________________
182
507 janań piva Ma, Ti, Bh, R; -mm iva W, -m iva Y, -im iva S, 609 duddha piva Ma, Ti, Tp, Bh, R, T; -rim ia , -mm iva Y,
-rm iva S, 782 chittań piva Ma, Ti, TP, R, T; -mm iva S.
Of these variants iva, which is found exclusively in the Vulgata MSS and S, is almost certainly an innovation. The change of the preceding Anu- . svāra into -mm or -rim represents an attempt to preserve the metre. Before iva, which begins with a vowel, the Anusvāra may, as in Skt, have changed into -m, in which case the syllable is short. As such the phrase vard imm iva has an exact parallel in Pāli mamm iva (see CPD II, p. 331; for alternative mām iva, see Geiger 71). The status of iva in Pāli, however, cannot be discussed here.
The MS of the archetype seems to have read via as well as piva. The former is not restricted to M. The single prevailing form in ś. and Mg. is via. It is, furthermore, undoubtedly identical to pali viya. In all these dialects and languages the particle occurs after vowels and after Anusvāra.
Geiger ($ 66) suggests that Pāli viya, through metathesis, goes back to *yiva, which would be a frozen sandhi-form. He proceeds from this form *yiva in order to be able to account for the medial -y-, the origin of which, however, need not detain us here. I largely agree with Geiger's explanation but should like to add some further remarks concerning the circumstances which may have been responsible for the supposed metathesis. An indication for this may be found by comparing the other most common enclitic particles, namely (t)ti, (v)va, (c)cia and vi/pi. It should be noted that all these begin with a consonant, some after having lost their original vowel. This apparent tendency of enclitics to begin with a consonant may have been responsible for the metathesis in via.
The second particle, piva, is otherwise exclusively found in Amg. and JM. texts. It occurs only after Anus vāra; after vowels these texts have viva. According to Pischel ($ 337) viva would consist of a kind of proclitic v- plus iva. The existence of such a proclitic v- is, however, doubtful. In the same paragraph Pischel has collected several more instances of this v-, most of which, however, should be explained differ
