Two major editions of the Iliad appeared at the end of the twentieth century: Helmut van Thiel's for Olms (1996), and Martin West's for Teubner (1998-2000). They are radically different in their methodological assumptions, and hence in the texts they offer. Helmut van Thiel trusts the direct transmission, i.e. the best medieval manuscripts.Footnote 1 He takes the position that ancient variants reported in the Homeric scholia are usually ‘suggestions’ of ancient scholars (for example Zenodotus) ‘towards the improvement of the text, or…deliberations about it’,Footnote 2 and that they are therefore of little significance when constituting the text. He also insists that modern editors not indulge in conjectures of their own. What they should do, rather, is represent the medieval transmission as faithfully as possible. He concedes that this is a modest aim, but one which he considers appropriate, given what can and cannot be known about the Homeric text. According to him, ‘laurels in textual criticism are not to be won from the text of Homer’.Footnote 3 Martin West would surely disagree: his edition offers a dazzling display of editorial ambition. He does not trust the medieval manuscripts, and sees his task as that of exposing and mending their shortcomings. In order to restore what he thinks was the original wording of the Homeric text, West makes use of weakly attested ancient variants; and, above all, employs his own critical acumen to weed out corruption and modernisation.Footnote 4
Unsurprisingly, these two editions have sparked a lively debate.Footnote 5 One important contributor is Gregory Nagy, who argues, in several successive publications, that the Homeric text evolved over a long period of time, from a stage of relative fluidity in the Dark Age to one of relative stability in the late Hellenistic period.Footnote 6 He consequently advocates an inclusive approach to variant readings, which he broadly regards as equally valid realisations of a developing multitext.Footnote 7 Nagy's approach has found many adherents, but also staunch critics. Margalit Finkelberg, among others, has pointed out that the degree of textual variation found in the Homeric poems is quite modest compared to the fluidity and multiformity that prevails in other oral traditions, including some ancient Greek traditions.Footnote 8 Quotations of Homer in classical authors display some divergences that go beyond single words—but, it must be said, not much beyond single words.Footnote 9 Early Ptolemaic papyri also differ from the medieval manuscripts in some of their readings (‘horizontal variation’) and in the number of lines (‘vertical variation’); though again even these so-called ‘wild papyri’ are not as wild as all that.Footnote 10 What we have, at least from the classical period onward, is an essentially stable Homeric text. More importantly, where variant readings do exist, it is often possible to discriminate between them—or at least that was our experience when commenting on Iliad 6.Footnote 11 Editors see their task precisely as making distinctions; Nagy's multitext approach shifts the onus of critical assessment to the reader, or rather the academic user of his online collection of ancient and medieval sources.
Both van Thiel and West seek to separate the Homeric text from ancient attempts to ‘explain’, ‘improve’, and ‘modernise’ it (to use the terms they employ in setting out their task), even if they do so in different ways. One way of conceptualising their work is to say that they attempt to draw a line between the composition of the text and its later receptions. In this article, we wish to explore that notional line, and hope to show that such an exploration improves our understanding of the Homeric text.Footnote 12
At a general and rather theoretical level, it is possible to argue that any attempt to separate composition from reception is doomed to failure—not least because a modern edition is itself an act of reception. As Vallance argues, ‘textual criticism is a particularly artful and radical form of commentary’, in which ‘the critic comments on texts by rewriting them’.Footnote 13 This point, however, is unlikely to carry much weight with textual critics. They may reasonably object that, although what they do is indeed a form of reception, they must hold on to the aim of recovering an original text, in order to be able to work at all. The form of reception (or composition) in which they engage depends, they would argue, on a separation between the ancient text and later corruptions, as well as later attempts to explain, improve, and modernise it (even if their own work constitutes precisely one such attempt). There is, in this approach, the tendency to privilege the relationship between the original text and its latest edition, which seeks to establish itself as the most faithful rendition of that text. What happened in between the original and its latest reconstruction is often acknowledged, but is typically subordinated to that primary relationship.
It is not difficult to see how badly this model fits Homer, where debates about what might constitute ‘the original text’ are rife. Those who hold that Homeric epic developed over a long period of time, through a fluid process of recomposition in performance, may find Nagy's multitext approach instinctively attractive. Still, the task of (re)constructing a text out of ancient and medieval testimonies remains, at the very least so that modern readers are provided with something to read. Moreover, the differences between variants (many of which, as van Thiel points out, seem indeed to be Hellenistic attempts to improve a transmitted text, rather than equally valid realisations of a fluid tradition) risk being obscured by a multitext approach to editing. This article aims to explore the Homeric text by setting up a dialogue between fields of scholarship that seldom talk to each other: textual criticism and reception studies.Footnote 14
Composition as Commentary
It may be helpful to start by asking where composition of the Homeric text ends and explanation of its meaning begins. The question quickly reveals that even the earliest text of the Iliad we can possibly reconstruct is shaped by attempts to explain and comment on an earlier poetic tradition. Some epic expressions, for example, must have sounded obscure even to the earliest audiences of Homeric poetry, because internal glosses attempt to explain their meaning. The epithet δαΐφρων may serve as an example, since the Iliad betrays some uncertainty about what it might mean. Two popular etymologies are suggested in the text: one points to the meaning ‘warlike’ (cf. ἐν δαΐ = ‘in battle’); the other to ‘wise’ (cf. δαήμων = ‘knowledgeable, understanding’). At Il. 5.277, for example, we find this pair of epithets used as near synonyms: ϰαρτερόθυμε δαΐφρον; a little later in the text, at 6.162, the epithet is explained differently, in the sequence ἀγαθὰ φρονέοντα δαΐφρονα; and at 11.482 δαΐφρονα ποιϰιλομήτην again work as near synonyms, but point to ‘resourceful’ rather than ‘wise’ or ‘warlike’. Later poets exploited the ambiguities inherent in Homeric language. Cairns has shown how Bacchylides in Ode 5 built an entire cluster of themes around Homeric etymologies of δαΐφρων.Footnote 15 The Homeric scholia, too, pick up the different explanations of the word δαΐφρων offered in the Homeric text;Footnote 16 and LSJ follow suit, suggesting the following translations: ‘warlike’, ‘fiery’, ‘wise’, ‘prudent’. In the case of δαΐφρων, then, reception is clearly linked to composition already in the Iliad: the poem includes explanations of the epithet for the benefit of audiences during live performance. Internal glosses were seized on by ancient poets and scholars, and they eventually made their way into modern dictionaries.Footnote 17
In other cases, it is harder to establish the exact relationship between the way a word is used and framed in the Iliad and its subsequent reception. The term φύλοπις sounds grand, epic and essentially obsolete to later Greek authors: Theocritus, for example, uses it to characterise the subject matter glorified by the ancient bards, ἀοιδοί.Footnote 18 In Hesiod's Works and Days, it pertains to the age of the heroes;Footnote 19 and it seems that already in the Iliad the word is beginning to sound grand but obscure, and hence to attract attention. It regularly features in the standard formula ἔγειρε δὲ φύλοπιν αἰνήν; but otherwise is often accompanied by internal glosses such as μάχη or πόλεμος, suggesting that it can be conceived as battle, or an aspect of war. Ancient commentators felt reminded of the shouting that is characteristic of Homeric battle (φύλοπις > φῦλα, ‘tribes’ + ὄψ, ‘voice’; cf. common βοὴν ἀγαθός etc.);Footnote 20 but at Il. 16.255f. Achilles wants to watch (rather than hear) the φύλοπις of Achaeans and Trojans. More generally there seems to be a degree of fuzziness about the exact meaning of this word in the Iliad, and some experimentation as to how it might be used. Here we may be witnessing the beginnings of speculation about a specific word in the Homeric text.
Sometimes, the Homeric habit of explaining words, and offering etymologies for them, has direct implications for editors of the Iliad. West, as we have already suggested, is rather quick to emend the transmitted text on the assumption that Homer used correct Greek—by which he means ‘correct’ by the standards of modern linguistic scholarship. So, for example, he rejects the transmitted form πνεύμων (‘lung’) at Iliad 4.528 on the ground that it arose by popular etymology from older πλεύμων, which is morphologically correct and therefore what he prints.Footnote 21 He has slim textual support for his choice, however.Footnote 22 So, one straightforward question is whether West is right to print πλεύμων, or whether editors should follow the manuscript consensus and print πνεύμων, as van Thiel for one does. Grammatical correctness is an obvious criterion for judging the Homeric text, but the crucial issue here (and an issue which West does not address) concerns the text's own criteria of correctness; or, to put it differently, what early audiences may have considered acceptable in terms of morphological formation.Footnote 23 Were their criteria for what is grammatical the same as ours? That would be surprising. Assuming that πνεύμων is indeed a corruption, or rather an etymologising version of πλεύμων, it is still necessary to decide whether it could feature in the Iliad—or, to put it differently, whether the earliest audiences of the Iliad could have made sense of it, and perhaps even appreciated it as an etymologising version of πλεύμων. West rejects this possibility: popular etymology, he implies, is below the Iliad. But, as a matter of fact, there is plenty of it in the poem. Rank's extensive study of etymologising figures in Homer (1952) only discusses part of the evidence; there would in fact be room for an even more extensive study.Footnote 24
Leumann's brilliant study Homerische Wörter (1950) is relevant here, in that it examines in detail what we might call the creative aspects of Homeric word formation.Footnote 25 One of Leumann's prime exhibits is the adjective ὀϰρυόεις.Footnote 26 He builds on Payne Knight's observation that ὀϰρυόεις must have arisen from wrong word division in formulaic phrases such as (πολέμου) ἐπιδημίοο ϰρυόεντος, whence derived first ἐπιδημίο’ ὀϰρυόεντος and then ἐπιδημίου ὀϰρυόεντος (i.e. reinterpreting ἐπιδημίο’ as ἐπιδημίō). As Leumann observes, similar cases of wrong word division are common in Greek epic, and can also be observed in other traditions.Footnote 27 At one level, these are what we might call ‘mistakes’, which crept in as the spoken language moved away from the inherited idiom of epic; but at another level they can be seen as creative acts of reception, often motivated by a transparent rationale. What occasioned the invention of ὀϰρυόεις was, in the first place, the fact that uncontracted genitives in -οο became obsolete at some point in the development of the Greek language. As Chantraine demonstrates, this must have happened before the Iliad and Odyssey were written down for the first time: in those poems, there are only two forms that legitimately represent the genitive singular of the second declension, one in -ō/ου and one in -οιο.Footnote 28 In view of this, it seems safe to assume that the adjective ὀϰρυόεις was already a separate word when the Iliad was composed. Indeed, Leumann shows that it was beginning to develop a semantic range of its own: Homeric ϰρυόεις is used in association with the battlefield, where it means something like ‘dreadful’ or ‘cruel’ (it is used of ἰωϰή, ‘rout’, at Il. 5.740 and φόβος, ‘flight’, at Il. 9.2; cf. its application to πόλεμος in Hes. Th. 936). ὀϰρυόεις at Il. 9.64 conforms to this pattern, but at Il. 6.344 Helen uses ὀϰρυόεις to describe herself, in conjunction with the epithet ϰαϰομήχανος, ‘devising evil’, which functions as an internal gloss. What we see here is a specific act of reception within a larger pattern of linguistic evolution.Footnote 29 Leumann concludes: ‘it is historically wrong to read ϰαϰομηχάνοο ϰρυόεσσης at Il. 6.344’. West disregards the warning, and prints precisely ϰαϰομηχάνοο ϰρυόεσσης.
One feature of ὀϰρυόεις, which Leumann does not discuss, is its relationship to the similar adjective ὀϰριόεις, ‘rough’, or ‘jagged’, always of stones used as weapons.Footnote 30 This word also occurs at the end of the hexameter line, and must have encouraged the formation of ὀϰρυόεις. A general moral can be drawn from this: in order to understand the oddities of the Homeric text, it is crucial to pay attention to sound and metre, as well as ancient attempts to explain obsolete vocabulary on the go, in the course of live performance.Footnote 31
Spoken Word and Epic Diction
Milman Parry argued that, while adhering to time-honoured forms of expression, epic bards strove to bring their art in line with a changing linguistic environment.Footnote 32 A good example of the uneasy compromises they struck is diectasis, which accounts for artificially ‘extended’ forms such as ὁρόω, ὁράαις, etc.Footnote 33 At an early stage of development, Greek epic must have featured the uncontracted forms ὁράω, ὁράεις, etc. in line with the spoken language that surrounded it. But, at some point, spoken language switched to contracted forms (ὁρῶ, ὁρᾶις etc.), and this development then influenced the pronunciation of epic. Performers also started to use contracted forms, but since these did not fit into the epic hexameter, they artificially re-extended them, using the vowel colour that resulted from contraction but retaining the metrical shape of older uncontracted forms.
Partly as a result of Parry's work, diectasis is now so well understood that editors no longer feel tempted to eradicate examples of it, or to impose consistency. The Iliad occasionally testifies to ‘mixed’ forms such as the participle ναιετάωσα, which shows some of the features of diectasis (contraction of older ναιετάουσα and compensatory lengthening) while retaining the original stem vowel alpha.Footnote 34 Clearly, changes in the spoken language were implemented piecemeal, and according to what sounded possible to epic performers and their earliest audiences. The results can seem capricious, to modern eyes.
The famously inconsistent treatment of digamma is another case in point. Words such as ἄναξ, Ἴλιος or ἄστυ are often used metrically as though they were still pronounced wanax, Wilios, wastu. But, on occasion, they are used as if they started with a vowel. We can imagine that bards at some point ceased to pronounce digamma, but still knew that these words somehow behaved as if they started with a consonant. On occasion, for example when it caused metrical difficulties, they chose to ignore that inherited piece of knowledge.Footnote 35
The case of μοι and μευ after ϰλῦθι/ϰέϰλυτε (‘listen’) should, in our view, be treated in the same way as digamma, or instances of diectasis. The form μευ (> μεο) with Ionic contraction looks relatively recent,Footnote 36 and syntactically too the genitive appears to be late. West argues for restoring μοι everywhere, on the ground that it was replaced by μευ under the pressure of spoken language.Footnote 37 His analysis may well be correct in principle, but the issue is when that pressure started to be felt. As Janko points out in his review of West's edition, ‘one can disagree, not about the sequence of phonetic changes, but where Homer falls in relation to them’.Footnote 38
Equally important as the question of how we date particular phonetic and grammatical changes is the issue of whether the Homeric text was ever consistent in matters such as the use of digamma, or the dative after ϰλῦθι.Footnote 39 Medieval manuscripts report a mixture of forms, and there is no reason to suppose that such a mixture would have sounded unacceptable to early audiences of the Iliad. Babylonian readers of the first millennium were sanguine about inconsistencies of spelling and form in the classic texts of their tradition.Footnote 40 Similarly, in current English, some expressions admit the use of different prepositions (e.g. ‘on/at this level’). In Homer, metre played an important regulatory role—but it has already become clear that metre does not explain all unusual features of the text, and there are some cases where even the most basic rules of metre were jettisoned under the pressure of the spoken language. For example, at some point in the epic tradition, the formulaic line ἕως ὃ ταῦθ’ ὥρμαινε ϰατὰ φρένα ϰαὶ ϰατὰ θυμόν must have been ἧος ὃ ταῦθ’ ὥρμαινε… Then Ionic metathesis set in and changed the singers’ pronunciation to ἕως—and this created a serious breach of metrical regularity, because the resulting verse now starts with a short syllable.Footnote 41 Accordingly, some editors emend the transmitted ἕως,Footnote 42 but here West rightly argues that the issue is not one of faulty transmission, but rather of mistaken assumptions, on our part, about what is possible in Homeric metre: as he points out, the bards did not wonder whether their verses were ‘metrical’ in the sense of fitting into some abstract scheme.Footnote 43 They rather struck a balance between the tradition they inherited and the knowledge and expectations of their audiences. Modern expectations of consistency should not obscure the question of what sounded possible.
Pronunciation as Commentary
Since early bards and audiences seem to have converged on an unstable compromise between the epic tradition they inherited and the language they spoke, modern attempts to arrive at the purest, most consistent, and earliest sounding text seem misguided in principle. Precisely because general patterns cannot be assumed, each instance of what looks anomalous to us needs to be considered carefully before it is emended. There are, for example, two expressions in Homer which look similar but, according to the medieval manuscripts, were written and (more to the point) pronounced differently: ἀρηΐφιλος, and Διὶ φίλος. West objects to the apparent inconsistency and writes both ἀρηΐφιλος and διίφιλος as single words.Footnote 44 This, however, seems to us problematic not only because of the evidence in the manuscript tradition, but also because the expressions seem to mean different things in the Iliad. The epithet ἀρηΐφιλος is used in Homer primarily of Menelaos and the Achaeans. Neither is particularly ‘dear to Ares’, who of course fights on the Trojan side. The expression, then, has little narrative force in the poem, and in fact serves as a metrically useful alternative for the common epithet ἀρήϊος, ‘warlike’ (again most commonly used of Menelaos and the Achaeans).Footnote 45 The manuscript spelling ἀρηΐφιλος, as one word, thus rightly treats φίλος as quasi enclitic, soft-pedalling any suggestion of personal affection on the part of the god Ares. The situation is quite different for Διὶ φίλος. This expression is used of people who are actually dear to Zeus, primarily Achilles and Hector. It hardly ever occurs in the plural, because Zeus's affections tend to focus on individuals. There is only one exception, at Iliad 8.517, and it confirms the rule: heralds as a group are under Zeus's special care. So Διὶ φίλος has clear thematic resonance in the Iliad—it actually means ‘dear to Zeus’—whereas ἀρηΐφιλος does not mean ‘dear to Ares’, but rather ‘warlike’. Of course, there was no word division in the early texts of Homer but, as West recognises, the argument is not about how we divide up words on a page, but rather how they were pronounced in performance: φίλος loses its emphasis in ἀρηΐφιλος but retains it in Διὶ φίλος.Footnote 46
The different spellings of the manuscript tradition preserve knowledge about how these words were uttered in performance.Footnote 47 As every actor knows, pronunciation and interpretation go together; and here the manuscripts preserve an oral interpretation of the text which West sacrifices in the name of morphological consistency.
Sound and Grammar
A focus on sound can help shed light on some of the more localised inconsistencies in the morphology of Homeric words. One example is γόον at Il. 6.500. The context is important:
The form γόον is unique in Greek, and difficult to explain in purely grammatical terms. Formally, it represents a third person plural aorist or imperfect of the root *γο-, but elsewhere Homer uses forms of γοάω, which leads us to expect (ἐ)γόησαν or (ἐ)γόων here.Footnote 49 Neither is metrically possible, and there is no question of changing the transmitted text.Footnote 50 Chantraine notes that γόον looks as though it was derived directly from the noun γόος, and compares the verb ϰτυπέω, whose aorist ἔϰτυπε/ϰτύπε, ‘he thundered’, likewise seems to derive from a noun (ϰτύπος, ‘crashing noise, thunder’).Footnote 51 Leumann suggests it is a linguistic relic.Footnote 52 He may be right, historically, but for Homeric audiences the real point of γόον at 6.500 is surely its sound: it echoes the noun γόον in the line immediately above, and resonates with ζωόν too. To ancient scholars, the phenomenon was known as parechesis, the deliberate ‘echoing’ of one word by another.Footnote 53
Olaf Hackstein has argued, in detail, that such echoing effects were popular, and were sometimes created even at the expense of grammatical regularity.Footnote 54 Here is an example:
The form ἐδήδοται has been subject to emendation ever since antiquity.Footnote 55 The problem is that the reduplicated perfect stem *ἐδήδ- ought not to contain the theme vowel omicron before the person ending (ἐδήδ-ο-ται). From the point of view of modern grammar its intrusion is hard to defend. From the point of view of Homeric grammar, however, it is entirely transparent: in order to reinforce the thematic parallel between eating and drinking at the level of sound, ἐϰπέποται (with a reduplicated stem πεπο-) inspires the ad hoc formation of ἐδήδοται, which echoes it. Hackstein rightly points out that such manipulations are common in spoken language. More importantly, they are a hallmark of Homeric style. Ancient readers considered them so characteristic of Homeric epic that they sometimes detected them even where modern scholars see nothing unusual in the text.
The following example (not discussed by Hackstein) is a case in point:
From a modern perspective γηράς is simply the correct Homeric form of the aorist participle of γηράσϰω. Some Hellenistic scholars shared this view (ΣA ad Il. 17.197a), but not all. According to some, γηράς was a deliberately shortened version of an expected sigmatic aorist γηράσας:
By the standards of modern historical grammar it is impossible to interpret γηράς in this way, but the question that concerns us here is whether ancient audiences may, in fact, have heard or assumed precisely such an interpretation. From Aristotle onwards, scholars certainly knew—or thought they knew—that Homeric epic presented cases of ad hoc shortening of expected forms.Footnote 56 They disagreed about the details, but clearly recognised the general feature. It seems likely that early Homeric audiences, just like their Hellenistic counterparts, accepted that Homer could contract words for ad hoc metrical reasons, or poetic effect.Footnote 57 It is therefore worth asking whether they could have regarded γηράς as one such case.
The form γηράσας, or indeed any other form of the sigmatic aorist of γηράσϰω, never features in Homer. The text does, however, betray signs that the old aorist ἐγήρα was beginning to be reinterpreted as an imperfect.Footnote 58 In Herodotus, this trend leads to a restructuring of the verbal paradigm, whereby ϰατεγήρα (reinterpreted as an imperfect) comes to stand side by side with the sigmatic aorists ϰατεγήρασαν (3rd pers. pl.) and γηράσασαν (part.).Footnote 59 We do not know when these developments took hold in the spoken language, and therefore how exactly Homer's earliest audiences would have analysed the form γηράς. The punning grammar of Il. 17.197 may suggest that the form was remarkable already in the earliest history of the text, but Hesiod still used the root aorist of γηράσϰω competently and freely.Footnote 60 The alleged parallel ὑποφθάς does little to support the scholiast's claim: there is no reason to believe that early Homeric audiences regarded that form as derived from ὑποφθάσας.Footnote 61 The other parallel, however, does lend weight to the scholiast's position, and also helps us make sense of ἐπιπλώς in its own right.
The form ἐπιπλώς occurs at Il. 6.291 and comes with a well-known textual problem attached to it:
Grammatically, the transmitted ἐπιπλώς must be the participle aorist of ἐπιπλέω/-ώω, ‘sail across’. Like γηράς at Il. 17.197 it is best analysed as a root aorist, but the omega is irregular: we would expect ἐπιπλούς (cf. classical γνούς). West restores ἐπιπλούς and argues that an error of transcription meant that original ‘ο’ was misinterpreted as omega instead of intended ō/ου. This suggestion is, however, problematic for two reasons. First, as Alfred Heubeck has convincingly argued, the earliest texts of Homer did in fact distinguish between omicron and omega.Footnote 62 But even if it were true that ἐπιπλώς arose from confusion of ō and ω, perhaps at some later stage and under the influence of Athenian texts, we would still need to explain why the error crept in here and not elsewhere, and why it was allowed to persist.
An answer can be pieced together by considering an ancient explanation:
ἐπιπλώσας: ὁ Ἀπολλώνιος σχηματίζει τὸ ἐπιπλώσας οὕτως· πλέω ϰαὶ ἐπιπλώω, οὗ ἀόριστος ἐπέπλωσα, μετοχὴ ἐπιπλώσας, εἶτα ἀποϰοπὴ ἐπιπλώς. ϰαὶ το ἐπέπλως δὲ ῥῆμα “τοῦνεϰα γὰρ ϰαὶ πόντον ἐπέπλως” (γ 15) ὁμοίως ἐϰ τοῦ †ἐπιπλώσας [ἐπέπλωσας? Erbse] ἀπεϰόπη· ὅτι γὰρ τὸ ἐπέπλως οὐϰ ἔστιν ἀπὸ θέματος τοῦ πλῶμι δῆλον ἐϰ τοῦ μηδὲ<ν> εἶναι ὡς ἀπὸ τῶν εἰς μι ϰίνημα, οὐϰ ἀπαρέμφατον ἐπιπλῶναι, οὐϰ εὐϰτιϰὸν ἐπιπλοίην, οὐ προσταϰτιϰὸν ἐπίπλοθι, οὐϰ ἄλλο οὐδέν. ἔδει δὲ ϰαὶ τὴν μετοχὴν εἶναι ἐπιπλούς ὡς βιούς ϰαὶ γνούς· ὅτι γὰρ οὐ Δωριϰῶς ἐτράπη εἰς τὸ ω ὡς βοῦς βῶς, δῆλον ἐϰ τοῦ μὴ προϰεῖσθαι αὐτοῦ ἐν χρήσει τὸ ϰοινόν.
(ΣA ad Il. 3.47a)Apollonius [Dyscolus] analyses ἐπιπλώσας as follows: πλέω and ἐπιπλώω, of which the aorist is ἐπέπλωσα, the participle ἐπιπλώσας, whence the shortened form ἐπιπλώς. The form ἐπέπλως in the phrase τοῦνεϰα γὰρ ϰαὶ πόντον ἐπέπλως (Od. 3.15) is likewise a shortened form of ἐπιπλώσας [or rather ἐπέπλωσας? Erbse]. That ἐπέπλως does not derive from πλῶμι is evident from the fact that no forms of this verb exist which follow the conjugation of verbs in μι: neither the infinitive ἐπιπλῶναι, nor the optative ἐπιπλοίην, nor the imperative ἐπίπλοθι are attested, nor any other relevant form. Moreover, the participle should have been ἐπιπλούς as in βιούς and γνούς. That ἐπιπλώς is not a Doric form of ἐπιπλούς, cf. βοῦς > βῶς, can be seen from the fact that the normal form [i.e. ἐπιπλούς] itself is not attested.Footnote 63
Ancient grammarians were clearly interested in the problems posed by ἐπιπλέω/ἐπιπλώω and its various forms. Like West, Apollonius Dyscolus points out that the expected form of the aorist participle is ἐπιπλούς, but unlike West he thinks he can explain the transmitted ἐπιπλώς: he suggests that it is a shortened form of the sigmatic aorist ἐπιπλώσας—just like the scholiast who tried to explain γηράς. To the modern reader, this explanation is far from satisfactory, and even in antiquity not everyone agreed.Footnote 64 But there are some observations that lend strength to Apollonius' argument. First, sigmatic ἐπιπλώσας does occur in Homeric epic: it would have been the more familiar form to ancient audiences, and presumably closer to how they spoke.Footnote 65 Secondly, the Homeric text itself invites comparison between ἐπιπλώς and ἐπιπλώσας. The latter form occurs at Il. 3.47, where we hear for the first time about the voyage of Paris. When the same trip is mentioned again in Iliad 6, Homer uses ἐπιπλώς. Somebody reading, or listening to, the opening books of the Iliad moves from ἐπιπλώσας to ἐπιπλώς, precisely along the lines Apollonius suggests. Neither form occurs elsewhere in the epics (just as Paris' voyage is not mentioned again), so they are thematically linked. Parechesis between the participle ἐπιπλώς and the finite forms ἐπέπλως (Od. 3.15), ἐπέπλων (Hes. Op. 615), ἀπέπλω (Od. 14.339) and παρέπλω (Od. 12.69), all placed after the main caesura, helps justify the unusual form, if further justification were needed.Footnote 66 It is of course possible that, when pressed, a rhapsode might have explained ἐπιπλώς as a rare dialect form, just as later grammarians did.Footnote 67 And it is even possible that someone, at some point in the history of the text, simply made a mistake, which was then interpreted as an acceptable form. But Apollonius' analysis seems in tune with the experiences of audiences in performance: for them, the form ἐπιπλώς would have sounded plausible. That, and a more general sense of the malleability of Homeric language, suggests that the transmitted ἐπιπλώς at Il. 6.291 should not be emended. West refers to ancient spelling conventions in order to explain how ἐπιπλώς came about—but his arguments are historically problematic, and in any case fail to explain why ω obtained in this particular case. We may make better progress by considering the ancient reception of epic: pronunciation in performance, and ancient views about Homeric grammar.
Grammar in Character Speech
The suggestion that what sounds good is allowed to influence Homeric grammar finds confirmation in character speech, where the phenomenon seems even more prominent than in the main narrative. The words of characters are spoken out loud not just in the context of actual performances, of course, but also within the fiction of the narrative, so this may not be too surprising. More generally, Homeric characters speak in a more lively and personal way than the narrator. Ruth Scodel points out that they often treat myth in a tendentious and self-serving fashion: much like the lyric poets, but unlike the poet of the Iliad, they recognisably adapt traditional stories to suit their own ends.Footnote 68 In a landmark article, Jasper Griffin made a more specific point about the language of Achilles: the poet feels himself into this character to the point that he adopts a particular form of language whenever Achilles speaks, complete with its own distinctive vocabulary.Footnote 69 Effects of this kind, and especially the use of focalisation, have been further studied by Irene de Jong and others, and are now well understood.Footnote 70 But the possible implications for Homeric grammar, and the Homeric text, have not so far been much explored.
Do Homeric characters twist grammar more radically than the main narrator? The question can be explored by considering a well-known textual crux at Iliad 1.291. Achilles has just insulted Agamemnon, who now bitterly complains to Nestor:
The verb in line 291 is a problem: context requires a form of τίθημι, or perhaps ἵημι, but προθέουσιν is not easily derived from either. Aristarchus was puzzled (his unconvincing explanation is reported in ΣA ad Il. 1.291b), and modern scholars have not fared much better.Footnote 71 There may, however, be a way of explaining this enigmatic form. As Hackstein points out, προθέουσιν may be understood as an extreme case of parechesis:
Hackstein shows that the point of προθέουσιν is play on the root θε-, which suggests an association with τίθημι rather than ἵημι. Grammatically, that leaves us with two problems: first, προθέουσιν has a thematic ending (τιθέω etc.) when we would have expected an athematic one (τίθημι etc.). Secondly, προθέουσιν looks like a truncated version of προτιθέουσιν. The unusual ending can certainly be explained: the bards heard forms like τιθέω in the language around them, and occasionally used them in their own poetry (Il. 13.732, Od. 1.192). But the loss of an entire syllable from the middle of the verb is startling. We can concede that verbal forms in epic are often shortened or extended, and that parechesis helps that process.Footnote 72 It is also true that the exuberance of Homeric language may be relevant here: the verb τίθημι, in particular, takes so many different forms in Homer—some of them by no means straightforward—that one more may have managed to slip into the general variety. Still, the ad hoc creation of a present προ-θέ-ουσιν does seem like an extreme case of morphological violation.
It is the context of the speech that helps to explain the form. Hackstein describes it well: ‘Notons que l’échange violent des paroles entre Achille et Agamemnon est très rapide. Nous nous trouvons dans la dispute entre ces deux héros et nous y voyons une langue extrêmement émotionelle, pleine d'invectives.’Footnote 73 Agamemnon's language is shaped by his hatred, and his sense of impotence. Achilles has just called him a ‘drunkard with the eyes of a dog and the heart of a deer’ (Il. 1.225). Nestor did his best to soften the blow, but Agamemnon is outraged almost beyond words (or beyond words that make grammatical sense). A fourfold anaphora (πάντων…πάντων…πάντεσσι…πᾶσι) suggests the stuttering anger that has taken hold of him, an anger that boils over towards the end of his speech. The form προθέουσιν in the concluding line not only makes for a vivid contrast with preceding ἔθεσαν but also suggests, at the level of grammar, what Agamemnon feels and fears: Achilles is divinely favoured, to be sure, but the gods could not possibly condone his present insults. In point of fact, they do. That possibility is so abhorrent to Agamemnon that his very language becomes shortened and harsh, insisting on θε- (‘but the gods, the gods…’).
It makes sense for editors to follow the example of the poet, and feel themselves into the state of mind of Homeric characters. Hector's speech to Hecuba in Iliad 6 is another case in point. The situation is, again, fraught: Hector tells his mother that the only thing that would make him happy would be to see Paris dead and buried. Saying this out loud is an admission of defeat—moral defeat in the first instance: Paris caused the war, and Paris was wrong to do so. At the same time, Hector acknowledges his own personal defeat and impotence: although his task is to kill the Achaeans, he actually wishes that his own brother was dead. It is particularly painful that he admits this to Hecuba, who is after all the mother of both Hector and Paris. What is worse, according to one tradition, it was precisely Hecuba who saved Paris from death in infancy, when a prophecy revealed that he would bring about the destruction of Troy.Footnote 74 She should not have to hear Hector say he wishes his own brother was dead; and he should not have to feel what he is saying. As it turns out, his language is also harsh—particularly at the end of his increasingly desperate speech:
The form ἀτέρπου at Il. 6.285 is problematic, because it seems to derive from an adjective ἄτερπος. The expected form of that word, in Homer as throughout Greek literature, is ἀτερπής, ἀτερποῦς. Zenodotus proposed the alternative reading φίλον ἦτορ, probably in order to avoid the strange ἀτέρπου.Footnote 75 West follows Zenodotus, on the ground that φρέν’ ἀτέρπου is problematic by the standards of modern grammar, whereas φίλον ἦτορ is unobjectionable. Most editors, by contrast, retain the transmitted form. Although West's position seems sensible, it does actually raise some difficulties. At a basic level, it ignores the fact that Homeric adjectives often inflect according to more than one pattern.Footnote 76 More importantly, it fails to consider that Hector's speech contains several other oddities too. He has just said that he would like to see Paris go down to the House of Hades—‘the invisible one’ (A-ides) according to punning interpretations found in the Iliad.Footnote 77 Hector's longing to see Paris in ‘the invisible realm’ captures the desperate and impossible nature of his desires.
Earlier on in the speech, Hector claimed that he wanted the earth to swallow up Paris (Il. 6.281f.): there is nothing ungrammatical about that expression, except that Hector uses it in an unprecedented and unidiomatic manner. In Homeric epic, speakers otherwise apply it to themselves, to express extreme shame, and hence the desire to disappear from the face of the earth (cf. Il. 4.182, 8.150, 17.415-17). That Hector should apply those words to Paris rather than himself reveals his predicament: he identifies with his brother, and loathes him. At this point, normal use of language breaks down. ϰέ in line 281 is decidedly unorthodox Greek, even by the flexible standards of Homeric grammar. But that, as Kirk ad loc. points out, is precisely the point: Hector feels that he needs to say things which cannot rightly be said.Footnote 78 In the context of this speech, the strange form ἀτέρπου is not as out of place as it would be in a plain-sailing narrative context.
Paris has the capacity of twisting Helen's words, as well as those of his brother. A little after Hector's outburst, Helen also expresses the thought that Paris deserves a bad end:
Helen's entire speech is tinged with regret, self-pity and, increasingly, contempt for Paris. Two unreal wishes with ὀφέλλω set the tone, each of them drawn out beyond what is grammatically comfortable.Footnote 79 By the time we get to the lines cited above, some modern scholars have had enough, and demand a text that they can actually construe. Following Herwerden, West restores τοῦ for transmitted τῶ, to provide a genitive that goes with ἐπαυρήσεσθαι: ‘for that (τοῦ) he will get his reward’. West again raises the possibility of a transcription error from an older text in which omicron was used to spell both long ō (~ ου in later texts) and omega—but there are difficulties with arguments of this kind, as has already emerged. The choice here is then between a transmitted text that looks syntactically awkward, and one which is grammatically smooth but is unattested. Of course, a difficult but comprehensible text should never be emended in favour of something simpler. But is Helen's τῶ understandable? Much depends on what we think possible in Homeric Greek. Modern grammatical standards do not easily map onto ancient theories, so here it makes sense to try and explain Homer by reference to Homer, as Aristarchus insisted we should.
Helen relentlessly attacks Paris (Il. 6.349-53)—until she concludes with a thought about his comeuppance. She then turns her attention to Hector: she has already suggested that, given she is in Troy, she would much rather have a strong and dependable husband (like Hector) than the one she has—and now she invites Hector to sit next to her for a while, and find some respite in her company. She is not really interested in what Paris does at this point, or in the precise mechanisms of his punishment. What she wants, above all, is to be done with him and turn to Hector. Transmitted τῶ ϰαί has an immediate and powerful effect in this context. Andromache uses the same turn of phrase to sum up her feelings for her dead husband:
Andromache's speech shows that τῶ ϰαί is idiomatic in contexts where one character judges another. The expression at 6.353 is not a mistake, but rather articulates a shift of focus from Paris to Hector.
The text proposed by Herwerden and West does not have the same forward thrust: the referent of τοῦ remains unclear until we realise that it goes with ἐπαυρήσεσθαι and work out that it refers back to Paris' lack of understanding. That moment of uncertainty is brief, but it suffices to create a rhetorical weakness in Helen's speech, which is further underlined by the fact that elsewhere in epic ἐπαυρίσϰω takes concrete and well defined genitives: a thing, a person, not a complex verbal idea, as would be the case here.Footnote 80 This does not mean that τοῦ would be unidiomatic. It is even possible that ancient audiences who heard τῶ would, with hindsight, have taken the word to be τοῦ. Our point is that standards of grammatical correctness are contextual. The transmitted text is not a slip of the pen, or a simple error: it fits the resonant patterns of early Greek epic. Nor does West's emendation straightforwardly restore grammatical correctness. In truth, the real difference between τῶ and τοῦ is not that one is idiomatic and the other is not, but that they are idiomatic in different ways: τῶ sounds right in the heat of the moment, when it is said and heard, whereas τοῦ can be construed with hindsight, and with a level head. In his 1973 introduction to textual criticism, West suggests that the editor of a classical text should start by making a translation.Footnote 81 If he or she cannot construe a sentence, there will be something wrong with it. This advice is useful, of course, but privileges modern rather than ancient contexts of reception. Homeric linguistic usage is sometimes more expressive, and more rhetorically inflected, than scholars facing a translation task might be prepared to accept. Helen's speech needed to sound right in performance, rather than prove acceptable to classicists working at a desk.
Speech-Framing Lines as Commentary
In his unforgettable portrait of the rhapsode Ion, Plato gives a good impression of the degree to which Homeric performers identified with the characters they portrayed. Plato's Ion claims that when he performs a sad passage his eyes ‘fill with tears’, and when the narrative becomes frightening his own hair stands on end, and his heart leaps (Ion 535c). He also insists that he hardly perceives the here-and-now of his own performance, feeling rather that he is himself in ancient Troy—even if he later admits that he checks whether his audience are crying when he is because, if they are, they will pay him better (535b-e).Footnote 82 Later in the dialogue, when Socrates presses the rhapsode to identify, as precisely as possible, his area of expertise, Ion claims that he knows exactly how a man would speak, as well as the kind of thing a woman might say, or a slave, or a freeman, or someone receiving orders, or someone giving them (Ion 540b). Plato's dialogue suggests that rhapsodes paid great attention to Homeric speeches, and this makes sense. Given that speeches represent a large proportion of the Homeric poems, the ability to deliver them must have been crucial to the success of a rhapsode's career.Footnote 83
Performers, as well as scholars, must have devoted a lot of thought to the precise tone and thrust of particular speeches—and it seems that traces of their interpretations have left their mark on the textual transmission of the Iliad. In her 1967 book on the Ptolemaic papyri, Stephanie West observes that lines which introduce and round off speeches display a greater degree of textual variation than average. She does not explore the phenomenon, or attempt to explain it, but surely two factors are relevant.Footnote 84 The first is that speech-introductory and speech-concluding lines are usually composed on the basis of a nuanced system of formulae, and can therefore be easily modified on the hoof.Footnote 85 The other is that, by framing a speech, these lines provide a first form of commentary on it: they tell audiences something about the speech they are about to hear, and afterwards give some indication of the effect it had.Footnote 86 For example, at the beginning of Iliad 6, Adrestos grabs Menelaos' knees and begs him to spare his life. Menelaos, who is depicted as a rather soft man in the Iliad, is ‘persuaded’ or ‘moved’ (according to a different variant) to save him, but Agamemnon intervenes, with an exceptionally brutal speech, and thus either ‘turns Menelaos’ purpose’ or ‘persuades’ him that Adrestos should be killed, like all other Trojans. In order to investigate the interaction between these three characters, it is useful to report the passage in full:
Interpretation is difficult, partly because the authorial comment in line 62 is itself hard to fathom. On one reading, the poet seems to claim that Agamemnon speaks what is right (αἴσιμα παρειπών), but ancient and modern readers alike find his speech exceptionally savage; indeed some scholiasts accused him of ‘beastliness’ (θηριότης).Footnote 87 When Agamemnon insists that even male foetuses still in their mothers' wombs should be killed, it is hard not to think of Astyanax, especially as this passage immediately precedes the scene where Hector and Andromache smile at their baby boy.
Issues of interpretation are, we claim, related to the textual variants. At line 51, one unedited Oxyrhynchus papyrus (1044 West) reads ἔπειθε, as do some of the more important manuscripts. Most manuscripts, however, have ὄρινε. Elsewhere in the Iliad, θυμὸν ὀρίνω is used when a highly emotional act of supplication is successful (cf. 9.595 and 24.465-67), so the uncertainty here concerns the emotional impact of Adrestos' speech on Menelaos. Now, it may be that one reading is preferable to the other—ἔπειθε may be the more compelling option, because Adrestos does not make an especially strong appeal for pity; on the other hand, perhaps the point is precisely that Menelaos is easily moved. In any case, what the variants betray is a long-standing uncertainty about interpretation. Ten lines later, there is again variation in the speech-concluding line: Agamemnon ‘changes’ Menelaos' mind or—according to some manuscripts—‘persuades him’. Here too there is, perhaps, a way of preferring one reading to the other: παρέπεισεν is otherwise used in the Iliad when the speaker has a restraining effect on the addressee (cf. e.g. 7.120, 13.788, 23.606), and this is not the case here. Still, the variants at the end of both speeches suggest that their tone, and Menelaos' precise reaction to them, was debated. It seems that what we have here are early, rhapsodic variants, framing the two speeches in performance. These variants, then, are best taken as a flexible and somewhat fluid interpretative ‘frame’ around the speeches of Adrestos and Agamemnon. Variations in speech-introductory and speech-concluding lines may, more generally, be treated as evidence for the reception of Homeric speeches on the part of rhapsodes and their audiences.
Conclusion
The reception and composition of Homeric epic are intertwined. Explanations of obscure epic words like δαΐφρων are already contained in the Iliad itself: they affect its composition, therefore, and not just the ways in which the poem is explained by later scholars. Similar arguments can be made about other aspects of the transmitted text. The evidence may not support a model of multitextuality on a grand scale, but does suggest live explanation in performance ranging from pronunciation to the framing of controversial speeches. Editors who fail to engage with ancient reception—by asking, for example, what might or might not have sounded grammatical to ancient audiences—miss important evidence for the constitution of the text, and are in danger of excluding forms which in fact have a justification. Conversely, however, students of reception cannot simply take the latest edition of the Iliad as a given (as they often do): crucial evidence emerges from consideration of the apparatus criticus.
It is by aligning textual choices with a detailed understanding of the early reception of epic that research on the text of Homer may most profitably advance. This makes for slow work. One advantage of following general editorial principles and stable grammatical rules is that they allow for swift progress on individual problems. There is, however, nothing swift about the Homeric tradition. The Iliad and Odyssey have been the focus of intense interest for almost three millennia: well attested oddities in medieval manuscripts are not usually simple mistakes, and often reward attention. This is one of the lessons that can be learnt from van Thiel's cautious approach. West, for his part, offers the most complete and reliable account of the ancient testimonies, and therefore crucially enables further explorations of the text, including those presented here. The examples we offer suggest that, when confronted with Homeric epic, it makes little sense to draw a line between composition and reception according to strict principles (such as that of reconstructing the most consistent or earliest possible version of the poem)—or refuse to draw any line at all, again on principle. Rather, it seems to us that progress may be made by considering this line and, as Newton might have said, thinking on it.
