
V 2 t[ A ; c( F ( jc) x: ^ 2:) a A jy ( A ;r[ F ( A :) ^ ^ jy]

Axioms 16 through 18 assert that o is an integer, that the successor and

the predecessor of an integer are again integers, and that between an
integer and its successor there is no other integer. Note that the formula
called A S ig is not a formula of the theory of real numbers because it
contains the constant ‘F ’ and hence is not itself an axiom of this theory.
It enables us, however, to describe easily an infinity of axioms of the
theory, that is, all its instances within the theory. Thus, in view of A S ig ,
the following formula is an axiom of the theory of real numbers:
(i) V;c a; • ^ ^ I + I AVyAx(x • x ^ i - h i ^ x ^ y ) - >
Vz[Ax(x - x ^ i - h i - ^ x ^ z ) A
Ay(Ax[x • X < I + I X y] z y)]
We may refer to A S ig itself as an axiom schema; this accounts for the
designation ‘^ 5 1 9 ’ .
^ 5 1 9 is often called the Continuity Schema and has the following
intuitive content: every nonempty set of numbers that is bounded above
has a least upper bound. For example, the instance (i) asserts that on
the hypothesis that some number is such that its square is at most 2 and
there is a number greater than all numbers fulfilling this condition,
there is a least upper bound of the set of numbers whose square is at
most 2. (Such a least upper bound indeed exists; it is the square root of 2.)
It should be observed that all of - ^ 1 8 are true in the domain of
rational numbers as well as in the domain of real numbers. It is the Con­
tinuity Schema that distinguishes between these two domains; it has, for
example, the consequence that every positive number is a square {T izg).
On the other hand, the positive rational number 2 has no square root
among the rationals (!Ti28). Accordingly, we must henceforth regard
our variables as referring to the real numbers.
Section 6 Extensions of Theories; the Theory of Real Numbers . 287

Theorems of the theory of real numbers:

T39 I. Show-l{y) AO ^ y A '-^y = o -> I ^3;

4, T 21, ^ 3 , Az, L L
Ai8y Ai6y 2, 5
T 18, 6, AZy A^f L L

Since the theory of real numbers is an extension of the theory of com­

mutative ordered fields, all theorems of the latter are theorems of the
former ; this accounts for the use of T’21 and T 18 in the derivation above.

T40 I(y) Ay ^ OA y = O -> y < —I

It will frequently be convenient to consider theorem schemata as well

as axiom schemata. By a theorem schema of a theory T we understand a
formula containing, besides constants of T, some additional predicates or
operation symbols, and such that all of its instances within T are theorems
of T. For example, T S \ i - below are theorem schemata of the theory
of real numbers.
The derivation of theorem schemata of a theory T is simplified by con­
sidering a theory U that is an extension of T and that satisfies the following
two conditions: (i) every constant of U that is not a constant of T is
either a predicate or an operation symbol; (ii) every instance within
T of an axiom of 17 is a theorem of T, Then any theorem of U that is not a
formula of T will be a theorem schema of T, (This remark requires a small
proof, which will not, however, be given here.)
In connection with the theory of real numbers we consider a theory U
whose constants are those of the theory of real numbers together with the
i-place predicate letter and whose axioms are A i - ^ 1 8 and all in­
stances within U of the formula A S ig , Thus A S ig is itself an axiom of
Uy and so is the formula
V;cF( — A VyAx:(F( —j c ) ^ < 3/) - > V2:[A^(F( — ->
X ^ z) A Aj(A.r[F( — jc) a: ^ j] ^ 3^)]

Clearly, U is an extension of the theory of real numbers that satisfies

conditions (i) and (ii) above.
7**541 provides a simple illustration of these points. It is a theorem of
Uy as the proof below indicates, and hence, by the observations above, a
theorem schema of the theory of real numbers.
TS^i I. S A ^ w - V jc F ( jc) - > V jcF ( — jc)
288 . Definitions; Formal Theories Chapter VIII

2, E l
3, TiO, L L
4, e g

TS^2 is the dual of A S ig ; it asserts, roughly speaking, that every non­

empty set of numbers that is bounded below has a greatest lower bound.

TS42 I. SAot<^VxF(x) AVyAx(F(x) -> y ^ x) ->

V2:[A:r(F(:v) z ^ x) A Ay(Ax[F(x) ->3; < x]
y ^ -2T)]

2. VjcF(^) AVyAjc(F(jc) “>3; < jc)

3- Ax(F(x) -^u ^ x) 2, E l
4- SAofe-Ax(F( —x) - ^ x ^ —u)
F{ —x) ^ —w 3, U I, T io, L L ,
T30
6. yyAjc(F(—jc) ->x 4: y) 4, E G
7* V2r[A:r(F( —jc) jc < 2) A
Ay(A:c[F( —:r) -> x ^ y]
z < y 2, T S 4 1, 6, A S ig
Ax(F{ —x) -^ x ^ w) A
A3^(Ajc[F( —:v) -> x ^ y]
w ^ y 7, E l
9- Shew^Ax(F{x) —zv ^ x)
10. F (jc ) — w ^ X 8, U I, T io, L L ,
T30
11. Shew- Ay(Ax[F{x) ->3^ < jc]
y < -w )
12. Ax[F(jc) ^ jc]
13- A^(F( —.x) - > x ^ —y)
14. F( —jc) X ^ —y 12, U I, T i o , L L ,
T30
15- y ^ —w 8, U I, 13, T i o , L L ,
T30
16. V2f[Ajc(F(:x:) z ^ x) A
Ay(A:c[F(jc) -> j < :v]
y < «)] 9, II, EG

In the preceding derivation a new informal abbreviation was introduced.

The subsidiary derivation of line i i has the following form:
Section 6 Extensions of Theories; the Theory of Real Numbers . 289

Sk 9W‘ A(x{<f)
<f>

rather than the explicit form:

Shew-Ao(,{(f) ip)

This sort of abbreviation corresponds to mathematical practice and will be

used frequently in the sequel.
The following theorem schema, whose proof involves a new abbre­
viation that will be discussed below, makes approximately the following
assertion: a greatest lower bound of a class of integers belongs to that
class. T544 makes the dual assertion concerning least upper bounds.
These assertions, it should be observed, do not hold for arbitrary classes
of real numbers. Consider, for example, the class of real numbers greater
than o. It is easily seen that o is the greatest lower bound of this class
but not a member of it.
r* 5 4 3 I. ^Aд«e^AJc[F(Jc) - > I(jc)] a Ajc(F(:v) -> y ^ x) a
Az{Ax[F{x) - ^ z < : X ] - > z ^ y ) - > F{y)
Ajc[F(jc) I(jc)] A
A ^ ( F ( a :) -^y ^ x ) a

A s r ( A :v [ F ( ^ ) z ^ x ]-> z ^ y )

3 Skew F{y)
4
5 iSAew-Aa;(F(A;) ► 3^ + I < a;)

6 F( ac) a ~ 3/ 4- I < ;JC

7 ^AewA^(F(2) < 2r)

8 F(x) A ~ X ^ z

9 y < z 2, U I, 8
10 3» + I < ar + I 9. ^24
11 AC < j + 1 6, T 21
12 AC < « + I 10, I I , T20
290 . Definitions; Formal Theories Chapter VIII

13- z ^ X 8, T 21
14. I( ^ ) A I(;^) 6, 2, U I, 8, 2, U I
IS- X = ZWX = Z + I 14, 13, 12, ^ 1 8
16. ^ X = z 8, T38
17- X = Z -h 1 6, 10, T308 of ch.
VI
18. X = z + I 1 5 .1 6
19. X ^ y 7. 2, AV, U I
20. y ^ X 6, 2, U I
2 1. X = y 19, 20, T19
22. ^ F{x) 4, 2 1, L L
23- F{x) 6
24. y + I 4: y 5. 2, U I
25- y + I ^ y T27

In the preceding derivation we permit the following variant of indhect

derivation:
SAew-Aa. {<l>

which, executed in full detail, would read:

Shew-A(x(<f>

(An analogous variant of indirect derivation, in which the formula derived

is not
Aa(^ ->• 4>)
Section 6 Extensions of Theories; the Theory of Real Numbers . 291

but simply

is also occasionally useful and is in fact employed in the derivations of

7^53 and T55 below.) The new abbreviation is employed in the sub­
sidiary derivations of lines 5 and 7 above.
7544 Ajc[F(jc) l{x)] A Aa:(F(jc) a: ^ jy) a
A;s:(Ajc[F(^) F(y)

The next two theorem schemata are the natural principles of mathe­
matical induction over all integers. 7545 is a simple consequence of the
Continuity Schema, together with 754 3 and 7544. The stronger principle
7549 can be obtained in a simple way from 7^ 4 5; we leave the derivation
to the reader. 746 - 748 are applications of the induction principle
7 5 4 5 ; the proofs of 747 and 748 are left to the reader. Henceforth
we shall for the most part omit annotative reference to principles of
logic, that is, principles of chapters I - V II.

7*?45 I. Show F(o) Ajc[I(jc) a F{ x) -> F{x + l ) A

a

F (^ + -I)] A;c[I(j;) F(.r)]

F(o) A Ajc[I(;c) a F( x)
F{x -h i) A F{x + “ i)] ^
'^Ax[l(x) -> F(^)]
3- l{u) A ^F(w) 2, QN, E l
4* Show ^ o ^ u
5- o ^ u
6. ShewAx{o ^ X ^ I( jc) a FW
->o ^ ;c)
o ^ JC A I ( j c ) A F (jc )-^
o ^ X
V jc [ o ^ X a I ( .r ) A F (^ )] A
VyA;c[o ^ X A I ( j c ) A
'^F(jc) -> y ^ x] 3, 5, EG , 6, EG
Ax(o ^ X A I ( j c ) A F{x) - >
W ^ x) A
Aj (Aat[o ^ X a I( jc) a
^F(^) - > y ^ ^ w) 8, TS^z, E l
10. S h e w - A x [ o ^ X A I(;c) a

II. o ^ A l(x) A
^ F { x ) - > l{ x )
12. O ^ W A \(w ) A F ( ii;) 10, 9. F543
292 . Definitions; Formal Theories Chapter VIII

13- ^ O = V) 2, 12, T308 of

Ch. V I
H- l{w H---- 1) 12, A 17
15- O < TO + — I 12, 13, T39, T24,
^4, L L
i6. iSA««>-F(w + —i)
17- ~ F (w + —i)
18. < w + —1 9, U I, 15, 14, 17
19. ~ w ^ zo + — I T28
20. F((w H---- 1) + i) 2, U I, 14, 16
21. F(w) 20, A i, A 2, A^,
A 3, L L
22. ~ F (w) 12
23- Show ~ M ^ o
Derivation similar to that
of line 4, using A S ig and
T’iS44 in place of T1S42 and
T S 43-____________________
24 o < M 23, T 21

1 iSA«w I( a:) ^ I( —x)

2 I(-o ) A 16 , Ts
3 S how A;c[I(a:) a I( — ^) ->
I ( - ( ^ + l) ) A l( - ( ^ + -I))]
4 I ( a:) A I ( - ; c )
5 - ( a; + i) = - a; H--- 1 T il
6 I ( - ( ^ + I)) 4. A17, 5
7 - ( a : + - i ) = - a; + I T i l , T io
8 ! ( - ( * + - I)) 4. A17, 7
I( a:) ^ I ( - a:) 2, 3, T545

m I( ac) a I(j ) -> I(Ai + y)

I( a;) a I(j ) I(Ai *y)
T549 Va:[I(a:) a F(*)] a Aa:[I(a:) a F( a:) F( a; + i) aF( a: + —i )] ^
Aac[I(ap) F( ac)]
The inductive proofs of the next two theorems, which are included only
for use in proving T52, are left to the reader. According to T'52, an integer
is always either odd or even but never both. (Here and henceforth paren­
theses within an iterated sum or product will be dropped.)
T50 I(ac) Yy(I(y) A AC = y + y ) v Yy(I(y) AAC=y-f-J-|- l)
Section 6 Extensions of Theories; the Theory of Real Numbers • 293

T^i I(^) ^Yy(I(j;) ^ x ^ - x = y - \ - y ^ \ ^ i)

T52 I(^) -> [Vy(I(>^) f^x = y + y)<r-^ '^Yy(ICv) = y + y ^ i)\
We conclude this section on the Continuity Schema with a proof of the
Principle of Archimedesy which asserts the following: if jy is any number,
and X any number other than o, then y can be exceeded by an integral
multiple of X. This assertion is T54; it is convenient to prove first the
special case in which x is assumed to be positive.
T53 I. Show-o ^ X A X = o ->Vn(l(n) A y ^ n • x)
2. O ^ X A r^X = O A ^Wn{l (n) A
y ^ n • x)
3- AO = O • X
I(o ) A 16 , T6, A6
4* VzWn{l{n) A z = n • x) 3
5- ■ Shew-Az\yn(l{n) a z = n • x) ->
_____________________ <y]
6. V«(I(w) h z = n • x)
7- I(«o) A ar = «0 * Jc 6, E l
8. g < _____________ 2, 7. T 2 1

Aai[Vw(I(«) A ar = n • jc) < ®o] A

Aty(Aar[VM(I(w) ^ z = n • x) ^
ai < z)] < ®) 4, 5, ^ 5 1 9 , E l
10. Show- ~ Aar[V«(I(w) a z = n ' x)
g < fo + —^]
11. Aar[V«(I(w) AZ = n ' x)
g < ®o H---- x]
12. H---- X 9 (2nd conjunct),
II

13- Vo + X ^ Vo 12, T24, A 2, A3,

A4
H- Vo ^ Vo + X 2, r24, ^ 1 , ^ 2 ,
^3
IS- Vo = Vo + X 13, 14, T ig
16. X = o 15, A 3, T2
17- ~x = o 2
18. Vn(I(ii) A aro = n • a;) A 10, T242 of ch.
~go < wo H---- X Ill, E l
19. I (w i) A a o = « 1 • a; 18, E l
20. Vn(I(w) A (mi + 1) • AC = M • a; ) 19, A i-j
21. («1 + l ) • AC ^ Vo 9 (ist conjunct),
20
22. go + AC < 2 1, A6, Ag, 19,
Ay
294 • Definitions; Formal Theories Chapter VIII

23 «0 < i'o H----X 22, r 2 4 ,^ i ,^ 4 ,

^3
24. ~ ZO ^ Vo + —X 18

T’54 1. Show ~ X = o V«(I(n) A _y < n • ¿c)

2. ^ X = o
3- O ^ X -> V«(I(m) A y ^ n • x ) 2, 7^53
4- Shew-x ^ o -> V«(I(m) a ^; < « • a;)
5- X^ o
6. O < —X 5> ^ 31
7- ~ —X = o 2, T 9
8. I(no) A^ ^ no • (-.x) 6, 7, TS3, E l
9- y < (-« 0 ) • iX 8, T i3, T iz
10. I(-«o) 8, T46
11. Vn(I(n) A y ^ n • x) 10, 9
12. Vn(I(n) Ay ^ rt • x) 3» 4»

7. Definitions. Of considerable importance, both from the intuitive

point of view and for later purposes, is the notion of a definition of one
constant in terms of others. Loosely speaking, a definition is a formula,
more specifically, a biconditional or an identity, which elucidates the
meaning of the constant it defines. We shall exclude from consideration,
both here and in the next chapter, definitions of logical constants. These
symbols will be regarded as completely understood, and accordingly will
be used freely in the formulation of definitions. In this chapter we shall
further restrict ourselves to definitions of predicates and operation sym­
bols. Definitions of variable-binding operators present special difficulties
whose consideration we prefer to postpone till the next chapter.
Thus in the following S is assumed to be a nonlogical constant and
either a predicate or an operation symbol; that is, 8 is to be either an
operation symbol or a predicate other than ‘ = \
If S is an «-place predicate (other than ‘ and L any class of nonlogical
constants not containing S, then a possible definition of S in terms of L
is a formula of the form
S a i, , , (f> ,
where ai, . . . , are distinct variables, and 0 is a formula all of whose
nonlogical constants are members of L and which contains no free variables
beyond ai, . . . , oLn-
For example, let L be the class consisting of the symbols ‘ + and
‘o’. Then a possible definition of the 2-place predicate ‘ < ’ in terms of
L is the biconditional
X < y <r-^ y z{ ^ 2T = O A ^ + (^r • 2t) = y )
Section 7 Definitions . 295

Definitions of operation symbols will assume the form of identities

rather than biconditionals. Thus, if S is an w-place operation symbol and L
any class of nonlogical constants not containing S, then a possible definition
of S in terms o/ L is a formula of the form
S a i. . . i ,
where ai, . . . , oun are distinct variables, and i is a term all of whose
nonlogical constants are members of L and which contains no free vari­
ables beyond ai, . . . , a^.
For example, let L be the class whose only member is the symbol
Then a possible definition of the i-place operation symbol in terms
of L is the formula
^ = V[Cv ^y) ' y = X\ .
Generally speaking, the right side of a possible definition is regarded as
elucidating the meaning of the constant occurring on the left side; hence
the requirement that the latter constant not occur on the right side of the
definition.
It is often useful to introduce new constants into a theory T by defini­
tion. This consists in passing to an extension of T obtained from T by
adding possible definitions of the new constants in terms of the constants
of T, More exactly, we say that a theory (7 is a definitional extension of a
theory T just in case (i) U is an extension of T, (ii) every constant of U that
is not a constant of T is either a predicate or an operation symbol, (iii)
for each constant 8 of 17 that is not a constant of T, there occurs among
the axioms of U exactly one possible definition of 8 in terms of the con­
stants of Ty and (iv) every axiom of U that is not an axiom of T is a possible
definition, in terms of the constants of T, of some nonlogical constant
that is not a constant of T.
The distinctive properties of definitions stem from the following three
facts, which concern definitional extensions and do not hold for arbitrary
extensions.
(1) If is a formula of a definitional extension C/ of a theory T, then
there is a formula i/j of T such that
xfi

is a theorem of U, Thus the power of expression of a theory is not essen­

tially increased by the addition of defined symbols.
(2) If is a theorem of a definitional extension of a theory T and at
the same time a formula of T, then <f> is already a theorem of T. Thus
the deductive power of a theory is not essentially enhanced by the addition
of definitions.
(3) The assertion (2) has as an immediate consequence the fact that
the addition of definitions can never introduce a contradiction into a
zg6 . Definitions; Formal Theories Chapter VIII

theory. Accordingly, let us call a theory T consistent if there is no sentence

<f>such that both (f> and its negation are theorems of T. Then a definitional
extension of a consistent theory is always consistent.
We should also call attention to the fact that (4) if K is a definitional
extension of U and ¡7 is a definitional extension of J*, then V is equivalent
to a definitional extension of T. (Two theories are said to be equivalent
if they have the same constants and the same theorems.)
The theory of real numbers has as a definitional extension the theory
Ti, characterized as follows. The constants of T\ are those of the theory of
real numbers, together with:
(2-place predicate)
< (2-place predicate)
2 (individual constant)
2 (I-place operation symbol)
I I (I-place operation symbol)
— (2-place operation symbol)
— (2-place operation symbol) .
The axioms of T\ are those of the theory of real numbers, together with:
Di X ^ y X = y

Dz X<y<->X^ytif->^X=y

D3 2 = 1 +1
Z )4 X^ = X •X

DS \x\ = 1 ^ [(o ^ X A Z = x)w{ -X )]

D6 X — y = X + —y

1)7 -^ = X •y-^
y
The intended reading of the new constants should be clear from the
possible definitions D i - D y . For example, the operation symbol intro­
duced by Z?5 is read ‘the absolute value of x\ Z>y has as a consequence
X
- = X ■ ^-1
o
Thus division by zero, like the expression is regarded as meaningful.
Our axioms, however, do not determine the value of because of the
antecedent of A8. In deference to mathematical custom, we depart
further than usual from our official notation and use ‘ —* both as a i-place
and as a 2-place operation symbol. D6 defines binary ‘ —’ in terms of
singulary ‘ —’ and addition. The context will always determine which
of the two senses is intended.
Section 7 Definitions . 297

We list a number of theorems of Ti, proving a few and leaving the

others (which can be easily obtained from preceding theorems and defi­
nitions) to the reader.
I . Show-\{n) A I(m) Ml < m -> n + 1 ^ tn
I(w) AI(m) A w < m A ^ w + i ^ w
n ^ m 2, Dz
m ^ n + i 2, T z i
m = nw tn = n + I 3 .4 . ^ 1 8
^ m = n 2, I>2
n + I ^ m 5 . 6, T28
^ n + I ^ m 2

T’56 X — X = o

T’S7 - ( x - y) = y - X
T ss X — y = {z — y) — { z — x)

T59 {x - y) -{- {z — w) - {x z) — {y + w)
T60 {x + y Y = X^ zxy + 3^2

In Tbo, as in later statements, we adopt the mathematical practices of

omitting the multiplication sign and of regarding addition and subtraction
as marking greater breaks than multiplication.
T61 j 2 = I

T62 {x — y Y = x^ — zxy +
T63 {x • y Y = JC2 • 3/2
X \2 X^
T64 y^o
y j ~ y2

X X
T65 - + - = X
2 2

o
T66 - = o
X

X+y X y
T67 — = - + -
z z z

xy y
T68 XZ o — -
xz z
298 . Definitions; Formal Theories Chapter VIII

/ X \ —X /X X
T6^ - = ---- A - -
\yj y \y -y

T jo 0 X^ X ^ 1 -> X^ ^ X

T 71 0 ^ X^
T72 X 0 -> o < x^

T72 X ^ X +

T’74 0 < I

T75 I < 2

T76 ^ X < X
T77 x < y ^ y < z -^ x < z
T7S X^yAy<Z->X<Z

T79 X<yAy^Z->X<Z
T80 x^y<->^y<x
T81 x^y<r-^x<yvx=y
T82 x<y<-^x + z<y + z
T83 X < y A Z < W ‘^ X + Z<y +
T84 0 < X X < 2X
T85 X < y —y < —X
T86 X < 0 <-^o < —X
T87 0 < JC -> 0 <

T88 X < 0 x~^ < 0

T8<) 0 < z -> [x < y<-^ xz < yz]
Tgo z < 0 -^ [x < y<r-^yz < xz]
T91 0 <XA 0 < y - ^ [ x ^ y<->y~^

T92

T92 0 <XA 0 < y ^ ‘ [ x < y<r^y~^ -

T’94 J C < O A J < 0 - > [ : C < y<-^y~^

z z
7^95 0 < X A X ^ y A 0 ^ Z -^ -^ -
y X
Section 7 Definitions . 299

T96 0 < X A X < y t i 0 < Z - > —< -

y X

X y
Tgj X ^ y AO < z ^ -
z z

X y
TgS x<y^o<z->-< —

X X
T99 < X A X < y ^ O < - A - < I
y y
Tioo Shew-o ^ y Ax"^ ^ y^ ^ x ^ y
2 O ^ y A X^ ^ y^
3 Show-o < X A o < y ■X ^ y

4 0<XAO<yArsuX^y
5 y < X 4, T80
6 y -y < y •X 4. 5. ^’89
7 y •X < X •X 4. 5. TSg
8 -y2 < j(^2 6 , 7, T77, D4
9 y"^ < x"^ 2, T 8 0

10 Show X ^ o -> X y
11 X ^ o
12 X y I I , 2, T20
13 Show y = o -> x < y
H y = o
15 x^ ^ o 2, T6
1 4 , £>4,

16 x^ = o 15, T 71, T ig
17 X= o 1 6 , £>4, T '1 6

18 X^ y 1 7 ,1 4 , T I8

19 {0<XA0<y)Wx^0yy = 0 2, T81, T80

20 X ^ y 3, 10, 13, 19

X
T io i 0^XA0<y->0^-
y

T102 O < X /^o < y -> o < -

y
T103 ViA^[(o ^ x ^ z = x ) 'f{'^ o ^ x ^ z = —x)<r^z = t\

By the last theorem, the descriptive phrase in D5 is proper; using this

fact, it is easy to obtain the following two theorems.
300 . D efinitions; F o rm a l T h eo ries Chapter V IIl

Г104 О ^ л: |л;| = л:

T105 л: < о -> |л:| = — X

Having obtained T104 and 7*105, it will no longer be necessary to

refer to Z>5, which has a rather cumbersome form.
T106

3. T i o ^

3. 4

7. T ío s
7, Ш
8, 9, D2
2, 6, TSo
7 10 7 \x\ = 0<->ЛГ = о
7 108 X ^ |л:|

T109 — X ^ |д:|

7 iio |-дс| = \x\

T ill X\ ‘ ^ у < - ^ Х ^ у А —Х ^ У
T 1 12 I. ,§Ав«>-|д; + j | < |л:| + Ij I
2. X < |л:| Tio8
3- у <\y\ Т108
4- л: + j < |л:| + Ij I 2, 3, Г 26
5- -л ; < |л:] T i 09
6. -у < bl Г 10 9
7- - X -\-----j < |д:| + |у 1 5. 6, Т26
8. - Ь + J') < bl + \у\ 7. T i l
9- Ь + >'1 < Ь| + Ь| 4, 8, T i l l

T i l 2, which is an immediate consequence of T i 12, is the familiar triangu­

lar inequality.

T112 1^ - j»! < |;r - 2| + 1« - y\

T 114 +
T iis \x-y\ = \y-x\
Section 7 Definitions . 301

T ii6 \x-y\ = \x\ -\y\

T i iy y # o
bl
T ii8 1. — y\ K Z ' i - ^ x — z < y / ^ y < x + z
2. Shew-x — z < y f ^ y < x + z - ^
y\ < b-
3- X — Z < y Ay < x + Z
4- —{y — x) < z 3, T82, D6, A2,
A i, A4, A3,
T85, T io
5- y —X < z 3, T82, Z)6, A2,
A i , A4, A3
6. |y — *1 = y — a; V
|y - jc| = - ( y - x) T104, T105, T80
7- b — y| < z 4. 5. 6
8. 5 Ae«>-b —jy| < z ^
X — z < y f y y < x - \ - z

9- \x - - y | < ar

10. X —y < z 9, Tio8, T^8

11. X —z < y 10, T82, D6, A 2,
A i , A4, A 3
12. y “ X< z 9, T ii5 ,iT io 8 ,r7 8
13- < X+ z 12, T82, D(>, A2,
A i , A4, A 3
14- X —Z < y h y < X ' { ' Z 11, 13

We pass now to a definitional extension Tz of Ti, obtained by adjoining

the I-place predicates ‘N ’ and ‘R ’ to the constants of Ti, and the following
possible definitions to the axioms of 7 i:
D8 N(x) I(x) A o < a:

Dg R(a;) VyV2r[I(y) A Ib ) AO<2TAx: = -]

z

D8 introduces the notion of a natural numbery Dg that of a rational number.

By remark (4) on page 296, T2 is equivalent to a definitional extension of
the theory of real numbers.
The following theorem schema (of is the minimum principle for
natural numbers: every nonempty set of natural numbers has a least
member.
302 . Definitions; Formal Theories Chapter VIII

T S iig I. iSAew-VatF(^) a Ajc[F(iii;) N (a;)]->

Vjf(F(*) A Ay [F(>') < j] )
2 VicF(A:) A A*[F(*) -> N(*)]
3 ^A©«>-Aa![F(a;) - > o < ;c]
4 F(^)
5 O^ X 2, 4, m
6 A :v [ F ( ^ ) -> z o ^ x] A
A jv ( A :v [ F ( ^ ) - > jy < x] - ^ y < ^o) 2 ,3, T542, E l
7 iSAw'A^[F(^) -> I(^)]
8 F(^)
9 I(^) 2, 8, D8
10 F(2To) 7, 6, T543
II V jc(F (*) a A y [ F ( j) -^ x ;y]) 10, 6 (ist con­
junct),A V ,EG

We now state two induction principles for natural numbers, both of

which are immediate consequences of the minimum principle. T S120,
whose proof is left to the reader, is the principle of strong induction; it
asserts that every natural number has the property F, on the assumption
that an arbitrary natural number has F if every smaller natural number has
F. T S i2 \ is the more familiar principle of weak induction; if o has the
property F, and whenever a natural number x has F, so does ic + i,
then every natural number has F.
T S 120 Ajc(N(;c) a Ay[N(y) Ky < x ^ FCv)] F( jc))
Aa;[N(a:) ^ F(x)]
T S 12 1 I. ¿¡A«wF(o) A Ajc[N(ji:) A F(jf) -^F(a; + i)] ->
A*[N(x) F( a;)]
2. F(o) A AAf[N(ic) A F(af)
F(a; + i)] A ~ A;e[N(ie) ^ F( a;)]
3- Va:[N(ic) A ~ F(a;)] 2 (3rd conjunct.)
4- A«:[N(a:) A -F (ic) ^N (ic)]
5- N(a;o) ^ ~ F(^o) ^
A;v(N(a;) a ~ F ( jc) ->-«0 < *) 3, 4, r 5 i i 9 , E l
6. JfO # 0 2, 5
7- N(ato - i) 5, D 8 , A 1 7 , D 6 ,
6, ?’39
8. F(iCo - i) 5, 7, T28, D 6
9- F(:ro) 2, 7. 8
10. ~ F(a;o) 5

In the annotations of lines 7 and 9 above, we have omitted reference

to certain arithmetical principles that are by now completely familiar;
Section 7 Definitions • 303

this practice will be pursued in the sequel. Also, obvious theorems of

logic, such as line 4 above, will sometimes be used even if not listed in
chapters I-V II.
The theory of natural numbers was axiomatized for the first time in
Peano [1]. Peano’s axioms consist of T S 12 1, together with T 1 2 2 - T125
below. T126 expresses the fact that the set of natural numbers is closed
under addition and multiplication.
T 122 N(o)
TI23 N(.v) N(^ + i)
Ti24 N(jc) A N (j) A .v 4- i = j + i X = y

T125 N(.v) a; + I # 0
T126 N(a;) a N(j ) -> N(a; + j ) a N(.

The next theorem, which is a lemma for T12S, asserts that if is

an even integer, then so is x. T128, whose proof is due to Pythagoras
and appears in Euclid [i], states that 2 has no rational square root.
Tl2"J A Vy(I(y) A = 2y) -» Vy(I(y) /^x = 2y)
I(ic) A yy(I(y) hx^= 2y) A
~Vy(I(y) h x = 2y)
I(yo) f'X = 2yo + I 2, T52
X^ = (2yo)^ + 2 • 2 - yo + I 3, T6o, T6i
= 2 •2 •yo^ + 2 •2 •Jo + I 4 , T'63
_ 2 •(2 •Jo^ + 2 •Jo) + I 5, ^ 9
1(2 •JO^ + 2 •Jo) 3, T48, ^ 16 , J 1 7 ,
T'47
Vj(I(j) A x 2 = 2 -J + l) 7, 6, E G
~ Vy(I(j) Aa:2 = 2 - J + 1) 2, T48, Ts2

T128 Show R(jc) - » # 2

R(x) A x^ = 2
y
V«Vj[I(j) A I(sr) A o < « A .V = -] 2, £>9
Show-f\.z(^yW{y) A I(ar) AO < z a

.V = -] ■ N(^))

Elementary

V j[I(j) A I(iro) A o < 3T0 A ic = y A

As:(Vj[I(j) A I(s) AO < Z A

304 • Definitions; Formal Theories Chapter V III

^ = ■ ] -^ 20 < 2) 3, 4, T 5 i i 9, E I
z
yo
I(jo) A I(;Sro) AO < Zo^X = — 5 (ist conjunct), E I

yo^
7- = 2 6, 2, T6\
Zo^
8. yo^ = 2 • Zo^ 7, 6, r i 6
9- V v { l( v ) A yo^ = 2 • V) 6, T48, 8, E G
10. I(Wo) Ayo = 2 - Wo 6, 9, Ti27, E I
11. 2 • 2 • Wo^ = 2 • Zo^ 8, 10, T63
12 . 2 • Wo^ = II
13- Wn{l{n) A Zo^ = 2 • W) 10, T48, 12, E G
14. I(wo) A Zo = 2 •mo 6, 13, TI27, E I
2 • Wo
15- X = 6, 10, 14
2 • mo
Wo
16. X = 14, 6, 15, T68
Wo
17- S h o w o < mo
18. o < mo
19. 2 • Wo < o 18, T8o, T33
20. zo o 14, 19
21. ^ zo ^ o 6, T8o
22. Zo ^ Wo 10, 14, 17, 16, EG ,
5 (2nd conjunct)
23- Wo < Wo + Wo 17
24. Wo < ^0 23.14
25- 2To ^ Wo 24, T80

While many positive numbers resemble 2 in having no square root

among the rationals, it is a consequence of the Continuity Schema that
every positive number does have a square root:
7 ’i 29 I. Sime-o < x ■ S yx —
2. o < a:
3- < ar
4 W yy^ < X

5 Show Ay(y 2 ^ X < IVy < »)

6
7 6
8 7
9 8, 6
10 6
Section 7 Definitions . 305

1 1. Show 1 ^ X
WzAy(y^ ^ X -> y ^ 2)
12. l <: X

13- Show-Syiy^ < X -> y < x)

Elementary, from 5
and 12
H- W zh y {y ^ ^ X - ^ y ^ z) 13
IS- Show X ^ I ->
yzSy{y'^ ^ X -^ y ^ z)
Similar to proof of 11
16. V z A y (y ^ ^ X -^ y ^ z) II, 15, T 21
17- Ay(y2 ^ X -> y < 2o ) A
Aw[Ay(y2 ^ X -^ y ^ ii)
Zo < m] 4, 16, A S i^ , E l
18. Show 1 ^ X -^ o < zo
19.
20. 19
21. i7(istconjunct), 20
22. 21

23- ■ O < Zo
24.
25- 24, 2
26. i7(ist conjunct),
25
27. 2, 26
28. O < Zo 18, 23, T2I
29. ICTTTtnMr ^0
30- ^ Zo^
SI- Zo^ < X 30
32. Wh T 32 1 of ch. V I

33- 32, E l
34- iSAew^o < ho ho < i
iTo^ a:o2
35- O < --- A -- < I 28, T j2 , 3 1, T99
X X
Zo^
36. — I < ------A
X
Zo‘‘
< o 35. TSs
3o6 . Definitions; Formal Theories Chapter VIII

37-

36

38. 37, 33

39- 38, 774, 775,

796
40.

41 ^ - f. - ^ )
X \ X /
42 = I — zho 41. 33
43 I — zho < I — zho + ho^ T72
44 I — 2 /(o < (i — ^0)^ 43, 2^62
Zo^
45 < (i - ho)^ 42, 44

46 -s-o^ < » • (i — ho)^ 4 5 ,2

47 o < 1 — ho 34
48 o < (i — ho)^ 47, T jz
Zo‘‘
49 < X 46, 48, T g j
(I - hof

50 49, 47, i ’64

-S^o
51 ^ Zo i7(ist conjunct),
1 — ho 40
52 — ho < o 34
53 I — ho < 1 52
54 o < 1 — ho 34
Zo
55 Zo < 54, 53, 28, 796
1 — ho
Zo
56 2^0 < 51
I - ho

57 Shew-zo^ < X
58 ~ Zo^ ^ X

59 X < Zo^ 58
60 hi T 3 2 1 of ch. VI,
El
Section 7
Definitions . З07
61. Skew-0 ^ hi h hi < i
^ X
62. о < Л — < I 2, 59. 7^99
X X
63- - I < ^“ --- A — — < 0 62, Г85
гр2
X X
64 о<I
^ I “ — < I 63
65 A ni <. ^ 64, 60
66 ° < I
65
67
. 3 ; ^ го • (i - Л1))
68 X

69 л:
^ = го2 . 28, Г72
«2
70 ~ ■2^0^ • ^ I — ^I — 69
71 ~ 2 q2 . _ 2 h i)
¿)) 70, 60
72 i - 2 A 1 ^ (l _ A i)2 T'73, T62
73 Л < 2 q 2 . (i _ /ij)2
72, T’71. 71
74 ^ < ( 2 0 • (l - Ai))2 73. Т’бз
75 < (^0 . (j _ 68, 74
76 О < I - Ai 6i
77 О ^ го •(i — h i ) 28, 76, ^ 1 4
78 У ^ s : o - ( i - h i )
75. 77. T’loo
79 ^0<i 2To • (l — hi) i7(2nd con­
junct), 67
80 —hi < о 61
81 i — hi < 1 80
82 «0 • (i - hi) < го 81, 28
83 <^гр • (i — hi) < го 79
84, X = гр2 29. 57. ^ 19
85. \y X = 84
It is a simple consequence of T22 and T129 that a number is non-
negative just in case it is a square:
T130 o ^ X=
The following theorem, which follows immediately from J'130 , shows
how might have been defined in terms of ‘ 4-’ and
T 131 X^ y X = y
For historical remarks pertaining to this chapter, see section 4 of
chapter IX .
 Chapter IX
Variable-binding operators

I. Definitions reconsidered. We turn now to the treatment, deferred

from chapter V III, of definitions of variable-binding operators. Roughly
speaking, in order to define a nonlogical constant S of degree </, m, w, p >,
we must provide for each term or formula
Sa i . .. cLfii .. . in
(where ai, . . . j ccm are distinct variables, ii, . . . , in are terms, and
<l>ij ^ • y <f>p are formulas) a synonymous term or formula not containing S.
To do this precisely, we shall make use of operation and predicate letters
and definitional schemata.
For simplicity, consider for a moment the special case of constants of
degree <i, i, o, i ). Let S be a nonlogical constant of this kind, and L
a class of nonlogical constants not containing S. Then a possible defini­
tional schema for S in terms of L will be a formula of the form
SaTra ^ <l> ,
where a is a variable, tt is a i-place predicate letter not in L, and is a
formula containing no free variables and no nonlogical constants beyond
77 and those in L,
More generally, consider an arbitrary nonlogical constant S, and again
let L be a class of nonlogical constants not containing 8.
If 8 is a nonlogical constant of degree <o, w, w,/>), possible defi­
nitional schema for 8 in terms of L will take the form
8a i . . . (Х.ГП i i a i . . . а ш ... •••OLm 'Л '!«! . . (X-m TTpOLi. . . —V
Here Oily. . . у (Xm are to be distinct variables, ii, . . . , iw are to be distinct
m-place operation letters, tti, . . . , ттр are to be distinct m-place predicate
letters, and rj is to be a term containing no free variables, and no non­
logical constants beyond ^i, . . . , Cn> • • • у'^ру those i nL ; we assume
in addition that Ci, . . . , Uy ^ri, . . . , ттр are not in L, and are distinct
from 8.
If, on the other hand, 8 is a nonlogical constant of degree /w, w, /)>,
then a possible definitional schema for 8 in terms of L will be a formula
Section I Definitions Reconsidered • 309

S a i... am Ciai •. • a7/i • • • in ^i • • • Tr^ai •

As before, ai, . , . , are to be distinct variables, ii, . . . , are to be
distinct m-place operation letters, and tti, . . . , are to be distinct m-place
predicate letters. In addition, is to be a formula containing no free
variables, and no nonlogical constants beyond ^ i ,. . . , tti, . . . , tt^, and
those in L ; again ^1,. . . , ^ 1,. « • , are to be distinct from S and
the members of L ,
For example, a possible definitional schema for the operator ‘lim’
(mentioned on page 273) in terms of
o, < , - , I I
would be the following:
lim n A(w) = 1jcA2'(o < z ^ VMw[N(w) h k < n IA(n) — jc| < 2r]) .

(This schema in fact reflects customary mathematical usage and will be

included in theories developed below.) A s another example, consider the
operator (also mentioned on page 273). The following formula is a
possible definitional schema for this constant in terms of the empty set
of nonlogical constants:
<-> YyAjc[F(jc) <r-^x = y'\ .
Observe that the notions just introduced comprehend the case in which
S is a predicate or operation symbol, which arises when m = o. Thus we
can dispense with the possible definitions of chapter V I I I in favor of a
uniform system of possible definitional schemata.
In order to accommodate the definitional introduction of variable­
binding operators, we must modify the characterization of a definitional
extension of a theory; we now use the notion of a possible definitional
schema.
Indeed, a theory U is now said to be a definitional extension of a theory
T if (i) U is an extension of T, and, for some Z), (ii) Z) is a class of possible
definitional schemata, in terms of the constants of T, for constants of U
that are not constants of T, (iii) for each constant 8 of U that is not a
constant of Ty there is in D some possible definitional schema for 8 in
terms of the constants of T, (iv) each constant of U that is not a constant
of T occurs in at most one member of Z), and (v) the axioms of U consist
of the axioms of T together with all instances within U of members of D.
As an example, we may consider a theory T 3 that closely resembles
the theory T2 of chapter V I II (p. 301). The constants of T 3 are those of
T2, and the axioms of T 3 consist of those of the theory of real numbers,
together with all instances within T 3 of the following formulas (‘A ’ and ‘B ’
are here to be o-place operation letters):
ly S I A ^ B <—> ^ A = B
DS2 A •< B i—>A ^ B a ^^A = B
310 • Variable-Binding Operators Chapter IX

DS^ 2 = 1 + 1
Z)54 A2 = a • a

DSs |A| = l 2:[(o ^ A a 2t = A ) v (^ O ^ \ A z = —A)]

DS6 A - B = A + -B
A
D Sj — = a • B -i
^ B
DSS N(A)<->I(A) AO < A
DSg ^ V3^V2r[I(j) A I(2t) A O ^ z a O = -Sr A A = J • z~^^] .

It is clear that the formulas D S i - DSg are possible definitional sche­

mata in terms of the constants of the theory of real numbers, and hence
that T 3 is a definitional extension of that theory. Further, T3 is easily
seen to be equivalent to T2. Accordingly, in developing T 3 and its exten­
sions, we may employ T i - T 1 3 1 of chapter V III.
(The construction of T3 illustrates a general principle: if i7 is a defini­
tional extension of a theory T in the sense of chapter V III, then there is a
definitional extension of T in the present sense that is equivalent to U,)
Definitional extensions continue to enjoy the properties ( i) - ( 4 )
mentioned on pages 295-96.

2. The theory of convergence. We are now in a position to extend

T 3 by introducing definitions of variable-binding operators; it is essential
for this purpose to employ possible definitional schemata and not merely
possible definitions. Accordingly, we construct as follows a definitional
extension T4 of T3. The constants of T4 are those of T 3 together with the
formula-maker ‘o ’, which is to have degree <(i, i, 2, o ), and the axioms
of T4 consist of those of T 3 together with all instances within T 4 of the
formula:
D S 10 A{n) o B(w) ^

Az{o < z ykAn[N{n) A k < n |A(w) — B(w)l < -sr])

(This symbolism departs from our official notational style in the direc­
tion of mathematical practice, which often introduces theoretically super­
fluous parentheses and which, generally speaking, requires the variables
accompanying an operator to be written as subscripts. Strict adherence
to the style of chapter V III, section i, would convert the left side of the
biconditional D S io into ‘o n An Bw*.)
The formula
A(n) ^ B(w)

is read 'A{n) and B(n) converge (to one another, as n approaches infinity)’.
Section 2 The Theory of Convergence . 3 1 1

Intuitively, the terms ‘A(w)’ and are regarded as representing two

infinite sequences of numbers,
A(o), A(n),.
and

according to D S io , these sequences are regarded as converging if, for any

positive number z, there is a point beyond which the difference between
any two corresponding terms of the two sequences is less than z.
The next two theorems give simple examples of convergence and non­
convergence.

T 132

3. 5

T67

7. 6

10
312 . V ariable-Binding O perators Chapter IX

12. 9, II

13* 3» 5

H* 6 ,1 3

15- 12 ,14

l6.
«2 + I
— n < Z\ 4, E G

«2 + I
О n 2, D S io
n ^

^133 Show ~ я + i o n
n

2 n I on
n
3 WkAn[N{n) Ak < n ^ |n + I -n|<i] D Sïo , Г74
4 An[N(n) A Ло < и ^ |я + I — я| < i] 3. E I
5 Show о ^ ko ^ Уя[К(я) A Äo < я]

6 о ^ ko
7 1(яо) А ^0 + I ^ Яо • I Г54, A io , E I
8 ko < по 7
9 Щпо) 6, 7, 8
10 V H [N (n) А Äo < я ] 9, 8, E G

11 Show ko < о Vя[N(я) А Äo < я]

12
13 T 122
14 12, 13, E G

15 У я [ К ( я ) AÄo < я ] Sr ”
16. N(ni) А ko < n i 15. E I
17- |я1 + I - т\ < I 4, 16
Section 2 The Theory of Convergence . 313

18. tti + I — til = I

19. |I| < I 17, 18
20 . I < I 19
21 . ^ I < I T76

Nonlogical operators, like the logical operators ‘A’, ‘V’, and ‘ 1 ’,

satisfy general principles of alphabetic variance and interchange. Particular
cases of these principles are given in the next two theorem schemata.
(By a theorem schema of T 4 is understood, as before (see p. 287 of
chapter V III), a formula containing some additional predicates or opera­
tion symbols beyond the constants of T 4 and such that all of its instances
within T4 are theorems of T4. In showing particular formulas to be
theorem schemata of T4, we shall consider an extension T 4 ' of T 4 that
contains the additional i-place operation symbols ‘C ’, 'Ci*, ‘D ’, ‘Di*, and
‘E ’, together with the i-place predicate T * (which figures in ^ 5 1 9 ) , and
whose additional axioms are the instances within this enlarged vocabu­
lary of A Si() and D S i - D S io , Then each theorem of T 4 ' will be a
theorem schema of T4.)

T 5 i 34 I. Show C{n) ^ D(n) C{m) D(w)

2. C{n) D(w) <-> C{n) D(w)

3* C{n) o D (w )^ C(w) o D(m) 2, AV

T S i^ S Show-KnC{ri) = Ci(n)A Aw D («) = Di(w)

[C(m) o D(n) <-> Ci(n) o Di(n)]
n n

2. hnC{n) = Ci{n) A A«D(n) = Di(n)

3. C(w) D(w) Ci{n) D(n) 2, Int

C(w) D(n) <-> Ci(/i) Di(w) 2, Int, 3

According to the next three theorem schemata, the relation of con^

vergence is an equivalence relation, that is, reflexive (T1S136), symmetric
(T aS i 37), and transitive (T*Si38). The proofs of T*Si36 and T Si^ 'J are
very simple, and are left to the reader.
T 5 i 36 C{n) o C{n)

T S 12 7 C{n) ^ D{n) D{n) ^ C{n)

314 • V a ria b le -B in d in g O p erato rs Chapter I X

7 5 13 8 I. Show C(;z) D(w) a D(«) E(;z) C(w) ^ E(w)

C(n) D(w) A D(w) ^ E(n)

Shew-hz{p < z -> V/jA w[N ( w) hk < n->

1C(«) - E(«)| < z])

o < z

o < -
2

Anj^N(K)A^0 < « -^|C(7?) —D(n)| < - j 5,2, D5 io, E l

7- A«j^N(K)A < tt ^|D(n) —E(n)| < -j 5, 2, D S io , E l

8. Show ko < ^1 -»VMn[N(n) A A < M->

9-
10.

11.

12. 6, II, 9

13- 7. I I
14.
12, 13
15- T 112 , 14
16.
10, E G

17- Showki ^ ko->-VkAn[N(n)Ak < M->

|C(«) - E(«)| < z]

Similar to derivation of line 8

18. VMn[N(n) A ^<«-^|C ( m) —E( m)| <sr] 8, 17

19. C{n) E( m) 3, Z)5 io

Section 2 The Theory of Convergence . 315

The relation of convergence is not only an equivalence relation but also,

according to the next theorem schema, a congruence relation under the
operation of addition. The proof of TSi2<)y which is similar to that of
TSi^Sy we leave to the reader.

T S 129 C(n) Ci(n) A D(w) Di(n) ->

C{n) + D(w) Ci(«) + Di(w)

If two sequences are ultimately identical (that is, differ only in an

initial segment), then they converge:
T S i\o VMw[N(w) h k < n -> C{n) = D(n)] C(w) D(w)
This is a simple consequence of D S io .
The next theorem schema is a lemma for T S ij{2\ it asserts that if E(w)
is a strictly increasing sequence of natural numbers, then, for every
natural number w, n ^ E(n).
7*5141 I. Skew-\nKm\N{n) a N(w) kh < m -> E(w) < E(m)] a

An[N{n) -^N(E{n))]->An[N{n) < E(n)]

AwAm[N(«)AN(m)AW < m ->E(w) < E(/w)]

A Aw[N(w) -> N(E(w))]
3- o ^ E(o) 2, DS8
4* Shew-An[N{n) a w ^ E(w)->
w + I < E(n + i)]

5- N(w) A w < E(w)

6. w < w + I 727, 7 8 o
7- E(w) < E(w + 1 ) S> 6, 2
8. n < E(w -f i) S>7
9- I(w) A I(E(w + i)) 5, DS8y A ijy 2
10. w + I ^ E(n + i) 9> 8, 755

1 1. Aw[A^(w) w ^ E(w)] 3,4 , 7 5 i2 i

According to 7 5 14 2 , if two sequences converge, then so do any two

corresponding infinite sub-sequences.
7 5 14 2 I. [C(w) D(w)] A
AwAm[N(w) A N(m) Ati < m -> E(w) < E(m)] a

Aw[N(w) N(E(w))] -> [C(E(w)) o D(E(w))]

3i6 . Variable-Binding Operators Chapter IX

[C (« )^ D (« )]a
AnAm[N(M)AN(OT)A« < m-^E(n) < E(m)]A
An[N(w) ^ N ( E ( m))]
Show hz{o < z -> VMn[N(n) a H < n -^
lC(E(n)) - D(E(«))1 < z])

o < z
An[N(«) A ko < n 1C ( m) — D ( m)| < z] 2 (iS tcon­
junct),
DSlOy 4,
El
6. Show A«[N(n) Ako < n
|C(E(«)) - D (E («))1 < z]

7- N ( m) Ako < n
8. n ^ E(n) 2, r 5 i 4 i, 7
9- ko < E(n) 7>8
10. N(E(n)) 2, 7
1 1. 1C(E(«)) - D(E(n))| < 5 . 10 . 9

12. VkAn[N(n) Ak < n

1C(E(«)) - D(E(n))| < z] 6, E G

13 - C (E (n ))^ D (E (« )) 3. ^ ^ lo

We now consider the familiar notion of the convergence of a sequence to

a number, which emerges as a special case of the convergence of two
sequences. T'5'143, which is simply an instance of D S io , gives a necessary
and sufficient condition for a sequence to converge to a number x.
7'5' i 43 C ( m) 0 A2t(o < z ^ VAA«[N(n) a k < n -> |C(n) — jcj < ar])
Tl
Thus the sequence
C(o), C (i),

is regarded as converging to the number x if, for any preassigned interval

surrounding X (no matter how small), all terms occurring sufficiently late
in the sequence lie within that interval.
We state the following theorems as examples of convergence to a
number.
^ w+ I
Ti44 ------ ^ I
n ”
Section 2 The Theory of Convergence .

I
Th s
n ^
The infinite sequences involved in these two theorems are respectively
1 2 3 4
I > > ) •••
0 1 2 3
and
I I I I
> » > “ > ••• •
0 1 2 3
The assertions T144 and T’i45 are thus intuitively true. (The fact that our
axioms do not determine the value of 5 does not affect the truth of
J'144 and T’i45; indeed, questions of convergence are always independent
of the initial terms of the sequences involved.) We prove r i4 4 and leave
T145 to the reader.
w + I
I. Show ------- o I
n «

iS^wee-Aar^o < z

n + I
WkAn[N(n) A A < M ^ —I < 2

3- o < z

4-

5-

6. T6^

7- 6

8. 3. 5

9- 8
3i8 • Variable-Binding Operators Chapter IX

10.

11. 7 ,1 0

12. 3> 5

13- 8, 12

14. i 3> II

IS-

4, E G

W -f- I
16. o I 2, T Si^2
n

If two numbers converge, then they are identical:

r i4 6 X X= y

An immediate corollary of 7146, T'5137, a n d T 5 i3 8 is that a sequence

can converge to at most one number:
r 5 i4 7 C{n) ^ ^ A C(/i) ^ y -^ x y

Some sequences do not converge to any number; for example:

T i 48 r^Wxnox
^ n

Proofs concerning specific instances of convergence are simplified

by T 5 i 3 4 - r 5 i 3 9 ; for example:
n(n + i)
T'149 I. •Skew-------; ---- 0 1

2. I O I T 5 13 6
n

3- -0 0 Th S

4- 1 + - o 1 + o 2, 3, T S129
n «
Section 2 The Theory of Convergence . 319

4>

A9

8 ,7

5, 6, T S i^ o

The relation of convergence is not a congruence relation under the

operation of multiplication. For example,
I
-0 0
n »
and
non
n
yet it is not the case that

- *n o o • n
n ^
for if it were, we should have
I o o

However, if two sequences converge to numbers, their product converges

to the product of those numbers.
T 5150 I. ¿¡Aew-C(w) o X A D(w) o j y C(w) • D(«) o ^ j

2. C { n ) o x !.T > { n ) o y

Shew- Az{o < z WkAn[N{n) a ^ < w ->

\iC{n)‘ D { n ) ) - { x - y ) \ < z ] )
320 • Variable-Binding Operators Chapter IX

o < z

o <
2{\y\ + l)

A « j^N(n) A Ao < n ^

|C(n) - < 2, D S io , s, E l
2{]y\ + l)

O< 4 ' ■
2\x\ + Z

Anj^N(n)AA:x < w ->

|D(m) - 3»! < 2, D S io , 7, E l

2\x\+Z J
Shot» k o ^ k i - ^ W k A n [ N { n ) A k < n

|(C(w) • D(n)) - (jc-3;)| < z]

10 . ko ^ ki
11. Show A«[N(n) A < M ->

l( C ( « ) - D ( ^ ) ) - ( x - j) | < ^ ]

12 . N(n) Kk\ < It

13- |C(«) - Jc| < 10, 12, 6

2{\y\ + l)

14. \y • (C(n) - x)\ ^

2
\y\ 13, T'104, J '1 16
2{\y\ + l)
z
15- |D(«) - y\ < 12, 8
2|ic| + z

16. IC («)-(D («)-3.)|<

|C(n)| 15, T104, T 116

2\x\ + Z

17- 2 ^ 2(|>'l + l) T106

Z
18. < - 17. 4. ^95
2{\y\ + 1) 2
Section 2 The Theory of Convergence . 321

19. |:«| + |C(n) - ^1 < |;k| + - 13.18

2

20 . C(w) < |2:1 + - 19, T 112

2

21 . |C (n )-(D («) - j ) | <

7 ,2 0 ,1 6

22 . |C (n )-(D (« )-> .)! +

\ y . (C(n) - ;k)| <

(("•3 (5 ir .) l
14, 21
2(|j 1 + i )]
23- (C («)-D («))-(^-> ;) =
(C(n) . (D(«) - y ) ) +
(>>-(C(n) - x:)) A9
24. |(C (« )-D (« ))-( a;- j ^)1 <

23, T '112 ,22

2(|j| + l))
\y\
25- < I T106, T27,
Ij I + I
T80, T99
z z
26. \y\-------------- < - 4. 25
2{\y\ + 1) 2
I I 2\x\ + Z
2
27. f I+ - = — ;— T67, T68
2 2

28. 4, T106, 27
( '■ '• 3 (sVi)-:
29. | ( C ( « ) - D ( n ) ) - ( ^ - j) |<
z z
— I— 24, 26, 28
2 2
322 . Variable-Binding Operators Chapter IX

30. I |(C(w) • D(w)) - (.уу)! < g 29, Г65

3 1. WkAn[N{n)sk < n ^
\[C (n )-D {n ))-{x -y )\< z ] II, EG

32. Show Ao-> VAAw[N(w)aA < и ->

|(С(и) • D(«)) - (.r • j-)| < z]

Similar to derivation of line 9

33- VAA/i[N(m) л a < и

|(С (й )-В (« ))-(^ -3;)| < ^ ] 9» 32

34- С(и) • D(w) ^ X • у 3, D S io

T*Si5i and T S 1^2 have straightforward proofs, which are left to the
reader. The latter provides a means of expressing the convergence of two
sequences in terms of the convergence of a sequence to a number.
T5 i 5 i C{n)<^xKT>{n)<^y ^УkKn\N{n)^k < n ->C(w) ^ D(w)]
X^ y
T 5 15 2 C(w) о D (w )^ C(/i) — D (w )o o

The last two principles of this section depend very heavily on the
Continuity Schema. T*Si53 asserts that every nondecreasing sequence
whose terms have an upper bound converges to a number; T5 i 54 is
the analogue for nonincreasing sequences whose terms have a lower
bound, and can be derived simply from
TiSi53 I. -Shew-KnKm\N(n) a N(w) a h ^ m C{n) ^ C(m)] a

VyAw[N(w) C(w) ^ y] C(n) z

AwA/w[N(w) A N ( w ) A n ^ m

C{n) < C(w)] A

VyAw[N(«) -> C(n) ^ y]

VmVw[N(w) Am = C{n)] T 12 2
Skew-Wyhm{Wn[]Si{n) A m = C(«)]->
m ^ y)

Elementary, using 2nd conjunct of 2

Section 2 The Theory of Convergence . 323

Am(V«[N(M) a o t = C ( m)] ->■ m ^ 20) a

Ay[A»i(V«[N(«) A m = C(«)] -^m ^ y) 3,4, A S ig , E l
^Zo ^ y]
6. Skew-Az[o < z ->VAA/z(N(w) a /j < w->

\C{n) - zo\ < z)]

o < z
^Zo ^ Zo - Z
AmQ/n[N{n) Mfti = C(;z)] -> 5 (2nd con­
/w < 2:0 — junct)
Z{^ ^ z ^ - z
10 . '^(Vw[N(w) Amo = C(w)]->
mo < ;s:o - z) 8, 9, E l
11. N(wo) Amo = C(«o) A2To — -2: < mo 10, E l
12 . Aw(N(w) Amo <
|C(w) - 2:0! < z)

13* N(w) A Wo < w

14. C(wo) < C(w) .1 1 , 13, 2 (ist
conjunct)

15- zo ~ z < C(w) 1 1, 14

16. C(w) < zo 13, 5 (ist con­
junct)

17- C{n) < zo z 7 ,1 6

18. \C{n) - zo\ < ^ 15, 17, T iiS ,
7^115
19. VAAw(N(w) a k < n ->
|C(w) - 2:0! < z) 12, E G

20. Wz C(n) o z 6, r 5 i 43, E G

^' n

TiSi54 AwAm[N(w) AN(m) Aw < m ->C(m) ^ C(w)] a

VyAw[N(w) -^ y ^ C(w)] ->yzC {n)
324 • Variable-Binding Operators Chapter IX

3. A sketch of further developments. The theory T4, with its

definitional extensions, can be identified with what is generally called
differential calculus» We offer, in partial justification of this claim, defini­
tions of four of the most fundamental concepts in this subject.
If the sequence A(w) converges to a number Xy this number is called
the limit of A(n) as n approaches infinity. Accordingly, the mathematical
notion of a limit at infinity may be introduced by the following definitional
schema:

D Si 1 lim A(n) = lx A(n) o x

(A more customary notation for fiim A(«)* is ‘lim A(«)'.)

n n-*co

By 7*148 it is clear that the definite description on the right side of D S 11

is not always proper; but if the sequence represented by ‘A(n)’ converges
to some number, then by 7 5 i 4 7 the definite description is proper. The
basic theorems concerning limits at infinity have essentially been obtained
in the previous section; for example, T i4 4 , ^ 14 5 , and 7 5 i 5 0 can be
respectively expressed as follows:

w + I
lim -------- = I

lim - = o
w n
Wx C{n) o X A y y D(n) o y

lim [C(«) • D (« )]= [lim C(«)] ■ [lim D(«)]

Let 'A (xy represent any term of T4, for instance, \x^ 4* i)'. Then, for
each value of 'x' among the real numbers, the corresponding value of
‘A(^)’ will also be a real number. It may happen that as the values of ^x*
approach a fixed number /, the corresponding values of ‘A(;c)’ will also
approach some number. In this case, we denote the latter number by

lim A(jc) ,
x -*l

which is read ‘the limit of A(.v) as x approaches l\ Thus, for example,

choosing / as 2 and ‘A(;c)’ as the term + i)’, we have

lim (jc2 + i) = 5 .
X-^2
Let us examine the situation somewhat more carefully. W e say that
a number u is the limit of A(:c) as x approaches I if the values of ‘A(jc)’ can
be brought as close to u as we wish by bringing the values of ^x" sufficiently
Section 3 A Sketch of Further Developments . 325

close to / (without actually taking / itself as a value for ‘at'). Thus, within
T4, the notion can be characterized as follows:
(i) lim A(:x:) = luAz{o <

y y [ o < y h / \ x { o < \x — l \ ^ \ x — l \ < y - ^ |A(^) — w| < jsr)])

Strictly speaking, the formula (i) is not a possible definitional schema.

For simplicity, in section 2 of this chapter we excluded free variables
from appearing in definitional schemata, and in (i) the variable 7 ’ is free.
But the effect of (i) can be achieved by the following definitional schema,
which introduces a constant 7 im’ of degree <0, i, 2, o > :

D S 12 lim A(jc) = ’\uKz(o < z->

x^B(x)
Wy[o < y A Ax{o < \x — B(o)| A 1^ — B(o)| < y -> |A(^) — u\ < 2t)])
Then (i) is an instance of D S iz and can be used instead of the latter in
developing the mathematical theory of limits at a point.
To say that A(;e) is continuous at a number a is to say that the number
A(a) is the limit of A(:v) as x approaches a ; recalling (i), we see that the
condition of continuity can be expressed as follows:

(2) Cont [A(:v), a] hz{o < z

Wy[o < y A Ax(p < \x ~ a\ A \x — a\< y ^ |A(:v) — A(iz)| < 2r)])

Thus we regard ‘Cont’ as an operator of degree <1, i, 2, o ) . On the basis
of (2) it is a simple matter to show that, for example,
Aa Cont \z • x d\ ,

Cont
X
&»] •
Cont [Ty[I(^) Ajy = l) V { '^ l(x ) A y = o)], a] <r-> 1(a) .

Let us for the moment identify real numbers with instants of time. If
A{x) is understood as the distance at the instant x oi 2 certain object
from an initial position, and the object is assumed to be moving in a
straight line, then the derivative of A(:v) at a (or Der [A(;v), a]) will
be defined in such a way as to be the velocity of the object at the
instant a. To arrive at a definition of the derivative, we consider first
the familiar procedure for computing average velocities: the average
velocity of an object, between two instants a and 6, is the distance traveled
in the interval (which may be positive or negative depending on the
direction of motion), divided by the elapsed time. Thus we can charac­
terize average velocity as follows:
326 . Variable-Binding Operators Chapter IX

A{b) - Aja)
(3) Av [A(a:), a, b\ =
CC b —a
here ‘A v’ is to be an operator of degree <o, i, 3, o>. The velocity at an
instant a is simply the limit of the approximations obtained by taking
average velocities over smaller and smaller time-intervals surrounding a.
Thus v^e have the following characterization of the derivative (which in
this case is interpreted as the instantaneous velocity):
(4) Der [A(jc), d\ = lim Av [A(^), a — a + K\\ ;
X h-^0 X
here ‘Der’ is an operator of degree <0, i, 2, o ). There are of course
other interpretations of the derivative, depending on the interpretation
assigned to ‘A(:v)’. Many of them have the common feature that
Der [A(:c), d\ is the rate of change of the values of A(jc) at the instant a.
The formulas (2 )-(4 ) correspond, like (i), to definitional schemata,
which we may call D S i^ -D S i^ , By T5 we shall understand the definitional
extension of T4 obtained by adding the instances of D S ii - jD 5 i 4, and
by Te the definitional extension of T5 obtained by adding the instances
of D S i^ .
The passage from differential to integral calculus is achieved by adding
to Te the general notion of finite summation, that is, the operation that
associates with any numbers A(o), . . . , A(n — i), the sum of A(o),
. . . , A(n — i). This sum is usually denoted by
A(o) 4- . . . + A(« - i)
or
n -l
2 A (i) ;
i=0

our notation will be officially

(5) 2 * ” A(i) ,
and informally

(6) Z A (0 .
X
We shall regard ‘2 ’ as a constant of degree (p, i, 2, o ) ; thus the expres­
sions (5) and (6) are terms in which the variable ‘f ’ is bound and the variable
‘w’ is free. ‘2 ’ may most conveniently be introduced by two new axioms,
neither of which has definitional form. Accordingly, T7, or integral
calculus, is to be that extension of Te which is obtained by adding ‘2 ’ to
the constants of Te and all instances (within the present theory) of the
following schemata to the axioms of Te;
0
A S20 2 M f) = o
Section 3 A Sketch of Further Developments . 327
w+ 1 Y n 1
A Szi N ( « ) - ^ 2 A(0 = [ 2 A(t)J + A(«)

A S20 and A S 2 1 constitute the usual so-called ‘recursive definition’

of ‘S ’, but it should be observed that T? is not a definitional extension of
Te. ‘2 ’ couldy however, have been introduced by definition. In other
words, there is a definitional extension of Te having A S20 and A S 2 1
as theorem schemata (see Montague [3]). But the construction of the
necessary definitions is rather involved, and relies on properties of the
real numbers deeper than those considered here.
Before turning to the integral, it is convenient to introduce notation
for the least upper bound and the greatest lower bound of a set of real
numbers; we use
sup F(;v)

and
inf F(^) ,

after the Latin ^supremurrC and Hnjimum\ to denote respectively the least
upper bound and the greatest lower bound of the set of all objects having
the property F. (The reader will recall from chapter V III the conditions
under which such bounds exist.) We regard ‘sup’ and ‘inf’ as operators
of degree <0, 1 , 0, i > and introduce them by the following definitional
schemata:
D S16 sup F(^) = l;2r[A:v(F(:Jc) x ^ z) h
X

Ay(A*[F(jc) ^ j ] -> sr ^ >')]

Z)5 i 7 inf F(jc) = 1s:[A*(F(a:) - ^ z x) k
X
Ajv(Aa:[F(^) ^ 5t)]
By Tg we shall understand the definitional extension of T7 obtained by
adding the instances of D S 16 and D Si^,
One purpose of the integral is to assign numerical measures to such
areas as that shaded below.

FIGURE I
328 • Variable-Binding Operators Chapter IX

In this figure the line marked ‘X ’ is called the X-axis. Each of its points
is identified with a real number, in particular, with the distance (in some
fixed system of units) of that point from the point marked ‘O’ . The curve
lying above the X-axis is assumed to satisfy the following description: for
any point X of the X-axis, the length of the segment joining x to the curve
and perpendicular to the X-axis is the number A(jc). In this sense we may
speak of the curve as given by the term ‘A(;c)’ . The shaded area may
then be described as that enclosed by the X-axis, the perpendiculars to it
at the points a and 6, and the curve given by *A(jc)’ .
An approximation to the shaded area can be computed in the following
way: divide the segment of the X-axis between a and b into three equal
parts. For each of these parts, construct the largest possible rectangle
with that part as base which lies completely beneath the curve. Then
compute the total area of the three rectangles so obtained. In this way
we arrive at the doubly shaded area in the following figure.

FIGURE II

a b ^

In the light of the familiar definition of area for rectilinear figures, it

is seen that the doubly shaded area is given by the number
b - a , f ^ b - al
(7) ------- * inf V;c A(jc) = y A a ^ x A x ^ a - \ - -------- +
3 2/ L 3 J
6 — . . . r[ . , . b —a 6 - a]
A(x) = V A a H---------- ^ XÁX ^ a -h 2 • --------- +

b —a , 3 3 J
b - a b — al
3 y
[
------- • inf V jc
A(x) = y A a + 2 ----------- ^ x A x ^ a + ^ • -----------

Now this approximation is not very close,

3 3 J by
but it can be improved
increasing the number of subdivisions of the interval between a and b (and

FIGURE III
Section 3 A Sketch of Further Developments • 329

hence of inscribed rectangles). For instance, if we consider six subdivi­

sions, we arrive at the approximation in Figure i i i .
On the model of (7) above, and with the aid of ‘S ', we can express in
general the area of the inscribed approximation resulting from n sub­
divisions; it will be the number
n
b —a

?(
f b - a b — ar\
inf yx I A(vV) = jy A a + i --------- < , x ^ x < , a + { i + i ) ----------- I j .
y

In a completely analogous way, we can obtain a family of circumscribed

rectilinear approximations to the desired area. For instance, in the case
of four subdivisions, we obtain the dotted area below.

F I G U R E IV

The general expression for the area of the circumscribed approximation

resulting from n subdivisions is

b —a

[ b —a 6 — a] \
A(.x) = y h a - V i • -------- ^ < a + (/ + i) • ------- I .
n n \]

It is intuitively clear that the shaded area, if it exists at all, is at least

equal to each of its inscribed approximations and at most equal to each
of its circumscribed approximations. Thus, if z is the desired area, we
have:

(8) An|^N(n)A I < n ^

i

, b - a b - a-\\
a + + i ) * ------- I^ Z A
n n ]/
330 • Variable-Binding Operators Chapter IX

/b - a T a / x , b - a
^ > I ----------sup V X A(^) = V A a + i • -------- ^ XA
^ \ n y l n
I
X ^ a + {i + i) • jj

N ow it turns out that in case A(;c) is continuous at every number between

a and b (and even in certain additional cases) there will be exactly one
number z satisfying (8). The shaded area can then be defined as this
number; accordingly, we arrive at the following characterization of the
integral of K{x) between a and 6, which we identify with this area.

(9) J A(;v)d:v = ’\zSn |^N(«)a I ^ ^------infV^ j^A(:v)=3;

, b - a X * “ ^ 1\
a + t • --------< ^ A ^ < a + ( i + i ) ‘ --------- J I ^ z A
n
/b — a r ^ b —a
^ ^ N I ------- . sup V jc A(:v) = jy a a + z •------ ^ xa
n y I n

X ^ a + { i + i ) ’ — jj
Here we regard the composite symbol

J
as an operator of degree <o, i, 3, o > ; on the left side of (9), *a* and 'b*
are free, and ^x^ is bound. A s usual, (9) can be replaced by a definitional
schema in standard form; *a* and *b' would be replaced by, say,
and W e should thus arrive at a definitional extension of Tg, and
hence of integral calculus, in which the usual Riemann integral is available.

4. * H isto rical rem ark s. The metamathematical notions of this and

the last chapter— definition^ theory, definitional extension, and the like—
constitute an extension of the treatment in Tarski, Mostowski, Robinson
[i] beyond the domain of theories with standard formalization, in partic­
ular, to theories containing variable-binding operators. The present
treatment was developed in collaboration with Professor Dana Scott,
and will also appear (along with its model-theoretic rationale) in Montague,
Scott, Tarski [1].
For a discussion of fields and ordered fields, see, for instance, van
der Waerden [1].
The Continuity Schema is due to Dedekind [i], and the present for­
mulation of the theory of real numbers to Montague [2]. The theory
Section 5 Appendix: List of Constants in Chapters VIII and IX . 331

of real closed fields (see Tarski [3]) is equivalent to the theory axiomatized
by our A i - Ai^y together with A S ig . The addition of the predicate T '
(denoting the set of integers), together with axioms governing it, makes
a significant difference: as Tarski has shown, there is an automatic pro­
cedure for determining whether a formula is provable in the theory of
real closed fields; but this is not true of the. theory of real numbers, in
view of Church [1]. A theory closely related to the integral calculus of
the foregoing section is presented informally, without emphasis on its
logical basis, in Landau [1].
It should not be supposed that, because the Principle of Archimedes
(T54) is provable in the theory of real numbers, every model of this
theory is what algebraists call an Archimedean field; indeed, A. Robinson
has shown in [i] that the class of Archimedean fields cannot be charac­
terized by any first-order theory (that is, any theory having the logical
basis described in this book).
The definition of the integral given in the foregoing section, together
with its intuitive justification, was strongly influenced by a conversation
with Dr. J. D. Halpern.

5. Appendix: list of constants used in chapters VIII and IX.

CONSTANTS DEGREE INTRODUCED

<1, 0, 0, I> p. 4
<1, 0, 0, 2> P- 4
A <1, 0, 0, 2> P- 39
V <1, 0, 0, 2> P- 39
<-> <1, 0, 0, 2> P- 39
A <1, I, 0, I> p. 86
V <1, I, 0, I> p. 86
= <1, 0, 2, 0> p. 213
1 <0, I, 0, I> P- 233

OPERATION AND
PREDICATE L ETTERS
A <0, 0, 0, o > p- 3 0 9
B <0, 0, 0, o > p- 3 0 9
C <0, 0, I, 0 > p- 3 1 3
Cl < 0, 0, I , 0 > p- 3 1 3
D <0, 0 , I, o > p- 3 1 3
Di <0, 0, I, o > p- 3 1 3
E < 0 , 0, I, o > p- 3 1 3
F <1, 0, 1, 0> p. 287
332 • Variable-Binding Operators Chapter IX

OPERATION SYMBOLS DEGREE INTRODUCED ON

+ <0, O, 2, 0> P- 279
<o, O, 1, o> P- 279
<o, o, o, o > P- 279
<0, O, 2, 0> P- 279
-1 o>
< 0 , O, I , P- 279
1 <0, o, o, 0> P- 279
2 <o, o, o, o> P- 296
2 <o, O, I , o> P- 296
<o, o, I , o> P- 296
<0, O, 2, 0> P- 296
<0, O, 2, 0>- p. 296

PREDICATES

<1, o, 2, o> P- 279

I < 1 , o, I , o> P- 286
<1, O, 2, 0> P- 296
< < 1, O, 2 , 0 > P- 296
N < 1, O, I, 0 > P- 301
R < 1, O, I , 0 > P- 301

VARIABLE-BINDING
OPERATORS
0 <1, I, 2, 0> p. 310
lim <0, I, I, o> P- 324
lim <0 , I, 2 , 0 > P- 325
->
Cont < 1, I, 2, 0> P- 325
Av <0, I, 3. o> p. 326
Der <0, I, 2, 0> p. 326
S <0 , I, 2, 0> p. 326
sup <0, I , 0, I> P- 327
inf <0, I , 0, I> P- 327

id <0, 1, 3> o> P- 330

6. Appendix: list of axioms, definitions, and theorems of chap­

ters VIII and IX.

AXIOMS:

Ai a: + (j + 2:) = (;v 4- j) + sr
Az X + y y + X

A3 X o = X
Section 6 Appendix: Axioms, Definitions, Theorems; Chaps. Vlll and IX • 333

X + —X = 0
As x - { y z) = { x -y ) - z
A6 X •y = y • X
A 'j X *I = X
A8 ^ X = o ^ X • x~^ = I

Ag X • (y + z) = (x • y) -h (x • z)
A io ^ 0 = 1
A ll o ^ XV o ^ —X
A i 2 ' - ^X = 0 - > ' ^ 0 ^ X V ' - ^ 0 ^ —X

A 13 O^XAO^y-^O^X-hy
A14. O ^ X A O ^ y O ^ X •y

A is X ^ y<-> o ^ y -h —X

A 16 I(o)
A iy I(^) -> I(jc + l) A I(jc + — i)

A 18 I(.x:) A I ( j ) A X ^ y A y ^ X + I -> y = X Vy = X + I

A S ig VxF(x) AVyAx(F(x) -> x ^ y) ->

Vz[Ax(F(x) -> x ^ z) A Aj (Aa;[F(jc) - > x ^ y] -> z ^ y)]
0
A S20 2 A (0 = o

n+l rn T
A S21 N(n) ^ 2 A(i) = [ 2 A ( i ) J + A(«)

DEFINITIONS (in the form given in chapter IX ) :

BSi
D S2 A < B ^ A < B a ~A = B
DS2 2=1 + 1
DS4. A2 = A • A
D S$ |A| = lz[(o < A as: = A)v ( ~ o < A ax = —A)]
DS6 A - B = A + -B

D Sj — = A • B -i
' B
DSS N (A)^-^I(A) ao ^ A
DSg R (A )^ Y y V 3i[I()') a I(s ) Ao<siA~o = 2rAA=j - « - i]
334 • Variable-Binding Operators Chapter IX

D S io A(n) oB(w)'«-^/\z{o < z VMw[N(«)hk < n

|A(n) - B(n)l < 3])
D Si I lim A(w) = lx A(n) o x

D Si2 lim A{x) = luAz{o < z ->Vy[o < y k Ax {o < |;c — B{x)\
x-*B{x)
A |x — B(a;)| < y -> |A(a;) — m] < s)])
Z)5i3 Cont [A(:c), B(jc)] As:(o < z ^ Vj[o < y a Ax (o <
* |:e - B(^)l A |;c - B(:c)| < ^ |A(;r) - A(B(ic))| < s)])
A(C(o)) - A(B(o))
D Sh A v [A(x ), B(^), C(^)] = ^ J
a: C(o) - B(o)
D S i¡ Der [A(a:), B(;c)] = lim [A?; [A(x), B(^) — h, B(^) + A]]
X ft-*0 X
D S i6 sup F(x’) = 1s;[A^(F(;e) -> x z) a
^ A>'(Ají[F( íc) ^ 3^] -> a ^ jy)]
D5 i 7 inf F( aj) = l2:[Aa;(F(jc) ^ a;) a
Ay(A*[F(a:) ^ j ^ a;] -> y ^ ar)]

D S i8 J A(a;) (1a; = 1^A«I N(n)

r a i ^
^/ C(o) - B(o)
^ Í ---------------•
B(a:)

r . C(o) - B(o)
infVAC A(A’) = V a B ( o) + t • ----------------- ^ XI
y I n
C(o) - B(o)
X ^ B(o) + (i + i)
]) ^ Zh
Z ^2 ( ‘^ ^ ° * ~ ° ‘ °> su p V » [A (« )= y » B (o ) + .■■
i ^

THEOREMS

Tl x + z = yA-z->x = y
Tz z-\-x=z-\-y-^x = y
Tz r^ Z = O A X ‘ Z = y - Z - ^ X = y
Ta ~^ = o a s - a; = z 'y X= y
T$ — 0 — 0
Section 6 Appendix: Axioms, Definitions, Theo
T6 •0 = 0

Tl X + —y = Oir^ X = y
T8 X = -y<r->y = — JC

T9 x = o < - > —x = o
Tio ---- X = X

Til - ( a; + = —X H--- y
Tl2 - X - y = -(A i-j)
Ti 2 --y = -( * -3 / )

th —X • —y = X • y
Ti s X= 0 ^( —
Ti 6 x - y = o<r^x = o y y = o
T1 7 X • y = 0 -^{x • y)~^ = x~^ •
T18 a: < :v

T19
T20 X ^ yA y^ Z -^ X ^ Z
T21 X ^ y yy ^ X
T22 0 ^ X•X
T23 0 ^ 1
T24 x^y->x + z^y-b z
T’25 x + z ^ y + z-> x^ y
T26 X ^yA Z ^W ->X + Z ^ y +
T27 y + I ^ y
T28 ^y^ y + -1
T'29 X ^ y ^ —y ^ —.V
7’30 —y ^ —X - ^ x ^ y
T31 X ^ 0 <-> 0 ^ —X
T22 O ^ X A y^O -^X -y^O
T33 X ^ y AO ^ Z -> X •Z ^ y •Z
T’34 X ^ y A Z ^ O - ^ y - Z ^ X - Z

T 3S 0 ^ Z A ^ Z = 0 -> 0 ^ Z" ' ^

T36 X • Z ^ y • Z AO ^ Z A Z = 0
T37 y - Z ^ X - Z A Z < , 0 A ^ Z = 0

T38 X = y ^ X ^ y
T39 l{y) A 0 ^ y A ^ > ^ y = 0 - > l ^
336 . Variable-Binding Operators Chapter IX
T^o l( y ) A y ^ 0A ~ y = 0 -* y ^ — I

TS^ i W xF{x) - > y x F { - x )

T S i2 V jcF( jc) a VyA:!c(F(.v) ->>’ ^ x) ^

V2[Ajf(F(;r) -^sr < a;) A A jv(A x '[F(a!:) -> jy ^ Ac] -> y < «)]

TS^3
A a:[F(ac) ^ I(a :)] a AAf(F(Af) ->jy ^ ac) a

A « ( A ac[F(ac) 2 ^ ac] -^ z ^ y ) -> F(jy)

T’5 4 4 A at[F(a;) I(Af)] A A a ;(F(a:) - ^ x ^ y) a

A2(AAc[F(Af) -> Af ^ 3■ ]-^y < s) ^ F(>-)

T S iS
F(o) A A a;[I(a:) a F( ac) -> F( a; -|- i ) a F( a; -|- -- 1)]->-

A a :[I(a :) - > F ( a :)]

T46 l{ x ) - ^ l{ - x )

T'47 I(AC) A I(y) ^ I(^ + y)

T^S l{ x ) A l( y ) ^ l{ x - y )

TS^g VAc[I(Air) A F( ac)] a AA:[I(Ar) A F( ac) ->F( a : + l) a F( ac H--- 1)]->

A ac[I(ac) ^ F ( aa)]

T 50 I(.v) -i» Vy(I(j) A a: = j + j) V Vj(I(j) A a: = j 4- J 4- 1 )

T 51 I(Ai) -> ~ y > ’(I(j) AA: + Ar = >'-)-J-l- 1 )

T 52 I(a :) ^ [Y)’(I(j) A AC = J -H y ) i r - * ~y>-(I(>') A X = y + y + 0]

T 53 o < a:a ~ X= 0 Vn(I(n) Ay ^ tt- x )

T S i
^ x == 0 Vn(I(tl) A y ^ n • x)

TSS
I(«) A I(ffi) Ati < m ->n + I ^ m

T5(>
X — X = 0

T 57 - [ x - y) = y - X

TsS X - y {z ~ y) - {z - x)

T 59 (a ; - j ) 4- (sr - H>) = (^ + s;) - (y -h w)

T 60 (a : -I- j)2 = a ;2 -H 2x y 4- y -

T 61 l2 = I
T 62 (aC - y Y = Af2 - 2x y 4-

T 63 {x •yY = x^ • y^
, .0 -,o
/ x y ^
T 64

X X
T 6s o 0
Section 6 Appendix: Axiom s, D efinitions, T h e o re m s; Cl

0
T66 - = 0
X

X -h y X y
T6^
z z z

xy y
T68 xz ^ 0 — = -
xz z

/X\ —X / X' \ ^
T6g y 0 ^ - l-\ = ---- A - - = —
\y/ y 1 - y

T jo
T71 0 ^ x"^

T72 X ^ 0 - > o < x^

T73 X ^ X + y^

T’74 0 < I

T75 I < 2
T76 X < X

T77 X < y A y < Z ~ ^ X < Z

T7Z X ^ y Ay < Z ^ X < Z

T79 X < y A y < ^ Z ~ ^ X < Z

TSo X ^ y<^ y < X

T 81 X y ^ x < y s X = y

T82 X < y < r ^ X - \ - Z < y - \ - Z

TS3 X < y A Z < W - ^ X - \ - Z < y - \ - w

T84 0 < X -> X < z x

T85 X < y < r -^ — y < —X

T86 X < 0 <r-> 0 < —X

T87 0 < Z 0 < z~-

T88 2 < 0 -> < 0

T8g 0 < Z ^ [ x < y < r -^ x z < y z ]

Tgo z < 0 - > [ x < y < -^ y z < x z ]

T91 0 < X A 0 < y - ^ [ x ^ y<r^y~^ < x-i]

T92 X < 0 A y < 0 - > [ x ^ y < r - > y -" ^ < x-^]
T93 0 < X A O < y - ^ [ x < y < r -> y ~ ^ < a: - 1]

794 X < 0 A y < 0 - ^ [ x < y < r -^ y ~ ^ < * - l ]

33^ • Variable-Binding Operators Chapter IX
z z
T95 0 < ; C A J C < j y A 0 < 5 r - > - < -
y X

z z
19 6 0 < X A X < y A 0 < Z - > - < -
y X
X y
T9I X^ y A O < Z ^ -
z z
X y
T98 X < y A O < Z -
z z
X X
T99 0 < X A X < y - > 0 < - A - < I
y y
T\oo O ^ y AX^ ^ y^ -> X^ y

T io i O^XAO<y->0^-
y
X
T102 O < X A O < y ->0 < -
y
r i0 3 WtAz[(o ^ X A Z = x ) v { ' - - > O ^ X A Z = — x ) ^ z = t ]

T 104 O ^ JC -> |;c| = ;c

Tio$ < o - > |jc| = — X

T106 o ^ \x\
Tioy \x\ = 0<r-^ X = o
T108 X < 1^1

Tiog —JC ^ \x\

T iio l-jcl = \x\
Till
T1 1 2 k + >'1 < \x\ + |>’|
T1 13 k - j| < k - ^1 + k - y\
7' i i 4 k - yl ^ ki + i:v|
T1 1 5 \x - y \ = \y - arl
T1 1 6 \x-y\ = \x\ • \y\

Ti i y y ^ O^
\y\
T ii8 |a ; — j l < z*-^ X — z < y h y < x + z
T S iig Va:F(jc) a A^[F(;c) -> N(;c)] -> V;c[F(ji) a f^y(F(y) < 3'])
Section 6 Appendix: A xiom s, D efinitions, T h e o re m s; Chaps. V III and I X . 339

TS120 A 4 N(^) a A>'(N(>')Ay < a; H-FCv)) -» F(*)]->

Aa:[N(a:) -^F(Af)]

T S 12 1 F(o) A Ax [N(^) F( a;) -> F(ac + i)] ^ Aaí[N(a;) -> F( ac)]

T 122 N(0)
T i22 N{x) -^N( ac+ i)

T iz \ N(;e) A N(:v) aac + i =3 ' + i ->a;=.3^

T iz$ N(a;) ->a; + I # o

T126 N(a:) a NÍj ) -> N(ac + y) a N(a: •y)

T izy I(a:) a V>i(I(y)) A Af2 = 2y) V>-(I(>') ^X = 2 y )

T128 R(a:) ^ x ^¥^ 2

T '1 2 9 O < X - ^ S y x --
T130 o < x < -*y y X = y'^
T1 31 X ^ y ^ - ^ ' ^ z x + sr2 = JV

«2+1
T1 32 -------- <s> n
n «
Tm -^ « + I <:>«
n
r 5 i 34 C{n) o D(w) <-> C(m) ^ D(w)

TSns Aw C(w) = Ci(w) A Aw D(w) = Di(w)

[C(w) D(w) <-> Ci(w) ^ Di(w)]

T S136 C(w)^C(w)

T S 137 C(w) ^ D(w) -> D(w) C(w)

TS138 C(w) <i> D(w) AD(w) o E(w) -> C(w) o E(w)

T S 139 C(w) o Ci(w) AD(w) o Di(w)

C(w) + D(w)oCi(w) + Di(w)

TS140 V/fAw[N(w) A k < n - > C(w) = D(w)] - > C(w) ^ D(w)

TS14.1 AwAm[N(w) A N(wг) A n < m E(w) < E(m)] a

Aw[N(w) -> N(E(w))] -> Aw[N(w) w < E(w)]

T Sl^2 [C(w) D(w)] AAwAtw[N(w) a N(m) aw < m -> E(w) < E(m)] a

A «[N («) N (E («))] -> [C (E («)) ^ D (E («))]

340 . V ariable-Binding O perators Chapter IX

T 5 i 43 C(w) o x <- ^ A z {o < z

WkAn[N{n) Ak < « -> |C(n) — x\ < 5r])

n + I
r i 44 ----- o I
n ^

Ths -0 0

Ti^e XO y -> X = y
n
TS14.7 C(n) <^XK C(w) <^y = y

T148 •'Wx n o X
n
n{n + l)
7149 o I
n
T S is o C(n) ^ x ^ D(n) ^ y - > C{n) • D (w ) ^ x - y

T S is i C(w) A D(w) o j aV M w[N ( w) A A < n - > C ( n ) ^ D (w )]->

71 71
X ^ y

T S is z C («) D(n)<-> C («) - D(n)

TS1 5 3 A n A »j[N («) A N(m ) a n ^ m ^ C(w) ^ C(»i)] a

VyAn[N(M) -> C(w) < -> V a C(n) o z

T5 i 54 AwA»i[N(n) a N(m) a « ^ m -^C{m) < C(w) ] a

YyAn[N(«)^>’ < C(w)] ->WzC{n)<>z
 BIBLIOGRAPHY

A c k e r m a n n , W.
[i] Solvable cases of the decision prohleniy Amsterdam, 1954.
B e h m a n n , H.
[i] Beiträge zur Algebra der Logik, insbesondere zum Entscheidungs­
problem, Mathematische Annalen, vol. 86 (1922), pp. 163-229.
B e r n a y s , P ., and S c h ö n f i n k e l , M.
[i] Zum Entscheidungsproblem der mathematischen Logik, Mathematische
Annalen, vol. 99 (1928), pp. 342 - 72.
B o o l e , G.
[1] The mathematical analysis of logic, London and Cambridge, 1847.
[2] An investigation of the laws of thought, London, 1854.
C a r n a p , R.
[1] Logische Syntax der Sprache, Vienna, 1934. English translation: New
York, 1937.
[2] Meaning and necessity, Chicago, 1947.
C h u r c h , A.
[1] An unsolvable problem of elementary number theory, American
Journal of Mathematics, vol. 58 (1936), pp. 345 - 63.
[2] A note on the Entscheidungsproblem, Journal of Symbolic Logic, vol. i
(1936), pp. 40 - 41 ; Correction, ibid,, pp. loi - 02.
[3] Introduction to mathematical logic, vol. i, Princeton, 1956.
Copi , I. M .
[i] Symbolic logic. New York, 1954.
COUTURAT, L.
[1] La Logique de Leibniz, Paris, 1901.
[2] Opuscules et Fragments inédits de Leibniz, Paris, 1903.
[3] Les Principes des Mathématiques, Paris, 1905.
D e d e k i n d , R.
[i] Stetigkeit und irrationale Zahlen, Braunschweig, 1872. English trans­
lation in Essays on the theory of numbers. La Salle, Illinois, 1901.
D e M o r g a n , A.
[i] Formal logic, London, 1847.
342 • Bibliography

E uclid
[i] Book X, T h e t h i r t e e n books of Euclid's e l e m e n t S y translated by SirT . L.
Heath, Cambridge, 1908; second edition, 1926.
F r e g e , G.
[1] Begriffsschrift, Halle, 1879.
[2] Über Sinn und Bedeutung, Zeitschrift für Philosophie und Kritik, vol.
100 (1892), pp. 25-50.
[3] Grundegesetze der Arithmetik, vol. i, Jena, 1893.
G e n t z e n , G.
[i] Untersuchungen über das logische Schliessen, Mathematische Zeit­
schrift, vol. 39 (1934 - 35), pp. 176 - 210 and 405 - 31.
G ö d e l , K.
[1] Die Vollständigkeit der Axiome des logischen Funktionenkalküls,
Monatshefte für Mathematik und Physik, vol. 37 (1930), pp. 349 - 60.
[2] Über formal unentscheidbare Sätze der Principia Mathematica und
verwandter Systeme I, Monatshefte für Mathematik und Physik, vol.
38, (1931), pp. 173-9 8 .
H e r b r a n d , J.
[1] Sur la théorie de la démonstration. Comptes Rendus des Séances de
VAcadémie des Sciences, vol. 186 (Paris, 1928), pp. 1274-76.
[2] Recherches sur la théorie de la démonstration (Travaux de la Société
des Sciences et des Lettres de Varsovie, Classe III, N0. 33, 1930,
128 pp.).

H i l b e r t , D ., and A c k e r m a n n , W.
[1] Grundzüge der theoretischen Logik, Berlin, 1928.
[2] Grundzüge der theoretischen Logik, second edition, Berlin, 1938.
[3] Grundzüge der theoretischen Logik, third edition, Berlin, 1949.
[4] Grundzüge der theoretischen Logik, fourth edition, Berlin, 1959.

H i l b e r t , D ., and B e r n a y s , P.
[i] Grundlagen der Mathematik, vol. i, Berlin, 1934.

J a s k o w s k i , S.
[i] On the rules of suppositions in formal logic, Studia Lógica, no. i
(Warsaw, 1934).

K a l m â r , L.
[i] Über die Axiomatisierbarkeit des Aussagenkalküls, Acta Scientiarum
Mathematicarum, vol. 7, 1934-35, pp. 222-43.

K l e e n e , s . C.
[i] Introduction to metamathematics, Princeton, 1952.

L a n d a u , E.
[i] Differential and integral calculus, New York, 1951.
 Bibliography . 343

L e i b n i z , G. W.
[i] Scientia Generalis. Characteristica XIX and XX, in Gerhardt’s Die
Philosophischen Schriften von G, W. Leibniz, vol. 7 (1890), pp. 228 - 35.
English translation in A survey of symbolic logic, by C. I. Lewis,
Berkeley, 1918, pp. 373 -7 9 and New York, i960, pp. 291 -97.
L u k a s i e w i c z , J.
[1] Elementy logiki matematycz?iej (Elements of Mathematical Logic),
Warsaw, 1929.
[2] Zur Geschichte der Aussagenlogik, Erkenntnis, vol. 5 (1935 - 36),
pp. Ill -31.

L u k a s i e w i c z , J . , and T a r s k i , A.
[i] Untersuchungen über den Aussagenkalkül, Comptes Rendus des Séances
de la Société des Sciences et des Lettres de Varsovie, Classe III, vol. 23
(1930), pp. 3 0 - 50. English translation: Tarski [4], Article IV.
M acC oll, h .
[i] The calculus of equivalent statements and integration limits. Proceedings
of the London Mathematical Society, vol. 9 (1877-78), pp. 9 - 2 0 and
177-86, vol. 10 (1878-79), pp. 16-28, vol. I I (1879-80), pp.
113-21.
M a t e s , B.
[i] Stoic logic (University of California Publication in Philosophy, vol.
26), Berkeley and Los Angeles, 1953 and 1961.
M o n t a g u e , R.
[1] On the paradox of grounded classes. Journal of Syjnbolic Logic, vol. 20
(1955), p. 140.
[2] Semantical closure and non-finite axiomatizability I, in Infinitistic
Methods, Proceedings of the Symposium on Foundations of Mathe­
matics, Warsaw, 1959, pp. 45 - 69.
[3] Deterministic theories, in Decisions, values, and groups, vol. 2, Oxford,
1963.
M o n t a g u e , R., and K a l i s h , D.
[i] Remarks on descriptions and natural deduction, Archiv fü r mathe­
matische Logik und Grundlagenforschung, vol.* 3 (1957), pp. 50-64;
vol. 3 (1957), pp. 65-73.
M o n t a g u e , R. , S c o t t , D. , T a r s k i , A.
[i] An axiomatic approach to set theory, Amsterdam, forthcoming.
P a g e r , D.
[i] An emendation of the axiom system of Hilbert and Ackermann for the
restricted calculus of predicates. Journal of Symbolic Logic, vol. 27
(1962), pp. 131-38*
P e a n o , G.
[i] Sul concetto di numero. Revista di matemática, vol. i (1891), pp.
87 - 102, 256 - 67.
344 • Bibliography

P e i r c e , C. S.
[1] Collected papers of Charles Sanders Peirce^ edited by Hartshorne and
Weiss, Cambridge, Mass., 1933; see various papers from 1870 to 1903
in vol. 3.
[2] On the algebra of logic: a contribution to the philosophy of notation,
American Journal of Mathematics, vol. 7 (1885), pp. i8o - 202.
[3] The logic of relatives. The Monist, vol. 7 (1897), pp. 161 - 217.
P o s t , E. L.
[i] Introduction to a general theory of elementary propositions, American
Journal of Mathematics, vol. 43 (1921), pp. 163 - 85.
Q u i n e , W. V.
[1] A system of logistic, Cambridge, Mass., 1934.
[2] Mathematical logic. New York, 1940; revised edition, Cambridge, Mass.
(Harvard University Press), 1951.
[3] Methods of logic. New York, 1950; revised edition, 1959.
[4] From a logical point of view, Cambridge, Mass. (Harvard University
Press), 1953.
[5] A proof procedure for quantification theory. Journal of Symbolic
Logic, vol. 20 (1955), pp. 141-* 49-
R o b i n s o n , A.
[i] On the metamathematics of algebra, Amsterdam, 1951.
R o s s e r , B.
[i] Logic for mathematicians. New York, 1953.
R u s s e l l , B.
[1] On denoting. Mind, vol. 14 (1905), pp. 479 - 93.
[2] The theory of implication, American Journal of Mathematics, vol. 28
(1906), pp. 159-202.
[3] Introduction to mathematical philosophy, London, 1919.
S c h o l z , H.
[i] Metaphysik als strenge Wissenschaft, Cologne, 1941.
S c h r ö d e r , E.
[i] Algebra der Logik, vol. i, Leipzig, 1890.
SUPPES, P.
[i] Introduction to logic, Princeton, 1957.
T a r s k i , A.
[1] Über einige fundamentale Begriffe der Metamathematik, Comptes
Rendus des Séances de la Société des Sciences et des Lettres de Varsovie,
Classe III, vol. 23 (1930), pp. 22-29. English translation: Tarski [4],
Article III.
[2] Projecie prawdy w jezykach nauk dedukcyjnych (The concept of truth
in the languages of the deductive sciences) (Travaux de la Société des
Sciences et des Lettres de Varsovie, Classe III, no. 34, 1933, vii, 116
pp.). German translation in Studia Philosophica, vol. i (1936), pp.
261 - 405. English translation: Tarski [4], Article VIII.
 Bibliography. 345

[3] A decision method for elementary algebra and geometry, second edition,
Berkeley and Los Angeles, 1951.
[4] Logic, semantics, metamathematics, Oxford, 1956.
T a r s k i , A ., M o s t o w s k i , A ., R o b i n s o n , R. M.
[i] Undecidable theories, Amsterdam, 1953.
VAN DE R W a ERDEN, B. L .
[i] Modern algebra. New York, 1949.
W h i t e h e a d , A . N ., and R u s s e l l , B.
[i] Principia Mathematica, vol. i, London, 1910; 2nd edition, 1925.
 IN D E X OF P R O P E R NAMES

Ackermann, W., 38, 128, 178, 212 MacColl, H., 79

Archimedes, 293, 331 Mates, B., X, 37, 79
Behmann, H., 128, 178^ Montague, R., ix, x, 170, 212, 226, 264,
Bernays, P., 128, 178, 212, 230 327» 330
Boole, G., 79 Mostowski, A., 330
Carnap, R., 178, 230, 264 Pager, D., 128, 178
Church, A., X, 170, 178, 212, 331 Peano, G., 303
Copi, L, 38 Peirce, C. S., 37, 79, 178
Couturat, L., 79 Post, E. L., 79, 178
Dedekind, R., 330 Pythagoras, 303
De Morgan, A., 66, 79 Quine, W. V., 38, 178, 212, 230, 264
Euclid, 224, 303 Riemann, B., 330
Frege, G., 37, 79» 128, 230, 264 Robinson, A., 331
Gentzen, G., 38, 212 Robinson, R. M., 330
Godel, K., 174, 179, 230 Rosser, J. B., 264
Halpern, J. D., x, 331 Russell, B., 79, 128, 152, 178, 230, 258, 264
Herbrand, J., 38, 212 St. Anselm, 4
Hilbert, D., 38, 128, 178, 230 Scholz, H., 230
Jaskowski, S., 38 Schönfinkei, M., 128, 212
Kalish, D., ix, 170, 212, 264 Schröder, E., 79
Kalmar, L., 212 Scott, D., X, 330
Kaplan, D., x Suppes, P., 38
Kleene, S. C., 200 Tarski, A., X, 37, 38, 79, 128, 174, 226,
Landau, E., 331 330, 331
Leibniz, G. W., 52, 7 9 » 223, 230 van der Waerden, B. L., 330
Lukasiewicz, J., 37» 79 Whitehead, A. N., 79, 128, 178, 230, 258
 IN D E X OF SU B JEC T S

Abbreviation, 8-9, 92-93, 139, 216 Constant (of a theory), 279

Abbreviatory clauses, 57-58 Constituent (of a biconditional), 39
Addition, rule of, 47 Contradiction, law of, 54
Adjunction, rule of, 47 Continuity, 325
Alphabetic variance, 155-56, 249, 276-77; Continuity schema, 286, 330
immediate, 155-56, 248-49, 276-77; Contradiction: explicit, 192; law of, 54;
laws of, 1 12-13 truth-functional, 192
Antecedent (of a conditional), 4 Convergence, 310 if.; of a sequence to a
Antecedent line, 21 number, 316; of two sequences, 310
Apparent variable, 92, 138-39 Converse (of a conditional), 50
Archimedes, principle of, 293, 331
Argument, 13, 95; English, 24, 95; false Decision procedure, 208
English, 30; symbolic, 14, 95 Deduction theorem, 38
Aristotelian logic, 91, 113 Definition, 294; of an operation symbol,
Association, laws of, 51, 65 295; of a predicate, 294
Assignment, see Truth value Definitional extension, 295, 309
Average velocity, 325-26 Definitional schema, 308-09
Axiom (of a theory), 279 Degree (of a constant), 271-72
Axiom schema, 286 De Morgan’s law, 66
Derivability: of an English sentence from
Biconditional, 39 an English sentence, 17 1; of a symbolic
Biconditional-conditional, rule of, 47 formula from a class of symbolic for­
Bound occurrence: of a symbolic term, 136, mulas, 103, 279; of a symbolic sentence
237-3^» 275; of a variable, 89, 237, 274 from a class of symbolic sentences, 23
Bound symbolic term, 137, 237-38, 275 Derivation, abbreviated: in connection
Bound variable, 90, 237, 274 with formal theories, 278-79; in the
description calculus, 250-52; in the full
Cancellation, laws of, 280-82 quantifier calculus, 164-65; in the iden­
Closure: of an argument, 126; of a formula, tity calculus, 221; in the monadic quan­
126 tifier calculus, 106-07; in the sentential
Closure axiom, 172 calculus, 58-60
Commutation: laws of, 50, 65; principle of, Derivation, complete, 23
35 Derivation, forms of: conditional, 16-17;
Complete derivation, see Derivation, com­ direct, 16; generalized indirect, 192;
plete indirect, 17; universal, loi
Completeness, 179, 230, 264 Derivation, unabbreviated: in connection
Composition, principles of, 63, 66 with formal theories, 278-79; in the
Conclusion (of an argument), 13, 95 description calculus, 244; in the full
Conditional, 4 quantifier calculus, 149; in the identity
Conditional-biconditional, rule of, 47 calculus, 221; in the monadic quanti­
Conditional-disjunction, rule of, 65 fier calculus, 102-03; in the sentential
Confinement, laws of, 109-10 calculus, 20-21, 50
Conjunct, 39 Derivative, 326
Conjunction, 39, 72 Description, definite: 234; improper, 234;
Consequent (of a conditional), 4 proper, 234
Consistency (of a theory), 296 Description calculus, 244
 Index of Subjects . 349

Descriptive operator, 233 Instance, 58, 106, 160, 276; within a theory,
Descriptive phrase, 233 285
Descriptive term, 236; vacuous, 258 Integral, 330
Differential calculus, 324 Integral calculus, 326
Dilemma, principles of, 53, 63 Interchange, 251-52; of equivalents, 164-
Disjunct, 39; added, 48 65
Disjunction, 39, 72 Iterated proper substitution, 276
Distribution: laws of, 66, 103-04, 108-09;
principles of, 35, 52 Leibniz’ law, 223
Double negation: law of, 36; rule of, 15 Limit: at infinity, 324; at a point, 325
Logical constant, 273
Elimination column, i94~95
Equivalence relation, 313 Minimum principle, 301
Equivalent formulas, 104 Modus ponenSy 15
Equivalent terms, 248 Modus tollendo ponenSy 47
Equivalent theories, 296 Modus tollenSy 15
Euclid’s law, 224, 253 Molecular formula, 185
Excluded middle, law of, 66 Monadic formula, 15 1
Existential generalization, 152; rule of, 99, Monadic quantifier calculus, 101
148-49
Existential instantiation, rule of, 100 Name: of English, 85; symbolic, 137, 238
Exportation, law of, 51 Natural deduction, 38
Extension (of a theory), 285 Negation, 4
Extensionality, 215-16 Noninitial occurrence (of a quantifier
phrase), 201
Factor, principles of, 52 Nonlogical constant, 273
Fallacy, 30; of begging the question, 30; Normal form: conjunctive, 72; disjunctive,
of unwarranted assumptions, 30 72; prenex, 154
Finite summation, 326 Number: complex, 279; natural, 301;
Formula, 87, 134, 213-14, 235-36, 272; rational, 279, 301; real, 279
of English, 86; molecular, 185; monadic,
15 1; in the official sense, 90; symbolic, ‘Occurs among’ (defined), 21
88, 135, 214, 236; of a theory, 280 Operation letter, 132, 273
Formula-maker, 271 Operation symbol, 274
Free occurrence: of a symbolic term, 137, Operator, see Variable-binding operator
237-38, 275; of a variable, 89, 237, Overlay, 153
274
Free symbolic term, 137, 237-38, 275 Paradox: of grounded classes, 226; of the
Free variable, 90, 237, 274 liar, 226; Russell’s, 152
Peirce’s law, 37
Generalization, 152; existential, 88; uni­ Phrase: of connection, 39; of description,
versal, 88 234; of quantity, 86
Praeclarum theoremay 52
Idempotence, laws of, 54, 62 Predicate, 274
Identity calculus, 220 Predicate letter, 86, 133, 273
Identity formula, 213 Premise (of an argument), 13, 95
Identity sign, 213 Prenex argument, 199
Improper descriptions, rule of, 243“ 44 Prenex form, 201
Indiscernibility of identicals, 223 Prenex formula, 193
Individual constant, 274 Prenex normal form, 154
Induction; for integers, 291; strong, 302; Proof, 35, 103
weak, 302 Proper descriptions, rule of, 242
Inequality, laws of, 284 f. Proper substitution, see Substitution
Inference rules, 14; of description, 242-44;
of identity, 220, 222-24, 253; for quan­ Quantifier: existential, 86; universal, 86;
tifiers, 99-100, 106, 148-49; sentential, vacuous, 1 1 1 - 12
IS» 47) 48, 64-65, loi Quantifier calculus: full, 149; monadic, loi
Infimunit 327 Quantifier-free formula, 194
350 . Index of Subjects

Quantifier negation, io6 Symbolization [continued]

Quantifier phrase: existential, 86; univer­ English sentence, 11- 12 ; interesting, 77
sal, 86 Symmetry: law of, 221 ; rule of, 222
Recursive enumerability, 200 Tautological implication, 76, 186
Reductio ad absurdum, 18; laws of, 37, 53 Tautological validity, 186
Reflexivity, law of, 221 Tautology, 74, 186
Relativization: of a symbolic argument, Term, 134, 235-36, 272; of English, 13 1;
173,261; of a.symbolic sentence, 172, 260 symbolic, 134, 236; of a theory, 280
Repetition, rule of, 15 Term-maker, 271
Russell’s paradox, 152 Theorem: of logic, 34, 103; of a theory,
280
Satisfaction, 85 Theorem schema, 287
Schema, see Axiom schema; Continuity Theory, 279; of commutative ordered
schema; Definitional schema; Theorem fields, 279; of convergence, 310 ff.; of
schema real numbers, 286
Scheme of abbreviation, 9 Transitivity: law of, 221; rule of, 222
Scope indicator, 264 Translation (into English): free, 10; literal,
Sentence, 6, 40; basic, 72; of English, 5, 9, 42, 93, i39“ 40» 216, 239; of a symbolic
85; in the official sense, 41; symbolic, 7, argument, 170
40, 90; of a theory, 280 Transposition, principle of, 36
Sentence letter, 5 Triangular inequality, 300
Sentential calculus, 47 Truth-functional contradiction, 192
Sentential connective, 39, 274; binary, 44 Truth-functional expansion, 124-25
Separation of cases, 64 Truth-functional incompatibility, 192
Signs, laws of, 282 Truth table, 74
Simple ordering, 283 Truth value, 73; assignments of, 73, 185;
Simplification, rule of, 47 with respect to an assignment, 73, 186
Stylistic variant, lo -ii, 43-44, 91-92,
140 ff., 217, 240; canonical, 44 Uniform replacement, 58
Subformula, 186 Universal generalization, 88
Substituend, 157 Universal instantiation, rule of, 99, 148-49
Substitution: of a formula for a predicate
letter, 1 57 ff.; of a term for an operation Validity: of an English argument, 25; of a
letter, 159-60, 250; of a term for a vari­ symbolic argument, 24
able, 148; of a variable for a variable, 99 Variable, 85; of generalization, 99, lo i; of
Summation, finite, 326 instantiation, 99, 100
Supremunty 327 Vacuous descriptive term, 258
Syllogism, principle of, 35, 51 Vacuous quantification, 111-12; laws of,
Symbolization, 24; of an English argument, 1 1 1- 1 2
24; of an English formula, 94; of an Variable-binding operator, 274
Í" ■ > '% i v - » . * Ä * * ' '''

J 4 II -> к Л »0 *В »^ •»

I- f 1 ' i 'X f .
'.Ьи!""т- tlfe '..:.# 1 11 \^ l

'^1^ '., lÿ»;

;, i' , ' f -П Ii"-' Í 1 i"lí . .И " ' " п' Щ ^Ji0.15-6611SO.i