¹ Heidegger, M., Sein und Zeit ¹² (Tübingen, 1972), §46–8, pp. 235–46Google Scholar, subjects this tendency to a remarkably acute and poignant phenomenological analysis.

² Gnilka, C., ‘Ultima verba’, JAC 22 (1979), 5–21Google Scholar. Schmidt, W., De ultimis morientium verbis (Diss. Marburg, 1914)Google Scholar provides a rather heterogeneous and poorly organized collection of Greek and Roman examples ranging from Homer to the New Testament (and beyond). On the ancient exitus genre, cf. Marx, F. A., ‘Tacitus und die Literatur der exitus illustrium virorum’, Philologus 92 (1937), 83–103CrossRefGoogle Scholar; Schunck, P., ‘Studien zur Darstellung des Endes von Galba, Otho und Vitellius in den Historien des Tacitus’, SO 39 (1964), 38–82CrossRefGoogle Scholar; Römer, F., ‘Ein Glanzstück römischer Memorabilienliteratur (Val. Max. 2, 6, 8)’, Wiener Humanistische Blätter 31 (1989), 52–65Google Scholar.

³ Homer, : Certamen 332–8Google ScholarAllen, , Vita V. 47–52Google ScholarAllen, : doubts on this score expressed polemically in Vita Herodotea 512–16Google Scholar Allen. Virgil, : Vita Donati 36Google Scholar (14.136–40 Hardie), Hieronymus, Chronicon 165Google Scholar h 21. ad Ol. 190.3 (37.7 Hardie); doubts on this score perhaps implied e silentio in Vita Probiana 27.18–21 Hardie.

⁴ Callimachus, 35 Pf. (= Hellenistic Epigrams xxx. 1185–6Google ScholarGow-Page, , A. P. 7.524Google Scholar); Leonidas xciii. 2535–40 Gow-Page, (= A. P. 7.715)Google Scholar; Meleager ii–iv. 3984–4007 (= A. P. 7.417–19). For some Latin variations, cf. Hor. Car. 3.30, Prop. 1.22.

⁵ Whether to our satisfaction is not at issue here.

⁶ To break the tension and heighten the dramatic pathos, Socrates even makes a joke before drinking the poison: see Burnet, J. (ed.), Plato's Phaedo (Oxford, 1911), pp. 116–17Google Scholarad 117b5, b6.

⁷ Phaedo's admission that he is weeping in reality not for Socrates but for himself is an allusion to (or a reminiscence of) a celebrated Homeric verse (Il. 19.302) which was to become a clichè in the Greek romances (Chariton 8.5.2; Ach. Tat. 2.34.7; Heliod. 1.18.1).

⁸ Already at the beginning of the dialogue (59a), Phaedo had referred to Apollodorus' lack of self-restraint as notorious: this is not the only example of ring-composition in this text (see below, nn. 13 and 69). On Apollodorus' reputation, cf. Symp. 173d.

⁹ ἤδη οὖν σχεδ⋯ν τι αὐτο⋯ ἦν τ⋯ περ⋯ τ⋯ ἦτρον ψυχ⋯μενα, κα⋯… 118a5–6. The construction (but not its emphatic force) is recognized by LSJ s.v. κα⋯ A.III.3; Gerth, R. Kühner-B., Ausführliche Grammatik der griechischen Sprache (Hannover-Leipzig, 1904 ³ = Hannover, 1976), ii. 2, p. 231Google Scholar; and Denniston, J. D., The Greek Particles ² (Oxford, 1954), p. 293Google Scholar s.v. κα⋯ I (10).

¹⁰ This is not stated explicitly, but can be inferred with certainty from the words a few lines later, ⋯ ἄνθρωπος ⋯ξεκ⋯λυψεν αὐτ⋯ν (118a12—13).

¹¹ So especially Baron, James R., ‘On Separating the Socratic from the Platonic in Phaedo 118’, CPh 70 (1975), 268–9, here p. 269Google Scholar; Schmidt, , op. cit., pp. 54, 55Google Scholar; von Wilamowitz-Moellendorff, U., Platan ² (Berlin, 1920), i. 172, 178 n. 1, 325Google Scholar; ii. 57. Despite the arguments advanced here for the authenticity of Socrates' words as reported by Plato, the anonymous referee for this journal and several of my friends are not convinced that they might not have been invented by Plato after all. If they are wrong, the central thesis of this paper still remains the most plausible solution; if they are right, it becomes virtually certain.

¹² That the Phaedo belongs to the period before the Republic is generally accepted, but its exact date is uncertain. Hackforth, R., Plato's Phaedo (Cambridge, 1955), p. 7Google Scholar, argues for 387 or shortly thereafter. Ledger, G. R., Re-counting Plato: A Computer Analysis of Plato's Style (Oxford, 1989), p. 224Google Scholar, places the Phaedo between the Gorgias (386) and the Protagoras (380?). Dixsaut, M., Platon: Phédon (Paris, 1991), pp. 26–8Google Scholar, suggests about 383–382. Thesleff, H., Studies in Platonic Chronology (Helsinki, 1982), pp. 140–4 and 237Google Scholar, sets it as late as 380–375. For a helpful survey of the history of attempts to determine the sequence of Plato's dialogues, see now Brandwood, L., The Chronology of Plato's Dialogues (Cambridge, 1990)CrossRefGoogle Scholar.

¹³ So e.g. Wilamowitz, , op. cit., ii. 58–9Google Scholar. Such an interpretation can be supported by reference to other passages in the same text which emphasize Socrates' piety, e.g. 60el–61b7, 85b3–7, 117cl–3; the ring-composition between the beginning and the end of the dialogue is thereby strengthened, see above, n. 8 and below, n. 69. But of course this explanation cannot exhaust the meaning of Socrates' words for Plato, as it cannot explain why Socrates' piety should be demonstrated with reference to Asclepius in particular rather than in any other way. Wilamowitz writes, ‘Charakteristisch ist, worauf seine letzten Gedanken gerichtet sind, das Spezielle daran ist ganz ohne Belang’ (op. cit., i. 178 n. 1): but in interpretation it is precisely the particular that matters in the end.

¹⁴ Yet so Lactantius, , Div. Inst. 3.20. 16–17Google Scholar, Int. Epit. 32.4–5.

¹⁵ Baron, op. cit.

¹⁶ For the divergent testimonia, see Herwig, W. (ed.), Goethes Gespräche, i–v (Zurich and Munich, 1965–1987), iii. 2, pp. 882–9, 902, nr. 7014–19, 7031Google Scholar; for commentary on this point, see iv. 573 ad nr. 7014. The question of what Goethe's last words really were is discussed by Schüddekopf, C. (ed.), Goethes Tod. Dokumente und Berichte der Zeitgenossen (Leipzig, 1907), pp. 26ffGoogle Scholar.

¹⁷ A partial collection of ancient testimonia is provided by Wyttenbach, D. (ed.), ΠλATΩNOΣ ΦAIΔΩN. Platonis Phaedo ² (Leipzig, 1825), pp. 318–19 ad 118aGoogle Scholar; a more complete one, together with some recent testimonia, by Heiberg, J. L., ‘Socrates' sidste Ord’, Oversicht over det Kgl. Danske Videnskabernes Selskabs Forhandlinger 1902: 4, pp. 105–16Google Scholar. I have not seen the following: Diez, H. F., ‘Über Sokrates’, Berlinische Monatsschrift 2 (1783), 281–6, 558–63Google Scholar; Lundgren, L. O., ‘Kriton, vi är skyldiga Asklepios en tupp’, Studiekamraten 54 (1972), 25–7Google Scholar; Wilder, A., ‘The Last Words of Socrates’, The Platonist 1 (1881), 39–42Google Scholar.

¹⁸ Lucian, , Bis accus. 5Google Scholar; Tertullian, , Ad Nationes 2.2, De Anima 1.6Google Scholar; Origen, , Contra Celsum 6.4Google Scholar. Tertullian, , Apol. 46.5Google Scholar also suggests that Socrates wished to honour Apollo indirectly by sacrificing to his son – implausibly: for why then did Socrates not choose to honour Apollo directly?

¹⁹ Gautier, R., ‘Les Dernières Paroles de Socrates’, Revue Universitaire 64 (1955), 274–5Google Scholar. Such views have been criticized on medical grounds (Baron, , op. cit., p. 269Google Scholar; Re, M. Del, ‘L'estremo voto di Socrate’, Sophia 25 (1957), 290–4, here 290–1)Google Scholar, but it is impossible to know how much weight to give such kinds of evidence (cf. Gill, C., ‘The Death of Socrates’, CQ 67 (1973), 25–8, here 27–8)CrossRefGoogle Scholar.

²⁰ Hugo, V., Cromwell, ‘Préface’, in Théâtre complet 1 (Paris, 1963), p. 426Google Scholar.

²¹ Dirlmeier, F. (ed.), Platon: Phaidon (Munich, 1959), p. 285 ad loc.Google Scholar; Friedländer, P., Platon. 1: Eidos Paideia Dialogos (Berlin and Leipzig, 1928), pp. 162–3 n. 2Google Scholar; Mitscherling, J., ‘Phaedo 118: The Last Words’, Apeiron 29 (1985), 161–5Google Scholar; Rouse, W. H. D., Greek Votive Offerings: An Essay in the History of Greek Religion (Cambridge, 1902), p. 204 n. 3.Google Scholar Such an interpretation merely acknowledges the discrepancy between the apparent meaning of Socrates' words and their dramatic context, without indicating why there should be such a discrepancy or what we should take its meaning to be; and in any case this view is precluded by the grammar of Socrates' words (see below, n. 63).

²² Westerink, L. G. (ed.), The Greek Commentaries on Plato's Phaedo, Vol. III, Damascius (Amsterdam, Oxford and New York, 1977), p. 285 (Ϟς' = 561); another scholium by the same author perhaps points in the same direction, p. 371 (ρνζ' = 157): δι⋯ τἰ ⋯φε⋯λειν ἔφη τῷ 'Aσκληπι⋯ τ⋯ν θυσ⋯αν το⋯ το⋯το τελευταῖον ⋯φθ⋯γξατο;…ἢ ὅτι Παιων⋯ου δεῖται προνο⋯ας ⋯ ψυχ⋯, ⋯παλλατομ⋯νη τ⋯ν πολλ⋯ν π⋯νων. Damascius' explanations are transmitted anonymously and used to be assigned to Olympiodorus; for the attribution to Damascius, now generally accepted, seeGoogle ScholarWesterink, L. G. (ed.), Damascius: Lectures on the Philebus (Amsterdam, 1959), pp. xv–xxGoogle Scholar, and Westerink, , op. cit., pp. 15–17Google Scholar.

²³ See Heiberg, , op. cit., pp. 108–9Google Scholar.

²⁴ Blanke, F. and Gründer, K. (eds.), Johann Georg Hamanns Hauptschriften Erklärt, iiiGoogle Scholar. Sokratische Denkwürdigkeiten, ed. Blanke, F. (Gütersloh, 1959), pp. 183–4Google Scholar; de Lamartine, A., ‘La Mort de Socrate’, in Guyard, M.-F. (ed.), Œuvres poétiques complètes (Paris, 1963), pp. 85–108Google Scholar; Nietzsche, F., Die fröhliche Wissenschaft, Viertes Buch, §340 ‘Der sterbende Sokrates,’ = Colli, G. and Montinari, M. (eds.), Nietzsche, Werke, Kritische Gesamtausgabe, v. 2 (Berlin and New York, 1973), pp. 249.15–250.6Google Scholar, and Götzen-Dämmerung, ‘Das Problem des Sokrates’, = ibid. vi. 3, pp. 61–7. On Nietzsche and Socrates, see Dannhauser, W. J., Nietzsche's View of Socrates (Ithaca and London, 1974)Google Scholar.

²⁵ e.g. Archer-Hind, R. D. (ed.), ΠΛATΩNOΣ ΦAIΔΩN. The Phaedo of Plato² (London, 1894), p. 146 ad lin. 12Google Scholar; Bluck, R. S. (ed.), Plato's Phaedo (London, 1955), p. 143 n. 1Google Scholar; Burnet, , op. cit., p. 147ad 118a7Google Scholar; Carafides, J. L., ‘The Last Words of Socrates’, ΠΛATΩN 23 (1971), 229–32Google Scholar; Capuder, A., ‘Note complémentaire au dernier mot de Socrate’, Ziva Antika 19 (1969), 21–3, here 23Google Scholar; Cumont, F., ‘A propos des dernières paroles de Socrate’, Comptes rendus de l'Académie des Inscriptions et Belles Lettres 1943: 112–26, here 121–6Google Scholar; Dixsaut, , op. cit., pp. 180 and 408–9 n. 382Google Scholar; Re, M. Del, op. cit., pp. 293–4Google Scholar; Re, R. Del, ‘Il gallo dovuto da Socrate ad Esculapio’, A&R 14–16 (1954), 85–6Google Scholar; Edelstein, E. J. L., Asclepius: A Collection and Interpretation of the Testimonies, i–ii (Baltimore, 1945), ii. 130–1Google Scholar; Gill, , op. cit., 27–8Google Scholar; Gnilka, , op. cit., pp. 9–10Google Scholar; Heiberg, , op. cit., pp. 115–16Google Scholar; Kerenyi, C., Asklepios: Archetypal Image of the Physician's Existence, trans. Mannheim, R. (Princeton, 1959), p. 59Google Scholar; Loriaux, R., Le Phédon de Platon: commentaire et traduction, vol. ii. 84b–118a (Namur, 1975), p. 162ad 118aGoogle Scholar; Minadeo, R., ‘Socrates' Debt to Asclepius’, ClJ 66 (1970/1971), 294–7, here 296–7Google Scholar; Reale, G. (ed.), Platone: tutti gli scritti (Milan, 1991), p. 130 n. 143Google Scholar; Robin, L. (ed.), Platon: Œuvres complètes, iv. 1. Phédon (Paris, 1926), 102 n. 3Google Scholar, and Platon: Œuvres complètes, i–ii (Paris, 1940–1942), i. 1334 n. 215Google Scholar; Tarrant, D., ‘Metaphors of Death in the Phaedo’, CR n.s. 2 (1952), 64–6, here 66Google Scholar; Vicaire, P. (ed.), Platon: œuvres complètes, iv. 1. Phédon (Paris, 1983), p. 110 n. 1Google Scholar; de Vogel, C. J., Pythagoras and Early Pythagoreanism: An Interpretation of Neglected Evidence on the Philosopher Pythagoras (Assen, 1966), p. 185Google Scholar; Wohlrab, M. (ed.), Platons ausgewählte Schriften für den Schulgebrauch erklärt. vi. Phaedon ⁴ (Leipzig-Berlin, 1933), p. 163ad2Google Scholar; Wyttenbach, , op. cit., p. 319Google Scholarad 118a (but in the first edition, Leiden 1810, p. 333 ad loc., he had rejected this view as a ‘mystica interpretatio aliena ilia ab hac Platonis narratione’); Zehnpfennig, B. (ed.), Plato: Phaidon (Hamburg, 1991), p. 206 n. 210Google Scholar.

²⁶ It is so much so that we may perhaps excuse Wilamowitz, for his peremptory declaration, ‘das Leben ist keine Krankheit, und Asklepios heilt kein Übel der Seele.… was hat der Heilgott mit dem Sterben zu tun? Wo erscheint er in einer ähnlichen Rolle? Wozu so weit schweifen?’ (op. cit., ii. 57, 58)Google Scholar – but not Stenzel, J. for the terms in which he endorses ‘v. Wilamowitz' wichtige Richtigstellung billigen ungnechischen Tiefsinns’ (RE III A 1(Stuttgart, 1927)Google Scholar, s.v. Sokrates [Philosoph], pp. 826–7): for genuine Greek examples, cf. Damascius' scholia, quoted above, and Phaedo 95dl–4, discussed below.

²⁷ , Pind.P. 3.55–7Google Scholar; , Aesch.Ag. 1022–4Google Scholar; , Eur.Alc. 127–9Google Scholar; Pherecydes, , FGrHist 3F 35aGoogle Scholar; etc. So too in Plato: Rep 3.408b–c. On the relation between the myth of Asclepius and his cult functions, see now Benedum, C., ‘Asklepios – der homerische Arzt und der Gott von Epidauros’, RhM 133 (1990), 210–27Google Scholar.

²⁸ Much of the material is helpfully collected and discussed in Edelstein and Edelstein, op. cit.

²⁹ Paus. 2.27.1, 6.

³⁰ 67a2–6: καì ⋯ν ᾦ ἂν ζ⋯μεν, οὓτως, ὡς ἒοικεν, ⋯λλυτ⋯τω ⋯σ⋯μεθα τοῡ εiδέναι, ⋯àν ὂτι μ⋯λιστα μηδ⋯ν ⋯μιλ⋯μεν τῷ σώματι μηδ⋯ κοινων⋯μεν, ὃτι μ⋯ π⋯σα ⋯νάγκη, μηδ⋯ ⋯ναπιμπλὠμεθα τ⋯ς τούτου φύσεως, ⋯λλà καθαρεύωμεν ⋯π' αὐτο⋯, ἒως ἂν ⋯ θε⋯ς ⋯πολύσῃ ⋯μἂς. 83d7–10: ⋯κ γàρ τοῢ ⋯μοδοξεῖν τῷ σώματι καì τοῖς αὐτοῖς χαίρειν ⋯ναγκάζεται οἶμαι ⋯μότροπός τε καì ⋯μότροφος γίγνεσθαι καì οἶα μηδέποτε εἰς 'Aιδου καθαρ⋯ς ⋯φικέσθαι, ⋯λλà ⋯εì το⋯ σώματος ⋯ναπλ⋯α ⋯ξιέναι….

³¹ So e.g. Bluck, , op. cit., p. 51Google Scholarad 61a (‘infect’), but cf. p. 83 ad 83a (‘contaminated’); Burnet, , op. cit., p. 37Google Scholarad 67a5 (‘nor suffer the contagion of’), but cf. p. 77 ad 83dlO (‘contaminated’, ‘tainted’); Places, E. Des, Platon: æuvres complètes, xiv. Lexique (Paris, 1964), i. 43Google Scholars.v. ⋯ναπιμπλάναι ‘infecter’; Hackforth, , op. cit., p. 48Google Scholarad 67a (‘bodily infection’), but cf. p. 93 ad 83d (‘taint’); LSJ S.v. ⋯ναπίμπλημι II. 2, ⋯νάπλεως II.

³² This is well seen by Parker, R., Miasma: Pollution and Purification in Early Greek Religion (Oxford, 1983), pp. 281–2Google Scholar. The usage was understood in antiquity, cf. Timaeus Soph. s.v. ⋯νάπλεως ⋯ναπεπλησμένος χρ⋯ται δ⋯ ⋯πì το⋯ μεμολυπμένου. Cf. also e.g. Dixsaut, , op. cit., p. 217Google Scholarad 67a (‘contaminer’), p. 250 ad 83d (‘infectée’, but cf. p. 357 n. 186 adloc.); Loriaux, R., Le Phédon de Platon: commentaire et traduction, vol. i. 57a–84b (Namur and Gembloux, 1969),p. 91Google Scholarad 67a (‘contaminé’).

³³ See especially Tarrant, op. cit.; also O'Brien, D., ‘The Last Argument of Plato's Phaedo, I and II’, CQ n.s. 17 (1967), 198–231CrossRefGoogle Scholar and 18 (1968), 95–106, and ‘A Metaphor in Plato: “Running Away” and “Staying Behind” in the Phaedo and the Timaeus’, CQ n.s. 27 (1977), 297–9CrossRefGoogle Scholar.

³⁴ Life is a prison: Phd. 62b, cf. Crat. 400c. Death is separation of body and soul: Phd. 64c, cf. Gorg. 524b. Death is a migration: cf. Apol. 40c, 40e–41c.

³⁵ σ⋯μα/σ⋯μα: Crat 400c, Gorg. 493a. Death is a dreamless sleep: Apol. 40c–e. See in general, de Vogel, C. J., ‘The SŌMA-SĒMA Formula: Its Function in Plato and Plotinus Compared to Christian Writers’, in Blumenthal, H. J. and Markus, R. A. (eds.), Neoplatonism and Early Christian Thought: Essays in Honour of A. H. Armstrong (London, 1981), pp. 79–95Google Scholar.

³⁶ The only fairly close one I know of is provided by Aristotle (Eudemus Frag. 5 Ross = Frag. 41 Rose ³ = Proclus in Remp. 2.349.13–26 Kroll); but it is an open question how far the formulation, and indeed even the content, of this fragment have been influenced by its Neoplatonic source, and in any case it scarcely suffices to support the mystical interpretation of Socrates' words. Nothing else even remotely similar is attested for Classical Greek literature, not even among the Pythagoreans, on whom see Burkert, W., Weisheit und Wissenschaft. Studien zu Pythagoras, Philolaos und Platon = Erlanger Beiträge zur Sprach- und Kunstwissenschaft 10 (Nürnberg, 1962), esp. pp. 98–142, pp. 271ffGoogle Scholar. Of course, in poetry those suffering intolerable pain can ask for death to come as a release from their torments (Θάνατος Φαιάν, , Aesch.Frg. 255Google Scholar, , Eur.Hipp. 1373Google Scholar): but Plato takes care to make clear that Socrates is not in pain, and anyway Socrates is expressing gratitude, not uttering a prayer (see below).

³⁷ Corpus Hermeticum 2, 6, 9, 14; Asclepius.

³⁸ Most of the late evidence, such as it is, is inconclusive. Thus Artemidorus 5.61 recounts that a man who dreamed that Asclepius killed him by striking him in the belly with a sword was later healed by abdominal surgery (but this is salvation in this life, not in the next one); at 5.13 a boy wrestler who dreams that he is rejected by Asclepius dies before the contest, ⋯ γàρ θεòς οὐ το⋯ ⋯γ⋯νος ⋯λλà το⋯ ζ⋯ν, οῧπερ εῗναι κριτής νομίζεται, ⋯ζέβαλεν αὐτόν (but the god of healing can certainly choose to refrain in hopeless cases, cf. e.g. Aristides, , Or. 28.132Google Scholar). , Orph.H. 37Google Scholar concludes ⋯λθέ, μάκαρ, σωτήρ, βιοτ⋯ς τέλος ⋯σθλòν ⋯πάζων (8); but cf. 3–4: μόλοις κατάγων ὑγίειαν / καì παύων νούσους, χαλλεπàς κ⋯ρας θανάτοιο. Porphyry, , Ad Marcellam 34Google Scholar (296.8–10) provides only a partial parallel: πολλάκις κόπτοπίτινα μέρη ⋯πì σωτηρία ‹τ⋯ν λοιπ⋯ν. σὺ; δ' ⋯πì σωτηρία› τ⋯ς ψυχ⋯ς ἒτοιμος ἒσο τò ὂλον σ⋯μα ⋯ποκόπτειν (296.8—10 Nauck, with Nauck's supplement). Proclus is said to have had a close, indeed intimate relation with Asclepius (Marinus, , Vita Prodi 29–31Google Scholar): yet he used it to save a girl from a deadly illness, and when he himself was dying he had to combat, by the strength of his own desire for death, Asclepius' attempts to save his life. Even Plotinus' last words (Porphyry, , Vita Plotini 2.25–7Google Scholar), which used until recently to be interpreted as a statement that Plotinus in his death was trying to bring the divine in him up to the divine in the universe, are nowadays read differently, as an admonition to those interested in philosophy to try during their lives to bring up the divine in them to the divine in the universe (precisely what Plotinus, who is dying, can no longer do): see especially Schwyzer, H.-R., ‘Plotins letztes Wort’, MH 33 (1976), 85–97Google Scholar, who proposes the text φήσας “πειρ⋯σθε τòν ⋯ν ⋯μῖν θεòν ⋯νάγειν πρòς τò ⋯ντῷ παντί θεῖον”, cf. Henry, P., ‘La Dernière Parole de Plotin’, SCO 2 (1953), 113–30Google Scholar, and Igal, J., ‘Una nueva interpretación de la últimas palabras de Plotino’, Cuademos de Filologia Clásica 4 (1972), 441–62Google Scholar.

³⁹ See in general Edelstein, and Edelstein, , op. cit., ii. 125–38Google Scholar.

⁴⁰ This point is recognized by Gallop, D., Plato: Phaedo (Oxford, 1975), p. 225ad 118a7–8 (who also suggests, with somewhat less cogency, that the mystic interpretation is incompatible with Socrates' argument in 90e2–91al); andGoogle ScholarMinadeo, , op. cit, p. 294Google Scholar.

⁴¹ Simmias is permitted to utter a residual doubt (107a9–b3), but only so that Socrates can respond with a myth of the after-life (107c–l 14d).

⁴² So e.g. Archer-Hind, loc. cit; Clark, P. M., ‘A Cock to Asclepius’, CR n.s. 2 (1952), 146Google Scholar; Cumont, , op. cit., pp. 121–4Google Scholar; Schmidt, , op. cit., p. 6Google Scholar.

⁴³ Damascius, , loc. cit., pp. 285 (ïνα τà νενοσκòτα τ⋯ς φυχ⋯ς ⋯ν τῇ γενέσει τα⋯τα ⋯ξιάσηται), 371 (ὃτι Παιωνίου δεῖται προνοίας ⋯ φυχή, ⋯παλλαττομένητ⋯ν πολλ⋯ν πόνων). So tooGoogle ScholarPrudentius, , Apotheosis 204–6Google Scholar: quamuis promittere et ipsi / gallinam soleant aut gallum, clinicus ut se / dignetur praestare deus morientibus aequum.

⁴⁴ The verb recurs in a similar context elsewhere in Plato (μηδ' αὖ όφεiλοντα ἢ θεῷ θυσίαςτωàς ἢ ⋯νθρώπῳ; χρήματα, Rep. 1.331b) and in Callimachus (τò χρέος ὡς ⋯πέχεις, 'Aσκληπιέ,τò πρò γυναικός / Δημοδίκης 'Aκέσων ⋯φελεν αὐξάμενος, / γιγνώσκειν, Ep. 54.1–3 Pf.); cf. also , Soph.Ant. 331Google Scholar, Theocr. 2.130. Hence a future obligation would derive from a hypothesized future-perfect benefit: ⋯στε τῷ 'Aπόλλωνι ῷλλα μοι δοκ⋯ χαρισтήρια ⋯φειλήσειν, Xen. Cyrop. 7. 2. 28.

⁴⁵ Cf. Burkert, W., Greek Religion, trans. Raffan, J. (Cambridge, MA, 1985), pp. 68–70Google Scholar; Rouse, , op. cit., pp. 204fGoogle Scholar. on thank-offerings to Asclepius, and cf. pp. 97, 191, 203, 350–1, and passim.

⁴⁶ At Artemidoru s 5.66, a dream is said to have meant δεῖν φυλάττεσθαι καì θύειν ⋯ποτρόπαια τῷ θεῷ: δεῖν, not ⋯φείλειν.

⁴⁷ IG iv², 1.126 (c. 160 A.D.). Cf. e.g. IG 14.967a and b (II—III A.D.); Herodas, , Mint. 4.15–18Google Scholar; Suda A.3893 s.v. 'Aρ⋯σταρχος (1.352.1–3 Adler = Aelian Frg. 101 Hercher); and Edelstein, and Edelstein, , op. cit., 1.294–306Google Scholar, T. 520–45. Indeed, a man who sacrificed before he had been healed could even come to seem suspect: Philostratus, , Vita Apollonii 1.10Google Scholar.

⁴⁸ Indeed, Reale, loc. cit., is even driven to the desperate suggestion, ‘Si noti come queste parole sembrino venire dall'oltretomba.…Platone fa dare a Socrate, come ultimo messaggio, la conferma che quanto aveva sempre sostenuto era esatto, ossia che stava passando alia nuova e vera vita’.

⁴⁹ Wilamowitz, , op. cit., i. 178Google Scholar n. 1, ii. 57–9.

⁵⁰ Lactantius, Div. Inst. 3.20.16–17, Inst. Epit. 32.4–5; Theodoret, , Graec. aff. cur. 7.47Google Scholar; Damascius, , op. cit., p. 371Google Scholar; Socratis et Socraticorum Epistola 14.9 (621 Hercher); Scholia ad Lucian Bis accus. 5 (138.14–17 Rabe); Suda σ.829 S.V. Σωκράτης (4.405.11–12 Adler).

⁵¹ e.g. Clark, loc. cit.; Gallop, loc. cit.; Geffcken, J., Griechische Literaturgeschichte, 2 (Heidelberg, 1934), pp. 88–9 n. 92Google Scholar; Grote, G., Plato, and the Other Companions of Socrates³ London, 1852), 2.195 n. d.Google Scholar; Hackforth, , op. cit., p. 190 n. 2Google Scholar; Wilamowitz, loc. cit.

⁵² Plato, , Apol. 28eGoogle Scholar, Loch. 181b, Symp. 221a–b.

^{page 53} Socratis et Socraticorum Epistola 14.9 (621 Hercher).

⁵⁴ Cf. Wilamowitz, , op. cit., ii. 58 n. 1Google Scholar.

⁵⁵ See especially Symp. 221b.

⁵⁶ Op. cit., i.178 n. 1, ii.58.

⁵⁷ So e.g. the ancient authors listed in n. 50 (except Socratis et Socraticorum Epistola 14.9), and among modern scholars Gallop, loc. cit., and Hackforth, , op. cit., p. 190 n. 2Google Scholar.

⁵⁸ Exceptions: e.g. Cumont, , op. cit., pp. 121–2Google Scholar; Dixsaut, , op. cit., pp. 180 and 408–9 n. 382Google Scholar.

⁵⁹ So Gallop, loc. cit., referring to 116d4. But in Greek prose this construction is relatively rare: cf. Kühner-Gerth, , op. cit., ii. 1, pp. 83–4Google Scholar; Schwyzer, E., Griechische Grammatik, ii. Syntax und syntaktische Stilistik ⁴ (Munich, 1975), pp. 243–4Google Scholar. And in fact it seems best to take the words ⋯μεῖς τα⋯τα ποιήσομεν (116d4) too as a genuine plural referring to Socrates and his friends: for he goes on in his very next words to address his disciples (καì ἂμα πρòς ⋯μ⋯ς,ibid.) and to include them in his first-person plural exhortation to obey (πειθώμεθα, d8).

⁶⁰ ἲσθι (115e4); χαῖρε (116d4); ἲθι, πείθου, ποίει (117a3).

⁶¹ ⋯γγυήσασθε (115d6, 9); ἂγεтε, καρτερεῖτε (117e2).

⁶² ποιήσομεν (116d4: see above, n. 59); πειθώμεθα (116d8).

⁶⁸ This, if nothing else, would suffice to preclude the interpretation that Socrates was speaking ironically (see above, n. 21): the notion that Socrates might have been indulging at this moment in a piece of whimsy of this sort might, just conceivably, be tolerable if Socrates used the singular forms and thereby indicated that the vow regarded himself personally; with the plural forms, however, and the implication that it affects the group of himself and his disciples as a whole, it can no longer be maintained.

⁶⁴ M Clark, op. cit.; but her views and mine diverge on a number of other points involving this identification. Cf. also Burger, R., The Phaedo: A Platonic Labyrinth (New Haven and London, 1984), p. 216Google Scholar. I have not seen Rouse, W. H. D., Great Dialogues of Plato (New York, 1956), p. 521, who seems to have made the same proposalGoogle Scholar(cited by Baron, , op. cit., p. 269 n. 5Google Scholar).

⁶⁵ Though scepticism is sometimes expressed concerning whether Plato was really sick or not (so even recently, e.g. Guthrie, W. K. C., A History of Greek Philosophy, 3. The Fifth-Century Enlightenment (Cambridge, 1969), p. 489 n. 2)Google Scholar, Wilamowitz is probably right in arguing that the illness was genuine (op. cit., i. 325 n. 1). If Plato invented his illness for the purposes of this dialogue and in fact was absent for some other reason, the probability attaching to the central thesis of this paper is enormously increased. But for the present argument what matters is not whether Plato's illness was real or fictional, but only the fact that it exists as a discursive reality within the Phaedo and hence forms part of the interpretative textual context for Socrates' last words.

⁶⁶ All other references in the dialogue to sickness are general and unspecific: cf. 66cl, 83cl, 86c4, 105c3–4, 110e6.

⁶⁷ Plato himself, of course, knew perfectly whether he was ill or not at the time of Socrates' death. If he has Phaedo imply some uncertainty here, this is probably not because in fact there was some uncertainty on this score, but because it is dramatically appropriate for a messenger to restrict his expressions of certainty to those matters to which he can attest on the basis of personal observation: thereby our impression of his veracity concerning all other matters is strengthened. So too, explaining the absence of Aristippus and Cleombrotus, Phaedo says, ⋯ν Aἰγίνῃ γàρ ⋯λέγονтο εἶναι (59c4), and here too there is no reason to subscribe to the ancient view that Plato is expressing an implied criticism of them (, Dem.De Elocut. 288Google Scholar). A further, interesting suggestion is made by the anonymous referee for this journal: ‘Phaedo's vagueness about Plato's illness guarantees that he himself will have missed the real meaning of Socrates' last words, and thus reassures us that the allusion to Plato (when we eventually grasp it) is not Phaedo's own imposition on the story, but part of the facts he unwittingly reports’.

⁶⁸ This was already recognized as a peculiarity of Plato's works in antiquity: see , Diog. Laert. 3.37Google Scholar.

⁶⁹ So too, both Socrates' first words in this dialogue (60a7–8) and his last words are, imperatives addressed to Crito, directing him to ensure that others carry out Socrates' wishes; and the Athenians' pious observation of their vow to Apollo at the beginning (S8bl–4) may correspond to Socrates' pious observation of a vow to Apollo's son Asclepius at the end. For other instances of ring-composition between these two sections of the dialogue, see above, nn. 8 and 13.

⁷⁰ Re, M. Del, op. cit., p. 290Google Scholar; Re, R. Del, op. cit., p. 85Google Scholar.

⁷¹ The Greeks tended not to distinguish, as we do, between prophecy directed to events later in time and clairvoyancy directed to events distant in space. Archaic Greek prophets are said to know τά τ' ⋯όντα τά τ' ⋯σσόμενα πρό τ' ⋯όντα (Horn. Il. 1.70, , Hes.Th. 38Google Scholar, etc.); in Herodotus, Apollo at Delphi proves his mantic powers by revealing what Croesus is cooking at that very moment in Sardis (Hdt. 1.46–8).

⁷² Il. 16.843–54, 22.355–60. Cf. Schol. ad Il. 16.851–54, 854a, 22.359–60a; Eust. 1089.60 ad Il. 16.851–4; and e.g. , Virg.Aen. 10.739–41Google Scholar, Servius, ad Aen. 10.740Google Scholar, Servius, (auctus) ad Aen. 4.613Google Scholar.

⁷³ Cf. also e.g. Xen. Cyrop. 8. 7. 21; Aristot. Frg. 10 Rose³.

⁷⁴ Other possible relations, of varying degrees of plausibility, between these two texts – the only ones in which Plato names himself – are proposed by Patzer, A., ‘Platons Selbsterwähnungen’, Würzburger Jahrbücher N.F. 6b (1980), 21–7Google Scholar.

⁷⁵ Cf. also , Xen.Apol. 30Google Scholar.

⁷⁶ Admittedly, the primary reference at Phd. 85b4–7 is to Socrates' preceding arguments concerning the fate of the soul, not to any words he might come to utter at the very point of death. But his dying words are even more truly his swan-song than those arguments; and whether his arguments were correct we shall not discover until after we have died ourselves, whereas the truth of his vision of Plato's recovery is immediately apparent. The veracity of the one prophecy may serve to make the other one seem more authoritative.

⁷⁷ There was a well-established connection between mantic visions and (often miraculous) medical cures in antiquity, cf. Brelich, A., Gli eroi greci: un problema storico-religioso (Rome, 1978), pp. 106–13, 113–18Google Scholar.

⁷⁸ Herodas 4.11–18, Libanius, , Orat. 34.36Google Scholar; cf. Headlam, W., Herodas: The Mimes and Fragments, ed. Knox, A. D. (Cambridge, 1922), p. 180Google Scholarad 4.16, and Rouse, , Offerings, op. cit., pp. 204, 297Google Scholar. So too, in the Issenheimer altar now in Colmar, Nicolas of Hagenau has sculpted within the altarpiece on the left a peasant who offers a cock, on the right a rustic gentleman who offers a piglet. Cocks are often associated with Asclepius anyway, see Artemidorus 5.9, Libanius, , Orat. 34.36Google Scholar, Suda A.4177 s.v. 'Aσκωλιάζοντες (1–385.19–23 Adler = Aelian Frg. 98 Hercher), A.1117 s.v. ⋯λεκτρυόνα (1.101.4–7 Adler), IG iv², 1.41 (c. 400 B.C.); and cf. Orth, F., ‘Huhn’, RE 8.2 (Stuttgart, 1913), p. 2533Google Scholar, and Cumont, , op. cit., pp. 122Google Scholar and n. 1, 124–6.

⁷⁹ See e.g. , Xen.Mem. 1.2.1, 4, 3.5ff., 5.1ffGoogle Scholar., etc.

⁸⁰ So too, various ancient anecdotes about Plato record his miraculous legitimation as Socrates' successor: see Riginos, A. S., Platonica: The Anecdotes Concerning the Life and Writings of Plato (Leiden, 1976), Anecdote 4, pp. 21–4Google Scholar. This appears to be the suggestion also of Burger, loc. cit.

⁸¹ See Derrida, J., ‘La Pharmacie de Platon’, in La Dissémination (Paris, 1972), pp. 69–197, especially 137–46, 170, 177, 189, 194–5Google Scholar.

⁸² It may be of interest in this connection that there is a certain tendency in Late Antiquity to view Asclepius and Plato as brothers and counterparts: Diog. Laert. 3.45 (= Anth. Pal. 7.109.1–2); Olympiodorus, , Vita Platonis 2.164–7Google Scholar, in Westerink, L. G. (ed.), Olympiodorus: Commentary on the First Alcibiades of Plato (Amsterdam, 1956), p. 6Google Scholar; , Anon.Vita Platonis 6.14–16Google Scholar, in Westerink, L. G. (ed.), Anonymous Prolegomena to Platonic Philosophy (Amsterdam, 1962), pp. 13–15Google Scholar; and cf. Riginos, , op. cit., Anecdote 9, pp. 28–9Google Scholar.

⁸⁸ 325c5–dl: σκοπο⋯ντι δή μοι ταῡτά τε καì τοὺς πράττονταςτà πολιτικά, καì τοὺς νόμους γε καì ἒθη, ὂσῳ μ⋯λλον διεσκόπουν ⋯λικίας τε εἰς τò προὒβαινον, τοσούτῳ χαλεπὠτερον ⋯φαίνεтο ⋯ρθ⋯ς εἶναί μοι τά πολιτικà διοικεῖν. This is not die place to attempt to decide whether this letter really is by Plato or not.

⁸⁴ Cherniss, H., ‘Plato (1950–1957)’, Lustrum 4 (1959), 5–308Google Scholar, here 132, dismisses Clark's solution (see above, n. 42) as ‘tasteless’. Certainly de gustibus non est disputandum: but the suggestion that, in his final moments, Socrates was not congratulating himself upon his own rescue but thanking the god for his friend's, may well be defended as more tasteful rather than less so.

⁸⁵ But in this connection the striking alternation of veiling and unveiling in this scene is perhaps worth noting. It was common practice in antiquity to cover the face of the dead. But it also formed part of a mystic initiation to veil the face of the initiand who was purified in preparation for the mystic vision, see e.g. Burkert, W., Ancient Mystery Cults (Cambridge, M A and London, 1987), pp. 94ffGoogle Scholar. It may also be noted in passing that the cock too had important symbolic meaning in the ancient mysteries: see Cumont, op. cit.

⁸⁶ So too in the deaths of Patroclus and Hector in Homer (see above, n. 72), the process of death is described realistically, the final words are simply reported as such, and their prophetic character is not mentioned in any way in the text itself but is left instead to be discovered by the reader's inference from his knowledge of the epic tradition.

⁸⁷ My suggestion that Plato might be trying to legitimate himself as Socrates' philosophical heir can perhaps receive further support from indications that he may have been facing competition on this score. T. D. Barnes suggests to me that the fact that Plato has Crito ask Socrates if he has anything else to say (118a9–10), but then has Socrates die without answering the question, might be intended as an implicit refutation of other alleged “last words” of Socrates currently in circulation. And A. A. Long points out that in a passage in Xenophon's Symposium, probably written a few years after the Phaedo, Socrates seems to designate Antisthenes as his philosophical heir, at least in certain regards (4.16). Many anecdotes concerning the mutual enmity betwen Plato and Antisthenes were told in antiquity (cf. Riginos, , op. cit., Anecdotes 43–48 and 103, pp. 98–101Google Scholar and 147–8): if they had any basis in fact, they may have arisen from rivalry for Socrates' philosophical succession.