expected to be cast upon the Elean dialect by the researches at Olympia, which are bringing to light almost every week forms of , great philological interest.
The Arcadian dialect rests almost wholly upon inscriptions, of which the most important is one found at Piali near Tegea in 1859, edited by Bergk, and afterwards by Michaelis, with valuable notes by Gurtius. The very careful examination of this dialect by Schrader (Curtius s Studien, x. 273-280) shows that it has more points in com mon with Doric than with ^Eolic, indeed that there is no single point in which it agrees with all the dialects of the latter, where it does not also agree with the former. Its agreement with Lesbian especially is only on minute points, which seem to be of independ ent origin. Hence its ^Eolic character may be definite^ given up. Among the more interesting phenomena which it presents, we may notice -au for the gen. sing. masc. of a-nouns, -01 as the dative (or possibly locative) sing, of o-nouns, lv used for els and ev, -rot as the inflexion of the 3d sing, middle (e.g., yivjroi, 8aTo<)and -r^.tvos as the ending of the participle of verbs in -e o> (dSi/cT^uei/os ; cf. a8a- Tjfj.fi cf = KaTa.8ir]ovfj.vc l i iii the Elean treaty ; Lesb. KaArj/xeyos, &c.).
The Cyprian dialect may be mentioned here ; for the results of its examination entirely confirm the statements of Herodotus (vii. 90) and Pausanias (viii. 5, 2) that Arcadians were among the colonists of Cyprus. This was first asserted by Bergk on the strength of a few glosses ; but recently the inscriptions have been deciphered by Dr Birch, followed with more complete success by Brandis, Schmidt, and Deecke and Siegismund. They are not written in Greek characters, but in an alphabet of their own, which is syllabic in its character, i.e., each sign represents a consonant followed by a vowel. Of these signs there are 56 as yet identified ; there is no distinction between tenues, medials, and aspirates, nor is there any mark of rough or smooth breathing ; the signs therefore stand for a, e, i, o, y, ka, ke, ki, ko, ky, pa, &c., ta, &c., ma, &c. , na, &c. , ra, &c., la, &c. , sa, &c. , and fa, &c. The number is made up by ja, je, ji (jo and jy not having yet been discovered), sse, za, and zo. If a word ends in a consonant, the sign of that consonant when followed by e is used ; but an article or a preposition is often treated as coalescing with its noun. When two consonants come together, the first is denoted by the sign of that syllable which it makes either with the vowel attached to the second consonant (e.g. , po-to-li-ne = ir-roiv) or with the preceding vowel (e.g., a-ra-ky- ro = apyvp(ji). A nasal is always omitted before an explosive (a-to- ro-po-sc = &vQp<inros}. Cyprian agrees with Arcadian in the geni tive in -an, in airv for air6 (sometimes followed in both by the dative), in the preposition lv (often with ace.), and in many less important points.
Doric.—The Dorian dialect was divided by Ahrens, following the Greek grammarians, into two main groups (1) the severer Doric, (2) the milder, the one being more closely connected with ^Eolic, the other with Ionic. To the former belonged the speech of Laconia, Crete, Gyrene, and the Greek colonies in Italy; to the latter the lan guage of Argolis, Messenia, Megara, and northern Greece, and the colonies of Asia Minor and Sicily. The basis of this distinction is the use of o> and TJ in the severer as against ov and i in the milder dialect. But the division can hardly be maintained in practice, and hence it is abandoned by most modern scholars. The northern Doric, for instance, which is ascribed by Ahrens to the second di vision, has been shown by Merzdorf (Sprachtvissensch. Abhandl. &c. Leipsic, 1874, pp. 23-42) to form a bridge between ^Eolic and Doric. Again, while we find ov in use at Thera, at Gyrene, a colony of Thera, u is retained ; hence this cannot point to a deep division. We may notice first the authorities for the particular dialects, and then the characteristics of Doric generally.
The Laconian dialect is known from few and unimportant in- criptions, from the fragments of Alcman, which, however, are in a language much modified for poetic purposes, and from the specimens in the Lysixtrata of Aristophanes and in other Attic comedies. There are also a large number of Laeonian glosses in Hesychius, and Thucydides (v. 77) gives a treaty in the Spartan dialect. Our knowledge is largely supplemented by the famous tables of Heraclea, a colony of Tarentum, which itself was founded by Sparta. These were found in the bed of the river Cavone in 1732 and 1735, and are now partly in the Museo Borbonico at Naples and partly in the British Museum.
From Crete there are numerous important inscriptions, chiefly treaties between various towns. It is curious that some of the most valuable of these were found in the ruins of the splendid temple of Dionysus in the island of Teos ; this temple enjoyed the rights of an asylum, and the inscriptions are mainly treaties acknowledging these rights on the part of various Cretan cities. They contain some highly interesting archaisms of form.
For Thera there is an important inscription containing the will of a wealthy lady Epicteta (C. I. G., 2448); for its more famous colony, Gyrene, there are only brief and fragmentary records.
The Argolic dialect appears on a very ancient helmet found at Olympia (C. I. G., 29) and on an inscription very recently dug up at the same place, as well as on several others of less importance.
From Messenia there is a long and very interesting inscription found at Andania, dealing with the cultus of certain deities ; it is of comparatively late date (probably 93 B.C.) and in a much modified Doric, but it contains some striking forms.
The Corinthian dialect is learnt mainly from inscriptions at its colonies of Corcyra and Syracuse, both of which cities supply some very ancient and valuable records. In the same way the Doric of Megara is preserved most fully on Byzantine inscriptions. For this we have also the Megarian in the Acharnians of Aristophanes.
For the Locrian dialect Ahrens had but few and fragmentary in scriptions and no literary evidence ; recently a bronae tablet con taining a treaty between Chaleion and (Eantheia (of the 4th century B.C.) has been dug up at the latter place; and also a tablet contain ing the regulations for founding a colony at Naupactus (cf. Curt., Stud., ii. 441-449, iii. 205-279). These throw much new light on. the dialect, and enable us to set it down with confidence as a link between Doric and ^Eolic.
The general character of the Doric dialect was that of a slow, deliberate, and emphatic speech ; it is the speech of the warrior and the ruler, not of the orator or merchant. The Tra.reia<rfj.6s, which the ancient authorities ascribe to the Dorians, is not distinctive of them, but was shared by the Boeotians and other .(Eoliaus ; it is to be re garded rather as a mark of an earlier stage of the language, which was retained like many other similar characteristics by the Dorians much more extensively than by contemporary lonians. It is quite the exception for any Doric characteristic to be of recent origin. A natural hypothesis finds in the full and broad sounds of the dialect of these " men of the mountain-forests " signs of the chest strength ened by mountain air and mountain life. To pass to details : In accentuation Doric showed no inclination to the barytone pro nunciation of Lesbos ; on the contrary, it has more oxytone forms even than Attic. In many words the Doric accent is of especial in terest as bearing valuable testimony to the origin of the inflexions ; we find not only ayyfoi, avdpunroi, and rvTn-0/j.fvot, but also eKtyov, e vcrav, ircu Sey, TTTCO/COS-, and d/iTreAoy (ace. plur. ), these forms all pointing back to a time when the final syllable was long, and thus demanding from philology an account of this length. In vowels a short a is often retained where Attic has e (lapos, rpd<f><a, T-paxco) or o^cucari cfcocrt): v in Laeonian became ov, but probably only as an indication that the earlier pronunciation of the vowel was retained, when in ordinary Greek it had sunk into ii. Wherever TJ in Ionic has come from an earlier d, Doric retains a, but where it has ori ginated in e, 77 is retained as in Lesbio-^Eolic (irar-fip, Bceot. irareip) ; it is also retained in augments (^px^^av), and as a contraction for ae (eVi/c?;). On the other hand ao and aw contract into a ( ArpttSa, yeAaV). The contractions of eo, eo, vary much in different dialects. The severer Doric gives t for ei and CD for ov. $s for efs, 3/j.fv for efffj-tf, &yiao~a, Maj<ra, fyfi] r}6(oi ri = t^fiKt}6<i)0 i, jSaiAa (Lesb- &6a), Kwpos, &c. A noteworthy phenomenon is presented by the shortening of long final syllables, almost exclusively where the length is due to compensatory lengthening in the place of a lost consonant (Sri/jLords, Kara rbs v6fj.os, irpd^ds, Tr6s [ = 7ro8-s], Ae -yes, TJKTEJ , re/cej ); of all forms of the dialect the Cretan especially favoured this. Of the consonants, the digamma was retained longer by the Doric than by any other dialect, but we find it gradually disappearing. It is used in the old Laeonian, Argolic, Corinthian, and Corcyrseaii inscriptions, but not in the Cretan, with the exception of the proper name /ct|ioi ; on the Heraclean tables it is very common, but there are some strange exceptions, as oiicid, fpydofj.ai, and &<pepy<o ; some have held that it is there wrongly inserted in /e|, but this is really a valuable confirmation of the labial spirant to which other lan guages also bear witness. The digamma is often changed to (as in Elean), but never before p, as in Molic; whether it ever actually passed into y, or whether the numerous forms which give this in the place of an earlier digamma are all due to the mistakes of copyists, is a question still xmder discussion (Curtius s Princ., ii. 229 T). T is constantly retained, where the lonians have weakened it into o", especially in TI : (pa.ri^=<prio"l ; so TvirTovTt =TVTrTovo~i, Ti6evTi = Te6etcri , cf. TrAoYios, irAourios, Sciris, SeAivownoi.
Three changes characteristic of Laeonian came in at a comparatively late date ; for they do not appear in the Heraclean tables, and consequently they must be later than the foundation of Tarentum. (1)0 becomes o- ; this is very common in the Spartan of the Lysistrata e.g., <r<=Ae<, ffiy^v ( = 6iyiiv o~i6s ( = 0e<k) ; cf. TW ffiu> 0-vfj.a.Tos (Time. v. 77), ffe~tos av-fip (Ar., Eth. Nic., vii. 1). (2) Final s becomes p ; this is still later, and does not appear in Aristophanes, but is very common in the more recent inscriptions. (3) Medial a between vowels becomes ; this is found in Aristophanes (Mia, tract, &c.) and in later inscriptions (nooiSavt), but not in Alcman.
The traditional change of y into S is denied by Ahrens and Curtius, who altogether reject, with very good reason, the asserted identity of Sa and 7?}. The appearance of | in the future and compound aorist of verbs in -&> (e.g., ooKifj.d&vTi, f^fpi^av, &c. in the Tabb. Hcrad., jutxn |ai in Aristoph. , Lysist.) has been rightly explained by Curtius (Princ., ii. 248) as a hardening of the original spirant y (j) before the er, the only possible alternative to its complete loss, which we find in the ordinary Greek SoKifido-w. The change of into o-S, ascribed by the grammarians to Doric, is more