2 South-west Italy and Sicily. They did however borrow the symbol x at some later date, as appears from its position at the end of the Roman alphabet, but before the period when έ was adopted in Greece, for there is no evidence that x in Latin was ever either an aspirate or a breath. The change of the value of the symbol in Latin is strange, and cannot be explained with certainty: Prof. Key1 conjectures plausibly that xs, which is found in the oldest Latin inscriptions may have been a copy of the same combination which, as before observed, was found in Greece instead of Ko: then, since x (the guttural aspirate) was never found in Latin except in this particular combination, it naturally suggested the following s; which thus became superfluous and was dropped. The later Greek symbol E does not agree with the Hebrew Samech. (in the place of which it stands) either in form or in power: I would again hazard the conjecture that it may be a relic of an older alphabet, put into the place of that first of the three sibilants, which were not all required by the Greeks. CH. VII. The Dorians had two symbols for s, Σ and M, which The sibiin form come nearest to the Phoenician equivalents of lants. the Hebrew Tsadi and Shin. The Umbrian also and Etrurian have two symbols: the Oscan and Latin dropped the second. The M, like Shin, was perhaps equivalent to sh, a sound not required by the Greeks (at least originally), nor by the Latins. The form and place of San correspond to shin, the name looks more like Zain, as Dr. Donaldson holds3. The history indeed both of the forms and names of the sibilants is obscure. Sigma was originally si san is preserved in oaupópas, and in the name Sampi of the symbol, which was also an easier way of writing sp, just as (which survived) was a simpler form of ps. 1 Ibid. p. 108. 2 E. g. Maxsumus. 3 See his account in the New Cratylus, p. 176, which differs to some extent from that here given. CH. VII. Zeta. Корра. The Latin F and H. Latin borrowed symbols. Z corresponds in place and moderately in form to Zain (sounded dz). The sound may have varied in Greek, where it always represents some phonetic corruption, as will appear in its place. It occurs (as weak s) in Oscan and perhaps in old Latin: this will be discussed below. Koppa was retained by the Dorians, and universally in the table of numerals for 90. It passed with the rest of the Doric alphabet into Latin, as Q. The Latin F and H stand for lost aspirates: the F representing in form the F (i.e. w), for which the Latins regarded v as a sufficient representative: and the H in its old force. The symbols Y and Z in classical Latin were borrowed. from the Greek, to express those Greek sounds which had no exact representatives in Latin. This new z differed from the old z mentioned above, i. e. from the French z: being the Greek, whatever that was, dz or the French or English j. I shall take the four heads of vowel-change in a slightly different order from that given in my sketch above, as follows: Substitution, Assimilation, Dissimilation, Loss. This will be found most convenient for the Latin. 1. A= a, e, 0. I. SUBSTITUTION. 1. Splitting of the A-sound. I shall commence with the most universal and most important change, that of Indo-European A into ă, ě, ŏ in Graeco-Italian, and ā into ā, ē, ō. I will give such evidence as can be given for the sound of these letters at the end of the section. There seems no reason to believe that this change was anything but phonetic. Short e and o are not raised powers of any of the simple vowels; and a, with which they are certainly connected, is heavier than either of them, the order in respect of weight being a, o, e; which is preserved in the conservative Latin. In Greek indeed, as I have mentioned above, the three vowels seem to be used in the a-scale, a being the radical; and e and o the first and second steps respectively. That o is heavier than e is best seen by comparing the same formations in Sanskrit and in Greek; thus jajāna = yéyova, where the short a of the Sanskrit is equivalent to e, and long a to o; similarly bhărāmi = pépw, but bhāra = φόρος. It was to be expected à priori that the strongest and by far the commonest vowel of the original speech would be more corrupted in use than any other. Instances of its variation are common enough within our own island. The full sound of the a is more commonly retained in Scotland, whilst in England it has been thinned down to the e-sound, though spelt generally as a, and sometimes. changed to o; in which case the Scotch frequently shews the e-sound, spelt as ae or ai. Thus the original na is still found in Scotland, but it is more commonly nae (e-sound); whilst as in England we have the same sound, nay, beside no. Scotch awa is English away. Twã is two, snaw is snow; but baith (e-sound again) is both, laith is loth, gae is go. In none of these changes is there anything dynamic: they are purely phonetic variations. CH. VII. Europe. These instances would shew that this variation was not Found confined to the Graeco-Italian. It extended over all the throughout European peoples; no doubt very gradually. I have already said it is not found in Sanskrit, which has no ě or o, and its ē, ō are equivalents of ai, au. It has therefore been inferred that the tendency was not felt till after the separation of the East from the West'. This seems at variance with the theory, that the North European families parted off from the Eastern, before the nations of Southern Europe had left them; since this change is certainly found in members of the Teutonic and Sclavonic groups. How is this difficulty to be solved? Was there some connec1 By Curtius, Gr. Et. p. 88. CH. VII. History of the change from A to e. 1 し tion again formed between the Germans and the Greeks after their first parting? Nothing would seem more improbable. Was this vowel-change the result of independent phonetic action in the separated peoples? This again is hard to believe, when we see the simple words and roots which exhibit the same variation: but it is not impossible. The vowels e and o are produced by the first change in Greek and Latin, and afterward sink to and v: but in North European nations it is generally e which is developed from i, and o from u. The ŏ of the Sclavonic must have been developed after the separation of that language from the Lithuanian. The Keltic agrees in the tendency of its vowel change mainly with the Teutonic: yet a in it could pass directly into e and o1. These facts shew that different languages could arrive at the same. sounds in different ways. Still I think it not impossible that a tendency to the change we are discussing may have. existed even before the first separation. We can well imagine that there may even then have been dialectical differences, distinguishing to some extent the fathers of the future nations. The causes producing these differences need not have acted uniformly: while the ancestors of the Hindus and Greeks agreed in most points, the ancestors of the Greeks and the Germans may have agreed This partial action would be quite in accordance with what I have said above about the formation of dialects within the Greek. in one. The weakening from a to e clearly was the first in time. It has spread more widely than that from a to o over the Western languages, and is also more prevalent in particular languages. Thus (to borrow one or two examples from Curtius) we find that from Indo-European dakan, "ten," have come Greek Séra, Latin decem, O. H. G. zehan, Gothic taihun (modified from tihun by the phonetic law of the language by which i and u become ai and au before r and h), Sclavonic desęti: also from sad, "to sit,” 1 Ebel, Celt. Stud. p. 121, 1 Greek edos, Latin sedes, Gothic sita, Lithuanian sedmi. It will be observed that the Sclavo-Lithuanian agrees with the Greek and Latin: in the Teutonic family while the Old High German has e the Gothic has i. This i is sometimes said to be weakened from e, as also u from o1; but it is strange that the middle step should have passed away without a trace of it being left, for there is no ĕ or o in Gothic. It is possible that this tendency was not fully developed within the Teutonic race till Goths were distinct from Germans, after which time each people carried out the change in its own way. At all events the use of i and u in Gothic is very parallel to that of e and o in Greek2. The tendency then to let a sink into e was clearly strong among the Western peoples before their separation. Traces of the change from a to o are much harder to find. Thus3 Bous and bos are the O. H. G. chuo, Slavonic govędo. But whilst aktan is octo in Graeco-Italian, no vowel-change is to be seen in the Gothic ahtau, or Lithuanian asztůni; ovis, "a sheep," is avis in Gothic and Lithuanian as much as in Sanskrit: the Sanskrit amsa is duos and umerus (where the u is derived from o), but remains amsa in Gothic: similarly we find λévn and ulna, but Gothic aleina, German elle. This tendency had scarcely begun to act at the time of the separation of the Northern and Southern peoples: it was then checked in the North, while circumstances, which we cannot now certainly ascertain, favoured its development among the Graeco-Italian people. It was more developed within the Northern races by the Sclavo-Lithuanian than by the Teutonic. Thus ost is Sclavonic for an "axle," which has a in all the other 1 As by Curtius, Gr. Et. p. 88. Thus in Greek we had beside radical тpap, the stems тpe and Tpop. Similarly in Gothic, the Indo-Eur. BANDH, "to bind," becomes band, whence bandi, a band" or bond," and bandja, a prisoner:" the present of the verb is binda (analogous to rpépw): and though band, 3 Curtius, Gr. Et. p. 88. P. E. 14 This CH. VII. |