1 Plato: Phaedo, translated with notes by Gallop, David Clarendon Plato Series (Clarendon Press, Oxford, 1975).Google Scholar

2 It is found in Olympiodorus, In Platonis Phaedonem Scholia, ed. Norvin, W. (Leipzig, 1913),Google Scholar 134.14, and derives from 72B 1; the name Argument from Alternation, is also used (cf. 71E 8, 72A 11, B 8). Both names are best restricted to the section from 71E 4 to 72E 1. The standard name for 69E - 72E in Olympiodorus is The Argument from Opposites; and that is certainly an appropriate denomination.

3 E.g., Furlong, E. J. (“Two Arguments in Plato's Phaedo,” Hermathena 55 (1940), 62-72), p. 64Google Scholar: “It requires no great ingenuity to discern the fallacies in the reasoning and the errors in the premisses of this argument”; Crombie, I. M. An Examination of Plato's Doctrines (London, 1963), vol I, p. 305Google Scholar: the Argument is “very weak”.

4 Op. cit., p. 104.

5 Op. cit., p. 104; cf. Bluck, R. S. Plato's Phaedo (London, 1955), p. 18.Google Scholar Bluck thinks that the Argument is “an argument of probability on mechanistic lines,” set down to please those who are impressed by “mechanistic arguments” (p. 22). I find nothing mechanistic in the Argument (see below for the mistaken view that the Argument turns on a Law of Nature); the reference to probability is presumably based upon the word at 70B 7 — but that word reflects the general reservation that Plato expresses more fully at 107AB (cf. 85CD) and does not point to a feature peculiar to the Cyclical Argument.

6 See Burnet, J. Plato's Phaedo (Oxford, 1911), p. 47;CrossRefGoogle Scholar Bluck, op. cit., pp. 20– 21; Crombie, op. cit., vol. I, p. 305; contra, Furlong, op. cit., p. 68. Gallop himself gives no grounds for a ‘dialectical’ reading of the Argument; nor does he say what he means by ‘dialectical’.

7 See especially Bluck, op. cit., pp. 20–21. I ignore the claim, based upon the word at 70B 6, that the Cyclical Argument is especially paraded as a or fairy tale. For a plausible explanation of Plato's choice of the verb see Gallop, op. cit., p. 228, n. 15.

8 See Gallop, op. cit., pp. 109, 111-12; and cf. n. 46 below.

9 All translations are Gallop's (but see n. 23).

10 This interpretation was first advanced by Bonitz, H. Platonische Studien (Berlin, 1886 ³). pp. 303-6;Google Scholar it is accepted by Archer-Hind, R. D. The Phaedo of Plato (London, 1894 ²). pp. 10, 71;Google Scholar Burnet, op. cit., p. 47; Bluck, op. cit., p. 57; Crombie, op. cit., vol. I, p. 308. Bonitz, p. 307, draws the corollary that all the arguments for immortality in the Phaedo depend upon the Theory of Forms. (Gallop, op. cit., p. 97, says, rather oddly, that “the doctrine of immortality is logically dependent upon the Theory [of Forms]. And all the major arguments except the first are based upon it.”) Friedlӓnder, P. Plato, trans. Meyerhoff, H. (London, 1958-69), vol. III, pp. 45–46Google Scholar, follows Bonitz, and concludes that the Cyclical Argument is thereby ‘depersonalised’: “it is not a matter of my death and yours, or your life and mine.” I do not see how that conclusion follows.

11 I agree wholly with Furlong, op. cit. (cf. Gallop, op. cit., pp. 135–36). Gallop, op. cit., p. 103, holds that “Plato does not offer a set of discrete, self-contained proofs of immortality, but a developing sequence of arguments, objections, and counterarguments. As the dialogue unfolds, the earlier arguments are criticised, refined, or superseded ….” I do not deny that the arguments in the Phaedo form a connected series; but I do maintain that the Cyclical Argument is self-contained, and is neither refined nor superseded in the rest of the dialogue.

12 Olympiodorus, op. cit., 125.30-34, 128.1-7, 206.3-5. I have not noticed the point in any later commentator. Furlong correctly observes that the Recollection Argument does not show immortality, but he holds that the Cyclical Argument “is offered by Socrates as a complete proof of the soul's immortality” (op. cit., p. 68).

13 Wolfe, J. (“A Note on Plato's ‘Cyclical Argument’ in the Phaedo,” Dialogue 5 (1966), 237-38), p. 237CrossRefGoogle Scholar, suggests that the opposites are, properly speaking, things designated by comparative adjectives; but that will not do logically, and it fails textually inasmuch as Plato's first examples of opposites are not comparative.

14 Olympiodorus, op. cit., 129.14-16 (cf. 52.9-11, 137.18-22) correctly notes: “Opposites are twofold — either the things that are shared [i.e. the properties] or the sharers [i.e. the objects]; and the latter are assumed here [i.e. at 70DE], as Socrates himself later tells us”; see also Loriaux, R. Le Phédon de Platon (Namur, 1969), p. 127.Google Scholar Gallop's italics at 103B 3, B 4, and B 6 correspond to nothing in the Greek (cf. 83B 2, 96E 9, 105C 1-3; contrast e.g. 71E 4, 73B 6, 74C 7, where English italics render Greek word order or the particle ). Gallop uses italics in three further ways: (i) they translate in phrases of the form (65D 5: seep. 226, n. 5); (ii) is translated ‘what it is’ (74B 3, D 6, 75B 2, etc.); (iii) the verb ‘to be’ is italicised when used of Forms (65C 3, 74B 1, 77A 4, etc.). Practice (i) is unnecessary; Gallop abuses it (see 74A 10); and once it threatens serious misunderstanding (see 105C 1-7). Practice (ii) is explained — if hardly justified — by the fact that the phrase is “semi-technical” (p. 130). Practice (iii), which is not adhered to with complete consistency, is quite unwarranted. I might note too Gallop's habit of capitalising ‘Being’ when it translates (e.g. 65D 12, 76D 7, 78D 1). He does that, he says (p. 94), to distinguish the noun ‘being’ from the participle. But he does not always translate by ‘Being’; the capital is unnecessary (English syntax does the distinctive work); and ‘Being’ inevitably conveys an imposingly metaphysical impression which lacks. Such criticisms of Gallop's translation are pedantic. But the translation of Greek philosophy is an extended exercise in higher pedantry.

15 For a list of some of them see Aristotle, Met ∆ 24.

16 E.g., Phys 205a6; see in general Phys A 5.

17 So Bonitz, op. cit., p. 303; Archer-Hind, op. cit., p. 9; Bluck, op. cit., p. 19; cf. Friedländer, op. cit., vol. III, pp. 44–45.

18 So Bonitz, op. cit., p. 308 (he finds (P1) a “commonplace of the philosophical consciousness” from early Pre-Socratic days); Bluck, op. cit., p. 56; Hackforth, R. Plato's Phaedo (Cambridge, 1955), p. 64;Google Scholar cf. Gallop, op. cit., p. 110.

19 See Williams, C.J.F. (“On Dying,” Philosophy 44 (1969), 217-30)CrossRefGoogle Scholar, pp. 217–18.

20 Williams, op. cit., p. 217, calls (P1) a “logically true generalisation”; Crombie, op. cit., vol. I, p. 306, calls it a “logical truism” (if ‘truism’ implies triviality, I dissent); cf. Gallop, op. cit., p. 109.

21 See also 92E - 95A. Gallop, op. cit., p. 158, sees there an intertwining of two independent arguments against Simmias's attunement theory of the soul: 92E 4 - 93A 10 contains the premisses for argument A; 93A 11 - C 10, the premisses for argument B; 93D 1 – 94B 3 contains argument B; and 94B 4 - 95A 3 contains argument A. In fact, however, the text exhibits no such neat symmetry; for each of the four sections Gallop distinguishes contains both premisses and inferences: the passage is a good example of but is not a schematic operation.

22 See in general Top 1.

23 Gallop translates here as “be born” rather than “come into being”: see below, n. 30.

24 For this paraphrase of “living things come into being” see below, p. 410.

25 Prior, A.N. Papers on Time and Tense (Oxford, 1968), p. 3.Google Scholar

26 Gallop, op. cit., pp. 107–8, appositely compares ‘alive’ and ‘dead’ to ‘married’ and ‘divorced'. Williams, op. cit., p. 218, construes ‘living’ as ‘existing’ (see below), and denies that ‘not yet existing’ forms a tertium quidbetween ‘living’ and ‘no longer living’. For, he says, “I can think of no way of attaching sense to the words ‘My father does not yet exist’ ”. Williams's example is odd; but it hardly proves his general thesis, which is quite certainly false. Beelzebub suggested to Satan that he tempt Moses: “No,” replied the Archfiend (who was omniscient though malevolent), “Moses does not yet exist — let us begin with Adam.”

27 Cf. Williams, op. cit., p. 218.

28 See Williams, op. cit., p. 221; cf. Gallop, op. cit., p. 106.

29 See 70c 9, D 3, 71D 11-2, D 14, 72A 4, D 1, D 9.

30 See Gallop, op. cit., pp. 105, 171, 220. Gallop translates by ‘be born’ at 70C 8, D 1, D 3, 710 15, 72A 5, A 8, D 9. He seems to treat ‘be born’ and ‘come to be alive’ as synonyms (see p. 105); and he equates coming to be alive with being conceived (p. 111). But ‘a is born’, ‘a comes to be alive’, ‘a is conceived’, and ‘a comes into existence’, all have distinct meanings; and in fact conception, quickening, and birth occur at different times. The Society for the Protection of Unborn Children does not hope to defend nonexistent people from misfortune. should always be translated ‘come to be’. (On the verb see further Gallop, op. cit., pp. 104–5, 170-71.)

31 Geach, P.T. God and the Soul (London, 1969), p. 59,Google Scholar argues that existence is a genuine property of individuals, at least in one of its forms; and then, he thinks, “ ‘to be’ … has the same reference as ‘to live’ …. This may confirm us against sophistical attempts to show that the verb ‘to be’ in this sense is not a genuine predicate of individuals. ‘Poor Fred was alive and is dead’; how could one argue that this is not a genuine predication about poor Fred?” Contra, Williams, op. cit., pp. 222–28.

32 Cf. 67D 4, D 9. (As Gallop notes, op. cit., p. 86 (cf. p. 110), ‘death’ here means ‘dying’ not ‘being dead’.) 64C 4-9 does indeed “prejudge the question in favour of the soul's survival” (Gallop, op. cit., p. 86); but the definition of death can easily be purified of that prejudgment. For the inconcinnities in Plato's use of ‘death’, ‘die’, etc. in the Phaedo see Gallop, op. cit., pp. 86–87.

33 “The remarkable thing is that the same Socrates who is so adamant about correctness of definition in ethical discussion in the so-called ‘Socratic’ dialogues, makes so little attempt to give a coherent and internally consistent definition of the soul in the Phaedo,” Robinson, T.M. Plato's Psychology (Toronto, 1970), p. 32.Google Scholar On in the Phaedo see Gallop, op. cit.,pp. 87–90; Robinson, op. cit., chap. 2.

34 Loriaux, op. cit., p. 113, says that the Cyclical Argument is “tied to the conception of the breath-soul”; but though that conception is used in Cebes' speech at 69E 6 – 70B 4, it does not appear in the Argument itself. According to Robinson, op. cit., p. 26, the Argument construes both as a ‘life-principle’ and as a cognitive principle, and fails to show that these two principles are identical; but I see neither principle in the text. Crombie, op. cit., vol. I, p. 308, asserts that “it is important to notice that Socrates’ problem is not to prove that men survive death, but to prove the immortality of the animating agent whose independence of his body has already been taken for granted.” But (i) the argument is not concerned with immortality (above, p. 401); (ii) the Argument tries to prove, and does not take for granted, the independence of the soul from the body; (iii) although Socrates does not believe that men survive death, he does believe that he will survive his death.

35 “[U]sually…self and soul are taken as synonymous” (Robinson, op. cit., p. 32; cf. Gallop, op. cit., p. 88). Robinson also talks of the as a sort of “counter-person” or “duplicate person” (op. cit., p. 22): that is a somewhat different notion.

36 Alcibiades I, 129A 7 - 130D 6, argues that “the man turns out to be nothing other than his soul” (130C 2). Plato's view here (if the Alc is his work) is different from that in the Phaedo; but the difference is terminological. Greek has no word for ‘person’; and it is easy to see how ‘man’ comes to do double duty, for ‘person’ and for ‘human being’. It is a tribute to Plato's insight that he was not misled by that fact into identifying persons and men; but of course any Greek acquainted with the doctrine of metempsychosis was implictly aware that persons are logically distinct from men.

37 But see 77D 4, 84B 2 (cf. Gallop, op. cit., p. 86).

38 Compare the remarks on ‘a does not exist’ in Plantinga, A. The Nature of Necessity (Oxford, 1974), p. 151.Google Scholar In Plantinga's terminology, (8a) is a predicative singular proposition, (8b) an impredicative singular proposition.

39 See Geach, op. cit., pp. 83–5, who draws on Aquinas, Summa Theologiae Ia, q.45, art.1-2.

40 See Aquinas, loc. cit., art. 2,2: “creari est aliquid fieri; sed omne fieri est mutari. Ergo creatio est mutatio.” Aquinas retorts that “creatio non est mutatio“ (loc. cit., ad 2). He diagnoses Aristotle's error by remarking that the ancient philosophers concentrated exclusively upon sublunary and particular comings into being (loc cit., ad 1); he observes that (P2*), in its Aristotelian version does not hold of creation (loc. cit., ad 2; cf. art. 1, resp.: “si consideretur emanatio totius entis universalis a principio, impossible est quod aliquod ens praesupponatur huic emanationi”; see also Spinoza, Cogitata Metaphysica, II, chap. 10); and he suggests that even so we are humanly bound to conceptualise creation on the model of change: “in creatione … non potest accipi aliquid idem aliter se habens nunc et prius, nisi secundum intellectum tantum, sicut si intelligatur aliqua res prius non fuisse totaliter et postea esse” (loc. cit., ad 2). For a defence of Aristotle, see Bennett, J. Kant's Dialectic (Cambridge, 1974), pp. 62–65.Google Scholar Bennett argues that “any existencechange can be regarded as an alteration,” and that “any existence-change ought to be reported in a manner which represents it as an alteration.”

41 Cf. Loriaux, op. cit., p. 131: “in other words, the coming to be which Plato analyses is always a coming to be of an accidental sort,” Loriaux describes that as “begging the question”, repeating a criticism that is older than Olympiodorus (op. cit., 53.6). Gallop, op. cit., p. 105, also accuses Plato of “begging the question”, but on a different count.

42 Gallop, op. cit., pp. 105–6, argues that the “central objection” to the Cyclical Argument is that “it insinuates a view of ‘birth’ in which the soul's discarnate existence is already covertly assumed”; for “the argument treats the predicate ‘alive’ as if it stood for an attributive capable of being acquired by an antecedently existing subject.” But ‘alive’ is such a predicate: Plato's error lies not in assuming that we can acquire the property of being alive, but in assuming that we do acquire that property; and that is a factual error, not a conceptual confusion.

43 See Robinson, T.M. (“Phaedo 70C: an Error and an Explanation,” Dialogue 8 (1969), 124-25;CrossRefGoogle Scholarcf. Gallop, op. cit., p. 110. Robinson's explanation, that Plato cheats or blunders by his use of the verb is unconvincing: Robinson thinks that the prefix in fact introduces the crucial notion of rebirth into the Argument, though it is apparently a harmless partner of the in . But the notion of rebirth is brought on quite openly by the use of at 69C 8 and 72A 5. Plato is not trying to slip anything past us by ; nor is he himself misled by the prefix. Wolfe, op. cit., p. 238, endorsed by Robinson, op. cit., p. 124, says that Plato's argument is invalidated by its failure to distinguish ‘be alive’ from ‘come alive’: that is certainly incorrect as far as 69E - 71 E goes.

44 The phrase “in that way too”, , at 72A 4 is puzzling. It can perhaps be interpreted in a way that defends Plato from the final criticism that I make against him. Thus: should be taken not with but with the clause that follows it: “We're agreed that living people come to be from the dead in that way too …. ” In what way? The way mentioned in 71E 14 - 72A 2, viz. coming to life again, . And Plato says “in that way too” in order to distinguish the way of 71E - 72A from the ordinary way of coming to be alive with which 69E - 71E was concerned. Thus 69E - 71E establishes that (i) we come to be from the dead and hence had prenatal existence, and 71E- 72A establishes that (ii) we come to be again from the dead and hence shall have post mortem existence. In our argument, 69E - 71E, there is no serious reference to being born again: that is reserved for the next stretch of the text. Our argument, then, aims only at conclusion (C).’ That interpretation has its attractions; but it is difficult to sustain textually, and it is philosophically uninteresting (for it throws all the weight of the Cyclical Argument onto Plato's lame assertion that nature is not lame (71E 9)).

45 E.g., 60C 8, “thanks for reminding me” (why not “thank you”?); 102B 8, “Well anyhow” (for : the slang hides the logical structure of the exchange); 82E 1, “philosophy … has been literally bound and glued to the body.” (Richard Robinson has almost convinced me, by 168 Shakespearean references, that “Plato was sick” may stand at 59B 10 even in British editions.)

46 I have already remarked upon Gallop's use of italics (above, n. 14) and of his translation of in 69E - 72E (above n. 30). Here are a few more notes on the translation. (i) Gallop holds, with the modern orthodoxy, that ‘cause’ is a “hallowed mistranslation” (op. cit., p. 169), and he prefers ‘reason'. In fact I think that ‘cause’ is exactly right; and it will seem so to anyone not bamboozled by Hume into thinking that causes must, logically, run to regular timetables (see J. Barnes, Aristotle's Posterior Analytics (Oxford, 1975), p. 96). (ii) at 101D 1 and D 3 is a crux: Gallop translates ‘hang on to’; at p. 188 he refers to n. 67 “for the meaning of ‘hang on to’”; but n. 67 (p. 235) leaves us none the wiser, and I still do not understand the received text at 101D. (iii) At p. 210, Gallop objects to Hackforth's translation of 105B 8-9 on the grounds that it “wrongly suggests that fire and fever are necessary conditions for heat and illness respectivesly”; but his own translation like Plato's Greek, carries exactly the same suggestion. (iv) At 93A 1 is rendered ‘compound’: that suggests a mistaken notion of the attunement theory of the soul. A here is, as Gallop makes plain on p. 149, a state of being compounded. (v) At 80C 6 is translated ‘even if’ (d. p. 230, n. 35); I doubt if the word can bear that sense. (vi) At 104D 5 surely refers to … , not to … ; that avoids the curious puzzle Gallop raises on p. 194.

47 E.g. pp. 80–83, where it is hardly worth listing four interpretations of 62A 2-5; pp. 148–49, where three construes of are discussed — and the philosophically exciting aspect of the theory gets a mere five lines of comment; pp. 202ff, where version B of the argument at 104C 7 - 107A 1 is open to simple refutation if Gallop's comment on p. 214 is correct.

48 E.g. pp. 77–78, 79-83, 98-102, 199.

49 Where he does use symbolism, it is introduced without explanation. On p. 83 its use is pointless; on pp. 208–9 the horseshoe of material implication gives quite the wrong sense to the theses it formalises.

50 I note 96E 2 (for “latter” read “former”); p. 80, four lines up (for “pleasurable” read “preferable”); p. 125, line 14 (for “Form Equality” read “Form Inequality”).

51 Robert Delahunty and Richard Robinson kindly read a draft of this paper and proposed several improvements to it.