^{page 220 note 1} Naukratis, Part I, published by order of the Committee of the Egypt Exploration Fund.

^{page 221 note 1} Petrie, op. cit. pp. 5, 19.

^{page 221 note 2} Op. cit. pp. 19, 20. They occur at level 230 in. and level 220 in. is 650 B.C., level 250 in. 600 B.C. approximately.

^{page 221 note 3} Op. cit. pp. 54–61.

^{page 223 note 1} I have said nothing as to direction of writing. The earliest inscriptions of course run, as a rule, from right to left; but in one or two the direction is reversed. It is obvious that such accidental reversion is just what one might expect in the use of a new and unfamiliar invention, as in a child just learning to print letters. One or two instances of βουστροφηδόν are distinctly later.

^{page 225 note 1} The names of the sibilants are of course a difficulty: but as these discoveries throw no light on them, they are outside the scope of the present paper. They at least can hardly be adduced as an argument against the view here adopted.

^{page 226 note 1} Daremberg and Saglio: art. ‘Alphabetum.’

^{page 227 note 1} Or perhaps it would be better to substitute for the expressions ‘earlier’ and ‘later’ in this discussion, ‘closer to’ and ‘more remote from the Phoenician prototypes.’

^{page 228 note 1} The Alphabet, II. 93.

^{page 229 note 1} Op. cit. II. 68.

^{page 229 note 2} ibid. p. 63.

^{page 229 note 3} Such, for instance, as those of M. Clermont Ganneau, deriving Ϝ from ε, χ from τ, φ from ψ from

^{page 230 note 1} See below. I think the earlier date now excessively improbable, but insert it here lest I seem to assume what some epigraphists do not grant.

^{page 230 note 2} Two vases found in Rhodian tombs are generally quoted; one with the Argive alphabet, another with a semicircular γ. But there is no necessity for them to represent the local alphabet at all. The evidence of coins only shows that the Ionic alphabet was in use in Rhodes in the fifth century.

^{page 231 note 1} I am indebted to Prof. Ridgeway for the suggestion of another possible explanation. The fullest form of the Ionic alphabet, with all the non-Phoenician signs, may have been in the earliest period restricted to Naucratis, while in Asia Minor, perhaps even in Miletus itself, the less complete form may still have remained in use. This suggestion does not affect the importance of the early alphabet found at Naucratis, since in any case it afterwards became universal.

^{page 231 note 2} The difference between the alphabets used upon the two is of course due to the fact that the one was dedicated by a Milesian, the other by a Rhodian; and each used his own local characters.

^{page 232 note 1} Strabo's words (XVII. 1, 18) will bear this meaning, if we do not assume the Inaros he mentions to be identical with the ‘Inaros son of Psammetichus’ who lived two centuries later. May not the name Inaros have come in here merely by a confusion of associations?

^{page 232 note 2} XV. 675. The date mentioned is hardly possible. It is against tradition, probability, and the evidence of excavation that Naucratis existed before the time of Psammetichus I.

^{page 232 note 3} For instance, the Therean κ and π, the Corinthian β and ε, the Chalcidian λ, the Naucratite ν and σ. It is impossible to refer these all to a primitive ‘Cadmean’ alphabet; unless by Cadmean we mean Phoenician. For the Corinthian β and ε see Taylor, , Alph. I. 103, 115.Google Scholar May we not infer similar forms in Phoenician, just as, were all Chalcidian, Boeotian, and early Attic inscriptions lost, we might infer L from Latin?

^{page 233 note 1} See his popular account in Baumeister, , Denkmäler, I. 51.Google Scholar He adds υ, but as that letter is not distinctively Ionic, it is more probably an independent Greek invention: besides υ is just the one letter whose later Greek forms are nearer to the Cypriote than its earlier.

^{page 233 note 2} See Deecke's table in Collitz, Griechische Dialekt-Inschriften I. One form of Κο identical with the later Greek Ω is given there, but with a query.

^{page 233 note 3} See the last form given on the table, page 221.

^{page 234 note 1} Of course they might, from this indication, have been borrowed earlier; but not later.

^{page 235 note 1} Or is the four-stroke form derived from shin, the three-stroke one from tsade? A suggestion confirmed by the form occupying the place of tsade in the abecedarium of Caere.

^{page 235 note 2} At Argos, for instance, the three-stroke form rests only on the evidence of Fourmont, the notorious forger of inscriptions. In the very rare cases when occurs earlier than Σ, both may have been in use together from the first.

^{page 235 note 3} The true relation of the two is pointed out by Mommsen, , Unterit. Dial. 5.Google Scholar

^{page 236 note 1} A dot was placed in the middle for ω (Thera), or half the symbol was used for ο, the whole for ω (Melos).

^{page 236 note 2} ξ must also have been a common inheritance, but it was not at first adapted to its later use.

^{page 236 note 3} Unless of course, as in Arabic and Sanskrit, the alphabet has been entirely rearranged on new principles.

^{page 237 note 1} It is true that this form is not known in Phoenician; but the various forms derived from it seem to imply that such must have been the original form.

^{page 237 note 2} So Taylor, II. 77.

^{page 237 note 3} or φσ.

^{page 238 note 1} Mr. E. S. Roberts, to whom I am indebted for kind permission to make use of the materials he has collected, suggests that there may be here also a similarity of sound, as is indicated by the fluctuations of early usage. It is hard to find a more certain instance.