The last chapter of the book, entitled Sense-data and Philosophical Analysis, is devoted to an examination of Phenomenalism. It begins with an elaborate critical account of the pronouncements of Moore, Miss Stebbing, Wisdom, Ayer and Duncan-Jones on the nature of Philosophical Analysis. It will suffice for me to mention one point which Marc-Wogau makes on the way and the conclusion which he eventually reaches.

(A) Philosophical Analysis in General.

      He thinks that he can detect in several of the writers whom he discusses a tendency to hover uncertainly between the two following views of the nature of what Wisdom calls 'new-level analysis'.

1. That we start with a fact F₁ (or with a sentence S₁ which expresses it), and then show that it is in a certain sense 'equivalent to' another fact F₂ (or to a sentence S₂ which expresses F₂).
2. That we start with a sentence S₁ which is a less clear and adequate expression of a fact F₁, and then substitute for it another sentence S₂ which is a clearer and more adequate expression of the same fact.

      He finds it difficult to reconcile the first alternative with the contention that the result of such an analysis is to give the meaning of the original sentence. Suppose, e.g., that S₁ is a sentence containing a nation-name, e.g., 'England'. Suppose that S₂ is a sentence which contains personal names, e.g., 'Churchill', 'Attlee', 'Bevin', etc., but no nation-name. If there is both a nation-fact and an equivalent person-fact, surely the former is the meaning of S₁ and not of S₂ and the latter is the meaning of S₂ and not of S₁.

      Suppose, on the other hand, that we accept the second alternative. Then it must be admitted that nation-names are 'superfluous', in a quite definite sense, in comparison with personal names. For every fact which is expressed by a sentence which contains a nation-name can be expressed more accurately by one which contains no nation-name and does contain personal names; whilst there are facts which can be expressed by sentences which contain personal names and cannot be expressed by sentences which contain nation-names.

      In that case, Marc-Wogau holds, Philosophical Analysis is not something radically different from the old notion of analysis as the definition of certain fundamental ideas. It merely makes that notion more precise in certain respects. In particular we find that 'definition' must be extended to cover what logicians call 'definition-in-use'. This is essential when we are concerned with terms which are logical constructions out of other terms The three kinds of analysis which Wisdom distinguishes under the names of 'same-level material', 'same-level formal', and 'new-level' analysis can all, under certain circumstances, be part of the subject-matter of philosophy; and they must all be regarded as instances of what the earlier philosophers called 'analysis of ideas'.

(B) Equivalence, Identity of Content, and Sameness of Meaning.

      It is contended by Phenomenalists that every sentence which contains a physical-object name, such as 'penny', can in some sense be 'translated into' a conjunction of sentences which do not contain any such name but do contain names or descriptions of sensations. This immediately raises the question: 'What is the relation between two sentences S₁ and S₂ which Phenomenalists have in mind when they say that the former can be translated into the latter?'

      Marc-Wogau distinguishes three symmetrical relations which may hold between two sentences, viz., equivalence, identity of content, and sameness of meaning.

      Two sentences have the same meaning if and only if 'each refers in the same way to the same aspect of the same fact'. Thus, e.g., the two sentences 'Tom and Dick are brothers' and 'Tom and Dick are male children of the same parents' have the same meaning; but the two sentences 'This is an equilateral triangle' and 'This is an equiangular triangle' do not.

      Two sentences are equivalent if they are either both true or both false. An example of two sentences which stand in this relation would be 'Cambridge is bigger than London' and 'New York is the capital of Sweden'.

      The notion of identity of content is ascribed by Marc-Wogau to Carnap. I find Carnap's account of the matter, as reported by Marc-Wogau, somewhat obscure; largely because I cannot make out whether it is a matter of entailment only or of both entailment and material implication. As the whole notion seems to me quite ridiculous on the second alternative, I shall assume that only entailment is involved. I shall also assume that the notion is primarily concerned with singular propositions. Both these assumptions may be wrong; but, if we make them, the notion of identity of content may be explained as follows.

      Consider any singular proposition p. We might define the 'formal content' of this as the class of all those propositions which are entailed either by p alone or by the conjunction of p with any one or more necessary propositions. The following would be an example. The proposition that the square of 1.414 is not equal to the ratio of 2 to 1 is part of the formal content of the proposition that 1.414 is a rational fraction. We might define the 'material content' of p as the class of all those propositions which are entailed by the conjunction of p with any one or more true contingent universal propositions. Thus, e.g., the proposition that Mr. Churchill is mortal is part of the material content of the proposition that Mr. Churchill is human, but it is not part of its formal content. We might define the 'total content' of p as the logical sum of its formal and its material content. I should suppose that two propositions which had the same formal content would necessarily have the same material content, and therefore the same total content. But one sees no reason why two propositions which had the same material content should not have a different formal content.

      If p and q mutually entail each other, it follows that they have the same formal content. If p and q have the same formal content, and if we use 'entail' in such a sense that every proposition entails itself, it follows that p and q mutually entail each other. (So far as I can see the second consequence would not follow from the hypothesis that p and q have the same formal content without the proviso which I have added.)

      Two sentences S₁ and S₂ would have the same formal content if the proposition which S₁ means and the proposition which S₂ means have the same formal content. Obviously this will be fulfilled if both sentences have the same meaning. But Mara-Wogau points out, quite rightly, that it may be fulfilled if they have different meanings. The following would be an example. The two sentences 'This is an equilateral triangle' and 'This is an equiangular triangle' quite obviously have different meanings; they 'refer either to different facts or to different aspects of the same fact'. But, assuming that 'this' denotes the same particular in both cases, the formal content of what is meant by the first is the same as the formal content of what is meant by the second.

      Now, so far as I can understand, Marc-Wogau maintains the following propositions about the conditions under which one sentence is 'translatable into' another in the sense required by Phenemenalists.

1. The two sentences must be identical in formal content, but need not have the same meaning.
2. If a sentence S₂ is, in this sense, a translation of two different sentences S₁ and S₁', then those two sentences must have the same meaning.
3. It follows from (ii) that identity of formal content between two sentences is not sufficient to ensure that one shall be a translation of the other in the sense required. For suppose that two sentences S₁ and S₁' have the same content but different meanings. Then any sentence S₂ which had the same content with either of them would have the same content with both of them. Therefore, if identity of content were a sufficient condition for one sentence to be a translation in the sense required, S₂ would be a translation of both S₁ and S₁'. But this is impossible, in accordance with Proposition (ii), since S₁ and S₁' by hypothesis have different meanings.

      Marc-Wogau does not profess to say what more than identity of content and less than sameness of meaning is needed to make one sentence a translation of another in the sense required. That he is right in thinking that identity of content is insufficient seems to me to be plain from the following consideration. Identity of content is a symmetrical relation. But, when Phenomenalists say that a sentence containing a physical-object name can be 'translated into' a conjunction of sentences which contain no such name but only names or descriptions of sensations, they are thinking of a relation which is asymmetrical. For they would not say that the former is a translation of the latter in the sense in which the latter is a translation of the former.

(C) Is Phenomenalism possible?

      Marc-Wogau uses the term 'M-sentence' to denote any sentence which contains the name of a physical object or class of such objects, e.g., 'There is a table in my room now'. He uses the term 'S-implication' to denote a sentence of the following form. 'If certain conditions (ultimately expressible entirely in terms of sensation) had been fulfilled in the past, certain sensations would have followed immediately and therefore would have occurred in the past or would be occurring now. And, if these conditions were fulfilled now or should be fulfilled in future, certain sensations would follow immediately and therefore would occur in the future'. (Marc-Wogau does not state the meaning of 'S-implication' nearly so elaborately; but I think that at least this amount of elaboration, and probably a great deal more, is needed.) He takes Phenomenalism to be the doctrine that every M-sentence can be 'translated into' a conjunction of S-sentences.

      Since identity of content is at any rate a necessary condition for one sentence to be translatable into another, any doubt as to whether an M-sentence is identical in content with any conjunction of S-sentences is ipso facto a doubt about the possibility of Phenomenalism.

      Marc-Wogau embarks on an elaborate critical discussion of the arguments and counter-arguments used by a number of philosophers who have debated this question. The most important of these objections fall under the following heads.

1. That the attempted translation is circular, because the S-sentences always contain or imply references to observer's bodies, i.e., to certain physical objects, which cannot be completely resolved into statements about actual and possible sensations.
2. That the attempted translation involves a vicious infinite regress; because the identity of content holds only on the supposition that the sensations are non-hallucinatory. This condition can be resolved at any stage into statements about the sensations which certain other observers would have or would have had; but these sensations must in turn be supposed to be non-hallucinatory.
3. That the phenomenalist analysis is incapable of dealing with the dispositional properties in general, and the causal properties in particular, which are an essential part of the notion of a physical object.
4. That, if the 'implication' in the S-implications be taken to be material implication, all sorts of absurd consequences follow because of the fact that a false proposition materially implies every proposition. On the other hand, the 'implication' cannot be entailment; and it is very difficult to give any satisfactory account of an intermediate meaning that it might have. (It has always seemed to me that persons who claim that Phenomenalism is completely 'empirical', with no a priori nonsense about it, should address themselves to giving an empiristic account of the notion of the consequences of unfulfilled conditions, which plays so important a part in Phenomenalistic translations of M-sentences.)

      As Marc-Wogau merely summarises and criticizes the arguments and counter-arguments of others, I shall not go into further detail about this part of his work. I think that his main conclusions may fairly be stated as follows.

1. Even if every M-sentence can be translated into a conjunction of S-implications, the two do not have the same meaning. They refer to different facts or to different aspects of the same fact.
2. Some of the arguments put forward to show that no M-sentence is even identical in content with any conjunction of S-implications are invalid, and some rest on presuppositions about causation which Phenomenalists cannot be expected to accept.
3. It is doubtful, however, whether any Phenomenalist has in fact produced a translation of M-sentences which gets rid of tacit or explicit reference to the bodies of observers; and, until someone has done this it remains doubtful whether it can be done.
4. No satisfactory answer has been made to the difficulties based upon the condition that the sensations referred to in the S-implications must be non-hallucinatory, and none has been made to the difficulties based on the paradoxes of material implication. So Marc-Wogau is very doubtful whether Phenomenalism is even possible.

(D) The relative Certainty of M-propositions and S-propositions.

      The last question which Marc-Wogau discusses is this. Let us suppose, for the sake of argument, that for every M-sentence there is a conjunction of S-sentences with the same content as it. Why is it thought to be specially desirable to make such a translation? One reason that is often given is that the S-propositions which enter into the phenomenalistic translation of an M-proposition are always more certain than the latter. This doctrine sometimes takes the special form that S-propositions are indubitable, whilst it is always possible to doubt any M-proposition; but it might be held by a person who did not admit that S-propositions are indubitable. So Marc-Wogau discusses the two questions

1. Are S-propositions completely certain?
2. Even if they are not, are the S-propositions which enter into the phenomenalist translation of an M-proposition always more certain than the latter?

      I think that the results of a long and complicated discussion may fairly be summarised as follows.

1. If any kind of S-proposition could be claimed to be indubitable, it could only be a singular categorical S-proposition, viz., one such as would be expressed by the sentence 'I am having such and such a sensation now'. But it is certain that no M-proposition can be translated into a conjunction of such S-propositions. If the translation can be made at all, the S-propositions involved are conditional and general, i.e., they are about the sort of sensations which any person would have had or would be having or would have if certain conditions had been or were now or should in future be fulfilled.
2. Marc-Wogau is not convinced that even singular categorical S-propositions are indubitable. But, even if they were, this would be irrelevant to the question at issue, for the reason just given:
3. He admits that any singular categorical S-proposition is more certain than any M-proposition. But it does not follow, and he does not admit, that every conditional and general S-proposition which enters into the phenomenalistic analysis of an M-proposition is more certain than the latter.

      I hope that I have now managed to give a fair and tolerably clear and accurate critical account of the main points in Marc-Wogau's book. In spite of the great length of this paper, there remain many interesting matters which I have left untouched. Whether or not one accepts Marc-Wogau's own arguments and conclusions or his criticisms on those of other philosophers, there can be no doubt that he has written a very valuable book. I hope that it will be read by many philosophers in England and the U.S.A. It is particularly useful for us to have an opportunity to see ourselves and our theories through the eyes of a very learned, intelligent and sympathetic stranger, brought up in a philosophic tradition which is different from, but not hopelessly alien to, our own.

      I will end by mentioning the few misprints which I have noticed in addition to those recorded on the last page of the book. They are as follows. Page 74 n., for Johnsson read Johnson. Page 154 n., for acht read Acht. Page 261, 1. 11, for Lichreflex read Lichtreflex. Page 320,1. 8, for Späre read Sphäre. Page 403 n., 1. 3, for Intuitiv read Intuitive. Page 431, 1. 11, for bezweifel read bezweifelt.