BMCR 2019.01.05

Homerus. Odyssea. Bibliotheca scriptorum Graecorum et Romanorum Teubneriana, 2026

, Homerus. Odyssea. Bibliotheca scriptorum Graecorum et Romanorum Teubneriana, 2026. Berlin; Boston: De Gruyter, 2017. lii, 519. ISBN 9783110425390. €99,95.

Publisher’s Preview

The posthumous publication of Martin West’s Odyssey complements his earlier edition of the Iliad (also in the Bibliotheca Teubneriana, 1998-2000), and crowns a long career of exceptional productivity and achievement.1 As the publishers point out, ‘this is likely to be the standard edition of reference for decades to come’.2 For this reason, it seems important to set out, for the general user as well as the specialist, the guiding principles governing West’s approach to the Homeric text and to illustrate, by brief and very selective example, how those principles affect the handling of specific editorial problems.3

The preface, written in lively and elegant Latin, focuses on specific issues of grammar and orthography: given that the overall approach to editing Homer has changed little since his edition of the Iliad, West frequently refers to the earlier work, ne molesta fiat materia repetita (p. VII). The overall aim is straightforward: to reconstruct the original text of the Odyssey, which West regards as the work of a single poet, who devoted himself to its creation over a long period of composition and revision, in the last three decades of the 7 th century BCE.4 The original text had no book divisions (hence no typographically marked divisions in this edition, as in West’s Iliad); it was written in an alphabet that did not distinguish between ε/η or ο/ω (hence the occasional emendation based on ancient theories of transcription from one alphabet into another), and was probably composed in Attica or Euboea. This last claim sets the Odyssey apart from the Iliad. For this reason, the Preface discusses Atticisms and warns readers that these will sometimes be retained (p. VII). In practice, however, several are not: for example, West prints ἄσσον, θάσσον, μάλλον, κρέσσων, μέζων rather than transmitted (and Attic) ἆσσον, θᾶσσον, μᾶλλον, κρείσσων, μείζων. The rationale is not made explicit, though presumably he considers these forms later than ‘the poet’.

In reconstructing his 7 th -century text of the Odyssey, West selects what he considers the best medieval manuscripts and supplements them with readings attested in the ancient papyri. The manuscripts have already been well reported by Ludwich and van Thiel, but West more than doubles the number of Odyssey papyri, mostly by harvesting the riches of the Oxford Oxyrhynchus collection.5 This is a major achievement in ‘both Homeric studies and papyrology’,6 though it does not result in substantial changes to the text: the new papyri broadly confirm what we already know about the Homeric paradosis – a point to which we return below.

West also includes in his edition an enormous apparatus of testimonia, which contribute little to the constitution of the text, as he himself points out: p. IX. The stated reason for including them is to give a clearer view of the tradition, but this seems problematic, for two reasons. If by ‘tradition’ we mean the diffusion of the Odyssey in antiquity, then the modern reader would be better served by a discussion that did not draw an artificial line between testimonia and other forms of paraphrase and allusion. If, on the other hand, the aim is to assess the relative weight of individual readings, it would be important to be selective: most of the testimonia originate in ancient scholarship and, as van Thiel points out, ‘ancient scholars display a strong tendency to depend upon one another’.7 West claims that his testimonia, as well as giving a general impression of the tradition, are useful for confirming medieval variants as old, and for illustrating levels of interpolation in antiquity. These aims, however, do not seem to us to require an apparatus of such length (and cost). More worryingly, the reader can easily be impressed by a long string of mutually dependent testimonia and grant them greater authority than they possess. West himself encourages this by using testimonia to boost minority readings. He also introduces otherwise unattested variants into his apparatus criticus on the basis of his collection of testimonia. His use of the Homeric Centos seems particularly problematic in this regard: there is no reason to assume that a line in Eudocia has to match the text of Homer in every last detail.8

Because West’s overall aim is to reconstruct the original, seventh-century text of the Odyssey, it becomes important for him to pinpoint ‘the poet’ in relation to a gradually evolving poetic language. He admits that forms which look late to us may pre-date composition: for example, the poet of the Odyssey may have used the artificial-sounding βῆ δ’ ἴμεναι, as well as the earlier βῆ δ’ ἰέναι (see Odyssea, p. XVIII). Still, the general tendency in West’s edition is to restore the oldest possible text, ensure its grammatical correctness (by modern standards), and minimize inconsistency. To return to the above example, West always prints the older and ‘more correct’ ἰέναι, despite his own misgivings – not least in order to ensure consistency with his Iliad.9 He likewise restores old genitives in -οο, although they almost certainly did not belong to the epic Kunstsprache when the Iliad and the Odyssey were composed.10 Similarly, he prints -εο- for transmitted -ευ-, despite Wachter’s recent (and to our mind convincing) warning not to do so.11 Taken together, West’s choices are reminiscent of restorative nineteenth-century scholarship, as exemplified by August Fick, Jan van Leeuwen, and Richard Payne Knight among others.

Payne Knight, in particular, is an important figure for West – a fact betrayed by surreptitious but unmistakable eulogy in the apparatus: iam Knight and corr. Knight, where a simple surname suffices for other scholars, and even fere Knight where the older scholar ‘almost’ arrived at the conclusion championed by West. Now, Payne Knight (1751-1824) presented himself as the British answer to the (more fully professionalised) German scholars of his day. His aim was to ‘restore’ Homeric epic to its ‘pristine clarity’ and ‘integrity’,12 thereby refuting the theories of Wolf and Heyne. For all their learning, these scholars lacked, in his view, the quintessential British quality of ‘common sense’ (Payne Knight, Prolegomena, Leipzig, 1816, p. 15: ipse hominum sensus communis, et experientia quotidiana reclamant). German scholars, for their part, accused Payne Knight of wilful eccentricity: they even coined the term Knightianismus to deride his idiosyncratic restorations.13 West is influenced by this particular chapter in the history of Homeric scholarship. Like Payne Knight, he seeks to restore the work of one great, original poet. Like him, he has a low opinion of the transmitted text and wants to revert to a ‘purer’ form of Homeric Greek. And again, like Payne Knight, he puts great faith in inscriptions and other external witnesses in recovering that form, as we illustrate below.14

The problem with privileging a nineteenth-century framework defined mainly by British and German traditions of scholarship, is that the arrival of the Americans can go almost unnoticed.15 To be blunt, the work of Milman Parry makes too little impact on West’s edition of the Odyssey.16 This is a serious shortcoming, for two reasons. The first is that formulaic analysis and oral poetics help to explain the Homeric paradosis and therefore offer important insights for establishing the text. The second is that Parry’s approach to Homeric epic posed new questions – some of which still cause debate and disagreement today. An edition alert to current controversies could make a major contribution to settling (or at least framing) them in a manner that increases clarity.

One controversy, as is well known, concerns the existence and nature of a so-called ‘Homeric multitext’. The idea here is that Homeric epic evolved over a long period of time, from a stage of relative fluidity and variation in the Dark Age to one of relative stability in the late Hellenistic period.17 To the proponents of this view, the aim of restoring a single, seventh-century text is not just difficult in practice but wrong-headed in principle. In his review of West’s Iliad, Gregory Nagy writes as follows: ‘there is no original text of the Iliad and Odyssey that Janko or West or anyone else can reconstruct on the basis of the existing textual tradition. The variations that survive in this textual tradition, many of which are transmitted by Aristarchus, prevent such a monolithic reconstruction.’18 This claim has been influential in Homeric scholarship, particularly in the United States, but is by no means universally accepted.19 Careful attention thus needs to be paid to the evidence that can be marshalled in order to support or refute it: assessment of the transmitted variants is, in other words, important.

The addition of many papyri in West’s edition confirms, among other things, that papyrus scraps do not necessarily preserve older readings than the medieval manuscripts, though in some cases of so-called ‘horizontal’ variation (i.e. small-scale differences within a line) they can tip the balance in favour of a specific reading (see, e.g., ad 5.124, 11.173, and 11.199). More important are those instances where the papyri attest ‘vertical’ variation, i.e. a discrepancy in the number of lines. Generally, if a papyrus does not report a line attested in the manuscripts, West brackets or omits that line.20 In this respect, as in several others, his edition differs from van Thiel’s, who adopts a maximalist approach and offers, at times, what seems to us a bloated text. Beyond specific insights about individual passages, the new papyri confirm that the degree of textual variation in Homer is modest compared to the multiformity attested in other oral traditions.21 As we have argued specifically in relation to the Iliad, even the ‘so-called “wild papyri” are not as wild as all that’. 22

Among attested variants, those known to the Hellenistic grammarians Zenodotus, Aristophanes, and Aristarchus have long been considered important, if controversial: did these ancient scholars collate ancient manuscripts (as Ludwich thought) or merely make recommendations of their own (as van der Valk argued)? Was it a combination of the two and, if so, can we tell the difference? Van Thiel dismissed the Alexandrians’ readings as mere conjectures and is censured by West for this: Odyssea, p. X. West’s own view, as set out in the Preface to the Iliad edition (pp. VI-VIII), is that Zenodotus based his text on a single inferior manuscript from Ionia; that Aristophanes used both Zenodotus’ text and other, better ones; and that Aristarchus still worked with a text that was ‘mostly inferior’ to the mainstream tradition.23 This means that West is sceptical of Alexandrian readings – at least in theory. In practice, he seems rather quick to admit into his text variants that are only attested in the work of Hellenistic scholars. This tendency is less obvious in the Odyssey, simply because fewer variants have been transmitted, but West’s decisions concerning Hellenistic variants can be questioned in this edition too – and even in cases where he follows well-established editorial decisions.24

For example, at Od. 1.208 the manuscripts and one papyrus transmit γάρ, whereas Aristophanes and Aristarchus recommend μέν. West, following Bekker, Ludwich, Ameis-Hentze and others, adopts the latter reading, but this is almost certainly wrong – and it is instructive to see why. It is standard in Homer to cap a request for information or action (ἀλλ’ ἄγε + imperative or hortative) with a reason for making it: ‘So come now … (I ask this) because, γάρ …’.25 In this case: ‘So, tell me: are you the son of Odysseus, so grown up as you are? For you look strikingly like him…’ The alternative, a μέν-clause after ἀλλ’ ἄγε … is unidiomatic in early Greek epic,26 and may even sound rude, since the request is not followed by an explanation, as is usual in such instances.27 Now Athena, especially in the guise of Mentes, is a paragon of good manners, so the question then becomes why Aristophanes and Aristarchus preferred μέν to γάρ in this passage. At a general level, Hellenistic scholars took an interest in Homeric particles and often championed alternatives to the received text. In this particular case, μέν, rather than transmitted γάρ, may have appealed to them because it creates a pleasing correspondence with δέ in line 212: ‘You look much like Odysseus when I still saw him regularly (μέν) – but (δέ) since the war I have not seen him’. The fact that the scholia describe μέν as having ‘a certain appeal’ (ἔχει τι εἶδος ἡ γραφὴ αὕτη) suggests that Aristophanes and Aristarchus argued along these lines. Replacing γάρ with μέν (even if, with modern scholars, we take μέν for μήν) involves reading the passage backwards, starting at the end, and articulating it as a prose-like complex period. This is not, on the whole, how Homeric epic worked: Egbert Bakker and others have shown how hexameter versification builds up by forward momentum, rather than backtracking rearrangement.28

The more general point here is that it is often possible, by patient, case-by-case consideration, to discriminate between different kinds of variants and the motivations behind them. Such detailed work does not settle the question of how exactly Homeric epic was composed and transmitted, simply because several models are compatible with the limited evidence available to us. As Albio Cassio points out, Homeric poetry is ‘likely to be the result of extremely complicated processes involving both orality and writing, which we can no longer reconstruct’.29 Still, it is possible to arrive at a more detailed understanding of Homeric composition and transmission by considering individual variants, their likely origin, and their significance.

West’s aim to reach beyond the transmitted text of the Odyssey to what he regards as its 7 th -century ancestor repeatedly leads him to make use of weakly attested variants and, indeed, to adopt modern conjectures in preference to the transmitted text.30 It also leads to intervention at the level of morphology and orthography. Now, of course, grammatical correctness is an obvious criterion for judging the transmitted text, but the crucial question (and this is a question West never addresses in a systematic fashion) is what early epic singers and audiences considered acceptable in terms of morphology and grammar.31 Were their criteria for what sounded correct the same as ours? That would be surprising.

For illustration, let us consider West’s decision, made already in his edition of the Iliad and maintained also for his Odyssey, to switch all third-person singular subjunctives in -ηισι to -ησι. West adduces two arguments for correcting the transmitted text. The first is linguistic: -ησι is the older form (cf. Vedic -āti), and there is no justification, in principle, for the added iota (thus Wackernagel and others since).32 The second is archaeological: the early inscription on the famous ‘Cup of Nestor’ found in Ischia displays the form ΠΙΕΣΙ, without iota (as first noted by Watkins).33 The aim for the Homeric editor, however, is not to print the oldest recoverable form of a word, but rather the Homeric form: as Ludwich pointed out in 1885, ‘Homerisch ist nicht Urgriechisch’. In order to reconstruct the Homeric form, further considerations become relevant. In the case of -ηισι vs -ησι, Martin Peters convincingly demonstrated that transmitted -ηισι should be retained: his study was published in 1998, too late to affect West’s edition of the Iliad, but in time to inform his work on the Odyssey.35 Here is Peters’ argument for why transmitted -ηισι, although typologically younger and ‘less good’ than -ησι, must be the Homeric form. Under pressure from the subjunctive in -ηι (already in Homer the more common form), the archaic ending -ησι acquired an additional iota and, probably as part of the same process of ‘updating’ inherited morphology, was reanalysed as vernacular -ηι plus the athematic ending -σι. On that basis, epic singers created other ‘extended’ subjunctive forms in -ωμι (-ω + μι, 1 st pers. sg.) and -ηισθα (-ηισ + θα, 2 nd pers. sg). The result was a typical proliferation of metrically useful alternatives:

The 3 rd person singular subjunctive in Homer

‘Short’ system (rhythmic shape ‒)


‘Extended’ system (rhythmic shape ‒ᴗ)
-ωμι (understood as -ω + μι)
-ηισθα; (understood as -ηισ + θα)
-ηισι (understood as -ηι + σι)

Since -ωμι and -ηισθα cannot be explained in Homer without transmitted -ηισι, West’s insistence on printing -ησι makes nonsense of the entire system.

The general insight underlying the observation above was set out by Milman Parry in an article published in 1932, where he demonstrated that epic singers, under pressure from a changing linguistic environment, constantly modified their traditional idiom.36 Creative adaptation, rather than textual corruption, accounts for ‘recent’ forms like the famous epic diectasis (ὁρόωντα < ὁρῶντα < ὁράοντα), which arose from the need to adapt the inherited morphology of epic to the contracted forms of the vernacular which the bards heard spoken around them. There can be no question of trying to ‘fix’ Homeric diectasis by reinstating uncontracted forms, as was sometimes attempted in the nineteenth century. More generally, it is important to resist the lure of archaising emendations: the ‘true’ Homeric form (to use one of West’s favourite critical terms: verum) may look morphologically younger than the earliest possible form we can reconstruct.

If we ask, systematically, how individual forms sit within the Homeric Kunstsprache, further progress can be made. As with the evaluation of variants, this needs to be done on a case by case basis. We offer here just one example, by way of conclusion: the much debated (and very Odyssean) phrase ἐν σπέσσι γλαφυροῖσι, ‘in hollow caves’, which has long exercised Homeric scholars. The transmitted form σπέσσι looks strange, to put it mildly.37 West, following Leo Meyer, confidently declares ‘σπέεσι verum est ’ (p. XX), on the ground that the latter is a better representation of original σπέϝεσι.38 Without going into too much detail here, we note that the paradigm of σπέος seems modelled by analogy: ἐν νηῒ γλαφυρῆι ~ ἐν σπῆϊ γλαφυρῶι; ἐν νήεσσι ~ ἐν σπήεσσι.39 We also note that, in the dative plural, Homeric epic includes forms in -έεσσι (ἐπέεσσι), -εσσι (ἔπεσσι) and -εσι (ἔπεσι), but not forms ending in -έεσι. Viewed from the perspective of traditional philology, this observation makes little sense: -έεσι is not, in itself, a meaningful unit of grammatical analysis (i.e. it is no ‘ending’). Early Greek epic singers and their audiences did not, however, read grammar books. They were interested in sound and repeated sonic patterns, which could be used to negotiate between their spoken, everyday language, and the inherited rhythms and formulations of hexameter poetry. In the case of our particular example, the form σπέσσι is the transmitted form, and it is transmitted because that is how it was pronounced. It is by attending to what may have sounded possible to early singers and their audiences that we gain a better understanding of the Homeric text.


1. Stephanie West saw the edition through to publication and, in a paragraph placed at the end of the Preface, explains that Martin West prepared the edition and wrote its Preface; she then checked the whole work and made some further corrections and additions: pauca quidem corrigenda vel supplenda repperi.


3. The importance of this is underscored by the fact that one of its predecessors, the Oxford Classical Text edited by Allen (2 nd edn 1917), is still widely used with little awareness of its shortcomings – despite the detailed criticism in N. Tachinoslis, Handschriften und Ausgaben der Odyssee. Mit einem Handschriftenapparat zu Allen’s Odysseeausgabe, Frankfurt am Main 1984. Allen’s Editio Major of the Iliad (Oxford 1931) is, as has often been noted, equally problematic: see H. van Thiel, Homeri Ilias, Hildesheim 1996, pp. VII-VIII and West himself, Homeri Ilias, vol. 1, Stuttgart and Leipzig 1998, p. X.

4. For details see M. L. West, The Making of the Odyssey, Oxford 2014, esp. pp. 92-142 (process of composition), pp. 35-43 (dating) and pp. 82-91 (background).

5. West acknowledges the help of Oxford graduate students in this enterprise: Odyssea, p. VIII.

6. We repeat here what M. J. Apthorp already pointed out in relation to West’s Iliad in ‘Homer’s winged words and the papyri: some questions of authenticity’, ZPE 128 (1999), p. 16: ‘It should be stressed that the appearance of Vol. I of West’s splendid new Teubner edition of the Iliad …is a major event in the history of both Homeric studies and papyrology.’

7. Van Thiel, Homeri Odyssea, Hildesheim 1991, p. xxiii.

8. See West, Odyssea, p. IX and, e.g., app. crit. ad 2.379a, 11.549.

9. West, Odyssea, p. XVIII: ‘ sed cum in Iliade, cuius traditio tanto firmioribus stat fundamentis, illud ἴμεναι fere nusquam sit, tutius esse arbitror, id etiam ex Odyssea excludere ’.

10. See P. Chantraine, Grammaire Homérique, vol. 1, 2 nd edn, Paris 1948, pp. 46-7 and R. Wachter, ‘Grammar of Homeric Greek’, in A. Bierl and J. Latacz, eds, Homer’s Iliad: The Basel Commentary. Prolegomena trans. B. W. Millis and S. Strack, ed. S. D. Olson (German edn 2000), Berlin and Boston, pp. 84-5, n. 24.

11. Wachter, ‘Grammar’, p. 85, n. 25.

12. Cf. R. Payne Knight, Prolegomena ad Homerum sive de carminum Homericorum origine, auctore et aetate itemque de priscae linguae progressu et praecoci maturitate, 2 nd printing, Leipzig 1816, p. 29, para. 32: neque alia ratione recentiora ab antiquis secernenda esse videntur nisi vera indole ac specie veterrimae linguae e tenebris eruta quam quidem certissimam normam in carminibus in pristinum nitorem ac formam integram restituendis adhibere licet.

13. F. J. Messmann, Richard Payne Knight: The Twilight of Virtuosity, The Hague 1974, pp. 131-2. For ‘Knightianismus’ as an insult in a polemic among competing German scholars, see A. Ludwich, ‘Knightianismus und homerische Textkritik’, Neue Jahrbücher für Philologie und Paedagogik 153 (1896), 1-16.

14. Payne Knight, Prolegomena, pp. 36-7: E brevibus titulis, in laminis et lapidibus insculptis, linguae veteris reliquiae eruendae erant atque ex iis norma aliqua constituenda, qua, exemplaribus inter se collatis, et metri analogiaeque justa ratione habita, e diversorum discrepantiis unum congruens et sincerum, ab omnibus rhapsodorum sordibus purgatum, confici potuisset.

15. West briefly dismisses ‘the oralists’ in The Making of the Iliad, Oxford 2011, pp. 4-5: his focus there is on theories of composition, rather than on the influence of oral poetics on the development of Homeric language.

16. R. Janko makes exactly the same observation in relation to West’s edition of the Iliad : ‘West’s Iliad ’, CR 50.1 (2000), p. 1.

17. G. Nagy, Homeric Questions, Austin 1996; id. Poetry as Performance. Homer and Beyond, Cambridge 1996; and several publications since. Milman Parry demonstrated that the Iliad and the Odyssey stem from a long tradition of composition and re-composition in performance, but did not claim that many written versions of those poems once existed. He first suggested and his direct collaborator, Albert Lord, fully articulated the so-called ‘oral- dictation theory’, cf. The Singer of Tales 1960, p. 153: ‘to Homer belongs the distinction of having composed the longest and best of all oral narrative songs. Their unusual length predicates exceptional circumstances of performance. If I be not mistaken, dictation to a scribe provides this opportunity.’

18. G. Nagy, Review of West, Homeri Ilias in BMCR 2000.09.12.

19. S. Reece points out, ‘Homer’s Iliad and Odyssey : from oral performance to written text’, in M. C. Amodio, ed. New Directions in Oral Theory, Tempe, AZ 2005, p. 52: ‘Among younger Homeric scholars, at least in the United States, the evolutionary model has almost achieved the status of orthodoxy’. He goes on to mount a patient and systematic case against the multitextual model.

20. E.g. Od. 2.393 (retained by van Thiel and von der Mühll; West brackets), 407 (van Thiel retains, von der Mühll brackets; West omits), 429 (van Thiel retains, von der Mühll brackets; West omits). The picture is complicated by the fact that West also adopts ancient and modern deletions without papyrus support, or even against papyrus evidence (as e.g. at 1.140).

21. See S. Reece, ‘ Iliad and Odyssey : from oral performance to written text’, in M. C. Amodio, ed. New Directions in Oral Theory, Tempe, AZ 2005, pp. 65-78.

22. B. Graziosi and J. Haubold, ‘The Homeric text’, Ramus 44 (2015), p. 2.

23. West attributes to the 1 st -century BCE scholar Didymus the work of collating manuscripts, where others credit Aristarchus himself with collation, see Homeri Ilias, vol. 1, Stuttgart and Leipzig 1998, p. VIII and the more elaborate restatement in Studies in the Text and Transmission of the Iliad, Munich and Leipzig 2001, pp. 36-7. On this issue see now F. Schironi, The Best of the Grammarians: Aristarchus of Samothrace on the Iliad, Ann Arbor 2018, pp. 63-78, who gives a more measured account, suggesting that Aristarchus collated in some cases. See also F. Montanari, ‘Alexandrian Homeric philology. The form of the ekdosis and the variae lectiones’, in M. Reichel and A. Rengakos, eds, Epea Pteroenta. Festschrift für Wolfgang Kullmann zum 75. Geburtstag, Stuttgart 2002, pp. 119-40; ‘Ekdosis: a product of the ancient scholarship’, in F. Montanari, S. Matthaios, and A. Rengakos, eds, Brill’s Companion to Ancient Greek Scholarship, vol. 2, Leiden 2015, pp. 660-71.

24. For some Iliadic examples, see B. Graziosi and J. Haubold, Homer: Iliad VI, Cambridge 2010, ad 6.112, 6.157, and 6.285.

25. Il. 24.197-8 (αἰνῶς γάρ) is a particularly close parallel; see also Il. 3.441-2 (γάρ), 6.354-5 (ἐπεί); 11.314-15 (γάρ), 22.254-5 (γάρ), 24.380-5 (γάρ), 522-4 (γάρ), Od. 1.169-73 (μὲν γάρ), 223-6 (ἐπεί), 2.212-14 (γάρ), 6.36-40 (καὶ δὲ … γάρ), 8.389-90 (γάρ), 572-86 (ἐπεί), 11.370-4 (δέ), 457-61 (γάρ), 492-503 (γάρ), 13.296-9 (ἐπεί), 17.190-1 (γάρ), 21.73-4 (ἐπεί, γάρ), 23.20-4 (γάρ), 171-2 (γάρ), 24.256-65 (ἐπεί, γάρ). See also J. D. Denniston, Greek Particles, 2 nd edn, Oxford 1950, p. 59 (‘[γάρ] after an expression denoting the giving or receiving of information, or conveying a summons to attention’).

26. K. F. Ameis and C. Hentze, Homers Odyssee für den Schulgebrauch erklärt, 12 th edn, Leipzig and Berlin 1908, p. 18, explain ‘μέν gleich μήν’, but neither αἰνῶς μέν nor αἰνῶς μήν can be paralleled in early Greek epic, whereas αἰνῶς γάρ is formulaic both in this position and elsewhere in the hexameter line: Il. 10.93, 24.198, Od. 4.597, 17.24 (αἰνῶς γάρ in verse-initial position); cf. HAp. 64 (‒ᴗᴗ αἰνῶς γάρ …), Od. 1.264, 4.441, 22.136, Hes. fr. 29.6 M-W (γὰρ αἰνῶς at line-end).

27. A. Bedke, Der gute Ton bei Homer: Ausprägungen sprachlicher Höflichkeit in Ilias und Odyssee, Münster 2016, shows how important it is to strike the right tone when it comes to giving orders or asking a favour in Homer.

28. E. J. Bakker, Poetry in Speech: Orality and Homeric Discourse, Ithaca, NY 1997. In the classical period, arguably under the influence of prose, we start to see rhapsodes playfully exploring the possibilities of backward rearrangement and manipulation of syntax: the stichomythic exchange in the Certamen Homeri et Hesiodi is our best evidence for that, see B. Graziosi, ‘Competition in wisdom’, in F. Budelmann and P. Michelakis, eds, Homer, Tragedy and Beyond. Essays in honour of P. E. Easterling, London 2001, 57-74; D. Collins, Master of the Game: Competition and Performance in Greek Poetry, Cambridge, MA 2004; and soon P. Bassino, Certamen Homeri et Hesiodi: A Commentary, forthcoming with de Gruyter.

29. A. C. Cassio, ‘Early editions of the Greek epics and Homeric textual criticism in the sixth and fifth centuries BC’, in F. Montanari, ed., Omero tremila anni dopo. Atti del congresso di Genova 6-8 luglio 2000, Rome 2002, p. 114.

30. Here we draw attention to an aspect of West’s editorial approach that seems to us problematic. As already in his text of the Iliad, West seeks to justify modern emendations by identifying ‘traces’ of them in the ancient and medieval transmission; see e.g. Odyssea, pp. XX ( Veri vestigia manent…), XXI ( Accedunt vestigia plura …), and XXV ( Vide quam recte agat papyrus …). The emendation, in other words, is presented as an act of preservation – even though a reading that looks attractive to modern readers may appear in the textual tradition for reasons wholly unrelated. For example, Aristarchus recommended the ‘correct-looking’ κακὰ μήσατο (printed by West) for transmitted κάκ’ ἐμήσατο in Il. 6.157: this need not be because he was concerned to preserve Hermann’s bridge, since we know that he preferred unaugmented forms. Likewise, a second-century Odyssey papyrus that preserves the forms τεινύμενος, ἔτεισαν, τεισόμεθ’, τείσεσθαι, ἐτείσατο ( Odyssea, p. XXV) is unlikely to preserve early Homeric forms: the spellings can, and should, be otherwise explained, see A. C. Cassio, ‘La più antica iscrizione greca di Cuma e τίν(ν)υμαι in Omero’, Die Sprache 35 (1991-1993), 186-207, esp. pp. 199-200.

31. The issue is raised with characteristic clear-sightedness in M. Leumann, Homerische Wörter, Basel 1950, pp. 24-6. On Aristarchus’ understanding of grammar, see S. Matthaios, Untersuchungen zur Grammatik Aristarchs: Texte und Interpretationen zur Wortartenlehre, Göttingen 1999 and now F. Schironi, The Best of the Grammarians: Aristarchus of Samothrace on the Iliad, Ann Arbor 2018, especially pp. 597-622.

32. J. Wackernagel, Vermischte Beiträge zur griechischen Sprachkunde, Basel 1897, pp. 50-1.

33. C. Watkins, ‘Observations on the “Nestor’s Cup” inscription’, Harvard Studies in Classical Philology 80 (1976), 25-40.

34. A. Ludwich, Aristarchs homerische Textkritik nach den Fragmenten des Didymos dargestellt und beurteilt, vol. 2, Leipzig 1885, p. 232.

35. M. Peters, ‘Homerisches und Unhomerisches bei Homer und auf dem Nestorbecher’, in J. H. Jasanoff, H. C. Melchert and L. Oliver, eds, Mír Curad: Studies in Honor of Calvert Watkins, Innsbruck 1998, 586-602, esp. pp. 594-596.

36. M. Parry, ‘The Homeric language as the language of an oral poetry’, Harvard Studies in Classical Philology 43 (1932), 1-50.

37. In the manuscripts, ἐν σπέσσι is often spelled without the double consonant, but this is not an issue: σσ is often spelled with a single σ in the manuscripts; the double consonant is guaranteed by metre.

38. West, Odyssea, p. XXIV, referring to L. Meyer, ‘Die homerischen anlautsgruppen δF und ϰF’, Zeitschrift für Sprachforschung 7 (1858), p. 204. West misreports Meyer’s argument: the suggestion that the Homeric form derives from *σπέ-εσ-σι is not in fact drawn from Meyer (who derives it from *σπέϝεσι) but from the ancient grammarian Herodian, who cites the otherwise unattested form *σπέεσσι to justify the accentuation of Homeric σπέσσι; see A. Lentz, Grammatici Graeci, vol. 3.2, Leipzig 1867, p. 129.

39. That the entire paradigm is problematic has long been recognized: P. Chantraine, Grammaire Homérique, vol. 1, 2 nd edn, Paris 1948, p. 7, for example, describes it as ‘embarrassing’: ‘La flexion de σπέος présente des difficultés caractéristiques. Ce mot …était embarrassant et les graphies de la vulgate trahissent cet embarras’. We argue here that the paradigm can be explained by morphological analogy and the generative power of formulaic language. For example, the accusative σπεῖος γλαφυρόν ( hapax legomenon at Od. 5.194) is ‘reverse-engineered’, as it were, on the basis of the genitive formula σπείους γλαφυροῖο ( Od. 5.68, 5.192; cf. 12.93).