[Comedy] differs from tragedy in its matter, in that tragedy is tranquil and conducive to wonder at the beginning, but foul and conducive to horror at the end, or catastrophe. .. .And for this reason some writers have the custom of saying in their salutations, by way of greeting, ‘‘a tragic beginning and a comic ending to you.’’
Dante, letter to Can Grande della Scala (Haller 1973, 100)
Of all literary genres, tragedy and comedy have come to be most identified with particular sorts of endings, the comic ending emblematized in the cultural imagination by united or reunited couples and the tragic by a body-littered stage. The expectation that the tragic ending will involve some kind of overturn further suggests (as in the passage from Dante above) that the beginning of a tragedy should be marked by serene good fortune. In fact, however, whatever the later development of the genre and of our conceptions of the tragic, Greek tragedy reveals a considerable variety of modes of ending and of beginning. A number of plays (for example, Aeschylus’ Eumenides, Sophocles’ Philoctetes, Euripides’ Iphigenia among the Taur-ians and Alcestis) end with deliverance, reunion, or reconciliation, even if these are qualified by what came before or will come after. And many tragedies that end with disaster begin not with happiness but with deep distress, among them Aeschylus’ Persians, Sophocles’ Women of Trachis, and Euripides’ Medea. It is in Aristotle’s Poetics (written some decades after the heyday of tragedy in fifth-century Athens) that we first find articulated the view that an unhappy ending, with a change from prosperity to misfortune, is, if not definitive of the genre, at least a mark of those plays that are best constructed and most essentially tragic.
Aristotle also identifies beginnings and endings as elements in a connected narrative sequence and as points of demarcation required by any plot that is a well-structured whole, and it is in this connection that he defines them:
A beginning is that which does not come necessarily after something else, but after which
It is natural for another thing to exist or come to be. An end, on the contrary, is that
Which naturally comes after something else, either as its necessary sequel or as its usual sequel, but itself has nothing after it. (Poetics 1450b27-30, tr. Hutton 1982)
The appearance of obviousness is (as often for Aristotle) deceptive. The problem of creating beginnings and endings in the continuous flow of events confronts all writers, but presents particular complexities for ancient authors, given that virtually all their stories are taken from longer myths familiar in general if not in detail to the audience. Aristotle here anticipates the interest of modern narratology in the workings of beginnings and endings as such (Richardson 2002). How does a beginning deal with what (inevitably) came before, so that it feels like an acceptable beginning? How does it prepare us for what follows? How does an ending complete the action of the play, and how does it deal with what (inevitably) comes after? How does it create what has come to be called closure, the sense of finality or conclusiveness at the end of a work (Kermode 1967/2000, Smith 1968, Fowler 1989, 1997)?
If beginnings and endings in Greek tragedy do not define the genre by a predictable trajectory from happiness to misery, they remain definitive in other respects. Beginnings have a programmatic authority in drawing the audience’s attention to the tragedy’s characters, choice of story, approach, and relationship to the genre (Dunn 1992, Segal 1992); endings have, or appear to have, an interpretative authority, since the point of closure may also be seen as the point at which the audience can finally look back at a completed action and read it fully, in a version of what Barbara Herrnstein Smith has called retrospective patterning (Smith 1968, 10-14, 212, 218). The ancient tragedians both enlist these modes of authority and undercut or complicate them, offering us at times obscure or misleading beginnings and open or disconcerting endings.
In what follows, I consider some of the varieties of tragic beginning and ending, drawing on examples from the three surviving Greek tragedians to explore some of the functions and effects of these parts of the drama, the development of conventions of beginning and ending, and the ways in which writers create, meet, or undercut audience expectations.
A tragedy may begin with the parodos (entry and first song of the chorus), but this is most often preceded by a prologue of some kind: a long opening speech, a dialogue scene, or a combination of the two (usually speech followed by dialogue). Some tragedies also include a monody (solo song by a single character) before the parodos, or a kommos (duet between the chorus and a character) in its place.
In the plays that have come down to us, Aeschylus’ beginnings are formally the most varied, perhaps in part because some of his plays (Eumenides, Libation Bearers, Seven against Thebes) form the second or third part of a trilogy and thus have a kind of secondary status as beginnings, but also perhaps because set conventions for the prologue develop gradually in the course of the fifth century. Of the plays that we know either begin a trilogy or stand alone, both Persians (perhaps our earliest extant tragedy) and Suppliants begin with the entrance of the chorus; Agamemnon begins with a monologue followed by a particularly long and complex choral sequence.1 In all three, however, the chorus dominates the opening moments of the play as it relates past events and expresses the emotional valence of these events, evoking a mood of shared apprehension.
All but one of Sophocles’ extant tragedies begin with a dialogue scene in which two characters speak, sometimes joined by a third (the exception is Women of Trachis, in which monologue is followed by dialogue). In all our extant plays the first two characters are joined by some sort of affective tie, such as patronage, friendship, or kinship: we see (for example) Athena and her favorite mortal, Odysseus (Ajax), Antigone and her sister Ismene (Antigone), Oedipus and the priest who speaks for Oedipus’ trusting subjects (Oedipus the King). But in spite of these modes of alliance the perspective evoked is inevitably double, not single, and tensions and differences characteristically emerge as the story begins to unfold.
Euripides begins almost all his plays with a monologue, of a type so marked that Aristophanes makes fun of it in his Frogs for its stock narrative of past events. The speaker may be the play’s central character (Helen in her name-play), a secondary character (Heracles’ old father Amphitryon in Heracles), a minor character (the nurse in Medea), or a god (Aphrodite in Hippolytus), and the speeches vary accordingly. But what distinguishes these Euripidean monologues in general from others is their relative detachment from the action and the thoroughness and directness (unmotivated by the story) with which they provide information. Consider, by way of contrast, the opening monologue of Aeschylus’ Agamemnon (1-39). The watchman, asking the gods for a respite, tells us where and who he is as part of a complaint about the task he has been forced to undertake, years of watching from the palace roof in Argos for the signal fires that will mark a Greek victory in Troy. He alludes to the circumstances, breaks off to announce that he has sighted the beacons, and concludes with some ominous and enigmatic words about conditions in Argos. This speech is part prayer, part soliloquy, and part news report. But Euripidean prologues, though they may reflect the circumstances and feelings of the speaker (the nurse in Medea, for example, narrates the chain of events that led to her mistress’s current predicament in the form of a wish that none of it had ever happened) are never really a part of the action. The lack of any clear motivation for the detailed account the speaker gives of the past and the absence (in many instances) of an obvious addressee gives the impression not of a soliloquy but of speech directed to the audience, one that selfconsciously fulfills the function of a prologue in providing necessary information. This impression is particularly strong in the plays in which a god speaks the prologue and never reappears: the combination of apparent authority over the play’s events and absence from these events makes these figures seem almost an approximation of a narrator.
All beginnings must of course provide a certain amount of information, however they do so. Athenian playwrights had an opportunity at the proagon (preliminary presentation of the playwrights and their casts) to give some kind of advance notice of their subject matter, but it may not have been extensive, and prologues generally seem to take pains to let the spectators know who the characters are, where the characters are, and what has happened by the time the story begins. To some extent, the last of these follows from the first, given tragedy’s repeated use of well-known myths; in a fragment from the fourth-century bce comic playwright Antiphanes (fr. 189 PCG) the speaker complains that his job is much harder than that of the tragedian, who need only introduce a character named Oedipus for the audience to be acquainted with both background and plot. But many figures in myth are associated with enough different stories, or long enough stories, to make many tragedies, and many of these stories exist in variant forms; furthermore, even if we can deduce story from character, we still need to know exactly where we are in the story and what has happened up to this point. (This is to some extent true even in a trilogy, since any amount of time may have passed between one play and the next.)
Compare two plays in which Oedipus is the central character. The beginning of Sophocles’ Oedipus the King shows us Oedipus as a ruler in conversation with his people, identified as the descendants of Cadmus (1); he refers to himself as ‘‘Oedipus, whose fame is known to all’’ (8). The spectators know, then, not only who is speaking, but where and when. We are in Thebes, and Oedipus is king; the riddle of the Sphinx, the killing of Oedipus’ father, and the marriage with his mother are all in the past, but not yet discovered; fame has not yet turned to infamy. But in Sophocles’ later Oedipus at Colonus, the opening words here (again spoken by the title character) situate the speaker at quite a different point in his story:
Child of the blind old man, Antigone,
What place have we come to, what people’s city is this?
Who will receive the wandering Oedipus today with scanty gifts... ? (1-4)
Oedipus’ address to his daughter serves also as a self-identification, and tells us that this play takes place after his fall, in changed circumstances, and in a new setting. But there is a further difference. The earlier play evokes a story and a setting found in the tradition as far back as Homer’s Odyssey (11.271-80); once the audience knows where and when we are, they know what the play must reveal. But the plot of Oedipus at Colonus seems to have been in part Sophocles’ own invention; the opening questions thus not only provide the audience with information, but also point to their own lack of knowledge and to the mystery of the particular significance of this setting (Dunn 1992, 2-3).
The beginning, then, may supplement or confirm the audience’s partial knowledge to clarify the point at which we enter the story; it may also limit clarification, and it may exploit that partial knowledge by playing with the audience’s expectations. In Aeschylus’ Prometheus Bound, at the end of the opening monody, Prometheus expresses alarm at the sound of approaching wings (124-25). The audience will presumably wonder whether this is the vulture, come to tear out his liver, but as it turns out, that is far in the future; what he hears now is the friendly chorus of the daughters of Ocean in their winged chariot.
Euripides’ prologues may also play with the spectators’ expectations in spite of their apparent straightforwardness in identifying the speaker (who generally names himself or herself without the dramatic pretext offered by the other two tragedians) and circumstances. In some plays (such as Ion, Helen, and Electra) he makes use of less well-known stories or variants an audience will be slow to recognize. Why, in the prologue of Helen, is Helen in Egypt and not in Troy or Greece? (She turns out to have spent the war there while the Greeks fought for a phantom.) Why is an unnamed farmer speaking the prologue in Electra? (It turns out that Aegisthus has married Electra to this man so that her children cannot be a threat to him.) In others, the very directness and completeness of the information may keep the audience guessing for some time exactly where this play will start. In Phoenician Women, for example,
Jocasta gives us a detailed history and genealogy of the house of Laius; an audience familiar with the tradition that she killed herself after discovering her marriage to her son will be surprised to learn that that discovery is far in the past and that she is a helpless spectator of the battle between the sons she bore to Oedipus.
As this last example suggests, the opening scene or song of a tragedy typically acquaints the audience with at least some of the events that have led up to this beginning. How this is done not only affects what version of events the audience presupposes and how much it knows from the outset, but also the perspective from which these events are viewed.
In Agamemnon, the watchman lets us know that the war in Troy is over, but that matters here at home are not as they should be. As he leaves the stage, the old men of the chorus enter, chanting in lyric anapests; with their opening words (‘‘It is now the tenth year since Priam’s great adversary, the lord Menelaus, and Agamemnon. ..,’’ 40-42) they take us back to the origins of the war and to its outset. Menelaus and Agamemnon are pictured as justly taking vengeance, with divine support, for Helen’s abduction by Paris. But Helen herself (‘‘a woman of many men,’’ 62) is a dubious reason for the suffering on both sides. Nor has the suffering been limited to death in battle. As they move into song the old men go back in memory to a troubling portent that ultimately (and somewhat mysteriously) led to the sacrifice of Agamemnon’s daughter Iphigenia to gain a fair wind for the fleet. The singers omit the actual moment of sacrifice (‘‘what came next I neither saw nor speak of,’’ 248), but even as they hope that everything will turn out well they do not conceal their fears or the words of the prophet Calchas about ‘‘child-avenging Anger’’ (155).
Aeschylus has not yet given us the whole of the relevant past. At later moments in the play other speakers (Cassandra, Aegisthus) will reveal the terrible earlier history of the house of Atreus and its connection with present events. But the chorus has already set before us not only past events themselves but the complex and unexpected connections of these events. One thing leads to another in unforeseen ways; no story is complete in itself; no end can be relied on. The trilogy will bear out these conclusions.
The beginning of Sophocles’ Antigone tells the story of what came before first through a conversation between Antigone and her sister Ismene, and then in the voice of the chorus. Because the past is initially presented through dialogue it is made known only gradually and piecemeal. Antigone speaks of the suffering that has come from their father and that dishonors the family, and wonders whether Ismene has heard the latest news; Ismene replies that she knows of their two brothers’ death, but no more than that. Antigone then tells her sister of Creon’s decree that Eteocles alone will receive funeral honors, while Polynices remains unburied, and proposes that the two of them bury their brother. In her reply, Ismene reminds her sister of their father’s self-blinding and death and of their mother’s suicide.
This way of introducing the past not only allows it to emerge with a certain naturalness but also presents it as the differentiated experience of individual characters. Antigone sees the past in terms of family dishonor, and will not allow any additional dishonor; Ismene, in contrast, sees the family history as one of death and destruction, and hopes that she and her sister can escape with their lives. When the chorus enters, we are given yet another perspective; in describing the battle just ended, and hailing the Theban victory rather than lamenting the dead brothers, these citizens tell the history of a royal family that posed a danger to the city and its inhabitants (Roberts 1989, 163-64).
If the beginning of Agamemnon adumbrates a past of linked and overlapping stories, and the beginning of Antigone constructs a past that is a different story for different characters, the beginnings of certain Euripidean tragedies suggest - in different ways - the existence of several possible pasts, only one of which the poet has chosen to write. Medea begins with a narration of the past as something that ought never to have occurred but did. In Helen, the title character relates a past that departs from the usual version of her story. The prologue of Orestes, like that of Phoenician Women, gives such an extensive retelling - going back to the distant past of the house of Atreus, and including almost every character associated with the myth - that it gives an impression of inclusiveness and authority. But the prologue ends with an unexpected twist: Orestes has indeed killed his mother, but instead of escaping the Erinyes by the judgment of an Athenian court (the version in the Oresteia) faces the prospect of being condemned to death by an Argive court - unless Menelaus rescues him.
Unless Menelaus rescues him. The beginning of a tragedy does not simply include or allude to the past; it generally looks ahead to what is to follow as well. If audience knowledge precludes major surprises it leaves room for both minor surprises and suspense (when and how will the expected ending come about?). It also fosters the use of foreshadowing, since the sorts of clues that in invented plots are accessible only the second time through may in traditional, myth-based tragedy be noticed at the initial presentation. Sometimes the beginning of a play anticipates the future only by the ordinary expectations and fears of characters: the Electra of Euripides’ Orestes hopes Menelaus will rescue her brother; the chorus in Aeschylus’ Persians dreads bad news of Xerxes’ army, long since gone to Greece; Sophocles’ King Oedipus promises to do all he can to find out what has caused the plague in his city. But a play may also begin with a prophecy known to one or more characters. The opening chorus of Agamemnon includes the prophecies of Calchas about the sacrifice of Iphigenia, already fulfilled except for certain ominous words about secret anger and the child that is to be avenged. Four of Sophocles’ prologue scenes (in Oedipus at Colonus, Women of Trachis, Electra, and Philoctetes) include or refer to the words of oracles.
The famous ambiguity of oracles finds an analogue in the allusive language of foreshadowing or prolepsis, a mode of anticipation particularly prominent in Aeschylus (Lebeck 1971). The opening lines of Persians speak of‘‘the Persians gone to the land of Greece’’ (1) but the word translated as ‘‘gone’’ can also have the meaning of‘‘dead’’ or ‘‘ruined,’’ and thus anticipates both the outcome the audience knows must follow and the play’s gradual shift from uninformed anxiety to mourning. The opening passages of Agamemnon are still more densely proleptic. The images of fire, the images of birds and their young, and the patterns of three evoked by the chorus all anticipate subsequent developments in the trilogy.
As in so many other respects, Euripidean prologues are distinctive in the mode of anticipation they sometimes exhibit. In five plays in which a god speaks the prologue (Alcestis, Hippolytus, Trojan Women, Ion, and Bacchae) that god not only sums up the past but also predicts the future. These predictions are not like ordinary prophecies: they are apparently straightforward declarations of the god’s plans and they are available only to the audience (Hamilton 1978). But their apparent straightforwardness - as with other aspects of Euripidean prologues - is misleading: in at least one of these (Ion) and some would say in two others (Bacchae and Hippolytus), although the god’s will is ultimately fulfilled, the details of its fulfillment are not quite as stated, and in one case (Ion) the god’s plan goes seriously astray and the mortals behave in unpredictable but not surprising ways. In the beginning of Ion, Hermes tells how Apollo raped Creusa, the princess of Athens, how she abandoned the child out of shame, and how that child (now named Ion) was raised in Delphi. He also anticipates the eventual happy reunion, through Apollo’s machinations, of Creusa and her son. As the play proceeds, however, mortal resentment, longing, and disappointment (all natural and understandable) cause Apollo’s projected plot to go off-track; another divinity, Athena, must be introduced at the close of the play to return things to their proper course.
This brings us to the endings of tragedies. These have received more critical attention than beginnings, for several reasons: the traditional association of tragedy with a particular type of ending, the sense that it is the end that confirms or enables interpretation of the drama as a whole, the general interest in issues of closure in literary studies in the past few decades, and the existence of certain particularly problematic endings.
With tragic endings as with tragic beginnings we see somewhat more formal variety in Aeschylus than in his successors, but the chorus plays a prominent role in all of his endings (often as a partner in a concluding exchange, spoken, chanted, or sung) and has the last word in all but two. In Sophocles and Euripides the choral role in the ending is diminished but formalized; dialogue exchange or long speech and response dominate their final scenes, but in all of their extant plays (with one possible and debated exception, Sophocles’ Women of Trachis)2 the chorus still speaks the last words in what seems to have gradually evolved into a conventional coda, a few lines of verse, usually anapestic, that mark (in a curtainless theater) the ending of the play. This coda is (not surprisingly) most formulaic in Euripides: two of his choral tags, whose sentiments are broadly applicable, are repeated with little or no variation in several plays. In other respects, too, Euripidean endings are most likely to make use of conventional features that in one way or another underscore closure (Dunn 1996); the most prominent of these is the deus ex machina, the god who appears above the action, sometimes suspended by the mechane or crane, sometimes on the top of the stage building.
As I noted above, the endings of Greek tragedies are not necessarily unhappy, although they are generally fairly solemn (a word that better conveys the sense of the adjective tragikos, ‘‘tragic,’’ in the fifth century). Aeschylus’ Oresteia ends with the acquittal of Orestes on the charge of matricide and the reconciliation of the Erinyes with the city of Athens; Sophocles’ Philoctetes ends with the return of Philoctetes from pain and isolation to companionship, victory, and healing; and in a set of endings that have tempted critics to look for a different generic term (tragicomedy, romance, or melodrama) Euripides has his characters reunite (Helen, Ion, Iphigenia among the Taurians), escape from danger (Helen, Iphigenia among the Taurians, Orestes), or even return from the dead (Alcestis).
Both these endings and the endings that fit the preferred Aristotelian model of a change of fortune from good to bad contribute to our sense of closure by resolving questions raised in the play and bringing the play’s action somehow to a successful conclusion. Different modes of dramatic closure may be associated with recurrent plot patterns or motifs: vengeance completed (Aeschylus’ Oresteia, Euripides’ Medea and Hecuba, the Electras of both Sophocles and Euripides), mysteries solved (Sophocles’ Oedipus the King, Euripides’ Ion), lost relations found (Euripides’ Alcestis, Ion, Iphigenia among the Taurians), danger escaped (Aeschylus’ Suppliants, Euripides’ Iphigenia among the Taurians and Helen), disaster realized (Euripides’ Trojan Women), prophecy fulfilled (Sophocles’ Oedipus the King, Ajax, and Women of Trachis), arrogance or strength brought down (Aeschylus’ Persians, Sophocles’ Ajax, and Euripides’ Bacchae).
Many of these types of endings reinforce closure not only by meeting expectations raised in the course of the play but also by evoking natural or cultural markers of closure in human lives: departure, reunion or reconciliation, solution or fulfillment, death, and ritual (Smith 1968, 101-2, 172-82). Not surprisingly, mourning is the most common of concluding rituals; of twenty-three plays that end in one way or another with death, nineteen end with some form of or reference to burial or mourning ritual. We find other rituals less often, but Eumenides, the last play in the Oresteia, ends with a celebratory procession to install the Erinyes - now become kindly - in their new home, and some Euripidean endings allude to a form of ritual that is to be established in the future. In Hippolytus, for example, Artemis declares that in the future young Athenian girls will offer up their hair in Hippolytus’ honor before their wedding day.
Such markers may, however, also be used in such a way as to interfere with closure. In several plays the ritual of burial is deprived of some of its effect by the exclusion of a would-be participant (Roberts 1993). In Euripides’ Medea, for example, Medea forbids Jason any part in his children’s burial - which she alone, their murderer, will carry out; a couple united only in their grief for their dead children will remain divided in the ritual expression of that grief.
Division and uncertainty undercut or qualify closure at the end of a number of tragedies. The effect is perhaps most striking in those Aeschylean plays whose endings are only provisional, since they come first or second in a trilogy: Agamemnon ends with a quarrel, Libation Bearers ends with a series of questions (including, tellingly, ‘‘where is the end?’’ 1075), and Suppliants ends with the emergence of a division of opinion between two parts of the chorus (or two choruses; scholars differ on this point). But plays that have no sequel may also incorporate in their endings such anti-closural topoi as division, disbelief, and uncertainty. Orestes and Electra are separated forever at the end of Euripides’ Electra, as are Agave and her father Cadmus at the end of his Bacchae. At the end of Euripides’ Orestes, Apollo’s concluding arrangements meet with an acceptance strongly tinged with skepticism from the characters involved. At the end of Oedipus the King, a scene explored by a number of scholars in recent decades (Taplin 1983a, Kitzinger 1993), we do not have any clear sense of the outcome that is to follow the play’s dreadful discoveries. Will Oedipus be exiled, as he requests? Will Creon force him to linger on in Thebes? The presence of this uncertainty is particularly striking in a play that in other senses exhibits such strong closure, revealing the fulfillment of all the story’s oracles and with them fulfilling the audience’s expectations.
So far I have been speaking of the audience’s expectations as aroused and fulfilled by the events of the plot, and as confirmed or disrupted by the author’s use of various closural modes. With endings as with beginnings, however, audience expectations in Greek tragedy are conditioned not only by the unfolding events of the plot, but also by their knowledge of the myth from which that plot is taken. To some extent, this knowledge may further reinforce closure: we expect in Oedipus the King that Oedipus will prove to have killed his father and married his mother not only because the prophecies in the play predict this outcome, but also because the traditional story demands it. When Agamemnon’s corpse is revealed at the end of Aeschylus’ play, his death is not only overdetermined by a variety of causes the drama has set before us (Clytemnestra’s anger at Iphigenia’s death, her love for Aegisthus, Aegisthus’ own desire for vengeance, the curse on the house of Atreus, Agamemnon’s actions at Troy); it is also the expected end of a very old and very familiar story, already a part of the tradition in Homer’s Odyssey (1.35-36, 3.193-94, 4.519-35, 11.404-434, 24.20-22, 96-97, 199-202).
In some plays, however, the traditional ending is made to conflict with the direction the play has been taking; it thus seems to enforce (rather than reinforce) closure and to some extent to disrupt the audience’s expectations. In Sophocles’ Philoctetes, Odysseus has brought Achilles’ son Neoptolemus to a deserted island so that he may trick Philoctetes (marooned there years before) into rejoining the Greeks and helping them take Troy. As the play draws to an end, Neoptolemus has abandoned deception for friendly persuasion but has failed to overcome Philoctetes’ rage at the Greeks who abandoned him. Philoctetes reminds Neoptolemus of his earlier (false) promise to take him home, and Neoptolemus prepares to make this promise good; the two of them set off together. It appears, and this impression is reinforced by stylistic elements of the scene (Hoppin 1990), that the play is going to depart quite radically from tradition; if these two men simply return home, the Greek expedition will fail. At this point, however, Heracles appears, and commands them to embark for Troy and capture the city; the myth is back on track.
Readings of what some have called the two endings of the Philoctetes vary, but it is at least arguable that the ending the myth requires is the right ending in other respects and does no violence to the drama (Schein 2001). Philoctetes, who suffers from a hideous wound, will be healed, and will regain his rightful place. Furthermore, his final conversation with Neoptolemus has suggested that he would like to do as the younger man asks, but cannot bring himself to let go of his anger. The deus ex machina may be read as enabling such a letting go. In Euripides’s Orestes, however, we find the tension between mythical outcome and plot pushed to an incomprehensible extreme. In the final scene of a play that derails the traditional story even as it echoes every earlier version (Wolff 1968, Zeitlin 1980), Orestes and Pylades (condemned to death for matricide along with Electra and disappointed of Menelaus’ help) have tried to kill Helen and have kidnapped Hermione, the daughter of Helen and Menelaus. Orestes, confronting Menelaus, stands on the roof with his sword at Hermione’s throat and prepares to set fire to the palace. At this point Apollo - whose oracle has been thoroughly discredited in the course of the play for commanding matricide - arrives as deus ex machina and commands an outcome appropriate to the traditional story, with more or less happy endings for all, including a marriage between Orestes and his erstwhile hostage Hermione. This disjunction between the direction of the plot and the imposed (if traditional) conclusion makes the ending one of the most disconcerting in all of Greek tragedy, and it is hard not to see it in
Part as a comment on the struggle between the inherited myth and the natural direction in which the playwright wishes to take the play.3
A myth known to an audience can also qualify closure by leading the audience to think of what lies beyond the end of the play. We know that the death of Agamemnon at Clytemnestra’s hands, traditional though it is, cannot really be the end of Aeschylus’ play, since the same tradition includes Orestes’ vengeance for his father’s death. Here the open-endedness (created not only by audience knowledge but also by various forms of prolepsis in the drama itself) simply looks forward to the rest of the trilogy. But there are also single plays in which audience knowledge of the myth qualifies or complicates closure. At the end of Philoctetes, for example, Heracles promises Philoctetes and Neoptolemus victory at Troy, but warns them that when they sack the city they must respect the temples of the gods. An audience that knows the tradition knows that the Greeks in general will disregard this command and that Neoptolemus - who in this play has been an almost entirely sympathetic, if at first misguided, character - will commit a striking act of sacrilege in killing Priam at his household altar.
This type of ending, in which a subtle allusion to a familiar story opens up a future beyond and even at odds with the mood of the play’s conclusion, seems to be particularly characteristic of Sophocles (Roberts 1988). Euripidean endings, in contrast, often seem to be trying to avoid such openness by including (almost enclosing) the future in a speech by a deus ex machina. Most of these concluding speeches prophesy the future of the surviving characters, often in such detail as to appear to leave little to the audience’s imagination. Apollo’s speech at the end of Orestes tells us that Helen will become a star; that Menelaus will remarry; that Orestes will be exiled for a year, will be acquitted at Athens, and will marry Hermione; that Neoptolemus, who expects to marry Hermione, will be killed at Delphi; that Electra and Pylades will marry; that Menelaus will rule in Mycenae and Orestes in Argos. This speech thus includes not only elements traditional in the inherited myth or familiar from its treatment by others (Orestes’ marriage with Hermione, the trial at Athens) but bits of other, tangential stories (Neoptolemus’ death) and unusual elements (Menelaus’ remarriage). It seems designed to give the audience everything it could expect and more, so that what is outside the drama (exo tou dramatos, in Aristotle’s phrase, Poetics 1453b32, 1454b3) is actually contained within it, and constitutes no sort of distraction or opening-up.
In other respects too Euripides appears to make regular use of certain conventions, ‘‘closing gestures’’ in Francis Dunn’s term (Dunn 1996), to effect or reinforce closure. The authority of the deus ex machina, the prophecy that contains the future, the regular use of the aition, or aetiology (an explanation of some later institution or monument known to the audience in terms of the play’s events) - all of these to some extent have the effect of confirming the end, in part by inducing a kind of stasis. The god’s authority declares the proper end, the prophecy prevents us from wondering about the future, and the aition displaces the movement of historical narrative with the stasis of contemporaneous presence or repeated practice. His choral codas too, by their generality of application and their stock morals, seem to cut short speculation in favor of a simplified understanding of the events.
But as Dunn has argued, ‘‘this blaze of clarity leaves the process of closure all the more uncertain’’ (Dunn 1996, 7). Euripides’ devices are by their very nature subversive, since they may be said to draw attention to themselves as conventions, as self-conscious rhetorical gestures. But they also lend themselves to particular modes of subversion in individual plays. The god who appears (or whose authority is evoked) at the end may already have had that authority compromised. Apollo fares particularly badly in this respect, both in Euripides’ Orestes, where earlier criticisms of the matricide he commanded linger to taint the resolution he brings, and in Ion, where he avoids facing any criticism of his behavior and sends Athena in his place. When Artemis appears at the end of Hippolytus, her promise that Aphrodite too will suffer the death of a favorite seems trivial next to the picture of human suffering and reconciliation provided by Theseus and his son. The prophecy that appears to present a settled (and generally familiar) future may frame that future in such a way as to make it seem problematic: at the end of Euripides’ Electra, Castor and Pollux predict for Orestes the familiar Aeschylean sequence of pursuit by the Erinyes and ultimate acquittal in Athens; Electra, as expected, will marry Pylades. But this resolution is deprived of any evident justice by the accompanying declaration that what Apollo did was wrong, and likewise of any redemptive force by the exile and separation imposed on the wretched and remorseful siblings. Euripidean aetiologies, as Francis Dunn has argued, include puzzling ambivalences and discontinuities (Dunn 1996, 45-63, 9496, 136-37; Dunn 2000). The rite Artemis establishes at the end of Hippolytus in compensation for the hero’s sufferings is also said to commemorate Phaedra’s love, a central cause of that suffering; when at the end of Helen the Dioscuri (brothers of the heroine and dei ex machina) announce that mortals will in the future call by Helen’s name the island where she stopped on her way to Egypt, the insignificance of the link between plot and later fact seems merely to trivialize the aetiological gesture. And Euripides’ choral codas, especially the ones that recur at the ends of several plays, have seemed so abrupt and simplistic as to provoke readerly resistance and scholarly claims of spuriousness. We encounter the following at the end of five very different tragedies (Alcestis, Andromache, Helen, Bacchae, and with a different first line, Medea):.
Many are the shapes divinities take,
Much that’s unanticipated the gods accomplish,
What we expect goes unfulfilled,
And the god finds a way for the unexpected.
Such was the outcome of this matter.
A number of the plays to which I have already alluded end in ways that challenge any simple reading. When Heracles makes Philoctetes abandon his resistance and rejoin the other Greeks, should we see this as a redemptive move that reintegrates him into a society exemplified by the honest and sympathetic young Neoptolemus, or the betrayal of a principled hatred for a treacherous enemy exemplified by the opportunistic Odysseus? When Apollo returns the plot of Orestes to its traditional outcome in a manner quite at odds with the preceding action, should we see this as an ironic critique of the god, as a reflection on the mess that mortals make of things when they act without divine guidance, or as a metatheatrical comment on tragedies and how they end?
One of the most controversial endings is that of Sophocles’ Electra. This play tells essentially the same story as Aeschylus’ Libation Bearers: Orestes returns from exile in disguise, and with his sister Electra’s help takes vengeance on Clytemnestra and her lover Aegisthus. Electra, not Orestes, is the central character here, and there are other significant departures from the Aeschylean version, but the most striking comes at the end. Orestes (spurred on by Electra) has killed Clytemnestra first, in a reversal of the Aeschylean order, and now forces Aegisthus into the house to meet his death. The chorus, left behind on stage with Electra, chants a short coda, celebrating the success and freedom of the family of Atreus, and the play ends. The reversal of the deaths allows Sophocles’ play to end with the less problematic of the two killings; the chorus’s words suggest that we should regard this as a happy outcome; and, most striking of all, the Erinyes neither appear (as in Aeschylus’ play) nor are mentioned at the end (as in Euripides’ version). On the other hand, the chorus’s words, like many choral codas, seem inadequate to what preceded; Aegisthus has made an ominous reference to the future evils of the house, and the Erinyes have been mentioned in a general way earlier in the play (112, 276, 491,1079). What are we to make of this? In particular, given that the audience knows the events of the Aeschylean ending, should we take their omission by Sophocles to signal that we should assume that ending (the Erinyes will pursue Orestes), or that ending plus its resolution in Eumenides (the Erinyes will pursue Orestes but will eventually be reconciled to his acquittal), or neither of the above (the Erinyes will never bother this Orestes)? Alternatively, should we simply assume that since this is Electra’s play we are not meant to be concerned with an ending that features Orestes? And how will each of these choices affect our reading of the play as a whole?
The endings I have mentioned so far are debated in the sense that there is not clear agreement on how to read them, but they are all at least accepted as the work of the playwright. Several Greek tragedies present us with endings whose authenticity has at some time or another been questioned, most notably the final scenes of Seven against Thebes and Iphigenia atAulis and the closing lines of Oedipus the King and Phoenician Women (which partially replicate each other).
I will not enter into these debates here, but will just note that the claim of spuriousness itself points to the way in which individual tragedies were read as part of a continuing myth and in relation to previous tragic versions of that myth, since each instance involves a supposed interpolation of material from another version or a continuation of the story, sometimes one familiar from a well-known tragedy. The ongoing debate about these endings also reflects our own lack of certainty (and our occasional presumptuousness) about ancient closural convention. It is a similar lack of certainty that has led scholars to be suspicious of the codas at the end of Sophocles’ and Euripides’ tragedies. But though we may have particular reasons to reject some of these, a general rejection is uncalled for. These endings are formulaic, and rarely shed any particular light on the tragedies they conclude, but they mark in several ways the boundary between the world of the play and the various worlds to which it is tangential: the world of continuing myth, the larger world of discourse, the world outside the theater (Roberts 1987; Dunn 1996, 13-25).
Debated endings should remind us that just as the endings of Greek tragedies may confound our sense of the tragic, they may also confound our sense of endings. They should also remind us of the role played by the spectator or reader in the creation of closure. Our interpretation of particular tragedies thus calls for an understanding of the general conventions and strategies of tragic beginnings and endings and for a
Certain self-consciousness in our own assumptions about what it means for a tragic plot to have Aristotle’s required beginning and end. (I leave the middles to others in this volume.)
World History
