RELATION TO OTHER TEXTS [Introduction Sarvānukramaṇi proves some at least of the additional matter in A to have belonged to the original text of the Brhaddevata ¹. But what method have I followed with regard to that part of the text which is common to both recensions, viz. six-sevenths of the whole work (or 1073 out of 1224 ślokas)? Now in this common portion of the text the divergence of the two recensions does not in the vast majority of cases go beyond a word or two in the line, A having as a rule the correct reading, while B has a corruption. Sometimes, however, the divergence cannot be accounted for by the same original reading; for while in these instances the matter is usually identical, the wording of part, or of all of the line, is entirely different 2. In this latter case I have, owing to the much greater correctness of the A MSS., followed the principle of adopting, when other things are equal, the readings of A. I have been confirmed in this by an examination of Sadguruśisya, who, in his commentary on the Sarvānukramani, quotes about fifty ślokas from the Brhaddevată". I find that he predominantly agrees with the readings of A, though he frequently quotes passages which are only to be found in B. From this I conclude that the readings of the B recension, as known to Sadgurusisya ¹, have, in the parts which the two recensions have in common, on the whole been better preserved 5 by A. xx] It has, however, not by any means been my aim to supply here the text of the A recension pure and simple. I have endeavoured to constitute, by means of all the evidence that can be brought to bear on it, a text approxi- mating as nearly as possible to the original form of the Brhaddevatā. 8. External evidence bearing on the text of the Brhaddevatā. I doubt whether, with the exception of the Vedic Samhitãs, there is any other work of Indian literature upon the text of which so much evidence out- side its own MS. material can be concentrated. From the point of view of textual criticism the task of editing the Brhaddevata has therefore proved 1 See Appendix vi and note on Brhadde- vatā viii. 137. 2 The number of cases, in the common text, in which the divergence of wording extends more or less to the whole line, amounts, I think, to hardly twenty. There is no instance in the first Adhyāya, in which the two re- censions can hardly be distinguished (there being also no omissions). 3 Cp. especially the critical notes on v. 50- 78. For a list of these quotations see Appendix V, PP. 134, 135. Sadgurusisya once quotes a passago (v. 64- 67) which does not even occur in all the MSS. of the B family. 4 Sayaṇa, who quotes nearly as much (seo Appendix v) of the Brhaddevata as Sadguru- śisya does, on the whole follows B, though he occasionally agrees with A. His relation to the recensions is not very clear, as he some- times has a different reading when A and B are identical. This may have been due to quoting from memory. I have already men- tioned that the Nitimanjarī follows B.
