¹ Buck-Petersen, , Reverse Index, pp. 636–8Google Scholar; Schwyzer, , Gr. Gram, i.497 f.Google Scholar; Chantraine, , Études sur le vocabulaire grec (1956), pp. 97–171, esp. pp. 103–5, 150Google Scholar, on the distinction between ethnika and kiētika and the precise function of the latter.

² Chantraine, , op. cit., p. 120Google Scholar, draws attention to this peculiarity, though he does not refer to Grammont's explanation of it discussed below.

³ Grammont, M., Traité de phonétique³ (1946), pp. 331–;7Google Scholar; Schwyzer, , Gr. Gram. i.262–5.Google Scholar

⁴ Even this mention has disappeared from the 9th edn. (1940), though vestiges of it remain at the end of the entry under , contr. , ‘red’. Lexicographers and editors in the nineteenth century were much influenced by the dismissal of in the 1822 edition of Stephanus, Thesaurus Graecae Linguae, elaborated by Dindorf, W. in the later edition viii (1865), 985 f.Google Scholar; but the inadequacy of Dindorf s argument will be apparent when his range of examples is compared with the list given here.

⁵ Cf. also the variants at 1.10.54 (Mras p. 54).

⁶ Cf. also the variants at 3.4.2 (Kramer, i.243).Google Scholar

⁷ For this date see the Introduction to Mras's edition of Eusebius, , Praep. Evang., pp. xiii–xvGoogle Scholar. It is unfortunate that P. Oxy. 2.227 (saec. A.D. i), with a passage from Xen. Oec, does not begin until 8.17, and that P. Oxy. 13. 1619 (saec. i–ii), which contains a text of Hdt. 3.26–72, has a gap in the relevant place (3.37).

⁸ Morpurgo, , Myc. Graec. Lex., pp. 28 f.Google Scholar; Ventris, and Chadwick, , Documents² (1973), pp. 91, 532.Google Scholar

⁹ Grammont, , Traité de phonétique, p. 335 n.l.Google Scholar

¹⁰ Such a back-formation seems more plausible than the alternative explanation that attested in Aristophanes might conceal the existence of the name as early as the fifth century B.C., of which there is no trace in the texts of writers from that period.

¹¹ Hamm, E.-M., Grammatik zu Sappho und Alkaios, p. 63Google Scholar; Stolz-Leumann, , hat. Laui und Formenlehre, p. 243Google Scholar, and the Prolegomena to Bailey, C.'s edition of Lucretius, p. 136.Google Scholar

¹² So also in Greek: see Schwyzer, , Gr.Gram, i.262 f., though the examples do include some possible exceptions, like Cretan veóra for veórāra.Google Scholar

¹³ A parallel case is Welsh erledigaetb ‘persecution’ for erl(id)edigaeth from the verb erlid ‘persecute’ and the two suffixes-edig, which forms verbal adjectives, and -aeth, forming abstract nouns (Jones, J. Morris, Welsh Grammar (1913), pp. 55, 230, 396)Google Scholar. A variant erlidigaeth, i.e. erli(de)digaetl is attested (cf. Thomas, R.J. (ed.) Geiriadur Prifysgol Cymru i (1950–1967), 1234 f.)Google Scholar, but in the form generally preferred it appears that analogy with words having the same suffix has caused -ed- to be retained (contrary to Sturtevant's formulation) at the expense of the -id- in the verbal stem. Enough remains of erlid in erledigaetb for the word to be readily intelligible, and the same would be true of beside .

¹⁴ Diet. étym. de la langue latine s.v. Poenus.

¹⁵ Diet. étym. de la langue grecque s.v. the suppositious goes back to Saalfeld, G.A., Tensaurus Italograecus (1884), p. 907Google Scholar, following Hehn, V., Kulturpflanzen und Haustbiere² (1874), p. 518.Google Scholar

¹⁶ CIL I.ii, 25; the inscription as it survives is thought to have been worked over at a later date, perhaps in the time of Augustus (cf. Warmington, E.H., Remains of Old Latin iv. 128 f. with references)Google Scholar. It includes some forms which are false archaisms, but there are no grounds for suspecting Poenieas in line 8. Clearly Poenieus came into Latin before the regular development of oe to ū (on which see Buck, C.D., Comparative Grammar, p. 87Google Scholar), but neither this feature nor the absence of aspiration (see Sturtevant, , Pronunciation of Greek and Latin, p. 157Google Scholar) enables us to determine a more precise date for the form's origin.

¹⁷ ‘landede sproghistoriske bidrag’, Nordisk Tidsskrift for Filologi 5 (1916), 100–2Google Scholar, rejecting the non-existent Jensen however would regard Poenicus as a modified form of phoimiks (), with the possibility that the formation could have been influenced by he does not consider whether might itself have existed in Greek.

¹⁸ Similar cases may seem to exist, but in each of them special factors are present which provide alternative explanations: ancora beside , azymus beside beside acc. , cunila? (Plaut. Trin. 935 Ritschl) beside (for references to discussions see the etymological dictionaries, esp. Ernout and Meillet⁴ (1967), s.w.). An interesting example is schěma (Plaut.) beside which Wackernagel, (Vorl. Über Syntax ii.45) and others explain as derived from a Hellenistic form . attested only in Hesychius; this would provide an approximate parallel to the derivation of Punicus from a Hellenistic put forward here.Google Scholar

¹⁹ G.P., and Edwards, R.B., ‘Red letters and Phoenician writing’, Kadmos 13 (1974), 48–57Google Scholar, esp. 51, and ‘The meaning and etymology of ’ forthcoming in Kadmos 16 (1977)Google Scholar, which includes a critical consideration of the theory recently advanced by Beattie, A.J. in ‘Some notes on the Spensitheos decree’ (Kadmos 14 (1975), 8–47)CrossRefGoogle Scholar that is to be connected with rather than .

²⁰ I wish to thank the editors and their referee for helpful criticisms of the first version of this article. The above list of examples of is now given in a much abbreviated form, but I should be glad to supply fuller information about any of the instances to readers who lack access to the editions cited.