second edition

Publication of this book was assisted

of by a Tgr~nt from the Publications Program the NatlO~al Endowment for the Humanities, an mdependent federal agency.

LOGIC, SEMANTICS, METAMATHEMATICS

PAPERS FROM 1923 TO 1938

BY

ALFRED TARSKI

TRANSLATED BY J. H. WOODGER

Second edition edited and introduced by JOHX CORCORAX

HAC K E T T P L' B LIS H I x G C 0 ~I P A x Y

i r

"; I' J _ r /

u,/ ...... ' ~ "'" ~.-)

First edition published in 1956 by Oxford University Press

Copyright © 1983 by Hackett Publishing Company, Inc.

All rights reserved

f Printed in the United States of America

1/lc.. Second edition, first printing 1983

l f 0 Cover design by Richard L. Listenberger

TO THE MEMORY OF HIS TEACHER

TADEUSZ KOTARBINSKI

THE AUTHOR

For further information, please address Hackett Publishing Company, Inc.

P.O. Box 44937 Indianapolis, Indiana 46204

Library of Congress Catalo,ing in Publication Data Tarski, Alfred.

Logic, semantics, metamathematics.

Includes bibliographical references and indexes.

1. Logic, Symbolic and mathematical- -Addrosso», essays, lectures. 2. Semantics (Philosophyj-> Address, essays, lectures. 1. Corcoran, John.

II. Title.

BC135.T35 1983 160 83-10850

ISBN 0-915144-75-1

ISBN 0-91S144-76-X (pbk.)

The paper in this book meets the guidelines for permanence and durability of the Committee on Production Guidelines for Book Longevity of the Council on Library Resources.

EDITOR'S PREFACE

TO THE REVISED EDITION

THE revised edition is intended to supersede the first edition in several respects. Two considerations which guided the design of the revisions were (1) the desire to increase the accuracy and clarity of the articles without altering their usefulness as historical documents and (2) the need to make Tarski's early thought more accessible to the current generation of linguists, logicians, mathematicians, philosophers, and historians. The Editor's Introduction and the Analytical Index are new with this edition. The Editor's Introduction shows how Tarski's development of much of the modern framework of logic and methodology can be traced in the articles in this volume. The Analytical Index is much more than a guide to locations of terms; it is as milch an index of ideas as of terminology, and it lists, besides terms used by Tarski, not only current synonyms for the terminology of the articles, but also current terms for ideas used by but not named by Tarski.

Included in this edition are all seventeen articles collected in the first edition. All of the front material of the original has been retained unchanged, Several items have been added to the bibliography. Over l[,lO pages have bf'l'1l altun-d in various wnvs, The fad thai Turski had been unable to fI'ad a !'ollsilil'rahlt, part of t ho original text lx-Ioro publication accounts for many of till' changes. Romp ehnngos in wording were needed for sake of clurity, uniformity, or precision. Typographical errors, in prose and ill mathematical formulas, have been corrected (although there were fewer of these than would have been expected given the complexity of the work). In several places, whole sentences have been rewritten in order to eliminate ambiguity or misleading connotations and in some places, where more extensive rewriting would have been called for, footnotes have been added indicating the nature of the difficulty.

The bibliographical note attached to the "Wahrheitsbcgriff'" (art iclo VIII) has been rewritten as has the first paragraph of this art ide's his-

viii

EDITOR'S PREFACE

torical note. Postscri h

.' . os scripts ave been added to articles IV and V V. .

new historical references have been inserted f t t .' anous

Th MOOM~

ere was cross-referenci he ori , '

nificantlv increased in th ntg amI °tI~g t e anginal articles which was sig-

. c trans a Ions and inc d '

present edition. rease even more in the

All the rewriting was done b T ki

has received Tarski's expli it y ars and the final form of every change

lCI approval Tarski d

spaced pages of changes which have b ,I prepare ?ver fifty single-

text. In addition, with the hel of M e~n Incorporated into th~ present H, G. Bohnert (Michiga 8t Pt U" 8c~nlan (Oregon State University),

n a e ruversity) and H R" (U' .

Pennsylvania), I prepared hundreds of ,'IZ. ruversity of

sponded in writing by ph ' suggestIOns to which Tarski reassisted Tarski.' one, or m conversation, S, Givant (Mills College)

Various methods to identif chan

found to be too awkwa d t Y ~ed passages were considered and

, . r, 00 expensive and t ti '

cially III light of the fact th t h " 00 ime consunung, espe-

had been introduced alre d a , c t~ng~s m the articles as originally written times difficult to determ~ y mh he rst edition, For example, it is some-

o me w et er hr ' ,

turing of a sentence t c f P asmg, termmology, and restrue-

o on arm to the needs f diff

eonceal subtle changes' '. 0 a I erent language could

, . m vlewpomt F h

edition can be regarded, ... or most sc olarly purposes this

Hasan authontatlve sub tit t f

owevor, writers iuvolv d I d I' Slue or the first edition,

e m e leate historic I '

consult the original pape th I a quesuons may need to

. ffici rs emse ves as even th fi di ,

msu cient for such p erst e ition may prove

urposcs.

TRANSLATOR'S PREFACE

.J. C,

THE setting free of Poland after the First World War was followed by intensive activity in her Universities. In the departments of philosophy and mathematics this took the form, in a number of places, of new and powerful investigations in the fields of mathematical logic, the foundations of mathematics, and the methodology of the sciences. Prominent in this movement was the Warsaw school led by Lukasiewicz, Kotarbillski, and Lesniewski. Under their skilled guidance a younger generation grew up and among these Alfred Tarski quickly distinguished himself.

Ever since I first enjoyed the hospitality of Professor Tarski in Warsaw in 1936, it has seemed to me that the importance and scope of the Polish school of logicians were insufficiently known and appreciated in the English-speaking world. Then came the Second World War, bringing ruin once more to Poland, killing men, destroying laboratories, and burning manuscripts and libraries. After this war it occurred to me that I should be performing a public service, as well as acknowledging in some small measure my debt to my Polish friends, if I prepared a collected edition of some of Professor Tarski's publications in English translation. When he visited England in 1950 to deliver the Sherman Lectures at University College, London, I mentioned my plan to him and received his approval.

This volume contains Tarski's major contributions to logic, semantics, and metamathematics published before the Second World War; their arrangement here corresponds to the chronological order in which they first appeared in print. With the exception of articles II and XI (which are too closely connected with, and too often referred to in, the other articles to be omitted), the volume does not include Tarski's studies in the foundations of special mathematical disciplines-set theory, group theory, etc. Neither does it contain his papers of a. pronouncedly mathematical character, dealing with special topics from the domain of set theory, measure theory, abstract

x

TRANSLATOR'S PREFACE

TRANSLATOR'S PREFACE

xi

algebra, elementary geometry, etc. Also excluded are short notes, abstracts, and preliminary reports which are closely related to some of the articles included in the volume and the contents of which are more fully presented in these articles. A longer paper from the domain of logic and methodology which has been omitted is ' Sur la methode deductive' in Travaux du IX" Congres International de Philosophie (Paris 1937); this paper is a purely expository one, and its ideas have been fully developed in Chapter VI of Tarski's book Introduction to logic (New York, 1941).

In a sense the present work is more than a volume of translations. Naturally an attempt has been made to remove the misprints and errors which occur in the originals. Moreover, the articles have been provided by the author with cross-references to other articles in the volume and with notes referring to later developments and recent literature. Occasionally some new remarks have been added for the purpose of clarifying certain passages in the original text. Articles II and VI contain more serious changes, 'I'arski having inserted in them several passages which, he hopes, will help to clarify and amplify their contents.

':"'hile the work of translation was in progress, the passages which seemed to me doubtful and difficult were noted down and sent to Professor Tarski in Berkeley, California. In this way it wa:'3 possible to adjust the text of the translations in many points, so as to meet the author's wishes. However in view of the time limit and the geographical distance between the residences of the author and the translator, it was impracticable to send the whole manuscript to Professor Tarski before it was set ~n print. Instead he received galley proofs, on which for ObVIOUS reaso~s he c.ould not suggest too extensive changes. It also. proved. ImpossIble to discuss the suggested changes in detail, and It was left to my decision which changes were actually to be carried through. Thus Professor Tarski is not responsible :or the final text or the technical aspect of the book.

Three articles in this volume are joint publications of Tarski and ~ther authors: Professor C. Kuratowski (article VII), Dr. A. Lllldenbaum (article XIII), and Professor J. Lukasiewicz

(article IV). We are greatly obliged to Professors K uratowski and Lukasiewicz for their permission to include the translations of the jointly published articles in the volume, J Jr. Linden baum fell a victim to the Gestapo during the war.

The papers included in the volume originally appeared ill tho following periodicals and collective works: Actes du Congnls International de Philosophie Scientifique (articles XV and XVI), Comptee Rendus de la Societe des Sciences et des Leitres de Varsovie (articles III and IV), Ergebnisse eines mathemaiischen. Kolloquiums (articles XIII and XIV), Erkenntnis (article X), Eundamenta 2lf·athematicae (articles I, VI, VII, XI, XII, and XVII), Ksieqa Pamiqtkowa Pierwszego Polskiego Zjazdu .l'r! atematycznego (article II), Monatshefte fur Mathematik und Physik (articles V and IX), Przeglq_d Filozoficzny (articles I, X, XV, and XVI), and Travaux de la Societe des Sciences et des Leures de Varsovie (article VIII). Acknowledgements should be made to the publishers and editors of these periodicals for their generosity.

I am obliged to Mr. S. W. P. Steen for kindly reading the proofs of article VIII and throughout the work I have received much help from my son, Michael Woodger. 'Ve are also indebted to several colleagues and students of Professor Tarski- Dr. C. C. Chang, Professor A. C. Davis, Professor J. Kalicki, Mr. R. Montague, Professor J. Myhill, Professor D. Rynin, and Mr. D. Scott-for their assistance in revising the original text of the articles and in reading galley proofs.

Finally it is a pleasure to acknowledge the courtesy and help which we have received from the staff of the Clarendon Press at all stages in the production of the book.

J.H.W.

AUTHOR'S ACKNOWLEDGEMENTS

IT is a rare privilege for an author to see a volume of his collected papers published during his lifetime, and especially so if the papers be translated into a single language from originals in a number of other languages. I cannot therefore but be deeply moved by the appearance of this volume and by the thought of the many and great sacrifices which its publication has laid upon my friend, Professor Joseph H. Woodger. For five long years he has devoted to this work an immense amount of effort and time, which otherwise could have been used for fruitful research in his chosen field, theoretical biology and its foundations.

The task of a translator is rarely a gratifying one. Circumstances have made it especially thankless in the present case; let me indicate some of them. The papers whose translations constitute the volume were originally published over a period of fifteen years and in several languages. They vary considerably in subject-matter, style, and notation. Under these conditions, the task of combining the papers in one book provided with a reasonable degree of terminological consistency and conceptual uniformity presents extreme difficulties. In a few cases (in particular, in the case of the monograph on the concept of truth, which occupies nearly one-third of the volume) the translation had to be based not upon the original, which was published in Polish, but upon the French or German version. This made it even harder for the translator to give a fully adequate rendering of the original intentions and ideas of the author. In a(ldit.ioll, due to the factors (If space and time, the translator was deprived of the benefit, IIf extensively discussing with the author even the major difficulties encountered in his work, and so achieving a meeting of minds before the text was set up in type. To illustrate this point I may mention that, for various reasons, I have been unable so far to read a considerable part of the present text, and it seems more than likely that I shall not have read it before receiving a copy of the published book. The realization of the difficulties involved makes me feel all the more indebted to him whose initiative, devotion, and labour have brought this volume into existence.

It is needless to say that I fully appreciate the assistance and consideration of all the persons mentioned in the translator's preface. But I should feel unhappy if at this place I did not make special mention of the man

xiv

AUTHOR'S ACKNOWLEDGEMENTS

who helped me more than anyone else in my part of the job-my younger colleague, the late Jan Kalicki. He spent the last two years of his short life in Berkeley; he generously offered his help on the day of his arrival and continued it untiringly and patiently, with the greatest devotion and conscientiousness, until his last day. He studied the originals of the articles, translated for Professor W oodger various passages from the Polish text, prepared references to recent literature, discussed with me the remarks which I planned to insert in the translations, assumed most of the burden of the extensive correspondence connected with the publication of the book, and read the first batch of galley proofs. His tragic and untimely death (in November 1953) was a cause of considerable delay in the publication of this volume.

EDITOR'S INTRODUCTION TO THE REVISED EDITION

I

University of California Berkeley, August 1955

.' . 1 t d areas who are already familiar with

SPECIALISTS in logic and logic-re a ~ L ' S ntics Uetamatilematics

, . f 'If d T ski's ooic eman ",'

the first edition 0 ,'\. re arski : , . . f

" ' ' ) do not need an introductory essay to m arm

(papers from 1923 to 1938 . II . ted results are

. it mathematIc a Y onen

them of its contents. Moreover, I s d 'd' . of Tarski's

I k' nd Surma 1 an ISCUSSIon

ably summarized by Pogorze s 'I a ki 'll known article 2 This essay,

subsequent work is included in Mostowe I, s we k ttemp't to outline

. d d f . readers' It rna es no a

however, is inten e or new , d it does not treat Tarski's

, I hi evements an I

Tarski's more mathematIca ac I, . d" t h . Tarski's development

R h it urpose IS to mrnca e 0\\

later work, at er, I s purpi h d 10 • of deductive science can

of the coneeptual framework of the met 0 0 gy

h h rti I in this volume,

be traced throng tea IC es d t ding of deductive science

In the hundred years 1879-1979 our un, ~rs a~. ot only has this period

, vth and reVISIon, l'

has undergone extensIve grov h' d in the two millennia

It th n had been ac ieve 1

contrihuted more new re~u R ' ~ J Fre e's BegrijJsschrijt (1879},

A ' I' ~ ol "/11'8 lc 3.)0 R,C. to g

from ristot e s 11 II!, ",' ". d th basis the significance,

1 ' tl I st hundre years ' e ,

hut it is also true t, lilt III .ne as h . b greatlv clarified, Indeed,

I' , ti f tic e'lrlipr results ave een, f h

and the IIIl11 a ions () ,1 • " . d s the century 0 t e

f . t this recent peno a

we are justified in re erring ,0

methodology of deduetive science. 'of deductive science to analyze and It is the task of the me:hodo.log) b th formalized and unformalized, investigate existing deductIve sCIences'r 0 I' tion of deductive sciences,

t' and orma iza

to contribute to the construc IOn he ori iples and methods used

and to formulate and critically evaluate t e prInCI

. c- tics If flamathematicB, J auruai

S S review of Loqic, .,eman ,," '

I Pogorzelski, W, and . urma, •

of Symbolic Logic 34 (1969) 99-106;, Th E cyclopedia of Philosophy, New York, 1967,

2 Mostowski, A, "Alfred Tarskl, e n

vol. 8, 77-81.

POSTSCRIPT

Tho present, second edition of Logic, Semantics, jJ.f etamathematics does not differ essentially from the first. The work on the original texts of till' papers included, as outlined in the translator's preface (page xiii, lines ))-)1-;), has been continued and, of course, extended to the text of the first edition. Thus, new misprints and errors which have been noticed in the meantime have been corrected; some cross-references to other papers and references to later developments have been added; and certain changes have been carried through to clarify various passages of the earlier texts and, in particular, to remove a number of translation defects of the first edition. A more thorough and essential revision of the work (which might be desirable for several different reasons) was not feasible.

I wish to express here my most genuine and cordial gratefulness to Professor John Corcoran for his work related to the publication of the present volume. The new edition of Loqie, Semantics, AI etnmalhematics is almost exclusively due to his initiative. In serving as the editor, he fulfilled his task with great fervor and enthusiasm. We collaborated in searching out and trying to remove various defects and weak spots of e~lrlier texts. The new indices at the end of this volume, prepared by Corcoran, are much superior to those in the first edition and hopefully will f:H'ilitatp t he ~tlld," and tlw usc of the work. 'Vita! is more valuable, he has provided I lIP volume with his own introduction. While I do not always a!l;ree with Corcoran's judgments expressed in that introduction I sincerely enjoy having it included in the volume and I feel confident that it will prove interesting and stimulating to man)' readers of my work.

Bakdey, California January 1982

ALFHED TAHSKI

xvi

EDITOR'S INTRODUCTION

EDITOR'S INTRODUCTION

xvii

a An elementary and clear ti I' .

is contained in his Introducti par la o~ervlew of Tarski's conception of methodology

(New York 1941) e . II on. tOttOgtC and to the Methodology of Deductive Sciences communications o;er SPf)C1~J I~ e preface. and in chapter 6. In several personal chapter not despite b~t ~rl 0 r~ars, Tarskl has emphasized the importance of this that thi~ chapter "should altozeth. Its elementary character. Church wrote in a review accounts" (Journal of Sy a bo~ge L er.s~l(rsede older and more familiar, but less accurate, Metamathematics was 'wr~te tC f oqu: . 1~4I) 30-~2). The material in Logic, Semantics, various elementary and ;un;a or :Ple~~ahzed ~lIdll'nCeS and it presupposes or avoids a more general audience in the~~~: ~~as which are expressed in a form suitable for

• Compare Tareki's remarks pa s 170 d

balance vis a vis formalized a 'd g~ .an 211 below. It is important to maintain

the fact that certain method:lo ui~a~rmll:hz:d language. There is no conflict betwt'cn languages and the fact that s . g tifi principles can be applied only to formalized guages. Indeed, Tarski's axio~~~n ~. c statements are normally made in natural Ianlargely in natural language Mo: Ions of farts o.f methodology involve axioms stated Tarski unless it is at least ade over, a ormal~zed language has little interest for "Truth and proof" Scieniitie qAuatel.y transllatable Into natural language. cr. Tarski, A.,

• 'J~ mencan vo 220 (1969)6' 63-77

One of the obstacles to acce tin th '.. . '..' esp. 68.

method was the idea that I P g e possihillty of SCientific treatment of deductive

systematic attempt to expr==: can not apply to. language itself, i.e., that any to things would inevitably res It . Ions among expressions or relations of expressions was the idea that formalizatio~ i~~ :::;nsense. or ~If-eontradiction. Another obstacle through formalization are fictional; ulcesfdlstortlOns so severe that results gained

. ee a so n. 12 below and, e.g., p. 267.

addition, one must recall that at the onset of the century several deductive sciences had already been constructed, none of which had been formalized, These include geometry (Euclidean and non-Euclidean), number theory, analysis (or the theory of real numbers), algebra of magnitudes, and algebra of classes, The Cartesian logician Arnauld [1602] had evcn attempt cd to axiomatize part of what would now be called the metatheory of Aris-

totelian logic,

In comparing the most recent work in the methodology of deductive

science (e.g, model theory) with the earliest work in the century of methodology (in particular, Frege's BegrijJsschrijt), one notes a clear qualitative difference, Roughly speaking, the difference can be seen to involve three interrelated ideas: (1) explicit concern with metatheory per se, signalized by self-conscious use of the metametalanguage; ~2! explicit formulation and use of syntactical laws, including the defimtlOn of formal languages by means of recursive (or "generative") gramm,ars;, and ~3) explicit formulation and use of semantic (or model-theoretic) Ideas, I,neluding satisfaction truth and consequence, If we consider only the dIS-

.' ' h . I f 'nference new independence

covery of new logical trut s, new ru es 0 I : '

results, and even the new axiomatizations of varIous, SCIences, ,th:re

, 'II 'I di g perceptIon of contIllUlty

emerges an Important, but potentIa Y nus ea n , ,

" I' ti diff e ce is not III the character of

WIth previous work, The qua Ita ive I er n

, hal" f ideas and methods

such results themselves but rather III t e an ,) SIS 0 , ' '

, ' f I' t' f the presuppositlOns WhICh

used and in the artICulatlOn and orma iza ion 0

formed the basis of previous results and methods,

It is probably futile to survey the literature of methodology from Frege

, . I' k tt mpts to discover the three

forward trying to find III the ear Y w or s a e ,

, ' , 'if G metry [1899J WhICh partly

Ideal', For example, Hdhert's FoundatIOns 0 reo '" '

h ' f r less awareness, implicitly

concerns the metatheory of geometry, S 0\\ sa, F' B

, ' 'I' , f f lization than does rege s eq-

or explicitly of the pos:.ni)lltles 0 orma , '

. , ,I lit t re of this period one

rijJsschrijt [lK79j.6 Neverthelesl'l, III the ear Y I era u ,

I s clear Ideas (completeness

does find an accumulating stock of more or es I

"I' d f rth) which can be proper Y

of theories, categoricity, deCldabl Ity, an so 0

, inted out that Whitehead and

8 Tarski, commenting on ~ draft o! this essa~tr ample, Since Hilbert's axioms Russell's Principia M athemattca [1913] IS an even . th ex ay of awareness of methodare written in natural language one. ex~.cts le~~:~ati:: :here the object-language is ological requirements than with Pnnctpta Ma. . M th matica is even loss advanced written in symbols, Still in some respects P_rt~ctPta h I a e dvanced than JJc(Jri.ff.~.~chrift. than Hilbert's Foundations of Geometry and It IS muc ess a

in the constructions and formalizations, 3 This field thus includes, as a part, what is sometimes called metamathematics, the study of the structural (or syntactic) properties of formal systems, The modern conception, and perhaps the very existence, of the methodology of deductive science as a separate science and in particular as a formalizable' deductive science is lar~ely due to Tarski in that it was he who collected, analyzed, ~nd codified emerging concepts and results, and it was he who most clearly and forcefully articulated and defended the possibility of such a science, Althou~h such a defense is no longer needed today, major mathematicians and philosophers of the time held that scientific treatment of deductive s~ience was impossible." It was Tarski's development of an increasingly rich and definite conception of methodology as a separate science-a development which can be traced in the present volume-that made such defense superfluous,

To un~er~ta~d Tarski's contribution to the modern conception of methodology I~ IS I~p~rtant to realize that there had already been several a~te~pts HI a Similar direction before the century of methodology, beP;1l111I11g, of course, with Aristotle's Posterior Analytics and in some respects, temporarily culminating in Bolzano's W issenschajtslehre [1837], In

xviii

EDITOR'S INTRODUCTION

EDITOR'S INTRODUCTION

xix

II

. I e (unformalized and without explicit semantics).

;:~~;;~~:a~rt~c~:Ue:~licitlY refers to the deductive science resented ,;:

t th . and thus the paper is written in a metamets anguage.

~:es:m:~~~her indication ~f the co~tent .of this ;:~~~~i:: ~;~t::~~~ t!:~~

it explicitly distinguis~es finitely aXiO~atlZa:::ms which are not finitely it demonstrates the existence of deductive s~d f odern methodology,

· d it bsence of some leas 0 m

axiomabzable, Thus espite a f b ic idea and it contains

. h I wareness 0 one asi

this early article sows a c ear a " and even

· . h time were regarded as surprlsmg

crucial results which, at tel. ' . I' itlv . tended to establish an

• l\. 1 . th article IS exp ICI . .} in ,

paradoxical. xioreover. e " hi h tl odological investiga-

. k" vithin w IC me 1

abstract "conceptual framew or '. di l' es can be carried out

• 1. deductIve ISCIP III

gations concermng concre e .

(III, 36-37), . ' summarizes and svnthe-

Article IV, written by Tarski ancltukaSlbewlcz, i the Warsa\\' School of

. d t] ds of several mem ers 0 , .

sizes results an me 10 IS • • ' f h resent point of view IS

. ' t i this article rom t e p

Formal LogIC. Prolllluen . 111 l' d rccur~ive grammar (of a

. t but unforma Ize , '

tho construction of a correc , f viltactic operations. In

d licit awareness 0 s" .

very simple language), an exp ICI· h' db. I'nterrelating axiomatIc

. . sults ac leve ,

addition, the article contalllS res . 1 lei li an interrelation that

. f d fi ' g sententIa ca cu ,

and matrix methods or e run the ·lllt.errelation of proof-

'11 I t be seen as

clearly foreshadows what WI a er.' ic) ethods.1o

. d d I theoretic (semantiC m

theoretic (syntactic) an mo e - . 1 'III b t published in the same

In article V written latcr than artrc e . u f III Here there are

'. d I the Ideas o· ,

year (1930), Turski further evc ops . . 0 h that Tarski first begms

, I . Its and It 1::< ere d

several IIC\Y metitodologlca resu .,' . 1 riving logic) of a de ue-

. I tl I 'ral baSIS (or unc e. .. .,'

to articulate tho idcn (I tie Ofll . II 0'. that tho dIstInctIOn

, ., . f t I . id('~l of logIC:l Ja~IS < • • )

t IVe fly:;t em. It I:; hy Ilw.tIIs 0 ie . d substantIve sCIence

, .. ' ( r bet ween logic an .

bet W(,PII lu"de lUi( I ll111tll4'llmt U'" () licit bj eet l:lllguage syntax

. . .1 ' V "t ill lacks cxp ICI 0

I~ drawn. N ('\,('I't h(·II'~s, nrt It e .

and semantics. . , Tarski becomes more explicit ab~ut

III article VI, first published 111 1931,· informal generative

I . I b sis, ~Ioreover, an . d t

the exact nature of the oglCa a; d reference IS rna e 0

1 biect language, an . b

grammar is given for a eornp ex 0 ~ ." (f sentential functIon Y a

flni "atlsfactIOn 0 a

the crucial problem of de mng s

articulated only within a metatheory having an explicitly formulated syntax and semantics, 7

A more fruitful route is to trace the above ideas backward. In this way one can identify the period between the two world wars as the watershed interval during which the qualitative transformation was achieved.

A very early attempt, perhaps the first, to codify a part of methodology of deductive science itself as a deductive science is contained in Tarski's 1928 lecture to the Polish Mathematical Society (first printed in 1930, and translated as article III below)." In this article, Tarski recognizes that formalized deductive sciences form a field of research in methodology "roughly in the same sense that spatial entities form the field of research in geometry" (III, 30). He then proceeds to present a deductive science each of whose intended models contains deductive systems,9 It is important to make some observations on this point. First, no semantics is presented for the object languages of the deductive systems in the intended models. Second, no explicit or implicit grammar is given for thCHC object languages. Third, the language of the theory is part of ordinary

7 This point becomes particularly clear on reading works by Dedekind, Padoa, Huntington, Veblen, Sheffer, and Langford which are cited by Tarski and listed in the bibliography, pp, 456-462 below.

8 Methodology was not thought of as a single, unified science to be reduced ultimatcly to a single comprehensive deductive discipline. Moreover, Tarski docs not indicate any single scheme for dividing it into separately codifiahlo subsoiencos. In fact, one aspect of Tarski's approach that adds clarity and flexihility is his willingness to isolate manageahlc suhsciences according to the needs at hand and to aim at relntively eornprehensivr, treatment of a limited subject matter instcnd of hcing foreet! to I'{'ttle for a more limited treatment of a eomprehensive slIbj(·ct, matt . .,r. He!', ('.g., V, f)(). Homan numerals indicate articles in Dogie, Semantics, IlJctatlwthcmali(·.~. Arabic numerals indicate years or pages.

9 Dogie, Semantics, Metamathematics, pp. 30-33. An intended model of the first theory presented is a pair (S,Cn) where S is the set of all sentences of a language and Cn. is the d~rivability relation. construed as a function which, when applied to a set of sentences, Yields the set of derivable consequences of the set: if a sentence x is a consequence of a set Y of sentences then x .Cn( V). A "deductive system" is defined as a set of sentences closed under derivability. Under an intended interpretation an individual variable ranges over sentences and a set variable ranges over sets of sentences thus including in its range all deductive systems consisting of sentences from S. Also note that the multitude of intended interpretations makes this a modern (rather than classical) axiom system in the SC?se of J. R. Shoenfield, Mathematical Logic, 1967, p, 2. Compare this remark with Tarski's eontrast of Boolean algebra with mereology (XI, 333n).

ki's Undecidable Theorie8 (with A. Mostowski 10 See for example, pp. 10 a.nd 1l of Tars

and R. 1\1. Robinson), Amsterdam, 1958.

xx

EDITOR'S INTRODUCTION

EDITOR'S INTRODUCTION

xxi

~I Strictly speaking thc definition simply introduces the sign Tr into the theory by fixln~ Its ~eam~g In terms of the primitive concepts. The formal correctness of the dcfin!tlon 18 o~vlO~a and ,there can be no issue concerning the material correctness of a definition which sl!llply Introduces a (new, previously meaningless) term. The issue is whether the extension of the term so-defined coincides with the extension of the concept of true ~~tenc.e. Although Tarski's argument for coextensionality is convincing to many 10~C1ans It ~ust be noted that "the correctness of the definition" does not admit o~p~f In the strict sense anymore than Church's thesis (the proposition that every e ectJvely computable function is recursive). It would be less misleading therefore to shpeak of the ~orrectness of Turski'» thesis (the eoextensionality of Tr and tr~e sentence) rat er than of the correctness of the definition "

12 It' h .

Th ~8 not t e cam:,. however, that the "Wahrheitsbegriff" was uniformly welcomed,

ere I ~ open hostility toward use of the classical concept of truth even beyond a gene~ distrust ?f semantic~ wit~in segments of philosophic and mathematical comm~l~l':'. Accor~mg to th~ historian, John Passmore, " ... some of the more obdurate ~s~~ts;emalned c~nvmced that Tarski was a metaphysician in formal dress" (A f 11 TeaTS o! PhiWsophy, .London,. ~957, p. 401). A surprising amount of the dis-

avor seemed to mvolve confusing definltion with criterion This di ti ti ire d

presupposed in lass' al hil ph' ,IS me Ion was a a y

tion of "decisionc pro~let!,,~, y andhim mathematics, ~specially, e.g. in formula-

, "IS one t ng to know what It means to say that a sen-

tence IS true and quite another to know of a criterion of truth i e to have a method ~hereby the trut~-value of a sentence may be determined (VIn' i97 238 254) I deed

:~~;yt~rie:~:~~~~ !~~~~~:;rt i~s ~~ effective criterion for determining iruth-~a~es of

of arithmet ic " Cf C • k cessary to presuppose a definition. of "true sentence ed. 19(3)," .v.ernap s remar s, p, 61, The Philosophy of R, Camap (P, A. Schilpp,

In order to treat the syntax of the object language within a deductive metasystem it is necessary to present a formal definition of the set of sentences of the object language. Such a definition has become known as a recursive grammar, and, more recently, as a generative grammar. In order to present a formally correct generative grammar it is necessary to conceive of string theory (the laws governing the interrelations among strings of signs) and to codify string theory as a deductive science. In the "Wahrheitsbegriff," Tarski isolates as a primitive notion the f~ndamental operation of concatenation of strings and he presents, employmg concatenation the first axiomatic codification of string theory (VIII, 173), thereby pr~viding deductive foundations of scientific syntaxY Article VIII also contains the first formal presentation of a generative grammar (VIII, 175, 176). It is to be regretted that many lin!);Uists, philosophers, and mathematicians know so little of the history of the methodOlogy of deductive science that they attribute the basic ideas of generative gra~~ar to linguists working in the 1950s rather than to Tarski (and other logiCian/

methodologists) working in the early 1930s. Cf. VIII, 267. . '

. 'I th b concepts in distingUlshmg

In the same article, Tarski emp oys e a ove .

between truth and provabilitv (VIII 237) as well as in raising the quest ion

.. , . "d .' h losi I consequence (VIn 252 ff,).

of whether dedUCibilIty eOinCI es wit oglCa '

. '" 3 T ki rsues the idea of formal-

In article IX, first published ill 19 3, ars I pu . iff"

. , b in fh "Wahrheltsbegrl

ization of the metatheorv that he had egun III e d

., "I h . able to assert that e-

and, via the Idea of omC'ga-lIlcomp eteness, e IS h it '

d . " ,,' hi' I "equence. Even thoug I IS

ucibilit.v does not coincide Wit oglca ron,,· II

, . r rtiuct both conceptua y

IIOW clear that dt'durihiiitv nud ('OIlSC'C]uellCC afC ( I~ . . , tifi

, 'C('SSarv for the SCiCn I c

and extcnsinnnllv. till' ('()lH'ppt. of rOTlseqll(,lIr(',!H ... , ", d I ' thi

f . ' ' f I I ,t ill It been arllCulate ' t IS IS

oundation of lIld('P('IHiC'IW!' pro» l'!, IIV ~ I tH

. ' 'I X'T) f 'I I)uhlished in 1936.

tusk which i)o1 ac('olHplislll'd III art!!' e. ~ • irs , lns th fi t

. 'I ' '[ ce " contams t e rs·

Artielo X V I, "Ou t.he ('OIH'(~pt of lo!(Jeu COlIl'cqllc 1 ., ,,'

" countermodels notion,

careful formulat ion of Tar'ski's now famous no f ence

and it also contains clear distinctions of the proper concept 0 consequ

, ' ' . kl ' LukasiewiCZ, Thue, Post, and

.~3 Other authors (including Lesmewskl, Adiu lewl~~ B t these authors presuppose:I ~Odel) had used ideas about scientific syntax before ~9 . u f I deductive setting, This !n an informal way the ideas that Tarski presented In a I~e Udations of ideas already ~ one more instance where Tarski provides the f0fl:ll~ oun lication of the principles Infonnally in use and it is a particularly clear and stnkmg odapp uo« to Logic and to the

a d h " h t 6 of his Intr uc "' .•

n met ods outlined by TarskI III c ap er 3) Methodology of Deductive Sciences (cited above in footnote '

sequence of objects in the intended interpretation). More importantly, article VI employs for the first time the precise methods later used in the definition of truth (VIII, 277), The immediate concern, both of this article and of the next (VII, written with Kuratowski) is the concept of definable property. VI contains explicit demonstration of the existence of undefinable properties, while VII articulates the distinction between explicit and implicit definitions as well as containing results concerning elementary definability.

The next article, VIII, is the famous "Wahrheitsbegriff" (Truth-concept) which can be seen to be the watershed between early and modern methodology of deductive science. In this article, Tarski articulates the idea of the extension of the concept of a true sentence (of a fixed, interpreted formal language). He then defines the set Tr and argues convincingly that Tr is identical to the extension of the concept of a true sentence." Tarski's definitions of truth and of its ancillary notion, satisfaction, have been widely accepted and widely celebrated." But article VIII contains much more: the first presentations of deductive foundations for semantics (including key ideas of model theory) and for syntax (including generative grammar).

xxii

EDITOR'S INTRODUCTION

EDITOR'S INTRODUCTION obstacles to it (and whether the obstacles were genuinel~ overcome. or

. d ). how it was emploved (and whether It serves essen-

merely set to one SI e " . 1 t k f then

tially different uses today); and what ~ts role.was ,int t ie I~l:.t~~;r ~o:ple~c<~

existing ideas (e.g. whether it fitted lIke a Ple~e m o~) ~Iorcover such puzzle or whether it started a new puzzle, so 0 spea " .heor .

di ft id in "demvstifving" aspects of currently accepted t ,J

stu res 0 en all. ' I ' h it has been shaped by extrinSIC

b h ' the extent to w lIC 1 . .

y S owing, e.g., . . vriters debated issues long SIllCC

factors such as interests of partIcular , 'd' . may find that

F' llv through such stu res we

forgotten, and so forth, <ina y, "th the full range of

currentlv accepted theory has not yet come todtermsd"1 d in this wav we

. "II dresse an, . J'

problems to which the idea was ongma, Y ~

may be led to new ideas or to new apphcatIOns,

xxiii

from two notions still sometimes confused with it, viz" deducibility (XVI, 412-413) and "no expressible countermodels" (XVI, 415-416), The need for these two distinctions, consequence vs. deducibility and consequence V8, "no expressible counterrnodels," is now widely accepted, but their philosophic importance is often overlooked. The conflation of consequence with deducibility amounts to replacement of a concept (consequence) with one of its criteria (deducibility), Such replacement undermines the basis for extending, refining, criticizing, and justifying the criterion, and it sets forth the criterion as if it were its own basis, In particular, completeness and soundness questions become meaningless, and rules of inference become arbitrary or purely conventional. On the other hand, conflation of consequence with "no expressible countermodels" makes consequence contingent on the expressive capacity of the language and opens the possibility that a consequence holding relative to a given language would fail were new terms added.t! As in many other cases, Tarski did not see himself as inventing or creating (or even defending) distinctions, but rather as analyzing and articulating distinctions already implicit, perhaps only vaguely, ill methodological practice (XVI, 409-10, 414),

Thus, during a period of eight years (l92S-1936), Tarski analyzed and coilifi!'d, ill a formal setting, the foundational ideas of syntax (viz" axiomat i4~ st ring theory and formally stated recursive grammars) as well as the foundational ideas of semantics (satisfaction, truth, and consequence). Thcse ideas are so familiar today that it i" often difficult to realize t hat it was necessary both to analyze t hem and to articulate the results of the analysis within a deductive metutheorv.

X evertheless, perhaps t he best way' to gain an underst finding of these ideas is still to read Turski's articles which present the original statements of what are now widely regarded as definitive formulat ions. To begin wit h, it is oftcn revealing to study the origin of an idea in order to v;rasp the needs for it (and the extent to which those need" have becn met); the

III

U This is implicit in !arski's remarks (XVI, 416-417) but he did not see fit to give an example. The following example is due in all essential respects to George Boolos ~tnL~cond-order Logic," Journa.l of Philosophy! 72 (1975), 509-527, esp. 525-527:

be a first;>rder language with the usual arithmetic constants interpreted in the natural numbers III the usual way and let L2 be obtained from Ll by adding one unary pr('dl:,~te T mterpretr-d as th(~ set of Godel numbers of the true sentences of Lt . Using Tarski s theorem It. can he seen that tho relation "no eounterrnodels expressible in L2" IS II proper ~llhspt of t hr- relation "no countt'rmodcls expressible in Lt ,"

, . 'Metamathematics can overlook Tarski's

No treatment of LogIC, Semantics, h l'lshments mentioned

fd fi iti n In t eaccomp

contributions to the study 0 e III 10 ' f thods of definition, and

" th development 0 me

above, the use of definition, . e t' I d rather prominent.

, . d fi ' ti vere all essen ia an

various results involving e III IOn' ith f Tarski's studY of

, b id here el er 0 .' •

But, almost nothmg has yet een s,a d f d fi abilitv,

, 'I f hIS stu Y 0 en.

definition itself or, III particu ar, 0 • I in Tarski's work,

bTt I v important ro es '

Two concepts of defina I I Y P a. , '1 tic definability (XIII,

l d fi bilit vane seman.

Tarski refers to them as forma e na, " merelv a "formal

, h ,,,ever that one I~ , ,

31;6), It IS by no means thc case, 0 ., .', d It "'I'th only in artIcle X,

I d fi I 111 tv IS ea .

analogue" of the other, Forma . c nao . "I ' VI (1931) VII (1931),

, I ' I Tf ,',; dealt Wit I III ." ,

(1934) while semunt ie ( chlla)l 1.~ J. '" f ' itives from an umn-

, 'I' ' I I tv 0 prIml'

VIII (1 \1:1I), nnd XI II (l n:l4-.'i), h mlllH~ )1 I . t t formal definability

, '~W'IOn) IS relevan 0 .

t('rprc~tc>d formal t IlI'ory (X,.·h' relevant to semantic

f t I Hich'\rcl type are ,

only, whoreu» paradoxes 0 lP .• '. h t there are no Important

1· , , Thi t t· implv of course, t a

definnbilit.v onlv. liS IS 110. 0 . ,

relationships between the two notions, II . in the context of an

bilit ise most natura ) , I

Issues of formal defina 1 1 yam, d t K of non-logic a

T' lving a fixe se " h

axiomatized deductive theory mvo d underlying logiC Wit

, d) nd base on an ,

constants (possibly unIllterprete a I f d finition, The questIOn

, f and ru es 0 e 1 t

fixed logical axioms rules of III erence, t Kl and K2 am .0

" .' K' t two par s .

arises whether it is possible to diVide III 0 . s" stated III tcrms

A 1 of "proper axIOm. ,

reaxiomatizc T by means of a set

xxiv

EDITOR'S INTRODUCTION

EDITOR'S INTRODUCTION

xxv

of KI only, in such a way that T is exactly the set of sentences derivable from Al using the logical axioms, the rules of inference, and the rules of definition (applied to introduce the constants in K2 in accordance with the way that they were used in T), In other words, is it possible to reaxiomatize T so that only some of the "primitives" remain undefined and the rest become defined? If this is so in a particular case then, since definitions appear in the equivalent reaxiomatization, these same definitions must have already been derivable in T as originally presented, Accordingly a constant, c, is said to be formally definable in terms of a subset KI (excluding c) of the set K of non-logical constants on the basis of a set S of sentences if a definition of c in terms of KI is derivable from S, It is useful to realize that this concept, formal definability, is entirely "formal" in the sense that discussion of it does not require any semantic notions (satisfaction, truth, assignment of meanings to non-logical constants, and so forth),

Issues of semantic definability arise in the context of an interpreted language L based on an interpreted set K of non-logical constants, If a is an object (individual, function, set, or the like) appropriately built on the interpretation of L, then one can ask whether a is expressible (in anyone of several appropriate senses) by means of a formula of L, For example, in VI (1!J:31) Turski takes L to be a certain language interpreted in the domain of real numbers and he takes a set S of real numbers to be expressed by each formula F(x) having x as its only free variable and such that a real number is in S if and only if it satisfies F(x). (It is useful to realize that the context of semantic definability does not require axioms, rules of inference, or rules of definition). It is obvious that there arc uncountably many sets ?f real numbers that are not definable in any single given language L havmg a countable set of formulas. It is also obvious that a test for formal defin~bility p~r se simply does not apply to the class of problems: given L and g~ven S, IS t~ere a formula of L which expresses 81 TllUs, in order to establish semantic non-definability results, other methods needed to be de~eloped. For example, in XIII (1934-5), it is observed that every 10gJCall~ definable set, i.e., a set which is semantically definable using no non-logical constants, is invariant under certain transformations of the ~niverse ~f discourse and that, therefore, any set not invariant in this way IS not logically definable,

Another difference between formal and semantic definability is that the

first is not "iterable" but the s d i Th

e secon IS, e set D of constants of Teach

formally definable in terms of a certain given subset of the set of all con-

stants of T is not itself a constant of T; thus it is meaningless to ask whether D is formally definable in terms of the given subset. However, the class D' of all subsets of the domain of an interpreted language L which are (individually) semantically definable in L is itself an object appropriately built on the interpretation of L. Thus, assuming that L contains a variable X ranging over sets, it makes sense to ask whether D' is semantically definable inL, i.e., whether there is a formula G(X) in L which is satisfie~ by a s~t S if and only if S is semantically definable in L (i.e. if and only If there ~s a formula F(x) in L which is satisfied by an individual b if and only If ~ is in S). In fact, it is the iterability of semantic definabi~ity that T~rskl exploits to obtain various results in VI (1931). Once the Idea of the Iter-

bi . ie definabili d j the context of higher-order

a ilitv of semantic defina I ity IS graspe 111

" . devi ". I te" the effect in first-order

languages one can, by vanous evrces, simu a

languages.

. ., f I I " f a variant of the method

In article X Turski gives a orrna ana )SIS 0

, bili hi h h d been presented informally

of establishing formal undefina I ity W IC a ..

. P d' thod yields a conditIOn for

by Padoa over thirty years earher. a oa s me .

. . . 'I t t i not definable 111 terms of

establishing that a given non-logica cons an IS .

. I ti to a given theory. Tarski

the rest of the non-logical constants re a rve .

d h P d Tarski method which

generalizes this to what might be calle tea oa- .

, ., h ' on-logical constant IS not

Yields a condition for estabhshmg t at a gIVen n . t

h t f the non-lOgIcal constan s

definable in terms of a given subset of teres. 0 d' .

" T ki i dicates that the con ition

relative to a given theory. In addition, ars 1 In difi ti

d b th t a minor mo I ca IOn

as stated by Padou is stronger than nee e so a

of Padoa's own method is much more broadly applicable. . ' . .

, . 'f d to as "definabIhtY-lll-a-

Today these two not ions, commonly re erre d d

' .. '\ I" . , 'idclv understoo an em-

theory" and ''lh'fillalnht.y-JIl-a-mot e, .1R w .. ' iod.seri

, ' r , ki , kIt he IJreceding perlo ,serIOus

ployed, lurgely ns u result of l'urs 'I i'! "or , II . • d t

, " 'I' , ,'!ted from f!l1lure to atten 0

pilliosoplll('nl lUlII 1l\llthelllllt.J(~:t ('ITOI!' resu ., 'h t

i 15 11 'ver despite t e presen

details of delinnhility that ure now c ear. 0\\ e , '

. ' I in notions of definability and, in

15 In philosophy there has been eonfusiOn mvo V; 'b . terms of other sorts of particular, concerning "definability" of one sort 0 t, ~g ms of sets). It is nOW clear things (e.g, "values" in terms of "facts" ?r numb~rs In derm ot involve definability in that in many cases arguments put forth m such ISBUes 0 n uments which do involve either of the above 'two senses and it also clear tha~ manYj a~ge In mathematiCS, inade-

hil hi ally tnconc USIV , ,(X

one or the other of these senses are p 1 ?SOP IC ibilit of sets of primitive notIOns " ,

quate beliefs had been formed concernmg reduci IldY , f Geometry," Munt8t 13

306-307 d V bl 0 "Hl'lbert's Foun atlOns 0

an compare e en, .

(1903) 303-309, esp, 306),

IV

EDITOR'S INTRODUCTION

. rtant to scholars actively engaged in original

tinues to be as useful and impo k .h t has happened and how it

research as it is to those concerned to now w a

has happened. f th stud" of any great book,

. b k is perhaps true 0 . e oJ

Study of this 00, as I h h to know before one can

. f 11 • how muc one as

helps one to realIze more u J know-a necessary step, of course,

know what it is that one does not yet

.' ..·t· that is not yet known.

in commg to know "hat 1 IS '11 f '1 t be profoundly impressed by

f thi -ork WI ai 0

No serious reader a . IS " • t 11 creates in these pages; but

. f h di . line he Vir .ua Y

Turski's mastery 0 t e lSClP d in his own research by

. d f il to be encourage . .

neither can the serIOUS rea er au i d d 1 human determmatlOn

ki' t dfast sober, an eep Y di

the example of Tars ISS ea, 1 everyone who stu ies

. . I' problems that perp ex

to come to grIPS With t re same

1 .h inspires explorers.

logic. Tarski is an exp orer w 0

xxvii

xxvi

EDITOR'S INTRODUCTION

accessibility of these ideas, they have still not made their full impaetespecially in technical philosophy.

The Pogorzelski-Surma review of the first edition of this work concludes as follows.

Most of Tarski's papers contained in ... this ... volume are classical for metalogic and metamathematics. A great many notions and theorems found here have already become standard and are a most essential component in logical education at all levels, beginning with the most elementary. Looking ... at the book as a whole, the thing that we found most striking is that there ... is hardly another scientist in the history of the exact sciences whose part in the construction of notions for a large domain of science was as powerful as the contribution of Tarski to the creation of conceptual apparatus for logic, metalogic, and even metamathematics. In fact, the conceptual structure of these disciplines is due to Tarski.

d gestions made by

. ifi tl from criticisms an sug. H B hnert

NOTE: This essay has benefitted ~I~I ca~ y ists and historians includmg . 0 and

logicians, philosophers, mathem~tIClans,~m~~z i. Hullett, J. Kearns, M. ~c~nla::veral N, D. Goodman, I. Grattan-Gumness, .: 's read previouS drafts an '. In heG. Weaver. In addition, Professor Ta~sk~~~st of the essay as well as varl~~:t~~~ive conversations, discussed with, me ~h~ ~~:ilS These eonveTSl\tions were V~~I for them.

matical philosophical, and hlstorlea h ' and I am grateful to . f the

, t ' tant c anges, T ki's own View 0

to me; they brought abou nnpor ort of Professor ars I 's mv sole

Nevertheless, the essay abc;>ve is not a r~fo . Not only are the abov,e :~\ did not development of his conceptIOn of !"et~od es ~~ whieh Professor ~arskl tn other cases responsibility, but there are sp.ec~fic ISS~ter of emphasiS or wordmg ~u I' IS in the fact seek agreement-in some cases It IS a ~a . One important disagreeme.n t I: cumulative, more fundamental matt('rs are a~ sta. ~ a development more conSIS e: 'reement with that he believes dUll my essay Imphe .\Hhough I am much ~or~ m gees are bound and directed than WI\8 act.llally the cas~'i slphl'r IIf logic, ecrtam dlffele~ose associates Professor Turski than wjt.h any nth(I'r P IC)'~ '1llstances hiB students hlln, ho~ be prefers

. I I av n uUln),' . ki d t 18 18 "

to he renl'eh'cl m t.1I' II HIVl' 1''''" .' .r with l'rofes8vr 1ars 1,lln

do not. Illways IIp;rc'(' wit.h I'l1l"h other (

it to 1)(,.

Despite the importance of this work as a historical document, it would be a mistake to regard it only as an archive of past achievements, or as a record of what has taken place during perhaps the most fruitful period thus far in the journey of modern logic. The articles in this volume are rich in ideas, methods, and results which have been applied to only a small part of their potential domain within both mathematical logic and philosophY (and hardly at all outside of these fields). Throughout this hook, moreover, there are suggestions of problems yet to be solved anti of important ideas yet to be analyzed. IS Loqic, Semantirs, lIfl'lama.thl'lnatics thus (:00-

J.C.

Ie There are passages which explicitly suggest problems and problem areas (e.g., 56, 174, 317, 382, 391, 412, 413, 419). In other passages Tarski expresses or implies dissatisfaction with his own characterization of various concepts suggesting in some cases that changes may be needed or at least that further justification is needed (e.g, "semantic category," 215JJ.; "structural property," 254JJ; "logical consequence," 418). Moreover, there are "programs" which either admit of being continued in various ways or for which there is no reason given why they cannot be continued. For example, in III and V there is a "grammar-free" characterization of "consequence." (Axioms 1-5, p. 31) These axioms can be seen to concern validity only, in the sense that they are satisfied hya logic in which every argument is valid (i.e. wherein every sentence is a consequence of each and ('v('ry set of sentences). This situation gives rise t~ several types of questions including: (I) do these axioms imply all of tho principles of the type sought by Tarski? (2) how can t h .. axiom Ret he extended in order to take account of invaliditv without losing its g~n(,r:l.l applicability? (3) besides "consequence," which other primitive notions should he addr-d in ordr-r to morr fully treat the type of situation at issue? Study of Turski's mode of inquiry generutes questions beyond itself.

State University of New York Buffalo, January 1982

CONTENTS

Editor's Preface to the Revised Edition TranslalQr' 8 Preface

Author's Acknowledgements

Editor's Introduction to the Revi8ed Edition

vii

ix

xiii

xv

I. ON THE PRIMITIVE TERM OF LOGISTIC

II. FOUNDATIONS OF THE GEOMETRY OF SOLIDS 24 III. ON SOME FUNDAMENTAL CONCEPTS OF

METAMATHEMATICS 30

IV. INVESTIGATIONS INTO THE SENTENTIAL

CALCULUS 38

(by Jan LUKASIEWICZ and Alfred TARSKI)

V. FUNDAMENTAL CONCEPTS OF THE METHODO-

LOGY OF THE DEDUCTIVE SCIENCES 60

VI. ON DEFINABLE SETS OF REAL NUMBERS llO

VII. LOGICAL OPERATIONS AND PROJECTIVE SETS

(by Casimir KURATOWSKI and Alfred TARSKI) 143

• VIII. THE CONCEPT OF TRUTH IN FORMALIZED

LANGUAGES 19

Introduction 152

§ 1. The Concept of True Sentence in Everyday or Colloquial

Language 154

§ 2. Formalized Languages, especially the Language of the

Calculus of Classes 165

§ 3. Tho Concept of True Sentence in the Language of the

Calculus of Clssaes 186

§ 4. Tho Concept of True Sentence in Languages of Finite

Order 209

§ 5. Tho Concept of True Sentence in Languages of Intinite

Order 241

§ 6. Summary 265

§ 7. Postscript 268 .

IX. SOME OBSERVATIONS ON THE CONCEPTS OF

",-CONSISTENCY AND ",-COMPLETENESS 279

X. SOME METHODOLOGICAL INVESTIGATIONS O~

THE DEFINABILITY OF CONCEPTS 296

XI. ON THE FOUNDATIONS OF BOOLEAN ALGEBHA 320

XII. FOUNDATIONS SYSTEMS

OF THE CALCULUS

OF

342

XIII. ON THE LIMITATIONS OF THE MEANS OF EXPRESSION OF DEDUCTIVE THEORIES

(by Adolf LINDENBAUM and Alfred TARSKI) 384

xxx

CONTENTS

XIV. ON EXTENSIONS OF INCOMPLETE SYSTEMS

OF THE SENTENTIAL CALCULUS 393

I

XV. THE ESTABLISHMENT

SEMANTICS

OF

SCIENTIFIC

401

ON THE PRIMITIVE TERM OF LOGISTICt

XVI. ON THE CONCEPT OF LOGICAL CONSEQUENCE 409

XVII. SENTENTIAL CALCULUS AND TOPOLOGY ABBREVIATIONS

BIBLIOGRAPHY

AN ALYTICAL INDEX

INDEX OF NAMES OF PERSONS INDEX OF SYMBOLS

421

IN this article I propose to establish a theorem belonging to logistic concerning some connexions, not widely known, which exist between the terms of this discipline. My reasonings are based on certain sentences which are generally accepted among logisticians. But they do not depend on this or that particular theory of logical types. Among all the theories of types which could be constructed! there exist those according to which my arguments in their present form are perfectly legitimate.s

The problem of which I here offer a solution is the following: is it possible to construct a system of logistic ·in which the sign of equivalence is the only primitive sign (in addition of course to the quantifierssj ?

This problem seems to me to be interesting for the following reason. We know that it is possible to construct the system of logistic by means of a single primitive term, employing for this purpose either the sign of implication, if we wish to follow the

1 The possibility of constructing different theories of logical types is also recognized by the inventor of the best known of them. cr. Whitehead, A. N., and Russell, B. (90), vol. 1, p. vii.

2 Such a theory was developed in ) 920 by S. LeSniewski in his course on the principles of arithmetic in the University of Warsaw ; an exposition of the foundations of 6 system of logistic based upon this theory of types can be found in Lesniewski (46), (47), and (47 b).

3 In the BOllS{) of Peirce (Reo Peirce, C. S. (58 a), p. 197) who gives this name to tho symbols 'IT' (universal quantifier) and '~' (particular or existential quantitlor] roprosont.ing abbreviations of the expressions: 'for every aignification of tho torms .. .' and 'for some signification of the terms .. .'.

455

456

463 503 505

t BIBLIOGRAPHICAL NOTE. This article constitutes the essential part of the author's doctoral dissertation submitted to the University of Warsaw in 1923. The paper appeared in print in Polish under the title '0 wyrazie pierwotnym logistyki' in Przeqlqd Filozojiczny, vol. 26 (1923), pp. 6S-89. A somewhat modified version was published in French in two parts under separate titles: 'Sur Ie terme primitif de la Logistique', Fundamenta Mathemalicae, vol. 4 (1923), pp. 196-200, and 'Sur les truth-functions au sens de MM. RUSBeIl et Whitehead', ibid., vol. 5 (1924), pp. 159-74. The present English translation is based partly on the Polish and partly on the French original.

VIII

THE CONCEPT OF TRUTH IN FORMALIZED LANGUAGESt

TIm I'NTBODuCTIoN

present article is almost h II

lem-the definition .I' W 0 Y devoted to a single prob-

reference to a en oJ t: Its task is to construct-with

F. 0' ven anguage--a mat . II

Jormally correct definition ,ena y adequate and

lem, which belongs t thO! the = true sentence'. This prob-

. 0 e elaseical t'

raISeS considerable difficuI . ques IOns of philosophy,

the term 'true se t ' .tIes. For although the meaning of

n ence In col] . I I

quite clear and intelli lbl oquia anguage seems to be

ing more preoisely ha! ;;~hall attempts ~o define this meanvestigations in whi h this erto been frUItless, and many inwith apparently evicdent ~r~ has been used and which started and antinomies (for whi 1 ~nusses have often led to paradoxes solution bas been found~' owever, a more or less satisfactory respect the fate of th . The ooncept of truth shares in this

th 0 er analogous t .

e semantics of Ian concep s In the domain of

Th guage.

e question how a certain .

rectly formulated nl if . concept IS to be defined is cor-

f 0 y 1 a list i .

o which tbe required d finiti s ~ven of the terms by means

definition is to fulfil its e ro hon IS to be constructed. If the

t n p per task, the sense of the terms in

lDLIOGRAPHICAL N

most part (ro 192 OTF., The reAult.R pn- • t I' ,

under tho f 'Urn" 9, 1 diM~IIMfI!'d f.hl'm . I I'll n :!'C In t.hll' paper date for the

tiveMcil'n' .: o On the (!nncl'pt. nf fr f'l In pllrf"""lnr. ill t.wo ,peturl'lI given ber 8, i9~) at. tho Logic Median or t.h;. ~I'h~,111 r"r"r"n",' f ... rormuliz('" dl'duc- 1930) A and lit the l'uliMh Ph'l . I flll4lphu"LI Hodd.y ill W'Lrl4l1W (Octo-

prese~ted (~hoJrt report: of these 11~:~:;~:':~III. M(~(~il'l . .Y .in ~,W6w [December 5, 1931 F y. l';uk8llleWicz) to th W III I arsk] [731. Thl~ paper was

. or reasons b d e arallW Seie t'fi So . 21

In the IDea' eyon my Control b . . n I C ciety on March ,

addit' ntllDe the original t t ' pu licatlOn WIIB deillyed by two years,

Ions (see p 247 r ex WIIS sup I d

paper WIIB .' '.lootnote 1). Also p emente by some substantial

The POli;~~~h~d III Taraki [76J. ' a SUmmary of the chief results of the

It WIIS subse grnal of the paper II

176b]) lind J!uent~y translilted in se~pe~~ finally in print as Tarski 176aJ, (Tllrshl [84b)) er, III addition to the era anguage~J first in German (Tarski provided with and Fren~h (Tarski [84~)~nt EnglIsh translation, in Italian hllve und II postscnpt in which . a,ch of these translations has been

In two e~~me a rather essential re:or:ne Views stated in the Polish original outline th a er articles, Tarski 182J .sldon and modification

I I' main ideas d an Tarsk' 184dJ I h' d

n the first of th an achievement . I , ave attempte to

objeetiotll! who h: articles I have a~ of this paper in a non-technical way. re ve been raised to the ~xpre~ed my views regarding some

Inveshgation presented here.

VIn CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 163 this list must admit of no doubt. The question thus naturally arises: What terms are we to use in constructing the definition of truth? In the course of these investigations I shall not neglect to clarify this question. In this construction I shall not make use of any semanbieal concept if I am not able previously to reduce it to other concepts.

A thorough analysis of the meaning current in everyday life of the term 'true' is not intended here. Every reader possesses in greater or less degree an intuitive knowledge of the concept of truth and he can find detailed discussions on it in works on the theory of knowledge. I would only mention that throughout this work I shall be concerned exclusively with grasping the intentions which are contained in the so-called classical conception of truth ('true-corresponding with reality') in contrast, for example, with the utilitarian conception (,true-in a certain respect useful'j.!

The extension of the concept to be defined depends in an essential way on the particular language under consideration. The same expression can, in one language, be a true statement, in another a false one or a meaningless expression. There will be no question at all here of giving a single general definition of the term. The problem which interests us will be split into a series of separate problems each relating to a single language.

In § 1 colloquial language is the object of our discussion. The final conclusion is totally negative. In that language it seems to be impossible to define the notion of truth or even to use this notion in a consistent manner and in agreement with the laws of logic.

In the further course of this discussion I shall consider exclusively the scientifically constructed languages known at the present day, i.e. the formalized languages of the deductive sciences. Their characteristics will be described at the beginning of § 2. It will be found that, from the standpoint of the present problem, these languages fall into two groups, the division being based on the greater or less stock of grammatical forms in a. particular language. In connexion with the 'poorer' langu~gC8 the problem of the definition of truth has a positive solution: there is a uniform method for the construction of the required

I Cf. Kot.arbinnkj, T. (37). p. 126 (in writing the pr.""mt nrtido I hav« repeatedly consulted this book and in many point<! adhered to tho wrmmol01O' there suggested).

PT OF TRUTH IN FORMALIZED LANGUAGES )55

VIII. § 1 CONCE . f

. d efforts which the constructiOn 0 II.

Amongst ~~e malllfol or the sentences of colloq uiallanguage

correct defillltlOn of truth f t tural is the search for a

c th haps the mos nil.

has called ror ,per. d finition which we can

8emantical definition. By this I mean a e

11 • words'

express in the fo owmg wvo=- he tat "atl'airs is so

h' 11, ays that t s e OJ JJ'

(1) a true 8entence is one w ~~ s. d d' and so.t

and so, and the state of aJfa~rs ~n ee ~s so t s clarity and

f· f formal correc nes , '

From the point 0 VIeW 0 . occurring in it, the

b· iby of the expressiOns

freedom from am IgID ch to be desired. Never-

la . b 'ously leaves mu b

above formu twn 0 VII' tention seem to e

.. ning and genera in .

theless its intUItIve mea k thi intention more definite,

d i t lli ible To rna e s . 1

quite clear an III e gi '. . el the task of a. semantica

and to give it a correct form, IS precie Y

definition. . t nces of a special kind present

As a starting-point certam sen e . 1 d finitions of the truth

. h ld serve as partIa e .

themselves whio oou I ations of vanous con-

orrectly as exp an , Th

of a sentence or more c " tJ.'1le sentence . e

crete turns of speech of the type x IS a be depicted in the

f hi kind of sentence can general scheme 0 t s

following way: .

(2) x is a true sentence if and only s: bstitute in the

t defimtlons we su

In order to obtain concre e .

. alIJlost everything w~ch

. . lies in particular, to antical antinolIlleB•

private con versatJOns; thiS ~pp ~ation marks and thesem. [ve LeimieWBki I shall say about expression~ I~ :~ fact does not in the least .,:v;recise fonn in

It remains perhaps to add t a I h and perhaps not qUI

in the responsibility for the sketcr.!ented.

which the following remarks are P 27 and

. und in KotarbiI1ski, T .. (37), P::~teIY

t Very similar formulatiOns are :mentaI'ieB which explaill app

136, where they are treated as co are for exa.IDple. the classical view of truth. . are not eesentiaUy n?w ~oI?t is n~t, or of what

Of course these formulatl?OS t1 .' To say of what 18 t. ~ I r of what is not the well.kn,:WJ?- w~rds of Ar;:: ;:;8&Y of what ~ thrt it 1B~i7~ Works, vol. 8, is not that It lB,.18 false: w Aristotle, MetGphy81Ul, • 7,

that it is not, is true. ( R Oxford 1908.)

English translation by W. D. 08S, •

154 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII definition in the case of each of these languages. In §§ 2 and 3 I shall carry out this construction for a concrete language in full and in this way facilitate the general description of the above method which is sketched in § 4. In connexion with the 'richer' languages, however, the solution of our problem will be negative, as will follow from the considerations of § 5. For the languages of this group we shall never be able to construct a correct definition of the notion of truth.'] Nevertheless, everything points to the possibility even in these easea-c-in contrast to the language of everyday life--of introducing a oonaietent and correct use of this concept by considering it as a primitive notion of a special science, namely of the theory of truth, and its fundamental properties are made precise through axiomatization.

The investigation of formalized languages naturally demands a knowledge of the principles of modern formal logic. For the construction of the definition of truth certain purely mathematical concepts and methods are necessary, although in a modest degree. I should be glad if this work were to convince the reader that these methods are now necessary tools even for the investigation of some purely philosophical problems.

§ 1. THE CONCEl'T OF TRUE SENTENCE IN EVERYDAY OR COLLOQUIAL LANGUAGE

For the purpose of introducing the reader to our subject, a consideration-if only a fleeting one-of the problem of defining truth in colloquial language seems desirable. I wish especially to emphasize the various difficulties which the attempts to solve this problem have encountered.'

1 The considerations whioh I shall put forward in this connexion are, for the. most part, not the result of my own studies. Views are expressed in theJD which have been developed by S. Lesniewski in his lectures at the Universi~y of Warsaw (from the year 1919/20 onwards), in scientific discussions and in

t Regarding this statement compare the Postscript.

156 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII. § 1 place of the symbol 'p' in this scheme any sentence and in the plac~ of 'x' any individual name of this sentence. '

GIven an individual n f

. arne 0 a sentence, we can construct an

expla~atIOn of type (2) for it, provided only that we are able ~ write down the sentence denoted by this name. The most Im~ortant and common names for which the above condition is sa~Isfied are the so-called quotation-mark names. We denote by this term every nam f

. eo a sentence (or of any other even mean-

m~' '

. SS, expression) which consists of quotation marks, left- and

ng~t-hand, and the expression which lies between them, and which (expression) is th hi t d

eo Jec enoted by the name in question.

As an example of S h . .

. . uc a name of a sentence the name 'It IS

snowing' will serve I thi

f . . n s case the corresponding explanation

o type (2) 18 as follows:

(3) ,. .

~t ss snowing' is a true sentence if and only if it is snowing·l

Another category of names of sentences for which we can

construct analogou 1 . .

s exp anations IS provided by the so-called

structural-descript' W

hi 'We names. e shall apply this term to names

w ch describe the d hi h

I wor s w c compose the expression denoted

Statements (sentences) I

expression, and thus as r are, a ,ways ,t~eated here as a particular kind of

'expression' 'state t' Inguistic enbities. Nevertheless, when the teJ1IlB

, men etc are' t ted ' f

printed sizna , ' ., m erpre as names of concrete serIes 0

,,_, varIOUS formulatio hi h "

to be quite correct d' ns w c occur In thfs work do not appear

consists in identif .' an glv~ the nppearance of a widespread error which sentence (3), sinc~:~~ih:essIOns ~f like sh8~, This applies especially to the be regarded 88 g I above interpretation quotatlon.mark names must the series of sign8o~e~~ (and n?t individual) names, which denote not only shape. In order tou ,~6uotll.tl~n marks but al .. o every Rorif's of signs of like of supertluoue co aVt?1 t' oth ,0hJections of thi" kind and alRo the introduction

mp rca Ion" mto tho di " ted

among other thin ith t ' IRCUaBIOn, which would bo connee

shape, it is convenf 7lt t?O necellluty of liKing tho concept of liken0t!8 of tence' etc d de 08 ipulato that terms liko 'wor,I' 'oxproaHion', 'son'

, ,,0 not enote concrct ' r' ' b

series which are of lik h ' e series 0 Signs hut the whole cl&88 of sue

we regard quotation_r!a~kape With th~ ~ri~s given; only in this ,*:nse shall Whitehead ANd R names as individual names of expresBIODB- cr. interpretat;o~ oi't': te usa,ell. B. A. W. (90), vel, 1, pp. 661-6 and-for other

Ie rm sentence' K t binski T ( 12" 5

take this opportunit . --: 0 ar msar, . 37), pp. .,-.

'denote' (like the words r ~~ m~n~lOnmg that I use the words 'name' and distinct senses becaus I 0 [ect ', class " • relation') not in one, but in manY (i.e. to individ~8Is) ~ tPPly them both to objects in the narrower sense the standpoint of tho.;: tI a so to all kinds of classes and relations, etc. From Russell, B, A, W, (90) v~orr of types expounded in Whitohead, A. N., and lIB systematically amb'( " pp, 139-68) those expressions are to be regardod

19uOUS.

VITr, § 1 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 157 by the name, as well as the signs of which each single word is composed and the order in which these signs and words follow one another. Such names can be formulated without tho help of quotation marks. For this purpose we must have, in the language we are using (in this case colloquial language), individual names of some sort, but not quotation-mark names, for all letters and all other signs of which the words and expressions of the language are composed. For example we could use 'A', 'E', 'Ef', 'Jay', 'Pe' as names of the letters 'a', 'e', 'f', 'i'. 'p'. It is clear that we can correlate a structuraldescriptive name with every quotation-mark name, one which is free from quotation marks and possesses the same extension (i.e, denotes the same expression) and vice versa. For example, corresponding to the name "'snow'" we have the name 'a word which consists of the four letters: Es, En, 0, Double-U (in that order)'. It is thus evident that we can construct partial definitions of the type (2) for structural-descriptive names of sentences. This is illustrated by the following example:

(4) an expression consisting of 'three words, of which the first is composed of the two letters I and Te (in that order), t~ second of the two letters I and Es (in that order), and the third of the seven leiters Es, En, 0, Double-U, I, En, a*! Ge (in that order), is a true sentence if and only if it is snOWtng.

Sentences which are analogous to (3) and (4) seem to be clear

. . h th n,n<7 of the word

and completely III accordance wit e mea~

'true' which was expressed in the formulation (1). In .regard

. . d th ctness of their form

to the clarity of therr content an e oorre

( . g of course that no

they arouse in general, no doubt assunun h

such doubt: concern the sentences which we substitute for t e

symbol 'p' in (2». su

But a certain reservation is nonetheless necessary here.. I ua-

tions are known in which assertions of just this type, III ~mbination with certain other not less intuitively clear p~emLSSes, lead to obvious contradictions, for example the an:!1I.umfYh~f

1 . le formulatIOn 0 t JII

the liar. We shall give an extreme Y SImp

antinomy which is due to J. l.ukasiewicz.

158 CONCEPT OF TRUTH IN

For the sake of moo t FO~~IZED LANGUAGES Vill,§l

, I grea er persplCUItv h 11

c as a typographical abb '. J we s a use the symbol

t· reVIatlOn of th .

ence prmted em this' e expression 'the sen-

th £ II . page, line 5 from th top' .

e 0 OWIng sentence: e . Consider now

c is not a true sentence

Having regard to th . '

establish empirically: e meanmg of the symbol 'c', we can

(0:) , .

c ~s not a true sentence I' id '

F ts t entteal with c.

or the quotation-mark name

other of its name) of the sentence c (or for any

s We set up an 1 .

(ft) 'c . exp anatlOn of type (2):

ss not a true sentence' .

c is not a true sentence. t8 a true sentence if and only if

The premisses (0:) and ('R

diction: ,..) together at once give a contra-

c is a true sentence if and onl' .

The Y if c '1.8 not a true sentence

source of this contradic' '. .

to construct the assertio tion IS easily revealed: in order

symbol 's' in the soh n (ft) We have substituted for the the term 'true sente::~ <.!!h an expression which itself contains contrast to (3) or (4)- ( ence the assertion so obtained-in of truth). Neverth I can no longer serve as a partial definition she ess no rational

uo substitutions should be f, . gr~und can be given why

I shaU restrict my If h orbldden m principle.

at' se ere to th f

n momy and w·u e ormulation of the above

q I POstpone dr .

uences of this fact till 1 aWI.ng the necessary conse-

shalJ next try t ater. I~avmg this difficulty aside I

o construct a d fi iti

generalizing expla t' e IU Ion of true sentence by

no. IOns of t (3)

may seem quite e . ype . At first sight this task

e te aSY-especIally fj

x nt mastered the te hni or anyone who has to some

It might be thought t1 c t Jue of modern mathematical logic. any sentential variable aC a We need do is to substitute in (3) can be sUbstitUi-~..1)' ll.e. a symbol for which any sentence

hi h ""U In p ace of th

w c OCCurs there t . e expression 'it is snowing'

formula holds for ev:e, :d then to assert that the resulting v ue of the variable; we would thus

vnr, § 1 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 159 reach at once a sentence which comprehends all sentences of type (3) as special cases:

.(5) for all p, 'p' is a true sentence if and only if p.

But the above sentence could not serve as a general definition of the expression '» is a true sentence' because the totality of possible substitutions for the symbol 'x' is here restricted to quotation-mark names. In order to remove this restriction we must have recourse to the well-known fact that to every true sentence (and generally speaking to every sentence) there corresponds a quotation-mark name which denotes just that sentence.1 With this fact in mind we could try to generalize the formulation (5), for example, in the following way:

(6) for all z, X is a true sentence if and only if, for a certain p, x is identical with 'p', and p,

At first sight we should perhaps be inclined to regard (6) as a COrrect semantical definition of 'true sentence', which realizes in a precise way the intention of the formulation (I) and therefore to accept it as a satisfactory solution of our problem. Nevertheless the matter is not quite so simple. As soon as we begin to analyse the significance of the quotation-mark names which OCcur in (5) and (6) we encounter a series of difficulties and dangers.

Quotation-mark names may be treated like single words of a language, and thus like syntactically simple expressions. The single constituents of these names-the quotation marks and the expressions standing between them-fulfil the same function as the letters and complexes of successive letters in single Words. Hence they can possess no independent meaning. Every quotation-mark name is then a constant individual name of a definite expression (the expression enclosed by the quotation marks) and in fact a name of the same nature as the proper name of a man. For example, the name" 'p' " denotes one

1 For example, this fact could be formulated in the following way:

(5') for all X; if x is a true ,entence, then,for a certain p. xu idenliaJl with '1"; from the premisses (5) and (6') the sentenoe (6) given below can be derived aft a conclusion.

160 CONCEPT OF TRUTH IN FORMALrz

ED LANGUAGES vm § 1

of the letters of the al habet W' " '

seems to be the most nP t I' Ith this Interpretation, which with the customa :a ura one, and completely in accordance definitions of the;- (;) of USIng quotation marks, partial generalizations In ype cannot be used for any significant

accepted as 8u~h a g:~er:e ;,an the sentences (5) or (6) be the rule of sUbstitution to za ion. In a?pl~g the rule called anything at all for the let~;r w.e,are ~ot Justified in substituting of a quotation-mark ~ which occurs as a component to substitute anything n:me ~Just as we are not permitted Consequently we obt' or t e letter 't' in the word 'true'),

am as conclusion t (5) b

sentence: 'p' is at, no ut the following

rue sentence if and l ;1" ,

see at once from this that on y ~j ~t M snowing. We

formulations of the thou ht the s~ntences (5) and (6) are not are in fact obviousl gl we WIsh to express and that they

y sense ess Mor th

at once to a contrad' ti ' eover, e sentence (5) leads

, IC Ion for we ca btai f "

easily, in addition t th' ,n 0 am rom It Just as

di 0 e above given c

ctory consequence' , , , onsequence, the contra-

not snowing S te ' p 'IS a true sentence if and only i I' it is

, en nee (6) al I ad 'j

the obviously senseless 1 o~e e s to no contradiction, but

'p'is the only tru teno- USlon follows from it that the letter e sen ence

To give greater clarity t~ th

be pointed out that ith e above considerations it may

WI Our cone ti f

names they can be I' , ep Ion 0 quotation-mark

e lllllnated and I d

example the corr di rep ace everywhere by, for

, espon mg st t I

nevertheless we 0 id ruc ura -descriptive names, If,

, onsi er expl ti

by the use of such nam ana IOns of type (2) constructed

then we see noway f es (asw_a~done, for example, in (4) above),

(~) 0 generalIzmg tJ I' , ,

,) or (6) we replace th . 10He exp unations. And if in

descriptive name 'P , e(. quotation-mark nume hy the structural-

l e or 'the word I'}

etter Pe'} we see at W HC I consists of the single

tion, once the absurdity of the resulting formula-

In order to rescue th

must seek quite a diff e se~ of sentences (5) and (6) we

k erent Interp t t'

mar names We re a Ion of the quotation-

, . must treat the

pOSlte expressions of hi h se names as syntactically COID-

expressions withi~ t: lC both the quotation marks and the em are parts, Not all quotation-mark

VIII, ~ 1 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 161 expressions will be constant names in that case. The expression " 'p' " occurring in (5) and (6), for example, must be regarded as a function, the argument of which is a sentential variable and the values of which are constant quotation-mark names of sentences. We shall call such functions quotatirm-/undirmll. The quotation marks then become independent words belonging to the domain of semantics, approximating in their meaning to the word 'name', and from the syntactical point of view they play the part of functors.! But then new complications arise. The sense of the quotation-function and of the quotation marks themselves is not sufficiently clear. In any case such functors are not extensional; there is no doubt that the sentence "for all p and q, in. case (p if and only if q), then 'p' is identical with 'q' " is in palpable contradiction to the customary way of using quotation marks. For this reason alone definition (6) would be unacceptable to anyone who wishes consistently to avoid intensional functors and is even of the opinion that a deeper analysis shows it to be impossible to give any precise meaning to such functors." Moreover, the use of the quotation functor exposes us to the danger of becoming involved in various semantical antinomies, such as the antinomy of the liar. This will be so even if-taking every care-we make use only of those properties of quotation-functions which seem almost evident. In contrast to that conception of the antimony of the liar which has been given above, we can formulate it without using the expression 'true sentence' at all, by introducing the

1 We call such words as 'reads' in the expression 'x reads' functors (this is 8 sentence-forming functor with one individual name as argument); also 'sees' in the expres. s ion 'x sees y' (8 sentence-fonning functor with two name arguments), and' father' in the expression' the father of z" (8 name-forming functor with one name argument), as well as 'or' in the expression 'p or q' (a sentenceforming functor with two sentence arguments); quotatlon marks provide an example of a name-forming functor with one sentence argument. The term 'functor' we owe to T. Kotarbiriski, the terms 'sentence.fonning Iunctor ' and

'name-forming functor' to K. Ajdukiewicz; cf, Ajduki~icz., K (3). .

• I shall not discuss the difficult problem of extensIOnality ill more detail here; cf. Carnap, R, (8) where the literature of the problem i~ given, and especially Whitehead, A. N., and Russell, B, A, W. (90), vol. I, pp, 65~-66. It should be noted that usually the terms 'extoWlional' and "intension ... l' are applied to sentence-forming functors, whilst in the toxt they am ... ppliod to quotation marka and thus to name-Iorming functors,

162 CONCEPT OF TRUTH ~

. ~ FOR:\L.-\LIZED LAXGUAGES vrn, § 1

quotatIon-functions with "

sketch of this formulati~n~anable arguments. "We shall give a

Let the symbol 'c' be at.

expression 'the sentenc . ypograprucal abbreviation of the

W e pnnted on th' Z·

e consider the f II . ~s page, me 6 from the top'.

o OW1ng statement:

for all p, if c is identi 1 . h

(if we ca unt the sentence 's', then not p

accept (6) as a definit" f

lllent asserts that. IOn 0 truth, then the above state-

c IS not a true te

We est bli h . . sen nce).

a s empIrIcally:

(0:) the sentence 'for all . ."

'p', then not p" 'deP" if c M tdentical with the sentence

M t nttcal with c

~n addition We make onl a sin .

which COncerns th y. gle Supplementary assumption doubts: e quotatIOn-function and seems to raise no

(P) for all p and q if th

sentence 'q' tM' :~ e sentence 'p' is identical with the

B ' np tJ and only ifq.

y means of elementa 1 .

diction from the . ry oglCallaws we easily derive a contra-

I sho . preIDlsses (0:) and (fJ).

uld like to draw attenti .

to which the consistent f on, ill passing, to other dangers

tion marks expo USe 0 the above interpretation of quota-

expressions (for e;:s u;, ~mely to the ambiguity of certain in (5) and (6) m t mb p e, t e quotation-expression which occurs

t". us e regarded it· . .

ron WIth variabl n oer am situations as a func-

e argument wh .

name which denot I ' ereas III others it is a constant

. ea a etter of tho I h b

pOint out the necessit f d . . a p a et). Further, I would

tions Whoso agleom ; 0 . a mlttlllg oortaln linguistio construeat least doubtful en WIth ~ho fundarnontallaw8 of syntax is

m' , e.g. meanmgful

eanmgless express' oxpressions which contain

IOns as syntaot' I

name of a meanin I rca parts (every quotation-

F all g ess expression will

or these reasons th serve as an example).

the new interpret t' e correctness of definition (6) even with

a Ion of t· '

tremely doubtful. quo atlOn marks, seems to be ex-

Our dis .

CUSBIOns 80 far entitle .

attempt to construct a US In any ease to say that the

'true correct Stmantical d,/: " .

sentence' meets with v " eJ.nztum of the ezpressum

ery real difficulties. We know of no

VIII, § 1 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 163 general method which would permit us to define the meaning of an arbitrary concrete expression of the type 'x is a true sentence', where in the place of 'x' we have a name of some sentence. The method illustrated by the examples (3) and (4) fails us in those situations in which we cannot indicate for a. given name of a sentence, the sentence denoted by this name (as an example of such a name 'the first sentence which will be printed in the year 2000' will serve). But if in such a case we seek refuge in the construction used in the formulation of definition (6), then we should lay ourselves open to all the complications which have been described above.

In the face of these facts we are driven to seek other methods of solving our problem. I will draw attention here to only one such attempt, namely the attempt to construct a structural definition. The general scheme of this definition would be somewhat as follows: a true sentence is a sentence which possesses such and such structural properties (i.e. properties concerning the form and order of succession of the individual parts of the expression) or which can be obtained fram such and such structurally described expressions by means of such and such structural transformations. As a starting-point we can press into service many laws from formal logic which enable us to infer the truth or falsehood of sentences from certain of their structural properties; or from the truth or falsehood of certain sentences to infer analogous properties of other sentences which can be obtained from the former by means of various structural transformations. Here are some trivial examples of such laws: et'ery expression consisting of four parts of ulhich the first is the word 'if', the third is the u1QT'd 'then', and the second and fourth are the same sentence, is a true sentence; if a true sentence cansiBts of f(YUr parts, of which the first is the word 'if', the second a true sentence, the third the word 'then', then the f(YUrth part is a true sentence. Such laws (especially those of the second type) are very importa.nt. With their help every fragmentary definition of truth, the extension of which embraces an arbitrary class of sentences, can be extended to all composite sentences which can be built up from sentences of the given class by combining them by means

]64 CONCEPT OF TRUTH

IN FORMALIZED LAXGUAGES VITI § I

of such expressions as 'if then' 'if d '

, t' . . . . ., an only if' , , , d'

no ,In short, by means of ex' . ,or, an ,

tential calculus (or theo pres~lOns belongmg to the senof setting up suffic' tl ry of deductIOn). This leads to the idea

len y numerous P rful

for every sentence to fall ' Owe , and general laws

we should reach under one of them. In this way

sentence. Yet this aw!eneral structural definition of a true least as far as naturail:!so see~s to be almost hopeless, at guage is not somethin fi ~age IS concerned. For this lanlimits. It is not laid :ow:~ed, closed, or bounded by clear language and th' hat words can be added to this

us ill a certain

Potentially W sense already belong to it

. e are not able to if

expressions of the la a e . SpeCI Y structurally those

can We distinguish ngu g which we call sentences, still less up a structural de-t:arr:t~ng them the true ones. The attempt to set

J<nl wn of the term 't

to colloquial la"1fl' rue sentence' -applicable

.. ~ uage 1,8 confronted with .

The breakdown of ll' lm1lperable difficulties.

that there is no satis~ct~revlOus attemp~ leads lIS to suppose portant arguments f ry way of SolVIng our problem. Im-

• 0 a general natur . s:-

ID support of till . . e can ill fact be invoked

s sUPposItIon a I h II

A characteristI'c c. t s s a now briefly indicate.

lea ure of coll . 11

various scientific Ian a '. oqu~a anguage (in contrast to

in harmony with th gu g_e~) IS Its uruversaIity. It would not be language a word oc e SPdInt ?f this language if in some other

· t curre whICh Could t b

I Could be claimed that 'if no e translated into itj

anything at all we 1 We can speak meaningfully about

, can a Bospeak b t i

If we are to maint· thi . a ou It m colloquiallanguage'.

am 118 urnv rt f

conncxion with Be t' . ersa I yo everyday language in

· man ical tnVUKt' t"

Sistent, admit into th I . Iga ions, wo rnuKt. to be con-

other express' 1° nngungtl, III addition to its sentences and

IOns, a 110 tho nil. f

pressions and sent mes 0 thoso sontellCOIl and ex-

, ences containi th

semantic expressions ' ng ese names, as well as such

B t i . as true sentence" , 'd ' t

u It IS jJresumabl' '. ,name, enote, e c.

which is the prim Y Just thIS unIVersality of everyday language

. ary source of all .

antmomies of th Ii semantwal antinomies, like the

. e ar or of het 1 . .

nOlllles seem t· ero oglCal words. These anti-

· 0 prOVIde a Proof th .

llIUversal in th b at every language which IS

e a ove sense, and for which the normal laws of

VIII, § 1 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 165 logic hold. must be inconsistent. This applies especially to the formulation of the antinomy of the liar which I have given on pagcs 157 and 158, and which contains no quotation-function with variable argument. If we analyse this antinomy in tho above formulation we reach the conviction that no consistent language can exist for which the usual laws of logic hold and which at the same time satisfies the following oonditions: (I) for any sentence which occurs in the language a definite name of this sentence also belongs to the language; (II) every expression formed from (2) by replacing the symbol 'p' by any sentence of the language and the symbol 'x' by a name of this sentence is to be regarded as a true sentence of this language; (III) in the language in question an empirically establiahed premiss having the same meaning as (ex) can be formulated and accepted as a true sentence.!

If these observations are correct, then the t'ery possibility of a consistent use of the expression 'true sentence' which is in harmony with the laws of logic and the spirit of everyday language seems to be very questionable, and consequently the same doubt attaches to the possibility of constructing a correct definition of this expression.

§ 2. FORMALIZED LANGUAGES, ESPECIALLY THE LANGUAGE OF

THE CALCULUS OF CLASSES

For the reasons given in the preceding section I now abandon the attempt to solve our problem for the language of everyday life and restrict myself henceforth entirely to formalized Zanguago;.2 These can be roughly characterized as artificially con-

I The antinomy of hetorologica l words {which 1 shall not describe here-er, GreHing, K., ~d Nelson, L. (24). p. 307) is simpler than th~ antinomy of the liar in so far as no empirical premiss analogous to (.:x) appears ill Its formulation; thus it leads to the correspondingly stronger consequence: there can be no consistent language which contains the ordinary laws of ~ogic and satisfies two conditions which are analogous to (I) and (n). but differ from them in that they treat not of sentences but of names, and not of the truth of sentences but of the relation of denoting. In this connexion compare the discussion in § 5 of the present articlo--the beginning of the proof of Th. I. and

in particular p. 248. footnote 2. . ..

2 The results obtained for formalized language also have a certain validity for colloquial language. and this iH owing to its universalit~, if we tra1l.91ste into colloquial lan~a~o any definition of a tru~ Mntencfl which hM been constructed for Rome formalized lang1la!l:", we obtain B rrB~'mtary "~finlt"m or truth which embracee a wider or narrower cat-egory of 8en«mc"".

166 CmWEPT OF TRUTH IN F

st ORMALIzED LANGUAGES V11I §2

ructed languages in 'hi h '

, , '" c the sense f

IS uruquely determined b ' t f 0 every expression

completely ~~1.au ti y 1 S orm, Without attempting a

<dU1 s rve and prec' d "

matter of considerable d.iffi lise escnpbon, which is a

to some essential propert' z: I shall draw attention here possess: (11:) for each f Itehs w ich all the formalized languages

" . 0 ese langu Ii

IS gIven in structural te ages a st or description

" rms of all th' . .

pre8mans of the la e 8~gns wtth whu:h the ex-

. nguage are fOrmed' (fJ)

expressIOns which can b r ,among all possible

t e ~ormed with th "

sen ence8 are distill . h ese SignS those called

t' guis ed by means f I

per les. Now formali d I 0 pure Y structural pro-

stMleted exclusiv I !e anguages have hitherto been con-

. e Y lor the p

8c~ence8 formaliz d th . urpose of studying deductive

de on e basiS of su h 1

an the science grow together . c anguages. The language

of the language of "to a smgle whole, so that we speak

. a partIcular fo li d

lDstead of this or th t e. • rma ze deductive science,

f rth a ~ormalized IF'

u er characteristic . anguage. or this reason

" properties of fo liz d

in connexion with the wa' . rma e languages appear

up; (I') a list, Or structu r:; w~c~ deductive ecienees are built called axiom8 or pr' 't ~a escnptlOn, is given of the sentences

les ot ; ~ml, l,ve statement . (8) . .

ru €8 of l.njerence, certain 0 "8, III special rules, called

bodied which penn' t th peratlOns of a structural kind are em-

te 1 e transformat' f

sen nces; the senten hi Ion 0 sentences into other

t b eea w ich can be bt .

ences y one or m ." 0 ained from given sen-

ore applIcatIOn f th

consequences of the gi . a 0 ese operations are called

Iven sontenc I

quences of the axiom es. n particular the conse-

I t s are called provable

. • remains perhaps to add or a8,~crtcd sentences»

In formal' languages d . thnt wo are not interested here Word 'formal' nn I an. RClonce!'! in one special sense of the

hi h ,mo Y SClencos to tt .

w IC no meaning is tt h' 10 aigns and expressions of

Iem here discussed h a ac ed, For such sciences the prob-

1 as no relevan 't'

,The formalization f. ce, I IS not even meaningful.

ducmg new s' . 0 a SCIence USUaU adm' -

Bet. These ~ into that science whO h y Its of the possibility of introapPear in t~I~-cal~ed defined ai~cr Were not explicitly given at the outcalled de'" ,~Sclence ill the first Ins .__ tn. contrast to the primitive signs)-

~.nuW1l8 whi h ...... ce m eltp .

the rules Of d ,.,' ., c are constructed ' rtlSBJOns of a special structure

tences of th ,?mtwn. Definitions D ill accordance with special rulee-«

e SCIence This f are sometimes d d ted

considered' th' eature of th f " regar e as Maar sen-

m e sequel. e ormaIizatton of languages will not be

VIII, § 2 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 167 We shall always ascribe quite concrete and, for us, intelligible meanings to the signs which occur in the languages we shall ccnsider.! The expressions which we call sentencea still remain sentences after the signs which occur in them have been translated into colloquial language. The sentences which nrc distinguished as axioms seem to us to be materially true, and in choosing rules of inference we arc always guidcd by the principle that when such rules are applied to true sentences the sentences obtained by their use should also be true.2

In contrast to natural languages, the formalized languages do not have the universality which was discussed at the end of the preceding section. In particular, most of these languages possess no terms belonging to the theory of language, i.e. no expressions which denote signs and expressions of the same or another language or which describe the structural connexions between them (such expressions I call-for lack of a better term-structural-descriptive). For this reason, when we investigate the language of a formalized deductive science, we must always distinguish clearly between the language about which we speak and the language in which we speak, as well as between the science which is the object of our investigation and the science in which the investigation is carried out. The names of the expressions of the first language, and of the relations between them, belong to the second language, called the metalanguage (which may contain the first as a part). The description of these expressions, the definition of the complicated concepts, especially of those connected with the construction of a deductive theory (like the concept of consequence, of provable sentence, possibly of true sentence), the determination of the properties of these concepts, is the task of the second theory which we shall call the metatheory.

For an extensive group of formalized languages it is possible 1 Strictly speaking this applies only to the signs called constants, Variables and technical signs (such as brackets, dots, etc.) possess no independent meaning; but they exert an essential influence on the meaning of the expresaions of which they fonn parts.

• Finally, the definitions are FlO oonatructed that they elucidate or determine the meaning of the Bign" whi .. h are int.rodueod into the IlUIgtll\g6 hy me .. nx e,f primitive Bigns or lIign~ previously defined [cf. P: 166, note I).

168 CONCEPT OF TRUTH IX

, - FOR~L-\LIZED LAXGFAGES \

to grvs a method by which 11r, § 2

a correct d fini '

constructed for each of th e tion of truth can be

of this method and f th em I ' The general abstract description

ld 0 e anguages to hi h i ,

wou be troublesom d w c It IS applicable

th c e an not at all '

erewre to introduce th perspIcUOUS, I prefer

e reader to thi th d '

way, I shall construct a d fini ' s me 0 m another

ith e tlon of this kind '

~ a particular concrete Ian in eonnexion

I~portant consequences, Th;.ag~ an~ show ~ome of its most grve in § 4 of this art' I m mdicatlons which I shall then

th IC e WI I hope b ffiei

e method illustrated b thi ,e su cient to show how

languages of similar log' y I s example can be applied to other

I rea constr t'

choose, as the object of m uc .lOn. ,

a deductive science f th y con~lderatlOns, the language of be well known t th: e utmost SImplicity which will surely

Th 0 e reader-that f th

e calculus of classes . fr 0 e calculu.s of classes.

can be regarded as Isfa ~gment of mathematical logic and

hi one 0 the rnt tati

w ich is commonl II d erpre atlOns of a formal science

A Y ca e Boolean al b

mong the signs co . , ge ra or the algebra of logic,l

I di ti mpnslllg the ex '

s mguish two ki d preSSlOns of this language

ole n s, cvn.stant8 and ' bl

n y LOUr constants' th " vana es,2 I introduce

(d'" . e negatton SIgn 'J.l' he si ,

tSJunctwn) 'A' th' ,t e SIgn of logtcal sum

. I . ,e untversal t ;J; , ,

tnc URton sign 'l' 3 I quan tJtCr Il ', and finally the

, regard thes' . . .

1 Cf. Schroder E (6 e signs as bemg equivalent III

and Ru II B ' . 2), vol. 1 (eapec' II

, II sse '. . A. W. (90), vol 1 12e. y pp. 160-3) and Whit.ehead, A. N .•

. Y mnkmg Use of a . I " pp, 05-12.

81gJlij (Iiko hr k n Ie en of },uk!l.8ioWicz I ". .

the fnet th' ac c~, dots, etc.j into th I - aVOid mtroducmg any teehnicsl

mcaningf 1 ut, I always write> tho funo t anguugo, and t.his is due chiofly to

~ U u ex prOHHion; ef I ukusl . c -or beforo t.ho arguments in every

Bunlly m l ., I\[OWWZ ,J ('>1) -

exj"terw . 'ny othor ~onHtt\nt" ,..., H"p''''U,lly pp. v un,1 40.

equi r·e "Ign, t)", Hign of im 1< to~~ur In th" ,·,,In'lh'M of ,,In''H''" o.g, tho

uivn Oll{·., of . [. p I<-n 'lnll of I . I'

pro,hl,.t of ' I. "lltlty. 11M woll .~ (f: t ('1<1"11 produnt (oonjunorion}, of

ealeu!.;" ) r<"l'I"''''''' (~{l" 1'. 16H, nut" I). 'j' I"., '·lHllpl"."."."t, thu Hum, nwl tho

o (~U~H('H C f ' ur t ULt r I h

under co - J _ all- __ Ormally ~r", k' (11 .... on "II Y u frngmont of t, 0

lU<l< oratIOn I - r,a mg-bo e t 1 .

calculus f .' tIS, howover t buns nwtO( In tho language

if we co';:' ~elass?s could be introduc:X1 uin e nO~ed the.t all constant .. of the possible b p te Its fonnalization by m k .to this l~gue.ge as defined terms, fragmente.Y n:eana of definitiollB (see p ~6mg the mtroduction of new signs can be fo ry I e.n~ge already suffices f, 6. note I), Owing to this fact our out that .::u ated ~ the complete Ian ~r the expre~ion of every idea which interpretin e: the Sl~ of incluaion • I' ~ng~ of ~~ SCience. I would also point of 'x' and ff ;x~resslOns oftlle type 'xy' wh e elllIllnated from our language by expression'} ) :n the same way in Whic~ . ere any va.riables occur in the place

xy . m the sequel we shall interpret the

VIII, § 2 CONCEPT OF TRUTH IN FORMALIZED LANGL' AGES 109 meaning respectively with the expressions 'not', 'or', 'for all ' (in the sense in which this expression was used in statement (tl) of § I, for example) and 'ill included in'. In principle IlllY arbitrary symbols could be used as varinblcs, provided only that their number is not limited and that they are dist.inet in form from the constants. But for the further course of uur work it is technically important to specify the form of t.hcso signs exactly, and in such a way that they can easily be ordered in a sequence. I shall therefore use as variables only such symbols as 'x,', 'x/, 'XII,', and analogous signs which consist of the symbol 'x' and a number of small strokes added below. The sign which has k such small strokes (k being any natural number distinct from 0) will be called the k-th oariable. In the intuitive interpretation of the language, which I always have in mind here, the variables represent names of classes of individuals. As expressions of the Janguage we have either single constants and variables or complexes of such signs following one another, for example: 'x.Nx;', 'Nlx,xw', 'Alx,x.lx.x,', 'fix, " Tlz, 1 XH x'" " 'Tx; x",' and so 00, Expressions of the type 'Np", 'Apq', 'ITxp', and 'Ixy', where in the place of 'p' and 'g' any sentences or sentential functions (this term will be explained below), and in the place of 'x' and 'y' any variables, appear, are read: 'not p' (or 'it is not true that p'),l 'p or q', 'for all classes x we havep', and 'the class x is included in the classy', respectively, Regarding composite expressions, i.e. those which are not signs, we can say that they consist of two or more other, simpler expressions. Thus the expression 'NIx, x,' is composed of the two successive expressions 'N' and 'Lx.x;' or of the expressions '.LVI' and 'x.a;' or finally of the expressions 'NIx,' and 'x,',

But the proper domain of the following considerations is not the language of the calculus of classes itself but the corresponding metalanguage. Our investigations belong to the metacalculus of classes developed in this metalanguage. From this springs the need to give the reader some account-if only a very brief

1 For stylistic reasons we somctimes use the expression 'it j .. not true that' instead of the word 'not', tho wholo expression being regarded "" .. Hingl" word, no independent meaning being given to the lICparato parte, and in particular to tho word 'truo', w hich occur in it.

170 CONCEPT OF TRUTH IN

. . FORM.4cLIZED LANGUAGES VIn § 2

one-of the structure of the metala . '

I shall restrict myself to the tw ngua!5e and of the metatheory. enumeration of all th' 0 most Important points: (1) the in the metalan e Sl~ and expressions which will be used importance in t:::ge, WIthout .explaining in more detail their up of a system of o.urse ofhit~e rnvestigation, and (2) the setting

axioms w ch suffice f th b

the metatheory or at least . s or ~ esta lishment of

obtained in this a t' I Thwill form a foundatIOn for the results

r IC e. ese tw . t

with our fundam t I 0 porn s are closely connected

en a problem'

should not be abl to ' were we to neglect them, we

correctly definin e assert either that we had succeeded in or that the defi ~t~ny concept on the basis of the metalanguage,

III Ion constructed

sequences. But I h II possesses any particular con-

h s a not attempt at II t . h

t e character of a stri tl f a 0 give t e metatheory

content myself wi~hnc ~ ormalized deductive science. I shall mentioned_th sayrng that--apart from the two points

e process of formali' th

specific pecuIiarit I . zrng e metatheory shows no

d· y. n particular th If'

efirution do not cliff ' e ru es 0 inference and of

er at all from th I d . .

other formalized ddt' . e ru es use m constructmg

A e uc ,lve SCIences.

mong the expressions of th

guish two kinds T th fi e metalanguage we can distin-

• 0 e rst belon .

cal character dr f g expresszcms of a general logi-

, awn rom any ffici

of mathematical 10 . 1 Th su ciently developed system

giC. ey can be di 'd d . ., .

expressions and d fi d VI e mto primitive

I· e ne expression b t thi .

ess In the prese t . s, u IS would be point-

. n case. FIrst we h . .

which have the . ave a series of expressions

same meamng as th

We are considcrin . thus ' , t » e constants of the science

and 'is included ~',_.s not or ltufnottr1lctll4t',2 'or', 'for all',

"n m symbols 'r' 'I'} k ..

stance we are abl t t . -"'. 1IU1 S to this circum-

• 0 0 ransJu.to eve .

mto the metalang "Ii ry ~xpressIOn of tho language

uage . For ex I

(or for all classes) ,. amp e, the statement 'for all a

'IT a a C a IS the t l'

xlIx x' To th - rans ation of the expression

I I. e same categ b I .

ory e ongs a senes of analogous

1 For exampl f

(90} e, rom the work Whi

fro' t<:;ut I do not intend to USe h tehead. A. N., and Russell, B. A. W.

Of:lI e ~xceptjona which I shall eerel~y special logical symbolism. Apart

. oqulal language.} For th xP ,wltly mention I shall UBe expressions

gwen below O~ C e meanmg of th . .

• See ~ amap, R. (8). e general logical expressions

p. 169, note 1.

VIII, § 2 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 171 expressions from the domain of the sentential calculus, of the first order functional calculus and of the calculus of clasHeH, for example, 'if ... , then', 'and', 'if and only if', 'for somex' (or 'there is an x such that ... '), 'is not included in'-in aymbols 'ct', 'ill identical with'-in symbols '=', 'i8 distinct from ',---in symbols '=1=', 'is an element of'-in symbols 'E', 'i,., not an element of ',--in symbols 'E', 'individual', 'class', 'null class'; 'class of all x such. that', and so on. We also find here some expressions from the domain of the theory of the equivalence of classes, and of the arithmetic of cardinal numbers, e.g. 'finite class', 'infinite class", 'power of a class', 'cardinal number', 'natural number' (or 'finite cardinal number'), 'infinite cardinal number', '0', '1', '2', '<', '>', '~', '~', '+', '-', .... Finally I shall need some terms from the logic of relations. The class of all objects x, to which there corresponds at least one object y such that xRy (i.e. x stands in the relation R to y) will be called the domain of the binary or two-termed relation R. Analogously, the counter domain of the relation R is the set of all objects y for which there is at least one object x such that xRy. In the case of many-termed relations we do not speak of domain and counter domain, but of the 1st, 2nd, 3rd, ... , n-th domain of the relation. A relation having only one element x in its domain and only one element y in its counter domain (a relation which thus holds only between x and y and between no other two objects) is called an ordered pair, where x is the first and y the second member. Analogously using many-termed relations we define ordered triples, quadruples, and in general ordered n-tuples. If, for every object y belonging to the counter domain of a two-termed relation R, there is only one object x such thatxRy, then the relation Riscalledone-many. The concept of sequence will playa great part in the sequel. An infinite sequence is a one-many relation whose counter domain is the class of all natural numbers excluding zero. In the same way. the term 'finite sequence of n terms' denotes every one-many relation whose counter domain consists of all natural numbers k such that 1 ~ k ~ n (where n is any natural number distinct from 0). The unique x which satisfies the formula xRk (for a given sequence R and a given natural number k) is called tho

1 i2 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 2 k-th term of the sequence R, or the term of the sequence R with index k, and is denoted by 'Rk" "Ve say that the sequences R and S differ in at mast the k-th place, if any two corresponding terms of these sequences R/ andS/ are identical with the exception of the kth terms Rk and Sk which may be distinct. In the following pages we shall deal with sequences of classes and of natural numbers, i.e, with sequences all of whose terms are either classes of individuals or natural numbers. In particular, a sequence all of whose terms are classes which are included in a given class a, will be called a sequence of subclasses of the class a.

In contrast to the first kind of expression, those of the second kind are specific terms of the metalanguage oj a structural-descriptice character, and thus names of concrete signs or expressions of the language of the calculus of classes, Among these are, in the first place, the terms 'the negation sign', 'the sign oj logical.sum', 'the sign of the universal quantifier', 'the inclusion sign', 'the k-th variable', 'the expression which consists oj two euccessioe expressions x and y' and 'expression'. As abbreviations of the first six terms I shall use the symbols »«, 'sm', 'un', 'in', 'vk" and 'xr-.,y' (the sign 'v' thus denotes a sequence, the terms of which are the successive varia bles t'!, v2, v3" •• ). These terms have already been used in introducing the reader to the language of the calculus of classes. I hope that, thanks to the explanations already given, no doubt will remain conccrning the meaning of these terms. 'Vith the help of t.hese terms (and posRibly genflrnl logical termR) all ot.her concepts of the metalanguage of a Htructural-deHcriptive kind can he defined. It is caHY to Reo thnt every Kimplo or cnmpol'lil;n oxpression of tho language undor invt'HtigntioH haM an indi v iii I I!Li nume in the metalanguage HimiJar to tho stl'uctuml-uol:lcriptivo names of colloquial language (cf, pp. 156 and 157). For example, the symbolic expression '(ng~in)~vl)~V2' can serve as a name of the expression 'N I x, x:', The fact that the metalanguage contains both an individual name and a translation of every expression (and in particular of every sentence) of the language studied will playa decisive part in the construction of the definition of truth, as the reader will see in the next section.

EFT OF TRUTH IN FORMALIZED LANGUAGES 173 VIII, § 2 CONC

. bl . the metalanguage I shall use the symbols

As vafla es In , , , , (4)'l' • , ow'

' , 'b' (2) '1' 'g' 'h'; (3) 'k', 'l', '111:, n, p ; , u, •

(1) a,. , ) :X' 'Y' In this order they represent tho

' , , , 'Z'" and (5 , . 1

x, y" f i di iduals of an arbitrury churnoter,

ames of (1) classes 0 in IVI

n f 1.1 co (:3) natural 1I11miwI'H und HPljllPIlt:I'H

(2) sequences 0 sue I c ass "', , ,

I b· (4) expressions anti i'icqul'lIC'PS of cxpreSSJOW', of natura Hum ers,

d (") lasses of expressions. I"

an .J C .. . . . . '1 f t he met alall~uall;e. .' Irsi ,

ncr t n now to the aXlom sys em () ~ .

"e ur. , ding to the two kinds of ex-

. . t b oticed that+-correspon

It ISO en this stem contains two quite pressions in the metalanguage- sy 1 logical axioms which

distinct kinds of sen. tences: thhe g~nera zstem of mathematical suffice for a sufficiently compre ensive sy" hi h describe

:-F; 'oms of the metalanguage w ic

logic, and the epeequ: UXl. fthe above structural-descriptive certain elementary properties o. it.i It is unnecessary to . te t with our intui tons,

concepts eODSlS n II kn xioms of the first kind."

licitl the we - own a

introduce exp ci Y . d t the following statements:" As axioms of the second kind we a op ., h' h

sm un and in are expressions, no two oJ W tC

AXIOM 1. ng, , ,

are identical, .. __ T ;, k 's a natural number

. ''''''''esswn if and VIUY tJ t

AXIOM 2. Vk IS an €X.l'" 'n and also 'rom t'l

. d' t"nctfrom 119 81n, un, t , J'

distinct from 0; Vol: tS lSt ,

if k *' l; ~. . ession. i/ and only if x and yare ex-

AXIOM 3. z y is an expr . 'n and from each of the

preseions; x~y is distinct/rom ng, sm, un, l ,

expressions vk• ressions then we have

Ir, z and tare exp, .

AXIOM 4_ J x, y, . ' ., h fi Uou.;ing conditions is satls-

x~y = z~t i/ and only if one OJ ht e .0 an expression u such that

fled: (0:) x = z and y = t; (~)ht eT~ tS expression u such that x = z~u and t = u""y; (y) t ere fS an

"" and Y = u~t. faa

z = xu. . le of induction.) Lei X be a c 8

AxIOM 5. (The prmcip ondi . () ow E X 8m EX, un E X

,-F; h r. llowing c ltWns:::t'-:1 ,

which satisfie« t e J 0 I U58 dist.inct variables I here treat

1 Although in the cases {l{~~ "~~a!ll namely ss classea of concrete scrres

ex ressions as special classes 0 m 1~1 •

Of~rinted signs (cf: p. 156, nO~r~! Whitehead, A. S" and liW<ll<>l1, B. A. W.

, They may agam be taken ,

(90), cf. p. 156, not" I. . I • has< never 1".ffJ1'fl 1,,,"n ~vpn In tI", ff,nn

• AK far a .. , 1 know th" m.·tBt 1001')

of au axiornut.ized H-yl:ih~JI"

174 CONCEPT OF TRUTH

IN FORMALIZED LANGUAGES VITI § 2

and' X ' '

tn E ; (fJ) if k is a natur l b "

"k EX' ( ) f X anum er d~stmct from 0, then

, l' t X E and Y E X th ......

pressirrn belonq« to the class X, ' en x y E X, Then every ex-

The intuitive sense ofAxs 1--4 '

Ax, 5 gives a precise f ulat , requires no further elucidation,

consists of a finit ormb ah~n of the fact that every expression

. e num er of SIgnS. t

It is possible to pro th

This fact guarantees ~e at the .above axiom system is categorical.

sufficient basis fio th 0 a certam degree that it will provide a. r e constructio f th

Some of th b . no e metalanguage.1

e a ove axIOms h

character and involve f ave a pronounced existential

Noteworthy amon th urther consequences of the same kind. class of all ~ e~e ~onsequences is the assertion that the

expresslOns 18 fifinit

able). From the' t 't' e (to be more exact, denumer-

and hardly evid fit ill IV de standpoint this may seem doubtful

en ,an on that t

system may be s bt . accoun the whole axiom-

u ject to senou 't"

would restrict this iti . s en icism. A closer analysis

ti en aeism entIrely to Ax 2 d

ial sources of thi infini s, an 3 as the essen-

not pursue this diffi ul te character of the metatheory. I shall

1 I c t problem any further here.s The con-

use the term 'categorical' .

not propose to explain in md the sense given in Veblen 0 (86) I do

axiom RV te more etail why I . h ,. .

-'" s m an objective guarant see m t e categoricity of an

m.ent of the corresponding ded t' ee ~hat the system suffices for the establish-

will be DO d l uc IveSClence' a ' f .

• un in Fraenkel, A. (16). ' series o remarkBonthisquestion

e Fo~ example, the following trul s .

xpresslons are regarded as tl Y ubtle pomts are here raised. Normally

such products). From this st id p~oducts of human activity (or as classes of ?Jany oxprCflHions appears to a~ p~n~ the supposition that there are infinitely mterpretation of the term' 0 0 viously nOllilonsical. But another possible pbYRical ~odilJl! of a particul::~=ion' ~r_nts itlIDlf: we oould consider all ~robhlm 18 thon transforrod to t and 81.W lUI expl'CAfliow.. The kernel of the mfinity of the numher of ex ~o domain of phYlli(lfl. Tho RHIIOrtion of the may not conform to modorn Pp~lIII, onl8 III then no longer IM!n8Clellll although it

YSICS and cOIImological theories.

. t The axiom set formulated h

~n the Polish original of the ere was published for the first time in 1933,

In .the German Original of r:l'ellefi paper. In the same year it also appeared ;xlom set is usually referred t~a: th s:s. 282]. The theory based on this

~m a mathematical point of vie: ':"":" 9. oj 8trings or theory oj concatenation. (With ,a fixed, finite or infinite It 18 Simply the theory of free semigroups and bibliographic references ~onum~r of generators). For further information Corcoran-Frank-Maloney (14t)~cermng the axiomatization of this theory see

VIII, § 2 CONCEPT OF TRUTH IN FORMALIZED LA~GUAGES 175 sequences mentioned could of course be avoided if the axioms were freed to a sufficient degree from existential assumptions. But the fact must be taken into consideration thlLt the elimination or weakening of these axioms, which gua.rantce tho existence of all possible expressions, would considerably increase tho difficulties of constructing the metatheory, would render impossible a series of the most useful consequences and 80 in troduce much complication into the formulation of definitions and theorems. As we shall see later this will become clear even in the present investigations. For these reasons it seems desirable, at least provisionally, to base our work on the axiom system given above in its initial unweakened form.

Making use of the expressions and symbols of the metalanguage which have now been enumerated, I shall define those concepts which establish the calculus of classes as a formalized deductive science. These are the concepts of sentence, axiom (primitive sentence), consequence and provable sentence. But first I introduce a series of auxiliary symbols which will denote various simple types of expression and greatly facilitate the later constructions.

DEFINITION 1. x is the inclusion with Vk as first and VI as second term-in symbols x = 'kJ-if and only if x = (in ..... ttc) ..... vl·

DEFINITION 2. x is the negation of the expression y-in symbols x = y-if and ooly if x = nlJ~y·

DEFINITION 3. x is the logical sum (disjunction) of the expressions y and z-in symbols x =Y+Z-l! and only if x = (sm ..... y) ..... z,

DEFINITION 4. X is the logical Bum of the expressions h, t:, ... , t .. (or a logical sum of a finite n-termed sequence t of expreesioney-«

n

in symbols x = ~ tk-if and only if t is a finite n-termed

sequence of expressions which satisfies one of the foUQUJing con-

,.-1

ditians: (IX) n = 1 and x = tl, (fJ) n > 1 and x = ~ tk+t".l

1 As will be seen, Def. 4, is a recursive definition which, as such, ~ ('.ert.ain methodological misgivings. It is, however, well known that with the help of a general method, the id_ of which we owe to O. Frega and R. Dedekind, every reeursive definition can be tranKformll<i into an e<llJivalent nonnAl dnflnition

176 CONCEPT OF TRUT

D H IN FORl\IALIZED LAc'fGUAGES VIII § 2

EFINITION 5. X is the I . I . '

expressUms Y and . ogIea product (conjunction) of the

_ Z-m symbols x = if .

x = Y+i. y.Z-} and tmly if

DEFINITION 6 x is the .

Y under the var;abl ~lllversal quantification of the expreeeio« x = (un ...... I7.,Y ... y. e vi-m symbols x = Ok y-if and only if

DEFI~ITION 7 x is th .

. e umversal qua tif ti

y under the var-abl ~ n 1 ca Ion of the expression

• esv v .

and only if' ~l' 1'."'" Vpft-tn symbols x = nk.;;ny-il

• 'J P M a jin$te n-termed s Pl 'J

whtch satisfies one of th f II . equence of natural numbers

x = n1' Y (.Q) n > 1 e 0 oU'Ing conditions: (a) n = 1 and

1 , I" and x = nk"n-1 n

D ru h~

EFINITIO~ 8 .

! ~ . X M a universal qua tifi .

Y t and only if either x _ or t . n 1 cation of tM expression

of natural numbers such -tl.~ here 18 a finite n-termed sequence p fw-t x = nk';;ny

1'), •

DEFINITION 9 .

• • X ss the existent' I '.

sum y under the va '_1.[ .1a quantrficatIOn of the expres-

Tluu e vic -zn -'''''mb 1 U

x = nkY. ~y 0 s x = ky-if and only 1/

We have thus introduced t .

means of which co hree fundamental operations by

mpound expressi L"

ones: negation 10' I . . ons are ~ormed from simpler

(L . ,gl.Ca additIOn and .

oglCa! addition is f ' lUllversal quantification.

Ji . ,0 course the 0 ti . ..

ornung logical sums f' ' pera Ion which consists III

find 'univ~rsRl nuont.ofj gt.~en expressions. The terms 'negation'

'j ,. I calIon' a I

opcrations on cxpr' • re user to refer both to certain

t h . . ~I'IRlOnl< and to" .

. CAC oPcTlLt.ions.) If I ." . CxprCI'ISlOns resulting from

t} l ' J~P;llll11ng With t h . I .

Ie a JOve opcrnt.i ' ~ me U!'IIOIlS 'A: I' we perform

• OIlA any num] f'. '.

extensive class f '. ier () tImNI we arrive at an

f· 0 expressIOns 'h' h

unctwns. 'Ve obtai th \\ IC are called sentential

f h· am e concept f

otiS notion. 0 sentence as a special case

(cf. Dedekind R (I5

(90) vall" ), pp. 33-40 and Wbi he

far .:.s th i pp. 550:-'. and vol, i. p. 244 te ~,A. N., and Russell. B. A. W. are I Ie ormulatloIlS so obtained h ). This, however. is unpractical in so

Il$I! o ear as .feg!irda • ave a more r .

tiona. For th then- content, and com~ lCated logical structure.

sequel. ese rea.sona I do not prop are I~ suitabls for further derivsose to avoId recursive definitions in the

VIII, § 2 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 177

DEFINITION 10. X is a sentential function if and only if x is an expression which sati,'ifies one of the four following conditions: (0:) there exist natural numbers k and l such that x~c tk I; (f3) there exists a sentential function !J such that x =c, y; (y) t1U'r~ ('xi.~t sentential functions y and z such that x c.~ Y-I-z; (.'5) there erist« IL. natural number k and a sententialfunction y such that x c nk y.l

The following expressions will serve as examples of sentential functions according to Def. 10: 'Lr.x;'; 'lVIx,xm', 'AIx,xmlx",x,', 'IT x,NIx,x.', and so on. On tho other hand the expressions 'I', 'L»,', 'A Lx, x.; " 'IT Lx.x;', etc., arc not sentential functions. It is easily seen that for every sentential function in the language we can automatically construct a structural-descriptive name of this function in the metalanguage, by making use exclusively of symbols which were introduced in Defs. 1, 2, 3, and 6. For example, the following symbolic

1 Def. 10 is a recursive definition of a somewhat different type from that of Def. 4 since the UBuBI 'transition from n-l to n' is lacking in it. In order to reduce this to an ordinary inductive definition we must, first induct.ively define the expressions 'x ia a 8entential function of the nth degree' {inclusions 'k.! would then be functions of the Oth degree, the negations and 10gi" .. 1 Hums of these inclusions, as well as their universal quantification under any variable, functions of the 18t degree, and so on}, and then simply stipulate that 'x ia a sentential function' means the same as 'there is a natural number n Buch tlia: z is a aententt'al [unction. of the nth deoree', Def. 10 could also be transformed into an equivalent normal definition in the following way:

x is a sentential function if and only if the formula. x E X holde for eve'"'j clos« X which satisfies the following four condiiions : [u] if k and I are natural numbers distinct from 0, then 'k.! EX; (Il) if Y E X, then ii EX; (y) if Y E X and Z E X 1M.. Y + Z EX; (8) if k is a natural number distinct from 0 and y EX, then ntyE X.

It should be emphasized that recursive definitions of the type of Def, 10 are open to much more serious methodological objections than the usual inductive definitions, since in contrast to the latter, statements of this type do not always admit of a transformation into equivalent normal definitions (see p. 175, note I). The fact that such a transformation is possible in the present case is owing to the special nature of the concepts occurring in the definition (to the fact, namely, that every expression bas a finite length and also that the operations given in eonditions (P)-(6) always lead from shorter to longer expressions). If. nevertheless, I sometimes give definitions of this kind in the present article in the place of equivalent normal definitions (Defs, 10, 11, 14,22, and 24), I do so because these deflnit.ions have important advantages of quite another kind: they bring out the COD tent of the concept, defined more clearly than the normal definition does, anrl-in con traet to tho usual recursive definition-they require no previous int.roduction of "",,COHW.ry con"cpts which are not WMJ<i elsewhere (e.g. the aceOHHory concept of Ii .. mlcntial funr-t.ion of the nth deW'oo t-

------,-----

178 CONCEPT OF TRUTH IN FORM.ALIZED LANGUAGES vrn, § 2 expressions serve as names of the above examples of sentential

f ti "'-" + ' d 'n -, une Ions: t1,2, L1,3, L3,1 L3,1, an 1L1)2 ,

DEFINITION II. vk is a free variable of the sentential function x if and cmly if k ie a natural number distinct frCJm 0, and x is a sen· tential function which satisfies one of the follawing ffYUr conditions: (rx.) there is a natural number l sudi that x = 'k,lor X = 'I,"; (f3) there is a sentential functicm y such that vk is a free variable of y and x = y; (y) there are sentential functicms y and z such that Vk is a free variable ofy and x = y+z or x = z+y; (8) there is a numberl distinct from k and a sentential function y such that Vk is a free variable of y and x = nl y,

Variables which occur in a sentential function but are not free variables of this function, are usually called bound (apparent) variables ,I

DEFINITION 12, X is a sentence (or a meaningful sentence)in symbols x E S-if and cmly if x is a sentential function and no variable Vk is a free variable of the function x.

Thus the expressions: nI'l.I' nl nz '1,2' nl U2 '1,2' nl('l,l+nI UZ'2,l) are sentences, but the functions: 'I,V n:h.B' 'I,l+nl U2L2,l are not sentences because they contain the f~e~ variable v!' By virtue of the above definition the symbol S denotes the class of all meaningful sentences,

The system of primitive sentences of the calculus of classes will contain two kinds of sentences.s The sentences of the first kind are obtained by taking any axiom system which suffices as a basis for the sentential calculus and contains the signs of negation and logical addition as the only constants-for example, the axiom system consisting of the following four axioms:

'ANAppp', 'ANpApq', 'ANApqAqp',

and 'AN ANpqAN ArpArq' ,a

1 cr. Hilbert, D., and Ackermann, W. (30). pp. 52-54. .

t Concepts which I Ilhall discuss in the further course of § 2 do not oc~ur III the definition of true sentence itself. I shall. however, make use ofthero I~ ~be preparatory discussions at tbe beginning of § 3 which establish the definitl,:,e form of the definition. I shall also use them in the formulation of c~a~ consequences of tbis definition (The. 3-6 of § 3) which express characterIStIC and materially important properties of true sentences. . f

a This axiom system is the result of a modification and simplificatIon 0 tbe axiom system which is found in Whitehead, A. N., and Russell, B. A. W. (90), vol. I, pp. 96-97; cf. Hilbert. D., and Ackermann, W. (30), p. 22.

ALlZED LANGUAGES 179

F TRUTH IN FORM: , ' , , and

vrn, § 2 CONCEPT 0 tential variables p : q, nus

eplace the sen h expressions t

In these axioms w~ r unctions, and then to tea 1 the opera· 'r' by any sentential f t lready sentences, we pp ~er of times

, d'f they are no a fficient num 'U

obtame ,I antitication a su The following WI

tion of universal q~ S have disappeared.

until all the free vanable ,

serve as examples: n x Iz, x, n x, [x, x, ,

'ANA n x, [x,X, , x [x.x,', etc,

n x AN Ic. x. Alx, • d kind we shall

'n x,· f the secon t

tences 0 f the as ye

btain the sen ystem 0

In order to o. . t some axioJl1 s . the inclusion

take as out startlng-po~ classes which cont~ns translate the

unformalized calculU:e~ed sign,l and wef !h:npresent article. sign as the only un . t the language 0 hich are defined

hi ysteDl In 0 nstants w . to

axioms of t IS S liuUnate all co flllS belongmg

Naturally we mu~t fir::i:n sign, as well ~ a~ :lculuS which ~re by means of the mel d the fUllctlona 'ft the negatlOn

. 1 lculus an 1 uantl or, s

the sententla ca the universa q 1 s of sentence

distinct in meaning fro~ 1 ddition. As exaOlP e

. si n of 10gwa a ,

slgnandthe g. we have ANlx.x,"Ix,x",.

of this second kind n x n x", AN Ix, x. ) 'f and onLy

, d 'rr x,' .' entence t _.1

'II x, Ix, x, an , rn (priOlltlve s ( ) xES anu<

'sanaXIO_.litions: ct l

DEFINITION 13. z ~ joUQWi1l!J cm"'" is a universa

of the tWO nd such that Xl)

if X satisfies one. l unctiQ'My,z,a u. ---+i/+Y, y+(Y+z,

there exist sentent~a f j()'l,l.r funch(Y1l.S Y , . d ntical with

. if one of the . n:~) x 18 1 ,e

quantijicatwn 0 _ (-+y+(u+z», I.t'

nd -+z+ u

y+z+(z+y), a Y _ )

one of the jive sentences 01 02 n3~,+~a+L1,3 '

n1 '1.l' n (~+L!'4+t3.t))' n1n,Us(L1,3.L2,3' '/:_+~+'''3»)'

n·\L&,1 "

n1 nz Us(La,l' 'a.S'

_.1 - - +L..,)· )))

allAN _ - +L3,.).{'1,3+~3 l,).<'" 2·iI,i· ... '· .

n1U2(nan,,{(L3,1+L3,2 n.(U,l+. ~nHunti~' late!! which is gt;:;\n ptortiC\lW by

h ""stelIl of paetu \)eerI lIimph

h here t e ~J however. ture)·

I 1 have c oeen . stem l\a8. jst.ential n.

E. V. (3'2). p. ~ <:::ptionll of an eJl eliminating certaIn

180 CONCEPT OF TRUTH

IN FORMALIZED L.A.NGUAGES VIII, § 2

In the formulation of the d fini .

sequence I shall e tion of the concept of con.

'u is an expressir:mus~'ea ~mong others, the following expression: stitutingthe 'abl 0 moo from the sentential function w by sub.

van e v for th iable c:

this expression is Ike ~art e VI' The intuitive meaning of

tion has a s h ear and simpla, but in spite of this the definiomew at complicated form:

DEFINITION 14 x is .

juncit'on y b bs . an expressIOn obtained [rom. the senienUal

y su StItutlllg th (f ) .

able VI ij and only ij k and 1 e ree vanablc e, for the (free) vari-

and . are natural numbers distinctjrom 0 and x

yare sententtal funct . h . h '

cOnditiQns: (IX) x = L a:::;s ~ tc. satisfy one ~f the following six

ber m distinctjr l k.k Y - LI,I' (fJ) there exist« a natural num-

om ,such that x = d

Y = 'm,l; (oy) v is not a . . 'k.man Y = 'I,m{ or x = Im,kand (8) there exi t I • free varwble oj the function y, and x = y;

and z .s an s sente~hal functions z and t such that x = i Y = t

• CXprCSstOn obt . d ' ,

vdor the variable . ( ) he ame . from t by substituting the variable .... '~h tl.-t VI' e t re extst sentential 'unctions z t u andw u<ov IGq X = z+u _ t J', " ,

and w respectively b'~ !-w, where z and u are obtained from t

m there exist sente~tial ~~tuti~ the variable vk!or the variable t'l; distinctfromkandls h hat nciions z, t and a natural number m from t by substituti ~h t ~ = nm z, y = nm t, and z is obtained

ng e vanable Vkfor the variable v.1

. For example, it follows fr . . . I

SlOns , n ( om thIS defillItlOn that the expres-

1,1' 3 '31+(13) and, +n

functions:, h (' 1,3 2 '2,3 are obtained from the

SUbstituting2~' £0 3 '3,2B+'2,a) and '2,3+ n2 '2,3 respectively by

1 r Va' ut the e . n

tained in this wa fi xpresslon 1 '1 3 cannot be ob-

n y rom the funct' n' .

1 '1,1 from tho function n, ion 2 '2,3 1I0r the expression

1I 2,1'

I The following iN

recurs' a normal t1"flllit' } .

.lveone(cf.p.177,notel): Ion wHeh ill oquiVldt'ut to tho above

z t.8 an e.:cpruai b'

tXlriabl on 0 lamed from til .

f: e Vol: for the variable V .., nd e 8ententtal junction y by substituting the

rom. 0 and .. , th I OJ a only tif Ie a d l . .

.fi II . . tJ e jormula zR h ids n are natural numbers dtattnct

00 OWing 81.1: conditiO'1t8. (<Xl ~ 0 for every relation R which satisfies the a and I, th~n 'k.m R'I.m ~nd , 'k,~, '1.,.; (fl) ~f ~ is a natural number distinct from

free oorwble oj e, then zR~oI:. (a·l: (,,) ij z UI a sentential function and "I is not z+u R t+w; <n if m is a ~ ) ij zRt. then em, (d if zRt and uRw. then !.\en nmz R nmt. ural number distinct from 0, k, and l and zRI,

The definitions of SUbstitution . ,.

and (47), p. 20 (T.E, nvnO) dep ~ LllSlllewski. S, (46), p. 73 (T.E. xr.vn), en on a totally different idea.

VIII, § 2 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 181

Among the consequences of a given class of sentences we include first all the sentences belonging to this class, and all the sentences which can be obtained from these by applying, an arbitrary number of times, the four operations of Hubstituiion, detachment, and insertion and deletion. II/ OW uniurrsal. quantifier.l If we had wished to apply these operations 1101, only to sentences, but to arbitrary sentential functions, obtaining thereby sentential functions as results, thcn thc meaning of the operation of substitution would be complctely determined by Def. 14, the operation of detachment would correlate the function z with the functions y and fi+z, the operation of introduction of the universal quantifier would consist in forming the function y + nk z from the function y + z (provided that Vk is not a free variable of the function y), the operation of removal of the universal quantifier would proceed in the opposite direction-from the function y+~ to the function y+Z.l Here, however, we want to restrict ourselves exclusively to sentences (in the sense of Def. 12), and therefore we modify the above four operations by referring them, not to the sentent ial functions involved, but rather to the sentences that are universal quantifications of these functions.

In order to simplify the construction I first define the auxiliary concept of consequence of the n-th degree.

DEFISITlON 15. x is a consequence of the nth degree of the class X of sentences if and only if XES, X s; S, n is a natural number and either (a) n = 0 and x E X, or n > 0 and one of the following five conditions is satisfied: (fJ) x is a consequence of the n-lth degree of the class X; (I') there exist sentential functions u and w, a sentence y and natural numbers k and l euch. tlud x is the universal quantification of the function u, y is the universal quantification of the junction w, u is obtainable from the function w by auhstituting the variable vkfor the variable VI, and y is a consequence of the class X of the n-lth degree; (8) there exist sententialfunctions u and w as well as sentences y and z such that z, y. and z are universal quantifications of the functions u, w+u, and w reIJpectiveiy, and y and z are consequences of the class X of the n - I th deqree; (e) there exisi senieniiol function« uand w, f18cnitmce yand f1 natural number k such that x is a uniuersal quantification. of tlu: [unction. u + n" 11', Y is a universal quantification of 1hI; [unction u+ w, t'k iii not f1 free variable of u. awl y is a consequence of llu: class X of the n-lth

1 cr. Lukasiewicz. J. (51). pp. 159-63; IV, p .. ~I}.

1 The concept of consequence could also be introduced directly (i.e. without the help of consequonce of the nth degree) in the following way:

a: E UnIX) if and only if X ~ S and if the formula a: E Y Iwlda for every clau Y which Batillfiea thefollou,ing cowlitiona: (,,) X I: 1'; ({3) if yES and iaa univerBCJl quantification of the function u, z ill a univr.rNal 'l1,umtifimli,m of the function w, u i8 obtainable from the function w by 8Ubstitutinll the variable v.l: fOf' the tJariabU "I and Z E Y, then y E Yo (y) iJ YES, y, z, and tare un·iver8al quantificatWna oj 1M Junctions u, w+u, and w reapectillely and ZE YandtE y, then yE Yo (8) i/y e_ 8, u and w are ~enten!i?l functiona, 11 iB a univerBal quantification of the func/.lOR u+~.I: W, Z lB a u~tver8al quantification of the function u+w, "" is not a fr~ va~!eofthe.1unct~uand Z E Y,thenyE Y; (~)ijYE 8, u andwareaen~ junctu:ns, y '" a unwerBalquantijWation oj the ju7W'ion u+w, z.a unwerBCJl quantification oj the function u+ nl: w·and z E Y. then y E Y.

It ia, however. to be noted that by tra.nsformation of the definition jUlit given ~to 8,recursive sentence of the type of Def. 10 we obtain a sentence which 18 eq~~valent neither with the above definition nor with any other nonnal

definition (cf. p. 177. note I). .

VIII, § 2 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 183 sentential calculus. Translating the proof of this theorem,l we show successively from Def. 13 that

n (-+ +,) nl('I,I+('I,l+'l,l))'

1 '1,1 '1.1 1,1'

and nl( ~l,l +'1,1 +'1,1 + (Ll,l+ ('1,1 +'1.1)+ ('1,1 +'1,1)))

are axioms; consequently by Def, 15

n1('l,l+hl+'I,l)+h1+LI,I))

d d n (- + L ) is a eonse-

is a consequence of the Ist egree an 1 '1.1. 1.1

quence of the second degree of the class of all aXiOTIlB. Hence, by

Defs. 16 and 17,nl(~+h'1) is a provable sen~nce'lt' be

I f such inferences the ddlicules can

From examp es 0 . h d t eliminate

. . d which would at once arise if we WIS e. 0 •

rmagme h the assumptIOns which are

from the axioms of the metat eory . Id

Th fact that the axIOms wou no

of an existential nature. e ticular sentences

. t e of some par 1 "" ,

longer guarantee the exis ene ..... h

bli h' t very relevant. .nUC

whose provability we wish to esta IS , IS no, the existence of

. h f t that even assummg .

more important IS t e ac . ht be unable to establish its prov-

some concrete sentence, we mig t refer to the exis-

, , . in th of it might be necessary 0

abili ty; since m e pro 'ted tences (as is seen from

hie more eomplica , sen .

tence of ot er, as a ru _) E Pr' which was sketched

the proof of the theorem O(th1 +I~l'l ith special theorem') of

) S I as we are dea mg WI

above . 0 ong k asures to provide these

h 'E Pr' we can ta e me

of t e type x : . h tee the existence of the

ith isses whic guaran

statements W1 prem of The difficulties would increase

sentences necessary for the pro. t f a general character

.' I 'f. assed to statemen s a

significant Y 1 we P certain kind are provable

which assert that all sentences of a f a ziven class of

lly are consequences a "'.

--or, still more genera , nr to include among

I uld then often be necessa'J

sentences. t wo . ' m tions which would not be

the premisses general eXIstential. ass~ , P we had elimiweaker than those which, for lPtUltlVe reasons,

nated from the axioms, 2

182 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VlII. § 2 degree; (~) there exist sentential Junctions u and w, a sentence y and a natural number k, such that x is a universal quantification oj the Junction u+w, y is a universal quantification of the fUnction U+nk wand y is a CQ'fUJequence of the class X of the n-lth degree.

DEFINITION 16. x is a consequence of the class X of sentence8- symbolically z E Cn(X)-if and only if there is a natural number n such that x is a consequence of the nth degree of the class X.1

DEFINITION 17. x is a provable (accepted) sentence or a theorem-in symbols x E Pr-if and only if x is a consequence of the set of all axioms.

From this definition, it is easy to see that we shall have, among the provable sentences, not only all the sentences which can be obtained from the theorems of the sentential calculus in the same way in which the axioms of the first kind (i.e. those satisfying the condition (ex) of Def. 13) were obtained from the axioms of the sentential calculus, but also all known theorems of the unformalized calculus of classes, provided they are first translated into the language under investigation. In order to become convinced of this we imitate in the metatheory, in every particular case, the corresponding proof from the domain of the sentential calculus or of. the calculus of classes. For example, it is possible in this way to obtain the sentence

nl ('1,1+'1.1) from the well-known theorem 'ANpp' of the

11 B A W (90). vol. I, p. 101. ·2.1. 1 cr Whitehead, A, N" and Ru8IlO, ' f ,;..__ 11 12 24, and 2R in , 3.

J, h plfll! 0 H.... , '

• This is easily seen from t e oxam

184 CONCEPT OF TRlJTH IN FORMALIZED LANGUAGES VIII. § 2

For these reasons the standpoint might be taken that Def. 17, in case the existential assumptions are rejected, would no longer embrace all the properties which we ascribe to the concept of theorem. The problem of a suitable 'correction' of the ~bove definition would then arise. More precisely expressed, It would be a question of constructing a definition of theorem which would be equivalent to Def, 17 under the existential ft.ssumptions and yet-quite independently of these assump~,~ons-would have as consequences all theorems of the type

if the sentence x exists, then x E Pr', provided the corresponding theorem 'x E Pr' could be proved with the help of the existential assumptions. I shall give here a brief sketch of an attempt to solve this problem.

It can easily be shown that the axiom system adopted in the metatheory possesses an interpretation in the arithmetic of the natural numbers. A one-one correspondence can be set up between expressions and natural numbers where operations on numbers having the same formal properties are correlated with the operations on expressions. If we consider this correspondence, we can pick out, from the class of all numbers, those which are correlated with sentences; among these will be the 'primitive' nu~bers. We can introduce the concept of a 'consequence' of a grven class of numbers, and finally define the 'accepted' numbers as 'consequences' of the class of all 'primitive' numbers. If we now eliminate the existential assumptions from the axioms, the one-on.c correlation disappears: to every expression a natural number still corresponds, but not to every number an expression.

n' '

lit .we enn RtIII pf(l811rVC tho concept of 'nec(lptNl' number

previously established and define tho theorems U .. 'l those whioh are c~rrelated with 'accepted' numbers. If we try, on the basis of this new definition, to prove that a concrete sentence is a theorem, we shall no longer be compelled-as is easily seen-to refer to the existence of any other sentences. Nevertheless the pr~of ~I still require-and this must be emphasized-an ex~tentlal assumption, the assumption, namely, that there exist sufficiently many natural numbers or-what amounts to the same thing-sufficiently many distinct individuals. Thus

VIII, § 2 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 185 in order to derive all desired conclusions from the new definition, it would be neccssary to include in the metatheory the axiom of infinity, i.e. the assumption that the class of all individuals is infinite." I know of no method, be it even less natural and more complicated than the one just discussed, ~hi~h ~ou1d lead to a satisfactory solution of our problem which IS mdependent of

the above axiom.

In connexion with the concepts of consequence and of theorem

I have mentioned rules of inference. When we have in mind the construction of a deductive science itself, and not the investigation of such a science carried out on the basis of the metatheory, we give, instead of Def. 17, a rule by which we may add to the science as a theorem every consequence of the axioms. ~n our case this rule can be divided into four rules-oorresponding to the four operations which we use in the construction of con-

sequences. II

By means of the concepts of sentence and of CO~quence a

the most important methodological concepts can be mtroduc~d into the metatheory, in particular the concepts of deducttve

system., of consistency and of completeness.'!.

DEFINITION 18. X is a deductive system if and only if

Cn(X) S; X S; S.

D N 19 X is a consistent class of sentences if and only if

EFINITIO . - - 0 (X)

X s S and if, far every sentence x, either x E Cn(X) or x En.

DEFINITION 20. X is a complete class of sentences if ~nd only if X s Sand if,for every sentence x, either x E Cn(X) or x E Cn(X).

h cept will prove useful:

In the sequel yet anot er con

and yare equivalent with

DEFINITION 21. The sentences x S

ir nd only if x E S yES, X S

respect to the class X of sentences lJ a '

and both x+y E Cn(X) and ii+x E Cn(X).

. ced i this

A more detailed analysis of the concepts mtrodu m

section would exceed the limits of the present work.

1 Cf. Whitehead, A. N., and Russell. B. A. W. (90). vol. 2. p. 203. ~ ce. pp. 70. 90. and 93 of tho prosont volume.

186 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES vnr, § 3

VIII, § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 187

Let us try to approach the problem from quit~ ~ differ~nt angle, by returning to the idea of a semantical definition as In § 1. As we know from § 2, to every sentence of the language of the calculus of classes there corresponds in the metalanguage not only a. name of this sentence of the structural-descriptive kind, but also a sentence having the same meaning. For example, corresponding to the sentence 'II x, II x.AIx,x. Ix, x,' is the name 'n n ( + )' and the sentence 'for any classes a and b we

1 2 L12 '21 f h

h C 'b b' c a' In order to make clear the content 0 t e

ave a _ or _ .

concept of truth in connexion with some one concrete sentence

of the language with which we are dealing we can apply the same method as was used in§ 1 in formulating the sentences (3) and ~4~ (cf p 156). We take the scheme (2) and replace the symbol. x in it by the name of the given sentence, and 'p' by its translatIOn into the metalanguage. All sentences obtaine~ in this way, e.g. 'n n (L +, ) is a true sentence iJ and ooZy iJ Jor any classes a

nd1b 2 001,2 a2~ b or b c a' naturally belong to the metalanguage

a we ve - -, . . t'

and explain in a precise way, in accordance with lingutB I~ US:f~

the meaning of phrases of the form :x i~ a ~rue sentence ; d:f occur in them. Not much more in principle IS to be deman e.

I d flnition of true sentence than that it should satisfy a genera e 1 d . elude the usual conditions of methodological correctness an . m h uld

all partial definitions of this type as special cases; that It s 0

be so to speak, their logical product. At most we. can t~O require that only sentences are to belong to the exte~l~n 0 e defined concept, so that, on the basis of the definitIOn ~~-

f th type 'x is not a true sentence, m

structed all sentences 0 e bit

which in the place of 'x' we have the name of an ar I rsry ex-

h bi t) hioh is not a sentence, can

pression (or of any ot er 0 lec w

be proved. II te

U. th b I "I'r' to denote the class of a true sen noee,

sing e sym 0 d i th f 11 wing conventhe above postulate can be expresse in e 0 0

tion:

T A 1.0000maUy correct definitioo oj the symbol

CoNVENTION • J'

. . uall done In tho l!outS6 of this work I shall

thoorems, although. thIS 18 not WI to ~h6 r~blcm of thc mutual rclr.tionH of

have several OCCBRlons to return p

these two concepts: of thoorem and of true sentence.

§ 3. THE CONCEPT OF TRUE SENTENCE IN THE LANGUAGE OF THE CALCULUS OF CLASSES

I pass on now to the chief problem of this article-the construction of the definition of true sentence, the language of the calculus of classes still being the object of investigation.

It might appear at first sight that at the present stage of our discussion this problem can be solved without further difficulty, that 'true sentence' with respect to the language of a formalized deductive science means nothing other than 'provable theorem', and that consequently Def. 17 is already a definition of truth and moreover a purely structural one. Closer reflection shows, however, that this view must be rejected for the following reason: no definition of true sentence which is in agreement with the ordinary usage of language should have any consequences which contradict the principle of the excluded middle. This principle, however, is not valid in the domain of provable sentences. A simple example of two mutually contradictory sentences (i.e, such that one is the negation of the other) neither of which is provable is provided by Lemma E below. The extension of the two concepts is thus not identical. From the intuitive standpoint all provable sentences are without doubt true sentences (the Defs. 13-17 of § 2 were formulated with that in mind). Thus the definition of true sentence which we are seeking must also cover sentences which are not provable. 1

1 The fact must also be takon into oonsldornt.ion that--in contrast to the concept of true senteneo-c-tho conccpt of provable I<entonco hM a purely accidental chnrncter when applied to Rome doduotlvo RoicrtlWR. which is chieflY connect", I with the hiRtorical development of tho "('iollce. It ill 8omotimes diffic_ult to give ohjoctivn grotnulH for nnrr-owing r)f widening tho (lxtollMi.on of tins concept in a p,uti(luJlI.f direction, For exnmplo, when wo IUO dealing With

the calculus of classes the sentence n In"I., whioh Htipulates tho existence of at least two distinct classes, is not accepted on the basis of the definitions of § 2-which will be expressed in Lemma E. Moreover this sentence cannot be derived from the formal hypotheses upon which the work of Schroder is baaed, although in this case the matter is not quite clear (cf. Schroder, E. (62), vol. i, pp. 245 and 246; vol. 2, Part I, p. 278; vol. 3, Part I, pp. 17 and 181; but ill many works this sentence occurs 11.8 an axiom of the algebra of lagro or forms an obvious consequence of these axioms (cf. Huntington, E. V. (32), p. 297, Post. 10). For quite different reasons which will be discussed below in connexion with Th. 24 (cf. especially p. 207: footnote), it would be desirable

to include the sentence nl(n"l,.+U.('J, r- n.(n., •.• +, •.• + •• ,.})) aIDons the

188 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VITI.f3 "I'r', formulated in the metalanguage, will be called an adequate definition of truth if it has the following consequences:

(0:) all sentences which are obtained from the expression 'x E Tr if and only if p' by substituting for the symbol 'x' a structuralde8criptive name of any sentence of the language in question andfor the symbol 'p' the expression which forms the translation of this sentence into the metalanguage;

(fJ) the sentence 'for any x, if x E Tr then XES' (in other words "Tr S 8').1

vnr, ~ 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 189 or in any other language. Our task is thus greatly complicated.

The idea of using the recursive method suggests itself. Among the sentences of a language we find expressions of rather varied kinds from the point of view of logical structure. some quito elementary, others more or less complicated. It would thus be a question of first giving all the operations by which simple sentences are combined into composite ones and then determining the way in which the truth or falsity of composite sentences depends on the truth or falsity of tbe simpler ones contained in them. Moreover, certain elementary sentences could be selected, from which, with the help of the operations mentioned all the sentences of the language could be constructed; ~hese selected sentences could be explicitly divided into true and false, by means, for example, of partial definitions of the type described above. In attempting to realize this idea we are however confronted with a serious obstacle. Even a superficial analysis of Defs. 10-12 of § 2 shows that in general composite sentences are in no way compounds of simple sentences. Sentential functions do in fact arise in this way from elementary functions, i.e. from inclusions; sentences on the contrary are certain special eases of sentential functions. In view of this fact: no method can be given which would enable us to define the required concept directly by recursive means. The possibility sugge.sts itself however of introducing a more general concept which is a~plicable t~ any sentential function, can be recursively defined, and, when applied to sentences, leads us directly ~o the concept of truth. These requirements are met ~y the ~ohon of the satisfaction of a given sentential function by glt'en objects, and in the present case by given classes of individuals.

Let us try first to make clear by means of so~e ex.a~ples the usual meaning of this notion in its customary linguistic us~ge. The way in which we shall do this represe~ts a natural generalization of the method which we have previously used for the con-

cept of truth. . .

The simplest and clearest case is that in which the given sen-

. . t . I ""'e free variable We can then

tential function con ams on y VH, •

It should be noted that the second part of the above convention is not essential; so long as the metalanguage already has the symbol 'Tr' which satisfies the condition (ex), it is easy to define a new symbol 'Tr" which also satisfies the condition {Pl. It suffices for this purpose to agree that Tr' is the common part of the classes Tr and 8.

If the language investigated only contained a finite number of sentences fixed from the beginning, and ifwe could enumerate all these sentences, then the pro blem of the construction of a correct ~efinition of truth would present no difficulties. For this purpose It wo.u]~ suffice to complete the following scheme: x E Tr if and only if e~ther x = Xl and PI> or x = x2 and P2'''' or x = Xn and Fn' the ~~bo)s 'Xl" 'Xz', ... , 'xn' being replaced by structuraldescnptlve names of all the sentences of the language investi-

gated and' , " ., b . f PI, P2 , ... , P« Y the corresponding translation 0

these sentences into the metalanguage. But the situation is not like this. Whenever a language contains infinitely many sentences, the dofinition constructed automnt,icctlly according to the above scheme would have to cOllsist, ofinfinitoly mUIlY words, and such sentences cannot be formulated either in the metalanguage

. ~t If we wished to SUbject the metalanguage and the metatheory expressed ~ I ~ th~ process of ~ormalization. then the exact specification of the mean. Ing 0 varrous expreSSlO1lJI which occur in the convention T would present no great difii It'

. cu 100. e.g, the expressions 'JormaUy correct deRnitiun oj the

gWen 81Jmbol' • st-·~t l d '. ~.

stud' " • • ~ ~ra - escnptwe name oj a given expression oj the language

~::. the tran8lat~on oj a given sentence (oj the langtlage studied) into the me~. ng~e'. After unimportant modifications of its formulation the coo. thee Ion Itself would then become a normal definition belonging to the metseory,

190 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 3 significantly say of every single object that it does or does not satisfy the given function. 1 In order to explain the sense of this phrase We consider the following scheme:

for all a, a satisfies the sentential function x if and only if p and substitute in this scheme for 'p' the given sentential function (after first replacing the free variable occurring in it by 'a') and for 'x' some individual name of this function. Within colloquial language we can in this way obtain, for example, the following formulation:

for all a, a satisfies the sentential function IX is white' if and only if a is white

(and from this conclude, in particular, that snow satisfies the function 'x is white'). A similar construction will be familiar to the reader from school algebra, where sentential functions of a special type, called equations, are considered together with the numbers which satisfy these functions, the so-called roots of the equations (e.g. 1 is the only root of the equation 'x+2 = 3').

When, in particular, the function belongs to the language of the calculus of classes, and the corresponding explanation of the expression 'a satisfies the given sentential function' is to be formulated wholly in the terms of the metalanguage, then in the above scheme we insert for 'p' not the sentential function itself, ~ut the expression of the metalanguage having the same meanmg, and for 'x' we substitute an individual name ofthis function which likewise helongs to tho metalanguage. For example, this method gives the following formulation in eonnoxion with the function 'IT x" l x, x" ':

!or all a, a satisfies the sentential function n3 'I,ll if and only if lor all classes b we have a s b

(Whence it follows at once that the only class which satisfies the function 'II XII Ix, XII' is the null class).

In cases where the sentential function has two distinct free variables we proceed in an exactly analogous manner. The only

I 1. PrOvisionally I ignore problema connected with semantical categories (or ogtcal types); these problems will be discussed in § 4.

vm, § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 191 difference is that the concept of satisfaction now refers not to single objects but to pairs (more accurately to ordered pairs) of objects. In this way we reach the following formulations:

for all a and b, a and b satisfy the sentential function 'x sees r. if and only if a sees b ; for all a and b, a and b satisfy the sentenhal lunctian '2,3 (i.e. 'Ixwxm') if and tmly if a s b.

Finally we pass to the general ease, where the given sentential function Contains an arbitrary number of free variables. For the sake of a uniform mode of expression we shall from now on not say that given objects but that a given infinite sequence 01 objects satisfies a given sentential function. If we restrict ourselves to functions from the calculus of classes, then the establishment of an unambiguous explanation of this expression is facilitated b! the fact that all the variables which occur in the language of this science are ordered (enumerated) in a sequence. In considering the question of which sequences satisfy a given sentential function, we shall always have in mind a one-many co~espondence of certain terms of a sequence I with the free variables of the sentential function where with every variable corresponds the

, I' will

term of the sequence with the same index (i.e. th.e term J k

be correlated with the variable Vk)' No account will be taken of the terms which are not correlated with any variable.! We can explain the procedure best by means of. concrete .examp~e8. Consider the function n2 '1 2 already mentioned. This function

. free vari 'bl so that we consider only the

contains only one ree varia e VI'

first terms of sequences. We say that the in~nite seque~ f 01 classes satisfies the sentential function n2 '1,2 if and only il the

1 This is a sim lification of. a purely techni~l nature. Even if we could

p . I m a sequence (e.g. because we

not order all the variables of a given ~ uld still number all the

used symbols of arbitrary shapes as variables), we co. . of

signs and thus all the variables, of every given express~onihe.g. on the. h:~he

' rd' hich they follow one another m e espressio .

the natural 0 er m w called the first the next the second, sign standing on the extreme left co~d be rtain correlation between the and so on. In this way we could again set up ace ..... ,_ correla-

. f ti d the tenns of the sequence. u .....

free variables of a given nne IOn ~ . h t) ould obviously vary with

tion (in contrast to the one descn~ m: e te~d w with it J'ft,ther serious the form of the function in question ; th18 wo . ca~1 and espocially of

complications in the formulation of Def. 22 given ow

conditions (y) and (8).

192 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VITI, § 3 class fl satisfies this function in the former sense, i.e. if for all classes b, we haoe]; s:; b. In an analogous way the infinite sequence f of classes satisfies the sentential function £2,3 if and only if the classes f2 and f3 satisfy the function in the previous sense, i.e. if f'1. £ f3' This process may be described in general terms as follows.

We consider the following scheme:

f satisfies the sentential function x if and only 1f I is an infinite sequence of classes, and p.

Given any sentential function, say s, from the calculus of classes, we replace in the above scheme the symbol 'x' by an individual (structural-descriptive) name of s constructed in the metalanguage; at the same time, for all the free variables Vk t VI' etc. occurring in s we substitute the corresponding symbols 'Ik" 'It', etc. and we replace 'p' in the scheme by the expression thus obtained from s (or by its translation into the metalanguage).

We shall use a recursive method in order to formulate a general definition of satisfaction of a sentential function by a sequence of classes, which will include as special cases all partial definitions of this notion that are obtained from the given scheme in the way described above. For this purpose it' will suffice, bearing in mind the definition of sentential function, to indicate which sequences satisfy the inclusions £k I and then to specify how the notion we are defining behaves when the three fundamental operations of negation, disjunction, and universal quantification are performed on sentential functions.

The operation of universal quantifioation calls for speoial considerat.ion. Lot x be any sentontinl function. and assume that we already know which S()CI'WlJ(l{~S lmtisfy tho function x. Considering tho meuning of tho oporution of IlnivorslLl 11110.n- t~ficationt we shall say that the sequence 1 satisfies the function nkX (where k is a particular natural number) only if this sequence itself satisfies the function x and does not cease to satisfy. it even when the kth term of this sequence varies in any way; . III other words, if every sequence which differs from the grven sequence in at most the kth place also satisfies the function. For example, the function n £ is satisfied by those,

d nl h 2 1.2

a~ 0 y t ose, sequences f for which the formula 11 S f2 holds

WIthout regard to the way in which the second term of this

VIII. § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES IlI3 sequence is allowed to vary (as is easily seen, this is only possible when the first term is the null class).

After these explanations the understanding of the following definition should not be difficult.

DE1<'INITION 22. The sequence f satisfies the sentential lund ion x if and only if 1 is an infinite sequence 01 classes and x is a 8enlwlinl function and il I and z are such that either (a) there ezisi natural numbers k and 1 such that x = £k,1 and it <: II ; ((3) there is a sententia! function y such that x = y and f does not satisfy the function y; h') there are sentential functions y and z such that x = y + z and f either satisfies y or satisfies z; or finally (~) there is a natural number k and a senteniial functioti Y such that x = n k y and every infinite sequence of classes which differs from fin at most the k-th place satisfies the

function y.l

The following are examples of the application of the above definition to concrete sentential functions: the infinite sequencef satisfies the inclusion £1,2 if and only if 11 £ f2' and the function

'23+'32 if and only if f2 *13; the functions n2'1,2 and n2'2.3 a~e satisfied by those, and only those, sequences 1 in which I, is the null class and fa the universal class (i.e. the class of all individuals) respectively; finally, every infinite sequence of classes satisfies the function '1.1 and no such sequence satisfies

the function '1 2' £1 2'

The concept just defined is of the greatest importance for

investigations into the semantics of language. With its help the meaning of a whole series of concepts in this field can easily be

I The normal definition, which is equivalent to the above recursive one, is as follows (cr. pp. 177, ISO, and 182):

The sequence j satisfies the sentential junction X ij, and (mly ij we Iun:« jRx jor

every relation R which satisfies the jollowing condition: , '

For any '.I and y, in order that gRy it is necessary and euffiClent that '.I fS an infinite sequence oj Ckul8CS, Y is a sentential junctwn and elihu, (~) there are natural numbers k and 1 such. toot y = 'k,1 and gk <;: fll or (fi) there 18 U sentential junction z such toot y = z and the jormula gRz does not hold; or (y) t},~reare sentential junctions z and t such that y = z+t and gR: or (JRt; or firwlly (0) th.cre is a natural number k and a sentential [unctior» z such thu; 'J nk Z {JTvi hflz

, ' , __ h'h' 1'1' "jrQm'Jatlhek'lhl)l{u~

[or evenJ infinite sequence h oj CIA.U>Se8 w ie tJI { IN tnr~

a-t most,

194 CONCEPT OF TRUTH IN }'OR:\fALIZED LANGUAGES VIII, § 3 defined, e.g. the concepts of denotation, definability,' and truth, with the last of which we are especially concerned here.

The concept of truth is reached in the following way. On the basis of Def. 22 and the intuitive considerations which preceded it, it is easy to realize that whether or not a given sequence satisfies a given sentential function depends only on those terms of the sequence which correspond (in their indices) with the free variables of the function. Thus in the extreme case, when the function is a sentence, and so contains no free variable (which is in no way excluded by Def. 22), the satisfaction of a function by a sequence does not depend on the properties of the terms of the sequence at all. Only two possibilities then remain: either every infinite sequence of classes satisfies a given sentence, or no sequence satisfies it (cf. the Lemmas A and B given below). The sentences of the first kind, e.g. UI '1,1> are the true sentences; those of the second kind, e.g. ni tI,l' can correspondingly be called the false sentences. t

1 To say that the name x denotes a given object a is the same as to stipulate that the object a (or every sequence of which a is the corresponding term) satisfies a sentential function of a particular type. In colloquial language it would be a function which consists of three parts in the following order: a variable, the word • is' and the given name x. As regards the concept of definability, I shall try to explain its content only in a particular case. If we consider which properties of classes we regard as definable (in refere~ce to the system of tho calculus of classes discussed here), we reach the foliowlDg formulations:

We. say that the 8entential function x defines llie property P of claB~e8 if ana only if for a natural number k (<xl x contains VA; a,. its 0,,1.'1 free oariable; and (P) in order that an 'infinite sequence f of classe •• should 8alisfy x, it i.'t necessary and suffirient that fk should have the property I'; Wil Sll.'l that the property P of elaese« is dcfinllble iJ and only if there ia a lI,mte"tial Junction x w!Jicl~ drfillf,s P,

On tho bllJ<iH of those HtipulutinllK it cun be shown, for oxamplo, that such propertios of cluases £I!I omptlness of contaiuiu" only one two, throe, etc.,

, ... ' . . 't I

elements aro definable, On tho other hund the property of eontaining wfinl ~ Y

m~ny eloments is not dofinahlo (cf. the remarks givon below in conlloXlO1l with Ths. 14-16). It will also be seen that with thil< interpretation the concept of definability does not depend at all on whether the formalization of the science investigated admits of the possibility of constructing definitions. More exact discussions of definability will be found in articles VI and XIII of the present volume.

t A method of defining truth which is essentially equivalent to the method developed in this work. but is based upon a different idea, has recently bee? suggested by J. C. C. McKinsey in his paper 'A new definition of truth, Syntltese. vol, 7 (1948-9), pp. 428-33.

VIII, § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 195 DEFINITION 23. X is a true sentence-in symbols x E T'r-« if and only if XES and every infinite sequence of classes 8atiHfil'S X.l

The question now arises whether this definition, about tho formal correctness of which there is no doubt, is also mat.erinlly correct-at least in the sense previously laid down in the eonvention T. It can be shown that the answer to this question it! affirmative: Def. 23 is an adequate definition of truth in the 8w·~e of convention T, since its consequences include all those required by this convention. Nevertheless it can be seen without difl~culty (from the fact that the number of these consequencet! 18 infinite) that the exact and general establishment of this fact has no place within the limits of the considerations so far bro~ght forward. The proof would require the setting up of an entirely new apparatus: in fact it involves the transition to a level one step higher-to the meta-metatheory, which would h~ve to be preceded by the formalization of the metatheory which .forms the foundation of our investigations.2 If we do not WIsh to depart from the level of our previous discussions, only one

1 In the whole of the above construction we could operat.e with finite

. b ft' tead of WIth mfirute sequences.

sequences with a variable num er 0 erms ms . ' I

It would then be convenient to generalize tho concept ?f finite neqllence. n

. hi te ence which has an nth term

the usual interpretanon of t IS rm a sequ 0 Iinnui rh

must also have all terms with indices less than n-we m~8t now re illqfU~t

elation as a finite sequence I ,~ this postulate and regard any many-on re I be dist.inct from 0 counter domain consists of a finite number of natura num. r~ }l; th~ The modification of the construction would consist ill el.'mma~:~:~:rms, sequences which satisfy the given sentential function all sri,e ·f. e etc which have no influence on the satisfaction of the fWlctJOn. us I onl I' h ., occur as free variables in the function (of course in finite numberh),oh y:. ::

• • J, l tc ld remain in the sequence w IC sa 15

terms with tho indices s, ,e ., wou f lasses would

this function. For example, those. ~d only those, sequejces ~ f ~.er~~ing the satisfy the function " .• which constst of onlY!W~.tern;sro~ a~he 'standpoint of formula f. ef •. The value of such a modi -ocedu r., clear but when we naturainess- and conformity wit~ the ~al proc . c~:a~ure sh~w themselves: come to carry it out exactly certam de~ect. of a lag! Re arding the concept of

Def. 22 then takes on a more comphcated form. g I

. . ted that---according to the above treatment--on Y one

truth, ,t 18 to be no hi h h no member at all can

I th 'mpty' sequence w ic as '

sequence, name y e e .' variables' we Khould then have

satisfy a sentence. i.e. a functIOn ,:",thout f~uall HBti:fietl hy the . empty' to call those sen~nces true whIch a~ 8C to t~iJl definition will douhtl""" sequence. A certam artlficUlhtYffia~t~) nfg 'Iiar with tho Hpe<:ific prO<:(>Ilu"'"

displeas« all th080 who are not RU cien y nrm .

which arc commonly IlM') in rn"themati"al co""trllc:tlUn~.

I See P: 1 RH, [oot"otfJ.

196 CONCEPT OF TRUTH IN FOR~L'\LIZED LANGUAGES VIII. § 3 method, the empirical method, remains-the verification of the properties of Def. 23 in a series of concrete examples.

, Consider, for example, the sentence nl U2'l,2' i.e. n z, N Tl XH Nix, x. '. According to Def. 22 the sentential

function '1,2 is satisfied by those, and only those, sequences f of classes for which II S 12 holds, but its negation, i.e, the function '1,2' only by those sequences for whichj', '* 12 holds. Consequently a sequence 1 satisfies the function n2~' if every sequence g which differs from 1 in at most the 2~d place satisfies the function '1,2 and thus verifies the formula gl 'f;. g2' Since gl = 11 and the class Y2 may be quite arbitrary, only those sequences! satisfy the function n2 '1,2' which are suchthatj', !jb bfor any class b. Ifwe proceed in an analogous way, we reach the result that the sequence 1 satisfies the function U2 '1,2' i.e. the negation of the function n2 '1,2' only if there is a class b for which II S b holds. Moreover, the sentence nl U2 '1,2 is only satisfied (by an arbitra?" sequence I) if there is for an arbitrary class a, a class b for which a S b. Finally by applying Def. 23 we at once obtain one of the theorems which were described in the condition (0:) of the convention T;

n. U2'1,2 E Tr if and only if for every class a there is a class b such that a r;;; b.

VIII, § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 197 question. Thus,forexample, nl U2('I,2+'l,2)isshown to be a true and n 1 nll'~ a false sentence. With respect to other sontoncca, e.g. the sentence nl n2 n3 (tI,d 'z,aba,l) or itH negution, the analogous question cannot he decided (at least HO long aH WI' do not have recourse to the special existentlal aasuruptions or the metatheory, cf. p. 174): Dof. 2:l11lonc gives no geneml criterion for the truth of a sentence. 1 Nevertheless, through the theorems obtained, the meaning of the corresponding oxpressions of tho type '» E Tr' becomes intelligible and unambiguous. It should also be noted that the theorem expressed in the condition (f3) of the convention T is also an obvious consequence of our definition.

\Vith these discussions the reader will doubtless have reached the subjective conviction that Def. 23 actually possesses the property which it is intended to have: it satisfies all the conditions of convention T. In order to fix the conviction of the material correctness of the definition which has been reached in this way, it is worth while studying some characteristic general theorems that can be derived from it. With a view to avoiding encumbering this work with purely deductive matter, I shall give these theorems without exact proofs. 2

THEOREM I (The principle of contradiction). For all sentences z, either x E Tr or i E Tr.

This is an almost immediate consequence of Defs. 22 and 23. THEOREM 2 (The principle of excluded middle). Far all sentences z, either x E Tr or i E Tr.

From this we infer without difficulty, by using the known theorems of the calculus of clul:lscs, that, nJ U2 '1,2 is a true sentence.

Wo can proceed in an ()xuet,ly nnalogou» way with overy other sentencn or the IfLngultgt, we nre oonsidering , I f for Huoll n. sentenc~ "" construct a corresponding assertion described in the condItIOn (ct) and then apply the mode of inference used above, we can prove without the least difficulty that this assertion is a consequence of the definition of truth which we have adopted. In many cases, with the help of only the simplest laws of logic (from the domain of the sentential calculus and the calculus of classes), we can draw definitive conclusions from theorems obtained in this way about the truth or falsity of the sentences in

• At least when it is regarded from the methodological viewpoint this is not a defect of the definition in question; in this respect l~ does not d~er at all from the greater part of the definitions which occur ill the deductive

seiences. . hill' of

• The proofs are based on the general laws of logic, t : 5~ c axioms

the meta.seience and the definitions of the concepts oee~mg m the theorems, In some cases the application of the general propertIes of the ~o;~Pt ~f cons uence, of deductive system, etc., which are given in artrc (> 0 t ie

eq I .. di t ! We are able to use the r""ult~ obt.ainod t1wre

present \.'0 umo l~ In len f)( • i

because it can easily be shown that the concepts of ... -ntonce urH "~)!l'·e<I'l.mr., . d I I ti f nil the axiom. "f,on which the .. hove·rn"ntlOn.,,1 work

mtro UCef loro sa. IS y ouo " . •

w .... baeod.

198 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 3 In the proof the following lemma, which follows from Defs, II and 22, plays an essential part:

LEMMA A. If the sequence f satisfies the sentential function x, and the infinite sequence g of classes is such that for every k, fk = Yk if Vk is a free variable of x, then the sequence g also satisfies the function x.

As an immediate consequence of this lemma and Def. 12 we obtain Lemma B which, in combination with Defs. 22 and 23, easily leads to Th, 2:

LEMMA B. If XES and at least one infinite sequence of classes satisfies the sentence x, then every infinite sequence of classes satisfies x.

THEOREM 3. If X £: Tr then Cn(X) £: Tr; thus in particular Cn(Tr) £: Tr.

This theorem is proved by complete induction based chiefly on Defs. 15, 16, 22, and 23; the following simple lemma is also useful in this connexion:

LEMMA .C. If Y is a universal quantification of the sentential function x, then in order that every infinite sequence of classes should satisfy x, it is necessary and sufficient that every infinite sequence of classes satisfies y.

The results contained in Ths. 1-3 may be summarized in the follOwing (obtained with the help of Defs. 18-20):

THEOREM 4. The class Tr is a consistent and complete deductive system.

THEOREM 5. Every prooable 8cntence is a true sentence, in otber words, Pr £: r-.

This theorem follows immediately from Def. 17, from Th. 3, and from Lemma D, the proof of which (on the basis of Def. 13 and Lemma C among others) presents no difficulty.

LEMMA D. Every axiom is a true sentence. Th. 5 cannot be inverted:

THEOREM 6. There exist true sentences which are not provable, in other words, Tr 't Pro

VIII, § 3 CONCEPT OF TRUTH IN FOR:'fALIZED LA~GlJAGEl'l 1!1!1 This is an immediate consequence of Th. 2 and the following lemma, the exact proof of which ill not quito caRy:

LEMMA E. Both nl nZ'I.2 r r- and h~n;'1:2 (: /'r.1

As a corollary from Ths. 1,0, and 6, I give finally the following theorem:

THEOREM 7. The class Pr is a consistent, but not a com-plete deductive system.

In the investigations which are in progress at the present day in the methodology of the deductive sciences (in particular in the work of the Gottingen school grouped around Hilbert) another concept of a relative character plays a much greater. part than t~e absolute concept of truth and includes it as a special case. This is the concept of correct or true sentence in an individual domain a.2,3 By this is meant (quite generally and roughly speaking) every sentence which would be true in the usual sense if we restricted the extension of the individuals considered to a given class a, or -somewhat more precisely-if we agreed to interpret the terms 'individual' 'class of individuals', etc., as 'element of the class a', 'subclass ofthe class a', etc., respectively. Where we are dealing with the concrete case of sentences from the calculus of classes we must interpret expressions of the type 'n xp' a.'l 'for every subclass x

, If we wish to include the sentence n. n., •.• among the acceptable

<_ [as ] ften the case ef P 186 footnote) we could use here, instead of

8en~nces as liS 0 ~ ,. •• ,

Lemma E, the following Lemma E':

Both nln'('II+'I,') E Pr and n=-,-::n:-.7"(.-, .• ";+- .. - . .) E Pro

The idea of the roof of both of these leIIllDSB is the same ~ that of the proofs of the cons!tency and incompleteness of t~e lower f~:.tI~~1 calculus which is found in Hilbert, D., and Ackermann, '''. (30), PP: a- .

2 The discussion of this relativised concept is not ~ntial for the under-

standing of the maintedth~me of !:s::l~an: :: :~~ t: ==t:

who are not mteres m spec ". 20S-9 are in closer of the deductive sciences (only the discussions on pp.

connexion with our. main th~is). Ackermann, 'V. (30), especially

• In this eonnexion see Hilbert, D"""nfink_and I 'I (5 ) But it should be

"1 d Be y P and Scho e. .., . a .

pp. 7~8 , an rna s"'. hi cept not to sentences

emphasized that the authors mentioned" relate ~ con. tt e ISlIguagc of the

hut to sentential functions with free var1abhlC!< ( ause ~~n"~n~- in tt". "trid

. h" h th t ere arc no ~ .., •. '~ .

lower funccional calculus w IC ~y UMh. th the term 'w-norally valid '

f th rd) and connected With t Ifl. fly woo ., L

~ense 0 e wo .' t -" •. , • (Of the HCCond of th~ Wink ... riv',d RfJI)\"r.

instead of the term correct. or true. .

pp.347-8.

200 COXCEPT OF THL"TH IX FOR:\L-\LIZED L.-\....'..-oUAGES VIII, § 3 Of the class a lce have p', and expressions of the type 'Ixy' as 'the subclass x of the class a is contained in the subclase y of the class a'. We obtain a precise definition of this concept by means of a modification of Defs. 22 and 23. As derived concepts we introduce the notion of a correct sentence in an indiridual domain with k elements and the notion of a correct sentence in every individual domain. It is worthy of note that-in spite of the great importance o.f these terms for meta mathematical investigations-they have hItherto been used in a purely intuitive sense without any attempt to define their meaning more closely."

DEFINITION 24. The sequence f satisfies the sentential function x in th~ in~vidual domain a 1] and O'nly if a is a class of individuals, Jan infinite sequence of subclasses of the class a and x a sentential fU~tctwn satisfying one of the fO'llou·ing four conditions: (ex) there exis! natural numbers k and I such that x =v e and] cr. (Q) there

. .. kJ k - Jl: I-'

M a sentential function. y such that z = fi and the sequence f does not

s~tisfY y in the individual domain a; (y) there are sententialfunc~~on.s ~ and z such that x = y+z and f satisfies either y or z in the tndlVtdual domain a; (8) there is a natural number k: and a sentential function y such that z = nkY and every infinite sequence g of subclasses O'f the class a which differs from f in at most the k-th place satisfies Y in the individual domain a.

DEFINITION 25. x is a correct (true) sentence in the individual domain a if and only if XES and every infinite sequencc of s'UbclaS8CS of the class a suti.'lfics the sentence x in the illdividual domain a.

l)rr.I"INITroN .,(' .. . . .. I I

• • - I. X M a COln~rJt, (I,nw) R('nt.(,lHlO HI nn lIIdlVIC un

dO.J1l1ull With Ic O[l'lnl'I!I;H in 8!JIIII)(}/8.r ( ('/~,lf and onl!ll] there ~X't81.~ a clU88 it such. that k is tlte cardinal 'ILl/mba IIf tlu. cla88 a and. x ~s a correct 8entence in lite indil"idtlal domain a.

1 An exception iH furnished by Herbrand J (:!6) in which tho author

dofines the con t f t ' •

. ' cop 0 rue sentence in a finite domain (pp. l08-12). A com-

fa~son of Herbrand's definition with Defs. 25 and 26 given in the text will ea t?e reader at once to the conclusion that we have to do here with like?oun~bgl terms rather than with a relationship of content Nevertheless, it

1S POSSI e that with resp t t ..... '. d

. 1 ' eo 0 cenam concrete deductive SCIences, and un or

hspeelal SS8umptions for tho corresponding motatheory Herbrand's concopt as t ]0 same exteru'ion (a d I th·' . I . , " ' n a so e Barno nnportance for metamathematlCa In>estlgatlOns) as a certai . I f 25

m specie case of the concept introduced in De. .

VIII, § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 201 DEFINITION 27. X is a correct (true) sentence in every individual domain-in symbols x E Ct-if and O'nly if fOT every class a, x is a correct 8entence in the individual domain a.

If we drop the formula 'x E S' from Def. 25, and thereby modify the content of Defs. 26 and 27, we reach concepts of a more general nature which apply not only to sentenccs but also to arbitrary sentential functions.

Examples of the application to concrete sentences of the concepts defined will be given below. In the interest of more convenient formulation of various properties of these concepts, I introduce some further symbolical abbreviations.

DEFINITION 28. x = Ek if and only if

x = nk+l 'k,k+1· nk+l(nk+2 'k+1,k+2+'k+1,k+Lk,k+l)· DEFINITION 29. z = 0: if and only if

x = nl(n2'1,2+U2('2.1·E2))·

As is easily seen, the sentential function Ek states that the class denoted by the variable Vk consists of only one ele~ent; .the sentence ex, which plays a great part in subsequent investigations, states that every non-null class includes a one-element class as a part.

DEFINITION 30. x = f3n if and anly if either n = 0 and

n+1 n I )

X - n € orn '* Oandx = ni.;;n+1( ~ Ek+ ~ ~ (LkJ+l·L{+1,k) .

- 1 I. k l k

DEFINITION 31. x = Yn if and anly if either n = 0 and x = f3o, or n '* 0 and x = f3n-l·f3n·

It follows from these definitions that the sentences f3n and Yn (where n is any natural number) respectively assert that there are at most n and exactly n distinct one-element classes or, what amounts to the same thing, distinct individuals.

DEFINITIO~ 32. xisa quantitative sentence (or asentenceabout the number of individuals) ifand anly if there exists afinite 8eq~,!:~e

n n

p of n nat ural numbers such. that ~ither x "-~ t YP. or x t )'P.·

202 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 3

I shall now give a series of characteristic properties of the defined concepts and the more important connexions which relate them with notions already introduced. This is the place for some results of a more special nature which are connected with the particular properties of the calculus of classes and cannot be extended to other disciplines of related logical structure (e.g. Ths. 11-13, 24, and 28).

THEOREM 8. If a is a class of individuals and k the cardinal number of this class, then in order that x should be a correct sentence in the individual domain a it is necessary and sufficient that x E Gtk•

The proof is based on the following lemma (among other things) which follows from Def. 24:

L~MMA~. Let a and b be two classes of individuals and R a relat~~n. whtch satisfies the following conditions: (lX) for any I' and g" if j'Rg' then f' is an infinite sequence of subclasses of a, and g of subclasses of b; «(3) if l' is any infinite sequence of sub~lass~s of a, then there is a sequence g' such thatf'Rg'; (y) if g' is any '/,nfim~e sequence of subclasses of b, then there is a sequence l' such thatf Rg'; (8) for allf', «.r. g", k and l, if j'Rg', j"Rg", and ~ a~d l a:e natural numbers distinct from 0, then f k ~ f; if and only if?k s:.: g~. I[ {Ry and the sequencef satisfies the sentential function x tn t~e l~lVtdual domain a, then the sequence g also satisfies this Junetzan tn the individual domain b.

J From this lemma, with the help of Def. 25, we easily obtain ..emma G which, together with Def. 26, nt, on co gives Th. 8:

L~~MMA G. If the d(MIlI~.~ a and b oj illdi1Jitlul,l~ luu»: lite same

cardinal number and . ., "d 1

• ' X '/,8 a correct sentence tn tlu: lndtm ua

domain. a, then x is also a correct sentence in the ·intlioidual domain b.

According to Th. 8 (or Lemma 0) the extension of the concept 'a sentence which is correct in the individual domain u' depe.nds entirely on one property of the class a, namely on its cardinal number Thi bl I

. s ena es us to neglect in the sequel al

results concerning thi . d

'. s concept, because they can be derive

Immediately from the' h

correspondmg theorems relating to t e classes oi;

VIII, § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 203

With the help of Defs, 24 and 25 the ThR. 1-6 and Lemmas A-D can be generalized by replacing the exprCHHiOTlII 'infinue sequence of classes', 'the sequence ... saiisjie« the sentential [unclion .. .', 'true sentence', and so on, by 'infinite sequence of sul)» classes of the class a', 'the sequence . . . 8Itli.'ifr,l'.s th« senumtial function ... in the individual domain a', 'correct senlrnce in the individual domain a', and so on, rcspcctively. Ai! a consequence of Th. 8 the results so obtained can be extended to sentences which belong to the classes Gtk• In this way we reach, among other things, the following generalizations of Ths. 4-6:

THEOREM 9. For every cardinal number k the class Gtk is a consistent and complete deductive system.

THEOREM 10. For every cardinal number k we have Pr s:.: elk' but ci; cfo Pro

In reference to Th. 10 the following problem presents itself: how is the list of axioms in Def, 13 to be completed, 80 that the class of all consequences of this extended class of axioms may coincide with the given class Gtk? Ths. 1 I and 12 which follow immediately below contain the solution of this problem and also prove that-with respect to the language ofthe calculus of classes -the definition of a correct sentence in a domain with k elements (Def. 26) can be replaced by another equivalent one which is analogous to the definition of provable sentence (Def. 17) and therefore has a structural character.

THEOREJlI II. If k is a natural number, and X the class consisting of all the axioms together with the sentences ex and Yk, then elk = Cn(X).

THEOREM 12. If k is an infinite cardinal number, and X the class con.sisting of all the axioms together aith the sentence ex and all sentences Yl (where l is any natural number), then Gtk = Gn(X).

The proof of these theorems is based chiefly on Ths. 9 and 10 and the three following lemmas:

LEMMA H. For every cardinal number k we have ex E Glk• LEMMA I. If k is a natural number and l a cardinal number distinct from k, then 'Yk E ct, and 'Yk E Ctk, but ;y~ E a, anti 'Y. E ci;

204 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 3

LEMMA K. If XES and X is the class consisting of all the axioms together with the sentence a, then there is a sentence y which is equivalent to the sentence x with respect to the class X and such that either y is a quantitative sentence, or y E Pr or g E Pro

Lemmas H and I are almost immediately evident, but the proof of the very important and interesting Lemma K is rather difficult.!

By means of Th. 9 and Lemma I it is possible from Th. 12 to derive the following consequence which combined with Th. 11 brings out the essential differences existing in the logical structure ofthe classes Ctk according to whether the cardinal number k is finite or infinite:

THEoRE~r 13. If k is an infinite cardinal number, then there is no class X which contains only a finite number of sentences which are not axioms, and also satisfies the formula

Ctk = Cn(X).2

From Lemma I and Ths. II and 12 we easily obtain the following consequences:

THEOREM 14. If k is a natural number and 1 a cardinal number distinct from k, then o; ¢. a, and en, ¢ a;

THEOREM 15. If k and 1 are infinite cardinal numbers, then etk = Ot;

THEOREM 16. If k is an inJinite cardinal number and. x E Ctk, then there is a natural number l ,'rUch tho: xc ef, (in other tl'ords, the Cla.'t8 Glk i.~ included in the Hum of (tll the claesc« Ct,).

According to Ths, 14-1Il (or Lorrunn I) t.horo (lxiHtH for every natural number k a sentence which iH correct in every domain

I In its essentials this lemma is contained in the results to be found in Skolem, Th. (64), pp. 29-37.

• The idea of the proof of this theorem is the same as that of the proofs of Ths. 24 and 25 in article V of the present volume, pp. 78-9. If we take over from the latter Def. 3, p .76, and at the same time extend our present concept of consequence by adding the words '01' x is an axiom' to the condition (ex) of Def. 15, then we could derive the following consequence from Ths. 11 and 13:

In order that the class Gtk should be an axiomatizable deductive system, it ia necessary and B'Ufficient that k be a nalural number.

VIII, § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 205 with k elements and in no domain with any other cardinal number. On the other hand, every sentence which in correct in one infinite domain is also correct in every other infinite domain (without reference to its cardinal number) I1H well all ill certain finite domains. From this we infer that the language ill question allows us to express such a property of classes of individuals I~H their being composed of exactly k elements, where k iH ILny natural number; but we find in this language no means by which we can distinguish a special kind of infinity (e.g. denumcrability), and we are unable, either with the help of a single or of a finite number of sentences, to distinguish two such properties of classes as finiteness and infinity."

By means of Ths. 9, 11, and 12 we can prove

THEOREM 17. If X is a consistent class of sentences which contains all the axioms together with the sentence ex, then there is a cardinal number k such that X s;: Glk; if X is a complete deductive system, then X = Clk•

If we combine this theorem with Ths. 11 and 12, we obtain a structural description of all complete deductive systems which contain all the axioms and the sentence a. It should be noted that the presence of the sentence a is essential here; the multiplicity of the systems which do not contain this sentence is significantly greater and their exhaustive description would not at all be a simple matter.>

The remaining considerations concern sentences which are correct in every individual domain, i.e. belong to the class Ct.

I These results, as well as Th, 19 given below, we owe to Loweriheim ; cf.

Lowenheim, L. (49) (especially Th. 4, p. 459) and Skolem, Th. (64).

• I have occupied myself in the years 192&-8 with problems of this .type, i.e. with the structural description of all complete systems of a given SCIence, in application to various elementary deductive sciences (algebra of logic, arithmetic of real numbers, geometry of straight lines. theory of order, theory of groups); on the results of these investigations, reports were made in the seminar exercises in the methodology of the deductive sciences which I conducted in Warsaw University in the years 1927/8 and 1!128i9. Cf. Pre»burger, M. (61) (especially note 4 on P: 95), and XII, § 5. For a dctatl,~1 discussion of certain closely related problems (611 well QH for further 1,,"110· graphical references} see abo tho recent publicationa of the author, Tarsk}, A. (II") und (li4 '1).

206 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 3 THEOREM IS. In order that x E Ot it is necessary and sufficient that, for every cardinal number k, x E Otk (in other words, the class Ot is the product of all the classes Otk).

This theorem, which is an immediate consequence of Def. 27 and Th. S, can be essentially sharpened bymeansofThs. 9 and 16:

THEOREM 19. In order that x E Ot it is necessary and sufficient that, for every natural number k, x E Otk•

The correctness of a sentence in all finite domains thus entails its correctness in every individual domain.

The following two corollaries are derivable from Ths. 9, 14, and IS:

THEOREM 20. For every cardinal number kwe have Ot ~ Gtk,but oi; ¢ a.

THEOREM 21. The class Ot is a consistent but not a complete

deductive system.

THEOREM 22. Pr ~ Ot, but Ot ct Pr.

This theorem follows from Ths. 10 and 18 and Lemma L:

LEMMA L. 0: E Ot but 0: E Pro

That 0: E Ot follows at once from Lemma Hand Th. 18. The exact proof of the second part of the lemma is considerably more difficult.

THEOREM 23. If x is a quantitative sentence then x E Ot.

The proof, which is based on Lemma I, Th. 18, and Def. 32, offers no difiiculties.

TH~OM;M 24. If X i.~ the clas« ron.<ri.~tinfl of all the (lximTts togetlter with the sentence ex, then: Ot _" Cn(X).

This theorem is most easily proved with the help of 'I'hs. 11, 12, and 18. By using Lemma K we obtain from it at once:

~HEOREM 25. If XES, x E Ot and x E Ot, then there is a quantitatwe sentence y, which is equivalent to the sentence x with respect to the class Ot.

By reference to Lemma Land Th. 24 we notice that we have the following situation: the concept of a sentence which is correct in every individual domain has a larger extension than the

VIII, § 3 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 207 concept of provable sentence, since the sentence ,): belongs to the extension of the first concept but not to that of the second. But if we increase the system of uxioms 'by just this Mingle Ht'ntenCt!.-.:, thc two concepts become Identical in ex tension. Because it seems to me desirable that, with respect. to tho calculus (If classes, the concepts of theorem and of correct sentence in every individual domain should not he distinct in oxtonsion,' I would advocate the inclusion of the sentence <X among the axioms of this science.

The problem still remains of clarifying the relation of the absolute concept of truth defined in Def. 23 to the concepts we have just investigated.

If we compare Defs. 22 and 23 with Defs. 24 and 25 and apply Th. 8, we easily obtain the following result:

THEORE:\1 26. If a is the ClaS8 of all individuals then x E Tr if and only if x is a correct sentence in the domain a; thus if k is the cardinal number of the class a, then Tr = Gtk.

As an immediate consequence of Ths. 20 and 26 we have:

THEOREM 27. Ct s:; Tr, but Tr 1- ci.

If we bring together Ths. 26 and 14 or Ths. II and 12, we reach the conclusion that those assumptions of the metatheory which determine the cardinal number of the class of all individuals (and which do not intervene in the proof of Th. 26 itself) exert an essential influence on the extension of the term 'true sentence'. The extension of this term is different according to whether that class is finite or infinite. In the first case the extension even depends on how big the cardinal number of this class is.

1 This tendency will be discussed in the next paragraph. It should be mentioned that Schroder, although beginning with other Ideas, has made the suggestion of completing the system of hypotheses of the calculus of classes With the sentence" (and even with still other sentences which, how~lVer. 8.R can ea-,~~y be shown, follow in a simple way from the sentence a}; cf. Schroder, E. 16~). vol. 2, Part I, pp. 31S--49. In thL" connexion I may remark tMt It _rTLII U' me that the inclusion of the sentence !I in the' formal" Hy.t<,m of th" .. l~'·h ... of IOj,(ic (of which the calculus of d_ iM an inwrpretation) ,.,(mld not be usofu}, for many jn~r-pretIlJion.Ji. ofthilf Hy",tfJJn a."~ known III whwh tho """nt..l'~ncf'\l in (pu)t1tion iN not ~ntjH;fi6fJ.

208 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 3

Because we can show, on the basis of the system of assumptions here adopted, that the class of all individuals is infinite, Th. 26 in combination with Th. 12 makes a structural characterization of true sentences possible:

VIII, § 3 CONCEPT OF TRUTH IN FORMALIZED LA~GUAGES 200 to construct a formally correct and materially adequate semantical definition of the expression 'true sentence'.

Moreover, by making use of the special peculiarit.iox of the calculus of classes, we have been able to transform thi;! definition into an equivalent structural definition whieh oven yields a general criterion of truth for the sentences of tho language of this calculus.

THEOREM 28. In order that x E Tr, it is necessary and sufficient that x is a consequence of the class which consists of all the axioms together with the sentence a:: and all sentences 91' where 1 is any natural number.

§ 4. THE CONCEPT OF TRUE SENTENCE IN LANGUAGES OF FINITE ORDER

The methods of construction which I have used in the previous section for the investigation of the language of the calculus of classes can be applied, without very important changes, to many other formalized languages, even to those with a considerably more complicated logical structure. In the following pages the generality of these methods will be emphasized, the limits of their applicability will be determined, and the modifications which they undergo in their various concrete applications will be briefly described.

It is by no means my intention, in these investigations, to consider all languages that can conceivably be imagined, or which anyone at any time could or might wish to construct; such an attempt would be condemned to failure from the start. In what I shall say here I shall consider exclusively languages of the san:e structure as those which are known to us at the present day (in the perhaps unfounded conviction that they will form in the future, as they have done hitherto, a sufficient basis for the foundation of the whole of deductive knowledge). And even these languages show such great differences in their construction that their investigation in a perfectly general, but at the same time precise, way must encounter serious difficulties. These differences are, of course, rather of a 'calligraphical' nature. In some languages, for example, only constants and variables occur, in others it is not possible to avoid the use of so-called technical signs (brackets, points. and so on). In some I~nguages symbols of an exactly specified form arc used aM variables, 1i0 that the form of the variablr-s dl'W'mb on HII' part till'_\" play

This sentence could, in its form, obviously be regarded as a definition of true sentence. It would then be a purely structural definition, completely analogous to Def. 17 of provable theorem. But it must be strongly emphasized that the possibility of constructing a definition of such a kind is purely accidental. We owe it to the specific peculiarities of the science in question (to those peculiarities which, among others, have been expressed in Lemma K, which is the most essential premiss in the proof ofThs. 12 and 28) as well as-in some degree-to the strong existential assumptions adopted in the metatheory. On the other hand-in contrast to the original definition-we have here no general method of construction which could be applied to other deductive sciences.

It is worth noticing that by analysing the proof ofTh. 28 and of the lemmas from which this theorem follows we can obtain a. general structural criterion of truth for nil sentences of the l~nguage investigated. From Th. :.!R Much 11 criterion for quantitativo eentencos is eusily derivublo, und tho proof of Lomms K ullows IIH ofr(J(~tivoly to oorrolut« with (JVllI'Y H(JUt,('IHlO of tho lunglln.ge It sentenoo which is equivulent to it, and which, if it is not quantitative, is manifestly true or manifestly false. Au an~logous remark holds for the concept of correctness in a given, or III every, individual domain.

Summarizing the most important results obtained in this section we can say:

We have succeeded in doing for the language of the calculus of classes what we tried in vain to do for colloquial language : namely

210 CONCEPT OF TRUTH IN FORMALIZED LA.~GUAGES VIII, § 4 and their significance. In others quite arbitrary symbols may be used as variables, so long as they are distinguished by their form from the constants. In some languages every expression is a system of linearly ordered signs, i.e. signs following one another in a line, but in others the signs may lie at different levels, not only alongside but also below one another. This calligraphy of the language nevertheless exerts a fairly strong influence on the fO.rm of the constructions in the domain of the metalanguage, as will doubtless be seen from a brief survey of the preceding para" graphs.' For those reasons alone the following exposition will have the nature of a sketch; wherever it takes a more precise form, it is dealing with concretely described languages which are constructed in the same way as the language of the calculus of classes (i.e. languages without technical signs, with variables ?f an exactly specified form, with linear arrangement ofthe signs ill every expression and so on).2

Before we approach our principal task-c-the construction of the definition of true sentence-we must undertake, in every concrete case, the construction of a corresponding metalanguage and the establishment of the metatheory which forms the proper field of investigation. A metalanguage which meets our requirements must contain three groups of primitive expressions: ~~) expressions of a general logical kind; (2) expressions having

e same meaning as all the constants of the language to be dis" cuss?d or which suffice for the definition of such expressions (takmg as a basis the rules of definition adopted in the meta-

~ cr.. for examplo, p. iur, footnote.

ant In order. t? give tho following exposit.ion B complotoly prociso, conorete, f ~ also .'lUfllcLOntly general form, it would lIuJllee if we l.ho"" as tho objoct ~ ~nvoKtlgation, the language of some one complete system of mathomatical t~glc. Such a language can be regarded 88 a universal Ianguege in the sense at ?,ll other formalized languages-apart from "ealligraphical ' differences-

are either frArnnents of 't b' . t

bad' ..... ': I , or can e obtained from it or from Its fragmen s

y ding certain constants, provided that the semen tical categories of these c~nstants (cf: below, pp. 215 ff.) are already represented by certain expresSIOns of the given language. The presence or absence of such constants exerts, which shallsh~w, only a minimal influence on the solution of the problem in ~ t~ we are mterested. As such a language we could choose the la.ngusge be e ~neral theory of sets which will be discussed in § 5 and which might

enrIChed by f' b 'f

m~ans 0 varia lea representing the names of two. and 0

many-termed relat.ions (of arbitrary semantical categories).

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 211 theory); (3) expressions of the structural-descriptive type which denote single signs and expressions of the language considered, whole classes and sequences of such expressions or, flnally, tho relations existing between them. That the expressions of tho first group are indispensable is evident. 'file expressions of the second group enable us to translate every concrete sentence or, more generally, every meaningful expression of the language into the metalanguage, and those of the third group provide for the assignment of an individual name to every such expression. These last two circumstances taken together play an essential part in the final formulation of the desired definition. Corresponding to the three grou ps of primitive expressions, the full axiom system ofthe metatheory includes three groups of sentences: (1) axioms of a general logical kind; (2) axioms which have the same meaning as the axioms of the science under investigation or are logically stronger than them, but which in any case suffice (on the basis of the rules of inference adopted) for the establishment of all sentences having the same meaning as the theorems of the science investigated;' finally, (3) axioms which determine the fundamental properties of the primitive concepts of a structuraldescriptive type. The primitive expressions and axioms of the first group (as well as the rules of definition and inference) n:ay be taken from any sufficiently developed system of mathematical logic; the expressions and axioms of the second gr~up na~urally depend on the special peculiarities of the language mvestIgated; for the third group suitable examples are provided in the ~n:s~ntation of§ 2. It is to be noted that the two first groups ofpnnutIve

• It has already heen mentioned (p. 16&-7) that we are h~re inte~ed ~x-I

I'· .' hi h t 'formal' III a quite speeia

c uSlvely In those deductive sciences weare no . di .

meaning of this word. I have, moreover, brought forward vanous c~n tLOllS-h

f hi h satisfied by the sciences ere

o an intuitive not a formal nature--w IC are . f th

investigated: a. strictly determinate and ~d~~andable :::::;::!f:rence~ constants, the certainty of the axioms, the rehability of the h ng the

An . dooi t i iust the fsct tat. arno

external characteristic of this stan POID IS l . OWl and

primitive expressions and the a.xioms of the metatheory the express. resaions

. F we regard certrun expo ~

aXIOms of the second group occur. or 88 soon ~ obstacle exiBt8

88 intelligible, or believe in the truth .of cert~m sentenc::~ ';::Ies of inference to using them as the need arises. This applies also to h tathoory In

hi h f f the theory to t o me . .

W IC we may if need be. trans er rom II . h in the

the sequel we shall convince ourselves that this need actua y AXIi! .

cases given.

212 CONCEPT OF TRUTH IN FOR~IALIZED LANGUAGES VIII, § 4 expressions and axioms partly overlap one another and in those ca~es in ~hic~ rna thematicallogic, or a fragment of i~, is the object of mvestlg~tlOn (as is the case with the calculus of classes), they even combme to form one group.

The establishment of the meta theory having been completed,

our next ta k j t di ti ,

S IS 0 s mgrush from the totality of expressions

of th~ language the especially important category of eentential.

functton8 and in t' I f

par ICU ar 0 senienee», The expressions of the

language investigated consist of constante and variablee. Among the constants, which are usually finite in number we find as a

rul t"· "

e, cer am SIgnS belongmg to the sentential calculus and the

predicat~ calculus: for example the signs of negation, logical su~, logical product, implication, equivalence, as well as the umv~rsal and existential quantifiers, which we have already met in § 2 I ddit'

. . n a Ion to these we somctimes find other signs

which are connected with the individual peculiarities of the language and denote concrete individuals, classes, or relations; such, for example, as the inclusion sign of the language of the calculus of classes, which denotes a particular relation between

classes of individu lUll h '. .

. a s, sua y t ere are infimtely many van-

ables. According to their form, and the interpretation of the l~nguage, they represent names of individuals, classes, or rela-

tions (sometimes th . I . .

ere are a so variables which represent sen-

tences, .i.e. the so-called sentential variablesj.! Among the e~p~essl~ns which are formed from the signs of both kinds, we

dlBtmgUlsh first of all tl '. . .

. ie pnrmlwe sentential functions, corre-

Rpoluhnp; to tho incl .: f h

" ,1I1'101I!'! 'It. 1 I) t (! cnlculu ... of clll ..... 'leR.i Thc

uxact deserilltioll of t 11(' f f II ' , h

. . orm 0 ,I(lse 1'4'11 tent.iul functions ILIIlI t C

Hpeciflca.t ion of th ' . t iti

. .'. orr III 111 IVO HO(J!Ul will d(I(lOlld upon tho special

peculmntles of tho I . .

. anguago III quesuou. In any case they are

certam complexes of constants which are names of individuals,

classes or relatio d f' .

, ns, an 0 varIables which represent these

1 In many languages 'ver-i h '

OCcur e forrni arrous ot er categorles of constants and variables

, .g, name, orrnmg functors who h . bi

composite exp , b ' IC ,ill com ination with variables, form

ressrons y which names f' d' 'd I I .

are represented (th ' ,0 ill IVI ua s, classes, and re ations

oomplementation e_,gih e word father' ill colloquial language, or the sign of note 1 and p 168 ill .~e3coTmhPlete language of the calculus of classes-c-cf. p. 161,

, , ,no.., e langua 'd '

no signs and e' ges consi ered in the present article contam

xpressions of this kind,

------

VIII, § 4 CONCEP,], OF TRUTH IN FOHMALIZED LANGUAGES 213 names. The first sign of such a complex is always the name of a. class or a relation or a corresponding variable, and is called n (sentence forming) functor of the given primitio« .vI'TIII·1J(.i,1l [unction;l the remaining sigUl-l am called arquments, IIILJIlely l st., 2nd, ... , kth argument-according to the place t.lu-y OC!lllpy. For every constant and variable of the language studied --with the exception of the constants of the sentential culculus arul tho universal and existential quantifiera=-a primitive function can be constructed which contains this sign (the sentential variables, even when they appear in the language, do not occur in tho primitive functions as functors or arguments, but each is regarded as a separate primitive function). Next we introduce the fundamental operations an ezpressions by means of which composite expressions are formed from simpler ones. In addition to the operations of negation, logical addition and universal quantification, which we have met with in§ 2 (Defs. 2,3, and 6), we consider here other analogously defined operations, such as logical multiplication, formation of implications and equivalences, as well as existential quantification. Each of these operations consists in putting in front of the expression considered, or in front of two successive expressions (according to the kind of operation), either one of the constants of the sentential calculus which belongs to the language, or one of the two quantifiers together with the variables immediately following it, The expressions which we obtain from the primitive functions by applying to them any number of times and in any order any of

1 Thus sentence-forming functors which have names as arguments are here identified with the names of classes or relations (in fact the one-argument functors with names of classes and the rest with names of two- or manytermed relations), This interpretation 800mB artificial with the interpretation of the tenn 'functor' which was given by some examples on p. 161, note I; in any case it certainly does not agree with ~he :'J>irit an~ formal structure of the language of everyday life. ·Without going mto deta~18~ It S,"'6mB to me for various reasons to be neither necessary nor useful to distinguish between these two categories of expressions [i.e, sentence-forming functors and names of classes Or relations), Moreover, the whole question is rather of a terrninological nature and is without influence o~ RU~ile<J.U"nt developrnenta. \V" may either regard the definition of functor grven m the text .... purely forma! IL",I disregard the current interpretation of the torrn, or IiO ext<md the mt<:rpr,·t .. , tion of terms I ike I namo of n. plfl.loUi " 4 name of R rf-.l1,t;oH! thllt ws Hwlud,_, exprossions whirh are not nnrnl;,'i in the uHua1 Kfm.o;'~.

f It would he more prop t II h

tary) , In his later bli ti er 0 ca t ese functions fundamental (or elemenpu rea IOns the author refers to them as atomic formulas,

214 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 4 the fun~amental operations, we call sentential functions. Among the variables which Occur in a given sentential function we can distinguish-by means of recursive definitions-free and bound variables. Sentential functions without free variables are called sentences (cf. Defs. 10-12 in § 2).

Next ,:e define yet other concepts which are closely connected WIth the deductive character of the science under investigation, namely the concepts of axiom, consequence, and theorem. Among the axioms we include as a rule certain logical sentences which are constructed in a manner similar to that used for the first kind of axioms of the calculus of classes (cf. § 2, Def. 13). Moreover the definition of axiom depends wholly on

the individual uli iti

. pee Ian res of the science investigated, some-

tImes even on accid tal f t hi . .

'. en a rae ors w ioh are connected WIth Its

histOrIcal development, In the definition of the concept of consequence we follow-mutatis mutandis-the pattern of § 2.

The operations by mea f hi h r

, ns 0 w c we rorm the consequences of a

given ~lass of sentences differ in no essential points from the operations which were' . D f - f given In e. 10. The consequences 0

the axioms are called provable sentences or theorems

After this preIimina k " ,

ry wor we turn now to our principal task

-th,e construction of a correct definition of true sentence. As we

saw III § 3 the method of trueti ,

, cons ruction available to us presupposes

first a definition of aneth f .

. er concept 0 a more general kind which

IS of fundamental impo t r· ". ,

r anee ror investigatIOns In the semantics

of language, I mean th ti f h .., ,

. e no ion 0 t e 8(ltt.,!/actwn of a sentential

[unction. by a sequence .1 of. 'r

' " "oJ J)1'ct8., n the same section I have

attempted to clarify tJ t '. .

". ' ie CUB omnry mnanmg of tlllA nxprcsRlon

III Its ordinary usage I h . J

. ave pomter out that, ill drawing up B

cofrrect d~finition of the concept of satisfaction use can be made

o recursIve definition F thi .

th . . or IS purpose It suffices-recallIng

e recursres, definition f . I' , .

. d 0 sententla function and bearing III

mI~ t~hef intuitive sense of the primitive sentential functions

tan c. e undamental operations on expressions-to establish wo tacts: (1) hi h

ti . w c sequences satisfy the fundamental func-

Ions, and (2) how the f

th Ii' concept 0 satisfaction behaves under

e app catIon of any of th f d

e un amental operations (or to put

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 215 it more exactly: which sequences satisfy the sentential functions which are obtained from given sentential functions by means of one of the fundamental operations, assuming that it has already been established which sequences satil;fy the sententiul functions to which the operation is applied). All soon as we have succeeded in making precise the sense of this concept of satiafaction, the definition of truth presents no further difliculty: the true sentences may be defined as those sentences which are satisfied by an arbitrary sequence of objects,

In earryingout the plan just sketched in connexion with various concrete languages we nevertheless meet with obstacles of a fundamental kind; in fact, just at the point where we try finally to formulate the correct definition of the concept of satisfaction. In order to make clear the nature of these difficulties a concept must first be discussed which we have not hitherto had an opportunity of introducing, namely the concept of semantical category.

This concept, which we owe to E. Husserl, was introduced into investigations on the foundations of the deductive sciences by Lesniewski. From the formal point of view this concept plays a part in the construction of a science which is analogous to that played by the notion of type in the system Principia M athematica of Whitehead and Russell. But, so far as its origin and content are concerned, it corresponds (approximately) rather to the well-known concept of part of speech from the grammar of colloquial language. Whilst the theory of types was thought of chiefly as a kind of prophylactic to guard the deductive sciences against possible antinomies, the theory of semantical categories penetrates so deeply into our fundamental intuitions regarding the meaningfulness of expressions, that it is scarcely possible to imagine a scientific language in which the sentences have a clear intuitive meaning but the structure of which cannot be brought into harmony with the above theory.!

1 Cf. Lesniewski, S. (46), especially pp. 14 and 68; Ajdukiewicz, K. (3), pp. 9 and 148. From the formal point of view the theory .of semantlc:-l categories is rather remote from the original th~~ of types of WhItehead, A: N., and Russell, B. A. W. (90), vol. 1, pp. 37 ff,; It differs less from the nlmphfied theory of types (cf. Chwistek, L. (12), PP', 12-14: Carnap, It, (8), pp. 19-22)

d ' te 'on of tho latter Hegardmg the new" e"prc..w~l In tho h .. ,t

an IS 8.11 eXTUU .' '. .. <)

paragraph of the text, compare the Postacrrpt. to thlH 6ft",]" (I" _fl~J,

216 CONCEPT OF TIll:"TH IN FOR:\L\LIZED LANGUAGES VIII, § 4

For reasons mentioned at the beginning of this section we cannot offer here a precise structural definition of semantical category and will content ourselves with the following approximate formulation: two expressions belong to the same semantical category if (1) there is a sentential function which contains one of these expressions, and if (2) no sentential function which contains one of these expressions ceases to be a sentential function if this expression is replaced in it by the other. It follows from this that the relation of belonging to the same category is reflexive, symmetrical, and transitive. By applying the principle of abstraction,l all the expressions of the language which are parts of sentential functions can be divided into mutually exclusive classes, for two expressions are put into one and the same class if and only if they belong to the same seman tical category, and each of these classes is called a semantical category. Among the simplest examples of semantical categories it suffices to mention the category of the sentential functions, ~~h~r the categories which include respectively the names of individuals, of classes of individuals, of two-termed relations bet:veen individuals, and so on. Variables (or expressions with vanables) which represent names of the given categories likewise belong to the same category.

In ~onnexion with the definition of seman tical category the followmg question arises: in order to estnblish the fact that two given expressions belong to one and the same semantical category, is it necessary to conRider all possible sentential functions which contain one of the ~iv(1n (lXIlrP8Hiofls and to investigate th . I I .

orr )(1 llWlOllr when one of thosn oxprosaiona if! replaced by the

nthor, or does it suffioo to mnko this oh8orvILtion in some or even in only one case? }'rom the standpoint of the ordinary usage of language the second possibility seems much more natural; in order that two expressions shall belong to the same seman tical category, it suffices if there exists one function which contains one of these expressions and which remains a function when this expression is replaced by the other. This principle, which can be called the first principle of the theory of semantical categories, is

1 cr. Carnap, R. (8), pp. 48-50.

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 217 taken strictly as a basis for the construction of the formalized languages here investigated.' It is especially taken into account in the definition of the concept of sentential function, I t also exerts an essential influence on the definition of tho operation of substitution, i.e. one of those operations with the help of which we form the consequences of a class of sentences. For if we wish that this operation, when carried out on any sentence, should always give a new sentence as a result, we must restrict ourselves to substituting for the variables only those expressions which belong to the same semantical category as the corresponding variables. 2 Closely connected with this principle is a general law concerning the semantical categories of sentence-forming functors: the functors of two primitive sentential functions belong to the same category if and only if the number of arguments in the two functions is the same, and if any two arguments which occupy corresponding places in the two functions also belong to the same category. From this it follows that, in particular, no sign can be simultaneously a functor of two functions which possess a different number of arguments, or of two such functions (even if they possess the same number of arguments)

I When applied to concrete languages the formulations gi~en in t~e t:ext-both the definition of seman tical category and the above-rnent.ioned principle-crequire various corrections and supplementations. They are in any case too general, for they also include expressions to which w~ do not usually ascribe independent meaning, and which we often include In the sa,me sernant ical categories to which meaningful expressions belong (for example, in the language of the calculus of classes, the expressions 'N'. 'II x,", and 'AIx,x,',wOluld belong to the same semantical category); in the case of these mearung,eSB

" 'I b h th t even the first prmciple of semantical

exprossrons, It can easr yes own a

categories loses its validity. This fact is of no essential importance for our investigations for we shall apply the concept of semantical category. not to

, '. I ' I t ri bles On the other hand the

composite expressrons, but exc usive y 0 va III • ,

, . th sequel show that the above formula-

examples which we shall encounter ill e, ,

tions admit of very far-reaching simplificatIOns In concrete cases. Th~ to a suitable choice of the signs used in the construction of the expressIOn:' of the language, the mere shape of the sign (and even of the ,CO~POSI~ expression] decides to which category it belongs. Consequently It IS possibfe that ill methodological and semantical investigations concemmg a, concrete Ian-

'I te does not exphcltly occur at

guage, the concept of aemantica ce gory

all. 'I hi t

2 In tho language of the ealculus of cia.",,", an,1 III U"'. angtJ"W·. W I" I

I "hall descrihe in more detail in the Aequel, Hur:h expr<>"'<l,7nf< r-nn only I", nther variublc-s ; t.his explain" the formulation of D"r. 14 m § -,

218 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, §4 in which two arguments which occupy corresponding places belong to different categories.

We require a classification of the semantical categories; to every category a particular natural number is assigned called the ord~r oj the category. This order is also assigned to all expressions which belong to this category.t The meaning of this term can be de~ermined recursively. For this purpose we adopt the followmg convention (in which we have in mind only those languages which we shall deal with here and we take account only of the semantical categories of the variables): (1) the 1st order is assigned only to the names of individuals and to the variables representing them; (2) among expressions of the n+ 1th order, where n is any natural number, we include the functors of all those primitive functions all of whose arguments are of at most the nth order, where at least one of them must be of exactly the nth order. Thanks to the above convention all expressions which belong to a given semantical category have the same order assigned to them, which is therefore called the order of that category. 2 On the other hand the category

1 or, Camap, R. (8), pp, 31-32.

• This classif c t' b' , . h

I a Ion y no means includes all semantic a] categories whic

are to be found in form li ed I " I d

, a IZ anguages. For example, It does not mc u e

se:~e:tlal v~riables and functors with sentences as arguments-Leo signs w IC occu~ m the sentential calculus-neither does it include functors which,

togethor with the corr di , ' I

to f espon mg arguments, form expressions WhICh be ong

r. o~e 0 the categories distinct from sentential functions, such as the nsmeormln~ functor~ mentioned on p. 213, foot.note,

. In view o~ this, the definition of ordor given in tho text could be widened

mthofollowlnO'way'(I)tot} 1 t I b 1 ' I' id 1

1 ,,,,. 10 s orr er . {long H{lnhm"os, numos of me IVI us S

am ,eXprlJftHIOnS ropro.mnting thom , (2) amollg "xprOF'~ion" of t.ho n + Ith order

we inelud« thoso funet)r 'tt t ' I

, . ( "WI I all nr ntrnry DllJllh,'r of nrgullumts of oro or

-;:: n, which togothol' with ttl r " I /' M

b to"" nrg\lmontll rorrn 01<pr"~HHm" 01 orr or ... ,.,

\I are not thomsolvCII ox . r' f I ' "'t'

d P ""HIOllS 0 t 10 tlth (Jr''''r. Evon t.Il1~ ,I" 101 IOn

~ not yet COVOI' 1\11 meaningful tlxpreHHion~ which occur in tho deduotive 8?IOnces. No Bi~8 which' bind' variables fall uruler thi>! dotlnition (thus such s~ 1108 the umversaland existential quantifiers the signH '1:' and 'Ll ' of the ! OOthryrOf sets and analysis or the sign of integration) 8i~ which-in contrast

o e unctors--can be called ' _A,'Qn8

in this' operators. (von Neumann speaks of abstr"""t

classifi c~~e~lOn, see Neumann, J. v. (54).) On the other hand the )8tte~ and kc~:n 18 co~pletely adapted to the system invented by LeSniewski

St e c ad by him in LeSniewski (46) and (47). This system contains no opteera ors eIxcept the universal quantifier which belongs to no sernsn tical

ca gory. may add that" , " ki

system t' , m my VIew, the lack of operators In Lesmews 19

the sen co~s It~te;; a deficiency which restricts its 'universal' character (in

so 0 p. 1, note 2) to a certain degree.

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 219 is by no means specified by the order: every natural number which is greater than 1 can be thc order of many different categories. Thus, for example, both the names of Cl!lt!HCK of individuals and the names of two-, three-, and many-termed relations between individuals arc expresaions of Ow 2nd order.

It is desirable to classify the sentential functions of the language according to the seman tical categories of the free variables occurring in them. We shall say of two functions that they possess the same semantical type if the number of free variables of every seman tical category in the two functions is the same (or, in other words, if the free variables of the one function can be put into one-one correspondence with the free variables of the other in such a way that to every variable a variable of the same category corresponds). The class of all sentential functions which possess the same type as a given function we can call a semantical type.

We sometimes use the term 'semantical category' in a derivative sense, by applying it, not to the expressions of the language, but to the objects which they denote. Such 'hypostatizations' are not quite correct from a logical standpoint, but they simplify the formulation of many ideas. We say, for example, that all individuals belong to the same semantical category, but that no classes or relations belong to this category. From the general law stated above concerning sentence-forming functors we conclude that two classes belong to the same category if and only if all their elements belong to one and the same category. Two twotermed relations belong to the same category if and only if their domains belong to the same category and their counter domains belong to the same category. In particular, two sequences belong to the same category if and only if all their terms rn:long to ~he same category. A class and a relation, or two relations having different numbers of terms never belong to the same category. It also follows that there can be no class whose elemcnts belong to two or more semantical categories; in an analogous way there can he no sequence whose terms belong to disti,nct seman tical cate-

. I I' 'duals arc somctimcs called objects of the 1st order,

goncs. n( IVI '

220 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 4 classes of individuals and relations between individuals objects of the 2nd order, and so on.

The language of a complete system of logic should containactually or potentially-all possible semantical categories which occur in the languages of the deductive sciences. Just this fact gives to the language mentioned a certain 'universal' character, and it is one of the factors to which logic owes its fundamental importance for the whole of deductive knowledge. In various fragmentary systems of logic, as well as in other deductive sciences, the multiplicity of the seman tical categories may undergo a significant restriction in both their number and their order. As we shall see, the degree of difficulty which we have to overcome in the construction of a correct definition of truth for a given concrete language, depends in the first ~lac~ on this multiplicity of the semantical categories appearmg l~ the language, or, more exactly, on whether the expressions and especially the variables of the language belong to a finite or an infinite number of categories, and in the latter case on whether the orders of all these categories are bounded above or not. From this point of view we can distinguish four kinds of languages: (1) languages in which all the variables ~lon~ to one and the same semantical category; (2) languages 10 WhICh the number of categories' in which the variables are included is greater than 1 but finite; (3) languages in which the variables belong to infinitely many different categories b~lt the order of these variables dOCH not exceed It previously given natural number 11; and tinnily (4) lunguages which contain vlLriablcs of arhitl'lLrily high ordor, Wo shnll call the languagos of the first throe kinds langwt(JI'8 of jinite order, in contrast to languages of the fourth kind, the lasujuaqes oj infinite order. The languages of finite order could be further divided into languages of the 1st, 2nd order and so on according to

th hi ."

e ghest order of the variables occurring in the language. By

w~y of supplementation of the sketch given at the beginning of this section of the construction of a metatheory it must be ~oted .here that the metalanguage, on the basis ~f which the lllvestigation is conducted, is to be furnished with at least all the

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 221 seman tical categories which are represented in the language studied. This is necessary if it is to be possible to trunslate any expression of the language into the metalanguugo.!

From the point of view of their logical structure t11!l language« of the 1st kind are obviously the simplest, The language of tho calculus of classes is a typical example. We have SCCII in § 3 that for this language the definition of the satisfaction of a sentential function by a sequence of objects, and hence the definition of true sentence, presents no great difficulties. The method of construction sketched there can be applied as a whole to other languages of the 1st order. It is clear that in doing this certain small deviations in detail may occur. Among other things it may be necessary to operate not with sequences of classes but with sequences of other kinds, e.g. with sequences of individuals or relations, according to the intended interpretation and the semantical categories of the variables occurring in the language."

A particularly simple example of a language of the l st kind which is worthy of attention is the language of the ordinary sentential calculus enlarged by the introduction of the universal and existential quantifiers. The simplici ty of this language lies, among other things, in the fact that the concept of variable coincides with that of primitive sentential function. In the metatheory of the sentential calculus two different definitions can be given of provable theorem, the equivalence of which is in no way evident: the one is based on the concept of consequence and is analogous to Defs. 15-17 of§ 2, the second is connected with the concept of the two-valued matrix. By virtue of this second definition we can easily determine whether any sentence is provable provided its structure is known." If we now construct for this language a definition of true sentence strictly according

I Here-mutat'$ mutandi8-the remarks of p. 21I, footnote, also apply.

• Certain complications, which I shall not discuss here, arise if in addition to variables, composite expressions of the same semantical category also occur in the language investigated; the complete language of the calculus of clnsses which was mentioned on p. 168, note 3, will serve a.. an exampl«, or th" benguage of a system of arithmetic invost.igated in Presburger, .\1. (61) (d. "I""

p. 212, footnote). . .

3 Cf, Hj lbr-r-t, D.; an,1 A"kermann, \\'. (30), I'p. '14 'I:;; I.lIka,'"'''''''I. •. J. 1,01).

1'1'. 1:;4 fT.; TV, § 4.

222 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII. § 4 ~o the pattern given in § 3, we can easily convince ourselves that It represents a simple transformation of the second of these definitions of provable sentence, and thus the two terms 'provable theorem' and 'true sentence' in this case have the same extension. This fact provides us, among other things, with a general structural criterion for the truth of the sentences of this language. The method of construction laid down in the present work could thus be regarded, in a ccrtain sense, as a generalization of the matrix method familiar in investigations on the sentential calculus.

Serious difficulties only arise when we consider languages of more Complicated structure, e.g. languages of the 2nd, 3rd, and 4th kinds. We must now analyse these difficulties and describe the methods which enable us at least partially to overcome them. In order to make the exposition as clear and precise as possible I shall discuss in somewhat greater detail some concrete formalized languages, one of each kind. I shall try to choose exam~les which are as simple as possible, are free from all less essential, subordinate complications and are at the same time sufficiently typical to exhibit the diffieultisr, mentioned to the fullest extent and in the most striking form.

The language of the logic of two-termed relations 'will serve as an ex~mple of a language the 2nd order.l The only constants of this language are: the sign of negation 'N' the sign of logical

sum 'A' d th . ,

.an e umversal quantifier 'IT'. As variables we can

Use theslgn~ 'x:, 'x/, 'x.;'; ... and '~\,', '.X,,', '~\I"·, .... Thcsign composed of the symbol 'x' and of k smallndditional strokes is called the k-th variable of the 1st order, and is denoted by the symbol' , Th .

, ,. Vk • 0 sign analogously constructed with the symbol

X IS called the k-th v . ble .r h . 'v. '

ana oJ t e 2nd order symbolically k'

The variables of the 1 t d '. .. I

s or er represent names of individua s,

~ho~e. of the 2nd order names of two-termed relations between mdiVIduals. From the material and also-in agreement with the further descri t' f I

pnon 0 the language-from the forma

VIII. § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 223 point of view, the signs 'vk' and '~' belong to two distinct semantical categories. Expressions of the form 'Xyz' are regarded as primitive sentential functions, where in the place of 'X' any variable of the 2nd order, and in the place of 'y' and 'z' any variables of the lst order may appear. These oxpressions are read: 'the individual y stands in the relation X to tho individual z' and they are denoted-according to the form of the variables-by the symbols 'Pk.l.m'. By the use of the sign ''"'' from § 2 we specify that Pk.I,m = (~,.... VI)'"' vm. The definitions of the fundamental operations on expressions, as well as those of sentential function, sentence, consequence, provable sentence, and so on, are all quite analogous to the definitions of § 2. But it must always be borne in mind that in this language two distinct categories of variables appear and that the expressions P play the part of the inclusions 'k)' In connexion with

k,l,m • f

the first of these facts we have to consider not one operation 0

quantification (Defs. 6 and 9) but two analogous operations: with respect to a variable of the lst order as well. as with respect to a variable of the 2nd order, the results of which are denoted by the symbols 'n"x', and 'nix' or 'U~x' and 'YkX' respectively. Correspondingly there will be two ope~atlOns ?f substitution. Among the axioms of the logic of relations we mcl~de the sentences which satisfy the condition (ex) of Def. 13, l.~. substitutions of the axioms of the sentential calculus, and umversal quantifications of these substitutions, and al~ all sentences which are universal quantifications of expressions of the

type Uk n; n;"(Pk,J,m·Y+Pk.l.m·Y),

where k, 1, and m are any natural numbers (l =F m) and y any sentential function in which the free variable Ji. does not occur. Considering their intuitive meaning the axioms of the last

category may be called pseudodejinitions.1 .

1 . to Lesni ski who has drawn attention to the necessity

This term we owe .. mew. . of the deductive sciences in

of including pseudodefinitions among the aXIoms. d t admit the those cases in which the formalization of the science ~ DO ) plIo possibility of constructing 'suitable defmitions (cr. p. I~. ~;n~~";ib:"yJ' of

d I bsritute for the Q..lW')m oJ ,.~

definitions can be regar ec 8R a au I I P 65 It It would not Whiwhead A Nand RUSHeIl. B. A. W. (90). vo .• p . . I f to difficult' to' show tho connoxion betwoon these senten..-" .... aJli & group 0 axioms adopted in Neumann. J. v. (54). p. 18.

1 This is a fragment of th 1 d

tiona of which Bre iven i e an.guage of the algebra of relations. the foun a· theleas suffi t g m SchrOder, E. (62). vol, 3-a fragment which neverces 0 express every idea which can be formulated in this language.

224 CONCEPT OF TRUTH IN FOR"L\.LIZED LA...~CUAGES VIII, § 4

To obtain a correct definition of satisfaction in connexion with the language we are considering we must first extend our ~owledge of this concept. In the first stage of operating with It we spoke o~ the satisfaction of a sentential function by one, tw~, three objects, and so on, according to the number of free variables o~curring in the given function (cf. pp. 189 ff.). From the semantICal standpoint the concept of sa tisfaction had there a strongly ambiguous character; it included relations in which the number of terms was diverse, relations whose last domain was a class of sentential functions, whilst the other domains-in the case of the language of the calculus of classea=-consiated of

objects of one and th . .

· . e same category, namely classes of indi-

vldua~s. Stnctly speaking we were dealing not with one concept,

but WIth an infinit b f .

· . e num er 0 analogous concepts, belongmg to

ddferent semantICal categories. If we had formalized the meta-

Ia '

.n~uage It would have been necessary to use infinitely many

dist~ct. terms instead of the one term 'satisfies'. The semantical

amblD'l11ty of thi . .

b~ S concept illcreases still more when we pass to

lang_uag~s of more complicated logical structure. If we continue

the mtwtive co id ti f .

nSI era Ions 0 § 3, analyse the examples gIven

there and constr t .

uc new ones after the same pattern, It soon

becomes clear that tri t ' .

th ' as ric semantioal correlation exists between

~ free vanables of the sentential function and the 0 bjects which

satIsfy these function' . f .

, s: every ree variable belongs to the same

semantlCal category th f' .

. as e name 0 tho object corresponding to

It, If, therefore at I, t t Iiff

. ,cas wo ( 1 . eront categories occur among

tho vanables of the JUI . I ., .

· IgUl1go ... as III t ie case we nre investignting

It does not Huflic(J t t ' t '. . I

. 0 rOH rIc c(JWu<ierahon to only U Rlllg e

category of ohje('(,s' I I' , I '.

'1'1 d ' ' , , 111 ( {'It IIIg wif. 1 tho concept 1)1' sat.isfaction.

ie ornams of the . I I' .

, . ~l1ng e re atlOlls which 111'0 covered by the

term satIsfaction' th

, us cease to be semantically unambizuoue

(only the last domai . t . 0 •

f. III consrs s as before exclusively of sentential

WIetIOns). But si th' .

nl mce e semantlCal category of a relation not

o y depends on the number of domains i e the number of

terms standing' th I' ' "

. ill ere ation to one another but also on the

categories of th d . '

t· r· ese omains, the category of the concept of

sa Is.actIOn or th th

' rarnsr e category of each single one of these

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LAXGUAGES 225 concepts, also depends on two circumstances. It depends on the number and also on the categories of the free variables which appear in the sentential functions to which the concept of satisfaction relates. In brief, it depends 011 what we have called the semantical type of the sentential function. To functions which belong to two distinct types two semunt.icully cliHLlneL concepts of satisfaction always correspond.! Some cxarnplcs will make this clear, We shall say that the objects R, a, and b satisfy the function Pl,Z,3 if and only if R is a relation and n an~1 b are individuals and we have aRb (i.e, a stands in the relation R to b). The function Pl,2,2,P3,2,2 is satisfied by the objects R, a, and S if and only if Rand S arc relations, a is an individual and

we have both aRa and aSa. The function n;n;(Pl,2,a+PI,3,Z) is satisfied by symmetrical relations and only by them, i.e. by relations such that, for all individuals a and b, if we have

aRb we also always have bRa. The function n;(p~.2,3+PI,3,Z) is satisfied by those and only those individuals a and b which satisfy the following condition: for every relation R, if a Rb, then bRa, i.e. individuals which are identical. In the above examples we have sentential functions belonging t~ four diff:rent semantical types, and we are, therefore, dealing with four different relations of satisfaction, in spite of the fact that the number of free variables and also the number of terms in the relations is the same in the first two examples.

The semantical ambiguity attaching to the concept of s~tisfaction in its original conception renders an e~aet ch~ractenzation of this concept in a single sentence, or even m a finite number of sentences, impossible, and so denies us the u~e. of the only method so far known to us of constructing a definition of a true sentence. In order to avoid this ambiguity, in dealin~ with the calculus of classes we had recourse to an artifice which 15 used by logicians and mathematicians in similar situations. Instead .of using infinitely many concepts of satisfaction of a sentential

1 . • f . of one sernant.ical type can (;O"""p,>",1 to w-vr-r a 1

Moreover, unctions .' rovi,Jn,1 tho fr'''', vanabl,,,; of

semantically distinct concepts of sat.isfact.ion , p. . ,. I r

. t t !<-lIBt two di"tinct ""mantI""l c·llu1(0rl"". In '" , I-

these funct-ions belong .o a, '. f the vllri"bJe,. tl",ir BrnHIj(',,,,,·,,t

tion to tho number and the ca\RgonoH 0 '

must ah~o be taken into eOfl!;ir).·rntion.

226 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, §4 function by single objects, we tried to operate with the semantically uniform, if somewhat artificial, concept of the satisfaction of a function by a sequence of objects, It happened that this concept is sufficiently more general than the previous one to include it-intuitively speaking-as a special case (to define the logical nature of this inclusion would, however, be a little difficult), It will easily be seen that this method cannot be applied to the present problem without further difficulty. Satisfaction in its new form is a two-termed relation, whose domain consists of sequences and counter domain of sentential functions, As before, there exists between the free variables of a sentential function and the corresponding terms of the sequences which satisfy it, a strict semantical correlation. Thus if the language of the logic of relations contains variables of two different semantical categories, we must likewise use two categories of sequences in our investigations, For example, the function n~ n;(Pl,2,a+PI,3,2) is satisfied exclusively by sequences of twotermed relations between individuals (namely by those and only those sequences F whose first term FI is a symmetrical relation), But the function n~(pI,2,3+Pl,3,2) is satisfied exclusively by sequences of individuals (i.e, by sequences 1 for which 12 = i3 holds). The domain of the relation of satisfaction and eo ipso the relation itself thus again becomes semantically ambiguouS, Again we are dealing not with one, but with at least two different concepts of satisfaction. But still worse, a closer analysis shows that the new interpretation of the concept of satisfaction can no longer as a whole be maintained, For one and the same sentential function often contains free variables of two different categories, To deal with such functions wo must operate with sequences whose terms likewise belong to two categories. The first term, for example, of the sequence which satisfies the function PU3 must be a relation, but the two following ones must be individuals, But it is known that the theory of semantical categories does not permit the existence of such heterogeneous sequences, Consequently the whole conception collapses, Thus changing the original interpretation of the concept of satisfaction has removed only one subsidiary cause of its ambiguity, namely the

T OF TRUTH IN FORMALIZED LANGUAGES 227

VIII, § 4 CONCEP ,

f t in the relations which arc

diversity in the number 0 othe: f more important factor,

, f th oncept: anot er ar

the obJect 0 e eoncepv, f the relations haa lost

the semantical diversity of the terms 0 '

none of its force, d i § 3 can be applied to t.he

Nevertheless the meth~ds use lIhn h with certain modili(~IL-

bei . stIgated a t oug

language no~ eing mve, , o~ible to find an interpretation of

tions, In this case, also ~t IS. P which this notion loses its scrnanthe concept of satIsfactIOn m , b comes so general that it tical ambiguity and at the sam he tll~e, el concept In fact, two

, 1ft e ongma '

includes all Specia cases 0 1 I h 11 call them the method of

h d vailab e: sa. .

different met 0 s are a '1.~.1.f mantical unijicatwn

d the metlt(J(N OJ se

many-rowed sequences an

oj the variables, , e should treat satisfaction

The first method reqUIreS that w d relation which holds

db t a three-terme

not as a two-terme , u as. I b tween sequences of two-

f individua s, e th

between sequences 0 ti 1 functions, We use e

termed relations and between senten ia f of individuals

, ,'the sequence .

following mode of expressI~n, h atisfy the sentential

and the sequence F of relatIOns toget er S easily be visualized f thi phrase can

function x'. The content 0 IS F mple the sequence f

pIes or exa, , f

by means of concrete exam . f I tions together satlS y

d th equence Fore a ds i th

of individuals an e s h ' d'vidual f2 stan m e

if d only if tern I al

the function PI,2,3 1 an I der to formulate a gener

relation FI to the individual 13'. nhor anner of Def. 22 in § 3,

d ctly III t em

definition we procee exa . the language we are con-

ember that, III , .' nten-

care being taken to rem th part of prIlllltlve se

sidering the expressions Pk),m plaYf e operation of universal

, h t . stead 0 one , . f

tial functions and tam , ecur The definitIOn 0

I ted operatIOns 0 '

quantification two re a to Def, 23.

true sentence is completely analdiofig~: some extent by treating

w be mo e lled two-

This method can no , between so-ca

termed relatIOn d ed pair

satisfaction as a two- , ti ns. Every or er

d tentlal fune 10 (}1.I;ed

rowed sequences an sen 1 d F is called a uco-r

, 'f t 0 sequences an fthesequence

WhICh conSIsts 0 w , ) where the kth term 0 d

sequence (or two-rowed ~tnx, h kth term of the first or secon 1 or of the sequence F IS called t e. 1 sCCJllcnec, J n the present

f th two-rowel -

row respectively 0 e

228 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES vnr, § 4 case we have to deal with ordered pairs which consist of a sequence of individuals and a sequence of relations, It is easily seen that this modification is a purely formal one and has no essential effect on the construction as a whole. It is to this modification of the method that the term 'method of manyrowed sequences' is adapted.

To understand the method of seman tical unification of the variables we begin with certain considerations which are not immediately connected with the language we are at present investigating. It is known that with every individual a a definite two-termed relation a* can be correlated in such a way that to distinct individuals distinct relations correspond. For this purpose it suffices to take as a * an ordered pair whose terms are identical with a, i.e. the relation R which holds between any two individuals band c if and only if b = a and c = a. On the basis of this correlation we can now correlate in a one-one fashion with every class of individuals a class of relations, with every manytermed relation between individuals a corresponding relation between relations, and so on. For example, to any class A of individuals there corresponds a class A * of all those relations a'" which are correlated with the elements a of the class A. In this way every sentence about individuals can be transformed into an equivalent sentence about relations.

Bearing these facts in mind we return to the language of the logic of relations and change the intuitive interpretation of the expressions of t.his languagc without in any way touching their formal structuro. All constantR will retain their previous meaning, whilHt all variableH hoth of tho IRt nru] 2nd order are from now on to reprcHent names of two-tent1mi relutions. '1'0 the primitive sentcntial functions of tho typo 'Xyz', whore instea~ of 'X' some variable l'k and instead of 'y' and 'z' any two vanabIes VI and Vm occur, we assign the following meaning: 'there exist individuals a and b such that a stands in the relation X to b, y = a *, and z = b*.' In this way the meaning ofthe composite sentential functions will likewise be modified. It is almost immediately evident that every true or false sentence in the earlier interpretation will remain true or false respectively in

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 229

h B virtue of this ncw interpretation all the vari-

t e new one. y ,

ables ofthe language now belong to one and the same se~lln~l~al

t . deed from the formal but from the mtuit.ive category, no mi'

. t f iew: they represent words of the same 'part of RIWCC I .

pom 0 VIe " I' t.i

Consequently the language we are considering can )e mvos I-

t d by exactly the same methods as all hmgllagl'K of tho IHt ~n~; in particular, satisfaction can be treated a.'! I~ tw()-t('~med relation between sequences of relations an:lllententmi functions. At the same time a complication of a techmcal natllre-~lthough

. rt nt one-presents itself. Since two frec variables of

an ummpo a d V. ay occur in

different orders but the same indices, e.g. v, an I' ill

the same sentential function, it is not clear without supplemen~ t r sti ulations which terms of the sequence are to correspon t: ~e v~riables of the Ist, and which to those of the 2nd order.

thi difficulty we shall stipulate that to every To overcome s I uence with an uneven index 2. k-I

variable vk a term of the seq , ble V a term with even index 2. k corresponds, and to every varia k F of relations satisfies

d For example the sequence

correspon, s. 'f and ~nly if there are individuals a and b the function Pk,l,m 1 , F. t b F = a'" and

h t ta ds in the relation 2,k 0 ,:U-l "

such t a a s n hi d t '1 the definitions of satisfaction

F - b* Apart from t IS e ai .

2,m-l -. diff in no essential point from the defini-

and of true sentence er I

tions given in § 3. ibed be a plied to all languages of

The two methods describe can p studied belong to

th 2 d kind 1 If the variables of the language .. d

en. . we regard satisfaction-vun er

n different semantical categones, n+ I-termed

owed sequences-as an

the method of many-r of the corresponding

h ldi b tween n sequences ed

relation 0 mg e .' Lfuncti ns or as a two-term

. d the semantica unc 10 , ed

categories an , ' ts of n-rowed sequences (i.e. order

relation whose domam consIS ' , ' bles occur which are not

' Ian es in which varia d I ith

1 This holds even for guag f 218 note 2). I shall not ea WI

included in the classificat~ononp. 218d(,';B !:ities 'which may occur here. But certain (not particularly unport~t) \hC t sentential variables, even If they I take this opportunity of mentlOnmlg ; the construction at all, and thatf.

do not comp rca h . the process 0

occur in the language, h -hile to include t em to

in particular, it would not be won hi h contain such varrables can ,L<, ex-

ifi ti Sentences w IC Ii hi an eouivulent

eementical urn ca IOn. h in one- many M ron, -'S I

cluded by correlating with each ,of t e:;'tial variables (cr. Hilbert, D., &Dc sentence which does not con tam !!en

Ackermann, W. (30). pp. 84--85).

230 CONCEPT OF TRUTH I

. N FORMALIZED LANGUAGES VIII, § 4

n-tuples of ordmary sequences)

consists of sentential f . and whose counter domain method form th unctIOns. Constructions based on this

emostnaturalgener Ii ti f

in § 3 and th . . a rza IOn 0 the constructions

en matenal corr t

doubts. ec ness appears to leave no

In applying the method of . '.

variables the ohoi f h semantlCal umficatIOn of the

, oice 0 t e unit, .

part, i.e. that sema ti I 'Jytn{J category plays an essential

n rca category' hi h

the language studied can be' m w c all the variables of

required of the unif . mterpreted. Only one thing is seman tical categor yIn~ .c~e~ory: that with all objects of every the given languag: wfi IC . IS represented by the variables of can be correlated . ' e ective objects of the chosen category

m a one-one f: hi .

objects distinct obt t as IOn (i.e. so that to distinct

, jec s correspond) N h .

of the unifying cat. . evert eless, the choice

di egory IS not alway . I .

scussed above' . s so simp e as m the example

m connexlOn with th I

relations' this ch . e anguage of the logic of

, oice cannot al b

which OCCur in th I ways e made from the categories

the language repr: antguage. If, for example, the variables of

. . sen names of tw te d .

mdividuals and n f I 0- rme relations between

. ames 0 c asses who h .

vlduals then th . I IC consist of classes of indi-

, e SImp est unlf .

category of two-term d I . ymg category seems to be the

I do not propose t et r~ ations between classes of individuals.

[i 0 en er into a furth I' .

It would presup kn er ana YSIS of this problem

set theory) I adPdose la owledge of certain facts belonging to

. on y the follo .

category cannot b fl wmg remarks: (1) the unifying

he 0 ower order tha

t ose OCcurring I'll th I ' n anyone category among

, e anguage' ('» f

kind It unifying c t ' .. or every Inngllage of the 2nd

It egory CILn hI' f(lunI '. .

categoritls and in f t ' . even mhnttely many such

• 110 nmong catt1g . ft'

IS the highest 0 d f .' ones 0 t ie nth order, where n

r er 0 the variabl .

As SOOn as the unif in . es Ocourrmg in the language.

sentential funCtI'O y g category IS specified, and the primitive ns correspondin I .

course of the wo k d g Y mterpreted, the further

r oes not diff

construction used f 1 er at all from the methods of

I or anguages of the 1st kind

n contrast to the meth d f .

no doubt that the d 0 0 many-rowed sequences, there is

eecon method' ha

theless the definit' IS somew t artificial. Never-

Ions constructed by this method prove, on

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 231 closer analysis, to be intuitively evident to a scarcely less degree than the constructions based on the first method, At the flame time they have the advantage of greater logical simplicity. In particular, when we are dealing with the definition of trill! Hentence the proof of the equivalence of its two form u III tion« prmHlIltH no difficulty in any concrete case. The easentlal ndvantagca of the method of unification of the variables only become clear, however, in the investigation of languages of the 3rd kind, since the method of many-rowed sequences here proves to be quite useless.

As a typical example of a language of the 3rd kind we choose the language of the logic ofmany-termedrelatioos.1 In this science We deal with the same constants 'N', 'A', and 'II' and with the same variables of the Ist order Vk, as in the logic of two-termed relations. But we also find here variables of the 2nd order in greater multiplicity than before. As variables of this kind we shall use such signs as 'X:', 'X~', 'X:II""" 'X:', 'X;', 'X~,', ... , 'X;"', 'X;~', 'X;:;', ... andsoon. ThecompositesymbolcoIL'ltructed from the sign 'X' with k small strokes below and lliuch strokes above will be called the kth variable functor u,-ith I arguments, and denoted by 'V1'. Intuitively interpreted, the variables vk represent, as before, names of individuals, whilst the variables VL represent names of l-termed relations between individuals, in particular for 1 = 1 names of one-termed relations, i.e. names of classes. Both from the intuitive and the formal points of view the signs vk, Vt, Vi, ... belong to infinitely many distinct semantical categories of the Ist and 2nd orders respectively. The primitive sentential functions are expressions of the type 'Xxy ... z ', where in place of 'X' any variable functor with 1 arguments and in place of 'x', 'y', ... , 'z' variables of the 1st order, 1 in number, occur. These expressions are read as follows: 'the I-termed relation X holds between the 1 individuals x, y, ... , z.' According to the number and form of the variables we denote the primitive functions by the symbols 'Pk.m', 'Pk.m.,.', .. ·,

1 This is 1\ language which resembles the langusgo of the lower. predir"w calculus of Hilbert, D., and Ackennann, W. (30), pp. 43 fl., but IK richer than tho Illtwr because variable functors can occur in it both lUI (roo and ..... bound varinbles.

232 CO~{cEPT OF TRl}TH IX FOR!\IALIZED LANGUAGES VIII, § 4

putting p - VI ~ l' (F? ~ ) ~

k.m --- k "I' Pk,m,n = riC vm Vn, and so on. In

order to obtain a unified symbolism, which is independent of the number of variables, we shall use symbols of the type

'I , (h "

Pk,p were p represents the name of a finite sequence of

natural numbers), the meaning of which is determined by the f~r~ula plp = (((V~ -'"' vpJ '" t'p.r--' ... ) -'"' t'p/ The further defilllho~~ of the m:tatheory do not differ at all from the analogous definitIOns relatmg to the logic of two-termed relations and even ~o the calculus of classes. As operations of quantification we mtroduce quantification with respect to the variables v and the

'W ~ k

varia es k and denote the result of the operations by the

symbols Tls=' and 'n~x' respectively. The list of axioms includes those which satisfy the condition (ex) of Def. 13 of § 2,

and pseudodefi . ti hi h f

III ions W Corm a natural generalization of

the .pseudodefinitions from the logic of two-termed relations. Their more detailed description seems to be unnecessary,

. We turn n~w to the problem of how the concept of satisfaction IS to be conceIVed and the definition of truth to be constructed for the language we are now considering. Any attempt to apply the meth~d of many-rowed sequences in this case fails completely. In this method the term 'satisfaction'-in whatever formexp.resses the relation of dependence between n sequences of varIOUS categories and the sentential functions, where n is exactly equ.al to the number of semantical categories represented by the variables of the given language. In the case We arc investigating th~ IHlm~er n is indefinit.ely lnrgo and tho metalanguage we arc u8mg:-lIke nil other actunlly (1xisting formalized IanguageM--cprovJ(le,'{ no moans for <!o:tling with tho relation of

mutual d(ll)(lnJclI(,(, hutw I'· I . I 1 - . . I

. . ...~ (Jell () 'Jpe..,.1 w lie I lulong to infinite y

many distinct semanticaJ cat(1gorie8. z

1 Strictly ki h

. I spea mg t e meaning of the symbol 'pI 'should be defined

recursive y. k.p

2 In those cases in which . I . I

deal with th tid' in ogica and mathematical constructions, we d te . d e mu ua ependence between an arbitrary not previously

e rmme number of objects f d h '

mostly use di 0 one an t e same semant.ical category, we

f di t' t or mar~ sequences. For objects which belong to a finite number

o IS me categones many d .

But in th k f . -rowe sequences fulfil the analogous function.

e nown ormahzed lang find - . ith

infinitely many , ( f di uages we nothmg like 'sequences WI

rows 0 istinet seman tical categories).

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED I,ANGUAGES 233

The method of semantical unification of the varia bles can, however, be applied to this language with complete 811C(~et'H. To sec this it suffices to note that we can correlate in It one-one fashion, with every n-termed relation R between individuuls, a dUHH u* which consists of n-tcrmcd sequences of individuals, IHlIIIPly Hill class of all sequences I which satisfy the following condition: the relation R holds between the individuals t.. 12"'" In' For example, the class of all sequences f with two terms II and 12 such that 11 Rf2 corresponds to the two-termed relation R. Consequently every sentence concerning many-termed relat.ions can be transformed into an equivalent sentence which asserts something about classes of sequences. It will be remembered that by sequences of individuals we mean two-termed relations between individuals and natural numbers. Accordingly all sequences of individuals, whatever the number of their terms, belong to one and the same semantical category and therefore the classes of these sequences, in contrast to many-termed relations, likewise belong to one and the same category.

On the basis of these considerations we now partially unify the semantical categories of the variables in the following way. To the variables Vk we give-at least provisionally-the same significance as before. But the variables Vi now represent the names of any classes which consist of finite sequences of individuals or of other objects of the same category (i.e. the names of objects of at least the 3rd order, according to the order which we assign to the natural numbers).' The primitive functions of the form 'Xxy ... z', which begin with a functor with 1 arguments and hence contain l variables of the Ist order, are interpreted by phrases of the type: 'the sequence of individuals the first term of which is x, the second y, ... and the lth (the last) is z, belongs to

1 In systems of mathematical logic, e.g. in Whitehead, A_. N., an~ RUBSell, B. A. W. (90), vol. 2, pp. 4 ff., the cardinal numbers and m particular the natural numbers are usually treated as classes consisting of c1~ of individuals (or other objects), namely as the classes ~f all those classes which arc similar (in the Principia Mathematica sense) to a gIven class. F or example, the number 1 is defined as the class of all those clesses which have exactly one element. With this conception the natural numbers are thus objects of ( .. t least) the 3rd, sequeneee of individuals of the 4th, and cl_ of th_ llequenc08 of the 6th order.

234 CONCEPT OF TR1JTH IN

FORMALIZED LANGUAGES VIII. § 4

the class X which . t

intuitive althou h consis s of i-termed sequences'. From the

from no~ on stilgl bnolt from the formal, standpoint, the variables

e ong to only t di t'

gories; in view of this . wo IS mct semanticaI cate-

course of our work th CIrcumstance we can use, in the further

gating languages of t:es:: :~~Ods as we employed in investi-

By means of the ph . 'th

sequence F h rase. e sequence! of individuals and the

, w ose terms form cl f fi .

individuals t th' asses 0 mte sequences of

, oge er sahsfy the . .

can brin . t. given sentential function' we

g mto sel'V.!ce the meth d f •

To use this c t' 0 0 many-rowed sequences,

oncep conSistently . fi

correlatio b t we must rst set up a one-one

n e ween the van bl VI d

sequence ]I' . h a es k an the terms of the

In suo a way th t t .

correspond t diff . a erms with different indices

Putti 0 1. erent vanables. This is most easilv done by

ng every vanable VI . J

the index (2. k-I) 21-1 k ill correspondence with a term having

F. }I' . • For example, the terms F. F: F. F.

6' 4" .. correspond to th . bl l' 3' 5' 2'

W· h e varia es VI V! VI V2 V2 V3 1

It this con t· h 1, 2, 3, 1> 2. 1 .... ·

above phra .ve~ Ion t e establishment of the meaning of the se m Its ap Ii t'

tion, and even the p ca IO~ to any concrete sentential func-

concept in . constructIOn of a general definition of the concerning ;huestlO.n" ~resents no further difficulties. Thus

e pnmitive functi h

sequences! d F f IOns, t ose and only those

an (0 the cat . .

satisfy the fi t' egones given above) will together

th unc IOn Pk,m which satisfy the following condition'

e sequence g of individ I h . .

with I hell) t tL ua s, W oso smglo term (It is identical

J "" ngs 0 110 ChtSHF' J

functions! and jI"1I ' 2.k--t...1l an analogous way, those

satisfy Jil" Ii II . WI together Hlttllilfy the function Pk III which

,I ". I) OWUlg <'onliif' . tl .. n

two terms h .' non: ,1('Il(lqUllIwn (/ of individunls with

, W lift) y.. f 1\ d r

F. r ~ -- Jm n (/2 J". IlllltmgH to tho olaes

(2.k-I).2· n general In orde th t th

together sati f th' f . rae scquencos f and }t' should

s y e unctIOn I it I

that the sequ f Pk,p' 1 18 necessarv and sufficient

g =]. ence g 0 individuals with l terms: where g = f ,

2 - P,"'" g, = !PI' should bel on 1 .p,

consists of sequences with th g to the class Ji(2.k-I).21-1 (which

1 Instead of th fi' e same number of terms).

functionf(~. l) W;;Ch~C::Zl:t!kt~) = (2.k-1). 21-1 we could use any other ordered. pairs of natural numbers e natural numbers in one-one fashion with correlations; cf. Fraenkel A (16)' Set theory offers many examples of such

,. • pp. 30 fl. and 96 fl.

VIII. § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 235

If we wish to apply the method of unification of the variables we again make use of the fact that It one-one correlation can be set up between any individuals and certain cll1HHeH of finite sequences, and in such a way that to overy individual a there corresponds the class a* containing as its only element a sequence whose only member is just the givcn individual. Beginning in this way we next modify the interpretation of the variables of Ist order in exactly the same direction in which we formerly modified the interpretation of the variables of the 2nd order. The primitive functions of the form 'Xxy ... e ', containing 1+ I signs, we now regard as having the same meaning as expressions of the type 'the I-termed sequence g of individuals which satisfies the conditions: gt = x, gi = y, ... , gi = z, belongs to the class X, which consists of sequences with I terms", With this intuitive interpretation all variables now belong to the same semantical category. The further construction contains no essentially new features and the reader will encounter no serious difficulties in carrying it out.

The method of semantical unification of the variables can be applied with equal success to the investigation of any language of the 3rd kind. Determining the unifying category may sometimes be more difficult. As in the case of languages of the 2nd kind it is here impossible to restrict consideration to categories occurring in the language studied. In contrast to those Janguages it is never possible to make the choice from among the categories of one of the orders represented in the language. This difficulty is not, however, essential and exclusively concerns languages of the lowest order. For it is possible to prove that for those languages in which the order of the variables does not exceed a given number n, where n > 3, any category of the nth order can serve as the unifying category.

In this way the various methods at our disposal enable us to define the concept of satisfaction and with it to construct a correct definition of truth for any language of finite order. We shall see in the next section that these methods do not extend further; the totality of languages of finite order exhausts the domain of applicability of our methods. This is therefore the place in which

236 CONCEPT OF TRUTH IN FORl\B.LIZED LANGUAGES VIII, § 4 to summarize the most important consequences which follow =: the definitions we have constructed. t

Flr~t, the definition of true sentence is a correct definition of

truth zn the sense 01' conrention T of § 3 It b . 1

• 'J • em races, as specia

cases, all partial definitions which were described in condition

(0:) of this convention and which elucidate in a more precise

and te . II

rna ria Y correct way the sense of expressions of the

type. 'x is a true sentence'. Although this definition alone provl~es no general criterion of truth, the partial definitions ment~oned do permit us definitely to decide in many cases the questIon of the truth or falsity of the sentences investigated.

In particular, it can be proved--on the basis of the axioms of the s~cond group adopted in the metatheory (cf. p. 211 )-that ~ll ~xwms of the science under investigation are true sentences. In a similar manner we can prove, making essential use of the fact that the rules of inference employed in the metatheory are not ~ogically weaker than the corresponding rules of the science Itself, that all consequences of true sentences are true. These two facts together enable us to assert that the class of true sentences contains all provable sentences of the science investigated (cf. Lemma D and Ths. 3 and 5 of § 3).

~mong the most important consequences of a general nature w~c~ follow from the definition of truth must be reckoned the prznczple of contradiction and the principle of the excluded middle.

These two theorems, together with the theorem on the consequences of true sentence.'! already mentioned, show that the

clas» or all true ~('.ntenc~oji' .

'J ' ' • . "0 orm.~ a conmsten; and cornplrte dcduct,ve

8!/8trm ('!'hs. I, 2, and 4).

AH an irnnwdiltto ultl I . .

, • loug 1 It Homowhat subsidiary, con-

sequence of these facts we obtain tho theorem that the class of all

pro~vable sentences likewise forms a consistent (although not neoes-

sarily complete) ded tin .

uc ne system. In this way we are able to

produce a proof of the . t '. . h

consis ency of vanous sciences for whic

we can construct definitions of truth. A proof carried out by

t Some further conseque f his f

the author 'On d id bI nces 0 t type are discussed in the article 0

concept of truth ~ eCI a e statements in enlarged systems of logic and the particular sect 9' oulIma1 I of SymbOlic Logic, vol, 4 (1939), pp. 105-12; cf. in

. ,po •

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 237 means of this method does not, of course, add much to our knowledge, since it is based upon premisses which are at least IlH strong as the assumptions of thc science under invest.igat.ion.! Nevertheless it seems to be worthy ofnotc that such It gcnerul met.hod of proof exists, which is applicable to an extensive range of deductive sciences. It will be seen that from the deductive standpoint this method is not entirely trivial, and in many cases no simpler, and in fact no other, method is known. t

In those cases in which the class of provable sentences is not only consistent but also complete, it is easy to show that it coincides with the class of true sentences. If, therefore, we identify the two concepts-that of true sentence and that of provable sentence-we reach a new definition of truth of a purely structural nature and essentially different from the original semantical definition of this notion.s Even when the provable sentences

1 As Ajdukiewicz has rightly pointed out in a somewhat different, conncxio.n (cf. Ajdukiewicz, K. (2), pp. 39-40) it does not at all follow from this that tbis proof is not correct from the methodol?gical .standpoint-~hat It cont~illa ill some form a petitio principii. The assert ion which we prove, i.e. the consietency of the science, does not occur in any way among the hypotheses of the proof.

• In the course of this work I have several times contr~u.~1 sernan t.icul definitions of true sentence with structural definitions. But t.hlH does not mean that I intend to specify the distinction betw~n the two kinds of ,defLIlItLOn,~

. t From the ill' tuitive standpoint these differences seem to be

ill an exac way. , , . h

tolerably clear. Def. 23 in § 3--as well as ot?~r definitions ,coru;tructed in t .e same way-I regard as a semantical definition because lJl a certain sense

t I . ith the problem discussed in the last three paragraphs see n connexion WI , , '" H (87 ) F om

the recent publications: Mostowski, A. (53 e) ~ well as , ang, ,: C ~~ed

tho results of these authors it is seen that ill some cases, hav tn,g su~ t

in constructing an adequate definition of truth for a theory T ill rts me a, theor ,we may still be unable to show that all the provable sentences of T are t~e in the sense of this definition, and hence we may also be unable~o car out the consistency proof for T in M, This phenomenon can roug y be ~plained as follows: in the pro?f tha~ a~1 prova~le se.nt~nc~ of ~ a; true a cortain form of mathematicallOductlOn 18 essentially ill' ~I~d' ' ,anf t~'~ formalism of M may be insufficiently powerful to secure the 'a , It) 0 L, inductive argument. Hence a certain clarification of the 835umptl~.1l3 (on PIP'

. dati f the metatheory may be desirahle. n

174 ff) concerrung foun ations 0 h '1

. 'f sufficiently developed system of mat ernatrca

particular the phrase rom any, hi h dOOR not deprive the

I "( 170) h ld be understood ill a way W IC

ogic p. S ou all a lied modes of inference. If the theory T j;< meta~heory of any norm Y'llPbP f lly achieved if we dc('ido to prov,<!" '. h"

of finite order our pUrp080 WI e u f I

metatheory M with a logical basia BII strong WI the genoml U"",ry " C 1lMf'll discu ss ed in the following section .

238 CONCEPT OF TRUTH IN

• FORMALIZED LANGUAGES VIII §.

do not form a complet t h '

of a structural definiti e s?,s em t e question of the construction

Ion IS not a priori ho IS' "

possible by add" pe ess. ometimes It IS

extend t'h ,mg certain structurally described sentences to

e aXIOm system of th ' , '

it b e SCIence ill a suitable way so that

I ecomes a system in which th I

coincides with th I e c ass of all its consequences

no question of e c ass of all true sentences, But there can be the attempt t a general method of construction. I suspect that

o construct a structural d finiti

simple eases=-e.ev i , e ron , even in relatively

.g. m eonnexron with th I .

relations studied' th . e OglC of two-termed

serious diffic It' mT e pr~ceding section-would encounter

u res, hese difficult' ld

much great h . res wou certainly become

er w en It came to th ti

structural crite . f e ques IOn of giving a general

with two J non 0 truth, although we have already dealt

anguages, that of the calculus of classes and that of

(which would be difficult to d fin '

so to speak an "infinite 1 ' 1 e e) It represents a. • natural generalization'

d ' I ogica product' f those narti '

eseribed in convention T d hi ,0 ose partial definitions which were

the sentences of the I an w ich establish a. direct correlation between structural definitions ~~tghe and the names of these sentences. Among the

st t d • e other hand I' I d .

rue e according to the followin ,~c u e those which are con-

other expressions is described' g :heme, FIrst, a class C of sentences or pression it is possible to k m ~c a way that from the form of every exSecondly, certain operationson w ethe~ it belongs to the given class or not, o~ any ~ven expression a an~ ~~~:,:slOns are listed such that from the form

bg,ven finite collection of expressions waf all me~bers el , .. "en belonging to a y performing on e ' e can decide whether a can be obtained

t I, ... ,e" anyone of th li t d

sen enees are defined as th .' e IS e operations, Finally the true

operations to the ('xpr(,~RI" ORe f\' hhlCh arc obtained bv applying'the listed

h . t d h . onR" t p e18.~R (' ','

.' 110,:- t at such a struct.uml def ' " " ~ny. num ,pr of times (it is to

eral crIteTlon of "rut h) (', tai , nition ~hll III no wav provides a gen-

niz . I , . A r .am <hff .. n-ne f f .

• 1( II'tw"4'n t,hI'R(' tw k ind . (,ll 0 a ormnl nature can he recog-

r" h ,,0 m,,, "f rlofinit i '1'1

,( '1111"">1 t " II"" of krill" f I' I I 10n14. J(' ""manticlLl definitiun

III V"Ht igut.".!, ",1(. t.lH' 1I",~o fllIR, l','r , "rdf'r thnn 1111 vnrinhlos of t.h« lanKllogll

of II I I I 0 ,H' u-rm '"ati f 'I f

II .ruc .urn d .. finit.iou Ih, I ' r ' " 11'''; 1111. or IIII' f"rrlluilltion

..... IN" Hllllie,'. 111 t ,.., f'or HI' '; I'm>! 0 (>!'rhILJlI! (.WII or I hrp,' of I.lw Low""t plicitly or irnplidtly~t' I. ,1'1..,1011 41.( IL ""l11llntio'IIL d"linitioll we u"" .. -cxof like meaning with th lOMe ",XPTl'8HlOflS of tlw Illl't.llianglla""· whi('h are

th e eXpr('HHi, f h Po ' '

ey play no part in the const tior 0 t e language investigated, whilst

to see that this distinction rish IOn of a structural definition; it is easy ment of logic. Moreover the du t ~s ~hen the language studied is a frsgsharp, as is shown by th~ fact th mct~on as a whole is not very clear and the semantical definition e I at With respect to the sentential calculus ~he structural definition OO:d re regarded as a formal transformation of It must be remembered that ~~ the matri~ method. At the same time based on the methods t e construction of semantlcsl definitions the st t I a present known t . '

rue ura definitions of se t 0 us, 18 essentially dependent upon

n ence and sentential function.

VIII, § 4 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 239 the sentential calculus, for which this problem could be relatively easily solved.l

In all cases in which we are able to define satisfaction and the

notion of true sentence, we can-by means of a modification

of these definitions-also define two still more gcncral concepts

of a relative kind, namely the concepts of 8atisfaction and correct sentence-both with re8pect to a given individwll domain a.2 This modification depends on a suitable restriction of the domain of objects considered. Instead of operating with arbitrary individuals, classes of individuals, relations between individuals, and so on, we deal exclusively with the elements of a given class a of individuals, subclasses of this class, relations between elements of this class, and so on. It is obvious that in the special case when a is the class of all individuals, the new concepts coincide with the former ones (cf. Defs. 24 and 25, and Th. 26). As I have already emphasized in § 3 the general concept of correct sentence in a given domain plays a great part in present day methodological researches. But it must be added that this only concerns researches whose object is mathematical logic and its parts. In connexion with the special sciences we are interested in correct sentences in a quite specific individual domain for which the general concept loses its importance. Likewise it is only in connexion with sciences which are parts of logic that some general properties of these concepts, which were proved in § 3 for the language of the calculus of classes, preserve their validity. For example, it happens that in these sciences the extension of the term 'correct sentence in the individual domain a' depends exclusively on the cardinal number of the class a. Thus in these investigations we can replace this term by the more convenient term 'correct sentence in a domain with k elements' (Def. 26, Th, 8). The theorems previously discussed concerning the concept of truth, such as the principles of contradiction and the excluded middle can be extended to the concept of correct sentence in a given domain. The concept of correct sentence in every

1 Cf. the remarks on pp. 207 f. and 221; I shall return to this problem in § 5 (cf. p. 254, footnote).

• Soo p. 199, noto 2.

240 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 4 individual domain (Def. 27) deserves special consideration. In its extension it stands midway between the concept of provable sentence and that of true sentence; the class of correct sentences in every domain contains all theorems and consists exclusively of true sentences (Ths. 22 and 27). This class is therefore in general narrower than the class of all true sentences; it contains, for example, no sentences whose validity depends on the magnitude of the number of all individuals (Th. 23). If it is desired to transform the system of the provable sentences of every science into a complete one, it is necessary at the outset to add sentences to the system which decide the question how many individuals exist. But for various reasons another point of view seems to be better established, namely the view that the decision regarding such problems should be left to the specific deductive sciences, whilst in logic and its parts we should try to ensure only that the extension of the concept of provable sentence coincides with that of correct sentence in every individual domain. For a supporter of this standpoint the question whether the extension of these two concepts is actually identical is of great importance. In the case of a negative answer the problem arises of completing the axiom system of the science studied in such a way that the class of provable sentences thus extended now coincides with the class of sentences which are correct in every domain. This problem, which properly is equivalent to the question of structurally characterizing the latter concept, can be positively decided only in a few cases (cf. Th. 24).1 Generally speaking the difticulties presented hy this qll{'~tion are no less essent-ial than thORO connected wit.h t,!10 unnlogous probl(lID of a structural definition of true sentence. Wn meet with Himilar difficulties when we attempt to define structurally the concept of correct sentence in a domain with k elements. Only in the case where k is a finite number is it easy to give a general method, modelled on the method of matrices from investigations on the extended sentential calculus, which makes a structural definition

D LA)/Gl'AUES 241 VIII, § 4 CONCEPT OF TRUTH IN FORMALIZE

. t ible In this way we even obtain a general

of this canccp POSSI . . • .•

iteri hich enables us to decide from the form of any HC.IltCIll.L

en erIOn w . I . fi IIi ttl

whether it is correct in a domain with a. prcvlOliH y given

number of elements.l . . f

. h t ter hcre into a more det,llih~d dlHCUHHIOII II

I do not WIS a en . 1 j H to

. . ti ti n the concept!! just con!!1l ereu . • oil e

speCIal inves Iga ions 0 f I

. I t h relating to the calculus () c ILHHPH,

results which are re ~van ere'm les in § 3. I will only mention

have already been gIven as exa p It have been obtained which

th t . t years numerous resu s .

a In recen t f certain scntence!! 10

bl to i nfer from the correc ness 0

ena c us 0 1 . f m their structural properties

special individual domaills or r.o d thus their truth.2 It is

. t in every domam an

their correc ness . clear content and can

evident that all these results only receIveta d precisely forrnu-

tl ed if a concre e an

only then be exac Y prov, . eepted as a basis for the

lated definition of correct sentence IS ac

investigation.

. 1 In the case of the lower functional calculus this problem, which is raised in Hilbert, D., and Ackermann, W. (30), p. 68, has recently been decided by Godel, S99 Godel, K. (20).

OF TRUE SENTENCE IN LANG CAGES OF § 5. THE CONCEPT

INFINITE ORDER

f th 4th kind hence to those

We come now to lang~ag~: °ond ~hC scope ;fthc methods of

of infinite order and so lying y. ti The language of

h d i the precedmg sec ion.

construction sketc e III will as an example. This

h .f classes serve

the general t wry OJ • ite of its elementary

. te rthy because, III sp J:'

language IS no wo ti I forms it suffices lor

structure and its poverty in gramma rca ,

h"nfinkel M. (5 a), p. 352. tain

1 Cf. Bemays, P., and Sc 0 'IDS of LOwenheim and Skolem, cer

• According to the well.known t~eo:_e domain provided they are correct categories of sentences are correct ~ riese sentences include. for examp~

. 11 finite and denuxnerable doroams. te ed relations, described in t

in a f the lozic of two, or many- .ial f t.i ns in which ,arlables

all sentences 0 e- . f sententla unc 10

t'on which are generalizatIOns 0 . bles In the case of the sentences

sec I , . I as free varia . 9 f § :J--can of the 2nd order occur exclu:llve Y It-as is shown in Tha. 15 and 1 0 d Acker.

of the calculus of classes th~= results of Bernays, Schiinfinkel, an ~late a be essentially sharpenedd . of application. They allow us to c:, such a mann have a narrower omam ith sentences of a speeial structure in with k: particular natural nuxnber k : of a given sentence in the dO~rti"" of the

way that from the corre= 1m w-from purely structural prope W (I)

elements (thus-as we alr y 0 d in follow". Cf. Ackermann. . II'

)'t correctnes:. s in every oma I , (26) J",w"ob",rn. I •. (4 r,

sentence IS. fink I M. (5 a). Horhrao( .' .• , f tI ... "",ult.Jo

Bernaye P., and Schon e. • .: .. 1A.matlc p".w.ntllt,"n ,

Skolern 'Th. (64). (6;'). and (66). 1-or ~ ;:n ..... ",." Chup·h .. \. (II "I·

in thi~ :lirection inducling more rN'f'Ji . ,

242 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 5 the formulation of every idea which can be expressed in the whole language of mathematical logic. It is difficult to imagine a simpler language which can do this.'

In the general theory of classes the same constants occur as in the previously investigated sciences, i,e, the signs of negation and of logical sum, as well as the universal quantifier. As variables we use such symbols as 'X~', 'X~', 'X;', and so on, i.e. signs composed of the symbol 'X' and a number of small strokes ~bove and below. The sign having n strokes above and k below IS called the k·th variable of the n-th order and is denoted by the symbol 'V~'. The variables Vi, Vi, V~, ... represent respectively names of individuals, objects of the Ist order; classes of individuals, objects of the 2nd order; classes of such classes, objects of the 3rd order, and so on. These variables obviously belong t? infinite_Iy many semantical categories. As primitive sentenbial functions we have expressions of the type 'XY' where in the place of 'X' any variable of the n+ 1 th order, and instead of 'Y' a variable of the nth order occurs. This expression is

1 The language of the general theory of classes is much inferior to the language of Whitehead, A. N., and Russell, B. A. W. (90) in its stock of 8ema~tl~al ca~~gor~e8, and still more inferior in this respect to the language used ~l. ~esmew8ki m hIS system (cf. p. 210, note 2; p. 218, note 2). In particular, in

IS ~nguage no sentential variables and neither names of two- or many-termed relations, nor variables representing these names, occur. The dispensability of sentential varia.bles depends on the fact mentioned on p. 229, footnote: to every sentence which contains sentential variables there is a logically equivalent ~entence which does not contain such vurjablos. The results of § 2, esp~clUIly l~efH. 13-17, suffice to show how such vnriables are to he avoided in settmg up IIstK of uxioms nnd in the derivation of theorems : cf. also Neumunn, J'I v .. (54) (especially note 9, p. 38). The pOIiKibility of eli minuting two. termed re utions result» from tho following eunaidorat.ion. With flvery relation U we can correlate in one-one f hi I . h Ino~

f I '. 8H lion, a C I\IIR of ordered pRlrH, rlJunoly, t 0 c ~

o al. o~erod purrs whose terms (C and y satisfy tho formulu, x1l1J. If the relation HI homogeneous, i.e. if the domain and eounter donmin of this relation belong to the .same semantieal category, then the ordered pair can be inter· p[eted otherwise than we have done on p. 171, namely as classes having twO c asses as elements: the class whose only element is::c and the class consisting he the two elements (C and y. In order to apply an analogous method to in-

omo~enoous relations we must first correlate homogeneous relations with ~hem m one-one fashion, and this presents no great difficulty. We proceed ab an analogous way with many-termed relations. In this way every statement ~ out two- ?r many-termed relations of arbitrary category can be transformed l~tO an equivalent statement about individuals classes of individuals classes o(13B)uch CI~lesl ' and so on. Cf. Kuratowski, C. '(38), p. 171, and Chwistek, L.

,especla y p. 722.

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 243 read: 'the class X (of n+ Ith order) has as an element the object Y (of nth order)', or 'the object Y has the property X'. For the designation of the primitive functions we employ the symbol '€" , setting €" = V"+l'" V". The further development of the

k,l' k,l k I . f

science differs in no essential way from that of the logiC () two-

or many-termed relations. The quantifications of the sentential functions x with respect to the variable V~ are dcnoted by the symbols 'n~ x' and 'U~x'. The axioms consist of (~) sentences which satisfy the condition (ex) of Def. 13 of § 2, WhICh are t~U8 derived from the axioms of the sentential calculus by substltu· tion, sometimes also followed by generalization; (2) pseudodefi~itions, i.e. statements which are quantifications of sententIal

functions of the type

u~+lnr(E~,Z.Y+E~'I·fi),

. . hi h d es not contain the

where Y is any sententiel functIOn w C 0

free variable V~+I; (3) the laws of extensionality, i.e. sentences

of the form

_ _ p+l+ £+1)

n£+2nf+In~+1(U~(Er.".€1:t,,,+Er.,,.E1:t,")+€k'Z E ,m '

which state that two classes which do not differ in their .elem~ntls

. rti d are thus IdentICa .

do not differ in any of their prope ies an h t b

In order to obtain in this science a sufficient basis for t . eels a ~

h ti and in partlcu ar 0

lishment of various parts of mat ema ICS b

. t add to the a ove

the whole of theoretical arithmetIC, we mus . h te ce

• ,f • ,-I1nity 1 e t e sen n

still one more axiom: (4) the axtOffl oJ tnJ' ,., -

U3(U2 2 n2(r+U2(E22.nH€~I+Eb).UHEl.t.€b))))'

1 1 €1,1' 1 ~1,1 2 I, ' .nv i diVl'duaIs 1

f i fin' tel many ill .

which guarantees the existence 0 III I Y . IDS we apply

I .' from the axio

n the derivation of consequences d the intro-

the operations of substitution, detachmen~, an al oUS to

h . al quantIfier, an og

duction and removal of t e umvers Def 1 ~ in § 2.

the operations described in conditions ('Y)-~') o~ .' C:nnexion

h t ofsatlSfactIOn ill

When we try to define t e concep _,,~ lties which we

. nter UlJl1CU I

With the present language we encou di 'ty of seman-

h . finite versl

cannot overcome. In the face of t e III tulate

. .' admittedly givo up the. poH .

I In adopting the axrom of mfimty we t in every mdlvldual

hich are cerrcc

according to which only the sentences w. f "40)

domain are to be provable sentences of logic (c . p. ~ .

244 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 5 tical categories which are represented in the language, the use of the method of many-rowed sequences is excluded from the beg~g, just as it was in the case of the logic of many-termed relations. But the situation here is still worse, because the method of seman tical unification of the variables also fails us. As we learnt in § 4, the unifying category cannot be oflower order than anyone of the variables of the language studied. Sequences whose terms belong to this category, and still more the relation of satisfac~ion, which holds between such sequences and the corresponding sentential functions, must thus be of higher order than all those variables. In the language with which we are now dealing variables of arbitrarily high (finite) order occur: consequently in applying the method of unification it would be ne~essary to operate with expressions of 'infinite order'. Yet ~eIthe~ the metalanguage which forms the basis of the present mvestlgation~, nor any other of the existing languages, contains such expressions, It is in fact not at all clear what intuitive meaning could be given to such expressions.

These considerations seem to show that it is impossible to construct a general, semantically unambiguous concept of satis~action for the language we are studying which will be applicable to all sentential functions without regard to their s~manti~al type. On the other hand there appear to be no d.IfficultIes which would render impossible in principle a coneistent application of the concept of satisfaction in its original formulation or ruther . . fl' .' f

, '" -m view 0 t ie semantioal ambiguity 0

that forrnulat,ion--·of all infinite number of such concepts. Each of t~~80 concepts is, from tho aemnntioul Htnwlpoil1t, nlready Hpoclhe~llllld would relnte oxeluaivoly tu funct.ions of 11 specific semantical type (e.g. to functions which contain a variable of the

Ist order as the 0 1 f . bl .

n y ree varia e). Actually-mdependently

of the logical structure of the language-the intuitive sense of

none of these e . .

. . xpressions raises any doubt. For every par-

tieular sentential function we can in fact define this meaning ex~ctly by constructing for every phrase of the type 'the ob!~cts a, b, c, ... satisfy the given sentential function' an in-

t.uit.ivel'v equivale t h hi h .

• n prase w IC IS expressed wholly in terms

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 245 of the metalanguage. Nevertheless the problem of the construction of a correct definition for each of these concepts again presents us with difficulties of an essential nature. On the hl1His of the languages which wc have previously studied it was easy to obtain each special concept of satisflwtion by a certain specialization of the general concept; in the present case this way is clearly not open to us. A brief reflection shows t.hat the idea of using the recursive method analogously to the definition of sentential function proves, in spite of its naturalness, to be unsuitable. It is easily seen that the composite functions of a particular seman tical type cannot always be formed from simpler functions of the same type. On the contrary, if we are to be able to construct arbitrary functions of a given type, we must use for that purpose all possible semantical types.' It would, therefore, be necessary, in the recursive definition of anyone of the special concepts of satisfaction, to cover, in one and the same recursive process, infinitely many analogous concepts, and this is beyond the possibilities of the l.anguage.

The central problem of our work, the constructIOn of the definition of truth, is closely eonnected with these considerations. If we were successful in defining, if not the general, at least anyone of the special concepts of satisfaction, then this problem would not offer the least difficulty.2 On the other

1 An external expression of this state of affairs is that in the definition of satisfaction not only is it essential to tak.e free v~riables into account ~ut a~so all the bound variables of the function ill quest.on, although these' ar-iab es have no influence on the semantical type of the functIOn; and whether the relation of satisfaction holds or not does not depend in any wa~§o~ the :;;ms of the sequence which correspond to these variables (cf. Def, 220 ,can :,.;~n

(S» It is to be remembered that analogous difficulties to those .mendtlon .. ill . . t to truct a recurslve efirutlOn

the text appeared earlier ill the attemp cons

of truth by a direct route (cf. p. 189). ..

, For example let us imagine that we have succeeded III some wayh· illh

, ... h f tential funcUons w rc

defining the concept of satisfactIOn III t e case ~ sen, ld h rate

contain a variable of 1st order as the only free vanabl~. "e cou t en ope

f I it.h h of the type • the individual a satlBfies the sentent.al func-

rOO y Wl P rases . . U' •

t.i , If consider some one concrete sentent.al functIOn. e.g. I <i.i-

~~7c~ is sa::fi:~tbY every arbitrary individual. ,,:e obtainln atjonce thelfoJ11()~; . t ntence ,j and 0 Y I et·ery "" tV" ......

ing definition of true sentence: x VI a rue Be. . 1 tL

. 2 1 ( . th CDnJunctwn oj the ".,..yw:e x arn ""

a satisfies the Junctwn x. U. fl.' s.e. e coniw»: f am ever other

junction U2.1 ). In an exactly analogous way we can p- r Y

• 1.1 . . h pt of truth.

specific concept of satISfactIOn to t "conce

246 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VITI, § 5 hand we know of no method of construction which would notdirectly or indirectly-presuppose a previous definition of the concept of satisfaction. Therefore we can say-{)onsidering the failure of previous attempts-that at present we can construct no correct and materially adequate definition of truth for the language under investigation. t

In the face of this state of affairs the question arises whether our failure is accidental and in some way connected with defects in the methods actually used, or whether obstacles of a fundamental kind play a part which are connected with the nature of the concepts we wish to define, or of those with the help of which we have tried to construct the required definitions. If the second supposition is the correct one all efforts intended to improve the methods of construction would clearly be fruitless. If we are to answer this question we must first give it a rather less indefinite form. It will be remembered that in the convention T of § 3 the conditions which decide the material correctness of any definition of true sentence are exactly stipulated. The construction of a definition which satisfies these conditions forms in fact the principal object of our investigation. From this standpoint the problem we are now considering takes on a precise form: it is a question of whether on the basis of the metatheory of the language we are considering the construction of a correct definition of truth in the sense of convention T is in principle possible. As we shall see, the problem in this form can be definitely solved, but in a negative sense.

It is not difficult to see that this problem exceeds the bounds of our previous discussion. It belongs to the field of the metametatheory. Its definitive solution, even its oorreot formulation, would require new equipment for investigation and especially the formalization of the metalanguage and the metatheory which uses it. But without going so far, and still avoiding

t The problem of the possibility of defining satisfaction and truth for ~he language under investigation will be conaiderably clarified by the discUSSion in the Postscript. It should be mentioned that the method of defining truth recently suggested in McKinsey, J. C. C. (53 b) is not based on a prel~ definition of satisfaction. Instead, McKinsey has to consider formahzed languages with non-denumerably many constants and has to use a metalanguage which is provided with a very strong set-theoretical apparatus.

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGE~ 247 various technical complications, I believe. I. am able to} gttve a

hi f positive nature t in can

fairly clear account of everyt ing 0 a II

at present - esta~I~~~~ :e~o;~;~:~:~: :~:l~ba:;h:~~~:' l~:~

In o~eratI~g w~ §§ 2 d 3 To simplify the further develop-

symbolism given m an . II

id ible misunderstandings we sha suppose

ments and avoi pOSSI

t ted that the language we are

the metalanguage to be so cons rue . fthe language

f nt of it . every expreSSiOn 0

studying forms a ragme .' f the metalanguage, but not

is at the same time an expr~sSlon o. ( g in the formula-

. bl s m certam cases e. .

vice versa. This ena es u . T to s ak simply of the

tion of condition (0:) of con:en~o:'st~ad oiexpressions of the

expressions of the language itse , .

'hi h h ve the same meanmg.

metalanguage w c a. d ventions we tum to the

After these reservatIOns an con I It

. f f the fundamenta resu .

formulatIOn and proo 0 ~_M1 "I'r' deooting a

I ( ) I hatever way the iYyu.vv. ,

THEOREM . 0: n w . he metatheory it will be possible

. is de-l:ned ~n t '

class of express~.ons, ~ J' • one 01 the sentences which were

to derive from ~t the negatton of 'Jnt, T'

cond" (0:) of the conve ton ,

described in the tt~on .1 all ooable sentences of the meta-

(f3) assuming that the clas~ oJ !:z to construct an adquate

theory is consistent, it is ~mpOS8t ent· T on the basis of the . he nse 01 conve ton

definition of truth tn t se 'J

metatheory. his theorem can be expressed in the

The idea of the proof of t. I . terpretation of the meta-

d 1 (1) a part1cu ar in

following wor s: .. h h8B employed it for other

od used here to Godel, w 0 iall pp 174-5

I We owe ~he meth blished work, Godel. K. (22). cf. ~~~ing' article

purposes in his recently pu . exceedingly important ~d in with strictly or 187-9.0 (proofofTh~)';"i;:::he theme of our work-It deals of deductive is not directly connec consistency and comple~ part also the methodological problems: h t~e be able to use the methods and in

s stems' nevertheless we s a urpose. f 'ts

y It 'f G6del's investigations for ~m: p hat Th. I and the sketch 0 I

resu s 0 't of mentlOOlDg t had aJready gone to press.

I take th~ o~:!tu: rhe present work after it Society of Sciences (21

proof was 0 Y resented at the Wa.rsaw t a peared. In At the time the work wa:s P fi as I know-had not 1.e. p ulta only

M h 1931) Godel's artlcle--'lO ar -" instead of POfllt.ve res ,

arc , ad rigine.lly exp"""""" baaed rtly on my own this place theref~re I. h 0 direCtion, which were pa hi h bsd been certain suppositIOns m the samhe h rt report GOdel. K. (21). w c

" d artly on t e 8 0 '

inVestigatiOns an Previously. tion",1 &rlid" r con-

published some months Painted with the above moo h""ry whirh n(~I",1

After I had b<lcome acqu. that the rI",h.ct.ve t

f nz other thmgR.

vincod my""! , amo "

had chosen as the object of his studies, which he called the 'system P', was strikingly similar to the general theory of classes considered in the present section. A part from certain differences of a • calligraphieal' nature, the only distinction lios in the fact that in the system P. in addition to three 10gLcai constants. certain eonstunta belonging to tho ar it.hmot i« of tho natural numbors also occur (a fur-rem-lung nnulogy 81,,0 f1xi"tH hf1tw",,>n tho "ystem P and tho system of urithrnot.ic Hkotrhfld in VI (HIlO I'l'. 113·1 !I). COllHe'llwnt,ly the results obtuinod for tho .. y.<t,I'm P ,'all ""Hily lI" parried nvor to Om I'rt, ... mt, <li"(111"8.on. Muroovnr. !.I,,, ub .. t,rllf't ohuruet,'r or t.l1O nwl,hod .. 11"".1 hy niHI,,1 renr!M", t~e validity or hi .. 1",,,,ult,,, in<i"Il"lld"lIt I,u 1\ high d"gr .. " of 1,],1'1 KI" ... iU(l pcx1uharltlO8 of tho .. eiflncIl invtlHtigatfld.

1 cr. Fruonkel. A. (16). 1'1'. 48 ff.

• If we analyse the sketch of the proof given below we easily note that ~ analogous reconstruction could be carried out even on the basis of colloqUlal language, and that in consequence of this reconstruction the antinomy of the liar actually approximates to the antinomy of the expression 'heter~logical'. For a rather simple reconstruction of the antinomy of the liar in th~s direction see Tarski, A. (82), note II, p. 371. It seems interesting that in thiS reconstruction all the technical devices are avoided which are used in the proof o~ Th. 1 (such as interpretation of the metalanguage in arithmetic or the diagonal procedure). In eonnexion with the last paragraph of the text cf. the concluding remarks of § I, pp. 164 i., and in particular p. 165, note 1.

VIII. § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 249

We shall sketch the proof a little more exactly. 1 , ,. _

c the moment to use the symbol n mstcad of

Let us agree lor . I f -t'

, t' I t'fiation of the sententm unc JOn y

'XIII' The existen ia quan 1 I

, . h' bl 'n' will be denoted by the Hymho

with respect to t e varia e f

. bl ' , thus rCJlrcsentA IIIlJllI\H 0

'U3 ' as before. The varia en. .

lY f hieh are classes of indlVldualli, Among

classes the elements 0 w th thi gs the natural numbers

these classes we find, among 0 er n, 2

ki the cardinal numbers.

and generally spea ~~ d that all facts belonging to the arith-

I have already men ione b reseed in the language of

metic of the natural numbers can e;:p lar if a natural number the general theory of. classes .. In pa. I::sil; constructed in this k is given, a sentential function ~k/~ , as the only free variable language which contains the syrn 0 n . . represented by this

h t th lass whose name IS

and which asserts t a e c (i ith the class con-

. I'th the number k i.e., WI

symbol is identic a WI . d' . daIs which have exactly k

sisting of just those classes of m IVI u

elements).2 For example: - - 1

nZ( 2 U1 n1 n~(£} l' (£l,z+4.1+£2.2))+

'1 = 1 £1.1' 1 lI· _- 1 -)))

2 n1 U1 U!(£ll +£1.:. £2.1' E2.: •

£1.1' I 2 • • .

. 'tion of the sequence of functIOns 'k

A general recursive defini t difficulty.

. . 1 ge presents no grea

within the meta angua . 2 ( 184) a one-one corre-

d [nted out III § p.

As I have alrea Y po. diffi Ity between the expres-

b t up WIthout ou . th

spondence can e se ural numbers ;wecandefinem e

sions ofthe language and the nat fexpressions in which every

, fi 'tesequenceq,o . ,IT'th

metalanguageanm m d only once. .vn

I age occurs once an .

expression of the angu I te with every operatIOn

I ti we can corre a

the help of this corre a Ion 1 mbers (which possesses

t'ononnatura nu .

on expressions an opera I . with every class of expressIOns a

the same formal properties), I this way the meta-

b nd so on. n . f th

class of natural num ers, a . . the arithmetIc 0 e

. terpretatlOn III

language receives an III laces e"Pres' ourselves as

. licity we shall in roany p roetatheory and not

1 For the sake of sunt, p hich folloWS belonged to.thethat a given ""nten""

th h the demonstra IOn w . tesd of sapng '. If 1

oug . in particular. 10S 'he senten..,,,.t<Ie· n

to the meta_metatheory ~h ry we shaH simply assert t h oof ill given b"..., is provable in the meta eo tte that only a sketch of t e pr

an case it must not be forgo n

an~ ono which is far from complete.

I Soe P: 233. note 1.

248 CONCEPT OF TRUTH IN FOR)lALIZED LANGUAGES VIII, § 5 language is established in the language itself and in this way with every sentence of the metalanguage there is correlated, in onemany fashion, a sentence of the language which is equivalent to it (with reference to the axiom system adopted in the metatheory); in this way the metalanguage contains as well as every particular sentence, an individual name, if not for that sentence at least for the sentence which is correlated with it and equivalent to it, (2) Should we succeed in constructing in the metalanguage a correct definition of truth, then the metalanguage-with reference to the above interpretation-would acquire that universal character which was the primary source of the semantical antinomies in colloquial language (cf. p, 164). It would then be possible to reconstruct the antinomy of the liar in the metalanguage, by forming in the language itself a sentence x such that the sentence of the metalanguage which is correlated with x asserts that x is not a true sentence. In doing this it would be possible, by applying the diagonal procedure! from the theory of sets, to avoid all terms which do not belong to the metalanguage, as well as all premisses of an empirical nature which have played a part in the previous formulations of the antinomy of the liar. 2

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 251 the theorem. We have proved that among the consequences of the definition of the symbol 'Tr' the negation of one of the sentences mentioned in the condition (a:) of the convention T must appear. From this the second part of the theorem immediately follows.

The assumption of consistency appearing in the part (~) of this theorem is essential. If the class of all provable sentences of the metatheory contained a contradiction, then every definition in the metatheory would have among its consequences all possible sentences (since they all would be provable in the metatheory), in particular those described in the convention T. On the other hand, as we now know," there is no prospect of proving the consistency of the metatheory which we are working with, on the basis of the meta-metatheory. It is to be noted that, in view of the existence of an interpretation of the metatheory in the science itself (a fact which has played such an essential part in the proof sketched above), the assumption of the second part of Th. I is equivalent to the assumption of the consistency of the science investigated itself and from the intuitive standpoint is just as evident.

The result reached in Th. I seems perhaps at first sight uncommonly paradoxical. This impression will doubtless be weakened as soon as we recall the fundamental distinction between the content of the concept to be defined and the nature of those concepts which are at our disposal for the construction

of the definition.

The metalanguage in which we carry out the investigation

contains exclusively structural-descriptive terms, such as names of expressions of the language; structural prope~ies of these expressions, structural relations between expresslO~, a~d so on, as well as expressions of a logical kind among which (in .the present case) we find all the expressions of the language studied. What we call metatheory is, fundamentally, the morplwlogy of language-a science of the form of expressions-a correlate of such parts of traditional grammar as morphology, etymology,

and syntax.

250 CONCEPT OF TRUTH IN

FORMALIZED LA."GUAGES VIII § 5

natural numbers and indirectl . '

theory of classes. y ill the language of the general

Let Us suppose that we hav d fi

in the metalangu Th e e ned the class Tr of sentences

age. ere would th d

class of natural numb hi. . en correspon to this class a

of arithmetic C .ders w oh IS defined exclusively in the terms

. onsr er the expression 'U3( .1..) - T ' ..

a sentential function f th 1 'n· 'f'n E r. This IS

the only free variable 0 Fr e metalang.uage which contains 'n' as with this f ti . om the prevIous remarks it follows that

uno IOn we Can correl t h

equivalent to it cae anot er function which is lor any value of' , b t hi

completely in te f. n, u w ieh is expressed

rms 0 anthmet· Vr h I

function in the h. 10. '¥ e s al write this new

sc ematlC form '.f'(n)' Th h

( I) 'f' • us we ave:

for any n U3( .1..) - •

. ' ! 'no 'f'n EO Tr if and only if I/J(n).

SInce the language of th

for the found t· e general theory of classes suffices

a Ion of the arith ti f

can assume that 'I/J( )'. me lC 0 the natural numbers, we

The function '0/( )' :mIS one of the functions of this language. the term with thn. d thus be a term of the sequence <p, e.g.

e ill ex k 'o/(n)' ,./..

'n' in the senten (1) , . = 'f'k· If we substitute 'k' for

ce we obtain:

(2) U3(-

1 'k·<Pk) E Tr if and only if I/J(k).

The symbol 'U3(L rp)' d

language und .1 k·.k • enotes, of course, a sentence of the

er mvestIgatlOn B I .

condition (0:) of th .. y app ymg to this sentence

s: e conventIOn T w bt·

rorm 'x E Tr if and l ., e 0 am a sentence of the

structural-dc8cript. on y if p , where 'x' is to be replaced by 3 statement U'3( ,./..Iv)eb°l" any other individual name of the

I 'k·"'k, ut 'r' hy thi t t .

statemcnt wl .. h . . . us fI ,a .emont Itself or by any

lie IS tlqUIvalcnt t ·t I

stituto 'U8(, "/")' I" "OJ. n pnrtillulllr wo oan SU b·

1 k· 'f'k lor x and for' • . .. .

the symbol' , th 1) --m VICW of tho meaning of

'k - e statement 'tl .

and I/J(n) , or, simply' k' I . iere IS an n such that n = k

formulation: I/J( ). n this way we obtain the following

(3)

U~('k·<Pk) E Tr if and only if I/J(k).

The sentences (2) and (3) .

one another· the s te stand III palpable contradiction to

, en nee (2)· . .

negation of (3). In this IS ill fact directly equivalent to the

way we have proved the first part of

1 Cf. Godol, K. (22), p. IIl6 (Th. XI).

"-,, C

_0_ OXCEPT OF TRUTH . ~

IN FORMALIZED LANGUAGES VIII § 5

The fact that the language t di d '

carried out in this I Sue and the deductive science

anguage are f Ii

interesting phenom ' orma ized has brought about an

enon , It has bee ibl

tural-descriptive n paSSI e to reduce to struc-

concepts certain th

different kind whi h ' 0 er notions of a totally

, ,c are distin ish d '

their origin and i thet guis e from the former both ill

n err usual mea '

consequence togeth ' h nmg, namely the concept of

er WIt as' f

been Possible to est bli h enes 0 related notions.! It has

given science as a p at fS what may be called the logic 0/ the

Ear 0 morphology ncouraged by thi '

s SUccess we hav tt

and to construct in th av e a empted to go further

concepts belongin t e metalanguage definitions of certain

g 0 another do "

semantics of mngua e-" main, namely that called the

noting, truth defi gbil' i.e. such concepts as satisfaction, de-

f ,na Ity and so A

o the seman tical ' on, characteristic feature

, concepts is that th' ,

certain relations betw h ey give expression to

hi een t e exp "

o jects about whi h th resslOns of language and the

c ese exp .

of such relations they h r~sslOns speak, or that by means

c aractenze certai I

or other objects W ul e am c asses of expressions

. e co d also say ( ki

materialis) that th rna ng use of the suppositio

esc concepts

between the nam f serve to set up the correlation

selves. es 0 expressions and the expressions them-

For a long time th

" e sernantical .

reputatIon among sp 'I" concepts have had an evil

. eoia IStslO thest d fl

reSIsted all attempt t d. U yo anguage. They have

8 0 efine th . .

properties of tho err rneamng exactly and the

, A'lC concepts a '

have led to pfl,rn, J. " pparently 80 olonr in their content,

•. ( • (OXeH and antin .

"mdolley to reduc tJ ' omws. For t,hat reason the

,0 WHO cOnC(I}lts t t . .

must, H.'Nn quito n t I ,08 ructllrnl-doscrlptJvc ones

n urn and well f ) I "

seemed to favour th . " - oum (l( .1 hn following fact

it has always been e ibk SI-lIbIllty of realizing this tendency:

t . POSSI e to repl

ains these eema ti I ace every phrase which con-

n rca terms and hi h

1 The reduction of th 'w io concerns particular

the morph I e concept of cons

stages of dO ogy of language is a result f :~uence to. concepts belonging to

10 f evelopment. When in 0 e deductIve method in its lateBt

ws rom oth ' everyday lifo

the . t er sentences we no do bt e, we say that a sentence fol-

exis ence of crt· u mean s th' .

light of th e am structural relat' orne mg quite different from

has b e latest results of God I't Ions between these sentences. In the een effected without rema' de I seems doubtful whether this reduction mer.

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 253 structurally described expressions of the language, by a phrase which is equivalent in content and is free from such terms, In other words it is possible to formulate infinitely many partial definitions for every semantical concept, which in their totality exhaust all cases of the application of the concept to concrete expressions and of which the sentences adduced in condition (r;) of convention T are examples. It was with just this end in vicw that, as a rule, we included in the metalanguage, with regard to the content of the semantical concepts, not only the names of expressions but all expressions of the language itself or expressions having the same meaning (even when these expressions were not of a logical kind, cf. pp. 210 f.), although such an enrichment of the metalanguage has no advantages for the pursuit of the 'pure' morphology of language.

In the abstract the fact mentioned has no decisive importance; it offers no path by which an automatic transition from the partial definitions to a general definition is possible, which embraces them all as special cases and would form their infinite logical product.! Only thanks to the special methods of construction which we developed in §§ 3 and 4 have we succeeded in carrying out the required reduction of the semantical concepts, and then only for a specified group oflanguages which are poor in grammatical forms and have a restricted equipment of sernantical categories-namely the languages of finite order. Let it be remembered that the methods there applied required the use in the metalanguage of categories of higher order than all categories of the language studied and are for that reason fundamentally different from all grammatical forms of this language. The analysis of the proof of Th. I sketched above shows that this

1 In the course of our investigation we have repeatedly encountered similar phenomena: the impossibility of grasping the Simult:anOOU3 dependenM between objects which belong to infinitely many semantical ca~gor"~:3; the lack of terms of • infinite order'; the impossibility of mcludmg, in one process _of definition, infinitely many concepts, and so on (pp. 188 f., 232 f., 243, 24"1· I do not believe that these phenomena can be viewed 80S a symptom of the formal incompleteness of the actually existing languages-their ("au"" i. to be sought rather in the nature of language itself: language, which '" .. pr<.dud of human activity neccH""rily P08seRSCS a • tinitistic' character, .. lid canm.t

, .' . f f tH or for tho ("Ofll<itruf"tUJU

servo aA an adequato tool for the mVMtlgatlon 0 ac •

of concept", of an eminently' infinitiHtic' character.

1 From bhi . .

(p 187 91)18: or ImmedIately from certain results contained in GOdel, K. (22)

p. - It can eSH'1 b . f, h

. th ' . lye III erred that a structural dofinition of trut -

:n8t~u~t~e disoussed on pp. 230 rr., espeeially on p. 237. note 2-eannot be ti e. even for most languages of finite order. From other investigam~~~aof this al~t.h()r .(op. cit., p. 193, Th. IX) it. ("II"ws t.hat in certain ele-

I. ry, leMl'1I ~n which we can cOlJst.ruct IIllCh a deflnitkm it. ill lI('vertheleS8

mpossuu« t.o jl;IVP a.. I I " '

Th ,. I ( '. .,!'lIcra II .rueturnl erll.erlOIl fur t.hl' truth of a M~lIt.ellce.

c IrM. 0 t.hl·'''' rl'lIl1lt" I' r . ~

a I .. . apr> 11'11, fiT Intlt.aIlC'· I.n H,,' IOjl;ic "f I.wn-t."rmru

ne many-t.C'rnlf't1 rf'l" I I' I' 1 .' . r

I . " . <>IIM 'IHCIIMl't ItI I 4. '1"", H['""n,1 rt·HII!t. II"phl'/I, or

oxamp (0 tu t ho l wer I'·' I

A k ' . uwe prf\( lea ... , ell. eulus (',·ngtlrt' ~'lIl1kli()lJ(lIIkll.lklll') of Hilhert-

c ermann (30) I'P 43 ff·· thl . . I' 1

t t h .,. ., III H Oll.>!t·, however the rC'Blllt 18 apII It'< ,

no 0 t e notion of a tr t b ' . II

I'd (' II .... ue sen ence, lit to the related notion of a uJlIversa Y

va I ~ ge~emgultlg') sentential function.

At this point we should Iik t I

th t' , e 0 ca I attention to the close connexion between

er~~:~o: tf thS~r;ctural definition of truth', and of 'general structural merability id iacussed in this work, and the notions of recursive enu-

an general roo . . (

for exam I . ursrveneas known from the recent literature see,

true sent~n~ ~ostO~skl, A. (53!), chap. 5). In fact, whenever the set of all enumerabl S III a given formalized theory has been proved to be recursively thea Oe, whe may say that there is a structural definition of truth for this

ry. n t e other hand w tha h . I '1. .

of truth h d ,e say t t ere IS a general structura en eTlon

recursive. wen, an only when, the set of all true sentences is (general)

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 2M ment of the class of provable sentences to form a complete and consistent system in itself presents no difficulties.'

An interpretation of Th. I which went beyond the limits given would not be justified. In particular it would be incorrect to infer the impossibility of operating consistently and in agreement with intuition with semantical concepts and especially with the concept of truth. But since one of the possible ways of constructing the scientific foundations of semantics iR closed we must look for other methods. The idea naturally suggests itself of setting up semantics as a special deductive science with a system of morphology as its logical substructure. For this purpose it would be necessary to introduce into morphology a given semantical notion as an undefined concept and to establish its fundamental properties by means of axioms. The experience gained from the study of semantical concepts in connexion with colloquial language, warns us of the great dangers that may accompany the use of this method. ~or th~t reason the question of how we can be certain that the ax~oma~Ic method will not in this case lead to complications and antmomles

becomes especially important.

In discussing this question I shall restrict myself to the the~ry

of truth and in the first place I shall establish the followmg theorem: which is a consequence of the discussion in the pre-

ceding section:

THEOREM II. For an arbitrary, preriously given natural number k, it is possible to construct a definition of the symbol 'Tr' on the basis of the metatheory, which has among its consequenc_es all. tho~e sentences from the condition (ct) of the convention T in whteh In the place of the symbol 'p' sentences uith 1;ariables of at mo~ ~he k-th order occur (and moreover, the sentence adduced in thecondthon

(f3) of this convention).

. uffi to k that this theorem no

By way of proof It s res remar

died' its whole extent but only

longer concerns the language stu mI. h' h

a fragment of it which embraces all those express.lOns w 1C

contain no variable of higher order than the kth. This fragment

• 9A of tlw pr'··...nt \'olum").

I Cf. V, Th. GO, R rCHIIlt of Lin1lenbllum" ("~ p ..

254 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 5 circumstance is not an accidental one. Under certain general assumptions, it proves to be impossible to construct a correct definition of truth if only such categories are used which appear in the language under consideration." For that reason the situation had fundamentally changed when we passed to the 'rich' Ian~uage~ of infinite order. The methods used earlier proved to be rna pplicable; all concepts and all grammatical forms of the metalanguage found an interpretation in the language and hence we were able to show conclusively that the semantics of the language could not be established as a part of its morphology. The significance of the results reached reduces just to this.

But, apart from this, Th. I has important consequences of a methodological nature. It shows that it is impossible to define in the. metat~eory a class of sentences of the language studied which consists exclusively of materialIy true sentences and is at t~e sam~ time complete (in the sense of Def. 20 in § 2). In par~lCu1ar, ifwe enlarge the class of provable sentences of the science rn~estigated in any way-whether by supplementing the list of axioms or by sharpening the rules of inference-then we either add false sentences to this class or we obtain an incomplete system. This is all the more interesting inasmuch as the enlarge-

256 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 5 is clearly a language of finite order and in fact a language of the 2nd kind. We can therefore easily construct the required definition by applying one of the two methods described in § 4. It is to be noted that the definition obtained in this way (together with the consequences given in Th. II) yields a series of theorems of a general nature, like the Ths. 1-5 in § 3, for example, if the formulations of these theorems are suitably weakened by restricting the domain of their applicability to sentences with variables of at most the kth order.

Hence it will be seen that, in contrast to the theory of truth in its totality, the single fragments of this theory (the objects of investigation of which are sentences which contain only variables whose order is bounded above) can be established as fragments of the metatheory. If, therefore, the metatheory is consistent we shall not find a contradiction in these fragments. This last result can be extended in a certain sense to the whole theory of truth, as the following theorem shows:

THEOREM III. If the class of all provable sentences of the metatheory is consistent and if we add to the metatheory the symbol 'Tr' as a new primitive sign, and all theorems which are described in condition.s (iX) and (fJ) of the convention T as new axioms, then th~ class of provable sentences in the metatheory enlarged in this way wtll also be consistent.

. T~ prove this theorem we note that the condition (iX) contains lllfimtcly many scntcnc{'s which are taken as axioms of the theory of truth. A finite number of tJws(' axioms -oven in union with th:l ~illglo axiom from condit.ion (fJ) eallllot, load to 1\ contmdlCtlOIl (so long ItK thero iK no ooutrudictiun alrc.ndy in tho mctatheory). Actually in the finite number of axioms obtained from (iX) only a finite number of sentences of the language studied appears and in these sentences we find a finite number of variables. There must, therefore, be a natural number k such that the order of none of these variables exceeds k. From this it follows, ~y Th. II, that a definition of the symbol 'Tr' can be constructed In the metatheory such that the axioms in question become consequences of this definition. In other words, these axioms, with

. ..

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 257 a suitable interpretation of the symbol 'Tr', become provable sentences of the metatheory (this fact can also be established directly, i.e. independently ofTh. II). If any class of sentences contains a contradiction, it is easy to show that tho contradiction must appear in a finite part of this class.' Since, however, no finite part ofthe axiom system described in Th, III contains a contradiction, the whole system is consistent, which was to be proved.

The value of the result obtained is considerably diminished by the fact that the axioms mentioned in Th. III have a very restricted deductive power. A theory of truth founded on them would he a highly incomplete system, which would lack the most important and most fruitful general theorems. Let us show this in more detail by a concrete example. Consider the sentential function 'x E Tr or i E Tr'. If in this function we substitute for the variable 'x' structural-descriptive names of sentences, we obtain an infinite number of theorems, the proof of which on the basis of the axioms obtained from the convention T presents not the slightest difficulty. But the situation changes fund~mentally as soon as we pass to the universal quantification of this sentential function, i.e. to the general principle of contradictio~. From the intuitive standpoint the truth of all those theorems IS itself already a proof of the general principle; this principle represents, so to speak, an 'infinite logical product' of those special theorems. But this does not at all mean that ~e can actually derive the principle of contradiction from the ~xlOms or theorems mentioned by means of the normal modes of inference

h t b a slight modification in

usually employed. On t e con rary, y .'

the proof of Th. III it can be shown that the pnnclple of contra-

" . ( tit' the existing sense of the

dICtIOn IS not a consequence a eas ill .

word) of the axiom system described. . b

'Ve could of course, now enlarge the above axlO~ system Y

. . '. I te hich are independent

adding to It a series of genera sen nces w . .

of this system. 'Ve could take as new axioms the prmclples of contradiction and excluded middle, as well as those sentences which assert that the consequences of true sentencc:; arc true,

1 Cf. V, Th. 4R, p. !JI of the [,r".,f,nt volume-

258 CONCEPT OF TRUTH IN FOR"HLIZED LANGUAGES VIII, § 5 and also that all primitive sentences of the science investigated belong to the class of true sentences. Th. III could be extended ~o the axiom system enlarged in this way.' But we attach little Importance to this procedure. For it seems that every such enlargement of the axiom system has an accidental character depending on rather inessential factors such, for example, as the ac~ual. state of knowledge in this field. In any case, various objective criteria which we should wish to apply in the choice of further axioms prove to be quite inapplicable. Thus it seems n~tural to r~~uire that the axioms of the theory of truth, together WIth th~ original axioms of the metatheory, should constitute a categorIcal eystem.s It can be shown that this postulate coinc~des in the present case with another postulate according to whic.h the axiom system of the theory of truth should unamblg~ou.sly determine the extension of the symbol 'Tr' which OCCurs III It, and in the following sense: if we introduce into the metatheory, alongside this symbol, another primitive sign, e.g. the symbol 'Tr" and set up analogous axioms for it then the

statement "I'r - T " b '

- r must e provable. But this postulate

cannot be satisfied. For it is not difficult to prove that in the contrary case the concept of truth could be defined exclusively by.means of terms belonging to the morphology of language, which would be in palpable contradiction to Th. I. For other

reasons of a more gen I t h '

, era nil. ure t ere can be no question of an

axiom system that would be complete and would consequently suffice for the solution of every problem from the domain of the

theory under consid ti This ! d

, era IOn. III IS an 101I11oc1i1l.W metho 0-

logical conse<lu('nc f 'J'h I (. h

. 0 0 '. apt) led not to the lnnguage of t e

goneral theory of 01 but I .

asaes ut to t 10 richer lunguage of tho mota-

theory a~d the theory of truth (cf. the remarks on p. 2M).

There IS howeve it diff d

. ' r, qui e a I erent way in which tho foun a-

tlOns of the the f t h

ory 0 rut may be essentially strengthened.

1 For this purpose we t

misses of the theorem b mus n~vertheless to some extent sharpen the pre'

of the metath . y 88SlllIllllg that the class of all provable sentences

eory IS not only co ist t b "

of Godel K (22) . D8lS en, ut also w,conslStent ill the sense

after a ;ingie a ,t It~7, or ill other words, that this class remains consistent cussed below. PP lea Ion of the rule of infinite induction, which will be dis-

2 See p. 174, nota 1.

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 259 The fact that we cannot infer from the correctness of all substitutions of such a sentential function as 'x E Tr or i E 'l'r' tho correctness of the sentence which is the generalization of thia function, can be regarded as a symptom of a certain imperfection in the rules of inference hitherto used in the deductive acicncea. In order to make good this defect we could adopt a new rule, the so-called rule of infinite induction, which in itH application to the metatheory may be formulated somewhat a.'i follows: if I.L given sentential function contains the symbol 'z ', which belongs to the same semantical category as the names of expressions, as its only free variable, and if every sentence, which arises from the given function by substituting the structural-descriptive name of any expression of the language investigated for the variable 'x', is a provable theorem of the metatheory, then the sentence which we obtain from the phrase 'for every x, if x is an expression then p' by substituting the given function for the symbol 'p', may also be added to the theorems of the metatheory. This rule can also be given another formulation which differs from the foregoing only by the fact that in it, instead of speaking about expressions, we speak of natural numbers; and instead of structural-descriptive names of expressions, the so-called specific symbols of natural numbers are dealt with, i.e. such symbols as '0', '1', '1+1', '1+1+1', and so on, In this form the rule of infinite induction recalls the principle of complete induction, which it exceeds considerably in logical power. Since it is possible to set up effectively a one-one correspondence betw~~ expre: eions and the natural numbers (cf. the proof ofTh. I) It 18 easy t see that the two formulations are equivalent on the basis of the metatheory. But in the second formulation no spec~c ,conce~ts of the metalanguage occur at all, and for this reason It 18 applic-

d d . . In the case where we are

able to many other e uctive SCIences. .

dealing with a science in the language of w~ch there ~re no specific symbols for the natural numbers this form~latlOn re-

. . I difi ti For example 111 order to

quires certain externa mo ca Ions. , , f

formulate the rule for the general theory of classes, instead 0

. t ti I function we must ooerate

substitutions of a gIven sen en Ill, ,.-

'U3( )' where in the place of

with expressions of the type 1 'k·l) , ,

260 CONCEPT OF TRUTH IN FOR~L~LIZED LANGUAGES VIII, § 5 'p' the function in question occurs and the symbol "k' has the same meaning as in the proof of Th. 1.1

On account of its non-flnitist nature the rule of infinite induction differs fundamentally from the normal rules of inference. On each occasion of its use infinitely many sentences must be taken into consideration, although at no moment in the development of a science is such a number of previously proved theorems effectively given. It may well be doubted whether there is any place for the use of such a rule 'within the limits of the existing conception of the deductive method. The question whether this rule does not lead to contradictions presents no less se~io~s difficulties than the analogous problem regarding the eXIStmg rules, even if we assume the consistency of the existing rules and permit the use of the new rule not only in the theory but also in the corresponding meta theory and in particular in an!. attempted proof of consistency. Nevertheless from the inbuibive standpoint the rule of infinite induction seems to be as reliable as the rules normally applied: it always leads from true sentences to true sentences. In connexion with languages of finite order this fact can be strictly proved by means of the de~nition of truth constructed for these languages. The fact that this rule enables many problems to be solved which are not solvable on the basis of the old rules is in favour of the acceptance of the new rule, not only in the theory but also in the metatheory. By the introduction of this rule the class of provable sentences is enlarged by a much greater extent than by any supplementation of the lil'lt of ll.XiOIllH.2 r n tho case of certain clerncntury dllliuctivo IIcicl1l(!el'l, thill onlurgomont ill 80 gfllut that the cluas of theurems lweolllcl:! I~ complete system and coinoides

1 I have previou~ly int d I ' , .' .

d ti . porn (J out t 10 Importance of the rule of mfuute in-

uc 1O~ m the year 1!l26. In a report to the Socond Polish Philosophical Congress, ill 1927 I have '

. t t ded" given, among oth~r things, a simple example of a con-

illS en uctJve system who h f . to

be . IC a ter a smgJe application of this rule ceases

consIStent and is therefi t ' Is IX

282' ore no w'conslStent (cf. p. 258, note 1 ; see a 0 ,

f29), p~~~~_~' Some remarks on this rule are to be found in Hilbert. D.

- "I ThdinUB,. f~r example, if we adopt this rule in the metalanguage without mc u g It m the languag t es

of th '. e, we can prove that the class of provable sen enc

e SCIence IS consistent hi h . 'on

with th - bl ,w IC prevIOusly was not possible. In connes!

IS pro em cf. Godel, K. (22), pp. 187-91 and 196.

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 281 with the class of true sentences. Elementary number theory provides an example, namely, the science in which all vurinbles represent names of natural or whole nurnhera and tho eOIlHtantfil are the signs from the sentential and pr!'di('ate calculi, tho HigliH of zero, one, equality, Bum, product and possibly other fligliH defined with their help. t

If it is decided to adopt the rule of infinite induction in tho metatheory, then the system of axioms to which Th. III refers already forms a sufficient foundation for the development of the theory of truth. The proof of any of the known theorems in this field will then present no difficulty, in particular the Ths. 1-6 in § 3 and the theorem according to which the rule of infinite induction when applied to true sentences always yields true sentences. More important still, these axioms, together with the general axioms of the metatheory, form a categorical (although not a complete) system, and determine unambiguously the extension of the symbol "I'r' which occurs in them.

Under these circumstances the question whether the theory erected on these foundations contains no inner contradiction acquires a special importance. Unfortunately this question cannot be finally decided at present. Th. I retains its full validity: in spite of the strengthening of the foundations of the metatheory the theory of truth cannot be constructed as a part of the morphology of language. On the other hand for the present w.e ca~ot prove Th. III for the enlarged metalanguage. The premiss which has played the most essential part in the original proof, i.e. the reduction of the consistency of the infinite axiom system to the consistency of every finite part of this system, now completely loses its validity-as is easily seen--<>n account of the .content of the newly adopted rule. The possibility that the question ca~not be decided in any direction is not excluded (at least on the basis of a 'normal' system of the meta-metatheory, which is constructed

t This last remark enables us to construct a rather 8imp~e de~~IO;;f truth for elementary number theory without using our genera met . fi e definition thus constructed can be further simplified, In fact we can lJ1It.

structurally describe all true s~ntencea which con:I~r::: ;t,:!'I:I~ri?:~~~!~ flers) and then define an arbitrary sentence to I' H I f be obtained from those elNnentary true sentencee by app ywg ,e ruse 0

infinite induct ion nrbitmrily rnnny t.irnox.

262 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES vm, § 5 according to the principles given at the beginning of§ 4 and does not contain the semantics of the metalanguage). On the other hand the possibility of showing Th. III to be false in its new interpretation seems to be unlikely from the intuitive viewpoint. One thing seems clear: the antinomy of the liar cannot be directly reconstructed either in the formulation met with in § 1 or in the form in which it appeared in the proof of Th. I. For here the axioms adopted in the theory of truth clearly possess, in contrast to colloquial language, the character of partial definitions. Through the introduction of the symbol 'Tr' the metalanguage does not in any way become semantically universal, it does not coincide with the language itself and cannot be interpreted in that language (cf. pp. 158 and 248).1

No serious obstacles stand in the way of the application of the results obtained to other languages of infinite order. This is especially true of the most important of these results-Tho 1. The languages of infinite order, thanks to the variety of meaningful expressions contained in them, provide sufficient means for the formulation of every sentence belonging to the arithmetic

I This last problem is equivalent to a seemingly more general problem of a methodological nature which can be formulated 88 follows. We presuppose the ?onsistency of the metatheory supplemented by the rule of infinite induction, We consider an infinite sequence t of sentences of the metatheory; furthe,r we take into the metatheory 0. new primitive sign' N', and add 88 uxrorns those and only those sentencos which are obtained from the Bcheme 'n E N if and only if p' by substituting for tho sign' n ' the kth specific symbol of the nntural numbers (i.e. the expression composed of k signs '1' separntod from one another by tho signA' + ') and for tho sign 'p' the ~th term of."'le sequence I (k h"ing ilPre nn Brbit.rary natural number). The question now I\rlM.'K whether the cl8HR of prov .. hl" Mont,eIll'OK of the metatheory, when "nll.rgo,1 in ,UliR, W"!' rem .. in. "oll.i .. tont., Thi» I'rnhlNll mny Ito <'nlll'.1 t~" lJro/,km of infinii« tnriurli,,,, ,/rfinili,mR, Th,' ,,"'urn IIYIII"'m d • .,.r,rihl'd in It can-from th" Intuitive standJloint, ~11tl r('K&rtl(~1 &H ,,' definition sui ge11e1tS of tho symbol 'N'. which il< ditltingui .. I .. ,,1 from normal dllHnitioll8 only by the fact that i,t is formulated in infinitely many sentences. In view of this character of the axroms the possibility of a negative solution of the problem does n?t seem very ~robab ~e., From Th. II and the interpretation of the metathoory ~ the t~e,ory Itself,. It IS not difficult to infer that this problem can be solved III ~ positive sense ill those cases in which the order of all variables which ~cur ill the sentences of the sequence t is bounded above. It is then even pOSSible to ~mnstruct '!' definition of the symbol' N' in the metatheory such that all the aDorns ~entlOned follow from it. This problem obviously does not depend on the specl~C properties of the metatheory 88 such; it can also be presented in the same or ill a somewhat modified form for other deductive sciences. e-s- for the general theory of classes.

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 263 of natural numbers and consequently enable the metalanguage to be interpreted in the language itself. It is thanks to just this circumstance that Th. I retains its va.lidity for all langunges of

this kind.'

Some remarks may be added a.bout those casos in which not

single languages but 'whole classes of languages are investiga.ted. As I have already emphasized in the Introduction, the concept of truth essentially depends, as regards both extension and content, upon the language to which it is applied. We can only meaningfully say of an expression that it is true or not if we treat this expression as a part of a concrete language. As soon as ~he discussion concerns more than one language the expression 'true sentence' ceases to be unambiguous. If we are to avoid this ambiguity we must replace it by the relative term 'a true sentence with respect to the given language'. In order to make the sense of this term precise we apply to it essentially the same procedure as before: we construct a common metalanguage for all the languages of the given class; wit~in th~ metalanguage we try to define the expression in questiOn WIth the help of the methods developed in §§ 3 and 4. If we are not successful we add this term to the fundamental expressions of the ~etalangu~ge and by the axiomatic method determine its mearung according to the instructions of Th. III of this section. On account of ~h~

I t· . t' of this term we should nevertheless expect a pnon

re a IVlza Ion ' '

that in carrying out the plan sketched abov~ the earlier ~ffi-

culties would be significantly increased and quite new com~lications might arise (connected for example with the necessity of

.' here' if we choose 8B our starting'point the

I A reservatlon IS necessary "ke hed <>18 note 2 then we

" t' I tegorles 8 tc on P: ~, '

ci888lficatlOn of seman rca ~ainfinite order for which Th. I loses its validity.

again encounter lan~Bges ,0 the Ian e of LeSniewski's Protothetic A typical example IS furrushed by f ~finitistic' character of all the (cf. LeSniewski, S. ~46)). In. conseque:cei~ is easy to construct, in the metasemantical categories o~ ~his Janft guthBgb' choosing 88 model the matrix method I aen> a correct definitIOn 0 ru , Y d fini . be

anguege, t' I al ulus 1I0reover such a e non can

from the extended senten ia TC_~ c, ·ki' has sho~n the cl888 of provable

, ed' th ways' 88 J.A><>IUewS ~, I

obtain 111 0 er ", I te and therefore the concept of provab e

sentences of the protothetlc 18 comp e'th' that of true sentence, Th. 1 on the

, id ' 'te extensIOn WI

sentence comer ee 111 ,I triction to all langullg"" in which the ord"r of

other hand ~ppheA wlth,out fros thA domain of Leimi"wMki'. O,,'ol,,'J'J (,.f, the semantlral cat .. gor .. '" rom

LeSniewski. S. (47)) iA not bounded above,

264 COXCEPT OF TRrTH IX FOR~L\LIZED L-\SGr_-\CES VIII, § 5 defining the word 'language'). I do not propose to discuss the problem touched upon in more detail in this place. The prospects for such investigations at the present time seem to be rather limited. In particular it would be incorrect to suppose that the relativization of the concept oftruth=-in the direction mentioned above-would open the way to some general theory of this concept which would embrace all possible or at least all formalized languages. The class of languages which is chosen as the object of simultaneous study must not be too wide. If, for example, we include in this class the metalanguage, which forms the field of the investigations and already contains thc concept of truth, we automatically create the conditions which enable the antinomy of the liar to be reconstructed. The language of the general theory of truth would then contain a contradiction for exactly the same reason as does colloquial language.

In conclusion it may be mentioned that the results obtained can be extended to other semantical concepts, e.g. to the concept of satisfaction. For each of these concepts a system of postulates can be set up which (1) contains partial definitions analogous to the statements described in condition (eI:) of the convention T which determine the meaning of the given concept with respect to all concrete, structurally described expressions of a given class (e.g. with respect to sentences or sentential functions of a specific scmantical type), and (2) contains a further postulate which corresponds to the sentence from the condition (f3) of tho Ramo convent.ion and st.ipulntes that, HlP concept in qll£'st.ion may be applied only to (1X pn'Hl-lioTis of t.ho given e1nss. \Vo I-Ihould ho pr()paI'1141 to f(·gn.rd RIIOh n dofinition 411' t.l14l (,ulw(·)!I, SI.IIr1i4,tIILR II. mnteriully ruloquuto 01\(\ if its ('OIlSNIIH'IIGOH indtl4l(.d 1111 tho postulates of tho above syHtem. Methods which are similar to those described in §§ 3 and 4 enable the required definition to be constructed in all cases where we are dealing with languages of finite order, or, more generally, in which the semantical concept studied concerns exclusively linguistic expressions in which the order of the variables is bounded above (cf. Th. II). In the remaining cases it can be shown-after the pa ttern of the proof of Th. I-that no definition with the properties mentioned can be

VIII, § 5 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 265

I t d J the metalanguage.l In order to construct the

formu a e mI' t I (J in

theo of the concept studied in these oases It AO, It must n _

ry . f . itive concepts. awl the postllirLto

eluded in the system 0 prinu f tI

d ib d hove must he included in the axiom HYHt('In (I .ne

cscn e a Iff 'I'h I II

th r A procedure analogous to t re pro!) (I .' •

meta co y the s stem of the metalanguage HupplcJn(,J1.teri III t1UR proves tha~ . [ Ily consistent. But the deductive power way remains merna . d 'l'h do not suffice

f th dded postulates is very restricted- ey .

() e a f f the most important general theorems concerm.ng

for the proo 0 • Th do not determine its extenSIon

th ept in questIOn. ey

e conc btained is not complete, nor

bi uously and the Bystem 0 t

unarn ig thi d fect we must strengtllen

. . I To remove IS e

is It categonca. heorv i t If by adding the rule of

the foundations of the metat eOfr~ £1 ~e But then the proof

. d ti to its rules 0 ill erence.

infinite m uc IOn t difficulties which we are not

of consistency would present grea

able at present to overcome.

§6. SUMMARY .

. I he summarized In the The principal results of this artie e may

following theses:. .r finite order a formally

F rormahzed language OJ b

A. or every J' d ,f: 't'on or true sentence can e

. II adequate eJ,m t 'J

correct and matena Y e making use only of expressions of

constructed in the metalanguag., .r th language itself as icell as

. l ki d expresswns OJ e .

a generallogtca m, 0 0 language, i.e. names of lin-

terms belonging to the morpb logy t1 l relations existing benceen

. d .r the struc ura c.

guistic expresstons an OJ

them. . es of infinite order the construction

B. For formahzed languag .

of such a definition is imposSlble. .

t of definability (although in this caee

1 This especially concerns the ~oncefself 8B well as the method of solut~on,

both the formulation of the p~ob emp'aris~n with the scheme put forwabrl tin

dill ti ns In com 't' 'mpossl e 0 require certain rno ca 10 ad the conjecture that 1 19 1 I can

I VI I have express . f the metalanguage.

~:~t;~~;s c:nce~t in its full ext~nt ~~h~~:-"; ~ll the more n';t" w orthy i~

. his conjecture exact y'. . le mentioned-to ('OnHtrur.

~~~ .f:fin~~.etP~S:ibol~= ~a~i:~I:~o;8-:,:n o~h:h:~fl~~~('fr:::e~~I~;t~~~~, :::;:~.

the de I ions 1 e, but to any 0 I~ .C' •..

Iy not to the whole lang. ag I . the langll&gll Iw·lf.

app, . h talang.lsgo but a "0 In

not only In t e me

266 CONCEPT OF TRUTH IN FORMALIZED LANG"cAGES VIII, § 6

C. On the other hand, even with respect to formalized languages of infinite order, the consistent and correct use of the concept of truth is rendered possible by including this concept in the system of primitive concepts of the metalanguage and determining its fundamental properties by means of the axiomatic method (the question whether the theory of truth established in this way contains no contradiction remains for the present undecided).

Since the results obtained can easily be extended to other semantical concepts the above theses can be given a more general form:

VIII, § 6 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 267 problem-the construction of the definition of true sentence and establishing the scientific foundations of the theory of t~I1,t~ +-belon s to the theory of knowledge and forms one of tho dll~f

'0 ble';' of this branch of philosophy. I therefore hopo tI.,,1 tin: !ork will interest the student of the theory of knowJcd~o al~o~1, all and that he will be able to analyse the results contam~d 10 ~t critically and to judge their value for, further researches III thl: fi ld without allowing himself to be discouraged by the appara

e , d thods used here, which in places have been

t~s of concedPhts an mtehitherto been used in the field in which he difficult an ave no

works, . I' Philosophers who are not accustomed

One word m cone USlOn, . I' d t

ive methods in their daily work are me me 0 to use deduct, a es with a certain disparagement, regard all formalized langu g , " t ti ith the t th se 'artificial cons rue ions WI because they con!r:~th: colloquial language, For that reason one natural langu g . d the formalized lan-

e th t th results obtaine concern

the fact a e , will eatly diminish the value of

guages almost exclusively gr" of many readers It

' , ti ations in the opmIOn ,

the foregoing mves Ig h thi view In my opinion the

would be difficult for me to s areh t~:allY ~hat the concept of

id ti of § 1 prove emp a I

consi era IOns ti I c ncepts) when applied to

h ( II as other seman ICa 0 ,

trut as we iuncti ith the normal laws of logic

' II e in conjuno Ion w .

colloquia anguag , d trsdictions. Whoever

' it bl to confusions an con f

leads mevi a y 'It' to pursue the semantics 0

. h . spite of all difficulties, will be

WIS es, m ith the help of exact methods

colloquial language WI the thankless task of a reform of driven first to undertake . ry to define its structure,

' He will find It necessa , ,

this language. izuit of the terms which occur m rt,

to overcome the ambigui y . to a series of languages of and finally to split the languaghe IDf hich stands in the same

d ter extent, eac 0 w ,

greater an grea ich a f lized language stands to Its relation to the next in which a orma be doubted whether the

It may however, ,

metalanguage. , ' after being 'rationalized' in this way,

language of everyday_ life, I and whether it would not

ill erve Ita natura ness I' ed

would st pres . t' features of the forma lZ

rather take on the characteris IC

languages.

A', The semantics of any formalized language of finite order can be built up as a part of the morphology of language, based on correspondingly constructed definitions,

B/, It is impossible to establish the semantics of the formalized languages of infinite order in this way.

C'. But the semantics of any formalized language of infinite order can be e8tablished as an independent science based upon its oum primitive concepts and its oum axiorns, possessing as its logical fOUndation a system of the morphology of language (although a full guarantee that the semantics constructed by this method contains no inner contradiction is at present lacking).

From the formal point of view the foregoing investigations have been carried out within the boundaries of the methodology of the deductive sciences. Some so to speak incidental results will perhaps ho of intorest to s}lceiaIi8t,H in thiH ti(,ltl. I would draw uttention to UIO filet. t'Ult with f,ho dl,tinition of true sentence for !lnductivo sciences of tinite order It general method has been ohtained for proving their ConsiHf,ency (u method which, however, does not add greatly to our knowledge), I would point out also that it has been POssible to define, for languages of finite order, the concepts of correct sentence in a given and in an arbitrary individual domain-concepts which playa great part in recent methodolOgical studies.

But in its essential parts the present work deviates from the main stream of methodolOgical investigatioDs. Its central

268 CONCEPT OF TRUTH IN FOR:lIALIZED LANGUAGES VIII, § 7

VIII, § 7 CONCEPT OF TRUTH IN FOR~ALIZED LANGUAGES 269 axiom, consequence, and provable theorem. Thus, for example, we include as a rule among the axiollls·--jusl as in the language of the general theory of classes in § a-the ~lIlbHtituti()IIH of Ow a.xioms of the sentential calculus, the PH(~lId()·ddinili()nH and tlu. law of extensionality (perhaps ulso other Sl'ntmH't'H, IltT~tr~lillg to the specific peculiaritics of the language). In deterlllllilng t.ho concept of consequence wc take as our model Def. I!) in § ~:

The concept introduced in § 4 of the order of an. eXprl'HHIIIII plays a part which is no less essential t~Ian .before .10 th~ ~onstruction of the language we are now consl(lerlIl~. It IS adv isable to assign to names of individuals and to the variables repreHenting them the order 0 (and not as before the order 1_l .. T.he order of a sentence-forming functor of an arbitrary (prII~utIve) sentential function is no longer unambiguously determm.ed .by the orders of all arguments of this function. Since the pnncl!J!es of the theory of the semantical categories no longer hold, It m~y

h t d the same sign plays the part of a functor m

happen t a one an .

tential functions in which arguments occupymg

two or more sen di r t

ti 1 the same places nevertheless belong to If eren

respec ive y . t t, k

d Thus in order to fix the order of any Sign we m~18 a e

~r ers. t th orders of all arguments in all scntentlal func-

mto accoun e . If th

. . hich this sign is a sentence-formmg functor. e

trons ill WI· I at.urs I t is smaller than a particu ar na ura

order of all these argumen s I . f .

. h in at least one sentential unction

b n and If t ere occurs 1

:~:r;m~nt which is exactly of order n -1, then we assi~ to ~he

. uestion the order n. All such sentence-Iofllllng

symbol ill q f i dividuals and the variables

functors-as well as the names 0 in . d

. I d d among the sizns of finite or er.

ti them-are me u e 0'-

represen mg of the ssibility that yet other

But account must also be taken ~ to hich

. f t ay occur ill the language w

sentence-formmg unc ors mI' " a

. finite order must be assigned. If, for ex amp e, ~ SIgn ~

an ill 1_ . functor of only those sentential functIOns which sentence for~g f finite order, where, however, these have all theIr arguments 0 " b an natural number, then this

orders are not bounded abov e Y Y

. '11 b of infinite order. f

SIgn WI e '. f i fi ite ()rdl"r W(' make \HW I,

I ··f the signs 0 m iru ...

In or.der tfo cl~ns:l:umh;r, taken from the tlwl)ry ofHclH, whidl the notion () or; ~

§ 7. POSTSCRIPT

In writing the present article I had in mind only formalized languages possessing a structure which is in harmony with the theory of seman tical categories and especially with its basic principles. This fact has exercised an essential influence on the construction of the whole work and on the formulation of its final results. It seemed to me then that 'the theory of the semantical categories penetrates so deeply into our fundamental intuitions regarding the meaningfulness of expressions, that it is hardly possible to imagine a scientific language whose sentences possess a clear intuitive meaning but whose structure cannot be brought into harmony with the theory in question in one of its formulations' (cf. p. 215). Today I can no longer defend decisively the view I then took of this question, In connexion with this it now seems to me interesting and important to inquire what the consequences would be for the basic problems of the present work if we included in the field under consideration formalized languages for which the fundamental principles of the theory of semantical categories no longer hold. In what follows I will briefly consider this question.

Although in this way the field to be covered is essentially enlarged, I do not intend-any more than previously-to consider all possible languages which someone might at some time construct. On the contrary I shall restrict myself exclusively to languages which-apart from differences connected with the theory of aernnnt.ical categories-exhihit. in their fltrnct.ure the greatest possible analogy with the Innguag('fl proviollHly studied. In pnrt.icular, for the sake of simplicity • .1 shall consider only those languages in which occur, in addition to the universal and existential quantifiers and the constants of the sentential calculus, only individual names and the variables representing them, as well as constant and variable sentence-forming functors with arbitrary numbers of arguments. After the manner of the procedure in §§ 2 and 4 we try to specify for each of these languages the concepts of primitive sentential function, fundamental operations on expressions, sentential function in general,

~70 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 7

IS a ge li ti

As is w:;ra za Ion of the usual concept of natural numbsr.!

b ~own, the natural numbers are the smallest ordinal

then ers. Since, for every infinite sequence of ordinal numbers

ere are numbers te h '

th . grea r t an every term of the sequence

ere are m part' 1 ' , lcU ar, numbers which are greater than all

natural numbers W II th

1m .' e ca em transfinite ordinal numbers. It is

own that In every n t 1 .

. on-emp y c ass of ordinal numbers there

IS a sbmalleshi~ nu~ber. In particular there is a smallest transfinite

num er w ch IS denoted b th b"

numb . y e sym 01 w. The next largest

er IS w+ I, then follow the numbers +2 +3 2

w.2+1 w.2+2 w , w , ... , w. ,

d '. "", w. 3, ... , and so on. To those signs of infinite

or er which are functo f .

exclusi I rs 0 sentential functions containing

srve y arguments of finite d

as thei d or er we assign the number w

err or er A sign whi h . f

functi ··hi c 18 a unctor in only those sentential

aons In w ch the a .

order ( d' . rguments are either of finite order or of

w an In which at least

actually of d' one argument of a function is

d finiti or er w), 18 of the order w+ 1. The general recursive

e IOn of order is as f II h

the small t di 1 0 ows: t e order of a particular sign is

es Or Ina number hi h i

a' w c 18 greater than the orders of all

rguments In all sentential functions in which the given sign occurs as a sentence-forming functor 2

Just as in § 4 we ca <list' . .

order W '. f n mgmsh languages of finite and infinite

. ecanm act as . to

ordinal nu b . sIgn every language a quite specific

m er as Its order n 1 th

which ds ' arne y e smallest ordinal number

excee the orders of II . bl

guage (th f I a varia es occurring in this lan-

e ormer anguages of th th d

shown-ret· thei e n or er-as can easily be

the order fatIDh eir former order under this convention because

o e namcs f i I' . 1

language of tl 0 ID( Wit uals has boon diminished. The

I t does not 16 :e~~~aJ theory of classes has the order w). variable in th al a oll0--:V from these stipulations that every

e anguages m qu ti ,

the contra 't es IOn IS of a definite order. On

ry I seems to me (b .

siderations) 1 t y reason of trials and other con-

a mos certain that .

the use of variables of de' we .cannot restrict ourselves to

finite order if we are to obtain languages

1 Cf Fr

2 • ~nkel, A. (16), pp. 185 ff

Cf. the Introduction of th .

(p. 186 in English tra.nslation~ system oflevels in Carnap, R. (10), pp. 139 fl.

VIII, § 7 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 271 which are actually superior to the previous languages in the abundance of the concepts which are expressible by their means, and the study of which could throw new light on tho problems in which we are here interested. We must introduce into ttw languages variables of indefinite order which, 1'10 to speak. 'run through' all possible orders, which can occur as functors or arguments in sentential functions without regard to the order of the remaining signs occurring in these functions, and which at the same time may be both functors and arguments in the same sentential functions. With such variables we must proceed with the greatest caution if we are not to become entangled in antinomies like the famous antinomy of the class of all classes which are not members of themselves. Special care must be taken in formulating the rule of substitution for languages which contain such variables and in describing the axioms which we have called pseudodefinitions. But we cannot go into details here. I

There is obviously no obstacle to the introduction of variables of transfinite order not only into the language which is the object investigated, but also into the metalanguage in which the investigation is carried out. In particular it is always possible to construct the metalanguage in such a way that it contains variables

1 From the languages just considered it is but a step to languages of another kind which constitute a much more convenient and actually much more frequently applied apparatus for the development of logic and mathematics. In these new languages all the variables are of indefinite order. From the formal point of view these are languages of a very simple structure; according to the terminology laid down in § 4 they must be counted among the languages of the first kind, since all their variables belong to one and the same semantical category. Nevertheless, as is shown by the investigations of E. Zermelo and his successors (cf. Skolem, Th. (66), pp. 1-12), with a suitable choice of axioms it is possible to construct the theory of seta and the whole of classical mathematics on the basis provided by this language. In it we can express so to speak every idea which can be formulated in the previously studied languages of finite and infinite order. For the languages here discussed the concept of order by no means loses ita importance; it no longer applies, however, to the expressions of the language, but either to the objects denoted by them or to the language as a whole. Individuals, i.e. objects which are not seta, we call objects of order 0; the order of an arbrtrary set is the smallest ordinal number which is greater than the orders of all elements of this set; the order of the language is the small .... t ordinal number which exceeds the order of all seta who"" existence follow. from the axioffill adopted in the language. Our further eXpoHition alJtO applw. WIthout "",tnr. tion to the Ianguegoe which have just been ,hJiCu""",J_

272 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 7 of higher order than all the variables of the language studied. The metalanguage then becomes a language of higher order and thus one which is essentially richer in grammatical forms than the language we are investigating. This is a fact of the greatest importance from the point of view of the problems in which we are interested. For with this the distinction between languages of finite and infinite orders disappears-a distinction which was so prominent in §§ 4 and 5 and was strongly expressed in the theses A and B formulated in the Summary. In fact, the setting up of a correct definition of truth for languages of infinite order would in principle be possible provided we had at our disposal in the metalanguage expressions of higher order than all variables of the language investigated. The absence of such expressions in the metalanguage has rendered the extension of these methods of construction to languages of infinite order impossible. But now we are in a position to define the concept of truth for any language of finite or infinite order, provided we take as the basis for our investigations a metalanguage of an order which is at least greater by 1 than that of the language studied (an essential part is played here by the presence of variables of indefinite order in the metalanguage). It is perhaps interesting to emphasize that the construction of the definition is then to a certain degree simplified. We can adhere strictly to the method outlined in § 3 without applying the artifice which we were compelled to use in § 4 in the study of languages of the 2nd and 3rd kinds. We need neither apply many-rowed sequences nor carry ont the sernanticnl unificnt.ion of the variables, for having uhandonnd the principle» of thn t.hoory of Honmntical categories we can freely operate with He41UI'IWml whose terms are of different orders. On the other hand the c01l8iticflttions brought forward in § 5 in connexion with Th. I lose none of their importance and can be extended to languages of any order. It is impossible to give an adequate definition of truth for a language in which the arithmetic of the natural numbers can be constructed, if the order of the metalanguage in which the investigations are carried out does not exceed the order of the language investigated (cf. the relevant remarks on p, 253),

VIII, § 7 CONCEPT OF TRUTH IN FORMALIZED LAXClUAGES :173

Finally, the foregoing considerations show the necessity of revising, to a rather important extent, the Theses A and B giv('n in the conclusions of this work and containing !L sunuuury of its

chief results:

A, For every formalized lanquaqe aformllll!J correct and matcrially adequate definition of true sentence can be. ron8Irurtr(~ in ilu: metalanguage with the help only of qeneral loqica! e.rprrSSI(J1/8, (if expressions of the language itself, and of terms from the morp/tolo(}y of language-but under the condition that the melalrz1UJ~g(' rr:: a higher order than the language which is the object of incesiiqulum;

B, If the order of the metalanguage is at most equal to that of the language itself, such a definition cannot be constructed.

From a comparison of the new formulation of the two theses with the earlier one it will be seen that the range of the results obtained has been essentially enlarged, and at the same time ~he

diti f r their application have been made more precise.

con 1 ions 0 , . C

In view of the new formulation of Thesis A the former 1 hesis

I its i t ce It possesses a certain value only when the

oses 1 s Impor an . .. . .

, t' ti carried out in a metalanguage which has the mves Iga IOns are I

d th language studied and when, having abandoner

same or er as e ,

h t t' f a definition of truth the attempt IH made to

t e cons rue IOn 0 ' ,

build up the theory of truth by the axiomatic method. It IS ea~y

to see that a theory of truth built up in this way cannot conta.m

. t diction provided there is freedom from contra-

an Inner con ra , . f h' h

diction in the metalanguage of higher order on the.basIs _o w ic

an adequate definition oftruth can be set up and ill WhICh t~lOse theorems which are adopted in the theory of truth as axioms

can be derived.' Th :\ d B

J st 'the conclusion of this work, the eses - an

us as ill ,t din them

be zi more general formulatIOn by ex en g

can e gIven a

to other semantical concepts:

f. mal' ed language can be established

A', The semantics of any Jor lZ , ed

1. ol .r la uage based on SUitably wnfjtructe

as a pari of the morprw ogy OJ ng

. h 1 '''II h"" n T,(,Hitiv" answor. n,,,

h uestlOn broBl" '" on P: - I

I In particular. t o q , f ' f 'to indU'ti\,lI d.·t,rut.IO'lJI <O"nl,(,"("

snrno n l-o holds for tho prohlern 0 rn ".1

on p. ~f)~, footnote.

274 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 7

de-/: .. .

J.mtt~, prm:~ded, however, that the language in whick the mor-

phology ~s earned out has a higher order than the language whose morphology it is.

.B/. It ~s impossible to establish the semantics of a language in th~s way if the order of the language of its morphology is at most equal to that of the language itself.

The Thesis A in its new generalized form is of no little importance for the methodology of the deductive sciences. Its con-

sequences run parallel ith th . . ..

WI e Important results which Godel

has reported in this field in recent years. The definition of truth

allows the oonsisten f d dueti .

. cy 0 a e uctive SCIence to be proved on the

~asls of a mctatheory which is of higher order than the theory It~~lf (cf. pp. 199 and 236). On the other hand it follows from

Godel's . t" . . . '

~ves igations that It IS ill general impossible to prove

the consistency of a theory if the proof is sought on the basis of a metatheory of equal or lower order.' Moreover Gadel has given a method for constructing sentences which-assuming the theory ~onc~rned to be consistent-cannot be decided in any direction

in this theory All se t ." l'

. n ences constructed accordmg to Gode 8

method possess the property that it can be established whether they are true or false on the basis of the metatheory of higher

order having a corre t d fi iti . .

. c e ru Ion of truth. Consequently It IS

possible to reach ad' .. .

. eCISlOn regarding these sentences, I.e. they

can be either pro d di

ve or sproved. Moreover a decision can be

reached within the' . .

SCIence Itself, WIthout making use of the

concepts and as t'

sump IOns of tho mctathcory-provided, of

course, that we hav . I .

I . e previous y enriched the language and the

~glCal. foundations of the theory in question by tho introduction

o variables of higher order.s Let us try to I' hi

. exp am t 8 somewhat more exactly. Consider

an arbitrary deductive science in which the arithmetic of

natural numbers ca b .

the i ., n e constructed, and provisionally begm

e investigation on the basis of a metatheory of the same order asthetheoryitself G"d l'

t . 0 e s method of constructing undecidable

sen ences has been outlined implicitly in the proof of Th. I in ; cr. Godel, K. (22), p. 196 (Th XI)

cr. GOdel K (22) ..

,. , pp. 187 B., and in particular, p. 191, note 48 a·

VIII, § 7 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 276 § 5 (p. 249 fl.). Everywhere, both in the formulation of the theorem and in its proof, we replace the symbol "I'r' by tho symbol 'Pr' which denotes the class of all provable HcntellC(11i of the theory under consideration and can be defirwd in tho metatheory (cf. e.g. Def. 17 in § 2). In accordance with the firHt part of Th. I we can obtain the negation of one of the scntenceH in condition (ex) of convention T of § 3 as a consequence of tho definition of the symbol 'Pr' (provided we replacc "I'r' in this convention by 'Pr'). In other words we can construct a sentence x of the science in question which satisfies the following

condition:

it is not true that x E Pr if and only if p or in equivalent formulation:

(1) z E Pdf and only if P

where the symbol 'p' represents the whole sentence x (in fact we may choose the sentence Ui(tk.,h) constructed in the proof

of Th. I as x). .

We shall show that the sentence x is actually undeCidable

and at the same time true. For this purpose we shall ~asB ~ a metatheory of higher order; Th. I then obviously remal~s valid.

. . t t on the basiS of the

According to TheSIS A we can cons rue , .

. t d finiti n of truth concernmg

enriched metatheory, a correc e 10

all the sentences of the theory studied. If we denote the class

of all true sentences by the symbol 'Tr' then-in aecorda: with convention T -the sentence z which we have construe

will satisfy the following condition:

(2) x E Tr if and only if p;

from (1) and (2) we obtain immediately

(3) x E Pr if and only if z E Tr.

. f the sentence x by the

Moreover, if we denote the negatiOn 0 h d fini-

r II . theorems from tee

symbol 'i' we can derive the 10 owmg

tion of truth (of. Ths. 1 and 5 in § 3):

(4) either x € Tr or i € Tr;

(5) if x E Pr, then z E Tr;

(6) if:i E r-, then :i E Tr;

276 CONCEPT OF TRUTH IN FOR:\{ALIZED LA...·'WUAGES VIII, § 7

From (3) and (5) we infer without difficulty that

(7) x E Tr

and that

(8) x E Pro

In view of (4) and (7) we have x E Tr, which together with (6) gives the formula

(9) x E Pr.

The formulas (8) and (9) together express the fact that z is an undecidable sentence; moreover from (7) it follows that x is a true sentence.

By establishing the truth of the sentence x we have eo ipso -by reason of (2)-also proved x itself in the metatheory. Since, moreover, the metatheory can be interpreted in the theory enriched by variables of higher order (cf. p. 184) and since in this interpretation the sentence x, which contains no specific term of the metatheory, is its Own correlate, the proof of the sentence x given in the metatheory can automatically be carried over into the theory itself: the sentence x which is undecidable in the original theory becomes a decidable sentence in the enriched theory.

I should like to draw attention here to an analogous result.

For every deductive science in which arithmetic is contained it is possible to specify arithmetical notions which, so to speak, belong intUitively to this science, but which cannot be defined on the basis of this science. 'With the help of methods which are complct(1iy annlogoliH to thos« uHl'ti ill UIO construction of tho dofinition of truth, it ill rUlV('rt,JwJI'HH pOI'IHiblo to show that theso concepts can be so dofiutltl provided t.ho soienco is enriched by the introduction of variables of higher order.'

In conclusion it can be affirmed that the definition of truth and, more generally, the establishment of semantics enables us to match some important negative results which have been obtained

1. C!. my summary, 'TIber definierbare Mengen reeller Zahlen,' Annales de 16 Sor:tete Poionaiee de Mathematiqm, t. ix, annee 1930, Krakow, 1931, pp. 206-7 (report on a lectu~e give~ on 16 December 1930 at the Lemberg Section of the rohs~ Mathematical Society); the ideas there sketched were in part developed ater III VI. Cf, VI, p. HO, Bibliographical Note.

VIII, § 7 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES 277 in the methodology of the deductive sciences with parallel positive results, and thus to fill up in some measure tho gapli thereby revealed in the deductive method and in the ediflco of deductive knowledge itself.

HISTORICAL NOTES. During the six-year period, I"'ginnillg with 111211 wln-n I arrived at the final formulation of the definition of truth along with mo"t of the remaining results presented in this work and ending with 193[, when the whole work appeared for the first time in a language inh'rnatirmally accepted for the exchange of scientific ideas, the problems investigated h .. rf' were discussed several times. Thus in the German language, in addition to my summary, Tarski (76), works by Carnap have appeared in which quite similar ideas were developed (cf. Carnap (lOa), 1934, and Camap (II), 1935).

It was to be expected that, in consequence of this lapse of six y nrs,

and of the nature of the problem and perhaps also of the lan~uo.g of

the original text of my work, errors re~ard~ng the historical connexiona might occur. And in fact Carnap Writes ill the second of the abovomentioned articles regarding my investigations that they have be-r-n carried out' ... in connexion with those of Good ... ', It will t Iu-r« fore not be superfluous if I make some remarks in this placr- about tho dependence or independence of my studies.

I may say quite generally that all my methods and results, Wjithh tho exception of those at places where I have expressly cmphasizer t Ill--

f<" t t 154and247-wercobtainedbyrnefl'utemd'·Twwl<-ntly.

c .100 no es, pp. .... I .

The dates given in footnote, p. 154, provide, I brliev«, HllfTWlf"lJt Ja>i1!1

for testing this assertion. I may point out further that my art icl« which

d i F h (VI) about which I had already report--d m Df:cr·rn·

appeare ill renc, . . '. '" •

ber 1930 (cf. the report in Gennan III A. 'I'arski (74» contains PTf.CISr])

those methods of construction which were used there for othe-r purposes

b t· th se t work for the construction of the definition of truth.

u In e pre n . '.

I should like to emphasize the independence of my illvestlg~tlOns

. h r 11 .' g points of detail: (1) the general formulation of

regarding t e ,0 owin ..

bl f d fining truth cf especially pp. 18;-8; (2) the positrva

the pro em 0 e , . f h f h

. f I bl . e the definition of the concept 0 trut or t e

solution 0 t re pro em, I. . . .

h th eans available in the metalanguage are sufficiently rich

case were e m ,~ , I' 1 •

. 11 this definition becomes that of the term ana ytica

(for logica anguages d f .

C ) Cf pp 194 and 236' (3) the metho 0 proving con-

used by arnap. . .' f d ?36

. b . f the definition of truth, cf, pp. 1 ~9 an - ;

sistency on the asis 0 f 1-3 ff d .

. ti truction of the metasvstern, c . pp. I ., an ill

(4) the aXlOma Ie cons • t, •

. . h thi (5) the discussions on pp. 184 f. on tile interpreta-

connexion Wit 18

2.78 CONCEPT OF TRUTH IN FORMALIZED LANGUAGES VIII, § 7

tron of the metasystem in arithm t' .

called 'method of arithm t" th e ic, which already contain the 80.

e lZmg e metalangua • hi h

far more completely and ite j ge W lC was developed

should like to draw att t':!w et mdependently by Godel. Moreover, I en IOn 0 results not relati t th

truth but to another sem t' I mg 0 e concept of

on p. 276. an ica concept, that of definability reported

In the one place in hi h .

GOdel-in the n . w lC. my work IS connected with the ideas of

for the case Wh:~:t~~: s~~:~~:n of the problem _of the definition of truth investigated_ I h guage 18 not richer than the language

ave naturally expressl hasi

p. 247, footnote}' it may b . Y emp asized this fact (cf.

very much com~leted e me:tlOned that the result so reached, which

to the otherwise alread m~ ':°hr d' ~as t~e o~ly one subsequently added

y rus e mvestlgatlOn.

IX

SOME OBSERVATIONS ON THE CONCEPTS OF w-CONSISTENCY AND w-COMPLETENESSt

IN an extremely interesting article GlJdeP introduces the concept of w-consistency, and constructs an examplc of a deductive system which is consistent in the usual sense, but is not w-consistent. In the present article I propose to give another simple example of such a system, together with some general remarks on the concept mentioned as well as on the corresponding concept of w-completeness.z

The symbolical language in which I shall construct this system is closely related to the language of the system P used by Godel. It is also the result of an exact formalization and simplification, as far as possible, of the language in which the system of Principia Mathematica of Whitehead and RusselJ3 is constructed. In spite of its great simplicity this language suffices for the expression of every idea which can be formulated in Principia M athematica.4

1 See Godel, K. (22).

2 Already in the year 1927, at the Second Conference of the Polish Pbilosophical Society in Warsaw in the lecture 'Remarks on some notions of the methodology of the deductive sciences' (listed by title in Rueh FiUJzojiczny, vol. 10 (1926-7), p. 96), I had pointed out the importance of these concepts, and the rule of infinite induction which is closely related to them and about which more is said in the text, but I had not suggested special DAmes for these concepts. I also communicated the example of a consistent and yet not ... - consistent system which I give in the present article in a slightly altered form. Naturally it is not hereby claimed that I already knew then the results later obtained by GOdel or had even foreseen them. On the contrary, I had personally felt that the publication of the work of GOdel cited above W&8 a most exciting scientific event.

• Whitehead, A. N., and Russell, B. A. W. (90).

• cr. articles VI and VIII of the present work, where I have used the same or a very similar language.

t BIBLIOGRAPHICAL NOTE. Thill article first appeared under the title 'Einige Betrachtungen iiber die Begritfe w.WidersprucMfreih~t und de!" w-Vollstii.ndigkeit', MoniJUMfk fur MalMmatilc und Phylilc, vol. 40 (1933), pp. 97-112. For the historical inronnation about the ,..,.ult. or thia &rt.icle see footnote 2 above.

400

OK EXTENSIONS OF INC011PLETE SYSTEMS

XIV

'Ve turn now to those systems of the sentential calculus in which two constants occur, the implication sign and the negation sign. For these we have

THEOREM 3. If X is a consistent set of sentences which contains the four sentences CpNNp, CqCpq, CNpCpq, and CpCNqNCpq as elements, then the ordinary (two-valued) system of the sentential calculus is the only consistent and complete system which includes the set X.

The proof is completely analogous to that of Th. 2 and is based on the following lemma which corresponds exactly to Lemmas 6 and 7:

LEMMA. 8. If x, = {CpNNp, CpCqp, CNpCpq, CpCNqNCpq} and Z = {CpNNp}, then: (a) for every sentence a E Sbz(S) either a E Cn(Xo) or Na E Cn(Xo), (b) Xo is complete with respect to the set Sbz(S), and (c) the set Xo+Sbz(S) is inconsistent.

Finally we consider systems of the sentential calculus in which, in addition to the signs of implication and negation, two other constants occur, the signs of disjunction and conjunction:

THEOREU 4. If X is a consistent set of sentences which contains the fallowing ten sentences as elements: CpNNp, CqCpq, CNpCpq, CpCNqNCpq, CpApq, CqApq, CNpCNqNApq, CpCqKpq, CNpNKpq, and CNqNKpq, then the ordinary (two-valued) system of the sentential calculus is the only consistent and complete system which includes the set X_

To the systems X which satisfy the hypothesis of Th. 4 (or 3) belong again all many-valued systems of Luknaiowiez and also, among others, the intuitioni.~lic s!/stem of tho RllIIj,onj,jal cnleulus of A, Heyting.' For oach of thet!<l syatoms t.lw ordiuury twovalued system constitutes the only consistent and complete extension,s

1 Heyting. A. (28).

• Thi,s result was previously obtained for the many. valued systems of the sente~tlal calculus by another method. cf. IV. p. 49. Th, 21 and the accompanymg footnote.

XV

THE ESTABLISHMENT OF SCIENTIFIC SEMANTICSt

. , . . a narrower sense than usual.

THE word 'semanblea IS used he~e fih t t lity of considerations

t d b semantiCS teo a 1

We shall unders an Y hl speaking express

. ts which roug Y , d

concermng those concep '. of a language an

. b tween the expressIOns

certain connexions e . d to b these expressions.

the objects and states of affairs referre : we may mention

If semantical concep shih

As typical examp es 0 . ,f. t' and definition, w c

the concepts of denotation, sahsJac ~on, ts:

. th following statemen .

appear, for example, III e xt ~rJeon' snow

• .F J ena' denotes l.' a1"'. ,

The expression 'the mctor OJ t' 'x3 = 2' defines

., , 's white" the equa ~on

satisfies the conditton x ~ b 't J the number 2.

. . 1) the cu e roo OJ • d

(determines unique y . monly recogruze

1 d this IS not com .

The concept of truth a so-an " 1 'cal interpretatIOn,

h t least III Its c assl di

-is to be included ere, a.. h me as 'correspon mg

. h 't ' SI~;ties t e sa

according to whic rue 5'~

with reality'. . semantics have traditionally

Concepts from the domaIll of. . ns of philosophers,

rt in the discusslO b n

played a prominent pa T rth less they have long ~

hi! 1 . ts Neve e - 1 point

logicians, and p a OgIS. .' III From the histonca

regarded with a certain.sceptlllcflS 'ded' for, although the cO~-1

t· . m IS we ounuev. . lloquia

of view this seep ICIS they occur m co .

tent of the semantical concepts, as ts to characterize this

h yet all attemp .' ons in

language, is clear enoug , . d d various discuss1 .

. I have faile , an d on quite

content more preCIse y d hich were base

which these concepts appea~ed an wmisses, have often led ~

. gly eVIdent pre . h the antl-

Plausible and seemID ffi to mentlOn ere

. . It su ces bi h

paradoxes and antIllonues. 'SUJ11IDBXY of an addresS ~9~5

This article 18 a. bilosophy in PsrJS. .:

t BIBLIOGRAPHICAL N~TE. Congress of Scientdi: P ntowaniu naukow~J

was given at the InternatlOnal r h under the title 0 ugru 5()-57 and Later ,n The article first appeared i~ Po '~ vol. 39 (1936). PPiichen Scmantik', in sema.ntyki' in Przeglqd. Fdozofi Y, g der wiasenschaft I 3 (Aclua1ill ...

, . 1 'Grundlegun S· lijiqut. vor,

German under the tit e. cU Phi/.osophie cttn 1-8.

Acte.o du Congrls Intemat.onal I 390) Paris. 1936, pp'

Scientifiques et Industrielles• vo . ,

402 ESTABLISHMENT OF SCIENTIFIC SEMANTICS XV nomy of the liar, the Grelling-N elson antinomy of heterological terms and the Richard antinomy of definability.

The main source of the difficulties met with seems to lie in the following: it has not always been kept in mind that the semantical concepts have a relative character, that they must always be related to a particular language. People have not been aware that the language about which we speak need by no means coincide with the language in which we speak. They have carried out the semantics of a language in that language itself and, generally speaking, they have proceeded as though there was only one language in the world, The analysis of the antinomies mentioned shows, on the contrary, that the semantical concepts simply have no place in the language to which they relate, that the language which contains its own semantics, and within which the usual logical laws hold, must inevitably be inconsistent. Only in recent years has attention been given to all these facts (as far as I know Lesniewski was the first to become fully aware of them).

As soon as we give full recognition to the above circumstances, and carefully avoid the errors so far committed the task of laying the foundations of a scientific semantics, i.e, ~f characterizing precisely the semantieal concepts and of setting up a logically unobjectionable and materially adequate way of using these concepts, presents no further insuperable difficulties. Naturally, in doing this we must proceed cautiously, making full use of the apparatus which modern logic provides and carefully attending to the requirements of present-day methodology.

In the solution of this problem we cart distinguish several steps. We must begin with the description of tho language whose semantics we wish to construct. In particular wo must enumerate the primitive terms of the language and give the rules of definition by which new terms distinct from the primiti:e _one~ can be introduced into the language. Next we must distlnguish those expressions of the language which are called sentences, separate the axioms from the totality of sentences: and finally formulate the rules of inference by means of which theorems can be derived from these axioms. The

H)lENT OF SCIENTIFIC SEMANTICS 403

XV ESTABLIS - n1 'fit is purely

. exact and clear 0 Y I

description of a language IS l' ·t only those concepts

. t y if we emp oy In I d

structural, that IS 0 sa , ment of the signs an

h fi and arrange

which relate to t e orm Not every language can

compound expressions of the langua;~, manner, The languages be described in this purely structu. are called formalized for which such a description can be gtf venactitude of all further

, th degree 0 ex , .

languages. Now, smoo e . II on the clarity and preCI810n

investigations depends essentIa y t;"s Of formalized languages

. 't' only the seman "" ~ J'

of this descriptlOn, i is thros .

which can be constructed by exact ~e fthe language on the baSIS

Th t step is the constructIOn 0 . to be developed

e nex h . n language 15

of which the semantics of t e ~ve we shall call the metalanguagei

d zhi h for the sake of brevlty, t' 15' the problem 0

an " IC , . . his construc lOU u1

The most important point In t'th a sufficiently rich vocab , ~~~ eq nipping the metalanguage WI, escribed by the partIe ts But the solution of this problem IS PIr r:aet semantica.l conce~ )

ical ncepts. n 1· , f aifall'S

nature of the semantIc co ob)'eets (and staWS a f the

, 1 t' ons between ssions 0

express eeTtam re a I'd and expre , h

I ge diseUSSC nts whic

referred to in the anguli. 't Hence the staWme t

language referring to these obJet~ s. of semantical concepts m(t~:S

t' 1 proper ies t: rred to .

establish the essen ia , f the objects rete hich

, h d 'gnatIOn 0 th wrIllS W

contain both t e eSI 'tself) and e The

, f the language I ' h language,

the expressIOns 0 I d cription of t e Jwlogy of

are used in the structura es, f the so-called morp , ns of

th domaIll 0 I pres8lO

latter terms belong to e. ti ons of individua es f tructural

d the deslgna 1 ions 0 s

language an are rties of express 'aln'''J'11age

f t tural prope The JIlet ~-

the language. 0 s ruc , and sO on, , ust thus

'preRslons. fgatlOns m

relations between ex" nticalinves 1 h rimna1

h b sis for sema , ns of teO b-

which hi to form tea " the expresSlO f 1 ~mlage,

k' d of expresSIOn, h logy 0 3"'f)-

contain both ill s , f the morp 0 I nguage,

h pressIODS 0 ryother a

language, and t e ex language, like eve 'cal expres-

In addition to these, the meta 11 r stock of purely lag! age which

must contain a l~rger or :~ whether a met~ mentioned sions, The questIOn now ith expressions of the ~ DB ,Ve shall

is equipped exclusively wi tical investigatIO '

b ' for seman forms a sufficient aslS

return to this question later,

ESTABLISHMENT OF SCIENTIFIC SEMANTICS

xv

XV ESTABLISHMENT OF SCIENTIFIC SEMANTICS "'05

I ua e being investigated. We

described expressions of the ang. g d finition of) the se-

d way of using (or a. e

then agree to regar at' lly adequa.te if it enables

t . stion as ma ena .

mantical concep ill que 11 the partial definitions Just

us to prove in the metalanguag~ a 'here such a partial

t' d By way of illustratlOn we give

men ioneu. isf ti n:

d finiti of the concept of satis ac 10 .

e Ion . 'X and Yare

'if: the sentential functwn

John and Peter sat~s, y h nd Peter are brothers.

brothers' if and only if Jon a d . bed

t . ctly speaking, the eson

It should also be noted that, a ~ 1 dequaey of the usage of

din the matena a I e

conventions (regar g ted' the metameta anguag

t) e formula ill

semantical concep s ar . If

and not in the metalanguage Itse'b of a preparatory and

far have een

All the problems so linun' aries can we ap-

1 ft r these pre

auxiliary nature. On y a e his is the establishment of .a ma-

proach our chief problem ". T th semantical concepts m the teria11y correct way of us~~ tw: procedures come into conmetalanguage. At this pam ti al concepts (or at least s~m.e sideration. In the first th~ OO:::e I:etalanguage as ne~ pn~lof them) are introduced rot . perties are established Y

d their basIC pro d d all state-

tive concepts an 'oms are inclu e

. A ong these an f th concepts

means of axioms- m 'all T adequate use 0 e c nt

ments which secure the maten ) tics becomes an independ~ t

in question. In this way ::~:: morphology of l~ngua~~d o:t

deductive theory based up asy and simple, IS wor .

when this method, which seems ent themselves. The se. tttnghUIP

bi tions prese t of the woe

in detail various 0 lec·. r the developmen .

uffiClent lor F rtaro reasons,

of an axiom system s id ble difficulties. or ce . always

of semantics offers conslin~r: here, the choice of ~=~tial faowhich we shall ~ot gOI character, depending on ~e) Yarious has a rather aCCIdenta t 1 state of our knowl . . prove to

the ac ua . his conneXlon

tors (such as e.g. uld like to use III t . hether the

. hi h we sho . snsee w

criteria w c r the questIOn . The problem

he inapplicable. Mor:~es;mantics is conslBten~. matic method axiomatically construcf e whenever the axro .... we see

. . es 0 coors , 'a1' portance,-..

of consIstency ansee. . a SpeCl un

. lied but here it acquil'eB

IS app ,

404

Our next task is to determine the conditions under which we should be inclined to regard a way of using semantical concepts as materially adequate and in accordance with ordinary usage. We shall explain this more exactly by reference to the concept of truth. 'Ve regard the truth of a sentence as its 'correspondence with reality'. This rather vague phrase, which can certainly lead to various misunderstandings and has often done so in the past, is interpreted as follows. ,"Ve shall regard as valid all such statements as:

the sentence 'it is Bna-wing' is true if and only if it is Bna-wing; the sentence 'the world war will begin in the year 1963' is true if and only if the world war will begin in 1963.

Quite generally we shall accept as valid every sentence of the form

the sentence x is true if and only if p

where 'p' is to be replaced by any sentence of the language under investigation and 'x' by any individual name of that sentence provided this name occurs in the metalanguage. (In colloquial language such names are usually formed by means of quotation marks.) Statements of this form can be regarded as partial definitions of the concept of truth. They explain in a precise way, and in conformity with common usage, the sense of all special expressions of the type: the sentence x is true. Now, if we succeed in introducing the term 'true' into thc metalanguage in such a way that every statement of the form discussed can be proved on the basis of the axioms and rules of inference of the metalanguage, then we ahall say that the way of using tho concept of truth which hall thus been ostnbliahod iR matcri(llly adequate. In particular, if we succeed in introducing such a concept of truth by means of a definition, then we shall also say that the corresponding definition is materially adequate. We can apply an analogous method to any other semantical concepts as well. For each of these concepts we formulate a system of statements, which are expressed in the form of equivalences and have the character of partial definitions; as regards their contents, these statements determine the sense of the concept concerned with respect to all concrete, structurally

406 ESTABLISHMENT OF SCIENTIFIC SEMANTICS XV from the sad experiences we have had with the semantical concepts. in colloquial language. Apart from the problem of consistency, a method of constructing a theory does not seem to be very natural from the psychological point of view if in this method the role of primitive conoepte-vthus of concepts whose meaning should appear evident--is played by concepts which have led to various misunderstandings in the past. Finally, should this method prove to be the only possible one and not be regarded as merely a transitory stage, it would arouse certain doubts from a general philosophical point of view. It seems to me that it would then be difficult to bring this method into harmony with the postulates of the unity of science and of physicalism (since the concepts of semantics would be neither logical nor physical concepts).

In the second procedure, which has none of the above disadvantages, the semantical concepts are defined in terms of the usual concepts of the metalanguage and are thus reduced to purely logical concepts, the concepts of the language being investigated and the specific concepts of the morphology of language. In this way semantics becomes a part of the morphology of language if the latter is understood in a sufficiently wide sense. The question arises whether this method is applicable at all. It seems to me that this problem can now be regarded as definitely solved. It proves to be closely connected with the theory of logical types. The chief result relevant to this question can be formulated as follows:

It is possible to construct in the metalanguage mrthorlologically correct and materially adequate definitions of the scmantical concepte if and only if the metalanguage is equiJljJftl with variables of higher logical type than aU the variables of the language which is the subject of int'cstigation.

. It would be impossible to establish here, even in general out~e, the thesis just formulated. It can only be pointed out that It has been found useful, in defining the semantioal concepts, to deal first with the concept of satisfaction· both because a definition of this concept presents relatively fe.:v difficulties, and

XV ESTABLISHMENT OF SCIENTIFIC SEMANTICS 407 because the remaining semantical concepts are easily reducible to it It is also to be noted that from the definitions oflse~an-

. , rt t theorems of a genera na ure

tical concepts various Impo an be d . d For example

concerning the concepts defined can enveu. tho laws of

fi iti f truth we can provo

on the basis of the de ill Ion 0 • ddle (in their motalogica.l

contradiction and of excluded nn

formulation). .' em antics on a scientific

With this the problem of establishing S d finitely say

I I d Only the future can e

basis is complete y so ve: . .s field will prove to be fruit-

whether further investigatIOns ill thl. and what place d th special SClences,

ful for philosophy an . e. h tot lity of knowledge, But

semantics will win for Itself III tea hed J·ustoy a certain

h th nlts hitherto reao . 1

it does seem t at e res f t that it has been posslb e

optimism in this respect. The very a: I t for formalized Ianto define the semantical concepts, a eas seems to be not

t d adequate manner

guages in a eorrec an hi! ophical standpoint.

'. . rt nee from the p os f

entirely WIthout nnpo a" f truth for example, has 0 ten

The problem of the definitIOn 0 d' ntal problems of the

of the fun arne .

been emphasized as one 'licatioIlll of semantiCS,

I dg Also certaIll a pp , f th

theory of know e he. th ry of truth, in the doIll&Ill 0 ti e

and especially of t e eo . metamathema ICS,

d 1." SCIences, or h

methodology of the de ue l~e Using the definition of trut

seem to me to deserve attentIOn. h roof of coIlllistency for

.. to any out t e p te ees

we are in a positdon c (terially) true sen n

deductive theories in whiC~ on; be~:ne under the conditio~ are (formally) provable; this c~ th proof of consistency IS

. whiCh e h II

that tho metalanguage l~ . bles of higher type t an ~

carried out is equipped Wlt.h v:nasentences of the theory disthe variables which occur III t e has no great cognitive value cussed. Admittedly such a proof misseS than the assump-

. it rests on logically stronger pre. ed Nevertheless,

smce I • tency IS prav . h t it

tions of the theory whose conslS. tereat for the reason ~ a If

the result seems to be of Borne III from the investigatIOns 0 cannot be improved. For it folloWBcannot be carried out if;h,e Gadel that the proof of consist_-enbcl~ of higher type. The. de ""

tains no vana hi h iR bkew)i!C

metalanguage con equence w 1C

tion of truth has yet another cons

ESTABLISHMENT OF SCIENTIFIC SEMANTICS

xv

408 connected with the investigations of Gadel. As is well known, Gadel has developed a method which makes it possible, in every theory which includes the arithmetic of natural numbers as a part, to construct sentences which can be neither proved nor disproved in this theory. But he has also pointed out that the undecidable sentences constructed by this method become decidable if the theory under investigation is enriched by the addition of variables of higher type. The proof that the sentences involved actually in this way become decidable again rests on the definition of truth. Similarly-as I have shown by means of the methods used in developing semantics-for any given deductive theory it is possible to indicate concepts which cannot be defined in this theory, although in their content they belong to the theory, and become definable in it if the theory is enriched by the introduction of higher types. Summarizing, we can say that the establishment of scientific semantics, and in particular the definition of truth, enables us to match the negative results in the field of metamathematics with corresponding positive ones, and in that way to fill to some extent the gaps which have been revealed in the deductive method and in the very structure of deductive scienee.!

1 More detailed information about many of the problems discussed in this article can be found in VIII. Attention should also be called to my later paper, Tarski, A. (82). While the first part of that paper is close in its content to the present article, the second part contains polemical romarks regarding various objections which have been raised against my investigations in the field of semantics. It also includes some obeervuticnn about the applicability of semantics to empirical sciences and thoir methodology,

XVI

ON THE CONCEPT OF LOGICAL CONSEQUENCEt

. one of those whose intro-

1 . l onsequence IS t

THE concept of ogtca c . t formal investigation was no ~

duction into the field of stnc h rt of this or that investlmatter of arbitrary decision on t; :: were made to adhere to gator; in defining this concept, e 0 of everyday life. But these the common usage of the lang~ahgeth difficulties which usually

nfr ted Wit e 1 it

efforts have been co on With respect to the can Y

1 in such cases. . . no way

present themse ves t of consequence 18 ill

of its content the common concep day language. Its extension superior to other concepts of.every ge fluctuates. Any attempt is not sharply bounded and Its ~~~e vague, sometimes contr~to bring into harmony all POSSI ected with the use of thIS

. hich are conn cile our-

dictory tendenCies w f il e We must recon

, . in! doomed to a ur . . definition of

concept, IS certa Y h f t that every preCise I

selves from the start to t e. ac features to a greater or e88 this concept will show arbitrary

d th t they had

degree. 10 'cians believe a ts

Even until recently many} rgtelY meagre stock of concep ~

d b means of a re a 1V f the common concep

succeeded, Y tl the content 0 hi h coin-

in grasping almost exac ~ d fining a new concept w C il

ather ill e belief could cas Y

of consequence, or r mon one. Such a of de-

cided in extent with the chi~n:ements of the methodOlOt~allogiC

. d t the new ac e f ma.thema 1C

arise aIDI s t the progress 0 ad how to

ductive science. Thanks 0 urse of recent dec es, li.zed

we have learnt, during ~h~ :es in the shape of)mfOr: the

h tical disClP . well 0 ,

present mat ema h theories, as 18

deductive theories. In t ese .. ..v of an addreSS gITh'ven ~:~I:

.• a~J . 1935. e

NOTE. ThUl UI. hy in pans. 'kania jogiez,

t BIBLIOGR.U'WCAL . tjfic PhilO8OP • 0 jciu wyru

International Congr.ess .of :c~~ under the title 5~S. and then in ~e~

din pnnt m 0 39 (1936). pp. • AcW du JfJ''lI'~

first ~~pe;e g/.qd Filozojiczny, ,:ol.d logischen Folgeru~gth 8ciantiliquetl at

nego m r::e .mer den Begrlff ~r vol. 7 (Actual

under the title . hie ScientijiqUt.

Internatiorwl de Phtlosop . 1936 pp. I-II.

IndustrielJe8, vol. 394), pan8, •

410 ON THE CONCEPT OF LOGICAL CONSEQUENCE XVI proof of every theorem reduces to single or repeated application of some simple rules of inference-such as the rules of substitution and detachment. These rules tell us what transformations of a purely structural kind (i.e. transformations in which only the external structure of sentences is involved) are to be performed upon the axioms or theorems already proved in the theory, in order that the sentences obtained as a result of such transformations may themselves be regarded as proved. Logicians thought that these few rules of inference exhausted the content of the concept of consequence. 'Vhenever a sentence follows from others, it can be obtained from them-so it was thought-in more or less complicated ways by means of the transformations prescribed by the rules. In order to defend this view against sceptics who doubted whether the concept of eonsequence when formalized in this way really coincided in extent with the common one, the logicians were able to bring forward a weighty argument: the fact that they had actually succeeded in reproducing in the shape of formalized proofs all the exact reasonings which had ever been carried out in mathematics.

. N~vertheless we know today that the scepticism was' quite justified and that the view sketched above cannot be maintained. Some years ago I gave a quite elementary example of a theory which shows the following peculiarity: among its theorems there occur such sentences as:

Ao· 0 pO,98e88es the given property P, AI' I poseesses the given property P,

and, in general, all particular sentencee of the form An· n pOS8e.~Be.9 the given pro'perty P,

where In' t d f

s an s or any symbol which denotes 1\ natural number

in ~ given (e.g. decimal) number system. On the other hand the universal sentence:

A. Every natural number possesses the given property P, cannot be proved on the basis of the theory in question by means of the normal rules of Inference.! This fact seems to me to speak

th I ~or a ,detailed description of a theory with this peculiarity see IX; for

e diseussion of the closely related rule of infinite induction see VIII. pp. 258 ff.

XVI ON THE CONCEPT OF LOGICAL CONSEQUENCE 411 for itself. It shows that the formalized conc.e~t of consequence, as it is generally used by mathematicallo~Clan.s •. b~ n~t =~:

incid ith the common concept. Yet mtmtlve y

comer es WI 1 . th usual sense

certain that the universal sentence A fol ows m e Pro-

from the totality of particular sentences An. A1'A' ~~~~'~ISO be vided all these sentences are true, the sentence

true. . . f h kind iust described it has

In connexion with SItuatIOns 0 tel J of inference which ibl t f. rmulate new ru es

proved to be POSSl e 0 0 . their logical structure, are

do not differ from the old ones 1m ell ad from true sentences . f llibl . e a ways e

intuitively equally mae, 1.. red ed to the old rules. An

to true sentences, but cannot be educ 1 f infinite induction

rul . the so call ru e 0

example of such a e IS - A be regarded as proved

according to which the sentence ca~ have been proved provided all the sentences Ao' ~l""used' "':" the same sense as

bol 'A' 'A' etc. bemg m .,' t

(the sym sO' 1 " t f lts infirutIstIc na nre,

thi Ie on acCOun 0 1 I

previously). But IS ~' from the old rules. It can on y

is in essential respects different h if we have first suebe applied in the construction of a t eoter:ces of this theory-a.

. infinitely many sen hi

ceeded in provmg liz d in practice. But t IS

state of affairs which is never rea e fa. cert,ain modifica-

defect can easily be overcome by means °onsider the sentence

this purpose we c

tion of the new rule. For A A "., A", ... are

II the sentences o· It used

B which asserts that a f inference hitherto

provable on the basis of the rules 0 ed) We then set up

all been prov .

(not that they have actu y ce B is proved, then the corre-

the following rule: if the senten ted as proved. But here spending sentence A can be accep tenee B is not at all a it might still be objected that thetrusec~ion but belongs ~_~e)

h nder cons ' discU1!l!"'L

sentence of the t eory u th ry of the theory .

so-called meta theory (i.e- the act::al application of the rulet: and that in consequence apr. ti from the theory to .

. uire a transl IOn hall restnct

question wIll always reg . d this objection we s

metatheory- In order to avOI

412

ON THE CONCEPT OF LOGICAL CONSEQUENCE XVI

XVI ON THE CONCEPT OF LOGICAL CONSEQUENCE

413

consideration to those deductive theories in which the arithmetic of natural numbers can be developed, and observe that in every such theory all the concepts and sentences of the corresponding metatheory can be interpreted (since a one-one correspondence can be established between expressions of a language and natural rrumbersj.! 'Ve can replace in the rule discussed the sentence B by the sentence B', which is the arithmetical interpretation of B. In this way we reach a rule which does not deviate essentially from the rules of inference, either in the conditions of its applicability or in the nature of the concepts involved in its formulation or, finally, in its intuitive infallibility (although it is considerably more complicated).

Now it is possible to state other rules of like nature, and even as many of them as we please. Actually it suffices in fact to notice that the rule last formulated is essentially dependent upon the extension of the concept 'sentence provable on the basis of the rules hitherto used'. But when we adopt this rule we thereby widen the extension of this concept. Then, for the widened extension we can set up a new, analogous rule, and so on ad infinitum. It would be interesting to investigate whether there are any objective reasons for assigning a special position to the rules ordinarily used.

The conjecture now suggests itself that we can finally succeed in grasping the full intuitive content of the concept of consequence by the method sketched above, i.e. by supplementing the rules of inference used in the construction of deductive theories. By making use of the results of K. Gi_ido12 we can show that this conjecture is untenable. In eV(lry deductive theory (apart from certain theories of a particularly dNllontary ~ature), however much we supplement the ordinary rules of inference by new purely structural rules, it is possible to construct sentences which follow, in the usual sense, from the theorems of this theory, but which nevertheless cannot be proved in this theory on the basis of the accepted rules of

't ~~r the concept of metatheory and the problem of the interpretation of a ~4; ~. eory in the corresponding theory see article VIII, pp. 167 ii., 184, and

• Cf. Godel, K. (22), especially pp. 190 f.

inference.' In order to obtain the proper concept of consequence~ which is close in essentials to the common conce~t, we mus

h ds d Ply quite different conresort to quite different met 0 an ap

. inc it It' perhaps not superfluous

ceptual apparatus III defimng I . IS.. he

to point out in advance that-in comparIson WIth the hnOW-t~ n 1 nly used by mat emil. ic ...

old concept of consequence as commo . ill

it . ortance This concept W logicians in no way loses I.S. Imp. 'fica~ce for the practical

Probably always have a deCISive Signt. -hi h

. h' 8 an mstrument \l JC

construction of deductive t oones, a t f these

. rti cular sen ences 0

allows us to prove or disprove ~a I id t'ODS of a general

h that ill oonsi era J

theories. It seems, owever, f quence must be

theoretical nature the proper concept 0 conse

placed in the foreground.2 • definition of the

The first attempt to formulate a precI8eC 3 But this

was that of R. arnap.

Proper concept of consequence h of application of the

. 1 bi tions t e range .

1 In order to anticipate p08Slb e 0 l~ d re exactly and the logical nature

result just formulated should be determme ~o ly , in particular it should be of the rules of inference exhibited moretc e~urai character of these rules.

. h t' eant by the s rue . I I' pointed out

exactly explamed w a IS III ts in question IS c ear Y •

• An 0 position between the two conc.ep t lIBt to my present standpo",t. . rt.i 1 PIX 293 If. Nevertheless, In con r . manner about the poss" 10 a IC e , pp. . a decidedly negative t of con"'" I have there expressed myself 10 1 definition for the proper concep h I was

bility of setting up an exact forma. ._. d by the fact that, W en .

. . h t time IS exp ... me f onstructlOn

quence, My positron at .t a I . hed to avoid any means 0 c. I fonus; writing the article mentIOned, WlS. I t,,.,..,, in any of Its cIasslca f

h th Y of loglca s r -- r concept 0

which wenl beyond t e ear. . possible to define the prope. ible in the

but it can be shown that it IS ~ exclusively the means adnllssl'de~tions

t I hilst USIng li . t our CODS' • ~

consequence adequa e Y w I e should thus rni baracter (to

classical theory of types; un e;s ':Iementary and fragroe~ta0~ especiallv solely t-o formalized languages 0 an of finite order; cf, artie e (IO) the te~ be exact, to the so·called langua~e: resting book, Camap. R. ~n.sequence

pp "68 ff) In his extremely m e lied to the old concept of '''"tiD-

". ~ ':. . ab'lily is app I . . in order to "'"

(logIcal) denvatwt~ or derw I t' n of deductive theories, . 'The opposition

I\I'! commonly llSO<1 in the const-ruc I:te BS the proper conceP\ di 'erse derived guish it from tho concept of CO,,~eq~ed by Csmap to the m~s 194'1f). he also botweon tho two concepts is ex en p'.' cf pp. 88 ff., an - Tv>r' c~ncept of

, d 'a·conce ... ,' f the pror-

concepts (' [.concepts . an rrectly-the importance 0 ral theoretical disCUS'

emphasizes-to my mmd CO ts derived from it, for gene

consequence and the concep islly P: 181.

sions (cf. e.g. p. 128). 88 f and carnap, R. (Hi :quence "'bieh

• Cf. Camap, R. (10), pp. . ., t another definition 0 c "'his definition

kg there rs ye bAI'SCter. ~

In the first of these w?r e of an ele[Dcntaryc ages of a more ~om'

is adapted to a formalized lan~tag.~~ot be applied to langu t of 1(llI:i,.111 con ... ··

. . h because I e ... ~· fi the conCf'p· h

18 not consIdered ere t pta to de ne . k of what "

C mapat em > hi the frame"or

plicated logical structure. a but also WIt m . • I fI Jl('\<' I>

. allanguages, bo t thl~ on 1" , •

quenee not only for ~pec~ hall have more to I!6Y a u

calls' general ttyntax . '" e a

4.l( ON THE CONCEPT OF LOGICAL CONSEQUENCE XVI attempt is connected rather closely with the particular properties of the formalized language which was chosen as the subject of investigation. The definition proposed by Carnap can be formulated as follows:

The sentence X follows logically from the sentences of the class K if and only if the class consisting of all the sentences of K and of the negation of X is contradicWry.

The decisive element of the above definition obviously is the concept 'contradictory'. Carnap's definition of this concept is too complicated and special to be reproduced here without long and troublesome explanations.!

I should like to sketch here a general method which, it seems to me, enables us to construct an adequate definition of the concept of consequence for a comprehensive class of formalized languages. I emphasize, however, that the proposed treatment of the concept of consequence makes no very high claim to complete originality. The ideas involved in this treatment will certainly seem to be something well known, or even something of his own, to many a logician who has given close attention to the concept of consequence and has tried to characterize it more precisely. Nevertheless it seems to me that only the methods which have been developed in recent years for the establishment of scientific semantics, and the concepts defined with their aid, allow us to present these ideas in an exact form."

Certain considerations of an intuitive nature win form our starting-point. Consider any class K of Rcntene{'H and a sontence X which follows from the sentenoes of this cIItHH. From an intuitive standpoint it can never happen tlmt 1I01.h t.1J(1 class K consists only of true sentences anti tho sentence X is falso. Moreover, since we are concerned here with tho concept of logical, i.e, formal, consequence, and thus with a relation which is to be uniquely determined by the form of the sentences between which it holds, this relation cannot be influenced in any way by empirical knowledge, and in particular by knowledge of the

1 See footnote 3 on p, 413.

t • The met?ods ~d concepts of semantics and especially the concepts of ruth and satlsfactlOn are discusaed in detail in article VIII; see also article XV.

XVI ON THE CONCEPT OF LOGICAL CONSEQUENCE 416

X or the sentences of the class objects to which the sentence . t be affected by re-

K refer. The consequence relatiOn ~anno f, rred to in these placing the designations of the objects re e bi ts The two

. ti of any other 0 Jec .

sentences by the designs IOns. t be very charac-

. t' dicated which seem 0

circumstances lUS Ill, t of consequence, may

teristic and essential for the proper concep t

d i th rollowing statemen : be jointly expresse III e 1.1

K nil in the sentence X, the (F) If, in the sentences of the c.lassz a t nts-are replaced by

constants-apart from purely logica cans a here replaced by like any other constants (like signs being everyw thus obtained fr()ffl

h Iss 0' sentences

signs), and if we denote t e c a .~ rom X by 'X", then the

K by 'K", and the sentence ~ta~n:l~ that all sentences of the sentence x' must be true provided

class K' are true. . . . dental

. . the discussion certalO mel

[For the sake of simplifymg h d in what folloWS. . d d both ere an d complications are disregar e. ' the theory of logical types, ~n

They are connected partly WIth. defined signs whiCh

. f liroinatIIIg any I .

partly with the neceSSIty 0 e med i.e. of rep acing

- th sentences eonee ,

may possibly occur III e ..

them by primitive signs.] bt . ed a necessary condItIOn

In the statement (F) we have 0 am ce of the class K. The

X t be a consequen 1 fficient.

for the sentence 0 hi ndition is a so su

. hether t 8 co f -e the

question now arises w d i the aflirnla 1\ ,

t be aDSwere m h neept

If this question were 0 d finition of t e co

. adequate e nly diffi-

problem of formulatmg an a: ....... atively. The 0 .

ld be solved auu ill , hi h occurs III

of consequence woll 'true w c

ted with the term d adequately

culty would be connec be exactly an

F) B t this term can

tho condition ( . U d

defined in semantics.1 .' at 80 favourable. It ~y, ~

Unfortunately the situatiOn IS n how this by COllSldenDg

.' t difficult to S X does not

it does, happen-It IS no that the sentence Iaas K

special formalized languages- the sentences of the c .

~ II in the ordinary sense from. ed This condition may 1D

o ow (F) is satlSfi . h' h we are

although the condition the language with w IC

fact be satisfied only because 41f.

1 See footDote 2 00 P:

1 These last remarks constitute a criticism of some earlier attempts to define the concept of formal consequence. They concern, in particular. Carnap's definitions of logical consequence and a series of derivative concept:a (~.con~equences and Lvconcepta, cf. Carnap, R. (10), pp. 137 If.). These definl' trons, III so far as they are set up on the basis of 'general syntax'. seem to me to be materially inadequate, just because the defined concepts depend essentHl;ly: III their extension, on the richness of the language investigated.

Kee footnote 2 on p. 414.

LOGICAL CO~SEQUENCE 417 XVI ON THE CONCEPT OF

. L b corresponding variables,

occur in the sente~ces belong:t to ilk/variables, and unlike by like constants being replace. y I L' of sentential functions.

thi we obtam a c ass ti 1

unlike. In sway . hich satisfies every senten la

An arbitrary sequence of objects w II d a model or realization oj

1 L' '11 be ca e f

function of the c ass , WI. his sense one usually speaks 0

the class L of sentences (in Just t t' theory) If in parti·

. t m of a deduc ive ., I

models of an aXIOm sys e . I ntence X we shall e. FlO

L . ts of a sing ese'

eular, the class eonsis odel of the sentence X.

refer to a model of the class L as a m d fine the concept of logical In terms of these concepts we can e

consequence as follows: "the class

. all from the sentences OJ

The sentence X follows logic Y ,_. K is also a model oj the

, model oj tke cuws

K if and only if every

sentence X. t t ds the content

one who unders an . n

It seems to me that every ad 't that it agrees quite ~e

fi ·t· n must Jlll it anOUS

of the above de m 10 till clearer from I s v

This becoIlles s h basis of

with common usage. . be proved, on t e

I rticular It can te ces must

consequences, n pa 'uence of true sen n

this definition; that every cODSeq snce relation which holds be true and also that the conseq~ independent of the seIlBe betwee~ given sentences is comp:e: occur in these sentence~ of the extra-logical constan~ ; t:e condition (F) formulate In brief it can be shown t a X is to follow from the ~e~-

, if the sentence l.:~ ondition 18 111

above is necessary th other hand, two c here

f hIs K On e f oIlBequence

tences 0 t e c as '. the concept 0 c . ken) is

general not sufficient, Sl~~ the standpoint we have tae being

defined (in agreement WI. oepte of the languag

h . chness m con

independent of t e ri ed definition

investigatcd. . It to reconcile the proPOs f ntenceB

Finally, it is not difficu tr'1'AO to call a class 0 sa

For we can a6·- ~ .

with that of Carnap. . B Scholz in his

. bad appeared in pnn~ert:Betrs.cbtung' •

. . at of thIS paper Eine JBhrhun 9-4"" (see III

t After the ~~lgw chaftslehre Bolzan08• ries vol. 6, pp. 39 :;een this article 'Die "tsse;. 'chen Schule, new s: ~aching analogy be about .. Abhandlungen der f'1£.'I 8 58) pointed out a sr- ted by B. Balzano

particular p. 472, footnote d the one sugges

definition of consequence an .

hundred years earlier.

416 ON THE CONCEPT OF LOGICAL CONSEQUENCE XVI dealing does not possess a sufficient stock of extra-logical constants. The condition (F) could be regarded as sufficient for the sentence X to follow from the class K only if the designations of all possible objects occurred in the language in question. This assumption, however, is fictitious and can never be realized,l We must therefore look for some means of expressing the intentions of the condition (F) which will be completely independent of that fictitious assumption.

Such a means is provided by semantics. Among the fundamental concepts of semantics we have the concept of the satisfaction of a sentential function by single objects or by a sequence of objects. It would be superfluous to give here a precise explanation of the content of this concept. The intuitive meaning of such phrases as: John and Peter satisfy the condition 'X and Yare brothers', or the triple of numbers 2, 3, and 5 satisfies the equation 'x+y = z', can give rise to no doubts, The concept of satisfaction-like other semantical concepts-must always be relativized to some particular language. The details of its precise definition depend on the structure of this language . Nevertheless, a general method can be developed which enables us to construct such definitions for a comprehensive class of formalized languages. Unfortunately, for technical reasons, it would be impossible to sketch this method here even in its general outlines.s

One of the concepts which can be defined in terms of the concept of satisfaction is tho concept of model. Let lIB assume that in the language we are conaidering oortain varinblos correspond to every extra-logicul constant, and in such 11 wny that every sentence becomes 11 sententiul function if tho COIlHt!1nts in it arc replaced by the corresponding variables. Let L be auy class of sentences. 'Ve replace all extra-logical constants which

418 ON THE CONCEPT OF LOGICAL CONSEQUENCE XVI contradictory if it possesses no model. Analogously, a class of sentences can be called analytical if every sequence of objects is a model of it. Both of these concepts can be related not only to classes of sentences but also to single sentences. Let us assume further that, in the language with which we are dealing, for every sentence X there exists a negation of this sentence, i.e. a sentence Y which has as a model those and only those sequences of objects which are not models of the sentence X (this assumption is rather essential for Carnap's construction). On the basis of all these conventions and assumptions it is easy to prove the equivalence of the two definitions. We can also show -just as does Carnap--that those and only those sentences are analytical which follow from every class of sentences (in particular from the empty class), and those and only those are contradictory from which every sentence follows.!

I am not at all of the opinion that in the result of the above discussion the problem of a materially adequate definition of the concept of consequence has been completely solved. On the contrary, I still see several open questions, only one of whichperhaps the most important--I shall point out here.

Underlying our whole construction is the division of all terms o~ ~~e la?guage discussed into logical and extra-logical. This division ~s certainly not quite arbitrary. If, for example, we ~ere to mclude among the extra-logical signs the implication SIgn, or the universal quantifier, then our definition of the concept of consequence would lead to results which obviously

contradict ordinar 0 I . .

Y usage. n t lO other hand, no objective

grounds are known to me which permit 1If1 to draw a sharp

I Cf. Carnap It (10) l3li II . •

(11) p. 18') Ti' • pp. . :. ""p'''ll"lIy 'I It". li:!.7 nru] I\:!.H; ('urnap. H.

d fi' iti -f' L!I. 10 and 11. IncHiontally I should Iiko to remark that tho

e m ion 0 tho concept of h

limits of s' consequence oro proposed doe s not exeeod tho

mittedl trtax in Carnap's conception (cr. Carnal'. It. (10). pp. 6ff.). Adsynt :b te general concept of satisfaction (or of model] does not belong to

ax, u we use only a spec' I f hi . . f

sentential f ti . .m case 0 t IS concept-the satisfaction 0

case can b r ch ions w?ich contain no extra-logical constants, and this special

e c aracterlzed using nl al 1 . . . _.

concepts B thO Y gener ogieal and specific synt.actIc .....

this . e ween t e general concept of satisfaction and the special case of

concept used here app . tel

8emantic~' t f roxima y the same relation holds as between the

.... concep 0 true senten d th' ~, '-'

sentence. ee an e syntactical concept of an .... ytlC ....

XVI ON THE CONCEPT OF LOGICAL CONSEQUENCE .19 boundary between the two groups of terms. It seems to be possible to include among logical terms some which arc usually regarded by logicians as extra-logical without running into consequences which stand in sharp contrast to ordinary uaage. In the extreme case we could regard all terms of the language as logical. The concept of farmal consequence would then coincide with that of material consequence. The sentence X would in thiS case follow from the class K of sentences if either X were true or at least one sentence of the class K were false."

In order to see the importance of this problem for certain general philosophical views it suffices to note that the divisi?n of terms into logical and extra-logical also plays an e~entlal part in clarifying the concept 'analytical'. But according to many logicians this last concept is to be regarded as the exact formal correlate of the concept of tautology (i.e, of a statement

1 It will perhaps be instructive to juxtapose the three concepts: 'deri\'~-

, d 'mlltonal conaequence •

billty' (cf.~. 413. notehe2)'h'formall _;o=U:~:h::e given sentence X Iollows, for the apecial case w n t e c ass , 1': Let us denote consists of oni II finite number of sentences: Y" Y., ... , .. -

by the symbol ~ Z' the conditional sentence (the implication) whose anttlCt cdcthnt

Y Yo 1': and whose consequen l8 e

is the conjunction of the sentences ,. ...... " blisbed:

sentence X. The following equivalences can then be esta -

'vahle! the sentenCU of thoe clas8 K if and only

the sentence X i8. (log~y) den ~ derivable from the a.riomB of logic);

if the sentence Z t8 l<Jgically provable (u. of thoe class K if and only if

the sentence X folWws formally from the senlenU8

the sentence Z is analytical; . eu oj thoe class K if and only if

the sentence X follows matenally from the senten

the sentence Z is tnoe.

t arouse certain objections; ef,

OC the three equivalences only the firs can . n with these eqw'valence6

'all 346 In connexJO

article XII. pp. 342-64. especl y . 4 14 and 42.

cf. also Ajdukiewicz. K. (~). p. 19. a;:t~;!P~he several variants of the con-

In view of the analogy mdl~tod nts itself whether it would not be useful copt of consequence, the questIOn p~ ta a general concept of a relative to introduce. in addition to the Spool concep • ce with rup«l to a class L of character, and indeed the concept of cons~us no'"tion (limiting o~lves

. of the previo .,.,

sentences. If we make use ag~n. we can define this concept as follows:

to the case when the cless K IS finito).

" the class K with respect to the the sentence X follows from »: senknU8 0 Z belongs to the class L.

class L of sentences if and only if th.e aentenU .

. . . b'lit would coincide WIth consequence

On the basis of this defimtlOn. derIVe. I Ybl '-n~- formal consequen.-

all I . e.llyprova e senre ~.

with respect to the cl8BS o~ ogre t to tho claB8 of all analyticalllCntencell. and

would be consequences WIth ~poo t to the clua of all true ""nUln.,.,. .

material consequences those WIth respoo

420 ON THE CONCEPT OF LOGICAL CONSEQUEXCE XVI which 'says nothing about reality'), a concept which to me personally seems rather vague, but which has been of fundamental importance for the philosophical discussions of L. Wittgenstein and the whole Vienna Cirele.!

Further research will doubtless greatly clarify the problem which interests us. Perhaps it will be possible to find important objective arguments which will enable us to justify the traditional boundary between logical and extra-logical expressions. But I also consider it to be quite possible that investigations will bring no positive results in this direction, so that we shall be compelled to regard such concepts as 'logical consequence', 'analytical statement', 'and 'tautology' as relative concepts ~hich must, on each occasion, be related to a definite, although III greater or less degree arbitrary, division of terms into logical and extra-logical. The fiuctuation in the common usage of the concept of consequence would-in part at least-be quite naturally reflected in such a compulsory situation.

1 Cf. Wittgenstein, L. (91), Carnap, R. (10), pp. 37-40.

XVII SENTENTIAL CALCULUS AND TOPOLOGYt

IN this article I shall point out certain formal connexions between the sentential calculus and topology (as well as some other mathematical theories). I am concerned in the first place with a topological interpretation of two systems of the sentential calculus, namely the ordinary (two-valued) and the intuitionistie (Brouwer-Heyting) system. With every sentence m of the sentential calculus we correlate, in one-one fashion, a sentence m1 of topology in such a way that m is provable in the two-valued calculus if and only if m1 holds in every topological space. An analogous correlation is set up for the intuitionistic calculus. The present discussion seems to me to have a certain interest not only from the purely formal point of view; it also throws an interesting light on the content relations between the two systems of the sentential calculus and the intuitions underlying these systems.

In order to avoid possible misunderstandings I should like to emphasize that I have not attempted to adapt th~ m~t~o~s ~f reasoning used in this article to the requirements of mtu~tIOmshc logic." For valuable help in completing this work I am mdebted to Professor A. Mostowski.

I Most results of this article were obt.ained in the year 1935. The conne~jonf

." .. lid Boolean al<YAbra (or the theory 0

he tween the int.nifionistic ea cu us an .,-. I . 1931

. §) I· ed by me still earlier, name y 10 •

• Iuduet.ivu RVRtlllTIS BOO Ii was cnscovere bo k d

~nm" rntnll.;ks to ,.hiS effect can be found in article XII of the present 0" atedn

. I . did I become aequam

in Turski, A. (HO). Only .. Itor eomplet.ing t ~U, pa~ H 70 In spite of an

with I,h" work. then "owly pubhshed, of Stone, .. . ~ ). .

" I . there is certamly some connexion

nnl,i""ly ditT.'rt",t view of Brouwtlfllm ogre . "

h ks as can sasily be soon companng

between particular results of t e two wor '" h ti al content these

, • ~ 2" d Th 4 11 In their mat erna IC

Stone s Th. " p. -. an my .., . d t t all apply to the two

two theorems are c108el~ related. But ~hl~':c~;~ :he kernel oftbis paper,

works as wholes. In particular, ~h. 4.24, Ill, id t"

tends in quite a different direction from Stone s Consi era ions,

Thi rti le is the text of an address given by

t BIBLIOGRAl'HICAL NOTE. IS a ic ird Polish Mathematical Congress in

the author on 30 September 193~ to theTh~ e de mnthimatique. vol. 16 (1937). Warsaw (see Annales de lo Soc.ete polonahlS t'tl 'J)er Au""""'enkalkUI und die

I ) Th . I fi t eared under tel e ........

p. 92. e arttc e rs app ." I 31 (1938) pp. 103-34.

Topologie', in Fundamenta Mathematuxu, vor. ,

BIBLIOGRAPHY

d M MALONEY 'String Theory', J. Sym-

(14t) J CORCORAN W. FRANK, anu m- '

. . ' ) 625-37.

bolic Loaic, 39, ~197~, e La 10 ique (Paris, 1905).

(14a) L. COUTURAT, Lalgebre.dd undgwas sollen die Zahlm? flLh. od. (15) R. DEDEKIND, Was stn

(Braunschweig, 1923). C McKINRF:Y, 'AIJ.("hraH and t1l1'ir

(15a) A. H. DIAMOND and J. C. 'th Soc liii (1947), 1J5!HI2. u

b ' Bull Amer. ma.·, 3 d ed (llor UI,

subalge ras, .'. in die Mengenlehre, 1'. •

(16) A FRAENKEL, Etnlettung

. . . d'onscmhlllfl

1928). £ ill s et fonctions additives

(17) M FR11:CHET, 'Des am e. 1923) 328-65. .'

. b t its ' Fund. Math., IV ( .. ' d logische SchhoJ3fln ,

a s rai , , tersuchungen uber as

(18) G. GENTZEN, "?n (1934), 176-210 and 405-:31. d M Brouwer',

Math. Z., XXXIX oints de 180 loglque e . (1929)

(19) T GLIVENKO,' Sur quelqu~s P Bull Clas8e des Sciences, xv ,

. Academie Royale de Belgtque, . unk

1 . chen F .

183-8. . . der Axiome des oglS

G.. 'Die Vollstandlgkeit .. (1930), 349-60.

(20) K. ODEL, .. 'Mh Math. Phys., XXXVll ultate iiber Ent·

tionenkalkuls , . t mathematische Res ., A ,-~" der

'E" me a f iheit fWU·

(21) K. GODEL, ml~~eit und Widerspruc~s rei haftiiche K!asse,

scheidungsdefim h tisch.naturwt8SenBC J'

. W· Mat ema 4-15

Wiss. tn ten,. lsvii (1930), 21 .: or Principia

Akademisch~r Anzetge:l unentscheidbare ,Sa1;Z ~7Ifoth. l'hy8.,

(22) K. GODEL, 'trber form dter Systeme I , .

. und vel'Wan

mathemattca .' n AUil.<>aW·n·

... (1931) 173-98. . d . tuitionlstlSche '

XXXVlli , . ' retatlon eS ill

(23) K. GODEL, Erne In~koll., iv (1933), 39-40. den Parsc:ioxif'n

kalkiils', Erg. mat NELsoN, 'Bemerkungen ze~ der Fries'BCwm (24) K. GRELLING and dLBurali.Forti', Abhandlung

von Russell un .. 1908), 301-34. . i 1927).

Schule neue Folge, 11 ~lehre (Berlin und ~IP\!' demonstration " (25) F. HAUSDORFF, Menge h s sur 180 tbeone de K 33 (1930).

(26) J. HERBRAND, '.R~~;.rcT':rsovie, Classe I!~;g:chen ~Iatrizen',

Trav. Soc. S~t. 'fheorie der aUSSSg ,

(26(1) H. HERMES, ... Zt;~l) 414-18. __ b liebige Satzsysteme,

Moth, Z., h.~1 ( ~x'iomeIlsysteme fur c . ,

(27) P. Tl~~ltT7., 'Ubf;r 929) 457-514. . tuitionistischen Logik,

Math Ann., CI (I 'I Regeln der 1Il WisS PhY8.·rnaJh.

.. • • I)' 1 forma (,II . Akad der .,

(2H) A. J Iir.Y'l'lNO,. I( PreufJiscJ;en . ,

Sitz111t[Jsbenchte dc~ talen Zahlenlehre •

l(lassc (1930),42-06. dl gung der elemen

'Die Grun e ik' Hath.

(29) D. HILBERT, . , (1931), 485-94. der Mathemat .'

Math. Ann., CIV • chen Grundlagen - d.

(29a) D. HILBERT, '!?i~i~~, 151-65. (}rundzftge der theore11.8 ,en

Ann., lxXxVlll W. ACKERMANN', _ - Ikrlin.

(30) D. HILBERT and \lath<:mnt,k,d

L 'k (Berlin, 1928). GrundZagcnder. .

ogt d P BERNAYS, _ - X\II

(31) D. HILBERT an . . , J 8'1m"o/>r TN.pr.

1934). ., termS of logIC, . . d,"

(31a.) H H' 'On primitive . tulat'·~ for

. IZ, 1 nd"nt 1'''. . .,~" 3'~·1.

(1952), 156. 'Sets of inf "pc S v OllOI). - .. .-If''

, HuNTINGTON,' r 1/Uuh.· 0<;.. t".tra<:t Ir"',m

(32) E. V. . , Trans. Ame . I tN! for a. ' \lath .in'"

algebra of logic, 'A set of postu _a - of inc!U.·lOn , .

(33) E. V. HUNTINGTON, f the simple relalton

expressed in terms 0

lxxiii (1913), 522-59.

BIBLIOGRAPHY

(1) W. ACKERMANN, 'tiber die Erfiillbarkeit gewisser Ziihlaus· driicke', Math. Ann. c (1928), 638---49.

(2) K. AJDUKIEWIcz, Z metodologji nauk dedukcyjnych (From the

methodology of the deductive sciences) (Lw6w, 1921). .

(3) K. AJDUKIEWICZ, Gloume zasady metodologji nauk: i logikiformalneJ (Main principles of the methodology of the sciences and of formal logic) (Warszawa, 1928).

(4) K. AJDUKIEWICZ, Logiczne podstawy naucaania (The logic_a!

foundations of teaching), Encyclopedja Wychowania, ii (Warszawa, 1934), 1-75.

(5) P. BERNAYS, 'Axiomatische Untersuchung des Aussagenkalkiils der Principia mathematica', Math. Z., xxv (1926), 305-20.

(5a) P. BERNAYS and M. SCHONFINKEL, 'Zum Entscheidungsproblem der mathematischen Logik', Math. Ann., xcix (1928), 342- 72.

(6) B. A. BERNSTEIN, 'Whitehead and Russell's theory of deduction as a mathematical science', Bull. Amer. math. Soc., xsxvii (1931), 480-8.

(6a) E. W. BETH, 'On Padoa's method in the theory of definition', Inda-gationes Mathenuuicae, xv (1953), 330---9.

(7) G. BmKHoFF, 'On the combination of subalgebras', Proc- Cam·

bridge phil. Soc., xxix (1933), 441-6. .

(7a) G. BmKHoFF, Lattice Theory, Amer. Math. Soc. Colloquium Publications XXV (revised ed.). (New York, 1948).

(7t) G. BOOLE, The Mathematical Analysis of Logic (Cambridge & London 1847).

(7f) G. BOOLE, An Investigation of the Laws of Thought (London, 1854). (7b) S. N. BUIlRI8 and M. K. KWATINETZ, 'Some cardinality results for certain classes of mathematical structures'. Noticr« of the Amer. math. Soc. 19, (1972), A--46.

(7c) S. N. RURI!!" and M. K. KWATJNET1., 'A metamnthernetlcal OhRcrvation and its application to tho eardlnalit.v of eertnin claHl'l('" (If rnRt.h~matir.alstrllcturcs', NolirCR of the An"'I'.·nllllh. Soc. 10, (I972), A-2IMI.

(8) H. CAltNAP, Abrip der LogiHtik (Wiml, 11l21l).

(9) R. CARNAP, 'Bericht iib"r Ulll.for>uwhung"l1 zur' UllglllllOillOIi Axiomatik', Erkenntni._~, i (1930),303-7.

(10) R. CARNAP, Logische Syntax del' Sprache (Wion, 1934).

(lOa) R. CAllNAP, Logical Syntax of Language (London, 1937), translation of Carnap (10) by A. Smeaton, incorporating Carnap (lOb) and Carnap (11).

(lOb) R. CARNAP, 'Die Antinomien und die Unvollstiindigkeit der Mathematik', Mh. Math. PhY8., xli, (1934),263-84.

(11) R. CARNAP, 'Ein Gultigkeitskriterium fur die Satze derklassi· schen Mathematik', Mh. Math. Phys., xlii (1935), 163-90.

(1111) A. CIIU_RCH, 'Special cases of the decision problem', Revue Philo· sophique de Louuain, xlix (1951), 203-21.

(12) L. C~WISTEK, 'The theory of constructive types (Principles of ~?glC and mathematics)', Part I, Ann. Soc. polon .. Math., 11 (1924), 9-48.

(13) L. CHWIXTEK, 'Neue Grundlagen der Logik und Mathematik " Math. Z., xxx (1929). 704--24.

(14) L. CHWISTEK, 'Dber die Hypothesen dor Mengenlehre', lrlath.

Z., xxv (1926), 439-73.

BIBLIOGRAPHY

459

458

BIBLIOGRAPHY

. en Systems der Grundlagen

(47b) S. LE§NIEWSKI, '?r~mdziige e~~~:e:~ea Logica. i (1938), 61-

der Mathematik, § 12, 0

144. l" 10 ie (Berkeley, Calif., 1918).

(48) C.!. LEWIS, A survey of 8~0 ~h,g(On logical matrices) Trav. (48a) J. {.os, '0 matrycach logICZl_ly B Nr. 19 (1949).

Soc. Sci. Lett. W:~oclaw, .~e:~keiten im RelativkalkUi', Math.

(49) L. LOWE:NHEIM, 'Uber Moghc .

Ann. lxxvi (1915), 447-70 ". lk ~ ., (On the concept of mBglu,

, . . wie o"Cl

(50) J. f,UKASlEWICZ, '0 ~OJ'i:CIU xix (1916), 35 If. .

tude) Przeglqd F~lozojiczny, ~_. a' (Two.valued lOgIC),

, 'L ik dwuwarto""low

(50a) J !..UKASIEWICZ, ogi a ... 921) 189-205.

• Przeglqd Filazojiezny, XXUl (1. . ~matycznej (Elements of

(51) J E.,UKASIEWICZ, Elementy log~h .,

. mathematical logic) (1929).. t ebachlogiki IJl&tematyczn~ (51a) J.l..UXASlEWICZ, '0 znaczcnlU IdeO 7 mathematical logic), Nau (On the significance and nee 0

Poleko x (1929), 604-20. fil B.a' (Logistic and Philosophy),

, 'Logistyka a ozo 31

(52) J. LUXASlEWICZ, . (1936), U5-· .' cal.

Przeglqd Filozojiczny, xxxut . m of the imphcatlO~ A

'The shortest aXIO Awd Sedwn ,

(52a) J. LUKASIEWICZ, ., 'Proc. Roy. Irish .,

culus of prOpOsitIOns , heinlich-

Iii (1948), 25--33. . die Grundlagen der Wah:

(53) S. MAZUBK1EWI?Z'~~;at;. Phya., sl (1::\ 34!:;ts'inclosure

kcitsrechnung , . A T ARSKI, 'On clo e em 2 62

(53a) J. C. C. McKINSE~ anif~a~hematica, xlvii (1946)'t~~ -S~M8e, vii

algebra', Anna .... 0 definition of tru ,

K 'A new

(53b) J. C. C. Mc INSEY, d ihre An·

1948-9) 428-33. he Karper un I _.~

( , 'Ab "hlbare Boolesc ik' Fund. Mu.n.,

(53e) A MOSTOWSKI, za . Metamathemat ,

. uf di aUgememe

wendung a Ie t'cal Logic)

xxix (1937), 34-53"'ka matemo.lyczfID (Mathema I

(53d) A. MOSTOWSKI, Log~ 1948). . . in the axiomatic

(Warszawa.Wroclaw,. d' ative definitIOns d xxxviii

(53e) A. MOSTOWSKI, 'Some ImPhre ~~xvii (1950), 111-24, an

, Fund Mat ., _ ....

set·theory " . Qr11Iillized arithJf.,..1i:

(1951), 238. es undeciJoble tn j .

(53!) A MOSTOWSK1, Sentenc . , Math. Z., XSVl

. (Amstordam, ~952\Iilbertschen Beweistheone , ..

(54) J. v. NEUMANN, Zur rimitivepropoBltiODB

(1927),1-46. . nin thenumberoft~e~917)' 32--41.

(55) J. NICOD, 'A redu~l~e phiZ. ss«, ~~ (Paris, 1924).

of logic', Proc- a. dans le f7IOfIIle Be 'ble de postulatB pour

(56) J NICOD La geometrW >.l.me jrrMuctl '-~' du MotU-

. "U uveau 8YS~ 1. 1~WfW'"

(57) A. PADOA, n no De'f.l,XierM CO'fIlJ'''''' ,

l'algebre', C. !1'1902) 249--56. id Hilbert"L'E"""~'

maticiens (ParlS, 'N 2 de M. l)aVl

'Le probleme o.

(58) A. PADOA, (1903), 85-91.

ment M(~th., v

(34) E. V. HUNTINGTON, 'New sets of indep€'ndent postulates for the algebra of logic', Trans. Amer. math. Soe., xxxv (1933), 274- 304, 551-8, and 971.

(35) E. V. HUNTINGTON, 'Independent postulates for the "informal" part of Principia Moihematica", Bull. Amer. math. Soc., xl

(1934), 127-36. . . .,

(36) S. JASKOWSKI, 'Recherches sur Ie systeme de. 180 log~que ~t~tIon. iste " Actes du Congres International de Phd080phte Sc;wntifique, vi (Paris, 1936).

(36a) S. JASKOWSKI, 'Une modification des definitions fondamentales de III. geometrie des corps de M. A. Tarski ', Ann. Soc. polan.

Math., xxi (1948), 298-30!. ..

(36b) S. JASKOWSKI, 'Trois contributions au calcul des proPOSitiOns bivalent', Stwlia Societatis Scieniiarum Torunensis, Section A. i (1948), 3-15. , (36e) B. J6NSSON and A. TARSKI, 'Boolean algebras with operators ,

part I, Amer. J. Math., lxxiii (1951), 891-939. . .

(37) T. KOTARBlNSKI, Elementy teorji poznania, logiki formalneJ ~ metodoZogji nauk (Elements of the theory of knowledge, formal logic, and the methodology of the sciences) (Lw6w, 1929).

(38) C. KURATOWSKI, 'Sur la notion de l'ordre dans 180 tMorie des

ensembles', Fund. Math., ii (1921),161-71. , .

(39) C. KURATOWSKI, 'Evaluation de 180 classe borelienne ou projective d'un ensemble de points i:t I'aide des symboles logiques', Fund. Math., xvii (1931), 249-72. . . , (40) C. KURATOWSKI, 'Uber eine geometrische Auffassung der LoglStlk , Ann. Soc. polon, Math. ix (1931), 201.

(41) C. KURATOWSKI, Topologie I (Warszawa·Lw6w, 1933).

(42) C. KURATOWSKI, 'Sur l'operation A de I'analysis situs', Fund.

Math., iii (1922), 192-5.

(43) C. KURATOWSKI, 'Les ensnmblos projoct.ifs ot t'induction trans, finio ', Fund, Math., xxvii (1936), 209-711.

(44) C. H. LANGFORD, 'Analytical cornplotoneaa of pORhll!lf,t, R(4s', Prot.

London math. Soc., xxv (1926). 11542.

(45) C. H. LANGFOltD, 'SOTnfl UWOrtHll>1 011 dndlldhilit.y', A nnrds of Mathcfl'ULtic.s, xxviii (11127). 16 ,to ; '1'h""""I1I>1 .. ;1 .Inducibility (Second paper) ', ihid., 459-71.

(45a) S. LESNIEWSKI, PodstaW1j ogolnej tcorii mnogo.4ci 1 (Foundations of general set theory I) (Moscow, 1916).

(46) S. LESNIEWSKI, "Grundzugs eines neuen Systems der Grundlagen der Mathematik', §§ 1-11, Fund. Math., xiv (1929), 1-81.

(47) S. ~SNlEWSKI, 'fiber die Grundlagen der Ontologie', C. R. Soc.

Sci, Lett. Varsovie, Ulass« III, xxiii (1930), 111-32. .

(47a) S. LESNIEWSKI, '0 podatawach matematyki ' (On the foundatiOns of mathe'?~tics, in Polish) Przeglqd Filozojiczny, xxx (1927): ~6.4-206; ibid., xxxi (1928), 261-91 ; ibid., xxxii (1929), 6(}..-I05, ibid., xxxiii (1930), 77-105; ibid., xxxiv (1931), 142-70.

460

BIBLIOGRAPHY

BIBLIOGRAPHY

" B ' rachunek zdan' (The full

(66b) J. SE.UPECKI, "Pelny t~oJwa~~ cW)"'lnnales Universitatis Mariae three-valued sentential co. c us,

, klod ka i (1946) 193-209, , '

Cune-S ows, " d d keji' (Some investigatIOns

, 'Zb d 'nadteorJ~ e u

(66e) B. SOBOCINSKI, a an 'Prze I d Filozojiezny, xxxv

upon the theory of deductwn) 9 q

(1932),171-93., 'wnych wielowartosciowych

(66d) B. SOBOCINSKI, :~sJoma~~,zai:i:~atization of certain many~ system6w teorJI dedukeli ( d d tion in Polish), Roczmkl valued systems of the theor,y of A e B~~t~w 'Uniwersytetu JrJzcfa Prac Naukowych Zrz~8z~nt~36 8%99--419, , Pihudskiego w Warszawte, I (1 ), ti for Boolean algebras ,

h f representa ions

(67) M. H. STONE, 'The t eory 0, 6 37-111.

Trans. Amer. math., So~., XI (~9:h~' theory of Boolean rings ;0 (68) M. H. STONE, 'ApplIcatwns 0 er math. s»; xii (1937), 3~5--48 . general topology', Trans. A~ ~he theory of Boolean rmgs to

(69) M. H. STONE, 'Applications 0 xxix (1937), 223-303. ,

general topologv', Fund- Math., t tions of distributive lattices

T• I . cal represen a , J Matematiky a (70) M. H. STONE, ' ,opo og~ , Caso is pro pestovam

d Brouwerlan logics , P

~ysiky lxvii (1937-8), 1-25. "_:a' Fund. Math., vi (1925),

, I embles ll1llO ,

(71) A. TABSKI, 'Sur es ens la

. f damentales de

45--95. ur les notIOns on Math vii

(72) A TABSKI, 'Remarques s ti es' Ann. Soc, polon. "

'methodologie des matMma iqu , 'h

, d Ii rmahzowanyc

(1929) 270-2. " wd w odniesiemu 0 so, reference to

(73) A. TABSKI, '0 PO~licJU~ra(O ~he notion of truth ill .. (1930-1),

nauk dedukcYJnych ,n ) Ruch Filozojiczny, xu

d d t.ive SCIences,

formalized e ue I A Soc,

Z hlen ' nn.

210-11... fin' bare Mengen reeller a '

(74) A. TABSKI, 'Ub~r de ier 206-7. ar rapport a

polon. Maih-, IX (1931), d'ensembles closes ~ ri (1930),

(75) A TARSKI 'Sur les classes , 'Fund. Jlat " xv

- , 'elementalTes ,

certaines operatwns dedukti\'en

S achender

181-304. ,iff in den pr , ch_natur-

(76) A. TAItflKI, 'D(lr WahrlJ(ld·ltsbwcg:s. in Wien, Jla_the71l~~ (1932),

, 'I' 'Akad cr , cJ AIIZ€lger,

DIs:l:lp men, : Akademt8 ler

wissm8ch(ljtliche ]\lasse, , h (On the eon-

23-20. d 'pzykach nauk dedukCY1.!:t...,.w 1933,

P , , praw Y W J> ' ces) " ... - , h n

(700) A, 'l'AIlI'KI, oJ~oe s of deductive Belen 'Jjsierten Spr&C e •

cept of truth in language, b 'ff in den fonna

Wahrhelts egn

(76b) A TARSKI, Der 935) 261--405,

'Studia Philosop~ica, Ii ~~as (New York"1949jik der Boole'schen (77) A. TABSKI, Cardtnal a 9 l_mgen zur Axioma III xxxi (1938),

'E' 'ge Bemer~ .... • 'Cwsse'

(78) A. TABSKI, iru S; Lett. VarsOVW' f

b 'C R Soc. c.. tl'TflJI "

Alge ra, . . , Jar"",l ~\'R

5 nts In en " , (I9:lfl),

33-3 . d 'dable statcme mholic Log!", I'

(79) A. TABSKI, 'On un ec\ of truth', J. Sy 1

logic and the concep <'~ Till/lIn, .l1,,1.,.

, A71n.,..,f,._.

105--12. , d Mengenkorpem, ..1

(80) A. TABSKI, 'Jdealem en , ',~I.'n;,( .. nklJrr.'r IITI'

xv (1937), 186-9, " d rnultiphkatJ\C

To, dditl\'e un

(81) A.TARSKI, 'ubcra

461

(58a) C. S. PEIRCE, 'On the algebra of logic: a contribution to the philosophy of notation', Amer. J. Math., vii (1885), 18~202 .. (59) M. PIERI, 'La. geomctria elementare instituita sull_e n_O~lOne di "punto" e "sfera''', Memorie di Matemaiica e di Eisica della Societe Italiana delle Scienze, Serie Terza, xv (1908), 345--450. (60) E. L. POST, 'Introduction to a general theory of elementary

propositions', Amer. J_ Math., xliii (1921), 163-85. '.

(60a) W. A, POGORZELSKI, 'Przeglgd twierdzeti 0 dedukcji dla raehunku zda~ , (A survey of the deduction theorems for sentential calculus, III Polish), Studia Logica, 15, (1964), 163-79.

(61) M_ PRESBURGER, 'Dber die Vollstandigkeit cines gewissen Sy~te~s der Arithmetikganzer Zahlen, in welchem die Addition als e~~Ige Operation hervortritt', C.R, Premier Conqres des l'rlathhnatwtens des Pays Slaves, IVarszawa 1929 (Warszawa, 1930), 92-1?l.,

(61a) A. ROSE, 'The degree of completeness of lU-valued LukaSIewICZ propositional calculus', J. London math. Soc., xxvii (1952), 92-102.

(61b) J. B. ROSSER and A. R. TURQUETI'E, • Axiom schemes for 711- valued propositional calculi', J. Symbolic Logic, x (1945), 61-~2. (61e) J. B. ROSSER and A_ R. TURQUETTE, 'A note on the deductn:e completeness of m-valued propositional calculi', J. Symho'tw Logic, xiv (1950), 219-25.

(61d) B. A. W. RUSSELL, The principles of mathematics (London, 1903). (61e) B. A, W. RUSSELL, Introduction to mathematical philosophy (London, 1920).

(62) E, SCHRODER, Vorlesungen iiber die Algebra der Logik (exa~!~ Logik) (Leipzig; i, 1890; ii, part 1, 1891; ii, part 2, 1905; Ill, part 1, 1895). , , (62a) K. SCHRi:iTER, 'Deduktive abgeschlossone Mengen ohne BasIS,

Mathematische Nachriehien, vii (1952), 293-304. ,

(63) H.M.SHgFFER,Tlwgeneraltheoryojnotalionalrelntivity(CambrIdge, Mass" 1921),

(63a) H. M_ SHEFh'ER, 'A Aet of five post.ulntos for Boolean algebras with application to logical cunat.nnt.s '. Trans. A mer. malh. Soc"

xv (1913), 4!H--8, .

(64) TH, I'iKoLl'M, 'UnLI'rHuehllllJ.("" iihl'l' dill Axion«: ,h'lI H.luRAt'nkalkuls und iibnT I'rodllkt,nt.ioll!'l' lIud HIIJlllltnt,j(lIlHprohlnJlIl', wolcho j,(OWiH"" KlIlHH"n von AIIHIlItJ.("" ""1,1',,1'1',,,,', Skri]: r ide1l-

sk(tplJsclskulJei, I. Mal.-nul. kl<uJse }919, N 0_ 3 (lUlU). ..

(65) TH_ SKOLEM, 'Logisch-kornhinatoriseho UnLorsuchwlgen uber die Erfiillbarkeit oder Bewcisbarkcit muthemati::lcher satze nebst einem Theorem tiber dichte Mengen ', Skrift Videnskapsselskapet, I. ~at.-nat. klasse 1920, No, 4 (1920). , ,

(66) TH. SKOLEM, 'Uber einige Grundlagenfragen der Mathematik , Skrijter utgitt av Det Norske Videnskaps-Akademi i Oslo, I. Mal,-nat. kla.sse 1929, No.4 (1929).

(600) J, ,&UPECKI, 'Dow6d aksjomatyaowalnoset pelnych Byste~6w w~e,lowartosciowrch rachunku zdari' (A proof of the axiomatlz; bill~Y of full systems of many-valued sentential cal~ulus, PolIsh), C_ R_ Soc. Sci. LeU_ Varsovie Classe III, XXXIl (1939),

110-28. '

462

BIBLIOGRAPHY

Mengenfunktionen', C. R. Soc. Sci. LeU. Var8ovie, Cla8se III, xxx (1937), 151-81.

(82) A. TARSKI, 'The semantic conception of truth and the founda· tions of semantics', Philosophy aM Phenomenological Research, iv (1944),.341-76. ., , (83) A. TARSKI, 'A problem concerning the notion of definabihty, J.Symholic Logic, xiii (1948), 107-Il.

(84) A. T ARSKI, A decision. method for elementary algebra and geometry,

2nd ed, (Berkeley, Calif., 1951). .

(84a) A. TARsKl, 'Some notions and methods on the borderlm.e of algebra and metamathematics', Proc; Interna.tional Congres8 of Mathematicians, Cambridge, Mass., 1950, i (Providence, Rhode

Island, 1952), 705--20. .

(84b) A. TARSKI, 'II coneetta di verita nei linguaggi formali:lizati' ,Italian translation of [76bJ by Franoesea Rivetti Barbo, printed as part. of Franeesca Rivetti Barbo, L'antinomia del mentuore nel penstel'O amtemporaneo, Da Peirce a Turski. Societa editrice Vita e Pensiero, Milano, 1961,391-677.

(84c) A. TARSKI, Logique, Semaniique, MetamatMmatique, Librairie Armand Colin, Paris, 1972.

(84d) A. TARSKI, 'Truth and Proof', Scientific American, 220, (6), 1969,63-77. (84e) A. TARsKr, 'A problem concerning the notion of definability', J. Sym-

bolic Logic, 13, (1948),79-104. .

(85) A. TARSKI and A. LINDENBAUlIl, 'Sur l' independance des notions primitives dans Ies systemes mathemat.iquea', Ann. Soc. polon. math., v (1927), 111-13.

(86) O. VEBLEN,' ASY9temofaxiomsforgoometry', Trans, Amer.math.

Soc., v (1904), 343-81.

(87) O. VEBLEN, 'The square root and the relations of order', Trone-

Amer. math. Soc., vii (1906), 197-9. .,

(87a) M. W AJSBERG, 'Aksjomatyzacja trdjwartosclowego rachunku zdan (Axiomatiaation of the three-valued propositional calculus)', C. R. Soc. Sci. LeU. Varsovie, Classe lII, xxiv (1931).

(87b) M. W M8DERO, "Beitrage zum MetauuElFlagf'llkalkiil I', Mh. Math.

PhY8., xlii (1935), 221-42.

(870) H. WANO, "Rornarks on thn COIllpRril'!flll of axiom RYlltIlIllEl'. Proe.

Nat. Acad. Sei., W(Ulhingtfffl, xxxvi (IWjO), H~ r.:J. l

(88) A. N. WHITEHEAD, An enquiry C(>nc:"nliTl{l till! 1Jri1lril'it:1I (Jf 7lllturt' knowledge (CamLritig'J, 1919).

(89) A. N. WHITEHEAD, The concept of nature (Cambridge, 1920) .. (90) A. N. WHITEHEAD and B. A. W. RUSSELL, Principia Mathernot~ca 2nd ed., i-iii (Cambridge, 1925--7).

(91) L. WI'lTGENSTEIN, Tractat'U8 loqieo-philosophicus (London, 19~2~. (92) E. ZERMELO, 'fiber den Begriff der Definitheit in der Axioroatik , Fund. Math., xiv (1929), 33~4.

AN ALYTICAL INDEX

and reorganizes the Suh.je~t

. . d x corrects, elrpands, f terminology, It I~

NOTE TO INDEX: ~h.IS III e much an index of i?eas. as 0 nd historians; hilt Index of the first edltl~n. ~;ul to philosophers, hngulsts, arehcnsive than its primarily designed to e ~~ I viewpoint it is more comp chosen to reflect even from the mathemahlc~l'ngs and subheadings were t cros&-referencing predecessor. par:ticular ki':~~rk. Redun~an.cy and f~e~u~arski's terminology

current interests m Tars ate for val'latlOns bot

h been included to compens . ts

a:~ein current logical terminology. b ttention to the following POI~ .

Use of the index will be facilitated I t'mYe: in one article the inb~exp~f;:nn~:

. sed severa I . tting An IC .... -

1. When a term or idea !S u eral of the a~icle, omi ticles is conveniently

only the boldface Roman num The paginatIOn of the ar

. I ccurrences. ted

orals of partlcu ar 0 f ch inde)[ page. f term is indica

printed at the foot 0 ea rthy occurrence 0 • ~ ed page numbers

. lly notewo I itahclZ h

2. In many cases, an c:speclae number. For. ex~mp e, ontre.sts, and 80 fort .

by italicizing :~e Arabic ~~ions, characterizatIOns, C brackets. Square indicate definitIons, expla . without square used by him in the

3. Terms actually used by Ta~s~ a6r;~rski ~r e~se ~tbY '(SOUNDNF..';SJ:

brackets [ I enc~c ter:~I:~!lthOUgh the id~~ ::~r~, the te~ ::~~:I

relevant sense. ;N~~j' do appear in Ta~ I were introduced 10

and '[COMPAC t occur because t ey

and 'compactness' do n? I s were written. nillation and ease usage long after the artlc e . t for reasons of .~Tgs under a given

it is conveOlen, . t of subh"""ln . k and once

4. In several instances ave two separate ~IS s once with an. aste~hich classify

of cross-reference, to ~e term is listed ~wlce'finds subheadmgsd W subheadings term. In these case\. under 'AXIOM on~AXIO)I.' one.fin :nd so forth)

without. For examp, xioms and under. extensionality,

Tarski's remarks about a . rns (e g., chOice,

t· I r a)[IO . term

indicative of par ICU a to nees of the .

that Tarski IIReS or refers· . d not contain occ~ used by Tarski.

I cations Cited. 0 that is actua ,y use the Id~

5. In several cases thhe °t rm indexed IS on\eSIS LOGICAL beca thOugh it 18 . I when t e ed'S:' "g< even __ ~ Indexer even . is cited un er ars on page "": ·t is n"'--=-'J

For example, pagtl 3~ I 'logical basis' aP-h to use this IDd)e~ I get the idea that Tarski expresses Y n page 384. . us, (see 2 above 0 nees of the

db that term 0 ed latiOnS . occurre

not referre to y f the italiciz oC d not contain

to read one or two 0 di locations that 0 tisted

referred to before rea 109 . erminology are n?~ut on

term indexed. . r variations in ,tISO\lORPHIS~1

ted that mlno . ed nder .

6. It should also be no le page 309 is Cit 0 ~ism' . n indn

separately. For exa~P 'rphic' not 'isom rp .1. ' followed by a fltr

. fi ds '!Somo h '8M _0 thc!ll'l!·.

this page one n . d by 'see' or Ybh -,ling in pMpn'mlllar.tlI,n

. . dICate 811 eaw h 11011: ' ..

7. Cross-reference IS III e necessary, to. a, under the CSllio~ ',.xl'r' .... 'oo

heading adjoined, wh~h ding 'operation on' of the h~ c"' ...... rdr""nrMl example, after the sUsu::eading 'o~rat~o;:)'. 'See' indlcah H '

cross-reference to the ions (operation

is indicated by 'see express