1 Why and How Tragedies Were Selected in the 5th Century BCE

On principle, selection by competition, in classical Greece, had nothing to do with canonization. Selection was everywhere: in athletic games (e.g. the Olympic games), in the way the city functioned (selecting juries and leaders), in poetic performances (choruses, dramas, comedies, odes). Competition and selection were not used primarily to achieve better efficiency, they had no utilitarian value – this is a big difference with the way competition is seen and used today in our free-market society. They were valued primarily as a means to honor and please the gods. They had the value of an ordeal.

Selection was aimed at for its own sake (or for the gods’ sake), whatever the means of selection. Actually, there were different ways of selection, apparently (from our modern point of view) quite incompatible with each other: objective criteria (e.g. athletic performances: who runs the fastest, who wins at fight, etc.), votes, chance or fate. One could even reasonably argue that the democratic regime in Greece originated in the religious will to introduce selection and competition everywhere in the city in order to better please the gods – or, to put it in a less provocative and teleological way: the democratic process adapted very well to the general ordeal system that was then in use.

Fate – that is the most obvious and the least disputable contribution of the gods’ hand – was everywhere in those selection processes: when juries or civic councils were drawn up in Athens, for instance, and also in tragic competitions. The intervention of fate took place at two levels:

• in selecting the jury: “Before the festival (or before the particular contest) the Council drew up a list of names selected from each of the ten tribes. […] The names were then placed in ten urns, each containing the names selected from one tribe. These urns were sealed […] and deposited in the Acropolis […]. At the beginning of the contest for which the judges were required the ten urns were placed in the theatre, and the archon drew one name from each.”¹

• in selecting the winner poet: each of the ten jury members threw a ballot in an urn, and the archon drew five ballots (and possibly more in cases of egality) to determine the winner (the exact process is still largely conjectural, and it may have taken various forms).²

The intervention of fate in the process means that the question of the intrinsic quality or value (whatever this means) of the plays was not of primary importance. The selection had a value in the moment it was made and for the very context it took place in: an instantaneous value, not an enduring one.

Moreover, the prize was given to a poet for the tetralogy (or the tragic trilogy) he had presented to the public and the god, it was not given for a single play. This is a big difference with our current tragic canon. For when tragic canons were eventually constituted, they were made of single plays, not of trilogies, even less of tetralogies. There is then a fundamental discrepancy between the selection processes at work in Athenian tragic contests and the tragic canon as we now know it. This is the point we want to make first: one must not infer a direct link between the tragic contests of the 5th century BCE and the canon of Greek tragedies as we now have it, consisting of 32 tragedies. Not all competitions, not all selections lead to the constitution of a canon. Best example of this: there were contests for dithyrambs too, and we have kept nearly nothing of them.

As soon as we have made this point, however, we must add some nuance. There was in a way something more than an instantaneous value of the prize awarded to the poet. Many other lesser tragic festivals were organized in Athens and outside Athens, in Attic or elsewhere, the rural Dionysia for instance, and we know that in those festivals some of the plays that had been performed at the Great Dionysia, and perhaps preferably those that had been awarded the prize, were performed again. Success in the major Athenian festivals could then produce larger success in Athens and outside Athens. This may have been the beginning of a canonization process as soon as the 5th century BCE.

Another parameter to be taken into account: monumental lists of winners and of rankings were erected in the city, reminding everyone of the history of the festivals, with a cumulative effect. Easy calculations could show that Sophocles, for instance, had won more contests than any other playwright. Aeschylus came second. One century later, however, Euripides was eventually considered as the best playwright of all, which still marks a disconnection between the rankings of the 5th century and the canonization process that took place later on.

Copies of the texts of the plays may have circulated as soon as the 5th century: that may explain the intertextual effects we can observe between plays from different playwrights, and also explain how Aristophanes could make such precise quotations in The Frogs. After Aeschylus’ death in 456 Athens had given permission for the continued production of his plays in festivals in competition against living writers. Although we do not know from the extant record whether such revivals really happened in the 5th century, Aristophanes’ claim in The Frogs that Aeschylus’ poetry did not die with him (in contrast with Euripides’) may allude to that revival practice.³

Patrimonialization effects may also have been favoured by the importance of families or clans in the field of theater: families of actors and of dramatists (both were linked together). We know for instance that Sophocles’ grandson, Sophocles the Younger, premiered in 401 the play Oedipus in Colonus, since his grandfather had died in 405. The dramatist Astydamas, who was very popular in the 4th century (he was credited with 15 victories) belonged to Aeschylus’ family: his father, his grandfather, his great-grandfather (who was Aeschylus’ nephew), his brother worked all in the theater business as poets and actors.⁴ Those families kept the original texts of the plays, and their interest was to promote themselves as a brand, which may have contributed to the perennialization of the older plays of the family.

2 Patrimonialization and Repertoire of Tragedies in the 4th Century BCE

From 386 onwards, the Great Dionysia included the revival of an old tragedy by an actor, although we do not know whether the practice was regular before 341. Here again we can see how single tragedies were favoured against trilogies. Those revivals coincided with the progressive rise of actors as the major players of tragic contests, although in the 5th century this role was devoted to dramatists. In 341 and 340 the actor Neoptolemus organized the revivals of plays by Euripides: his fame and his wealth never stopped rising from that time.

We shall leave aside here the role played by stateman Lycurgus in choosing publicly Aeschylus, Sophocles and Euripides as the three major playwrights of the classical period: our reader is invited to refer to André Lardinois’ paper on this topic in the same volume.

Aristotle’s Poetics (written around 340) is the best extant testimony we have of the shift in the use of tragedies in the period. For Aristotle, tragic performances were quite superfluous; they provided only a pleasure of a vulgar order. Tragedies interested him more as texts than as performances and, according to him, it is as mere texts that tragedies could achieve their best effects. The shift from a living art (that is performance) to the practice of reading made easier, of course, the constitution of a canon, for texts can be conveniently stored and reproduced.

In the Poetics however the tragic canon proposed was still more diverse than ours: Aristotle quoted 15 playwrights and alluded to 60 plays, many more than our current canon. One could intuitively presume that Aristotle’s preferences had an impact on our canon, but this is not so clear. Sure, our three great playwrights, Sophocles, Euripides and Aeschylus, were also the most quoted playwrights in the Poetics, but Aristotle did not stay at all at the origin of this choice: he only followed the fashion of his time. And if we look more closely at Aristotle’s assessments of the plays he is commenting upon, then it becomes clear that his influence was quite shallow, if any.

For instance, in the Poetics the most quoted plays were Oedipus Rex (seven times), Iphigenia in Tauris (four times), Medea (three times) and Orestes (twice), and Medea and Orestes were quoted only as bad examples of tragedies. However, Euripides’ canon in the 2nd century CE retained Medea and Orestes, which Aristotle disliked, and left Iphigenia in Tauris aside, although this play was heavily praised by the philosopher. The fact that Aristotle also praised another play by Euripides, Cresphontes, did not prevent the text of this play to be completely lost, and if we eventually preserved the text of Iphigenia in Tauris, it is by chance alone, as we shall see below. This lack of influence of Aristotle’s Poetics on the canonization of tragedies is actually congruent with the largely acknowledged fact that “neither before nor after the alleged loss of Aristotle’s esoteric writings does the Poetics seem to have been widely read”.⁵

True, our current tragic canon is dependent on the choice made by Athenians in the 4th century of choosing Aeschylus, Sophocles and Euripides as their best playwrights, although there is a long way between that moment and our own small and residual corpus of tragedies, reduced to 32 tragedies. We have not even kept the whole corpuses of the three playwrights: Aeschylus wrote 90 plays, Sophocles 123, and Euripides 92.

Our oldest preserved tragedy is Aeschylus’ Persians, created in 472. The most recent is from 401: it is Oedipus at Colonus, by Sophocles, which was performed posthumously. We can easily calculate the total number of tragedies performed during the major festival, the Great Dionysia, between 472 and 401. Three playwrights took part in the competition each year, and each one of them presented a tragic trilogy, that is three tragedies, plus a satyr play. Nine tragedies every year for 72 years make 648 in all, but we must add also all the tragedies that were performed in less important festivals like the Lenaia, and all those that were shown outside of Athens – and we have also to remember that the history of tragedy did not start in 472 and did not stop in 401. So the total of 648 should be easily doubled or tripled or even more. With our 32 preserved complete tragedies we have less than 5% of the production of tragedies in ancient Greece, and a figure around 1% or 2% is much more likely.

What happened in the meanwhile? Although the prominence of the three great tragedians was never questioned, it is sure that after the 4th century many more playwrights were performed, read or studied in the Greek world. In order to get a general view of the popularity of plays, a careful study should be made of the reception of tragedies in the Hellenistic period, taking into account not only the textual quotations, but also such material documents as vase paintings, whatever difficult they are to interpret.

3 The Canonic Choice of Tragedies of the 2nd Century CE

The second stage of the canonization process for tragedies, as we know it, took place much later than the 4th century BCE: at the end of the 2nd century CE (maybe prepared by a progressive selection process which lasted for centuries). At that time an anthology of 24 plays was used in schools, which contained:

• seven tragedies by Aeschylus, in this order: Prometheus Bound, Seven against Thebes, The Persians, The Oresteia (Agamemnon, The Libation Bearers, The Eumenides), and The Suppliants;

• seven by Sophocles (the first three were Ajax, Electra, and Oedipus Rex; we are not sure of the order of the last four: Antigone, Oedipus at Colonus, The Women of Trachis, and Philoctetes);

• ten by Euripides, in the following order: Hecuba, Orestes, Phoenician Women, Hippolytus, Medea, Alcestis, Andromache, Rhesus (not considered as the work of Euripides anymore), The Trojan Women, and Bacchae.

This anthology was copied, it spread out, and saved eventually most of what we know of Attic tragedy.⁶ The hypothesis of the existence of such an anthology was made in the 19th century, and it was afterwards accepted by most scholars. It explains why in the manuscripts the plays come most often in the same order, with scholia, and why after the 2nd century the papyri with plays not included among those 24 tragedies became suddenly scarce.

The academic choice of the 2nd century must have obeyed a lot of criteria – educational, aesthetic, moral, philosophical, religious, political or ideological in general –, but in no way can it be seen today as a neutral reflection of reality, which only a random choice could have provided. The authors of this anthology may have retained the texts they preferred, or those that tradition preferred, or the easiest ones, or the least inappropriate, or even, as we would like to believe, the most representative of the variety of all tragedies. The problem is that we will hardly be able to ever know this, because no introduction, no preface, and no instructions were supplied with this selection. It is only likely that the choice was primarily educational, since it allowed references to the Homeric canon, which had already been read by the students (this explains why the tragedies drawing their subject from the Trojan war are so frequent in our current corpus), it arranged a progressive level of difficulty from one play to another, and it provided parallels between the works of the three authors.

According to André Lardinois, the hypothesis could be made that the number of 24 plays may have something to do with the 24 songs into which each Homeric poem was divided: the tragic canon was modelled somehow on the epic canon, and the number 24, like the 24 letters of the Greek alphabet, was a way to anchor the new canon in the tradition. The fact that more plays by Euripides were preserved, compared to Aeschylus and Sophocles, may also be owed to the use of Euripides’ speeches as models in rhetorical education.

There is a way, however, to get a clearer picture of the implicit principles of this anthology: by looking at the plays that were not transmitted by it. For we do not have only the 24 tragedies aforementioned, but no less than 32 in all: to the 24 chosen for the school eight others were added, all of them by Euripides.

There are indeed a few medieval manuscripts that, in addition to the ten plays by Euripides selected for the school, include nine more plays, which come without any scholia, unlike the others. Strangely enough, those nine plays are classified more or less according to the initials of the Greek titles: epsilon with Helen; eta with Elektra, Heracles and Heracles’ Children; kappa with Cyclops; iota with The Suppliants (Hiketides), Ion, Iphigenia in Tauris and Iphigenia at Aulis. This odd classification has an explanation: it follows approximately the alphabetical order of an ancient edition of Euripides’ works, whose partial reproduction was still to be found in Thessaloniki in the 14th century; although that manuscript itself has now disappeared, we still have two copies of it in Florence and Rome, made in the 14th century.⁷ So here we are with nine additional plays, including eight tragedies, miraculously recovered and narrowly escaping full oblivion.

Eight tragedies, this is not much of course, compared with all the tragedies produced in Athens or even with Euripides’ tragedies alone. Yet we should not underestimate the importance of those eight alphabetical plays, which by a rare set of circumstances form a collection totally independent of any academic choice: those plays were not arbitrarily selected for reasons mostly unknown to us, but preserved by accident, according to the random order of the alphabet, and just because an edition of Euripides was in the right place at the right time. That makes a huge difference: it gives us at last a statistically representative sample that may allow us to know Euripides’ tragedies in a more objective way.

The comparison of those alphabetical or random tragedies with the school tragedies gives a striking result indeed. We need not take more than one glance at the list to see that the alphabetical plays include the most atypical tragedies in the whole corpus, and in Euripides’ in particular: Helen, Ion, and Iphigenia in Tauris notably. More specifically, among the eight alphabetical tragedies there is only one drama that ends badly: Heracles. The proportion is exactly the opposite in the ten selected plays, among which two only end well, Orestes and Alcestis (although Alcestis should not be classified among ordinary tragedies since it was performed originally as a satyr play). We could hardly get a more glaring contrast.

4 Euripides and the Modern Idea of Tragedy

Such a discrepancy between the two groups is highly instructive. It explains in particular why Euripides is generally considered in modern times as the least tragic playwright. Modern philosophers considered him as the least tragic of the three playwrights: according to Schlegel and Nietzsche, he was a real destroyer of tragedy – a strange judgment indeed, since Aristotle saw him as the “most tragic poet” (tragikotatos ton poieton) (the most efficient in writing tragedies). The main problem they found with Euripides was the fact that he wrote too many “happy” tragedies, that is tragedies with happy endings, and this did not conform to the idea of tragedy (or tragic) the philosophers conceived at the end of the 18th century – or, to be more precise, to the ideas of tragedy.

Two main conceptions had emerged indeed. One saw tragic as the confrontation between man and destiny. Schelling saw in Oedipus the best illustration of the metaphysical question of freedom: although Oedipus had committed his crimes in spite of himself, he chose to expiate those crimes that were actually committed by fate alone, and in so doing he affirmed the value of human freedom. That commentary by Schelling was the origin of all interpretations that see tragic as the confrontation between man and destiny, an idea that August Wilhelm Schlegel popularized throughout Europe with his Lessons on Dramatic Literature. Still now, if one asks a student of letters about tragedy, there are all chances that he will come out with this definition of tragic – without knowing that it comes actually from Schelling and not from ancient writers.

Hegel brought another interpretation, a dialectical conception of tragedy: according to him the tragic conflict par excellence is not between man and fate, but between two equivalent moral authorities. The model here was less Oedipus the King than Antigone: Creon embodied the raison d’État against Antigone, who spoke for family values and piety. However, for Hegel, such a tragic conflict is only an appearance: it is the sign of a transcendent harmony that cannot be achieved in our world. So Greek tragedies aimed to teach the absolute nature of the moral norm.

In both conceptions, Schelling’s and Hegel’s, the tragedy must end with the hero’s death or his complete defeat, because this is the only solution to the antinomies of reason as revealed by Kant’s philosophy. For Schelling and Hegel, Greek tragedies were to give an answer to the philosophical problems of the German idealism of the 18th century, and this excluded happy tragedies from the picture, and Euripides with them.

However, Aristotle mentioned those happy tragedies, and said that Euripides’ “most tragic” (tragikotatai) plays (that is, according to him, the most efficient as tragedies, the closest to their own essence) were those “which ended with misfortune”. He even opposed explicitly those “who blamed Euripides” for giving an unhappy end to his tragedies, which means that there was in Athens at the time a debate on this question, and that maybe most people preferred happy endings – as in Hollywood movies today. In the 17th century, the French playwright Corneille mentioned the possibility of the tragédie heureuse too, which Italians called tragedia a lieto fine.

In fact, in the 18 tragedies we have kept under Euripides’ name, nine only have an unhappy ending, that is 50%, and without any doubt this relatively low proportion helped to relegate Euripides in the margins of the tragic reflection developed by philosophers in the eighteenth and nineteenth centuries. For Aeschylus, there are 71% of unhappy endings (five out of seven); for Sophocles, 57% (four out of seven). But we have just seen that Euripides’ relatively low rate of unhappy tragedies is actually a mere artefact of the manuscript tradition, which combined two radically different groups of plays: the first ones, arbitrarily selected by the grammarians of the 2nd century CE, end badly with a rate of 80%; but in the second ones, which were preserved in a totally random way, the rate falls to 12% only.

If you consider Euripides’ fragments too, which came also to us in a random way, you come to the same result: when it is possible to extrapolate the endings of fragmentary tragedies by Euripides (it happens in about 30 cases), we find that only 30% of them approximately have an unhappy ending. It is a figure much closer to the picture given by the alphabetic plays than by the canonical plays. If you consider Sophocles’ fragments too in the same way, you arrive to a similar result: around 50% of his tragedies had a happy ending. The calculation is more difficult to make with Aeschylus because he wrote “bound trilogies”, that is tragedies that followed the same story in a trilogy, which means that the end of a single tragedy is not as significant.

5 Stoic Influence on the Tragic Canon

That leads us to the question: why was the canonic choice made by grammarians in the 2nd century CE so unfavourable to happy tragedies?

Clearly, by selecting plays with common characteristics, the ancient academic choice of tragedies worked like a Procrustean bed. This much is certain, in particular: in Euripides’ case, the selected plays fit much more easily the modern concept of tragic than those that were left aside, and such a discrepancy may well help us to understand the origin and the reasons of the choice of the 3rd century CE.

We have seen already that the modern concept of tragic could be summarized very roughly as man’s struggle against transcendence and his crushing by fate. But although the idea of destiny appeared already in Homer, it was expressed in Greek thought and literature through a variety of words and concepts not quite synonymous with each other (ananke, tukhe, moira, heimarmene, aisa, pepromene, potmos, khreon, etc.) – until the 3rd century BCE.⁸ Then Stoicism was invented, and it is only with Stoicism that the concept of fate (heimarmene) began to play a central and structuring role in ancient thought and philosophy.

But significantly enough, the rise of Stoicism coincided with the canonization process of tragedies, which may have taken several centuries. The end of the process, when the canon was eventually stabilized, took place at the end of the 2nd century and the 3rd century CE, that is exactly at the peak of the Stoic influence in the Greco-Roman world, an influence that extended up to the head of State: Emperor Marcus Aurelius adopted Stoicism as his personal philosophy; the Stoic concepts became commonplace, and spread in all the philosophical schools, and in particular the ideas of fate and providence acquired much more meaning and importance than at the time of the three great tragedians.

It is well known that the Stoics liked to quote Oedipus’ story to explain the power of fate, and Seneca’s tragedies show an undisputable preference for unhappy and even disastrous endings:⁹ the choice that was eventually stabilized clearly reflected that preference for works and myths that could echo the philosophical concerns of the time. We can therefore formulate the hypothesis that the tragic conceptualization based in modern times on Sophocles’ and Aeschylus’ works was somehow prepared and facilitated by an academic choice that selected plays fitting more or less the contemporary vision of the world, which was at the time influenced by Stoicism, where fate and providence played some role. And it must also be noticed that the same Stoic influence can be found in the early theology of Christianity, which developed around that time.

Conversely, Euripides’ alphabetical plays fit much less easily that conceptual framework of Stoicism or of early Christianity: they show that the field of tragedy was even more diverse, aesthetically and ideologically, than we would think on the basis of the plays the grammarians preserved for us nearly two thousand years ago.¹⁰

…

Clearly enough, our modern conception of tragedy is dependent on the implicit ideological or esthetical bias which pervaded the long canonization process that culminated in the 2nd century CE. This later canonization was of a different order than the one that took place in the 4th century BCE. It was, of course, already dependent on the choice of the three great playwrights made in Athens in the 4th century, but its aim was less obviously political than pedagogical. We cannot relate it to any explicit decree, such as the one that was taken by Lycurgus in Athens. It is not even sure that the grammarians who made that choice were completely aware of the Stoician ideas that may have influenced them, so pervasive were those ideas at the time. The canonization of tragedies was dependent on a large process of cultural formation implying dramatic mutations in the vision of the world (Weltanschauung), in the conception of nature, Providence, and gods.

Something is sure however: a school choice was stabilized, and this choice propagated rather fast. This suggests at least two last hypotheses:

1. Cultural communication must have gone fast at the time of the greatest extension of the Roman empire, for this speed suited the need for homogenization at the scale of the Empire, at least in the Greek-speaking parts: one empire, one emperor, one culture (or two, for the empire was bilingual), and also, by way of consequence, only one set of Greek tragedies to study in most parts of the empire.

2. The stabilization of a specific set of tragedies was perhaps induced by the need for efficiency in education. Selecting tragedies helped to decrease the number of tragedies that were to be studied, and many reasons may have entailed such a restriction: either less time could be devoted to education, or education costs went higher, or more and poorer people had access to education, or all or some of those reasons together.

Homogenization, need for efficiency and for economy: one may hear in those terms a summary description of our current world, but they may also describe accurately the Roman empire at the end of the 2nd century and the beginning of the 3rd century. Here again Jan Assmann’s proposal seem to fit quite well: “In times of acute polarisation inside cultures, in times of broken tradition where it must be decided which order one should follow, then comes the formation of canons.”¹¹ Restricting educational corpuses was, and is, a way to adapt to an economic crisis, although it might lead later to an educational crisis.

The tremendous success of this school canon of the 2nd century contributed to the preservation of our current corpus of Greek tragedies, and to the formation of the culture of Modern Europe, and for this we may thank those grammarians of the 2nd century. However, the very success of those tragedies contributed also to the progressive oblivion of the other tragedies (a process that had already been going on for a long time before), and eventually to their definitive disappearance. This is both a success and a failure those grammarians of the 2nd century surely never dreamt of achieving.