¹ Wackernagel, J., ‘Indogermanische Dichtersprache’, Philologus 95 (1943), 1–19Google Scholar; reprinted together with other important articles on Indo-European poetry in Schmitt, R. (ed.), Indogermanische Dichtersprache (Darmstadt, 1968), pp. 83–101Google Scholar. See also Schmitt, R., Dichtung und Dichtersprache in indogermanischer Zeit (Wiesbaden, 1967), pp. 61–102Google Scholar, West, M. L., ‘Greek poetry 2000–700 b.c.’, CQ N.S. 23 (1973), 179–92CrossRefGoogle Scholar, Nagy, G., Comparative Studies in Greek and Indic Meter (Cambridge, Mass., 1974)CrossRefGoogle Scholar, and Floyd, E. D., ‘Kleos aphthiton: an Indo-European perspective on early Greek poetry’, Glotta 58 (1980), 133–57Google Scholar.

² Cf. Webster, T. B. L., ‘Early and late in Homeric diction’, Eranos 14 (1956), 34–48Google Scholar, Page, D., History and the Homeric Iliad (Berkeley, 1959), pp. 222ff.Google Scholar, Hoekstra, A., Homeric Modifications of Formulaic Prototypes (Amsterdam, 1965), pp. 24f.Google Scholar, West, M. L., CQ N.S. 23 (1973), 187Google Scholar, Horrocks, G. C., ‘The antiquity of the Greek epic tradition’, PCPS 206 (1980), 5Google Scholar.

³ Such as κλέος εὺρύ,μέγα κλέος, κλέα ⋯νδρ⋯ν et alia (for the full list see Schmitt, op. cit. pp. 71ff.).

⁴ See e.g. Schmitt, op. cit. pp. 65f., Nagy, op. cit. pp. 142, 229f. and passim, Floyd, E. D., Glotta 58 (1980), 133Google Scholar.

⁵ I follow J. B. Hainsworth in holding that any expression that occurs at least twice in Homer should be accounted a formula; see The Flexibility of the Homeric Formula (Oxford, 1968), p. 42Google Scholar.

⁶ Unless of course one claims that all Homeric expressions are formulae. That this is hardly probable was convincingly argued by Hainsworth, in ‘Structure and content in epic formulae: the question of the unique expression’, CQ N.S. 14 (1964), 155–64CrossRefGoogle Scholar and Hoekstra, op. cit. pp. 15f., 24.

⁷ Cf. Page, op. cit. pp. 224ff., Hainsworth, op. cit. pp. 7f. Actually, the number of so-called ‘equivalent formulae’ varies with the criteria applied for discerning the reasons underlying the duplication. The question has not yet been sufficiently explored (among recent studies see Janko, R., ‘Equivalent formulae in the Greek epos’, Mnemosyne 34 [1981], 251–61)CrossRefGoogle Scholar, but it is reasonable to suppose that Homer's motives for replacing one formula with another were much more subtle than many are ready to admit. The majority of the equivalent formulae are expressions which are especially current in Homer's usage, so it is quite possible that prevention of monotony furnished the oral poet with sufficient justification for retaining two mutually interchangeable formulae (cf. Hainsworth, op. cit. pp. 9f.). Similarly, such ‘equivalent’ formulae as βο⋯πις πότνια Ἣρη and θε⋯ λευκώλενος Ἣρη, attesting as they do two different attributes of the goddess, might have been considered semantically complementary and therefore functionally justified, as was demonstrated for the pair πολυɸλο⋯σβοιο θαλ⋯σσης and θαλ⋯σσης εὺρυπόροιο by Gray, D. in her ‘Homeric epithets for things’, CQ 41 (1947), 109–21, at p. 111Google Scholar.

⁸ 4.584, 7.333; ἄɸθιτος and ἄσβεστος are not metrical equivalents as was mistakenly held by B. Mader in Lexikon des frühgriechischen Epos i. s.v. ἄɸθιτος.

⁹ Though μοι, at the beginning of the apodosis refers to both its clauses, this does not turn ἄɸθιτον into the attribute of κλέος (μοι…κλέος ἄɸθιτον ἔσται ‘I shall have imperishable fame’): here, μοι is the dative of interest, not of possession (μοι at Il. 9.413 appears in P. Chantraine's list of examples of the dative of interest in Homer; see Grammaire Homérique ii. p. 71). Homeric background for the use of the dative of interest with predicative adjectives is furnished by Vivante, P. in The Epithets in Homer (New Haven & London, 1982), p. 109Google Scholar: ‘Whereas the noun–epithet phrase is self-contained, the predicate is quite relative to the context. …the quality [of predicates — M.F.] needs its immediate term of reference, needs to be strongly connected with the occasion; it is not expressed as an absolute endowment standing above the passing action. Hence the “dative of interest” which we so often find with such predicates…’

¹⁰ κλέος οὔποτ' ⋯λεῖται Il. 2.325, 7.91; Od. 24.196; ἄσβεστον κλέος εἴη Od. 4.584, 7.333; κλέος οὺραν⋯ν ἴκει Il. 8.325; Od. 9.20. Related expressions are αἰε⋯ / …κλέος ἔσσεται ⋯σθλóν Od. 24.93f., which need not be taken into account here because of its uniqueness and the separation between two verses, μέγα (μέγιστιν) ὺπονρ⋯νιον κλέος εἴη (⋯στίν) Il. 10.212; Od. 9.264, which is not fixed and is too long, and κλέος οὐραν⋯ν εὐρὺν ἴκανέν Od. 8.74, 19.108, which is an expanded version of κγέος οὐραν⋯ν ἴκέι Cf. also Il. 7.451, 458, 17.232; Od. 18.255, 19.128.

¹¹ Though neither ἄσβεστον κλέος έἴη nor κλέος οὐραν⋯ν ἴκέι is formulated in the future indicative, the difference is only crucial with regard to the latter. In the former έἴη may be replaced by ἔσται without causing any metrical change; however, putting κγέος οὐραν⋯ν ἴκέι in the future tense would involve a change in the metrics, because the epic future of ἴκω is ἴξομαι (see LSJ ² s.v. ἴκω, ἱκνέομαι).

¹² See Gray, , CQ 41 (1947), 116Google Scholar, Page, op. cit. p. 228, Hainsworth, , CQ N.S. 14 (1964), 163fCrossRefGoogle Scholar. and op. cit. p. 40, Hoekstra, op. cit. pp. 13f.

¹³ Page, op. cit. p. 226, Hoekstra, op. cit. pp. 26ff. After Homer κγέος ἄɸθιτον, with or without εἶναι, occurs in the inscription from Krisa, Schwyzer Nr. 316, Pseudo-Hesiod, fr. 70.5 Merkelbach and West, Sappho, fr. 44.4 L.-P., Ibycus, fr. 1.47 Page, and Ion of Chios, fr. 7.3 D.; see Schmitt, op. cit. pp. 62f., Floyd, E. D., Glotta 58 (1980), 140ffGoogle Scholar.

¹⁴ Cf. Page, op. cit. pp. 226ff.

¹⁵ Il. 2.46, 186 (applied to the sceptre of Agamemnon), 13.22 (the palace of Poseidon), 14.238 (the throne promised by Here to Hypnos). As Parry demonstrated in ‘Les Formules et la métrique d'Homère’ (‘Homeric formulae and Homeric metre’ in The Making of Homeric Verse [Oxford, 1971], pp. 198f.Google Scholar), association of the formula with this particular metrical position was strong enough to cause a metrical fault in Il. 13.22. ἄɸθιτον at Il. 9.413 is the only instance in Homer of the adjective occurring in its fixed position without being followed by αἰεί.

¹⁶ When used alone and not in its fixed position (three times in the fourth and once in the first foot), ἄɸθιτος is applied to the golden wheel-rim of Hera's chariot (Il. 5.724), to the palace of Hephaestus (Il. 18.370), to the marvellous grapevines on the Island of the Cyclops (Od. 9.133), and to the plans (μήδεα) of Zeus (Il. 24.88). That only two out of the nine cases in which ἄɸθιτος is found in Homer, with or without αἰεί, fall into the sphere of incorporeal objects, indicates that the concrete associations of the term must have been the original ones; at the same time, the fact that its application to incorporeal entities like κγέος or μήδεα (both hapax legomena in Homer) sharply increases in post-Homeric poetry seems to entail that the term's meaning evolved from concrete to abstract. The development of the term is so understood by H. Ebeling (not yet influenced by the Indo-European hypothesis!) in his Lexicon Homericum ² i. s.v. ἄɸθιτος (‘transfertur ad ea quae mente percipiuntur’); the same evolution of ἄɸθιτος from the concrete to the abstract meaning is traced in Treu, M., ‘Griechische Ewigkeitswörter’ Glotta 43 (1965), 10ffGoogle Scholar. However, in the article on ἄɸθιτος in Lexikon des frühgriechischen Epos i., B. Mader argues that the term underwent a peculiar evolution from the abstract to the concrete and again to the abstract, placing κλέος ἄɸθιτον at the beginning of its semantic development.

¹⁷ Il. 10.212, 22.514; Od. 4.584, 9.264. See Nagy, op. cit. pp. 105f.

¹⁸ This specific type of separation at the verse-end is described in Hainsworth, op. cit. p. 97.