________________
172
THE INDIAN ANTIQUARY.
[JUNE, 1892.
substituted for r, the nominative masculine, and usually the nominative neuter, end in é, and the locative in asi; the other distinguishes the cerebral n and the palatal h, retains the initial y and the r unchanged, makes the nominative singular of masculine a-bases end in 6, and the locative in amhi or in é. The first comprises all the inscriptions, except Girnar and Kapur di Giri, which alone constitute the second class. It is the more impossible to doubt the existence in this case of a dialectic difference, because certain of the peculiarities which denote the first group are quoted by the grammarians as proper to the Magadhi dialect. Such are the nominative in é and the substitution of 1 for r. It must be admitted that these are also the only points of agreement, and that neither in its omissions - the absence of , n, é, the elision of initial y - nor in certain peculiar usages - the retention of j, of chh, &c., - does the dialect of the inscriptions correspond with the Màgadhi of the grammarians. On the contrary, we have seen that the use of the group 8, attributed by the grammarians to Magadhi appears only in the orthography of Girnar.
Is it possible to trace subdivisions, to distinguish sub-dialects, within the limits of these two main groups? Between Girnar and Kapur di Giri, if we except the groups 8! and st on the one hand, and the use of the three sibilants on the other, both of which, in my opinion, should not be admitted into the calculation, the only differences of a somewhat general character, which I note, are the group tp at Girnar, which, according to my theory, corresponds to a pronunciation pp, and which is represented at Kapur di Giri by t; the locative singular, which is in mhi, and more rarely in é at Girnar, and in é, never in mhi at Kapur di Giri; and the genitive of bases in in, which is in inô at Girnar, and which, at Kapur di Giri, follows, by the formation isa, the analogy of the declension in a. It may also be noted that the group hm or mh, which is retained at Girnar, is unknown at Kapur di Giri, where bamhana is written bramaņa, and that the termination vya of the future participle passive, which usually at Girnar adheres to the spelling viya, is, at Kapur di Giri, generally assimilated to va (vva). Finally, we may add one or two other divergencies, such as the 3rd person plural in are used at Girnar, and the accidental substitution of y forjat Kapur di Giri. We can thas have no hesitation in holding that the two sets of inscriptions, to a certain degree at least, do reflect different shades of dialect, which are absolutely distinot.
I do not think that the case is the same with regard to the versions which constitute the former group. If we put out of consideration the alleged use of é and sh at Khálsi, regarding wbich I shall shortly take an opportanity of stating my opinion, and which has nothing to do with the present discossion, the only appreciable differences refer to the initial y, the use of r, and the termination of nominatives neuter in ani. Khálsi and the columnar edicts retain the initial y more frequently than the others, but as they present at the same time a number of examples of its elision, even in the same words, it is clear that no linguistic conclusion can be drawn from the fact, especially as in the versions which elide it most regularly, at Jaugada and Dhanli, examples are inversely found of its retention. In some instances Khâlsi makes the nominative of nenter bases in a, in an and not in &: but it also contains a more considerable number of nominatives in é of bases usually treated as neuters; on the other hand, in one instance, Jaugada writes anusásanari. Räpnâth writes chhavachharé and chirathitike, árád havé, pakaré, bat, alsó, sátiléké, apaladhiyêna, and ahálé; and if it is admitted that it throughout retains the initial y, it must not be forgotten that it is short, that it has only three such examples, and that it is impossible to assume the existence of a peculiar dialect from such a detail, in the face of its otherwise perfect agreement with those inscriptions which more nearly resemble Magadhi. It is plain, however, that we must not neglect sporadic discrepancies. They have a certain significance which should be cleared up. This problem appears to me to be an easy one. It will solve itself when we have elucidated one point, regarding which people have, I think, come to wrong conclusions.
It has been admitted hitherto that each of the versions of the edicts faithfully represents the dialect of the country in which it has been engraved. I believe that
