Hostname: page-component-74656f9884-npj9w Total loading time: 0 Render date: 2026-11-07T08:00:38.426Z Has data issue: false hasContentIssue false
  • English
  • Français

SHORT CONSIDERATIONS ON LENGTHENED NAMES: ARISTOTLE, POETICS 21, 1458A3–4

Published online by Cambridge University Press:  21 September 2026

Matteo Moretti Open the ORCID record for Matteo Moretti [Opens in a new window]
Matteo Moretti*
Affiliation:
Princeton University
*
mm9668@princeton.edu
Rights & Permissions [Opens in a new window]

Abstract

This article examines the two examples of ‘lengthened name’ (ἐπεκτεταμένον ὄνομα) given by Aristotle in chapter 21 of the Poetics. It argues that the generally printed text is unsatisfactory because it does not serve the purpose of Aristotle’s examples well. In the case of the first example, the article suggests that a branch of the manuscript tradition other than the one privileged by most editors should be followed, while in the case of the second, a conjecture is offered. An appendix is devoted to highlighting a detail of William of Moerbeke’s translation of the first example.

Keywords

AristotlePoeticsGreek philosophytextual criticismmanuscriptsGreek linguisticsWilliam of Moerbeke

Information

Type
Research Article
Information
The Classical Quarterly , First View , pp. 1 - 6
Check for updates
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2026. Published by Cambridge University Press on behalf of The Classical Association

1. INTRODUCTION

In chapter 21 of the Poetics, Aristotle draws several distinctions within ‘names’ (ὀνόματα).Footnote 1 Among them features the class of ‘lengthened names’ (ἐπεκτεταμένα ὀνόματα). Here is the relevant passage (excluding the parts on shortened names) as printed by most editors, accompanied by a selective apparatus criticus based on a direct reading of the manuscripts in question, which practically coincides with what KasselFootnote 2 and Tarán and GutasFootnote 3 report in theirs (1457b35–1458a1, 1458a1–2, a3–4):

ἐπεκτεταμένον δέ ἐστιν … ἐὰν φωνήεντι μακροτέρῳ κεχρημένον ᾖ τοῦ οἰκείου ἢ συλλαβῇ ἐμβεβλημένῃ … οἷον τὸ πόλεως1 πόληος καὶ τὸ Πηλείδου2 Πηληιάδεω …

1 πόλεως B: πόλεος Π || 2 Πηλείδου Par. gr. 2038: Πηλέος A: Πηλέως B

And a name is lengthened … when it has (i) a vowel longer than its own or (ii) an inserted syllable … for instance polêos for poleôs and Pêlêiadeô for Pêleidou.

Aristotle seems to be specifying two ways in which a name can be lengthened: (i) one of its vowels can be lengthened, or (ii) it can be lengthened by the insertion of a syllable. At first sight, it is not clear whether the disjunction should be taken as inclusive or exclusive, but the presence of two examples might suggest that each is tailored to explain one of the two ways in which a name can be lengthened.Footnote 4 Accordingly, many commentators have taken Aristotle’s first example, that is, saying πόληος for πόλεως, as a case of (i) using a vowel longer than the one proper to the name in question, and the second, that is, saying Πηληιάδεω for Πηλείδου, as a case of (ii) adding a syllable to the name. I believe that this move is correct, but that instead of πόλεως (B) we should print πόλεος (Π), and instead of Πηλείδου (Par. gr. 2038) we should print the conjecture Πηλείδεω. This article attempts to show that, contrary to what several editors and commentators have assumed, Aristotle in the two examples is not contrasting the lengthened form with the standard Attic one. To be sure, the use of the adjective οἰκεῖον in (i) suggests that he is considering (his) ordinary language—that is, the Attic dialect—in that case at least.Footnote 5 But in that context, what Aristotle refers to as οἰκεῖον is not the name’s form itself, but merely the vowel that becomes lengthened. In the example that he provides, this is undoubtedly the vowel ε (§2), which remains unaffected by the discrepancy in the manuscript readings reported above. Therefore there are no a priori reasons to prefer the standard Attic πόλεως to the non-Attic πόλεος.

First, let us sketch the manuscript tradition as outlined by Tarán and Gutas.Footnote 6 Ξ indicates the sub-archetype of the Graeco-Latin strand of the tradition, from which both Π and γ independently derive. On the one hand, Π stands for the lost manuscript written in majuscule script from which both A (Par. gr. 1741, c. mid tenth century) and Φ (the lost manuscript on which William of Moerbeke’s Latin translation [1278] is based) derive through two different transliterations into minuscule script (α and β respectively). Both A and Φ are primary witnesses to the text of the Poetics, and Par. gr. 2038 (fifteenth century) is ultimately derived from A. On the other hand, γ stands for the lost manuscript written in minuscule script from which B (Ricc. 46, c. mid twelfth century), the third primary witness to the text of the Poetics, derives through at least one intermediate copy in minuscule script. More precisely, in this reconstruction, γ represents the manuscript in which the transliteration from majuscule to minuscule script occurred.

Let us now begin by discussing the first of Aristotle’s examples.

2. THE FIRST EXAMPLE OF LENGTHENED NAME

As can be seen from the apparatus criticus in §1, πόλεως is the reading of B, πόλεος of Π (A + Φ). But while it is certain that A presents πόλεος, it is more difficult to determine what Φ had on the basis of Moerbeke’s translation; so it is not certain that Π had πόλεος. We will return to this later (§4), but for now it is safe to assume that Π did indeed have πόλεος.

Now, since B goes back to the sub-archetype Ξ independently of Π (§1), πόλεως and πόλεος seem to have the same evidential value (but more on this in a moment). Hence the editor has to decide what to print on the basis of reasons other than stemmatic ones. Two considerations favour πόλεος (Π).

First, πόλεος would better serve the purpose of the example, since it differs from πόληος in only one vowel, whereas the pair πόλεως–πόληος looks like a case of vowel length swapping or ‘quantitative metathesis’.Footnote 7 As BywaterFootnote 8 notes, Aristotle’s standard was Attic, and πόλεως is the standard Attic form, so he may have considered πόληος to be the lengthened form of πόλεως. Yet the example is clearer if we print πόλεος instead of πόλεως, since the pair πόλεος–πόληος differs in only one vowel, the difference being that -η- is the lengthened form of -ε- (the vowel referred to as ‘proper’ to the name in question: §1).

The second consideration is that πόλεος is a lectio difficilior. This form of the genitive of πόλις is attested before Aristotle in Ionic inscriptions from Oropos (IG VII 235.41 = GDI 5339.41) and Amorgos (IG XII 7.103 = GDI 5354), as well as in poetry—particularly Attic drama—often for metrical reasons.Footnote 9 It is a rare form, though. By contrast, πόλεως is the standard Attic expression. It is therefore likely that a scribe mistakenly wrote πόλεως instead of πόλεος. It may be objected that πόλεος is not a lectio difficilior in this context, since a scribe could have had the same considerations about the accuracy of the example that I have and decided to write πόλεος instead of πόλεως. But it requires a decent knowledge of Greek and a good understanding of the text of the Poetics to realize that πόλεος is an existing form—given its rarity—and that it might better serve the explanatory purpose of the example. Hence such an objection does not carry much weight.Footnote 10

The objection carries even less weight when we consider the pair Πηλέος (A)–Πηλέως (B) a few words later (the reading Πηλείδου, as argued in §3, is probably a conjecture). As Bywater explains, it is easy to suppose that this word, whatever its correct form, ‘got in by assimilation to the preceding πόλεως (or πόλεος, as it is in AC)’,Footnote 11 but it is worth asking which form—πόλεως or πόλεος—prompted the mistake in the first place. This is important because such an assimilation is shared by the entire Graeco-Latin strand of the tradition, meaning it must date back to the sub-archetype Ξ at least, along with the form that caused the error.

It seems likely that πόλεος caused the error. For if there could be a reason for changing πόλεως to πόλεος, that is, the presumption of making Aristotle’s example clearer than what he himself did, the same rationale however cannot be applied to changing Πηλέως to Πηλέος, since the pair Πηλέως–Πηληιάδεω hardly looks like a case of lengthening a name by (i) lengthening one of its vowels or (ii) inserting a syllable, and Πηλέος would not help. The only reason that I can think of is to achieve consistency with the preceding (and, in such hypothesis, emended) πόλεος, but this is rather weak—especially if one assumes that the same scribe is responsible for both πόλεος and Πηλέος, because this would imply that the scribe understood the first example well but neglected the second.Footnote 12 By contrast, the change from Πηλέος to Πηλέως would be easier to explain. We would have to suppose only that, just as a scribe who did not understand the ending -εος replaced it with the more familiar -εως in the first instance of such an ending, the same phenomenon occurred with the second instance too (either due to the same scribe or another).

It is therefore likely that πόλεος caused the error and was consequently the reading of Ξ. For this reason, and also because it is better than πόλεως in this context, we should follow Π and print it.Footnote 13

THE SECOND EXAMPLE OF LENGTHENED NAME

What about the second example, where we expect a name to be lengthened by inserting a syllable? The reading printed by most editors, Πηλείδου, is found in Par. gr. 2038 and is probably a conjecture, as is usual in this manuscript.Footnote 14 The other manuscripts have Πηλέος (Π) or Πηλέως (B), which cannot be correct because Πηληιάδεω is not the same word with an inserted syllable, but a different word from Πηλέος–Πηλέως.Footnote 15

Πηλείδου is in principle a standard Attic form, though unattested before Herodian. Moreover, it does not illustrate Aristotle’s point, namely the insertion of a syllable, in a straightforward way.Footnote 16 So we might think that it is better to conjecture another form of the genitive, namely Πηλείδεω. This form is attested four times in Homer, most notably in Odyssey 8.75, about the quarrel between Odysseus and Achilles, which was famous;Footnote 17 more generally, patronymics with such an ending are common in Homer and Herodotus. Most importantly, Πηλείδεω is preferable to Πηλείδου because it is closer to Πηληιάδεω, and the inserted syllable, -α-, is easily identifiable. Since Πηλείδου is no less a conjecture than Πηλείδεω, the latter deserves consideration.Footnote 18

Hence we should print Poetics 21, 1458a3–4 thus:

οἷον τὸ πόλεος1 πόληος καὶ τὸ Πηλείδεω2 Πηληιάδεω …

1 πόλεος Π: πόλεως B || 2 Πηλείδεω scripsi: Πηλείδου Par. gr. 2038: Πηλέος A: Πηλέως B

For instance polêos for poleos and Pêlêiadeô for Pêleideô.

APPENDIX: MOERBEKE’S TRANSLATION OF THE FIRST EXAMPLE

Both KasselFootnote 19 and Tarán and GutasFootnote 20 report that πόλεος is the reading of Π, in the belief that not only A but also Φ (the manuscript used by Moerbeke for his translation) presented πόλεος. This is probably correct, as will be shown below, but not obvious, since Moerbeke translates as follows:

superextensum quidem puta “poleos” (idest canus) per e breue, “polios” per ita longum.

Lengthened, on the one hand, is for example poleos (that is, white) with an epsilon, which is short—polios with an eta, which is long.

At first sight, poleos in Moerbeke’s text could be πόλεος or πόλεως, since he could not distinguish between omicron and omega writing in Latin. The explanation ‘idest canus’ is curious, however, and may provide evidence that Moerbeke did indeed read πόλεος.

In fact, there is little chance that such an explanation was added by anyone other than Moerbeke himself. For his translation was completed in 1278, and the two independent manuscripts that contain it, whose reading I have verified, can be dated close to that year, one around 1280—namely the Toletanus bibl. Capit. 47.10 (T)—and the other around 1300—namely the Etonensis 129 (O). Since the qualification ‘per e breue’ makes it certain that Moerbeke was reading a word with an epsilon, and the specification ‘idest canus’ would be inexplicable if he were reading the common Attic form πόλεως, we can suppose that he had the rare form πόλεος in front of him and mistook it for the nominative of the adjective πολιός, which means precisely ‘white’, ‘grey’.Footnote 21 All in all, then, it is more likely that πόλεος is the reading of the lost Φ and, since it is also A’s reading, of the lost Π.

Footnotes

*

This paper is based on a presentation I gave to the Princeton Classical Philosophy reading group on Aristotle’s Poetics. I thank the participants for discussion, especially Andrew Ford, Barbara Graziosi, Mirjam Kotwick, Hendrik Lorenz and Ben Morison. Special thanks to Mirjam Kotwick, Jesse Lundquist and Alexa White, who read and commented on earlier versions of the paper. Martina Buston and Lorenzo Salerno also helped with specific issues. Christina Kraus and Arianna Trecarichi played a key role in obtaining digitised reproductions of some important manuscripts. Lastly, I thank CQ’s reader and Editor for comments and suggestions.

References

1 There is some scholarly debate about how to translate the term ὄνομα in this chapter, with the three main alternatives being ‘name’, ‘noun’ and ‘word’. For my purposes, however, it makes no difference.

2 R. Kassel, Aristotelis De Arte Poetica Liber (Oxford, 1965), 35.

3 L. Tarán and D. Gutas, Aristotle Poetics (Leiden and Boston, 2012), 202. Unlike Kassel (n. 2), 35, however, they say that Πηλέος is not the reading of A but of Π, which is not entirely correct (see n. 12).

4 In principle, a name could be lengthened both by lengthening one of its vowels and by adding a syllable. The passage does not suggest this possibility, however, and the later grammatical tradition also seems to consider the two alternative ways of lengthening a name independently of each other: see Apollonius Dyscolus, Pron. 60.11 (lengthening of a vowel), 34.5, 37.9–10, 99.1 (insertion of a syllable).

5 In Aristotle’s classification, οἰκεῖον ὄνομα appears to have a similar meaning to κύριον ὄνομα, which is defined as an ordinary or prominent name in a given community in Poet. 21, 1457b3–6. For the similarity between the two categories see E.M. Cope and J.E. Sandys, The Rhetoric of Aristotle, vol. 3 (Cambridge, 1877), 18 on Rh. 3.2, 1404b31–7.

6 See Tarán and Gutas (n. 3), 3–76, 129–59. I will refer to their outline of the manuscript tradition and stemma throughout this article.

7 As defined by B. McCall, ‘Metathesis’, in G.K. Giannakis (ed.), Encyclopedia of Ancient Greek Language and Linguistics, vol. 2 (Leiden and Boston, 2014), 418–19, metathesis is ‘a linguistic phenomenon whereby two adjacent sounds are transposed’, and in particular quantitative metathesis occurs ‘where the length of the vowel metathesizes while the quality of the vowel remains’, as e.g. in the pair βασιλῆος–βασιλέως.

8 I. Bywater, Aristotle on the Art of Poetry (Oxford, 1909), 288 on Poet. 21, 1457b35.

9 Cf. E. Schwyzer, Griechische Grammatik, vol. 1 (Munich, 1939), 572 and n. 3 for epigraphic occurrences of πόλεος (and of other words with the genitive form -εος). A.C. Cassio, ‘L’épigramme pour Théotimos d’Atrax, CEG 637 (457 av. J.-C.): mise en page, dialecte, ambiguïtés’, RPh 92 (2018), 57–72, at 65–6 analyses another interesting occurrence of πόλεος in a Thessalian funerary epigram. (I thank CQ’s reader for emphasizing the importance of epigraphic evidence to me.) The form πόλεος is not found in the Homeric poems (with the possible exception of Il. 21.567), but is attested in Theognis 1.56—where, however, -εο- is scanned as one syllable—and occurs in each of the three great Attic tragedians; for passages see R. Kühner and F. Blass, Ausführliche Grammatik der griechischen Sprache (Hannover, 18903), §126.5; R. Jebb, Sophocles: The Plays and the Fragments. Volume 3: The Antigone (Cambridge, 1888), 40 on Ant. 162. In some cases, πόλεος was corrupted into πόλεως and then restored metri causa: see Eur. fr. 360.11, 21 TrGF. In Empedocles fr. D4.2 Laks–Most, πόλεος has been restored because Empedocles does not use the Attic ending (-εως) of the genitive case: R. Merzdorf, Commentationes Philologae (Leipzig, 1874), 41–56, at 53–4.

10 Moreover, at least with regard to MS A, there are no conspicuous instances of attempts to correct the text. Bywater (n. 8), xxx lists a few instances where A appears to emend, but in most of these cases it is not even possible to imagine an emendation rather than a scribal slip. More generally, it is more common to trivialize the text than to make it more complex. This observation also suggests that πόλεος is not, as one might suppose, a mere scribal error resulting from inaccurate subvocalization after vowel quantity was lost.

11 See Bywater (n. 8), 288 on Poet. 21, 1458a4.

12 To evade this counterargument, a supporter of πόλεως might argue that Π changed πόλεως to πόλεος, and Πηλέος was only written by A or its antigraph α. Indeed, we cannot be sure as to whether Φ had Πηλέος or Πηλέως, as Moerbeke’s Pileos could reflect either. Therefore, there is no guarantee that Πηλέος was already in Π. As mentioned above, however, there is no apparent reason why a scribe should have intentionally changed Πηλέως to Πηλέος, and it would be far-fetched to suggest that the Πηλέως resulting from the first assimilation was assimilated again by A (or α) to Πηλέος. Thus assuming that Πηλέος was already in Π seems to be the more economical explanation of the evidence. This would mean that the supporter of πόλεως would have to argue that one and the same scribe had a good grasp of the first example but thoroughly misunderstood the second.

13 Even if πόλεος were the reading of Ξ, it could be the result of an even more ancient mistake, and the correct form πόλεως could have been recovered by emendation. But such a complicated story would only have to be told if πόλεως were decisively superior to πόλεος, which it is not. In any case, T. Twining, Aristotle’s Treatise on Poetry (London, 1789), 169, is the only editor that I know of who has printed πόλεος. In his footnotes to the translation of the passage, J. Sykutris, Αριστοτέλους Περί ποιητικής (Athens, 1937), seems to have πόλεος in mind, but he does not print it or argue for it. Perhaps he only uses it for explanatory purposes.

14 See Tarán and Gutas (n. 3), 149–52, who argue that the readings peculiar to this manuscript are insufficient to postulate the existence of another lost manuscript representing an independent strand of textual tradition, but are best explained as humanistic conjectures. This view has been challenged by R. Janko, review of Tarán–Gutas, CPh 108 (2013), 252–7, at 255–7; but L. Tarán, ‘The text of Aristotle’s Poetics in the Codex Parisinus Graecus 2038’, Mnemosyne 69 (2016), 785–98 has provided a convincing response to Janko’s criticism.

15 C. Gallavotti, Aristotele. Dell’arte poetica (Milan, 1974), 182–3 defends Πηλέως by arguing that this expresses the patronymic in Attic usage and thus parallels the Homeric patronymic Πηληιάδεω. But this consideration is insufficient to justify retaining Πηλέως, which, as argued above, renders Aristotle’s text inconsistent. On the other hand, M. Schmidt, Aristoteles. Über die Dichtkunst (Jena, 1875), 52 retains Πηλέος and proposes printing τὸ Πήλεος (sic) <Πηλῆος καὶ τὸ Πηλείδου> Πηληιάδεω. As Bywater (n. 8), 288 on Poet. 21, 1458a4 notes, however, the pair Πηλέος–Πηλῆος would constitute a second instance of the first way of lengthening a name (see [i] in §1), making Aristotle’s text redundant. Moreover, as argued below, there are problems with the form Πηλείδου. Therefore, Schmidt’s addition should be rejected.

16 For the sake of clarity, despite Aristotle’s own definition of syllable in chapter 20 (1456b34–7), the inserted syllable here should be α (probably because, as Bywater [n. 8], 268 on Poet. 20, 1456b34 notes, the common use of the word συλλαβή differs from the technical one Aristotle uses in chapter 20, and usually a συλλαβή is simply a unit of sound into which a word, foot or verse can be divided).

17 This observation becomes even more compelling when we note that Homer is referred to repeatedly in these lines of the Poetics: ἀρητήρ (1457b35) probably derives from the famous Il. 1.11, or at least from the Homeric poems, since the word is basically not found outside them. The form Πηληιάδεω is obviously taken from Il. 1.1. Two of the three ‘shortened names’ given as examples, namely κρῖ and δῶ (1458a5), are clearly Homeric (the former is used almost exclusively in Homer). Finally, the line cited in 1458a7 as an example of an ‘altered name’ is Homeric (Il. 5.393).

18 Again, Sykutris (n. 13) seems to have Πηλείδεω in mind in his footnotes to the translation of the passage, but he does not print it or argue for it, using it merely for explanatory purposes.

19 Kassel (n. 2), 35.

20 Tarán and Gutas (n. 3), 202.

21 In his apparatus criticus, Gallavotti (n. 15), 80 seems to recognize the connection between canus and πολιός, but it is not clear whether he thinks that Moerbeke’s Greek manuscript had πολιός (and, if so, how he would explain the qualification ‘per e breue’).