30 SANSKRIT, PRAKRIT, AND APABHRANQA
a long vowel, or a short vowel and a consonant, when another consonant follows.
It is probable enough that literature of a secular character was composed in these Old Prakrits until the second century A. D., but about that date we have clear evidence of the fundamental changes which mark what may be called the Middle Prakrit of the grammarians and of most of the extant literature. This consists in the softening or disappearance of intervocalic con- sonants, carried to the furthest in Maharastri in the dominions of the Catavahanas of the south-west, but noteworthy also in the other Prakrits recognized by the grammarians, MagadhI, and Caurasenl. We see in the dramas of Bhasa, as compared with those of Acvaghosa on the one hand and of Kalidasa on the other, clear evidence of transition, the omission of intervocalic consonants, the softening of surds to sonants, the reduction of aspirates to A, the change of y into j, the substitution of n for n, the simplification of double consonants with compensatory lengthening. The evidence of inscriptions supports the view which assigns the loss of intervocalic consonants to the second century A. D., 1 in which century Maharastri lyric began its successful career, made known to us in the anthology of Hala. Once stereotyped by the grammarians at an uncertain date, the Prakrits rapidly lost in importance as they became more and more divorced from current speech, while they did not possess the traditional sanctity of Sanskrit or its clarity of structure and beauty of form.
Of the Prakrits Maharastri held pre-eminence by its use in drama, whence it was introduced perhaps by Kalidasa from lyric poetry, and by its adoption for epic poetry. Caurasenl was normally the prose Prakrit, though it appears to have been occasionally used in verse ; its employment in prose outside the drama was probably once much wider than was later the case when the Jains used a form of Maharastil for prose as well as for verse, though the presence of Caurasenl forms in prose suggests that Maharastri is here intrusive. 2 C aura senl wa s markedly more
1 Bloch, Milanges Llvi, pp. ] 2 fT. (kamdra, however, is from iarmdra). As regards lingualization cf. Turner, JRAS. 1924, pp. 555 ff., 58a ff. (danda, however, is not for dandra ; see Lklen, Stud. z. altind. und vergl. Sprachg., p 80).
3 Jacobi, Bhavisatta Kaha, pp. 88 ff. ; RSO. li. 231 ff.
