Published online by Cambridge University Press: 11 February 2009
A. H. Sommerstein has recently directed our attention away from the belaboured topic of the essential and original structure of Old Comedy to the more productive question of how the extant plays of Aristophanes are shaped. He begins with the question of the source of ‘the five-act principle, standard in Menandrian comedy’ (140). Correctly looking to the chorus as the key element in articulating a play's form, Sommerstein finds that the five-act format already dominates the shape of Old Comedy, although he argues that the number of acts ranges from four to seven and that only in the fourth century do the acts become about the same length. His analysis is largely correct, I think, but it is in danger of being ignored because his criteria are not objective and have not been applied systematically. Thus, B. Zimmermann has more recently described the structure of Aristophanic comedy as free
1 Sommerstein, A. H., ‘Act Division in Old Comedy’, BICS 31 (1984), 139–52.Google Scholar
2 ‘According to the criteria we have adopted, we find that Aristophanes’ fifth-century plays can be divided into a number of acts that ranges from four to seven but is most commonly five (in five plays out of nine) … The length of the first act varies only between 203 and 322 lines, and the last, except in Clouds, is always short (34 to 106 lines), but the length of intermediate acts varies very widely’ (150).
3 Zimmermann, B., ‘L'Organizzazione interna delle Commedie di Aristofane’, Dioniso 57 (1987), 49–64.Google Scholar
4 By ‘stasimon’ I mean strophic choral song.
5 See Taplin, O., The Stagecraft of Aeschylus (Oxford, 1977), pp. 110–12Google Scholar on empty stage (‘at a rough estimate an actor remains during one in three of all act-dividing songs’, p. 111)Google Scholar and pp. 291–4 on time-lapse (‘in the whole of surviving tragedy the only explicit time lapse is at A. Eum 234/5’, p. 291). Weissinger, R. T., A Study of Act Divisions in Classical Drama (Iowa Studies in Classical Philology 9 [1940]), p. 25Google Scholar gives the following statistics for empty stage: 4× during parodos, 17 during the first stasimon, 18 during the second, 18 during the third and 13 during the fourth. Since he is considering 33 tragedies, Taplin's one-third does not seem to be an exaggeration.
6 To take only the one virtually complete play, Dyskolos, we find empty stage with choral interlude four times and without choral interlude seven times. At least one of the latter, the rescue of Knemon (Dys. 665), involves the passage of time and there are several others which may as well (392, 521). By contrast only two of the four choral interludes involve time lapse (at 232 Daos fetches Gorgias; at 783 Sostratos talks to his father). It is probably significant that in both cases the time lapse is made obvious by the re-entry of the character who last exited. Thus, while a choral interlude always occurs with empty stage, the reverse is not at all true and, while time lapse sometimes occurs during a choral interlude, it does not always.
7 The equivalence in length of the acts of Dyskolos (232−194−193−164−186) makes such an assumption understandable, but we can see from the fragments of the Samia that its Act III (over 215 lines) was almost twice as long as its Act V (122). What Sommerstein says about the first four acts (150) applies as well to the fifth: the range in act-length decreases from Aristophanes' fifth-century plays (75–693) to his fourth-century plays (175–373) and in Menander is even smaller (122–280).
8 Hunter, R. L., ‘The Comic Chorus in the Fourth Century’, ZPE 36 (1979), 23–38, pp. 24–5Google Scholar. Sommerstein argues, against Hunter, that there is no time lapse here, which only shows how subjective this criterion is.
9 Gelzer says 1131 but he means 1113, as his analysis of the Clouds (1445) makes clear. Also he includes Peace 1127 in his analysis (1460) but not his list.
10 Zielinski, T., Die Gliederung der altattischen Komödie (Leipzig, 1885), pp. 213–15Google Scholar; Pickard-Cambridge, A. W., Dithyramb, Tragedy and Comedy (Oxford, 1962), p. 198Google Scholar; Gelzer, T., ‘Aristophanes der Komiker’, RE Supplement XII (1971), col. 1520.Google Scholar
11 See Hamilton, R., ‘Announced Entrances in Greek Tragedy’, HSCP 82 (1978), 63–82, pp. 72–3Google Scholar. Taplin, following Dale, is rightly critical of the definition of stasimon in the Poetics, but at the same time recognizes the validity of its modern definition as strophic choral song: ‘there is only one strophic choral lyric which definitely does not divide two acts (A. Hik 418ff.)’ (p. 51).
12 As Sommerstein notes, this is not true of the parodos, but that is immaterial since the parodos is defined as the entry of the chorus, not its first song (though, as a consequence, it is difficult to say when the parodos ends).
13 Sommerstein says Cl. 804 does not ‘amount to an act-break’, presumably because of its length, which is only seven lines.
14 Apparently this is one of the ‘scene-dividing as distinct from act-dividing, songs’ which ‘are not structurally significant and may be included or omitted at the author's option’ (148), though it is not said how we can tell.
15 ‘It might possibly be maintained that 655–686 should be classified as an act-dividing song … but there is never an act-break between the parodos and the parabasis’ (146).
16 And the more inclusive part since by definition empty stage requires that ‘actors do not take part’.
17 The problem of getting cooking paraphernalia on stage is hardly sufficient reason to posit Peisetairos' exit and re-entry without some marker in the text.
18 Sommerstein (148) speaks of a ‘break between 1310 and 1316 during which no actor is on stage’ but he must mean 1316 and 1328. Both before and after the song at 1316 Trygaeus invites the chorus to eat plakountes (1314, 1356). The absence of any mention of exiting to don a γαμικ χλανς (contrast Birds 1693) may be a further indication that Trygaeus does not exit.
19 Socrates must have exited before 866, when he is called out of the Thinkery, and he is addressed in the third person at 830. The chorus addresses him in the second person at 808, perhaps as he is leaving, or this may be like the rather numerous choral addresses in tragedy to a non-present actor (see Taplin, p. 281).
20 Empty stage is the only criterion marking two songs as act-dividers in Sommerstein's analysis (Th. 947, Frogs 814), though two others (Ach. 836, Wasps 1450) could be added. Weissinger (38) calculates that 10 of 17 comic stasima have empty stage.
21 In the cases where time has passed during a choral song there usually has been some off-stage action and one could argue that it is the action not the time that is being marked: Ach. 1143 (expeditions); Kn. 498 (Council), 1264 (Demos cooked); Cl. 510 (Strepsiades instructed), 1115 (Pheidippides instructed), 1303 (Strepsiades beaten); Wasps 1009 (Philocleon dressed), 1265 (symposium and komos), 1450 (party); Peace 729 (trip to earth); Birds 676 (meal), 1058 (sacrifice; wall building), 1694 (trip to get Basilinna); Lys. 1014 (Spartan embassy), 1189 (feast); Th. 947 (Mnesilochos bound), Frogs 674 (Dionysus judged), 1482 (feast).
22 We may compare Weissinger's statement (41) that 23 of the 26 time-lapses in comedy occur during a parabasis or stasimon, which means that time-lapse occurs during about two-thirds of the act-dividers (5 acts in 9 plays requires 36 act-dividers).
23 This is not certain for Birds, where an increasing minority of scholars think the first four birds, all fantastic and all carefully described, belong to the chorus. See Sifakis, G. M., Parabasis and Animal Choruses (London, 1971), p. 126 n. 5Google Scholar for a list of disputants and add Stone, L. M., Costume in Aristophanic Comedy (New York, 1981), pp. 355–6Google Scholar, Green, J. R., ‘A Representation of the Birds of Aristophanes’, Greek Vases in the J. Paul Getty Museum 2 (1985), p. 117Google Scholar and Taplin, O., ‘Phallology, Phlyakes, Iconography and Aristophanes’, PCPS 33 (1987), p. 94Google Scholar. We should not think that the 24 individual names in 297–304, which most critics take to be the 24 members of the chorus, require that the chorus be individually masked any more than the individually named chorus members of Wasps and Lysistrata need wear different masks. Likewise, we should not confuse the chorus of women in the Thesmophoriazusae with the individual actors.
24 In Frogs there is a second chorus and it is not absolutely clear, despite gallons of scholarly ink, how that is managed. If the frog chorus is offstage there will be no confusion; if it appears on stage, its parodos will be ‘duplicated’ by the parodos of the mystics fifty lines later and so the confusion is only momentary.
25 For the formal characteristics, see e.g. Pickard-Cambridge, pp. 197–9.
26 Wasps 1516 and Frogs 1099 are the only comic stasima that lack exit and entrance of an actor. Since the former (a ‘Choral finale’ according to Pickard-Cambridge, p. 220) ends its play, it seems reasonable to treat the latter as anomalous also, and so I have excluded it from the list of act-dividing stasima. Almost every example listed here (not Lys. 781) is labelled ‘stasimon’ by Pickard-Cambridge (213–28). Twice the strophe and antistrophe are separated by well over a hundred lines, effectively transforming their status (Birds 1553 = 1694, Lys. 1014 = 1189). One purely choral song is not strophic: Peace 1316.
27 It would be convenient for the five-act theory to exclude Ach. 836 but since SS (matching stanzas) would be perceived as SA (strophe-antistrophe) the continuation of the matching pattern of Ach. 836 and Frogs 814 is more a ‘bonus’ than an exception.
28 For the question-marks at the parodoi of Birds and Frogs see above nn. 23, 24.
29 As Sommerstein (150) notes regarding the second act, ‘it is often hard to specify precisely where that act begins’. He seems to depend on a shift in focus while I mark where the chorus engages an actor, hence the numerous slight and occasional large discrepancies in our numbers. These differences will affect only our interpretation of the length of the second act, which does not seem particularly noteworthy in either analysis.
30 The anomalous four-act Peace has no such obvious explanation, but the problem it poses seems a legitimate one.
31 Zimmermann (51) notes that the parodos always occurs after line 200; Sommerstein (150) that the first act ‘varies only between 203 and 322 lines’.
32 When there is disagreement, Sommerstein's numbers are placed second.
33 For 876 Hunter (28) argues that the interlude is guaranteed by the dinner ‘imagined to take place between 876 and 877’, which is accepted by Sommerstein. R. G. Ussher, on the other hand, in his commentary on Ecclesiazusae (Oxford, 1973), xxviiiGoogle Scholar, argued that ‘no lapse of time…is necessary in either place where manuscripts have χορο’.
34 So Ussher, p. 114: ‘This is technically the προδος… But in this play a formal προδος is dispensed with, and the following choric passages are sung as the women leave, not enter, the orchestra’.
35 ‘Another kind of entertainment’ (Sommerstein 141); ‘another style of performance’ (Handley 59).
36 The interlude at 770 was expiscated from the Venetus scholia by Handley, who also explained why V is to be preferred to R, which marks only 770 and 801 (in a later hand).
37 See above n. 6. Conversely, among the many times when the stage is empty within an act in New Comedy, a few may involve offstage action or passing time (Samia 55, Dyskolos 392, 521, 666). Samia 55 and Dyskolos 510 are both ‘covered’ by an intervening scene, but, since at Samia 120 we have a similar ‘covering’ scene just before a choral interlude, we cannot count both of them as making up for offstage action or passing time.
38 Dyskolos 232, 783, Samia 200, 420, Epitrepontes 418 and Perikeiromene 266. We also find re-entry after a ‘covering scene’ (Epi. 171, Dys. 619, Sa. 120, Asp. 390) or a short monologue (Peri. 1005). This leaves us with only Sa. 615, Dys. 426, Asp. 249 unaccounted for, but the validity of such extensions of Hunter's rule is undercut by the fact that the pattern of a short monologue separating exit and re-entry is very common within acts.
39 This is less often claimed for structure than for language, dramaturgy, theme and character, but even so there were early in the century a number of critics either affirming or denying that Euripides in particular or tragedy in general was the source of the five-act rule. So Weissinger p. 45, citing Norwood, says ‘episodes in comedy are not at all like those in tragedy’, while Harsh, P. W., A Handbook of Classical Drama (Stanford, 1944), p. 163Google Scholar argues that Euripides' ‘melodramatic plays … tend to have four responsive choral songs dividing the play into five sections and thus anticipating the later practice of having five acts’ (see also pp. 316 and 443 n. 26). Flickinger, R. C., The Greek Theater and its Drama (Chicago, 1926 3), p. 193Google Scholar is more cautious, ‘since the histrionic divisions in tragedy were more usually five and since comedy fell more and more under the domination of tragedy, the rigid principle was at last set up for both tragedy and comedy that each play should contain five acts, no more, no less’.
40 This analysis differs from the now standard treatment of Aichele, K., ‘Das Epeisodion’, in Die Bauformen der griechischen Tragödie (ed. Jens, W., Munich, 1971), pp. 50–1Google Scholar in the following ways: (a) epiprologue and epiparodoi at Al. 746, Hel. 385 and division at Or. 959 are not counted by Aichele; (b) Aichele adds act-divisions at Hi. 1267, Hec. 1022, HF 873, 1015, Ion 1228, Or. 1352, 1536, Ba. 1152, IA 1509; (c) Aichele counts choral anapaestic announcement as part of choral song rather than part of the following act. For a telling challenge to Aichele's ‘uncritical acceptance’ of Aristotelian terminology, see Taplin 470ff.
41 Cropp, M. J., Euripides Electra (Warminster, 1988), p. 157Google Scholar rightly points to Philoctetes 827ff. as a parallel for interruption of a stasimon by an actor speaking stichic verse.
42 Many editors treat the whole of 960ff. as Electra's monody, but Willink, C. W., Euripides Orestes (Oxford, 1986), pp. 240–1Google Scholar is entirely persuasive in arguing for it as a choral stasimon plus monody: ‘an act-dividing Lament for the Extinction of the Royal House of the Atreidae, comprising a traditionally patterned strophe and antistrophe and a long solo epode.’
43 In Sophocles too the number of acts decreases steadily, with the exception of the OC, though there is only one five-act play:
Note that the end of the first act is more consistent in Sophocles, ranging from 93 to 150 with a mean of 120 and a mean deviation of 17.0 (14.2%).
44 As in comedy, in tragedy there is sometimes difficulty in delimiting the parodos; here it has been carried to the end of the lyrics, i.e. Al. 935 (epiparodos), Med. 214, El. 213, IT 236, Ion 237.
45 Makaria exits at 601, giving Iolaus a seven-line monologue before the act ends.
46 It is doubtful whether the unmarked exit of a servant/messenger should be counted. Likewise El. 858.
47 The nurse cries from within, somewhat like the children in Med. 1270ff. (in the middle of a stasimon!).
48 Electra is called out only after the chorus has heard a sound and reacted (747–50).
49 It is not clear whether Orestes enters during the stasimon or only after.
50 It is difficult to tell at what point Hecuba is left alone; certainly by 796.
51 Taplin, thus, is wrong (e.g. 54) to try and elevate exit/entry to a major structural role in tragedy.
52 I am grateful to Gregory W. Dickerson, S. Douglas Olson and the journal's referee for their helpful suggestions.