________________
306
Amrita
Of these the form sānuni occurs once, while the forms sānavi, sānau and sāno occur 9 times each. Here in all the 9 cases of sāno (followed by avye 8 times and by avyaye once) in the IX Mandala, the Padapātha uniformly gives the original word as sānau, regarding the sandhi as irregular. This use is duly noted by the Prātiśākhya (II. 70) where Uvvata says that sano avye takes the place of the regular sānāvavye. Macdonell59 suggests that the regular form sānavyavye would give rise to a disagreeable sequence of syllables and therefore the endingless form is used, but without the lengthening for the sake of the metre. In all these 9 cases, the o occupies the 9th place of the Tristubh and Jagatī and is metrically short. In all the cases it is followed by a-. Now, if the demands of the metre are to be satisfied for a short syllable, one fails to see how the Guna form in place of the expected Vrddhi form could help, if both o and au are regarded as long. If on the contrary, the tradition, which regarded o followed by a- as metrically short, was still alive, one can readily understand that the use of o before the following a- without its elision would give the required metrical scheme. Other cases like go-agrayā IX. 71, sd; tiro-ahnyam III. 28. 3b; III. 28. 6c, dūre-ante, III. 54. 7a etc. should be judged in the same manner.
It is well-known that the Prākrit dialects show the short values of ě and , mostly before conjunct consonants and sometimes final e and o are regarded as short in the metrical scheme. These facts are noted by writers on Prākrit grammar and metre. Their frequency is much greater in the Apabhramsa language and is probably a part of the general tendency of the shortening observable in that language.
Two conditions circumscribe the short values of ě and 7 in Prākrit. Either they occur before a conjunct consonant as in ottha, chetta etc. or they are found as the final syllables of words like mālāe, dhammāð etc. In the first case, the metrical value of the syllables remains the same, both Sanskrit ostha and kşetra having the same scheme as the Prākrit ottha and chetta. In the second case, it is only under the influence of metre that we can be conscious of the short value of these vowels. It will, therefore, be more correct to say that these final e and o are pronounced short for the convenience of the metre. In both cases, the phonetic variation is not significant, for with the following conjunct the long values of ě and o cannot remain side by side with their short values in Prākrit; and in case of short ě and o at the end, there are no doublets with long ē and 7 with a difference of meaning. In other words, whether short or long e and o form the same phonemes. ..