Atharva-Veda Samhita/General Introduction/Part I

GENERAL INTRODUCTION, PART I.

BY THE EDITOR

General Premises

Scope of this Part of the Introduction.—As stated above, p. xxix, this Part contains much that might, but for its voluminousness, have been put into a preface. The main body of the present work consists of translation and commentary. Of the latter, the constituent elements are mainly text-critical, and their sources may be put under ten headings, as follows:

1. Vulgate. European mss.
2. Vulgate. Indian mss.
3. Vulgate. Indian reciters.
4. Vulgate. Commentator's readings.
5. Vulgate. Pada-readings.

6. Vulgate. Prātiçākhya and its comm.
7. Vulgate. The Anukramaṇīs.
8. Vulgate. Kāuçika and Vāitāna.
9. Kashmirian recension. Pāippalāda ms.
10. Parallel texts.

Of these sources, nine concern the Atharva-Veda, and the tenth concerns the parallel texts. Of the nine concerning the Atharva-Veda, eight concern the Vulgate or Çāunakan recension, and the ninth concerns the Kashmirian or Pāippalāda recension. Of the eight concerning the Vulgate, the first four concern both the saṁhitā- and the pada-pāṭhas,^[1] and the second four concern the ancillary texts.

Partly by way of indicating what may fairly be expected in the case of each of these elements, and partly by way of forestalling adverse criticism, it will be well to make certain observations upon them seriatim, under the ten headings. Under an eleventh, I desire to add something to what was said in the preface, p. xxxvii, about the commentary as a whole; and, under a twelfth, to add a few necessary remarks concerning the translation. Under a thirteenth, the explanation of abbreviations etc. may be put; and finally, under a fourteenth, a tabular view of previous translations and comments.

Scope of the reports of variant readings.—By "variant readings" are here meant departures from the printed Berlin text.^[2] Absence of report means in general that the mss. present no true variants, albeit Whitney does not rehearse every stupid blunder of every ignorant scribe. There is of course no clear line to be drawn between such blunders and true variants; and in this matter we must to a certain degree trust the discrimination of the learned editors.

The term "manuscripts" often used loosely for "authorities," that is, manuscripts and oral reciters.—S. P. Pandit, in establishing his text, relied not only upon the testimony of written books, but also upon that of living reciters of the Veda. Accordingly, it should once for all here be premised that Whitney in the sequel has often used the word "manuscripts" (or "mss.") when he meant to include both mss. and reciters and should have used the less specific word "authorities." I have often, but not always,^[3] changed "mss." to "authorities," when precise conformity to the facts required it.

The difficulty of verifying statements as to the weight of authority for a given reading may be illustrated by the following case. At iii. 10. 12 c, Whitney's first draft says, "The ṣ of vy àṣahanta is demanded by Prāt. ii. 92, but SPP. gives in his text vy àsahanta, with the comm., but against the decided majority of his mss., and the minority of ours (H.O., and perhaps others: record incomplete)." The second draft reads, "SPP. gives in his text vy às-, against the decided majority of all the mss." Scrutinizing the authorities, written and oral, for the saṁhitā (since for this variant pada-mss. do not count), I find that Whitney records H.O., and that SPP. records Bh.K.A.Sm.V., as giving ṣ, in all, seven authorities; and that Whitney records P.M.W.E.I.K., and that SPP. records K.D.R., as giving s, in all, nine authorities. Whitney's record is silent as to R.T.; and SPP's report of K. is wrong either one way or else the other. The perplexities of the situation are palpable. I hedged by altering in the proof the words of the second draft so as to read "against a majority of the mss. reported by him."

1. Readings of European Mss. of the Vulgate Recension

The reports include mss. collated, some before publication of the text, and some thereafter.—To the prior group belong Bp.B.P.M.W.E.I.H.; to the latter, collated some twenty years after publication,^[4] belong O.R.T.K. Op.D.Kp. Whitney's description of the mss. is given in Part II. of the Introduction (p. cxi), and to it are prefixed (pp. cx-cxi) convenient tabular views of the mss. The immediate source of these reports is his Collation-Book: see pages cxvii to cxix. In the Collation-BooK, the Berlin and Paris readings (B.P.) are in black ink; the Bodleian readings (M.W.) are in red; the London or "E.I.H." readings are in blue; and, excepting the variants of K.Kp. (which are also in blue), those of the mss. collated after publication (O.R.T.Op.D.) are in violet. The writing is a clear but small hand. The indications of agreement with the fundamental transcript are either implicit (the absence of any recorded variant), or else made explicit by the use of very small exclamation-points. The differences of method in recording are duly explained at the beginning of the Collation-Book, as are also the meanings of the various colored inks: and Whitney's procedure throughout the Book conforms rigorously to his prefatory explanations.

The interpretation of a record so highly condensed and not always complete was sometimes an occasion of error, even for Whitney who made the record and knew the circumstances of its making; and, as may well be imagined, such interpretation was positively difficult and embarrassing for the editor (who had not this knowledge), especially in cases where, after the lapse of years, the colors of the inks were somewhat faded. —Thus Whitney misinterprets his notes of collation at vi. 36. 2, where it is P.I.K. (and not Bp.²I.K., as he wrote it in his copy for the printer) that read víçvāḥ.—Again, at vi. 83. 3, it is W.O.D. (and not H.O.R., as he wrote it for the printer) that read galantás. —Again, in writing out his commentary for the printer so many years after making his collation, he frequently forgot that there was no Op. for books v.-xvii., and has accordingly often reported a reading in violet ink as a reading of Op. when he should have reported it as a reading of D. This slip happened occasionally through several hundred type-pages and remained unnoticed even until the electroplates were made; but I believe I have had all the instances of this error rectified in the plates. —Likewise, in writing out for the printer, the fact seems to have slipped from his mind that he had made his fundamental transcript of book v. from codex Chambers 109 (= Bp.²) and not, like all the rest of the first nine books, from Chambers 8 (= Bp.). I have accordingly had to change "Bp." into "Bp.²," or vice versa, some ten times in book v. (at 6. 8; 7. 3; 8. 3; 24. 3, 14; 27. 10; 30. 11). —I may add that in (the often critically desperate) book xix., Whitney seems to use such an expression as "half the mss." loosely in the sense of "a considerable part of the mss.": so at xix. 29. 1, where the record is presumably not complete for Whitney's authorities, and where "half" is not true for SPP's. —For my own part, in consulting the Collation-Book for manuscript readings, I have exercised all reasonable care, using a magnifying glass regularly and referring frequently to the prefatory explanations.

2. Readings of Indian Manuscripts of the Vulgate

By "Indian mss." are meant those used by S. P. Pandit.—No other Indian authorities are intended, in this section and the next, than those given in S. P. Pandit's edition; they include, as is fully and most interestingly explained in his preface, not merely manuscripts, but also oral reciters. Whitney had only the advance sheets of the parts with comment (books i.-iv., vi.-viii. 6, xi., and xvii.-xx. 37); but, although the remaining parts were accessible to me, I did not attempt for those remaining parts to incorporate S. P. Pandit's apparatus criticus into Whitney's work. I refrained with good reason, for such an attempt would have involved far too much rewriting of Whitney's copy for the printer.

S. P. Pandit's reports not exhaustive.—It is far from being the case that S. P. Pandit always reports upon all his authorities. For books i.-xvii. he had 12 saṁhitā and 6 pada authorities, besides the incomplete comm.; but at ii. 36. 4, note 2, for instance, he reports only 6 out of 13 authorities.^[5] In summarizing SPP's reports, Whitney often says "all of SPP's mss.," "all but one," "the majority," "half," and so on; and it must therefore here be noted that these expressions refer not to the totality of SPP's authorities concerned, but rather to the totality of those concerned and reported upon by SPP. in any given instance. Compare Whitney's notes to iii. 4. 5 (line 2 of the note); iv. 7. 3 (line 6); iv. 26. 5 and iii. 30. 3; ii. 36. 4 (line 9), with SPP's critical notes on the same verses.

3. Readings of Indian Oral Reciters of the Vulgate

By "Indian oral reciters" are meant those employed by S. P. Pandit.—It was from the lips of three living authorities that the Bombay editor took much of the testimony which he used in the establishment of his text. His Vāidikas were Bāpujī Jīvaṇrām (cited as Bp.), Keçava Bhaṭ bin Dājī Bhaṭ (K.), and Venkaṇ Bhaṭjī (V.), "the most celebrated Atharva Vāidika in the Deccan." The last two were authorities for the whole text in both pāṭhas, saṁhitā and pada. The remarks made in the preface to the Bombay edition by S. P. Pandit concerning his reciters are extremely interesting and suggestive.

Errors of the eye checked by oral reciters.—The student should bear in mind the especial weight of the oral testimony in cases where errors of the eye, as distinguished from errors of the ear, are probable. Thus the testimony of the reciters, at ix. 8 (13). 20, establishes the reading visalpa-, as against visalya- of the Berlin text. Save in AV., the word is otherwise unknown, and, as the ms.-distinction between lya and lpa in such a case is worthless, the instance is a typical one to show the value of the reciters' reading: see W's note to vi. 127. 1. The case is somewhat similar at iii. 12. 3, āsyand-, as against āspand- (see the note and my addition); so also at viii. 6. 17, spandanā́, as against syandanā́, where, although only V. is cited, his testimony is abundantly confirmed by the sense (see note). At xix. 66. i (see note), as between those mss. which give paid and the Vāidikas K. and V., who recited yāhi, there can be no question that we ought to follow the latter, although SPP. strangely rejects their evidence. Cf. the notes on çāyaya, at iv. 18. 4, and samuṣpalā, at vi. 139. 3. One of the clearest errors of visual or graphical origin is "Sāyaṇa's" idam, at vi. 37. 2, for hradam or hṛdam of the authorities, including K. and V. (cf. W's and SPP's notes). If this comm. was the real Sāyaṇa, the blunder does him no credit. At viii. 2. 1, çnuṣṭi is established (as against çruṣṭi) by the testimony of all the reciters; although the case is less clear at iii. 17. 2 and 30. 7 (see the notes). Upon their testimony, at x. 7. 16 (see notes), we ought to accept as the true Atharvan reading, prapyasā́s, albeit ἅπαξ λεγόμενον and of questionable meaning.

4. Readings of the Hindu Commentator

The critical value and the range of his variant readings.—Whitney has given full and well-reasoned expression to his low opinion of the exegetical value of the commentary and of the range and critical value of its variant readings, in an article in the Festgruss an Roth, pages 89-96. To that article, with its abundant lists and details, I call, as in duty bound, the especial attention of the reader. The commentator does indeed correct a good many surface-blunders, part of which the Berlin editors had also corrected; and his readings are occasionally supported (as against the two editions) by a parallel text:^[6] but his variants "consist almost exclusively of single words or forms," and of real critical insight he exhibits almost none.

Thus he fails to recognize the fact that the ordinary usage of the mss. makes no distinction between double consonants in groups where the duplication is phonetic, and those in groups where the duplication is etymological (cf. W's Grammar, §232); and is accordingly so obtuse as to misunderstand and explain tádyā́meti, at iv. 19. 6, as tád yám eti, although the slightest heed for the rules of accent would have shown him that it is impossible for the combination to mean anything but tád dyā́m eti. Similarly at iv. 28. 3, again with utter disregard of accent, he makes out of stuvánnemi (that is stuván emi: cf. Festgruss, p. 90-91) an untranslatable stuvan nemī: here, it is true, one of the wildest blunders of the pada-kāra was before him; but even a modicum of insight should have kept him out of that pitfall. Again, he seems never to have observed that past passive participles with a preposition accent the preposition (cf. Grammar, §1085 a), and accordingly takes saṁvṛ́tas at xviii. 3. 30 as if it were sámvṛtas. Despite accent and pada-kāra, he takes rajasā́, p. -sā́ḥ, at xi. 2. 25, as instr. of rájas! And so on.

The text used by the commentator is nevertheless notably different from that given by the mss. used for the Berlin edition, and from that given by S. P. Pandit's authorities. In books i.-iv. Whitney counts over three hundred peculiarities of the commentator's text, and in the Festgruss he gives several lists of them. He has intended in the present work to report all variants of the commentator's text throughout, and I trust that those which may have escaped his notice (or his and mine) will prove to be few indeed.

Was the commentator of the Atharva-Veda identical with the Sāyaṇa of the Rig-Veda?—I suggest that it might prove to be an interesting and by no means fruitless task to institute a systematic and critical comparison of the Mādhavīya-vedārtha-prakāça (or RV. -bhāṣya) with the bhāṣya on the AV., with special reference to the treatment of the accent in the two works, and to the bearings of these comparisons upon the question of the identity of the Sāyaṇa of the RV. with the "Sāyaṇa" of the AV. The latter^[7] does indeed sometimes heed his accents; but the occasions on which he takes notice of them expressly are of utmost rarity (see W's note to xix. 13. 9 and mine to verse 4).

If, by way of comparing the two comments, we take the accusative plural yamárājñas, we find that at RV. x. 16. 9 Sāyaṇa explains it quite rightly as a possessive compound, yamo rājā yeṣāṁ, tān; while at AV. xviii. 2. 46, on the other hand, in the half-verse addressed to the dead man, 'by a safe (?) road, go thou to the Fathers who have Yama as their king,' áparipareṇa pathā́ yamárājñaḥ pitṝ́n gacha, "Sāyaṇa" makes of the very same form a gen. sing, and renders 'by a safe road belonging to king Yama (tasya svabhūtena mārgeṇa) go thou to the Fathers'! Evidently, so simple a matter as the famous distinction between índra-çatru and the blasphemous indra-çatrú (cf. Whitney on TPr. xxiv. 5; Weber, Ind. Stud. iv. 368) was quite beyond his ken. Such bungling can hardly be the work of a man who knew his Rig-Veda as the real Sāyaṇa did.

5. Readings of the Pada-pāṭha

These were reported in the Index, and have since been published in full.— As elsewhere noted, these have been reported in the Index Verborum in such wise (see Index, p. 4) as to enable us to determine the pada-form of every item of the Atharvan vocabulary. An index, however, is an inconvenient vehicle for such information, and the complete pada-pāṭha, as published by S. P. Pandit, is accordingly most welcome. Some of his occasional errors of judgment in the establishment of that text are pointed out by Whitney in the places concerned; but the pada-pāṭha has deeper-seated faults, faults which are doubtless original with its author and not simple errors of transmission.^[8] Here again I may make a suggestion, namely, that a critical and systematic study of the palpable blunders of the pada-pāṭha would be an interesting and fruitful task. Even the pada-text of books i.-xviii. stands on a very different plane from that of the RV. (cf. Geldner, Ved. Stud., iii. 144). A critical discussion of its character is not called for here; but several illustrative examples may be given.

Illustrations of the defects of the Pada-pāṭha.—Verb-compounds give occasion for several varieties of errors. Thus, first, as respects accentuation, we find, on the one hand, incorrect attribution of accent to the verbal element (cf. v. 22. 11); and, on the other, denials of accent which are quite intolerable, as at xiv. 2. 73 (yé: ā́: agaman instead of ā॰ágaman) and xiv. 1. 9 (yát: savitā́: adadāt: where Çākalya resolves aright savitā́: ádadāt).^[9]

Secondly, as respects details of division, we find gross violation of the rule. The rule (a very natural one) for compounds with finite verb-forms is that the preposition, if accented, is treated as an independent word and has the vertical mark of interpunction (here represented by a colon) after it; but that, if accentless (proclitic), it is treated, not as an independent word, but as making a word-unit with the verb-form, and is accordingly separated therefrom only by the minor mark of separation or avagraha (here represented by a circle). Thus in AV. i. 1, we have ní: ramaya and pari॰yánti. Such a division as ní॰ramaya or pari: yánti would be wholly erroneous; and yet we find errors of the first type at vi. 74. 2 (sám॰jñapayāmi), 114. 2 (úpa॰çekima), xiii. 3. 17 (ví॰bhāti), xviii. 2. 58 (pári॰īn̄khayātāi), 4. 53 (ví॰dadhat).^[10]

Various combinations.—The combination of e or o (final or initial) with other vowels gives rise to errors. Thus at viii. 2. 21 cd = i. 35. 4 cd, ténu (= te ánu) is resolved by the pada-kāra as té ánu, and the comm. follows him in both instances. In matters concerning the combination of accents he is especially weak, as when he resolves saptā́syāni into saptá ā́syāni at iv. 39. 10 (see note). The errors in question are of considerable range, from the venial one of not recognizing, at xiv. i. 56, that ánvartiṣye means ánu: vartiṣye,^[11] to the quite inexcusable ones of telling us that yá stands for yáḥ in the verse x. 10. 52, yá eváṁ vidúṣe dadús, té etc., or that māyá stands for māyā́ḥ as subject of jajñe in viii. 9. 5. Perhaps his tát: yā́m: eti (iv. 19. 6) and stuván: nemi (iv. 28. 3), already noticed (p. lxvii) in another connection, may be deemed to bear the palm. Beside the former we may put his resolution^[12] of sómātvám (= sómāt tvám), at iv. 10. 6, into sómā: tvám.

6. The Prāticakhya and its Commentary

Character of Whitney's editions of the Prātiçākhyas.—In the preface to his edition of the Tāittirīya Saṁhitā, Weber speaks with satisfaction of the service rendered him in the task of editing that Saṁhitā by Whitney's critical edition of the appurtenant Prātiçākhya. Whitney's edition of that treatise is indeed a model; but even his earlier edition of the Atharvan Prātiçākhya was buttressed by such elaborate studies of those actual facts which form the topics of the Prātiçākhya, and by such complete collections of the different classes of those facts, that he could speak with the utmost authority in criticism of the way in which the maker of the Prātiçākhya, or of the comment thereon, has done his work, and could pronounce weighty judgment concerning the bearing of the treatise in general upon the constitution of the Atharvan text.

Bearing of the Atharvan Prātiçākhya upon the orthography and criticism of the text.—First, as for the orthography, a discussion of the importance of the Prātiçākhya for that purpose is superfluous for any student acquainted with the nature of the treatise; but the orthographic method pursued by the editors of the Berlin text and the relation of that method to the actual prescriptions of the Prātiçākhya are made the subject of a special chapter, below, p. cxxiii.—Secondly, the treatise does bear upon the general criticism of the text. That it ignores the nineteenth book is a weighty fact among the items of cumulative evidence respecting the original make-up of the text and the supplementary character of that book: see p. 896, line 6. In matters of detail also, the treatise or its comment is sometimes of critical value: thus the non-inclusion of iḍas pade among the examples of the comment on APr. ii. 72 (see note) arouses the suspicion that vi. 63. 4 (see note) was not contained in the commentator's AV. text.

Utilization of the Atharvan Prātiçākhya for the present work.—Whitney's edition is provided with three easily usable indexes (not blind indexes): one of Atharvan passages, one of Sanskrit words, and a general index. The first gives in order some eight or nine hundred Atharvan passages, and gives nearly twelve hundred references to places in the Prātiçākhya or the comment or Whitney's notes, in which those passages are discussed. Whitney has transferred the references of the first index with very great fulness, if not with absolute completeness, to the pages of his Collation-Book, entering each one opposite the text of the verse concerned. Very many or most of them, after they have once been utilized in the constitution of the text of the Saṁhitā, are of so little further moment as hardly to be worth quoting in the present work; the rest will be found duly cited in the course of Whitney's commentary, and their value is obvious.

7. The Anukramaṇīs: "Old" and "Major"

More than one Anukramaṇī extant.—At the date of the preface to the Berlin edition, it was probably not clearly understood that there was more than one such treatise. The well-known one was the Major Anukramaṇī, the text of which was copied by Whitney from the ms. in the British Museum in 1853, as noticed below, p. lxxii. In making his fundamental transcript of the Atharvan text, certain scraps, looking like extracts from a similar treatise, were found by Whitney in the colophons of the several divisions of the mss. which he was transcribing, and were copied by him in his Collation-Book, probably without recognizing their source more precisely than is implied in speaking of them as "bits of extract from an Old Anukramaṇī, as we may call it" (see p. cxxxviii).

The Pañcapaṭalikā.—The Critical Notice in the first volume of the Bombay edition made it clear that the source of those scraps is indeed an old Anukramaṇī, and that it is still extant, not merely as scattered fragments, but as an independent treatise, and that its name is Pañcapaṭalikā. That name is used by "Sāyaṇa" when he refers to the treatise in his comm. to iii. 10. 7. In the main body of this work the treatise ia usually styled the "quoted Anukr." or the "old Anukr." The word "old" means old with reference to the Major Anukramaṇī; and since the dependence of the latter upon the former is now evident (see p. 770, ¶4, end, p. 793, ¶1, end) it appears that the word "old" was rightly used. The excerpts from the treatise, scattered through Whitney's Collation-Book, have been gathered together on six sheets by him. I was tempted to print them off together here for convenience; but several considerations dissuaded me: they are after all only fragments; they are all given in their proper places in the main body of this work; and, finally, the Bombay editor (see his Critical Notice, pages 17-24) gives perhaps more copious extracts from the original treatise than do the colophons of Whitney's mss. For some of the excerpts in their proper sequence and connection, see below, pages 770-1, 792-3, and cf. pages 632, 707, 737, 814.

Manuscripts of the Pañcapaṭalikā.—Doubtless S. P. Pandit had a complete ms. of the treatise in his hands; and, if its critical value was not exhausted by his use of it, it may yet be worth while to make a critical edition of this ancient tract. It is not unlikely that the ms. which S. P. Pandit used was one of those referred to by Aufrecht, Catalogus catalogorum, p. 315, namely, Nos. 178-9 (on p. 61) of Kielhorn's Report on the search for Sanskrit mss. in the Bombay Presidency during the year 1880-81. Both are now listed in the Catalogue of the collections of mss. deposited in the Deccan College (Poona), p. 179. According to Garbe's Verzeichniss der Indischen Handschriften (Tübingen, 1899), p. 90, Roth made a copy of the treatise from a Bikaner ms., which copy is now in the Tübingen Library.

The Bṛhatsarvānukramaṇī.—This treatise is usually styled in the sequel simply "the Anukr.," but sometimes "the Major Anukr." The excerpts from the treatise which are given at the beginning of the introductions to the several hymns in this work are taken from Whitney's nāgarī transcript which he made in London in 1853 on the occasion of his visit there to make his London collations (p. xliv). The transcript is bound in a separate volume; and the edited excerpts are so nearly exhaustive that relatively little work remains for an editor of the treatise to do.

Manuscripts of the Bṛhatsarvānukramaṇī.—Whitney made his transcript from the Polier ms. in the British Museum which is now numbered 548 by Bendall in his Catalogue of the Sanskrit mss. in the British Museum of 1902. The ms. forms part of Polier's second volume described below, p. cxiii, under Codex I; and it is the one from which was made the ms. transcribed for Col. Martin and numbered 235 by Eggeling (see again p. cxiii). Whitney afterwards, presumably in 1875, collated his London transcript with the Berlin ms. described by Weber, Verzeichniss, vol. ii., p. 79, No. 1487, and added the Berlin readings in violet ink. The Berlin ms. bears the copied date saṁvat 1767 (A.D. 1711): it is characterized by Weber, Ind. Stud. xvii. 178, as "pretty incorrect"; but my impression is that it is better than the ms. of the British Museum.

Text-critical value of the Anukramaṇīs.—The most important ancillary treatise that an editor needs to use in establishing the text of the saṁhitā, is the Prātiçākhya; but the Anukramaṇīs are also of some importance, especially for the settlement of questions concerning the subdivisions of the text (cf., for example, pages 611, 628: or note to iv. 11. 7), as has been practically shown by S. P. Pandit in his edition, and in his Critical Notice, pages 16-24.—The pronouncements of the Anukramaṇīs concerning the verse-norms of the earlier books (see p. cxlviii) are also of value in discussing general questions as to the structure of the saṁhitā. In particular questions, also, the statements of the Major Anukr. are sometimes of critical weight. Thus iii. 29, as it stands in our text, is a hymn of 8 verses; but our treatise expressly calls it a ṣadṛca, thus supporting most acceptably the critical reduction (already sufficiently certain: see note to vs. 7) of the hymn to one of 6 verses, the norm of the book. —Here and there are indications that suggest the surmise that the order of verses (cf. p. 739) or the extent of a hymn (cf. p. 768), as contemplated by the Anukr., may be different from that of our text.—Its statements as to the "deity " of a given hymn are sometimes worth considering in determining the general drift of that hymn; and its dicta regarding the "seers" of the hymns are of interest in certain aspects which are briefly noticed below, pp. 1038 ff. —Then too, the manuscripts of the Anukr. may sometimes be taken as testimony for the readings of the cited pratīkas (cf. note to iv. 3. 3). And it happens even that the authority of the Major Anukr. may be pressed into service at x. 5. 49 (see the notes) to determine which pair of verses (whether viii. 3. 12-13 or vii. 61. 1-2) is meant by the yád agna íti dvé of the mss. (see below, p. cxx: and cf. the case at xix. 37. 4).

The author of the Major Anukramaṇī as a critic of meters.—The author shows no sense for rhythm. His equipment as a critic of meters hardly goes beyond the rudimentary capacity for counting syllables. Thus he calls ii. 12. 2 jagatī; but although pāda a has 12 syllables, its cadence has no jagatī character whatever. To illustrate the woodenness of his methods, we may take ii. 13. 1: this he evidently scans as 11 + 11: 10 + 12 = 44, and accordingly makes it a simple triṣṭubh, as if the "extra" syllable in d could offset the deficiency in c! For the spoiled c of the Vulgate, the Ppp. reading pibann amṛtam (which is supported by MS.) suggests the remedy, and if we accept that as the true Atharvan form of the verse, it is then an example of the mingling (common in one and the same verse) of acatalectic jagatī pādas with catalectic forms thereof. So far, indeed, is he from discerning matters of this sort, that his terminology is quite lacking in words adequate for their expression.^[13]

If the author of the Major Anukr. showed some real insight into Vedic meters, his statements might, as can easily be seen, often be of value in affecting our critical judgment of a reading of the saṁhitā or in determining our choice as between alternative readings. The contrary, rather, is wont to be the case. Thus at iv. 15. 4, his definition, virāṭpurastādbṛhatī, implies the division (given also by the pada-mss.) 10 + 8: 8 + 8, thus leaving the accentless parjanya stranded at the beginning of a pāda! An excellent illustration of the way in which he might help us, if we could trust him, is offered by iv. 32. 3 b, which reads tápasā yujā́ ví jahi çátrūn. Here Ppp. makes an unexceptionable triṣṭubh by reading jahīha, and the author of the Anukr. says the verse is triṣṭubh. His silence respecting the metrical deficiency in the Vulgate text would be an additional weighty argument for judging the Ppp. reading to be the true Atharvan one, if only we could trust him—as we cannot. Cf. end of W's note to iv. 36. 4.

Such as it is, his treatment of the meters is neither even nor equably careful. Thus he notes the irregularity of vii. 112. 1, while in treating the repetition of the very same verse at xiv. 2. 45 (see note), he passes over the bhuriktvam in silence. Throughout most of the present work, Whitney has devoted considerable space to critical comment upon the treatment of the meters by the Anukr. Considering the fact, however, that the principles which underlie the procedure of the Hindu are so radically different from those of his Occidental critic, no one will be likely to find fault if the criticisms of the latter prove to be not entirely exhaustive.

His statements as to the seers of the hymns.—The ascriptions of quasiauthorship, made by the author of the Major Anukr. and given in the Excerpts, are set forth in tabular form at p. 1040 and are critically discussed at p. 1038, which see.

8. The Kāuçika-Sūtra and the Vāitāna-Sūtra

The work of Garbe and Bloomfield and Caland.—As elsewhere mentioned (p. xxv), the Vāitana has been published in text and translation by Garbe, and the text of the Kāuçika (in 1890) by Bloomfield. Since 1890, a good deal of further critical work upon the Kāuçika has been done by Bloomfield^[14] and by Caland.^[15] —The value of these Sūtras is primarily as a help to the understanding of the ritual setting and general purpose of a given hymn, and so, mediately, to its exegesis. From that aspect they will be discussed below (p. lxxvii). Meantime a few words may be said about their value for the criticism of the structure of the Saṁhitā.

Bearing of the ritual Sūtras upon the criticism of the structure and text of the Saṁhitā.—Bloomfield himself discusses this matter in the introduction to his edition of Kāuçika, p. xli. He there points out instances in which briefer independent hymns have been fused into one longer composite hymn by the redactors of the Saṁhitā, and shows that the Sūtras recognize the composite character of the whole by prescribing the employment of the component parts separately. Thus (as is pointed out also by Whitney), iv. 38 is made up of two independent parts, a gambling-charm (verses 1-4) and a cattle-charm (verses 5-7). The Sūtra prescribes them separately for these wholly different uses, the former with other gambling-charms; and to the latter it gives a special name. Bloomfield's next illustrations, which concern vii. 74 and 76, have in the meantime given rise to the critical question whether vii. 74. 1-2 and 76. 1-2 did not form one hymn for Keçava.^[16]

The mss. of the Sūtras may sometimes be taken as testimony for the readings of the cited pratīkas. The like was said (p. lxxiii) of the mss. of the Anukramaṇīs. The mss. of the Kāuçika (cf. Bloomfield's Introduction, p. xxxix) are wont to agree with those of the Vulgate, even in obvious blunders.

Grouping of mantra-material in Sūtra and in Saṁhitā compared.—Many instances might be adduced from the Kāuçika which may well have a direct bearing upon our judgment concerning the unitary character of hymns that appear as units in our text. To cite or discuss them here would take us too far afield, and I must content myself once more with a suggestion, namely, that a systematic study of the grouping of the mantra-material in the ritual, as compared with its grouping in the Saṁhitā, ought to be undertaken. At Kāuç. 29. 1-14 the verses of AV. v. 13 are brought in for use, all of them and in their Vulgate order. The like is true of AV. ix. 5. 1-6 at Kāuç. 64. 6-16. Whether it would lead to clear-cut results is doubtful; but the relation of the two groupings is a matter no less important than it is obscure. The obscurity is especially striking in book xviii., where the natural order of the component rites of the long funeral ceremony is wholly disregarded by the diaskeuasts in the actual arrangement of the verses of the Saṁhitā. Thus xviii. 4. 44, which accompanies the taking of the corpse on a cart to the pyre, ought of course to precede xviii. 2. 4, which accompanies the act of setting fire to the pile. See my remark, below, page 870, lines 7-9, and my discussion, pages 870-1, of "Part III." and "Part V." of xviii. 4. As is noted at xviii. 1. 49 and 2. 1, the ritual group of verses that accompany the oblations to Yama in the cremation-ceremony wholly disregards even so important a division as that between two successive anuvāka-hymns. It is pointed out on p. 848 that verse 60 of xviii. 3 is widely separated from what appears (most manifestly and from various criteria) to be its fellow, to wit, verse 6.

Many difficulties of the Kāuçika yet unsolved.—It will very likely appear that Whitney has misunderstood the Kāuçika here and there; as also, on the other hand, he has in fact here and there corrected the text or the interpretation of Garbe or of Bloomfield. At the time of Whitney's death, Bloomfield's chief contributions (SBE. xlii.) to the interpretation of Kāuçika had not yet appeared, nor yet those of Caland. As I have more than once said, no one ought to be so well able to give a trustworthy translation of a difficult text as the man who has made a good edition of it; and for this reason one must regret that Bloomfield did not give us—in the natural sequence of the sūtras—as good a version as he was at the time able to make, instead of the detached bits of interpretation which are scattered through the notes of SBE. xlii. Caland observes, in the introduction to his Zauberritual, p. IV, that in using the Kāuçika he soon found that, in order to comprehend even a single passage, it is necessary to work through the whole book. The like is, of course, equally true of the Prātiçākhya. A commentator upon the Saṁhitā who wishes (as did Whitney) to combine in his comment the best of all that the subsidiary treatises have to offer, cannot of course stop to settle, en passant, a multitude of questions any one of which may require the investigation of a specialist. Thus Whitney, in his note to X. 5. 6, said in his ms. for the printer, "The Kāuç. quotes the common pratīka of the six verses at 49. 3, in a witchcraft-ceremony, in connection with the releasing of a bull." If Caland is right (Zauberritual, p. 171), the hocus-pocus with the "water-thunderbolts" does not begin until 49. 3, and the svayam is to be joined to the preceding sūtra (ZDMG. liii. 211), and the letting loose of the bull (49. 1) has nothing to do with the uses of x. 5. This is just the kind of error which we cannot fairly blame Whitney for making. Special difficulties of this sort should have been settled for him by the sūtra-specialists, just as he had settled the special difficulties of the Prātiçākhya when he edited that text.

Value of the ritual Sūtras for the exegesis of the Saṁhitā.—Estimates of the value of these Sūtras as casting light upon the original meaning of the mantras have differed and will perhaps continue to differ. The opinion has even been held by a most eminent scholar that there is, on the whole, very little in the Kāuçika which really elucidates the Saṁhitā, and that the Kāuçika is in the main a fabrication rather than a collection of genuine popular practices. The principal question here is, not whether this opinion is right or wrong, but rather, to what extent is it right or wrong. It is, for example, hard to suppose that, upon the occasion contemplated in kaṇḍikā 79 of the Kāuçika, a young Hindu, still in the heyday of the blood, would, at such an approach of a climax of feeling as is implied in the acts from the talpārohaṇa to the actual nidhuvana (79. 9) inclusive, tolerate—whether patiently or impatiently—such an accompaniment of mantras as is prescribed in sūtras 4 to 9. Whatever philological pertinence may be made out for them (cf. Whitney's note to xiv. 2. 64), their natural impertinence to the business in hand seems almost intolerable.

To this it may be answered that the Sūtra often represents an ideal prescription or ideale Vorschrift,^[17] compliance with which was not expected by any one, save on certain ceremonial occasions, the extreme formality of which was duly ensured by elaborate preparation and the presence of witnesses.

The data of the Kāuçika no sufficient warrant for dogmatism in the exegesis of the Saṁhitā.—There is every reason to suppose that the actual text of the saṁhitās is often a fragmentary and faulty record of the antecedent (I will not say original) oral tradition; and that the stanzas as we find them have often been dislocated and their natural sequence faulted by the action of the diaskeuasts. It is moreover palpable that questions of original sequence, so far from being cleared up, are often complicated all the more by the comparison of the sequences of the ritual texts (see p. lxxv). In these days of rapid travel and communication, it is hard to realize the isolation of the Indian villages (grāmas) and country districts (janapadas) in antiquity. That isolation tended to conserve the individuality of the several localities in respect of the details, for example, of their nuptial and funeral customs; so that the local diversities are sometimes expressly mentioned (uccāvacā janapadadharmā grāmadharmāç ca: AGS. i. 7¹). Astonishingly conservative as India is {see my remarks in Karpūramañjari, p. 206, ¶2, p. 231, note 2), it can nevertheless not be doubtful that her customs have changed in the time from the date of the hymns to that of the ritual books. Evidently, there are divers general considerations which militate strongly against much dogmatism in the treatment of these matters.^[18]

Integer vitae as a Christian funeral-hymn.—During the last twenty-four years, I have often been called to the University Chapel to pay the last tribute of respect to one or another departed colleague or friend. On such occasions, it frequently happens that the chapel choir sings the first two stanzas of the Horatian ode (i. 22), integer vitae scelerisque purus, to the solemn and stately music of Friedrich Ferdinand Flemming. Indeed, so frequent is the employment of these words and this music, that one might almost call it a part of the "Funeral Office after the Harvard Use." The original occasion of the ode, and the relation of Horace to Aristius Fuscus to whom it is addressed, are fairly well known. The lofty moral sentiment of the first two stanzas, however seriously Horace may have entertained it, is doubtless uttered in this connection in a tone of mock-solemnity. Even this fact need not mar for us the tender associations made possible by the intrinsic appropriateness of these two pre-Christian stanzas for their employment in a Christian liturgy of the twentieth century. But suppose for a moment that the choir were to continue singing on to the end, even to Lalagen aniabo, dulce loquentem! what palpable, what monstrous ineptitude! If only the first two stanzas were extant, and not the remaining four also, we might never even suspect Horace of any arrière-pensée in writing them; and if we were to interpret them simply in the light of their modern ritual use, how far we should be from apprehending their original connection and motive!

Secondary adaptation of mantras to incongruous ritual uses.—Let no one say that this case is no fair parallel to what may have happened in India. On the contrary: instances—in no wise doubtful and not a whit less striking—of secondary adaptation of a mantra to similarly incongruous uses in the ritual may there be found in plenty. This secondary association of a given mantra with a given practice has often been determined by some most superficial semblance of verbal pertinence in the mantra, when in fact the mantra had no intrinsic and essential pertinence to the practice whatsoever. For example, ÇGS. prescribes the verse ákṣan for use when the bride greases the axle of the wedding-car; here, I think, there can be no doubt^[19] that the prescription has been suggested by the surface resemblance of ákṣan 'they have eaten' to ákṣam 'axle.' Or, again, to take an example which has been interestingly treated by Bloomfield, the verses xiv. 2. 59-62 doubtless referred originally to the mourning women, who, with dishevelled hair, wailed and danced at a funeral; and they were presumably used originally as an expiation for such noisy proceedings. Secondarily, they have been adapted for use in connection with the wedding ceremonies, "in case a wailing arises," and doubtless for no better reason than that they contained the word for "wailing"; and they have accordingly been placed by the diaskeuasts among the wedding verses, where we now find them. See Bloomfield, AJP. xi. 341, 338: and cf. vii. 466.

9. Readings of the Kashmirian or Pāippalāda Recension of the Atharva-Veda Saṁhitā

General relations of this recension to the Vulgate or Çāunakan recension.^[20]—Just as, on the one hand, the minute differences between two closely related manuscripts of the same recension (for example, between Whitney's P. and M.) represent upon a very small scale the results of human fallibility, so, upon the other hand, do the multitudinous and pervading differences between the general readings of the manuscripts of the Vulgate and those of the birch-bark manuscript of the Kashmirian recension truly represent in like manner the fallibility of human tradition, but on a very large scale. The Çāunakan or Vulgate recension represents one result of the selective process by which the Indian diaskeuasts took from the great mass of mantra-material belonging to the oral tradition of their school a certain amount, arranging it in a certain order; the Kashmirian recension represents another and very different result of a similar process.

Since the birch-bark manuscript has thus far maintained its character as a unique, we shall perhaps never know how truly it represents the best Kashmirian tradition of this Veda; it is quite possible that that tradition was vastly superior to the written reflex thereof which we possess in the birch-bark manuscript, and which, although excellent in many places, is extremely incorrect in very many. Systematic search will doubtless reveal the fact that the Pāippalāda recension, even in the defective form in which it has come down to us, often presents as its variant a reading which is wholly different, but which, as a sense-equivalent, yields nothing to the Vulgate in its claim for genuineness and originality: thus for the Vulgate readings tátas (x. 3. 8), iyāya (x. 7. 31), yā́ ca (x. 8. 10), kṣiprám (xii. 1. 35), amā́ ca (xii. 4. 38), respectively, the Pāipp. presents the sense-equivalents tasmāt, jagāma, yota, oṣam, and gṛheṣu.

The material selected by the makers of the two recensions is by no means coincident. The Kashmirian text is more rich in Brāhmaṇa passages and in charms and incantations than is the Vulgate.^[21] The coincident material, moreover, is arranged in a very different order in the two recensions (cf. p. 1015); and it will appear in the sequel that even the coincident material, as between the Kashmirian and the Vulgate forms thereof, exhibits manifold differences of reading, and that the Kashmirian readings are much oftener pejorations than survivals of a more intelligent version.

This, however, is not always the case: thus, of the two recensions, the Kashmirian has the preferable reading at xii. 2. 30 d. Or again, at v. 2. 8 and xiv. i. 22, the Kashmirian recension agrees with the Rig-Veda, as against the Vulgate, and, at xi. 2. 7, with the Kaṭha reading. In this connection it is interesting to note that the conjectures of Roth and Whitney for the desperate nineteenth book are often confirmed in fact by the Kashmirian readings: instances may be found at xix. 27. 8; 32. 4, 5, 8; 44. 2; 46. 3 (two); 53. 5; 56. 4.

The unique birch-bark manuscript of the Pāippalāda text.—This is described by Garbe in his Verzeichniss as No. 14. It consisted of nearly three hundred leaves, of which two are lost and eight or more are defective. They vary in height from 14 to 21 centimeters; and in width, from 11 to 16; and contain from 13 to 23 lines on a page. The ms. is dated saṁvat 95, without statement of the century. If the year 4595 of the Kashmirian loka-kāla is meant, the date would appear to be not far from A.D. 1519. A description of the ms., with a brief characterization of some of its peculiarities, was given by Roth at Florence in Sep. 1878, and is published in the Atti del IV Congresso internasionale degli Orientalisti, ii. 89-96. Now that the facsimile is published, further details are uncalled for. A specimen of the plates of the facsimile is given in the latter volume of this work. The plate chosen is No. 341 and gives the obverse of folio 187, a page from which have been taken several of the illustrative examples in the paragraphs which follow.

Roth's Kashmirian nāgarī transcript (Nov. 1874).—A nāgarī copy of the original birch-bark manuscript was made at Çrīnagara in 1873. This copy is No. 16 of Garbe's Verzeichniss, and we may call it Roth's Kashmirian nāgarī transcript. It came into Roth's hands at the end of November, 1874. The year of its making appears from Roth's essay, Der Atharvaveda in Kaschmir, pages 13-14; and the date of its arrival in Tübingen, from p. 11 of the same essay. With great promptness, Roth gave an account of it in his essay, just mentioned, which was published as an appendix to an invitation to the academic celebration of the birthday (March 6, 1875) of the king.^[22] —It would appear that Roth's Kashmirian transcript was not the only one made from the birch-bark original in India: S. P. Pandit seems also to have had one; for he cites the Pāippalāda in his edition, vol. iv., p. 369. The copy used by him is doubtless the nāgarī copy procured by Bühler, and listed as VIII. 1 of the collection of 1875-76, on p. 73 of the Catalogue of the Deccan College manuscripts. See also Garbe's Verzeichniss, under No. 17, for the description of another copy (incomplete).

Arrival of the birch-bark original in 1876 at Tübingen.—The original seems to have come into Roth's hands in the early summer of 1876. The approximate date of its arrival appears from Whitney's note to p. xiii of the pamphlet containing the Proceedings of the Am. Oriental Society at the meetings of May and Nov., 1875, and May, 1876 (= JAOS. x., p. cxix): "As these Proceedings [that is, the pamphlet just mentioned] are going through the press, it is learned from Professor Roth that the original of the Devanāgarī copy, an old and somewhat damaged ms. in the Kashmir alphabet, on highly fragile leaves of birch-bark, has reached him, being loaned by the Government of India, which had obtained possession of it. It corrects its copy in a host of places, but also has innumerable errors of its own. It is accented only here and there, in passages."

Roth's Collation (ended, June, 1884) of the Pāippalāda text.—This is written on four-page sheets of note-paper numbered from 1 to 44 (but sheet 6 has only two pages); the pages measure about 5½ × 8½ inches, and there are some 9 supplementary pages (see p. lxxxii, top), sent in answer to specific inquiries of Whitney. As appears from the colophon added by Roth (see below, p. 1009), this Collation was finished June 25, 1884. Since Roth's autograph transcript described in the next paragraph was not made until some months later, I see little chance of error in my assuming that Roth made his Collation for Whitney from his Kashmirian nāgarī transcript, and that he used the birch-bark original to some extent to control the errors of the copy.^[23] Occasional suspicions of error in the Collation were not unnatural, and they led Whitney to ask Roth to reexamine the manuscript upon certain doubtful points. Whitney's questions extend over books i. to v., and others were noted, but never sent. Roth's answers form a valuable supplement to his Collation, and end in April, 1894.

Roth's autograph nāgarī transcript (Dec. 1884).—The end of the Collation which Roth made for Whitney was reached, as just stated, June 25, 1884. After the following summer vacation. Roth made a new transcript from the birch-bark, as appears from his letter to Whitney, dated Jan. 11, 1893: "Von Pāippalāda habe ich devanāgarī Abschrift, aber nicht vollständig. Die mit Vulgata gleichlautenden Verse, die nur durch Fehler Eckel erregen, habe ich bios citiert, z.B. die vielen aus RV., nehme mir aber doch vielleicht noch die Mühe, sie nachzutragen. Ich habe an der Abschrift unermüdlich vom 19. Sept. bis 28. Dez. 1884 geschrieben und diese Leistung als eine ungewöhnliche betrachtet." This transcript is doubtless far more accurate than the one used for the Collation. The badness of the latter and the fragility of the birch-bark original were doubtless the reasons that determined Roth to make his autograph nāgarī transcript: see p. lxxxv, top. ⌊☞ See p. 1045.⌋

The facsimile of the Tübingen birch-bark manuscript (1901).—A magnificent facsimile of the birch-bark manuscript has now been published by the care and enterprise of Bloomfield and Garbe.^[24] The technical perfection of the work is such as to show with marvellous clearness not only every stroke of the writing and every correction, but even the most delicate veinings of the bark itself, with its injuries and patches. Even if other things were equal, the facsimile is much better than the original, inasmuch as a copy of each one of 544 exquisitely clear and beautiful chromophotographic plates, all conveniently bound and easy to handle and not easily injured and accessible in many public and private libraries throughout the world, is much more serviceable than the unique original, written on leaves of birch-bark, fragile with age, easily injured, requiring the utmost caution in handling, and accordingly practically inaccessible except to a very few persons: but other things are not equal; for the transitory advantage of the brilliantly heightened contrast of color which is gained by wetting the birch-bark original, and which passes away as soon as the leaf is dry, is converted into a permanent advantage by the chromophotographic process, in which the plates are made from the freshly wetted original. Moreover, the owner of a facsimile is at liberty to use it at home or wherever he pleases, and to mark it (with pen or pencil) as much as he pleases. The facsimile may therefore truly be said to be in many respects preferable to the original.

Roth's Collation not exhaustive.—Now that the superb facsimile is published, it is possible for a competent critic to test Roth's Collation in respect 1. of its completeness, and 2. of its accuracy. As, first, for its completeness, it is sufficiently apparent from several expressions used by Roth,^[25] that he saw plainly that it would be the height of unwisdom to give with completeness the Kashmirian variants as incidental to a work like this one of Whitney's, whose main scope is very much broader. Roth was a man who had a clear sense of the relative value of things—a sense of intellectual perspective; and he was right.

Faults of the birch-bark manuscript.—The birch-bark manuscript is indeed what we may call in Hindu phrase a veritable 'mine of the jewels of false readings and blunders,' an apapāṭhaskhalitaratnākara, a book in which the student may find richly-abounding and most instructive illustrations of perhaps every class of error discussed by the formal treatises on text-criticism. Thus it fairly swarms with cases of haplography (the letters assumed, on the evidence of the Vulgate, to be omitted, are given in brackets): tăṁ tvā çāle sarvavīrās suvīrā [ariṣṭavirā] abhi sañ carema: ihāiva dhruvā prati [ti]ṣṭha çāle, folio 54 b^3-4 = iii. 12. 1 c, d, 2 a; vaṣaṭkāre yathā yaçaḥ: [yathā yaças] somapīthe, folio 187 a^15-16 = x. 3. 22 b, 21 a; āditye ca [nṛca]kṣasi, folio 187 a¹⁷ = x. 3. 18 b; apa stedaṁ^[26] vāsamathaṁ gotham uta [ta]skaram, folio 158 b¹ = xix. 50. 5 a, b. Confusions as between surd and sonant (cf. p. 749, p. 57) and between aspirate and non-aspirate and between long and short vowels are so common as hardly to be worth reporting: cf. uṣase naṣ pari dhehi sarvān rātrī anākasaḥ, which is found at folio 158 b⁴ = xix. 50. 7 a, b, and exemplifies all three cases (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^[27] 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 handling without injury. To copy the birch-bark leaves in their proper order is a process by which they need suffer no harm; and this is precisely what Roth did (see p. lxxxii) as soon as possible after finishing the pressing task of making the Collation for Whitney. ⌊☞ See p. 1045.⌋

Care taken in the use of Roth's Collation. Word-division.—In carrying this work through the press, I have constantly and with the most scrupulous pains utilized Roth's original Collation and his supplementary notes thereto, endeavoring thus to check any errors concerning the Kashmirian readings that might have crept into Whitney's copy for the printer. Since Roth's system of transliteration differs considerably from Whitney's, the chances for mistakes arising through confusion of the two systems were numerous; and I have taken due care to avoid them. It may here be noted that Whitney's system transliterates anusvāra before a labial by m and not by ṁ;^[28] but that in printing the Kashmirian readings, I have followed the Collation in rendering final anusvāra by ṁ (or ṅ), save before vowels. Furthermore, in making use of Roth's Collation, Whitney has habitually attempted to effect a satisfactory word-division. In many cases this is hardly practicable; and in such cases it was probably a mistake to attempt it. For examples, one may consult the readings at v. 29. 2, ‘syatamo; vi. 44. 2, sarogaṇaṁ; 109., 1, jīvātavā yati; 129. 3, vṛkṣe sārpitaḥ intending vṛkṣeṣv ār-; vii. 70. 1, dṛṣṭā rājyo, intending dṛṣṭād āj-.

The Kashmirian readings have not been verified directly from the facsimile by the editor.—As the facsimile appeared in 1901, it is proper for me to give a reason for my procedure in this matter. In fact, both my editorial work and the printing were very far advanced^[29] in 1901, so that a change of method would in itself have been questionable; but an entirely sufficient and indeed a compelling reason is to be found in the fact that it would have been and still is a task requiring very much labor and time to find the precise place of the Kashmirian parallel of any given verse of the Vulgate, a task which can no more be done en passant than can the task of editing a Prātiçākhya,—all this apart from the difficulties of the Çāradā alphabet.

Provisional means for finding Vulgate verses in the facsimile.—Whitney noted in pencil in his Collation-Book, opposite each Vulgate passage having a Kashmirian parallel, the number of the leaf of the Kashmirian text on which that parallel is found, adding a or b to indicate the obverse or the reverse of the leaf. These numbers undoubtedly refer to the leaves of Roth's Kashmirian nāgarī transcript (No. 16, Garbe) from which Roth made his Collation; but as there was no prospect of their being of any use, Whitney has not given them in this work.

One of Roth's first tasks, after the arrival of the birch-bark original, was doubtless to find the place therein corresponding to the beginning of each leaf of his Kashmirian nāgarī transcript. These places he has indicated by writing over against them on the side margin of the bark leaf the number of the leaf (with a or b) of that transcript.

This was most fortunate; for the added numbers, in Roth's familiar handwriting, although sometimes faint or covered up by a patch used in repairing the edges of the bark leaf, are for the most part entirely legible in the facsimile: and it has given me much pleasure during the last few days (to-day is April 21, 1904) to assure myself of the fact which I had previously surmised, that these pencilled numbers afford us an exceedingly useful, albeit roundabout, means of finding the place of any Kashmirian parallel in the facsimile,—useful at least until they are superseded by the hoped-for edition of an accurate transliteration of the facsimile with marginal references to the Vulgate. Whitney's pencilled reference-numbers were arranged by Dr. Ryder in the form of a table, which I have recast and given below: see pages 1013 ff.

What ought an "edition" of the Kashmirian text to be?—This question was privately discussed by Whitney and Roth in the letters^[30] exchanged between them in 1893. Whitney hoped that all that was peculiar to the Kashmirian text might be printed in transliteration in the Kashmirian order and interspersed with references to the Vulgate parallels of the remainder, also in the Kashmirian order, the whole to form an appendix to the present work. Roth's hope was that Whitney's strength might hold out long enough for him to finish this work without such a burdensome addition. Neither hope was fulfilled; and at that time, doubtless, even the thought of a facsimile reproduction was not seriously entertained. Bloomfield's difficult task of securing the needed funds once accomplished, the next step, unquestionably, was to issue the facsimile without any accessory matter. That too is now an accomplished fact; but the facsimile, apart from its large paleographic interest, is still, in default of certain accessories, a work of extremely limited usefulness. As to what should next be done, I have no doubt.

1. A rigorously precise transliteration.—First, the whole text, from A to izzard (as Roth says), should be printed in a rigorously precise transliteration. Conventional marks (other than those of the original), to indicate divisions between verses and pādas and words, need not be excluded from the transliteration, if only the marks are easily recognizable as insertions of the editor.

As to minor details, I am in doubt. In the prose parts, the transliteration might correspond page for page and line for line with the birch-bark original: the metrical parts might either be made to correspond in like manner line for line with the original; or else they might be broken up so as to show fully the metrical structure (and at the same time, with a little ingenuity, the Kashmirian vowel-fusions), in which case the beginning of every page and line of the bark leaves should be duly indicated by a bracketed number in its proper place. In case the transliteration corresponds with the original line for line throughout, then the obverse and reverse of each bark leaf might well be given together in pairs, the obverse above, and the reverse below it, on each page of the transliteration, since this would be especially convenient and would yield a page of good proportion for an Occidental book.

2. Marginal references to the Vulgate parallels.—Secondly, on the margin throughout, and opposite every Kashmirian verse that corresponds to a verse of the Vulgate, should be given the reference to the place in the Vulgate where the corresponding Vulgate verse is found.

3. Index of Vulgate verses thus noted on the margin.—Thirdly, in an appendix should be given, in the order of the Vulgate text, an index of all the Vulgate verses thus noted on the margin, with a reference to the birch-bark leaf and side (obverse or reverse—a or b) and line where its Kashmirian correspondent may be found.

These I conceive to be the essential features of a usable edition of the Kashmirian text, and I hold them to be absolutely indispensable. The text is often so corrupt that one cannot emend it into intelligibility without sacrificing too greatly its distinctive character. All conjectures, accordingly, should be relegated to a second and separately bound volume.

4. Accessory material: conjectures, notes, translations.—The accessory material of the second volume should be arranged in the form of a single series of notes and in the sequence of the Kashmirian original, and it should have such numbers and letters at the outside upper corners in the head-lines, that reference from the original to the notes and from the notes to the original may be made with the very utmost ease and celerity. This accessory material should comprehend all conjectures as to the more original Kashmirian form of manifestly corrupt words or passages, in so far as they point to readings not identical (compare the next paragraph) with those of the Vulgate; indications of word-division, especially the word-division of corrupt phrases and the resolution of the very frequent double sandhi; a running comment, proceeding verse by verse, giving any needed elucidatory matter, and explaining the rationale of the blunders of the Kashmirian version where feasible (as is often the case), pointing out in particular its excellences, and the many items in which it serves as a useful corrective of the Vulgate or confirms the conjectural emendations of the latter made in the edition of Roth and Whitney;—and all this in the light of the digested report of the variants of the parallel texts given by Whitney in the present work and in the light of the other parallels soon to be made accessible by Bloomfield's Vedic Concordance. An occasional bit of translation might be added in cases where the Kashmirian text contains something peculiar to itself or not hitherto satisfactorily treated.

For the cases (hinted at in the preceding paragraph) where corrupt Kashmirian readings point simply to readings identical with those of the Vulgate, a simple reference to the latter will sometimes suffice to show the true reading and sense of what the Kashmirian reciters or scribes have corrupted into gibberish. Thus the Kashmirian form of xii. 3. 36 b, found at folio 226 b¹ 3, is yāvantaḥ kāmān samitāu purasthāt. Apart from the aspiration (overlooked by Roth) of the prior dental of purastāt, each of these four words by itself is a good and intelligible Vedic word; but taken together, they yield far less meaning than do the famous Jabberwock verses of Through the Looking-glass.^[31] Their presence in the Kashmirian text is explained by their superficial phonetic resemblance to the Vulgate pāda yāvantaḥ kā́māḥ sám atītṛpas tā́n, of which they are a palpable and wholly unintelligent corruption. It is evident that, with the Vulgate before us, conjectural emendation of the Kashmirian text in such cases is an entirely gratuitous procedure. And as for such grammar as kenedaṁ bhūmir nihataḥ (a feminine noun, with neuter adjective pronoun and masculine predicate participle: folio 186 a¹⁵ = x. 2. 24^a),—to mend that would be to rob the Kashmirian text of its piquancy; and why should we stop with the genders, and not emend also the senseless niha- to the intelligible vihi-? Let all this be done, and we have the Vulgate text pure and simple.

10. Readings of the Parallel Texts

The texts whose readings are reported.—The principal texts included in these reports are: of the Saṁhitās, the Rig-Veda, Tāittirīya, Māitrāyaṇi, Vājasaneyi-, Sāma-Veda, and Atharva-Veda; of the Brāhmaṇas, the Āitareya, Kāuṣītaki, Tāittirīya, Çatapatha, Pañcaviṅça, and Gopatha; of the Āraṇyakas, the Āitareya and Tāittirīya; of the Upanishads, the Kāuṣītaki, Kaṭha, Bṛhadāraṇyaka, and Chāndogya; of the Çrāuta-Sūtras, the Āçvalāyana, Cān̄khāyana, Āpastamba, Kātyāyana, and Lāṭyāyana; of the Gṛhya-Sūtras, the Āçvalāyana, Çān̄khāyana, Āpastamba, Hiraṇyakeçi-, Pāraskara, and Gobhila. Other texts are occasionally cited: so the Kāṭhaka and the Kapiṣṭhala Saṁhitā, and the Jāiminīya Brāhmaṇa; and the names of some others may be seen from the List of Abbreviations, pages ci ff. I have added references to some recently edited parallel texts, without attempting to incorporate their readings into the digested report of the variants: such are the Mantra-pāṭha, von Schroeder's "Kaṭhahandschriften," and Knauer's Mānava-Gṛhya-Sūtra. Von Schroeder's edition of Kāṭhaka i. came too late. The information accessible to Whitney concerning the then unpublished Black Yajus texts was very fragmentary and inadequate; this fact must be borne in mind in connection with implied references to the Kāṭhaka and Kapiṣṭhala (cf. his notes to iii. 17; 19; 20; 21; v. 27; vii. 89).

The method of reporting the readings aims at the utmost possible accuracy.—Whitney has constantly striven for three things: that his reports should be characterized, 1. and 2., by the utmost attainable accuracy and completeness; and, 3., that they should be presented in a thoroughly well-digested form. First, as to the accuracy, little need be said. It may be well to remind the reader, however, that Whitney has used the most methodical precision in this matter, and that, accordingly, if, under a given AV. verse, he cites a parallel text without mention of variant, his silence is to be rigorously construed as meaning positively that the parallel text reads as does the AV. verse in question. As a matter of fact, I believe that it will be found possible in nearly every case to reconstruct the parallel texts with precision from the data of Whitney's reports.

It needs here to be noted that Whitney, in reporting variants from the Māitrāyaṇī, has disregarded what are (as explained by von Schroeder in his introduction, pages xxviii-xxix) mere orthographical peculiarities of that text. Accordingly, at iii. 14. 3, he treats the nā (= nas) ā́ gata of MS. as if it were na ā́ gata. Again, the MS. correspondent of iii. 19. 3 has, in saṁhitā, sváṅ, and in pada, svā́n; Whitney reports svā́ṅ, and quite properly, although it is neither the one thing nor the other. So at ii. 34. 3, he reports ṭā́ṅ, although MS. has, in s., ṭaṅ, and in p., tā́n.

The completeness of the reports far from absolute.—Secondly, as for its completeness, it may be asked whether Bloomfield's great work, the Vedic Concordance, will not show Whitney's parallels to be far from exhaustive. To this I reply that the primary purpose of Bloomfield's Concordance is to give the concordances, and to do so with as near an approach to completeness as possible, even for the less important texts, a task of which the preliminaries have required the assiduous labor of years. In Whitney's work, on the other hand, the giving of concordances is only one of many related tasks involved in his general plan, and is, moreover, only incidental to the discussion of the variants. I have tested the two works by comparison of random verses in the proof-sheets, and find (as I expected) that Bloomfield does indeed give very many references which are not given by Whitney; but that these references (apart from the Kāṭhaka) are concerned prevailingly with the numerous subsidiary or less important texts which fall within the purview of the Concordance. Whitney had excerpted all the texts, so far as published (see the list, above), which were of primary importance for his purpose. The parallels to which Bloomfield's additional references guide us will have to be reckoned with in due course by Whitney's successors; but I surmise that they are not likely upon the whole greatly to affect the sum of our critical judgments respecting the Atharvan text.^[32]

The reports are presented in well-digested form.—Thirdly, as to the form of the reports. It is one thing to give numerical references to the places where the pādas and their variants are to be found.^[33] It is another to rehearse, in full for each text concerned, the readings containing variants; and the result of this process is in a high degree space-consuming and repetitious for the author, and time-consuming and confusing for the user. It is yet another and a very different thing to compare these readings carefully, to note the points of agreement, and to state briefly and clearly the points on which they differ.^[34] The result of this last procedure is a well-digested report of the variants which is easily and quickly usable for the purpose of critical study. I call especial attention to this valuable feature of Whitney's work, partly because of its practical importance, and partly because it shows the author's power of masterly condensation and of self-restraint.

11. Whitney's Commentary: Further Discussion of its Critical Elements

Comprehensiveness of its array of parallels.—I have already called attention (p. xxxvii) to the fact that the Commentary expressly disavows any claim to finality; and have spoken briefly of its importance as a tool, and of its comprehensiveness. In respect of the comprehensiveness of its array of parallels, it answers very perfectly one of the requirements set by Pischel and Geldner in the Introduction (p. xxx) to the Vedische Studien: "Das gesamte indische Altertum kann und muss der vedischen Exegese dienstbar gemacht werden. In vorderster Linie wollen auch wir den Veda aus sich selbst erklären durch umfassenderes Aufsuchen der Parallelstellen und Combinieren zusammengehöriger aber in verschiedenen Teilen des Veda zerstreuter Gedanken." That Whitney's work will prove to be an instrument of great effectiveness in the future criticism and exegesis of the Veda I think no one can doubt. It will easily be seen that often, in the cases where the older attempts have failed, the fault is to be laid not so much to the learning and ingenuity of the scholars concerned, as to the lack of powerful tools. Such a powerful tool is this; such is Bloomfield's Concordance; and other such helpful tools are sure to be invented and made in the next few decades. The pratīka-indexes of Pertsch, Whitney, Weber, Aufrecht, and von Schroeder are admirable; and without them Whitney's work could not have been made. Their main use is to make feasible the systematic comparison of the texts one with another. This is what Whitney has done here, with the Atharvan text as starting-point, and the results of his comparison lie before us in the conveniently digested reports of the variants.

Criticism of specific readings.—Examples abound showing how the reports may be used for this purpose. They enable us to recognize the corruptness of a reading, which, although corrupt, is nevertheless to be deemed the genuine Atharvan reading, as in the case of yáç cárati at iv. 5. 5 over against the yáç ca cárati of RV. vii. 55. 6; or, again, to discover with certainty the true intention (cf. TB. ii. 4. 7¹⁰) of a lot of wavering variants, as in the case of those that disguise the sváravo mitā́ḥ of xix. 42. 1. They show us that the vastly superior tradition of the RV. corrects that of the AV. in many places (cf. the accentless asahanta of xi. 1. 2); but that the AV. occasionally scores a point even against the RV., as in the case of maghā́su at xiv. 1. 13 (RV. agā́su), or as in the case of nāu...nāu at xviii. 1. 4 (RV. no...nāu). What a puzzle is the phrase (xiv. 2. 72) janiyánti nāv ágravaḥ, 'The unmarried [plural] of us two [dual] seek a wife,' by itself, involving, as it does, a breach of the mathematical axiom that the whole is greater than any of its parts! but the comparison of RV. vii. 96. 4, with its nú for nāu, teaches us that the error lies in the nāu, even if it does not show us with certainty how that error is to be emended. Even with all the array of variants, we are (as Whitney notes at iv. 8. 1; vi. 22. 3; 31. 3) at times forced to the conclusion that certain verses were hopelessly spoiled before ever any of the various text-makers took them in hand.

Illustrations of classes of text errors.—I have already hinted at the variety of special investigations to which the mass of critical material here assembled invites. The various occasions of probable error in the transmission of Indic texts have not yet been made the object of a systematic and formal treatise. Here we have, conveniently presented, the very material needed for such an advance in the progress of Vedic criticism. By grouping suspected readings into clearly defined classes, it will become possible to recognize suspected readings as real errors with a far greater degree of certainty than ever before. Illustrations of this matter are so abundant as easily to lead us far afield; but several may be given.^[35]

Auditory errors.—A most striking example of a variation occasioned by the almost complete similarity of sound of two different readings is presented by the pratītya of AÇS. iii. 10. 11, as compared with the pratī́caḥ of AV. vi. 32. 3. Compare dyām of HGS. i. 15. 3, with jyā́m of AV. vi. 42. 1. —Confusion of surd and sonant is exemplified in the variant version of part of the familiar RV. hymn, x. 154, given at AV. xviii. 2. 14, where we have yébhyo mádhu pradhā́ ádhi, 'for whom honey [is] on the felly.' This may or may not be the genuine Atharvan reading; but it is certainly an unintelligent corruption of the pradhā́vati of the RV.: and it is very likely that we have the same blunder at vi. 70. 3, where the occasion for the corruption is palpable.^[36] The simplification of twin consonants is exemplified at xviii. 3. 3, where the editors of the Berlin text gave, with the support of all the mss. then accessible, the reading jīvā́m ṛtébhyas: that this is an error for mṛtébhyas is shown beyond all doubt by the TA. variant mṛtā́ya jīvā́m (cf. the note on p. 832).

Visual errors.—Several classes of errors are chargeable to "mistakes of the eye." Confusions such as that between pāhi and yāhi are simple enough, and are sometimes to be controlled by the evidence of oral reciters (cf. p. lxvi); but, considering the fragmentariness of our knowledge of Indic paleography, who may guess all the more remote occasions for error of this kind? —Of errors by haplography, yá ā́ste yáç cárati (just mentioned) is a good type: this is undoubtedly the true Atharvan reading, and it is undoubtedly wrong, as is shown by the meter, and the comparison of RV., which has yáç ca cárati: cf. notes to iv. 5. 5; vi. 71. 1; vii. 81. 1; xix. 42. 3; 55. 3. For a most modern case, see note to xiii. 2. 35.

Metrical faults. Hypermetric glosses and so forth.—Our suspicions of hypermetric words as glosses are often confirmed by the downright absence of those words in the parallel texts. Instances are: hástābhyām at AV. iv. 13. 7 (cf. RV. x. 137. 7); devó at RV. x. 150. 4^[37] (cf. RV. iii. 2. 8); asmábhyam at TS. ii. 6. 12² (cf. naḥ at RV. x. 15. 4); imám at AV. xiv. 2. 40 (cf. RV. x. 85. 43). —On the other hand, the damaged meter of our text often suggests a suspicion that some brief word has fallen out or that some briefer or longer or otherwise unsuitable form has been substituted for an equivalent suitable one; and the suspicion is borne out by the reading of the parallel texts. Thus in divó [vā] viṣṇa utá vā pṛthivyā́, mahó [vā] viṣṇa urór antárikṣāt, the bracketed vā's, missing at AV. vii. 26. 8, are found in their proper places in the TS. and VS. parallels. The pātu and īyús of AV. xviii. 2. 55 quite spoil the cadences of a and c, which cadences are perfect in their RV. original at x. 17. 4.

Blend-readings.—The blend-readings, as I have called them, stand in yet another group. A good example is found, at AV. xiv. 2. 18 (see note), in prajā́vatī vīrasū́r devṛ́kāmā syonā́; its genesis is clear, as is also the intrusive character of syonā́, when we compare the Kashmirian reading prajāvatī vīrasūr devṛkāmā with that of the RV., vīrasū́r devákāmā syonā́ (11 syllables). The like is true of asyá at VS. xii. 73, áganma támasas pārám asyá: cf. the oft-recurring átāriṣma támasas pārám asyá with the aganma tamasas pāram of the Kāṭhaka, xvi. 12, p. 235³. —The above-given examples suffice to show how rich is the material gathered in this work for an illuminating study of the fallibilities of human tradition in India.

12. Whitney's Translation and the Interpretative Elements of the Commentary

The Translation: general principles governing the method thereof.—The statements concerning the principles involved in the translating of the Upanishads, as propounded by Whitney in his review of a translation of those texts, apply—mutatis mutandis—so well to the translation of this Veda, that I have reprinted them (above, p. xix: cf . p. xxxvii); and to them I refer the reader.

The translation not primarily an interpretation, but a literal version.—Whitney expressly states (above, p. xix) that the design of this work is "to put together as much as possible of the material that is to help toward the study and final comprehension of this Veda"; accordingly, we can hardly deny the legitimacy of his procedure, on the one hand, in making his version a rigorously literal one, and, on the other, in restricting the interpretative constituents of the work to narrow limits. He recognized how large a part the subjective element plays in the business of interpretation; and if, as he intimates, his main purpose was to clear the ground for the interpreters yet to come, his restriction was well motived. It is, moreover, quite in accord with his scientific skepticism that he should prefer to err on the side of telling less than he knew, and not on the side of telling more than he knew: a fact which is well illustrated by his remark at viii. 9. 18, where he says, "The version is as literal as possible; to modify it would imply an understanding of it."

A literal version as against a literary one.—Let no one think that Whitney was not well aware of the differences between such a version as he has given here, and a version which (like that of Griffith) makes concessions to the demands of literary style and popular interest. Whitney's version of xviii. 1. 50, as given below, reads: 'Yama first found for us a track; that is not a pasture to be borne away; where our former Fathers went forth, there [go] those born [of them], along their own roads.' With this compare his version of 1859 (O. and L.S., i., p. 58):

Yama hath found for us the first a passage;
  that's no possession to be taken from us;
Whither our fathers, of old time, departed,
  thither their offspring, each his proper pathway.

Each version has its own quality; each method has its justification: to make a complete translation after the second method, one must inevitably waive the consideration of philological difficulties, a thing by no means licit for Whitney in such a work as this. The admirable version of Griffith illustrates the advantages of the second method, and also its inherent limitations.^[38]

Interpretative elements: captions of the hymns.—The preponderating elements of the commentary are of a critical nature, and these have been discussed by me at length in chapters i to ii of this Part I. of the General Introduction (above, pages lxiv to xciii); of the interpretative elements a few words need yet to be said. And first, it should be expressly stated that the English titles of the hymns (the captions or headings printed in Clarendon type throughout, just before the Anukramaṇī-excerpts) constitute, for the books of short hymns at least, a most important part of the interpretative element of this work. They have evidently been formulated by Whitney with much care and deliberation, and are intended by him to give briefly his view of the general purport of each hymn. In a few cases these captions were lacking, and have been supplied by me from his first draft (so at i. 35) or otherwise (so at ii. 12; v. 6; vii. 109: cf. books xv., xvi., and xviii., and p. 772, end). These captions are given in tabular form near the end of the work: see volume viii., p. 1024.

Interpretations by Whitney.—Where the text is not in disorder, a rigorously literal version is in many (if not in most) cases fairly intelligible without added interpretation. The need of such additions Whitney has occasionally, but perhaps not often, recognized. Thus after rendering the pādas i. 2. 3 ab by the words 'when the kine, embracing the tree, sing the quivering dexterous reed,' he adds, "that is, apparently, 'when the gut-string on the wooden bow makes the reed-arrow whistle.'" Similarly at vi. 125. 1. The text speaks at xviii. 1. 52 of an offense done puruṣátā: Whitney renders 'through humanity,' and adds "that is, through^[39] human frailty." Cf. note to vii. 33. 1.

It may be noted in this place (for lack of a better one) that Whitney, in reporting the conjectures or interpretations of his predecessors, passes over some in silence. Sometimes this appears to have been done intentionally and because he disapproved them. Thus at iv. 37. 3, he notes in his first draft the suggestions of BR. and OB. concerning avaçvasám; but ignores them in his second. Similarly, at ii. 14. 3, he omits mention of a translation of the verse given by Zimmer at p. 420.

Exegetical notes contributed by Roth.—It appears from the letters between Roth and Whitney that the former had written out a German version of this Veda, and that, although it was complete, its author did not by any means consider it as ready for publication. In order to give Whitney the benefit of his opinion on doubtful points. Roth made a brief commentary upon such selected words or phrases (in their proper sequence) as seemed to him most likely to present difficulties to Whitney. The result is a parcel of notes, consisting of 250 pages in Roth's handwriting, which is now in my keeping. From these notes Whitney has incorporated a considerable amount of exegetical matter into his commentary. It is yet to be considered whether the notes contain enough material unused by Whitney to warrant their publication, if this should appear upon other grounds to be advisable.

The translation has for its underlying text that of the Berlin edition.—With certain exceptions, to be noted later, the translation is a literal version of the Vulgate Atharvan text as given in the Berlin edition. For the great mass of the text, this is, to be sure, a matter of course. It is also a matter of course in cases where, in default of helpful variants to suggest an emendation of a desperate line, we are forced to a purely mechanical version, as at xii. 1. 37 a, 'she who, cleansing one, trembling away the serpent,' or at vi. 70. 2 ab. Even in the not infrequent cases where (in spite of the lack of parallel texts) an emendation is most obvious, Whitney sticks to the corrupted text in his translation, and reserves the emendation for the notes. Thus, at iv. 12. 4, ásṛk te ásthi rohatu māṅsám māṅséna rohatu, he renders 'let thy blood, bone grow,' although the change of ásṛk to asthnā́ would make all in order.

The translation follows the Berlin text even in cases of corrigible corruptions.—On the other hand, it may seem to some to be not a matter of course that Whitney should give a bald and mechanically literal version of the true Atharvan text as presented in the Berlin edition in those very numerous cases where the parallel texts offer the wholly intelligible readings of which the Atharvan ones are palpable distortions. Granting, however, that they are, although corrupt, to be accepted as the Atharvan readings, and considering that this work is primarily a technical one, his procedure in faithfully reproducing the corruption in English is entirely justified.

A few examples may be given. Whitney renders táṁ tvā bhaga sárva íj johavīmi (iii. 16. 5) by 'on thee here, Bhaga, do I call entire,' although RV.VS. have johavīti, 'on thee does every one call.' At v. 2. 8, túraç cid víçvam arṇavat tápasvān is rendered 'may he, quick, rich in fervor, send(?) all,' although it is a corruption (and a most interesting one) of the very clear line dúraç ca víçvā avṛṇod ápa svā́ḥ. So purudámāso (vii. 73. 1), 'of many houses,' although the Çrāuta-Sūtras offer purutamāso. At RV. vi. 28. 7 the cows are spoken of as 'drinking clear water and cropping good pasture,' sūyávasaṁ riçántīḥ: the AV. text-makers, at iv. 21. 7, corrupt the phrase to -se ruçántīḥ, but only in half-way fashion, for they leave the RV. accent to betray the character of their work. Even here Whitney renders by 'shining (rúçantīḥ) in good pasture.' The AV., at xviii. 4. 40, describes the Fathers as ā́sīnām ū́rjam úpa yé sácante; Whitney is right in rendering the line by 'they who attach themselves unto a sitting refreshment,' although its original intent is amusingly revealed by HGS., which has (juṣantām) māsī ’mām^[40] ūrjam uta ye bhajante, 'and they who partake of this nourishment every month.' For other instances, see the notes to iv. 21. 2 a; iii. 3. 1; iv. 16. 6 (rúçantas for ruṣántas), 8 (váruṇo); 27. 7 (viditám); vi. 92. 3 (dhā́vatu); ii. 35. 4; iii. 18. 3; iv. 2. 6; 15. 5; vii. 21. 1; and so on.

Cases of departure from the text of the Berlin edition.—These are always expressly stated by Whitney. They include, first, cases in which the Berlin edition does not present the true Atharvan text. An example may be found at xix. 64. 1, where the editors had emended wrongly to ágre and the version implies ágne. At xix. 6. 13, the editors, following the suggestion of the parallel texts, had emended to chándāṅsi the ungrammatical corruption of the AV. chándo ha (jajñire tásmāt); but since Whitney held that the latter reading "has the best right to figure as Atharvan text," his intentionally ungrammatical English 'meter were born from that' is meant to imply that reading.

Here are included, secondly, cases in which the Berlin reading, although it has to be recognized as the true Atharvan reading, is so unmanageable that Whitney has in despair translated the reading of some parallel text or an emended reading. Thus at vii. 57. 2 c it is assumed that ubhé íd asyo ’bhé asya rājataḥ is, although corrupt, the true Atharvan reading. The corruption is indeed phonetically an extremely slight distortion, for the RV. has ubhé íd asyo ’bháyasya rājataḥ; and from this the translation is made. —Other categories might be set up to suit the slightly varying relations of mss. and edition and version: cf. xix. 30. 1; xviii. 4. 87; and so on.

Whitney's growing skepticism and correspondingly rigid literalness.—At xiii. 4. 54, Whitney says: "Our rendering has at least concinnity—unless, indeed, in a text of this character, that be an argument against its acceptance." The remark is just; but one does not wonder that its author has been called der grosse Skeptiker der Sprachwissenschaft. That his skepticism grew with the progress of his work is clear from a comparison of the unrevised with the revised forms (cf. p. xxvii) of the early books. Thus at vi. 57. 2, as a rendering of jālāṣá, his manuscript at first read 'healer'; but on the revision he has crossed this out and put the Vedic word untranslated in its stead. With his skepticism, his desire for rigid literalness seems to have increased. At ii. 33. 5, the first draft translates prápada very suitably by 'fore parts of the feet'; but the second renders it by 'front feet.' Similarly, at vi. 42. 3, there is no reasonable doubt that pā́rṣṇyā prápadena ca means [I trample] 'with heel and with toe' (cf. viii. 6. 15; vi. 24. 2); but again he renders by 'front foot.' At iii. 15. 7, his prior draft reads 'watch over our life': 'life' is an unimpeachable equivalent of 'vital spirits' or prāṇā́s; but the author has changed it to 'breaths' in the second draft.

His presumable motive, a wish to leave all in the least degree doubtful interpretation to his successors, we can understand; but we cannot deny that he sometimes goes out of his way to make his version wooden. Thus he renders bhṛ, when used of skins or amulets (viii. 6. 11; 5. 13) by 'bear' instead of 'wear.' At iv. 21. 1,, he speaks of cows as 'milking for Indra many dawns,' although 'full many a morning yielding milk for Indra' can hardly be called too free. Cf. his apt version of úttarām-uttarām sámām at xii. 1. 33, 'from one year to another,' with that given at iii. 10. 1; 17.4, 'each further summer.' In a charm to rid the grain of danger, vi. 50. 1 d, 'make fearlessness for the grain' is needlessly inept. It is easy for Sanskritists, but not for others, to see that 'heroism' (vīryà), as used of an herb at xix. 34. 8, means its 'virtue' (and so he renders it at xii. 1. 2); that 'bodies' of Agni at xix. 3. 2 are his 'forms' (çivās or ghorās); and so on; but to others, such versions will hardly convey the intended meaning. The fact that svastíbhis, in the familiar refrain of the Vasiṣṭhas, is a plural, hardly justifies the infelicity of using such a plural as 'well-beings' to render it at iii. 16. 7; and some will say the like of 'wealfulnesses ' (iv. 13. 5), 'wealths,' and 'marrows.'

It lies entirely beyond the province of the editor to make alterations in matters of this kind. It is perhaps to be regretted that these infelicities, which do not really go below the surface of the work, are the very things that are the most striking for persons who examine the book casually and without technical knowledge; but the book is after all primarily for technical study.

Poetic elevation and humor.—The places in which the AV. rises to any elevation of poetic thought or diction are few indeed. Some of the funeral verses come as near it as any (among them, notably, xviii. 2. 50); and some of the philosophic verses (especially of x. 8 under Deussen's sympathetic treatment) have an interest which is not mean. The motive of xix. 47 is an exceptionally coherent and pleasing one. I presume that the idea of sending the fever as a choice present to one's neighbors (v. 22. 14) is intended to be jocose. Witchcraft and healing are serious businesses. If there is anything else of jocular tone in this extensive text, I do not remember that any one has recognized and noted it. The gravity of Whitney's long labor is hardly relieved by a gleam of humor save in his introduction to ii. 30 and his notes to vi. 16. 4 and 67. 2 and x. 8. 27, and the two cited at p. xcvii, line 4 from end, and p. xciv, l. 23.

13. Abbreviations and Signs explained

General scope of the list.—The following list is intended not only to explain all the downright or most arbitrary abbreviations used in this work, but also to explain in the shortest feasible way all such abbreviated designations of books and articles as are more or less arbitrary. The former generally consist of a single initial letter or group of such letters; the latter, of an author's name or of the abbreviated title of a work.

The downright abbreviations.—These are for the most part identical with those used by Whitney in his Grammar and given and explained by him on p. xxvi of that work: thus AA. = Āitareya-Āraṇyaka. —Whitney's omission of the macron proper to the A in AA., AB., AÇS., AGS., BAU., and TA. was doubtless motived by a purely mechanical consideration, the extreme fragility of the macron over a capital A; that he has not omitted it in Āpast. or Āp. is a pardonable inconsistency. —The sigla codicum are explained at p. cix, and only such of them are included here as have more than one meaning: thus, W. = Wilson codex and also = Whitney.

Abbreviated designations of books and articles.—For these the list is intended to give amply sufficient and clear explanations, without following strictly any set of rules of bibliographers. In the choice of the designations, brevity and unambiguousness have been had chiefly in mind. —An author's name, without further indication of title, is often used arbitrarily to mean his most frequently cited work. Thus "Weber" means Weber's Indische Studien. With like arbitrariness are used the names of Bloomfield, Caland, Florenz, Griffith, Grill, Henry, Ludwig, Muir, Winternitz, and Zimmer: cf. the list. —Where two coördinate reference-numbers, separated by a comma, are given (as in the case of Bloomfield, Grill, and Henry), the first refers to the page of the translation, and the second to the page of the commentary. Of similar numbers, separated by "or" (as on p. 286), the first refers to the original pagination, and the second to the pagination of the reprint.^[41] Explanation of arbitrary signs.—The following signs (and letters) are used in the body of this work more or less arbitrarily.

Parentheses are used in the translation to enclose the Sanskrit original of any given English word (see above, p. xx), such indications being often most acceptable to the professional student. For numerous instances, see xii. 1, where the added bhū́mi or pṛthivī́ (both are added in vs. 7) shows which of these words is meant by the English earth. They are also used to enclose an indication of the gender (m. f. n.) or number (du. pl.) of a Vedic word whose gender or number cannot otherwise be shown by the version.

Square brackets are employed to enclose some of the words inserted in the translation for which there is no express equivalent in the original.

Ell-brackets, or square brackets minus the upper horizontal stroke (thus: ⌊ ⌋) were devised by the editor to mark as portions of this work for which Whitney is not responsible such additions or changes as were made by the editor (cf. p. xxviii, end). These types were devised partly because the usual parentheses and brackets were already employed for other purposes, and partly because they readily suggest the letter ell, the initial of the editor's name.

Hand.—In order to avoid the expense of alterations in the electroplates, all considerable additions and corrections have been put together on pages 1045-46, and reference is made to them in the proper places by means of a hand pointing to the page concerned (thus, at p. 327, line 11: ☞ See p. 1045).

The small circle (thus: ॰) represents the avagraha or division-mark of the pada-texts. This use of the circle is common in the mss. (as explained at p. cxxii) and has been followed in the Index Verborum (see p. 4).

The Italic colon (:) is employed as equivalent of the vertical stroke used in nāgarī to separate individual words or padas. Both circle and colon are used in the note to vi. 131. 3. I regard both the circle and the colon as extremely ill adapted for the uses here explained.

The letters a, b, c, d, e, f, etc., when set, as here, in Clarendon type, are intended to designate the successive pādas of a Vedic stanza or verse.

Alphabetic list of abbreviations.—The downright abbreviations and the abbreviated designations of books and articles follow here, all in a single alphabetically arranged list. AA. = Āitareya-Āraṇyaka. Ed. Bibl. Ind. 1876.

AB. = Āitareya-Brāhmaṇa. Ed. Th. Aufrecht. Bonn. 1879.

Abh. = Abhandlungen.

AÇS. = Āçvalāyana-Çrāuta-Sūtra. Ed. Bibl. Ind. 1874.

In the ed., the 12 adhyāyas of the work are divided into two Hexads (ṣaṭkas), a Prior and a Latter, and the numbering of those of the Latter begins anew with 1. In Whitney's citations, the numbers run from i. to xii.: thus (in his note to iv. 39. 9) ACS. II. ii. 14. 4 is cited as viii. 14. 4.

AGS. = Āçvalāyana-Gṛhya-Sūtra. Ed. A. F. Stenzler in Sanskrit and German. Leipzig. 1864-5. Ed. also in Bibl. Ind. 1869.

AJP. = American Journal of Philology. Ed. B. L. Gildersleeve. Baltimore. 1880-.

Ak. = Akademie.

Amer. = American.

Anukr. = Anukramaṇī or, sometimes the author of it.

ĀpÇS. or Āp. = Āpastamba-Çrāuta-Sūtra. Ed. R. Garbe in Bibl. Ind. 1882-1902. 3 vol's.

ĀpGS. = Āpastambīya-Gṛhya-Sūtra. Ed. M. Winternitz. Vienna. 1887.

APr. = Atharva-Veda Prātiçākhya. Ed. W. D. Whitney in JAOS. (vii. 333-615). 1862. Text, translation, and elaborate notes.

Aufrecht. Das XV. Buch des AV. Text, translation, and notes. Ind. Stud. i. 121-140. 1849. See below, p. 769.

AV. = Atharva-Veda. AV. = also Atharva-Veda-Saṁhitā. Ed. by R. Roth and W. D. Whitney. Berlin. 1855-6. Ed. also by Shankar Pandurang Pandit. Bombay. 1895-8. 4 vol's,

-av. = -avasāna: see explanation following.

In the excerpts from the Anukr., the Sanskrit eka-, dvi-, tri-, etc., constantly recurring in composition with avasāna and pada, are abbreviated by the Arabic numerals 1, 2, 3, etc. Thus, at p. 727, the excerpt 3-av. 6-p. atyaṣṭi may be read as try-avasānā ṣaṭ-padā ’tyaṣṭiḥ.

B. = Brāhmaṇa.

BAU. = Bṛhad-Āraṇyaka-Upaniṣad. Ed. Otto Böhtlingk. Leipzig. 18S9. Other ed's: Calc, Bo., Poona.

Bāudhāyana = Bāudhāyana-Dharma-Çāstra. Ed. E. Hultzsch. Leipzig. 1884.

Bergaigne: see Rel. Véd.

Bergaigne-Henry, Manuel = Manuel pour étudier le Sanscrit védique. By A. Bergaigne and V. Henry. Paris. 1890.

Bibl. Ind. = Bibliotheca Indica, as designation of the collection of texts and translations published by the Asiatic Society of Bengal in Calcutta.

Bl. = Bloomfield.

Bloomfield (without further designation of title) = Hymns of the AV., together with extracts from the ritual books and the commentaries, translated by Maurice Bloomfield. Oxford. 1897. This book is vol. xlii. of SBE.

In this work Bl. sums up a very large part, if not all, of his former "Contributions" to the exegesis of this Veda, which he had published in AJP. (vii., xi., xii., xvii.), JAOS. (xiii., xv., xvi.—PAOS. included), ZDMG. (xlviii.). The "Contributions" are cited by the abbreviated designations (just given) of the periodicals concerned.

Bloomfield, Atharvaveda = his part, so entitled, of the Grundriss. 1899.

Bo. = Bombay.

BR. = Böhtlingk and Roth's Sanskrit-Wörterbuch. Published by the Imperial Russian Academy of Sciences. St. Petersburg. 1852-1875. Seven vol's. Often called the (Major) (St.) Petersburg Lexicon. Cf. OB.

Caland (without further indication of title) = Altindisches Zauberritual. Probe einer Uebersetzung der wichtigsten Theile des Kāuçika-Sūtra (kaṇḍīkās 7-52). By W. Caland. Amsterdam. 1900. From the Verhandelingen der Koninklijke Ak. van Wetenschappen te Amsterdam. Deel III. No. 2.

Caland, Todtengebräuche = Die Altindischen Todten- und Bestattungsgebräuche. Amsterdam. 1896. See p. 813.

Caland, Totenverehrung = Ueber Totenverehrung bei einigen der Indo-Germanischen Volker. Amsterdam. 1888.

Caland, Pitṛmedha-Sūtras = The Pitṛmedha-Sūtras of Bāudhāyana, Hiraṇyakeçin, Gāutama. Leipzig. 1896.

Calc. = Calcutta or Calcutta edition.

ÇB. = Çatapatha-Brāhmaṇa. Ed. A. Weber. Berlin. 1855.

ÇÇS. = Çān̄khāyana-Çrāuta-Sūtra. Ed. A. Hillebrandt. Bibl. Ind. 1888.

ÇGS. = Çān̄khāyana-Gṛhya-Sūtra. Ed. H. Oldenberg in Ind. Stud. (xv. 1-166). 1878. Skt. and German.

ChU. = Chāndogya-Upaniṣad. Ed. O. Böhtlingk. Leipzig. 1889. Skt. and German. Ed. also in Bibl. Ind., Bo., and Poona.

Collation-Book = manuscript volumes containing Whitney's fundamental transcript of the AV. text and his collations, etc. For details, see p. cxvii.

comm. = the commentary on AV. (ascribed to Sāyaṇa and published in the Bombay ed.); or, the author thereof.

Daç. Kar. = Daça Karmāṇi, a paddhati to certain parts of the Kāuç. See Bl's introduction, p. xiv.

Delbrück. Altindische Syntax. Halle. 1888.

Denkschr. = Denkschriften.

Deussen, Geschichte = Allgemeine Geschichte der Philosophie mit besonderer Berücksichtigung der Religionen. By Paul Deussen. Leipzig. The first vol. (part 1, 1894: part 2, 1899) treats of the philosophy of the Veda and of the Upaniṣads.

Deussen, Upanishads = Sechzig Upanishad's des Veda aus dem Sanskrit übersetzt und mit Einleitungen und Anmerkungen versehen. Leipzig. 1897.

Dhanvantari = Dhanvantarīya-Nighantu. Some references are to the Poona ed.; Roth's references are, I presume, to his transcript described by Garbe, Verzeichniss der (Tübinger) Indischen Handschriften, No. 230.

du. = dual.

ed. = edition (of) or editor or edited by or in.

et al. = et alibi.

f. or fem. = feminine.

Festgruss an Böhtlingk = Festgruss an Otto von Böhtlingk zum Doktor-Jubiläum, 3. Februar 1888, von seinen Freunden. Stuttgart. 1888.

Festgruss an Roth = Festgruss an Rudolf von Roth zum Doktor-Jubiläum, 24. August 1893, von seinen Freunden und Schülern. Stuttgart. 1893.

Florenz = his German translation of AV. vi. 1-50, with comment, in vol. xii. of Bezzenberger's Beiträge. Göttingen. 1887. See below, p. 281.

GB. = Gopatha-Brāhmaṇa. Ed. Bibl. Ind. 1872.

Geldner: see Siebenzig Lieder and Ved. Stud.

Ges. = Gesellschaft.

GGA. = Göttingische Gelehrte Anzeigen.

GGS. = Gobhila-Gṛhya-Sūtra. Ed. Friedrich Knauer. Leipzig. 1885. Text, transl., and comment: in 2 parts.

Grammar or (Skt.) Gram, or Gr. = Whitney's Sanskrit Grammar, 2d ed. Leipzig and Boston. 1889. There is a 3d ed. (1896), which is essentially a reprint of the 2d.

Grassmann = Rig-Veda. Uebersetzt etc. Leipzig. 1876—7. 2 vol's.

Griffith = The hymns of the AV., translated, with a popular commentary. By Ralph T. H. Griffith. Benares and London. 1895-6. 2 vol's. Cf. p. xcv, above.

Grill = Hundert Lieder des AV. By Julius Grill. 2d ed. Stuttgart. 1888. Translation and comment.

Grohmann = Medicinisches aus dem AV., mit besonderem Bezug auf den Takman. In Ind. Stud. (ix. 381-423). 1865.

Grundriss = Grundriss der Indo-Arischen Philologie und Altertumskunde. Begründet von Georg Bühler. Fortgesetzt von F. Kielhorn. Strassburg. 1896-.

Gurupūjākaumudī = Festgabe zum fünfzigjährigen Doctorjubiläum, Albrecht Weber dargebracht von seinen Freunden und Schülern. Leipzig. 1896.

h. = hymn or hymns.

Hāla's Saptaçataka: reference is made to A. Weber's treatise thereon (Leipzig. 1870) and to his edition thereof (Leipzig. 1881).

Hardy = Die Vedisch-brahmanische Periode der Religion des alten Indiens. By Edmund Hardy. Münster in Westphalia. 1893.

Henry (without further indication of title) = Victor Henry's French translation of books vii.-xiii. of the AV., with commentary. It appeared in 4 vol's (Paris, Maisonneuve) as follows: book xiii., 1891; book vii., 1892; books viii.-ix., 1894; books x.-xii., 1896. For precise titles, see below, pages 388, 471, 562, 708.

HGS. = Hiraṇyakeçi-Gṛhya-Sūtra. Ed. J. Kirste. Vienna. 1889.

Hillebrandt, Veda-Chrestomathie. Berlin. 1885.

Hillebrandt, Ved. Myth. = his Vedische Mythologie, Breslau. 1891-1902.

Hillebrandt, Ritual-litteratur = his part of the Grundriss. 1897.

IF. = Indogermanische Forschungen. Ed. by Brugmann and Streitberg. Strassburg. 1892-.

IFA. = Anzeiger für Indogermanische Sprach- und Altertumskunde. "Beiblalt" to IF.

Index Verborum = Whitney's Index Verborum to the published Text of the AV. Issued as JAOS., vol. xii. New Haven, Conn. 1881.

Ind. Streifen = A. Weber's Indische Streifen. Berlin and Leipzig. 1868. 1869. 1879. 3 vol's.

Ind. Stud. = Indische Studien. Ed. Albrecht Weber. Volume i. (Berlin. 1849-50) to volume xviii. (Leipzig. 1898).

JA. = Journal Asiatique. Publié par la Société Asiatique. Paris. 1822-. Cited by series, vol., and page.

JAOS. = Journal of the American Oriental Society. New Haven, Conn. 1843-.

JB. = Jāiminīya-Brāhmaṇa. Cited from Whitney's transcript, described by him at JAOS. xi., p. cxliv, = PAOS. for May, 1883.

JRAS. = Journal of the Royal Asiatic Society of Great Britain and Ireland. London. 1834-.

JUB. = Jāiminīya-Upaniṣad-Brāhmaṇa. Ed. H. Oertel in JAOS. (xvi. 79-260). 1896 (presented, 1893). Text, transl., notes.

K. = Kāṭhaka; or, sometimes the codex K. Von Schroeder's ed. of book i. of the Kāṭhaka appeared in Leipzig, 1900.

Kap. = Kapiṣṭhala-Saṁhitā.

KaṭhaB. = Kaṭha-Brāhmaṇa: see below, p. 903, ¶ 2.

Kaṭha-hss. = Die Tübinger Kaṭha-Handschriften und ihre Beziehung zum TA. By L. von Schroeder. Sb. der k. Ak. der Wiss. in Wien. Vol. 137. Vienna. 1898.

Kāuç. = The Kāuçika-Sūtra of the AV. With extracts from the commentaries of Dārila and Keçava. Ed. Maurice Bloomfield. Issued as vol. xiv. of JAOS. 1890. For concordance of two methods of citing this text, see p. 1012.

KB. = Kāuṣītaki-Brāhmaṇa. Ed. B. Lindner. Jena. 1887.

KBU. = Kāuṣītaki-Brāhmaṇa-Upaniṣad. Ed. E. B. Cowell. Bibi. Ind. 1861. Text and translation.

KÇS. = Kātyāyana-Çrāuta-Sūtra. Ed. A. Weber. Berlin. 1859.

Keç. = Keçava or his scholia on Kāuç. See Bl's introd., p. xvi.

Kuhn's Pāli-gram. = Beitrage zur Pāligrammatik von Ernst W. A. Kuhn. Berlin. 1875.

KZ. = Zeitschrift für vergleichende Sprachforschung....begründet von Th. Aufrecht und A. Kuhn. Berlin. (Now Gütersloh.) 1851-.

Lanman, Noun-Inflection = Noun-Inflection in the Veda. By C. R. Lanman. In JAOS. (x. 325-601). 1880.

Lanman, (Skt.) Reader = Sanskrit Reader, with Vocabulary and Notes. By C. R. Lanman. Boston. 1888.

LÇS. = Lāṭyāyana-Çrāuta-Sūtra. Ed. Bibl. Ind. 1872.

Ludwig (without further indication of title) = vol. iii. of his Der Rigveda in 6 vol's. Prag. 1876-88.

Vol's i.-ii. contain the translation of the RV., and iv.-v. contain the comment. Vol. iii. (1878) contains many translations from AV. and is entitled Die Mantra-litteratur und das alte Indien als Einleitung zur Ueb. des RV.—Where reference to the transl. of the RV. equivalent (in vol. i. or ii.) of an AV. passage is intended, that fact is made clear (as at p. 118 top, 113, 248, etc.).

Ludwig, Kritik des RV.-textes: see p. 860.

m. = masculine.

Macdonell, Ved. Mythol. = his Vedic Mythology in the Grundriss. 1897.

MB. = Mantra-Brāhmaṇa (of the SV.). Cited from ed. in periodical called Ushā. Calcutta. 1891.

MBh. = Mahā-Bhārata. Citations refer to Bo. ed. (or ed's), or to both Bo. and Calc. ed's.

Mém. Soc. Ling. = Mémoires de la Société de linguistique de Paris.

MGS. = Mānava-Gṛhya-Sūtra. Ed. F. Knauer. St. Petersburg. 1897.

MP. = Mantra-Pāṭha: or, the Prayer Book of the Āpastambins. Ed. M. Winternitz. Oxford. 1 897. Part of the material of MP. had already been given in the work cited below under Winternitz, Hochzeitsrituell, as explained also below, p. 738.

MS. = Māitrāyaṇī-Saṁhitā. Ed. L. von Schroeder. Leipzig. 1881-6.

Muir (without further indication of title) = OST., which see.

Muir, Metrical Translations from Sanskrit Writers. London. 1879.

N. = North.

n. = note; or, sometimes neuter.

Nāigeya-kāṇḍa of SV.: see below, under SV.

Nakṣ. or Nakṣ. K. = Nakṣatra-Kalpa. See Bl's introd. to Kāuç., p. xix.

Noun-Inflection: see above, under Lanman.

O. and L. S. = Oriental and Linguistic Studies. By W. D. Whitney. New York. 1873. 1874. 2 vol's.

OB. = Otto Böhtlingk's Sanskrit-Wörterbuch in kürzerer Fassung. St. Petersburg. 1879-89. Seven vol's. Often called the Minor (St.) Petersburg Lexicon. Cf. BR.

Oldenberg, Die Hymnen des RV. Band I. Metrische und textgeschichtliche Prolegomena. Berlin. 1888.

Oldenberg, Die Religion des Veda. Berlin. 1894.

Omina und Portenta: see under Weber.

OST. = Original Sanskrit Texts. Translated by John Muir. London. 1868-73. 5 vol's.

p. = pada-pāṭha.

-p. (as in 3-p., 4-p.) = păda (in the sense of subdivision of a stanza): see explanation above, under -av.

Pāipp. = Pāippalăda or Kashmirian AV. For details concerning the collation and its sources and the birch-bark original and the facsimile, see above, pages lxxx ff.

Pāṇ. = Pāṇini's Grammar.

Pandit, Shankar Pandurang: see below, under SPP.

PAOS. = Proceedings of the American Oriental Society.

They were formerly issued (with pagination in Roman numerals to distinguish them from the Journal proper) as appendixes to be bound up with the volumes of the Journal; but they were also issued in separate pamphlets as Proceedings for such and such a month and year. The citations below are so given that they can readily be found in either issue.

Pariç. = AV. Pariçiṣṭa: cf. Bl's introd. to Kāuç., p. xix.

PB. = Pañcaviṅça-Brāhmaṇa or Tāṇḍya-Mahā-brāhmaṇa. Ed. Bibl. Ind. 1870-74. 2 vol's.

Peterson, Hymns from the RV. Ed. with Sāyaṇa's comm., notes, and a transl. by Peter Peterson. Bombay. 1888.

Pet. Lex. = the Major St. Petersburg Lexicon. See BR.

Pet. Lexx. = the two St. Petersburg Lexicons, Major and Minor. See BR. and OB.

PCS. = Pāraskara-Gṛhya-Sūtra. Ed. A. F. Stenzler. Leipzig. 1876. 1878. Skt. and German.

Pischel, Gram, der Prākrit-sprachen = his part, so entitled, of the Grundriss. 1900.

Pischel, Ved. Stud.: see below, under Ved. Stud.

p. m. = prima manu.

Poona ed. = ed. of the Ānanda-Āçrama Series.

Ppp. = Pāippalāda AV.: see above, under Pāipp.

Prāt. or Pr. = Prātiçākhya of the AV.: see above, under APr.

Proc. = Proceedings.

R. = Roth; or, sometimes the codex R.

Rājan. = Rājanighaṇṭu. Cited no doubt from Roth's own ms., now Tübingen ms. 176. There is a Poona ed.

Rel. Véd. = Abel Bergaigne's La Religion védique d'après les hymnes du RV. Paris. 1878-83. 3 vol's. Bloomfield made an Index of RV. passages therein treated. Paris. 1897.

Rev. = Review.

Roth, Zur Litteratur und Geschichte des Weda. Stuttgart. 1846.

Roth, Ueber den Atharva Veda. Tübingen. 1856.

Roth, Der Atharvaveda in Kaschmir. Tübingen. 1875.

Roth, Ueber gewisse Kürzungen des Wortendes im Veda. Verhandlungen des VII. Internationalen Orientalisten-Congresses. Vienna. 1887.

Roxburgh, Flora Indica: the citations by vol. and page refer to Carey's ed. of 1832; but these can easily be found in the margin of the Calc. reprint of 1874.

RPr. or RV. Prāt. = RV. Prātiçākhya. Ed. Max Müller. Leipzig. 1869. Also by A. Regnier in JA.

RV. = Rig-Veda or Rig-Veda-Saṁhitā. Ed. Th. Aufrecht. Also by Max Müller.

RW. = Roth and Whitney.

s. = saṁhitā-pāṭha.

Sächsische Ber. = Berichte der königl. Sächsischen Ges. der Wiss.

ṢB. = Ṣaḍviṅça Brāhmaṇa. Cited presumably from ed. of Jībānanda Vidyāsāgara. Calc. 1881. Ed. of part by K. Klemm. Gütersloh. 1894.

Sb. = Sitzungsberichte. Those of the Berlin Ak. are usually meant.

SBE. = Sacred Books of the East. Transl. by various Oriental Scholars and ed. by F. Max Müller. Oxford. 1879-1904. 49 vol's.

Scherman, Philosophische Hymnen = Phil. Hymnen aus der RV.- und AV.-Saṁhitā verglichen mit den Philosophemen der älteren Upanishad's. Strassburg. 1887.

schol. = scholia of Dārila or of Keçava or of both, on Kāuç.: see Bl's introd., p. xi and p. xvi.

von Schroeder: see above, Kaṭha-hss., and below, Zwei Hss.

Siebenzig Lieder des RV. Uebersetzt von Karl Geldner und Adolf Kaegi. Mit Beiträgen von R. Roth. Tübingen.

s.m. = secunda manu.

Speyer, Vedische Syntax = his part of the Grundriss, entitled Vedische und Sanskrit Syntax. 1896.

SPP. = Shankar Pandurang Pandit as editor of the Bombay edition of the AV. It is entitled: Atharvavedasaṁhitā with the Commentary of Sāyaṇācārya. 1895-8. 4 vol's.

Sūrya-Siddhānta = Translation of the Sūrya-Siddhānta, a Text-book of Hindu Astronomy; with Notes; etc. In JAOS. (vi. 141-498). 1860.

SV. = Die Hymnen des Sāma-Veda. Ed. Th. Benfey. Leipzig. 1848. Text, transl., glossary.

The verses of the Prior ārcika are cited, by the numbers in natural sequence, as i. 1 to i. 585; similarly, those of the Latter ārcika, as ii. 1 to ii. 1225.—The verses of the Nāigeya supplement to the Prior ārcika are cited as SV. i. 586 to i. 641, and as edited by S. Goldschmidt in the Monatsbericht der k. Ak. der Wiss. zu Berlin, session of Apr. 23, 1868. Cf. note to AV. iv. 26. 1 and to xiii. 2. 23.

TA. = Tāittirīya-Āraṇyaka. Ed. Bibl. Ind. 1872. There is also a Poona ed.

TB. = Tāittirīya-Brāhmaṇa. Ed. Bibl. Ind. 1859-? There is also a Poona ed.

TPr. = Tāittirīya-Prātiçākhya. Ed. W. D. Whitney. In JAOS. (ix. 1-469). 1871.

Trans. = Transactions.

TS. = Tāittirīya-Saṁhitā. Ed. A. Weber. In Ind. Stud., vol's xi. and xii. Leipzig. 1871-2. There is also a Poona ed.

Vāit. = Vāitāna-Sūtra. Ed. R. Garbe. London. 1878. German transl. by him. Strassburg. 1878.

Ved. Stud. = Vedische Studien. Von R. Pischel und K. F. Geldner. Stuttgart. vol's. 1889. 1897. 1901.

VPr. = Vājasaneyi-Prātiçākhya. Ed. A. Weber. In Ind. Stud. (iv.). 1857-8. Skt. and German.

VS. = Vājasaneyi-Saṁhitā. Ed. A. Weber. Berlin. 1852.

vs. (never v., which is used as meaning 5) = verse: vss. = verses: of., for example, line 2 of note to iv. 12. 1.

W. = Whitney; or, sometimes the codex W.

Weber (without further indication of title) = Weber's Indische Studien: see above, Ind. Stud.

Weber, Omina und Portenta: in Abh. der k. Ak. der Wiss. for 1858. Berlin. 1859.

Weber, Rājasūya = Ueber die Königsweihe, den Rājasūya: in Abh. der k. Ak. der Wiss. for 1893. Berlin. 1893.

Weber, Sb.: for the meaning in book xviii., see below, p. 813.

Weber, Vājapeya = Ueber den Vājapeya: in Sb. der k. Ak. der Wiss. for 1892, pages 765-813. Berlin. 1892.

Weber, Vedische Beiträge.

Under this title was issued a series of 9 articles in Sb. der k. Ak. der Wiss. zu Berlin, from 1894 to 1901. They are usually cited by Sb. and the date. For the AV., the most important is no. 4 (1895, concluded 1896), treating book xviii., as explained below, p. 813.

Weber's Translations of books i.-v. and xiv. and xviii.: for these, see p. cvii.

Wh. or Whitney, Grammar: see above, under Grammar.

Whitney, Index Verborum: see above, under Index.

Whitney, O. and L.S.: see above, under O. and L.S.

Whitney, Roots = The Roots, Verb-forms, and primary Derivatives of the Sanskrit Language. Leipzig. 1885.

Whitney's other contributions relating to the AV.: for some of these, see Preface, pages xxiii, xxv, xxvi.

Winternitz or (in book xiv.: cf. below, p. 738) simply Wint. = his Hochzeitsrituell in the Denkschriften der k. Ak. der Wiss., vol. xl. Vienna. 1892.

Wiss. = Wissenschaften.

WZKM. = Wiener Zeitschrift für die Kunde des Morgenlandes. Vienna. 1887-.

ZDMG. = Zeitschrift der Deutschen Morgenländischen Gesellschaft. Leipzig. 1847-.

Zimmer (without further indication of title) = his Altindisches Leben. Berlin. 1879.

Zwei Hss. = Zwei Handschriften der k. k. Hofbibliothek in Wien mit Fragmenten des Kāṭhaka. By von Schroeder. In Sb. der k. Ak. der Wiss. for 1895 (38 pages). Vol. cxxxiii. Vienna. 1896.

14. Tabular View of Translations and Native Comment

Previous translations.—Native commentary.—It may prove useful to have, in convenient tabular form, a list of the most important or comprehensive previous translations, with dates; and also a list of those parts of the text upon which the native commentary has been published in the Bombay edition. The dates are taken from the title-pages of the volumes concerned; the dates of the prefaces, or of the parts of the volumes concerned, are sometimes considerably earlier. For bibliographical details, see the List, pages ci-cvi. The braces at the right show which of SPP's four volumes contains the text, or the text with comment,. of any given book.

I. Translation of the whole text.

Griffith, 1895, 1896: see p. cii.

II. Translations of a mass of selected hymns.

Bloomfield, 1897: see p. ci.⁠Ludwig, 1878: see p. civ.⁠Grill, 1888: see p. cii.

III. a. Translations of single books.						III. b. Books with comment of "Sāyaṇa."
Book	i.	Weber,	Indische Studien,	iv.	1858.	Book i., entire.	$\scriptstyle {\left.{\begin{matrix}\ \\\\\ \\\ \\\ \ \end{matrix}}\right\}\,}$ SPP's vol. i.
	ii.	"	"⁠"	xiii.	1873.	Book ii., entire.
	iii.	"	"⁠"	xvii.	1885.	Book iii., entire.
	iv.	"	"⁠"	xviii.	1898	Book iv., entire.
	v.	"	"⁠"	"	"
	vi.	1-50.	Florenz (see p. 281).		1887.	Book vi., entire	$\scriptstyle {\left.{\begin{matrix}\ \\\\\ \\\ \\\ \ \end{matrix}}\right\}\,}$ SPP's vol. ii.
	vii.	Henry,	Le livre vii.		1892.	Book vii., entire.
	viii.	"	Les livres viii et ix.		1894.	Book viii., 1-6.
	ix.	"	"⁠"		"
	x.	"	Les livres x, xi et xii.		1896.
	xi.	"	"⁠"		"	Book xi., entire	$\scriptstyle {\left.{\begin{matrix}\ \\\\\ \\\ \\\ \ \end{matrix}}\right\}\,}$ SPP's vol. iii.
	xii.	"	"⁠"		"
	xiii.	"	Les hymnes Rohitas.		1891.
	xiv.	Weber,	Indische Studien, v.		1862.
	xv.	Aufrecht,	Indische Studien, i.		1850.
	xvi.
	xvii.					Book xvii., entire.	$\scriptstyle {\left.{\begin{matrix}\ \\\\\ \\\ \ \end{matrix}}\right\}\,}$ SPP's vol. iv.
	xviii.	Weber,	Sitzungsberichte.		1895-6.	Book xviii., entire.
	xix.					Book xix., entire.
	xx.					Book xx., 1-37.

Chronologic sequence of previous translations and discussions.—In judging between the translations or opinions of different exegetes, it is desirable to know their chronological sequence. In giving the detailed bibliographical minutiae below, at the beginning of each hymn, I have always endeavored to arrange them chronologically; but the following brief table in addition will not be superfluous. The difference in time of the printing of the translations of Griffith and Bloomfield and Henry (x.-xii.) was so small that they must have been each independent of the others. For the places of publication etc., see the List, pages ci-cvi.

1850. Aufrecht, book xv.

1858. Weber, book i.

1862. Weber, book xiv.

1872. Muir, select., OST. v.

1873. Weber, 2d ed., book ii.

1878. Ludwig, selections.

1879. Zimmer, selections.

1885. Weber, book iii.

1887. Scherman, selections.

1887. Florenz, book vi. 1-50.

1888. Grill, 2ded., 100 hymns.

1891. Henry, book xiii.

1892. Henry, book vii.

1894. Deussen, Geschichte, i. 1

Henry, books viii.-ix.

1895. SPP's text, vol's i.-ii.

1895. Griffith, books i.-ix.

Weber, book xviii. 1-2.

1896. Weber, book xviii. 3-4.

Griffith, books x.-xx.

Henry, books x.-xii.

1897. Bloomfield, selections.

1898. Weber, books iv.-v.

SPP's text, vol's iii.-iv.

1. Doubtless the pada-pāṭha also is an ancillary text, and these headings are therefore not quite logical; but they will serve.
2. Here it is to be noted that, by reason of breakage of type, the last part of the "run" (as the printers say) is not always like the first; in other words, that not every copy of the Berlin edition is like every other (cf. note to i. 18. 4).
3. Thus in the note to iii. 7. 2, "a couple of SPP's mss." means two men, not books. Cf. notes to xix. 32. 8; 33. 1.
4. In discussing iii. 23. 6, Whitney says in the Prāt. (p. 442), "Every codex presents dyduḥ"; while in this work (below, p. 128) he reports O. as reading dyāuṣ. Since "every codex" means every codex collated before publication, this is no contradiction.
5. At iv. 26. 5, SPP. reports 8 out of 13 saṁhitā authorities, Sm. and V. being given on both sides, and of course wrongly on one or the other.
6. Thus at xix. 20. 4 b, vármā́har várma sū́ryaḥ, the comm. reads agnir for ahar, and is supported therein by AÇS. and Āp
7. A remark in his comment on ii. 4. 1 (Bombay ed., i, 210¹⁶), to the effect that the jan̄giḍa is a kind of tree familiarly known in Benares, suggests the surmise that his bhāṣya may have been written in that city.
8. The pada-text of book xix., which swarms with blunders (cf. p. 895, end, 896, top), is clearly very different both in character and origin from the pada-text of books i.-xviii.
9. If Whitney is right in supposing that vi. 1. 3 is a spoiled gāyatrī the first pāda of which ends with savitā́, then I believe that the accentlessness of sāviṣat is to be regarded as pointing to a false resolution and that the pada-text should be amended to ā́॰sāviṣat; but cf. vii. 73. 7 c and Çākalya's resolution of its RV. parallel.
10. In some of these cases, the rationale of the error is discernible: cf. the notes, especially the note to xiii. 3. 17.
11. Cf. the confusion between pātv ṛṣabhás and pātu vṛṣabhás at xix. 27. 1, Bombay ed.
12. Cf. note to xix. 50. 1, where nírjahyāsténa táṁ drupadé jahi, doubtless meaning nír jahi and ā́ stenám drupadé jahi, is resolved as níḥ: jahyāḥ: téna.
13. For the reader's convenience it may be noted that verses deficient by one or two syllables, respectively, are called by him nicṛt and virāj; and that verses redundant by one or two are called bhurij and svarāj.
14. See his seven Contributions to the interpretation of the Veda (below, p. ci), his Hymns of the AV. (SBE. xlii.), and his review of Caland's Zauberritual (Göttingische gelehrte Anzeigen, 1902, no. 7).
15. See his Altindisches Zauberritual, and his eight papers Zur Exegese und Kritik der rituellen Sūtras (ZDMG. li.-lvii.). Of the papers, those most important for the Kāuçika are the ones contained in vol. liii. See also WZKM. viii. 367.
16. See Bloomfield's note, SBE. xlii. 558; Whitney's introduction to vii. 74, and the note added by me at p. 440, top; and Caland's note 5 to page 105 of his Zauberritual. Hymn 76 of the Berlin ed. is in no wise a unity: see the introduction thereto.
17. I owe this suggestion to Professor Delbrück of Jena, who was my guest while I had this chapter in hand and was so kind as to criticize it. As a curious parallel to the case above cited, he told me of the verses prescribed for use in the Brüdergemeine of Count Zinzendorf:

    Mein mir von Gott verliehenes Weib!
    Anitzt besteig' ich deinen Leib.
    Empfange meinen Samen
    In Gottes Namen. Amen.

18. Caland's sketch of the funeral rites is a most praiseworthy and interesting one, and his description of the practices which he there sets forth in orderly and lucid sequence is well worth the while: but his descriptions are taken from many sources differing widely in place and time; and it is on many grounds improbable that the ritual as he there depicts it was ever carried out in any given place at any given time.
19. I had hesitatingly advanced this view, below, in my note to xviii. 4. 61; and I am pleased to see now that Bloomfield had unhesitatingly given it as his own opinion long before, at AJP. xi. 341.
20. Further reference is made to these general relations below, at p. 1013.
21. So Roth in the Atti (p. 95), as cited on this page.
22. My copy of Roth's essay was given me by my teacher, the author, Feb. 26, 1875.
23. In some cases, fragments of the birch-bark original seem to have become lost after Roth's Kashmirian nāgarī transcript was made, so that the latter, and the two other Indian copies mentioned on p. lxxxi, have thus become now our only reliance. Thus for avīvṛdhat of the Vulgate at i. 29. 3 b. Roth reports as Pāipp. variant abhībhṛçat, and adds "nur in der Abschrift vorhanden." This must have stood on the prior half of line 12 of folio 3 b of the birch-bark ms.; but a piece of it is there broken out.
24. The Kashmirian Atharva-Veda (School of the Pāippalādas). Reproduced by chromophotography from the manuscript in the University Library at Tübingen. Edited under the auspices of the Johns Hopkins University in Baltimore and of the Royal Eberhard-Karls-University in Tübingen, Württemberg, by Maurice Bloomfield, Professor in the Johns Hopkins University, and Richard Garbe, Professor in the University of Tübingen. Baltimore. The Johns Hopkins Press. 1901. The technical work by the firm of Martin Rommel & Co., Stuttgart.
25. Such are: "Verse, die nur durch Fehler Eckel erregen," p. lxxxii; "On y trouve, il est vrai, de très-bonnes parties, mais d'autres sont tellement défigurées, qu'on a besoin de conjectures sans nombre pour arriver à un texte lisible," Atti, p. 96; "das Kauderwelsch," "ganze Zeilen so unsicher dass man nicht einmal die Wörter trennen kann," p. lxxxvi.
26. To judge from stedam for stenam, we might suppose that the ms. at this point was written down by a scribe at the dictation of a reciter with a bad cold in his head.
27. Such as suryam at p. xxxvi, foot-note.
28. I am sorry to observe that the third (posthumous) edition of his Grammar (see pages 518-9) misrepresents him upon this point.
29. The main part of this book was in type as far as page 614 (xi. 1. 12) in Dec. 1901. The remainder (as far as p. 1009, the end) was in type Dec. 13, 1902.
30. Under date of Feb. 14, Whitney suggests to Roth: "Why not give a Pāipp. text, as an appendix to our volume ["our volume" means the present work], noting in their order the parallel passages by reference only, and writing out in full, interspersed with the former, the remainder?"—Roth makes answer, March 14: "Ich will nur wünschen, dass Ihre Gesundheit so lange Stand halte, um das Werk zu Ende zu führen. Weil das aber als ein glücklicher Fall zu betrachten ist, nicht als eine sichere Voraussicht, so wünschte ich alle Erschwerungen, also auch die Frage von einer Publikation der Paippal. Rec. gänzlich beseitigt zu sehen."—Whitney, June 16, expresses the hope that Roth may reconsider the matter, 1. because "a text of such primary importance will and must be published, in spite of its textual condition," and 2. because "there will, so far as I can see, no other opportunity present itself of producing it so modestly and unpretendingly, or in a method adapted to its imperfect state: the occasion is an ideal one."—Roth answers, July 2: "Mein lieber Freund, das ist kein erfreulicher Bericht, welchen Ihr Brief vom 16. Juni über Ihre Erlebnisse erstattet. Und ich sehe namentlich daraus, dass Sie die Geduld sich erworben haben, die durch Uebung im Leiden kommt....In einer Ausgabe der Pāipp. müsste das ganze gedruckt werden, von A bis Z....Wie wird sich das Kauderwelsch gedruckt ausnehmen? ganze Zeilen so unsicher, dass man nicht einmal die Wörter trennen kann....Daran bessern, was ja das einzige Verdienst wäre, dürfte man nicht....Für Sie wird die einzige angemessene Sorge in diesem Augenblick sein, wieder gesund zu werden, alsdanu die zweite, den Atharvan ans Licht zu bringen."—Whitney writes, Aug. 25: "I give up with reluctance the hope of the further inclusion of Pāipp. in our edition; but I will not bother you further with remonstrances or suggestions."
31. For the sake of fathers to whom English is not vernacular, it may be added that this classic of English and American nurseries is the work of Charles Lutwidge Dodgson ("Lewis Carroll") and is a pendant to Alice's Adventures in Wonderland.
32. In spite of its intrinsic importance, such is the case, I believe, with the ÇB., to which Whitney makes, I think, rather meagre reference.
33. And it is a large achievement to do it on such a scale as does the Concordance.
34. Whoever doubts it, let him take so very simple a case as AV. ii. 29. 3 or iv. 14. 1, write out the AV. text in full and then the three parallel Yajus-texts beneath it, compare them, underscore in red ink the points of difference, and then state them with brevity and clearness. Then let him examine Whitney's reports, and I think he will freely admit that they are indeed well-digested and are models of masterly condensation. More difficult cases are ii. 1. 3; 13. 1; iii. 10. 4; 12. 7; 19. 8; vii. 83. 2; 97. 1; xiv. 2. 71. The amount and intricacy of possible variation is well exemplified by vi. 117. 1. Perhaps Whitney has erred in the direction of overcondensation in his note to vii. 29. 2.
35. Others, taken from the Kashmirian text, are given above, p. lxxxiii.
36. Confusions of surd and sonant are discussed by Roth, ZDMG. xlviii. 107: cf. note to ii. 13. 3, below. The Kashmirian text swarms with them.
37. Here Bollensen long ago proposed (Orient und Occident, ii. 485) to athetize abhavat.
38. It would be idle presumption in me to praise the work of a man whose knowledge of the literature and customs and spirit of India is so incomparably greater than my own; but I may be allowed to repeat the judgment of my revered and beloved friend, M. Auguste Barth, concerning Griffith's Veda-translations: Elle [the RV. translation] se présente ainsi sans aucun appareil savant, ce qui, da reste, ne veut pas dire qu'elle n'est pas savante. L'auteur, qui a longtemps dirige le Benares College, a une profonde connaissance des langues, des usages, de l'esprit de l'Inde, et, pour maint passage, on aurait tort de ne pas tenir grandement compte de cette version en apparence sans prétentions (Revue de l'histoire des religions, year 1893, xxvii. 181). Elle [the AV. translation]...mérite les mêmes éloges (Ibidem, year 1899, xxxix. 25).
39. By a curious coincidence, "through human frailty" is precisely the rendering given by Griffith.
40. Perhaps the corruption is yet deeper seated, and covers an original māsi-māsy ūrjam.
41. Here let me protest against the much worse than useless custom of giving a new pagination or a double pagination to separate reprints. If an author in citing a reprinted article does not give each reference thereto in duplicate, or if his reader does not have at hand both the original and the reprint (and either of these cases is exceptional), the seeker of a citation is sure to be baffled in a large proportion of the instances concerned. It is amazing that any author or editor can be so heedless as to tolerate this evil practice.