________________
169
endings is entirely regulated by the metre, Weber notes that his MSS are highly irregular in that they add the Anusvāra at random. They do so in particular in the case of the plural endings which end in -a and in -i. This agrees with the state in the hemistich-final position where these Same MSS would often have -āin, -āņaṁ and -ehin, while -esuṁ (plural, but ending in -u) and -enań (ending in -a, but singular) are conspicuously absent. It would follow that the presence of the Anusvāra in -ain, -āņań and -ehim in hemistich-final position in Weber's MSS is secondary.
However, this is only part of the picture. Ma, Ti and Tp at the end of the hemistich indeed generally have the endings without Anusvāra, but exceptions occur, which are, however, usually restricted to one or two of these MSS. See, for instance, naaņāin (qaaņālt) in Ma in *5. The addition of the Anusvāra can hardly be explained as secondary, as otherwise, i.e. within the Gathā, it is not found at all. A more likely explanation is that the source of Ma, Ti and Tp occasionally did have -ain, -āņań and -ehim, but that most instances have been removed in a process of normalization in favour of the endings without Anus vāra. The reason behind the preference for the endings without Anusvāra is unclear. It might be a case of extending the endings without Anusvāra which prevail within the Gathā to the instances at the end of the hemistich. Note in this connection that in Tp the endings -e and -o, which mostly are to be scanned short, are represented by the characters ě and throughout, i.e. also where they are metrically long and also where the distinction does not matter, as at the end of the hemistich. A further illustration of the extent to which the MSS of the Third South-Indian recension were normalized will be given below in connection with the Ya-śruti.
Apparently the situation in the source of Ma, Ti and Tp was irregular. It is in any case no longer possible to determine whether in a particular case the archetype read, for instance, -āi or -ain. In the edition I have decided to follow the South-Indian MSS, which means that I have throughout adopted the endings without Anusvāra, except where Ma, Ti and/or Tp have an ending with Anus vāra.
As to these variants the following procedure has been adopted. First those MSS are mentioned of which the existence of a variant is beyond doubt, and then I have added between brackets those MSS of which the reading cannot be definitely ascertained, mostly R, V, Y and Bh, and T
