The Philosophy of Logical Atomism

The Philosophy of Logical

Atomism

Bertrand

Russell

The Philosophy of Logical Atomism

London and New York

by Fontana

First published in the Routledge Classics in 2010 by Routledge

2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN

Routledge is an imprint of the Taylor & Francis Group, an informa business

© 2010 The Bertrand Russell Peace Foundation Ltd Introduction © 1985 David F. Pears

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers.

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

ISBN 10: 0–415–47461–2 ISBN 10: 0–203–86477–8 (ebk) ISBN 13: 978–0–415–47461–0 ISBN 13: 978–0–203–86477–7 (ebk)

This edition published in the Taylor & Francis e-Library, 2009.

To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.

ISBN 0-203-86477-8 Master e-book ISBN

C ONTENTS

i n t r o d u c t i o n t o t h e 1985 edition by

d a v i d p e a r s vii

The Philosophy of Logical Atomism (1918) 1

1 Facts and Propositions 1

2 Particulars, Predicates, and Relations 15

3 Atomic and Molecular Propositions 32

4 Propositions and Facts with More than One Verb:

Beliefs, Etc. 47

5 General Propositions and Existence 61

6 Descriptions and Incomplete Symbols 77

7 The Theory of Types and Symbolism: Classes 92 8 Excursus into Metaphysics: What There Is 110

Logical Atomism (1924) 126

b i b l i o g r a p h y 151 c h r o n o l o g i c a l t a b l e s 155

i n d e x 159

I NTRODUCTION TO THE 1985 E DITION

The best way to understand a philosophical theory is nearly always to try to appreciate the force of the arguments for it.

Logical atomism is no exception. It is a theory about the funda- mental structure of reality and so it belongs to the main tradition of western metaphysics. Its central claim is that everything that we ever experience can be analyzed into logical atoms. This sounds like physics but in fact it is metaphysics.

In The Philosophy of Logical Atomism, given as a series of lectures in the winter of 1917–18 and republished in this volume,

Russell says that his reason for calling his doctrine logical atomism is because:

the atoms that I wish to arrive at as the sort of last residue in analysis are logical atoms and not physical atoms. Some of them will be what I call “particulars”—such things as little

1

They were first published in The Monist, 1918, and reprinted in Russell: Logic

and Knowledge, ed. R.C. Marsh, Allen & Unwin, London 1956.

patches of colour or sounds, momentary things—and some of them will be predicates or relations and so on. The point is that the atom I wish to arrive at is the atom of logical analysis, not the atom of physical analysis.

His negative point is clear enough: they are not physical atoms. Their positive characterization is not so obvious. What exactly are logical atoms? He tells us that they are particulars, qualities, and relations, and he is evidently relying on the fact that, when we look at reality from a logical point of view, it seems to reduce to particular things possessing certain qualities and standing in certain relations to one another. At least, that is how we usually think and speak about it, and so that is how it has to be. If reality did not in fact fall apart in the way in which we carve it up in thought and speech, everything that we think and say about it would be radically mistaken.

Here we have one of the premisses of Russell’s logical atom- ism: there must be a general correspondence between the ways in which we divide up reality in thought and speech and the ways in which it divides up in fact.

It is an appealing premiss which at first sight may seem undeniable. But what if the realism is challenged? Can it really be justified? An idealist would point out that we can never verify the correspondence between thought and reality, because we can never apprehend reality as it is in itself, unaffected by the medium of thought.

Perhaps what really happens is that we project our categorizations onto the

2

p. 3.

3

Cf. Plato: Phaedrus 265E, where an older analogy is used: philosophers should divide things at their natural joints as if they were dissecting sacri ficial victims.

4

This premiss is explicit in Lecture 8.

5

Cf. H. Putnam: Realism and Reason: Philosophical Papers Vol. III, Cambridge, 1983,

Introduction.

world. As Hume said, “The mind has a great propensity to spread itself on objects.”

Even if this premiss is conceded, it is not powerful enough on its own to support an argument for logical atomism. An atom is something indivisible or not further analyzable. A logical atom- ist, therefore, needs to show not only that the divisions traceable in logic correspond to real divisions in the nature of things, but also that the two corresponding processes of analysis do not continue indefinitely. If Russell is right, there must be a point at which words and things will be found to be not further analyzable. But why should we believe that?

In fact, he offers us more than one reason for believing it. In his later essay, Logical Atomism, published in 1924

and reissued in this volume, he says “. . . I confess it seems obvious to me (as it did to Leibniz) that what is complex must be composed of sim- ples, though the number of constituents may be infinite.”

This, of course, is an assertion rather than an argument, but it is quite a persuasive assertion. However, that may be only because it is a maxim useful for guiding our thoughts rather than a truth about the nature of things. That is how Kant thought of it.

But Russell certainly regarded it as a truth, albeit one that might never be established by the actual completion of a piece of logical analysis.

Much later, in 1959, defending his earlier ideas against J.O.

Urmson’s criticisms,

he said “As regards simples” (i.e. logical atoms) “I can see no reason either to assert or deny that they may be reached by analysis. Wittgenstein in the Tractatus and I on

6

D. Hume: A Treatise of Human Nature, Book 1, Part 3, § .

7

“Contemporary British Philosophy” ed. J.H. Muirhead, London 1924. The essay was first republished in Russell: Logic and Knowledge.

8

p. 143.

9

Critique of Pure Reason, Transcendental Dialectic, Book II, Ch. 2, §ii.

10

Philosophical Analysis: its Development between the Two World Wars, J.O. Urmson,

Oxford 1956.

occasion spoke of ‘atomic facts’ as the final residue in analysis, but it was never an essential part of the analytical philosophy which Mr. Urmson is criticizing to suppose that such facts were attainable.”

Russell went on to defend this interpretation of his earlier philosophy by quoting from the discussion at the end of Lecture 2: “. . . that is, of course, a question that might be argued—

whether when a thing is complex it is necessary that it should in analysis have constituents that are simple. I think it is perfectly possible to suppose that complex things are capable of analysis ad infinitum, and that you never reach the simple. I do not think it is true, but it is a thing that one might argue, certainly.”

Evidently there is some confusion in his recollection of his own earlier ideas. Given the general correspondence between lan- guage and reality, there were two views that he might have taken of logical atomism. One was that it had been verified by the actual completion of logical analysis. The other view was that it was self- evident, and so, even if nobody had yet verified it by completing any logical analysis, there could be no doubt of its truth. The first of these two views goes very well with the empiricism that he took over from Hume, while the second one, which he attributes to Leibniz, is more in keeping with rationalism. The view can- vassed in the discussion at the end of Lecture 2 is quite different from both. It is really the negation of logical atomism and he mentions it in the discussion only to point out that it is not his view. So he ought not to have cited it later as evidence that he had approached logical atomism in the second, rationalist way. He did in fact use that approach, but what he said in the discussion at the end of Lecture 2 cannot be taken as evidence of that fact.

It may be useful to introduce names for the three views that have just been distinguished from one another. The two main

11

My Philosophical Development, London 1959, p. 221.

12

The Philosophy of Logical Atomism, pp. 30–31.

lines of thought start from the assumption that there is a general correspondence between language and reality, which ensures that the complete analysis of words will match the complete analysis of things. One of them may be called “The Empirical Approach” and the other “The Rationalist Approach.”

The difference between them is not a difference of opinion about the nature of things, but only about the way to establish what their nature is. Both hold that reality is composed of logical atoms which are not further analyzable. A philosopher who uses the Rationalist Approach will claim that this conclusion is self- evidently true, or, at least, that it can be established by a priori reasoning. The Empirical Approach, on the other hand, leads to the claim that it is established by actual logical analysis. The idea is that, when we analyze the words in our vocabulary, we soon reach a point at which we find that we cannot analyze them any further, and so we conclude that we have reached the bottom line where unanalyzable words correspond to unanalyz- able things.

The two views may be combined without any incoherence.

They share the same conclusion, logical atomism, and they both incorporate the assumption of a general correspondence between language and reality. They differ only in their methods of establishing the conclusion. According to one view, it is estab- lished empirically, like the conjecture that every even number is the sum of two prime numbers, while the other view takes it to be provable, as it is hoped that the arithmetical conjecture will be proved one day. So Russell was not wrong when he allowed both views to be represented in his treatment of logical atomism.

However, their combination can produce a certain tension.

For suppose that we are developing the empirical argument for logical atomism. How are we to know that the words which we cannot analyze any further really are not further analyzable?

Is there not a danger that we may think we have reached the

bottom line before we have really done so? Perhaps our “logical

atoms” are not really logical atoms at all, but only a stage on the way to a complete analysis.

At this point it is possible for a divergence to open up between the two approaches, because the difference between their methods of argument may lead to a difference between their conclusions about the nature of things. In fact, this is what hap- pened to Russell and Wittgenstein. Russell used the empirical argument and claimed, in the spirit of Hume, that, when we find that we cannot push the analysis of words any further, we can plant a flag recording the discovery of genuine logical atoms.

Wittgenstein disagreed, because he did not think that the actual progress of logical analysis to date yielded such a reliable indica- tion. So he used an a priori argument instead and identified logical atoms outside the area explored by Russell.

The trouble was that, when Russell used the empirical argu- ment, the result was going to depend on whether he would make a premature identification of logical atoms, like Columbus in the Caribbean. Columbus believed that he had sailed far enough to reach India and Russell believed that he had taken the analysis of language far enough to reach logical atoms. Wittgen- stein, on the other hand, did not rely on the results of logical analysis and used an a priori argument instead. So the two philo- sophers located the object of their search in different places.

When we look at the matter in this way, it may be possible to explain how Russell became confused in his recollection of his own earlier ideas. What he says at the end of Lecture 2 is that reality may not after all be composed of logical atoms, and so, given the assumption of general correspondence between lan- guage and reality, words too may be interminably analyzable.

However, this third view, which may be called “The No Terminus

Theory” is very speculative and so, quite naturally, it produces a

pragmatic reaction. If, in fact, we soon reach a point where we, at

least, cannot push analysis any further, why not treat that as the

bottom line? Why not forget about the endless analyzability of

reality and identify logical atoms at the point at which logical analysis ends for us?

This pragmatic line leads to a considerable weakening of the assumption of general correspondence between language and reality. The correspondence would be perfectly preserved by an endless analysis of language, but the brief analysis with which we would have to content ourselves would not preserve it per- fectly. Our logical atoms would not be ultimate because there would not be any ultimate logical atoms. Incidentally, this No Terminus Theory comes very close to abandoning realism.

Logical atoms become a kind of projection of the only analysis that we find ourselves able to achieve.

It is possible that this pragmatic line of thought explains the confusion in Russell’s later recollections. For it is really only a more extreme development of the Empirical Approach. In fact, there is some evidence that Russell looked at the matter in this way in 1924. Just before the passage already quoted from Logical Atomism he writes:

When I speak of “simples”, I ought to explain that I am speak- ing of something not experienced as such, but known only inferentially as the limit of analysis. It is quite possible that, by greater logical skill, the need for assuming them could be avoided. A logical language will not lead to error if its simple symbols (i.e. those not having any parts that are symbols, or any significant structure) all stand for objects of some one type, even if these objects are not simple. The only drawback to such a language is that it is incapable of dealing with anything simpler than the objects which it represents by simple symbols.

This is cautious rationalism associated with tentative pragma- tism. The a priori argument for ultimate logical atoms is not

13

p. 143.

completely endorsed and the suggestion is almost made that there are two degrees of simplicity or atomicity. There are things that are simple for us, and there are, or may be, things that are really simple.

Be that as it may, the two main lines of thought, which must be kept in mind by anyone trying to understand Russell’s logical atomism, are the Empirical Approach and the Rationalist Approach. The former is dominant in The Philosophy of Logical Atom- ism and so it may be worked out in detail first, and the recessive Rationalist Approach, preferred by Wittgenstein, may be left for later treatment. That is the plan to be followed in the remainder of this Introduction.

When Russell applied his newly developed logic to metaphysics and the theory of knowledge, his main ambition was to rebuild empiricism on firmer foundations. Previous empiricists, from Hume to J.S. Mill, had relied on a theory of mind which stood on the shifting sands between philosophy and psychology. Rus- sell wished to replace this theory with something more robust.

His new theory would be concerned with the expression of thoughts rather than with their psychological structure, and so would make everything open to view and amenable to scientific treatment. It would be a new theory of language rigorously constructed on the framework provided by the new logic.

We are by now accustomed to this innovation. One way to ap- preciate its originality at the time is to contrast Russell’s empiri- cism with Hume’s. In his Enquiry Concerning Human Understanding Hume wrote:

14

See The Philosophy of Logical Atomism pp. 25–26. Cf. Our Knowledge of the External

World as a Field for Scienti fic Method in Philosophy, pp. 49–69. The first of these two

passages explains the relation between a logical language and ordinary lan-

guage, and the second contains the main message of a chapter entitled Logic as

the Essence of Philosophy.

Complex ideas may, perhaps, be well known by definition, which is nothing but an enumeration of those parts or simple ideas, that compose them. But when we have pushed up def- initions to the most simple ideas, and find still some ambiguity and obscurity; what resource are we then possessed of? By what invention can we throw light on these ideas, and render them altogether precise and determinate to our intellectual view? Produce the impressions or original sentiments, from which the ideas are copied.

He then went on to compare this method with the invention of “a new microscope or species of optics.”

Russell’s logical version of empiricism differs from this psy- chological version in many ways. Instead of dealing with com- plex ideas, he deals with complex symbols or words. Words do not copy things but designate them, and so designation replaces the replication, which in Hume’s system links ideas to impres- sions. But underneath these differences there are important structural similarities between the two versions of empiricism.

Hume uses definitions to resolve complex ideas into their elem- ents and Russell uses them to resolve complex symbols into theirs. When no further analysis is possible because the ideas or symbols are indefinable, Hume has recourse to impressions, the original input of the mind, and Russell has recourse to acquaint- ance, the basic relation between the mind and its objects.

Hume and Russell both use the empirical argument for atom- ism: when we analyze ideas or words, we soon find that we cannot proceed any further, because definition, the tool of analy- sis, can do no more. When this happens, we know that we have

15

Enquiry Concerning Human Understanding, §VII pt. i.

16

See D.F. Pears: Hume’s Empiricism and Modern Empiricism, in David Hume, a Sym-

posium, ed. D.F. Pears, London 1963, reprinted as Ch. 7 of D.F. Pears: Questions in

the Philosophy of Mind, London 1975.

reached the bottom line: psychological atoms in Hume’s theory, logical atoms in Russell’s. This is a clear example of the use of an empirical method in philosophy. What Russell hoped was that the method would be more perspicuous and would lead to firmer results if it were applied to language rather than to thought.

This never occurred to Hume, who claimed that his kind of examination of ideas was the “new species of optics.”

There are two distinct reasons why this method goes very well with empiricism. One is that it brings philosophy down to earth by substituting empirical investigation for a priori argument. The other is that it is natural to use this philosophical method, as Russell did, to put empiricism itself, a particular philosophical doctrine, on a sound basis. So when analysis could proceed no further, he appealed to acquaintance or direct experience. It is important not to confuse these two reasons with one another.

Empiricism is a philosophical theory about the sources and limits of human knowledge, and, though it goes very well with the use of an empirical method in philosophy, there is no necessary connection between the two.

Russellian analyses proceed by way of definitions, terminate with indefinables, and, at that point, base themselves on acquaintance. His procedure is fairly easy to follow when he is analyzing general words, but his extension of the same kind of analysis to singular words or expressions leads to notorious dif- ficulties. It is, therefore, best to start with his analysis of general words.

A hexagon is, by definition, a plane figure with six straight sides. Someone who had never seen one could learn the mean- ing of the word “hexagon” through its definition, provided that he had seen other kinds of plane figures with straight sides and could count. He would not need to be acquainted with hexagons themselves, so long as he was acquainted with the elements

17

See Russell: Our Knowledge of the External World, loc. cit.

required for the definition of the word. So far, this is close to Hume’s account of the composition of complex ideas and of the two ways in which we can acquire them: either we can build them up out of their elements, or else we can get the corres- ponding complex impressions. Russell’s account is, of course, overtly linguistic, whereas Hume’s was psychological in spite of his explicit use of definitions in the analysis of complex ideas.

But the similarity is close.

There is, however, an important new development in Russell’s treatment of the topic. The function of definitions in his theory is to allow us to learn the meanings of complex words without acquaintance with the things designated by them. But the defin- ition of a complex word may be perfectly correct without being the kind of definition that can perform this function. For example, he points out that colours can be defined by their wavelengths, but that kind of definition would not put us in a position to recognize without the use of scientific instruments colours that we had not previously seen.

So it would be quite unlike defining a hexagon as a plane figure with six straight sides. The difference is that the definition of a colour by its wave- length does not epitomize the way in which we actually apply the colour-word, but the definition of a hexagon does do that. So Russell distinguished between the two kinds of definition. Those that epitomize the application of words and put us in a position to recognize in the ordinary way things that we have not previ- ously seen are analyses and the others are not analyses.

To put the point in another way, a Russellian analysis of a word always follows a route which traces a possible way of learning its meaning. Each bifurcation marks a point at which it would be possible to stop and achieve the acquaintance that would impart the meaning of the word. If, for some reason,

18

See Enquiry Concerning Human Understanding, loc. cit.

19

See The Philosophy of Logical Atomism, pp. 20–22.

acquaintance is not available at that point, the would-be learner has to continue the analysis. Since Russellian acquaintance is direct experience, this theory of analysis is a straightforward version of empiricism. This comes out very clearly in his treat- ment of simple qualities and relations. In their case, there are no bifurcations and no further analysis is possible, and the need for acquaintance must be met immediately, because there is nowhere else to go.

Russell gives determinate shades of colour as examples of simple qualities, just as Hume had done before him.

The fact that the word “scarlet” can be defined by the wave-length of the colour is no help to the would-be learner, or at least not help of the right kind, and so he has to achieve acquaintance with the colour itself. This important doctrine may be called “The Theory of Forced Acquaintance.” If a would-be learner of the meaning of a word designating a quality is forced to achieve acquaintance with the quality itself, that indicates that it is a simple quality.

Here we have Russell’s dominant criterion of atomicity: the quality is a logical atom, because it is a logical atom for us.

Naturally, when he proposes this criterion, his suspicion, that the possibility of logical analysis might extend further than we are now able to exploit it, is recessive. It is really striking how close his dominant empirical criterion is to Hume’s criterion: a simple idea, according to Hume, is one that can only be derived from the corresponding impression.

Up to this point, though many of Russell’s doctrines are questionable, their meaning and interconnections are tolerably clear. However, his extension of the same doctrines to singular expressions and the particulars designated by them is much more difficult to understand. But before we go into those problems, it

20

D. Hume: A Treatise of Human Nature, Book I, Pt 1, §i.

21

The Philosophy of Logical Atomism, ibid.

might be useful to review the main doubts felt by other philo- sophers about the doctrines already expounded.

One doubt, mentioned earlier, is whether the general frame- work of Russell’s realism is acceptable. Perhaps it is only a projec- tion of our own habits of thought. There are also several points at which his theory of meaning has been questioned. Is the ability to recognize a quality or relation quite so large a part of knowing the meaning of the word designating it? A better alter- native might be to think of a word as a single knot in a net. Then it would be the whole net that captured reality, and the mean- ings of single words would be given by their positions in the net rather than by their separate connections with the things that they designated. Wittgenstein put this point by saying that the meaning of a word is given by its role in a language-game.

Atomism is the exact opposite of this kind of holism.

Even if Russell’s atomism were generally acceptable, there are difficulties in his account of the learning of meanings. Is it really true that the ability to recognize a determinate shade of colour can be acquired only through acquaintance with the colour itself? Hume’s remarks about a shade of blue not represented in his experience are enough to raise a reasonable doubt on this point.

Finally, Russell’s treatment of essence might be ques- tioned. As an empiricist he was preoccupied with learning the meanings of words by actual confrontation with the things designated by them and this led him to identify the essences of things with their accepted recognitional properties. But the advances of science since the Renaissance have made that identi- fication very questionable. For example, Putnam has argued that

22

See Philosophical Investigations, passim.

23

They occur at the end of A Treatise of Human Nature Book I, Pt 1, §i. Professor H.A. Prichard in his lectures on Hume at Oxford used to call Hume’s refusal to modify his theory of simple qualities in the light of these counter-examples

“brazen e ffrontery.”

the essence of a disease is its viral or bacterial causation rather than the clinical symptoms by which it is recognized, because the symptoms can vary over time without any change in the underlying identity of the disease.

When Russell extended his theory of logical analysis to singu- lar expressions, he encountered a number of further difficulties.

The apparatus of Humean empiricism had to be adjusted to fit the logic of proper names and definite descriptions like “The King of France.” The logic of these expressions depended on a lot of very subtle considerations, some of which had been explored by J.S. Mill

and G. Frege,

but there was no general agreement about the weight to be attached to them in a well- balanced theory. Proper names obviously do designate particu- lars, but there was (and still is) a lot of controversy about the way in which they do it and about the knowledge that is required for their effective use. Russell’s view was that ordinary proper names are complex expressions which can therefore be analyzed. However, he also believed that there must be unanalyz- able proper names which make the same kind of pinpoint contact with simple particulars that unanalyzable adjectives make with simple qualities.

So there are two layers of difficulties here. First, there are the basic problems about the logic of singular expressions. Second, on top of these problems which are inherent in the subject, there are special problems about Russell’s distinction between ordin- ary, analyzable proper names and unanalyzable proper names.

For the parallelism between his analysis of general expressions

24

E.g. the plague in Athens at the end of the 5th century B.C. was probably measles. See H. Putnam: Dreaming and “Depth Grammar”, especially pp. 310–315 of Philosophical Papers, Vol. II, Cambridge, 1975, where this article is reprinted. Cf.

H. Putnam: The Meaning of “Meaning”, especially pp. 245–257 of the same volume.

25

See J.S. Mill: A System of Logic, Book I Ch. 1 §5.

26

See G. Frege: Sense and Reference in Translations from the Philosophical Writings of Gottlob

Frege, Peter Geach and Max Black, Oxford 1966.

and his analysis of singular expressions is hard to establish in detail and it is not clear exactly how he saw it. Interpretation has to start from the contrast between ordinary proper names, like London, and the unanalyzable proper names which he thought that his theory required, and which he called “logically proper names.”

According to him, ordinary proper names are analyzable as definite descriptions: London is the capital of Britain, the largest port on the River Thames, etc. There are notorious difficulties about the choice of a particular definite description to serve as the analysis of an ordinary proper name, but the point to notice first about these descriptions is that they do not seem to divide the thing into its elements. In the example given they are just the ordinary kind of description that would help a tourist to identify the city, and Russell always uses examples of this kind in The Philosophy of Logical Atomism. It might be thought that this is because it was given as a set of popular lectures, but it will soon appear that there is more to it than that.

His view about logically proper names is that they cannot be analyzed in the same way as ordinary proper names or in any other way. They are, according to him, the names of particular sense-data: for example, “n” might be the name of the small patch of blue now in the centre of my visual field. At first sight, it is puzzling that he says that “n” is unanalyzable. Why should it not be analyzed by the definite description, “The small patch of blue . . . etc.”? What difference did Russell see between the logically proper name, “n”, and an ordinary proper name like

“London”? It is true that I could describe London by giving its internal structure, and perhaps it is true that I could not give that kind of description of a very small patch in my visual field. But that difference seems to be irrelevant, because Russell does not use that kind of description to analyze ordinary proper names.

27

The Philosophy of Logical Atomism, pp. 28–30.

London is the capital of Britain, Aristotle the teacher of Alexander, and so on. Why then does he insist that logically proper names are unanalyzable?

One possible justification for singling out logically proper names for special treatment would be a certain kind of essential- ism. The idea would be that some descriptions give the essential properties of things while others give their accidental properties.

The mere fact that a sense-datum satisfied a certain description would not be enough to show that its name was analyzable. For the description might attribute an accidental property to it—it might just happen to be blue—and that would not indicate that its name was analyzable. It would be analyzable only if it satisfied a description attributing an essential property to it; if no such description could be found, it would be unanalyzable. So Russell needed a way of distinguishing essential properties of things from their accidental properties.

This is exactly what one would expect after reading his account of the analysis of general words. It is an essential property of hexagons that they should have six straight sides, but only an accidental property that they should be the shape preferred by bees when they are constructing their cells. Incidentally, if he did have a way of distinguishing the essential properties of things from their accidental properties, one of the problems produced by his analysis of ordinary proper names could be solved. The name “Aristotle” would not be analysed by the definite description the teacher of Alexander, because it is only a contingent fact that Aristotle taught Alexander.

This kind of analysis would work for an ordinary proper name only when the description attributed an essential property to the person or thing named by it. If the thing possessed no essential properties, its name would be unanalyzable, a logically proper name.

28

This point is developed by S. Kripke in Naming and Necessity, Oxford 1980,

pp. 49–78.

How then did Russell draw the distinction that he needed between the essential properties of particulars and their acci- dental properties? The answer is strangely disappointing. In the case of ordinary proper names he seems to have been unaware of the need to put any restriction on the kind of definite description used in their analysis. This cannot be discounted as a feature of a popular exposition, because it is not confined to The Philosophy of Logical Atomism but recurs in all his treatments of the topic. Like Frege, he allowed any description satisfied by a complex particu- lar to serve as the analysis of its name.

It is true that essentialism is not the only way of imposing a restriction on definite descriptions acceptable as analyses. “The author of the Iliad” and “The man they are talking about at the next table”

are examples of accidental descriptions which may play a privileged role in introducing names. However, essential- ism was clearly the best basis for the kind of restriction that Russell needed, because it selects definite descriptions in a way that does not depend on the context but always remains constant for a given particular. Surprisingly, he made no use of it or of any other method for achieving a restriction in the case of ordinary proper names.

However, his theory of logically proper names does rely on a rudimentary version of essentialism. It must do, because otherwise the mere fact that sense-data possess properties would be enough to show that they cannot be simple particulars. He must be assuming that sense-data are simple because they have no essential individuating properties, or, at least, no specifiable ones. He must, therefore, have been using some criterion to

29

See S. Kripke, loc. cit., pp. 57–59 on the use of an accidental description to fix

the reference of a name. The second example illustrates what G. Evans calls “the

deferential use of a proper name”: see The Causal Theory of Names, Proceedings of

the Aristotelian Society, Supp. Vol. 1975, p. 205. But see also G. Evans: Varieties of

Reference, Oxford 1982, p. 387.

distinguish the essential from the accidental properties of particulars. What was it? It seems that he assumed that the essen- tial properties of an individual, if it has any, will be features of its structure or composition. This is an assumption that a prac- titioner of logical analysis would hardly notice that he was making. It is part of the standard picture of analysis.

This interpretation is not entirely inferential, because there is direct evidence for it in the texts. For example, at the beginning of Lecture 3 he expatiates on the self-subsistence of simple par- ticulars: “. . . each particular has its being independently of any other and does not depend upon anything else for the possibility of its existence.” He then develops a comparison between his simple particulars and “the old conception of substance.” His simple particulars possess

the quality of self-subsistence that used to belong to substance, but not the quality of persistence through time. [sc. because they are short-lived sense-data.] A particular, as a rule, is apt to last for a very short time indeed, not an instant but a very short time. In that respect particulars differ from the old substances but in their logical position they do not.

This passage, taken by itself, does not actually prove that what makes a Russellian particular simple is its lack of structure or composition. For compositeness is not the only impediment to logical independence or self-subsistence. One particular, A, might depend on another particular, α , for the logical possibil- ity of its existence, even though α was not a component of A. For example, A might be a Rembrandt and α the painter himself.

However, Russell’s conception of logical analysis made him blind to the importance of this kind of case. Analysis, as the word implies, is taking something to pieces, and so, if the essence of

30

The Philosophy of Logical Atomism, p. 32.

a thing is going to be revealed by the logical analysis of its name, it must be a feature of its structure or composition. Simplicity, therefore, is equivalent to lack of structure, or at least to lack of specifiable structure.

Russell’s view of simple qualities, like scarlet, is similar but not quite the same. It would have been exactly the same if he had held that there are no logical connections between simple qual- ities. That was the line taken by Wittgenstein, who therefore concluded that shades of colour are not simple.

Russell’s requirement for simple qualities was less demanding: there might be logical connections between them, but they must not be connections of a kind that would allow someone to learn the meaning of a colour-word without acquaintance with the colour itself.

In other words, he insisted only on the Doctrine of Forced Acquaintance. However, underneath this difference between his conception of the simplicity of particulars and his conception of the simplicity of qualities there is a striking simi- larity: simplicity is equivalent to lack of internal structure.

Russell never developed this version of essentialism in a sys- tematic way and he never applied it to ordinary proper names. It is, as it were, an important part of the drama which never quite gets on to the stage—something outside the tragedy, as Aristotle says in The Poetics. However, though this essentialism remained rudimentary in his writings, his logical atomism cannot be understood without it. If he had not relied on a distinction between essential and accidental properties, his claim, that the logically proper names of sense-data cannot be analyzed as defi- nite descriptions, would be unintelligible. If I cannot analyze

31

See Tractatus Logico-Philosophicus, 6.3751.

32

In Our Knowledge of the External World, p. 62, Russell seems inclined to deny that

there are any logical connections between atomic facts. However, the context

shows that he is thinking of pairs like “a is red” and “b is green” rather than

pairs like “a is red” and “a is green.”

“n” as “the small blue patch now in the centre of my visual field”, there must be some reason why this analysis is inadmis- sible, and the same reason ought to explain why I cannot analyze

“scarlet” as “the colour of geraniums.” The only reason that can plausibly be ascribed to Russell is structural essentialism.

The most direct evidence for this interpretation can be found in his essay, The Relation of Sense-data to Physics. It is commonly supposed that he regarded all sense-data as simple particulars.

However, in an important but neglected passage in this essay he points out that some of them are complex particulars.

He gives as an example a sense-datum consisting of a patch of blue to the left of a patch of red. This is a complex sense-datum, because it has an internal structure and two distinguishable components.

Wittgenstein used a similar example in his review of logical atomism in Philosophical Investigations

: the sentence, “The broom is in the corner” was supposed to be analyzed as “The brush is attached to the stick and they are both in the corner.” Here the complex particular is a material object rather than a sense-datum but the point made about its structure is the same.

If we turn from Russell’s theory of analysis to its application, the evidence for this interpretation becomes overwhelming. His main ambition was to analyze tables and chairs, cabbages and kings, as series of sense-data.

Any analysis of that kind would have two important peculiarities. First, unlike the example used by Wittgenstein, it would involve a shift of category. Second, as a consequence, it would be a branching analysis, moving from a single complex particular to a series of simple particulars. For sense-data are more short-lived than material objects and so the analysis of a material object will always require a very large number of them. The result will be quite unlike the analysis of

33

Mysticism and Logic, pp. 145 ff. See especially p. 147.

34

Philosophical Investigations, §60.

35

See The Philosophy of Logical Atomism, pp. 17–18 and pp. 110–125.

“Scott” as “The author of Waverley”

or the analysis of

“Bismarck” as “The ego which. . . .”

Russell pictured the analysis of a complex singular expression as a ramifying process which would terminate with the logically proper names of sense-data. The shift of category raises a num- ber of notorious problems. Can each of us really name his own sense-data and describe them without depending in any way on the external world or other people’s reactions? Wittgenstein gave a negative answer, arguing that the necessarily unteachable language required by this theory could not even be set up by a single person for his own private use.

Even if he could set it up, communication with anyone else would be impossible, because nobody could achieve the acquaintance with anyone else’s sense-data that he would need to understand their statements about them. These are problems on the borderline between semantics and the theory of knowledge. There are also some difficult metaphysical questions about sense-data. Are they really self-subsistent? And what account can be given of sense-data that are not actual but only possible?

However, if we look at Russell’s theory from a logical point of view, the most important question that has not yet been answered is about the link between a logically proper name and the simple particular designated by it. What is the nature of this link and how is it established?

Part of the answer is straightforward. Someone is confronted by a simple particular—in Russell’s theory, he is acquainted with a simple sense-datum—and he gives it a logically proper name, “n.” If he used “n” again later to refer to the sense-datum—

36

Loc. cit. pp. 77–82.

37

This is a rare example of an analysis of a complex particular which is non- branching but involves a shift of category. See Russell: Knowledge by Acquaintance and Knowledge by Description, in Mysticism and Logic, p. 218.

38

Philosophical Investigations, passim.

in Russell’s theory, he would have to be relying on experience- memory and perhaps even on short-term experience- memory

—he would mean that very same sense-datum. To put the point in Kripke’s way, “n” would be the rigid designator of that sense-datum, because in virtue of its meaning it would designate the same sense-datum in every possible world in which it existed.

Naturally, this does not imply than n exists necessar- ily, because it would be possible to plant the name on n in the actual world and then to speculate that in another possible world n might not have existed.

But why did Russell not extend the same treatment to ordin- ary proper names? Kripke’s theory is that ordinary proper names are rigid designators: an infant was christened “Richard Nixon”

and thereafter that name rigidly designates that person. Why did Russell not anticipate this theory of ordinary proper names?

In fact, he does half anticipate it. For he sometimes says that ordinary proper names can be used as names rather than as covert definite descriptions, and, when he says this, he means that they can be used as logically proper names, which are, in fact, the rigid designators of simple particulars. There is a clear example in Lecture 6:

I should like to make clear what I was saying just now, that if you substitute another name in place of “Scott” which is also a name of the same individual, say, “Scott is Sir Walter”, then

“Scott” and “Sir Walter” are being used as names and not as descriptions, your proposition is strictly a tautology.

If one

39

See Russell: On the Experience of Time, Monist 1915 and The Philosophy of Logical Atomism, p. 32, which imply that acquaintance with particulars does not reach beyond the limit of the specious present.

40

S. Kripke, Naming and Necessity, pp. 47–49.

41

The meaning of this sentence is unmistakable, but it would have been more accurately expressed if Russell had inserted the word “and” twice, once before

“say” and again before “your.”

asserts “Scott is Sir Walter”, the way one would mean it would be that one was using the names as descriptions. One would mean that the person called “Scott” is the person called “Sir Walter”, and “the person called ‘Scott’ ” is a description, and so is “the person called ‘Sir Walter’ ”. So that would not be a tautology. It would mean that the person called “Scott” is iden- tical with the person called “Sir Walter”. But if you are using both as names, the matter is quite different. . . . if I say “Scott is Sir Walter”, using these two names as names, neither “Scott”

nor “Sir Walter” occurs in what I am asserting, but only the person who has these names, and thus what I am asserting is a pure tautology.

This remarkable passage makes Kripke’s point, that, when two rigid designators flank the identity-sign, the resulting statement is necessarily true, if it is true at all.

For in such a case the identity will hold in virtue of the meanings of the two rigid designators.

It might be thought that Russell’s concession, that ordinary proper names may be used as logically proper names, is merely a feature of a series of popular lectures. In such a context it is natural for him to use material objects to introduce points that are really about sense-data, and so he warns his audience that in the sentence “This is white” the word “This” functions as a logically proper name only when it is used “quite strictly, to stand for an actual object of sense” rather than “for this piece of chalk as a physical object.”

However, there is more to it than that. For even if we chose our examples exclusively from sense- data and substituted two names of a complex sense-datum for the two names of the man, “Scott” and “Sir Walter,” his point

42

The Philosophy of Logical Atomism, pp. 82–83.

43

See Naming and Necessity, pp. 99–105 and pp. 143–144.

44

The Philosophy of Logical Atomism, p. 29.

about the use of ordinary proper names as logically proper names would still have a valid application: if two names of the same complex sense-datum were used as logically proper names, the resulting identity-statement would be “a pure tautology.”

This is an important point, because it shows that Russell’s logically proper names combine two different features which could easily come apart. A sense-datum does not have a specifiable individual essence and so its logically proper name is unanalyz- able. That is the first feature of Russellian logically proper names.

Their second feature, which follows from it given the Doctrine of Forced Acquaintance, is that they must be directly attached to the sense-data that they designate without the mediation of any essential description. Now Russell never drew the clear line that he needed between essential and accidental descriptions. He therefore regarded direct attachment as attachment unmediated by any description whatsoever. So according to him, when I use the logically proper name, “n,” it will designate this particular sense-datum, whatever descriptions it satisfies, and what I say about it will be said de re.

However, these two features might well come apart. For it would be possible to attach a name directly to a complex sense- datum when one was acquainted with it. It would not matter that this sense-datum satisfied both accidental and essential descriptions. For the attachment would not be mediated by them, just as it was not mediated by the accidental descriptions of the simple sense-datum in the other case. Here too the name

45

When Russell explains how the ego is known by description, it is important for him to be able to start from de re knowledge of sense-data. See On the Nature of Acquaintance in Logic and Knowledge, p. 168: “The subject attending to ‘this’ is called

‘g’. . . . ‘This’ is the point from which the whole process starts, and ‘this’ itself

is not de fined but simply given. The confusions and difficulties arise from

regarding ‘this’ as de fined by the fact of being given, rather than simply as given.” His

point is that if it were de fined as given, it would be defined as given to the ego,

which would be circular.

would be used to designate that particular sense-datum in de re statements. Or, to revert to the kind of example that Russell employed to make this point, an ordinary proper name might be directly attached to a person with exactly the same sequel.

It is worth looking more closely at what he says about the second feature of his logically proper names: “We are not acquainted with Socrates, and therefore cannot name him. When we use the word ‘Socrates’, we are really using a description. Our thought may be rendered by some such phrase as ‘The Master of Plato’, . . . but we certainly do not use the name as a name in the proper sense of the word.”

His reason for saying this is not that Socrates belongs to the external world, but rather that he died long ago. It is, therefore, a little confusing to find him allowing that “Scott” and “Sir Walter” may be used as names “in the proper sense of the word.” He should have made the point about his contemporaries.

The passage continues: “That makes it very difficult to get any instance of a name at all in the proper strict logical sense of the word. The only words one does use as names in the logical sense are words like ‘this’ or ‘that’. One can use ‘this’ as a name to stand for a particular with which one is acquainted at the moment. We say ‘This is white’. If you agree that ‘This is white,’ meaning the

‘this’ that you see, you are using ‘this’ as a proper name.”

He then issues the warning, already quoted, about the differ- ence between the piece of chalk and a sense-datum. Finally, he points out that “ ‘this’ has a very odd property for a proper name, namely that it seldom means the same thing two moments run- ning, and does not mean the same thing to the speaker and the hearer. It is an ambiguous proper name, but it is really a proper name all the same, and it is almost the only thing I can think of that is used properly and logically in the sense that I was talking of for a proper name.”

46

The Philosophy of Logical Atomism, p. 29.

This needs some explanation. Two features of logically proper names have been distinguished, their unanalyzability and their direct attachment to things. Russell, as has been explained, allowed that ordinary proper names sometimes possess the second of these two features, but, of course, never the first. Now direct attachment is really attachment unmediated by any essen- tial description. But because Russell never explicitly distinguished between essential and accidental descriptions, he regarded direct attachment as attachment unmediated by any description what- soever. So a directly attached name will designate the same thing in every possible world in which that thing exists, whatever descriptions it satisfies.

In the passage just quoted he says that the only words in ordinary language that function like logically proper names are demonstratives. He says this because he is thinking of the other feature of logically proper names, unanalyzability. Obviously, ordinary proper names do not designate simple particulars and so they fail the test of unanalyzability. If he had waived that requirement, he could have allowed that ordinary proper names fulfil the role of logically proper names to perfection. There is no mystery about this. They would fulfil the role of logically proper names to perfection because that role would only be the direct attachment to things that is needed for de re statements.

However, he does not waive the requirement of unanalyz-

ability. He, therefore, has to find in ordinary language singular

expressions directly attached to simple sense-data. Now sense-

data never recur, and so, of course, we do not feel any need to

name them, and our nearest approach to naming them is picking

them out by demonstratives. This is fairly close because a

demonstrative really is directly attached to its sense-datum and it

really does designate the same sense-datum in every possible

world in which that sense-datum exists. However, unlike a name,

it is not a rigid designator, because it does not perform this

function solely in virtue of its meaning, but partly in virtue of

the varying contexts of its use. So, judging it by the standard of a name, which it is not, Russell calls it “ambiguous,” which it is not.

The important thing is to understand how he gets himself into this position. He does so by asking too much of logically proper names. If he had insisted only on direct attachment to things, ordinary proper names would have passed muster too, and then the distinction between “ordinary” and “logically proper” would have collapsed. But he requires something more of logically proper names, unanalyzability. However, at the same time, he shows his feeling for the great semantic importance of direct attachment by allowing that ordinary proper names may be used as logically proper names. The trouble is that once he has added the second requirement, of unanalyzability, his failure to find any perfect examples of logically proper names in ordinary language is inevitable. We can use demonstratives directly attached to our sense-data, and, when we do so, they designate the same sense-data in every possible world in which those sense-data exist, but they do not perform this function solely in virtue of their meanings. So he is wrong to take the variations in what they designate as a sign of ambiguity.

Russell’s rehabilitation of empiricism relied on very close cooperation between his semantics and his theory of knowledge.

It is, therefore, necessary to go a little more deeply into the theory of acquaintance that supported his theory of logically proper names. One important question about Russellian acquaintance has already been raised. How far does a person’s acquaintance extend back into his own past experience? To put the question in another way, does it reach back any further than the limit of short-term memory of the specious present? The point of the question is that knowledge by description takes over where acquaintance peters out.

The answer is that Russell always allows acquaintance with

qualities and relations experienced in the more remote past, but

he is less permissive about acquaintance with particulars. For he was inclined to think that particulars move out of range of acquaintance when they move out of the specious present. It is true that he did not reach this conclusion in his first discussion of this topic,

but it is firmly established in his mind by 1917.

It is an important conclusion, because, if acquaintance pro- vides indispensable support for logically proper names, then any restriction imposed on the range of acquaintance will auto- matically restrict the situations in which logically proper names can be used and understood. Now this restriction will be imposed on logically proper names entirely because of the exi- gencies of their direct application to things. It would, therefore, remain in force even if logically proper names were not unana- lyzable. Even if their direct application to things were not unavoidable, as it is in Russell’s theory, but only one of two options—the other option being analysis into elements with which their users were acquainted—the restriction would still remain in force.

The reason for this is that Russell held that the direct applica- tion of a name to a complex particular also requires the support of acquaintance. To put the point in his way, acquaintance is needed when an ordinary proper name is used as a logically proper name. So if he had lived to read Kripke’s treatment of ordinary proper names as rigid designators, his first reaction would certainly have been to insist on the support provided by acquaintance. But, given the restriction imposed by him on the range of acquaintance, this would have produced the following result: when someone applies an ordinary proper name to a particular with which he has had acquaintance in the past, it can be used as a logically proper name or rigid designator only if his acquaintance with it belongs to the specious present.

47

See The Problems of Philosophy, pp. 114–118.

48

See The Philosophy of Logical Atomism, p. 32.

There are, however, objections to imposing this restriction on what Russell calls “the use of names as names,” and one of the aims of recent work on proper names has been to remove it and to find alternative ways of supporting their use as rigid designa- tors. A student of Russell’s philosophy might well begin to develop the objections by criticizing his account of memory. It is implausible to suppose that, if I talk to someone called “John” at a party, then, as the moment of the conversation recedes into the past, I am forced to use his name through the mediation of a definite description instead of using it as a name. What is needed here is some alternative kind of support for the continued use of his name as a name. A plausible candidate would be causation, specifically the kind of causation on which experience-memory depends.

When we look at the history of the theory of proper names from this point of view, we can see recent causal theories as natural developments of an obvious criticism of the restriction of “the use of a name as a name” which Russell based on his restriction of the range of acquaintance-memory. “The use of a name as a name” must reach outside the field of the user’s immediate awareness, beyond the limit of his short-term mem- ory and across the boundaries separating his consciousness from other people’s. Kripke deals with the last of these three extensions of range by suggesting that proper names—in his view, rigid designators—may be supported by a causal line running back through the oral tradition to the original christening

and Evans argues for another version of the causal theory.

There is a second question worth asking about Russell’s

49

See C.B. Martin and M. Deutscher: The Causal Theory of Memory, Philosophical Review, 1966.

50

Naming and Necessity, pp. 91–97.

51

The Causal Theory of Names, Proceedings of the Aristotelian Society

Supplement, 1975.

theory of acquaintance. Did he really regard it as a kind of pin- point contact with reality? If so, acquaintance would not involve any selection or interpretation. A person could be acquainted with a sense-datum without being acquainted with it as an instance of a specific type of thing, and he could be acquainted with one of its qualities without being acquainted with it as a determinate in a certain determinable range.

His definition of acquaintance in On the Nature of Acquaintance

and his informal treatment of it in The Problems of Philosophy,

both make it quite clear that this was his idea. Acquaintance is a kind of knowledge and yet it is a purely extensional relation between subject and object. This is confirmed by his review of his theory of acquaintance in 1959, long after he had given it up. He explains that he no longer holds that sensations automatically give us knowledge when we have them, as the term “sense- datum” implied: “ ‘Perception’ as opposed to ‘sensation’

involves habit based upon past experience. We may distinguish sensation as that part of our total experience which is due to the stimulus alone, independently of past history. This is a theor- etical core in the total occurrence. The total occurrence is always an interpretation in which the sensational core has accretions embodying habits.”

His criticism of his own earlier position is a just one. It is a criticism often made by Wittgenstein in his later writings, where it serves as the first step in his argument against the possibility of a private language. Certainly, it is very paradoxical to treat acquaintance as pinpoint contact. The problems inherent in this view can easily be seen in the case of qualities. When someone learns the meaning of a colour-word by being shown an object of that colour, he has to pick out the hue and ignore all the

52

Logic and Knowledge, p. 162.

53

Ch. V.

54

My Philosophical Development, p. 143.

object’s other features. Also, when he does pick it out, what really counts is not his momentary relationship with it, but its long-term effect: he should be able to recognize it next time he sees it. If we call this “acquaintance” with the quality, then acquaintance will not be mere pinpoint contact. It will involve selection at the time and a recognitional capacity later.

The same is true of acquaintance with particulars. You do not claim acquaintance with a person simply on the ground that he has passed through your field of vision without your noticing him and without your acquiring the ability to recognize him next time. In Russell’s theory this point is not so obvious as the parallel point about acquaintance with qualities and relations.

For his simple particulars are sense-data and so, unlike people and comets, they do not recur and there is no possibility of re-identifying them. However, that does not altogether cancel the usual requirements for acquaintance. You must still pick out your sense-datum, say as a particular member of the class of visual sense-data, and, even if you are never going to have an opportunity to re-identify it, you must know at the time which one it is and later which one it was. So this kind of acquaintance too is more than pinpoint contact. It has to plant a name in the world.

The marriage arranged by Russell between logic and empiricism gives The Philosophy of Logical Atomism its special character. Most of his emphasis is on the empirical argument for logical atomism:

if we think of a word and begin to analyze it by substituting definitions, we soon reach a point at which the supply of def- initions runs out and the analysis terminates. The empirical argument claims that this shows that we have reached logical atoms.

We may ignore the No Terminus Theory which makes a brief

appearance at the end of Lecture 2, because it is not any kind of

version of logical atomism. For though it employs the same

analytical method, it maintains that it never reaches any bottom line on which logical atoms might be identified. The only importance of this theory in the development of Russell’s phil- osophy is that it would imply that our powers of analysis are necessarily deficient.

This implication becomes more interesting when it is modified—our powers of analysis have not yet been developed as far as they might be—and when it is combined with the Rationalist Approach, which treats logical atomism as self- evident or provable a priori. This combination, as was pointed out earlier, does not necessarily lead to a rejection of the assumption of a general correspondence between language and reality. For a full analysis of language would still be as fine-grained as reality itself. However, it might lead to a certain weakening of the assumption, because it might seem to open up the possibility of distinguishing between absolute atomicity and atomicity relativized to our present analytical achievements.

Wittgenstein was opposed to any such dilution of pure atom- icity, even if it was only part of a compromise allowing for two degrees of the property. So, if we want to see the Rationalist Approach disentangled from Russell’s empiricism and followed to its true end, we should look at the early philosophy of Wittgenstein.

Wittgenstein did not take logical atomism to be self-evident

but tried to deduce it by an a priori argument from his theory of

language. He believed, like Russell, that an ordinary descriptive

sentence depends for its meaning on the fact that at some point

in its analysis there are words actually standing for things. Now

on the No Terminus Theory all things would be complex. There-

fore, at the point where designation occurred, it would have to

be designation of complex things. But from this it follows that at

that point further sentences would have to be true. For it would

have to be true that the elements of those complex things were

put together in a way that was required for their existence.

So, to pick up the beginning of the argument again, if the analysis of ordinary descriptive sentences never terminated on logical atoms, their meaning would always depend on further factual truths. But that would not be acceptable, because it would make meaning indeterminate. The meaning of any descriptive sentence would depend on a truth, whose meaning would depend on a further truth, and so on ad infinitum.

Russell probably became aware of this argument only when he read the Tractatus just after the First World War. It is, therefore, not surprising that the Rationalist Approach immediately became more prominent in his writings after having remained recessive in The Philosophy of Logical Atomism. In the lectures the spirit of Wittgenstein’s argument does sometimes almost suc- ceed in speaking through Russell’s sentences, but the message is always transformed, because the point becomes epistemic rather than semantic.

There is a good example at the beginning of Lecture 3: “There is, as you know a logical theory . . . according to which, if you really understood any one thing, you would understand every- thing. I think that rests upon a certain confusion of ideas. When you have acquaintance with a particular, you understand the particular itself quite fully, independently of the fact that there are a great many propositions about it that you do not know, but propositions about the particular are not necessary to be known in order that you should know what the particular itself is. It is rather the other way round. In order to understand a prop- osition in which a name of a particular occurs, you must already be acquainted with that particular. The acquaintance with the

55

See Tractatus 2.0211.2.022 and 3.23–3.24. Cf. Notes Dictated to G.E. Moore in

Norway in Notebooks 1914–16, p. 116: “The question whether a proposition has

sense (Sinn) can never depend on the truth of another proposition about a

constituent of the first,” cf. Wittgenstein’s 7th and 8th entries on 18 June

1915, Notebooks 1914–16, p. 64.