1 The Philosophy of W.V. Quine, expanded ed., eds. L.E. Hahn and P.A. Schilpp (La Salle: Open Court, 1998) 19. “Two Dogmas of Empiricism” was published in Philosophical Review 60 (1951), 20-43, and reprinted with revisions in W.V. Quine, From A Logical Point of View, 2^nded. (Cambridge: Harvard University Press, 1980), 20-46. Page references in the text are to the revision. Except where noted, quoted passages also occur in the original.

2 The Time of My Life, (Cambridge: MIT Press, 1985), 226, also 150.

3 “Two Dogmas in Retrospect”, in W.V. Quine, Quintessence, ed. R. Gibson (Cambridge: Harvard University Press, 2004), 57.

4 The Philosophy of W.V. Quine, 19.

5 W.V. Quine, Word and Object, (Cambridge: MIT Press, 1960). 46.

6 Op. cit., 66-67. In this work Quine grants that “the stimulus meanings of ‘Bachelor’ and ‘Unmarried man’ are ... identical for any one speaker” (46), treats terms like “bachelor” and “unmarried male” as “stimulus synonymous” (55) and characterises sentences judged true by competent speakers of English “come what stimulus ... may” as “stimulus analytic” (along with sentences obtainable “from logical truths by [stimulus] synonymy substitution”) (66-67).

7 La Salle: Open Court, 1974, 79. Quine reckons this notion “a somewhat nearer approximation to the analytic sentences uncritically so called” than the notion of “stimulus analytic” in Word and Object (80).

8 “Two Dogmas in Retrospect”, 62. Significantly, there is no discussion of analyticity in W.V. Quine and J. Ullian, The Web of Belief, (New York: Random House, 1970).

9 The Ways of Paradox, rev. and enlarged ed., (Cambridge: Harvard University Press, 1976), 107.

10 Word and Object, 67.

11 The Philosophy of W.V. Quine, 207. Also, in The Roots of Reference, Quine refers to “Carnap and other epistemologists” when proposing a substitute notion of analyticity (80) and in “Two Dogmas in Retrospect” writes: “I recognize the notion of analyticity in its obvious and useful but epistemologically insignificant applications” (61). I am not claiming Quine was as clear when he wrote “Two Dogmas” as he was later on, only that his concerns were epistemological both early and late.

12 For this interpretation, see, e.g., S. Soames, Philosophical Analysis in the Twentieth Century, vol. 1, (Princeton University Press, 2003), 355-360.

13 I disregard less important suggestions touched on in §§1-3, for instance, the idea that the predicates of analytic statements are conceptually contained in their subjects.

14 Analyticity and self-contradictoriness are also said to be “two sides of a single dubious coin” (23), definition said to put “the cart before the horse” (24), necessity sufficiently “narrowly construed” said to have the “air of hocus-pocus” (29) and interchangeability in the required sense said to be “intelligible only insofar as the notion of analyticity is already understood in advance” (31). In addition Quine disapproves of Carnap’s “tende[ncy] to explain analyticity by appeal to ... state-descriptions” (23).

15 In “Carnap and Logical Truth”, Quine says defining synonymy in terms of analyticity would be “circular” but does not take this to show analyticity is disreputable (129). The problem with Carnap’s doctrine of logical truth is rather that it is a “pseudo-doctrine”, one that “leaves explanation unbegun” (113). Also note that Quine should not be upbraided for failing to notice that semantic and ethical notions (even notions like “parent”) cannot be pinned down without reference to notions of the same general sort.

16 “Two Dogmas in Retrospect”, 61. In “Two Dogmas” Quine simply says he goes further than those like “Carnap, Lewis and others [who] take a pragmatic stand” since he repudiates the boundary they take to exist between the analytic and the synthetic. Burton Dreben, Quine’s closest collaborator, often stressed that it is insufficiently appreciated that Quine chiefly acquired his conception of epistemology from Lewis. Also see W.V. Quine, “Comments on Parsons” in Perspectives on Quine, eds. R. Gibson and R. Barrett. (Oxford: Blackwell, 1990), 292.

17 Compare Soames, op. cit., 355, 361 and 363.

18 W.V. Quine and R. Carnap, Dear Carnap, Dear Van, ed. R. Creath, (Berkeley: California University Press, 1990), 47. It is of some interest given what has already been noted that C.I. Lewis was in the audience when Quine lectured on Carnap and near the beginning of his remarks he referred to Lewis’s conception of the a priori as ‘definitive or analytic in its nature’ (68). The reference is to C.I. Lewis, Mind and the World Order, (New York: Scribner, 1929), 231.

19 The Ways of Paradox, 102.

20 Ibid., 101 and 106. Quine seems to have had misgivings about the distinction before 1934 since when visiting Carnap in Prague in 1933, he apparently questioned whether logical axioms are different in kind from empirical sentences. See “Two Dogmas in Retrospect”, 55. Even in 1935 he seems to have had hope for the dogma of reductionism. See “Truth by Convention”, 100, n.20.

21 “Lectures on Carnap”, 49 and 65, and “Truth by Convention”, 102.

22 At the beginning of “Truth by Convention” Quine says: “It is less the purpose of the present inquiry to question the validity of [the contrast between analytic/a priori and synthetic/a posteriori] than to question its sense” (77).

23 Quine allows that some sentences normally regarded as synthetic might end up being classified as analytic, the law of freely falling bodies, for example (“Lectures on Carnap”, 62).

24 In “Carnap and Logical Truth” Quine puts it this way: “The lore of our fathers ... is a pale grey lore, black with fact and white with convention. But I have found no substantial reason for concluding that there are any quite black threads in it, or any white ones” (132).

25 “Two Dogmas in Retrospect”, 55-56.

26 Nor does Carnap propound a “doctrine” or defend a “thesis” (“Lectures on Carnap”, 47 and 66).

27 Ibid., 47.

28 W.V. Quine, From a Logical Point of View, 130-138.

29 Dear Carnap, Dear Van, 430; The Philosophy of Rudolf Carnap, 919; and R.M. Martin, “On ’Analytic’” Philosophical Studies3 (1952), 42-47. In the revised version of “Two Dogmas”, Quine added a paragraph in response to Martin (35).

30 “Notes on the Theory of Reference”, 138.

31 Quine criticises the behavioural criteria for analyticity that Carnap offers in “Meaning and Synonymy in Natural Languages” Philosophical Studies 6 (1995), §3. See, e.g., Word and Object, 35.

32 Word and Object, 257.

33 The Pursuit of Truth, rev. ed., (Cambridge: Harvard University Press, 1992), §3.

34 “Two Dogmas in Retrospect”, 60. This remark, by the way, indicates that Quine does not repudiate analyticity because it is imprecise.

35 In conversation, Burton Dreben observed more than once that Quine appreciated that his argument was similar to Einstein’s but refrained from mentioning the fact for fear of seeming immodest.

36 “Lectures on Carnap”, 64, and “Truth by Convention”, 77.

37 In “Two Dogmas” Quine questions the possibility of explaining “analytic-in-L” for “variable L” (34), and in “Notes on the Theory of Reference” adds that we do not have “any systematic routine for constructing definitions for ‘analytic-in-L’, even for the various individual choices of L” (138).

38 Phlogiston and the ether, it hardly needs recalling, are discounted despite the fact neither has been demonstrated conclusively not to exist. For Quine meanings are in the same boat as the ether.

39 None of this would satisfy Carnap, his final argument for analyticity being that it is needed to understand Einstein’s development of the theory of relativity. See R. Carnap, Philosophical Foundations of Physics, (New York: Basic Books, 1966), 257. In Quine’s view, by contrast, Einstein’s manoeuvre is nothing more or less than an example of semantic ascent (Word and Object, 272).

40 Op. cit., 63 and 65. Also compare: “[The sentences] which we are not going to give up at all, so basic are they to our whole conceptual scheme ..., if any, are the sentences to which the epithet ‘a priori’ would have to apply” (65).

41 In Word and Object Quine laments that “[t]he title of ‘Two Dogmas’ ... has proved unfortunate in its unintended but very real suggestion that there is no empiricism without the dogmas in question” (68, n.7).

42 Compare Pursuit of Truth, 19: “The most notable norm of naturalized epistemology [Quine’s preferred view] actually coincides with that of traditional epistemology. It is simply the watchword of empiricism: nihil in mente quod non prius in sensu”.

43 See R. Carnap, The Logical Syntax of Language, (Chicago: Open Court, 2002), 318. In “Two Dogmas” Quine takes “the unit of empirical significance [to be] the whole of science” (42). Later he will take it, more realistically, to consist of “a critical semantic mass” of sentences (see, e.g., Pursuit of Truth, 17).

44 It is worth noting that C.I. Lewis, whom Quine couples with Carnap at the end of “Two Dogmas”, refers to analyticity as an alternative to self-evidence and intuition, op. cit., viii.

45 Quine did not believe—he will later underscore the fact—that the problem of underdetermination cannot be solved by noting that good theories are simple, general and the like. See, e.g., “On Empirically Equivalent Systems of the World”, in Confessions of a Confirmed Extensionalist, 242, and Pursuit of Truth, 99.

46 It is worth noting that whereas Quine holds that empiricists do not have to avail themselves of the notion of analyticity since they can regard putative a priori truths as central to our conceptual scheme (44), Carnap takes the sentences in question to be central to our conceptual scheme because they are analytic (and a priori). For Carnap firmness and centrality are what require explaining, not what do the explaining.

47 See, for instance, R. Creath, ‘Every Dogma has its Day’, Erkenntnis35 (1991), 348, 371 and 385.

48 Similar remarks, I believe, are in order regarding the dogma of reductionism. As I see it, a study of §5 (also §4) would complement the present examination of §§1-3 and §6. Also notice that when detached from Quine’s critical remarks, empiricism free of the dogmas is arguably inferior to empiricism supplemented by them. Armed with the analytic/synthetic distinction the modern empiricist can avoid the awkward conclusion that numbers exist in the same sense as electrons and regard sentences like “The number one weighs two kilograms” as nonsensical rather than obviously false.

49 While writing this paper I have had in mind discussions about Quine’s treatment of the analytic/synthetic distinction that I had with Burton Dreben during the last years of his life. In addition I want to thank Paul Forster for constructive criticism and useful advice, Peter Hylton for conversation and reading material, and two anonymous reviewers for comments and suggestions.