¹ I must also thank archaeological friends in Oxford, notably Dr. Paul Jacobsthal, for discussing the object with me.

² For Zeus Naios, see Farnell, , Cults I, pp. 38 ff.Google Scholar; Cook, A. B., CR XVII (1903), pp. 178–86.Google Scholar The same epiklesis is also found at Athens (IG II², 4707; Acrop. Inv. 4887 (cf. lemma ad IG II², 4643); IG II², 4643; 5113; I², 274, lines [11], [20], 197, 225, etc.; of these only 4707 is a dedication to Zeus Naios; 4643 is a dedication to Zeus Naios and Dione, his consort, and the remainder refer only to Dione. The cult was evidently established by the fifth century, and continued into the Roman period), and at Delos (Bekker, , Anee. Grace. I, p. 283Google Scholar, s.v. no documentary corroboration of this exists, so the cult can have been only of minor importance—if indeed the entry is not an error); in both places it doubtless derived from the Dodonaean cult; cf. Cook, op. cit. p. 186.

³ Carapanos, C., Dodone et ses Ruines (Paris, 1878), 2 vols.Google ScholarTexte and Planches. I refer to this work as C; Roman numerals refer to the plates. The other main collection of material is that published by Euangelides, D., Ἠπειρωτικὰ χρονικά X (1935), pp. 192–264Google Scholar, with πίννκες 1–28. I have not seen Friederich, J., Dodonaica, 1935Google Scholar, and know nothing of its contents.

⁴ See, for example, C. xxiii, 1, 3, 4, 6; xxiv, 1, 4, 5: all dedications; xxvii ff.: all documentary texts.

⁵ An epigraphical corpus of this material is badly needed. There is a useful list of inscriptions in bronze in Robert, , Coll. Frochner i, pp. 47–8.Google Scholar

⁶ Alexander, , SGDI, 1334–35Google Scholar; 1337; 1346. Neoptolemus, 1336. For the reproductions in C. see notes 12–13.

⁷ The most important discussions of the problem are those of Nilsson, , Lund Årsskrift, N. F. Afd. 1, Bd. 6 (1910), no. 4Google Scholar, ‘Studien zur Geschichte des Alten Epeiros’, pp. 59 ff.; Beloch, , GG III, 2², pp. 181 ff.Google Scholar; Cross, G. N., Epirus, 1932, pp. 109–14Google Scholar (a useful summary). The problem hinges mainly on the relationship of the mentioned in the documents which refer to King Alexander, to the symmachy of the Epirotes, which occurs in the decree which mentions King Neoptolemus. Nilsson regarded the Molossian κοινόν as the sovereign community of the Epirots which preceded the Epirot symmachy, primarily on the grounds of lettering (with which I agree, see below, note 13), and thus made Alexander the first Molossian king of that name. Beloch held that the was simply a unit of the Epirot symmachy, and contemporary with it. It is hardly possible to decide between these two alternatives, although I favour the earlier date on epigraphical grounds (though this does not mean that Beloch's explanation may not be factually correct). The argument of Cross, pp. 110–11, that the Aetolians were the first κοινόν in the west, originating ca. 314 B.C., and that the appearance of the must be later than this, is untenable in fact (though it could be right in principle; but cf. Nilsson, op. cit. p. 62, note 1), now that we know that the Aetolian koinon was already in existence by 367 B.C. (Tod, , GHI, 137Google Scholar). The constitutional question needs re-examination in the light of: (a) the proxeny-decree issued by with no reference to either king or Molossian προστάτης, published by Euangelides, op. cit. p. 245, no. 1, the letter-forms of which point to a date ca. 300 B.C. (very similar to the Neoptolemus-inscription and to our dedication): (b) the decree of the dated by the Molossian prostates, and apparently of the later third or second century, published ibid. p. 261, no. 1.

⁸ See Nilsson, op. cit. pp. 59–60; Tarn, , Antig. Gonat. pp. 55 ff.Google Scholar; Lenk, RE, s.v. Molossi, col. 22. Other references in Cross, op. cit. pp. 124 ff. Beloch, op. cit. pp. 181 ff., does not make it clear whether he accepts the identification with Alexander I or not. In ibid. I, p. 472 he seems to.

⁹ The later date is favoured by Klotzsch, , Epirot. Gesch. pp. 53, 173–4Google Scholar; Cross, loc. cit. For Alexander II see the dis cussion in Cross, pp. 124 ff.

¹⁰ For Neoptolemus see H. Berve, RE, s.v. Neoptolemos (4), cols. 2463–4. (Neoptolemus I was the son of Alcetas (see Berve, ibid.s.v. Neoptolemos (3)), and does not therefore come into question.) Cross, op. cit. pp. 106 ff., argues that Neoptolemus was not the son of Alexander I, since if he had been it must be presumed that he reigned as an independent monarch, while Arrian (FGrH, 156, F1 § 7) and Dexippus (ibid. 100, F8 (1), § (3)) ‘both speak of Epirus as forming part of Antipatros' government after Alexander's death’. In fact, however, both Arrian and Dexippus refer to Epirus in terms of specific boundaries (Arrian says while Dexippus has ), and though the implications of these definitions for political history cannot, in the absence of other evidence, be determined, they clearly admit the possibility that Epirus was only in part a Macedonian protectorate, while the rest of the country remained, as it had been in the time of Alexander the Great, independent.

¹¹ Compare, for example, the bronze plaques from Lusoi, , JÖAI, IV (1901), pp. 64 ff.Google Scholar nos. 4–7 (= IG V, 2, 392–395, and tab. v, 2; Kern, , Inscr. Gr. pl. 21Google Scholar).

¹² Facs. of SGDI, 1336: C. xxvii, 1.

¹³ Facs. of SGDI, 1334 in AEMÖ V (1881), p. 132; ibid. 1335, C. xxvii, 3; ibid. 1337, C. xxxii, 5; ibid. 1346; C. xxxi, 1. It should be noted that the first two of these four inscriptions are not in dotted style, and the normal engraved stroke lends itself less readily to slight curves. In spite of this I feel fairly confident that they are of the fourth century, and not the third, as would be necessary if Alexander II was the king in question.

¹⁴ C. xxiii, 2; cf. ibid.Texte, p. 40: ‘L'emploi des lettres longues η et ω indique que Terpsiclès est un Ionien’. A dialect form of the present tense occurs in SGDI, 1369 = C. xxv, 1:

¹⁵ C. xxiii, 3, 5–6.

¹⁶ Thus Δί occurs in C. xxiii, 1–2. The first probably of the fifth century, the second perhaps even earlier. C. xxvii, 1, where it also occurs, is probably of the sixth century. The latest instance with it is C. xxiv, 4, a double dedication (both clearly inscribed at the same time): (sic; as often). This piece appears to be slightly earlier than our snake (note the sigma with widely divergent hastae). Some of the early pieces (C. xxiii, 3; xxiv, 5) and the majority of the Hellenistic ones (xxiii, 5–6; xxiv, 1–6; xxv, 2) have Διί. Both Δί and Διί occur in the oracular texts: Δί: C. xxxiv, 3 = SGDI, 1557; C. xxxvii, 3; Euangelides, op. cit. p. 255, no. 13: Διί: C. xxxv, 2; xxxvi, 5 = SGDI, 1564. Hoffmann's interpretation of the form Διεί in the oracular text, SGDI, 1582 = C. xxxiv, 3, as a variant of is accepted by Bechtel, , Gr. Dial. ii, pp. 78 ff.Google Scholar, but, as Salonius, , De dialectis Epirotarum, etc., p. 148Google Scholar, § 3, pointed out, the presence in the same text of ἰς = εις indicates that this is probably only an etacism. (So also, I suspect, in the Corcyraean, dedication, IG ix, 1, 718Google Scholar, where Hoffmann, ibid., claims Διεί is This is a dedication in the koine to Zeus Hypsistos.)

¹⁷ For this see Meisterhans-Schwyzer, , Gram. der alt. Inschr. p. 111.Google Scholar

¹⁸ C. xxxvii, 1 = SGDI, 1573 = Michel, 846 = Syll. ³ 1164. Blass, , Rh. Mus., XXXIV 1879, p. 160Google Scholar, considered that the mu had been omitted accidentally. But the omission is not uncommon in this position.

¹⁹ For instances of the normal ethnic Στράτιθς see e.g. Syll. ³ index, s.v. The ktctikon Στρατικός occurs as an ethnic in IG, IX, i², 3A, line 25.

²⁰ For example: (Syll. ³ 380; 422, line 8; 500, line 9), (ibid. 417, line 11), (IG VII, 287; 2467a); (IG VII, 1760), (ibid. 420), (ibid. 361), (IG V (1), 948); (Syll. ³ 417, line 5), (IG VII, 12); (Syll. ³ 492, line 30), (IG VII, 2848), Μακεδὼν ἐκ ⊙ετταλονίκης (Syll. ³ 492, lines 34), (ibid. line 37), (IG II², 8404), (IG VII, 415); (IG ii², 7973); (IG II², 10293), (ibid. 10292), (ibid. 10291). Examples of this usage could, of course, be multiplied; cf. Reusch, , Gramm. der delph. Inschr. i, pp. 268–9.Google Scholar

²¹ Instances are: (Syll. ³ 314, vi, line 28); (ibid. 444A; cf. Reusch, op. cit. p. 269, § 4); (IG II², 8826; here, however, the use of the normal ethnic to denote a cleruch would perhaps be incorrect); ΚοσμΙα ἐκ Κελσινῶν (IG II², 9009); (IG II², 9334. Here the form may be influenced by the more usual for which see previous note.) There is nothing to suggest that the individual thus designated possessed a lower status (e.g. that of a metic) than an individual designated by the true ethnicum.

²² The ethnic occurs only in C. xxiii, 2, and C. xxiv, 3,

²³ E.g. C. xxiii, 2: C. xxiii, 5: xxiii, 6: and others.