________________
187
31 3 Ma, 4 Tp; 2 Bh, 3 R; K, U, B, Y, P, G, 5; 5 T; 8 S; 3 V)
sattasaāin kaivac- hūleņa viraiaiṁ
chalena kodte majjhaarão sālakārāṇa gāhāna.
In Ma lacuna between h(ālena)...rāņa gāhāņa. saīsin K -- kaivachalena K, vatthalena , kavatsalena y -- kotiye Ma; kodta U, B, Y; koţte S -- majjhaārammi R; U, B, P, masjhaārammi K, majhjhaărammi Y; majjhaārammi 5 -- sālaiņa T -- viraiāna Bh -- salamkārāna K -- gāhāņa R; K, U, B, P; T; S; (Bh).
From among ten millions of Gathās full of Alankāras, sevenhundred were collected by Hala, the patron of poets.
The ending -e in kogte is to be scanned short. In most NagarT MSS this -ě, for lack of a separate character, is represented by -e. In some MSS, and practically throughout in B, Y and S, it is represented by -i after -ā (-ai) and -ū (-ūi) and by -a after -T (-Ta). Bh, which usually has -te, sometimes has -Ti; see, for instance, * 22' muhti and tfi. Similarly, - að and -1o are written as - du and - Tu respectively. -ūu for -ūő is not found in the Sattasaf.
i and u as graphic representations of ě and Ở are also found in closed syllables (Pischel $ 119). The origin of -a in, for instance, kodta, is as yet unclear. It may, however, have arisen from confusion between - i- and -ya- : kodTe + kosti → kogfya + kogfa. The recognition of this ending -a was facilitated by the fact that an ending consisting of a vowel similar to the stem-vowel, as in - Ti, was apparently considered awkward. Thus, while Var. V 22 accepts the ending -a, the commentator Bhāmaha, ad Var. V 23, explicitly forbids it after a-stems, where it would give forms like mālāa.
In the Gathās of the Sattasat the endings -e and -o occur practically only in positions where they have to be scanned short. Long -e and - are found at the end of the hemistich, which is metrically neutral. Furthermore they are found at the end of the uneven Pādas where (u) - U does not occur. Otherwise they are rare: see (*22), *43, 226, 228, 322, 492, 513, 593, 599, 739, 789, 797 (ex conj.) and 810. The same is also
