1 See Russell, Bertrand Logic and Knowledge, Marsh, R.C. ed. (London: Allen & Unwin 1956), 175-282, 321-44Google Scholar.

2 All references in this style are to Bradley’s The Nature of all Being.

3 Citations (like this one) in the form of a number with a decimal point give Tractatus passage numbers from Wittgenstein, L. Tractatus Logico-philosophicus, trans. Pears, D.F. and McGuinness, B.F. (London: Routledge & Kegan Paul 1961)Google Scholar.

4 See Klemke, E.D. ed., Essays on Frege (Urbana: University of illinois Press 1968), 507-36Google Scholar. For objections to this sort of interpretation, and for what I think are better accounts of the Tractarian Satz, see Shwayder, David ‘Review of Stenius,’ Mind 72 (1963) 275-88CrossRefGoogle Scholar; Shwayder, ‘Gegensttinde and Other Matters,’ Inquiry 7 (1964) 387-413; and Schwyzer, H.R.G. ‘Wittgenstein’s Picture-Theory of Language,’ in Copi, Irving and Beard, Robert eds., Essays on Wittgenstein’s Tractatus (London: Routledge & Kegan Paul 1962) 271-88Google Scholar.

5 Russell, Bertrand ‘Knowledge by Acquaintance and Knowledge by Description,’ in Mysticism and Logic (New York: Barnes & Noble 1970), 159Google Scholar

6 Where I speak of obtainings and non-obtainings, Bradley speaks of states of affairs as ‘existing’ or ‘failing to exist.’

7 Few commentators would disagree with this. A recent exception is Hintikka, Merrill and Hintikka, Jaakko Investigating Wittgenstein (Oxford: Basil Blackwell 1987), 55Google Scholar. Bradley’s arguments against them (75ff.) are convincing and illuminating.

8 I have not written these in symbols or attempted to specify whether pis a proposition or a putative fact. This is because— as I will suggest presently— I don’t think the Tractatus allows for any adequate formulation of the axioms.

9 Bradley offers another argument to show why Wittgenstein should accept these axioms (50-2).

10 What I have to say about this expression will apply to any alternative which uses a truth functional connecter to connect modal expressions equivalent to ‘◊p’ and ‘ò(◊p)’.

11 What I mean by this is that there are ways for things to be which make the proposition true. When Wittgenstein says tautologies lack truth conditions (4.461) he uses ‘truth conditions’ in a different way such that on his usage, a condition for the truth of a proposition is a ‘truth condition’ only if it can fail to be satisfied (ibid). My usage is closer to 4.463, ‘¶1,2.

12 And as suggested in the last section above, the meaningless of modal propositions seems to be a consequence of what the Tractatus says about showing and saying at least for expressions containing essential occurrences of expressions e.g., of the form ‘Possibly p’ where p is an elementary proposition.

13 This sort of argument is due to Lewis, David On the Plurality of Worlds (Oxford: Basil Blackwell 1986), 83-4Google Scholar.

14 Bradley knows that the difference between abstract and concrete objects is not terrifically clear. In the spirit of Lewis, he argues that whatever you think the distinction amounts to, if donkeys are concrete in the actual world, they are concrete in possible worlds too (210ff.).

15 See Merrihew Adams, Robert ‘Theories of Actuality,’ in Loux, Michael J. ed., The Possible and the Actual (Ithaca: Cornell University Press 1979), 201Google Scholar

16 I am indebted to Jay Atlas for conversation on this point.

17 This is from Tapscott’s e-mail response to my description of Bradley’s view; if it misses the point, my description is to blame.

18 For a summary of these, see Black, Max A Companion to Wittgenstein’s Tractatus (Ithaca: Cornell University Press 1964), 247-8Google Scholar, 344-66). I discuss some of Wittgenstein’s position in my ‘Wittgenstein’s Tractatus,’ in Shanker, S.G. ed., The Routledge History of Philosophy, vol. 9 (London: Routledge, Chapman, and Hall forthcoming)Google Scholar.

19 In Copi and Beard, eds., 31-7; originally published 1929.

21 A familiar problem with Russell’s account is that claims about swords don’t really reduce to claims about sense data. A less discussed problem is that if 11. is false, but not necessarily false, it’s analysis must—by Russell’s account of possibility —lead us to functions whose variables are sometimes true, i.e., true of some sense data actually experienced in this world. Russell doesn’t tell us how the relevant functions can some true propositions as values without making 11. true.

22 The identity sign would be used for the same purpose in the further steps required to complete the translation of 11. into a quantified expression whose variables range over sense data.

23 Fogelin, Robert J. Philosophical Interpretations (Oxford: Oxford University Press 1992), 171-2CrossRefGoogle Scholar

24 For example, see Black, 111-12.

25 And yet he also acknowledges that Wittgenstein believed expressions which include what seem to be names for complexes require analysis into expressions which do not (59-60, 136). This is puzzling because it’s hard to see why any such analysis would be necessary if expressions like ‘Excalibur’ can be treated as names for non-actual individuals. Bradley addresses this very briefly at page 136, saying that Wittgenstein’s analysis of the name for a complex reveals its structure in a waywhich is illuminating in something like the same way as is the analysis of water into oxygen and hydrogen.

26 This is why we can understand propositions without having their meanings explained to us if we understand their constituent parts, and why propositions can use old expressions to express new meanings (4.021-4.024, 4.027ff.). As the parenthetical remark in 3.143 suggests, Wittgenstein thought his claim that propositions are not names marked a crucial distinction between his and Frege’s theories of language.

27 I take the indeterminacy mentioned at 3.24, ‘¶ 3 to consist in the fact that a quantified expression can claim that at least one object fits some description and leave it open just which object that would be. What this passage means by ‘generality sign’ is a quantifier.

28 For the discussion in the notebooks, see Notebooks 1914-1916, 50ff., 59ff., 64, 69). The Tractatus passages are inconclusive at best (136). One is 5.02; its point is not to give an example of a genuine name whose referent is a complex, but rather, to illustrate the difference between an argument and an index. The other, 3.323, concerns unanalyzed natural language uses of the sign ‘Green’ to talk sometimes about a person, and other times, about a color. Contrary to Bradley, the point of this example (which Wittgenstein suggests is on a par with the use of ‘is’ as a copula, and an expression for existence as well as identity) is not that ‘Green’ is a name which refers to a complex. Its purpose is to illustrate how the surface grammar or ordinary speech obscures the logical form of the propositions it expresses.

29 Very roughly, determinacy requires that for any fact whatever, the sense of a proposition completely determines whether that fact satisfies or fails to satisfy any of its truth conditions. For details see my Wittgenstein’s Philosophy of Language (London: Routledge & Kegan Paul 1972), 41-7,53-4.

30 He thought analysis is needed, not to tell us what they say, but rather, to help us understand how they say it.