¹ This sequence of lines appears at the end of Alcestis, Andromache, Helen, and Bacchae; in Medea only the first line is different, appearing as ‘ Zeus in Olympus is steward of many things’ ().

² Exceptions to the anapaestic rule are Sophocles‘ Oedipus Tyrannus (trochaic tetrameters), Euripides' Ion (trochaic tetrameters), and Euripides' Trojan Women(lyric iambics). The closing lines of Sophocles' Trachiniae are assigned by some MSS to the chorus, by others to Hyllus, and a scholium on Oedipus Tyrannus 1523 attributes the last lines of the play to Oedipus.

³ There has been to my knowledge only one attempt at a complete account of all closing lines in Greek tragedy, Mayerhoefer, F.'s Uber die Schliisse der erhaltenen griechischen Tragodien(Diss. Erlangen, 1908)Google Scholar. Mayerhoefer begins by discussing some issues of genuineness, notes the increasing standardization of closing lines after Aeschylus, and categorizes the passages as Bemerkungen ad hoc, Sentenzschliisse, and Schematische Schliisse (Euripides' repeated endings). He concludes that we cannot expect to find in these endings the Idee of the drama. Kremer, G.'s ‘Die Struktur des Tragodienschlusses’ (Die Bauformen der griechischen Tragodie, ed. W. Jens, Munich, 1971, pp. 117–41)takes almost no account of the place of theconcluding lines themselves in the structure of the play's conclusion. An early work by U. von Wilamowitz-Moellendorff, In wieweil befriedigen die Schliisse der erhaltenen griechischen TrauerspieleGoogle Scholar, has recently been edited with introduction and notes by W. M. Calder III (Leiden, 1974). This long essay, Wilamowitz's final school paper, is concerned rather with the outcome of the action of plays than with their final lines or scenes. R. Kannicht briefly discusses the genuineness of the Sophoclean and Euripidean endings in his commentary on Euripides' Helen (Heidelberg, 1969) ad 1688–92. For other treatments of the issue of these endings see works cited in notes 4–10, 14, 15, 22 below; a recent dissertation on Euripidean endings by F. Dunn (Yale, 1985) was not yet available as I was completing this essay. I omit here comments on specific endings, except in some instances where these lead to more general remarks.

⁴ Bowra, C. M.,Sophoclean Tragedy(Oxford, 1944), pp.9–10.Google Scholar

⁵ Burton, R. W.,The Chorus in Sophocles' Tragedies(Oxford,1980), p.184. I here adopt his term ‘coda’.Google Scholar

⁶ I take the term ‘closure’ (as applied to the ending of a literary work) from Smith, B. H.'s Poetic Closure(Chicago and London, 1968), which, likeGoogle ScholarKermode, F.'s The Sense of an Ending (New York and Oxford, 1966), has been particularly influential in the recent rise of critical interest in problems of closure. Among the articles on ancient literature that acknowledge a debt to Smith areGoogle ScholarSchrijvers, P. H., ‘Comment Terminer une Ode?’, Mnemosyne 4(1973), 140–59CrossRefGoogle Scholar, Nagle, B. R., ‘Open-ended Closure in AeneidT, CW 16(1983),257–63Google Scholar, Reeve, M. D., ‘Tibullus 2.6’, Phoenix 38(1984),235–9CrossRefGoogle Scholar, and Santirocco, M., “The Poetics of Closure: Horace Odes III. 17–28’,Ramus 13(1984),74–91.CrossRefGoogle Scholar

⁷ Philologus17 (1861), 422–36.Google Scholar

⁸ Barrett, W. S.(ed. and comm.), Euripides' Hippolytos(Oxford,1964)Google Scholar, commentary ad 1462–6. For a more balanced consideration, see Kannicht's note on Helen 1688–92 (op. cit., n. 3 above); cf. also Rees, B. R., ‘Euripides, “Medea,” 1415–19’, AJP 82(1961),176–81Google Scholar, and Lloyd-Jones, H.'sreview of Barrett's commentary,JHS 85(1965),171.Google Scholar

⁹ Both Ritter and Barrett envision interpolation at a fairly early date. Ritter (op. cit., n. 7) imagines a post-Euripidean, pre-Alexandrian poet of weak ability who adds the Sophoclean endings for new productions, and Barrett (op. cit., n. 8), less cautious in this respect than Page, holds that ‘the most likely culprits [for the repeated Euripidean codas] are the actors’, appealing to the taste of a ‘public addicted to sententious commonplaces’. The lines that appear in the repeated coda of Alcestis and the others are found as early as a papyrus of the third century B.C.E. (see Barrett ad loc); references to the codas of plays occasionally appear in the scholia.

¹⁰ R.Dawe,Studies on the Text of Sophocles,1(Leiden, 1973), pp. 174–5, 203–4, 266–73, has recently given some support to Ritter's doubts on Sophocles’ Ajax and Electra, but the only Sophoclean coda he deletes without hesitation is the widely distrusted conclusion of Oedipus Tyrannus, where the awkwardness is particularly pronounced and where there are other difficulties as well (see n. 12 below). Dawe raises the possibility that in this play we have lost the ending referred to by the scholium (see n. 2) and that our ending is quite late op. cit., pp. 267, 273, and cf. his edition with commentary (Oxford, 1982). Kannicht (op. cit., n. 3), shares Barrett's doubts about style and usage in the codas of Hippolytus and Iphigenia in Tauris, but see Lloyd-Jones (op. cit., n. 8) for a defence of the coda of Hippolytus. The coda of Iphigenia in Aulis is rejected by a number of scholars as part of a final scene they consider an interpolation.Google Scholar

¹¹ Ritter, op. cit. (n. 7), 422. He supports this general argument with others which are more particular; perhaps the most noteworthy is that in several cases the final lines are directly addressed to the audience and are therefore suited to comic rather than to tragic choruses (p. 428). For further comments on this claim and on his point about the chorus see pp. 61–3 below.

¹² It is clear, of course, that the final lines of tragedies, especially where those lines are loosely connected with the preceding action, would have been easily tampered with, and there is some evidence that points to such tampering, although even Page, D. L.calls it ‘confessedly meagre’ (Actors‘ Interpolations in Greek Tragedy[Oxford,1934], p.95). Page accepts two such pieces of evidence as ’perhaps traces of a common practice‘ of transference of lines, not necessarily by actors. (1) The closing lines of Oedipus Tyrannus bear a close resemblance to Phoenissae 1758–63 (lines immediately preceding the coda), and many have taken this resemblance to suggest interpolation in one or both plays. (2) In two MSS of Hippolytus, we find added to the usual ending the lines that conclude Phoenissae, Orestes, and Iphigenia in Tauris (see Barrett's apparatus ad 1466a-c). The first piece of evidence suggests that we should be wary of assuming that we know the endings of Oedipus Tyrannus or Phoenissae (whose entire closing scene is problematic); but if the similar lines belonged originally to either play, they can still tell us something about the characteristics of fifth-century endings. And although the second piece of evidence may support the widespread view that Euripides' repeated codas were sometimes added to plays to which they did not originally belong, we need not assume that the coda in question was not Euripides'. (See p. 57 for discussion of a very different type of interpolation, one that is unmistakably an addition to the text without any possibility of being a displaced but genuine coda.)Google Scholar

¹³ There is clear evidence for the use of formulaic closing lines in Menander's comedies; see Handley, E. W.'s note on 968ff. in his edition with commentary of the Dyscolus(Cambridge, Mass., 1965). Given Barrett's reaction to Euripidean repetition, it is worth noting Handley's comment on the presence of such lines at the ends of several of Menander's plays and of something very similar in another playwright: ‘…although we do not know that [Menander] invented the formula, the fact that he uses it repeatedly suggests so‘.Google Scholar

¹⁴ Cf. Barrett's criticism of the staging argument (op. cit., n. 8); see also Rees (op. cit., n. 8), who argues that the use of the codas to accompany the departure of the chorus need not rid them of all interpretive significance, and suggests a parallel with liturgical practice at the end of an Anglican church service, where recessional hymn and organ voluntary are ‘not to be regarded as a mere device or stopgap…’ (p. 178).Google Scholar

¹⁵ Hermann, G., ed. and comm., Euripides‘Bacchae(Leipzig,1823) ad 1383. Hermann's comment is so often referred to that it should perhaps be quoted in full: ’Qui factum sit, ut Euripides quinque fabulas iisdem versibus flnierit, non memini me a quoquam interpretum indicatum legisse. Scilicet, ut fit in theatris, ubi actorum partes ad finem deductae essent, tantus erat surgentium atque abeuntium strepitus, ut quae chorus in exitu fabulae recitare solebat, vix exaudiri possent. Eo factum, ut illis chori versibus parum curae impenderetur.‘ Hermann is speaking specifically of the repeated codas here. The comment on Oedipus Tyrannus 1524–30 in his edition (Leipzig, 1833) is evidence enough that he did not consider all codas negligible: ’Perinepta est scholiastae ad praecedentem versum adnotatio:. Nam fine careret fabula, nisi aut chorus aut Oedipus aliquid adiiceret, ut quo tenderent ista omnia, quae in scena acta sunt, intelligeretur.‘ Rees, op. cit. (n. 8), 178, asks why the audience should be departing at all before the end of a trilogy. We do not know how long an interval there was between plays, or between trilogy and satyr play, but it seems improbable that the audience departed en masse after each play (although the often-cited Birds 78592, which offers the audience the prospect of a quick flight home from the theatre during the tragedies, does suggest a certain restlessness).Google Scholar

¹⁶ There is a modern example of what might be called the deliberately insignificant close: the portion of a film, usually still but sometimes in motion, that serves as background for the final credits. Most film-goers rise and make their way out as soon as the credits appear (Rees, op. cit. (n. 8), 178, notes the resemblance to Hermann's imagined audience) and cinema-owners reinforce this behaviour and the attitude it entails by bringing up the lights and opening the doors (to the irritation of the dedicated). But the film that continues behind the credits is not a true analogue to our codas, since what is shown is normally only a pictorial backdrop to information that comments not on the action itself but on the mimesis. Furthermore, if it should come to be more common to show a surprise bit of interesting action while the credits roll, as some recent film-makers have done, audiences will presumably become more attentive, and film-makers will begin to count on this attentiveness.

¹⁷ Hester, D. A.,‘Very Much the Safest Plan, or Last Words in Sophocles’,Antichthon 1 (1973),8–13. W. B. Stanford cites without reference a similar view of the conventional endings ‘as a deliberate device for easing the transition of the audience's mind from the world of the play back to real life’ in his commentary on Ajax (London and New York, 1963), ad 1418–20.CrossRefGoogle Scholar

¹⁸ Op. cit. (n. 17), 9.

¹⁹ Cf. Bowra, op. cit. (n. 4), p. 9, and Hester, op. cit. (n. 17), 9.

²⁰ See for example Smith, op. cit. (n. 6), pp. 10–14, 36–7, 119, 212, 218, on what she calls ‘retrospective patterning’.

²¹ Aristotle, Poetics 6, 1450a38–9.

²² Cf. for exampleDodds, E. R.'s commentary on Euripides' Bacchae(2nd ed., Oxford,1960) ad 1388–92: ‘A modern producer would bring down the curtain on 1387, but a Greek dramatist had to get his Chorus out of the orchestra.’Google Scholar

²³ The idea of process is important here. The coda is not as abrupt as the curtain, and the lines that precede the coda also as a rule contain indications of the approaching end (see pp. 58–9, 61 below).

²⁴ Dawe, op. cit. (n. 9), 1, p. 118 and 2, p. 50, notes the presence of this couplet (constructed in part from Ajax 41) in N and O; it is followed in O by two unmetrical lines. M. Haslam discusses it in his article ‘O Suitably-Attired-in-Leather-Boots: Interpolations in Greek Tragedy’, in Bowersock, G. W., W. Burkert, and Putnam, M. C. J., eds., Arktouros: Hellenic Studies Presented to Bernard M. W. Knox (Berlin, 1979), pp.91–100.CrossRefGoogle Scholar

²⁵ There are of course passages that refer to the preceding action at the ends of tragedies, but only one of these (Cho. 1065ff.) really approaches a summary, and it is clearly a descriptive comment on recent events by someone involved in them, whereas the lines in question here sound like the words of a narrator. See Tarrant, R. J.'s commentary on Seneca's Agamemnon (Cambridge,1976)ad 1006ff.; it is interesting that in the summary passage Tarrant is discussing, Cassandra is declaring her eagerness to narrate recent events to the Trojans who have died before her.Google Scholar

²⁶ Fr. 590 in Pearson, A. C.,The Fragments of Sophocles,2(Cambridge, 1917)Google Scholar and Radt, S.,Tragicorum Graecorum Fragmenta, 4 (Gottingen, 1977). The first line as here given is Grotius' emendation for the unmetrical .Google Scholar

²⁷ The form and content of Nauck 446 suggest that it too is the coda of the play to which it belongs, Euripides' earlier Hippolytus:

²⁸ I do not mean that the features cited here belong to mutually exclusive categories of coda; they are often combined. Farewells or references to departure (in which the verb is particularly common) are to be found in Sophocles' Trachiniaeand Philoctetesand in Euripides' Hecuba, Heraclidae, Suppliants, Trojan Women, Heracles, Electro, Ion, Iphigenia in Aulisand Rhesus;commands, prayers or wishes in Sophocles' Trachiniae, Oedipus Tyrannus, Philoctetesand Oedipus at Colonusand in Euripides' Heraclidae, Hecuba, Trojan Women, Iphigenia in Tauris, Phoenissae, Orestesand Rhesus.Lamentations occur in Euripides' Hippolytus, Heraclesand Trojan Women,and emphasis on finality, authority, or necessity in Sophocles' Trachiniaeand Oedipus at Colonusand in Euripides' Hecuba.Gnomic statements are to be found in the codas of Sophocles' Ajax, Antigone, Oedipus Tyrannusand Tereus,in the shared conclusion of Euripides' Alcestis, Medea, Andromache, Helenand Bacchae,and in his Hippolytus,his lost earlier Hippolytus, Electraand Ion.

²⁹ Interestingly, some of these elements are also prominent, combined with certain formal features of endings, in what critics have described as false or premature endings - for example, in Antigone's anapaests at Antigone 937–942 and in the tetrameter exchange between Neoptolemus and Philoctetes at Philoctetes 1402–1408. On the Antigone, see Kremer, op. cit. (n. 3), p. 132; for a discussion, with examples, of’false endings’ in general, see Taplin, O.,Stagecraft in Aeschylus (Oxford,1977), pp. 180–4, 327. (Smith, op. cit. (n. 6), p. 212, uses the term ’false closure’ in a different sense.)Google Scholar

³⁰ In the discussion that follows I am especially indebted to Smith's fourth chapter (‘Special Terminal Features’) particularly the sections ‘ Closural Allusions’,‘ Unqualified Assertion’, and ‘The Poetic Coda’ (op. cit. (n. 6), pp. 172–95.) Smith's subject is closure in lyric, not in narrative, and although she occasionally notes parallels, she stresses the difference between the two literary types: ’Whereas the structure of a play or novel is related to the structure of events, the structure of a lyric poem is related to that of personal discourse…’ (p. 122). In the codas of our tragedies, however, we are clearly dealing not so much with the end of the action, since the structure of events is already complete, as with the end of the play on the level of speech, and we might therefore expect the presence of devices like those Smith mentions.

³¹ On informal conversational closure, see Smith (op. cit. (n. 6), p. 187).

³² For a brief comment on the relation between convention and significance in the references to funerals that end Shakespearean tragedies, see Brower, R.'s introduction to the Signet edition of Coriolanus(New York,1966). On our need to punctuate our lives with indications of endings, see especially Kermode (op. cit., n. 6), passim.Google Scholar

³³ The difficulty we have with the gnomic is of course not confined to the endings of works, nor to the Greek theatre. A memorable attempt to cope with the presence of gnomic utterances may be seen in Ingmar Bergman's film of Mozart's Magic Flute. Whenever a piece of proverbial wisdom appears in a song, the director has his characters hold up large signs with the words calligraphically inscribed - thus, I take it, making sure the audience will see these sayings as analogous to mottos on samplers or shields, and not confuse them with ‘ the meaning of the action’. See also n. 36 below.

³⁴ Burnett, A. P.,Catastrophe Survived(Oxford,1971), p.4. Burnett is not here speaking specifically of the codas, but in arguing that the playwrights worked in various ways against ‘the excessive unity of the discrete tragedy’ she remarks that they ‘would from time to time dwell expressly on the theme of the whirl of fate, as if to remind us that the rigid finale to come was only an illusion’. The ‘whirl of fate’ is in fact a frequent theme in the conclusion itself.Google Scholar

³⁵ See Electra 1497–1500, Oedipus at Colonus1769–72. On such Sophoclean allusions and their force, see, in addition to the commentatorsGoogle Scholar, Easterling, P. E.,‘Philoctetes and Modern Criticism’, ICS 6(1981),39Google Scholar, Taplin, O.,‘Sophocles in his Theatre’, Sophocle, Entretiens Hardt 29(Geneva, 1983),155–74Google Scholar, and Winnington-Ingram, R. W.,Sophocles: An Interpretation(Cambridge,1980), esp. p.302and n. 70.1 have dealt with the subject at some length in another paper, ‘Sophoclean Endings: Another Story’, forthcoming in Arethusa.CrossRefGoogle Scholar

³⁶ Critics' difficulties with Webster's general use of gnomic material or sententiae seem to be analogous to the difficulties of Classical scholars with the gnomic endings of tragedy.Berry, R., The Art of John Webster(Oxford,1972)Google Scholar, cites Jack‘The Case of John Webster’, Scrutiny 16(1949),39: ’…this background of moral doctrine has nothing to do with the action of the plays: so far from growing out of the action, it has all the marks of having been superimposed by the poet in a cooler, less creative mood than that in which the Duchess and Flaminio had their birth. There is no correspondence between the axioms and the life presented in the drama. This dissociation is the fundamental flaw in Webster [italics mine].’ Note that the material in question is being treated as a kind of interpolation by the author himself. Berry's defence of the artistry of Webster's use ofsententiae (see esp. pp. 28–31,77–78) includes an interesting comparison with the Greek chorus.Google Scholar

³⁷ Note the marker of the approaching end of this paper.

³⁸ Even if the ending of Seven Against Thebes as we have it is not by Aeschylus, it is more than probable that the play ended with the words of the chorus, since the proposed excisions either exclude Antigone and Ismene altogether or reduce them to minor voices. The chorus that speaks last in Eumenidesis not the chorus of Erinyes that has been present throughout, but this secondary chorus of celebrants does accompany (in both senses) the original chorus in its departure. On Aeschylus' final processions see Taplin, op. cit.(n. 29), pp.127–8,411–15. The standard choral anapaests of later tragedy replicate these processions in miniature.

³⁹ The observable facts about the chorus's role in tragic endings have been explained (or explained away) differently by others. Ritter, as noted above, op. cit. (n. 7), 422, argues that the diminished role of the chorus in Sophocles and Euripides, particularly in prologue and exodus, means that their plays should not (and in fact do not) end with the chorus. Mayerhoefer (op. cit. (n. 3), pp. 43–5) takes the opposite point of view: in trying to explain why the use of choral endings becomes the rule as time goes on he argues that because the chorus as a general rule is more detached from the action in Sophocles and Euripides than in Aeschylus, it is particularly well suited to conclude the play. He thus sees (as I do) a particular appropriateness in the use of choral endings, although he sees it as deriving not from Aeschylean practice but from a move away from that practice. But his view (based largely on the Agamemnon)that choruses should only have the last word when they have not just previously been strongly involved is hard to support given the end of Suppliants(or indeed of Persians)..

⁴⁰ There are well-known difficulties with the term ‘dramatic illusion’; Bain, D. makes careful note of them in the first chapter of his book Actors and Audience: a Study of Asides and Related Conventions in Greek Drama(Oxford,1977), pp.1–12. We are not of course to assume that the usual audience is really under any serious illusion as to the reality of the events portrayed on stage, but Bain suggests that we may continue to use the term ‘illusion’ if we are conscious of its limitations. As he puts it elsewhere, ’All that I mean when I say that an actor preserves the illusion is that he pretends to be a character other than himself and that his pretence is accepted by the audience’ (‘Audience Address in Greek Tragedy’, CQ 25 [1975], 1). Bain, like Taplin (op. cit. (n. 29), esp. pp. 129–34, 394–5), holds that there is no real rupture of the illusion, so described, anywhere in Greek tragedy. Both scholars do note the existence of some features approximating such a rupture (see, for example, Bain [1975], p. 22 and Taplin, p. 133), but deny these any significant effect; the limitation of their treatment is thus its insensitivity to the possible effects of different degrees and kinds of dramatic self-referentiality. No one would claim that fifth-century tragedy does what fifth-century comedy does, but given that in general tragedy firmly maintains its own boundaries, Euripides' occasional forays are of particular interest.Google Scholar

⁴¹ Op. cit. (n. 7), 428.

⁴² Bain [1975] (op. cit. (n. 40), p. 22) grants partial recognition to this effect.

⁴³ The closing lines of the Rhesusare: Compare Aristophanes' Acharnians, which ends with a victory song which is also motivated within the story.

⁴⁴ On the Orestes see esp. Zeitlin, F., 'The Closet of Masks: Role-Playing and Myth-making in the Orestes of Euripides’, Ramus 9(1980),51–77; on PhoenissaeCrossRefGoogle Scholar, Foley, H., Ritual Irony- Poetry and Sacrifice in Euripides(Ithaca and London, 1985), pp.106–46Google Scholar. On Euripides' general tendency to sophisticated play with what his audience knows or expects, cf. also Winnington-Ingram,'Euripides: Poietes Sophos’, Arethusa 2(1969),127–42.Google Scholar

⁴⁵ On Medea see Eliot, T. S.,‘ Seneca in Elizabethan Translation’, Selected Essays(New York, 1932, 2nd ed. 1950), p. 73Google Scholar, cited by Costa, C. D. N.in his commentary on Seneca's Medea(Oxford, 1973), ad loc.Google Scholar

⁴⁶ For a brief account of the debate, with recent bibliography, see E. Fantham's introduction to her edition with translation and commentary of Troades. She draws primarily on Zweierlein, O.'s defence of the recitation theory in Die Rezitationsdramen Senecas(Meisenheim am Glan, 1966), but her own final conclusion is that ‘only the readers would experience the plays as complete works’ (p. 48).Google Scholar