1 Wittgenstein, Ludwig, Philosophical Investigations (Macmillan, 1953)Google Scholar. All references by section number are to Part I of the Philosophical Investigations.

2 Wittgenstein's German is ‘wissen’ and not ‘kennen'; hence (1) is not the claim that one can only be acquainted with, or directly aware of, his own sensations. That claim may, of course, be part of the justification offered for (1).

3 I regard the consequence “So another person cannot understand the language” as one licensed by some view of the person who holds that only he can know his sensations and not by a principle Wittgenstein holds. The inability of anyone but the speaker to understand his language is supposed to be a consequence of a philosophical theory about sensations and meaning; whether one counts it as “logically necessary” or not depends upon one's view of the “logical necessity” of the supporting theories.

4 I use ‘private object’ to mean the kind of thing a sensation would be if ( 1) and (2) were true.

5 A I though this argument was suggested to me by some of Wittgenstein ‘s remarks-§ 270 and “Always get rid of the idea of a private object in this way: assume that it constantly changes, but that you do not notice the change because your memory constantly deceives you” (p. 207)-1 do not claim that it is his argument. In VII below, there is some discussion of the relationship between the argument presented in this paper and Wittgenstein 's attack on private language.

6 See VI below.

7 In general, the connections between the objects the diarist is speaking of and public phenomena are limited by the correct account of the conditions which would enable us to know what kinds of objects he is speaking of and to understand the names he assigns them. When the possibility of a private language is at stake, it does not matter whether that account is acceptable to a philosopher who thinks sensations are private objects. I think this is why Wittgenstein strips sensations of their natural expressions (see § 256).

8 By this I mean that a judgment that something is an E is always followed by a diary entry or a blush; an E unjudged or mistakenly judged is not so followed.

9 The reader will soon find that this argument may not be good enough either, but its (hopefully) successful heirs are much less perspicuous.

10 I am assuming that having the sense of similarity he in fact has─at least as far as private objects are concerned—is not an essential property of the diarist.

11 The point made here is not that the description given of the diarist's case does not entail that he does recognize a kind of private object and so it is possible that one have the same grounds and the diarist not recognize a kind of private object. The point is much stronger: if one takes the description given in the diarist's case as sufficient to warrant saying that he has recognized a kind of private object, then in some, and probably most, identically described cases, one will get a false result. So the description cannot provide sufficient, although admittedly non-entailing, grounds for the claim that the diarist recognizes a kind of private object. I have assumed, throughout, that the fact that P^O is a description of the diarist's actual performance does not warrant assigning any privileged status to the judgments of sameness and similarity made there.

12 See V below for a defense of this claim.

13 I do not think that, if we are to favour the diarist's actual case over the variations, each kind supposedly recognized need connect directly with public phenomena. The richer those connections, the more assurance that the diarist's judgments of similarity and sameness are not awry. Given a fairly rich network of connections between the putative kinds and public phenomena, the diarist would be entitled to trust his apparent recognition of other kinds for which there appeared to be no such connections. The cases in which such connections are discovered constitute grounds for thinking that his intuitions of sameness are trustworthy.

14 Here I am indebted to Carl Ginet.

15 The most common interpretation of the private language argument turns upon the claim that the diarist cannot determine whether his memory of private objects─usually those figuring in original ostensive definitions of private kind names─is defective. Even if one waives any qualms about the verification ism latent in the resulting argument, it is hard to see why the diarist cannot determine, in any reasonable sense of ‘determine', that his memory of private objects is correct. More importantly, the common version of the private language argument allows that the diarist can recognize that a private object he now has is of the same kind as one he putatively remembers figuring in an original ostensive definition. If this ability to recognize kinds is left untouched, only a wholesale use of a verification principle or scepticism about memory will sustain the claim that the diarist cannot determine that he is not misremembering the kinds his terms designate.

16 Cf. § 202 and § 380.

17 I wish to thank all those who commented on earlier versions of this paper, especially Robert Coburn, Lawrence Crocker, Carl Ginet, Sydney Shoemaker, and Stephen Thomas.