RUDOLF CARNAP

THE UNITY
OF SCIENCE

Ke y T e x t s
Classic Studies in th e H istory of Ideas
RUDOLF CARNAP

TH E UNITY
OF SCIENCE
 Printed and Bound by
Antony Rowe Ltd., Chippenham, Wiltshire
Key Texts
Classic Studies in the History of Ideas

THE UNITY OF SCIENCE

Rudolf Carnap

Translated with an Introduction by

M . Black

THOEMMES
PRESS
 © Thoemmes Press 1995

Published in 1995 by
Thoemmes Press
11 Great George Street
Bristol BS1 5RR
England

ISBN 1 85506 391 3

This is a reprint of the 1934 Edition

Publisher’s Note
These reprints are taken from original copies of each book.
In many cases the condition of those originals is not perfect,
the paper, often handmade, having suffered over time and
the copy from such things as inconsistent printing pressures
resulting in faint text, show-through from one side of a leaf
to the other, the filling in of some characters, and the break
up of type. The publisher has gone to great lengths to ensure
the quality of these reprints but points out that certain
characteristics of the original copies will, of necessity, be
apparent in reprints thereof.
 CONTENTS

PAGE
I ntroduction by M. B lack . 7
A uthor ' s I ntroduction 21
P hysics as a U niversal Science
1. T he H etero g en eity of
Science . . . . 31
2. L anguages . . . . 37
3. P rotocol L anguage . 42
4. T h e P hysical L anguage as
an I ntersubjective
L anguage 52
5. T he P hysical L anguage as
a U niversal L anguage 67
6. P rotocol L anguage as a
P art of P hysical
L anguage 76
7. U n ifie d Science in P hysical
L anguage 93
 INTRODUCTION
B y M. B lack

1. Origins of the Viennese Circle1. The

so-called Viennese circle of philosophers,
to which Professor Carnap belongs, in­
herits a tradition of empirical and anti-
metaphysical thought, continuous in
Vienna since the middle of the Nineteenth
Century and fostered by a long series of
eminent university teachers of Philosophy
and Science. The growth of this trend
of thought coincided with the rise of
Liberalism in Austria-Hungary (especially
from 1848 onwards) and drew much of its
inspiration from the empirical and utili­
tarian elements of progressive thought in
England at the same period. Of the
academic teachers prominently in sym­
pathy with this movement, some of the
1 For the subject-matter of this section I am
chiefly indebted to a pamphlet, “ Die wissen-
schaftliche Weltauffassung. Der Wiener
Kreis ”, in the series Veroffentlichungen des
Vereines Ernst Mach (Vienna, 1929). The same
pamphlet also contains a useful detailed biblio­
graphy.
7
best known are Th. Gomperz (1869-80)1,
the translator of J. S. Mill, Mach (Privat-
dozent 1861-4, Professor 1895), and
Boltzmann (Mach’s successor, 1902-6).
Parallel to the work of these men were the
attempts made to reform the traditional
Logic of Aristotle and the Scholastics, of
which the beginnings can be seen in
Bolzano (especially in the Wissenschaft-
slehre, 1837), and a fuller development in
Brentano (Professor of Philosophy in the
Theological Faculty, 1874-1880, after­
wards Dozent in the Philosophical Faculty)
and Hofler (1853-1922). Of the many
who took an active part in the philo­
sophical discussions of Brentano's circle
in the late Nineteenth Century, we may
pick out von Meinong (in Vienna 1870-82,
afterwards Professor at Graz).
Subsequently, the most effective in­
fluences on this trend of thought were
the researches, in Logic and the Foundat­
ions of Mathematics, of Russell and others
of the ‘ logistic ' school2 (especially through
Principia Mathematica, 1910). Russell's
influence has been since reinforced by
1 The dates given refer in each case to the
period spent in Vienna.
a Cf. M. Black, The Nature of Mathematics, p. 7
and pp. 15 ff. for an account of these theories.
8
Wittgenstein’s Tractatus Logico-Philo-
sophicus (published in book-form in 1922)
which solved one of the major problems
of an empiricist outlook by providing a
more satisfactory solution of the nature
of Logic and Mathematics. This book
has been the chief inspiration of many
distinctive features of the contemporary
positivist1 movement in Vienna.
The ‘ Viennese circle in its present
form, originated in informal discussions
dating from the appointment of Moritz
Schlick, in 1922, as Professor of Philo­
sophy in Vienna. Out of these arose the
“ Verein Ernst Mach ” (formally created
in 1928 with Schlick as chairman), an
association " for propagating and further­
ing a scientific outlook ”, and for ” creat­
ing the intellectual instruments of modem
Empiricism ” . The first number of
Erkenntnis, a periodical devoted to further­
ing the aims of the circle2, appeared three
years ago (1931). The long article by
Professor Carnap which has been translated
1 This and other descriptions occurring in the
above are misleading if taken too literally. Cf.
page 29 below.
aErkenntnis contains reports of the various
conferences, lectures, etc., arranged by the
group, in addition to papers on Philosophy and
the methodology of Science.
9
for this book appeared in Erkenntnis
(Vol. ii, 1932, pp, 432-465) under the title
“ Die physikalische Sprache als Universal-
prache der Wissenschaft ” , and has been
revised, by the author, for this edition.
An authoritative statement1 of the
programme of the Viennese circle declares
that its outlook is “ characterized, not so
much by special assertions, as by its
fundamental attitude, its point of view
and by the direction of its researches.
Its goal is unified Science : its endeavours
are to relate and harmonize achievements
of individual researchers in the various
branches of Science. From this choice of
subjects arises the emphasis on collective
work ; hence also the prominence allotted
to communicable knowledge; these aims
inspire the search for a neutral system of
symbols, free from the dross of historical
languages, the search for a complete
system of concepts. We strive for order
and clarity, reject all dim vistas and
fathomless depths. In Science there are
no ‘ depths ’, all is on the surface . . . the
scientific outlook knows no insoluble rid­
dles. Analysis of the traditional problems
1 “ Die Wissenschaftliche Weltauffassung. Der
Wiener Kreis ”, p. 15.
10
of Philosophy reveals them to be, in
part, problems in appearance only (pseudo-
problems) and, for the rest, able to be
transformed into questions subject to the
verdict of the empirical sciences. The
clarification of such problems and state­
ments constitutes the object of philo­
sophic activity

2. Relations to Wittgenstein. Active

co-operation is sufficiently rare in Philo­
sophy to deserve attention. More so when
the social activity is based on the inspira­
tion of the most solipsist, therefore,
by conventional implication, the most
anti-social of all philosophers. But if the
Viennese circle is very deeply indebted
to the opinions of Wittgenstein, it would
be a mistake to emphasize that connection
so far as to underestimate the considerable
influence of Mach and Russell on the
circle, or the many elements of novelty
owing nothing to either of these predeces­
sors.
The derivation from Mach explains
many of the paradoxes involved in the
descent of a belligerent group from a
philosophy as quietist in temper as
11
Wittgenstein’s1. From Mach comes, by
direct descent, that belief in the value of
group activity which has marked the work
of the Viennese circle; from him, also,
and by indirect influence through Russell,
those elements of Pragmatism already
present, by implication, in Wittgenstein,
which go far to mitigate the asperities of
a marriage between Empiricism and
Solipsism. If legitimate, the results of
this union are of the highest importance ;
for it has fallen to many to exorcise
Metaphysics from Philosophy, but it has
been left for the Logical Positivists, as they
are sometimes called, to behave as if they
had succeeded. It is not for the trans­
lator to estimate their success, or to rob
the reader of the pleasures either of in­
vective or applause. But professional
philosophers, who have heard with un­
failing equanimity their treatises described
as compendia of ‘ nonsense ’, may be
interested to find here detailed and ingeni­
ous arguments for refutation ; and scient­
ists, who have always found time for a
1 By Wittgenstein’s opinions I mean always
to refer to the views expressed in Tractatus
Logico-Philosophicus. In the absence of any
subsequent writings it is not yet possible to say
to what extent these views have since been
revised.
12
malicious grin at the expense of Philo­
sophy, are bound to welcome this latest
attem pt to bring order into a disgraceful
muddle of mutually intolerant opinions.
The analytic method adopted by the
Viennese circle culminates in the judgment
that there are no distinctive philosophical
problems. Speculative philosophy must
be transformed into a new methodology,
the analysis of linguistic forms. But to
say so much is to lay too much emphasis
on the iconoclastic aspects of these opinions.
Their chief concern is to consolidate the
achievements of scientific discovery by
analyzing the limitations and essential
structure of the language in which all
knowledge must be expressed. Involved
in this programme is the demarcation of
boundaries between various departments
of linguistic expression, an aim which may
receive as much applause from theologians
and artists as from scientists, and abuse
only from those who find pleasure in the
choice of their own labels. Divisions of
any sort imply principles of justification ;
in the underlying principles on which the
Viennese circle base their separation
between Sciences and the domains of
‘ nonsense ’ the critic may expect to find
13
their most constructive contributions;
here also should be the origin of their
divergence from Wittgenstein. For the
latter, also, establishes criteria of sense,
purges Philosophy of Metaphysics and
separates both from the Sciences; but with
different consequences.

3. The notion of sense. The theory of

Tractatus Logico-Philosophicus centres
round the notion of ‘ sense whose specifi­
cation is linked with and reveals that con­
ception of the essential structure of
language on which is based the doctrine
of the untenability of traditional philo­
sophy. Since the sense of statements is
defined in terms involving reference to
‘ atomic ’ statements, or ' atomic ' facts,
the latter notion is the hub of Wittgen­
stein’s account. The same notion receives
a distinct modification in the theories of
the Viennese circle.
In the Tractatus the world (i.e. the
subject-matter of philosophical analysis)
is conceived to consist ultimately of simple
irreducible ‘ objects', occurring in com­
plex arrangements or ‘ configurations
and thereby constituting facts or states of
14
affairs. To each ultimate complex arrange­
ments corresponds the simplest kind of
true statement, viz., an ‘ atomic ' state­
ment, in such a way as to reflect the form
of the fact by a one-one correspondence
between objects and words. In virtue of
the correlation between constituents of
the atomic fact and elements of the true
atomic statement, the latter are combined
in the same structure as the form er; and
all statements of more complicated refer­
ence to the world than atomic statements
must, in so far as they are statements and
not mere collocations of words, be reduc­
ible to logical conjunctions and disjunc­
tions of atomic statements. This doctrine
can also be expressed in an alternative and
more striking m anner: in order to have
sense, a statement must be verifiable in
(my) experience i.e. the words of which it
is composed must be definable in terms of
words which refer to (my) immediate exper­
ience. Unless, indeed, the process of definit­
ion ends in this way, instatementscomposed
of words with immediate reference, it will
never be possible to know what is meant
by the group of words, which will then be
just nonsense. Hence the importance of
atomic statements ; they are the elements
15
on which all other statements are based;
all assertions which are not downright
nonsense are either truth functions (logical
combinations) of atomic statements, or
else hypotheses, i.e. rules for constructing
atomic statements. Strictly speaking, only
atomic statements and truth functions of
atomic statements have sense.
In this necessarily inadequate summary
of Wittgenstein’s teaching three important
aspects can be isolated : (I) recognition of
the importance of logical structure, (II)
the exclusion of logical structure from
being itself the subject-matter of state­
ments1, and (III) verifiability in (my)
experience as a necessary criterion of the
sense of statements. Easily deducible
from these principles is the nonsense of
most traditional questions in philosophy,
e.g. concerning the existence of other
minds, the reality of the external world,
etc. Without entering into the detailed
criticism which these views deserve, it is
possible to point out their dogmatic
11 have spoken of statements rather than of
facts in accordance with recent practice of the
Viennese circle. The whole of the above account
can be easily transformed into an ‘ objective ’
account in terms of facts, and postulates e.g.
no belief in subjective idealism.
16
character and to hint at their consequences.
It is clear, in the first place, that
Wittgenstein’s doctrine of nonsense cannot
be refuted; for any attempted refutation
in other than its own terminology must
be ‘ nonsense ’, hence ipso facto empty of
all assertion ; on the other hand, there are
no internal inconsistencies to be found
in the theoretical formulation of what is
essentially a privative doctrine. The
most serious consequences of the prohibi­
tion of certain combinations of words can
be to restrain its defenders, in consistency,
from making any assertions at all. That
this may in fact be the logical outcome of
accepting the system of Tractatus Logico-
Philosophicus is shown by any attem pt
to specify in detail the nature of atomic
statements (out of which all genuine state­
ments must be constructed). For it has
been seen that atomic statements must
have immediate reference; hence they
must have instantaneous verification.
(I can be acquainted with the ultimate
‘ objects ’ only at the moment of acquaint­
ance and reference to them at any other
moment, e.g. by description, must be
indirect.) Such statements would there­
fore necessarily be composed of (logically)
17
proper names for tne objects of acquaint­
ance during the instant at which the state­
ment is made. Atomic statements would
contain no descriptive terms and must
therefore be unintelligible except to the
speaker and to him only at the moment
of utterance ; since, moreover, its elements
would be merely distinguishable signs for
objects of unknown logical form, the
atomic statement would be bare of all
except the most trivial structure. It is
questionable how far such ' statements ’
have any right to the name ; and
‘ hypotheses or statements involving any
element of generality, are in still worse
case, since they are merely prescriptions
for the manufacture of atomic statements
and can at most be supported. Thus
atomic statements can be verified but
express nothing; while all other state­
ments express but can never be verified.
Consistent adherence to the principles
of the Tractatus is thus seen to have radical
consequences; the philosopher is either
to wag his finger, like Kratylos, in lieu of
speech, or escapes silence at the expense of
discord between theory and practice.
Hence the Tractatus cannot be said to have
succeeded in exhibiting the structure of
18
 language and drawing the line between
Science and Metaphysics ; for the adopted
criterion of sense is so stringent as to exclude
the whole of Science from the region of
sense in order to share a crowded limbo
with the bulk of everyday knowledge and
the speculation of metaphysicians.
The absence of detailed discussion of
everyday knowledge embodied in
' hypotheses' is the weakest point of the
treatment of the Tractatus. Here is
where the advantages of the pragmatic
approach of the Viennese circle are chiefly
to be found. Their modifications (for
an adequate account the reader must
refer to their own works1) can be summar­
ized as the rejection of the substratum
of atomic facts while preserving Wittgen­
stein's insistence of the importance of
logical structure. The theory of atomic
statements is replaced in Carnap’s essay
(translated in this book) by the theory of
the ‘ p r o t o c o l t h e direct record of ex­
perience, whose form now remains in­
determinate (cf. especially Sections 3, p. 42
1 E.g. in Erkenntnis. Cf. also “ Logical
Positivism and Analysis" (Henriette Hertz
Lecture) by L. S. Stebbing, London, 1933, and
Blumberg and Feigl, “ Logical Positivism ",
Journal of Philosophy XXVIII, 281-296 (1931).
19
below). Rejection of the metaphysical
presuppositions of the Tractatus is then
pursued to the extreme limit of excluding
all reference to the ‘content ’ of statements
and the practice of a special ‘ form al>x
mode of speech from which all such refer­
ence has been eliminated. The absence
of the doctrine of atomic facts permits the
retention of a wider criterion of verifi­
ability in sense experience while allowing
a pragmatical sense to general statements,
natural laws and hypotheses. This solu­
tion brings its own difficulties; it blurs the
very definite outlines of the notion of
structure in Wittgenstein and leaves truth
in an uncomfortable half-way house
between correspondence and coherence.
For if the truth of statements is provisional
some account is needed that does not make
their truth dependent on human conveni­
ence or human prejudice. This is the
difficulty that earlier Pragmatism had
to meet and could never answer satisfac­
torily. For Logical Positivists also, it
is a pressing question, but the absence
of a final answer cannot detract from the
exceptional interest and importance of
their work.
1 Cf. also R. Carnap, “ On the Character of
Philosophical Problems ”, Philosophy of Science,
I. pp. 5-19, 1934.
20
 AUTHOR’S INTRODUCTION
THE VIENNESE CIRCLE DOES NOT PRACTICE
PHILOSOPHY

The reader may find it easier to under­

stand the main article if I preface it by
some remarks on the general nature of the
views held by the Viennese Circle to which
I and my friends belong.
In the first place I want to emphasize
that we are not a philosophical school and
that we put forward no philosophical
theses whatsoever. To this the following
objection will be made : You reject all
philosophical schools hitherto, because you
fancy your opinions are quite n e w ; but
every school shares this illusion, and you
are no exception. No, there is this
essential difference, must be the answer.
Any new philosophical school, though it
reject all previous opinions, is bound to
answer the old (if perhaps better formu­
lated) questions. But we give no answer
to philosophical questions, and instead
reject all philosophical questions, whether
21
of Metaphysics, Ethics or Epistemology.
For our concern is with Logical Analysis.
If that pursuit is still to be called Philo­
sophy let it be so ; but it involves excluding
from consideration all the traditional
problems of Philosophy. In origin,
Philosophy included Mathematics and
also, until recently, the sciences of Sociol­
ogy and Psychology. At the present time,
these studies have been separated from
Philosophy in order to constitute indepen­
dent branches of Science. Both General
Logic and the Logic of Science, i.e. the
Logical Analysis of scientific terms and
statements, must be separated from
Philosophy, in the same fashion, in order
to be pursued according to an exact, non-
philosophic, and scientific method. Logic
is the last scientific ingredient of Philo­
sophy ; its extraction leaves behind only
a confusion of non-scientific, pseudo
problems.
Metaphysicians—whether they are sup­
porters of Monism, Dualism or Pluralism,
of Spiritualism, Materialism or some other
‘ -ism ’ propound questions concerning
the essence of the Universe, of the Real, of
Nature, of History, etc. We supply no
new answers but reject the questions
22
themselves as questions in appearance only.
Epistemology claims to be a Theory of
Knowledge, to answer questions as to the
validity of knowledge, the basis on which
knowledge rests. Here again are to be found
many answers from various ‘ -isms ’ ;
naive and critical Realism, subjective,
objective and transcendental Idealism,
Solipsism, Positivism, etc., have as many
different answers. We supply no new
answer but reject the questions themselves
since they seem to have the same character
as those of Metaphysics. (The case is
altered if the questions are formulated
not as philosophical enquiries but as a
Psychological enquiry concerning the origin
of knowledge; in the latter form the
question is proper to Science and can be
investigated by the empirical methods of
Psychology; but such an answer has
nothing to do with the philosophical theses
of the -isms mentioned.) If ‘ Epistemology *
is understood to denote unmetaphysical,
purely logical analysis of knowledge, our
work certainly falls under that classification.
Ethics raises the question of the basis
of validity of moral standards (prin­
ciples of value) and of the specifica­
tion of valid norms. Answers are given
23
b y Idealists, Utilitarians, Intuitionists,
etc. Here again we reject the ques­
tions themselves in view of their
metaphysical character. (The case is
otherwise in psychological or sociological
investigations of the actions and moral
judgments of mankind ; such a method is
certainly both unobjectionable and scienti­
fic, but its results belong to the empirical
sciences of Psychology and Sociology, not
to Philosophy. It is better to avoid the
term ‘ Ethics ' for such investigation in
order to avoid confusion with normative or
regulative Ethics.)
As against the preceding subjects, our
own field of investigation is that of Logic.
Here are to be found problems of pure
Logic, i.e. questions relating to the con­
struction of a combined logical and mathe­
matical system with the help of symbolic
Logic. Further, the problems of applied
Logic, or the Logic of Science, i.e. the
logical analysis of terms, statements,
theories, proper to the various department
of science. Logical Analysis of Physics,
for example, introduces the problems of
Causality, of Induction, of Probability, the
problem of Determinism (the latter as a
question concerning the logical structure
24
of the system of physical laws, in divorce
from all metaphysical questions and from
the ethical question of freedom of will).
Logical Analysis of Biology, again, in­
volves the problems of Vitalism, to take
one example (but here again in a form free
from Metaphysics, viz. as a question of the
logical relations between biological and
physical terms or laws). In Psychology,
Logical Analysis involves, among others,
the so called problem of the ‘ relation
between Body and Mind' (here also a
non-metaphysical question, concerned not
with the essential nature of two realms of
being but with the logical relations between
the terms or laws of Psychology and
Physics respectively). In all empirical
sciences, finally, Logical Analysis involves
the problem of verification (not as a
question concerning the essence of Truth,
or the metaphysical basis of the validity
of true statements, but as a question con­
cerning the logical inferential relations
between statements in general and so
called protocol or observation statements).
In this fashion we use Logical Analysis
to investigate statements of the various
kinds proper to the various departments
of Science. The statements of traditional
25
Philosophy can also be subjected to the
same treatment. The result is to reveal
the absence of that logical relation (of
implication) to empirical statements and,
in particular, to protocol statements,
whose presence is a necessary condition
for the verifiability of the statements in
question and is therefore usually, and with
justice, required in the findings of all
scientific procedure. All statements be­
longing to Metaphysics, regulative Ethics,
and (metaphysical) Epistemology have
this defect, are in fact unverifiable and,
therefore, unscientific. In the Viennese
Circle, we are accustomed to describe such
statements as nonsense (after Wittgen­
stein). This terminology is to be under­
stood as implying a logical, not say a
psychological, distinction; its use is
intended to assert only that the statements
in question do not possess a certain logical
characteristic common to all proper
scientific statements ; we do not however
intend to assert the impossibility of
associating any conceptions or images
with these logically invalid statements.
Conceptions can be associated with any
arbitrarily compounded series of words; and
26
metaphysical statements are richly evoca­
tive of associations and feelings both in
authors and readers. It is precisely that
circumstance which so hinders recognition
of their non-scientific character.
In traditional Philosophy, the various
views which are put forward are often
mixtures of metaphysical and logical
components. Hence the findings of the
Logical Analysis of Science in our circle
often exhibit some similarity to definite
philosophic positions, especially when these
are negative. Thus, e.g., our position is
related to that of Positivism which, like
ourselves, rejects Metaphysics and requires
that every scientific statement should be
based on and reducible to statements of
empirical observations. On this account
many (and we ourselves at times) have
given our position the name of Positivism
(or New Positivism or Logical Positivism).
The term may be employed, provided it is
understood that we agree with Positivism
only in its logical components, but make
no assertions as to whether the Given is
real and the Physical World appearance,
or vice versa ; for Logical Analysis shows
that such assertions belong to the class
of unverifiable pseudo-statements. Our
27
views are related, in similar fashion, to
those of Empiricism, since we follow that
theory so far as to reject a priori judg­
ments ; Logical Analysis shows that every
statement is either empirically verifiable
(i.e. on the basis of protocol statements),
analytic, or self contradictory. On this
account, we have at times been classified,
both by ourselves and by others, as
Empiricists.
The following article is an example of
the application of Logical Analysis to
investigating the logical relations between
the statements of Physics and those of
Science in general. If its arguments are
correct, all statements in Science can be
translated into physical language. This
thesis (termed ‘ Physicalism' by Neurath)
i9 allied to that of Materialism, which
respectable philosophers (at least in
Germany, whether in other countries also
I do not know) usually regard with
abhorrence. Here again it is necessary
to understand that the agreement extends
only as far as the logical components of
Materialism; the metaphysical compon­
ents, concerned with the question of
whether the essence of the world is mater­
ial or spiritual, are completely excluded
28
from our consideration. In the final
section of the article it is shown that
methodical Materialism and methodical
Positivism are not incompatible; in the
terminology which I have been using here,
this is as much as to s a y : the logical
components of Positivism and Materialism
are mutually compatible. This same
example shows how great is the need for
caution in classifying the opinions of the
Viennese Circle under any of the old -isms.
Between our view and any such traditional
view there cannot be identity—but at
most agreement with the logical compon­
ents. For we pursue Logical Analysis,
but no Philosophy.
Prague, January 1934. R. C.

29
 A d v ic e to t h e R e a d e r .

Some of the words of most frequent occurrence

in the following paper are unfortunately without
exact English equivalents; the translations
adopted are likely to be misleading without
some explanation.
‘ Determination ' (for * Bestimmung ’) :
a description,
or any indeterminate sym­
bol whose exact value
(usually numerical) is
obtained as the result of
definite operations,
or the result of such oper­
ations.
Singular statement (for ‘ Einzelsatz ’): state­
ments describing particular
states of affairs in contrast
to general statements.
Physical language is used technically and does
not denote the terminology
customary in Physics
(cf. p. 95).
Nonsense (or psewdo-espression) is intended to
carry none of its usual
abusive connotation. Tech­
nical use = whatever can­
not be verified in experi­
ence.

30
 PHYSICS
AS A UNIVERSAL SCIENCE
B y R udolf Carnap

1. T he H eterogeneity of Science.
2. Languages.
3. P rotocol L anguage.
4. T he P hysical Language as an
I ntersubjective Language .
5. T he P hysical Language as a
U niversal L anguage.
6 . P rotocol L anguage as a part of
P hysical L anguage.
7. U nified Science in P hysical
Language.

1. T he H eterogeneity of Science .
Science in its traditional form con­
stitutes no unity, and can be separated
into Philosophy and the technical sciences.
The latter can be classified again as formal
sciences (Logic and Mathematics) and
empirical sciences. I t is usual to sub­
divide the last class further and to under­
stand that it includes, in addition to the
'n a tu ra l' sciences, Psychology and the
Geisteswissenschaften (social sciences) gen­
erally.
31
 The basis of these various divisions is
not merely convenience; rather is the
opinion generally accepted that the various
sciences named are fundamentally distinct
in respect of subject matter, sources of
knowledge and technique. Opposed to
this opinion is the thesis defended in this
paper that science is a unity, that all
empirical statements can be expressed in
a single language, all states of affairs are
of one kind and are known by the same
method.
Very little will be said here concerning
the nature of Philosophy and the formal
sciences. The author's views on this point
have already been sufficiently explained
by others on several occasions. Detailed
attention will however be given to the
question of the unity of the empirical
sciences.
It is to modern developments in logic
and particularly in the logical analysis
of language that we owe our present in­
sight into the nature of Logic, Philosophy
and Mathematics. Analysis of language
has ultimately shown that Philosophy
cannot be a distinct system of statements,
equal or superior in rank to the empirical
sciences. For the activity of Philosophy
32
consists rather in clarifying the notions
and statements of science. In this way
does cleavage of the field of knowledge
into philosophy and empirical science
disappear ; all statements are statements
of the one science. Scientific research
may be concerned with the empirical
content of theorems, by experiment,
observation, by the classification and
organization of empirical m aterial; or
again it may be concerned with establish­
ing the form of scientific statements,
either without regard for content (formal
logic) or else with a view to establishing
logical connections between certain specific
concepts (Konstitutionstheorie and theory
of knowledge considered as applied logic).
Statements in Logic and Mathematics
are tautologies, analytic propositions,
certified on account of their form alone.
They have no content, that is to say,
assert nothing as to the occurrence or
non-occurrence of some state of affairs.
If to the statem ent: “ The (thing) A is
black” we add “ or A is blue”, the
supplemented statement still conveys
some information though less than at
first. If, however, we replace the supple­
mentary phrase previously chosen by ‘ or
33 B
A is not black ’ the compound statement
no longer conveys any information at all.
It is a tautology, i.e. is verified by all
circumstances. From such a statement
no knowledge of the properties of the
thing A can be derived. Theorems in
Logic and Mathematics have, nevertheless,
in spite of tautologous character and lack
of content, considerable importance for
science by virtue of their use in trans­
forming statements having content. For
the present thesis it is important to
emphasize that Logic and Mathematics are
sciences having no proper subject matter
analogous to the material of the empirical
sciences. Postulation of ‘ formal ’ or
‘ ideal ’ objects to be set against the ' real ’
objects of empirical sciences is unnecessary
in the theory here briefly sketched.
Statements having content, i.e. state­
ments, as is usually said, expressing some
state of affairs, belong to the field of
empirical sciences. Our chief question is
whether these statements, or to speak
more conventionally, whether the states
of affairs expressed by such statements
axe divided into several mutually irre­
ducible kinds. The traditional answer
is in the affirmative ; and it has been
34
usual to make the chief distinctions
between the subject matters of natural
science, History, the social sciences, etc.
(Geisteswissenschaften), and Psychology.
On the basis of observations and ex­
periments, the natural sciences describe
the spatio-temporal events in the system
which we call ‘ Nature From the
individual accounts thus obtained arise
general formulae, so-called ‘ laws of
n a tu re ' (the process of ' induction ').
These in turn make it possible to obtain
new specific statements, e.g. predictions
(the process of ‘ deduction ').
History, and the social sciences also use
the method of observing material events.
The usual view maintains, however, that
observation in such fields is merely a
subordinate method, the proper method
being ‘ understanding ’, empathy (‘ Ein-
fiihlung ') projection of oneself into histori­
cal monuments and events in order to
grasp their ' essence The further ques­
tion arises, so it is maintained, in all
sciences dealing with culture in the widest
sense as well as in specifically normative
disciplines such as ethics, of comprehend­
ing ‘ values ', of establishing ‘ norms '.
The usual view therefore is that the subject
35
matter of such branches of knowledge,
the Geisteswissenschaften as Germans say,
whether they are ‘ significant forms ’ or
systems of values, are of a nature funda­
mentally different from the subject matter
of natural science and cannot be under­
stood by the methods of natural science.
As to the nature of Psychology widely
divergent views are prevalent. Experi­
ments are made, measurements often
taken of factors capable of quantitative
determination. Many psychologists there­
fore include their science among the natural
sciences, but while doing so accentuate
the difference between their respective
subject matters. Psychology, they say,
deals with the ‘ psychical ’, with the
phenomena of consciousness, perhaps
also of unconsciousness, while other
natural sciences treat of the ‘physical’.
Other psychologists, again, lay the em­
phasis on the relation between their
science and the moral sciences. In
Psychology also, they say, knowledge is
gained by ‘ understanding ’ and empathy.
The difference consists in the fact that
Psychology does not deal with works of art
and institutions, as Ethics and Sociology
do, but with the regularities to be found
36
in the phenomena of consciousness. These
various conceptions yet agree in the answer
they furnish to the questions which we
wish to discuss. Psychology is a science
with its own fundamentally distinct and
isolated subject matter.
We shall not need to discuss in further
detail at this point divergent views of the
relations between the various sciences. It
is sufficient to remember that it is usual
to speak of fundamentally distinct kinds
of objects ; it matters little for our purpose
whether the distinction is made in the
manner described above (e.g. ‘ ideal ’ and
're a l' objects ; physical, psychical objects ;
'values’) or in some other. All such accept­
ed views are contrary to the thesis of the
unity of Science.

2. L anguages.
In formulating the thesis of the unity
of Science as the assertion that objects
are of a single kind, that states of affairs
are of a single kind, we are using the
ordinary fashion of speech in terms of
' objects ’ and ‘ states of affairs ’. The
correct formulation replaces ‘ objects ’ by
' w ords' and ‘ states of affairs' by
‘ statements ’, for a philosophical, i.e. a
37
logical, investigation must be an analysis
of language. Since the terminology of
the analysis of language is unfamiliar we
propose to use the more usual mode of
speech (which we will call ‘ material ’)
side by side with the correct manner of
speaking (which we will call the ‘formal ’).
The first speaks of ‘ objects ‘ states of
affairs of the ‘ sense ’, ‘ content ’ or
' meaning ’ of words, while the second
refers only to linguistic forms.1
In order to characterize a definite lan­
guage it is necessary to give its vocabualry
and syntax, i.e. the words which occur in
it and the rules in accordance with which
(1) sentences can be formed from those
words and (2) such sentences can be trans­
formed into other sentences, either of the
same or of another language (the so-called
1 A strictly formal theory of linguistic forms
(‘ logical syntax ’), will be developed later. The
‘ thesis of syntax ’ which has only been sketched
in the above will there be explained in detail
and justified. It asserts that all propositions
of philosophy which are not nonsense are syn­
tactical propositions, and therefore deal with
linguistic forms. (So-called propositions in
metaphysics, on the other hand, can be only
the subject-matter of suntactical statements,
e.g. of a statement which asserts their syntactical
invalidity, i.e. which asserts that they are
nonsense.)
(The book here announced is Logische Syn­
tax der Sprache, Vienna, 1934.)
38
rules of inference and rules for translation).
But is it not also necessary in order to
understand the ‘ sense ’ of the sentences,
to indicate the ‘ meaning ’ of the words ?
N o ; the demand thereby made in the
material mode is satisfied by specifying
the formal rules which constitute its
syntax. For the ‘ meaning ’ of a word is
given either by translation or by definition.
A translation is a rule for transforming a
word from one language to another,
(e.g. ‘ cheval' = ‘ horse ') ; a definition
is a rule for mutual transformation of
words in the same language. This is true
both of so-called nominal definitions (e.g.
* Elephant ’ = animal with such and such
distinguishing characteristics) and also, a
fact usually forgotten, for so-called osten-
sive definitions (e.g. ‘ E lephant' = animal
of the same kind as the animal in this or
that position in space-time) ; both defini­
tions are translations of words.
At the expense of some accuracy we may
also characterize a language in a manner
other than in the formal mode above
and, using the more ‘ intuitive ’ material
mode, say a language is such that its
statements describe such and such (here
would follow a list of the objects named in
39
the language). The alternative formula­
tion is permissible provided the writer and
the reader are clear that the material mode
is only a more vivid translation of the
previous description in the formal mode.
If this is forgotten the danger may arise
of being diverted by the material mode of
speech into considering pseudo-questions
concerning the essence or reality of the
objects mentioned in the definition of
a language. Nearly all philosophers and
even many Positivists have taken the
wrong turning and gone astray in this
way.
As an example we may take the language
of arithmetic. In the formal mode, this
particular language might be character­
ized as follows:—
Arithmetical statements or sentences are
compounded of signs of such and such
a kind put together in such and such a way;
such and such (specified) rules of trans­
formation apply to them.
Alternatively, using now the material
mode, we could s a y :—
Arithmetical thorems state certain pro­
perties of numbers and certain relations
between numbers.
Though such a formulation is inexact
40
it can be clearly understood and is permis­
sible if carefully handled. One must not,
however, be led by this formulation into
considering pseudo-questions about what
kind of objects these ‘ num bers' are,
whether they are ‘ re a l' or ‘ ideal extra­
mental or intra-mental, etc. If the formal
mode is used, in which ‘ numbers ’ are
replaced by ‘ numerical s y m b o ls s u c h
pseudo-questions vanish.
In the rest of the paper we shall at all
times help the reader by using both modes
of expression and write the formal, and,
strictly speaking, only correct, expression
of our thought in a parallel column on the
left of the more usual formulation.
Various ‘ languages ’ can be distin­
guished in science. Let us for example
consider the language of economics which
can be characterized in somewhat the
following fashion : i.e., by the fact
that its sentences that its propositions
can be constructed describe economic
from expressions : phenomena such as
‘ supply and de­ supply and demand,
mand ’, ‘ wage etc.
' price ’, etc. . . .
put together in
such and such a way.
41
 We will call a language a universal
language
if every sentence if it can describe
can be translated every state of
into it, affairs,
and if this is not the case, a ‘ p artial'
language. The language of economics
is a ‘ partial ’ language since e.g.
a theorem in phy­ the state of an
sics concerning the e le c tro -m a g n e tic
vectors of an elec­ field in some region
tro-magnetic field cannot be described
cannot be trans­ in economic terms.
lated into the lan­
guage of economics.

3. P rotocol L anguage.
Science is a system of statements based
on direct experience, and controlled by
experimental verification. Verification in
science is not, however, of single statements
but of the entire system or a sub-system
of such statements. Verification is based
upon ‘ protocal statements ', a term whose
meaning will be made clearer in the course of
futher discussion. This term is understood
to include statements belonging to the basic
protocol or direct record of a scientist’s (say
a physicist’s or psychologist’s) experience.
42
Implied in this notion is a simplification of
actual scientific procedure as if all experi­
ences, perceptions, and feelings, thoughts,
etc., in everyday life as well as in the
laboratory, were first recorded in writing as
' protocol’ to provide the raw material for a
subsequent organization. A ‘ primitive '
protocol will be understood to exclude all
statements obtained indirectly by induc­
tion or otherwise and postulates therefore
a sharp (theoretical) distinction between
the raw material of scientific investigation
and its organization. In practice, the
laboratory record of a physicist may have
approximately the following form :
‘ Apparatus set up as follows
arrangement of switches pointer
readings of various instruments at various
t i m e s s p a r k i n g discharge takes
place at 500 volts ’. Such a set of state­
ments is not a primitive protocol in view of
the occurrence of statements
deduced with the which describe
help of other state­ states of affairs not
ments from the pro­ directly observed.
tocol.
A primitive protocol would perhaps run
as follows: " Arrangement of experiment:
at such and such positions are objects of
43
such and such kinds (e.g. ' copper wire ' ;
the statement should be restricted perhaps
to ‘ a thin, long, brown body ’ leaving the
characteristics denoted by ' copper ’ to be
deduced from previous protocols in which
the same body has occurred) : here now
pointer at 5, simultaneously spark and
explosion, then smell of ozone there ” .
Owing to the great clumsiness of primitive
protocols it is necessary in practice to
include terms of derivative application
in the protocol itself. This is true of the
physicist's protocol and true in far greater
measure of the protocols made by bio­
logists, psychologists and anthropologists.
In spite of this fact, questions of the justi­
fication of any scientific statement, i.e.
of its origin in protocol statements, in­
volve reference back to the primitive
protocol.
From now onwards ' protocol state­
ments ' will be used as an abbreviation for
‘ statements belonging to the primitive
protocol ’ ; the language to which such
statements belong will be called the
*protocol-language ’. (Sometimes also
termed ‘ language of direct experience ' or
' phenomenal language ' ; the neutral term
‘ primary language ’ is less objectionable.)
44
In the present state of research it is not
possible to characterize this language with
greater precision, i.e. to specify its vocabul­
ary, syntactical forms and rules. This is,
however, unnecessary for the subsequent
arguments of this paper. The analysis
which follows is a sketch of some of the
views as to the form of protocol statements
held at the present day by various schools
of thought. Though the author will take
no sides in the issues involved the inciden­
tal discussion will elucidate still further the
meaning of the term ‘ protocol-language ’.
The simplest state­ The simplest state­
ments in the pro­ ments in the pro­
tocol-language are tocol-language refer
protocol-statements to the given, and
i.e. statements need­ describe directly
ing no justification given experience or
and serving as phenomena, i.e. the
foundation for all simplest states of
the remaining state­ which knowledge
ments of science. can be had.
Question : What Question : What
kinds of word occur objects are the ele­
in protocol state­ ments of given,
ments ? direct experience ?
45
First Answer : Pro­ First answer : The
tocol statements are elements that are
of the same kind a s : directly given are
‘ joy now ’, ‘ here, the simplest sensa­
now, b lu e; there, tions and feelings.
red
Second ans we r : Second answer:
Words like ‘ blue ’ Individual sensa-
do not occur in sions are not given
protocol statements directly but are the
but appear first of result of isolation.
all in derived pro­ Actually given are
positions (they are more complex ob­
words of higher jects such a s :—
type). Protocol
statements on the
other hand are of
forms similar to the
following :—
(a) ' Red circle, (a) Partial gestalts
now ’ of single sensory
fields, e.g. a visual
gestalt,
or ( b ) .................... or (b) Entire sen­
sory fields, e.g. the
visual fields as a
unity,
46
° r ( c ) ..................... or (c) The total ex­
perience during an
instant as a unity
still undivided into
separate sense - re­
gions.
Third answer : Pro­ Third answer : Mat­
tocol statements erial things are ele­
have approximately ments of the given :
the same kind of a three dimensional
form as ‘ A red cube body is perceived as
is on the table ’. such directly and
not as a series of
successive two-
dimensional pro­
jections.
These are three examples of contem­
porary opinions which are, of course,
usually expressed in the material mode.
The first can be termed ‘ Atomistic
Positivism ’ and is approximately Mach’s
standpoint. Most present-day critics
regard it as inadequate, for objections
brought against it by subsequent psy­
chologists and especially followers of
Gestalt psychology are to a great extent
justified. Opinion on the whole tends to
choose between the variations included in
47
the second of the answers given above.
The third view in our classification is not
often held to-day; it is however more
plausible than it appears and deserves more
detailed investigation, for which this is
however not the place.
Statements of the system constituted
by science (statements in the language of
that system) are not, in the proper sense
of the word, derived from protocol state­
ments. Their relation to these is more
complicated. In considering scientific
statements, e.g in physics, it is necessary
to distinguish in the first place between
‘ singular ' statements (referring to events
at a definite place and time, e.g. ‘ the
temperature was so much at such and such
a place and time ') and the so-called ‘ laws
of nature ', i.e. general propositions from
which singular propositions or combina­
tions of such can be derived (e.g. ‘ the
density of iron is 7 ‘ A (always and every­
where ’). In relation to singular state­
ments a ' law ’ has the character of an
hypothesis ; i.e. cannot be directly deduced
from any finite set of singular statements
but is, in favourable cases, increasingly
supported by such statements. A singular
statement (expressed in the vocabulary
48
of the scientific system) has again the
character of an hypothesis in relation to
other singular statements and in general the
same character in its relation to protocol
statements. From no collection of protocol
statements, however many, can it be
deduced, but is in the most favourable case
continually supported by them. In fact
deduction is possible but in the converse
direction. For protocol statements can
be deduced by applying the rules of infer­
ence to sufficiently extensive sets of singu­
lar statements (in the language of the
scientific system) taken in conjunction
with laws of nature. Now the verification
of singular statements consists of perform­
ing such deductions in order to discover
whether the protocol statements so ob­
tained do actually occur in the protocol.
Scientific statements are not, in the strict
sense, ' verified' by this process. In
establishing the scientific system there is
therefore an element of convention, i.e.
the form of the system is never completely
settled by experience and is always
partially determined by conventions.
We will now consider the case of a
person A undertaking, with the help of
his protocol, verifications of scientific
49
statements in the manner described above.
The question whether each person has his
own protocol language will be discussed
later. For the present A's own protocol
language will be referred to as ' t h e '
protocol language.
Whenever the rules If a state of affairs
of transformation described by p can
state the conditions be reduced to facts
under which state­ about given, i.e.
ments in the proto­ direct, experience of
col language can be A, A has in theory
deduced from a the possibility of
statement p, it is verifying p. A then
always possible, in knows the ‘ sense ’
principle, for A to of p, for the ‘ sense ’
verify p. Whether of p, or what is
A can actually expressed by p, con­
do this depends on sists of the method
empirical circum­ of verification, i.e.
stances. If, how­ in the reduction to
ever, there is no the given. If some
such inferential re­ statement p is not
lationship between in this inferential
a statement p and relation to state­
statements of the ments concerning
protocol language the given, p cannot
then p is not verifi- be understood by
50
able for A ; p has no A, i.e. p is nonsense.
sense, is formally
incorrect.
In such a case A For if A is to under­
cannot understand stand a statement
the statement p, he must know what
for to ' understand ' states of affairs in­
means to know the volving the given
consequences of p, (what possible di­
i.e. to know the rect experiences)
statements of the are the case if p is
protocol language true.
which can be de­
duced from p.
If an inferential re­ If the state of affairs
lation of the kind expressed by a
described holds be­ statement p is veri­
tween a statement fiable in the manner
p and each of the described by several
protocol languages persons
of several persons
p has sense for each such person. In
such a case p will be said to have sense (for
those persons) inter-subjectively. This term
of course is relative to the persons who
understand the statement in the manner
described. By a language ‘ inter-sub­
jective ' (for certain persons) will be under­
stood a language whose statements are
51
inter-subjective (for those persons). A
statement p, which is inter-subjective (for
certain persons), is said to be inter-sub-
jectively valid if p is valid for each person,
i.e. if it is supported, in sufficient measure,
by the protocol statements of each such
person.
It will be proved in the following para­
graphs that the physical language is
inter-subjective and can serve as a univer­
sal language, i.e. as a language in terms of
which all states of affairs could be ex­
pressed. Finally, an attem pt will be
made to show that the various protocol
languages also can be regarded as partial
languages, in the sense defined above, of
the physical language.

4. T he P hysical Language as an I nter-

subjective L anguage .

The physical language is characterized

by the fact that statements of the simplest
form (e.g. the temperature of such and
such a place at a specified time is so
much),
attach to a specific express a quantita­
set of co-ordinates tively determined
(three space, one property of a
52
time co-ordinates) a definite position at
definite value or a definite time.
range of values of
a coefficient of phys­
ical state.
Quantitative determination can also be
replaced by qualitative, as is usual in
science as well as in everyday life, for
reasons of brevity and ease of understand­
ing. Qualitative determinations can
therefore be included in the physical
language provided
rules are set up for they can be under­
translating all such stood as determina­
statements into tions of physical
quantitative deter­ states of affairs or
minations so that occurrences so that
e.g. the statement e.g. “ It is rather
“ It is rather cool coolhere ” and “ The
here ” might be temperature here is
translated into the between 5 and 10
statement “ The degrees centigrade”
temperature here is are taken as state­
between 5 and 10 ments of identical
degrees centigrade” . sense.
This characterization of the physical
language corresponds to the traditional
form of physics (for the sake of simplicity
53
we are neglecting the coefficients of prob­
ability which occur in the physical state­
ments). We wish however to interpret
the term ‘ the physical language ’ so widely
as to include not the special linguistic
forms of the present merely but also such
linguistic forms as physics may use in any
future stage of development. It may be
that physical position will eventually be
determined by more or less than four
co-ordinates; perhaps it will not be possible
to regard the co-ordinates simply as tem­
poral and spatial magnitudes. Such
modifications are of no importance for
present purposes. The physical language
will certainly continue to be so constituted
that every protocol that every fact of
statement composed perception in every­
entirely of words day life, e.g. every­
which can be (quite thing that can be
crudely) described learnt about light
as sensation-, per­ or material bodies
ception-, or thing- (in the naive inter­
words, can be trans­ pretation) can be
lated into it. expressed in the
physical language.
This property of physical language is
sufficient for our further discussion. It is
54
unnecessary to specify further the exact
form of physical languages which may
possibly arise in the course of the future
evolution of physics. In order to facilitate
direct understanding the spatio-temporal
linguistic forms will always be used in the
following paragraphs. On the basis of the
property of the physical language just
mentioned, our thesis now makes the
extended assertion that the physical
language is a universal language, i.e. that
every statement can every state of affairs
be translated into it. can be expressed
in it.
In addition to the simplest form of
statements previously described, namely
of singular statements, there are now vari­
ous compound propositional forms to be
considered. The most important is the
general implicative statement expressing
a general implication : if at some point P
in space-time there is a determination a
(i.e. a is the value of a certain magnitude
at a certain place and time denoted by P)
then there will be some other point P 1,
standing in such and such a spatio-
temporal relation to P, at which (P1)
there will be with such and such probabil­
ity a determination a1 — f(a) functionally
55
dependent on a. This is the general form
of a natural law in its widest meaning.
P and P 1 often coincide. An example
involving qualitative determinations:
‘ Blood is red ’ ; a : distinguishing char­
acteristic of blood assumed not to include
its colour; P = P 1 ; a1 : the colour red.
An example involving quantitative deter­
minations : the second of Maxwell’s equa­
tions :

a : the determination of the spatial dis­

tribution of the electrical field in the
neighbourhood of P which is denoted by
‘ curl E ’ ; P 1 = P ; a1 : the rate of change
of the magnetic field at P, denoted by-^j-
The possibility of applying science, i.e.
of making predictions concerning sub­
sequent occurrences, depends upon the
formulation of laws of nature.
The concepts of physics are quantitative
concepts, numerical determinations. This
is a fact of decisive importance in per­
mitting prediction on the basis of exact
natural laws. Another peculiarity of
physical concepts, which is of importance
for the present discussion, consists in their
56
abstractness and the absence of qualities
from their enunciation. This is to be
interpreted as follows: The rules of
translation from the physical language
into protocol language are of such a kind
th at no word in the physical language is
ever correlated in the protocol language
with words referring only to a single sense
field (e.g. never correlated with determin­
ations of colour only or sound only).
It follows that a physical determination
permits the inference of protocol state­
ments in every sensory field. Physical
determinations are ' inter-sensory ’ in a
sense which will be immediately explained.
Moreover, they are also ‘ inter-subjective
in agreement with the experiences of the
various subjects; this will be discussed
later.
The determination: ‘ A note of such and
such pitch, timbre and intensity ’, in the
protocol language or in the language of
qualities (which we need not at present
distinguish) corresponds to the following
determination in the physical language :
‘ Material oscillations of such (specified)
basis frequency with superimposed addi­
tional frequencies of such (specified)
amplitudesBut
57
 a physical state­ the presence of such
ment containing oscillations can be
these determina­ determined not only
tions is correlated by auditory sensa­
not only with state­ tions (the sound of
ments containing such a note) but
the corresponding also, with the help
determination in the of suitable instru­
auditory field but ments, in the form
also under certain of visual and tactile
conditions with sensations.
statements contain­
ing determinations
from other sensory
fields.
There are no coefficients of physical
state exclusively correlated with quantita­
tive determinations in a single specific
sensory field. This is a fact of fundamental
importance. For any qualitative deter­
mination in some sensory field, we can
determine, with the help of qualitative
determinations from other sensory fields,
the class of the correlated physical deter­
minations. As shown by the illustration
used above, qualitative determinations
in the auditory field can be translated into
physical statements of a particularly
58
simple form. The process is more com­
plicated when determinations of colour
are involved, e.g. ‘ green of such and such
a kind ’ (denoted by a number, from
Ostwald's colour atlas say). Correlated
in such a case is not a single physical
state but a set of physical states. Every
state of this class consists of a definite
combination of frequencies of electro­
magnetic oscillations (e.g. for a definite
‘ green ’ this class includes a combination
of a wave-frequency of high intensity from
the green part of the spectrum and, a
‘ red ’ frequency of feebler intensity as
well as a combination of blue and yellow
frequencies of medium intensities, etc.).
It is an important fact that the com­
position of the set of physical determin­
ations correlated to a qualitative deter­
mination can be established experimentally
by using the fact that the physical
determinations are correlated to qualita­
tive determinations in other sensory fields.
Thus, e.g. the composition of the set of
combined combinations of frequencies
referred to above can be established only
in virtue of the fact that the frequencies
in question can be recognised by signs
other than their respective colours, e.g
59
by the position of the corresponding line
in the spectroscopic image. The colours
of the spectrum are redundant in the
implied experiment since a photograph
will furnish all the information required.
Hence a person completely blind to colours
could still establish frequencies occurring
at a definite position in space-time. So
far, we have remained inside the region
of visual sense, but it is possible to extend
this reasoning to other senses. It would
be possible for example to build into the
spectroscope an electrical apparatus for
exploring the spectrum, so constructed
that a radiation of sufficient intensity set
into motion a pointer which could be felt
or a microphone which could be heard.
By such means a person completely blind
would still be capable of determining the
frequency of an electro-magnetic oscilla­
tion.
From these arguments follows the
theoretical possibility of establishing re­
sults of the following three kinds :—
1.Personaldeterminations : A can discover :
which physical de­ under which physi­
termination (or class cal conditions he ex­
of physical deter- periences a definite
60
minations) corres­ quality (e.g. a defin­
ponds to a definite ite green).
qualitative deter­
mination in his pro­
tocol language (e.g.
‘ green of such and
such a kind ’).
That determinations of this kind are
theoretically always possible is due to the
fortunate circumstance (an empirical fact,
not at all necessary in the logical sense)
that
the protocol the content of ex­
perience.
has certain ordinal properties. This
emerges in the fact of the successful con­
struction of the physical language in such
a fashion that qualitative determinations
in protocol language are single-valued
functions of the numerical distribution
of coefficients of physical states.
On applying this to our example it
follows that the scales of the tactile,
visual and photo-spectroscopes can be
calibrated in such a fashion that these
instruments give the same reading for
every given case. In short the same
physical determinations correspond to
61
the qualitative determinations of every
sensory field; we shall use the abbrevi­
ation : physical determinations are valid
inter-sensorily.
2. Determinations by other persons : An
experimenter E (e.g. a psychologist) can
discover by using another person S (sub­
ject of an experiment) :
which physical de­ what are the physi­
termination (or set cal conditions in
of physical deter­ which S experiences
minations) corres­ a definite quality
ponds to a definite (e.g. a definite
qualitative deter­ green).
mination in S’s pro­
tocol language (e.g.
‘ green of such and
such a colour ’).
The procedure used is the following :—
E varies the physical conditions (e.g.
the combinations of various frequencies of
oscillations) and discovers the conditions
to which S reacts with the protocol state­
ment containing the qualitative determin­
ation in question. The possibility of such
a discovery is independent of
62
whether the corres­ whether E can also
ponding qualita­ sense the corres­
tive determinations ponding qualities,
(names of shades of
colours, etc.) occur
in E ’s protocol lan­
guage,
or of the possibility that E is colour blind
or completely blind. For in this case, as
in the case of his own experiences, E, as
previously stated, receives the same result
whether he uses the tactile, auditory or
photo-spectroscope. The discovery of the
set of these physical determinations cor­
responding to a definite qualitative deter­
mination will be called the ‘ physicalizing '
or physical transformation of this qualita­
tive determination.
The result of our discussion can now be
formulated as follows: A person can
physicalize the qualitative determinations
both of himself and of another person.
3. Determinations on other persons made
by several experimenters. If the experi­
ments on a single subject, S, as described
in the preceding subsection, are performed
not by a single experimenter E x but by
several experimenters E x, E a, . . . the
various results obtained are in mutual
63
agreement. This is due to the following
fact.
The determined value of a physical
magnitude in any concrete case is indepen­
dent not only of the particular sensory
field used but also of the choice of the
experimenter. In this we have again a
fortunate but contingent fact, viz. the
existence of certain structural correspon­
dences between
the protocols series of experi­
ences
of the various experimenters. A differ­
ence of opinion between two observers con­
cerning the length of a rod, the tempera­
ture of a body, or the frequency of an
oscillation, is never regarded in physics
as a subjective and therefore unresolvable
disagreement; on the contrary, attempts
will always be made to produce agreement
on the basis of a common experiment.
Physicists believe that agreement can in
principle be reached to any degree of
exactitude attainable by single investi­
gators ; and that when such agreement is
not found in practice, technical difficulties
(imperfection of instruments, lack of time,
etc.) are the cause. In all cases hitherto
64
 where the matter has been investigated
with sufficient thoroughness this opinion
has been confirmed. Physical determin­
ations are valid inter-subjectively.
Under headings (1) and (3) above we
have spoken of a ‘ fortunate accident’ ;
the state of affairs mentioned under (2)
is however a necessary consequence of the
others. It may be noticed however that
these facts, though of empirical nature,
are of far wider range than single empirical
facts or even specific natural laws. We
are concerned here with a perfectly general
structural property of experience which
is the basis of the possibility of intersensory
Physics (fact (1)), and intersubjective
Physics (fact (2)), respectively.
The question now arises whether another
language exists which is intersubjective
and can therefore be considered as a
language for Science. The reader’s
thoughts may turn perhaps to the language
of qualities, used say as a protocol lan­
guage. In virtue of the previously men­
tioned fact of the possibility of giving a
physical interpretation to the language of
qualities, the latter must be a sub-language
of the physical language. According to
customary philosophical opinion, however,
65 c
there can (or even must) be another
non-physical interpretation. It will be
shown later that there are objections to
such a non-physical interpretation, and
that, in any case, the language of qualities,
when so interpreted, is not inter-sub­
jective.
It will also be demonstrated that all
other languages used in science (e.g.
Biology, Psychology or the social sciences)
can be reduced to the physical language.
Apart from the physical language (and its
sub-languages) no intersubjective language
is known. The impossibility of an inter-
subjective language not included in the
physical language has certainly not yet
been proved ; there are however not the
slightest indications to suggest that such
a language exists. Further, not a single
determination, of any kind, is known
which, established intersubjectively, is
incapable of translation into the physical
language.
It is a just demand that Science should
have not merely subjective interpretation
but sense and validity for all subjects who
participate in it. Science is the system
of intersubjectively valid statements. If
our contention that the physical language
66
is alone in being intersubjective is correct,
it follows that the physical language is the
language of Science.

5. The P hysical L anguage as a U niver ­

sal L anguage .

In order to be a language for the whole

of Science, the physical language needs to
be not only intersubjective but also
universal. It follows, therefore, to con­
sider whether this is the case, i.e. whether
the physical language has the property
that
every statem en t every possible state
(whether true or of affairs (every con­
false) can be trans­ ceiv ab le s ta te ,
lated into it. whether actually oc­
curring or not) can
be expressed by it.
We will begin by considering the subject
m atter of the inorganic sciences, of Chemis­
try, Geology, Astronomy, etc. In these
regions doubt can hardly arise as to the
applicability of physical language. The
terminology employed is often different
from that of Physics but it is clear that
every determination arising can be reduced
to physical determinations. For the
67
definition of such determinations is always
in terms either of physical determinations
or else of qualitative determinations (e.g
results of observations) ; even in the
latter case no objections will be raised to
the physical interpretation of qualitative
observations in these sciences.
The first serious doubts will arise in
connection with Biology. For the issue
of vitalism is still violently controversial
at the present time. If we extract the
kernel of sense and reject the metaphy­
sical pseudo-questions which are usually
confused in the controversy, the essence
of the m atter in dispute can be stated as
the question whether the natural laws
which suffice to explain all inorganic
phenomena can also be a sufficient explan­
ation in the region of the organic. A
negative answer to this question, such as
supplied by vitalists, necessitates the form­
ulation of specific and irreducible biological
laws. The Viennese circle is of the opinion
that biological research in its present form
is not adequate to answer the question.
We therefore expect the decision to be
made in the course of the future develop­
ment of empirical research. (Meanwhile
68
our presumption tends more to an affirma­
tive answer.) It is however important
to notice that the question of the univer­
sality of the physical language is quite
independent of the vitalist-mechanist con­
troversy. For the former is a question of
reducing not biological laws to physical
laws but biological concepts (i.e. deter­
minations, words) to physical concepts.
And the fact of the latter reducibility can,
in contrast to that of the former, be easily
demonstrated. This will perhaps appear
obvious immediately the confusion of the
two issues is eliminated. Biological deter­
minations involve such notions as species,
organisms and organs, events in entire
organisms or in parts of such organisms,
e tc .; (notions such as ‘ will ’, ‘ image ’,
' sensation ’, etc. can be referred to
Psychology and omitted from consider­
ation here). Such notions are always
defined in Science by means of certain
perceptible criteria, i.e. qualitative deter­
minations capable of being physicalized ;
e.g. ‘ fertilization ’ is defined as the union
of spermatazoon and egg; ‘spermatozoon’
and ‘ egg ’ are defined as cells of specified
origin and specified perceptible proper­
ties ; ‘ union ’ as an event consisting of a
60
specified spatial redistribution of parts,
etc. It is possible to define with the help
of similar physical determinations the
meanings o f' metabolism ‘ cell-division
‘ growth ‘ development ‘regeneration',
etc. The same is true in general of all
biological determinations, whose defini­
tions always supply empirical and percept­
ible criteria.
(This is not the case however, for such
words as ' entelechy ’, but terms of this
sort belong to a vitalistic philosophy of
nature rather than to Biology and can
occur only in ‘ nonsensical ’ statements.
It can be shown that these terms represent
pseudo-concepts, incapable of formally
correct definition.1)
The preceding arguments show that
every statement in Biology can be trans­
lated into physical language. This is true,
in the first instance, of singular statements
concerning isolated events; the corres­
ponding result for biological laws follows
immediately. For a natural law is no
more than a general formula used for
deriving singular statements from other
1 Cf. Carnap, “ Ueberwindung tier Metaphy-
sik durch logische Analyse der Sprache",
Erkenninis, Vo1. II, p. 219.
70
singular statements. Hence no natural
law in any field can contain determinations
absent from the singular statements in
the same field. The question set by
vitalism of the relation of biological laws
(which laws the foregoing shows to be
translatable in all circumstances into
physical language and therefore to
belong to the general type of physical
law) to the physical laws valid in the
inorganic realm, does not even arise for
consideration here.
The application of our principles to
Psychology usually provokes violent op­
position. In this department of Science,
our thesis takes the form of the assertion
that all psychological statements
can be translated refer to physical
into physical lan­ evfents (viz. physi­
guage. This applies cal events in the
both to singular body, especially the
statements and to central nervous sys­
general statements tem, of the person
( ‘ p sy ch o lo g ic al in question) whether
laws ’). In other of definite single
words, the definition events or in general
of any psychological of events of specified
71
term reduces it to type in a specified
physical terms. person or, more gen­
erally still, of such
events in any per­
son. In other words,
every psychologic
concept refers to
definite physical
properties of such
physical events.
The problems raised by these statements
are to be dealt with in another paper and
will therefore not be discussed further in
this place.1
If the assertion of the possibility of
translating psychological statements into
physical language is well grounded, the
truth of the corresponding assertion con­
cerning the statements of (empirical)
Sociology easily follows. Sociology is
understood here in its widest sense to
include all historical, cultural and economic
phenomena ; but only the truly scientific
and logically unobjectionable statements
of these sciences belong to this classifica­
tion. The sciences mentioned often in
their present form contain pseudo concepts,
1Carnap, “ Psychologie in physikalischer
Sprache ”, Erkenntnis, Hi. 107-142, 1933.

72
viz. such as have no correct definition,
and whose employment is based on no
empirical criteria;
such words stand in such (pseudo-) con­
no inferential rela­ cepts cannot be re­
tion to the protocol duced to the given,
language and are are therefore void
therefore formally of sense.
incorrect.
Examples : ' objective spirit ’, ‘ the mean­
ing of H istory', etc. By (empirical)
Sociology is intended the aggregate of the
sciences in these regions in a form free
from such metaphysical contaminations.
It is clear that Sociology in this form deals
only with situations, events, behaviour
of individuals or groups (human beings or
other animals), action and reaction on
environmental events, etc.
These statements These events may
may contain physi­ be in part physical
cal and also psycho­ (so-called) and in
logical terms. If the part mental (so-
foregoing thesis, of called). If the fore­
the possibility of going thesis, of the
converting psycho­ possibility of reduc­
logical determina­ ing psychological
tions into physical notions and state-
73 D
determinations, is ments to physical
valid, then the same terms is valid, socio­
must be true of all logical events must
sociological terms be entirely physical.
and statements.
These principles were first enunciated in
fundamental outline by Neurath,1 who
1Neurath, *' Soziologie im Physikalismus ",
Erkenntnis, Vol. II, cf. also his " Physikalismus ”,
Scientia, Nov. 1931, " Empirische Soziologie.
Der wissenschaftliche Gehalt der Geschichte
und Nationalokonomie ” . Schriften z. wiss,
W eltauff.,Vol.Y , Vienna, 1931. Neurath was also
the first both in the discussions of the Viennese
circle and, later, in the first article mentioned,
to demand constantly the rejection of formula­
tions in terms of ‘ mental experience '. He
rejected the comparison between statements
and ‘ reality ’, insisting that the correct mode
was in terms only of statements and stated the
thesis of physicalism in its most radical form.
I am indebted to him for many valuable sugges­
tions. By distinguishing between the ' formal '
and the ‘ material ’ modes, rejecting the pseudo­
questions which use of the latter provokes, prov­
ing the universality of physical language,
and in the consistent application of the formal
mode to the construction of syntax (only
sketched in the present article) I have arrived
at results which wholly confirm Neurath's
views. Moreover the demonstration (par. 6),
in the present article, that the protocol language
can be included in physical language, disposes
of our previous difference of opinion on this
point (the question of the *phenomenal lan­
guage ’) which is mentioned in Neurath's article.
Neurath's suggestions, which have often met
with opposition, have thus shown themselves
fruitful in all respects.
74
has also discussed in detail their bearing
on the problems and methods of sociology ;
and his papers will be found to include
many examples of the possibility of formu­
lation in physical terms and of the elimin­
ation of pseudo concepts. We shall
therefore be able to omit any further
discussion on this point.
The various departments of science have
now been inspected. The standpoint of
traditional philosophy would demand the
inclusion of Metaphysics. But logical
analysis arrives at the result (cf. f.n. p. 70)
th at so-called metaphysical statements
are no more than pseudo statements,
since they stand in since they describe
no inferential rela­ no state of affairs,
tion (either positive either existent or
or negative) to pro­ non-existent. This
tocol statements. is due to the fact
They either contain that they either con­
words irreducible to tain (pseudo-) con­
protocol words or cepts which cannot
are compounded of be reduced to the
reducible words in a given and therefore
manner contrary to denote nothing, or
the laws of syntax. are compounded of
sensible concepts in
nonsensical fashion.
75
 Our investigations of the various depart­
ments of Science therefore lead to the
conclusion
that every scientific that every fact con­
statement can be tained in the sub­
translated into phy­ ject m atter of sci­
sical language. ence can be describ­
ed in physical lan­
guage.
We must investigate whether statements
in protocol language can also be converted
into physical language.

6. P rotocol Language as a part of

P hysical L anguage.
To what extent do statements in
protocol language conform to our thesis
of the universality of the physical lan­
guage ? That thesis demands that
statements in proto­ given, direct experi­
col language, e.g. ences are physical
statements of the facts, i.e. spatio-
basic protocol, can temporal events.
be translated into
physical language.
76
 Objections will certainly be raised to
these assertions. It will be said
“ Rain may be a
physical event but
not my present
memory of rain. My
perception of water
which is falling at
this moment and my
present joy are not
physical events ” .
This objection is in the spirit of usual views
on this question, and would be accepted
by most writers on the Theory of Know­
ledge. If this objection is considered
more closely it will be remarked, in the
first place, that it is directed only against the
material formulation of our thesis (in the
right hand column). We have previously
seen that the material mode is a mere trans­
formation of the correct formal mode of
speech and easily leads to pseudo-problems.
We shall therefore, regard this objection
critically in view of the fact that it can be
formulated only in the terminology of
the right hand column i.e. in the material
form, but for the moment, however, we
will leave such criticisms on one side and
77
adopt the (fictitious) procedure of regard­
ing the m atter from the standpoint of our
opponent: we shall, in the first place,
use the material mode quite freely
and, secondly, suppose that the objection
and the grounds on which it is based in
its material formulation are justified. It
will then appear that we are led into
insoluble difficulties and contradictions.
This fact will disprove the supposition
and dispose of the objection.
Let p be a singular statement in the
protocol language of a person Sx, i.e. a
statement about the content of one of
Sx’s experiences, e.g. ‘ I (i.e. Sx) am thirsty ’
or, briefly, ' Thirst now ’. Can the same
statement of affairs be expressed also in
the protocol language of another person
S8 ? The statements of the latter language
speak of the content of Sa's experiences.
An experience in the sense in which we
are now using the word is always the
experience of a definite person and cannot
at the same time be the experience of
another person. Even if Sx and S2 were,
by chance, thirsty simultaneously the two
protocol statements ‘ Thirst now ' though
composed of the same sounds would have
different senses when uttered by Sx and
78
 S2 respectively. For they refer to different
situations, one to the thirst of the other
to the thirst of S2. No statement in S2's
protocol language can express the thirst
of Sj. For all such statements express
only what is immediately given to S2 ;
and Si’s thirst is a datum for Sx only and
not for S2. We do say of course that S2
can ‘ recognise ’ the thirst of Si and can
therefore also refer to it. What S2 is
actually recognising however is, strictly
speaking, only the physical state of Sj's
body which is connected for S2 with the
idea of his own thirst. All that S2 can
verify when he asserts ' Sj is th irsty ' is
that Si’s body is in such and such a state,
and a statement asserts no more than can
be verified. If by *the thirst of Sj ' we
understand not the physical state of his
body but his sensations of thirst, i.e.
something non-material, then Si’s thirst
is fundamentally beyond the reach of
S2's recognition.
A statement about Si's thirst would
then be fundamentally unverifiable by Sa,
it would be for him in principle impossible
to understand, void of sense.
In general, every statement in any
person's protocol language would have
79
sense for that person alone, would be
fundamentally outside the understanding
of other persons, without sense for them.
Hence every person would have his own
protocol language. Even when the same
words and sentences occur in various
protocol languages, their sense would be
different, they could not even be compared.
Every protocol language could therefore be
applied only solipsistically; there would
be no intersubjective protocol language.
This is the consequence obtained by con­
sistent adherence to the usual view and
terminology (rejected by the author).
But even stranger results are obtained
by using, on the basis of our supposition,
the material terminology which we regard
as dangerous. We have just considered
the experiences of various persons and
were forced to admit that they belong to
completely separated and mutually dis­
connected realms. We will now consider
the relations between the content of my
own experiences say, as described by
statements in my protocol, and the cor­
responding physical situation as described
by singular statements in physical lan­
guage, e.g. ‘ Here the temperature is
20 degrees centigrade now ’. We have
80
on the one side the content of experience,
sensations, perceptions, feelings, etc., and
on the other side constellations of electrons,
protons, electro-magnetic fields, etc. ;
that is, two completely disconnected
realms in this case also. Nevertheless an
inferential connection between the protocol
statements and the singular physical
statements must exist for if, from the
physical statements, nothing can be
deduced as to the truth or falsity of the
protocol statements there would be no
connection between scientific knowledge
and experience. Physical statements
would float in a void disconnected, in
principle, from all experience. If, how­
ever, an inferential connection between
physical language and protocol language
does exist there must also be a connection
between the two kinds of facts. For one
statement can be deduced from another if,
and only if, the fact described by the first
is contained in the fact described by the
second. Our fictitious supposition that
the protocol language and the physical
language speak of completely different
facts cannot therefore be reconciled with
the fact that the physical descriptions can
be verified empirically.
81
 In order to save the empirical basis of
the physical descriptions the hypothesis
might perhaps be adopted that although
protocol language does not refer to physical
events the converse is true and physical
language refers to the content of experi­
ences and definite complexes abstracted
from such content. Difficulties then arise
however on considering the relation be­
tween the several persons' protocol lan­
guages and physical language. S1’s pro­
tocol language refers to the content of
Sx's experience, Sa's protocol language to
the content of S8’s experience. What can
the intersubjective physical language
refer to ? It must refer to the content
of the experiences of both Sx and S t.
This is however impossible for the realms
of experience of two persons do not over­
lap. There is no solution free from
contradictions in this direction.
We see that the use of the material
mode leads us to questions whose dis­
cussion ends in contradiction and insoluble
difficulties. The contradictions however
disappear immediately we restrict our­
selves to the correct, formal mode of
speech. The questions of the kinds of
facts and objects referred to by the
82
various languages are revealed as pseudo­
questions. These led us, in turn, to
further unanswerable pseudo-questions
such as the question how the reciprocal
convertibility of physical language and
protocol language is compatible with the
‘ fact ’ that the first refers to physical
situations and the second to experienced
content. These pseudo-questions are auto­
matically eliminated by using the formal
mode. If, instead of speaking of the
‘ content of experience ’, ‘ sensations of
colour ' and the like, we refer to ‘ protocol
statements ’ or *protocol statements in­
volving names of colours ' no contradic­
tion arises in connection with the inferen­
tial relation between protocol language and
physical language. Should then, those
expressions in the material mode not be
iised at all? Their use is in itself no
mistake, nor are they senseless, but we
see that the danger involved is even
greater than previously stated. For com­
plete safety it would be better to avoid
the use of the material mode entirely,
although it is the terminology usual
throughout the whole of Philosophy (also
in the Viennese circle). If this mode is
still to be used particular care must be
83
taken that the statements expressed are
such as might also be expressed in the
formal mode. That is the criterion which
distinguishes statements from pseudo­
statements in Philosophy. [Although the
danger that pseudo-questions may arise
in using the material mode is always
present, the contradictions can be avoided
by using the material terminology mon­
istically, i.e. by speaking exclusively of
the content of experience (in the spirit of
solipsism) or else exclusively of physical
states (in the spirit of materialism). If,
however, a dualist attitude is adopted,
as is customary in philosphy, if one speaks
simultaneously of ‘ content of experience '
and ‘ physical states ’, (' m atter ’ and
‘ spirit ’, ‘ body ’ and ‘ soul ’, ‘ m ental'
and ‘ physical ’, ‘ acts of consciousness ’
and ‘ intentional objects of consciousness ’)
then contradictions are unavoidable].
When all contradictions and pseudo­
questions have been eliminated by using
the formal mode, the problem still remains
of analyzing the reciprocal inferential
relations between physical language and
protocol language. We have previously
mentioned that if a sufficient number of
physical statements are given, a statement
84
in protocol language can be deduced. A
more precise consideration now shows that
the simplest form of such deduction is
found when physical statements describe
the state of the body which belongs to the
person in question. All other cases of
deduction are more complicated and can
be reduced to this case. (In describing
the state of the body, the state of the
central nervous system and especially the
brain is the most important, but further
details are unnecessary for our argument.)
For example, a protocol statement p :
“ red now (seen by S) ” can be deduced
from a definite description of the state of
S’s body.
The reader may still hesitate, feeling
that such a deduction is utopian and would
need full knowledge of the physiology
of the central nervous system for its
performance. This is not however the
case ; derivation of the required physical
statements is already possible and is
achieved in everyday life whenever com­
munication occurs. It is true that what
we know in such cases of the physical
situation of other persons’ bodies cannot
as yet be formulated as a numerical dis­
tribution of physical coefficients of state
85
but it can be formulated in other expres­
sions of the physical language which are
just what we require. Let us, e.g. denote
by ' seeing red ' that state of the human
body characterized by the fact that certain
specified (physical) reactions appear in
answer to certain specified (physical)
stimuli. (For exam ple; Stim ulus; the
sounds ' What do you see now ? ’
reaction : the sound, ‘ red Stimulus,
the sounds, ‘Point out the colour you have
just seen on this card ’ ; reaction : the
finger points to some definite part of the
card. Here all those reactions must be
counted that are usually regarded as
necessary and sufficient criteria for anyone
to be ‘ seeing red now '). It is true that
we do not know the numerical distribution
of the physical coefficients which character­
ize the human body in this state of ‘ seeing
red ’ but we do know many physical
events which often occur either as cause
(e.g. bringing a poppy before the eyes of
the person concerned) or as effect of such
a state. (Examples of effects: certain
speech-movements ; applying a brake in
certain situations.) Hence we can first
recognize that a human body is in that
86
state and then predict what other states
of this body may be expected to occur.

If P be a physical statement: ‘The body

S is now seeing red '; P is, in the first inst­
ance, distinguished from a singular physical
statement, in describing not a single point
of space-time but an extended spatio-
temporal region, viz. the body ; it is further
distinguished by corresponding, not to a
definite numerical distribution of the
coefficients of physical states involved in
natural laws, but to a large class of such
distributions (whose composition is as
yet unknown). If a physical statement
is singular in the strict sense, no statement
of the protocol language can be deduced
from it, nor conversely. But if P is the
statement described, the protocol state­
ment p : ' Red (is being seen by S) now '
can be deduced from P and also conversely
p from P. In other words p can be trans­
lated into P, they both have the same
content. (The syntactical concept ' of
the same content ’ is defined as ‘ recipro­
cally inferable ’.)
Hence, every statement in the protocol
language of Scan be translated into a phys­
ical statement and indeed into one which
87
describes the physical state of S’s body.
In other words there is a correlation be­
tween S's protocol language and a very
special sub-language of the physical lan­
guage. This correlation is such that if
any statement from S's protocol lan­
guage is true the corresponding physical
statement holds intersubjectively and
conversely. Two languages isomorphic in
this fashion differ only by the sounds of
their sentences.
On the basis of this isomorphy we can
say the protocol language is a sub-language
of the physical language. The statement
previously made (in the material mode at
the time), that the protocol languages of
various persons are mutually exclusive,
is still true in a certain definite sense :
they are, respectively, non-overlapping
sub-sections of the physical lan­
guage. The reciprocal interdependence
of the various protocol languages which
could not be explained in terms of the
previous material account is now seen to
be a result of the rules of transformation
inside the physical language (including
the system of natural laws).
If the result thus obtained, of the
88
identity of content of P and p, be formu­
lated once again in the two modes, i.e.
“ P can be inferred “ P and p describe
from p, and con­ the same state of
versely ” affairs ”
the material formulation will again pro­
voke the old criticisms. Our previous
arguments have prepared us to take a
critical attitude towards this formulation.
But we will now consider in greater detail
the materially formulated objections, for
this is the critical point in the argument
on which our thesis is based.
Let us assume that S2 makes a report,
based on physical observations, of the
events in S /s body yesterday. Then (in
the material mode), Sx will not accept this
report as a complete account of yesterday’s
section of his life. He will say that
although the report describes his move­
ments, gestures, facial expressions, changes
in his nervous system and in other organs
it leaves out his experiences, perceptions,
thoughts, memories, etc. He will add that
these experiences must necessarily be
lacking in S2’s report since S2 cannot
discover them or at least cannot obtain
them by physical observation. Now, we
89
will assume that S2 introduces by defini­
tions, terms such as ' seeing red ’ (cf. the
example above, p. 86), into the physical
language. He can then formulate a part
of his report with the help of such expres­
sions in such a way that it runs identically
with Sj’s protocol. In spite of this Sx will
not accept this new report. He will
object that although it is true th at Sa now
uses expressions such as ‘ joy ‘ red ’,
‘ m e m o r y e t c . he means something else
them Sx does by the same words in his
protocol; the referends of the expressions
are different. For Sa, he says, they denote
physical properties of a human body, for
himself, personal experiences.
This is a typical objection whose form is
familiar to all those occupied with the
logical analysis of the statements and
concepts of Science. If we succeed in
demonstrating that some scientific term
or other reduces by virtue of its definition
to some complex of other determinations
and therefore denotes the same as the
latter, the objection is always made against
us that “ we mean something else ”. If
we show that two definite propositions
can be deduced one from the other and
therefore have the same content, or (in
90
the material form) say the same thing,
we hear again and again ‘ but we mean
something different when we use the first
and when we use the second \ We know
th at this objection rests upon a confusion
between what is expressed by a proposition
and the images we associate with the
proposition (between ‘ (logische) Gehalt ’
and ‘ Vorstellungsgehalt ’) (cf. Carnap :
Scheinprobleme in der Philosophic).
The same can be said of the present
objection. Sj connects different associa­
tions with the statements P and p respec­
tively for, on account of their linguistic
formulation, P is thought of in connection
with physical statements whereas p is
associated with the protocol. This differ­
ence in associations is however no argu­
ment against the thesis that the two
propositions have the same content (i.e.
express the same), for the content of a
proposition is constituted by the possibility
of inferring other propositions from it. If
the same statements can be inferred from
two given statements they must both have
the same content, independently of the
images and conceptions that we are
accustomed to associate with them.
We must now throw more light on the
91
question of the relation between the
protocol statement p to the corresponding
physical proposition P x where both are
about physical objects. Let us choose p
to be “ A red sphere is lying on the table
here ” and, for P x, “ A red sphere (i.e
having certain physical properties) is lying
on the table p has not the same content
as Pj, for it is possible to have an hallu­
cination of a sphere when there is none on
the table, or, conversely, the sphere can be
on the table unseen. But p has the same
content as another physical statement P a,
viz. “ S’s body is now in physical situation
Z ' '. The situation Z is specified by various
determinations including e.g. (1) The
stimulus “ What do you see ? ” is followed
by the reaction consisting of the move­
ments, etc. belonging to the sounds ‘ a red
sphere on the table ’ ; (2) If a red sphere
is laid on the table and S is put in a suitable
situation Z occurs. P x can in certain
cases be inferred from P 2 ; this necessitates
using the definition of Z and suitable
natural laws. The argument is from an
effect to an habitual cause as used both
in Physics and in everyday life. Since P a
can be inferred from p (because they have
the same content), Px can be indirectly
92
inferred from p. The usual interpretation
of the protocol statement as referring to a
certain condition of the person’s environ­
ment is therefore an indirect interpretation
compounded of the direct reference (to the
state of the body) and an appeal to
causality.
The conclusion of our discussion is that
not only the languages of the various
departments of Science but also the proto­
col languages of all persons are parts of the
physical language.
All statements whether of the protocol, or
of the scientific system consisting of a system
of hypotheses related to the protocol, can he
translated into the physical language. The
physical language is therefore a universal
language and, since no other is known, the
language of all Science.

7. U nified S cience in P hysical

L anguage.
Our view that protocols constitute the
basis of the entire scientific edifice might
be termed Methodical Positivism. Similarly,
the thesis that the physical language is the
universal language might be denoted as
Methodical Materialism. The adjective
‘ methodical ’ is intended to express the
93
fact that we are referring to a thesis
which speaks simply of the logical
possibility of certain linguistic trans­
formations and derivations and not
a t all of the ‘ reality ' or ‘ appearance ’
(the ' existence ' or the ' non-existence ')
of the ' given the ‘ mental ’ or the
‘ physical Pseudo-statements of this
kind occasionally occur in classical formu­
lations of Positivism and Materialism.
They will be eliminated directly they are
recognized as metaphysical adm ixtures;
this is in the spirit of the founders of these
movements who were the enemies of all
Metaphysics. Such admixtures can be
formulated only in the material mode and
by eliminating them we obtain Method­
ical Positivism and Methodical Materialism
in the sense defined. When the two views
are so purified they are, as we have seen,
in perfect harmony, whereas Positivism
and Materialism in their historic dress have
often been regarded as incompatibles.1
Our approach has often been termed
‘ Positivist ’ ; it might equally well be
1 Cf. Carnap, Der Logische Aufbau der Welt,
p. 245 ff.
Frank, “ Das Ivausalgesetz und seine Gren-
zen ”, Schr. z. wiss. Weltauff., Vol. VI, Vienna,
1932, p. 270 &.
94
termed ' Materialist No objection can
be made to such a title provided th at the
distinction between the older form of
Materialism and methodical Materialism—
the same theory in a purified form— is not
neglected. Nevertheless, for the sake of
clarity we would prefer the name of
‘ Physicalism ’1. For our theory is that
the physical language is the universal
language and can therefore serve as the
basic language of Science.
The physicalist thesis should not be
misunderstood to assert that the termin­
ology used by physicists can be applied
in every department of Science. It is
convenient, of course, for each department
to have a special terminology adapted to
its distinct subject matter. All our thesis
asserts is that immediately these termin­
ologies are arranged in the form of a
system of definitions they must ultimately
refer back to physical determinations.
For the sake of precision we might supple­
ment or replace ‘ physical language ’ by
the term ‘ physicalistic language ’ ; denot­
ing by the latter the universal language
which contains not only physical terms
(in the narrow sense) but also all the
1 Neurath, loc. cit.
95
various special terminologies (of Biology,
Psychology, Sociology, etc.) understood
as reduced by definitions to their basis in
physical determinations.
If we have a single language for the
whole of science the cleavage between
different departments disappears. Hence
the thesis of Physicalism leads to the
thesis of the unity of Science. Not the
physicalist language alone but any univer­
sal language would effect a unification of
Science but no such language other than
the physicalist is known. The possibility
of setting up such a language must not,
however, be excluded. Its construction
would involve the determination of its
vocabulary and of its syntax, including
rules for transformations inside the lan­
guage and for inferring protocol state­
ments. Moreover, in accordance with
our previous discussion, every proposition
P of this language in order to have any
sense must allow protocol statements to
be inferred according to stated rules. In
that case it would be possible, in view of
the inferential connection between physical
language and protocol language, to con­
struct a statement P a of the physical
language in such a way that all those
96
statements of the protocol language could
be inferred from it which could be inferred
from P. The two propositions P and P x
of the two different systematic languages
would then be so related that in every case
where P was true P x would also be true,
and conversely. Hence P could be trans­
lated into P lf and conversely.
In general,
every statement in every statement in
the new language the new language
could be translated could be interpreted
into statements of as having the same
the physical lan­ sense as a statement
guage and con­ of the physical
versely. language, i.e. every
statement of the
new language would
refer to physical
facts, to spatio-
temporal events.
Hence, every systematic language of
this kind can be translated into the
physical language and can be interpreted
as a portion of the physical language in
an altered dress.
Because the physical language is thus
the basic language of Science the whole of
Science becomes Physics. That is not to
97
be understood as if it were already certain
that the present system of physical laws
is sufficient to explain all phenomena. It
means
ev ery scientific every scientific fact
statement can be can be interpreted
interpreted, in prin­ as a physical fact,i.e.
ciple, as a physical as a quantitatively
statement, i.e. it determinable pro­
can be brought into perty of a spatio-
such a form th at it temporal position
correlates a certain (or as a complex of
numerical value (or such properties).
interval, or prob­
ability distribution
of values) of a co­
efficient of state to
a set of values of
position coordinates
(or into the form of
a complex of such
statements).
An explanation, i.e. Every scientific ex­
the deduction of a planation of fact
scientific statement, occurs by means of
consists of deducing a law, i.e. by means
it from a law of the of a formula which
same form as physi- express the fact that
98
cal laws, i.e. from a situations or events
general formula for of specified kind in
inferring singular any spatio-temporal
statements of the region are accom­
kind specified. panied by specified
events in associated
regions related in
specified fashion.
It is specifically for explaining state­
ments (or facts) by means of laws that a
unitary language is essential. It is
theoretically always possible inside the
total system of Physics
to find an explana­ to find an explana­
tion for every singu­ tion for every single
lar statement, i.e. a fact, i.e. a law in
law by means of accordance with
which this state­ which this fact is
ment (or a corres­ required (with some
ponding probability degree of probabil­
statement) can be ity) by the existence
inferred from other of other, known,
propositions based facts.
on the protocol.
For our discussion, it is of no import­
ance whether these laws take the form
of unique determinations as assumed
in classical Physics (determinism) or,
99
alternatively, as assumed in present day
Physics, determine the probability of
certain value distributions of parameters
(statistical laws of Quantum Mechanics).
In contrast to the universality of
Physics cases arise in every partial lan­
guage which can be expressed in that
language but are fundamentally incapable
of explanation in that language alone,
e.g. in Psychology e.g. a psychological
where no explana­ event such as a per­
tory law can be ception can be
formulated of a described but not
statement of the explained; for such
kind "Mr. A is now an event is con­
seeing a red circle ” ditioned not only
since the explana­ by other mental
tion must deduce events but also by
this statement from physical (physio­
statements such as logical) events.
“ A red sphere is
lying before Mr. A ”
and "Mr. A has his
eyes open ” , etc.
The prediction of an unknown is similar
to the explanation of a known truth or
event, viz. derivation with the help of laws.
Hence sub- or partial languages are not suf­
ficient for prediction and a unitary language
100
 is necessary. If our thesis that there is a
unitary language were false, the practical
application of Science to most regions
would be crippled. It is the fact that
physical language supplies the basis for
unified Science which first ensures the
thorough applicability of Science.
The thesis of the unity of Science has
nothing to say against the practical
separation of various regions for the
purposes of division of labour. It is
directed only against the usual view that
in spite of the many relations between
the various regions they themselves are
fundamentally distinct in subject m atter
and methods of investigation. In our
view these differences of the various
regions rests only upon the uses of
various definitions, i.e. of various
linguistic forms, of various abbreviations.
While
the statements and the facts and ob­
words jects
of the various branches of Science are
fundamentally the same kind. For all
branches are fa rt of the unified Science,
of Physics.

101
Ke y Te x t s
C lassic Studies in th e H istory of Ideas

Thoemmes Key Texts are facsimile paperback editions of books

which have not been available for many years.
This series gives particular emphasis to less popularized works
of major thinkers in the history of ideas.
We also feature more recent classic monographs which are
now regarded as key textbooks for the humanities.

THE UNITY OF SCIENCE

Based on the belief that statements are meaningful only if they
are empirically verifiable, in The Unity o f Science Carnap
endeavours to work out a way in which the observation
statements required for verification are not private to the
observer. The work shows the strong influence of
Wittgenstein, Russell and Frege.
This, the first English translation, was revised by Carnap for
this edition.

L O G IC , P H IL O S O P H Y O F S C IE N C E

ISBN 1 85506 391 3

ISBN 1-85506-391-3
£9.99 U K net

Published by Thoemmes Press

11 Great George Street
Bristol BS1 5RR 9 781855 06391