• • on
xvnij
TWO EECENSIONkS
[intro (1 net ion
University Library. With the exception of a few small lacunae ^ it is complete. It is dated 1786 (=A.D. 1864). It is described in Wickie-* masinghe’s Catalogue of the Max Muller Sanskrit MSS., p. 64a, no. 61,
Accordino- to Mr. Keith these seven MSS. fall into two groups : A, comprising a, b, m ; and B, comprising c, d, g, h. A represents the more modern group, and shows a somewhat simplified text. Of this gioup in is perhaps the best. In the B group, g and h correspond on the whole, and have a general resemblance to c and d. The extract of the Ulwax MS. given by Peterson agrees with g (beginning) and h (end).^ ^
In the (quoted Brhaddevata passages, the Mltimaiijail MSS. follow tlie text of the longer recension (B) in nearly all decisive passages , but the quotations seem to have been very carelessly copied.
6. Two recensions of the Brhaddevata.
An examination of these two main groups of MSS. shows that B (or the longer recension) contains 133 slokas not to be found in A, while A has 18 not to be found in B, The question here arises, what is the relation of these two groups or recensions? A few years ago I read the Brhaddevata critically with three of my advanced pupils (one of whom was Mr. A. B. Keith), in order to teach tliom how to collate MSS. and to constitute a correct text from them. At that time I formed the opinion that the shorter recension was the original one, as the e^i ten- sions in B produced the general impression of superfluous matter. Con- tinued study has, however, convinced me that the additions of the longer recension belonged, on the whole, to the original text, and that abridgement b
This conclusion rests chiefly on two arguments. In the first the text of the Sarvanukramani is not only based throughout on Brhaddevata as one of its main sources, but actually contains, although itself a Sutra work, a large number of metrical passages from vata embedded in its text: several of these passages are taken from the additional matter peculiar to B
The second argument is founded on the length of tlie vargas into which each Adhyaya is divided. The normal number of llokas in a varga is five^. The fifty-eight vargas of the first two Adhyayas (in which the text of
^ My text of the Brhaddevata contains | looted in Appendix vi, Belation of the Brhad- 1224 slokas, of which 1073 are common to hotli devatS, to other Texts, pp. 147-153. recensions ; Ahas ioqIj B 1206. The original ^ This is derived from the division of the form of the text may have contained nearly text of the Bg-veda, where the normal number ■■■3^200. oftho stanzas contained in each varga is live, as
2 The evidence for this will be found col- far as the length of the hymn will admit of ii
