ON TRUTH

Episteme
A SERIES IN THE FOUNDATIONAL,
METHODOLOGICAL, PHILOSOPHICAL, PSYCHOLOGICAL,
SOCIOLOGICAL, AND POLITICAL ASPECTS OF THE SCIENCES, PURE AND APPLIED

VOLUME 16

Editor: Mario Bunge

Foundations and Philosophy of Science Unit, McGill University

Advisory Editorial Board:

Raymond Boudon, Sociology, Masion des Sciences de I'Homme, Paris

George Bugliarello, Engineering, Polytechnic Institute of New York
Bruno Fritsch, Economics, E. T.H. Zurich
Ivan T. Frolov, Philosophy and Social Sciences, USSR Academy of Science, Moscow
Erwin Hiebert, History of Science, Harvard University
Philip Kitcher, Philosophy, University of California, Davis
Nicholas Rescher, Philosophy, University of Pittsburgh
Michael Ruse, Philosophy and Zoology, University of Guelph
Raimo Tuomela, Philosophy, University of Helsinki
Hao Wang, Mathematics, Rockefeller University, New York
Paul Weingartner, Philosophy, Salzburg University

The titles published in this series are listed at the end of this volume.
 FRANK PLUMPTON RAMSEY

On Truth
Original Manuscript Materials (1927-1929)
from the Ramsey Collection
at the University of Pittsburgh
edited by
NICHOLAS RESCHER
University of Pittsburgh, U.S.A.

and
ULRICH MAJER
Georg-August University, Gottingen, Germany

..
SPRINGER SCIENCE+BUSINESS MEDIA, B.V
'Library of Congress Cataloging-in-Publication Data

Ramsey. Frank Plumpton. 1903-1930.

On truth I Frank Plumpton Ramsey edited by Nlcholas Rescher and
Ulrich Majer.
p. cm. -- CEpisteme ; v. 16)
ISBN 978-94-010-5662-5 ISBN 978-94-011-3738-6 (eBook)
DOI 10.1007/978-94-011-3738-6
1. Truth. 1. Rescher. Nicholas. II. Majer. Ulrich. III. Title.
IV. Series: Episteme <Dordrecht. Netherlands) ; v. 16.
B1649.R254052 1990
121--dc20 90-40393

ISBN 978-94-010-5662-5

Printed on acid-free paper

Ali Rights Reserved

1991 Springer Science+Business Media Dordrecht
Originally published by Kluwer Academic Publishers in 1991
Softcover reprint of the hardcover 1st edition 1991
No part of the material protected by this copyright notice may be reproduced or utilized in
any form or by any means, electronic or mechanical, including photocopying, recording or by
any information storage and retrieval system, without written permission from the copyright
owner.
 CONTENTS

Preface Vll

Editor's Introduction IX

ON TRUTH

Introduction: Logical Values 3

I. The Nature of Truth 6
Appendix: Two Variant Elaborations
Appendix 1 to Chapter I 17
Appendix 2 to Chapter I 21
II. The Coherence Theory of Truth 25
1. Coherentist Arguments Against Absolute Truth 25
2. A Further Objection from the Philosophy of Science 33
3. Arguments for the Coherence Theory Refuted 36
III. Judgment 43
IV. Knowledge and Opinion 55
V. Judgment and Time 67

APPENDIX I: OLDER DRAFT VERSIONS

Intrc:1uctory 81
I. The Nature of Truth 84
II. The Coherence Theory of Truth 95
III. Judgment 98
V. Judgment and Time 103

APPENDIX II: SUPPLEMENTAL MATERIAL

1. The Nature of Proportions (1921) 107

2. "On Justifying Induction": Paper to the Society (1922) 120
3. "The Long and Short of It" 124
Name Index 129
 FRANK PLUMPTON RAMSEY
"On top of Red Pike in the Lake District" 1925 (Photo by Letitia Ramsey).
 PREFACE

The present publication forms part of a projected book that F. P. Ramsey

drafted but never completed. It survived among his papers and ultimately
came into the possession of the University of Pittsburgh in the circumstances
detailed in the Editor's Introduction. Our hope in issuing this work at this
stage - some sixty years after Ramsey's premature death at the age of 26
- is both to provide yet another token of his amazing philosophical creativity,
and also to make available an important datum for the still to be written
history of the development of philosophical analysis. This is a book whose
appearance will, we hope and expect, be appreciated both by those interested
in linguistic philosophy itself and by those concerned for its historical
development in the present century.
 EDITORS'INTRODUCTION

1. THE RAMSEY COLLECTION

Frank Plump ton Ramsey (22 February 1903 -19 January 1930) was an extra-
ordinary scholarly phenomenon. Son of a distinguished mathematician and
President of Magdalene College, Cambridge and brother of Arthur Michael,
eventual Archbishop of Canterbury, Ramsey was closely connected with
Cambridge throughout his life, ultimately becoming lecturer in Mathematics
in the University. Notwithstanding his great mathematical talent, it was
primarily logic and philosophy that engaged his interests, and he wrote original
and important contributions to logic, semantics, epistomology, probability
theory, philosophy of science, and economics, in addition to seminal work
in the foundations of mathematics. His original editor spoke the unvarnished
truth in saying that Ramsey's premature death "deprives Cambridge of one
of its intellectual glories and contemporary philosophy of one of its profoundest
thinkers,"l and J.M. Keynes characterized one of his papers as "one of the
most remarkable contributions to mathematical economics ever made."2
Considering the scope and variety of his achievements, it is astonishing to
contemplate that Ramsey died Uust) before attaining his 27th birthday.
In 1982 the University of Pittsburgh acquired through the mediation of
Nicholas Rescher a substantial collection of Ramsey's manuscripts, consisting
of notes, lectures, and various unfinished writings. The collection was
augmented in 1986 by a gift of further autograph material from Ramsey's
daughter, Mrs. Jane Burch, who generously vested the publication rights to
the Ramsey manuscripts in the University of Pittsburgh. With these subsequent
additions, the collection currently (1988) comes to some 120 items amounting
to some 1200 holograph pages. This Ramsey Collection forms part of the
Archives of Scientific Philosophy in the Twentieth Century housed in the
University's Hillman Library, and bears an impressive silent witness to the
astonishing versatility and intellectual power of this English prodigy.
The Ramsey Collection consists of manuscripts of variable time of origin,
x EDITOR'S INTRODUCTION

length, and value. The earliest manuscripts are notes and essays dating [rom
his undergraduate days; the latest represent substantial writings on which
he was still working at the time of his death. The shortest manuscripts are
brief notations and memoranda of only a few lines; the longest are substantial
chapters or essays ranging in length up to a manuscript of 45 pages devoted
to issues on The Foundations of Mathematics, which was published under that
title as the lead essay in the Braithwaite anthology of Ramsey's work. 3 The
least significant papers are notes and abstracts on books Ramsey was reading;
the most valuable are drafts or even virtually completed versions of researches
that Ramsey himself destined for publication.
Topically, the collection covers the whole diverse range of Ramsey's
extensive interests, ranging from logic, probability, and the foundations of
mathematics to the theory of knowledge and the philosophy of language
and beyond to issues of economics and social philosophy. The overall spectrum
has a rather wider range than is represented in the Ramsey material published
to date (there is, for example, an essay on sex, sexuality, and perversion).
But even when the same ground is covered, the contentions and views put
forward in the manuscript often supplement and sometimes differ significantly
from those that have seen the light of print.
Although the manuscripts only rarely carry notations of date, they can
in most cases be dated at least approximately, even without using physical
evidence of paper or ink. Internal evidence apart, Ramsey often worked in
interaction with the published literature and treats issues whose topicality
can be dated from other sources.
The manuscripts present no graphological difficulties. They are written
in a good English school hand, without use of special notations or abbreviations.
The distribution and ordering of the manuscripts in the elegant cloth-bound
boxes in which most of them reached the University of Pittsburgh is neither
chronological or topical. To some extent, the material was arranged by genre,
but no very systematic order was established. Ramsey himself was certainly
not responsible for this arrangement; such order as there is is due to those
who collected his scattered papers together after his death.

2. THE TRUTH PROJECT

The most substantial unpublished item in the Ramsey Collection relates to

a projected book on Truth and Probability. It is a portion of this project that
is at issue in the present publication.
The manuscript contents of Box I of the Ramsey Collection come to
210 foolscap pages, most of them tightly filled with Ramsey's flowing script.
They differ from the materials contained in the other boxes in that they
 EDITOR'S INTRODUCTION Xl

deal with a connected and clearly interrelated way with one overarching
question: the nature of truth. Unlike the other materials in the collection,
they constitute a single large-scale thematic unit.
The box contains eighteen folders of manuscripts in all. Two of them
(nos. 10 and 16) consist of brief memoranda. The rest, however, constitute
a unifed whole. The thematic of the material and its complex interrelationship
is set out in Appendix A to this Introduction.
The overall project bears the title "Truth and Probability," as Ramsey
himself indicated in the Alpha table of contents. For Ramsey presented several
somewhat divergent table-of-contents sketches for the project, which are set
out in Appendix B. The latest of them (Delta) agrees with the manuscript
material as we have it, except for one missing final chapter on "The Object
of Judgment. "
The development of Ramsey's overall project was foreshadowed in the
1926 essay "Truth and Probability," subsequently published by Braithwaite,
whose table of contents stood as follows:
1. The Frequency Theory
2. Mr. Keynes's Theory
3. Degrees of Belief
4. The Logic of Consistency [= Probability Logic]
5. The Logic of Truth [= Inductive Logic]
As these rubrics indicate, this range of themes largely coincides with the
second (probabilistic) part of the projected work on Truth and Probability in
the version of the Alpha table of contents, which is roughly contemporary
(1926127). The book was presumably planned to extend the materials of this
essay into new areas (chance, causality), as well as adding a wholly new
initial treatment of truth, judgment, and logic.
It is thus clear that Ramsey initially contemplated a very extensive project
- a substantial book on Truth and Probability with an initial section on semantics
and epistemology, a midsection on logic, and a final section on probability
and induction. However, with the passage of time, and the unfolding realization
of the initial parts of the project, its character changed. The initial, semantic-
epistemological section expanded from three chapters into six. The mid-section
on logic also expanded substantially (compare Alpha with Gamma) and
increasingly assumed a life of its own as Ramsey came to project a logic
book for which we also possess a projected table of contents (see MS 002-
22-03). And with the expansion of the first two sections, the third section
on probability, which had given the initial impetus to the whole project,
was left aside. 4 The newer interests became absorbing and displaced the old.
Ramsey's original project On Truth and Probability thus evolved into two: a
book On Truth and a treatise on Logic.
XlI EDITOR'S INTRODUCTION

What we have in Box I, then, is an effectively complete version of Ramsey's

book On Truth, except for a missing sixth chapter on "The Object ofJudgment."
Additionally we have (in two versions) the opening discussion of the Logic
book, and this has been included here because it is eminently suitable as
an introduction to Ramsey's overall project.
In relation to the survey of the Box I material given in Appendix A,
the principal offerings of the present publication are as follows.
1. The aforementioned Introduction in its later version.
2. Chapters I, II and III in their later versions.
3. Chapter IV, in its sole versions.
4. Chapter V in its later version.
To this material we append also the earlier versions of all of the preceding
items, with the exception of Chapter IV, which exists only in one single
version and, of course the missing Chapter VI on "The Object of Judgment."
However, this "missing chapter" is really no mystery. It was unquestionably
intended to deal primarily with "propositional reference" and its materials
were absorbed into the revised version of chapter III, "Judgment", which
was now no longer qualified as Preliminary, and which discussed "the object
of judgment" (viz., propositional reference) at considerable length. It is thus
probable that the On Truth material as we have it represents an effectively
complete, albeit unpolished version of Ramsey's book.
The project On Truth is far and away the most substantial item in the
Ramsey Collection that has not yet been published. This failure is not without
its reasons. The state of the manuscript itself indicates that Ramsey's thought
on matters in the book was developing rapidly, with the result that his ideas
about the book were also in a state of flux. In the end, Ramsey's first editor,
R.B. Braithwaite, who prepared Ramsey's unpublished papers for the press
soon after his death at the invitation of his widow, decided against publishing
this obviously unfinished material on the ground that "he <Ramsey> was
profoundly dissatisfied with it, and the preliminary matter that remains is
quite unsuited for publication" (Braithwaite, pp. xiii-xiv). However, with
the wisdom of two generations of hindsight it is clear that the value of
this material is far greater than was apparent at the time and its preliminary
character nowadays seems far less of an obstacle. Moreover, Ramsey's own
dissatisfaction with the material- motivated by some radical further transitions
in his own philosophical development - is less significant for us than it was
for his friends. What matters for us are not his putative feelings about his
manuscripts but the obvious importance and interest of his ideas.
 EDITOR'S INTRODUCTION X111

3. INTELLECTUAL DEVELOPMENT

In point of time, Ramsey's work on the project On Truth fell during the
period from 1927 to early 1929. The manuscript itself makes various references
to material published in 1927 and 1928, and has for its latest citation the
paper on "The Growth of the Perception of the External World" by H.W.B.
Joseph in the January 1929 issue of Mind (see Chapter IV, footnote 15). Further
evidence for this dating is given by the close kinship between the issues
treated here and those treated in the "Fact and Propositions" paper of 1927
on the one hand and the so-called "Last Papers" of 1929 on the other (both
published by Braithwaite).
In both On Truth and in "Facts and Propositions" Ramsey defends his
redundancy theory of truth: "It is true that Caesar was murdered" means
no more than that Caesar was murdered, and "It is false that Caesar was
murdered" means that Caesar was not murdered" ("Facts and Propositions,"
p. 42). Indeed, Ramsey's commitment to this theory dates from his under-
graduate days and forms an ongoing Leitmotiv of his thought. 5 But in On
Truth Ramsey went on to maintain that the redundancy theory is not only
compatible with a correspondence theory of truth but actually constitutes
the heart and core of such a theory.
Internal evidence marks that "On Truth" as something of a way-station
in Ramsey's transit from his early logicism via a pragmatic position towards
the version of intuitionism he favored at the end of his brief life. For despite
its close relationship to the 1926 "Truth and Probability" essay published
by Braithwaite, the 1927-28 project On Truth betokens a significant shift
in Ramsey's thought. In "Truth and Probability" Ramsey viewed induction
as a useful habit which 'is reasonable because the world is so constituted
that inductive arguments lead on the whole to true opinions" (p. 197). (Note
that Ramsey thus foreshadows Braithwaite's own inductive justification or
vindication.) In On Truth, however, he accepts the pragmatic standpoint more
extensively and sees his own theory of truth as supporting a Peircean
justification of induction. Thus On Truth stands far closer to Ramsey's 1927
publication "Facts and Prospositions" than to his 1926 essay on "Truth and
Probability.' ,
The project On Truth shows the beginnings of Ramsey's conversion to
finitism and intuitionism as manifested in the "Last Papers." In "The Foundation
of Mathematics" (1925) and "Mathematical Logic" (1926), Ramsey endorsed
the logicist standpoint of Russell and Wittgenstein which regarded universal
and existential statements as propositions and specifically as infinite conjunctions
or disjunctions of particular statements. This is a view with which Ramsey
is now no longer satisfied, and which he later rejects explicitly in "General
Propositions and Causality" (1929). In On Truth Ramsey inclined to the
XIV EDITOR'S INTRODUCTION

"holistic" view of Norman Campbell and Heinrich Hertz that theories (and
the universal and existential statements they comprise) are not to be seen
as propositions subject to the tertium non datur principle of having to be either
true or false. Rather, they are "secondary systems" (to use Campbell's expression),6
propositional superstructures that are only capable of highly indirect veri-
fication, and are best characterized in terms of probability rather than truth-
status. In this respect, then, On Truth marks a break between the earlier
logicist Ramsey of "Facts and Propositions" and the later intuitionist Ramsey
of the "Last Papers. "

4. HISTORICAL CONTEXT

The 1927-1929 period when Ramsey wrote the material On Truth was not
a propitious time for a philosophical treatment of this subject. Between the
work of Russell and Wittgenstein in the period up to the close of W orId
War I and the publication in 1935 of Alfred Tarski's monumental "Der
Wahrheitsbegriff in den formalisierten Sprachen," the field belonged pred-
ominantly to the representatives of neo-Hegelian idealism. In the 1920's, truth
was simply not a theme that engaged the best philosophical minds of the
day. This circumstance, combined with the difficulty of producing, and indeed
even of seeing the need for constructing, a conceptually appropriate definition
of the expression "true statement" such as Tarski was to produce in 1935
was not as yet appreciated. People were still inclined to regard as mere
puzzles the semantic antinomies for whose resolution such a definition was
needed. Those few theorists who appreciated the significance of the issues
were inclined to see them as insuperably difficult. In Cambridge, a G.E.
Moorean attitude prevailed, anticipated by the thesis of Frege's Logical
Investigations: "It thus appears likely that the content of the word "true" is
altogether unique and indefinable." The new-model mathematical philosophers
did not yet see the elucidation of truth as a pressing issue and inclined to
leave the matters to the more traditional philosophers (then principally
idealists). Against this background it is less surprising that Braithwaite did
not appreciate the importance of Ramsey's project than that Ramsey himself
was impelled to pursue it with such vigor and commitment.
It is in fact remarkable how close Ramsey came to anticipating Tarski's
theory of truth - in spirit if not in letter:

We can say that a belief is true if it is a belief that p, and p. This definition
sounds odd because we do not at first realize that "p" is a variable sentence
and so should be regarded as containing a verb; "and p" sounds nonsense
because it seems to have no verb and we are apt to supply a verb such
 EDITOR'S INTRODUCTION xv

as "is true" which would of course make nonsense of our definition by

apparently reintroducing what was to be defined. (P. 13 of MS 001-02.)

The fundamental resemblance with Tarski's definition of truth accordingly

consists in the circumstance that Ramsey too proceeds by way of contextual
definition according to which "is true" applied to a proposition (belief, claim,
etc.) is equated with its assertibility pure and simple: "Suppose a man believes
that the earth is round, then this belief is true if and only if the earth is
round." In this particular regard, Ramsey - like T arski himself - follows
the lead of Frege who was the first to mention the redundancy thesis to
the effect that "is true" is seen as invariably eliminable in all contexts of
its application to propositional objects since any such application is simply
tantamount to asserting the proposition itself.?
To be sure, Ramsey's definition is more primitive than Tarski's in applying
only to propositional objects and leaving aside the difficult issue of propositional
functions (and thus of quantificational theory in general). And beyond this
strictly technical difference there is also a deeper philosophical difference
that should be noted.
For Ramsey, the analysis of truth begins rather than ends with its definition.
For as he sees it the pivotal question of the range of objects to which the
characterization "is true" can apply must also be addressed, and one must
ask just what it is that beliefs, opinions, claims, statements, and the like have
in common in virtue of which "is true" can be predicated of them. This
line of thought led Ramsey to the idea of "propositional reference" and to the
problem of its clarification. A great part of the early chapters of the On
Truth project were devoted to this issue.
Where Tarski rests content with remarking the equivalence of the sentence-
predication "p is true" with the objective assertability of 'p' itself, Ramsey
thus pressed on into an inquiry to clarify the very idea of assertability and
to determine the range of objects over which the variable at issue ranges.
And he approached the question "what constitutes propositional reference?"
by way of the conception of correspondence with fact - at least as a first
approximation. However, he saw this as a matter that needs further clarification
and refinement. And in this regard he considered the pragmatic theory of
truth as less a rival than a supplement to a correspondence approach. In
consequence, the discussion of On Truth shows that Ramsey regarded such
pragmatic factors as simplicity, definiteness, utility as crucial for clarifying
the basis for claiming the truth of scientific contentions when entire theories
rather than mere observation claims are at issue.
On the view of science which Ramsey shared with Hertz and Campbell,
scientific theories are not propositions but complexes replete with condi-
tionalized and hypothetical propositional commitments whose assertibility
XVI EDITOR'S INTRODUCTION

conditions we cannot verify through (inevitably limited) observations, and

whose links to cognitive1y manageable reality proceeds less through their
verifiability than through their utility for the production of verifiable predictions.
Clarifying the way in which the idea of truth bears on such theories is something
that cannot be achieved by logic alone but requires the development of a
theory of induction, an inductive logic. This hard-gained conviction led Ramsey
on the one hand to push his truth project out beyond the standard range
of logical and linguistic philosophizing and on the other hand confronted
him with a task of such magnitude that its completion, even in preliminary
form, required a labor which, massive effort notwithstanding, was simply
unachievable given the time constraints imposed by an uncooperative fate.
But when everything is said and done, there is no question in our minds
about the importance of Ramsey's book. Since it remained unpublished, it
obviously exerted no historical influence. But its ideas are of the highest
historical and systematic interest. Ramsey's articulation of a redundancy theory
of truth greatly surpassed what had gone before and pointed the way towards
what lay ahead. His theory of "propositional reference" opens up semantical
teaching that has still not been adequately explored even now. And his critique
of the Coherence Theory of Truth as propounded by the English Idealists
- deeper and more serious than Russell's - is still among the most insightful
and probing evaluation of the position that we have.
However, the real value of Ramsey's work lies not in its critical but
in its constructive contributions. In his work in semantics and logic, Ramsey
both projected and realized a high ambition: "to show how truth and
reasonableness can be defined without assuming any unique unanalyzable logical
relations, and further to show that the current modes of explaining logic
by means of such relations as altogether untenable."8

5. EDITORIAL PRELIMINARIES

The printed text presented here reproduces the substance of Ramsey's

manuscript, but departs occasionally from its letter, as follows:
1. When Ramsey has struck a passage out, we generally include it, duly
placed within square brackets [ ], when this could conceivably provide
useful information for an interpreter.
2. We occasionally interpolate conjectural omissions, always indicating
this by placing them in pointed brackets < >.
3. Ramsey is totally unsystematic in using the single and double quotes.
We regularly use single quotes for single words and double quotes
elsewhere.
4. In many instances, we have filled in Ramsey's incomplete footnote
references.
 EDITOR'S INTRODUCTION XVll

5. We have occasionally emended Ramsey's punctuation and sometimes,

though rarely, even his paragraphing.
Throughout such changes, it has been our guiding principle to preserve Ramsey's
text intact while enhancing its readability. No printed version can present
a draft manuscript exactly as it stands, but we have tried to keep the clang
of editorial machinery from intruding too loudly - although in a project
of this sort its noise cannot be eliminated altogether.
The editors are grateful to David Carey, Laurie Eck, and especially Christina
Masucci for their help in checking the accuracy of the transcription from
Ramsey's manuscript.

NICHOLAS RESCHER
ULRICH MAJER
XVlll EDITOR'S INTRODUCTION

NOTES

1. R.B. Braithwaite, Editor's Introduction to Frank Plumpton Ramsey, The Foundation of

Mathematics and other Logical Essays (London, 1931), p. ix.
2. Quoted by Braithwaite, ibid, p. x.
3. The Foundations of Mathematics and other Logical Essays by Frank P!umpton Ramsey, edited
by R.B. Braithwaite (London: Routledge & Kegan Paul, 1931). The essay was subsequently
reissued in F.P. Ramsey, Foundations: Essays in Philosophy, Logic, Mathematics and Economics,
edited by D.H. Mellor (London, Routledge & Kegan Paul, 1978). This latter anthology
includes substantially the same material as Braithwaite's. (It deletes some discussions dealing
mainly with matters of probability and adds two important papers on economic issues).
4. To a considerable extent the material on probability is contained in "Truth and Probability"
(1926) which was read at the "Moral Sciences Club" and later published by Braithwaite.
5. The final page of MS 006-05-01 which contains a lecture on "The Nature of Propositions"
that Ramsey read to the Moral Sciences Club on November 18, 1921, at age 18, reads
as follows:

Lastly we come to truth, which I shall deal with as briefly as possible. The most certain
thing about truth is that 'p is true' and 'p', if not identical, are equivalent. This enables
us to rule out at once some theories of truth such as that "to be true" means "to
work" or "to cohere" since clearly 'p works' and 'p coheres' are not equivalent to
'p '. There are I think only three sensible theories of truth (1) that a true belief is defined
to be one which has a certain relation with a fact, (2) truth is indefinable and has
no connection with a relation between belief and fact, (3) truth is indefinable but as
a matter of fact true beliefs do have a certain relation to facts which false beliefs do
not have.

(3) I think we can dismiss; we have seen no reason to suppose there are facts and
if truth be indefinable, I think none can be drawn from the nature of truth; so if
truth be indefinable we have no reason to suppose there are facts and therefore no
reason for thinking true beliefs are related to facts in a way which false ones are not.
Of course in the special case of events we have seen that some true propositions and
therefore true beliefs are related to events in a way false ones are not, since false
beliefs have no quasi-subjects.

When he wrote this, Ramsey was still an undergraduate. (He took the Tripos examination
in June of 1923.)
6. Ramsey does not actually use Campbell's term until his 1929 paper on "Theories," but
already here he is drawn into the range of the "holistic" consideration of Campbell and
Hertz that constitute its natural habitat.
7. Frege's seeming self-contradiction in maintaining both the redundancy thesis and the
indefinability of truth is removed by the consideration that a merely contextual definition
is at issue which, as such, explains the use of the term rather than establishing its conceptual
content.
S. Introduction, 001-1S version, ad fin.
 EDITOR'S INTRODUCTION XlX

APPENDIX A

EXPLANATORY INVENTORY OF BOX 1 MATERIALS

RELATING TO THE BOOK ON TRUTH [AND PROBABILITY]

Folder Content Description Title Given by Ramsey

001-01 Older version of Chap. I Chapter I. The Nature of Truth [Preliminary

Considerations]
001-02 Newer version of Chap. I Chapter I
001-03 Older version of Chap. III

001-04 On some issues discussed in the

second half of MS 001-02
001-05 On some issues discussed
(first two towards the end of MS 001-02
pp.)
001-05 The coherence theory of truth
(rest)
001-06 Absolute vs. coherentist truth
001-07
Arguments for the coherent

001-08 1 theory refuted

001-09 Older version of Chap. II Chapter II. The Coherence Theory of Truth
001-10 Assorted notes and jottings
001-12 Chap. IV in its sole version Chapter IV. Knowledge and Opinion
001-13 Newer version of Chap. V Chapter V. Judgment and Time
001-14 Older Version of Chap. V Judgment and Time (The Problem of
Judgment)
001-15 A misconceived defense of the
coherence theory from the
philosophy of science
001-16 Assorted notes and jottings
001-17 Introduction to a treatise on Chapter I. Logical Values
logic
001-18 Older version of the preceding Chapter II. Introductory
introduction
xx EDITOR'S INTRODUCTION

APPENDIX B

RAMSEY'S SUCCESSIVE TABLES OF CONTENTS

ALPHA BETA GAMMA DELTA

(MS 001-01-01) (MS 001-15-02) (MS 001-01-02) (MS 001-15-02)

Truth Truth Truth Truth

Preliminary Preliminary
Truth and Truth and
Coherence Coherence
Judgment Judgment Judgment
Preliminary Preliminary
Knowledge and Knowledge and Knowledge and Knowledge and
Opinion Opinion Opinion Opinion
Theories of Judgment and Time Judgment and Time
Judgment
The Object of The Object of
Judgment Judgment
Representative Ideas Representation
Negation and Negation and Negation
Disjunction Disjunction
Disjunction
Generality Theories of
Generality
Formal Logic and Tautology and
Consistency Contradiction
Partial Belief
Definition
Universals and Particulars and
Particulars Universals
Existence Extistence
The Calculus of
Probability
Mr Keynes' Theory
of Probability
Identity and Number
Place of Formal
Logic and
Probability in
Theory of
KnowledgE
 EDITOR'S INTRODUCTION XXI

Appendix B (continued)

ALPHA BETA GAMMA DELTA

(MS 001-01-01) (MS 001-15-02) (MS 001-01-02) (MS 001-15-02)

Generality further
considered

Implication
Causality Causal Laws
Chance
Attempts to Justify
Induction
Nature of Degrees of Belief
Knowledge and of
Rational Belief
Scope of Inductive
Logic
[Theory of Statistics1

Yet another table of contents, which stands close to Alpha in its concern with probability,
is given in MS 002-22-01. It runs as follows:
1. Belief and Truth 6. Meaning of If
2. Terms 7. Causality
3. Logic and Consistency 8. Chance
4. Partial Belief 9. Justification of Induction
5. Keynes' Probability Relations 10. Rational Belief
ON TRUTH
 001-05--0/
1""~ 96v11TVV.> u.~"", Iv. c(1i,;. ... fi"3 rA.~ Ilk it, llii 10 111";;6
f/;,/Lt }lAtf~_o.iJ 1""-y.4. va., ..... to. rk 1___ MId}- Iv"Y ..... ~ "' .... ,u
we I<o~ r. ~l~$ ~I>\ _ H-!J>v .........1 I(e..ir.1 "'~I-',< ,,,.,, 4 ......
....r It" lie t. .. ltf"? ~.. ". e~F';" Ii ........ ,je.. ... l.e{ ' / t""", ""<.

'" )..J,?/ .. ~ Ik"f Iu: u i ... /~ (~J~rl,~ ... ;- "'''Y I<! SO va.)<&( ...,

hoI- 6.. tfA.;'IALIi~ <..t4,. fi......, <11 ;"'"H)'. B. ... r Ii... (,<, .. ,,1-1(-

)C_ F", ~ tt..... tt....'? cI)- v..~ '" ut.U H.I!. t.. (--- .A
~ W.,.....,6' ~l".",ko{ tf) (J&i~,-I7"'l,sl"~ ... -sy.tlt ... DJ' ""'~ J",dy'",,,tI'"
Iv"",Lc! ~c. l..jJ. v'<)"f. t.f.." it.. IIU. ... , , " ' , Ht{ .." II...)' " .. t""II-,<t
It,

/A,<f~ ~kL4 C\. lYl1it ... t.... ).. ..til. ") 1;.../1 I....... I..,~ cl.'-)t-~.Q-14,. ... ,'I.
~ ..... J~ ~ o~st. .,-~ Itt; J:~ Ir; 1t6$,,1.. lQ:" t;: ... t( c.~~ 'f"r/~i"
-.' . "'-')/oo.o.t--~
)< (',9 -Ail )'K.I't kO(":I)''' Utw,;dii..~ '1 '''-t./.. ')'-"-fAa.1 A ... s ~
'A-it 6Q.ft/ ~", ~ ~ /,=(';'4,"
V'Q.'~IUI$

TJu. 4\.g/tlt\Ul.tt 1V4id liiy do ;,,,-'/e11vb.,t/ I;"J"A"<";-~

It.. v.. t.11'>\ Ie", /(tIIt. S"-c,," ro 6c. ;u,..".
(~, n... fe" J t;.. ... ~ c.""'/'U-l /~ ic/i:vc.r.1r In-..It '< ... c(
u....lt..t.... 'Y (1- h<,Q. r J<:M ..r ~ c..4.1t ".,1" I'! 4.6 1l/... 4.~ tt..&;li,
t/;:o.l 14. RI.l J('c ....... ~ /u-t I. ..jv, 4. a: it! "<'
J IA {ll.<,!, ("-4 "" V\.6~1<J/
Jr...t4 ft.;"'J; ~t<.r tt~ ... ~J.CIl.~ c? tt.d41:-'~ de-q "of- AC\.~,
l~t>..l- !l ~ tt let I it w tt., r 6... J J. o/~ (f),;~ /i:q,t- -4 c/;f/,
6.~~(7- AAAh--.t/r-fi-.<.{., /t<A CU>1/w-l rOlf td Il.. <f!'),

Len-",. 1M.v>, ~ I~ cit/lie &" t- ).~ "'U w4u tUJ

/{tIlC ~ ,,<\(,

ItfA{- Ifu.., ""tL ~ ,.... c..l!c.l.... k Q. 6'.10 Itc. ~ L; t.tit 8<b.(\,!

 (001-17)

<INTRODUCTION>!

It is a commonplace that Logic, Aesthetics, and Ethics have a peculiar position

among the sciences: whereas all other sciences are concerned with the
description and explanation of what happens, these three normative studies
aim not at description but at criticism. To account for our actual conduct
is the duty of the psychologist; the logician, the critic, and the moralist tell
us not how we do but how we ought to think, feel, and act.
The triad of critical disciplines, Logic, Aesthetics, and Ethics correspond
to the three so-called fundamental values, truth, beauty, and goodness, but
the correspondence is by no means exact. For, whereas the chief question
in Ethics is undoubtedly "what is good?", and is Aesthetics "what is beautiful?",
the question "what is true?" is one which all the sciences answer, each in
its own domain, and in no way the peculiar concern of Logic. What Logic
studies is not so much the truth of opinions, as the reasonableness of arguments
or inferences. As the distinction is an important one it may be as well to
dwell on it.
Truth is an attribute of opinions, statements, or propositions; what exactly
it means we shall discuss later, but in a preliminary way we can explain
it as accordance with fact. (If a man thinks that Mr Baldwin is prime minister
he thinks truly because Mr Baldwin is in fact prime minister.) If we have
an opinion or statement by itself the most important point of view from
which we can criticise it is that of truth and falsity,2 and the proper person
to do this is not the logician but the expert on the particular matter with
which the statement deals. Opinions and statements however, generally occur
not by themselves, but as the result of some mental process, such as perception,
memory, inference, or guessing. The logician is concerned with the particular
method of forming opinions known as inference or argument, and the inferences
he approves of are not so appropriately called 'true', and 'valid', but 'sound',
or 'rational'. of course, since the whole purpose of argument is to arrive
at truth, there must be some relation between the soundness of arguments
4 INTRODUCTION

and the truth of opinions, but it is not easy to say exactly what the relation
is. It is not so simple as it would be if true opinions were never based on
unsound arguments, and inferences of high probability never led to error.
But whatever the relation may be, the fact remains that the primary subject
of the logician is inferences or arguments, not opinions or statements, and
his predicate of value is rationality not truth.
Nevertheless, before coming to his real point the logician is bound to
begin by preliminary investigations into the nature and forms of opinions
and statements, which must be conceded to belong properly to psychology
since they are concerned not with values but with the actual characters of
mental processes. Since, however, psychologists grossly neglect the aspects
of their subject which are most important to the logician, they are commonly
regarded as belonging to logic, and logic as the term is ordinarily used consists
to a great extent of psychology. In the same way, students of ethics and
aesthetics are obliged to undertake for themselves all sorts of psychological
preliminaries.
The three normative sciences: Ethics, Aesthetics and Logic begin, then,
with psychological investigations which lead up, in each case, to a valuation,
an attribution of one of the three values: good, beautiful, or rational, predicates
which appear not to be definable in terms of any of the concepts used in
psychology or positive science. I say 'appear' because it is one of the principal
problems of philosophy to discover whether this is really the case [whether,
that is to say, 'good', 'beautiful', 'rational' (or for that matter 'true') represent
undefinable qualities which ... ].
It is, of course, possible to take one view in regard to one kind of value
and the other view with regard to the other kinds; it could be held, for
instance, that whereas goodness and beauty could be defined in terms of
our desires and admirations, rationality introduced some new element peculiar
to logic, such as indefinable probability relations. But the arguments that
can be used are so much the same, that when the alternatives that can be
used are clearly stated, any normal mind is likely to make the same choice
in all three cases. It would be out of place to discuss goodness and beauty
in a book on 10gic,3 but it will be one of my chief objects to show that
the view, which I take of them, that they are definable in [ordinary factual]
natural terms, is also true of rationality and truth: so that just as ethics and
aesthetics are really branches of psychology, so also logic is part, not exactly
of psychology, but of natural science in its widest sense, in which it includes
psychology and all the problems of the relations between man and his
environment. 4 But this is not a matter which can be settled in advance: logic
tries to discover what inferences are rational; we all have some idea as to
what this means, but we cannot analyse it exactly until we have made
considerable investigations, which are commonly regarded as belonging to
5 INTRODUCTION

logic which is expected not merely to determine the application but also
the analysis of its standard of value.

NOTES

1. <Ramsey himself gave the rubric: Chapter I; Logical Values and in the earlier version Chapter I:
Introductory. That it was intended as the opening of the Logic book is evidenced by Ramsey's
table of contents for this project.>
2. Besides this we can of course criticize its verbal expression as felicitous or otherwise,
and we can say that it is important or trivial.
3. For an excellent treatment of these ideas see Mr LA. Richard's Principles of Literary Criticism
(London, 1924). See also Professor G.E. Moore's Principia Ethica (Cambridge, 1903).
4. [Logic cannot be altogether contained in psychology, because the soundness of our thought
depends on that thought agreeing with its object, and hence in part on the properties
of the object, to an extent not in general regarded as belonging to psychology.J
 (001-02)

CHAPTER I

<1. WHAT IS TRUTH?>

What is truth? What character is it that we ascribe to an opinion or a statement

when we call it 'true '? This is our first question, but before trying to answer
it let us reflect for a moment on what it means. For we must distinguish
one question, "what is truth?", from the quite different question "what is
true?" If a man asked what was true, the sort of answer he might hope
for would either be as complete an enumeration as possible of all truths,
i.e., an encyclopaedia, or else a test or criterion of truth, a method by which
he could know a truth from a falsehood. But what we are asking for is
neither of these things, but something much more modest; we do not hope
to learn an infallible means of distinguishing truth from falsehood but simply
to know what it is that this word 'true' means. It is a word which we
all understand, but if we try to explain it, we can easily get involved, as
the history of philosophy shows, in a maze of confusion.1
One source of such confusion must be eliminated straight away; besides
the primary meaning in which we apply it to statements or opinions, the
word true can also be used in a number of derived and metaphorical senses
which it is no part of our problem to discuss. Obscure utterances such as
"Beauty is truth, truth beauty" we shall make no attempts to elucidate, and
confine ourselves to the plain work-a-day sense in which it is true that Charles I
was beheaded and that the earth is round.
First we have to consider to what class of things the epithets 'true' and
'false' are primarily applied, since there are three classes which might be
suggested. For we use 'true' and 'false' both of mental states,2 such as beliefs,
judgments, opinions or conjectures; and also of statements or indicative
sentences; and thirdly according to some philosophers we apply these terms
 THE NATURE OF TRUTH 7

According to the philosophers who believe in them, it is these propositions

which are true or false in the most fundamental sense, a belief being called
true or false by an extension of meaning according as what is believed is
a true or a false proposition. But in as much as the existence of such things
as these propositions is generally (and to my mind rightly) doubted, it seems
best to begin not with them but with the mental states of which they are
the supposed objects, and to discuss the terms true and false in their application
to these mental states, without committing ourselves before we need to any
doubtful hypothesis about the nature of their objects.
The third class consisting of statements or indicative sentences is not a
serious rival, for it is evident that the truth and falsity of statements depends
on their meaning, that is on what people mean by them, the thoughts and
opinions which they are intended to convey. And even if, as some say,
judgments are no more than sentences uttered to oneself, the truth of such
sentences will still not be more primitive than but simply identical with that
of judgments.
Our task, then, is, to elucidate the terms true and false as applied to
mental states, and as typical of the states with which we are concerned we
may take for the moment beliefs. Now whether or not it is philosophically
correct to say that they have propositions as objects, beliefs undoubtedly
have a characteristic which I make bold to call propositional reference. A belief
is necessarily a belief that something or other is so-and-so,3 for instance that
the earth is flat; and it is this aspect of it, its being "that the earth is flat"
that I propose to call its propositional reference. So important is this character
of propositional reference that we are apt to forget that a belief has any
other aspects of characters at all, and when two men both believe that the
earth is flat we say they have the same belief, though they may believe
it at different times for different reasons and with different degrees of conviction
and use different languages or systems of imagery; if the propositional references
are the same, if they are both "beliefs that" the same thing, we commonly
ignore all other differences between them and call them the same belief.
It is usual in logic to express this resemblance between the two men's
beliefs not by saying as I do that they have the same propositional reference,
but by calling them beliefs in the same proposition; to say this is not however
to deny the existence of such a character as propositional reference, but
merely to put forward a certain view as to how this character should be
analysed. For no one can deny that in speaking of a belief as a belief that
the earth is flat we are ascribing to it some character; and though it is natural
to think that this character consists in a relation to a proposition; yet, since
this view has been disputed, we shall start our inquiry from what is undoubtedly
real, which is not the proposition but the character of propositional reference.
We shall have to discuss its analysis later, but for our immediate purposes
8 CHAPTER I

we can take it without analysis as something with which we are all familiar.
Propositional reference is not, of course, confined to beliefs; my knowledge
that the earth is round, my opinion that free trade is superior to protection,
any form of thinking, knowing, or being under the impression that - has
a propositional reference, and it is only such states of mind that can be either
true or false. Merely thinking of Napoleon cannot be true or false, unless
it is thinking that he was or did so and so; for if the reference is not propositional,
if it is not the sort of reference which it takes a sentence to express, there
can be neither truth nor falsity. On the other hand not all states which have
propositional reference are either true or false; I can hope it will be fine
to-morrow, wonder whether it will be fine to-morrow, and finally believe
it will be fine to-morrow. These three states all have the same propositional
reference but only the belief can be called true or false. We do not call
wishes, desires or wonderings true, not because they have no propositional
reference, but because they lack what may be called an affirmative or assertive
character, the element that is present in thinking that, but absent in wondering
whether. In the absence of some degree of this character we never use the
words true or false, though the degree need be only of the slightest and
we can speak of an assumption as true, even when it is only made in order
to discover its consequences. For states with the opposite character of denial
we do not naturally use the words true or false, though we can call them
correct or incorrect according as beliefs with the same propositional reference
would be false or true.
The mental states, [then], with which we are concerned, those, namely,
with propositional reference and some degree of the affirmative character,
have unfortunately no common name in ordinary language. There is no term
applicable to the whole range from mere conjecture to certain knowledge,
and I propose to meet this deficiency4 by using the terms belief and judgment
as synonyms to cover the whole range of [mental] states in question [although
this involves a great widening of their ordinary meanings] and not in their
ordinary narrower meanings.
It is, then, in regard to beliefs or judgments that we ask for the meaning
of truth and falsity, and it seems advisable to begin by explaining that these
are not just vague terms indicating praise or blame of any kind, but have
a quite definite meaning. There are various respects in which a belief can
be regarded as good or bad; it can be true or false, it can be held with
a higher or a low degree of confidence, for good or for bad reasons, in
isolation or as part of a coherent system of thought, and for any clear discussion
to be possible it is essential to keep those forms of merit distinct from one
another, and not to confuse them by using the word "true" in a vague way
first for one and then for another. This is a point on which ordinary speech
is sounder than the philosophers; to take an example of Mr Russell's, some-
 THE NATURE OF TRUTH 9

one who thought that the present Prime Minister's name began with B would
think so truly, even if he had derived his opinion from the mistaken idea
that the Prime Minister was Lord Birkenhead; and it is clear that by calling
a belief true, we neither mean nor imply that it is either well-grounded
or comprehensive and that if these qualities are confused with truth as they
are, for instance, by Bosanquet,5 any profitable discussion of the subject becomes
impossible. The kind of merit in a belief to which we refer in calling it
true can be easily seen to be something which depends only on its propositional
reference;6 if one man's belief that the earth is round is true so is anyone
else's belief that the earth is round, however little reason he may have for
thinking so.
After these preliminaries we must come to the point: what is the meaning
of 'true'? It seems to me that the answer is really perfectly obvious, that
anyone can see what it is and that difficulty only arise when we try to say
what it is, because it is something which ordinary language is rather ill-
adapted to express.
Suppose a man believes that the earth is round; then his belief is true
because the earth is round; or generalising this, if he believes that A is B
his belief will be true if A is B and false otherwise.
It is, I think, clear that in this last sentence we have the meaning of
truth explained, and that the only difficulty is to formulate this explanation
strictly as a definition. If we try to do this, the obstacle we encounter is
that we cannot describe all beliefs as beliefs that A is B since the propositional
reference of a belief may have any number of different more complicated
forms. A man may be believing that all A are not B, or that if all A are
B, either all Care D or some E are F, or something still more complicated.
We cannot, in fact, assign any limit to the number of forms which may
occur, and must therefore be comprehended in a definition of truth; so that
if we try to make a definition to cover them all it will have to go on forever,
since we must say that a belief is true, if supposing it to be a belief that
A is B, A is B, or if supposing it to be a belief that A is not B, A is not
B, or if supposing it to be a belief that either A is B or C is D, either
A is Bore is D, and so on ad infinitum.
In order to avoid this infinity we must consider the general form of a
propositional reference of which all these forms are species; any belief whatever
we may symbolise as a belief that p, where 'p' is a variable sentence just
as 'A' and 'B' are variable words or phrases (or terms as they are called
in logic). We can then say that a belief is true if it is a belief that p, and
pJ This definition sounds odd because we do not at first realize that 'p'
is a variable sentence and so should be regarded as containing a verb; "and
p" sounds nonsense because it seems to have no verb and we are apt to
supply a verb such as "is true" which would of course make nonsense of
10 CHAPTER I

our definition by apparently reintroducing what was to be defined. But 'p'

really contains a verb; for instance, it might be "A is B" and in this case
we should end up "and A is B" which can as a matter of ordinary grammar
stand perfectly well by itself.
The same point exactly arises if we take, not the symbol 'p', but the
relative pronoun which replaces it in ordinary language. Take for example
"what he believed was true." Here what he believed was, of course, something
expressed by a sentence containing a verb. But when we represent it by
the pronoun 'what' the verb which is really contained in the 'what' has,
as a matter of language, to be supplied again by "was true." If however
we particularize the form of belief in question all need for the words "was
true" disappears as before and we can say "the things he believed to be
connected by a certain relation were, in fact,S connected by that relation."
As we claim to have defined truth we ought to be able to substitute
our definition for the word 'true' wherever it occurs. But the difficulty we
have mentioned renders this impossible in ordinary language which treats
what should really be called pro-sentences as if they were pro-nouns. The only
pro-sentences admitted by ordinary language are 'yes' and 'no', which are
regarded as by themselves expressing a complete sense, whereas 'that' and
'what' even when functioning as short for sentences always require to be
supplied with a verb: this verb is often "is true" and this peculiarity of
language gives rise to artificial problems as to the nature of truth, which
disappear at once when they are expressed in logical symbolism, in which we
can render "what he believed is true" by "if P was what he believed, p".
So far we have dealt only with truth; what about falsity? The answer
is again simply expressible in logical symbolism, but difficult to explain in
ordinary language. There is not only the same difficulty that there is with
truth but an additional difficulty due to the absence in ordinary language
of any simple uniform expression for negation. In logical symbolism, for
any proposition sympol p (corresponding to a sentence), we form the
contradictory -p (or -p in Principia Mathematica); but in English we often
have no similar way of reversing the sense of a sentence without considerable
circumlocution. We cannot do it merely by putting in a "not" except in
the simplest cases; thus "The King of France is not clever" is ambiguous,
but on its most natural interpretation means "There is a King of France
but he is not clever" and so is not what we get by simply denying "The
King of France is clever"; and in more complicated sentences such as "if
he comes, she will come with him" we can only deny either by a method
special to the particular form of proposition, like "if he comes, she will
not necessarily come with him" or by the general method of prefixing "It
is not true that -", "it is false that - " or "It is not the case that -",
where [again] it looks as if two new ideas, 'truth' and 'falsity', were involved,
 THE NATURE OF TRUTH 11

but in reality we are simply adopting a round-about way of applying not

to the sentence as a whole.
Consequently our definition of falsity (to believe falsely is to believe p,
when -p) is doubly difficult to put into words; but to argue that it is circular,
because it defines falsity in terms of the operation of negation which cannot
always be rendered in language without using the word "false", would simply
be a confusion. "False" is used in ordinary language in two ways: first as
part of a way of expressing negation, correlative to the use of "true" as
a purely stylistic addition (as when "it is true that the earth is round" means
no more than that the earth is round); and secondly as equivalent to not
true, applied to beliefs or other states of mind having propositional references
or derivatively to sentences or other symbols expressing those states of mind.
The use we are trying to define is the second, not the first, which in the
guise of the symbol -p we are taking for granted and propose to discuss
later under the head of negation. 9
Our definition that a belief is true if it is a "belief that p" and p, but
false if it is a "belief that p" and -p is, it may be remarked, substantially
that of Aristotle, who considering only the two forms "A is" and "A is
not" declared that "To say of what is, that it is not, or of what is not,
that it is, is false, while to say of what is that it is, and of what is not
that it is not, is true."10
Although we have not yet used the word' correspondence' ours will probably
be called a Correspondence Theory of Truth. For if A is B we can speak
according to common usage of the fact that A is B and say that it corresponds
to the belief that A is B in a way in which if A is not B there is no such
fact corresponding to it. But we cannot describe the nature of this cor-
respondence until we know the analysis of propositional reference, of "believing
that A is B." Only when we know the structure of belief can we say what
type of correspondence it is that unites true beliefs and facts. And we may
well be sceptical as to there being any simple relation of correspondence
applicable to all cases or even if it is always right to describe the relation
as holding between the "belief that p" and the "fact that p"; for instance
if the belief is disjunctive as it is when Jones thinks that Smith is either
a liar or a fool, are we to say that it is made true by a "disjunctive fact,"
"the fact," namely, "that Smith is either a liar or a fool"? [If we believe
that reality contains no such mere "either-or" we shall have to modify our
account.] Or if we hold it absurd to believe that reality contains such a
mere either-or, what does the belief correspond to?
But the prospect of these difficulties need not distress us or lead us to
suppose that we are on a wrong track in adopting what is, in a vague sense,
a correspondence theory of truth. For we have given a clear definition of
truth which escapes all these difficulties by not appealing to a notion of
12 CHAPTER I

correspondence at all. A belief that p, we say, is true if and only if p; for

instance a belief that Smith is either a liar or a fool is true if Smith is either
a liar or a fool and 'not otherwise. It seems, indeed, possible to replace this
definition by a periphrasis about the correspondence of two facts; but if such
a periphrasis is not ultimately legitimate that does not prove that our definition
is wrong, but merely that it should not strictly be called a correspondence
theory and that a statement of it in terms of correspondence should be regarded
as merely an inaccurate popular explanation. Truth, we say, is when a man
believes that A is B and A is B, whether or not such an occurrence can
be accurately described as a correspondence between two facts; failure to
describe it in terms of correspondence cannot show that it never occurs and
is not what we mean by truth.
This account of truth is merely a truism, but there is no platitude so
obvious that eminent philosophers have not denied it, and at the risk of wearying
the reader we shall insist on our truism once more.
Let us take three statements like this:

The earth is round

It is true that the earth is round.
Anyone who believes that the earth is round believes truly.

It is really obvious that these statements are all equivalent, in the sense that
it is not possible to affirm one of them and deny another without patent
contradiction; to say, for instance, that it is true that the earth is round
but that the earth is not round is plainly absurd.
Now the first statement of the three does not involve the idea of truth
in any way, it says simply that the earth is round. [In the second we have
to prefix "It is true that" which is generally added not to alter the meaning
but for what in a wide sense are reasons of style [and does not affect the
meaning of the statements J. Thus we can use it rather like "although" in
conceding a point but denying a supposed consequence, "It is true that the
earth is round, but still ... ," or again we often use it when what we say
has been questioned: "Is that true?" "Yes, it is perfectly true." But in the
last case the phrase "it is true that the earth is round" is changing from
simply meaning that the earth is round ... ]
The meaning of the second, on the other hand, is less clear: it may be
a mere synonym for the first, but more often contains some reference to
the possibility of someone believing or saying that the earth is round. We
are thinking not merely that the earth is round, but that because it is round
anyone ll who believes or says that it is round believes or says truly. We
have passed from the first of our statements to the third. But the third amounts
in a sense to no more than the first, and it is merely the first thought of
 THE NATURE OF TRUTH 13

in connection with the possibility of someone saying or believing it. To take

a parallel case, we can say simple "The weather in Scotland was bad in
July", or we can think of that fact in reference to its possible effect on
one of our friends and say instead "If you were in Scotland in July, you
had bad weather." So too we can think of the earth being round as a possible
subject of belief and say "If you think the earth is round, you think truly"
and this amounts to no more than that the earth has the quality you think
it has when you think it is round, i.e. that the earth is round.
All this is really so obvious that one is ashamed to insist on it, but our
insistence is rendered necessary by the extraordinary way in which philosophers
produce definitions of truth in no way compatible with our platitudes,
definitions according to which the earth can be round without its being true
that it is round.1 2 The reason for this lies in a number of confusions of which
it must be extremely hard to keep clear if we are to judge by their extraordinary
prevalence. In the rest of this chapter we shall be occupied solely with the
defence of our platitude that a belief that p is true if and only if p, and
in an attempt to unravel the confusions that surround it.
The first type of confusion arises from the ambiguity of the question which
we are trying to answer, the question "what is truth?", which can be interpreted
in at least three different ways. For in the first place there are some philosophers
who do not see any problem in what is meant by 'truth', but take our
interpretation of the term as being obviously right, and proceed under the
title of "what is truth?" to discuss the different problem of giving a general
criterion for distinguishing truth and falsehood. This was for instance Kant's
interpretation13 and he goes on quite rightly to say that the idea of such
a general criterion of truth is absurd, and that for men to discuss such a
question is as foolish as for one to milk a he-goat while another holds a
sieve to catch the milk.
And secondly even when we agree that the problem is to define truth
in the sense of explaining its meaning, this problem can wear two quite
different complexions according to the kind of definition with which we
are prepared to be content. Our definition is one in terms of propositional
reference, which we take as a term already understood. But it may be held
that this notion of propositional reference is itself in need of analysis and
definition, and that a definition of truth in terms of so obscure a notion
represents very little if any progress. If a belief is identified as what Mr
Jones was thinking at 10 o'clock in the morning, and we ask what is meant
by calling the belief so identified a true belief, to apply the only answer
we have so far obtained we need to know what Mr Jones' belief was a
"belief that"; for instance, we say that if it was a belief that the earth is
flat, then it was true if the earth is flat. But to many this may seem merely
to shirk the hardest and most interesting part of the problem, which is to
14 CHAPTER I

find out how and in what sense those images or ideas in Mr Jones' mind
at 10 o'clock constitute or express a "belief that the earth is flat." Truth,
it will be said, consists in a relation between ideas and reality, and the use
without analysis of the term propositional reference simply conceals and shirks
all the real problems that this relation involves.
This charge must be admitted to be just, and an account of truth which
accepts the notion of propositional reference without analysis cannot possibly
be regarded as complete. For all the many difficulties connected with that
notion are really involved in truth which depends on it: if, for instance,
"propositional reference" has quite different meanings in relation to different
kinds of belief (as many people think) then a similar ambiguity is latent
in 'truth' also, and it is obvious that we shall not have got our idea of truth
really clear until this and all similar problems are settled.
But though the reduction of truth to propositional reference is a very
small part and much the easiest part of its analysis, it is not therefore one
which we can afford to neglect. [Not only is it essential to realize that truth
and propositional reference are not independent notions requiring separate
analysis, and that it is truth that depends on and must be defined via reference
not reference via truth. ]14 For not only is it in any event essential to realise
that the problem falls in this way into two parts,15 the reduction of truth
to reference and the analysis of reference itself, and to be clear which part
of the problem is at any time being tackled, but for many purposes it is
only the first and easiest part of the solution that we required; we are often
concerned not with beliefs or judgments as occurences at particular times
in particular men's minds, but with, for instance, the belief or judgment "all
men are mortal"; in such case the only definition of truth we can possibly
need is one in terms of propositional reference, which is presupposed in the
very notion of the judgment "all men are mortal"; for when we speak of
the judgment "all men are mortal" what <we> are really dealing with is
any particular judgement on any particular occasion which has that pro-
positional reference, which is a judgment "that all men are mortal." Thus,
though the psychological difficulties involved in this notion of reference must
be faced in any complete treatment of truth, it is well to begin with a definition
which is sufficient for a great many purposes and depends only on the simplest
considerations.
And whatever the complete definition may be, it must preserve the evident
connection between truth and reference, that a belief "that p" is true if
and only <if> p. We may deride this as trivial formalism, but since we cannot
contradict it without absurdity, it provides a slight check on any deeper
investigations that they must square with this obvious truism.
 THE NATURE OF TRUTH

NOTES

1. How difficult the problem is may be judged from the fact that in the years 1904-25
Mr Bertrand Russell has adopted in succession five different solutions of it.
2. I use "state" as the widest possible term, not wishing to express any opinion as to
the nature of beliefs etc.
3. Or, of course, that something is not so and so, or that if something is so and so, something
is not such and such, and so on through all the possible forms.
4. [It should perhaps be remarked that the late Professor Cook Wilson held that these
mental states do not in fact belong ... J It should, however, be remarked that according
to one theory this is not really a deficiency at all, since the states in question have
nothing important in common. Knowledge and opinion have propositional reference in
quite different senses and are not species of a common genus. This view, put forward
most clearly by J. Cook Wilson, (but also implied by others, e.g., Edmund Husserl)
is explained and considered below.
5. Bernard Bosanquet, Logic, 2nd ed., Vol II (Oxford, 1911), pp. 282 ff. Of course he sees
the distinction but he deliberately blurs it, arguing that an account of truth which enables
an ill-grounded statement to be true, cannot be right. His example of the man who
makes a true statement believing it to be false, reveals an even grosser confusion. He
asks why such a statement is a lie, and answers this by saying that "it was contrary
to the system of his knowledge as determined by his whole experience at the time."
Granting this, it would at most follow that coherence with the man's system of his
knowledge is a mark not of truth (for ex hypothesi such a statement would have been
false) but of good faith; and this is brought in as an argument in favour of a coherence
theory of truth!
6. It has been suggested by Professor Moore ("Facts and Propositions," Proceedings of the
Aristotelian Society, Supplementary Volume VII (1927), pp. 171-206; see p. 178) that the
same entity may be both a belief that (say) the earth is round and a belief that something
else; in this case it will have two propositional references and may be true in respect
of one and false in respect of the other. It is not to my mind a real possibility, but
everything in the present chapter could easily be altered so as to allow for it, though
the complication of language which would result seems to me far to outweight the
possible gain in accuracy. See ...
7. In Mr Russell's symbolism

B is true:=: '( :l.p). B is a belief that p & p. Df

8. In a sentence like this "in fact" serves simply to show that the oratio obliqua introduced
by "he believed" has now come to an end. It does not mean a new notion to be analyzed,
but is simple a connecting particle.
9. See below. <Presumably this is a reference to the unwritten chapter on negation.>
10. Metaphysics, Gamma, 6 1011b25, Mr Ross' translation.
11. For instance the man we are talking to may have just made the point and we concede
it. "Yes, it's true, as you say, that the earth is round, but -" or we may have made
it and be questioned "Is that true, what you were saying, that the earth is round?"
"Yes, it's perfectly true."
12. Thus according to William James a pragmatist could think both that Shakespeare's plays
were written by Bacon and that someone else's opinion that Shakespeare wrote them
might be perfectly true "for him." ("The Meaning of Truth," p. 274.) On the idea
that what is true for one person may not be true for another see below.
16 CHAPTER I

13. See Kritik der reinen Vernunft, "Die transzendentale Logik." Einleitung III (A57=B82): "Die
alte und beriihmte Frage ... Was ist Wahrheit? Die Namenerklarung der Wahrheit, dass
sie namlich die Obereinstimmung der Erkenntnis mit ihrem Gegenstande sei, wird hier
geschenkt und vorausgesetzt; man verlangt aber zu wissen, welches das allgemeine und
sichere Kriterium der Wahrheit einer jedem Erkenntnis sei." The reason why there can
be no such criterion is that every object is distinguishable and therefore has something
true of it which is true of no other object. Hence there can be no guarantee of truth
irrespective of the object in question.
14. [This might perhaps be denied if reference were something essentially different in the
cases of true and of false beliefs; e.g., if the precise way in which a man's belief today
that it will be wet tomorrow was a belief "that it will be wet to-morrow" depended
on how to-morrow's weather actually turned out. But this is absurd for it would allow
us to settle the weather in advance by simply considering the nature of the prophet's
expectation and seeing whether it had true-reference or false-reference.]
15. It might possibly be questioned whether this division of the problem is sound, not because
the truth of a belief does not obviously depend on its reference, i.e., on what is believed,
but because reference might be essentially different in the two cases of truth and falsity,
so that there were really two primitive ideas, true-reference and false-reference, which
had to be separately analysed. In this case, however, we could tell whether a belief
that A is B were true or false, without looking at A by simply seeing whether the
manner in which the belief was a "belief that A is B" was that of true-reference or
false-reference, and infer with certainty that to-morrow would be fine from the fact
that someone believed in a particular way, the way of false-reference, that it would
be wet. See below.
 APPENDIX: TWO VARIANT ELABORATIONS

(001-04)

<APPENDIX 1 TO CHAPTER 1>*

Objections to such a definition are of two kinds, [merely] formal and serious.
By a formal objection I mean one which does not deny that a belief that
A is B is true when and only when A is B, but says that this, though a
correct statement, is not the right definition or explanation of truth. For
instance it might be argued that a true belief is a belief in a true proposition,
and that though the proposition 'A is B' is true, when and only when A
is B, yet for the proposition 'A is B' to be true and for A to be B, are
two different (though equivalent) facts, and that the former, not the latter,
is the one which should be used in defining true belief. [This sort of niggling]
These formal questions I propose to leave till later, and devote the present
chapter to examining the objections of a more serious sort, which deny that
a belief is true if and only if it is a belief "that p" and p, and propose
instead definitions such as that a belief is true if it is useful, or if it is part
of a coherent system, definitions which are not merely formally but materially
at variance with ours.
I must confess I find it hard to see how anyone can dispute that a belief
that A is B is true if and only if A is B. Let us take the three sentences.

The earth is round

It is true that the earth is round.
Anyone who believes the earth is round believes truly.

It is clear that the first two of these are equivalent; anyone who said "The
earth is round but it is not true that it is," or "It is true that the earth
is round but the earth is not round" would be contradicting himself. I think
they have, in general, the same meaning, and that when we use "It is true
that the earth is round" in preference to the shorter formulation we do
so for what may be called in a broad sense reasons of style; for instance

*< Editor's note: This discussion elaborates matters treated in the second half of Chapter I.>
18 CHAPTER I

It is true that" can be used rather like "although" in conceding a point

but denying a supposed consequence, and again when what we say is questioned
and on reflexion we decide to reaffirm it. "Is that true?" "Yes its perfectly
true." In this last case, however, it could plausibly be said that "It's true
that the earth is round" was used not simply for "The earth is round" but
for the third of our sentences "Anyone who believes (or says) the earth
is round believes (or says) truly." But on our view this third statement amounts
really to no more than the first, just as "if you were in scotland in July
you had bad weather" comes to no more than "the weather in scotland
was bad in July." For "anyone who believes the earth is round believes truly"
means simply that the earth has the quality which anyone who believes it
to be round believes it to have, i.e., that the earth is round.
On our view therefore, or any view of the same sort, even if formulated
differently, to assert anyone of our three sentences and deny any other
would be self-contradictory, as every common man would take it to be;
but some philosophers have given theories of truth which if they mean anything
<mean> that this is not the case, and that the earth could be round without
its being true that it was round.
It is, indeed, hard to see how anyone could think such a thing, and one
suspects that the dispute is a mere misunderstanding and that by their theories
of truth, they are not trying to answer our question as to the meaning of
the term but presupposing our or some similar theory of what the term
'true' means and trying to discover to what beliefs it should be applied,
i.e., that they are discussing not "what is truth?" but "what is true?"l But
the study of their works soon dispels any such hope and shows that we are
really faced with serious disagreement as to the whole nature of truth. Indeed
of the three leading types of theory, the Correspondence Theory, the Coherence
Theory and Pragmatism, only the first agrees with us on the main issue
that a belief that A is B is true if and only if A is B, and our view belongs
undoubtedly to the class of correspondence theories, although we have not
yet used the word correspondence. [For instead of saying that the belief
is true, if A is B, we can say, if we like, that it is true if there exists
such a fact as that A is B, and this fact that A is B can then be said to
correspond to the man's believing that A is B. But the kind of correspondence
that obtains between them can only be discovered from the analysis of
propositional reference, of what we mean by believing that A is B. Only
then we can settle in what way believing that A is B corresponds to A's
being B, and that is a task we have not yet attempted; we have taken
propositional reference for granted and asked not "what constitutes propo-
sitional reference?," but, "given propositional reference, what constitutes
truth?"]
For if A is B, we can speak according to common usage of the fact that
 APPENDIX: TWO VARIANT ELABORATIONS 19

A is B and say that it corresponds to the belief that A is B in a way in

which if A is not B no such fact corresponds to that belief; but the kind
of correspondence can only be fixed when we know the analysis of "believing
that A is B" that is, of propositional reference. This we shall attempt to
give later, but so far we have simply taken "believing that" to be something
with which we are all familiar in spite of our inability to analyse it, and
have given our definition of truth and falsity in terms of this unanalysed
notion. Weare thus only in a position to speak vaguely of whatever
correspondence there may be between A's being B and a man's belief that
A is B, and not to make the notion precise. It is clear, moreover, that the
task of making it precise is faced with difficulties not only from the side
of the belief but also from the side of the corresponding fact. If Jones believes
that Smith is either a liar or a fool, are we to say that his belief, if true,
corresponds to a "disjunctive fact" that Smith is either a liar or a fool?
Or shall we deny that there is any such either-or in reality and say that
the belief is made true either by the fact that Smith is a liar, or by the
fact that he is a fool? Or is this question merely verbal and are both expressions
equally [good, valid] accurate? To my mind they are both equally inaccurate,
for this talk of correspondence, though legitimate and convenient for some
purposes, gives, in my opinion, not an analysis of truth but a cumbrous
periphrasis, which it is misleading to take for an analysis. To believe truly
is to believe that p when p, and there is no need [but many disadvantages
in restating] to recast this definition in terms of correspondence, unless indeed,
some sort of correspondence is essentially involved in the notion of 'believing
that p," a question which we must leave till we come to the analysis of
propositional reference.
Criticisms of the correspondence theory are nearly all directed to particular
simple forms of correspondence which are not such as any reasonable man
could suppose to hold between beliefs and facts. They obviously cannot
constitute truth, and are only suggested for the sake of destroying them.
For instance Professor Joachim defines correspondence thus: "Two different
factors ... 'correspond' when each of them is a whole whose inner structure
is teleological, when that structure is identical as the explication of the same
idea or purpose, and when, finally, for every distinctive part fulfilling a
determinate function within the one factor there is a part fulfilling the same
function within the other."2 And of course the view that it is this sort of
correspondence that holds between true beliefs and facts is not worth a
moment's consideration. A belief that grass is green is true, a belief that
grass is red is false, but no one could seriously suggest that this was because
the former belief "explicated the same idea or purpose" as the greenness
of the grass. It is the two beliefs which could reasonably be said to have the
same structure, the function fulfilled by red (or whatever makes it a belief
20 CHAPTER I

about redness) in the one being fulfilled by green in the other; so that if
one belief has the same structure as the fact so also does the other, and
the difference between truth and falsity clearly cannot be simply a difference
of structure.
But to knock down such men of straw is not to give any serious reason
for abandoning the obvious truism that a "belief that p" is true if and only
if p [which is what our opponents propose, and it behooves us to consider
what positive reasons they can have for their theories] and the motive forces
behind the Coherence and Pragmatist theories must be found elsewhere.

NOTES

1. So Kant, Kritik der reinen Vernunft, A57= B82.

2. Harold H. Joachim, The Nature of Truth (Oxford, 1906), p. 12.
 APPENDIX: TWO VARIANT ELABORATIONS

(OOt-05 First part)

<APPENDIX 2 TO CHAPTER 1>*

On the other hand, if a description <of truth> in terms of correspondence

is possible, then our definition might be criticised for failing to mention this
correspondence. That a belief that A is B is true if and only if A is B,
is, it might be said, a correct statement but it is not a correct definition
of exactly what we mean by a true belief, which is one corresponding to
a fact; no doubt the belief that A is B corresponds to a fact if and only
if A is B, yet these are different things and it is the former not the latter
which gives the meaning of truth. So also' believers -in proposihons ITllght
say that a true belief is a belief in a true proposition and that though the
proposition "A is B" was true if and only if A is B, yet for the proposition
to be true and for A to be B were different things and the former not
the latter gave the meaning of true belief. The difficulty, of course, only
arises if one, at least, of these rival formulations should turn out to be legitimate,
that is, if there are such things as propositions or such a relation as this
correspondence; [this we do not yet know,1 but supposing it does arise it
does not go very deep] and in any event it is not very serious since the
alternative definitions are ex hypothesi logically equivalent to ours and do
not really define different notions. [It is never or hardly ever possible and
if it were possible, it would be futile to chose between two equivalent definitions
except on the score of convenience. Does "great-uncle" mean grandparent's
brother or parent's uncle?]
A man, we may suppose, is believing that A is B; two cases are possible,
either A is B or it is not. In the first case in which A is B, the proposition
"A is B," if there is such a thing, is true, and the belief that A is B corresponds
to a fact, namely the fact that A is B. In the second case A is not B, the
proposition is false, the fact does not exist. Weare all agreed that the belief
is true in the first case and false in the second, and differ only as to whether
it is A being B, the proposition being true, or the existence of the fact

*<Editor's note: This discussion elaborates matters treated towards the end of Chapter 1. >
22 CHAPTER I

that gives our meaning when we say that the man's belief is true. [Since
by hypothesis they must all happen together always, I know not how one
can decide which we mean by ... ] It is surely a futile question, like asking
whether "great uncle" really means parent's uncle or grandparent's brother;
but if we have to choose there is every advantage in choosing the definition
which says that the belief is true simply if A is B, and avoids all the psychological
and metaphysical difficulties involved in the correspondence of beliefs with
facts or the existence of 'propositions.'
Not that these difficulties will not have to be faced. [If we wish really
to get to the bottom of the notion of truth we must not be content merely
to reduce it to propositional reference but must proceed with the analysis
of that in turn. But for many philosophical purposes such a reduction to
propositional reference will suffice; we often do not need to .... ] To give
a complete analysis of truth it is not enough merely to reduce it to propositional
reference but this too must be analysed [in turn J. So long as we accept it
[that notion] without analysis there is [still] an element of obscurity in talking
of "belief that p," a feeling of lurking difficulties which must not be shirked.
And since truth depends on and is defined in terms of [propositional reference]
this notion of "believing that," obscurity [and ignorance] with regard to
propositional reference means obscurity with regard to truth. For instance
we do not yet know whether there really is a general notion of propositional
reference so wide as to embrace all cases of believing, or whether the term
"believing that" may not really be ambiguous and its different meanings
have nothing important in common. Since in this latter case 'truth' too would
be ambiguous and the 'correspondence' between 'true' beliefs and facts would
be quite different for different kinds of beliefs, we shall clearly not have
finished with truth until we have got to the bottom of propositional reference.
[Nor till then can we conveniently answer a certain sort of objection
to our definition that a belief that A is B is true if A is B. I mean the
kind of criticism which allows that this is a correct statement about truth but
objects to it as a definition [of truth J. For instance, some people hold that
a true belief is a belief in a true proposition and that though the proposition
"A is B" is true when and only when A is B yet for the proposition A
is B to be true and for A to be B are two different (though equivalent)
facts; and, so they think, it is the former and not the latter of these facts
that should be used in defining true belief. Clearly this objection can only,
or at any rate can much more easily be disposed of when we have decided
whether propositional <reference exists> J.
But for many philosophical purposes it is not necessary to go so deep:
we need not go into the analysis of the notion of a "belief that p" and
can be content with a definition of truth which tells us what is meant by
calling a belief true when its propositional reference is given. At all events,
 APPENDIX: TWO VARIANT ELABORATIONS 23

the first step in the analysis of the idea of truth is to show that it depends
on propositional reference in such a way that when we know what a belief
is <as> a "belief that," we know in what case it is true, i.e., if it is a "belief
that p" it is true if p.2
This may seem such a truism as hardly to be worth enunciating, and
indeed a great many philosophers in discussing "what is truth?" take this
first step for granted and tackel straightaway the next and much more difficult
problem of what is propositional reference. Thus William James often3 insists
that truth means the agreement of our ideas with reality, and that the problem
is what this agreement means, i.e. in our terminology, a belief is true if
it is a belief that A is B and A is B. But what is this connection between
the belief and A's being B that makes it a belief that A is B?
Yet, surprising though it may seem, many philosophers have explicitly
or by implication denied this truism, and constructed theories completely
at variance with it. One suspects at first, that there is some misunderstanding,
that in asking "what is truth?" they are not, like us, seeking merely for
the meaning of the word, but for some criterion for truth, some way of
telling what is true. 4 But this hope can not survive an examination of their
works, and the number of philosophers who are clear on the essential point
is surprisingly small.
[By the essential point I mean this; when we are asking what is meant
by calling a belief true, we must be clear whether or not we are taking
its propositional reference for granted, whether we are asking merely what
is meant by calling true a belief that so-and-so ("so and so" being either
explicitly given or at least assumed to be theoretically definite) or taking,
say, "what is passing in Mr Smith's mind at 10 o'clock" <to be at issue>.]

NOTES

1. [But we shall come later to the conclusion that they are not legitimate. See ... ]
2. <The first page of the miscellaneous jottings of folder 001-10 reads as follows:>
And secondly, even if, however, we agree that our problem is to define truth in the
sense of explaining its meaning, there are still two different ways of taking the problem
according as we do or do not accept the notion of propositional reference as sufficiently
[precise, clear, and intelligible] understood to be used in our definition of truth.
If for instance we are discussing a belief identified as what Mr Smith was thinking at
ten o'clock in the morning, our definition of truth only tells us what is meant by calling
such a belief true, if we already know what is meant by saying that his belief was a
belief that so-and-so, perhaps that the earth is flat. His thoughts at that time were (at
least on one view) images and ideas of some kind whose connection with what they are
said to refer to, namely that the earth is flat, is by no means easy to describe.
Secondly, <truth> may be taken either as defined in terms of reference or in terms of
more primitive notions. We have only done the former, but the latter is more complete.
24 CHAPTER!

Nevertheless the former does for many purposes, and is all many have attempted to give.
A definition of truth in terms of reference must not bring in any extraneous idea, like
Peirce, Meinong, Hofler. The idea of this former <approach> might be questioned as not
on <the right> way, but is must be <accepted>. Reference cannot depend on truth or
truth <be> intrinsic <to it>. James agrees, but gets quite tied up: he gets a fragmentary
account of reference which makes him get completely <muddled>. Also he does not understand
<its> propositional nature. Russell sees <the> point but gets words involved and confuses
this muddle with C<orrespondence>.
3. E.g., William James, The Meaning of Truth (New York and London, 1906), Preface, p. v.
4. $0 Kant. "Die alte und beriihmte Frage ... Was ist Wahrheit? Die NamenerkIarung der
Wahrheit, dass sie namlich die Ubereinstimming der Erkenntnis mit ihrem Gegenstande
sei, wird hier geschenkt und vorausgesetzt; man verlangt aber zu wissen, welches das
allgemeine und sichere Kriterium der Wahrheit einer jeden Erkenntnis sei." Kritik der reinen
VernunJt, "Die transzendentale Logik," Einleitung III.
 (001-05)

<CHAPTER II>

<TRUTH AND COHERENCE>

d. COHERENTIST ARGUMENTS AGAINST ABSOLUTE TRUTH>

Let us take the Coherence Theory first; this holds that the truth of a belief
that A is B depends not on whether A is in fact B but on how far the
belief that it is forms part of a coherent system. It is a theory which is
very easy to reduce to absurdity and after Mr Russell's amusing essay on
"The Monistic Theory of Truth"! it is difficult to see how anyone can still
cling to it; but the defect of all refutations by reductio ad absurdum is that
they do not reveal where the line of thought which leads to the absurdity
first goes astray. According to Mr Russell the first mistake of the advocates
of the Coherence Theory lies in their assuming an abstract metaphysical axiom
called the Axiom of Internal Relations, but some of their arguments seem
to me not to depend on any such axiom but to arise from confusions of
a simpler sort such as ensnare not only the abstruse metaphysician but also
the common man.
The lines of argument in question are two; in the first place it is argued2
that truth is a property not of single judgments but of systems of knowledge,
so that no judgment can be wholly or absolutely true but "the truth of
'true' judgments is essentially the truth of a system of knowledge and ...
the 'truth' of systems of knowledge is borrowed from the Ideal Experience
which is struggling for self-fulfilment in them."3 And, secondly, it is argued
that truth cannot be a relation between our beliefs and facts, for it must
be something which we can test, and all we can test are the relations between
our beliefs. Truth must therefore lie within the circle of our beliefs and
not pass outside them to an unknowable reality.4
In giving our account of truth we have assumed that any belief or judgment
has a definite propositional reference or meaningS in virtue of which it is
absolutely true or false. The obvious reason for disputing this lies in the
possibility that judgments may be vague in the same sort of way as the
words we use to express them. How many hairs must a man have not to
26 CHAPTER II

be bald? There is no definite number, and if someone is judging that he

is bald, his judgment may be so vague as not be definitely either true or
false. 6 But this cannot be our opponents' ground of objection, since a system
of vague judgments7 would be no less vague than its members, and as they
contend that such a system is capable of truth in a higher degree than its
members the obstacle they see to absolute truth cannot be in vagueness.
The arguments which they do put forward in support of their contention
seem to be four.
(a) The first is a mere confusion between truth and certainty. It may
be that I cannot be absolutely certain that Julius Caesar crossed the Rubicon;
perhaps he did no such thing; but this absence of certainty does not prove
that if he did, it is not absolutely true that he did [and not merely partially
true]. This confusion is a very common one, for those different persons who
assure us that they make no claim to absolute truth generally mean merely
that they are not absolutely certain. S
(b) The second argument starts from the undoubted fact that the same
sentence can mean very different things to different people and that to any
one person its meaning may depend on its context. This fact is then transferred
from statements to judgements, which are said to have no definite meaning
in isolation but only in their context, so that since truth depends upon meaning
it is not applicable to isolated judgments but only to systems of judgments. 9
There is a certain plausibility in this argument but when we examine
it we shall see that it rests entirely on a confusion, and that what it seeks
to establish is something plainly absurd. The confusion arises in the use of
the word 'judgment.' A judgment is in the first place "a piece of concrete
thinking," something happening in Mr Smith's mind at 10 o'clock in the
morning; he may, for instance, be saying to himself "No, it won't." If, then,
we were to consider in isolation just that fraction of his mental process
represented by the words "No, it won't," it is plausible to maintain that
we should not be able to discover at all what its meaning was, as this entirely
depends on what has gone before. Nevertheless we ordinarily suppose that
as a result of this preceding context such words do express, at any rate
sometimes, a judgment with a definite meaning. We cannot say that the
judgment would have that meaning "in isolation"; a piece of concrete thinking
cannot happen in isolation; but in the context it actually has, that is its meaning,
and according to that meaning, as I think, it is true or false. I say we commonly
suppose that judgments have definite meanings in this way; but we can go
further and say that this is what our opponents too [the adherents of the
view we are discussing] must suppose, when they talk of systems of judgments.
For these "systems of judgments" can hardly be composed of such pieces
of concrete thinking as we have been considering. Euclidean space which
according to Professor Joachim is a system of judgments,lO can hardly be
 THE COHERENCE THEORY OF TRUTH 27

identified with the concrete thinking of those who learn or teach geometry.
The judgments which enter into such a system are not yours [Mr Smith's]
and mine [Mr Jones's], but, for instance, the judgment that the angles at
the base of an isosceles triangle are equal, and the judgment that all men
are mortal, which are abstractions from the actual judgments of men. We
do not [want] need here to take any particular philosophical view of abstraction,
for on all such views this at least is clear: in order that we can talk significantly
of the judgment that all men are mortal, be it abstraction, universal or what,
your, my and other people's judgments that all men are mortal must have
something in common in order that they may be instances of the same universal
or type; and this that they have in common, is not the words we use to
express them, but their meaning. If single judgments had not definite meanings,
to talk of the, or even of a judgment that all men are mortal, i.e., of a
judgment with a certain meaning would be nonsense. A "system of judgments "
must really mean a system of meanings of judgments, and if judgments had
not their own meanings the phrase would be sheer nonsense.!!
(c) The third argument used against absolute truth turns simply on confusing
two uses of words like "condition" or "basis"; thus it is argued by Prof.
Joachim!2 that a proposition of elementary geometry such as that the angles
of a triangle add up to two right angles is only true if we assume the basis
of Euclidean Space, and that similarly no scientific judgment is true unless
regarded as "the abbreviated statement of a meaning which would require
a whole system of knowledge for its adequate expression." Now it may
well be that the geometrical proposition is not true of physical space but
merely in abstract Euclidean geometry; i.e., that all that is really true is
that the proposition is a consequence of certain definite assumptions, and
this, which is a single hypothetical judgment, is an absolute truth. But this
is not, I fancy, what Prof. Joachim means: he does not doubt, or at least
did not then doubt, that physical space13 (so far as anything is true of it)
is Euclidean, and what he intended to say was that the proposition about
the angles of a triangle was only true because space is Euclidean. 14 But to
argue from this that the meaning of the proposition must be expanded so
as to include the whole nature of Euclidean space, would, I think, be a
sheer confusion, for just because space is (on this view) Euclidean the proposition
is true in its literal meaning, Euclidean space being a kind of space in which
the angles of a triangle are together equal to two right angles.
Precisely the same argument occurs at the crucial point in Bradley's Logic
when he is dealing with the singular judgment of sense,15 of which he claims
that "what it says is true, if it is true at all, because of something else ...
so we have a judgment which is really conditioned, and which is false if
you take it as categorical." He sees, however, that this argument confuses
two senses of absolute or categorical truth; first that in which absolutely
28 CHAPTER II

true means true, and not merely the consequence of a certain hypothesis,
in which, that is to say, absolute is opposed to conditional; and secondly
that in which absolutely true means true for no reason, absolute being opposed
to conditioned. That a judgment is true because of something else does not
mean that it is not true. [He nevertheless thinks his argument can be preserved
by complicated considerations about the nature of conditions, which seem
to me to have no cogency.16 His answer is to compare the series of phenomena
to a chain <with> each link fast to the next, but no fast link fastened to
firm ground to hold it. I do not see why this should be thought absurd
and why the links of such an endless chain should not be real. He has, however,
other reasons of a metaphysical character. In any event it is the sort of a
metaphor which only confuses the issue.]
His answer to this is to say that "the condition, on which the judgment
holds, is unknown, and it admits also the opposite of what is asserted. The
judgment therefore, in its present form, is at once both true and false. "17
This answer seems to depend simply on confusing a necessary with a sufficient
condition; our judgment p was said to be true only because of q; i.e., if
q were not true, p would not be; i.e., q is a necessary condition of p. But
this is in no way incompatible with q admitting also the opposite of p.
Thus suppose p is my reading Bradley: this only came about because of
q, my hearing him praised. But of course, q is perfectly compatible with
the opposite of p, I might have heard him praised without reading him,
and this is supposed to prove that I never have read him! At least, this is
the only sense I can attach to the argument; for if condition in the first
place meant sufficient condition, it is self-contradictory to say that it is
compatible with the opposite of that for which it is a condition. Nor is
this conclusion escaped by saying that the condition is unknown or unspecified:
for even the unknown cannot have self-contradictory properties. But from
Essays on Truth and Reality, p. 232, it is clear that Bradley means that two
sets of conditions are involved, but unknown, one sufficient for p, the other
for not-po This means that he proposes to argue: "For reasons unknown,
you read my book." (Remember we are starting from the hypothesis that
p is conditioned and trying to prove it conditional, not assuming it conditional
to begin with). "But under other unknown circumstances you would not
have read it. Therefore, it is not absolutely true that you did read it! "18

NOTES

1. In Bertrand Russell, Philosophical Essays (London, 1910), pp. 150-169.

2. See Harold H. Joachim, The Nature of Truth (Oxford, 1906), pp. 85-113.
3. Joachim, ibid., p. 89.
 THE COHERENCE THEORY OF TRUTH 29

4. See Bernard Bosanquet, Logic, 2nd ed. (London, 1911), Vol. II, pp. 265-267.
5. 'Judgment' and 'meaning' are Prof. Joachim's terms corresponding to our 'belief and
'propositional reference.'
6. For a theory of vagueness see below, p .....
7. E.g., a system of 'psychology' consisting of such general laws as "All bald men are
foolish. "
8. It is difficult to see that Mr Bradley's argument in Essays on Truth and Reality (Oxford,
1914), pp. 262-265, has any plausibility except on the basis of this confusion. He suggests
that there might be many worlds in some of which a Caesar crossed the Rubicon, in
others of which he did not. Even so, "at least one Caesar crossed the Rubicon" would
be an absolute truth (if true at all).
9. Cf. Joachim, op. cit., pp. 92-95.
10. Joachim, op. cit., p. 28. But he surely should have said Euclidean geometry.
11. The point is also clear if we take Prof. Joachim's own comparison of a system of knowledge
and a symphony; the possibility of a symphony being performed depends on its consisting
of notes with assigned pitches producible at will by sufficiently expert players in that
or any other context. And so too a science consists of judgments with definite meanings
which can be grasped by any sufficiently intelligent student and reproduced by him
in any other context. But cf. below pp .....
12. Joachim, op. cit., pp. 95-99.
13. Of course, physical space must be more precisely defined, by determining what is to
be meant by "straight line" etc., but these definitions only fix what we are to mean
by our terms, they do not form a "system of knowledge."
14. This is the only sense I can give to the phrase categorical basis.
15. F. H. Bradley, Principles of Logic, 2nd ed. (London, 1922), vol. I, pp. 97-98.
16. <In margin:> Confusion of if and because.
17. Bradley, op. cit., p. 113, note 54. In the text p. 100 there is an even less satisfactory
metaphorical argument.
18. [In the last resort his argument reduces to his theory that in judgment we ascribe predicates
to Reality. This seems to me clearly false for reasons to be given in Chap ...... But
even if it were granted his argument still requires us to suppose that Reality is a must
mean that a is its whole nature, and that is merely a. (Bradley, Essays, pp. 226-9.) This
is simply the elementary confusion of the "is" of predication and with the "is" of identity.
See below .... J <Against this footnote Ramsey first placed question marks, and subsequently
struck it out entirely.>
 CHAPTER II

(001-06)

<CHAPTER II, SECTION 1, CONTINUED>

(d) [The strongest ... ] The kind of judgment in regard to which it is

clearest that truth and falsity are absolute is the simple judgment of fact
such as "Caesar crossed the Rubicon." Prof. Joachim, however, maintains
that even this is not absolutely true in its literal meaning and would not
appear unmodified as part of the whole truth about the event regarded from
any systematic standpoint, e.g. as forming part of the life of Caesar or the
decline of the Roman Republic. When made in such a context the meaning
of the judgment, according to Professor Joachim, is the actual event which
was not

"a nucleus of 'brute fact' encased, solid and distinct, within a surrounding
complex of conditions. It was Caesar at the head of his army and animated
by conflicting motives of patriotism and ambition, who crossed. And he
crossed the Rubicon at this determinate political juncture, with a full
consciousness of the effect of his action on the political crisis at Rome.
This - and more - is the meaning of the historical judgment in its proper
context, its definite meaning. This concrete happening is 'the fact' affirmed
in the judgment if indeed you can arrest the expansion of its meaning
even here. We can be sure, at any rate, that the actual happening contains
no bare crossing of a stream by a man in the abstract as a solid grain of fact,
separable from a complicated setting which particularises it. "1

Now it is obvious that "Caesar crossed the Rubicon" is not the whole
truth about that event, but I cannot see that Prof. Joachim's arguments have
any tendency to show that it is not part of the whole truth; because he
was at the head of his army was he not still Caesar, because he was led
to cross by motives of ambition did he any the less cross? When we say
simply that he crossed, we do not particularise where, when and from what
motives, but we are not denying that he must have crossed at some definite
place and time for definite reasons. Because we call him simply a man and
say no more about him we do not mean he as a "man in the abstract"
without body parts or passions. If I asked someone whether there was a
dog in the house would it be reasonable for him to reply "No; there isn't
an abstract dog, but only a poodle "?
Nor do I see how a complete life of Caesar could be written which did
not say either that he crossed the Rubicon, or something else from which
that fact could at once be extracted, i.e. something which contains that fact
in its meaning. Professor Joachim is, indeed, prepared to allow that "Caesar
 THE COHERENCE THEORY OF TRUTH 31

crossed the Rubicon" is not wholly false, but since he thinks that every
judgment, e.g. "Caesar did not cross the Rubicon," involves some truth,
this concession seems hardly to do justice to the facts.
It seems to me that the refusal to admit that anything but the whole
truth can be wholly true comes from the confusion I mentioned before between
truth and intellectual merit in general, and that this in turn comes from
an emotional attitude towards the word truth which does not allow her
votaries to admit that she can ever be tedious or trivial. Thus Professor
Joachim speaks somewhere of "truth so trifling that serious falsehood is involved
in it"2 and Bosanquet asks, "could we seriously say that a judgment about
it is true in which its full significance and implications ae ignored?"3 I say
this is reverence for the word "truth" advisedly; for the idea of truth is one
which no one could reverence who clearly apprehended it; knowledge and
ingenuity one can admire, but mere truth can be attained by a fool by accident
and has no moral value.
Another argument sometimes used is that a historical work, which is accurate
in every detail, may be less true than one which although less accurate gives
a better idea of the period. The truth of the work cannot therefore consist
in the truth of the separate statements. The answer to this is simple; by
saying one work is truer than another we mean something about the relative
importance of the true and false judgments they express, including not only
judgments about details but also those about the relative importance of different
causal factors, etc., which are sometimes not expressed explicitly in statements,
but by the proportionate space given to the different topics, and if these
are misleading we get the paradox of a work not being regarded as wholly
true in spite of the accuracy of every statement.
Bradley argues4 that "Caesar crossed the Rubicon" is not absolutely true
because there may be many worlds, in some of which he crossed it and
in others of which he didn't, or again because Caesar might recur in this
present world. The possibility of identifying which Caesar you mean he rejects,
because he links that <ostensive> designations ('this', 'now', 'here', etc.) have
no place in a genuine judgment. This view seems to me plainly contrary
to fact [and it is defended by the merest rhetoric J.5 But even if it were
granted, the argument in no way shows that 'At least one Caesar crossed
a Rubicon' is not an absolute truth. Nor does it even show that "The one
and only Caesar crossed the one and only Rubicon" is not absolutely true
or absolutely false according as there are or are not many Caesars. In fact,
the argument's whole plausibility rests on the confusion between truth and
certainty.
32 CHAPTER II

NOTES

1. Harold H. Joachim, The Nature of Truth (Oxford, 1906), p. 107.

2. See Joachim, op. cit., p. 17.
3. Bernard Bosanquet, Logic, 2nd ed. (Oxford, 1911), p. 287. May it not be true of the Bible
that it contains at least 500 z's even if this ignores its "full significance and implications"?
4. F.H. Bradley, Essays on Truth and Reality (Oxford, 1914), pp. 262-5.
5. Bradley, op. cit., pp. 234-235. [Note how he passes without any justification from feeling
being "unintelligible" to its being inconsistent. Those who have no hope of finding everything
in the world intelligible in his sense do not therefore contradict themselves. J
 THE COHERENCE THEORY OF TRUTH

(001-15)

2. <A FURTHER OBJECTION FROM THE PHILOSOPHY OF SCIENCE>

Our commonsense assumption that single beliefs can have propositional

references according to which they are absolutely true or false remains,
therefore, unshaken by any of these arguments; but before we pass on, it
will be best to consider what might be urged against it by philosophers who
take their ideas from physical science.
Weare often told, for instance, that the question whether Space is Euclidean
is by itself meaningless and is only significant as part of a complete physical
theory. Thus the same facts, it is thought, could be explained either by saying
that space is Euclidean and rays of light are curved or by saying that space
is non-Euclidean and rays of light are straight (follow geodesics); and these
two theories would be only verbally different, and taken as wholes would
have the same meaning, the choice between them being one of convenience
only.
But this may be granted without any sacrifice of our fundamental
assumption; for in such a case, the paradox is merely due to the same terms
being in different senses in the two theories. The definitions or meanings
in terms of experience of "straight line," "congruent," etc. are different
in the theory on which Euclidean geometry is true from those in the theory
which makes it false. Owing to the ambiguity of these terms the same sentence
is true on one interpretation, but false on another; but this does not show
that in each theory it has not its own meaning and truth, independent of
the truth of the whole. Both systems of order, Euclidean and non-Euclidean
geometry are illustrated in nature by different relationships either of which
could be logically deduced from the other.
There is, however, a more radical philosophy! [of science which is harder
to deal with] on which science begins with observations and laws which
assert observed uniformities; and these laws are then explained by theories
which introduce undefined entities and relations. Some of the statements which
a theory makes about these undefined entities are to be interpreted by means
of a 'dictionary' in such a way that they can be proved true or false by
observation. But other statements about the undefined entities have no such
interpretation, and are regarded as having no 'truth', except such as can
be derived from the satisfactoriness of the theory of which they form part. 2
For the theory to be regarded as satisfactory, such of its statement as are
to be interpreted in observable terms must be true, and it must have certain
further merits, which may be called in a broad sense aesthetic. This philosophy
of science is, of course, disputed, but as it is a plausible view, let us consider
its bearings on our problem.
34 CHAPTER II

In spite of a superficial resemblance it is radically different from the

Coherence Theory, in that it allows independent truth to some propositions
in a theory, and so presupposes that some judgments have their own meaning
and truth. But what are we to say of the other propositions and the judgments,
if any, which they express? It is clear that since they are supposed to have
no meaning, they do not express any judgment or belief, except in a very
elliptical way; for instance they may be said to express the judgment that
the theory which contains them is satisfactory, or simply that, whether it
is satisfactory or not, it does contain them. On the latter interpretations they
resemble such sentences as "Cerberus has three heads," which are literally
meaningless (since Cerberus is not the name nor even thought to be the
name, of anything) but can be given the sense "such and such an author
narrated that Cerberus had three heads." And on the former interpretation
they are like the judgments made by writers in planning out their books
"It will be best for Cerberus to have three heads." These so-called propositions
do not therefore express judgments and constitute no exception to our view
that single judgments are true or false; but they are interesting as showing
that a large body of sentences, which appear to express judgments and are
manipulated according to the laws of formallogic 3 may not express judgments
at all. This is a fact which it is well to bear in mind, since this view can
be extended to include not merely what appear to be statements about facts
which could not be observed, but also all statements apparently about facts
which have not or will not actually be observed. So that questions about
cosmogony, or the back of the moon, or anything no one has ever seen
may not have any independent meaning, but only be about what it would
be best for us to say in order to get a satisfactory scientific system. If this
were so, "The back of the moon is made of green cheese" might be both
'true' and 'false', equally satisfactory 'theories' having been found, one of
them containing that sentence or rather allowing it to be deduced, and the
other containing its contradictory.
[This, I say, is a possible view; but it is one which I shall disregard in
the present work, in which we shall suppose that our ordinary statements
about the external world express definite judgments, which are true or false.
Much of what we say would still be true on the contrary hypothesis, for
instance our account of truth as we have just seen, and the rest would, I
think, survive with some systematic modification. But to prove this we should
have to make precise the theory of the external world [the view of science]
which we have to meet, and complicate our argument by references to it
every stage, a task which I feel is better postponed to another occasion].4
 THE COHERENCE THEORY OF TRUTH 35

NOTES

1. See especially Norman Campbell, Physics: The Elements (New York, 1919) and also Heinrich
Hertz, Prinzipien der Mechanik <Leipzig, 1894>, Introduction.
2. Many scientists wish wherever possible to eliminate unobservable entities, such as the orbit
of the electron in a hydrogen atom, but this is so far no more than an ideal.
3. Why they should be so manipulated is, on this view of science, rather a mystery.
4. <Ramsey first placed a question mark alongside the opening sentence of this paragraph
and subsequently struck the paragraph out entirely.>
 CHAPTER II

(001-07 AND 001-08)

d. ARGUMENTS FOR THE COHERENCE THEORY REFUTED>

We must now return to the Coherence Theory. We have seen that the
contention that truth is an attribute of systems of knowledge, not of single
judgments cannot be sustained, but there is still the possibility that the truth
of single judgments is to be defined by coherence. The arguments advanced
in favour of this, mainly consist of refutations of alternative doctrines. Thus
Professor Joachim disposes of the theories that truth is a quality of independent
entities and that it consists in correspondence "for a mind" and shows that
each of them leads ultimately to coherence.
We need not linger long on Prof. Joachim's refutation of the doctrine,
then held by Mr Russell and prof. Moore, that truth and falsity are qualities
of independent entities. This is not my view, and it has, I think, been abandoned
even by its authors. Nor can we defend the "fundamental postulate of all
logic" that "Experiencing makes no difference to the facts," in which, if
it is to be relevant to the present discussion, "experiencing" must be taken
widely enough to include judgment and belief. For the success of M. Cone
is alone enough to show that believing that a thing will happen can often
make it happen, at any rate in the medical field.!
It is, I feel, unfortunate that in Prof. Joachim's discussion this view that
truth is a quality of independent entities is the only one he considers on
which truth is not essentially "for a mind." Truth is, we may agree, an
attribute of mental states consisting, as I believe, in correspondence to fact;
but this does not make it in Prof. Joachim's sense 'for' a mind. As is shown
clearly by his example from Leibniz, by saying that a correspondence is 'for'
a mind he means not merely (or perhaps not at all) that one of the corresponding
entities is a mental state, but that some mind is aware of or recognizes these
entities as corresponding. Against the position which I adopt, that a belief
may be true by correspondence to fact, even when no one recognizes it
as corresponding, he says nothing no doubt because it had not then2 sufficient
authority to deserve his attention. [He dismisses <this view> in one place3
with a reference to his discussion of the quite different view that truth is
a quality of non-mental propositions, and in another with the dogmatic assertion
that the angles of a triangle would not be equal to two right angles if no
one had ever thought they were!]4
His refutation of the correspondence theory does not, therefore, claim
to apply to the version of that theory which I advocate, but it is nevertheless
worth considering as showing the sort of objections we have to meet. In
the first place we may grant that truth could not possibly consist in the
 THE COHERENCE THEORY OF TRUTH 37

only kind of correspondence which Prof. Joachimn considers [and that for
much simpler and more conclusive reasons than any he advances]. "Two
different factors," he says, "'correspond' when each of them is a whole
whose inner structure is teleological, when that structure is identical as the
explication of the same idea of purpose, and when, finally, for every distinctive
part fulfilling a determinate function within the one factor there is a part
fulfilling the same function within the other."5 It does not really seem to
me that the view that truth is constituted by a correspondence of this sort
deserves a moment's consideration. A belief that grass is green is true, a
belief that grass is red is not: could anyone seriously suggest that this was
because the former belief was a structure "explicating the same idea or purpose"
as is explicated by the greenness of the grass, and the second belief was
not such a structure? The structures of the two beliefs could reasonably be
regarded as identical, the function fulfilled by the part red6 in the one being
fulfilled by the part green in the other. But it would surely be absurd to
suppose that of these two beliefs of identical structure one had the same
structure as the fact but the other had not.
The question can be made clearer if we talk not of beliefs but of statements
or sentences, whose structures are more obvious. It is clear that there is
a sort of correspondence between the sentence "grass is green" and the fact
that grass is green, in as much as the elements of the fact are the meanings
of the terms in the sentence. But this correspondence does not consist in
identity of structure, for the sentence "grass is red" has precisely the same
structure as the sentence "grass is green" but does not correspond to the
fact in the same way since the word "red" does not mean green. Whether
or not structural identity is necessary for truth, it is clearly not sufficient for
it, and cannot be supposed to be the kind of correspondence which constitutes
it. The kind of correspondence which might be sufficient is that arising from
the relation of meaning. The kind of correspondence based on meaning does
not, as we shall see, make error the same as not thinking, as does the kind
considered by Professor JoachimJ
This is presumably dismissed by Prof. Joachim when he says that he assumes
that a purely external relation is in the end meaningless and impossible. What
mayor may not be the case "in the end" I do not know, but in the middle
or wherever it is that this life is situated, there is a relation between words
and what they stand for, capable of giving rise to a sort of correspondence,
and determining or at least helping to determine, the truth of statements.
That this may not be so "in the end", is to my mind irrelevant. In the
end, I understand, there is no time, and no events in time. Consequently,
since judgments are events in time there are no judgments, and no true
judgments. The problem of truth as an attribute ofjudgments cannot, therefore,
I conceive, be a problem as to what happens in the end, and arguments
38 CHAPTER II

drawn for this are no more relevant here than they are in aeronautics.
I have spoken above as if truth were constituted in part at least by a
kind of correspondence based on the meaning of words, but I do not want
to assert this at the present stage [nor is it exactly true]. It belongs also
to what we have called Question II, the question of what constitutes
propositional reference, not to Question I, what, given propositional reference,
<MS 001 - 08>
constitutes truth, over which we are primarily at issue with the Coherence
Theory. I introduced it as a type of correspondence, not considered by Professor
Joachim, and relevant here as the type [of correspondence] which would
constitute truth, if Question II were answered in a certain way. For it will
be remembered that although we decided that truth depended on correspon-
dence, we postponed the decision as to what type of correspondence it was,
saying merely that it was the kind [of correspondence] there evidently is
between believing that a thing is so and its actually being SO.8
Nor can I believe that this is not what Prof. Joachims means at any rate
when he pronounces anyone else's opinion true: suppose he reads Bosanquet's
logic and wonders whether Bosanquet's view that every hypothetical implies
a categorical ground is true or false. Surely he is comparing Bosanquet's
opinion with the state of the case: if the fact is as Bosanquet says, if every
hypothetical does imply a categorical ground, then and only then is what
he says true [and otherwise false].
To this I suppose it might be replied that in such a case when A is considering
whether B's opinion is true, it is wrong to say that A compares B's opinion
with the fact, since he himself has no sure knowledge of the fact but only
his own opinion. Consequently, we should say that A compares B's opinion
not with the fact but merely with his own opinion, and the relation which
he pronounces is not one between an opinion and a fact but between two
opinions. But such an objection would be invalid, as it is of the first importance
to see. According to the objector, A has no sure knowledge of the fact,
but then he equally has no sure knowledge as to what is really B's opinion,
and if through ignorance of the fact he can compare nothing to the fact,
then through ignorance as to B's opinion he can not compare that but merely
his own opinion about it; so that what he really does is to compare his
opinion about B's opinion with his opinion as to the fact. We can now see
that the point has no special relation to truth, but arises in connexion with
any kind of comparison, for instance, comparison of colour. Suppose the
question arises, whether two regiments of soldiers which I have seen or read
about have uniforms of the same colour, and I think "The uniforms of the
Northshires are red and the uniforms of the Southshires are red, so they
have the same colour". Then the relation which I affirm in so concluding
 THE COHERENCE THEORY OF TRUTH 39

is one between the actual colours or uniforms, not one between my opinions
about them. My conclusion is founded no doubt on my opinions as to what
each colour is, but [is not about those opinions but] states a relation which
I believe to hold not between my opinions (or at least not merely between
them) but between the real colours. 9 So also if A says B's opinion agrees
with the fact, the relation he assumes to hold is one between B's opinion
and the fact, not one between his own opinions about them: although, of
course, his making the assertion is the result of his holding these opinions.
I do not therefore see how it can be denied that when we say that other
people's opinions are true or false we mean that they do or do not correspond
to the facts; but how is it, we shall be asked, when we come to consider
the truth of our own opinions. For this is the case from which the advocates
of the Coherence Theory draw their strongest arguments, and the source,
as it seems to me, of all its plausibility.
On the correspondence theory, it is argued,lO truth means a correspondence
between our beliefs and a reality outside them, an external standard to which
we claim to conform. But we cannot put before ourselves such a standard
with any hope of attaining it; nor can we use it as a criterion. If I make
a judgment, and claim that is is true, this cannot be because I see that it
corresponds to a fact other than itself; I do not look at the judgment and
the fact and compare them; the fact as I see it is simply my judgment, and
all I can test it by is my other judgments. Wherever I turn I am bound
by the circle of my own ideas, and the truth I seek [cannot consist in a
relation between them and something outside but] must lie inside the system
of my beliefs, not in a relation between them and an unknowable reality.ll
The second line of reasoning employed by supporters of The Coherence
Theory is important as an example of a general fallacy, which is by no
means confined to this school of thought, and lies at the root of a lot of
mistakes in regard to truth and knowledge. In the present context the argument
runs as follows.
On The Correspondence Theory, it is said, truth means a correspondence
between our beliefs and a reality outside them, an external standard to which
we claim to conform. But we cannot put before ourselves such a standard
with any hope of attaining it; when, for instance, I say that someone else's
opinion is false I mean merely that it does not agree with my own, and
when I make a judgment myself and claim that it is true this cannot be
because I see that it corresponds to a fact other than itself; I do not look
at the judgment and the fact and compare them, but the fact (as I see it)
is simply my judgment, and all I can test it by is my other judgments. Wherever
I turn I am bound by the circle of my own ideas and the truth I seek must
lie inside the system of my beliefs and not in a relation to an unknowable
reality.1 2
40 CHAPTER II

Truth therefore in any sense in which we can know ourselves to possess

it, must be immanent, and this immanent standard must be coherence and
comprehensiveness for this is the test we all employ. There is no memory
or perception of which we are so certain that we would not abandon it
as an illusion if it were impossible to harmonize our beliefs otherwise. 13
These arguments, plausible though they are, seem to me to mistake the
nature of the problem, and to create an entirely illusory difficulty. The fact
is we can not claim truth for our judgments when making them, for if we
did, our judgment would be about itself, which is impossible. We cannot
think about our judgments until we are outside them [as we might think
about other peoples] and then we [can] discuss them just as we do those
of others people. For instance, when I discuss how far certain opinions I
held as a child were true, I am wondering how far the facts were as I
then thought. If I go on to ask how far the facts are as I now think, in
that asking a new thinking has begun and what I "now" think is already
in the past. It is, therefore, a fundamental mistake to suppose that in our
thinking, except when we are specifically thinking about thought, we use
the idea of truth at all; we do not set up truth as a standard and ask ourselves
"Are my thoughts about the earth true, do they correspond to fact?" we
ask simply what is what, for instance is the earth flat? And when we try
to find out whether the earth is flat, the nature of truth is irrelevant, for
we are concerned with the nature of the earth not with that of our thought:
of course it does not help us to be told "Your view that the earth is flat
will be true if it corresponds to a fact". To know the meaning of truth
could not possibly help us, and the phrase "test of truth" used by Bosanquet
is really a misnomer. For truth is strictly a property of beliefs, and what
we want are tests not for the properties of beliefs but for the properties
of the things we are studying.
The argument we are trying to refute is that when we decide that the
earth is round, we are not judging that our belief corresponds to a fact
but rather that it is the only one which is coherent with our beliefs about
ship's disappearing below the horizon, etc. The truth which we suppose our
belief to have must therefore consist not in correspondence with fact but
in coherence with other beliefs. The mistake lies, as we have seen, in supposing
that in judging that the earth is round we are thinking about beliefs at all;
neither our final judgment that the earth is round nor our initial beliefs that
ships disappear, etc., are objects of our thought at all. We are not thinking
about our own thinking but about ships disappearing and the earth being
round; if coherence comes into our thought at all it is the coherence of
reality not the coherence of our beliefs.
Coherence, in fact, cannot sensibly be put forward as a theory of truth
but at most as a theory of argument or proof:14 when so regarded it has
 THE COHERENCE THEORY OF TRUTH 41

an important contribution to make, which we shall consider when we come

to deal with inference.
[On our last argument against the coherence theory in which we urge
that a judgment cannot [judge of itself that is true] be about itself, it is
well to pause, as the matter is an important one and what we have said
about it will probably not be accepted without further discussion.] What
we have said above means that we must reject altogether the view [held
by some (e.g. Prof. Joachim)J15 that true means "true for a mind" and that
no judgment is true unless recognized as being true. On this view if I judge
that the earth is round, my judgment is not true unless I (or perhaps someone
else) recognize it to be true; simply to make the judgment is not enough,
I must also make a judgment about it. This necessary recognition that my
judgment is true [is clearly of the nature of a further true judgment] must
however itself be a true judgment, and the question arises whether or not
it is different from my original judgment; if it is different <then> since it
is itself true there will have to be yet another judgment about it and so
on ad infinitum, which seems plainly contrary to fact. If, however, the judgment
in which I recognize that my judgment is true is that judgment itself, then
it must be a judgment about itself. This I say is obviously impossible, and
leads to contradiction: for if it were possible to have a judgment which
was about itself, we could have a judgment which said that it itself was
false, and such a judgment would be true, if false and false if true, and
so both true and false and neither true nor false, which is impossible.16

NOTES

1. Cf. Mr R.B. Braithwaite in Mind, vol. 36 (1927), p. 468.

2. Since then Prof. Moore and Mr Russell have adopted it.
3. [H.H. Joachim, The Nature of Truth (Oxford, 1906), p. 8].
4. [Ibid., p. 14. These are not his exact words, but express his meaning, which is obscured
by the uncertainty as to what sort of thing he means by "a truth." ]
5. Ibid., p. 12.
6. Or if red and green are not parts of the judgments, whatever factors in the judgments
make them judgments as to redness and greeness respectively.
7. Joachim, op. cit., pp. 130-133.
8. F.H. Bradley (Essays on Truth and Reality (Oxford, 1914), p. 109) argues that truth cannot
copy fact because it goes beyond the given facts. We can show the inadequacy of this
argument by an illustration. A general may make a map showing what he imagines
to be the disposition of the enemy's forces; this map will not be copied from reality [but
based on conjecture], but if the general is clever it may correspond to and in that sense
be a copy of the enemy's real disposition. Only in this extended sense has anyone ever
held that truth should copy reality. But, of course, a disjunctive or negative belief cannot
as Bradley says conceivably be properly called a copy: but must be [described] analysed
in a more complicated way. Philosophers are, however, much too prone to suppose
42 CHAPTER II

that one simple [explanation] kind of analysis must be given for all kinds of belief; as
we shall see, copying and pragmatism are both elements in the true analysis which is
exceedingly complicated, too complicated for us to hope to give it accurately.
9. By "real colours" can be meant here either their real colours in a conventional sense,
the colours the uniforms would have in normal light etc., or the colours they really
appeared to me to have when I saw them. The contrast is not between real colour
and apparent colour, but between the colour as it really is or really appeared and the
colour as it is thought to be or remembered to have appeared.
10. See especially B. Bosanquet Logic, 2nd ed. (Oxford, 1911), Vol. II, pp. 265-267. [Also
F.H. Bradley, Essays on Truth and Reality (Oxford, 1914), p. 108; "If truth is to copy
fact, then truth at least seems to be in fact unattainable." But he is really discussing
a view that is not in question here, that truth should copy given facts.
11. <The following two paragraphs are taken from a misplaced MS page (numbered 34)
from MS 001-05-01. The continuity is guaranteed by the identity of the ensuing paragraph.>
12. <The next bit comes from 001-05-01/45! The page missing in the MS 001-08, 45, was
misplaced there.>
13. <The next sentence overlaps between the misplaced page 45 (see footnote 12) and the
continuation of the present manuscript MS 001-08, p. 46.>
14. This is virtually admitted by Bosauquet iu Mind, vol. 31 (1922), pp. 335-336. He admits
that "Caesar was murdered" is true, if the murder really took place but argues that
we are not entitled to say it is true without proof. But this is quite irrelevant to the
meaning of truth; we are equally not entitled to say a thing is useful without proof
that it is, but this does not [illuminate] bear on the meaning of the word "useful".
To use "true" to mean "proved true" is an atrocious misuse of language and leads
to the kind of confusion which brings philosophy into merited disrepute.
15. Joachim, op. cit., Ch. I, passim.
16. The question in what senses a judgment can and in what senses it cannot be about itself
is a difficult one, but in the sense required here it clearly cannot. On the whole question
see ....
 (001-11)

CHAPTER III

JUDGMENT

The truth and falsity of mental states we found to depend upon their
propositional references and our next task is to analyse this characteristic,
to answer, in fact, our Question II, what is propositional reference? But
before starting on this it is best to make certain preliminary distinctions between
the states of mind with which we have to deal.
To begin with there is a very elementary point about the way in which
we use the words knowledge, belief and opinion. In general, these refer
not to definite acts of thinking but to the persistent background of the mind.
If we say that a boy knows the date of the Norman Conquest, we do not
mean that he is at the moment thinking "1066" but that he would be able
to give that answer if required. Weare talking not about what he is actually
thinking, but about what he would think or say in certain circumstances.
of course, he would not know it if he had not learnt it and so thought
about it at least once; but we say he still knows it even when he hasn't
thought about it for years provided he has not forgotten and could still answer
our question.
By "knowledge," then, we usually mean something mere potential, which
would be manifested if occasion arose but may still be there without being
manifested, and the same is true of "belief' and "opinion." An opinion is
not an act of thinking; when we "form" or "change" an opinion we are
thinking, but the opinion so formed remains ours and is carried round with
us ready to be given to anyone who asks for it without our needing to
think of it at all in the mean time.
Knowledge, belief and opinion when the terms are used in this way [in
this ordinary sense], we shall call dispositional, because in their capacity for
remaining latent they resemble qualities of disposition or character in the
ordinary sense, for a man may be called brave or irascible without any
suggestion that he is at the moment actually displaying these qualities.
It must not be thought that these dispositional characteristics are only
to be found in connection with mind; for they are common enough in the
44 CHAPTER III

material world. When, for instance we say that a poker is strong, we mean
that it would support a considerable strain without breaking. This fact is
something purely hypothetical; for the poker in question may never be subjected
to such a strain, but still it is true of it, that if it were, it would not break,
and this is, in our sense, a dispositional characteristic.
In the case of the poker we suppose that whether it would or would
not break depends on the nature and arrangement of the particles of which
it is composed; and that its strength (which is a dispositional characteristic)
depends on some positive (i.e., non-dispositional) characteristics of the
component particles. But most of us have no idea what those positive
characteristics are, and we can perfectly well talk of the dispositional quality
strength, without knowing the positive qualities on which it depends. So
also in the case of the boy who knows the date of the Conquest, we must
suppose his knowledge to depend on some arrangement, 'trace' or 'record'
in his mind or brain, which is formed when he learns the date and persists
until he forgets it, his forgetting being simply the disappearance of this trace. 1
These traces, or in different cases other formations, constitute the positive
qualities from which dispositional knowledge and beliefs are derived, but
most of us have no idea as to what sort of structures or modifications the
traces are, and take them simply as unknown causes which bring it about
that if for instance we ask the boy for the date he tells us correctly. So
when we are trying to explain as at present what we mean by knowledge,
etc. we have no concern with the real nature of these traces but merely
with the kind of thoughts or actions which they are supposed to cause. Just
as in explaining the meaning of strength, we have only to explain what
is meant by supporting a strain without breaking, not what properties of
a body they are which enable it to support a strain.
To say a man has such and such knowledge, beliefs and opinions means
then generally something hypothetical, something about what he would think,
say,2 or do in suitable circumstances. It is, in my view, important to realise
that it is not only a question of what he would think or say but also of
what he would do, for many of our dispositional beliefs are manifested far
more in our actions than in our thoughts. For instance, I have a dispositional
belief (or perhaps I should rather say knowledge) that the Cambridge Union
is in Bridge Street; but this belief is very rarely manifested in an act of
thought: I do not often have occasion to judge that that is where the Union
is: I only do this when I have to inform a stranger, or just now when I
thought of it for an example. On the other hand, this belief of mine is frequently
manifested by my turning my steps that way when I want a book from
the Union Library, which I do without any process of thought which could
properly be called thinking that the Union is in Bridge Street. In Oxford,
I should have to think where the Union was, but in Cambridge, where I
 JUDGMENT 45

am at home, I go there habitually without having to think.

But although a man's dispositional beliefs influence his behaviour and may
sometimes be most easily recognised through their influence on his behavior,
it may still be objected, that when we say that he has such and such a
belief, we are not meaning anything about how he behaves but only about
how he thinks; that its effect on conduct is a usual, or even an invariable
accompaniment of belief, but not part of its essence, not part of what we
mean by the word. Such an objection, is not in my view sound, but I cannot
prove this at this early stage; we shall however find it impossible to give
any satisfactory account of belief or even of thought without making any
reference to possible resulting action. 3 For instance, the notion of a numerical
degree of belief which is important in the theory of probability can only
be explained as that of a belief leading to certain sorts of practical decision. 4
Our problem is to explain propositional reference, and in connection with
dispositional knowledge and beliefs, this means that we have to analyse the
meaning of saying that a man has a belief that such and such is the case,
for instance, that the earth is flat. This we have seen to be partly an assertion
about what he would think or say and partly (if I am right) one about how
he would behave. The assertion we make about his behavior is evidently
a very complicated one, for no particular action can be supposed to be
determined by this belief alone; his actions result from his desires and the
whole system of his beliefs, roughly according to the rule that he performs
those actions which, if his beliefs were true, would have the most satisfactory
consequences. 5
But this connection between beliefs and conduct we must leave till later
and for the present confine ourselves to the other parts of the notion of
a dispositional belief, the disposition it entails to think or speak in a certain
way. From this point of view the propositional reference of the belief is
evidently derived from that of the resultant thoughts or assertions; a man's
belief that the earth is flat is manifested by his thinking and saying that
the earth is flat, and the propositional reference of his belief to be defined
in terms of the reference of his thinking or the meaning of his words. 6
The propositional references of dispositional knowledge and beliefs are
therefore derived from those of definite acts of thought. For such acts of
thought we do not generally use the words 'knowing' or 'believing'; but
rather 'thinking', as in "I was just thinking it would snow to-morrow", or
of the terms 'judging', 'inferring', 'asserting', 'perceiving', 'discovering' and
'learning' which all refer to definite acts, not to dispositions.
Such acts of thought (or perception if that be distinguished from thought)
which have propositional reference and an affirmative character I shall call
judgments. I say "acts of thought" in order to exclude dispositional beliefs7
etc.; I insist on their havingpropositionai reference to exclude such an occurrence
46 CHAPTER III

as thinking of a number, without thinking that it has any particular property,S

;~nd I say they must have an affirmative character in order to exclude acts
0'[ wondering and doubt. The question of acts of denial will be discussed
under negation in ch ....
This use of the word Judgment' is common in works on logic, but as
it goes considerably beyond the way the term is used in ordinary language
it will be best to try to elucidate it further. It has been said that judgment
is a decision reached from doubt, and presupposes a preliminary process of
inquiry and indecision9; in ordinary language this may be so, but we shall
use the word much more widely so as to include any form of thinking that,
whether it be a reasoned conclusion or a guess or a prejudice or a memory
or a presentiment or anything else whatever of the same general type. Judgment
in our usage presupposes no process of reflexion or weighing of evidence;
we may reflect and weigh the evidence before we judge but only too often
we jump to a conclusion without any such process.
We have excluded doubt from judgment, and this raises the question whether
by judgment we necessarily mean an attitude of complete conviction, or
whether something short of this, a mere preference, say, for one opinion
rather than another can count as judgment. An adequate treatment of this
question of degrees of belief will only be attempted in Ch. IV, and for the
present we must be content with a vague answer. Suppose first [the judgment
results from] <that> we are concerned with a process of reflexion. Then
the most convenient course is not to confine the use of the word judgment
to those cases in which the evidence seems absolutely conclusive, but to include
the formation of any opinion, if it is formed with sufficient confidence for
us to take it as a basis for our future thought and action, and regard its
truth temporarily at any rate, as a settled question. On the other hand we
must clearly exclude those cases in which we merely decide that one possibility
is more likely than another, but do not feel the latter to be so improbable
that we can dismiss it from our minds and disregard it in our practical decisions.
When, secondly, we are not concerned with reflexion, but with a judgment
of perception or memory, the difficulty hardly arises. We say to ourselves
or otherwise think that such and such is or was the case without any feeling
of doubt and unless something made us pause to think we should be prepared
to act on our judgment. Such judgments vary in the degree in which they
could survive cross examination if anything were to make us sceptical, but
generally occur first of all [with such strength that] without any admixture
of doubt, so that we need have little hesitation on that score in awarding
them the title of judgments.
In many cases, however, there are other difficulties in the way of deciding
whether we have a genuine case of judgment. An excellent illustration of
this has been given by Cook Wilson.
 JUDGMENT 47

"We see at a little distance a person whom 'we mistake for an acquaintance'
and without hesitation perform some act which it would be a liberty
to take with anyone but an acquaintance, do something in fact which
we rightly say we should not have done if we had ever suspected he
was not an acquaintance. "10

For Cook Wilson it is incorrect to say that in such a case we judgedll or

believed that it was our friend, because he only uses these terms for cases
in which we come to a conclusion after a process of reflexion, so that they
are clearly inapplicable here.

"It is true that, if asked, we might say 'I thought it was my friend' -
'I believed it was my friend' - 'I was sure it was my friend', but," says
Cook Wilson, "these expressions are all inaccurate. The truth is, as will
be admitted, that in the given case, when I perceive the familiar characteristic
of my friend, it never 'enters into my head' that they could belong to
anyone else. I don't think about that at all and so the processes of judgment
belief and opinion are impossible."12

In his usage, therefore, [this is not an example of judgment;] we could not

have judged that it was our friend; but in my usage, in which judgment
does not imply reflection, can we not say that is was a case of judgment?
If not, what was it? According to Cook Wilson, it was a fact of consciousness
"simulating judgment"; I treated him as if he were my friend, not only
in the practical sense of acting as if he were, but also in my mental attitude.
Now to say that this mental attitude 'simulates' judgment must mean that
it resembles judgment in certain respects, that it has some of the characteristics
of a judgment formed by reflection, and as it seems to me, it is just these
characteristics which constitute the affirmative character by which in my
usage judgment is defined. Let us think what probably occurred. On seeing,
say, the man's back resembling his friend's, the man may have said to himself,
"Hullo, it's Jones", or, if not, perhaps he may have had an image of Jones's
face suggested by this back he saw. In the first case his saying to him self
"Hullo, it's Jones" would indicate to my mind that he was making ajudgment.
[To some this may seem impossible; "saying the words to yourself' they
will say could not be a judgment, at most it could express one. This does
not seem to me at all certain; when we think "in words" I should say that
the words were an essential part of our thinking, and not merely the expression
or the incidental accompaniment of some process outside themselves. "Can
you think in French" does not mean "can you express your thoughts in
French?"13 But even if I am wrong in this, and our thoughts never consist
in our saying things to ourselves, there still seems no reason why the man's
48 CHAPTER III

words "Hullo, it's Jones" should not be the expression of a thought.]

His words are not a meaningless formula; as he says them, they have
a definite meaning, referring to a definite man whom he sees so that they
must surely either be part of or express14 a thought; and this thought having
a definite propositional reference, namely that the man he sees is Jones, must
be on our definition a judgment. [If I understand him rightly, Cook Wilson
agrees that if I said it to myself that would be a sign that I was thinking,
but supposes that in the example we are considering I did not say such a
thing [to myself] but [merely] had an image of the man's face. [The state
of consciousness ... ]]
Let us take next the case in which he does not say anything to himself
but merely has an image of Jones' face. In this case, it still seems to me
that this image, just like the words in the last case, would be or express
a judgment. For it would not be a mere phantasy image since ex hypothesi
it moves him to action, and must therefore differ in some way from a mere
phantasy, perhaps through being accompanied by some related feeling, such
as Mr Russe1l1 5 calls a belief-feeling. In such a case, he would be thinking
in or by means of an image just as at other terms he might think in words;
this does not mean of course that he mistakes his image for a perception,
any more than he mistakes words for what they mean; but that whereas
at other times he says things to himself in words, so sometimes he says them
in imagery, a different medium or language, but not so different that the
process is not fundamentally the same, and that if there is an act of thought
in the one case so there is not also one in the other.16
But suppose he neither said anything to himself nor had an image, what
then? In this case there are, I think, two possibilities: first that he made
a judgment of some other kind or in some other way, and secondly that
he was moved to act directly by his visual sensations of the man's back
without any intervening words, imagery, or thought. [My action might have
become an immediate response to seeing that kind of back, a "conditioned
reflex".] If he very often came upon his friend from behind, the action of
slapping him on the back might have become an immediate response to seeing
that kind of back, a "conditioned reflex" working without any thought or
act of consciousness beyond his initial visual sensations. In this case alone
would it be correct to deny that he judged it to be his friend; of course,
in such a case, the man should still say e.g. "I thought it was Jones" but
this would not be literally true. The only possible sense in which he had
such a thought would be that there was in his mind or brain a conformation
analogous to that which would give rise to a dispositional belief that that back
must be Jone's back. [That is not an actual event but a hypothetical property
postulated to explain my conduct.] Such a "thought" would be nothing
observable, but something postulated to account for the man's conduct and
 JUDGMENT 49

for the fact that if asked "who's that?" he would undoubtedly have replied
"Jones" (unless something made him pause to reflect). Indeed we might even
say that this "thought" was a dispositional belief, and that by saying he
thought it was Jones' back, he meant that he would have behaved (in this
case actually did behave) as if it was, and if asked have said it was. [The
only difficulty in identifying it with a dispositional belief lies in its temporary
nature; for he has never seen this particular back before.]
There is, however, a certain difficulty in identifying this so-called thought
with a dispositional belief, for what exactly is to be the propositional reference
of this belief? On the one hand it appears to be about just this particular
back which he treated as Jones', but clearly the same mental disposition or
habit was manifested the day before when he slapped what was really Jones'
back, and the disposition has no particular reference to this back but to any
back of a certain sort. But on the other hand, we cannot say that it is a
dispositional belief that all backs of this kind are Jones' for though on each
occasion that we met a back like that he would probably jump to the conclusion
that it was Jones', he might easily shy at saying that all such backs were
Jones', since the form of this question probably make him reflect. The best
thing to say seems to be that strictly speaking we have not a dispositional
belief that all such backs are Jones', but a disposition to treat both in thought
and action any such back that may turn up as Jones', and this we might
call a dispositional belief function (by analogy with propositional function)
which is not in itself either true or false, since it has no determinate reference,
but in each manifestation is either true or false according to whose back
occasions it to be manifested.
The same phenomenon occurs in many other cases in which we are tempted
to speak loosely of a dispositional belief; thus in the example given above
of my belief that the Union is in Bridge Street, manifested by my turning
my own steps or directing a stranger's thither, what is relevant is everyday
something different namely than on that day the Union is in Bridge Street.
Here again we cannot get out of the difficulty by making it a belief that
the Union will always be in Bridge Street, since I am not at all sure this
might not cease to be true, and we must say that we are strictly concerned
with a dispositional belief-functionY
But we must return to our problem, of whether the man can properly
be said to have judged that it was Jones' back that he saw. The conclusion
we have come to is this: if his seeing the back led either to his saying to
himself "Hallo, there's Jones" or to his having an image of Jones' face of
such a kind or with such accompaniments that it issues in action, then we
must say he made a judgment. If on the other hand he acted directly on
seeing the back without any such intermediary process, then there was no
judgment, although we might perhaps say that his response manifested or
50 CHAPTER III

was due to a belief function. An immediate (conditioned reflex) response

to a stimulus can be in our view a judgment provided it is a response in
thought (e.g. words or images) and not in action. The distinction between
thought and action may not be absolutely precise but even if we take the
least favourable hypothesis and suppose with the Behaviorists that what we
call thought is nothing but small movements of the muscles of the throat
we must still recognize that there is an important difference between these
small movements and large movements such as slapping someone on the back,
for the former have meaning or propositional reference which enables them
to be true or false in a way the latter are not. [Illustrate by fly: "It's a
fly" is a judgment; brushing it off, not. ]
The same distinction can be applied to the problem of how far judgment
is involved in perception. That a sensation causes us to act, [or leaves a
trace which enables us to remember its quality] does not necessarily mean
that we judged it to have a certain quality; nor is this involved in its leaving
a trace which enables us to remember it afterwards. Whether indeed we
could properly say that we perceived that something was so and so, whether
we said to ourselves that it was or not, we have according to our definition
a judgment; but to determine in any given case whether there is a judgment,
and, if so, what is its propositional reference, is an extraordinarily difficult
task. According to Dr Broad the typical perceptual situation, such as is indicated
by the phrase, "I hear a bell" or "I see a chair," always involves what
he calls a quasi-belief which

"consists in the fact that certain specific bodily feelings (connected with
the automatic adjustment of the body), certain emotions, and certain feelings
of expectation, are related in an unique way to the apprehended sensum.
These are causally dependent on the traces left by past experience. When
a sensum of a sepecific kind is intuitively apprehended certain traces are
excited; there arouse certain emotions and induce certain bodily adjustments
which are accompanied by specific bodily feelings. They may in addition
call up certain images; and, even if they do not do this, they may evoke
a more or less vague feeling of 'familiarity.' These 'mnemic consequences'
of the apprehension of the sensum do not just coexist with it; they
immediately enter into a specific kind of relation to it, which I do not
know how to analyse further. And these 'mnemic consequences' in this
specific relation to this intuitively apprehended sensum constitute the quasi-
belief about the sensum, which gives the situation its specific External
Reference. Any situation constructed of such materials in such relations,
ipso facto, has such and such an External Reference. "18

Taking this account as substantially correct, is this "quasi-belief' judgment?

 JUDGMENT 51

It seems to me that the phenomena enumerated by Dr Broad as making

up the quasi-belief fall into two kinds only one of which can be elements
in a judgment. Suppose I see an apple; then my visual sensation (or if you
will the intuitive apprehension of the visual sensum) has two kinds of mnemic
effects. In the first place it may call up images of the taste of apples etc.
and perhaps the word 'apple'; these can I think be or express parts of a
judgment, just as the man's image of Jones' face was or expressed part of
a judgment that what he saw was Jones' back. But secondly the sensation
may cause grosser bodily changes or bodily actions; it may for instance make
my mouth water, and my limbs adjust themselves to grasp the apple. In
this case, the stimulus is exciting me directly to act, and my response is
an action not a thought; neither my mouth watering itself nor the feeling
produced by my mouth watering can be part of a judgment. My mouth
watering and the adjustment of my limbs are my getting ready to eat the
apple, they are part not of thinking the apple is good to eat, but of actually
eating it, which we normally take to be a consequence of the thought; only
in this case habit or instinct has made the intermediate stage of judgment
disappear; thought has been 'telescoped' away and the stimulus leads straight
to action. [As in so many cases of habitual response, the intermediate state
of judgment has disappeared; in a phrase sometimes used it has been
'telescoped'.] We have, therefore, to distinguish two levels of quasi-belief;
having an image of the taste is on the higher level of judgment or assertion,
that, we are virtually picturing to ourselves, is what the taste would be
like; mouth watering is on the lower level of action; the quasi-belief of
which it is part means no more, as Dr Broad says,19 than that we act as
it would be reasonable to act if we believed, manifesting what we have
called above a dispositional belief function.
I have now said what I can to explain the range of phenomena which
I intend the term judgment to cover, a range which we can also describe
as that of all those mental acts, as opposed to dispositions, which could be
expressed by statements; and we have now to consider the criticism made
by Cook Wilson on this wide use of the term which has in his opinion
been the source of a great deal of confusion. 2o He argues that as a matter
of language 'judgment' ought to have the narrower meaning of decision after
doubt, but this is a mere matter of convenience, and his more important
criticisms are directed towards showing that to use not merely 'judgment'
but any word in the proposed sense is to risk overlooking the most important
distinctions. For by a Judgment' that the earth is round we may mean, in
this wider use, anything from mere hearsay or prejudice, to a reasoned
conclusion. But the prejudice and the reasoned conclusion are states or processes,
which, in Cook Wilson's opinion, have nothing in common except the verbal
form "the earth is round" by which each of them might be expressed.
52 CHAPTER III

"Now, if this common form were an expression of the mental attitude

of the person using it, it would be reasonable to expect to find a common
and essential element in the mental attitude corresponding to the verbal
form. But the form merely states the nature of what we know to be,
or think to be, existent, with complete abstraction of the fact that it
is for us a matter of knowledge, conjecture, or belief. So far from being
an expression of our mental attitude, it says nothing about it whatever.
A is B means that a certain object has a certain nature or quality; it
doesn't matter whether the statement is true or not, that is what it means. "21

On this view, then, that two mental acts are expressed by the same form
of words only shows that they are concerned with the same object or objects
not that they have anything further in common. The obvious answer to this
is the one contained virtually in our definition of judgment by reference
to an "affirmative attitude"; namely, that there must be something in common
to those cases in which we express our attitude by "The earth is round"
which is not common to those in which we ask "Is the earth round?". The
fact with which we are concerned is in both sets of cases the roundness
of the earth; but in the one case we adopt to it an affirmative attitude,
in the other a merely interrogative one. It is the obvious difference between
thinking that and wonder whether, and it is absurd to deny that there is
something in common to the affirmative acts in virtue of which they are
not merely wonderings whether but thinkings that, capable of truth and falsity.
Nevertheless it can still be argued that this common character of affirmative
acts is only superficially so important, and for deeper investigations is trifling
in comparison with the differences between them. In grouping a prejudice
and reasoned conclusion together as judgment, we are recognising their
resemblance in two respects. In the first place they both have the felt quality
or absence of felt quality22 characteristic of assertion as opposed to doubt
or inquiry; and in the second place they have similiar effects on subsequent
thought and conduct. If we judge that A is B, then until we reconsider
the matter we take A to be B both in practice and in any subsequent theoretical
investigation, and this happens whether our judgment was based on good
reasons or on no reasons at all. Any two judgments that A is B have in
common, therefore, a certain felt quality and certain effects, and also, of
course, each of them is in some way concerned with A's being B, but this
need not prevent the relation between the judging self23 and A's being B
from being quite a different one in the two cases, so different that the two
relations are not properly speaking species of the same genus at all.
It is argued that this is actually the case when one of the judgments is
knowledge and the other merely opinion;24 that in knowing or apprehending
that A is B, the mind has quite a different relation to A's being B from
 JUDGMENT S3

that which it has in opmmg that A is B. If this is right, then our term
"propositional reference" is essentially ambiguous and has at least two quite
different meanings; for ajudgment's propositional reference is its characteristic
of being a judgment, say, that A is B; and to say that it has this characteristic
is a way of saying that either the judgment or25 the judging self has a certain
relation or relations to A and B or to A's being B; so that if in two cases
these relations are different, so must the two characteristics called "pro-
positional reference" be different. 26 Since our present inquiry is into the meaning
of "propositional reference" this possibility of an essential ambiguity is a
most important one for us, and we must examine carefully the arguments
used to support it. As regards terminology, however, the most convenient
plan seems to be to continue to use our terms 'judgment' and 'propositional
reference' in the wide senses we have given them, but to keep in mind their
possible ambiguity.

NOTES

1. This view is disputed; but even if there are ultimately no such things we can regard
traces as "logical constructions" and talk about them e.g. as in the text without serious
error. In the whole subject see C.D. Broad, Mind and its place in Nature (London, 1925),
Ch.X.
2. We neglect there and elsewhere the possibility of his lying in saying the opposite of
what he thinks (not necessarily saying what is false); it seems necessary to include "say"
because it may not always be possible to distinguish any process of thought apart from
his saying it either to himself or aloud. [But if he is lying there will be such a thought,
and what is relevant there will be that thought and not his words.]
3. See below.
4. See below.
5. This formula obviously requires modification to include the case of partial beliefs, and
is anyhow inexact as it takes no account of the impulsive element in action.
6. [Weare neglecting throughout the possibility of dying> ... ]
7. Although it is often hard to say in a particular case whether we are or are not confronted
with an act of thought, the distinction between such acts and mere dispositions is in
principle a perfectly clear one, since the former are real events and the latter mere
capacities or potentialities. The practical difficulty is not to distinguish an act of thought
from a disposition, but to distinguish it from some other occurence such as a mental
image or a bodily action. The difficulty is increased by the fact that a bodily action
may, as we have seen, resemble a thought in being a manifestation of the same dispositional
belief.
8. It is sometimes denied that there is any thinking which is not thinking that; if so there
is no need to exclude it, and in any case it is sufficiently excluded by the qualification
that the acts must have an affirmative character, since it is obviously impossible to affirm
without affirming that something or other is so and so.
9. J. Cook Wilson, Statement and Inference, Vol. I (Oxford, 1926), p. 92. His criticism on
<sic for of> the wider use we considered below.
54 CHAPTER III

10. Cook Wilson, op. cit., Vol. I, p. 109.

11. Another reason why he cannot call it judgment is that on his view a judgment cannot
be false. Op. cit., pp. 104-108. See also below, p .....
12. Cook Wilson, Op. cit., p. 110.
13. [See also below p.... where an attempt is made to answer objections to the view that
talking to oneself can be judgment. J
14. [The relation of words to thought will be discussed lateL]
15. Bertrand Russell, The Analysis of Mind (London, 1921), p. 250.
16. Cook Wilson, if I understand him rightly, makes some difference between words and
images, since he takes my having said so and so to myself as a sign that thinking really
occurred, but treats imagery as irrelevant. Op. cit., p. 110.
17. The value of this concept in theoretical psychology I cannot judge: I put it forward
merely as an attempt to analyse what is involved in some ordinary uses of language.
18. C.D. Broad, Mind and its Place in Nature (op. cit.), p. 215.
19. C.D. Broad, op. cit., p. 153.
20. Cook Wilson, Statement and Inference, Vol. I, pp. 78-97.
21. Cook Wilson, Statement and Inference, Vol. I, p. 87.
22. Some hold that assertion involves a positive belief feeling; others that there is a positive
feeling of suspense in doubt or inquiry whose absence constitutes assertion. It does not
matter here which of these views is taken.
23. Or perhaps we should say not the judging self but the judgment itself.
24. As we pointed out above (p .... ) 'knowledge' and' opinion' are generally used for dispositions,
not as here for judgments. We discuss this latter use below (p .... ).
25. All the 'or's' in this sentence refer not to different kinds of judgments, but to different
theories of judgment, some of which speak of a relation between the self and the objects,
others between the judgment and objects, etc. As so often, I have complicated the
phraseology in order not to have to adopt a particular theory prematurely.
26. Cook Wilson's other line of attack on the wide use of the term Judgment' is based
on his view that the division of logic into two parts the Theory of Judgment and the
Theory of Inference is utterly confused. (Op. cit., pp. 84-91). He holds that it is impossible
to separate judgment and inference in this way, since if I infer from some premises
that the earth is round, my judgment is the process of inference itself. Even if this
were true, which I do not think it is, the distinction can easily be defended; since the
theory of judgment will then treat of certain matters which occur both in connexion
with judgments which are inferences and in connection with those which are not. (Though
as Cook Wilson points out much that comes under this head is concerned either with
verbal forms or with the objects judged about.) And the theory of inference will treat
of matters which only occur in connection with inferences, for instance, the relations
of premises and conclusion.
 (001-12)

CHAPTER IV

KNOWLEDGE AND OPINION

We saw above that judgment in the wide sense in which we use the term
was held by some not to be a real genus, but to comprehend two essentially
different processes, knowledge and opinion; and our next step must be to
examine the meaning and validity of this distinction.
The words "knowledge" and "opinion", we have also seen are most
commonly used not of judgments but of dispositions; but since the distinction
which we are investigating is primarily one between different kinds of
judgments, (which can be extended to the dispositions arising from or
manifested by these two kinds of judgments) we shall use the words knowledge
and opinion in the present chapter to mean judgments and not dispositions.
It is not held that the two categories of knowledge and opinion, exhaust
the class of judgments, thus belief may be admitted as a third category, and
there may be others such as "being under the impression that" which according
to Cook Wilson! is neither knowledge, belief nor opinion. But the fundamental
distinction proposed is that between judgments which are cases of knowing,
and those which are not, opinions being taken as representative of all those
judgments which are not knowledge.
The importance attached to this distinction may perhaps be explained in
the following way. What we know, it is said, are facts, for instance the
fact that I am sitting at a table or that the earth is round, so that knowledge
is a relation between a mind and a fact. 2 On the other hand a false judgment3
cannot be a relation between a mind and a fact, for if the judgment is false
there is no fact to which the mind could be related. If a man judges that
the earth is flat, his judgment cannot be a relation between his mind and
the fact that the earth is flat, for there is no such fact as that the earth
is flat. What false judgment is, is a difficult question, but it is certain that
a false judgment that A is B cannot be a relation between the mind and
the fact that A is B. Hence it appears that knowledge and false judgment
are two utterly different things, and that the characteristics we have called
propositional reference must be quite different in the two cases, so different
S6 CHAPTER IV

that it is hardly reasonable to call them by the same name. For if a man
judges that A is B and his judgment is a case of knowledge, its propositional
reference consists in a relation between the mind and the fact that A is
B; but if his judgment is false the characteristic that makes it a judgment
that A is B cannot be any relation to the fact that A is B, since there is
no such fact.
It might be thought that this difference was not one between knowledge
and other judgments, but between true judgments and false ones. Such a
view would, however, be very strange: take, for instance, a man who believes
that it is going to rain next day; can we really suppose that the nature
of his belief, the sense in which it is true to say that he believes it will
rain, depends on whether it actually rains or not? So that by analyzing his
mental processes today we could tell what the weather would be like tomorrow?
Since this seems absurd, everyone agrees that believing is the same kind of
process whether the belief is true or false, and since the false belief that
A is B cannot be a relation between the mind and the fact that A is B,
a true belief cannot be such a relation either. Similarly no other species
of judgment, which is even false, can consist in a relation to the fact judged:
the objects of belief, opinion and conjecture cannot be facts.
It is however, supposed that in knowledge we have a kind of judgment,
which is infallible. This means not merely that we only call a judgment
knowledge when it is true, but that there is a class ofjudgments, distinguishable
otherwise than as true ones, which are not merely all in fact true, owing
to some law of nature, but are in the nature of the case incapable of being
false, since in them the mind is apprehending a fact, and were there no
such fact, there would be not a false judgment but no judgment at all. Let
us call judgments of this kind, if such there be, apprehensions,4 without however
committing ourselves to the view that there are any such things at all. The
theory that besides the fallible species of judgment, there is this infallible
species called apprehension arises from the use (in ordinary language) of the
term knowledge, a use which we must now investigate.
When we say that a judgment is knowledge, we imply at least two things;
that the judgment in question is true and that it is made with complete
conviction. I should not apply the term knowledge to a judgment that the
earth is round, unless I myself thought the earth was round, i.e. unless in
my view the judgment was true; nor could the term knowledge be properly
used, if the man who judged were not absolutely certain that the earth was
round, but merely regarded that as the more probable alternative.
The term knowledge, then, implies both truth and certainty,S and on the
other hand belief and opinion imply neither truth nor falsehood, but generally
imply at least a slight degree of uncertainty. Thus none of these terms are
applicable to the common case in which a man is certain of something false;
 KNOWLEDGE AND OPINION 57

for then we cannot say he knows it, because it is false, nor can we say
that he believes it because he is certain. We have, in fact, to say that he
is sure of it, or certain of it, or else to use the misleading expression that
he thinks he knows it, which suggests quite falsely that he is thinking about
is his own thinking or knowing and not the matter in question.
But although truth and certainty are part of what we mean by knowledge
they are rarely, if ever, the whole of it for in using the term knowledge
we nearly always imply that the thinker has sufficient reason or evidence
for his judgment: not merely that it is true and that he is certain of it,
but that in some sense he has a right to be certain of it. A man's hereditary
political or religious convictions may be true, but if he has never thought
the matter out, and seen that there is conclusive evidence for their truth,
they would generally be regarded not as knowledge, but merely as prejudices
which happened to be true;-in order to be knowledge, a judgment must
be not merely true and certain but also well grounded. 6
In what this well groundedness consists is a very difficult problem which
properly belongs to a later stage of logical inquiry: but something must be
said about it here, because it is one possible interpretation of the distinction
between well grounded judgments and others, which leads to the view that
we are here considering that judgments and propositional references are of
two essentially different kinds.
The usual account of knowledge is that it is of two kinds, direct and
indirect. Perception, memory and insight into self-evident truths are forms
of direct knowledge; they are not required to be and could not be justified
by argument. [I perceive that I am sitting at a table, I remember that I
had coffee for breakfast and I see that two straight lines cannot enclose
space. In none of these cases is my judgment founded on arguments, our
confidence in such judgments does not require arguments to justify it, nor
could any such arguments be found.] But the remainder of our knowledge
is indirect and obtained by inference or argument from what we know already,
and so ultimately from what we know directly; and for these judgments
to rank as knowledge, for our confidence in them to be justified, the arguments
on which they are founded must have conclusive or demonstrative force.
[Refer to Cook Wilson if possible.]
It is, however, clear that most of what we ordinarily call knowledge
does not satisfy this severe standard: we speak of historical or scientific
knowledge, although neither in history nor in any of the sciences except
pure mathematics can the truth of our judgments ever be rigidly demonstrated.
[All the arguments on which they are founded are of an inconclusive character,
being drawn from testimony or experience. It is very unlikely that we are
mistaken in such things, but it cannot be proved impossible [that we are
not.] We say a man knows, if his judgment is <appropriately> founded. The
58 CHAPTER IV

arguments in which they are founded all depend on testimony and experience
and so are necessarily inconclusive. ] However many witnesses assert that
something happened, it cannot be proved that they are not lying, and however
many experiments confirm a scientific hypothesis it may yet break down
later. By knowledge we mean justified confidence, but we regard ourselves
as justified in placing complete confidence in arguments that do not begin
to amount to strict demonstration. It is even clear that arguments of the
sort usually accepted as a sufficient foundation for knowledge not only may
but sometimes actually do lead to erroneous conclusions. For what scientific
proposition is better established than that a man once dead cannot come
to life again? And for what fact of ancient history is there better evidence
than for the Resurrection of Our Lord?
Yet one or other of these strong lines of argument must in this instance
lead to a false conclusion; the" contest of opposite improbabilities," whichever
way we resolve it, shows that it is on improbabilities and not on impossibilities
that our knowledge is founded. The truth is that we accept as giving knowledge
any argument of sufficiently high probability:7 a confident judgment based
on such an argument from known premises is regarded as knowledge when,
as is usually the case, it is true. But in the rare cases when the argument
leads to a false conclusion, the judgment being false can no longer be called
knowledge, although it is just the same sort of mental act as it would be
if it were true and were called knowledge. The process of judging on such
evidence is a fallible one and cannot even when true be supposed to be
the apprehension of a fact.
Knowledge, therefore, as we ordinarily use the word does not mean
apprehension; science and history are not forms of apprehension, and we
do not pursue them in the hope of apprehending, for no such hope could
possibly be justified.
The sphere of apprehension cannot therefore be so wide as that ofknowledge
in the ordinary sense, but its friends may still claim that it includes [all]
our direct knowledge and all that we have rigidly deduced from it. But
such a claim cannot be sustained, if rigid deduction or demonstration is used
in the ordinary sense in which anyone who constructs or follows a correct
mathematical proof is held to have demonstrated his conclusion; for although
the conclusion must be true if the premisses are true and the proof is really
correct, the proof may nevertheless seem correct to its author and yet its
conclusion be utterly mistaken; so that the final judgment that the conclusion
is true, depending as it does on the argument merely seeming and not actually
being correct, is not infallible and even when true is not a case of apprehension.
For instance a man may correctly conclude by simple addition that if the
inhabitants of the four quarters of a town number 2037, 3156, 1432, and
614 respectively, then the total population is 7239, but a judgment obtained
 KNOWLEDGE AND OPINION 59

in such a way would not be a case of apprehension, for a similar but false
state of mind could perfectly well arise from a mistake in arithmetic. If
any knowledge founded on argument is apprehension, it can only be that
founded on an argument which is not merely in fact valid, but whose validity
is itself apprehended. The apprehension of the validity of an argument would
be a case of direct knowledge to which we shall come shortly; it is obviously
something we can never achieve except perhaps in the simplest cases; in
no argument of any [interest or] complexity can we hold the whole process
before our mind at once and see that it is right with an infallible insight.
I at least cannot even see that 57 and 47 make 104. I can only work it out.
We come now to direct knowledge, of which the principal kinds 8 are
perception, memory and insight into abstract truths. The easiest one to begin
with is memory.
I remember having coffee for breakfast this morning: my memory is a
judgment, and according to the ordinary use of words it is knowledge; and
also it is direct knowledge in the sense that it is not inferred from any other
judgment. If I had forgotten it, I could [no doubt] have inferred it from
[many] other things which I remember, but if I had lost all my memories
I should have no way of inferring to the past at all, and memory must be
regarded as an independent form of knowledge. The question is whether
in such a case a memory can be an apprehension, for instance can I apprehend
the past fact that I had coffee for breakfast? It seems to me clear that memory
cannot be apprehension, simply because it is fallible: not merely do we often
fail to remember (which has no bearing on the present question), but we
sometimes remember wrong. If now remembering right were apprehension,
it would have to be a mental process of a quite different kind from remembering
wrong, and it would be impossible to explain why it is so hard to tell one
from the other. For the man who confidently remembers wrong always thinks
that he is remembering right; and on the view in question this means that
he not only misremembers, but also misintrospects, since by examining the
nature of his present experience he ought to be able to tell that it is not
a memory at all, but merely a mis-memory.
The alternative view is that the experience of remembering that you had
coffee for breakfast is one which could conceivably occur when you had
not really had it, although the constitution of the mind or brain is such
that this very rarely happens. The clearer the memory and the smaller the
time interval between the event remembered and the remembering the less
often do the two fail to correspond. It may be the case that sufficiently
clear memories of sufficiently recent events are never mistaken, but even
so this would be merely a law of nature depending on empirical evidence
not a logical necessity. Those who hold that such memory is apprehension
imply not merely that it is never wrong, but that it could not be wrong
60 CHAPTER IV

from the nature of the case, the experience of remembering being something
that could not happen without the existence of the fact remembered, in the
same sort of way as there could not be a battle without combatants.
The only support for the view that memory is apprehension is derived
from the way we use the word "remember," as implying truth. So that if
a man remembers wrong, we generally say that he doesn't really remember
but merely thinks he does. The word remember thus means to make a certain
sort of judgment truly, and we have no word for the sort of judgment itself:
a peculiarity of language which makes us think that truth belongs essentially
to such judgments, and that if the judgment is false, some quite different
mental process must be occurring. It is however obvious that in this case
language cannot possibly be a safe guide, since the phrase we are driven
to, "he thinks he remembers,"9 does not properly express what we want
at all: it refers strictly not to the primary memory-like (or on the apprehension
view memory-"simulating") judgment that such and such happened in the
past but to a subsequent introspective thought about this first judgment. We
have remarked on this difficulty before in connection with the words know,
perceive, etc., which also imply truth, but in the case of know we avoided
it by substituting "be sure (or certain) of' which has no such implication.
With regard to most other kinds of direct knowledge we can argue similarly:
perception in the ordinary sense of knowledge of the external world is evidently
a fallible process. Although I am completely certain that I am now sitting
at a table and writing with a pen, it is clear that the mental processes which
constitute or produce this knowledge are of a kind which sometimes lead
astray. There are many illusions which show that such perceptions are fallible
and can not be apprehensions. But when we come to what Dr Broad calls
"inspection," our knowledge of our own sense data and bodily feelings, the
case may well be different. Thus Dr Broad comes to the conclusion that
there is no reason to doubt that inspection so far as it goes is infallible10
and if so, it would seem that in inspection we have at last found a process
of apprehension. But the value of Dr Broad's arguments on this point must
be considerably discounted by his subsequent assertion that what is called
Stumpfs argument that it is possible to have a series of colours or other
qualities each indistinguishable from its neighbors, though the two ends of
the series are evidently different "proves with almost complete certainty that
some pairs of sensa, which seem to us to be exactly alike in quality or intensity
must really differ in these respects."l1 For this seems at complete variance
with his previous conclusion since it follows that judgments of inspection
relating to pairs of sensa may perfectly well be mistaken, and inspection
at least of pairs of sensa, but if so surely of single sensa also, cannot be
an infallible process. [Without wishing to be dogmatic] It seems to me that
we here have a perfectly good reason for rejecting, let alone doubting, the
 KNOWLEDGE AND OPINION 61

infallibility of inspection, and deciding that here agam our search for
apprehension is fruitless.
The last possibility of our finding it seems to be in our insight into self-
evident truths; but here again the prospect seems unfavourable: the feeling
of self-evidence can be no guarantee of the truth of a judgment, for the
progress of science is continually obliging mankind to abandon what it took
for self-evident truths. [The axiom of parallels and the uniqueness of the
time system are sufficient examples from our own time to illustrate the recent
destruction of them.]
The Theory of Relatively has overthrown those "necessary forms of
thought" Euclidean Geometry and the single time series, and the Quantum
Theory has thrown doubt on the Law of Causation itself. It is true that
no similar revolution has subverted the laws of logic and arithmetic but that
is because these are not genuine judgments but mere tautologies.1 2
But though all judgments which feel self-evident cannot be true, it might
still be argued that some of them were real acts of insight and in our sense
apprehensions; others mere pretenders simulating generalizations from insight
or self-evidence which could naturally be prejudices or results of experience
sometimes true, but sometimes false. This hypothesis has, however, nothing
to recommend it; for the two classes of really self-evident acts of insight
and apparently self-evident judgments are not really distinguisable. The man
who makes the judgment cannot tell which it is, or he would not be liable
to think it an act of insight when in fact it is false, and the outside observer
can indeed tell that if it is false it must be a pretender and not really an
insight, but if it is true he has no way of telling one from the other. There
seems therefore no reason to hold that there is this separate class of really
self-evident judgments and not simply [true and false prejudices,] judgments
which feel self-evident <but> which (like all other kinds of judgments) are
sometimes true and sometimes false.
We have now been through the principal kinds of knowledge and in none
of them have we found a plausible place for the apprehension of facts, so
that the conclusion is suggested that in spite of its importance in some
epistemologies there is simply no such thing as this apprehension. This is
not to deny that there may be apprehension of terms, whether particular
or universal. We can (perhaps) apprehend the colour red and a particular
place in our visual field; against such apprehensious our arguments have no
force since in these cases there is no possibility of error: were there (per
impossible) no such colour as red, we should not be thinking falsely of it;
we simply couldn't think of it at all.!3 But when we come to the judgment
that a certain place is red, the possibility of falsehood shows that our judgment
cannot (even if true) be a relation between us and the fact that the place
is red, for to give this account of it is to make <no sense of> thinking of
62 CHAPTER IV

nothing or not thinking at all. What other account we should give of judgment
is a difficult problem, with which we must shortly deal; but it is a problem
for everyone even those who believe in apprehension. By separating off
apprehensions or infallible judgments, they may narrow the scope of the
difficulty to fallible judgments (since for infallible judgments they suppose
the answer is easy), but the difficulty still remains.
In our view no judgments are apprehensions and we shall argue later14
that this is not merely a limitation of the human mind, that it is not merely
beyond our capacity to apprehend facts, but that such a thing is in the nature
of the case impossible. But even if we are wrong in this, if there <indeed>
are apprehensions, their relative importance can easily be exaggerated. Even
if inspection or perception, for instance, are infallible the knowledge we
obtain by them must be remembered, if we are to use it for inference or
any other process of thought; and with memory the door is opened to fallibility,
and apprehension must give place to ordinary judgment. The typical judgment,
the premisses and conclusion of the typical inference; made in ordinary life,
and not apprehensions and a logic which regards inference as an advance
from apprehension to apprehension15 cannot <should?> have much bearing
on most ordinary thought. For our part we shall dismiss this apprehension,
and confine ourselves to ordinary fallible judgments. These, various as they
are, can be regarded as species of one genus: they do not fall into two classes,
knowledge and opinion whose propositional references are of quite different
kinds.
But before passing on we must consider for a moment the sort of objection
which would be made to our conclusion by some philosophers; if there is
no apprehension, no infallible mode of knowledge, they would ask, what
right we can have to be certain of anything? Thus Cook Wilson16 argues
that it is impossible for a man to decide falsely that the evidence for a proposition
is conclusive, for if such a false judgment were possible "we should never
be sure that any demonstration was true and therefore there would be no
such thing as demonstrative knowledge."
This kind of argument seems to me entirely mistaken. In the first place,
there is no doubt that people do make mistakes as to the conclusiveness of
evidence: and not merely stupid people but even the cleverest, as any teacher
or mathematical analysis knows. This fact, with the risk of error which it
involves, must simply be faced; it cannot be denied; nor can it be escaped
by making a distinction between the genuine apprehension of the conclusiveness
of evidence and the mere fallible belief in its conclusiveness. For even if the
genuine apprehension were infallible that would not banish the risk of error,
unless we had an infallible way of telling whether our conviction were a
genuine apprehension or a mere belief, and the facts show that we have no
such way. We sometimes make mistakes and it's no use pretending we don't.
 KNOWLEDGE AND OPINION 63

Our fallibility cannot, therefore, be explained away; but are its consequences
really so disastrous as Cook Wilson seems to suppose? I see clearly before
me a book-case; does the fact that men are occasionally the victims of illusions
mean that I have no right to be certain that what I see is indeed a bookcase?
Does the fact that men sometimes make mistakes in addition mean that if
I and an opponent have arrived independently at the same totals for one
bridge scores, we may not be certain that we are right? To such questions
common sense gives us a perfectly clear answer; illusions are so infrequent
that it is far best for men to be certain that their judgments of perception
are true, and to act accordingly. The possibility of illusion might be allowed
for by their having not a complete conviction but a conviction just a minute
fraction, say one part in a million, short of certainty; and in a sense that
is what they all have. I say, for instance, that I am certain that what I
see is a book case, but I am not so absolutely certain that I could not conceivably
be persuaded that I was wrong, if anyone could bring forward sufficiently
strong arguments to that effect. If one of my friends explained to me that
he had been conducting experiments on optical illusions at my expense, I
might be convinced by him that what I had seen was not a book case at
all. If, however, I put an absolute trust in my judgments of perception, I
could only come to the conclusion that my friend was a liar, and I should
have to take the <b>ally tricks of conjurers for demonstrations of <the> most
extraordinary physical phenomena. So also my arithmetical fallibility means
that r should not be wise to put too much trust in an unchecked addition,
but if two people get the same answer the chance that they got the same
wrong answer by coincidence is negligible and we may be reasonably certain
that they are right. But again not so certain as to refuse to listen to anyone
who claimed to have discovered the contrary.
Our so called direct knowledge, perception, memory, etc. is all of this
nature; we make such judgments with practical certainty which is not however
so complete, that we might not be brought to abandon them if they came
into conflict with other beliefs. In general I trust my memory, but if I
remembered writing something yesterday and found today that there was
something different in my m.s., I should be more likely to conclude that
my memory was defective than that the m.s. has changed in the night. Such
cases, however, are rare and the mass of our perceptions and memories are
never called into question. They are true and made with practical certainty,
a certainty which is justified in the sense that the instinct to judge in this
way leads to truth in the overwhelming majority of cases, and is therefore
one which it is advisable or rather essential for men to trustY Such judgments
being (nearly always) true, certain and justified may properly be called
knowledge, even though the processes which lead to them are not infallible.
Of course, on this view on the rare occasions when they lead to false judgments,
64 CHAPTER IV

those judgments will be justified even though false; but this is not really
a paradox, for everyone must admit that there are circumstances in which
error is humanly speaking unavoidable.
Some of our abstract or universal judgments may seem to be more certain
than this: nothing, we feel, could make us abandon them: but we must not
forget that some of the greatest revolutions in science are made by those
who manage to doubt what everyone has previously thought unquestionable.
The laws of logic are, in our view, in a different position: they are not
genuine judgment and offer no scope for doubt. The same is true of pure
mathematics, but mathematical propositions soon become too complicated
for us fully to appreciate and we easily make mistakes about the properties
of symbols when we cannot clearly conceive their meaning?18
It is essential in the theory of knowledge to realize that our ultimate
premisses are not absolutely certain but liable to be abandoned or modified
in the course of the inquiry. Some have denied this because it conflicts with
their narrow theory of inference as a passage from apprehension to appre-
hension. The syllogism, they say, would be impossible if our conviction of
the truth of one premiss could be upset by our subsequent conviction of
the truth of the second. 19 Quite so; but in fact we sometimes cannot draw
the conclusion of our syllogism, because we are sure that it is false, and
are driven back on our premisses, one of which must be abandoned. More
fundamental than the syllogism is what Mr Johnson calls the antiiogism20
which forbids us to conjoin in thought the three propositions all A are B,
S is A, S is not B. Generally we start by believing two of the three and
use the antilogism to reject the third; but sometimes this third proposition
too is something we believe, and we are given a hard choice as to which
to abandon. But unless we are willing sometimes to reconsider our premisses,
we are in the plight described by Hobbes as that of those who trust to
books, who "do as they, that cast up may little summs into a greater, without
considering whether those little summes were rightly cast up or not; and
at last finding the error visible, and not mistrusting their first grounds, know
not which way to clear themselves; but spend time in fluttering over their
books; as birds that entering by the chimney, and finding themselves inclosed
in a chamber, flutter at the false light of a glass window, for want of wit
to consider which way they came in. "21
Equally serious is the opposite error made by supporters of the Coherence
Theory of Truth, who think that there are no ultimate premisses but only
the need for a coherent system. But evidently we require of our coherent
system that it should embody as many as possible of the things we instinctively
believe. We should not regard as plausible a historical system, however
coherent, which contradicted all our memories: and held that memory was
an illusory faculty or even that it "went by contraries."
 KNOWLEDGE AND OPINION 65

NOTES

1. J. Cook Wilson, Statement and Inference (Oxford, 1926), p. 113.

2. This is not quite accurate; either the time of knowing must be introduced as a third
term to the relation or mind must be replaced by a mental event at that time. I neglect
this refinement throughout this chapter, and also speak inaccurately but, I hope, intelligibly
of a judgment as a relation [a term to which there are many objections (e.g., at least
we should say not relation but relational fact)]. More accurate language would, I think,
in this connection be unnecessarily clumsy and pedantic. The necessary distinctions will
all be made in good time and can be left till it becomes necessary later.
3. [I use 'a judgment' here for what is more properly called "the fact that the man judges
that so-and-so is the case in the particular way in which he does judge it" a fact of
Dr Moore's second class.
(See "Facts and Prepositions," Proceedings of the Aristotelian Society, Supplementary Volume
7, 1927.) I also use 'relation' loosely for relational fact.]
4. Cook Wilson uses 'apprehension' not only for our knowledge that something is the case,
but also for our relation to the terms of a judgment, the things we are knowing about.
It is only the former with which we are here concerned.
5. To avoid possible misunderstanding it should be said that I use certainty in its so-called
"psychological" sense in which it merely means that the thinker is certain not that the
matter itself is in any logical sense certain.
6. This term is meant to include so-called direct knowledge for which no grounds are
supposed to be needed; it might be more accurate to say "not insufficiently grounded,"
but this seems a needless pedantry.
7. What is meant by an argument of high probability will be discussed in Ch..... Roughly
it is an argument of a kind which when applied to true premisses nearly always gives
a true conclusion.
8. There may be other kinds such as inspiration and telepathy, but I cannot discuss these
here.
9. We also say "I think I remember" when we have an image and are not sure whether
it is a memory image or not, or if a memory image, whether it is an image of something
we experienced or merely of something we were once told about. In this case we are
not making a confident memory judgment, nor are we thinking we are making such
a judgment so that [in fact J we neither know or think we know.
10. C.D. Broad, The Mind and its Place in Nature, (London, 1925), pp. 300-303.
11. Ibid., p. 409.
12. See p ..... Ref. to Whately.
13. But suppose, the reader may suggest, we are thinking of a unicorn; there is no such
thing as a unicorn, but yet we can think of it. We shall see below (p. .. .. ) that in
such a case we are really thinking of the description or characteristic of having one
horn; which there certainly is even though there is nothing answering to the description
or having the characteristic.
14. See p.....
15. Cf. H.W.B. Joseph "The Growth of the Perception of the External World," Mind,
Vol. 38 (1929), pp. 26-42; see p. 39.
16. Cook Wilson, Statement and Inference, op cit., Vol. I, pp. 107-108.
17. For a further treatment of the nature of this justification and knowledge see below.
18. See below p .....
19. Cf. Cook Wilson, Statement and Inference, op. cit., Vol. II, p. 452. He is arguing against
66 CHAPTER IV

the coherence theory of truth, but the argument seems to apply as well against our
theory of certainty.
20. Cf. W.F. Johnson, Logic (3 Parts; Cambridge, 1921122124). Part II, p. 78.
21. Thomas Hobbes, Leviathan, Part I, ch. 4.
 (001-13)

CHAPTER V

JUDGMENT AND TIME

[or? TIME AND THE MIND]

In this chapter I propose to interrupt the main course of our argument in

order to develop some very general considerations about events in time. These
have a bearing on our present problem because a judgment is an occurrence
or an event in time, and anything that can be said about events in general
will apply to judgments in particular, and may help us to elucidate the logical
category to which they belong. We shall, in fact, find that the word event
is used in two different senses, and that there is no agreement as to which
of these senses is applicable to judgments.
It is easiest to begin with physical events: to the unsophisticated mind
the world consists of enduring substances whose qualities and relations are
liable to change. Thus the same tree may be green in the summer and brown
in the autumn, and when we say the tree is green we are apt to think that
we are expressing a fact which has only two terms, the tree and its quality.
But from these two terms no complete fact can be formed for our sentence
"The tree is green" has no definite truth unless we understand a particular
time at which it is green, in general, of course, the time at which we are
speaking. In the same way when we say "The tree was green" we either
mean that it was green at a definite time or that it was green at some
time or other, i.e. that there was a time at which it was green. In any
case, the fact to which we refer has at least three terms, a substance, a
quality and a time. This means that its colour is not really a quality or
characteristic of the tree in the way that being prime, for instance, is a
characteristic of the number seven. "Seven is prime" has a complete sense
without any reference to a time, the 'is' being a pure copula. From "the
tree" or "green" we cannot make a complete sense in this way. We appear
to do so when we say "the tree is green", because we have smuggled in
reference to a time in the word 'is' which is no longer a pure copula but
means "is now." The same is of course true of spatial or other relations
between bodies. "The cat is on the mat" expresses a fact with four terms:
the cat, the mat, the time and the relation expressed by 'on'.
68 CHAPTER V

In ordinary language, which is based on this view of the world as a collection

of substances with changing qualities and relations, we speak of an event
whenever there is an important change in these qualities or relations. Thus
when the moon comes between the earth and the sun we say there is an
eclipse, and instead of saying "at five o'clock on June 10th the moon was
between the sun and the earth"; we can say "at five o'clock on June 10th
there was an eclipse of the sun"; Although such usage is familiar, intelligible
and convenient, there is still a certain difficulty in saying what exactly an
eclipse (or any other event) is. The eclipse, we might perhaps say, was the
moon's being between the sun and the earth, or the fact that the moon
was between the sun and the earth, but it would not always be good English
simply to substitute either of these phrases for "the eclipse" in any sentence
in which it occurs. For instance, the fact that at a certain time T the moon
was between the sun and the earth, is a fact about the time T, whereas
we should never sayan eclipse was [a fact] about a time, but that it happened
at a time. 1 This is because we forget that such facts as that the moon was
between the sun and the earth, have times among their constituents; we
think and speak of the fact or eclipse as consisting in a spatial relation between
the sun, moon and earth, and then of the further fact that this first fact
happened at a certain time. 2 But we have seen that this conception, which
lies at the bottom of our linguistic usage, is essentially erroneous, and that
no fact can consist simply in a spatial relation between the sun, moon and
earth, since the time at which the relation holds is always another constituent.
Our ordinary language cannot really be made consistent in any way; we
say "The eclipse happened at 5 o'clock," meaning "The moon was between
the sun and the earth at 5 o'clock", in which since "at 5 o'clock" occurs
in each sentence. "The eclipse happened" must stand for "the moon was
between the sun and the earth." But when we speak of "two eclipses" we
don't mean "two moons between the sun and the earth," but two times
at which the same moon was between the same sun and earth. And since
it is the times which are two, we seem obliged to identify the eclipse with
the time, i.e. with 5 o'clock in the above example, which was the one thing
it seemed before that it was clearly not to be identified with.
But these vagaries of language are of no real importance, except to warn
us not to take grammatical usages as an accurate reflection of the meaning
they express. On all the occasions when we talk of an eclipse of the sun
we are concerned with the fact that at a certain time the moon is or will
be between the sun and the earth, and if we are to use the word in philosophical
analysis, it seems to me that it is such a fact 3 which we must take the word
'eclipse' to mean.
Events, then, on this view of the world are facts about the qualities and
relations of substances at times. That there are such facts is certain; what
 JUDGMENT AND TIME 69

is not so clear is that this analysis, which makes them consist in the holding
of relations between substances and times, is a complete one; that substances
and times are among the ultimate constituents of the world. Of course, no
one would think that the bodies of ordinary life, cows and chairs and socks,
could not be analysed further; or that the fact that a sock is on a chair
was a simple relationship between two entities (and a time) and not a
complicated fact about the spatial relationships of millions of particles. 4
But it might be held, indeed at sometime it used frequently to be held,
that some of the ultimate physical facts were that such and such particles
(atoms or what not) had such and such spatial relations to one another at
such and such times, and that physical events consisted in the holding of
these relations. This view was compatible both with the relative and with
the absolute theory of space, (since the spatial relations between particles
might either be direct or derived indirectly from the spatial relations of the
points in absolute space which the particles occupied), but it presupposed
inevitably an absolute theory of time, although this was not always realised.
For if time is relative, the relations of earlier and later hold primarily
between events, and to say that an event happened at a certain time means
that it happened simultaneously with or before or after certain other events,
times being defined in terms of events. But on the view we are considering,
events are facts about times, and times cannot be defined in terms of events
without circularity. Besides this it can be seen that time cannot be derived
from relations between (or qualities of) substances; for the time series is
essentially asymmetrical: if tl is before t2, t2 is not before t 1, but after it.
But if one set of relationships between substances occurs before another,
there is nothing to prevent it also occurring again after the other. 5
The absolute theory of time is, however, no longer tenable after Einstein's
discoveries, and we have to replace it by a conception of the world as a
four-dimensional manifold or 'space-time'. The elements of this manifold,
which are called point instants, are generally taken to be derived by a process
which Dr Whitehead has called "extensive abstraction" from events, which
are, as it were, regions or volumes of space time, and are taken to be the
ultimate constituents of the world. These events are the terms of spatio-
temporal relations, and instead of substances we have long events or series
of events with certain persistent characters, which belong to each event of
the series or each temporal slice of the long event. It may seem that this
is a mere change of name, for what other difference is there between a
"long event" and a "substance"? But the change of name involves a change
of conception corresponding to the modern assimilation of time and space;
for the old theory of substance treats time and space differently, by allowing
a substance spatial but not temporal parts. If a wall is red in one place and
blue in another, that is because one part of the wall is red and another
70 CHAPTER V

blue; but if the wall is red at one time and blue at another we do not
think that this means that one temporal part of it is red and another blue,
for we suppose, or at any rate talk as if, it were the very same thing which
was both red (at one time) and blue (at another). We have seen how this
refusal of temporal parts to substances involves us in [the absolute theory
of time] the difficult view that the fact that a thing is red at a time is
an ultimate kind of triple relationship between a thing, a colour, and a time;
and so in the absolute theory of time. Now this <is a> difficulty which simply
does not arise in the case of space, since that a wall is red at a certain
place, obviously means that part of the wall is red, and that that part is
at the place, i.e. (on the relative view) has certain spatial relations to other
things. No one would postulate an ultimate kind of relationship between
the whole wall, the colour and the place; the colour so obviously belongs
not to the whole wall but only to part of it.
The change from "substance" to "event" means that we can adopt the
same analysis for time as for space. The new term is chosen because an
"event" can be supposed to have temporal parts, without any straining of
language; for instance, though we cannot speak of the earlier part of the
wall we can perfectly well speak of the earlier part of its history; and the
new "event" which we substitute for the wall is supposed to have temporal
parts like the wall's history as well as spatial parts like the wall. For a wall
to be red at one time and blue at another, means that one part event in
its history is blue and another red; and to say that these events happened
at such and such times is to say that they have certain temporal (strictly
spatio-temporal, since space and time cannot be separated) relations to certain
other events. Time is now defined in terms of events, for events are temporal
parts of substances supplying the terms required for the relations of a relational
theory of time to hold between. Before this was impossible, since substances
had no temporal parts and what were called events were facts about substances
and times, and times could not be defined in terms of such events without
circulari ty.
On this view the identity of a substance through time is constituted by
spatio-temporal continuity and the permanence of certain characters (or
sometimes their systematic variation) throughout its history. What makes
the different temporal parts of a fountain pen all parts of the same thing,
is much what makes the spatial parts of a lawn all parts of the same lawn,
namely continuity and permanence of character, in the one case through
time and in the other through space. 6
The belief in the abstract identity of a substance through time has been
the source of many fallacious arguments purporting to prove the immortality
of the soul, which as Mr Russell has remarked would ... J
Dr Broad in Mind and its place in Nature (pp. 588-598) argues that the
 JUDGMENT AND TIME 71
theory that the notion of "material event" can be taken as fundamental,
and the notion of "material substance" as derivative, owes its plausibility
to the tacit assumption of Absolute Space; and he shows, as if it were a
matter of some difficulty, how it can be maintained without assuming
"substantival" but only "adjectival" Absolute Space and Time. Since his view
appears to be contrary of that taken in the present work according to which
the theory that events are primary does not, and the theory that substance
is primary does require Absolute Time, some consideration of his arguments
seems to be required.
As regards time nothing that he says show the slightest need for an Absolute
Theory, whether substantival or adjectival, since nothing is easier than to
replace the temporal qualities which Dr Broad ascribes to events by the
relations of simultaneity, before and after. (The last two he must have in
any case in some form or other.)
As regards space the position is more complicated: the difficulty about
the motion of a homogeneous liquid is a genuine one but easily overcome
in the "Field Theory of Matter" in which the motion is represented by
appropriate terms in the matter-energy tensor.
Besides this, there is a difficulty which only occurs if you start with
"instantaneous punctiform particulars" and not with extended events. This
difficulty is as follows; if you reject both persistent substances and absolute
space, it may seem hard to see how two instantaneous states of the world
are in any way connected together,s though the difficulty is at once removed
if we postulate direct spatial relations between events at different times. But
even the resemblance of a difficulty vanishes when we adopt the method
of extensive abstraction (called by Dr Broad himself "the Prologomena to
every future Philosophy of Nature").9 For we can now define the different
events which form [part of] the history of a thing as a series whose terms
overlap, instead of a series with continuously varying spatial qualities,lO and
as this is the only connexion in which these absolute spatial qualities are
required, they can evidently be dispensed with altogether.
We must now consider the relation of the events, which, on the new
view, are the ultimate constituents of nature to what we previously called
events when we were discussing the substance theory. An event we then
decided, was a fact about the relations (or qualities) of substances at a time:
for instance, an eclipse was the fact that at a certain time the sun, moon
and earth were in line. For such facts, let us quite arbitrarily use the word
"occurrence", in order to distinguish them from the events of the new theory,
which we will continue to call "events".
Of course, the abandonment of the substance theory does not mean that
there are no longer any such things as occurrences; no theory can deny that
there are eclipses, i.e. that sometimes the moon comes between the earth
72 CHAPTER V

and the sun. What has changed is our view as to the ultimate analysis of
such facts; for instance the eclipse is no longer held to consist in a relationship
between substances and a time, but in there being events in those series
of events known as the histories of the sun, moon and earth, which have
certain spatio-temporal relations both to one another, and to other events
by reference to which they are dated. The occurrence is, thus, not an event,
but a fact about events, and when, in questions of philosophical analysis, we
are concerned with something which we naturally call an event, we must
consider carefully whether it is in our special terminology, an event or an
occurrence, for these are fundamentally different categories.
In the case of an eclipse the occurrence is a fact about several events,
which makes the distinction clearer than when, as often happens, the occurrence
is a fact about one event only. In such cases, we are accustomed to use
the same phrase for both event and occurrence without any consciousness
of ambiguity. For instance, "the murder of Caesar" often stands for an
occurrence, and could be replaced by "the fact that Caesar was murdered."
Thus in, "he was aware of the murder of Caesar," or "the murder of Caesar
was due to his ambition," what the man was aware of, and what was caused
by Caesar's ambition, was not an event, i.e. a region of space-time, but
a fact, the fact that Caesar was murdered. This fact, properly analysed, consists
in their <sic for there> being an event at a certain time and place having
certain characteristics and being part of such and such series or long events.
And this event is also called "the murder of Caesar" when we say "the
murder of Caesar preceded the accession of Augustus by - years", for an
interval of time is a relation between two events not one between facts
or occurrences.
It should be realized that our distinction between the occurrence or fact
that there is an event of a certain sort, and the event itself, is not in any
way pedantic or unnecessary; in other cases similar distinctions are so obvious
that no one would dream of bothering to point them out. It requires, for
instance, little subtlety to distinguish between Signor Mussolini and the fact
that Italy has a Duce, and events and occurrences are really as distinct as
the dictator and his dictatorship.
We must now return to the mental events such as judgment with which
we are primarily concerned. Here again no philosophical system can deny
the existence of what we may call mental occurrences. People at various
times judge that various things are the case, and such a fact as the fact
that so-and-so judges such-and-such at a certain time we may call a mental
occurrence, just as the fact that the moon was between the earth and the
sun at a certain time is called a physical occurrence.
The existence of such occurrences is therefore, beyond dispute, but as
soon as we try to analyse them, and ask what are the terms and relations
 JUDGMENT AND TIME 73

involved in them, we get a difference of opinion analogous to that between

the substance and event theories of physics. On the substance theory of mind
the occurrence called judgment is a relation between an ego, the object or
objects judged or judged about, and, we must add, the time at which the
judgment happens; the theory being only tenable in conjunction with an absolute
theory of time. The alternative theory, however, introduces into the analysis
a mental event, which is at the time in question, i.e. has certain temporal
relations to other events; [and which] is owned by or belongs to the person
judging, and has certain relations to the object or objects judged or judged
about. Thus that Mr Smith judged yesterday that the earth is flat is to be
analysed, on this view, as meaning that there was a certain sort of event
temporarily related to the present in the way indicated by yesterday, related
in a certain way to the earth and flatness, and further it was an event in
Mr Smith's mind. The meaning of this last clause, that it was an event in
Mr Smith's mind is still left vague: it may mean that the event was owned
by a certain "pure ego," or that it is connected by certain relations to certain
other mental events or to a certain body, or as Dr Broad has suggested
it may mean that the event has a certain "mental position." But whichever
of these views be taken the occurrence consists not in a relation between
the objects Mr Smith's pure ego and the time, but in one between the objects
and an event which happens at the time and is part of Mr Smith's mental
history.
It must be noticed that this distinction between the substance and event
theories is not the same as that between "pure ego" theories and others;
a view which rejects the pure ego must indeed, be in our sense an event
theory, but one which accepts it need not be in our sense a substance theory,
for the pure ego may be introduced to 'own' and so to collect together
the mental events which belong to one mind. The essential point is rather,
that on the event theory when someone judges that the earth is flat, there
is a mental event which is his judging and which is not merely the fact
that he judges; an existent is as Dr Broad would say, not a subsistent. ll
Between these two views of mind, the substance and the event, we do
not need at present to decide. When we have got further with our analysis
we shall find it quite impossible to give a satisfactory account of judgment
on the basis of the substance theory, so that we may forbear to urge here
the general reasons which make that theory unplausible. What we have to
do now is to point out that if the event theory be adopted, we must guard
against an ambiguity in the word 'event' analogous to that which we found
in discussing physical events. Again we must distinguish the 'occurrence', e.g.
the fact that Mr Smith judges that the earth is flat and the 'event' involved
in the fact: the occurrence being, as before, the fact there is such an event.
Which of these shall we call his "judgment"? Language gives us no
74 CHAPTER V

unambiguous guide. His judgment is at a certain time, an attribute which

belongs properly to the event, for the fact is a fact about the time. But
if he simultaneously judges two different things, we should speak normally
of "two judgments" without implying that there were two distinct events,
instead of, perhaps, only one event with relations of reference to two different
objects or sets of objects. 12
On the whole, it seems best to use ''judgment'' for the occurrence i.e.,
for the fact that the thinker judges; then whatever view we take of the
dispute between substance and event theories, whenever anyone judges that
A is B at a time T, we shall have a judgment, namely the fact that he
so judges at that time.!3 On the event theory this fact will consist in there
being an event of a certain sort in his mind at that time having relations
of a certain sort to A and B; this event will not be, in our terminology,
the judgment, and we will call it, arbitrarily, the content of the judgment.
If, however, the event theory be false there will be no such thing as this
content, and we shall not be able to make much use of the notion until
that theory has been established.
Before we leave the subject of "judgment and time" it may be as well
to discuss a supposed difficulty that has led McTaggart14 to deny that time
is real, and Dr Broad15 to assert that judgments about the future are neither
true nor false.
Dr McTaggart distinguishes what he calls the A series defined by past,
present, and future, from the B series defined by before and after, and contends
that the A series is essential to Time, that it is not analysable in terms of
the B series, and further that it is self contradictory since every event has
the three incompatible characteristics of pastness, presentness and futurity.
It seems to me that to see that all this is nonsense we have only to reflect
on the meaning of the words "present" or "now". By "the present time"
a man means the time at which he is speaking, and by an event "now"
he means one simultaneous with his speech. Hence at every time at which
they are used these terms have a different meaning; they may fully compare
to the words I and you, which have a different meaning for every person
who uses them. Every person calls himself '1', the man he is speaking to
'you', and other people 'he'; and similarly every time calls itself 'the present',
earlier times it calls 'past' and later ones 'future'. There is no more mystery
in an event being both 'present', (when mentioned at one time) and "past"
(when mentioned at another) than in a man being both '1' (when spoken
of by one man (himself) and 'you' (when spoken to by another). To infer
that the time relationship is unreal is like concluding that the relationship
of conversation must be unreal because in it people have the incompatible
characteristics of Iness and youness. 16
Really the whole idea is so silly that it is hard to see what makes it
 JUDGMENT AND TIME 75

plausible, for it can hardly arise simply from the ambiguity of 'now' though
this may well be a contributing factor. It is due mainly, I think, to the
inevitable difficulty of imagining a temporal series of events. One way of
doing this is to go through the events one after another in the order in
which they happened, as when one mentally rehearses a tune. But this method
is often unsatisfactory as we may want to have the events in our mind not
one after another, but all at once so as better to see their relations. We
then imagine them spread out before us along a line, replacing the time
order by a space order. This method of representing a temporal series of
events by a spatial series of images or symbols correctly reproduces the
properties of the time series with the exception of its 'sense', the difference
between 'before' and 'after'. This can only be represented by the <sic for
an> arbitrary convention such as that "to the left of' is to stand for 'before',
and not for 'after'. The representation being thus imperfect, forfeits to
convenience much of the essence of the time series, and we try to correct
this by combining it with the other mode of representation (a representation
by a time series) and imagine the events spread out before us being let up
in turn by the bullseye lantern of presentness. We then suppose presentness
to be a real quality which moves down the series, and invent with Mr Dunne!7
another time series for it to move in, or adopt the less logical hypotheses
of McTaggart or Broad. But the whole difficulty comes from combining
two disparate modes of representation. We can represent the time series
of events, by a time series of images which come one by one in succession;
this gives the time quality correct but is inconvenient for purposes of study.
Or we can represent the time series by a space series which we can contemplate
at our leisure, but which is after all spatial and not temporal. But if we
combine the two ways, if we both have all the images spread before us and
light them up one by one, we are representing one series by two and langing
ourselves in a hopeless fog if we use such an inaccurate symbolisation in
any philosophical discussion;!8 but in practical matters the double representation
may be the most convenient, since we can attend to either aspect at will,
and enjoy now the sense of succession, now the repose which enables us
to look backwards and forwards till we find what we want. So long as
we merely alternate the two methods no harm is done, but ideas based on
their combined and simultaneous use are inevitably nonsensical.

<Editors' note: The following paragraphs are evidently an older version

of the text beginning three manuscript pages earlier with "Before we
leave ... " and were replaced by Ramsey himself. The page at issue originally
bore the corresponding number in his manuscript (no. 20). Nonetheless
the discussion differs sufficiently to be worth adding here.>
76 CHAPTER V

Before we leave the subject of "judgment and time" it may be as well to

discuss certain difficulties which have led Dr MacTaggart19 to deny the reality
of Time and Dr Broadzo to assert that judgments about the future are neither
true nor false, contentions which are clearly fatal to the opinions we have
been advancing. The difficulty as put by Dr Broad is that every event has
the three incompatible characteristics of pastness, presentness, and futurity,
and changes from being future to being present and from being present to
being past. As Mr Braithwaite21 and others have urged, this difficulty seems
to me a mere mistake.
Change has been defined by Mr Russell22 thus: "Change is the difference,
in respect of truth or falsehood, between a proposition concerning an entity
and the time T, and a proposition concerning the same entity and the time
T, provided that these propositions differ only by the fact that T occurs
in the one where T occurs in the other." Thus if a poker is hot at time
T, and cold at time T it has changed.
The fact that events change from being future to being past can, of course,
be dealt with in this way. To sayan event is future at time T means that
that event is later than the events which occur at time T; which is in no
way incompatible with its being later than the events at time T. The difficulty
is supposed however to arise when we investigate the meaning of present;
the future is what is later than the present, the past what is earlier, but
what is the present? What is this quality of presentness which is passed on
from event to event?
The answer is that presentness is not a quality at all, any more than "you-
ness" is a quality.

NOTES

1. Cf. G.E. Moore, "Facts and Propositions," in "Mind, Objectivity and Fact," Proceedings
of the Aristotelian Society, Supplementary Volume VII (1927), p. 176.
2. See for an instance of this W.E. Johnson, "Analysis of Seeing," Mind, vol. 27 (1918),
p. 142 and G.E. Moore, Proceedings of the Aristotelian Society, vol. 21 (1920-21), p. 136.
But reference above shows that he has changed his view since then.
3. But not necessarily exactly the fact mentioned, which is to some extent indeterminate
or general, but the absolutely specific fact that the particles of the sun, moon and earth
are in such and such definite relations.
4. It will, no doubt, be objected that since we can know that the sock is on the chair
without knowing anything about the particles, these cannot come into the analysis of
the fact we know, the fact, namely, that the sock is on the chair. This is perfectly
true, and what is said refers strictly not to the fact about actual or possible sensations
which a man knows when he knows that a sock is on a chair, but to the specific physical
fact, which is the possibly unknown necessary and sufficient condition for what he knows
to be true.
 JUDGMENT AND TIME 77

5. Cf. the first part of Mr Russell's article on "Is Position in Time and Space Absolute
or Relative," in Mind, vol. 10 (1901), pp. 293-317. It need hardly be mentioned that
this article does not give his present views.
6. For an admirable discussion of the relations of events and objects see A.N. Whitehead's
An Enquiry Concerning the Principles of Natural Knowledge (Cambridge, 1919) and Concept
of Nature (Cambridge, 1920).
7. <Ramsey broke off at this point.>
8. This difficulty seems to be the one exploited by Kant in the proof of the First Analogy.
9. C.D. Broad, Scientific Thought (London, 1923), p. 4.
10. Cf. C.D. Broad, Mind and its Place in Nature (London, 1925), p. 597.
11. Broad, Mind and Its Place in Nature (op. cit.), p. 559. Of his two forms of the Pure Ego
theory the first is in our sense a substance theory, the second an event theory.
12. Cf. G.E. Moore, "Facts and Propositions," in "Mind, Objectivity, and Fact," Proceedings
of the Aristotelean Society, Supplementary Volume VII (1927), pp. 176-7.
13. A fact of Prof. Moore's first class. Op. cit., p. 174.
14. McTaggart, The Nature of Existence (Cambridge, 1927), Vo!' 2, Ch XXXIII.
15. Broad, Scientific Thought (London, 1923), Ch. 2. See also R.B. Braithwaite, Proceedings of
the Aristotelian Society, Supplementary Volume VIII (1928), pp. 162-174.
16. Dr Broad's theory that the future is unreal is, of course, no solution at all, supposing,
that is, that the difficulty is geuuine. For the contradiction that the death of Queen
Anne was both future (up to 1714) and past (after 1714), he merely substitutes the
contradiction that there both was no such event as Queen Anne's death (up to 1714)
and was such an event (after 1714).
17. J.W. Dunne, An Experiment with Time. (London, 1927).
18. Russell's argument against Neutral Monism based on Emphatic Particulars (Monist, July
1914) seems to involve just this muddle. Since his conversion to Neutral Monism no
more has been heard of it, although in 1919, it was the argument on which he placed
most reliance (Monist, 1919, p. 55).
19. MacTaggart, Nature of Existence, Vo!' 2, Ch. XXXIII.
20. Broad, Scientific Thought, Ch. 2.
21. R.B. Braithwaite, Proceedings of the Aristotelian Society, Supplementary Volume VIII (1928),
pp.162-174.
22. Bertrand Russell, Principles of Mathematics, sect. 442; quoted by McTaggart, op. cit., p. 14.
 APPENDIX I

OLDER DRAFTS
 (001-1S)

CHAPTER I

INTRODUCTORY

It is a commonplace that logic, aesthetics and ethics have a peculiar position

among the sciences; whereas other sciences are concerned with the description
and explanation of what happens, these three normative studies aim not at
description but at criticism. To account for our actual conduct is the duty
of the psychologist; the logician, the critic and the moralist tell us not how
we do, but how we ought to think, to feel and to act.
Nevertheless, in order to criticise, it is first necessary to understand; we
cannot approve or condemn mental processes without carefully analysing their
nature, and it is both customary and reasonable to include under logic the
investigation of many psychological preliminaries, about which the psycho-
logists themselves have very little to teach us - in fact, the aspect of thought
with which logical criticism is concerned, what is called below its propositional
reference, is one which psychologists have only dealt with in the vaguest
possible way.
But for the logician the analysis of thought is only a prelude to criticism,
to an examination, that is, of how far different processes of thought conform
to some standard of value, how far they are true or valid. And just as the
most fundamental problems in the other normative studies are to define their
standards of value, to explain what is meant by 'good' or 'right' or 'beautiful',
so the logician has not only to say what thoughts are true and valid, but
above everything to explain what he means by these epithets.
Indeeci, in regard to the primary logical value of truth, all the logician
can do is to determine its meaning; it is for him [alone] to tell us what
truth is, but which opinions are true we shall learn not merely from logic
but from all the sciences, each in its own domain. If logic were to undertake
to distinguish true judgments or opinions from false ones it would have to
include the whole of knowledge. This, of course, has never been regarded
as its function; it has criticised not opinions in regard to their truth; but
inferences or arguments in regard to their validity or reasonableness. It is
interested not so much in our beliefs themselves as in the arguments which
82 APPENDIX I: OLDER DRAFT VERSIONS

lead us to hold them; the man who is led to a true opinion by a bad argument
finds no favour with the logician, whose criticism is directed not to the
primary value of truth in judgment, but to the secondary value of reasonableness
in argument. 1
That the value of truth is primary and of reasonableness secondary is
a point on which we must insist; for it follows from the fact that the whole
purpose of argument or inference is to arrive at true judgments, that the
reasonableness of an argument must be connected with some tendency in
it to lead to true conclusions. In fact, the validity of arguments is [derived
from] related to the truth of opinions, in somewhat the same way as the
rightness of actions is related to the intrinsic value of their consequences.
But the precise nature of the connection must be left to be determined later
when we come to the actual analysis of the concept 'reasonable'.
With regard to the meaning of any concept of value such as goodness,
[beauty,] truth or validity there are three main schools of opinion which
may, perhaps, be called idealist, realist and naturalist. With the idealists such
as Green, Bradley and Bosanquet I shall not deal; their writings seem to
me to be almost entirely nonsense; the living issue is between the realists
and the naturalists. [It is, of course, possible to take different views of two
different kinds of value, but the arguments in all cases are so much the same
that if they were set out clearly, this would only rarely happen.]
Some philosophers, of course, are realists for one sort of value and naturalists
for another, but this is generally due to confusion, since the arguments for
realism are so similar in all three fields, that if they were clearly apprehended
they would be found convincing either in all cases or in none.
In regard to ethics the opposition has been clearly set out in the writings
of Dr Moore; in his Principia Ethica he convicted of "the naturalistic fallacy"
all those who sought to define 'good' in terms of any natural attribute such
as the pleasurable, the approved or the desired. He argued that 'good' stood
for none of these things but was the name of a simple [indefinable] unanalysable
quality, which was very different from all ordinary qualities, stood in a sense
'outside' the natural world, and formed the distinctive subject matter of ethics.
Dr Moore's attitude, in contrast to the naturalistic views which he condemns,
I shall call, for want of a better word, realistic. [Its essence does not lie
exactly in the declaring that good is unanalysable, because, for instance, another
ethical predicate, right (as applied to actions) was said by Dr Moore to be
analysable in terms of good, an action being right if it has the best possible
consequences. ]
By this term <'realism'> I shall not mean his general tendency to declare
<as> unanalysable concepts which others have claimed to be able to analyse
[although that is certainly a general tendency of this school of thought];
but [simply] his belief that ethical valuation involves a distinctively ethical
 INTRODUCTORY 83

predicate not belonging to the natural world. It is this rather than the simplicity
of 'good' that I take to be characteristic; thus I shall call realists in logic
not only people who hold that truth or validity are unanalysable, but also
those who think that they involve in their ultimate analysis one or more
distinctively logical predicates or relations, such as a unique kind of cor-
respondence or indefinable probability relations.
[The principal upholders of naturalism in logic in recent years have been
the pragmatists. ]
Opposed to this almost universal view, is the position defended by the
pragmatists and behaviourists, and in his recent works by Mr Russell2 , that
all the notions used in logical criticism can be defined in terms of natural
qualities and relations, such as are involved in ordinary psychological in-
vestigations. This naturalistic view of logic will be taken in the present work,
in which I hope to show how truth and reasonableness can be defined without
assuming any unique unanalysable logical relations, and further to show that
the current modes of explaining logic by means of such relations are altogether
untenable.

NOTES

1. [It remains to be examined later whether the reasonableness of arguments can be determined,
independently of the knowledge of actual fact provided by actual fact. J It remains however
to be seen whether, as the above paragraph suggests, the validity of arguments is really
independent of the positive knowledge provided by science. We shall show later that
certain types of valid arguments are in this sense a priori while others are not.
2. The position of Mr Russell, as we shall see later, is peculiar; in general he is a naturalist
but he appears to accept Mr Keynes' probability relations in a way altogether at variance
with his <OWll> main position.
 (001-01)

CHAPTER I

THE NATURE OF TRUTH

(Preliminary)

What is truth? What character is it that we can ascribe to an optmon or

a statement when we call it true? This is the first question we have to ask,
and before trying to answer it, it may be as well to reflect for a moment
on what it means. Above all we must distinguish clearly our question, what
is truth, from the quite different question, what is true? If we asked what
was true, the sort of answer we should hope for would either be as complete
an enumeration as possible of all true opinions [the truth about everything],
i.e., an encyclopaedia, or else a test or criterion of truth, a method by which
we could know a truth from a falsehood. But what we are asking for is
neither of these things, but something much more modest; we do not hope
to learn an infallible means of telling true statements from false ones, but
simply to know what it is that this word "true" means. It is a word which
we all understand, but if we try to explain it, we can easily get involved,
as the history of philosophy shows, in a maze of confusion.
One source of this confusion can be eliminated straight away. Besides
the primary meaning in which we apply it to statements, beliefs and opinions,
the word true can also be used in a number of derived and metaphorical
senses, which it is no part of our problem to discuss. Obscure utterances
such as "Beauty is Truth, Truth Beauty" we shall make no attempt to elucidate,
but confine ourselves to the plain workaday sense in which it is true that
Charles I was beheaded and that the earth is round.
We might at first sight suppose that the things which are true or false
in the most fundamental sense are indicative sentences or statements, but
very little reflection is needed to see that whether statements are true or
false depends on what people mean by them, that is to say on the thoughts
or opinions which they are used to express; and we are obliged to regard
truth as primarily an attribute of thoughts and opinions and only derivatively
of sentences. 1 What are ultimately true and false are therefore states of mind,
using that term in a very wide sense to include both mental acts and mental
 THE NATURE OF TRUTH 85

dispositions. 2 But, of course, not all states of mind are true or false but only
such as have what I shall call a propositional reference.
As typical of a mental state with a propositional reference we may take
a belief; a belief is necessarily a belief that something or other is so and
so, for instance that the earth is flat, and it is this aspect of it, its being
"that the earth is flat", that I propose to call its propositional reference.
So important is this character of propositional reference, that we are even
apt to forget that belief has any other aspects or characters at all. Thus
if two men both believe that the earth is flat, we say that they have the
same belief in spite of the fact that they may believe it at different times
for different reasons and with different degrees of conviction, and express
their beliefs in different languages, words or images; if the propositional
references are the same, if they are both beliefs that the same thing, we
commonly ignore all other differences between them and call them the same
belief. It is usual in philosophy to express this resemblance between two
men's beliefs, not by saying, as I do, that they have the same propositional
reference, but by calling them beliefs in the same object, objective or
proposition, and there have been endless disputes about what kind of entity
or fiction this object or proposition is, as it is only the hardiest verbalists
who can persuade themselves that "that the earth is flat" is the name of
something real. Our change of language wi11 not, of course, enable us to
escape these difficulties, but it will make it possible to postpone them for
a while and to conduct our present discussion without presupposing any
particular solution of them. Whatever the objective of a belief may be, its
character of reference is undoubtedly real, and by basing our treatment on
this character, as something with which we are all familiar, we can [postpone
the difficulty until later when the character of reference must be analysed
<and so>] simplify the problem considerably.
Propositional reference is not, of course, confined to beliefs; my knowledge
that the earth is round and my opinion that free trade is superior to protection,
any form [in fact] of thinking of knowing or being under the impression
that - [has a propositional reference, and all of these are either true or
false, except knowing which is ... ], any mental state in fact which is either
true or false has a propositional reference, and in discussing truth and falsity
it is with such states alone that we sha11 be concerned. On the other hand
it is not the case that any state with propositional reference must be either
true or false. I can wish it would be fine to-morrow, wonder whether it
will be, and finally believe that it wi11 be; these three states a11 have the
same propositional reference but only the belief can be called true or false.
We do not call wishes, desires or wonderings true; not because they have
no propositional reference, but because they lack what may perhaps be ca11ed
an affirmative or assertive character, the element that is present in thinking
86 APPENDIX I: OLDER DRAFT VERSIONS

that, but absent in wondering whether. At the opposite pole to this affirmative
character, there is perhaps a negative character present in acts of denial or
disbelief; these we do not usually call true or false, though we can call them
correct or incorrect, words which can for the present be regarded as definable
in terms of true and false as applied to affirmative states.
Such an account may not be ultimately correct and will have to be
reconsidered when we come to deal with negation, but for our present
preliminary purposes we can say that denying that so and so is the case
is correct or incorrect, according as asserting that it is the case is false or
true, and confining our attention to affirmative states with propositional
references, say that it is these and these alone which are either true or false. 3
And of these we shall take, for convenience, beliefs as typical.
[If, then, a state of mind has an affirmative character and a propositional
reference it will be either true or false; and we have to discover the meaning
of the ... J
It is important to realise that the words "true" and "false," when properly
used, are not vague terms indicating praise or blame of any kind, but have
a quite definite meaning. There are various respects in which a belief can
be regarded as good or bad: for instance it can be held for good or for
bad reasons, and it can either be isolated or part of a coherent system of
knowledge, but it is not to these merits or defects that we refer when we
call the belief true or false. To take an old example of Mr Russell's,4 someone
who thought that the present Prime Minister's name began with B would
think so truly, even if he had derived his opinion from the mistaken idea
that the Prime Minister was Lord Birkenhead; and it is clear that by calling
a belief true we neither mean nor imply that it is either well grounded,
or comprehensive, and that if these qualities are confused with truth, as they
are by Bosanquet,S any profitable discussion of the subject becomes quite
impossible. The kind of merit in a belief to which we refer in calling it
true can be easily seen to be something which depends only on the beliefs
propositional reference;6 if one man's belief that the earth is round is true
so is anyone else's belief that the earth is round, however little reason he
may be able to give for thinking so.
After these preliminaries we must come to the point: what is the meaning
of 'true '? It seems to me that the answer is really perfectly obvious, that
anyone can see what it is, and that difficulty only arises when we try to
say what it is, because it is something which ordinary language is rather
ill-adapted to express.
Suppose a man believes that the earth is round; then his belief is true
because the earth is round; or generalizing this, if he believes that A is B,
his belief will be true if A is B, and false otherwise, [and it is clear, that
this connection is not a mere coincidence but arises from and constitutes
 THE NATURE OF TRUTH 87

the nature <of truth> and it is clear that what is meant by 'true' is simply
this - "believing that A is B when A is B." To believe truly is to believe
that A is B, when A is B, or to believe that A is not B, when A is not
B, or to believe that either A is B or C is D, when either A is B or C
is D, and so on, through all the possible forms of belief.]

<Editor's note: The next four pages of MS 001-01, namely numbers 8,

9, 10, and 11 in Ramsey's pagination, were moved by him into the later
MS 001-02 as pages 12, 13, 14, and 15 thereof. We also give them here.>

It is, I think, clear that in this last sentence we have the meaning of truth
explained, and that the only difficulty is to formulate this explanation strictly
as a definition. If we try to do this the obstacle we encounter is that we
cannot describe all beliefs as beliefs that A is B since the propositional reference
of a belief may have any number of different more complicated forms. A
man may believe that all A are not B, or that if all A are B, either all
Care D or some E are F, or something still more complicated. We cannot,
in fact, assign any limit to the number of forms which may occur, and must
therefore be comprehended in a definition of truth; so that if we try to
make a definition to cover them all it will have to go on forever, since
we must say that a belief is true, if supposing it to be a belief that A is
B, A is B, or if supposing it is to be a belief that A is not B, A is not
B, or if supposing it to be a belief that either A is B or C is D, either
A is B or Cis D, and so on ad infinitum.
In order to avoid this infinity we must consider the general form of a
propositional reference of which all these forms are species; any belief whatever
we may symbolise as a belief that p, where "p" is a variable sentence just
as "A" and "B" are variable words or phrases (or terms as they are called
in logic ). We can then say that a belief is true if it is a belief that p, and
p.7 This definition sounds odd because we do not at first realize that "p"
is a variable sentence and so should be regarded as containing a verb; "and
p" sounds nonsense because it seems to have no verb and we are apt to
supply a verb such as "is true" which would of course make nonsense of
our definition by apparently reintroducing what was to be defined. But "p"
really contains a verb <already>; for instance, it might be "A is B" and
in this case we should end up "and A is B" which can as a matter of ordinary
grammar stand perfectly well by itself.
The same point exactly arises if we take, not the illustrative symbol 'p',
but the relative pronoun which replaces it in ordinary language. Take for
example "what he believed was true." Here what he believed was, of course,
something expressed by a sentence containing a verb. But when we represent
it by the pronoun 'what', the verb which is really contained in the 'what'
has, as a matter of language, to be supplemented again by "was true." If
88 APPENDIX I: OLDER DRAFT VERSIONS

however we particularize the form of belief in question all need for the
words "was true" disappears as before and we can say "the things he believed
to be connected by a certain relation were, in fact,S connected by that relation."
As we claim to have defined truth we ought to be able to substitute
our definition for the word 'true' wherever it occurs. But the difficulty we
have mentioned renders this impossible in ordinary language which treats
what should really be called pro-sentences as if they were pro-nouns. The only
pro-sentences admitted by ordinary language are "yes" and "no", which are
regarded as by themselves expressing a complete sense, whereas "that" and
"what" even when functioning as short for sentences always require to be
supplied with a verb: this verb is often "is true" and this peculiarity of
language gives rise to artificial problems as to the nature of truth, which
disappear at once when they are expressed in logical symbolism, in which we
can render "what he believed is true" by "if P was what he believed, p".
So far we have dealt only with truth; what about falsity? The answer
is again simply expressible in logical symbolism, but difficult to explain in
ordinary language. There is not only the same difficulty that there is with
truth but an additional difficulty due to the absence in ordinary language
of any simple uniform expression for negation. In logical symbolism, from
any proposition symbol p (corresponding to a sentence), we form the
contradictory -Ip (or-p in Principia Mathematica): but in English we often
have no similar way of reversing the sense of a sentence without considerable
circumlocution. We cannot do it merely by putting in a "not" except in
the simplest cases; thus "The King of France is not clever" is ambiguous,
but on its most natural interpretation means "There is a King of France
but he is not clever" and so is not what we get by simply denying "The
King of France is clever"; and in more complicated sentences such as "if
he comes, she will come with him" we can only deny either by a method
special to the particular form of proposition, like "if he comes, she will
not necessarily come with him" or by the general method of prefixing "It
is not true that ... ", "It is false that - " or "It is not the case that ... ", where
[ again] it looks as if two new ideas, "truth" and "falsity," were involved,
but in reality we are simply adopting a round-about way of applying not
to the sentence as a whole.
Consequently our definition of falsity (to believe falsely is to believe p,
when -Ip) is doubly difficult to put into words; but to argue that it is circular,
because it defines falsity in terms of the operation of negation which cannot
always be rendered in language without using the word "false", would simply
be a confusion. "False" is used in ordinary language in two ways: first as
part of a way of expressing negation, correlative to the use of "true" as
a purely stylistic addition (as when "it is true that the earth is round" means
no more than that the earth is round); and secondly as equivalent to not
 THE NATURE OF TRUTH 89

true, applied to beliefs or other states of mind having propositional references

or derivatively to sentences or other symbols expressing those states of mind.
The use we are trying to define is the second, not the first, which in the
guise of the symbol _IP we are taking for granted and propose to discuss
later under the head of negation. 9

<Editors' note: MS 001-01 resumes here with

MS page 12 in Ramsey's pagination.>

In the last pages we have been defining truth and falsity in terms of propositional
reference, which we have taken for granted. For instance, our analysis of
what is meant by "At 10 o'clock Mr. Jones had a true belief' is roughly
"At 10 o'clock Mr. Jones was believing that certain things were related in
a certain way, and they were related in that way," in which we are leaving
altogether unanalyzed the notion of believing that - , i.e., of propositional
reference. The analysis of this will be our task later, and for the present
all I wish to do is to mention the possibility that this analysis will throw
the nature of truth in a new light; we have seen, we may say, how any
analysis of believing that can be extended at once to give an analysis of
believing truly; but it may turn out that we have really been putting the
cart before the horse and that we can only analyse believing that by reference
to believing truly. I do not think myself that this is so, but since it is the
view of some philosophers 1o it cannot be ruled out in advance.
[But while we must for this reason regard our definition of truth as to
some extent merely provisional, it ... ]
But this doubt does not take us very far. At worst it means that our
definition of truth is redundant and capable of simplification, for nothing
can make us doubt that a true belief is one which is a belief that p, when
p. Even if this is not the correct definition, it still expresses something which
arises necessarily from the nature of truth, something with which the proper
definition is certainly compatible. And so this is all we require to take our
bearings with regard to some famous doctrines, such as Pragmatism and the
Correspondence and Coherence Theories. For this purpose we may sum up
our own discussion of Truth as follows: If with regard to a belief we know
what it is a belief that, for instance, that it is a belief that the earth is
round, then we know in what case it is true and in what false, in the instance
given it is true if the earth is round, false otherwise. This is absolutely obvious
and a theory of truth which contradicts it is quite certainly wrong.
On the other hand, there is another question, the question as to what
constitutes propositional reference (i.e., what is meant by saying a belief
is a belief that so-and-so), which we have not discussed, so that we are
not yet in a position to reject any answer to it which may be suggested.

89
90 APPENDIX I: OLDER DRAFT VERSIONS

Let us call these two questions I and II, Question I being "Given propositional
reference, what determines truth?" and Question II "What constitutes
propositional reference?"
The first main type of theories of truth are Correspondence Theories;
and they have primarily to be divided according to whether it is Question I
or Question II that they hold should be answered by reference to cor-
respondence. Thus Mr Russell in Philosophical Essays and his Home University
Library volume <on The Problems of Philosophy> only uses correspondence to
answer Question I, but in his The Analysis of Mind, in definite disagreement
with his earlier view, he introduces correspondence in his answer to Question II.
To Question I the correspondence theorists give substantially the same
answer as ours. They too think that Mr Smith's belief that A is B is true
if A is B, and they say that in such a case the truth of the belief is constituted
by a relation between the fact that Mr Smith thinks that A is B on the
one hand, and the fact that A is B on the other.!1 This relation they call
"correspondence," but nothing precise can be said about its nature until we
know how to analyse the fact that Mr Smith thinks that A is B, i.e., until
we have answered question II. It is clear that this theory amounts in a sense
to the same thing as ours, and we shall have to discuss later whether [it
is a more correct way to word it] and how it is preferable.
We shall find that ours is a superior method of statement in that we
are able to avoid mentioning either correspondence or facts. One advantage
of this can be easily seen: suppose what is believed is a disjunction, as when
Smith thinks that Jones is either a liar or a fool: this belief, we say, is true
if Jones is either a liar or a fool, and surely this is a better way to explain
truth in such a case than one which introduces "the fact that Jones is either
a liar or a fool", since although such a phrase is permissible as a matter
of language, there is no such fact, no such either-or in reality but either
the fact that he is a liar or else the fact that he is a foo1. 12
The correspondence answer to Question I we shall, therefore, regard as
a cumbrous and possibly inexact method of expressing our answer, but it
is a useful phrase, which leads to very little harm, to speak of a true belief
as one which corresponds to a fact, [leaving vague] defining the type of
correspondence no further than as being that type (or types, since there may
be different with different forms of belief) of relation that there [obviously]
must be between thinking that such and such is the case and such and such
actually being the case.
But the correspondence theory is as we have said, [also] sometimes put
forward as an answer to Question II; it is held that to believe that A is
B means to have in your mind an entity or entites corresponding to or
representing A being B; by "correspondence" being here meant some specific
relation such as resemblance. When we come to discuss it in Ch. II we
 THE NATURE OF TRUTH 91

shall find that in spite of its unpopularity this view is part, though not the
whole, of the truth.
[As regards Question I we can, then, say that we agree with the
correspondence answer, but propose so to simplify it as to eliminate all
mentioning of correspondence, but as regards Question II we must postpone
the discussion of it; we shall find later that the true answer is very complicated
and contains correspondence only as one of its elements. [If further we are
asked what kind of correspondence it is that occurs is referred to in ... ]]
The next theory of truth to be considered is Pragmatism, the view that
by a true belief is meant one which is useful. By the inventor of Pragmatism,
C.S. Peirce, this was, I think, given primarily as an answer to Question II,
a belief that A is B, being roughly a belief leading to such actions as will
be useful if A is B, but not otherwise.!3 Although this is not the whole
analysis of propositional reference, I think that like correspondence it gives
one element in it, and is not simply to be mocked at. And when stated
in the above crude form it leads to the conclusion that a belief is true if
and only if it is useful, and this conclusion is by no means ludicrous but
in complete harmony with my account of truth. For a belief that A is B,
means on this view, a belief which is useful if and only if A is B; such
a belief will therefore be useful if and only if A is B, i.e., if <and only
if> it is true; and so conversely it will be true if and only if it is useful.
What is ludicrous, is not the general idea, although in such a crude form
this is obviously incorrect, but the way in which William James confused
it especially in its application to religious beliefs. This application depended
on using the theory in just those cases in which its crudeness and need for
complication and refinement are most glaring and so turning a sensible sort
of answer to Question II into an absurd answer to Question 1.
Let us take, for example, the belief in hell. To say a man believes in
hell means, according to the pragmatists that he avoids doing those things
which would result in his being thrown into hell, and which he would not
avoid on account of any other consequences they may have. Such conduct
will be useful to the man if it really saves him from hell, but if there is
no such place it will be a mere waste of opportunities for enjoyment. But
besides this primary utility there are other ways in which such conduct may
or may not be useful to the man or others; the actions from which a belief
in hell would cause him to abstain might bring disasters in their train either
for him or for others even in this present life. But these other consequences
of the belief, whether useful or not, are clearly not relevant to its propositional
reference or truth, and if the pragmatist definition makes them relevant this
only shows that the definition must be corrected by qualifying clauses or
even abandoned altogether. William James, on the other hand, included
explicitly these further kinds of utility and disutility, which must obviously
92 APPENDIX I: OLDER DRAFT VERSIONS

be excluded if pragmatism is to have any plausibility, and thought that the

truth of the belief in hell depended not on whether hell in fact existed but
on whether it was on the whole useful for men to think it existed. Thus
even a belief that hell exists, he made its truth depend not on whether hell
really exists but on something quite different, and so gave an answer to
Question I incompatible with what is obviously the case.
Such absurdities, however, form no part to the essential pragmatist idea,
even if they constitute its chief attraction to some minds; and although it
cannot be appealed to to contradict what we have said in answer to Question I,
in regard to Question II, we shall see that pragmatism has a considerable
contribution to make.1 4
<But on our main issue, that of Question I, pragmatism has nothing to
contribute. For it maintains "that A is B is true if and only if it is useful
for us to adopt "a belief that A is B."> It is obvious, however, that this
cannot be the meaning of "a belief that A is B"; for example in saying
that a man believes that Bacon was the author of Shakespeare's plays we
cannot mean that he has habits whose utility depends on Bacon being the
author. Since only then would the belief that A is B be true (i.e., useful)
if and only if A were B. But such a definition of reference <sic for truth>
could not conceivably be right unless we limited the notion of utility in
a way which would destroy most of the attractions of James' views. Let
us take, for instance, the belief that hell exists; if hell does exist, such a
belief has a direct utility in saving people from it by preventing them from
sinning. And according to the pragmatists when we say that a man believes
in hell we mean that he avoids doing those things which would result in
his being thrown into hell and which he would not avoid on account of
any other consequences that they may have. Such conduct will be useful
to him if it really saves him from hell, but if there is no such place it will
be a mere waste of opportunities for enjoyment.
And whatever the complete definition <of truth> may be, it must preserve
the evident connection between truth and reference that a belief "that p"
is true if and only if p. We may deride this as trivial formalism, but we
cannot contradict it without absurdity, and we should value the necessity
of conforming to it as a check on our deeper investigations; for the disregard
for this check is one of the main sources of the errors of the pragmatists.
For the pragmatists had a laudable desire for an account of truth which
went deeper than the mere formal reduction of truth to reference, and they
contended that the meaning of an idea (or as we might say the reference
of a belief) depended on and was to be defined by reference to the habits
of action to which it led, and that in the case of a true idea these habits
were such as would pay. Truth was therefore defined by William James
as "the expedient in the way of our thinking".1 5 But such a definition could
 THE NATURE OF TRUTH 93

only be compatible with the connection between truth and reference, if we

assumed a very strange definition of reference. For we should have to say
that calling a belief, a belief "that A is B" we meant that it was such as
would be useful if A were B but not useful if A were not B.
But, whatever may be the relative values of a mere reduction of truth
to propositional reference, and a definition of it in terms of more primitive
ideas, they are two problems which it is essential to distinguish; for if they
are not distinguished, and if the essential relation of truth to propositional
reference is overlooked, philosophers are liable to give a vague and inaccurate
definition of truth, in terms not of propositional reference but of the supposedly
more primitive notions of utility and habits of action, and then use this definition
either to establish a false relation between truth and propositional reference,
or to deny that there is any close relation between them, and in either case
falling into patent absurdity.

NOTES

1. This may need some qualification in view of the possibility that when we think in words,
the words may be not so much an expression of our thought as a part or even the
whole of the thought itself.
2. See infra, Ch. dII "Judgment''>.
3. Assumptions entertained without belief are also called true or false by an extension of
meaning easy to explain. For refutations of other theories on which truth is ultimately
independent of the affirmative character of mental acts, see below.
4. <Bertrand Russell, The Problems of Philosophy (Oxford, 1912), p. 205.>
5. [Bernard Bosanquet, Logic, Vol. II, 2nd ed. (Oxford, 1911), pp. 282 ff. He even confuses
a true statement with one made in good faith.] Footnote to go in here on next page.
d t is given in Chapter I as note 5 thereof.>
6. It has been suggested by Professor Moore that the same entity may be both a belief
that (say) the earth is round and a belief that something else; in this case it will have
two propositional references and may be true in respect of one and false in respect
of the other. I think that everything I say could easily be altered so as to allow for
this possibility, but the complication of language which would result seems to me to
outweight the possible gain in accuracy.
7. In Mr Russell's symbolism

B is true:=: (3p). B is a belief that p & p. Df

8. In a sentence like this "in fact" serves simply to show that the ratio obliqua introduced
by "he believed" has now come to an end. It does not mean a new notion to be analyzed,
but is simply a connecting particle.
9. See below. <Presumably a reference to the projected chapter on negation is intended.>
10. E.g., the late Dr McTaggart (The Nature of Existence, Vol. I (Cambridge, 1921), p. 11,
sect. 10), who held that a true belief was one which corresponded to a fact, and a
belief that p was one which either corresponded (if true) or merely "professed to
94 APPENDIX I: OLDER DRAFT VERSIONS

correspond" (if false) to a certain fact. On this view believing truly that p is evidently
a simpler notion than believing that p; [but in the absence of any account of the meaning
of "professing" it is not one which need be taken very seriously] and much the same
would follow from some of Mr Russell's language in the The Analysis of Mind (London,
1921), pp. 272-273.
11. The fact that A is B, is something that exists if and only if A is B. Hence if the belief
is false there is no such corresponding fact.
12. <The paragraph originally reads as follows:> [<It is clear> that this is decisive in its
favour. One simple consideration in this sense can be urged at once: suppose that what
is believed is a disjunction as when Smith thinks that Jones is either a liar or a fool;
this belief as such is true if Jones is a liar or if he is a fool, and surely this is a better
way to explain its truth in such a case, than one which introduces "the fact that Jones
is either a liar or a fool," since although such a phrase is permissible as a matter of
language, there is a sense in which there is [evidently1 no such [disjunctive1 fact, no
such mere either-or in reality, but <only> either the fact that he is a liar or else the
fact that he is a fool.]
13. Wanted: Note on Peirce. <Ramsey's principal access to Peirce was via Morris R. Cohen
(ed.), C.S. Peirce: Chance, Love, and Logic (New York, 1923).>
14. <At this point, Ramsey initially proposed to close the chapter with the following,
subsequently deleted, passage: [The last well known theory of truth is the Coherence
Theory; unlike the others this affords us no assistance whatever in answering either
of our questions but derives its plausibility entirely from [certain] confusions. It is evident
that coherence does not in general determine either the propositional reference or the
truth of beliefs. The beliefs of a man suffering from persecution mania may rival in
coherence those of many sane men but that does not make them true; besides, coherence
is a character [property] of a system, and truth and reference are characters of single
beliefs.]>
15. William James, Pragmatism: A New Name for Some Old Ways of Thinking (New York and
London, 1907), p. 222.
 (001-09)

CHAPTER II

THE COHERENCE THEORY OF TRUTH

The third well-known theory of truth is the coherence theory which is

absolutely irreconcileable with ours, since it holds that the truth of a belief
that A is B depends not on whether A is B, but on how far the belief
that it is forms part of a coherent system. The absurdities which result from
this view have been most amusingly set out by Mr Russell in his essay "On
the Monistic Theory of Truth" in Philosophical Essays,! but although this reductio
ad absurdum is a conclusive refutation of the theory, it is still, I think, instructive
to consider some of the arguments used by its adherents both in defending
their own and in attacking rival positions. According to Mr Russell the fallacy
in all these arguments lies in their assuming the axiom of internal relations
[and are therefore invalid], but it seems to me that part of what his opponents
say does not depend for its plausibility on this axiom and may be worth
a fresh investigation. Without going into any great detail it will, I think,
be possible to show that the Coherence Theory is largely based on a serious
confusion the unfortunate consequences of which extend far beyond the present
context. Since, unlike the Correspondence Theory and Pragmatism, the
Coherence Theory is almost entirely concerned with Question I and hardly
at all with Question II, I propose to say all I have to say about it in the
present chapter.
The Coherence Theory, as usually expounded,2 is not confined to the
contention that truth is to be defined as, or by reference to coherence, but
makes an assertion whose consequences are much more far-reaching; for it
maintains that no single judgment3 can be absolutely true, in the sense of
having a meaning which would be included without modification in the whole
truth, so that "the truth of 'true' judgments is essentially the truth of a
system of knowledge and the 'truth' of systems of knowledge is borrowed
from the Ideal Experience which is struggling for self-fulfillment in them."4
[This contention evidently goes far beyond the mere statement that truth
is to be defined by coherence, which would still allow truth to be a predicate
of single judgments although not ... ]
96 APPENDIX I: OLDER DRAFT VERSIONS

This contention that no single judgment can be wholly true must, of course,
be distinguished from the much more plausible contention that no single
judgment can be made with complete certainty. It may be, that I cannot
be really certain that Charles I was beheaded, but this does not prove that
that he was beheaded is not wholly true and part of the whole truth.
Nevertheless truth and certainty are often confused and those different persons
who "make no claim to absolute truth" frequently mean no more than that
they are not absolutely certain.
The contention, then, that we have to consider is that no single judgment
can be absolutely true, and if this connection is right all that we have hitherto
said must be wrong, or at the least, require serious modification, for we
have assumed that any belief or judgment has a definite propositional reference
or meaning in virtue of which it is without qualification true or false. The
most obvious reason for doubting this assumption lies in the possibility that
judgments may be vague in the same sort of way as the words we use to
express them. How many hairs must a man have not to be bald? There
is no definite number, and if someone is judging that he is bald, his judgment
may be vague and not definitely either true or false. 5 But this cannot be
our opponents' ground of objection, since a system of vague judgments would
be no less vague than its members; for their contention is that such a system
is capable of truth in a higher degree than its members, and the obstacle
they see to absolute truth can not lie in vagueness. 6
It will be replied no doubt? that single judgments have some meaning,
but not the full meaning that they have in a system of knowledge: and so
to assess the value of this reply we may use the comparison made by Prof.
Joachim between a system of knowledge and a symphony. The notes in a
symphony, it is contended, are not what they would be apart from the
symphony, but are determined by their mutual relations. This is true in the
sense that what note comes next depends on what would best continue the
melody etc., but it is not true if it means that each note has not its own
definite pitch. If pitch were not a property of single notes we could not
talk of a symphony at all in the sense in which we speak of Beethoven's
Fifth Symphony, as opposed to certain particular performances, for the Fifth
Symphony is essentially a system of notes each of a definite pitch, which
could occur also in other symphonies or even in tuning up.s [Similarly, if
each judgment had not its own meaning. To say that the pitch, a note has
in the symphony, was different from that which it would have outside is
absurd; for what is it?] So also each judgment in a system of knowledge
must have a definite meaning, which could also occur outside that system.
Otherwise "system of knowledge" would be just a meaningless phrase.
 THE COHERENCE THEORY OF TRUTH 97

NOTES

1. <Bertrand Russell Philosophical Essays (London, 1910), pp. 150-169; see esp. pp. 156-160.
See also C.D. Broad "Mr Bradley on Truth and Reality," Mind, vol. 23 (1914), pp. 349-
370; see esp. pp. 361-363.
2. See especially Harold H. Joachim's The Nature of Truth; (Oxford 1906), and Bernard
Bosanquet's Logic (2nd ed.; Oxford, 1911), especially Vol. II, Chap. IX, which will be
cited as Joachim and Bosanquet respectively. Also see F.H. Bradley, Essays on Truth and
Reality (Oxford, 1914).
3. 'Judgment' is the word used by these authors, where we have been using 'belief, and
its 'meaning' is what we have called "propositional reference".
4. Joachim, op. cit., p. 89. For his defense of the position see p. 85-113.
5. Vagueness must not be confused with indeterminateness. "He has fewer than 50 hairs"
is less determinate than "He has exactly 37 hairs," but not vague in the way that "He
is bald" is.
6. The following analogy may perhaps be helpful, though it should not be pressed too far.
The dependence of the meaning of a judgment on its context may be compared to the
way in which the colour of one part of the visual field may be affected through contrast
by the colour of a neighboring part. And just as each part of the field has its own colour,
whatever may be the cause of its having that colour, so each judgment has its own meaning
and according to that meaning is true or false.
7. Joachim, op. cit., p. 102.
8. This is not to say that a symphony is a collection of notes; what kind of a system it is
is irrelevant; all we require is the plain truth that the elements of the system are notes.
 (001-03)

<CHAPTER lIb

<JUDGMENT>

The truth or falsity of a mental state we found to depend upon its propositional
reference, and our next problem is to analyse or explain this characteristic;
but before starting on this, it is as well to consider certain preliminary
distinctions between the states of mind in question.
The most important of these is that between acts and dispositions. In ordinary
language we use knowledge, belief, and opinion primarily of dispositions;
"he knows he's got to leave at the end of the term," "he believes that when
he's dead he'll go to heaven," "in his opinion Protection would cure
unemployement"; in all these cases we are not speaking of an act in the
man's mind with a definite date, but of a general state or disposition. When
we say he knows he's got to leave or he knows his multiplication table,
we are talking of enduring dispositions of his mind, manifested at times in
particular acts of knowing, but conceived as existing even when not so
manifested, just as a man is called courageous even when not at the moment
displaying his courage. Again by an act or acts of thinking at a certain date
we "form an opinion," and the opinion so formed remains our opinion until
we 'change' it; although we may never think about the matter in the meantime,
our opinion is supposed still to be there and ready to be given to anyone
who happens to ask for it.
But we also have other words which refer not to dispositions but to definite
dateable (but not necessarily instantaneous) acts of mind. Thinking, as in
"I was just thinking that its going to rain," (but we should never say "I
was just knowing that 2 + 2 = 4"), judging, inferring, asserting, perceiving,
discovering and learning all refer to acts not to dispositions. And we can
also use knowledge and opinion in this sense in the combinations "coming
to know" and "forming the opinion."
It is clear that in common language both acts and dispositions can be
called true or false, and that both have in some sense propositional references.
But it seems also clear that the fundamental use of truth and reference is
that in which they are applied to acts, for whatever is the correct account
 JUDGMENT 99

of dispositions, they must obviously be defined by reference to the acts in

which they are manifested (or would be manifested if occasion arose), and
the truth or falsity of the disposition arises from that of the acts and not
vice versa. In what follows we shall, therefore, be primarily concerned with
mental acts not with mental dispositions, but before passing on there are
various characteristics of the latter which require our notice.
I have a dispositional belief, or perhaps I should rather say knowledge,
that the Cambridge Union is in Bridge Street. This belief is very rarely
manifested in an act of thought; I do not often have occasion to judge that
that is where the Union is; I only do this when I have to inform a stranger,
or just now when I thought of it as an example. On the other hand my
belief is frequently manifested by my turning my steps that way when I
want a book from the Union library. In general, a belief is a disposition
not only to make a certain kind of judgment on suitable occasions but also
to behave in certain ways in pursuing the pursuit of certain ends. It shows
itself not only in acts of thought, but also in acts of body or will. Now
suppose I left Cambridge for some time, and while I was away the Union
moved to a new site. On my return I might turn my steps to Bridge Street
for the same purpose as before and when I got there exclaim with surprise,
"But I thought the Union was here."1 However, I probably shouldn't really
have thought about it at all, but merely gone there according to myoId
habit. So without ever saying it to myself or otherwise actually thinking
it, I can behave as if the Union were in Bridge Street; and this behaviour
can be regarded as true or false according as the Union is in Bridge Street
or not. In the phrase regarded by Cook Wilson2 as appropriate to such a
case, I can be under the impression that it is in Bridge Street and this impression
can be true or false.
It might seem from this that we had to consider thoughts and actions
in the ordinary sense as two quite different kinds of acts having propositional
reference; [namely] that my going to Bridge Street and my saying to myself
"The Union is in Bridge Street" could both be true and must both be
distinguished as definite acts from my general states or dispositions. But this
would, I think, be a mistaken view; it is not possible to take a piece of
my conduct and regard it as having a definite propositional reference in
the same way as a piece of my thinking has. Take my going to Bridge
Street; in <doing> this we said I behave as if the Union were there, but
also as if the Union had a library, and as if the book I wanted were contained
in that library but in no other nearer one from which I could borrow <it>,
and as if the library would still be open and so on indefinitely. My conduct
is the result or manifestation of my whole system of dispositional beliefs.
The idea that a piece of conduct has one particular propositional reference
only arises when one of the beliefs on which it is based turns out to be
100 APPENDIX I: OLDER DRAFT VERSIONS

mistaken. In general the beliefs on which we act are true, but when just
one of them turns out to be false, as for instance when the Union has moved,
our attention is fixed on that one and our conduct condemned as erroneous
in one particular respect.
Weare thus led to the following conception; we have on the one hand
particular acts of thinking, knowing or judging that, and on the other our
dispositional beliefs or knowledge. These dispositional beliefs manifest them-
selves in two ways: firstly by giving rise to corresponding judgments when
occasion arises for making them, and secondly by governing our actions,
roughly according to the rule that we perform those actions which if our
beliefs were true would have the most satisfactory consequences.
Dispositional beliefs themselves can be caused by acts of thought, as when
we form an opinion by thinking the matter over; but they can also be caused
in other ways. Thus if a man sees what looks like a lump of iron and stoops
to pick it up, he makes the bodily adjustments necessary for lifting a heavy
weight, and if it turns out to be a light shell he is surprised and says he
thought it was a solid lump of iron. But save in exceptional cases he didn't
really think about it at all, his bodily adjustments being due to 3 a merely
dispositional belief that it was solid, which caused him to behave as if it
were solid, and would have caused him if asked to say it was. (At least
that would be his first impulse, though on reflection he may have become
doubtful.) Now this dispositional belief, or as it is called by Dr Broad4 quasi-
belief, is not caused by a corresponding judgment or other thinking in the
past. If the man has never seen this lump before he has never thought about
this lump before, and his quasi-belief about it cannot be caused by any previous
thought about it. It is obviously due to his experience of similar lumps in
the past, but it is not even necessary that he should have thought about
those, since he might evidently get into the way of making such adjustments
without ever thinking at all. In view of its temporary nature, it may seem
absurd to call such a belief or quasi-belief a disposition at all, especially
as a way we suggested of distinguishing acts from dispositions was by their
definite dates. But it is evident that this is not the real ground of distinction
and the man's belief that the lump is solid is fundamental[ly], analogous to
my belief that the Union is in Bridge Street. Each of them is known by
its influence on actions and neither of them is an act of thinking; nor is
my belief about the Union strictly always the same enduring disposition.
For what is relevant is every day something different, namely that on that
day the Union is in Bridge Street, and the belief which primarily influences
my action is every day a new one.
To give an adequate account of dispositional beliefs or dispositional
knowledge would be a very difficult task. Their logical status is far from
clear. Dr Broad speaks as if they were identical with the actions to which
 JUDGMENT 101

they lead. This cannot be quite right even in the simplest cases. But it can
well be held that in speaking of them we are really talking about actions
and thoughts, a statement about a dispositional belief having some complicated
meaning of that sort. In this case they would be what Mr Russell calls "logical
constructions"; alternatively they may be entities or characters inferred or
hypothecated to explain our actions and thoughts; [and] the problem of their
status is very analogous to that of the unobservable entitites in physics, and
I may perhaps be forgiven for not discussing it here, especially as our primary
concern is with acts of thought, and not with dispositions. The reason I
have lingered so long on the latter is that I do not think it is possible to
understand the propositional references or truth or falsity of thoughts without
considering the effects they have on our acting either directly or indirectly
through dispositional beliefs. In particular, the phrase "degree of belief' as
it occurs in the theory of probability can only be understood by taking into
account the effects of belief on action.
There is, however, an objection that may be made somewhat in the following
terms. "The distinction you make is radically unsound. The bodily adjustments
of the man lifting the weight you ascribe to a dispositional belief, on the
ground that he did not think about it.S But suppose he had thought about
it, how would the case have been different? When you say he thought about
it all you mean is that he said to himself 'Its solid', and his saying this to
himself is simply a movement of the muscles of the throat not radically different
from the movements of the muscles in his arms?" One reply I could make
to this would be to deny that a thought could possibly be merely a movement
of the throat, or even of the brain, but I do not wish to assume even this,
and prefer to try to find a reply which might be convincing even to such
an objector. Let us, therefore, admit for the purpose of argument that thought
is only a movement of the throat; still, I would contend that although there
may possibly be cases which fall between our two alternatives the distinction
is valid and important.
If the man says to himself "The lump is solid," all that takes place may
be movements of his throat, but these words or movements have a fairly
definite meaning or propositional reference in virtue of which thay are true
or false; whereas the movements in his arms have no such meaning; they
have in a vague way reference to the environment, he would not make
them if he knew how light the lump really was or if he thought it was
too heavy to lift or fastened to the floor, but they have not the same precise
or relatively precise meaning as words have. The distinction is not necessarily
that between thoughts and bodily movements, but between actions, whether
of mind or body, which have definite or fairly definite propositional references
or meanings and those which have no such references.
After this preliminary discussion, let us set aside dispositional beliefs and
102 APPENDIX I: OLDER DRAFT VERSIONS

knowledge and confine ourselves to acts of thought6 with propositional

references. And of these again we can set aside such as have not an affirmative
character, such as wonderings, doubts or denials. Those that remain are
commonly called judgments.
The use of the word judgment in this sense has been severely criticized
by the late Professor J. Cook Wilson,7 and before settling to use it it would
be well to consider his objections. Some of these are based on considerations
of language; in ordinary speech, judgment implies decision after doubt, and
is therefore inapplicable to perception in which there is not doubt, and to
forming opinions without complete conviction when the only thing we decide
definitely is that the evidence is insufficient. Nevertheless if in logic we need
a word to include all these cases, there seems little reason why we should
not follow many distinguished writers in taking the word 'Judgment" and
arbitrarily extending its meaning.
But he also makes the more serious objection that the use of any term
in the proposed sense has been and is a source of nothing but confusion,
since it fails to distinguish between different kinds of mental acts which have
nothing in common except the verbal forms by which they are expressed.
It is proposed, in fact, to call a "judgment" any mental act which is expressed
by an indicative sentence such as "the earth is round", and if two such acts
were expressed by the same sentence they would be regarded as instances
of the same judgment. But, as Cook Wilson argues, the sentence reveals
nothing about the character of the mental act but only the object or objects
with which it is concerned. "The earth is round" may express anything from
a mere guess to a fully reasoned conclusion, and the differences between
it and other sentences correspond to no distinctions between different kinds
of thinking, but only to differences in what is thought about.

NOTES

1. Cf. Bertrand Russell, An Outline of Philosophy (London, 1927,) p. 272.

2. J. Cook Wilson, Statement and Inference, Vol. I (Oxford, 1926), pp. 109-113.
3. Or perhaps the belief is these adjustments and not merely a cause of them.
4. CD. Broad, Mind and Its Place in Nature (London, 1925), p. 153.
5. <Alongside this passage, Ramsey wrote in the margin: Strong!>
6. Or it may be movements in the throat or brain, what we have to say will hold equally
well or even better on this view, which we shall not bother to mention explicitly again
for some time.
7. J. Cook Wilson, Statement and Inference, Vol. I (Oxford, 1926), pp. 86-97.
 (001-14)

CHAPTER V

[THE PROLEM OF JUDGMENT]

Judgment and Time

In the last chapter we indicated, as well as we could, the range which we

took our term 'judgment' to cover, and defended it against those who make
an essential difference between knowledge and opinion. We have now to
go back to the "Question II" which we left over from Chapter I and try
to answer it in the case ofjudgments. This question was, it will be remembered,
"What is propositional reference?", that is to say, what do we mean when
we speak of a judgment that A is B, or say that so-and-so judges that A
is B.
Let us suppose for definiteness that Mr Smith judges that the earth is
flat; the sentence we use to express such a fact is of the simple form subject,
verb, object, and this suggests that the fact itself consists in the holding a
relation called judging between a mind (Mr Smith) and what we may call
a proposition, that the earth is flat; and that this relation of judging is one
about which nothing further can be said: it cannot be analysed or explained
to anyone who is not acquainted with it, anymore than we can analyse the
relation of before. On such a view the complete analysis of the fact would
be that the relation of judgment unites the thinker and the proposition judged.
If this were all there was to say, our problem would be easy indeed;
but unfortunately there are objections to any such simple account of the
matter, and before we have finished we shall have to consider much more
complicated hypothesis. For one thing, in giving our simple analysis we forgot
an essential element in the fact, the time at which the judgment took place.
[In language the time is often left to be understood; if we use the present
tense and say <that> Mr Smith judges that the earth is flat, we mean, of
course, that he is doing it either now or in the very recent past; if, however,
we use the past tense we generally refer to any definite time in the past
but merely mean that he did it some time or other. Such general fact as
that Mr. Smith judged it sometime or other is obviously dependent on there
being a particular fact that he judged it at a definite time T. ]
A judgment has a date; in other words it is an event or occurrence and
104 APPENDIX I: OLDER DRAFT VERSIONS

this is part of its nature which requires careful discussion, for the harmless
looking word 'event' conceals an ambiguity which has been the source of
considerable confusion.
[The problems to which this chapter is devoted concern not only judgment
but many other kinds of events both mental and physical. The treatment
of questions of such generality may seem out of place in a discussion of
the particular nature of judgment, but it seems indispensable [if certain
confusions are to be avoided] and <if we are to get> rid of certain general
misconceptions which frequently prevent the recognition of the truth.]
In order to get a true view of judgment it seems necessary to get rid
of certain general misconceptions, which concern not merely judgment but
many, if not all, other kinds of events both mental and physical. Such a
discussion takes us rather far afield from our proper subject, but cannot be
avoided since the points in question have not to my knowledge been set
out fully by any other author,l and it will help us later too when we come
to deal with the distinction between substance and accident2 and the connotation
or non-connotation of proper names. 3

NOTES
1. The best discussion I know is by C.D. Broad in Mind and Its place in Nature (London,
1925), pp. 588 ff., but I cannot altogether agree with his argument.
2. Vide infra, p ... ..
3. Vide infra, p... ..
 APPENDIX II

SUPPLEMENTAL MATERIAL
 (006-05-01 )

THE NATURE OF PROPOSITIONS

Paper read to the Moral Sciences Club in Cambridge on 18 Nov. 1921

Mr. Russell in The Analysis of Mind divides propositions into two main kinds,
word propositions and image propositions. I want it to be clear that I do
not use propositions in this sense; what Mr. Russell calls word propositions
I shall call sentences, reserving proposition for those things, if any, which
sentences stand for. This paper is divided into four parts; first I shall examine
the view that the phrases which express propositions are incomplete symbols,
which is put forward in those words in the Introduction to Principia Mathematica,
and is clearly also held by Dr. McTaggart in The Nature of Existence, although
he expressed it differently. I shall give reasons for rejecting this view and
in the second part of the paper put forward the alternative in which I believe.
Then I shall discuss the relations between propositions and facts, and lastly
truth.
We may notice first that in Principia Mathematica it is said, not that sentences,
but that phrases expressing propositions are incomplete symbols. It is not
asserted that when I say, for example, "Socrates is mortal," that sentence
in that case is an incomplete symbol; such a view would be absurd. What
we are concerned with are propositions about propositions, or sentences
containing propositional phrases such as "Aristotle believed that plato was
mortal," where "that plato was mortal" is a propositional phrase. The Principia
view holds that in this sentence the propositional phrase "that plato was
mortal" stands for nothing. The proposition, Aristotle believed that plato
was mortal, is apparently of the form aRb, asserting that the relation of
belief holds between Aristotle and the proposition "plato was mortal";
Whitehead and Russell assert that it is not really of this form but of some
other, perhaps the triply relational; that is, it asserts a three term belief relation
between Aristotle, Plato and mortality. These three with the belief relation
are its only constituents and none of them are expressed by "plato was mortal."
So we see that what is meant by the assertion that propositional phrases
108 APPENDIX II: SUPPLEMENTAL MATERIAL

are imcomplete symbols, is that in any sentence containing such a phrase,

this phrase stands for nothing, because the proposition expressed by the sentence
is not of its apparent form and has no constituent for which the propositional
phrase stands.
To examine this view we must consider first its scope; that is, what sorts
of sentences there are which contain propositional phrases. They seem to
be mainly, if not solely, of two kinds; they either express a relation between
a proposition and a thinker, like "I believe p" and "he desires p," or they
express a relation between two propositions, like "p implies q" and "p or
q". It is remarkable that this second kind is almost entirely neglected by
adherents of the incomplete symbol view, who, concerned merely with the
analysis of judgment, concentrate on the proposition "I believe p," which
it is relatively easy for them to analyse, and neglect the much more difficult
kind, typified by "p implies q."
For "p implies q" apparently asserts the relation of implication between
two propositions; but those who hold that the symbols p, q are incomplete
cannot accept this analysis, and it is therefore incumbent on them to provide
an alternative. This can, so far as I can see, only be done in one way; by
saying that "p implies q" is of the multiple relation form, the relata being
the several constituents of p and q. Thus, to take some examples, in "If
he comes, I go" the relation would be four termed, the relata being "he,
coming, I, going"; in "Had Lloyd George been assasinated, Germany would
have beaten England" the relation would be five termed, the relate being
"Lloyd George, assasinated, Germany, beat, England." So, on this analysis,
we should have not one dual relation of implication but many multiple ones,
whose multiplicity would depend on the complexity of the premiss and
conclusion of the implication; and since there is no limit to this complexity,
they would be infinite in number.
The difficulty in this view is obvious; one wonders what common property
this infinite set of relations can have, by which we group them as implicative
relations, so that we know logical laws which they obey and other relations
do not. (It may well be supposed that such laws would be nonsense by the
theory of types, since they profess to embrace all instances of implication
and therefore all these implicative relations, which since some are two termed,
some three, some four, and some are not all of one type, and therefore
form an illegitimate totality. But we do not need to appeal to the theory of
types.) It is possible quite simply to prove that this view is false, by considering
any logical truth expressed in terms of implication, for example p ::J q. q::J r: ::Jp ::J r.
This is asserted of any three propositions p, q and r: on the theory of
types, they must indeed be of a definite type, but this does not matter to
the argument since in one type are contained propositions of all forms subject,
predicate, dual relational, etc. Now what on the view we are considering,
 THE NATURE OF PROPOSITIONS 109

is the analysis of this proposition? Take first the simplest case in which p,
q, r are all subject-predicate propositions, e.g. "If he comes, I go and if
I go, he dies; imply Ifhe comes, he dies." The first of the constituent propositions
"If he comes, I go" expresses on this view the assertion of a relation which
we may call 14 between the four terms "him coming, I, going" and similarly
for the other constituent propositions; so the whole is to be interpreted as
a relation which we may call 115 between the fifteen terms He, coming,
14, I, going, I, going, 14, he, dying, he, coming, 14, he, dying.
The point is that this analysis, which in the simplest case is so complicated,
will vary with the forms of p, q, r; so that the analysis of the general proposition
"for all p, q, r, p:J q p r: :J p:J r" will be of the form for all a, b, c, P, Q,
R, 115 holds of a, P, 14, b, Q, b, Q, 14, c, R, a, P, 14, c, R, and a similar
thing for the next form in which p, say, is relational, and similar things
for all the infinity of other forms possible for p, q, r; so that this general
proposition would be seen on analysis to be infinitely complex. Now it is
generally agreed, and is, I think, evident, that no proposition entertainable
by us can be infinitely complex. Such infinite complexity is, in fact, what
is generally meant by a vicious infinite regress, which I have thus shown
to occur in the only analysis of this proposition possible on the incomplete
symbol vi~w. It may be remarked that this whole argument from the analysis
of implication, can be applied still more forcibly if based on the analysis
of probability relations; if, instead of considering "p implies q" we had
considerer "q has to p the probability relation of degree d," i.e., "the probability
of q on hypothesis p is d," we should have been still more astonished at
the bewildering multitude of probability relations, not dual, but of indefinite
multiplicity, infinite in number for each degree of probability, and possessing
no obvious properties whereby we could collect together those for each degree
of probability.
The reasoning, by disproving one of its necessary consequences, has I consider
shown that the incomplete symbol view of propositions must be rejected,
and we have now to consider possible alternatives to it.

II

All accounts of propositions and judgment which I have read are variations
of either the propositional theory of judgment or the incomplete symbol
theory of the propositions. The propositional theory of judgment asserts that
belief is a dual relation between something mental, perhaps a self, perhaps
a mental event, and a proposition. This view avoids the difficulties due to
supposing propositional phrases incomplete symbols; but its opponents doubt
whether there is any reason to believe that there are such things as propositions
110 APPENDIX II: SUPPLEMENTAL MA TERrAL

and no reason can be given except that no one has seen any alternative
except the incomplete symbol theory which I think I have disproved. But
it seems to me easy to construct a mixed theory, which satisfies the two
apparently incompatible requisites of a really satisfactory theory, which are
that it should hold: (1) that propositional phrases are not incomplete symbols,
(2) that there are no such things as propositions in the usual sense. That
is, (1) propositional phrases stand for something, but (2) they do not stand
for propositions in the usual sense since there are no such things. Therefore
they stand for things other than propositions. It may sound very paradoxical
to say that propositional phrases such as P, q, "p implies q" stand for things
other than propositions, but when you hear what I suppose it is they stand
for, you will probably say that it would have been misusing words to call
them propositions.
Consider, for a moment, a somewhat similar case, the discovery that numbers
are classes of classes; which always seems absurd until you are accustomed
to it. People naturally say that it is a misuse of language to call classes of
classes numbers; but what Frege discovered was that classes of similar classes
had all the arithmetical properties of numbers, so that it was reasonable to
suppose that in such statements as 2+2=4 the symbols 2, 4 stood for such
classes of classes. It was then possible to put his discovery in two ways;
he could either say "I have discovered what numbers are, they are classes
of classes"; or in order to avoid being told he was misusing language he
could have said, "I have discovered the meaning of 2+2=4; 2, 4, etcetera, do
not stand for numbers; they stand for classes of classes; I see no reason to
suppose that these strange metaphysical things called numbers exist." This
second way is the way I prefer to put my theory; it seems to me that in
such statements as "p implies q" p, q stand for things as certainly existent
as classes of classes seemed. These can be called propositions when you are
sufficiently accustomed to the theory not to be misled by the old associations
of the word, but till then had better be given some other name, since they
are not at all the kind of thing logicians suppose propositions to be; it would
be ridiculous to call them unities or constructs as propositions are called
by Mr. Russell in the Principles of Mathematics and Mr. Johnson respectively.
If we consider those mental events we call thoughts, we see that there
are two kinds of characters which they can have which are of great importance.
A thought may be a belief, a doubt, or an assumption, etc. These I call
pistic characters. I think that they consist in the presence or absence of feelings,
such as Mr. Russell calls belief feelings, in connection with the thought; but
their analysis is not very relevant to the matter in hand and I mention them
to distinguish them clearly from the other fundamental kind of characters
of thoughts, which I call referential characters or references. A referential
character is a character like that which we assert of a belief in saying that
 THE NATURE OF PROPOSITIONS 111

it is a belief that Socrates was mortal or of a doubt when we say that it

is a doubt, whether God exists. Generally in asserting of a belief that it
is a belief that p, where p is a sentence, we are asserting of it some character,
and such characters I call references. It is therefore indubitable that there
are references.
Those who believe in propositions support a particular way of analysing
references, holding that to have a reference is to have a certain relation
to a proposition; that is to say, that to assert that a belief is a belief that
p, is to assert that the belief has this relation to the proposition p. In particular,
Mr. Johnson supposes that the relation in question is of the kind expressed
by saying that the proposition is a factor in the belief. This way of analysing
references is undoubtably favored by the English language, for we do talk
of people believing or doubting propositions. But this in itself is a weak
argument, clearly inadequate as a basis for supposing there to be these
propositions, so unlike anything else in the world; and careful consideration
of the implications of this linguistic custom shows that no good argument
can be based upon it. It might be thought for example that propositions
were indispensable in as much as there are sets of thoughts which clearly
have something in common, namely that they consist in the adoption of various
attitudes by various thinkers to the same proposition. But this can be expressed
without begging the question by saying that there are sets of thoughts which
though thoughts of different thinkers with different pistic characters have
the same reference; an obvious fact which necessitates no particular analysis
or references. Again it might be urged that whether a belief is true or false
depends only on the proposition believed and that therefore there are
propositions. But clearly all that is certain is that the truth or falsehood
of a belief depends only on its reference, which again does not necessitate
the analysis of a reference into a relation to a proposition. I think there
only remain two arguments that could be used in favour of the propositional
view; one is negative; that there is no better way of analysing references;
the other, is that since in p ~ q the symbols p, q cannot be incomplete there
must be propositions for p, q to stand for. Take the latter argument first;
undoubtably p and q stand for something; the question is whether the things
they stand for are such that references are to be analysed into a relation
to them. I think that this is not the case but that the things p,q stand for
are the references "being a belief, (or a doubt whether, etc.) that p," "being
a belief (or a doubt whether, etc.) that q" themselves. For short I shall in
the future call the reference "being a belief that p." This view arises in
the following way.
It is I believe frequently held that such statements as p ~ q are to be analysed
in terms of beliefs. What this means is not clear, but it sounds as if what
the analysis meant was "If anyone believed p, he would be justified in believing
112 APPENDIX II: SUPPLEMENTAL MA TERrAL

q" an effort which is apparently ridiculous since it leads at once to a vicious

infinite regress; for this second statement is again of the implication form
and must be analysed as meaning "If anyone believed that anyone believed
p, he would be justified in believing that that man was justified in believing
q" and so on. But the analysis might be defended thus. It can be written
"For all x: x believes p implies x is justified in believing q" which is the
way Whitehead and Russell analyse "All who believe p are justified in believing
q." Now it has been held that Whitehead and Russell are mistaken in their
analysis of such propositions, that "All S are p" really asserts a relation
between the properties Sand P; and even if Russell and Whitehead are right
"all S are P" though not identical with is equivalent to some proposition
asserting a relation between Sand P. If then we take not "For all x: x
believes p implies x is justified in believing q" but the equivalent proposition,
in which a relation, which after Russell we may call formal implication,
is asserted to hold between the concepts, "believing p" and "justified in
believing p," we have an analysis of "p ~ q" namely "believing p formally
implies justified in believing q" which leads to no such infinite regress. We
can however easily see that this analysis is false; i.e. that "p ~ q" and" 'believing
p' formally implies 'justified in believing q'" are not equivalent verbal forms:
For "p ~ q" is true if and only if either q is true or p is false; but since
a property which nothing has formally implies any property - i.e., since
if there are no S "all S are P" is true whatever P may be; " 'believing
p' formally implies Justified in believing q' "will be true even if q is false
and p is true and therefore "p ~ q" false, provided no one has ever or will
ever believe p. For then "believing p" is a property which nothing has.
If however we take "p ~ q" as asserting a relation other than formal
implication between the references "being a belief-that in p," "being a belief-
that in q" we have an unobjectionable analysis. For even on the propositional
view there is some relation holding between these references, when either
q is true or p is false, so that if this relation be substituted for implication
and references for propositions and so on, we have a calculus of reference
completely parallel to the calculus of propositions. The case is exactly like
the substitution of classes of classes for numbers. We have these indubitably
real references which have all the formal properties commonly ascribed to
propositions; we shall see shortly that references can be easily analysed without
assuming propositions; we have therefore no reason except unfamilarity which
quickly wears off against supposing that when we say "p implies q" we
assert a relation between the references "being a belief that p" and "being
a belief that q" instead of between those mysterious entities propositions or
constructs.
How then are we to analyse references? This question is of course ambiguous;
for what is generally called the analysis of propositions, e.g. Russell's analysis
 THE NATURE OF PROPOSITIONS 113

of "The Author of Waverley is Scotch," becomes on my view a question

of analysing references. But this I postpone; what I am asking now is what
are references? They are properties of thoughts but what sort of properties?
Assuming that the propositional view is wrong, i.e. that to have a reference
is not to be related to a proposition, there seem to remain two possible
views. One is that of Russell in his Analysis of Mind. It is that to say that a
belief is a belief that p is to say that the belief or "content" as he calls it is (1)
if "p is true" points towards the fact p or (2) if "p is false" points away
from the fact not p, where 'points towards,' and 'points away from' are
complicated relations which he only pretends to have analysed in simple cases.
This analysis seems to me unnecessarily cumbrous, and also it is based
on a view of facts which I do not share, as will appear later. The correct
analysis is that references are multiple relational properties. There is no doubt
that there is some multiple relation which holds between a belief and the
things which would ordinarily be called the constituents of the proposition
p when and only when the belief is a belief that p, but the holding of this
relation may not be identical with but only equivalent to what is asserted
by saying that the belief is a belief of p. For an example take a belief that
this is green. On the propositional view this would assert a relation between
the belief and the proposition "This is green," which would be equivalent
to a multiple relation between the belief, this, and green, derived from the
relation between the belief and the proposition and the relations between
the proposition and its constituents, this and green. But much the simplest
view which supposes no unobservable entities and/or propositions is that the
holding of this multiple relation is what is asserted in "the belief is a belief
that this is green." To this view I can see no objections and I conclude
that to have a reference p is to be multiply related in a certain manner
to the individuals and universals which would usually be called the constituents
of the proposition p.
On this view to analyse a proposition or, more accurately a reference
is to substitute for the verbal form given an equivalent which exhibits clearly
the things in a multiple relation to which the reference consists. Thus to
analyse "All S are p" into "For all x: x is S implies x is p" or "the character
if S, then P characterises everything" is to say that a belief that all S are
P is a belief a certain multiple relation to the character "if S, then p" and
to the property of characterising everything.
As I explained earlier, I might have stated my view as that propositions
are references, instead of as "there are no propositions," but in "p implies
q" p, q stand for references." I have put it as I actually have stated it because
I thought that course would make it more intelligible. Besides, had I said
that propositions are references, I might have got tied up over such a statement
as "I believe p." This I analyse as "my belief has reference p" or for short
114 APPENDIX II: SUPPLEMENTAL MA TERrAL

"1 have a p-belief," thus making p what is, 1 believe, called a cognate accusative.
So just as Mr. Johnson said that "I experience red" means "I have a red
experience," I say that "I believe p" means "I have a belief with the reference
or referential character p." So that once the theory is understood, it is better
stated as that propositions are characters of beliefs, just as Frege's theory
of numbers is best stated as that numbers are classes of classes.
The conclusions reached may be summarised thus; when I believe p, my
belief is multiply related to the things ordinarily called the constituents of
p; if then 1 write 'p', the symbol stands not for a proposition which is before
my mind when I write, but for the property of my belief, expressed by
saying that it is a belief that p. Similarly "p implies q", "p or q" assert
relations between these referential properties of beliefs.

III

Hence forward I shall use proposition to mean what I have hitherto meant
by reference.
We have now to consider truth and facts. It is generally agreed that there
are facts; dispute is mainly about whether there is a different fact associated
with each true proposition; we can in English put 'the fact' before any
propositional phrase so that "the fact that 2+3 >4" and "the fact that 2+3=5"
appear to be different, but it is held by Mr. Johnson that they are identical,
the same fact being characterised determinately by 2+3=5 and less deter-
minately by 2+3>4. It may be remarked that on my view of propositions
namely that they are references, it is impossible to hold that they literally
characterise facts. For references are characters of beliefs and therefore the
relation between proposition and belief is characterisation; it is impossible that
this should be also the relation between proposition and fact; for if that
were the case we should be unable to distinguish the evidently different
statements "my belief is a belief that p" and "p is about my belief' since
both would be analysed as "p characterises my belief. "
Let us consider what arguments there are which can throw light on the
question of facts? They are of two kinds. Some are based on an analysis
or the supposed need for an analysis of truth; these 1 postpone; others are
based on supposed implications of our use oflanguage. We often make remarks
apparently about facts "it is a fact that King Charles was executed," "Newton's
Theory of Gravitation is contrary to the facts," "this view is forced upon
us by the facts." And "The severity of the Puritans caused the licentiousness
of the Restoration" apparently asserts a causal relation between two facts:
the fact that the Puritans were severe and the fact that the restoration was
licentious.
 THE NATURE OF PROPOSITIONS 115

But it seems to me disputable whether the correct analysis shows that

such propositions have facts as their constituents. In fact I think they have
not, except in those cases in which the facts could naturally be called events.
For example when I say that it is a fact that King Charles was executed,
I never mean more than that he was executed; when I say the facts demand
such and such a view, I only mean that the view is implied by certain propositions
which I know. Apart from considerations about truth, there seems to be
only one argument in favour of facts other than events; namely that there
is clearly some relation which holds between such pairs of propositions as
"2+3>4","2+3=5" and "Love is at least as good as Faith," "Love is better
than Faith." This relation it may be urged is that such a pair of propositions
characterise the same fact more or less determinately; but a more obvious
analysis of the relation is that such propositions have the same subject or
subjects of which they assert more or less determinate characters of relations
under what Mr. Johnson calls the same determinable. This is a completely
adequate analysis which does not presume the existence of facts.
Turning now to occurrent facts or events, the case is rather different;
for there are propositions which apparently cannot be analysed except so
as to have events as constituents, e.g., "it was what A said that made B
lose his temper" which clearly asserts a causal relation between two events.
Besides I agree with Dr. Whitehead in thinking that there are events from
which space and time are constructed by extensive abstraction. We have
now to consider the logical nature of these events; it is frequently supposed
that they are complexes, that is neither individuals nor characters nor relations
but things which in some sense consist in the possession of a property by
a thing or the holding of a relation between some things. This view is supported
by the fact that we can only identify events by means of propositions and
in consequence, talk as if the view were correct, speaking of "the event
which is the existence of the great pyramid during a certain day" or "the
passage of a bicycle down the street." There is an obvious objection to this
view; the execution of Charles I and the death of Charles I are certainly
the same event but it is difficult to see that the same event can consist in
the possession by Charles I of the property of dying and can also consist
in the possession by him of the property of being executed. [This objection,
together with the evident undesirability of supposing that events are things
of a new kind, neither individuals nor characters, .... J
And there is a stronger objection; those who suppose that events are
complexes would agree probably with Dr. Moore in the symposium on Mental
Acts in the latest volume of the Aristotelian Society that not all properties
and relations are such that the possession of the property by a thing or the
holding of the relation between two things constitute events, but that a necessary
condition for this that the property or relation should be such as is possessed
116 APPENDIX II: SUPPLEMENTAL MATERIAL

or holds at times. For example blushing is a suitable property since a man

can blush at one time and not at another, and so we should say that his
blushing was an event or occurrence. In propositions which have such properties
or relations as constituents, there is necessarily a time-reference, although
this may be concealed by the verbal expression. The time-reference may
be definite or indefinite; when I say "The plate was on the table" I probably
refer to some definite occasion; when I say "charles I was executed" I may
well mean that at some time or other he was executed. Making explicit
the time reference we have therefore the proposition "A had the property
P at time t (not an instant but an interval)" where A may be a person
and P blushing.
Not the only things of which times are ultimately predictable are events;
objects are only at times in virtue of their relations to events; so the thing
which was at time t if the proposition is true must be the event. If events
are complexes, we have two alternatives; the event is either (1) the possession
of P by A, (2) the possession of P by A at time t. Alternative (1) would
supply us with the analysis of "A had P at time t," namely that "the event
which is A having P was at time t, i.e., had certain temporal relations to
given events, perhaps the present or perhaps the birth of Christ." But
nevertheless it must be rejected, for a man may blush everyday; and these
events occurring at intervals cannot be identical; yet if each was identical
with the possession by the man of the property of blushing they would be
identical with one another.
Weare left with the second alternative, that the event is the possession
of P by A at time t. But this too seems liable to objections, if we reject
as we clearly must any Absolute Theory of Time. For it leaves us at a loss
for any plausible analysis of "A has P at time t"; for "at time t" can only
mean at time interval d from some fixed event E. We can then analyse
"A has P at time t" plausibly enough as asserting a relation between A,
P, d and E, or a relation of magnitude d between A, P, and E; both of
these analyses presuppose some analysis of E.
Since E, being an event, must be a complex; [how can it be analysed]
if we suppose that it also consists in the possession of P by A at time to.
We cannot analyse E like any other event, into the holding of a relation
between A', P', to and E, since E reappears in the analysis of itself. It follows
that E must be fundamentally different from other events, which seems to
me an implausible assumption. The preceding objections, together with the
fact that if events be not complexes, there is no evidence for the exist-
ence of complexes, except such as may follow from considerations about
truth (and I hope to show later that none will so follow), allow us to get
rid of the category complex altogether, seem to me to make the theory
that events are complexes decidedly inferior to the alternative theory that
 THE NATURE OF PROPOSITIONS 117

events are individuals. In order to explain this alternative I must first explain
what I mean by subjects and quasi-subjects.
When a proposition is of the subject predicate form, i.e. when to assert
it is to assert of one thing that it has a certain property, then I call the
thing of which the property is asserted the subject of the proposition. If
the proposition is of a relational form, i.e. if to assert it is to assert that
a certain relation holds between certain things, then those things are the
subjects of the proposition. It is easily seen that if an individual is a constituent
of a proposition it must be a subject. The definition of quasi-subject is more
difficult. The only propositions which have quasi-subjects are those which
assert existence, i.e. those which assert that there is a thing or things having
a certain property. Thus including such [singular] propositions as the Author
of Waverley was Scotch which asserts that there is a thing which has a
property of being the only writer of Waverly and of being Scotch. Thus
"Lions Exist," "Some animals are lions," "The most interesting animal in
the zoo is a lion," all assert existence. We then define "the class of quasi-
subjects of the proposition (Ex)P(x) is the class defined by the property p"
i.e. the quasi-subjects of "there are lions" are the lions themselves; the quasi-
subject of "The Author of Waverley is Scotch" is Scott. And it is an immediate
consequence of the definition that a false proposition has no quasi-subjects.
The view about events which I wish to advocate is that they are individuals,
and that the relation between an event and a time proposition which is,
as one might say, 'about' the event, is simply that the event is the only
subject or quasi-subject of the proposition. This is often obscured in language.
For example, neither "The table fell over just now" nor "Old St. Paul's
was then standing" seem to have events as subjects or quasi subjects. But
when they are analysed this is seen to be the case; the first should be analysed
as "There was an event x such that x shortly preceded this present and
was a situation of this table in which the table was in motion in a certain
manner." This has no subjects and if it is true only one quasi-subject, the
event x, (unless of course the table fell over twice). "Old St. Paul's was
then standing" should be analysed "There was an event x such that x had
the temporal relations indicated by 'then' and was a situation of Old St.
Paul's." Again "Yesterday I believed that" should be analysed "There was
an event with the temporal relations indicated by yesterday which was a
belief was mine and had the reference indicated by 'that'. It might be asked;
"why if events are individuals can we only identify them by means of
propositions?" The alternative would be to name them but we can only
name them when they are immediately before us by such a name as 'this';
we can name objects otherwise because they recur in experience and can
be recognised. Being unable to identify an event by a name we do so by
means of a proposition of which it is the only quasi-subject. There is no
118 APPENDIX II: SUPPLEMENTAL MATERIAL

real necessity for this; we could equally identify events by descriptive phrases;
instead of talking of the event which is the existence of the Great Pyramid
through a certain day, we could talk of the event extending through that
day in which the Great Pyramid was situated.
The only remaining objection to this theory that I can think of is that
though it explains how some time propositions are about events, and in
particular how a true proposition with only one event as subject or quasi-
subject is about one event; it does not explain what would be meant by
saying that a false proposition such as "the bicycle went 50 mph" when
it really only went 20 mph is about an event, in fact, the same event as
"the bicycle went 20 mph" is about. But this explanation can easily be given.
Let us recall the definition of a quasi-subject "The class of quasi-subjects
of the proposition (Ex) Px is the class defined by the property P." If (Ex) Px
is false as I said before there will be no quasi-subjects. Suppose now P is
a compound property, i.e. to have P is to have Pj and P2 and P3 etc. Then
in such a case the event or events which we should say (Ex) Px was about
would be the quasi subject or subjects of (Ex)Pjx or (Ex)Pzx or etc. one
of these at least being supposed true. In particular if we require a unique
event which (Ex) Px is about we choose one of (Ex) PjX, (Ex) Pzx, etc. which
is true and has a unique quasi-subject. Thus in our example the unique quasi-
event which "The bicycle went 50 mph" is about is the unique quasi-subject
of the "The bicycle went" namely the event in which the bicycle was located.
To sum up this account of facts and events; we have seen that there
are events and that they are individuals; we have seen so far no reason to
suppose that there are any such things as facts other than events or any
things such as complexes which are neither individuals nor properties nor
relations.

IV

Lastly we come to truth, which I shall deal with as briefly as possible. The
most certain thing about truth is that "p is true" and 'p', if not identical,
are equivalent. This enables us to rule out at once some theories of truth
such as that "to be true" means "to work" or "to cohere" since clearly
"p works" and "p coheres" are not equivalent to 'p'. There are I think
only three sensible theories of truth (1) that a true belief is defined to be
one which has a certain relation with a fact, (2) truth is indefinable and
has no connection with a relation between belief and fact, (3) truth is
indefinable; but as a matter of fact true beliefs do have a certain relation
to facts which false beliefs do not have. (3) I think we can dismiss; we
have seen no reason to suppose there are facts and if truth be indefinable,
 THE NATURE OF PROPOSITIONS 119

I think none can be drawn from the nature of truth; so if truth be indefinable
we have no reason to suppose these are facts and therefore no reason for
thinking true beliefs are related to facts in a way which false ones are not.
Of course in the special case of events we have seen that some true propositions
and therefore true beliefs are related to events in a way false ones are not;
since false beliefs have no quasi-subjects.
 (007-06-02 )

PAPER TO THE SOCIETY - AUTUMN 1922

I am going to discuss one of the most important philosophical questions,

which is of general interest and not, I think, difficult to understand. What,
however, is so difficult that I have abandoned the attempt, is to explain
the reasons which are to me decisive in favour of the view which I shall
put forward, namely that it is the only one compatible with the rest of
Mr. Wittgenstein's system.
"The process of induction," he says, "is the process of assuming the simplest
law that can be made to harmonize with our experience. This process, however,
has no logical foundation but only a psychological one. It is clear that there
are no grounds for believing that the simplest course of events will really
happen." This is the view which I wish to defend, but I shall begin by
considering the only plausible account of an alternative view, with which
I am acquainted; namely, that of Maynard Keynes in his Treatise on Probability.
He there introduces an hypothesis, which he calls the hypothesis of limited
variety, which is roughly that all properties of things arise from the various
combinations of absence and presence of a finite number of fundamental
or generator properties. He argues that the assumption of this hypothesis
would justify the inductive process used by scientists; that is, not that it
would justify our attributing certainty to their conclusions, but that the
appropriate degree of probability would approach certainty as more and more
observations were made which confirmed the conclusions.
It seems to me that his argument for the adequacy of this hypothesis contains
a mistake; but this is too complicated to explain now. I may however mention
that what is required is not the hypothesis that variety is limited, but that
it has some definite limit. We can justify induction by supposing that there
are only 1000 generator properties, or by supposing that there are only 1,000,000;
but it is not enough to suppose merely that they are finite in number. To
explain why I think this modification necessary, is not possible without going
into difficult details.
But whether I am right in this or not, I see no logical reason for believing
 PAPER TO THE SOCIETY 121

any such hypotheses; nor are they the sort of things of which we could
be supposed to have a priori knowledge, for they are complicated generalisations
about the world which evidently may not be true. To this it may be answered
that it is not necessary to suppose that we know the hypothesis for certain
a priori, but only that is has a finite a priori probability. (It may be explained
that the alternative to a finite probability is not an infinite, but an infinitesimal
one, like the probability that a cushion has one definite shade of colour out
of an infinite number of possible ones.) For inductive arguments would have
some force if the initial probability of the hypothesis were finite, and could
then be applied to the hypothesis itself, and so the probability of the hypothesis
would be increased which would again increase the force of inductive argument.
This would correspond to our feeling that induction derived its validity in
part at least from our experience of its success.
So we have to consider the a priori probability of an hypothesis of limited
variety; how are we to determine whether it is finite? Presumably, by direct
inspection; but owing to the abstract nature of the hypothesis this is difficult;
it seems to me easier to approach the question if we notice, that if the
hypothesis has a finite a priori probability, so has any generalisation such as
"all swans are white"; this indeed is the sole point in introducing the hypothesis.
If then we can see, as I think we can, that the a priori probability of "all
swans are white" is infinitesimal, so must be that of the hypothesis of limited
variety.
I do not see how to proceed any further without discussing the general
nature of probability; according to Keynes there holds between any two
propositions some objective logical relation, upon which depends the degree
of belief which it is relational to have in the one proposition, if the other
is what is known already. This clear objective theory is however blurred
by his saying in one or two passages that "probability is relative in a sense
to the principles of human reason." With the word human we pass from
a purely logical notion to one which is in part, at least, psychological, and
in consequence the theory becomes vague and muddled. Keynes tries to identify
the logical relations between propositions, with the psychological ones which
express the degree of belief which is rational for men to entertain, with
the result that the probability relations of which he speaks cannot be clearly
identified, whence the difficulty in deciding whether or not they can always
be compared with one another or measured by numbers.
Between propositions there are indeed logical, or as I should prefer to
say formal relations; some of these enable us to infer one proposition from
the other with certainty, others only with probability. For example if p,
q are two elementary propositions, i.e. such as assert atomic facts, we can
see what probability the proposition p V q gives to the proposition p in the
following way:
122 APPENDIX II: SUPPLEMENTAL MATERIAL

There are 4 conceivable cases:

p true and q true
p true and q false
p false and q true
p false and q false
Of these the last (p and q both false) is excluded by the hypothesis p V q;
of the remaining three cases p is true in 2; so p V q gives to p the probability
2/3. Such probabilities are inevitably numerical and arise in a clearly statable
way from the logical forms of propositions; and they seem to me to be
the only logical probabilities, and they alone can provide logical justification
for an inference.
It is clear that they do not justify induction; for they in no way allow
inference from one lot of facts to other entirely distinct ones; there is no
formal relation of this sort between the proposition that certain examined
swans are white and the proposition that some other swan is white. This
may also be seen by taking up the problem of induction in the form in
which we left it. We saw that if induction is to be justified there must
be a finite initial probability that all swans are white, and this there will
not be owing to the infinite number of things, which, for all we know a
priori, may be black swans.
To this theory of probability it will probably be objected, first, that it
is difficult to give it practical application, because we do not know the logical
forms of the complicated propositions of everyday life; to this I answer that
it may nevertheless be the true theory, and that this is supported by the
contradictions to which applications to daily life of theories of probability
almost invariably lead.
A second and more serious objection is that it does not justify such inductions
which we all regard as reasonably, and which must be in some sense reasonable
or there is nothing to distinguish the wise man from the fool. But I would
suggest that this sense in which they are reasonable need not be that they
are justified by logical relations.
There seems to me to be some analogy between this question and that
of objective or intrinsic good. In the latter we consider the justification of
our actions, and are at once presented with the simple solution that this
lies in their tendency to promote intrinsic value, a mysterious entity not
easy to identify; if now we turn to the justification of our thoughts we
have the equally simple solution that this lies in their following certain logical
probability relations, equally mysterious and difficult to identify, as the only
ones discoverable are evidently unsuitable. I think that both these simple
solutions are wrong, and the true answers are in terms not of ethics or logic,
but of psychology; but this is the end of the analogy; actions are justified,
 PAPER TO THE SOCIETY 123

if they are such, that to them or to their consequences we or people in

general have certain psychological reactions such as being pleased. But we
cannot give this account of the justification of inferences. Two accounts seem
to me possible; one, suggested by Hume, is that good inferences are those
proceeding from those principles of the imagination which are permanent,
irresistible and universal, as opposed to those which are changeable, weak
and irregular. The distinction between good reasoning and bad is then that
between health and disease.
The other possible account has only just occurred to me, and as I am
tired I cannot see clearly if it is sensible or absurd. Roughly it is that a
type of inference is reasonable or unreasonable according to the relative
frequencies with which it leads to truth and falsehood. Induction is reasonable
because it produces predictions which are generally verified, not because of
any logical relation between its premiss and conclusion. On this view we
should establish by induction that induction was reasonable, and induction
being reasonable this would be a reasonable argument.
 THE "LONG" AND "SHORT" OF IT
or
A FAILURE OF LOGIC*

All exact reasoning, especially mathematical reasoning, rests on the supposition

that a precise statement is either true or false, and cannot be both true and
false. Consequently mathematicians have in recent years been forced to give
a lot of attention to certain cases in which simple and apparently flawless
reasonings can be used to establish contradictory conclusions; for the situation
which arises is as uncomfortable to the mathematician as it would be if he
could prove that two sides of a triangle were together greater than the third
side, and also prove that they were together less than it.
A simple illustration of this is provided by the following example, from
which all mathematical elements have been removed. The word "short" is
a short word, but the word "long" is not a long word: this suggests a division
of adjective words according as they do or do not have the property which
they connote. Words like "short," which apply to themselves, let us call
autological; and words like "long," which do not apply to themselves, let us
call heterological. Now suppose we put the question, "Is the word 'heterological'
a heterological word?" Then we at once obtain contradictory answers. For
if it is heterological, that means that it does not apply to itself, i.e., that
it is not heterological; but if it is not heterological, then it does apply to
itself, i.e., it is heterological. How can these conclusions be reconciled with
the Law of Contradiction?
Let us take another example, one which actually forced the great Frege
to admit that his system contained some fundamental mistake. Classes we
can divide according to whether they are or are not members of themselves.
Thus the class of men is not a man, and so not a member of itself: but
the class of things other than men is also not a man, and is therefore a
member of itself. Now consider the class of all classes which are not members
of themselves; is this class a member of itself or not? If it is a member
of itself, then it is one of its members, i.e., one of the classes which are
not members of themselves, and so it is not a member of itself. But if it

* Paper first published in the American Philosophical Quarterly, vol. 24 [1987], pp. 357-359.
 "THE LONG AND SHORT OF IT" 125

is not a member of itself, then it is not one of the classes which are not
members of themselves, and so must be a member of itself. We have thus
proved both that it is and that it is not a member of itself.
About a dozen such contradictions have been found of various kinds, from
the absolutely non-mathematical one about "heterological" to an undeniably
mathematical one about infinite ordinal numbers. And it is probable that
only a little ingenuity is required to invent as many more. The result is
a serious crisis for mathematical and philosophical thought, which has not
however received the attention it deserves, owing to the scarcity of philosophers
who are good enough mathematicians, and of mathematicians who are good
enough philosophers to get a firm grasp of the situation. Nevertheless some
of the greatest mathematicians in the world have been led to investigate
anew the fundamental principles of their science in order to avoid not merely
the possibility but the actuality of contradiction with which they are threatened.
The situation is not altogether a new one in the history of thought, for
at least twice before the intellectual situation has been over-shadowed by
apparently insoluble contradictions. In ancient Greece, Zeno of Elea pro-
pounded his paradoxes about motion, which cannot be regarded as having
been finally cleared up until the latter half of the last century. The most
famous of these is the race between Achilles and the tortoise. Achilles, who
runs ten times as fast as the tortoise, gives the tortoise ten yards start. By
the time Achilles has run this ten yards the tortoise is a yard further on;
by the time Achilles has run this yard, the tortoise has gone 1 yard further.
By the time Achilles has gone this .1 yard, the tortoise will have gone another
.01 yard; and so on, the tortoise being always ahead and Achilles never catching
him up: The solution of this paradox requires an adequate theory of the
infinite divisibility of space which was only achieved in recent times.
Fortunately, however, men were prepared to trust their common sense before
they were in possession of a theoretical justification for their conclusion.
The second great thinker who invented contradictions of this sort was
the philosopher Kant, whose "antinomies" were an essential part of his system.
He produced proofs that, for instance space is infinite in extent, and again
that it is only finite: that it is infinitely divisible and again only finitely
divisible. From this possibility of proving contradictory results he deduced
that the sphere of human reason was limited, and that contradiction was
the natural consequence of arguing about things beyond the range of possible
experience. His position has, however, been overthrown by the general
recognition that the arguments by which he reached his contradictions are
fallacious, and that so far from it being possible to prove both that space
is infinite and that it is finite, it is not possible to prove by a priori reasoning
either of these results.
Now again we are faced with contradictions, but the present situation
126 APPENDIX II: SUPPLEMENTAL MATERIAL

is different in several ways from either of the preceding ones; then the
contradictions were invented by philosophers to support their systems, whereas
the modern ones were, to begin with, stumbled upon by mathematicians
who had nothing to gain for their theories from them and could only regard
them with dismay. The modern contradictions, especially the more mathe-
matical ones, therefore appear as obstacles inevitably encountered in the pursuit
of science, not as the artificial creations of philosophers. Another difference
is that the paradoxes of Zeno and the antinomies of Kant related principally
to space, time and motion, whereas the modern "paradoxes of the Theory
of Aggregates," as they are called, are matters of pure logic, and demand
for their solution not merely a new theory of space and time, but a reform
of logic; that is, some new idea as to what is a good argument and what
a fallacy. Hence although history teaches us that such paradoxes will be
solved in time, we must expect that in this case the task will be decidedly
more difficult than it was in the previous ones.
Although there is a considerable literature about the contradictions of the
theory of aggregates, none of it can be said to contribute materially towards
a solution except the work of Mr. Bertrand Russell, which although it is
not universally accepted, and does not profess to be completely satisfactory,
is by many regarded as giving an approximately correct account of the matter.
His solution is given by his "Theory of Types." The most important idea
in this is that a series of words which form a grammatical sentence may
nevertheless be nonsense. For instance "virtue is triangular" is a grammatical
sentence, but according to Mr. Russell nonsense. It is not merely that it
is false to say that virtue is not triangular, but that this statement is literally
nonsense, for it is only things and not qualities like virtue that can significantly
be said to be triangular. It is therefore necessary to distinguish different "types"
of things, and to declare that the result of applying an adjective to a thing
of the wrong type is not falsehood but nonsense. In particular, a class is
always of a different type from its members, so that if the members are
men it is nonsense to say that the class either is or is not a man; and it
is nonsense to say that a class is or is not a member of itself, so that the
arguments used in establishing the contradiction about the class of classes
which are not members of themselves are all literally nonsense and there
is not really a contradiction but simply a nonsensical arrangement of words
which looks like a contradiction.
In this way, by the principle that the rules of grammar do not guarantee
significance, we can escape about one half the contradictions of which the
one about classes is typical. But the other contradictions like the one about
"heterological' are not so easy, because "heterological" is not a nonsensical
word, but expresses a real property of an adjective word, and since it is
itself an adjective word it is the right type of thing to be autological or
 "THE LONG AND SHORT OF IT" 127

heterological, and the question "Is 'heterological' heterological?" is not

nonsense. In order to deal with a contradiction of this sort we have to make
further "typical" distinctions among the adjective which are applicable to a
given sort of thing. The need for this can be made apparent in the following way.
Suppose we have a set S of characteristics; for instance, let them be,
if possible, all characteristics significantly predictable of things of a certain
type. Then we could ask whether a given thing has all the characteristics
of the set S, and if it had, we could regard this as a further characteristic
of it, and we could introduce an adjective A to mean "having all the
characteristics of the set S." As the characteristics of the set S will be named
by a set of adjectives, the question is raised whether the adjective A can
be an adjective of this set, or whether, since it presupposes the totality of
the set, it cannot be regarded as a member of it without a vicious circle;
so that whatever set of characteristics and adjectives we take, there will
always be other characteristics and adjectives which presuppose the totality
of the set taken and so cannot be members of it. If we adopt this view
we shall reject the possibility of an all-embracing set of characteristics
predicable of a given thing, and construct a hierarchy of different orders
of characteristics, each presupposing the totality of those of lower orders.
In this way we can solve the remaining contradictions: for instance, "short"
and "long" are adjectives of the lowest order, but "heterological" is found
to be one of higher order because it means "meaning a characteristic not
predicable of itself', and the indefinite a in this means that a totality of
characteristics is involved in it. These we must suppose to be of some definite
order, say the lowest, so that "heterological" will mean "meaning a
characteristic of lowest order not predicable of itself," and "autological"
similarly "meaning a characteristic of lowest order predicable of itself," and
"heterological," which involves in its meaning the totality of lowest order
characteristics means therefore a characteristic of higher order and is therefore
neither autological nor heterological.
Such is Mr. Russell's solution, but it cannot be regarded as completely
satisfactory, because it has the unfortunate consequence of invalidating an
important type of mathematical argument, on which a good deal of modern
mathematics rests. This difficulty has again thrown the subject into the melting-
pot, and some eminent mathematicians have expressed the opinion that parts
of their subject (which has always been regarded as the only part of science
that never needs revision) are radically unsound. It seems, however, to the
present writer that Mr. Russell's theory can be amended in a rather subtle
way which will entirely overcome the difficulties. The key to the situation
is found to be an ambiguity in the notion of meaning to which logicians
are being obliged to pay more and more attention. But the matter is too
intricate for non-technical discussion.
 INDEX

Aristotle 11, 107 Joachim, H. H. 19, 20n, 27, 28n, 29n, 30,
31 , 32n, 36, 37, 38,41, 41n, 42n, 96,97n
Bacon, Francis 15n, 92 Johnson, W. E. 66n, 76n, 110, 111, 114, 115
Baldwin, Stanley 3 Joseph, H. W. B. xiii, 65n
Beethoven, Ludwig van 96
Birkenhead, Lord 9, 86 Kant, Immanuel 20n, 24n, 77n, 125, 126
Bosanquet, Bernard 9, 15n, 29n, 31, 32n, 38, Keynes, J. M. ix, 83n, 120, 121
40, 42n, 82, 93n, 97n
Bradley, F. H. 27,28, 29n, 31, 32n, 41n, 42n, Leibnitz, G. W . von 36
82,97n Lloyd George, David 108
Braithwaite, R. B. x , xi, xii, xiii, xiv, xxin,
41n, 76, 77n McTaggert, J. M. E. 74, 76, 77n, 93n, 107
Broad, C. D. 50, 51, 53n, 54n, 60, 65n, 70, Mellor, D. H. xxin
71, 73, 74, 75, 76, 77n, 97n, 100, 102n, Moore, G. E. xiv, 5n, 15n, 36, 41n, 65n, 76n,
l04n 77n, 82, 93, 115
Buonaparte, Napoleon 8
Peirce, C. S. 91, 94n
Caesar, Augustus 72 Plato 107
Caesar, Julius xiii, 26, 30, 31, 32, 42n, 72
Campbell, Norman xiii, xiv, xv, xxin, 35n Richard, l. A. 5n
Cohen, Morris R. 94n Ross, W. D. 15n
Cook, Wilson J. 15n, 46, 47, 48, 50, 53n, Russell, Bertrand xiii, xiv, xvi, 8, 15n, 24n,
54n, 55, 57, 62,63, 65n,99, 102n 25, 28n, 36, 41n, 54n, 70, 76, 77, 77n, 83,
Cone, M. 36 83n, 86, 90, 93n, 94n, 95, 97n, 101, 102n,
107,110,112,113,126,127
Dunne, J. W. 75,77n
Socrates 107, 110
Frege, Gottlob xiv, xv, xxin 114 Stumpf, Kar! 60

Green, T. H. 82 Tarski, Alfred xiv, xv

Hertz, Heinrich xiii, xv, xxin, 35n Whitehead, A. N. 69, 77n, 112, 115
Hobbes, Thomas 66n Wittgenstein, Ludwig xiii, xiv, 120
Hume, David 123
Husser!, Edmund 15n Zeno 125, 126

James, William 15n, 23, 24n, 91 , 92, 93, 94n

 Episteme
A SERIES IN THE FOUNDATIONAL,
METHODOLOGICAL, PHILOSOPHICAL, PSYCHOLOGICAL, SOCIOLOGICAL, AND
POLITICAL ASPECTS OF THE SCIENCES, PURE AND APPLIED

1. W.E. Hartnett (ed.): Foundations o/Coding Theory. 1974 ISBN 90-277-0536-4

2. J.M. Dunn and G. Epstein (eds.): Modern Uses of Multiple-Valued Logic. 1977
ISBN 90-277-0747-2
3. W.E. Hartnett (ed.): Systems: Approaches, Theories, Applications. 1977
ISBN 90-277-0822-3
4. W. Krajewski: Correspondence Principle and Growth of Science. 1977
ISBN 90-277-0770-7
5. J.L. Lopes and M. Paty (eds.): Quantum Mechanics, a Half Century Later. 1977
ISBN 90-277-0784-7
6. H. Margenau: Physics and Philosophy. Selected Essays. 1978 ISBN 90-277-0901-7
7. R. Torretti: Philosophy of Geometry from Riemann to Poincare. 1978
ISBN Hb 90-277-0920-31 Pb 90-277-1837-7
8. M. Ruse: Sociobiology: Sense or Nonsense? 2nd ed. 1985
ISBN Hb 90-277-1797-4 / Pb 90-277-1798-2
9. M. Bunge: Scientific Materialism. 1981 ISBN 90-277-1304-9
10. S. Restivo: The Social Relations of Physics, Mysticism, and Mathematics. Studies in
Social Structure, Interests, and Ideas. 1983
ISBN Hb 90-277-1536-X / Pb (1985) 90-277-2084-3
11. J. Agassi: Technology. Philosophical and Social Aspects. 1985
ISBN Hb 90-277-2044-4/ Pb 90-277-2045-2
12. R. Tuomela: Science, Action, and Reality. 1985 ISBN 90-277-2098-3
13. N. Rescher: Forbidden Knowledge and Other Essays on the Philosophy of Cognition.
1987 ISBN 90-277-2410-5
14. N.J. Moutafakis: The Logics of Preference. A Study of Prohairetic Logics in Twentieth
Century Philosophy. 1987 ISBN 90-277-2591-8
15. N. Laor and J. Agassi: Diagnosis: Philosophical and Medical Perspectives. 1990
ISBN 0-7923-0845-X
16. F.P. Ramsey: On Truth. Original Manuscript Materials (1927-1929) from the Ramsey
Collection at the University of Pittsburgh, edited by N. Rescher and U. Majer. 1991
ISBN 0-7923-0857-3
17. H.A. Shenkin: Medical Ethics. Evolution, Rights and the Physician. 1991
ISBN 0-7923-1031-4

KLUWER ACADEMIC PUBLISHERS - DORDRECHT / BOSTON / LONDON