² Preface to Lyrical Ballads, edd. Brett, R. L. and Jones, A. R. (revised edition, 1965), pp. 243Google Scholar, 245, 254, 250, 251.

³ cf. Silk, M. S., ‘Aristophanes as a lyric poet’, YCIS 26 (1980), 123–5Google Scholar.

⁴ Po. 1458 a 18–22: the other characteristic was clarity. It is arguable that archaism must always be the most characteristic source of poetic diction: cf. Barfield, Owen, Poetic Diction³ (1973), 152Google Scholar.

⁵ So, rightly, Whatmough, J., Poetic, Scientific and other Forms of Discourse (1956), 105 f.Google Scholar; the point is usually obscured by commentators (see e.g. Bywater's long discussion at 1457b4).

⁶ Lex. Hippocr. 19. 66 Kühn ⋯ γλ⋯ττα παλαι⋯ν ⋯στιν ⋯νομα τ⋯ς συνηθειας ⋯κπεπτωκός: γλῶττα in this sense occurs as early as Aristophanes (fr. 222).

⁷ And conceivably before: see Pfeiffer, R., History of Classical Scholarship from the Beginnings to the End of the Hellenistic Age (1968), 78 fGoogle Scholar., 90–2.

⁸ Ars Poetica 68 ff.

⁹ Near the end of Part I (p. 214 in the standard editions).

¹⁰ Martin, G. D., Language, Truth and Poetry (1975), 196Google Scholar.

¹¹ cf. Silk, M. S., Interaction in Poetic Imagery (1974), xii fGoogle Scholar. and 261 (‘LSJ: deficiencies’), and n. 86 below.

¹⁸ cf. below, pp. 327 f.

¹³ Leumann, M., Homerische Wörter (1950)Google Scholar; Harvey, A. E., ‘Homeric Epithets in Greek Lyric Poetry’, CQ n.s. 7 (1957), 206 ffCrossRefGoogle Scholar.

¹⁴ Dichtung und Philosophic des friihen Griechentums (1962), 30; cf. Antidoron (Festschrift Wackernagel, J., 1923), 274 ffGoogle Scholar.

¹⁵ Dichtung und Philosophic, 30, n. 5.

¹⁶ Homerische Wörter, 231–3.

¹⁷ Antidoron (Festschrift Wackernagel), 274 f.

¹⁸ The most notable complications arise from the existence of an Arabic version of Galen's commentary on Hp. Epid. 6 alongside an often defective Western manuscript tradition (see Wenkebach's, preface to CMG V 10Google Scholar. 2. 2, pp. x ff.).

¹⁹ Galen's discussion is long and not as clear-cut as it might be: CMG V 10. 2. 2, 46–54.

²⁰ For whom Wenkebach conjectured Pamphilus (I A.D.): see Philologus 86 (1931), 325Google Scholar, and Smith, W. D., The Hippocratic Tradition (1979), 153–5, 160–2Google Scholar; on Hippocratic glossography in general, see Wellmann, M., Hippokratesglossare (1931)Google Scholar.

²¹ CMG V 10. 2. 2, 47 f.

²² Xenocritus of Cos, Bacchius of Tanagra and Philinus of Cos: see the succinct comments of Smith (n. 20 above), 202 f., 211 f. and Pfeiffer (n. 7 above), 92, n. 2. v

²³ And also, incidentally, alluded to in Nicander's Φλύκταιναι.

²⁴ On δυσπέμΦελος see West on Hes.Th. 440.

²⁵ See Interaction in Poetic Imagery (n. 11 above), especially pp. 27–56. The abnormality discussed there was metaphor, but mutatis mutandis the discussion applies here too.

²⁶ Notably twentieth-century behaviourist attempts to construct models of language that ignore mental processes, and not least the spurious antimentalist equation (associated, for instance, with Wittgenstein), ‘the meaning is the use’. Whatever the theoretical interest of this equation, its pragmatic value for the lexicographer is akin to the value of Wordsworth's ultra-mentalist proposition, ‘poetry is the spontaneous overflow of powerful feelings’, for the practising critic. Poetry may or may not have something to do with spontaneous overflows, but it certainly cannot be one itself. Nor can meaning be the use; cf. Martin (n. 10 above), 14 ff., and n. 71 below. On linguistic behaviourism see further John Lyons, Semantics I (1977), 121 ff., and Martin, 9 ff. My discussion of meaning owes a good deal to Martin's admirable book and a little to Benveniste's distinction between ‘signifiance’ (a word's meaning within a language) and ‘sens’ (its meaning in a context), on which see conveniently Todorov, T. in Mélanges linguistiques offerts a Emile Benveniste (1975), 509 ffGoogle Scholar.

²⁷ The qualifications are designed to cater for abstract expression: cf. Martin (n. 10 above), 68 f. On reference and referent see Lyons (n. 26 above), 95 ff., 174 ff., and Martin, 8 f., 25 ff., 68 ff.

²⁸ This essentially literary-critical use of ‘denotation’ and ‘connotation’ differs from those favoured by semantic theorists, on which cf. Lyons (n. 26 above), 175 f.

²⁹ The model is a modified version of one discussed by Martin (n. 10 above), 23, 42.

³⁰ This view of metaphor is essentially the one taken in Interaction in Poetic Imagery (n. 11 above), especially pp. 9–12, 103 f.; see also Nowottny, W., The Language Poets Use (1962), 49 ffGoogle Scholar.

³¹ On this point, cf. p. 313 below.

³² In a letter to Robert Bridges, 8 October 1879.

³³ CMG V 10. 2. 2, 49.

³⁴ For what it may be worth, I note that haloes are roughly the same shape as parts surrounding pustules. It could equally well be argued that the referents are different - and Pearson for one (on S. fr. 338) believes that the Sophoclean words ‘refer to the brightness of the golden fleece’.

³⁵ From this point I broadly follow principles and procedures set out in Interaction in Poetic Imagery (n. 11 above), chs. 2 and 4. Inter alia, I do not appeal to material later than the fourth century B.C. either as prospective iconyms or as evidence about them (cf. Interaction, 38–42, 82) – even though Hellenistic usage in particular often shows discussible developments. Such was the artificial linguistic milieu of the Hellenistic litterateur (pertinently represented by his access to - and his academic interest in – lists of γλῶτται from earlier literature compiled by others like himself), that only confusion can result from taking pre-Hellenistic and Hellenistic together.

³⁶ In Lyons'terms (n. 26 above, I. 18 ff.), the two would be different ‘lexemes’. The distinction between polysemy and homonymy is hard to draw on any level (ibid. n. 550–69).

³⁷ Which current linguistics finds problematic in itself: e.g. Lyons (n. 26 above), II. 553 f.

³⁸ cf. Silk (n. 11 above), 35 and 37, n. 7.

³⁹ Whereas Pindar's application of the word to ιππος is more or less relatable to the Homeric applications of and itself to a lion (/. 12.46 and 24.42 respectively). For a useful discussion ofayqvwp in some of its other aspects, see A. Sideras, Aeschylus Homericus (1971), 42.

⁴⁰ So E. Degani, Helikon 2 (1962), 43, in the course of a compendious study of the word (pp. 37–56).

⁴¹ The missing citations are Pi. /. 3. 2. (withKopos), A.fr. 442. 2 Mette (withfiios), andMel. Adesp. S 458. 1. 3 (apparently quasi-adverbial, Svafj.oy£wv alyves). The adjective was also conjectured by Ahrens at A.Ch. 68, and one derivative is attested in Aeschylus:Svaaiavfi fioav, Pe. 281. It may be noted that the Corcyra inscription cited by LSJ(IG 9 (1) 886. 2) is post-classical.

⁴² Interaction in Poetic Imagery, 59–63,233–5. See further the essays collected by Molina, D.Newton-de in On Literary Intention (1976)Google Scholar, although I confess that some of the contributors, especially the ‘pro-intentionalists’, seem to me deficient in the sharp focusing required to assist scrutiny of the question.

⁴³ In terms of quantity of evidence, as a matter of fact, LSJ tends to be very much better with rare words than ordinary ones (on which I refer to my comments in Interaction in Poetic Imagery, 83). The instances of⋯λίαστος not cited by LSJ are newly discovered and/or scrapings: B.(?) fr. 60. 10 f.⋯λὑπò;ɩ [άστοɩς] ὑπò πένθєσɩν Trag. Adesp. P. Ox. 3216. 7 ⋯λίαστον; (no context), E.fr. 1123 dub.⋯λίαστον (no context, but perhaps unjustly suspected by Nauck).

⁴⁴ AtOr. 1479 speculatively emended (essentially on metrical grounds) to ἄλαστος by Biehl (following Wilamowitz) in his 1965 Berlin edition and the 1975 Teubner.

⁴⁵ Glotta 32 (1953), 236–8.

⁴⁶ The presumption of an original etymological connection between αʅθοψ and αἲθ-, ‘burn’, does not justify the decision of Lex. fr.-gr. Ep. to make a composite article out of αîθοψ, αἵθων et al.

⁴⁷ Interaction in Poetic Imagery, 239 f. and 101 (onP. 10. 53 f).

⁴⁸ Ibid.

⁴⁹ As O. 1. 15μοʋσɩκ⋯ς ⋯ν ⋯ώτῳs: cf. 3. 3 f., P. 10. 53, I. 1. 51, I. 7. 18, I. 8. 16a, fr.6b (f), fr. 52 f. 59, and, more distantly or tenuously, O. 5. 1, O. 8. 75, O. 9. 19, I. 5. 12 (in which instance there is perhaps a faint allusion to ‘wool’ throughποɩμαίνοντɩ Stoneman, cf. R., Maia 33(1981), 130)Google Scholar.

⁵⁰ KN Dd 1157 has the sequence a-wo-ti-jo, which Ruijgh, C. J., Etudes sur la grammaire et le vocabulaire du grec mycènien (1967), 158Google Scholar, interprets as ʹΑFώτɩος or ʼΑFωτίων, a shepherd's name (and certainly found in the vicinity of the sign for’ ovis'),’ dérivé de ἂFωτος (˃ ἂωτος), flocon (de laine), laine fine’.

⁵¹ , Hes.fr. 9. 2Google Scholar; PY Jn 389 and K N Ch 900 (ko-so-u-to-); the proper name occurs also at Dialexeis 5. 11, on which see n. 55 below.

⁵² Winds are themselves sometimes represented as flying creatures, but usually in visual art, rarely in words: Page, cf. D. L., Further Greek Epigrams (1981), 365Google Scholar, Roscher, Lex. Myth., s.w. Boreas, Windgotter,Neuser, K., Anemoi (1982), 232 fGoogle Scholar.

⁵³ Dale, on E. Hel. 1111Google Scholar.

⁵⁴ Fraenkel, on Ag. 1142Google Scholar; Wilamowitz, on E.HF 488Google Scholar.

⁵⁵ loc. cit.; cf. e.g. Kannicht, on E.Hel. 1111Google Scholar. Perhaps the best evidence for a classical association with colour (though not for a classicalmeaning) is the sole occurrence of the word not cited by LSJ, , Dialexeis 5. 11 (cf. n. 51 above)Google Scholar. Here the writer gives a list of three pairs of words, the members of which differ by accent (⋯ρμονία):Γλα⋯κος καì уλαʋκòς καì Ξάνθος καìξανθòς καì Ξο⋯θος καì ξονθός. In each pair the first word is a proper name, the second an adjective; both the first two adjectives, уλαʋκός and ξανθός, are colour terms; the likely implication is that the writer tookξοʋθός the same way. The apparent association of ξοʋθός with ξανθός is also noteworthy in view of the confusion between the two in our text of Empedocles (cf. LSJ s.v.ξοʋθός II).

⁵⁶ Dürbeck, H., Mïinchener Studien zur Sprachwissenschaft 24 (1968), 9–33Google Scholar. There is a more recent, but no more adequate, discussion by Xanthakis-Karamanos, G., Studies in Fourth-Century Tragedy (1980), 83 fGoogle Scholar.

⁵⁷ E.g.Stanford, W. B., Phoenix23(1969), 3Google Scholarff., lrwin, E., ColourTermsinGreekPoetry(1914) 205 ffGoogle Scholar.

⁵⁸ P. Hib. 172. 114. The form λєɩρός (λєɩρɩόєɩς) referred to by LSJ occurs only on a late inscription (cited s.v.).

⁵⁹ Bacchylides seems to have been peculiarly fond of -πρʋμνος compounds: not only the rare epic form єύπρʋμνος (13. 150: II. 4. 248), but the otherwise unattested αἰολόπρʋμνος (1. 114) and λєπτόπρνος(17. 119) as well. Outside Bacchylides, ⋯μϕίπρʋμνοςand єὒπρʋμνος occur in tragedy (see LSJ s.w.).

⁶⁰ Above, p. 312. The identification with coral has been doubted: see e.g. Irwin (n. 57 above), 206–8.

⁶¹ For recent discussions of the word see West, on Hes.Op. 182Google Scholar and Athanassakis, A. N., Rh.M. 119 (1976), 4–7Google Scholar. The latter derives ⋯μοίɩος (A) from ⋯μός + * íς, crediting it with an original meaning ‘levelling' (gleichmachend), and sees in (B) an uncomprehending equation of this compound with ⋯μοīος.

⁶² In this instance it might be argued that μαλєρός is transferred (metonymically) from πʋρός toуνάθος -engendering, in consequence, that intrusive effect associated with tenor adjectives attached to vehicle nouns: see Silk (n. 11 above), 142, 144. Butμαλєρός, as an iconym, is hardly capable of such a distinct effect.

⁶³ Strictly their ‘roofs', a stock fifth-century poetic metonym. To suggest thatμαλєρά and μέλαθρα belong together should not have to be controversial, when position and parallel sound and rhythm, as well as the appearances of inflexion, force them together. In restrospect we might think the adjective has the feel of a transferred epithet or a kind of predicative force, but the notion (in LSJ, Lee ad loc., et al.) that μαλєρά is adverbial is a rationalizer's response, not a reader's or listener's. The Euripidean phrase is bracketed (evidently as an intrusive gloss) in the new OCT, but the editor, Diggle, gives no sign of supposing that the words themselves are corrupt or untragic or indeed un-Euripidean.

⁶⁴ Aspiration (like etymology) uncertain. As LSJ observes, s.fin., Aristarchus took the word to be ⋯δ-, from which fact alone one could infer that the matter was in doubt. In my citations I have simply followed the different spellings in the editions used.

⁶⁵ cf. the discussion in Leaf ad loc.

⁶⁶ It is noteworthy that epic uses phrases like πʋκν⋯…στєνάϰων (II. 18. 318, cf. 10. 9, 21. 417), and that the instance inIliad 18 occurs in the vicinity of ⋯δɩνός (see list (c)). The words occur again in close proximity at Od. 19. 516 (also listed under (c)).

⁶⁷ With ⋯δɩνώτєρον a weak neutral ground-term: Silk (n. 11 above), 19–21, 88.

⁶⁸ δάκος in classical Greek always denotes a wild or dangerous animal (cf. LSJ i). The received opinion that in Pi.P. 2 the word means ‘bite’ is arbitrary. It should be taken as ‘monster', withκακαуορɩ⋯ν a defining genitive, like (e.g.)ἄχθος…уʋναɩκ⋯ν at S.El. 1241 f.

⁶⁹ A θ⋯μα need not be, but characteristically is, an animal victim: cf. LSJ s.v.

⁷⁰ Above, p. 314.

⁷¹ In such cases the distinction between ‘meaning’ and ‘use’ is indispensable and the folly of attempting a theoretical equation of the two terms (see n. 26 above) most apparent.

⁷² The word is usefully discussed by Sideras (n. 39 above), 66 f.

⁷³ InTh. 856 (cf. LSJ s.v. 3a), whereas LSJ's classification of Pers. 976 under the ‘oar’ sense (1) is somewhat arbitrary.

⁷⁴ Mostly involving words attested in Homer, but (outside Hp.) not in pre-Hellenistic prose. A typical example is⋯λθαίνєɩν see LSJ s.v. and Brock, N. van, Recherches sur le vocabulaire medical du grec ancien (1961), 198 ffGoogle Scholar. Cf. also Leumann (n. 13 above), 308 ff.

⁷⁵ For a fuller discussion of the applications of the word, see Barratt, on E. Hipp. 1464Google Scholar.

⁷⁶ So, rightly, Lex.fr.-gr. Ep. although the article is spoiled by an arbitrary ascription of the sense ‘heftig’ to Aristophanes'⋯ρπαλέως (cf. LSJ's ‘vehemently'). Renehan, R., Greek Lexicographical Notes (1975), 41Google Scholar, offers some criticisms of LSJ's article, but appears to accept its interpretations.

⁷⁷ Frisk and Chantraine (etymological dictionaries s.v.) interpret⋯ρπαλλ as a folk-etymologized version of an earlier ⋯λπαλλ (preserved in Hsch.), related toἔλπομαɩ ˂ *wel-p-˃ Lat. volup. This etymology is both plausible and compatible with my thesis, but not a prerequisite of it.

⁷⁸ Presumably because the superior methods of communication of those ages, however unsophisticated by our standards, would have tended to standardize current usage. For a list of such words in Spenser, see McElderry, B. R. Jr., PMLA 47 (1932), 144 ffGoogle Scholar.

⁷⁹ And was still so understood by some writers as late as the seventeenth century: see the citations in OED s.v.

⁸⁰ Capgrave, : OED s.v. II. 7. aGoogle Scholar.

⁸¹ From Lytton's, King Arthur (1849)Google Scholar: OED s.v.(s.fin.).

⁸² See LSJ s.v. πρόμος and Sideras (n. 39 above), 38.

⁸³ Above, pp. 320 and 324. It is hardly accidental that Pindar should be especially active in these linguistic manoeuvres. Not only is he the most sophisticated manipulator of words among the poets: he is also particularly fond of words of ‘elastic meaning', as Stanford, W. B. puts it(Ambiguity in Greek Literature [1939], 132)Google Scholar.

⁸⁴ With an additional, but very faint, allusion to fire, by virtue of a reference to κєραυνός, parallel to τρɩόδονς, in the previous line.

⁸⁵ Where, however, the gap betweenπόντοɩο and ῤɩπάν is a papyrus gap, in view of which it is not certain thatῤɩπάν and ⋯μαɩɩμακέταν belong together.