¹ The investigation has been made as part of the work of the department under Prof. Ernest Gardner at University College, London, with whom the present writer has had the advantage of discussing the epigraphic and other evidence—as also with Mr. G. F. Hill, Dr. Head, Dr. Kenyon, and other gentlemen of the British Museum, Prof. Conway and Mr. Witton (to all of whom the thanks of the writer are due)—so that the résumé here offered is something more than a statement of the writer's. own findings.

² Such as Ethiopic forms.

³ A number of mutilated readings I have set aside as doubtful even though the context makes the meaning clear.

⁴ For this sign with another meaning, note occurring on a group of four or five ostraka all temp. Domitian which Wilcken, (Ostraka i. p. 96Google Scholar) thinks proceeded from one bureau. He notes that it occurs with proper names and may mean Πι(κῶς) a name which is common. Again Greufell, , Hunt, , and Smyly, , Tebtunis Papp., Pl. I., London 1902Google Scholar, note in index (Pap. 5. 103).

⁵ The stone stands at present near the entrance to the Reading Room.

⁶ There is a third in Roehl's, reading of 1. 15Google Scholar, but the letters are not legible in the inser.

⁷ Prof. Percy Gardner argues for Mesembria as = Midday.

⁸ So Pape, Wörterbuch, s.v. Μεσημβρία.

⁹ This or may be (Mr. Hill suggests) an actual modification of π to represent the surd labial corresponding to just as in Indian coins of Gondophares we get

which seems equally to imply a palatalized form of Gamma (passing through the intermediate consonant-y into the pure vocalic sound, as in gestern = yesterday); and here too the modified Gamma passing into comes at last to take the place of a Upsilon, even in situations where there was never a Gamma-value at all, that is it comes to be a new letter as in on coins of Kadphises II.

¹⁰ Strabo, calls it Μεσημβρία (vii. 319)Google Scholar, Herodotus, Μεσαμβρίη (iv. 93)Google Scholar; and see (vii. 108) another town on the Aegean coast of Thrace.

¹¹ Papa (Handwörterb. s.v.) actually says: auf Münzen Μεταμβριανοί but I think this is a misreading of ΜΕΤΑ for ΜΕΤΑ.

¹² ἐν τῷ μεταξὺ δὲ διαστήματι τῷ ἀπὸ Καλλά τιδος εἰς ᾿ Απολλωνίαν Βιζώνη τέ ἐστιν, ἦς κατε τόθη πολὺ μέρος ὑπὸ σεισμῶν καὶ Κρουνοὶ καὶ ῾ Οδησσός Μιλησίων ἄποικος καὶ Ναύλιχος Μεσημβριανῶν πολίχνιον —Strab. vii. 319.

¹³ See Larfeld, , Griech. Epigr. 1901Google Scholar in Muller's, Handbuch i. pp. 505, 510Google Scholarsqq. and Kirclihoff, , Studien zur Geschichte des Gr. Alph. Gülersloh, 1887, pp. 168Google Scholarsqq. See also Bergt, , Griech. Lüteraturgesch. i. 189⁹Google Scholar for definite reasons against Μ=Τ Μ=Τ Taylor, I., The Alphabet i. p. 93Google Scholar, etc., unhesitatingly abides by this genealogy, viz.:

¹⁴ Cp. for the rounding perhaps ρ of the Chaldean Alphabet. See D. H. Müller, Epigraphische Denkmäler aus Arabien; and the Tables of Semitic Alphabets.

¹⁵ Followed by Dr.Glaser, , and stated by Prof.Sayce, : The Higher Criticism and the Monuments, Oxford, 1893Google Scholar.

¹⁶ See articles by Mr.Thompson, R. C., and Mr.Hall, H. R. in Nature, Sept. 25, 1902Google Scholar, and June 26, 1902 respectively. Epitome of the results by Dr. Budge in his Hist. of Egypt, vol. vi. Intro.

¹⁷ They have not been answered, though Dr.Winckler, dealt again with the matter in the Hibbert Journal in 1904Google Scholar.

¹⁸ In the Samaritan and Rabbinic alphabets the form of capital is = 90. This is of course very remote from ancient Greek letter-forms; but it shows once more a striking analogy for the development of

¹⁹ ‘Aus der Alphabetreihe von Cäre (vgl. meine Griech. Epigraphik S. 505) bekannten Zeichen für Ssade.’

²⁰ Deecke says (Bursian, , Jahresb. Supplbd. 87 p. 27)Google Scholar: Das pränestinische ist auch venetisch, kampanisch, Sabellisch = kapenatisch eine Art s (etwa śś? s. etr. anaśśes).

But this seems rather to be ξ, cp. the old Italian inscription in the Necropolis of Este (near Venice): a, e, v, z, h = III i, k, l, m, n, s = ⊳⊲, o, p, ś, r, s, t, u, φ, χ. Bursian, ibid. p. 121 (Pauli, C., Altitalische Forschungen iii. ‘Die Veneter und ihre Schriftdenkmäler, Leipzig, 1891, p. 186)Google Scholar.

²¹ The oft-quoted saying of Herodotus i. 139 has been, oddly enough, brought in to support various Tsade theories; whereas it plainly says that San and Sigma are equivalents, and this agrees with the facts which directly associate Sigma with Shin. Why not then accept this plainer meaning? San is very improbably Tsade, while it most probably is Shin.

²² The statement that ‘the Phoenicians’ invented our alphabet is inexact. Phoenician is, according to Kautzsch's Gesenius' Hebrew Grammar, strictly only a branch of the Middle Semitic or Canaanitish, which itself is only one of three great branches using this alphabet. Similarly vague is the statement that the Greek alphabet is derived from ‘the Phoenician’ (see for example Kirchhoff, Stud. zur Gesch. des Gr. Alphab. 1887, p. 168)Google Scholar. Perhaps even the Greek signs go back to others ‘in some respect earlier in form … than any extant monument’ (E. A. Gardner, , The Early Ionic Alphabet, 1886, p. 15Google Scholar; and see passim for Uses of the term).

²³ The Hebrew names are not necessarily the Semitic names any more than the numerical values are Semitic. A petitio principii seems to me to lie in the repeated argument one meets from the place of Tsade and other letters. There is no ‘place’ of a missing Greek letter known, except by the assumption that the Hebrew alphabet order was also the Semitic. Once gain the point that was eighteenth in an assumed Semitic alphabet, such as Phoenician or that of the Moabite stone, and one has immediately a (perhaps false) premise for many deductions about in Greek.

²⁴ Quoting from Henrich's, Gr. Epigr. pp. 361–375Google Scholar.

²⁵ Every year brings iresh confirmation. See article on ‘Archaeological discoveries in Crete and Egypt,’ Nature, July 9th, 1903.

²⁶ In some forms this first stroke inclines considerably.

²⁷ Bursian, , Jahresb. Supplbd. 87Google Scholar.

²⁸ = 100 is doubtful.

²⁹ To this effect: that both the order and names of the letters, together with their numerical values have passed over from the Phoenicians to the Greeks in whose language the letters Α—ϒ are borrowed from the old Semitic; so also Old Italic Alphabets. That in default of special arithmetical figures the consonants were also used as numerical signs. The earliest traces of this usage are, however, first found on the Maccabean coins [i.e. of John Hyrcanus and his successors, from 135 B.C.].

But I note that this is no more than judgment by default, there being very little of Old Hebrew at all. The remains are, in fact:

(1) The Meša stone, 9th B.C.

(2) The Siloam-inscription, perhaps 8th B.C.

(3) Twenty seal-stones, some pre-exiiic but bearing little except proper names.

(4) The Maccabean coins, late 2nd B.C.

Dr. Lionel Barnett of the Oriental Department of the British Museum kindly remitted to me in 1903 the following statement which may be taken as the view at present accredited: ‘As the Greeks received the Semitic alphabet, already in a fixed order, and are found already using it for numerical purposes at least by 800 B.C. it is probable that the Semites also used it numerically before them.’ As this contribution strives to show, every one of these statements is at present hypothetical.

³⁰ Lenormant died without having had the assistance of some monuments which have since made possible such advance as has been made; so that his conclusions must reluctantly be put aside as out of date. The Meša stone is not considered in his article on the origin and formation of the Greek alphabet, in 1873. This was discovered by Clermont-Ganneau in 1870 and published by him in 1873 in the Revue Archéologique.

³¹ Apart from this, much value in an investigation so intricate must be attached to sound theory, so that a brief bibliography of the topic for the last twenty years may be welcome:

Wilamowitz-Moellendorff, , Homerische Untersuchungen, pp. 288 ff. 1884Google Scholar.

Gardner, E. A., ‘The Early Ionic Alphabet,’ J.H.S. 1886Google Scholar.

Kirchhoff, , Studien zur Geschichte des Griechischen Alph. 1887Google Scholar.

Szanto, E., ‘Zur Gesch. des griech. Alph.’ in the Mittheilungen (Athens), 1890Google Scholar.

Kalinka, E., ‘Eine Boiotische Alphabetvase,’ in Ath. Mitth. 1892Google Scholar.

Larfeld, W., in section Greek Epigr. of Von Müller's, Handbuch, 1891Google Scholar.

Schmid, W., ‘Zur. Gesch. d. griech. Alph.’ in Philologus, 1893Google Scholar.

Kretschmer, P., ‘Die Sekundaren Zeichen des griech. Alph.’ in Ath. Mitth. 1896Google Scholar.

Earle, M. L., ‘Supplementary signs of the Gk. Alph.’ in Am. J. Arch. 1903Google Scholar.

The last named reviews all the foregoing and adds his own views. He makes a valuable classical reference to Aristotle, Metaph. 1093Google Scholar a with Syrianus, , Schol. Arist. Metaph. p. 940bGoogle Scholar (the arguments of Archinus in commending to the Athenians the introduction of the Ionic alphabet).

³² Gardner, E. A., Inscriptions from Naukratis, 1884–1885Google Scholar.

³³ If we accept [ϑ]αλαΤης (see above).

³⁴ Originally put by Clermont-Ganneau, in Mélanges Graux, Paris, 1884, pp. 415–460Google Scholar.

³⁵ See Müller's, Handbuch, vol. i. pp. 505–511Google Scholar.

³⁶ Sadé he does not explain, as perhaps it did not pass into Greek.

³⁷ As shown above, pp. 344 f.

³⁸ In the Hebrew order = 90.

³⁹ Tsade, Larfeld says, p. 149, was a living letter in Corinth still in the sixth century, at Melos in the second half of the sixth, and at Sikyon even in the fifth.

⁴⁰ So Kirchhoff, , Stud. z. G. d. g. A. pp. 134Google Scholarsq.

⁴¹ In criticism of Gundermann's, (worthless) Die Zahlzeichen in Ephemeris für Semit. Epigraph. p. 106Google Scholar.