Page:Atharva-Veda samhita.djvu/92

lxxxiv (dh for d, ī for i, k for g).—Of variety in the character of the Kashmirian variants there is no lack. Thus we see the omission of a needed twin consonant (cf. p. 832) in yad [d]aṇḍena, folio 91 b $5$ = v. 5. 4 a; interesting phonetic spellings in mahīyam of folio 264 b $6$ for mahyam of iii. 15. 1 d, and in e te rātriy anaḍvāhas of folio 158 a $17$ for ye te rātry anaḍvāhas of xix. 50. 2 a; inversion in the order of words in sa me kṣatraṁ ca rāṣṭhraṁ ca of folio 187 a $4$ = x. 3. 12 c. Not one of these examples was reported, though probably all were noticed, by Roth. In his Collation for v. 6, he notes for verses 11-14 "unwesentliche Differenzen," without specifying them. We may regret his failure to report such an interesting reading as yathāhaṁ çatruhāsany, folio 3 b $14$, where çatruhā is a correct equivalent of the çatruhas of the Vulgate, i. 29. 5 c; but with such a blunder as asăni in the very next word, and such grammar as ayaṁ vacaḥ in the preceding pāda, we cannot blame him. In an incomplete collation, there is no hard and fast line to be drawn between what shall be reported and what shall not.

Collation not controlled by constant reference to the birch-bark ms.—Secondly, as. for the accuracy of Roth's Collation in the variants which he does give,—I do not suppose that Roth attempted to control his Kashmirian nāgarī transcript (No. 16, Garbe) on which he based his Collation, by constant reference to the original. Thus far, I have hardly come upon inaccuracies myself; but it is not improbable that occasional slips on his part may yet come to light. It is proper here, therefore, partly by way of anticipating ill-considered criticism, to explain the situation.

Such reference would have ruined the birch-bark ms.—As any one can see from the table, pages 1018 to 1023, the Kashmirian correspondents of the Vulgate verses are to be found in the birch-bark manuscript in an entirely different order. Thus, if we take for example the six Vulgate verses iii. 12. 1, 6, 8; 13. 1; 14. 1; 15. 1, we shall find their Kashmirian correspondents at the following places (leaf, side, line) respectively: 54 b $2$, 276 b $7$, 225 a $10$, 50 a $1$, 32 b $8$, 264 b $5$. From this it is evident that the mechanical process of referring, as one proceeds verse by verse through the Vulgate, to the parallel verses of the birch-bark original, for the purpose of checking step by step the transcript used for the Collation, would have involved an amount of handling of the fragile birch-bark leaves (nearly 300 in number) which would have ruined them. The leaves are now about 400 years old, and some idea of their fragility may be gained from the remarks in the preface to the facsimile, page II. It was doubtless this difficulty that impressed upon Roth the necessity of making a copy which should be at once accurate, and also strong enough to endure