APABHRAN^A 35
confused condition reflected in the grammarians. Hemacandra, who belonged to the western school which goes back to the Vdlmiki Sutras, describes one kind of Apabhrarica, but alludes to others; in the eastern school we find a division as Viacata, Nagara, and Upanagara, in all of which r after consonants is kept while in the first r before consonc^ts also. Faint traces of the observance of this rule may be found in a few verses cited by Hemacandra ; the great poems, Bhavisattakaha and Neminaha- cariu assimilate r, and thus belong to a later type of Apabhrarica. In Bengal we find a type of Apabhrarica long in use in Buddhist texts, and a much degraded form, Avahattha, is evidenced in the Prdkrta Pihgala (14th cent.), but the basis even of this Apabhrarica is Maharastrl, not Magadhi, testifying to its ultimate western origin.
From the nature of Apabhrarica it follows naturally that in Old Gujarat! we find a considerable amount of resemblance in inflexion to Apabhrarica, as was to be expected from the fact that the vernacular is a descendant in considerable measure of that vernacular which was applied to Prakrit to form the early Apabhrarica. In other cases we could not expect to find any such important coincidences ; thus in Bengal the Apabhrarica used was not formed by applying vernacular inflexions to the local Prakrit ; at most some local colour was given to a speech which came from the west, and the same remark clearly applies in other cases. Sir G. Grietson's efforts 1 to establish a Maha- rastra Apabhrarica as a connecting link between Prakrit and MarathI are clearly unsuccessful. Nor indeed, it must be added) is there yet any adequate proof even of the relations suggested by him between the Prakrits and the vernaculars ; 2 thus traces of Magadhi in Bengali are extremely difficult to establish with any cogency. 3
There is no reason to suppose that Apabhrarica formed a necessary step towards composition in vernaculars, and in Maharastra and Kashmir Apabhrarica appears to have been
1 BSOS. f. iii. 63.
2 E. g. his view (JRAS. 1925, pp. 228 ff.) as to single consonants in the Noith-West Prakrit is clearly improbable.
3 M. Shahidullah, IHQ. i. 433 ff. Bloch {Formation de la langue marathe ; JA. '9 12 , i. 336) insists that the modern dialects presuppose a Prakrit koine.
D %
