Chapter 6:
The Composition of the Iliad

Introduction

In the previous chapters I have looked at the Iliad as we have it without paying much attention to questions of origin. My working hypothesis has been that the text can be interpreted according to the same canons and procedures that we use in the analysis of other large-scale works. The hypothesis gained strength from the emergence of an architectonic design and from the discovery that much narrative detail, however traditional if considered on its own terms, assumes a specific structural function through its placement or elaboration. The discriminating use of lion similes (above, p. 116) is typical of the poet's way, and even in the battle scenes, the loosest and most formulaic part of the poem, detail is never allowed to proliferate for long without some control that relates it to the poet's larger narrative ends.

In this chapter I turn to what a long tradition of scholarship has identified as the Homeric Question: How did the Iliad come into being? I ask the question in a modified form: What processes of composition can plausibly account for the highly organised structure modern criticism has discovered in the text? This version of the question has a strong ‘unitarian’ bias and reveals my affinity with those scholars who use the unified nature of our Iliad as their main argument for the position that it must have been the work of a single poet.

Any plausible hypothesis about the process of composition, however, must satisfy conditions beyond its compatibility with the architecture of the poem. It must do justice to Parry's demonstration that the Iliad is rooted in a tradition of formulaic language and must respect more generally those aspects of Homeric art that oral criticism has taught us to see. Finally, a plausible hypothesis must take into account the observations of the analysts, the scholars who have used narrative discontinuities or contradictions as evidence for the position that our Iliad is the work of several hands. This approach to Homer has never found much favour in the Anglo-American world, if only because the zeal of its predominantly German practitioners has often been singularly devoid of common sense. But, however offensive this position is to the intuitions of the common reader, neither oral criticism nor unitarian theories offer satisfactory explanations for the major stumbling-blocks on which the analytical case rests. (See J. A. Davison for a survey of the Homeric Question.)

Repetition and Contextual Surplus

In looking for a theory of composition that is compatible with the findings of major branches of Homeric scholarship, I begin with the analysis of a class of repetitions that previous scholars have either ignored or used without any concern for their status as evidence. I refer to whole lines orlonger passages that are repeated once or twice in the poem. Analysts have blithely derived theories of multiple authorship from such doublets or triplets on the assumption that they could discriminate between original and copy. Oral critics, on the other hand, have not bothered to differentiate such repetitions from the mass of unmistakably formulaic repetitions. But the behaviour of many doublets is incompatible with assumptions of formulaic origin, and the task of constructing a plausible hypothesis of composition is greatly narrowed down if we recognise the co-existence in the Iliad of formulaic and non-formulaic repetitions.

The formula that reveals the oral nature of the Homeric epic is typically a fraction of a line, but there are longer passages as well that are clearly formulaic in usage. There is the entire system of speech-framing lines with its rigid combinations of a saying-phrase with a name-phrase. In addition some forty-five lines occur four times or more often in the Iliad; another twenty lines have four or more occurrences in the Iliad and Odyssey. Their frequency and their occurrence in fixed contexts with random distribution suggests that such lines are formulaic, although we cannot tell, nor does it matter very much, whether these lines or invented some or all of them. Examples of such lines are

ἀλλ᾽ ἄγεθ᾽ ὡς ἂν ἐγὼ εἴπω πειθώμεθα πάντες
Come then, do as I say, let us all be won over (8)
ἀνέρες ἔστε φίλοι, μνήσασθε δὲ θούριδος ἀλκῆς
Friends, be men and remember fierce battle (7)
αὐτίκα δ᾽ ἐξ ὀχέων σὺν τεύχεσιν ἆλτο χαμᾶζε
At once in all his armour he leapt to the ground from his chariot (8)
ἂψ δ᾽ ἑτάρων εἰς ἔθνος ἐχάζετο κῆρ᾽ ἀλεείνων
to avoid death he shrank into the host of his own companions (7)
δούπησεν δὲ πεσών, ἀράβησε δὲ τεύχε᾽ ἐπ᾽ αὐτῷ
He fell thunderously, and his armour clattered about him (6)
A line of this type will occur on average once every forty-five lines, a sufficient frequency to contribute to the formulaic character of the poetry. But, apart from speech-framing and otherwise formulaic lines, there are about 600 passages, ranging from one to nine lines, and adding up to a little over a thousand lines, that are repeated once or twice and account for almost 2,000 correspondences within the Iliad and another 350 correspondences between the Iliad and Odyssey. These figures do not include some forty instances in which a message is repeated or an order is given and executed in the same words. This convention is a striking feature of Homeric narrative, but it has no bearing on questions of composition.

Roughly speaking, every eighth line in the Iliad is a doublet or triplet, although the distribution is quite uneven, ranging from less than ten percent to almost thirty percent for individual books. Some doublets are no doubt formulaic, i.e. they are pre-existing lines that because of the limitations of the Homeric corpus do not reveal their formulaic nature beyond a reasonable doubt. There are other doublets that cannot be formulaic. The clearest examples are the twenty-odd doublets five lines or longer. We may exclude from the outset the possibility that such passages were generated independently. Since a five-line passage will consist of fifteen to twenty phrasal elements, it defies all probability that independent generation, using the same formulas and rules of combination, would produce exactly the same sequence of words over five or more lines on an average of once every 400 lines. The other possibility, that the whole passage is a formula, runs counter to the principle of economy. The longer a passage the more limited its field of application. It is in the interest of the oral poet to travel lightly. Why should he burden his memory with the heavy luggage of whole verse paragraphs when simpler modules and the rules of combination achieve the same result?

With these general considerations in mind let us turn to some examples. In the second book Agamemnon says:

ηομ. ιλ. 2.111 Ζεύς με μέγα Κρονίδης ἄτῃ ἐνέδησε βαρείῃ,
ηομ. ιλ. 2.112 σχέτλιος, ὃς πρὶν μέν μοι ὑπέσχετο καὶ κατένευσεν
ηομ. ιλ. 2.113 Ἴλιον ἐκπέρσαντ᾽ εὐτείχεον ἀπονέεσθαι,
ηομ. ιλ. 2.114 νῦν δὲ κακὴν ἀπάτην βουλεύσατο, καί με κελεύει
ηομ. ιλ. 2.115 δυσκλέα Ἄργος ἱκέσθαι, ἐπεὶ πολὺν ὤλεσα λαόν.
ηομ. ιλ. 2.116 οὕτω που Διῒ μέλλει ὑπερμενέϊ φίλον εἶναι,
ηομ. ιλ. 2.117 ὃς δὴ πολλάων πολίων κατέλυσε κάρηνα
ηομ. ιλ. 2.118 ἠδ᾽ ἔτι καὶ λύσει: τοῦ γὰρ κράτος ἐστὶ μέγιστον.
Hom. Il. 2.111Zeus son of Kronos has caught me fast in bitter futility.
Hom. Il. 2.112He is hard; who before this time promised me and consented
Hom. Il. 2.113that I might sack strong-walled Ilion and sail homeward.
Hom. Il. 2.114Now he has devised a vile deception, and bids me go back
Hom. Il. 2.115to Argos in dishonour having lost many of my people.
Hom. Il. 2.116Such is the way it will be pleasing to Zeus, who is too strong,
Hom. Il. 2.117who before now has broken the crests of many cities
Hom. Il. 2.118and will break them again, since his power is beyond all others.
2.111-118 He continues for another twenty-one lines and concludes:
ηομ. ιλ. 2.139 ἀλλ᾽ ἄγεθ᾽ ὡς ἂν ἐγὼ εἴπω πειθώμεθα πάντες:
ηομ. ιλ. 2.140 φεύγωμεν σὺν νηυσὶ φίλην ἐς πατρίδα γαῖαν:
ηομ. ιλ. 2.141 οὐ γὰρ ἔτι Τροίην αἱρήσομεν εὐρυάγυιαν.
Hom. Il. 2.139Come then, do as I say, let us all be won over; let us
Hom. Il. 2.140run away with our ships to the beloved land of our fathers
Hom. Il. 2.141since no longer now shall we capture Troy of the wide ways.'
2.139-141 At the beginning of Book 9 the same lines occur, where they now serve as the whole of Agamemnon's speech. The hypothesis that the doublet depends on a formula for a leader in despair is surely much less plausible than the alternative that either a short version was expanded or a longer version cut. Exactly the same cut/expansion relationship obtains between the scenes that describe the departure of Hera and Athene from Olympus (Il. 5.719-21 =Il. 8.381-3; Il. 5.733 -7= Il.8.384 -8; Il 5.745-52= Il. 8.389-96). In Books 4 and 8 we find the following passage:
ηομ. ιλ. 4.20 ὣς ἔφαθ᾽, αἳ δ᾽ ἐπέμυξαν Ἀθηναίη τε καὶ Ἥρη:
ηομ. ιλ. 4.21 πλησίαι αἵ γ᾽ ἥσθην, κακὰ δὲ Τρώεσσι μεδέσθην.
ηομ. ιλ. 4.22 ἤτοι Ἀθηναίη ἀκέων ἦν οὐδέ τι εἶπε
ηομ. ιλ. 4.23 σκυζομένη Διῒ πατρί, χόλος δέ μιν ἄγριος ᾕρει:
ηομ. ιλ. 4.24 Ἥρῃ δ᾽ οὐκ ἔχαδε στῆθος χόλον, ἀλλὰ προσηύδα:
ηομ. ιλ. 4.25 αἰνότατε Κρονίδη ποῖον τὸν μῦθον ἔειπες:
Hom. Il. 4.20So he spoke; and Athene and Hera muttered, since they were
Hom. Il. 4.21sitting close to each other, devising evil for the Trojans.
Hom. Il. 4.22Still Athene stayed silent and said nothing, but only
Hom. Il. 4.23sulked at Zeus her father, and savage anger took hold of her.
Hom. Il. 4.24But the heart of Hera could not contain her anger, and she spoke forth:
Hom. Il. 4.25'Majesty, son of Kronos, what sort of thing have you spoken?
4.20-25=8.457-62

Shall we argue that the tradition possessed a six-line formula for the situation in which Hera and Athene are angry at a remark by Zeus, but Athene bites back her anger whereas Hera cannot control herself? Surely the existence of a formula for three specified characters in a specified sequence makes nonsense of all assumptions about the usefulness of a formula system as an improvisational device. The inescapable conclusion for the three cases cited, as well as similar ones, is that passage A is a copy of passage B or vice versa. We may not be able to establish which passage is the original or whether the copyist was the same or a different poet. But the presence of such doublets establishes beyond a reasonable doubt that at least with regard to these passages we are dealing with a process of composition in which there is a fixed text (not necessarily written down) from which a poet copies a passage for use in another context. The implications of that conclusion for the composition of the Iliad are considerable. It is, however, possible that the limited number of long and self-evident instances of copying are the result of interpolation or other special circumstances and tell us little about the composition of the poem as a whole. For this reason it is important to establish whether traces of a model-copy relationship, so apparent in long doublets, can also be found in the much larger number of short doublets.

The evidentiary value of doublets emerges very clearly if we treat them as formulas and check whether their behaviour is compatible with formulaic origin. A formula R may appear in contexts r1, r2... .rn if each context shares the minimum requirements for its correct use. According to a theory of formulaic origin, r1 and r2 are separately related to R, but their relationship to each other is a by-product of formulaic composition, and they stand in a random relationship to one another. Even the most cursory examination of doublets shows that this is not the case. On the contrary, doublets show a high degree of contextual surplus, by which I mean resemblances between r1 and r2 that are not required by the meaning of the line(s) in question. Some contextual surplus will occur randomly or will be the result of specific narrative constraints compatible with formulaic composition. Thus, if we insisted on treating Agamemnon's speech in Books 2 and 9 as two occurrences of the speech of the discouraged leader formula, we should not attach any significance to the fact that Agamemnon is the speaker on both occasions, for that similarity arises from the most general constraints of the narrative. But with the Hera and Athene scenes of Books 5 and 8 we must assume either that there is a departure of Hera and Athene formula or we must account for the contextual surplus that makes them, as opposed to some gods, the subjects of the departure of the gods formula on two occasions.

Contextual surplus has little value as evidence in any individual case, where it can always be dismissed as random. But its cumulative value is high. If we can show that many doublets show contextual surplus of varying kinds and degree, and that its incidence in the aggregate is higher than one would expect on a random basis, the hypothesis that doublets derive independently from formula types becomes less plausible than the hypothesis that many doublets derive from each other. Thus the cumulative weight of contextual surplus is strong evidence for a model-copy relationship among doublets.

Contextual surplus is most striking when it attaches to short passages that do not in themselves raise any suspicions about their formulaic origin. Take the following example. Hippothoos drags the body of Patroklos by the heel

ηομ. ιλ. 17.291 Ἕκτορι καὶ Τρώεσσι χαριζόμενος: τάχα δ᾽ αὐτῷ
ηομ. ιλ. 17.292 ἦλθε κακόν, τό οἱ οὔ τις ἐρύκακεν ἱεμένων περ.
Hom. Il. 17.291for the favour of Hektor and the Trojans, but the sudden evil
Hom. Il. 17.292came to him, and none for all their desire could defend him.
17.291-292=15.449-50 Aias kills him, and the poet speaks a little necrologue:
ηομ. ιλ. 17.301 . . . . οὐδὲ τοκεῦσι
ηομ. ιλ. 17.302 θρέπτρα φίλοις ἀπέδωκε, μινυνθάδιος δέ οἱ αἰὼν
ηομ. ιλ. 17.303 ἔπλεθ᾽ ὑπ᾽ Αἴαντος μεγαθύμου δουρὶ δαμέντι.
Hom. Il. 17.301and he could not
Hom. Il. 17.302render again the care of his dear parents; he was short-lived,
Hom. Il. 17.303beaten down beneath the spear of high-hearted Aias.
17.301-303=4.477-9 The first two lines recur in Book 15, where Teukros, at the command of his half-brother Aias, kills Kleitos. The second doublet is applied to Simoeisios, an earlier victim of Aias. The mutually reinforcing association of these passages with Aias, their contextual surplus, is very strong.

The restriction of a doublet to the same person or to a set of closely related characters is the commonest and clearest form of contextual surplus. Menelaos is a particularly good example because he turns up in a number of doublets that are restricted either to himself, or to himself and his brother, or to himself and Nestor or Antilochos. (In the following examples, Lattimore does not always use the same translation for different occurrences of the same line.)

Restriction to Menelaos

Restriction to Menelaos and Agamemnon

Restriction to Antilochos and Menelaos

Antilochos is a prominent minor character, who usually appears in the company of Menelaos. In Book 5 he comes to the latter's rescue when he is - threatened by Aeneas; in Book 15 he follows Menelaos' encouragement to do something special and kills Melanippos. In Book 17 he is told by Menelaos to bring the news of Patroklos' death to Achilles; in Book 23 he tries to outwit Menelaos in the chariot race and is upbraided by the latter after the race. The narrative connections are probably suggested by the relationship of the houses of Menelaos and Nestor in the tradition. Antilochos or Antilochos and Menelaos form a strong contextual surplus in a number of doublets.

Implications of Contextual Surplus

The passages discussed above show strong and mutually reinforcing contextual surplus. They raise the question why seventeen disparate and often quite nondescript doublets should be restricted in their use to contexts that involve the presence of Menelaos, Menelaos and his brother, or Menelaos and the family of Nestor. Similar networks of doublets with marked context restrictions exist for Hektor, Hektor/Paris, Hektor/ Patroklos/Sarpedon, Aias, Aias/Menelaos, Aeneas, Deiphobos, Meriones/ Idomeneus. By far the most extensive network, however, links Achilles to himself and to his substitutes Patroklos and Diomedes. There is not enough space to develop this network in detail here; I conclude instead with a set of Odysseus-Sarpedon doublets that are remarkable both for their quirkiness and because they extend into the Odyssey

This contextual surplus relating Odysseus, Sarpedon and Menestheus is remarkable both for its specificity and its pointlessness. Unlike the Menelaos passages, the Odysseus-Sarpedon passages cannot be coordinated with any narrative tradition or purpose. But the very quirkiness of the association strengthens its evidential value.

Contextual surplus can take other forms than identity or association of characters. Thus the line ‘stabbed with the bronze-pointed spear and unstrung his sinews’ 4.469 = 11.260 both times refers to a victim who had been trying to drag another body by the foot. The line ‘raged on among the champions until so he lost his dear life’ 11.342 = 20.412 refers to warriors trusting in the swiftness of their feet. More strikingly, the line both times occurs in the vicinity of a reluctant father. In Book 20 we hear that Priam would not let his favourite son Polydoros join the fighting, but he disobeyed his father. In Book 11, where Agastrophos is the victim, the motif occurs in the immediately preceding killing, where the sons of the prophet Merops fall to Diomedes after ignoring their father's advice.

A third example looks far-fetched at first sight but gains some force from the dense network of Diomedes-Achilles doublets:

ηομ. ιλ. 5.590 τοὺς δ᾽ ἕκτωρ ἐνόησε κατὰ στίχας, ὦρτο δ᾽ ἐπ᾽ αὐτοὺς
ηομ. ιλ. 5.591 κεκληγὼς: ἅμα δὲ τρώων εἵποντο φάλαγγες
Hom. Il. 5.590Hektor saw them across the ranks, and drove on against them
Hom. Il. 5.591crying aloud, and with him followed the Trojan battalions
5.590-591=11.343-4 This doublet occurs both times in the context of a Hektor-Diomedes encounter, but I am interested here in the phrase ‘drove on against them’ (ôrto d'ep' autôi), which occurs only in this doublet and in the river fight, where the river does not let go of Achilles but ‘rose on him’ (ôrto d'ep' autôi 21.248). If we now return to Book 5, we find Diomedes frightened by Hektor like
ηομ. ιλ. 5.597 ὡς δ᾽ ὅτ᾽ ἀνὴρ ἀπάλαμνος ἰὼν πολέος πεδίοιο
ηομ. ιλ. 5.598 στήῃ ἐπ᾽ ὠκυρόῳ ποταμῷ ἅλαδε προρέοντι
ηομ. ιλ. 5.599 ἀφρῷ μορμύροντα ἰδών, ἀνά τ᾽ ἔδραμ᾽ ὀπίσσω,
Hom. Il. 5.597and like a man in his helplessness who, crossing a great plain,
Hom. Il. 5.598stands at the edge of a fast-running river that dashes seaward,
Hom. Il. 5.599and watches it thundering into white water, and leaps a pace backward,
5.597-599 The association of the rare theme of death by water with the humdrum phrase ôrto d 'ep 'autôi/autous in two of its three occurrences is noteworthy, all the more so since the coincidence fits the firmly established pattern according to which Diomedes undergoes the trials of Achilles in a palliated form.

A weak form of contextual surplus appears in doublets that are restricted to the same book or neighbouring books. Such doublets are especially frequent in the fighting scenes of Books 13-15. A much more interesting problem is posed by clusters of doublets that relate two books to each other. A good example is the relationship of Books 3 and 19. These books share five doublets for a total of ten lines. The longest is the standard arming passage, which refers here to Paris and Achilles but also occurs in a varied form in other books (Il 3.330- 2, 334-5 = Il. 19.369-73). Then there are three doublets that refer to a sacrifice made in confirmation of an oath (Il. 3.271 2 = Il. 19.252 3, Il. 3.279=Il. 19.260, Il 3.292= Il. 19.266). That arming and oath scenes should occur in close proximity on two occasions and produce verbal resemblances is not in itself remarkable. But an additional doublet puts the others in a new light and generates strong contextual surplus. Helen speaks of ‘my own brothers, born with me of a single mother’ 3.238, whereas Briseis in her lament for Patroklos speaks of ‘my three brothers, born with me of a single mother’ 19.293.

The pervasive evidence of contextual surplus permits certain minimal conclusions to be established with great certainty. Contextual surplus is the product of an individual mind that betrays its presence in the text by the persistence and sometimes quirkiness of its associations. This mind does not operate with a stock of formulas but copies parts of a particular text inscribed in its memory. Contextual surplus by itself cannot tell us whether the mind remembers its own productions or those of another, but it is sufficient proof that the remembered passage is a part of a fixed text, which may or may not have been written down. It follows that there are two types of repetition and memory that prove with equal force the existence of two different processes of composition. The system of noun-epithet formulas and related phenomena points to a traditional diction that served the needs of oral poets in whose work improvisation played a dominant role. Doublets with contextual surplus point to a process of composition in which new lines are created by re-using or adapting fragments from a previously existing text. The collective memory of formulaic diction points to a process of composition to which the ordinary concept of a text is quite inappropriate. The individual memory of contextual surplus presupposes on the part of the author a sense of textual identity that is fundamentally like ours, although it is characterised by a very high tolerance to verbatim repetition. As it turns out, the account of composition that reconciles these two types of memory and repetition also permits us to integrate into one hypothesis the major insights of analytical, unitarian and oral theories about the composition of the Iliad.

The Composition of the Iliad

The Development of the Epic Poet

A plausible account of the composition of the Iliad must respect the constraints that different aspects of our knowledge of the poem exercise on each other. The evidence for an individual memory and for the model copy relationship of many doublets is of itself compatible with theories of single and multiple authorship. The large-scale structural coherence of the work, on the other hand, is a strong ‘argument from design’ for single authorship. The formulaic elements of the style finally point to the rootedness of the poem in an established tradition. The mutual constraints of these elements point to a process of gradual composition, the stages of which are to some extent recoverable.

The author of the Iliad grew up in a tradition of oral versemaking that provided him with an extraordinarily powerful instrument, the language of hexametric narrative. He learnt his craft and assembled a repertoire of familiar stories to be recited on festive occasions of various kinds. A typical performance would last an hour or two; perhaps some were as long as Odysseus' narrative at the court of the Phaeacians. But it is difficult to imagine an institutional context that would support more extended narratives. The poet's repertoire in all probability did not consist of fixed texts, but we may also assume that two performances by him of a given story would resemble one another more closely than either one would a telling of the same story by another singer. Such resemblances would include the scope of the story as a part of the total repertoire, the arrangement and weighting of elements within that scope, and elements of diction and style. We should allow the young Homer one to two decades to develop into a master of his craft, with a complete repertoire and a style and excellence recognisable by his audience.

Whether or not we want to attribute innovative tendencies to Homer, we must think of him not only as a master of his craft but also as a supremely gifted man who, like Shakespeare, Bach or Mozart, could not help transforming whatever he touched. Aristotle saw the particular excellence of the Homeric poems in their control of large-scale narrative, and since this quality distinguishes them from all other heroic poems it is reasonable to look for a specific quality of Homer's genius in his sense for large-scale narrative. We arrive thus at the situation of a master-poet who decides at a certain point to extend the scope of heroic narrative and to compose a work of encyclopaedic dimensions that would incorporate much of his repertoire. The context and motives of that decision are shrouded in darkness, but some speculation may be permitted. One aspect of a poet's greatness is his ability to give authoritative expression to the deepest concerns of his age. Dante and Milton, the most autobiographical of the great epic poets, were open about the relationship of their poetic ambition to the religious and political conflicts of medieval Florence and seventeenth-century England. Did Homer respond similarly to the sense of pan-Hellenic identity of which we find other traces in the eighth century BC? Did this response trigger the break with the tradition that led to the Iliad as a poem much longer and more complex than any of its predecessors, or should we look for the cause of this transformation merely in the individual excellence and ambition of a monumental composer? (Kirk, 1962, 280) Was the Ionian poet from Asia Minor attracted to the story of Troy because, like the author of the Aeneid, he wanted to account for his cultural indebtedness to a homeland beyond the sea? Such questions are of course unanswerable, but it is a fact that within a generation, perhaps even in the poet's lifetime, the Iliad became a founding text of Greek culture. It is tempting to relate this historical achievement to an intention and to see Homer's ambitious design for a monumental poem prompted by the desire to give shape to the growing Hellenic consciousness of his age. At the very least, there is no reason to use his status as an oral poet to deny the presence in his career of motives that are more clearly articulated and more easily traced against a known historical background in the works of Vergil, Dante and Milton.

If it took Homer the better part of two decades to become a master-poet, the execution of his design for a grand poem must have been the work of further years, perhaps decades. Let us assume, for argument's sake, that the composition of the Iliad lasted as long as the Trojan War. We can hardly imagine Homer delaying publication to the moment of perfection. Nor can we imagine him reciting from work in progress, as Vergil did with the Aeneid. Neither of these alternatives is easily reconciled with what we know from the Odyssey and other heroic traditions about the circumstances of a professional singer. Instead we should think of a gradual process of composition in which the poem was always complete after a fashion, having a beginning, a middle and an end. Thus, during Homer's maturity there was always something like a complete Iliad, but it was always expanding. (A distant analogue would be the composition of Goethe's Faust, of which there also was a complete/incomplete version over a period of fifty years.) We may assume that the plan and scope of the work underwent modifications over the years but that it was part of the poet's genius to have conceived of a design capable of such expansion.

The assumption of an evolutionary process in which an always complete Iliad was always expanding is eminently compatible with the one distinctly oral feature of Homeric narrative that may justly be said to operate with equal force at all narrative levels. G. S. Kirk (1976, 78), who has given the best description of this feature, calls it the cumulative style (other names for it are additive or paratactic). Unlike a Latin period, the Homeric sentence is complete from a very early stage but can be greatly elaborated through additions. The first paragraph of the Iliad is a good example.

‘Mênin aeide thea’ is a complete sentence to which the poet adds elements each of which leaves the sentence complete but capable of further expansion:

ηομ. ιλ. 1.1 μῆνιν ἄειδε θεὰ πηληϊάδεω Ἀχιλῆος
ηομ. ιλ. 1.2 οὐλομένην, ἣ μυρί᾽ Ἀχαιοῖς ἄλγε᾽ ἔθηκε,
ηομ. ιλ. 1.3 πολλὰς δ᾽ ἰφθίμους ψυχὰς Ἄϊδι προΐαψεν
ηομ. ιλ. 1.4 ἡρώων, αὐτοὺς δὲ ἑλώρια τεῦχε κύνεσσιν
ηομ. ιλ. 1.5 οἰωνοῖσί τε πᾶσι, Διὸς δ᾽ ἐτελείετο βουλή,
ηομ. ιλ. 1.6 ἐξ οὗ δὴ τὰ πρῶτα διαστήτην ἐρίσαντε
ηομ. ιλ. 1.7 Ἀτρεΐδης τε ἄναξ ἀνδρῶν καὶ δῖος Ἀχιλλεύς.
ηομ. ιλ. 1.8 τίς τ᾽ ἄρ σφωε θεῶν ἔριδι ξυνέηκε μάχεσθαι;
Hom. Il. 1.1Sing, goddess, the anger of Peleus' son Achilleus
Hom. Il. 1.2and its devastation, which put pains thousandfold upon the Achaians,
Hom. Il. 1.3hurled in their multitudes to the house of Hades strong souls
Hom. Il. 1.4of heroes, but gave their bodies to be the delicate feasting
Hom. Il. 1.5of dogs, of all birds, and the will of Zeus was accomplished
Hom. Il. 1.6since that time when first there stood in division of conflict
Hom. Il. 1.7Atreus' son the lord of men and brilliant Achilleus.
1.1-7 The forward surge and cumulative force of this magnificent paragraph are not arrested by any suspension of meaning except for the very end, where the postponement of the subject into the next line creates a two-line semantic unit with great closing force. This procedure, in which modification is achieved without revision or loss of forward motion, clearly has its origin in a technique of oral versemaking.

A procedure that allows for continuous forward composition also tolerates subsequent insertions of additional materials: the joints of a narrative stretch composed in this fashion are rarely so tight that something cannot be squeezed in. Take the example of the Catalogue of the Ships, which shows the technique operating at several levels. This is almost certainly a piece of narrative that existed once in a somewhat different form and was inserted in its present place after minor modifications that themselves took the form of additions. At one time the Catalogue may have been a simple muster of troops. The text acquires its naval character through the addition after each contingent of a line specifying the number of ships. We would not miss these lines if they had not been transmitted. Moreover, at the beginning of the Catalogue the poet invokes the Muse to tell him the names of the leaders rather than the crowd, but the invocation ends with a detachable specification of the leaders as the leaders of the ships 2.493.

The Catalogue certainly was composed to reflect the Achaean forces at the beginning rather than at the end of the war. For instance, it lists Protesilaos, the first man to die on Trojan soil, and Philoktetes, who was left behind on Lemnos. The resulting editorial problem was solved by addition rather than subtraction: to the mention of these two leaders is in each case added a detachable commentary (2.699-709, 721-8) that brings the story up to date. Once adjusted by means of appropriate additions, the whole Catalogue becomes an addition that is inserted at a joint in the narrative.

The epic that accumulated over ten or more years as a consequence of the poet's expansion of an elastic design differed from his previous repertoire in being a fixed text to be recited verbatim rather than composed anew with each performance. It is very likely that the concept of a fixed text entered Greek culture through the creation of the Iliad as a poem of such scope, excellence and complexity that it had to be protected from the vagaries of performance. Most Homeric scholars believe—and, I think, rightly—that the existence of the Iliad as a fixed text is closely related to the introduction of writing into Greek culture some two generations before the life of Homer. But the concept of a fixed text does not depend on writing, and it would be misleading to think of the availability of writing as the chief cause of the transformation of an oral tradition into the monumental epic.

Shortly after Homer, the heroic tradition went into a tailspin from which it never recovered. Following a line of argument that W. J. Bate in The Burden of the Past (p. 82) has traced back to Velleius Paterculus, G.S. Kirk (1976, 2-3) has argued that the cause of that decline lies in the very excellence of the Iliad and Odyssey. These poems exhausted the possibilities of the genre, discouraged creative imitation, and replaced innovation with preservation. Like tragedy three centuries later, the epic died after a burst of stupendous energy only to survive as a fixed monument. If we accept this plausible analysis, for which the history of culture provides many analogues, the concept of a fixed text is the product of a collusion between Homer and his audience. The author, after developing from the fluid tradition a repertoire that bore the imprint of his personality and over the years had grown more stable, took the further step of refashioning and freezing this repertoire into a monumental poem. The audience responded to this extraordinary ambition and recognised the poem as a classic in need of preservation. The birth of the Iliad was the death of the heroic tradition.

The role of writing in the creation of the Iliad as a fixed text is much disputed. Kirk believes that Homer did not write and that the Iliad and Odyssey were transmitted orally in a substantially unchanged form for several generations until they were written down in the middle of the sixth century. This is a minority view, but Kirk's arguments have the great virtue of stressing that the concept of a fixed text is not dependent on writing. On the other hand, it is tempting to see in the availability of writing a powerful contributing cause to the growth of the monumental epic, not to speak of the problems of transmission.

The Mycenaean Greeks used writing for administrative and commercial purposes (Linear B). After the collapse of their civilisation, writing disappeared from Greece for about four centuries. Towards the end of the ninth century BC, the Greeks adapted an alphabet from the Phoenicians. The date is suggested by the resemblance of the Greek alphabet to Phoenician scripts current during that period. We know very little about the spread or degree of literacy in Greece until roughly 600 BC when inscriptions become slightly more frequent. Only a handful of documents can with any confidence be dated before 700 BC. The most famous of these - the early date is not uncontested - is a cup found in 1953 in Ischia. It is a plain piece of pottery with a three-line inscription, partly iambic and partly hexametric. The inscription is fragmentary; a plausible reading attributes to the modest cup aphrodisiac powers and therefore ranks it above the ‘cup of Nestor’ (Nestoros poterion), about which Book 11 of the Iliad has much to say. It is a striking feature of the inscription that its orthography and punctuation imply a system of prosodic notation for hexametric verse. The few other inscriptions from around 700 RC also are in hexameters.

Because of the archaising tendency of heroic poetry, writing is never mentioned in the Iliad and Odyssey, with one exception. In Glaukos' narrative of Bellerophon we read that Proitos sent Bellerophon to Lykia ‘and handed him many murderous symbols, which he inscribed in a folding tablet’ (6.169-9). Most scholars think of these lines as a heroically stylized description of a letter, both because of the emphasis on many signs and because of the use of the word graphein, which usually means scratch or graze in Homer but here is used in a manner that anticipates its later usage as the standard for writing.

The evidence, slender as it is, is sufficient to establish the fact that writing was known in Homer's day. There is, however, no positive evidence to suggest that Homer's contemporaries had the resources to record, store and transmit a document as long as the Iliad. The early fixing and transmission of the Homeric texts will forever remain a mystery. Our conviction that we do indeed have ‘Homer's Iliad’ rests on one of two assumptions, both of which involve a leap into the dark and beyond the available evidence. On one assumption the fixed text was transmitted by memory over several generations. There is no analogue for verbatim transmission of texts of such complexity, with the possible exception of the RigVeda. On the other assumption, we must attribute to Homer's contemporaries the ability to handle lengthy and complicated written documents. The chief difficulty with this assumption is not technological but historical. If the Greeks of the late eighth century BCE had this capacity, why does the historical memory of classical Greece not extend back to Homer but blurs shortly before 600BCE?

On balance, the assumption of a written Iliad causes fewer difficulties than the assumption of its oral transmission over several generations. The history of the world is full of inventions and discoveries whose obvious fields of application were long ignored. Thus it does not strain belief (and is compatible with the epigraphic evidence) that writing, which was adapted from the Phoenicians, presumably for trading purposes, was used by Homer to create and record the monumental epic, but that the technological capacity implicit in this achievement remained specialised for several generations and was not applied to other forms of extended discourse until much later. In the ancient biographical tradition about Homer we read that his dowry to his daughter was a copy of the Cypria, a lost epic about the events preceding the Iliad. The anecdote may be historical in pointing to a copy of the text as the professional poet's chief capital.

It is likely, then, that Homer saw the possibility that writing offered to his ambitions and that he used it to create a text that differed from earlier heroic poetry in being longer, more complex, and fixed. But Homer the writer did not unlearn the skills of oral versemaking. On the contrary, the technical constraints of writing in his days reinforced the procedure of cumulative composition. Even in an age of Xerox machines and word processors revision of a text in the light of subsequent changes is a tedious, complicated and error-prone business. In Homer's day, when papyrus must have been scarce and writing laborious, a text once written represented an investment one would touch only with reluctance. On the basis of such considerations G. P. Goold has developed a model of the progressive fixation of the Homeric texts, and some such procedure most readily accounts for the abundance of doublets in the text. Above all, such a hypothesis resolves the paradox that the Iliad is both a magnificently designed and, by our standards, a poorly edited poem.

Homer possessed a copy of the Iliad, to which he added over the years until it reached its present shape. The existence of a text that was at any one time fixed but subject to further additions had an effect on the nature of the additions. The poet did not mind repeating himself. Oral versemaking had accustomed him to a very high level of tolerance to repetition at the phrase level, and such conventions as the verbatim repetition of messages established precedents for the repetition of longer passages. The move to a fixed text did not immediately diminish the tolerance to repetition, but created new forms of it. The poet who had composed one speech by a discouraged Agamemnon had no scruples in using a cut version of it when he needed a similar speech for the subordinate purpose of introducing the Embassy. On many other occasions, the poet unconsciously remembered his own combinations of formulaic phrases, as with rigêsen d'ar'epeita (then shuddered), which occurs only with anax andrôn Agamemnon (the lord of men Agamemnon). In other words, through convenience and inertia, the fixed text replaced the formulaic repertoire as the source of new lines. The force of that development increased with time. Let us imagine Homer in the eighth year of his Iliad, with a text that has grown to 12,000 lines, or 75 per cent of its final length. For the poet at this stage of his career, the existing Iliad was a much more powerful determinant than the heroic tradition. The last 4,000 lines of the Iliad are Iliadic rather than heroic hexameters, composed for a specific place in a magnum opus and reflecting the pressures of that work both in their sameness and difference. But Homer the writer did not yet face any incentives to avoid repetition, and the conditions of recitation ensured that even extended doublets would, for the most part, lie below the threshold of recognition.

The bulk of doublets, including the very long ones, cannot be interpreted as the product of a poetic intention but are a carry-over of formulaic habits into a new mode of composition. Of the splendid lines portraying the clash of armies (4.446-51 = 8.60-5) one can only say that the poet used them again, presumably because he thought well of them. Nothing is gained by reading one context in the light of the other. Sometimes the contexts get in each other's way. The famous prophecy about the fall of Troy is spoken by Agamemnon to Menelaos and by Hektor to Andromache (4.163-5 = 6.447-9). Most modern readers will find the echo disturbing and wish the words had been reserved for Hektor, in whose mouth they resonate with pathos and irony, an effect that is undercut by the vindictive certainty of the hysterical Agamemnon. There are, however, occasions when the doublets establish a significant link between their contexts. Whether the poet intended these links or whether they are a byproduct of his associative memory we cannot tell, but the effects yield to interpretation. While such semantically charged doublets are a minority it would be a mistake to disregard them simply because the majority of repetitions are inert.

Occasionally, it seems possible to speak of a deliberate effect. Thus, the moment of death for Patroklos and Hektor is described in an identical three-line sequence. When Thetis visits Achilles for the first time she says:

ηομ. ιλ. 1.362 τέκνον τί κλαίεις; τί δέ σε φρένας ἵκετο πένθος;
ηομ. ιλ. 1.363 ἐξαύδα, μὴ κεῦθε νόῳ, ἵνα εἴδομεν ἄμφω.
Hom. Il. 1.362child, do you lament? What sorrow has come to your heart now?
Hom. Il. 1.363Tell me, do not hide it in your mind, and thus we shall both know.
1.362-363 She repeats the same question, with a slight variation of the second line, when she visits Achilles the second time 18.73. A precise echo of the second line, however, occurs when Achilles addresses the weeping Patroklos (1.363=16.19. The sympathetic question almost becomes a refrain articulating the stages of Patroklos' career. It is also deeply ironic that, in the conclusion of his fighting speech to the Myrmidons, Patroklos repeats the words of Achilles to Thetis:
ηομ. ιλ. 1.411 γνῷ δὲ καὶ ἀτρεΐδης εὐρὺ κρείων ἀγαμέμνων
ηομ. ιλ. 1.412 ἣν ἄτην ὅτ᾽ ἄριστον ἀχαιῶν οὐδὲν ἔτισεν.
Hom. Il. 1.411that Atreus' son wide-ruling Agamemnon may recognize
Hom. Il. 1.412his madness, that he did no honour to the best of the Achaians.
1.411-412 = 16.273-4

When Hektor reaches the ships, the Achaeans are pushed back to the point from which they started in Book 2. Both passages include a bird image with the identical line ‘of geese, and of cranes and of swans longthroated’ 2.460=15.692. This is not the only time that the narrative uses an identical line to return to the same spot. At the end of Book 15 the poet turns from the retreating Aias to Achilles and Patroklos. When he returns to Aias, he repeats himself: ‘the volleys were too much for Aias, who could no longer hold his place’ 15.727=16.102. When Patroklos encourages the Myrmidons to attack the Trojans, they cheer just as the Achaeans had cheered the words of Odysseus (2.333-4 = 16.276-7); the doublet marks the recovery from a nadir. Twenty lines later the flight of the Trojans ‘along the hollow ships’ takes the same form as that of the Achaeans when Hektor smashed the gates (12.471=16.296).

The Stages of the Iliad: A Rough Sketch

The evidence of the doublets, together with our knowledge of the poetic craft in which Homer's artistry is rooted, suggests that the Iliad grew over a period of many years and in a cumulative fashion. Is it possible to go farther and to reconstruct the stages of its genesis? The answer depends on the extent to which we believe that Goold's ‘progressive fixation of the text’ took place without revision of previous elements. If the text grew simply by addition, then it should in principle be possible to recover earlier stages through a process of subtraction, using as evidence the narrative discontinuities to be expected when insertions strain the original narrative joints. On the other hand, there is a great difference between no revision and some revision. We may well believe that the poet was reluctant to adjust his earlier narrative in the light of subsequent additions. We may also concede, without calling into question the excellence of his design or brilliance of his execution, that he was not a very good nuts-and-bolts editor of his own work. But that the progressive fixation of the text proceeded entirely without deletion or revision is highly improbable, given the size and complexity of the work. Since even a modest amount of deletion and revision, with its possibilities for feedback between early and late passages, can effectively block the path to the origin of the text, precise reconstruction, the dream of much nineteenth- and twentieth-century German scholarship, is a chimerical enterprise. On the other hand,the rough stages of the poem's growth are visible and have been well known since the nineteenth century, although scholars of that period invariably attributed the stages to different hands. Gottfried Hermann was the first scholar to develop a systematic expansionist theory early in the nineteenth century. In the English-speaking world, the theory gained currency through its inclusion in Grote's History of Greece and through the great commentary by Walter Leaf, which still provides the sanest and most comprehensive discussion of the narrative stumbling-blocks that have given rise to theories of multiple authorship.

In addition to uncertainty about the extent of revision, the task of reconstruction is made more difficult by the nature of the evidence. Homer's language is a linguistic and cultural amalgam of several centuries. It is possible to identify early and late elements in this amalgam with great confidence, but of course the evidence thus obtained is quite irrelevant to the task of identifying individual strata in a work that may have grown into its current form over the much shorter span of a single lifetime. Linguistic evidence may be helpful in identifying some limited post-Homeric interpolations, but for the bulk of the Iliad we must assume that the amalgam did not change significantly during the poet's life.

The evidence for different stages of the Iliad lies in various forms of narrative discontinuity resulting from the poet's indifference to some unintended consequences of his additions to the text. There are two difficulties with such evidence. First, it is much more difficult to reach agreement on what counts as a narrative discontinuity than to identify a Mycenaean relic or an Aeolian morphological form. Second, narrative discontinuity engenders critical ingenuity. Interpretative strategies of all kinds, from allegorical exegesis onward, have been developed to make sense of textual difficulties. There is no narrative discontinuity that a skilled interpreter cannot fill with meaning; whether it should be filled is a matter of judgement and rests on one's assessment of the range of narrative conventions operative in the work. My own sense is that Homeric narrative, while very subtle in its use of juxtaposition and implicit contrasts, is very straightforward in its concern for causes and consequences—witness the opening lines of the Iliad. Any interpretation that violates the straightforwardness of Homeric narrative may miss the mark as easily as one that shrugs off problems of coherence as inevitable by-products of oral composition.

The two major forms of discontinuity are cracks in the narrative joints and poor cross-references. With both forms, there are many instances in the Iliad where interpretation puts a much greater strain on the conventions of Homeric narrative than does the assumption that the discontinuity results from the poet's deficiency as an editor. A cardinal example of a crack in the narrative occurs in the opening of the second book. Zeus has taken steps to honour his promise to Achilles and has sent a dream promising victory to Agamemnon. One expects that Agamemnon will act on this deceptive dream and will come to grief, but nothing of the kind happens. Instead he calls a council of his elders and informs them that he will first test the army ‘as is proper’ 2.73 and make a speech urging them to go home. The failure of this plan, the mutiny of the army and Odysseus' restoration of order lead to a new beginning of the war in which, as countless critics have pointed out, we move back in time to the first year. The derailment of narrative at the beginning of Book 2 corresponds to some odd features of what may have been the other end of the original joint. Book 11, the third day of battle, marks a very strong beginning. It is, to be sure, appropriate as the overture to this decisive day of fighting, but a number of correspondences between Books 11 and the first two books suggest that the third day of fighting was once the first and that with Book 11 the narrative returns to the position at which it had been interrupted at the beginning of the second book.

Another crack in the narrative is remarkable for showing signs of careful splicing on the surface. Book 13 ends with a description of the noise of battle. In the opening lines of Book 14, Nestor responds to the noise of battle, leaves his tent, and shortly encounters the injured warriors Agamemnon, Diomedes and Odysseus. Their ensuing conversation, however, does not reflect the situation at the end of Book 13, where the Achaeans have the upper hand; rather, it responds to the crisis of Book 12, when Hektor smashes the gates of the wall. Thus Nestor's response to the battle noise covers up a significant discontinuity. If we return now to the end of Book 13, we see that an encounter of Aias and Hektor is elaborately prepared through an exchange of flyting speeches, but is left dangling in the air. On the other hand, towards the end of Book 14, there is an encounter of Aias and Hektor that begins unusually abruptly 14.402.

A third and very glaring example occurs in Book 20. The book opens with an assembly of the gods in which Zeus urges the gods to take sides in the imminent battle. The gods line up on the battlefield, and there is an expectation of cosmic terror as sky and earth are shaken and Hades fears for the safety of his infernal realm. But nothing happens. Instead the narrative turns to the unusually digressive encounter of Achilles and Aeneas, which does contain the structurally important lion simile in which the chain of lion similes culminates but is otherwise only loosely related to the narrative (20.164; above, p. 119). If we look for a continuation of the narrative that pitches the gods against each other, we must go to the second half of Book 21, where we can very clearly see the poet's splicing of different pieces. Hera sends Hephaistos to assist Achilles in his fight with the river 21.330; this confrontation of water and fire in their most elemental form leads to the free-for-all of the gods that was expected (if not quite in this style) after the impressive line-up of Book 20. The transition is smooth enough, but in retrospect the concluding line of the prologue looks strange:

ηομ. ιλ. 20.73 ἄντα δ᾽ ἄρ᾽ Ἡφαίστοιο μέγας ποταμὸς βαθυδίνης,
ηομ. ιλ. 20.74 ὃν Ξάνθον καλέουσι θεοί, ἄνδρες δὲ Σκάμανδρον.
Hom. Il. 20.73and against Hephaistos stood the great deep-eddying river
Hom. Il. 20.74who is called Xanthos by the gods, but by mortals Skamandros.
20.73-74 If they are poised to battle already, why does it take Hera to set Hephaistos against Skamandros? More fundamentally, what business does the lowly local river have among the immortal gods in the first place? The answer is that the lines were added to the prologue in the light of changes that (a) separated it from its original continuation and (b) merged the river fight with the fight of the gods.

Many students of Homer attribute the problem of odd, absent or inaccurate cross-references to the imagination of pedantic readers and appeal to an alleged rule of oral poetics under which the attention span of the poet and his audience is limited to the immediate narrative context. For instance, in Book 8, Zeus predicts that there will be fighting over the body of Patroklos epi prumnêisi. . . steinêi en ainotatêi (8.475 6). Aristarchus deleted the line on the grounds that the prediction does not come true: the fighting over the body of Patroklos does not take place ‘near the bows of the ships’, let alone ‘in a very narrow place’. Schadewaldt (1943, 110 n) excused such inaccuracies by postulating an uncertainty principle that results partly from the poet's desire for progressive revelation and partly from the exigencies of traditional diction, which did not always permit the required precision. But the trouble with the phrase is not that it is insufficiently accurate; rather, it is falsely specific, and there is no evidence that it is formulaic. We may not want to make much of the discrepancy, but it will not do to see in such instances an unproblematic byproduct of oral composition. Precise cross-references are not uncommon in the Iliad. Thus Antenor and Hektor refer to the broken truce (7.69, 7.351), Hektor 8.177 and Achilles 9.349 to the recent building of the wall. Athene refers to Thetis' interview with Zeus 8.370, and Poulydamas remembers yesterday's defeat 13.745. Similarly, Hektor remembers his flouting of Poulydamas' advice on the previous night 22.100. Patroklos describes Sarpedon as the man who first breached the wall 16.558. And, with a precision that borders on pedantry, Glaukos, after the death of Sarpedon, prays to Apollo to heal the hand-wound he had suffered 2,500 lines earlier (16.511, 12.387). Who would have noticed had Homer nodded on this occasion? Given the frequency of quite precise cross-references, false connections and the conspicuous absence of expected connections pose a problem that requires explanation.

A very instructive case of missing or misleading cross-references occurs in the career of Pandaros, which spans Books 4 and 5. Armed with the weapon of Paris, he re-enacts the original offence and wounds Menelaos. His death, on the other hand, resembles that of Hektor: he is the victim of Diomedes' spear guided by Athene. His career has thematic coherence as a foreshortened version of the war which begins with Paris and ends with Hektor. Against the strong presence of this pattern we must set the spectacular absence of any explicit statement to the effect that his death is the punishment for his foolish transgression. When Pandaros reappears in Book 5, introduced by his patronymic, he injures Diomedes, who prays to Athene that she may assist him in his revenge. The goddess grants the prayer, and we see its fulfillment 160 lines later when she guides Diomedes' spear 5.290. This is an intelligible sequence of promise and implicit fulfilment, and it tells a story that concerns only Diomedes and Pandaros. The second encounter of the two has an unusually broad prelude. Aeneas goes in search of Pandaros and scolds him for not using his bow. Pandaros gives a lengthy answer in which he describes this unsuccessful attack on Diomedes, whom he correctly supposes to stand under the protection of some god. Arguing that he foolishly took his bow when he should have followed his father's advice and come to Troy with his horses, he continues:

ηομ. ιλ. 5.206 ἤδη γὰρ δοιοῖσιν ἀριστήεσσιν ἐφῆκα
ηομ. ιλ. 5.207 Τυδεΐδῃ τε καὶ Ἀτρεΐδῃ, ἐκ δ᾽ ἀμφοτέροιϊν
ηομ. ιλ. 5.208 ἀτρεκὲς αἷμ᾽ ἔσσευα βαλών, ἤγειρα δὲ μᾶλλον.
Hom. Il. 5.206For now I have drawn it against two of their best men, Tydeus'
Hom. Il. 5.207son, and the son of Atreus, and both of these I hit
Hom. Il. 5.208and drew visible blood, yet only wakened their anger.
5.206-208 These lines are doubly problematic. First, Pandaros refers to Diomedes as if he had not told the story of their previous encounter twenty lines earlier in the same speech. Second, the reference to Menelaos is assimilated to the Diomedes scene as if it had been an injury in open battle. Thus lines 206 8 are both redundant and inaccurate, but together with the general fuss Pandaros makes about his bow they strongly link the Pandaros of Books 4 and 5. Why did the poet not take the obvious step and have Diomedes exult over the body of Pandaros, saying something like: ‘There you are, you foolish braggart, and may other truce-breakers come to no better end’? The gloating speeches of Book 13 (above, p. 93) show that such words are entirely within the conventions of Homeric narrative. Why are they absent here? The most plausible answer is that the poet developed the role of Pandaros in our Iliad from a simpler version that survives in Book 5 but saw no need to make explicit the changed significance the death of Pandaros acquired in the elaborated version.

Inadequate cross-references also appear between the duel of Paris and Menelaos and the Paris Hektor scene of Book 6, where Hektor accuses his brother of dodging war because of an unspecified cholos, ‘anger’ 6.326. This allusion to an event not reported in the Iliad is especially odd since we might reasonably expect Hektor to say something about the fiasco of the duel in Book 3. Some scholars have heard such a reference in Paris' reply in which he says that he is full of grief rather than anger but will go and fight since ‘victory passes back and forth between men’ 6.339. Such an echo requires very subtle ears, and the fact remains that Hektor's visit to Paris does not in any real sense follow on the events of Book. 3 (Heitsch).

A somewhat similar problem is raised by the duel of Hektor and Aias and its relation to the duel of Paris and Menelaos. Here there are several passages that refer explicitly to the breaking of the truce, but the links are perfunctory, and do not answer the question how anything like the Hektor-Aias duel is at all possible following the disaster of the broken truce (Kirk, 1978).

Yet another problem appears in the assembly of the gods in Book 20, where Zeus tells the gods to join the fighting. This is usually taken as the lifting of his earlier order in which he forbade the gods to join the fighting. But Zeus does not say: ‘I hereby revoke my previous order.’ Indeed, nothing in the scene compels us to assume that the gods previously were absent from the fighting for an extended period—a strange fact in view of the occasional references to the enforced idleness of the gods elsewhere in the poem (11.73, 13.523, 15.113). It is not even possible to say that the scene by its structure implicitly refers to the earlier injunction. Homeric scenes will sometimes recall one another through a set of correspondences and reversals. Thus, the oblivious Andromache of Book 22 reverses the solicitous Andromache of Book 6; the boasts of Hektor in the council of the third night recall those of the second night, etc. But the opening of Book 20 neither refers to nor corresponds to the opening of Book 8, and this absence is all the more conspicuous since in the battle of the gods in Book 21 there is a very explicit cross-reference on a much less important matter: Ares attacks Athene with the express purpose of avenging his defeat at her hands in Book 5 21.394.

From a structural perspective the most striking discontinuity is the absence of any back-references to the Embassy at places where such a reference would seem a good deal more functional than Glaukos' memory of his injured hand. Indeed, in Books 11 and 16 the text is in open conflict with the Embassy. In Book 11, Achilles says to Patroklos:

ηομ. ιλ. 11.608 δῖε Μενοιτιάδη τῷ ἐμῷ κεχαρισμένε θυμῷ
ηομ. ιλ. 11.609 νῦν ὀΐω περὶ γούνατ᾽ ἐμὰ στήσεσθαι Ἀχαιοὺς
ηομ. ιλ. 11.610 λισσομένους: χρειὼ γὰρ ἱκάνεται οὐκέτ᾽ ἀνεκτός.
Hom. Il. 11.608'Son of Menoitios, you who delight my heart, o great one,
Hom. Il. 11.609now I think the Achaians will come to my knees and stay there
Hom. Il. 11.610in supplication, for a need past endurance has come to them.
11.608-610 In Book 16 he says that the Trojans would not be victorious ‘if powerful Agamemnon treated me kindly’ 16.72, but urges Patroklos to
ηομ. ιλ. 16.84 ὡς ἄν μοι τιμὴν μεγάλην καὶ κῦδος ἄρηαι
ηομ. ιλ. 16.85 πρὸς πάντων δαναῶν, ἀτὰρ οἳ περικαλλέα κούρην
ηομ. ιλ. 16.86 ἂψ ἀπονάσσωσιν, ποτὶ δ᾽ ἀγλαὰ δῶρα πόρωσιν.
Hom. Il. 16.84win, for me, great honour and glory
Hom. Il. 16.85in the sight of all the Danaans, so they will bring back to me
Hom. Il. 16.86the lovely girl, and give me shining gifts in addition.
16.84-86 These words are not easily construed as the words of a man who had on the previous night rejected a most elaborate attempt at reconciliation. A whole literature has grown up around attempts to explain Achilles' silence on various grounds, none of them persuasive. Schadewaldt (1943, 81) argued that when Achilles imagines the Achaeans' supplication he implicitly contrasts this with the Embassy in which Aias and Odysseus merely asked him to settle out of court. As for the phrase ‘if powerful Agamemnon treated me kindly’, he sees it as referring to a fundamental attitude unaffected by recent events. And he points out rightly that the same speech that appears to ignore Agamemnon's offer contains another statement that has a passage in Book 9 as its antecedent. When Achilles says
ηομ. ιλ. 16.61 .... ἤτοι ἔφην γε
ηομ. ιλ. 16.62 οὐ πρὶν μηνιθμὸν καταπαυσέμεν, ἀλλ᾽ ὁπότ᾽ ἂν δὴ
ηομ. ιλ. 16.63 νῆας ἐμὰς ἀφίκηται ἀϋτή τε πτόλεμός τε.
Hom. Il. 16.61and yet I have said
Hom. Il. 16.62I would not give over my anger until that time came
Hom. Il. 16.63when the fighting with all its clamour came up to my own ships.
16.61-63 his words refer as plainly as anything in the Iliad to his final words to Aias:
ηομ. ιλ. 9.650 οὐ γὰρ πρὶν πολέμοιο μεδήσομαι αἱματόεντος
ηομ. ιλ. 9.651 πρίν γ᾽ υἱὸν πριάμοιο δαΐφρονος ἕκτορα δῖον
ηομ. ιλ. 9.652 μυρμιδόνων ἐπί τε κλισίας καὶ νῆας ἱκέσθαι
Hom. Il. 9.650that I shall not think again of the bloody fighting
Hom. Il. 9.651until such time as the son of wise Priam, Hektor the brilliant,
Hom. Il. 9.652comes all the way to the ships of the Myrmidons, and their shelters,
9.650-652 But the silence of Achilles about the offer of Agamemnon does not yield to any interpretation that is compatible with the conventions of Iliadic narrative, and we may conclude that his silence is not an intentional and interpretable aspect of the narrative but a by-product of the cumulative process of composition.

Discontinuities in the joints and cross-references of the Iliad are the result of an editing technique that did not keep pace with the poet's architectonic ambition. The creative genius and the wretched editor of the scholar's imagination are one and the same person, composing a major work at an intersection of two modes of textual production. Editorial deficiencies, however, are not structural flaws. This is the point usually overlooked by the analysts in whose scheme of values editorial neatness ranks next to, perhaps even above, godliness. It is a characteristic experience of reading the Iliad that as soon as the reader adjusts to the sloppy tolerances of the text its most glaring cracks disappear and its structural coherence comes into full view. But because Homer was not, by our standards, a very good editor of his text we can sketch the development of the Iliad through several phases. Homer began his career by developing a repertoire of songs, which existed in his memory in a fluid or semi-fluid form. The earliest Iliad is that structure which eventually proved capable of absorbing much of his repertoire into the fixed text of a grand epic. This Ur-Iliad was very much like the hypothetical mênis-poem of the nineteenth-century analysts. It moved from the quarrel of Agamemnon and Achilles and Achilles' prayer to Zeus to the defeat of the Achaean heroes in what is now the battle of the third day. It continued with Hektor's attack on the ships, the Patrokleia, and the return of Achilles to battle. This earliest version did not have the elaborate and expansive finale of our Iliad. Perhaps in it Hermes did steal the corpse of Hektor - a possibility that is raised by the gods in our text, only to be discarded as inappropriate (24.24, 71).

The poet expanded this Ur-Iliad by adding two large blocks of narrative, roughly Books 2-7 and 12-15 in our Iliad. Each block shows traces of internal stratification. The derailment of narrative at the opening of the second book is the chief evidence for the addition of the first block, but the relation of Paris-Menelaos to Aias-Hektor, Paris and Hektor in Books 3 and 6, the oddities in the career of Pandaros, and numerous other features make it certain that the first day of fighting reached its final shape in several stages. The fact that key scenes in this block are dated early in the war is the strongest argument for the hypothesis that these books existed in some form prior to their inclusion in the Iliad. It is very easy to see how a poet would shrug off as insignificant the anachronism that results from putting scenes from the beginning of the war at the beginning of his poem, even though it begins late. It is much harder to see why a poet would invent such anachronisms. (It is tempting to see in the flashback narrative of the Odyssey a solution to technical problems of narrative raised by the construction of the Iliad.) The books existed in the repertoire of the poet. When he recast them for inclusion in his epic he probably gave them for the first time a genuine fixed shape, but his version of, say, Helen on the walls had sufficiently solidified over the years to resist a seamless incorporation into the fixed text.

The second block of narrative is identified as a later insertion by the opening of the Patrokleia. When Patroklos returns to Achilles he reports to him the situation of the war as it existed at the end of Book 11. It could be argued that since Patroklos had spent his time tending to the wounded Eurypylos his report accurately reflects the limited knowledge he has of the war. But it is not like Homer to distinguish between his own and his characters' knowledge of plain facts, and what, in any event, would be the point of such perspectivism? Patroklos' report at the opening of Book 16 is out of date because it was composed at a time when Books 12-15 did not yet exist. The curious splicing of Books 13 and 14 is only one of many features pointing to the internal layering of this block.

Book 9 represents a late stage in the growth of the Iliad, as is shown by the lack of back-references in Books 11 and 16. The impulse for the composition of this book may have come from the poet's recognition that in his transformation of a mênis poem into an Iliad he was beginning to lose sight of Achilles, whose absence from battle threatened to turn into disappearance from the poem. Hence the decision to foreground the absence of the protagonist by his renewed refusal to fight (above, p. 44). It is quite in keeping with the assignment of Book 9 to a late stage that Adam Parry (1956) has found in the speech of Achilles evidence of a personal Achillean and Homeric style.

The addition of the Embassy required the creation of a situation that would plausibly motivate it. It is an old insight of analytical criticism that Book 8 serves this function. The second day of fighting in the Iliad is remarkable for the brevity of description and for the extremely high percentage of repeated lines. Given the previous elaborations of the first and third days of fighting the poem did not stand in need of further fighting scenes. Thus the poet composed a minimal version of a major Achaean defeat. Only the end of the book rises above a utilitarian minimum: the famous tableau showing the Trojans outside the wall is carefully composed with a view to the contrasting despair of the Achaeans, and the transition from Book 8 to Book 9 is the most artful and deliberate in the whole poem.

There is powerful evidence that our version of Book 9 is itself an elaboration of an earlier form in which Aias and Odysseus were the only delegates. When the delegates walk to the tent of Achilles, the poet repeatedly uses dual forms of pronouns and verbs. There are some dozen instances in the Iliad in which dual verb forms are used with plural nouns, sometimes because there is a residual notion of pairs and sometimes for no apparent reason at ail. But in Book 9 dual pronouns and verbs reinforce each other and continue for some twenty lines so as to convey a strong vision of two delegates. There are, however, three delegates: Aias, Odysseus and Phoinix, the old tutor of Achilles. Further oddities in the narrative relate to the manner in which Phoinix is introduced. There are many attempts to interpret the dual forms, but they are all counsels of despair and do not explain the flagrant violation of ordinary narrative norms, for which much the simplest explanation is that the poet wanted to add the more intimate appeal of the old tutor and did not bother to revise his previous narrative in the light of his addition.

Book 9 is evidence of Homer's growing interest in the consciousness of his protagonist. The same interest transformed the death of Patroklos without completely obliterating traces of an earlier version. The narrative oddities of Patroklos' death concern the fate of his armour and the relationship of Kebriones and Sarpedon as his major victims. Like the death of Hektor, the Patrokleia enacts the rule that the killer is killed: Patroklos kills Kebriones, Hektor's charioteer, and after a bitter fight over the body Hektor kills Patroklos. The Patroklos - Kebriones - Hektor triangle is supported by a sequence of two similes. Hektor and Patroklos fight over the body of Kebriones like two lions over a dead deer 16.756. When Hektor kills Patroklos he is like a lion who has defeated a boar after a long fight over a spring 16.823. In this triangle Patroklos is a strong warrior who loses in the end to an even stronger one. But this version of Patroklos' death is overlaid by another version in which Patroklos is the helpless victim struck by a god. In this version his major antagonist is Sarpedon, who supersedes Kebriones as a warrior of greater stature and whose death brings greater glory to Patroklos. That Sarpedon is an addition to Book 16 is apparent among other things from the overly explicit cross-references that refer to his presence in Book 12, where he is also a latecomer. There are also inconsistencies about the armour of Patroklos, sometimes explained by attributing the entire motif of the exchange of armour to another poet who wanted to insert his poem about the shield of Achilles. G. S. Kirk (1962, 220) attributed the loose ends to the oral composer's weaving of narrative fabric with the threads of different traditions. More probably, the loose ends result from a reinterpretation of the Patrokleia in the light of the deeper purpose visible in the Embassy: the poet developed the theme of Patroklos as the alter ego of Achilles (the exchange of armour) and chose for Patroklos a mode of death that would be reflected with greater pain in the consciousness of Achilles.

The expansion of the Iliad also reflected the aristeia of Achilles. It is a peculiar feature of the Iliad that some of its most awkward narrative joints occur in this section. Nowhere in the Iliad is narrative continuity as poor as in Books 20 and 21, the sequence of events that moves from the assembly of the gods via the ‘Aeneid’ to the scenes of mass slaughter, from there to the curious doublets of Lykaon and Asteropaios (the former one of the most magnificent scenes in the Iliad), and then to the river fight and theomachy. Why this should be so is hard to tell, but the fact is worth noting.

The first and last books of the Iliad were probably among the last sections of the work to receive their present shape. As time went on the poet faced the problem that Henry James lamented when he termed the novel a ‘baggy monster’. To counteract the sprawling tendencies of his epic, the poet designed a narrative frame, an elaborate set of correspondences that relate beginning to end. The design is deeply rooted in the poet's vision of the world as a cosmos of polar opposites, which led Whitman to analyse the entire Iliad as a geometric structure in which every part is balanced by a counterpart in a pattern of elaborate and total symmetry. But the geometric structure is much more apparent in the outer than in the middle sections of the work, and it reaches a peak of formalism in the virtual mirror images of beginning and end (above, p. 68).

That the Iliad developed roughly in the manner sketched above is plausible, but the nature of the evidence does not permit firmer conclusions. It is also plain that the unity of design, which is visible through the stages of evolution, is not a compelling argument for single authorship. Analysts are fond of pointing to Gothic cathedrals, in which the work of many hands over centuries has elaborated a single design. The honest unitarian should admit that there is an element of faith in his position. Additional support for the hypothesis of a single poet, however, comes from the virtual scholarly consensus that Book 10 of the Iliad, the so-called Doloneia, is a later addition. The 579 lines of his book relate how after the failure of the Embassy both the Achaeans and Trojans engage in spying missions, how Diomedes and Odysseus trap the Trojan spy Dolon, extract a confession from him, kill him despite their promises to the contrary, and conclude their mission with a massacre of the sleeping Thracians. The book is full of odd words and things, such as the boar helmet or the animal skins worn by the Achaeans, and it delights in a complex parallelism of scenes that has struck many scholars as mannered. The emphasis on the ruthless cunning of Diomedes and Odysseus differs markedly from the ethos of fighting that prevails elsewhere in the poem. Even in translation the reader senses that the Doloneia takes him into another world and speaks with a different voice. One could of course argue that the Doloneia is in more ways than one a night piece and that it contrasts intentionally with the remainder of the work. On the other hand, there is no reference to the events of the Doloneia anywhere in the Iliad, and the narrative moves without disturbance from the last line of Book 9 to the first line of Book 11. No other passage of comparable length can be cut from the Iliad without any consequences to the structure of the rest.

The Doloneia is the only part of the Iliad whose authenticity has been doubted since antiquity, and modern scholars agree for the most part with the position attributed by Eustathius to ancient critics that it was composed after the completion of the Iliad for inclusion in its present place. This scholarly consensus contrasts sharply with the notorious disagreements analysts have about the rest of the Iliad. In particular, there is no comparable agreement about differences of voice and narrative technique in other parts of the Iliad. The Doloneia thus is a test case: it shows what types of evidence or degree of convergence between such types is required to make a forceful claim that something has been added to a complete text. Since no other part of the Iliad comes close to meeting those requirements, the consensus about the Doloneia is an acknowledgement of sorts that the assumption of a single author working over a lifetime is the most plausible hypothesis to account for the co-existence in the Iliad of narrative discontinuities with an overriding coherence of voice, design and purpose.

Martin Mueller. The Iliad. The Unwin Critical Library. London: Allen & Unwin: 1984.

hide Display Preferences
Greek Display:
View by Default:
Navigational bar: