# Atharva-Veda Samhita/General Introduction/Part II

GENERAL INTRODUCTION, PART II.

ELABORATED BY THE EDITOR, IN LARGE PART FROM WHITNEY'S MATERIAL

General Premises

Contents of this Part.—While Part I. contains much that might be presented in a preface, the contents of Part II. are more strictly appropriate for an introduction. The contents of Part I. are briefly rehearsed at p. lxiii; and the contents of both Parts, I. and II., are given with more detail and in synoptic form at pages x—xv, which see. As was the case with the ten text-critical elements of the commentary in Part I., the subject-matter of Part II. also may be put under ten headings as follows:

1. Description of the manuscripts.
2. Their opening stanza.
3. Whitney's Collation-Book.
4. Repeated verses in the mss.
5. Refrains and the like in the mss.
6. Accentuation-marks in the mss.
7. Orthographic method of Berlin text.
8. Metrical form of the Atharvan sarhhita.
9. Divisions of the text.
10. Its extent and structure.⌋

Authorship of this Part.—While Part I. is wholly from the hand of the editor. Part II. is elaborated in large measure from material left by Whitney. Chapters 2 and 3, however, although written by the editor, are incorporated into this Part, because the most fit place for them is here, just after chapter 1. In the rewritten portions of the other chapters, it has not been attempted thoroughly to separate the author's part from the editor's; but paragraphs which are entirely by the editor are enclosed in ell-brackets, ⌊ ⌋. The whole matter has been carefully stated by me in the preface, at pages xxix-xxx, and these the reader is requested to consult.⌋

1. Description of the Manuscripts used by Whitney

The brief designations of his manuscripts (sigla codicum).—The sigla O. and L. seem to be arbitrary. It is helpful to note that Whitney apparently intended that all the rest should be suggestive. Thus B., P., R., T., and D. are the initials respectively of Berlin, Paris, Roth, Tanjore, and Deccan; small p. of course means pada-text; and small s. means saṁhitā-text; and K. was the first letter of Bikaner not previously employed as siglum. M. and W., which designate the mss. of the Mill collection and Wilson collection of the Bodleian, were chosen as being initials of Mill and Wilson. The letters E. I. H., as designating the mss. of the Library of the India Office in London, were plainly meant to suggest the name East India House, the designation of the London establishment of the Hon. East India Company previous to 1858. Observe that Whitney's "I." was first used by him to designate E.I.H. ms. No. 2142 (Eggeling's No. 234), but only until he discovered that that ms. was a mere copy of the Poller ms. in the British Museum; after that time Whitney collated the Poller original, retaining for it, however, the designation "I." The sigla of the mss. used by Whitney before publication are essentially the same as those given by him at the end of his Introductory Note to the AV.Pr., p. 338, which see.⌋

Synoptic table of the manuscripts used by Whitney.—It will be convenient to have, in addition to Whitney's description of his mss., a synoptic table of them, cast in such a form that the reader may easily see just what ones were available for any given book. The following table is essentially the same as one which Whitney made for his own use.

 SAṀHITĀ-MANUSCRIPTS PADA-MANUSCRIPTS ${\displaystyle \scriptstyle {\overbrace {\quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad } }}$ ${\displaystyle \scriptstyle {\overbrace {\quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad \quad } }}$ Berlin Paris Oxford "East India House" Haug Roth Tanjore Bikaner Berlin Haug "Deccan" Bikaner ${\displaystyle \scriptstyle {\overbrace {\quad \quad \quad \quad \quad } }}$ ${\displaystyle \scriptstyle {\overbrace {\quad \quad \quad \quad \quad \quad \quad } }}$ ${\displaystyle \scriptstyle {\overbrace {\quad \quad \quad \quad \quad \quad \quad } }}$ Mill Wilson Mus. i. P. M. W. E. I. H. O. R. T. K. Bp.a Bp.2a Op. D. Kp. ii. P. M. W. E. I. H. O. R. T. K. Bp.a Op. D. Kp. iii. P. M. W. E. I. H. O. R. T. K. Bp.a Op. D. Kp. iv. P. M. W. E. I. H. O. R. T. K. Bp.a Op. D. Kp. v. P. M. W. E. I. H. O. R. T. K. Bp.a Bp.2b′ D. Kp. vi. P. M. W. E. I. H. O. R. T. K. Bp.a Bp.2b″ D. Kp. vii. P. "M." W. E. I. O. R. T. K. Bp.a Bp.2b″ D. Kp. viii. P. "M." W. E. I. O. R. T. K. Bp.a Bp.2b″ D. Kp. ix. P. "M." W. E. I. O. R. T. K. Bp.a Bp.2b″ D. Kp. x. P. "M." W. E. I. O. R. T. K. Bp.b Bp.2b″ D. Kp. xi. B. P. M. W. E. I. O. R. T. K. Bp.b D. Kp. xii. B. P. M. W. E. I. O. R. T. K. Bp.b D. Kp. xiii. B. P. M. W. E. I. O. R. T. K. Bp.b D. Kp. xiv. B. P. M. W. E. I. O. R. T. K. Bp.b D. Kp. xv. B. P. M. W. E. I. O. R. T. K. Bp.b D. Kp. xvi. B. P. M. W. E. I. O. R. T. K. Bp.b D. Kp. xvii. B. P. M. W. E. I. O. R. T. K. Bp.b D. Kp. xviii. B. P. M. E. I. O. R. T. K. Bp.b Op. D. Kp. xix. B. P. M. W. E. I. O. R. T. K. D. L. Kp. xX. B. P. M. W. E. I. O. R. T. K. Bp.c Bp.2c D. Kp.

Berlin manuscripts of the Atharva-Veda.—A tabular view of the various numberings and designations of the nine Berlin mss., Weber, Nos. 331-339, will be found useful and is given here. The left-hand column gives the sigla used by Whitney, but with some marks (a, b, c, ′, ″) added for convenience of reference to or from the preceding table. The second column shows which books any given ms. contains. The third gives the numbers of the mss. as they stand in Weber's Catalogue; and the fourth gives the old numbers assigned to those mss. when they formed a part of the collection of Sir Robert Chambers. The right-hand column shows what book or group of books was transcribed by Whitney from the original ms. named in the same line.

 Books Weber-No. Chambers-No. Copied by Whitney Bp.a i.-ix. 332 8 Books i.-iv. and vi.-ix. Bp.b x.-xviii. 335 108 Books x.-xviii. Bp.c xx. 336 114 Book xx. Bp.2a i. 331 117 Bp.2b′ ${\displaystyle \scriptstyle {\left\{{\begin{matrix}\ \\\ \end{matrix}}\right.}}$ v. 333 109 Book v. Bp.2b″ vi.-ix. 334 107 Bp.2c xx. 337 116 B.′ xi.-xx. 338 115 Book xix. B.″ xi.-xx. 339 120

Manuscripts used by Whitney before publication of the text.—The following descriptions were written out by Whitney in such form as to require almost no changes.⌋

Bp. Under this designation are, for convenience's sake, grouped two Berlin pada-manuscripts, making together a complete pada-text to books i.-xviii. The first manuscript, Bp.2 (Chambers, No. 8; Weber, No. 332), is described on pp. 82-83 of Weber's Catalogue of the Berlin Sanskrit mss. It contains books i.-ix., written in a clear but rather rude hand, quite fairly correctly, and accented throughout in a uniform manner. At the end of book ii. is a colophon (given in full by Weber), stating the date as A.D. 1593-4; but this is probably copied from the scribe's original. At the end of the fourth book was perhaps another colophon; but, if so, it is lost, with the last word of the last verse in the book, by the omission of a leaf (leaf 125). The second manuscript, Bp.b (Chambers, No. 108; Weber, No. 335: see Weber's Catalogue, pp. 83-84), containing books x.-xviii., is defective at the end, lacking the last two verses of xviii. (except the first word of 4. 88), and of course also the colophon. It is written in three different hands, with fair correctness (Weber's note, "by the same hand as 334," is a mistake). It is accented in the same manner as No. 332.

Bp.2 This designation also applies to more than one manuscript: the first manuscript, Bp.2a, contains only book i. (Chambers, No. 117; Weber, No. 331: Cat., p. 82), is handsomely and very accurately written, and is quite independent of Bp. It is dated A.D. 1632. Its mode of accentuation changes soon after the beginning (see below, p. cxxi). The second manuscript, Bp.2b, contains books v.-ix. This manuscript, though one in paper, size, and hand, has by some means become separated into two parts, the one (Chambers, 109; Weber, 333) containing only book v., and the other (Chambers, 107; Weber, 334: both p. 83 of Cat.) containing books vi.-ix. They are less independent than Bp.2a, representing the. same proximate original as Bp. (though they are not copied from Bp., nor are they its original); but they are decidedly more accurate than Bp., and also more carefully corrected since copying. There is no colophon to either part, but they are as old, apparently, as Bp.2a, or as Bp.; their mode of accentuation agrees throughout with that of the latter.

B. or Bs. This is the Berlin manuscript (Chambers, 115; Weber, 338: pp. 84-85 of Cat.) of books xi.-xx. in saṁhitā-text. It is rather incorrect and somewhat worm-eaten. It bears the date A.D. 1611. In the Berlin Library is (Chambers, 120; Weber, 339: p. 85 of Cat.) a modern copy ⌊B.″⌋ of it, having value only as having been made before its original was so much worm-eaten as at present.

P. and M. These are virtually one manuscript, being two copies of the same original, by the same hand, and agreeing precisely in form and style. P. is in the Paris Library, and is in two volumes, marked D 204 and D 205. M., also in two volumes, belongs to the Mill collection in the Bodleian Library at Oxford.[1] By some curious and unexplained blunder, the copy of books vii.-x. that belonged to M. was sent by mistake to Paris with P., so that P's first volume contains books i.-x., and its second vii.-xx., while of M. the first volume contains i.-vi., and the second xi.-xx. In the references made in the notes below, the copy of vii.-x. included in the first[2] volume of P. is accounted as M. The differences of the two are not altogether such as are due only to the last copyist; since P. has been collated and corrected (winning thereby some false readings). P. is also more carefully copied than M., but both are rather inaccurate reproductions of a faulty original. A colophon copied in both at the end of book xi. gives saṁvat 1812 (A.D. 1756) as the date, doubtless of the original; the copies are recent, probably since the beginning of the nineteenth century. Their mode of accentuation is by strokes, not dots; that of P. is defective from xiii. 1 to xix. 10.

W. This also, like M., belongs to the Bodleian Library at Oxford,[3] and is a saṁhitā-manuscript of the whole Atharvan, excepting only book xviii. It has no colophon at the end, but is a modern copy, on European paper, and in part made from the same original as P. and M., as is shown both by accordances in minute peculiarities and errors of reading, and by containing at the end of book xi. the same colophon as they. In certain of the books, namely i., ii., vi.-x., xvi., xvii., it shows signs of greater independence. It is by far the most faulty and least valuable of all the manuscripts collated. Only the first book is accentuated, nearly in the familiar RV. method.

E. This is a saṁhitā-manuscript of all the twenty books (except the latter half of xviii., from 3. 6 on), belonging to the India Office Library in London. It is described in Eggeling's Catalogue on p. 37 (now numbered 229 and 230; formerly 682 and 760 or 113). It has no date; Eggeling reckons it as of the 17th century. It is written on coarse rough paper, in a large and irregular hand, apparently by a scholar for his own use, and is fairly correct. The text is here and there a little mutilated at the edges by the reprehensible carelessness of the binder; otherwise it is in good preservation. Its method of accentuation is very various: see below, p. cxxii.

I. This is a complete copy of the saṁhitā-text, in large form (14¾ × 6¾ in.), being one of the set of Vedic manuscripts brought to Europe by Col. Polier, and now belonging to the British Museum in London. The Atharvan material is contained in two volumes: vol. i. gives first book xix., then xx., then i.-x.; vol. ii. gives the Anukramaṇī, then the Gopatha Brāhmaṇa, then books xi.-xvii., then xviii.—each division, in both volumes, being separately paged. There is no colophon; but the whole is evidently a modern copy, made for Col. Polier himself. It is on smooth paper, well written, and not especially inaccurate. It contains the verse çáṁ no devī́r etc. prefixed at the beginning, like some of the manuscripts compared later (see p. cxvi).

Of all this Atharvan material of Polier's, a copy was made for Col. Martin while it remained in the latter's keeping (as Prof. H. H. Wilson informed me that he personally knew it to have been for a time); and this copy now constitutes Nos. 233-236 of the India Office collection, being credited as presented by R. Johnson (No. 234, containing Books xi.-xviii., has W. D. W.'s note to this effect reported in the Catalogue; but Prof. Eggeling fails to notice that the other volumes are of the same character). The collation of No. 234 was begun, but abandoned on the discovery of its origin. Doubtless No. 232 (old number 901) is another copy of the Polier first volume, made at the same time for Colebrooke, or else ⌊made for Martin and⌋ later given ⌊to Colebrooke⌋ by Martin, as it is stamped "Claud Martin"; ⌊at all events, the one who gave it to the Library was Colebrooke⌋.

H. This manuscript, again, belongs to the India Office Library (No. 231; old No. 1137; Catalogue, p. 37). It contains only the first six books, and is handsomely but rather incorrectly written. It has no date, but its accentuation was added in A.D. 1708. Its mode of marking the accent varies: see below.

Manuscripts collated after publication of the text.—The following descriptions also were written out by Whitney, except those of mss. R. and T., which have been supplied by the editor.⌋

The above are all the manuscripts known to have been in Europe in 1853; and upon them alone, accordingly, the printed text was founded. Those that follow have been since collated, and their readings are reported in the notes to the translation.

O. In the possession at present of the Munich Library, but formerly of Prof. M. Haug (to whom they belonged at the time of their collation), are a parcel of Atharvan manuscripts containing a complete saṁhitā-text, with a pada-text of six books, variously divided and bound together, and in part mixed with other texts. The saṁhitā-text is designated as above: it is in five parts: 1. books i.-v., on European paper, 8¼ × 3 in., each book separately paged. The date at the end, çake 1737 (= A.D. 1815) may be that of the original from which this copy is made. It is written in a small but neat and clear hand. 2. Books vi.-xvii., 8¾ × 3¾ in., written in a good sizeable hand, by a Māṁnajī; dated saṁvat 1690 (A.D. 1634); the paper is in parts badly damaged, so as hardly to hold together, and of two leaves in book xii. only fragments remain. It makes great use of the virāma, and of ँ as anusvāra-sign. It numbers the verses only in vargas, making no account of the hymns (sūktas); nor does it notice the prapāṭhaka division. 3. Book xviii., 9¼ × 5 in.; in a large regular hand; dated çake 1735 (A.D. 1813). When collated, it was bound in one volume with pada-ms. of i.-iii. before it, and saṁhitā of xx. after it. 4. Book xix., bound up with i. (saṁhitā i.-v.), and in all respects agreeing with it, save that the (copied?) date is two years later; both are works of the same copyist. 5. Book xx., bound in (as above noted) after 3. The size is 8¾ × 4½ in., and it is dated çake 1735 (A.D. 1813).

Op. This designates the pada-text of the Haug or Munich manuscripts, as above described. They include books i.-iv., xviii., and xx., in three divisions: 1. books i.-iii., bound up (as noted above) with the saṁhitā-text of xviii. and xx. The books are paged separately, but all written by one hand; the date at the end is çake 1733 (A.D. 1811); size 9 × 4¾ in. The hand is large and clear, and the text (corrected by the accentuator) very correct. 2. Book iv.: size 8x4 in.; date çake 1736 (A.D. 1814). 3. Books xviii. and xx., bound with the preceding, and of same size; separately paged; date çake 1762 (A.D. 1840). From xx. are omitted the peculiar Atharvan parts, except hymn 2.

O. and Op. were not collated word by word throughout, because use of them was allowed only for the time of a limited stay in Munich. Books xv.-xix., and the peculiar parts of xx., also it paryāya hymns in the preceding books, and the pada-text, were collated thoroughly; in the metrical parts of vi.-xiv. the comparison was made by looking through the transliterated copy and noting readings on all doubtful points.

⌊These mss. are described in the Verzeichniss der orientalischen Handschriften aus dem Nachlasse des Professor Dr. Martin Haug in München, München, T. Ackermann, 1876. By the siglum O. are designated the mss. there numbered 12, 13, and 14; by Op., those numbered 15 and 17. The dimensions there given differ in part a little from those given by Whitney. It is worth while to report from JAOS. x., p. cxviii, W's critical remark about this material: "all in good and correct manuscripts, made by and for Hindu scholars (not copies by professional scribes for the use of Europeans)."⌋

R. ⌊This is a complete saṁhitā-ms., belonging at the time of its collation (1875) to Roth, and now in the Tübingen University Library. It is described by Roth, Der Atharvaveda in Kaschmir, p. 6, and by Garbe, in his Verzeichniss, as No. 12, p. 11. It is bound in two volumes, the one containing books i.-x., and the other, books xi.-xx. In the colophons to a number of the books (so viii., ix., x., xiv., xix.) is the date çake 1746 (A.D. 1824); but at the end of xx. is the date saṁvat 1926 (A.D. 1870). It was bought for Roth from a Brahman in Benares by Dr. Hoernle, and Roth judged from the name of the scribe, Paṭuvardhana Viṭhala, that it originated in the Deccan. Whitney says (JAOS. x., p. cxviii, = PAOS. Nov. 1875) that it has special kindred with the Haug mss. Roth adds that it is written and corrected throughout with the most extreme carefulness and is far more correct than the AV. mss. are wont to be.⌋

T. ⌊This also is a complete saṁhitā-ms., a transcript made from the Tanjore-mss. described on p. 12 of A. C. Burnell's Classified index to the Sanskrit mss. in the palace at Tanjore and numbered 2526 and 2527. The transcript was sent to Roth by Dr. Burnell and is described by Roth and by Garbe in the places just cited under codex R. Books i.-iv. of the transcript are unaccented; the rest are accented. According to Burnell, No. 2526 contains books i.-xx., is unaccented, and was written about A.D. 1800; and No. 2527 contains books v.-xx., is accented, and was written A.D. 1827 at Benares. I find no note stating the relation of Roth's transcript to its Tanjore originals: presumably the transcript of the unaccented books, i.-iv., was taken from the unaccented No. 2526; and that of the accented books, v.-xx., from No. 2527.⌋

D. This is a pada-manuscript belonging to the Deccan College at Poona, collated while in Roth's possession at Tübingen. It is unaccented in book xviii. It is very incorrectly written, and its obvious errors were left unnoted. It gives a pada-text even for book xix., but not for the peculiar parts of xx. ⌊The Index to the Catalogue of 1888 of the Deccan College mss. gives only two complete pada-mss. of the AV., to wit, the ms. listed as III. 5 on p. 13, and the one listed as XII. 82 on p. 174. The Catalogue gives as date of the latter saṁvat 1720; and as date of the former, saṁvat 1741. In the Collation-Book, Whitney gives at the end of book xx. the colophon of his D. with the dates saṁvat 1741, çake 1606. This agreement in date seems to identify his D. with the ms. III. 5. That ms. is a part of the collection of 1870-71, made by Bühler; it is booked as consisting of 435 pages and as coming from Broach or Bharūch.⌋

L. A pada-manuscript of xix. in the Berlin Library was apparently copied from D. while it was still in India (this copy is denoted by L.). ⌊It is described by Weber, Verzeichniss, vol. ii., p. 79, under No. 1486, with details confirmatory of the above.⌋

K. By this sign is meant a manuscript from Bikaner containing the complete saṁhitā-text; it was for some time in the hands of Roth at Tübingen, and was consulted by means of a list of some 1200 doubtful readings sent to Tübingen and reported upon. These concerned books i.-xviii. alone; xix. and the peculiar parts of xx., not admitting of treatment in that way, did not get the benefit of the collation. The manuscript claims to be written in saṁvat 1735, çake 1600 (A.D. 1678-9), by Eṁmvāgaṇeça, under king Anūpasiṅha, at Pattana-nagara.

Accompanying this is a pada-manuscript written by the same scribe, but without accents. Where there is occasion for it, this is distinguished by the designation Kp.

2. The Stanza çáṁ no devī́r abhíṣṭaye as Opening Stanza

It was doubtless the initial stanza of the text in the Kashmirian recension.—This stanza, which appears as i. 6. 1 of the Vulgate, doubtless stood at the beginning of the Pāippalāda text. In 1875, Roth, in his AV. in Kaschmir, p. 16, remarks upon the general agreement in the tradition according to which çáṁ no etc. was the initial stanza of Pāipp., and not yé triṣaptā́ḥ as in the Vulgate; and regrets all the more on that account that the first leaf of the Pāipp. ms. is lost.⌋

Çáṁ no as initial stanza of the Vulgate text.—Whitney notes that this stanza is also found prefixed to the text of the Vulgate in four of the mss. used by him, to wit, I. and R. and O. and Op. Thus at the beginning of I. we have the stanza çáṁ no devī́r abhíṣṭaye entire, and then yé triṣaptā́ḥ.

⌊In 1871, Haug had noted, p. 45 of his Brahma und die Brahmanen, that the Mahābhāṣya, in rehearsing the beginnings of the four Vedas, gives çam etc. as the beginning of the AV.,[4] and that both of his mss. (our O. and Op., as just stated) prefix the stanza. In 1873, Ind. St. xiii. 431-3. Weber again called attention to the fact concerning the Mahābhāṣya, and to a similar one concerning the Gopatha Brāhmaṇa. In 1874, Indian Antiquary, iii. 1 32, Bhandarkar speaks of our stanza as representative pratīka of the AV. in the Brahmayajña or daily devotional recitation of the Hindus. For further discussion of the matter, see Bloomfield, Kāuçika, Introduction, pages xxxvii and xxxviii, and the references there given, and his note to 9. 7, and his §§13-14 in the Grundriss, p. 14. We may add that in 1879 Burnell observed, on p. 37 b of his Tanjore Catalogue, that the real South Indian mss. of the Mahābhāṣya ignore the AV. and omit the çáṁ no devī́r.⌋

3. Whitney's Collation-Book and his Collations

Description of the two volumes that form the Collation-Book.—The Collation-Book is the immediate source of the statements of this work concerning the variants of the European mss. of the Atharva-Veda. It contains, in Whitney's handwriting, the fundamental transcript (in Roman transliteration) of the text, and the memoranda of the subsequent collations. It is bound in two volumes: of these, the second, comprehending books x.-xx., appears to have been written first, since it is dated "Berlin. Oct 1851-Jan 1852"; while the first, comprehending books i.— ix., is dated "Berlin. Jan-March 1852," and thus appears to have been written last. The volumes are of good writing-paper (leaves about 8 by 10 inches in size), the first containing 334 leaves, and the second, 372.⌋

Whitney's fundamental transcript of the text.—In copying book x. (the first book copied), Whitney has written the text on both sides of the leaf; but for the books subsequently copied, books xi.-xx. and i.-ix., he has written the text on the verso only and used the recto of the next leaf for various memoranda. For all the first eighteen books except book v., this fundamental copy is a transcript of the pada-text contained in the two Berlin pada-mss. (see the table on page cxi), called Chambers, 8 and 108, and designated above as Bp. For book v., he copied from one of the four mss. to which the designation Bp.2 is applied, to wit from Chambers, 109.⌋

⌊The fundamental copy of book xix. was made by Whitney from the saṁhitā-ms. Chambers, 115 = B. He appears to have copied the text first on the recto, without word-division, and using Roman letters, although applying to them the vertical and horizontal accent strokes as if to nāgarī; and afterwards to have written out the text on the opposite page, the verso of the preceding leaf, with word-division, and with accents marked in the usual European way. At xix. 27. 6 Whitney makes the note, "acc. from Paris ms. to the end of the book."—For book xx., the transcript was made from Chambers, 114 = Bp.c.⌋

Collations made before publication of the text. The Berlin collations (first collations).—In this paragraph, only books i.-xix. are had in view, and codex B.″, as being a mere copy of B.′, is disregarded. From the table on p. ex, it appears that for books ii.-iv., x., and xix. there was only one ms. at Berlin, and so none available for collation. The first collation of book v. (since this was copied from Bp.2 = Chambers, 109) was made of course (see the table) from Bp. = Ch. 8. The first collation of the books copied from Bp. was made (also of course) from Bp.2: that is, the first collation of book i. was made with Ch. 117, and that of books vi.-ix. with Ch. 107. For books xi.-xviii. the collation was made of course with B. = Ch. 115.⌋

The Paris and Oxford and London collations.—These, made in the months of March to July, 1853, were the last collations before the printing of the text. They were made in the order as named, and their sequence appears from the biographical sketch[5] above, p. xliv.⌋

Collations made after publication (that is, made in 1875 or later.)—Twenty years or more after the issue, in Feb. 1855, of the printed text of books i.-xix., were made the collations of the mss. enumerated below. See above, page xliv, and JAOS. x., p. cxviii.⌋

Collation of the Haug, Roth, Tanjore, and Deccan mss.—The collation with the Haug mss., O. and Op., was made at Munich, in June and July, 1875. Then followed, at Tübingen, the collations with the mss. D. and T. and R.⌋

The collation with the Bikaner ms., K.—This, as stated above, p. cxvi, was made by means of a list of doubtful readings sent by Whitney to Roth and reported upon.⌋

Other contents of the Collation-Book.—The various memoranda (mentioned above, p. cxvii, ¶3) are usually written on the blank page opposite the hymn or verse concerned. They include the excerpts from the Major Anukramaṇī, the citations of concordant passages gathered from an exceedingly comprehensive study of the other Vedic texts, very full references to the AV. Prātiçākhya and to the Kāuçika and Vāitāna Sūtras, references to the writings of Occidental Vedic scholars in which a given verse or hymn has been treated by way of translation or comment, schemes of the meters and criticism thereof, and finally miscellaneous notes.—I may add here that Whitney left a Supplement to his Collation-Book. It consists of 19 loose leaves containing statements of the variants of B.P.M.W.E.I.H. in tabular form. With it are about a dozen more leaves of variants and doubtful readings etc.⌋

4. Repeated Verses in the Manuscripts.[6]

Abbreviated by pratīka with addition of ity ekā etc.—There are 41 cases of a repeated verse or a repeated group of verses occurring a second time in the text and agreeing throughout without variant with the text of the former occurrence. These in the mss. generally, both saṁhitā and pada, are given the second time by pratīka only, with ity ekā (sc. ṛ́k). or íti dvé or íti tisráḥ added and always accented like the quoted text-words themselves. Thus ix. 10. 4 (= vii. 73. 7) appears in the mss. as úpa hvaya íty ékā. On the other hand, the very next verse, although it differs from vii. 73. 8 only by having 'bhyā́gāt for nyā́gan, is written out in full. So xiii. 2. 38 (= x. 8. 18) appears as sahasrāhṇyám íty ékā; while xiii. 3. 14, which is a second repetition of x. 8. 18 but contains further the added refrain tásya etc., is written out in full as far as tásya. The like holds good of xiii. 3. 18. See note to xiii. 3. 14.

List of repeated verses or verse-groups.—The 41 cases of repetition involve 52 verses. The list of them is given on p. 3 of the Index Verborum (where xix. 23. 20 is a misprint) and is given with the places of first occurrence. The list is repeated here, but without the places of first occurrence, which may always be ascertained from the commentary below. It is: iv. 17. 3; v. 6. 1 and 2; 23. 10-12; vi. 58. 3; 84. 4; 94. 1-2; 95. 1-2; 101. 3; vii. 23. 1; 75. 1; 112. 2; viii. 3. 18, 22; 9. 11; ix. 1. 15; 3. 23; 10. 4, 20, 22; x. 1. 4; 3. 5; 5. 46-47, 48-49; xi. 10. 17; xiii. 1. 41; 2. 38; xiv. 1. 23-24; 2. 45; xviii. 1. 27-28; 3. 57; 4. 25, 43, 45-47, 69; xix. 13. 6; 23. 30; 24. 4; 27. 14-15; 37. 4; 58. 5.

Further details concerning the pratīka and the addition.—The pratīka embraces the first word, or the first two, ⌊or even the first three, when one or two of them are enclitics: so vi. 94. 1; 101. 3; viii. 3. 22; ix. 1. 15⌋; but at xix. 58. 5 the whole first pāda is given with ity ékā added. Occasionally, in one or another ms., the repeated verse or group is given in full: thus by O.R. in the cases of repetitions in book xviii. Both editions give all the repeated verses in full.

The addition is lacking at v. 6. 1 and v. 6. 2; although these are consecutive verses, it is clear from the separate giving of two pratīkas that here repetitions of non-consecutive verses are intended, and that the addition in each case would be íty ékā. The addition is also lacking at xiv. 1. 23-24; where, however, the repetition of consecutive verses, vii. 81. 1—2, is intended. Here again the mss. give two pratīkas separately, pūrvāparám (= vii. 81. 1 and xiii. 2. 11) and návonavaḥ (= vii. 81. 2); and they do this instead of giving pūrvāparám íti dvé, because the latter procedure would have been ambiguous as meaning perhaps also xiii. 2. 11-12.

The addition íti pū́rvā is made where the pratīka alone might have indicated two verses with the same beginning. This happens at xiii. 1. 41 (where aváḥ páreṇa might mean either ix. 9. 17 or 18: see note, p. 716) and at xviii. 4. 43 (but as to this there is disagreement: see note). —By lack of further addition, the intended repetition is doubtful at x. 5. 48-49, where yád agna íti dvé might mean either viii. 3. 12-13 or vii. 61. 1-2 (see note, p. 585); there is doubt also at xix. 37. 4 (the case is discussed fully at p. 957).

5. Refrains and the like in the Manuscripts

Written out in full only in first and last verse of a sequence.—For the relief of the copyists, there is practised on a large scale in both the saṁhitā- and the pada-mss. the omission of words and pādas repeated in successive verses. In general, if anywhere a few words or a pāda or a line or more are found in more than two successive verses, they are written out in full only in the first and last verses and are understood in the others ⌊cf. p. 793, end⌋. For example, in vi. 17, a hymn of four verses, the refrain, being c, d of each of the four, is written out only in 1 and 4. Then, for verse 2 is written only mahī́ dādhā́re ’mā́n vánaspátīn, because yáthe ’yám pṛthivī́ at the beginning is repeated. ⌊That is, the scribe begins with the last one of the words which the verse has in common with its predecessor.⌋ Then, because dādhā́ra also is repeated in 2-4, in verse 3 mahī́ also is left out and the verse reads in the mss. simply dādhā́ra párvatān girī́n—and this without any intimation of omission by the ordinary sign of omission. —Sometimes the case is a little more intricate. Thus, in viii. 10, the initial words só ’d akrāmat are written only in verses 2 and 29, although they are really wanting in verses 9-17, paryāya II. (verses 8-17) being in this respect treated as if all one verse with subdivisions ⌊cf. p. 512 top⌋.

Such abbreviated passages treated by the Anukramaṇī as if unabbreviated.—The Anukramaṇī generally treats the omitted matter as if present, that is, it recognizes the true full form of any verse so abbreviated. In a few instances, however, it does not do so: such instances may be found at xv. 2, where the Anukr. counts 28 instead of 32 or 4 × 8; at xv. 5 (16 instead of 7 × 3); at xvi. 5 (10 instead of 6 × 3); at xvi. 8 (33 instead of 108 or 27 × 4): cf. the discussions at p. 774, ¶2, p. 772, ¶3, p. 793 end, p. 794 top. Such treatment shows that the text has (as we may express it) become mutilated in consequence of the abbreviations, and it shows how old and how general they have been. —One and another ms., however, occasionally fills out some of the omissions—especially R., which, for example, in viii. 10 writes só ’d akrāmat every time when it is a real part of the verse.

Usage of the editions in respect of such abbreviated passages.—Very often SPP. prints in full the abbreviated passages in both saṁhitā and pada form, thus presenting a great quantity of useless and burdensome repetitions. Our edition takes advantage of the usage of the mss. to abbreviate extensively; but it departs from their usage in so far as always to give full intimation of the omitted portions by initial words and by signs of omission. In all cases where the mss. show anything peculiar, it is specially pointed out in the notes on the verses.

6. Marks of Accentuation in the Manuscripts

Berlin edition uses the Rig-Veda method of marking accents.—The modes of marking the accent followed in the different mss. and parts of mss. of the AV. are so diverse, that we were fully justified in adopting for our edition the familiar and sufficient method of the RV. That method is followed strictly throughout in books i.-v. and xix. of the Haug ms. material described above at p. cxiv under O. 1 and 4, but only there, and there possibly only by the last and modern copyist. ⌊Whitney notes in the margin that it is followed also in book xviii. of O., and in books i.-iii. and iv. of Op., and in part of Bp.2a. In this last ms., which is Chambers, 117, of book i., the⌋ method of accentuation is at the beginning that of the Rik, but soon passes over to another fashion, precisely like that of Bp. ⌊see next ¶⌋ saving that horizontal lines are made use of instead of dots. The method continues so to the end.

Dots for lines as accent-marks.—The use of round dots instead of lines as accent-marks is a method that has considerable vogue. It is applied uniformly in the pada-mss. at Berlin (except in Bp.2a as just stated): a dot below the line is the anudāttatara-sign, in its usual place; then the sign of the enclitic svarita is a dot, usually not above, but within the akṣara; and the independent svarita is marked either by the latter method or else by a line drawn transversely upward to the right through the syllable. The dots, however, are unknown elsewhere, save in a large part of E. (from near the end of vi. 27 to the end of xix.) and also in large parts of H.

Marks for the independent svarita.—It was perhaps in connection with the use of the dots that the peculiar ways of marking the independent svarita arose. The simplest way, used only in parts of the mss., is by a line below, somewhat convexed downwards. Or, again, we find just such a line, but run up into and more or less through the akṣara, either below or through the middle. ⌊From this method was probably developed the method of⌋ starting with a horizontal bit below and carrying it completely through the akṣara upwards and with some slant to the right and ending with a bit of horizontal above. ⌊Cf. SPP's Critical Notice, p. 9.⌋ This fully elaborated form is very unusual, and found only in three or four mss. (in part of Bp.2a = Ch. 117, in D. and L., and occasionally in Kp.); ⌊its shape is approximately that of the "long ʃ": cf. SPP's text of ii. 14 and my note to iii. 11. 2⌋.

Horizontal stroke for svarita.—A frequent method is the use of the anudāttatara line below, just as in the RV., but coupled with the denotation of the enclitic svarita by a horizontal stroke across the body of the syllable, and of the independent svarita by one of the signs just noted. But even the independent svarita is sometimes denoted by the same sign as the enclitic svarita, to wit, by a dot or a horizontal line in the syllable itself. The last method (independent svarita by horizontal) is seen in the old ms. of book xx., Bp.c dated A.D. 1477, and in B′.

The udātta marked by vertical stroke above, as in the Māitrāyaṇī.—It is a feature peculiar to E. among our AV. mss. that, from the beginning of book vi. on, it marks the udātta syllable by a perpendicular stroke above,* while the enclitic svarita, as in other mss., has the horizontal stroke in the akṣara; but just before the end of vi. 27, both these strokes are changed to dots, as is also the anndāttatara-stroke.; while in xx. the accentuator goes back to strokes again for all three. *⌊Note that in SPP's mss. A. and E. the udātta is marked by a red ink dot over the proper syllable.⌋

Accent-marks in the Bombay edition.—SPP., in his edition, adopts the RV. method, with the sole exception that he uses the fully elaborated peculiar ʃ-sign, given by the small minority of the mss.,* for the independent svarita. No ms., I believe, of those used by us, makes this combination of methods; and it may safely be claimed that our procedure is truer to the mss., and on that as well as on other accounts, the preferable one. *⌊See, for example, his Critical Notice, p. 14, description of Cp.⌋

Use of a circle as avagraha-sign.—As a matter of kindred character, we may mention that for the sign of avagraha or division of a vocable into its component parts, a small circle is used in all our pada-texts, even of book xx., excepting in the Munich text of xviii. and xx., as stated on p. 4 of the Index Verborum. ⌊It is used also in SPP's pada-mss.: see his Critical Notice, pages 11-14.⌋ This special AV. sign has been imitated in our transliteration in the Index and in the main body of this work ⌊cf. page c⌋; but it may be noted that SPP. employs in his pada-text the sign usual in the RV.

7. Orthographic Method pursued in the Berlin Edition[7]

Founded on the manuscripts and the Prātīçākya.—Our method is of course founded primarily upon the usage of the manuscripts; but that usage we have, within certain limits, controlled and corrected by the teachings of the AV. Prātīçākya.

That treatise an authority only to a certain point.—The rules of that treatise we have regarded as authority up to a certain point; but only up to a certain point, and for the reason that in the AVPr., as in the other corresponding treatises, no proper distinction is made between those orthographic rules on the one hand which are universally accepted and observed, and those on the other hand which seem to be wholly the outcome of arbitrary and artificial theorizing, in particular, the rules of the varṇa-krama[8] or dīrgha-pāṭha. ⌊Cf. Whitney's notes to AVPr. iii. 26 and 32 and TPr. xiv. 1.⌋

Its failure to discriminate between rules of wholly different value.—Thus, on the one hand, we have the rule ⌊AVPr. iii. 27: see W's note⌋ that after a short vowel a final or or n is doubled before any initial vowel, a rule familiar and obligatory[9] not only in the language of the Vedas but in the classical dialect as well; while, on the other hand, we have, put quite upon the same plane and in no way marked as being of a wholly different character and value, such a rule as the following:

The rule ⌊iii. 31⌋ that after r or h an immediately following consonant is doubled; ⌊as to these duplications, the Prātīçākyas are not in entire accord, Pāṇini is permissive, not mandatory, and usage differs greatly, and the h stands by no means on the same footing as the r: cf. W's Grammar, §228; his note to Pr. iii. 31; and Pāṇini's record, at viii. 4. 50-51, of the difference of opinion between Çākaṭāyana and Çākalya.⌋

Another such rule is the prescription that the consonant at the end of a word is doubled, as in triṣṭupp, vidyutt, godhukk; this is directly contravened by RPr., VPr., TPr.—Yet another is the prescription that the first consonant of a group is doubled, as in aggniḥ, vṛkkṣaḥ, etc. ⌊See W's notes to these rules, at iii. 26 and 28.⌋ ⌊"The manuscripts of the AV., so far as known to me, do not, save in very infrequent and entirely sporadic cases, follow any of the rules of the varṇakrama proper, excepting the one which directs duplication after a r; and even in this case, their practice is as irregular as that of the manuscripts of the later literature." So Whitney, note to iii. 32.⌋

Items of conformity to the Prātiçākhya, and of departure therefrom.—Without including those general euphonic rules the observance of which was a matter of course, we may here state some of the particulars in which the authority of the Prātiçākhya has served as our norm.

Transition-sounds: as in tān-t-sarvān.—Pr. ii. 9 ordains that between n̄, ṇ, n' and ç, ṣ, s respectively, k, ṭ, t be in all cases introduced: the first two thirds of the rule never have an opportunity to make themselves good, as the text offers no instance of a conjunction of with ç or of with ; that of final n with initial s, however, is very frequent, and the t has always been introduced by us (save ⌊by inadvertence⌋ in viii. 5. 16 and xi. 2. 25). —The usage of the mss. is slightly varying ⌊"exceedingly irregular," says W. in his note to ii. 9, p. 406, which see⌋: there is not a case perhaps where some one of them does not make the insertion, and perhaps hardly one in which they all do so without variation.

Final -n before ç- and j-: as in paçyañ janmāni.—Pr. ii. 10 and 11 prescribe the assimilation of -n before a following palatal (i.e. its conversion into ), namely, before ç- (which is then converted by ii. 17 into ch-), and before a sonant, i.e. before j- (since jh- does not occur). In such cases we have written for the converted -n an anusvāra; there can hardly arise an ambiguity† in any of the instances. ⌊A few instances may be given: for -n j-, i. 33. 2*; ii. 25. 4, 5; iv. 9. 9*; 36. 9*; v. 8. 7; 22. 14*; vi. 50. 3; viii. 2. 9*; xii. 5. 44; for -n ç-, i. 19. 4*; iii. 11. 5; iv. 8. 3; 22. 6, 7; xviii. 4. 59. The reader may consult the notes to those marked with a star. —SPP. seems to allow himself to be governed by his mss.; this is a wrong procedure: see notes to viii. 2. 9; i. 19. 4; iv. 9. 9.⌊ †⌊But see xiii. 1. 22.⌋

Final -n before c-: as in yāṅç ca.—Rule ii. 26 virtually ordains the insertion of ç. Owing to the frequency of the particle ca, the cases are numerous, and the rule is strictly followed in all the Atharvan mss. and so of course in our edition. This is not, however, the universal usage of the Rik: cf. for example ii. 1. 16, asmā́n ca tā́ṅç ca, and see RPr. iv. 32.

Final -n before t-: as in tāṅs te.—The same rule, ii. 26, ordains the insertion of s. As in the other Vedas, so in the AV., a s is sometimes inserted and sometimes not; its Pr. (cf. ii. 30) allows and the mss. show a variety of usage. Of course, then, each case has been determined on the authority of the mss., nor do there occur any instances in which this is wavering and uncertain. ⌊The matter is fully discussed in W's note to ii. 26, and the 67 cases of insertion and the 28 cases of non-insertion are given on p. 417. Cf. also note to AV. i. 11. 2.⌋

Final -t before ç-: as in asmac cliaravas.—By the strict letter of rules ii. 13 and 17, the ç- is converted into ch- and the preceding final -t is then assimilated, making -cch-. In such cases, however, we have always followed rather the correct theory of the change, since the -t and ç- by their union form the compound -ch-, and have written simply -ch-, as being a truer representation of the actual phonetic result. The mss., with hardly an exception, do the same. ⌊The procedure of the edition and of the mss. is, I believe, uniformly similar also in cases like ṛchāt, gacha, yacha, etc.⌋

Abbreviation of consonant-groups: as in pan̄kti and the like.—By ii. 20 a non-nasal mute coming in the course of word-formation between a nasal and a non-nasal is dropped: so pan̄ti; chintam and rundhi instead of chinttam and runddhi; etc. The mss. observe this rule quite consistently, although not without exceptions; and it has been uniformly followed in the edition. At xii. 1. 40, anuprayun̄ktām is an accidental exception; and here, for once, the mss. happen to agree in retaining the k. ⌊Cf. the Hibernicisms siren'th, len'th, etc.⌋

Final -m and -n before l-: as in kaṅ lokam and sarvāṅ lokān.—Rule ii. 35 prescribes the conversion of -m and of -n alike into nasalized -l. In either case, the resultant combination is therefore, according to the prescription of the Pr., nasalized -l + l-, or two l's of which the first is nasalized. Thus kam lokam becomes ka + nasalized l + lokam, a combination which we may write as kal lokam or as kaṅl lokam or as kaṅ lokam.

⌊It is merely the lack of suitable Roman type that makes the discussion of this matter troublesome. In nāgarī, the nasalized l should properly be written by a l with a nasal sign over it. In Roman, it might well be rendered by an l with a dot as nearly over it as may be (thus ; in practice, a ṅ is made to take the place of the dot alone or else of the dot + l, so that for the sound of "nasalized l" we find either ṅl or simply .⌋

For the combination resultant from -m l-, the mss. are almost unanimous in writing ⌊not what the Pr. ordains, but rather⌋ a single l with nasal sign over the preceding vowel, as in kaṅ lokam at xi. 8. 11; this usage is followed by the Berlin text.

For the resultant from -n l-, the mss. follow the Pr., not without exceptions, and write doubled l with nasal sign over the preceding vowel, as in sarvāṅ lokān, x. 6. 16, etc., asmiṅ loke, ix. 5. 7, etc.; this usage also has been followed in the Berlin text (but not with absolute uniformity).—It would probably have been better to observe strictly the rule of the Pr. and to write both results with double l and preceding nasal sign.

Visarga before st- and the like: as in ripu stena steyakṛt, viii. 4. 10 = RV. vii. 104. 10. Our Pr. ⌊see note to ii. 40⌋ contains no rule prescribing the rejection of a final visarga before an initial sibilant that is followed by a surd mute. The mss. in general, although with very numerous and irregularly occurring exceptions, practice the rejection of the , and so does the Rik ⌊cf. RPr. iv. 12, TPr. ix. 1; VPr. iii. 12⌋; and the general usage of the mss. has been followed by us. ⌊For examples, see x. 5. 1-14: cf. also notes to iv. 16. 1 (ya stāyat: SPP. yas tāyat), i. 8. 3, etc.⌋

The kampa-figures 1 and 3.—Respecting the introduction of these figures between an independent circumflex and an immediately following acute accent in the saṁhitā, our Pr. is likewise silent. The usage of the mss. is exceedingly uncertain and conflicting: there is hardly an instance in which there is not disagreement between them in respect to the use of the one or of the other; nor can any signs of a tendency towards a rule respecting the matter be discovered. There are a few instances, pointed out each at its proper place in the notes, in which a short vowel occurring in the circumflexed syllable is protracted before the figure by all the saṁhitā-mss* Such cases seemed mere casual irregularities, however, and we could not hesitate to adopt the usage of the Rik, setting 1 after the vowel if it were short in quantity, and 3 if it were long. ⌊This matter is discussed with much detail by W. in his notes to APr. iii. 65, pages 494-9, and TPr. xix. 3, p. 362.⌋ *⌊See APr., p. 499, near end, and notes to AV. vi. 109. 1 and x. 1. 9.⌋

The method of marking the accent.—With respect to this important matter, we have adapted the form of our text to the rules of the Rik rather than to the authority of the mss. As to the ways of marking the accent, a wide diversity of usage prevails among the Atharvan mss., nor is there perhaps a single one of them which remains quite true to the same method throughout. Their methods are, however, all of them in the main identical with that of the Rik, varying only in unimportant particulars. ⌊The details have been discussed above (see p. cxxi), and with as much fulness as seemed worth while.⌋

8. Metrical Form of the Atharvan Saṁhitā

Predominance of anuṣṭubh.—The two striking features of the Atharva-Veda as regards its metrical form are the extreme irregularity and the predominance of anuṣṭubh stanzas. The stanzas in gayatrī and triṣṭubh are correspondingly rare, the AV. in this point presenting a sharp contrast with the Rig-Veda. The brief bits of prose interspersed among metrical passages are given below, at p. 1011, as are also the longer passages in Brāhmaṇa-like prose. ⌊In the Kashmirian recension, the latter are even more extensive than in the Vulgate: see p. lxxx.⌋

Extreme metrical irregularity.—This is more or less a characteristic of all the metrical parts of the Vedic texts outside of the Rig-Veda (and Sama-Veda). In the saṁhitās of the Yajur-Veda, in the Brāhmaṇas, and in the Sūtras, the violations of meter are so common and so pervading that one can only say that meter seemed to be of next to no account in the eyes of the text-makers. It is probable that in the Atharvan saṁhitā the irregular verses outnumber the regular.

Apparent wantonness in the alteration of RV. material.—The corruptions and alterations of Rig-Veda verses recurring in the AV. are often such as to seem downright wanton in their metrical irregularity. The smallest infusion of care as to the metrical form of these verses would have sufficed to prevent their distortion to so inordinate a degree.

To emend this irregularity into regularity is not licit.—In very many cases, one can hardly refrain from suggesting that this or that slight and obvious emendation, especially the omission of an intruded word or the insertion of some brief particle or pronoun, would rectify the meter. It would be a great mistake, however, to carry this process too far, and by changes of order, insertions, and various other changes, to mend irregularity into regularity. The text, as Atharvan, never was metrically regular, nor did its constructors care to have it such; and to make it so would be to distort it.

9. The Divisions of the Text

Summary of the various divisions.—These, in the order of their extent, are: pra-pāṭhakas or 'Vor-lesungen' or 'lectures,' to which there is no corresponding division in the RV.; kāṇḍas or 'books,' answering to the maṇḍalas of the RV.; and then, as in the RV., anu-vākas or 're-citations,' and sūktas or 'hymns,' and ṛcas or 'verses.' The verses of the long hymns are also grouped into 'verse-decads,' corresponding to the vargas of the RV. Besides these divisions, there are recognized also the divisions called artha-sūktas or 'sense-hymns' and paryāya-sūktas or 'period-hymns'; and the subdivisions of the latter are called paryāyas. In the paryāya-hymns, the division into gaṇas (or sometimes daṇḍakas: p. 628) is recognized, and the verses are distinguished as avasānarcas and gaṇāvasānarcas (see p. 472). A great deal of detail concerning the divisions of the books (the later books especially) may be found in the special introductions to the several books.⌋

The first and second and third grand divisions of books i.-xviii.—A critical study of the text reveals the fact that the first eighteen books are divided (see p. xv) into three grand divisions: the first (books i.-vii.) contains the short hymns of miscellaneous subjects; the second (books viii.-xii.) contains the long hymns of miscellaneous subjects; and the third consists of the books (xiii.-xviii.) characterized each by unity of subject. These divisions, although not clearly recognized in name (but cf. page clvii, below) by the text-makers, are nevertheless clearly recognized in fact, as is shown by the general arrangement of the text as a whole and as is set forth in detail in the next chapter, pages cxl-clxi. Concerning their recognition by the Old Anukr., see the paragraphs below, pages cxxxix f. In this chapter will be treated the divisions commonly recognized by the native tradition.⌋

The division into prapāṭhakas.—The literal meaning of pra-pāṭhaka is 'Vor-lesung' or 'lesson' or 'lecture.' This division, though noticed in all the mss., is probably a recent, and certainly a very secondary and unimportant one. It is not recognized by the commentary, and it does not appear in the Bombay edition. No ms. gives more than the simple statement, "such and such a prapāṭhaka finished"; no enumeration of hymns or verses is anywhere added. There are 34 prapāṭhakas, and they are numbered consecutively for the whole text so far as they go, that is, from book i. to book xviii. inclusive. The prapāṭhaka-division is not extended into books xix. and xx.

Prapāṭhakas: their number and distribution and extent.—First grand division (books i.-vii.): in each of the books i.-iii. there are 2 prapāṭhakas; in each of the books iv.-vi. there are 3; and in book vii. there are 2: in all, (6 + 9 + 2 =) 17. —Second grand division (books viii.-xii.): in each of the five books viii.-xii. there are 2 prapāṭhakas: in all, 10. —Third grand division (books xiii.-xviii.): each of the first five books, xiii.-xvii., forms 1 prapāṭhaka, while the sixth and last, book xviii., forms 2: in all, 7. —Sum for the three divisions, (17 + 10 + 7 =) 34. —In book iv. the division is very uneven, the first of the 3 prapāṭhakas containing 169 verses or over half the book; while in xii., on the other hand, in order to make an even division of the 304 verses as between the 2 prapāṭhakas, the end of the first is allowed to fall in the middle of a hymn (just after 3. 30), thus giving 148 verses to the first and 156 to the second. ⌊On comparing the verse-totals of the books of the first grand division with the number of prapāṭhakas in each book, an attempt towards a rough approximation to equality of length among them will appear. The like is true in the second grand division; and also in the third (note especially book xviii.), so far as is feasible without making a prapāṭhaka run over more than one book.⌋

Their relation to the anuvāka-divisions.—The prapāṭhaka-divisions mostly coincide with the anuvāka-divisions. Exceptions are as follows: prapāṭhaka 11 begins with v. 8, in the middle of the second anuvāka of book v.; 19 begins with viii. 6, in the middle of the third anuvāka of book viii.; 21 begins with ix. 6, in the middle of the third anuvāka of book ix.; 23 begins with x. 6, in the third anuvāka of book x.; 25 begins with xi. 6, in the third anuvāka of book xi.; and 27, as already noted, begins in the middle of the third hymn (and conterminous anuvāka) of book xii.

The division into kāṇḍas or 'books.—⌊The word kāṇḍa means literally 'division' or 'piece,' especially the 'division of a plant-stalk from one joint to the next,' and is applied to the main divisions of other Vedic texts (TS., MS., ÇB., etc.). The best and prevailing rendering of the word is 'book.' As to the length of the kaṇḍas and their arrangement within their respective grand divisions, see p. cxliii, below. The division into kāṇḍas is of course universal, and evidently fundamental.

The division into anuvākas.—The anu-vākas, literally 're-citations,' are subdivisions of the individual book, and are numbered continuously through the book concerned. They are acknowledged by the mss. in very different manner and degree. There is usually added to the anuvāka a statement of the number of hymns and verses contained in it, ⌊and those statements are reproduced in this work in connection with the comment.⌋ ⌊From these it appears that the anuvāka-divisions are sometimes very unequal: thus the last anuvāka of book vi., where the average is 35 verses, has 64.⌋ ⌊In the course of the special introductions to the books, there is given for each of the books vii.-xix. (except xiv. and xvii.) a table showing the number of hymns and the number of verses in each anuvāka: see pages 388 and so on. For xiv. and xvii. also the facts are duly stated, but not in tabular form, which was needless.⌋ The enumeration of verses is often made continuously through the anuvāka (cf. p. 388, end).

Their number, and distribution over books and grand divisions.—The pertinent facts may be shown by a table with added statements. In the table, the first couple of lines refers to the first grand division; the second, to the second; and the last, to the third.

 Books i. ii. iii. iv. v. vi. vii. contain respectively 6 6 6 8 6 13 10 anuvākas. Books viii. ix. x. xi. xii.* contain respectively 5 5 5 5 5 anuvākas. Books xiii.* xiv.* xv. xvi. xvii.* xviii.* contain respectively 4 2 2 2 1 4 anuvākas.

Thus the first grand division has 55 anuvākas; the second has 25; and the third has 15: sum, 95. Moreover, book xix. contains 7, and xx. contains 9. In the colophon to book xvii., neither printed edition has the note prathamo ‘nuvākaḥ; but it is found (cf. p. 812) in the mss. Each of the books viii.-xi. has ten hymns (p. 472), and so each anuvāka there consists of just two hymns. In book xii., of five hymns, the anuvāka is coincident with the hymn. The like is true in books xiii., xiv., xvii., and xviii. (p. 814). In the table, these five books are marked with a star. But furthermore: if, as seems likely (see p. cxxx, below), books xv. and xvi. are to be reckoned each as a book of two hymns (and not as of 18 and 9 respectively), then all the books from xii. on, to xviii., are to be starred, and regarded as having their anuvākas and hymns conterminous.⌋

⌊It is noted at p. 898, ¶2, that in book xix. there appears an attempt to make the anuvāka-divisions coincide with the sense-divisions or divisions between the subject-groups. I do not know whether the same is true in books i.-xviii., not having examined them with regard to this point; it is true in the case of the last anuvāka of book ix. (= RV. i. 164 = AV. ix. 9 and 10), where, as the RV. shows, the true unit is the anuvāka and not the AV. hymn. On the other hand, Whitney observes (at p. 194) that an anuvāka-division falls in the middle of the Mṛgāra-group, and (at p. 247) that another falls between v. 15 and 16 with entire disregard of the close connection of the two hymns.⌋

Their relation to the hymn-divisions in books xiii.-xviii.—In these books and in xii., the anuvāka is, as noted above, admittedly conterminous with the hymn everywhere except in the two paryāya-books, xv. and xvi. In the colophon to xiv. 1, a ms. of Whitney's speaks of the hymn as an anuvāka-sūkta; and it is possible that, for book xiv., at least, the author of the Anukr. did not recognize the hymn-divisions (see p. 739). That they signify very much less in books xiii.-xviii. than they do in the earlier books is very clear (see the third paragraph of p. cxxxi, and the third of p. clx); so clear, that it is not unlikely that they are of entirely secondary origin.⌋

⌊It is at the beginning of book xii. that the anuvāka-divisions begin to coincide with the hymn-divisions; and it is precisely at the corresponding point in the Anukr. (the beginning of paṭala viii.) that the author of that treatise apparently intends to say athā ’nuvākā ucyante. From book xii. on, therefore, it would seem that the saṁhitā was thought of by him as a collection of anuvākas, or that the subordinate division below the kāṇḍa which was alone worthy of practical recognition, was in his opinion the anuvāka and not the sūkta.⌋

⌊If this be right, then it would seem as if, in the series of books xii.-xviii., the books xv. and xvi. ought not to be exceptions. In them, also, the groups of individual paryāyas or paryāya-groups should be conterminous with the anuvākas. Book xv. will fall, accordingly, into two groups of 7 and 11 paryāyas respectively; and book xvi. into two groups of 4 and 5. This method of grouping the paryāyas receives some support from the fact that hymn xix. 23 refers to book xv. as "two anuvākas" (see note to xix. 23. 25), and from the fact that the Pañcapaṭalikā makes similar reference to book xvi. (see p. 792, ¶4, to p. 793), and speaks of our xvi. 5 as ādya, that is, 'the first' of the second group (p. 793). Moreover, the treatment of books xv. and xvi. by the makers of the Pāipp. text (see p. 1016, line 12) would indicate that the anuvāka is here the practically recognized unit subordinate to the kāṇḍa. As for the bearing of this grouping upon the citation of the text concerned and upon the summations, cf. p. cxxxvii, top, and p. cxlv, table 3, both forms⌋

The division into sūktas or 'hymns.'—The hymn may well be called the first considerable natural unit in the rising scale of divisions. Of the hymn, then, verses and pādas are the natural subdivisions, although single verses or even stock-pādas may also be regarded as natural units. Book and hymn[10] and verse are all divisions of so obviously and equally fundamental character, that it is quite right that citations should be made by them and not otherwise. However diverse in subject-matter two successive sūktas may be, we rightly expect unity of subject within the limits of what is truly one and the same sūkta. It is this inherent unity of subject which justifies the use of the term artha-sūkta (below, p. cxxxiii) with reference to any true metrical hymn; and our critical suspicions are naturally aroused against a hymn that (like vii. 35) fails to meet this expectation. The hymn, moreover, is the natural nucleus for the secondary accretions which are discussed below, at p. cliii.⌋

The hymn-divisions not everywhere of equal value.—It is matter of considerable critical interest that the hymn-divisions of different parts of our text are by no means of equal value (cf. p. clx). Thus it is far from certain whether there is any good ground at all for the division of the material of book xiv. into hymns (the question is carefully examined at pages 738-9). And again, the material of book xviii. is of such sort as to make it clear that the hymn-divisions in that book are decidedly mechanical and that they have almost no intrinsic significance (see p. 814, ¶6, p. 827, ¶2, p. 848, ¶8). The familiar Dīrghatamas-hymn of the Rig-Veda has been divided by the Atharvan text-makers into two (ix. 9 and 10), and doubtless for no other reason than to bring it into an approximate uniformity in respect of length with the hymns of books viii.-xi. (p. clvi). As Whitney notes, hymns xix. 53 and 54 are only two divided parts of one hymn: so 10 and 11; 28 and 29.⌋

The division into ṛcas or 'verses.'—This division is, of course, like the division into books and hymns, of fundamental significance. It is maintained even in the non-metrical passages; but the name is then usually modified by the prefixion of the determinative avasāna, so that the prose verses in the paryāya-hymns are called avasānarcas (p. 472).⌋

Subdlvisions of verses: avasānas, pādas, etc.—Concerning these a few words may be said. Avasāna means 'stop,' and so 'the verse-division marked by a stop.' The verse usually has an avasāna or 'stop' in the middle and of course one at the end. Occasionally, however, there are, besides the stop at the end, two others: and the verse is then called tryavasāna. Moreover, we have verses with more than three stops, and sometimes a verse with only one (ekāvasāna).—The next subordinate division is the pāda or 'quarter.' As the name implies, this is commonly the quarter of a four-lined verse or verse with two avasānas; but sometimes, as in a verse with an odd number of pādas (like the gāyatrī), a pāda may be identical with an avasāna. The division into pādas is recognized by the ritual, which sometimes prescribes the doing of a sequence of ceremonial acts to the accompaniment of a verse recited pāda by pāda (pacchas) in a corresponding sequence. —Even the pāda is not the final possible subdivision, as appears from KB. xxvi. 5, ṛcaṁ vārdharcaṁ vā pādaṁ vā padaṁ vā varṇaṁ vā, where the verse and all its subdivisions receive mention.⌋

Numeration of successive verses in the mss.—In this matter, the mss. differ very much among themselves, and the same ms. differs in different books, and even in different parts of the same book; so that to give all details would be a long, tedious, and useless operation. A few may be given by way of specimen. In books iii. and v. the enumeration in our mss. is by hymns only. ⌊Sometimes it runs continuously through the anuvāka: above, p. cxxix.⌋ In vi. it is very various: in great part, 2 hymns are counted together; sometimes 4; also 10 verses together, or 9, or 8. In book vii., some mss. (so P. and I.) number by decads within the anuvāka, with total neglect of real sūktas; and the numbering is in all so confused and obscure that our edition was misled in several cases so as to count 5 hymns less in the book than does the Anukr., or than SPP's edition. The discordance is described at p. 389 and the two numberings are given side by side in the translation.

Groupings of successive verses into units requiring special mention.—The grouping of verses into units of a higher degree is by no means so simple and uniform in the mss. as we might expect. It is desirable, accordingly, to discriminate between decad-sūktas and artha-sūktas and paryāya-sūktas. The differences of grouping are chargeable partly to the differences of form in the text (now verse, now prose) and partly to the differences in length in the metrical hymns.⌋

Decad-sūktas or 'decad-hymns.'—With the second grand division begins (at book viii.) a new element in the subdividing of the text: the metrical hymns, being much longer than most of those in the first division, are themselves divided into verse-decads or groups of ten verses, five or more odd verses at the end of a hymn counting as an added decad. The numbers in the final group thus run from five to fourteen: cf. pages 388, end, and 472, ¶5. Book xvii. divides precisely into 3 decads: p. 805. The average length of the decad-sūktas is exactly ten verses in book x. (35 decads and 350 verses: p. 562), and almost exactly ten in book xviii. (28 decads and 283 verses: p. 814). In the summations, these decads are usually called sūktas and never by any other name (as daçatayas), while the true hymns are called artha-sūktas.

⌊Although known to the comm. and to some mss. in book vii. (p. 388), the decad-division really begins with book viii.; and it runs on through book xviii. (not into xix.: p. 898, line 6), and continuously except for the breaks occasioned by the paryāya-hymns (p. 471, end) and paryāya-hooks (xv. and xvi.: pages 770, 793). In book vii., this grouping is carried out so mechanically as to cut in two some nine of the short sense-hymns of the Berlin edition. The nine are enumerated at p. 389, line 8; but in the case of five of them (45, 54, 68, 72, 76), the fault lies with the Berlin edition, which has wrongly combined the parts thus separated.⌋

⌊In the summations, as just noted, the decads are usually called sūktas; and they and the paryāya-sūktas are added together, like apples and pears, to form totals of "hymns of both kinds" (p. 561, line 8). The summations of the decad-sūktas and paryāya-sūktas for books viii.-xviii. are duly given below in the special introduction to each book concerned, and these should be consulted; but for convenience they may here be summarized.

 Book viii. ix. x. xi. xii. xiii. xiv. xv. xvi. xvii. xviii. Decads 24 21 35 27 23 14 14 3 28 Paryāyas 6 7 3 7 6 18 9 ⌋

Artha-sūktas or 'sense-hymns.'—⌊This technical term might be rendered, more awkwardly, but perhaps more suitably, by 'subject-matter hymns.' It is these that are usually meant when we speak of "hymns" without any determinative. The comm. very properly notes that hymns xix. 47 and 48 form a single artha-sūkta, and that the next two form another. The determinative artha- is prefixed in particular to distinguish the sense-hymns from the paryāya-hymns (p. 611, ¶5), and there is little occasion for using it of the short hymns of the first grand division.⌋ The verses of the artha-sūkta are sometimes numbered through each separate component decad or sūkta, and sometimes through the whole artha-sūkta, the two methods being variously mingled. In books xii.-xiv. and xvii. and xviii., as already noted, the artha-sūktas and anuvākas are coincident, the mss. specifying their identity.

Paryāya-sūktas or 'period-hymns.'—In the second and third grand divisions are certain extended prose-compositions called⌋ paryāya-sūktas. They are divided into what are called paryāyas, or also paryāya-sūktas, but never into decads. ⌊The term paryāya-sūkta is thus somewhat ambiguous, and has a wider and a narrower meaning as designating, for example, on the one hand, the whole group of six paryāyas that compose our ix. 6, or, on the other, a single one of those six (e.g. our ix. 6. 1-17). To avoid this ambiguity, it is well to use paryāya only for the narrower meaning and paryāya-sūkta only for the wider. The hymn ix. 7 is a paryāya-sūkta consisting of only one paryāya. For the word pary-āya (root i: literally Um-gang, circuit, περίοδος) it is indeed hard to find an English equivalent: it might, with mental reservations, be rendered by 'strophe'; perhaps 'period' is better; and to leave it (as usual) untranslated may be best.⌋

⌊The paryāya-hymns number eight in all, five in the second grand division (with 23 paryāyas), and three in the third grand division (with 33 paryāyas). They are, in the second division, viii. 10 (with 6 paryāyas); ix. 6 (with 6) and 7 (with 1); xi. 3 (with 3); and xii. 5 (with 7); and, in the third division, xiii. 4 (with 6); book xv. (18 paryāyas); and book xvi. (9 paryāyas). The paryāya-sūktas are marked with a P in tables 2 and 3. For further details, see p. 472.⌋

⌊It will be noticed that two books of the third division, xv. and xvi., consist wholly of paryāyas; and, further, that each book of the second division has at least one of these hymns (ix. has two such, and contiguous), except book x. Even book x. has a long hymn, hymn 5, consisting mostly of prose, but with mingled metrical portions; but despite the fact that the Anukr. divides the hymn into four parts, which parts are even ascribed to different authors (p. 579), it is yet true that those parts are not acknowledged as paryāyas. Moreover, the hymn is expressly called an artha-sūkta by at least one of Whitney's mss.⌋

Differences of the Berlin and Bombay numerations in books vii. and xix.—As against the Berlin edition, the Bombay edition exhibits certain differences in respect of the numeration of hymns and verses. These are rehearsed by SPP. in his Critical Notice, vol. i., pages 16-24. Those which affect book vii. are described by me at p. 389, and the double numberings for book vii. are given by Whitney from vii. 6. 3 to the end of vii. The Bombay numberings are the correct ones (cf. p. 392, line 4 from end). Other discrepancies, which affect book xix., are referred to at p. 898.⌋

Differences of hymn-numeration in the paryāya-books.—These are the most important differences that concern hymns. They affect all parts of a given book after the read paryāya of that book. They have been carefully explained by me at pages 610-11, but the differences will be more easily apprehended and discussed if put in tabular form. The table harmonizes the hymn-numbers, without going into the detail of the difference of verse-numberings, which latter, however, are not seriously confusing.

 Hymns of the Bombay ed. The underwritten hymns or parts of hymns of the Berlin edition correspond to the hymns of the Bombay edition as numbered in either margin. Hymns of the Bombay ed. Book viii. Book ix. Book xi. Book xii. Book xiii. 1 1 1 1 1 1 1 2 2 2 2 2 2 2 3 3 3 31-31 3 3 3 4 4 4 332-49 4 41-13 4 5 5 5 350-56 51-6 414-21 5 6 6 61-17 4 57-11 422-28 6 7 7 618-30 5 512-27 429-28 7 8 8 631-39 6 528-38 446-51 8 9 9 640-44 7 539-46 452-56 19 10 101-7 645-48 8 547-61 10 11 108-17 649-62 9 562-73 11 12 1018-21 71-26 10 12 13 1022-25 8 13 14 1026-29 9 14 15 1030-33 10 15⌋

⌊Since the two editions differ, the question arises, Which is right? The fourth paragraph of p. 611 (which see) leaves it undecided, but states the real point at issue plainly. I now believe that the methods of both editions are at fault and would suggest a better method. To make the matter clear, I take as an example the paryāya-sūkta xi. 3, which consists of a group of three paryāyas.

 Suggested method Berlin method Bombay method xi. 3. 11-31 xi. 3. 1-31 xi. 3. 1-31 xi. 3. 21-18 xi. 3. 32-49 xi. 4. 1-18 xi. 3. 31-17 xi. 3. 50-56 xi. 5 1-7

The four sets of numbers in the first column relate to the four text-divisions: the first set (xi.) to the book; the second (3) to the paryāya-sūkta or group of paryāyas; the third (1, 2, 3) to the individual paryāya; of that group; and the fourth (1-31, 1-18, 1-7) to the verses of the paryāyas.⌋

⌊In the Berlin text, on the one hand, we must admit that each of the three component paryāyas of xi. 3 is duly indicated as such by typographical separation and that the paryāya-numbers (1 and 2 and 3) are duly given in parenthesis. That text, however, practically ignores the paryāyas, at least for the purposes of citation, by numbering the verses of all three continuously (as verses 1-56) from the beginning of paryāya 1 to the end of 3. Thus only the group of paryāyas is recognized; and it is numbered as if coördinate with the artha-sūktas of the book.⌋

⌊In the Bombay text, on the other hand, each paryāya is numbered as if coordinate with an artha-sūkta, and the verses are numbered (of course, in this case) beginning anew with 1 for each paryāya. This method ignores the unity of the group of individual parydyas and throws previous citations into confusion.⌋

⌊Books xv. and xvi. consist wholly of paryāyas. Here, therefore, the case is not complicated by the mingling of paryāyas and artha-sūktas, and the Berlin text ignores the grouping[11] of the paryāyas, and treats and numbers each paryāya as if coördinate with artha-sūktas, and numbers the verses beginning anew with 1 for each paryāya (cf. p. 770, line 30).⌋

Whitney's criticism of the numbering of the Bombay edition.—⌊Whitney condemned, at p. 625, the procedure of the Bombay edition. In his material for this Introduction, I now find a few additional words on the matter, which may well be given.⌋

Each paryāya is reckoned, in the summations, as on the same plane as a real hymn or artha-sūkta. Hence SPP. is externally justified in counting, for example, the nine artha-sūktas and three paryāyas of book xi. as twelve hymns, numbering the verses of each separately; at the same time, such a deviation from the method pursued in our edition, throwing into confusion all older references to book xi. after 3. 31, was very much to be deprecated, and has no real and internal justification, since each body or group of paryāyas is obviously and undeniably a unitary one (see, for example, our viii. 10, and note the relation especially of its third and fourth and fifth subdivisions or paryāyas). In such matters we are not to allow the mss. to guide us in a manner clearly opposed to the rights of the case.

Suggestion of a preferable method of numbering and citing.—It is plain, I think, that both editions are at fault: the Berlin edition, in ignoring the individual paryāyas in books viii.-xiii. and in ignoring the paryāya-groups in xv.-xvi.; and the Bombay edition, in ignoring the paryāya-groups everywhere. Moreover, the procedure of the Berlin text is inconsistent (p. 770, line 27) as between books viii.-xiii. and books xv.-xvi., the unity of the groups in xv.-xvi. being no less "obvious and undeniable" than in the example just cited by Whitney.⌋

⌊The purpose underlying the procedure of the Berlin edition was that all references should be homogeneous for all parts of the Atharvan text, not only for the metrical parts but also for the prose paryāyas, and consist of three numbers only. But, as between the paryāyas and the rest, it is precisely this homogeneity that we do not want; for the lack of it serves the useful purpose of showing at a glance whether any given citation refers to a passage in prose or in verse.⌋

⌊For a future edition, I recommend that all paryāya-passages be so numbered as to make it natural to cite them by book, paryāya-group paryāya, and verse. The verse-number would then be written as an exponent or superior; and, for example, instead of the now usual ix. 6. 31, 45; 7. 26; xi. 3. 50, we should have ix. 6. 31, 51; 7. 126; xi. 3. 31. In books xv. and xvi. I should reckon the anuvāka as determining the limits of each group of paryāyas (p. cxxx); and thus, for example, instead of the now usual xv. 7. 1; 8. 1; 17. 1; xvi. 5. 1, we should have xv. 1. 71; 2. 12; 2. 101; xvi. 2. 11. The tables on pages 771 and 793 may serve for conversion-tables as between the proposed method and the Berlin-Bombay method.⌋

⌊The merits of this method are clear from what has been said: it avoids ignoring the paryāyas of viii— xiii. and the groups of xv.-xvi., and avoids the inconsistency of the Berlin method; it maintains the recognition of the uniformity of books viii.-xi. as books of ten hymns. each (p. 611, line 25); and it assimilates all references to paryāya-text in a manner accordant with the facts, and shows at a glance that they refer to paryāya-passages.[12] Moreover, it avoids the necessity of recognizing hymns of less than 20 verses for division III. (p. cxlv); and by it one is not inconvenienced in finding passages as cited by the older method.⌋

Differences of verse-numeration.—The differences of hymn-numeration, as is clear from the foregoing, involve certain differences of verse-numeration also; but besides these latter, there are certain other differences of verse-numeration occasioned by the adherence of the Bombay editor to the prescriptions of the Anukramaṇīs. They have been fully treated in the introductions to the books concerned; but require mention here because they affect the verse-totals of the tables considered in the discussion (pages clvii, clix) of the structure of the text. The five paryāya-hymns, affected are given in the first line of the subjoined table, and in the second are set references to the pages of this work where the Bombay totals are given. The third line gives the totals of avasānarcas for the Bombay edition, and the fourth those for the Berlin edition, and the fifth the differences. It may be well to remind the reader, that, in its proper place in the text, the second paryāya of xi. 3 is printed, both by RW. and by SPP. (at vol. iii., pages 66-83), as 18 numbered subdivisions; but that the Bombay editor prints it again (just after p. 356 of the same vol.), this time as 72 avasānarcas, as required by the Anukr. The matter is fully explained by me, pages 628-9. The totals for xi. 3 in the one ed. are 31 + 18 + 7 = 56, and in the other 31 + 72 + 7 = 110, a difference of 54. The sum of the plus items is 188.

 Paryāya-hymns viii. 10 ix. 6 xi. 3 book xv. book xvi. [See pages 516 546 632 771 793] Bombay totals 67 73 110 220 103 Berlin totals 33 62 56 141 93 Plus items 34 11 54 79 10⌋
Summations of hymns and verses at end of divisions.—These are made in the mss. at the end of the division summed up, and constitute as it were brief colophons; and the details concerning them are given in the notes at the points where they occur. ⌊For examples, see the ends of the several anuvākas and books: thus, pages 6, 12, 18, 22, 29, 36, and so on. The summations become somewhat more elaborate and less harmonious in the later books: see, for example, pages 516, 561, 659, 707, 737.⌋

The summations quoted from the Pañcapaṭalikā.—A peculiar matter to be noted in connection with the summations just mentioned is the constant occurrence with them, through books i.-xviii., of bits of extract from an Old Anukramaṇī, as we may call it: catch-words intimating the number of verses in the divisions summed up. ⌊For details respecting this treatise, see above, p. lxxi.⌋ These citations are found accordantly in all the mss.—by no means in all at every point; they are more or less fragmentary in different mss.; but they are wholly wanting in none of ours (except K. ⌊and perhaps L.⌋). The phrases which concern the end of a book are the ones apt to be found in the largest number of mss. In book vii. there is a double set, the extra one giving the number of hymns in the anuvāka.

Indication of the extent of the divisions by reference to an assumed norm.—In giving the summations of verses, it is by no means always the case that the Pañcapaṭalikā expresses itself in a direct and simple way. Sometimes indeed it does so where its prevailing method would lead us to expect it to do otherwise: thus in book vi., where the normal number of verses to the anuvāka is 30, it says simply and expressly that anuvākas 3 and 4 have 33 verses each (trayastriṅçakāu: p. 311) and that 5 and 6 have 30 each (triṅçakāu: p. 1045). Very often, however, the extent of a division is intimated by stating its overplus or shortage with reference to an assumed norm. One hardly knows how much critical value to assign to the norms (the last anuvāka of book vi., with 64 verses, exceeds the norm of 30 by more than the norm itself); but the method is a deviation from straightforwardness of expression, and that deviation is increased, as is so often the case, by the gratuitous exigencies of the metrical form into which the Pañcapaṭalikā is cast. Thus for book v. it says (pages 230, 236), 'the first [anuvāka] falls short of sixty by twice six and the next after the first by eleven.' So forty-two is in one place (p. 61) 'half-a-hundred less eight,' and in another (p. 439) it is 'twice twenty-one.' For anuvāka 3 of book vii. the total is 31 (norm 20); but here (p. 413) not even the overplus is stated simply as 'eleven,' but rather as 'eight and three.' This method of reference to a norm is used even where the departure from it is very large, as in the case of anuvāka 3 of book iv., which is described (p. 176) as having 21 verses over the norm of 30.⌋

Tables of verse-norms assumed by the Pañcapaṭalikā.—For the first grand division (books i.-vii.), on the one hand, this treatise assumes a norm for the verse-totals of the anuvākas of each book.[13] These may be shown in tabular statement as follows:

 For book i. ii. iii. iv. v. vi. vii. Verse-totals 153 207 230 324 376 454 286 Anuvākas 6 6 6 8 6 13 10 Averages 25 34 38 40 63 35 29 Anuvāka-Norms 20 20 30 30 60 30 20

The norm is spoken of (p. 92) as a nimitta, literally, perhaps, 'fundamental determinant.' Frequent reference has already been made to these norms in the main body of this work, either expressly (as at pages 220 and 388: of. also pages 6, 18, 22, 152), or implicitly at the ends of the anuvākas.⌋

⌊By combining (as in lines 2 and 3 of the table) a part of table 1 of p. cxliv with a part of the table on p. cxxix, the actual average of the verse-totals of the anuvākas may be found for each book (as in line 4). It is perhaps a fact of critical significance that for each book this average is greater than the norm assumed by our treatise.⌋

⌊For the second grand division (books viii.-xii.), on the other hand, our treatise assumes a norm which concerns the verse-totals of the hymns, and not (as in the first division) those of the anuvākas. They are, in tabular statement, as follows:

 For book viii. ix. x. xi. xii. Verse-totals 259 302 350 313 304 Hymns 10 10 10 10 5 Averages 26 30 35 31 61 Hymn-Norms 20 20 30 20 60

The lengths of the hymns are often (not always) described by stating the overplus or shortage with reference to these norms. This is oftenest the case in book x. (so with seven hymns out of ten: see p. 562); it is the case with all the artha-sūktas of book xii. (four out of five: p. 660); with hymns 1, 3, and 5 of book ix., and 6 and 8 of book xi.; and least often and clearly the case with book viii. (cf. the unclear citation, p. 502, ¶2).—Here again the actual averages are greater than the norms.⌋

The three "grand divisions" are recognized by the Pañcapaṭalikā.—Partly by way of example, and partly with ulterior purpose, we may instance the citations from the Pañcapaṭalikā which give the verse-totals of the six anuvākas of book iii. These totals are respectively 33, 40, 38, 40, 35, and 44. The citations are indeed to be found below, scattered over pages 92, 103, 113, 123, 131, and 141; but it will be better to combine them here (with addition of the "obscure" clause of p. 141, ¶8) into what appears to be their proper metrical form, with attempted emendation at the points[14] in which the verse was obscure to Whitney:

triṅçannimittāḥ ṣaḍṛceṣu kāryās
tisro daçā ’ṣṭāu daça pañca ca rcaḥ:
caturdaçā ’ntyā; anuvākaçaç ca

'Among the six-versed [hymns] (i.e. in book iii.), the verses are to be (made: i.e.) accounted [respectively] as three, ten, eight, ten, and five, with thirty as their fundamental determinant; and the last as fourteen: and one is to treat the number (anuvāka by anuvāka: i.e.) for each anuvāka as an overplus over the norm.'⌋

⌊In the section headed "Tables of verse-norms" etc., it was shown that, while the Pañcapaṭalikā's norms for books i.-vii. concern the anuvākas, its norms for books viii.-xii. concern the hymns. This distinction is observed also by the comm. in making his decad-divisions (see p. 472: l. 28). These facts are in entire accord with the explicit statements of the Pañcapaṭalikā: to wit, on the one hand, with that of the verse just translated; and, on the other, with the remark cited at the end of viii. i (p. 475, end), sūktaçaç ca gaṇanā pravartate, 'and the numbering proceeds hymn by hymn.' Here sūktaças is in clear contrast with the anuvākaças of our verse, and the remark evidently applies to the remaining books of the text that come within the purview of the Pañcapaṭalikā, that is (since it ignores books xix.-xx.), to books viii.-xviii. or to the second and third grand divisions.⌋

⌊Thus, between the first grand division on the one hand and the second and third on the other, our treatise makes a clear distinction, not only by actual procedure but also by express statement. But this is not all. As between the second and the third, also, it makes a distinction in fact: for, while a norm that concerns the verse-totals of artha-sūktas (and not of anuvākas) is assumed for the second, no norm is assumed for the third (cf. p. 708, line 12) and the verse-totals for each artha-sūkta or paryāya-sūkta are stated simply hymn by hymn.⌋

10. Extent and Structure of the Atharva-Veda Saṁhitā

Limits of the original collection.—It is in the first place clearly apparent that of the twenty books composing the present text of the Atharva-Veda, the first eighteen, or not more than that, were originally combined together to form a collection. There appears to be no definite reason to suppose that the text ever contained less than the books i.-xviii. It is easy to conjecture a collection including books i.-xiv. and book xviii., leaving out the two prose paryāya-books xv. and xvi. and the odd little book xvii. with the queer refrain running nearly through it; but there is no sound reason for suspecting the genuineness of these prose books more than of the prose hymns scattered (see below, p. 1011) through nearly all the preceding books; and in the Pāippalāda recension it is Vulgate book xviii. that is wanting altogether, books xv.-xvii. ⌊or rather, books xv.-xviii.: cf. p. 1015⌋ being not unrepresented.

Books xix. and xx. are later additions.—That these are later additions is in the first place strongly suggested by their character and composition. As for book xx., that is in the main a pure mass of excerpts from the Rig-Veda; it stands in no conceivable relation to the rest of the Atharva-Veda; and when and why it was added thereto is a matter for conjecture. As for book xix., that has distinctly the aspect of being an after-gleaning; if its hymns had been an accepted part of the main collection when that was formed, we should have expected them to be distributed among the other books; and the text is prevailingly of a degree of badness that sets it quite apart from the rest; while its pada-text must be a most modern production. ⌊For the cumulative evidence in detail respecting book xix., see my introduction, pages 895-8.⌋

Other evidences of the former existence of an Atharva-Veda which was limited to books i.-xviii. are not rare. That the prapāṭhaka-division is not extended beyond book xviii. may be of some consequence, but probably not much. The Old Anukramaṇī stops at the same point. More significant is it that the Kāuçika-sūtra ⌊does not, by its citations,[15] imply recognition of the text of book xix. as an integral part of the saṁhitā, and that it⌋ ignores book xx. completely. It is yet more important that the Prātiçākhya and its commentary limit themselves to books i.-xviii.

In the Pāippalāda text, the material of book xix. appears in great part, as we are bound to note, and quite on an equality with the rest. Of book xx., nothing ⌊or practically nothing: see p. 1009⌋ so appears. It is also noteworthy that Pāipp. (as mentioned above) omits book xviii.; but from this need be drawn no suspicion as to the appurtenance of xviii. to the original AV.—The question of the possible extension of individual hymns anywhere does not concern us here, ⌊but is discussed on page cliii.⌋

The two broadest principles of arrangement of books i.-xviii.—Leaving book XX. out of account, and disregarding also for the present book xix. as being a palpable supplement (see pages 895-8), it is not difficult to trace the two principles that underlie the general arrangement of the material of books i.-xviii. These principles are:⌋

1. Miscellaneity or unity of subject and 2. length of hymn.—The books i.-xviii. fall accordingly into two general classes: 1. books of which the hymns are characterized by miscellaneity of subject and in which the length of the hymns is regarded; and 2. books of which the distinguishing characteristic is a general unity of subject and in which the precise length of the hymns is not primarily regarded, although they are prevailingly long.[16] The first class again falls into two divisions: 1. the short hymns; and 2. the long hymns.⌋

The three grand divisions (I. and II. and III.) as based on those principles.—We thus have, for books i.-xviii., three grand divisions, as follows: I. the first grand division, consisting of the seven books, i.-vii., and comprehending the short hymns of miscellaneous subjects, more specifically, all the hymns (not paryāyas: p. cxxxiv) of a less number of verses than twenty[16]; II. the second grand division, consisting of the next five books, viii.-xii., and comprehending the long hymns of miscellaneous subjects, more specifically, all the hymns (save those belonging to the third division) of more than twenty verses; and III. the third grand division, consisting, as aforesaid, of those books of which the distinguishing characteristic is a general unity of subject, to wit, the six books, xiii.-xviii. —There are other features, not a few, which differentiate these divisions one from another; they will be mentioned below, under the several divisions.⌋

The order of the three grand divisions.—It is clear that the text ought to begin with division I., since that is the most characteristic part of it all, and since books i.-vi. are very likely the original nucleus of the whole collection. Since division I. is made up of hymns of miscellaneous subjects (the short ones), it is natural that the other hymns of miscellaneous subjects (the long ones) should follow next. Thus the last place is naturally left for the books characterized by unity of subject. This order agrees with that of the hymn-totals of the divisions, which form (cf. tables 1, 2, 3) a descending scale of 433 and 45 and 15.⌋

Principles of arrangement of books within the grand division.—If we have rightly determined the first rough grouping of the material of books i.-xviii. into three grand divisions, the question next in logical order is, What governs the arrangement of the books within each division? This question will be discussed in detail under each of the three divisions (cf. pages cxlix ff., clvii, clix); here, accordingly, only more general statements are called for. Those statements concern the verse-norms of the hymns for each book, and the amount of text.⌋

The normal length of the hymns for each of the several books.—For the first grand division these norms play an important part in determining the arrangement of the books within that division, as appears later, p. cxlix. For the second grand division it is true that the Pañcapaṭalikā assumes a normal hymn-length for each book (p. cxxxix); but that seems to have no traceable connection with the arrangement of the books within that division (p. clv). For the third, no such norm is even assumed (p. cxl, near end).⌋

The amount of text in each book.—Table.—This matter, in its relation to the order of the books, I must consider briefly here for the three grand divisions together, although it will be necessary to revert to it later (pages clii, clvii, clix). Since our saṁhitā is of mingled verse and prose, it is not easy (except with a Hindu ms., which I have not at hand) to estimate the precise amount of text to be apportioned to each separate book. If we take as a basis, however, the printed page of the Berlin text, and count blank fractions of pages, the 352 pages are apportioned among the 18 books as follows:

 Book i. has 13 pages Book viii. has 22 pages Book xiii. has 13 pages ii. 16 ix. 21 xiv. 12 iii. 20 x. 27 xv. 10 iv. 27 xi. 25 xvi. 5 v. 28 xii. 22 xvii. 3 vi. 40 xviii. 21 vii. 27 Division I. 171 Division II. 117 Division III. 64

From this it appears that, for division I., the amount of text is a continuously ascending one for each of the books except the last (book vii.); and that, for division III., it is a continuously descending one for each of the books except (in like manner) the last (book xviii.); and that, although the verse-totals of the Bombay edition for the books of division II. form a series (see p. clvii, line 11) which ascends continuously (like that of I.) for all books except (once again) the last, the books of division II. are, on the whole, most remarkable for their approximate equality of length.⌋

Arrangement of the hymns within any given book.—While the general guiding principles of arrangement of the books within the division are thus in large measure and evidently the external ones of verse-norms and amount of text, it is not easy to see what has directed the ordering of the

 Table 1. First grand division, books i.-vii., seven books Book viii. Book vi. Book i. Book ii. Book iii. Book iv. Book v. Verse-norms ${\displaystyle \scriptstyle {\left.{\begin{matrix}\ \end{matrix}}\right\}\,}}$ 1 2 3 4 5 6 7 8 contains Sum of hymns Sum of verses 56 hs. of 1 vs. 56 56 26 hs. of 2 vss. 26 52 10 122 hs. of 3 vss. 132 396 11 12 30 hs. of 4 vss. 53 212 3 8 1 22 hs. of 5 vss. 34 170 4 2 5 13 hs. of 8 vss. 24 144 3 1 5 6 21 hs. of 7 vss. 36 252 3 4 6 10 2 hs. of 8 vss. 25 200 1 1 2 3 4 hs. of 9 vss. 11 99 2 3 2 hs. of 10 vss. 7 70 1 1 6 hs. of 11 vss. 8 88 2 5 hs. of 12 vss. 7 84 1 3 hs. of 13 vss. 4 52 3 hs. of 14 vss. 3 42 3 hs. of 15 vss. 3 45 1 h. of 16 vss. 1 16 2 hs. of 17 vss. 2 34 1 h. of 18 vss. 1 18 118 142 35 36 31 40 31 hymns 433 286 454 153 207 230 324 376 verses 2,030
 Table 2. Second grand division, books viii.-xii., five books Book viii. Book ix. Book x. Book xi. Book xii. contains Sum of hymns Sum of verses 1 h. of 21 vss. 1 21 1 2 hs. of 22 vss. 3 66 1 h. of 23 vss. 1 23 1 2 hs. of 24 vss. 3 72 1 1 1 hs. of 25 vss. 3 75 3 1 P 1 3 hs. of 26 vss. 8 208 1 2 hs. of 27 vss. 3 81 2 1 hs. of 28 vss. 3 84 1 1 hs. of 31 vss. 2 62 1 h. of 32 vss. 2 32 1 P 1 hs. of 33 vss. 2 66 1 1 hs. of 34 vss. 2 68 1 h. of 35 vss. 1 35 1 h. of 37 vss. 1 37 1 h. of 38 vss. 1 38 2 hs. of 44 vss. 2 88 1 h. of 50 vss. 1 50 1 h. of 53 vss. 1 53 1 h. of 55 vss. 1 55 1 P h. of 56 vss. 1 56 1 h. of 60 vss. 1 60 1 P h. of 62 vss. 1 62 1 h. of 63 vss. 1 63 1 P h. of 73 vss. 1 73 10 10 10 10 5 hymns 45 259 302 350 313 304 verses 1,528
 Table 3. Third grand division, books xiii.-xviii., six books Rohita Book xiii. Wedding Book xiv. Vrātya Book xv. Paritta Book xvi. Sun Book xvii. Funeral Book xviii. contains Sum of hymns Sum of verses 2 hs. of 3 vss. 2 6 1 1 hs. of 4 vss. 2 8 2 hs. of 5 vss. 2 10 1 3 hs. of 8 vss. 4 24 2 1 hs. of 7 vss. 3 21 1 h. of 8 vss. 1 8 3 hs. of 9 vss. 3 27 1 hs. of 10 vss. 1 10 4 1 hs. of 11 vss. 5 55 1 h. of 12 vss. 1 12 2 hs. of 13 vss. 2 26 1 h. of 28 vss. 1 26 1 h. of 27 vss. 1 27 1 h. of 30 vss. 1 30 1 h. of 46 vss. 1 46 1 P h. of 56 vss. 1 56 1 1 hs. of 60 vss. 2 120 1 h. of 81 vss. 1 61 1 h. of 64 vss. 1 64 1 h. of 73 vss. 1 73 1 h. of 75 vss. 1 75 1 h. of 89 vss. 1 89 4 2 18 P 9 P 1 4 hymns 38 188 139 141 93 30 238 verses 874

⌊Such is Whitney's table; and it is well to let it stand, as it furnishes the best argument against treating the paryāyas of books xv. and xvi. each as a singel hymn (cf. p. cxxxvi, top). Treating the as explained at p. cix, it will appear as follows.

 Table 3, second form Rohita Book xiii. Wedding Book xiv. Vrātya Book xv. Paritta Book xvi. Sun Book xvii. Funeral Book xviii. contains Sum of hymns Sum of verses 1 h. of 28 vss. 1 26 1 h. of 30 vss. 1 30 1 P h. of 32 vss. 1 32 1 h. of 46 vss. 1 46 1 P h. of 50 vss. 1 50 1 P h. of 58 vss. 1 56 1 1 h. of 80 vss. 2 120 1 P 1 h. of 61 vss. 2 122 1 h. of 64 vss. 1 64 1 h. of 73 vss. 1 73 1 h. of 75 vss. 1 75 1 h. of 89 vss. 1 89 1 P h. of 91 vss. 1 91 4 2 2 P 2 P 1 4 hymns 15 188 139 141 93 30 283 verses 874⌋
several hymns within any given book. It is clear that the subject has not been at all considered; nor is it at all probable that any regard has been had to the authorship, real or claimed (we have no tradition of any value whatever respecting the "rishis"). Probably only chance or arbitrary choice of the arranger dictated the internal ordering of each book. ⌊On this subject there is indeed little that is positive to be said; but (in order to avoid repetition) I think it best to say that little for each grand division in its proper place under that division: see pages cliv, clvii, and clx.⌋

Distribution of hymns according to length in the three grand divisions.—Tables 1 and 2 and 3.—The distribution of the hymns according to their length throughout the books of the three grand divisions is shown by Whitney's tables 1, 2, and 3, preceding. The numbers rest on the numerations of the Berlin edition, and due reference to the differences of numeration of the Bombay edition is made below at p. cxlvii. A vertical column is devoted to each book and in that column is shown how many hymns of 1 verse, of 2 or 3 or 4 and so on up to 89 verses, there are in that book, by the number horizontally opposite the number of verses indicated in the column headed by the word "contains." To facilitate the summation of the number of hymns and verses in the Atharva-Veda, the last column but one on the right gives the number of hymns of 1 vs., of 2 vss. and so on, in the division concerned, and the last column on the right gives the total number of verses contained in the hymns of 1 vs., of 2 vss. and so on (the total in each line being, of course, an exact multiple of the number preceding in the same line). Accordingly we may read, for example, the sixth line of table 1 as follows: "Book vii. contains 10 hymns of 3 verses and book vi. contains 122. The sum of hymns of 3 verses in the division is 132, and the sum of verses in those hymns is 396."⌋

Tables 1 and 2 and 3 for divisions I. and II. and III.—These ought properly to come in at this point; but as their form and contents are such that it is desirable to have them stand on two pages that face each other, they have been put (out of their proper place) on pages cxliv and cxlv.⌋

Grouping of the hymns of book xix. according to length.—Table 4.—Apart from the two hymns, 22 (of 21 verses) and 23 (of 30), which are in divers ways of very exceptional character, it appears that every hymn of this book, if judged simply by its verse-total length, would fall into the first grand division, as being of less than 20 verses.[17] This fact is of critical interest, and is in keeping with the character of book xix. as an after-gleaning, and in particular an after-gleaning of such material as would properly fall into the first grand division (cf. p. 895, ¶2). The table:

 Table 4. The supplement, book xix., one book In book xix. there are 15 4 2 9 6 8 4 3 1 12 2 hymns, Containing respectively 1 2 3 4 5 6 7 8 9 10 11 verses. In book xix. there are 2 1 1 1 1 hymns, Total: 72 hymns. Containing respectively 14 15 16 21 30 verses. Total: 456 verses.⌋

Summary of the four tables.—Table 5.—Extent of AV. Saṁhitā about one half of that of RV.—The totals of hymns and verses of tables 1-4 are summed up in table 5. From this it appears that the number of hymns of the three grand divisions of the Atharva-Veda Saṁhitā is 516 or about one half of that of the Rig-Veda, and that the number of verses is 4,432 or considerably less than one half. If the summation be made to include also the supplement and the parts of book xx. which are peculiar to the AV., the number of hymns amounts to 598 or about three fifths of that of the RV., and the number of verses amounts to 5,038 or about one half of that of the RV. Table 5 follows:

 Table 5. Summary of Atharvan hymns and verses Grand division I., books i.-vii. contains 433 hymns and 2030 verses. Grand division II., books viii.-xii., contains 45 hymns and 1528 verses. Grand division III., books xiii.-xviii., contains 38 hymns and 874 verses. Totals for the three grand divisions: 516 hymns and 4432 verses. The supplement, book xix., contains 72 hymns and 456 verses. Totals for books i.-xix.: 588 hymns and 4888 verses. The Kuntāpa-khila of book xx. contains 10 hymns and 150 verses. Totals for books i.-xix. and khila: 595 hymns and 5038 verses.⌋

⌊The numbers of tables 1-5 rest on the Berlin edition. The differences between that and the Bombay edition do not affect the amount of text, but only the verse-totals. Even the verse-totals are not affected, but only the hymn-totals (p. 389, l. 10), by the differences in book vii. For the paryāya-hymns, the verse-totals of the Bombay edition amount to 188 more (see p. cxxxvii) than those of the Berlin edition. For the Bombay edition, accordingly, the grand total must be raised (by 188) from 5,038 to 5,226.⌋

First grand division (books i.-vii.): short hymns of miscellaneous subjects.—While the general considerations of length and subject are indeed sufficient for the separation of books i.-xviii. into three grand divisions as defined above, the first division shows yet other signs of being a minor collection apart from the other two. In the first place, the hymns that compose it are mostly genuine charms and imprecations, and wear on the whole a general aspect decidedly different from that of books viii— xviii., as is indeed apparent enough from the table of hymn-titles, pages 1024-37; they are, in fact, by all odds the most characteristic part of the Atharva-Veda, and this is tacitly admitted by the translators of selected hymns (see p. cvii), their selections being taken in largest measure (cf. p. 281) from this division. In the second place, the books of this division are sharply distinguished from those of the others by the basis of their internal arrangement, which basis is in part that of a clearly demonstrable verse-norm, a norm, that is to say, which, for each separate book, governs the number of verses in the hymns of that book.[18]

Evidence of fact as to the existence of the verse-norms.—A most pervading implicit distinction is made by the Major Anukramaṇī between this division and the next in its actual method of giving or intimating the length of the hymns. In division II., on the one hand, the number of verses is stated expressly and separately for every hymn. In division I., on the other hand, the treatise merely intimates by its silence that the number for any given hymn conforms to the norm assumed for that book, and the number is expressly stated only when it constitutes a departure from that norm. Thus for the 142 hymns of book vi., an express statement as to the length is made only for the 20 hymns (given at p. 281, lines 17-18) which exceed the norm of three.[19]—For convenience of reference, the norms may here be tabulated:

 Books vii. vi. i. ii. iii. iv. v. Norms 1 3 4 5 6 7 8⌋

Express testimony of both Anukramaṇīs as to the verse-norms.—The Major Anukr. (at the beginning of its treatment of book ii.: see p. 142) expressly states that the normal number of verses for a hymn of book i. is four, and that the norm increases by one for each successive book of the first five books: pūrvakāṇḍasya caturṛcaprakṛtir ity evam uttarottarakāṇḍeṣu ṣaṣṭhaṁ yāvad ekāikādhikā etc. Than this, nothing could be more clear or explicit. Again, at the beginning of its treatment of book iii., it says that in this book it is to be understood that six verses are the norm, and that any other number is a departure therefrom: atra ṣaḍṛcaprakṛtir anyā vikṛtir iti vijāntyāt. At the beginning of book iv. it has a remark of like purport: brahma jajñānam iti kāṇḍe saptarcasūktaprakṛtir (so London ms.: cf. p. 142 below) anyā vikṛtir ity avagachet. Moreover, it defines book vi. as the tṛcasūktakāṇḍam (cf. pages 281, 388), and adds to the definition the words tatra tṛcaprakṛtir itarā vikṛtir iti. Cf. Weber's Verzeichniss der Berliner Sanskrit-hss., vol. ii., p. 79.⌋

⌊In the recognition of the verse-norms, as in much else (p. lxxii, top), the Pañcapaṭalikā serves as source and guide for the author of the Major Anukr. Thus the older treatise calls book ii. 'the five-versed' (see the citation at p. 45), and book iii. in like manner 'the six-versed' (see p. cxl). Cf. also the statements of the next paragraph as to book vii.⌋

One verse is the norm for book vii.—The absence of any book in which two-versed hymns are the norm, and the frequency of two-versed hymns in book vii., might lead us to think that both one-versed and two-versed hymns are to be regarded as normal for book vii. (cf. p. 388, line 13); but this is not the case (cf. line 24 of the same page). The Major Anukr. speaks of book vii. as 'the book of one-versed hymns,' ekarcasūktakāṇḍām; and its testimony is confirmed by the Old Anukr., as cited by SPP. on p. 18 of his Critical Notice, which says, 'among the one-versed hymns [i.e. in book vii.], [the anuvākas are or consist] of hymns made of one verse,' ṛk-sūktā ekarceṣu. Further confirmation of the view that one (not one or two) is the true norm for book vii. is found in the fact that the Anukr. is silent as to the length of the hymns of one verse (cf. p. cxlviii), but makes the express statement dvyṛcam for each of the thirty[20] hymns of two verses.⌋

Arrangement of books i.-vii. with reference to verse-norms.—If we examine table 1 (p. cxliv), in which these books are set in the ascending numerical order of their verse-norms, several facts become clear. It is apparent, in the first place, that this division is made up of those seven books in which the number—normal or prevalent—of verses to a hymn runs from one to eight; secondly, that the saṁhitā itself begins with the norm of four; and, thirdly, that the number two as a norm is missing from the series. Fourthly, it is indeed apparent that every book shows departures from its norm; but also—what is more important in this connection—that these departures are all on one side, that of excess, and never on that of deficiency.⌋

⌊We may here digress to add that, if we compare table 1 with those following, it appears, fifthly, that in book vii. are put all the hymns of the three grand divisions that contain only 1 or 2 verses; sixthly, that neither in this division, nor yet in the other two, nor even in book xix., is there a hymn of 19 verses, nor yet one of 20.[21] From table 1, again, it appears, seventhly, that this division contains a hymn or hymns of every number of verses from 4 verses to 18 verses (mostly in books i.-v.) and from 1 verse to 3 verses (exclusively in books vi. and vii.).⌋

Excursus on hymn xix. 23, Homage to parts of the Atharva-Veda.—It is worth while at this point to recall to the reader's mind this remarkable hymn in its bearing upon some of the questions as to the structure of our text: see pages 931-4, and especially ¶6 of p. 931. As our saṁhitā begins with four-versed hymns, so does xix. 23 begin with homage "to them of four verses " (p. 931, line 29), and not with homage "to them of one verse." Again, grouping all hymns of four verses or more in this division according to length, there are 15 groups (not in the least conterminous with books) each containing a hymn or hymns of every number of verses from 4 to 18, and to these 15 groups the first 15 verses of xix. 23 correspond (p. 931, line 27). Again, of the fact that books i.-xviii. contain not one hymn of 19 verses nor yet one of 20, account seems to be taken in that the form of verses 16 and 17 differs from that of the 15 preceding (p. 931, line 37). Again, as in our series the norm two is lacking, so also is lacking in xix. 23 a dvyṛcebhyaḥ svāhā (but cf. p. 931, line 28, with p. 933, line 2). Finally the verses of homage "to them of three verses" and "to them of one verse" (xix. 23. 19-20) stand in the same order relative to each other and to the verses of homage to the 15 groups as do books vi. and vii. to each other and to the books containing the hymns of more than three verses, namely books i.-v.—Cf. further pages clvii and clix.⌋

We now return to the arrangement of the books within the division by norms.—The norms of books i.-vii. respectively, as the books stand in our text, are 4, 5, 6, 7, 8, 3, 1. From this point of view, the books fall into two groups: group X contains books i.-v., and its norms make a simple continuous ascending numerical scale beginning with four (4, 5, 6, 7, 8); group Y contains books vi. and vii., and its norms make a broken descending numerical scale beginning with three (3, 1). Here several questions arise as to group Y: first, why is its scale inverted, that is, why does not book vii. precede book vi? second, why does not group Y (and in the reversed order, vii., vi.) precede group X, so as to make the whole series begin, as is natural, with one instead of four, and run on in the text as it does in the table at p. cxliv? and, third, why is the scale broken, that is, why have not the diaskeuasts made eight books of the first division, including not only one for the one-versed hymns, but also another for the two-versed?⌋

⌊With reference to the last question, it is clear that the amount of material composing the two-versed hymns (30 hymns with only 60 verses: see p. cxlix, note) is much too small to make a book reasonably commensurate with the books of the first division; it is therefore natural that the original groupings of the text-makers should include no book with the norm of two.⌋

Exceptional character of book vii.—The first two questions, concerning group Y or books vi. and vii., are closely related, inasmuch as they both ask or involve the question why book vii. does not precede book vi. By way of partial and provisional answer to the second, it is natural to suggest that perhaps the scrappy character of the one-versed and two-versed hymns militated against beginning the Vedic text with book vii. And indeed this view is not without indirect support from Hindu tradition: for according to the Bṛhad-Devatā, viii. 99, the ritualists hold that a hymn, in order to be rated as a genuine hymn, must have at least three verses, tṛcādhamaṁ yājñikāḥ sūktam āhuḥ.[22] It may well be, therefore, that the diaskeuasts did not regard these bits of one or two verses as real hymns, as in fact they have excluded them rigorously from all the books i.-vi. From this point of view our groups X and Y have no significance except for the momentary convenience of the discussion, and the true grouping of books i.-vii. should be into the two groups, A, containing books i.-vi., and B, containing book vii.⌋

⌊The exceptional character of book vii. is borne out by several other considerations to which reference is made below. Its place in the saṁhitā is not that which we should expect, whether we judge by the fact that its norm is one verse or by the amount of its text (p. cxliii). If we consider the number of its hymns that are ignored by Kāuçika (see pp. 1011-2), again we find that it holds a very exceptional place in division I. Many of its hymns have a put-together look, as is stated at p. cliv; and this statement is confirmed by their treatment in the Pāippalāda recension (p. 1014, l. 15). Just as its hymns stand at the end of its grand division in the Vulgate, so they appear for the most part in the very last book of the Pāippalāda (cf. p. 1013, end). As compared with the great mass of books i.-vi., some of its hymns (vii. 73, for instance) are quite out of place among their fellows.⌋

Book vii. a book of after-gleanings supplementing books i.-vi.—It is very easy to imagine group A, or books i-vi., as constituting the original nucleus[23] of the saṁhitā (p. cxlviii, top), and group B, or book vii., as being an ancient supplement to that nucleus, just as book xix. is unquestionably a later supplement to the larger collection of the three grand divisions (cf. p. 895). This view does not imply that the verses of book vii. are one whit less ancient or less genuinely popular than those of books i.-vi., but merely that, as they appear in their collected form, they have the aspect of being after-gleanings, relatively to books i.-vi. This view accords well with the exceptional character of book vii. as otherwise established and as just set forth (p. cli).⌋

Arrangement of books with reference to amount of text.—If these considerations may be deemed a sufficient answer to the first two questions so far as they relate to book vii., there remains only that part of the second question which relates to book vi. One does not readily see why the saṁhitā might not have opened with book vi., the book of the varied and interesting three-versed hymns, so that the norms would have run in the order 3, 4, 5, 6, 7, 8 (1); and, since this is not the case, it may be that some other principle is to be sought as a co-determinant of the order of arrangement.⌋

⌊If we consult the table on p. cxliii, we see that, in division I., the scale of numbers of printed pages of text in each book (13, 16, 20, 27, 28, 40, 27) is a continuously ascending one for each book except the last (book vii.). The like is true if we base our comparison on the more precise scale of verse-totals for each book (153, 207, 230, 324, 376, 454, 286), as given at the foot of table 1, p. cxliv.⌋

⌊These facts, in the first place, strongly corroborate our view as to the exceptional character of book vii. By the principle of norms, it should stand at the beginning of the division; by the principle of amount (judged by verse-totals), it should stand between books iii. and iv. That it does neither is hard to explain save on the assumption of its posteriority as a collection. In the second place, these facts suggest at the same time the reason for the position of book vi. in the division, namely, that it is placed after books i.-v. because it is longer than any of those books.⌋

Résumé of conclusions as to the arrangement of books i.-vii.—Book vii., as a supplement of after-gleanings, is placed at the end of the grand division, without regard to amount of text or to verse-norm. Books i.-vi. are arranged primarily according to the amount of text,[24] in an ascending scale. For them the element of verse-norms, also in an ascending scale, appears as a secondary determinant. It conflicts with the primary determinant in only one case,[25] that of book vi., and is accordingly there subordinated to the primary one, so that book vi. (norm: 3) is placed after books i.-v. (norms: 4- 8).⌋

Departures from the norms by excess.—The cases of excess are most numerous in book v. (see p. 220), and concern over ${\displaystyle {\tfrac {14}{15}}}$ of all the hymns. On the other hand, the cases of conformity to the norm are most numerous in books vi. and i. and concern about ${\displaystyle {\tfrac {6}{7}}}$ of the hymns in each book. For books ii., iv., vii., and iii. respectively, the approximate vulgar fraction of cases of conformity is ⅗, ½, ½, and ⅖. For each of the seven books, in the order of closeness of conformity to the verse-norm, the more precise fraction is as follows: for book vi., it is .859; for i., it is .857; for ii., it is .61; for iv., it is .52; for vii., it is .47; for iii., it is .42; and for v., it is .06.⌋

Critical significance of those departures.—From the foregoing paragraph it appears that the order of books arranged by their degree of conformity (vi., i., ii.), agrees with their order as arranged by their verse-norms (3, 4, 5), for the books of shorter hymns. This is as it should be; for if the distinction of popular and hieratic hymns is to be made for this division, the briefest would doubtless fall into the prior class, the class less liable to expansion by secondary addition.⌋

We are not without important indications[26] that the hymns may have been more or less tampered with since their collection and arrangement, so as now to show a greater number of verses than originally belonged to them. Thus some hymns have been expanded by formulized variations of some of their verses; and others by the separation of a single verse into more than one, with the addition of a refrain. ⌊Yet others have suffered expansion by downright interpolations or by additions at the end; while some of abnormal length may represent the juxtaposition of two unrelated pieces.⌋

Illustrative examples of critical reduction to the norm.—⌊The instances that follow should be taken merely as illustrations. To discuss the cases systematically and thoroughly would require a careful study of every case of excess with reference to the structure of the hymn concerned and to its form and extent in the parallel texts,—in short, a special investigation.[27]

Thus in i. 3 (see p. 4), verses 2-5 are merely repetitions of verse 1 (and senseless repetitions, because only Parjanya, of the deities named, could with any propriety be called the father of the reed: cf. i. 2. 1); while verses 7 & 8 have nothing to do with the refrain and are to be combined into one verse: we have then four verses, the norm of the book.

Once more, in ii. 10 (see p. 51), no less evidently, the verse-couples 2 & 3, 4 & 5, and 6 & 7 are to be severally combined into three single verses, with omission of the refrain, which belongs only to verses 1 and 8: so that here we have five verses, again the normal number.

So, further, in iii. 31 (see p. 141), as it seems clear, 2 & 3, without the refrain, make verse 2; 4 with the refrain is verse 3; and 5 is a senseless intrusion; then, omitting all further repetitions of the refrain, 6 & 7 make verse 4; 8 & 9 make verse 5; and 10 & 11 make verse 6, six being here the verse-norm.

In book vi., a number of hymns which exceed the regular norm are formular and would bear reduction to hymns of three verses: instances are hymns 17, 34, 38, 107, 132. ⌊The cases are quite numerous in which the added verse is lacking in one of the parallel texts. Thus in book vi., hymns 16, 17, 34, 63, 83, 108, 121, and 128 (see the critical notes on those hymns and cf. p. 1014, l. 16) appear in the Pāippalāda text as hymns of three verses each.⌋

Besides these cases, there are not a few others where we may with much plausibility assume that the verses in excess are later appendixes or interpolations: such are i. 29. 4-5; ii. 3. 6; 32. 6; 33. 3 ab 4 cd, 6; iii. 15. 7-8; 21. 6, 8-10 (see note under vs. 7); 29. 7-8; iv. 2. 8; 16. 8-9; 17- 3; 39. 9-10; vi. 16. 4; 63. 4; 83. 4; 122. 3, 5; 123. 3-4. In book vii., moreover, the put-together character of many of the longer hymns is readily apparent (cf. hymns 17, 38, 50, 53, 76, 79, and 82 as they appear in the table on p. 1021).

But such analyses, even if pushed to an extreme, will not dispose of all the cases of an excess in the number of verses of a hymn above the norm of the book: thus iii. 16 corresponds to a Rig-Veda hymn of seven verses; iv. 30 and 33 each to one of eight; and v. 3 to one of nine. It will be necessary to allow that the general principle of arrangement ⌊by verse-norms⌋ was not adhered to absolutely without exception.

Arrangement of the hymns within any given book of this division.—In continuation of what was said in general on this topic at p. cxliii, we may add the following. The "first" hymn (pūrvam), "For the retention of sacred learning," is of so distinctly prefatory character as to stand of right at the very beginning of the text, or removed therefrom only by the prefixion of the auspicious çaṁ no devīr abhiṣṭaye (p. cxvi). It is noteworthy that books ii., iv., v., and vii. begin each with a "Mystic" hymn; that the five kindred hymns "Against enemies" are grouped together at ii. 19-23, as are the seven Mṛgāra-hymns at iv. 23-29. Hymns iii. 26-27 are grouped in place and by name, as digyukte; and so are the "two Brahman-cow" hymns, v. 18 and 19, and the vāiçvānarīya couple, vi. 35 and 36. The hymns "To fury" make a group in the AV. (iv. 31-32) as they do in the RV., from which they are taken.⌋

Second grand division (books viii.-xii.): long hymns of miscellaneous subjects.—As was said of the first division (p. cxlvii), there are other things besides length and subject which mark this division as a minor collection apart from the other two: the verse-norms do not serve here, as in division I., to help determine the arrangement, the norms assumed by the Pañcapaṭalikā (p. cxxxix) being for another purpose and of small significance; and the reader may be reminded of the fact (p. cxxxii) that the grouping of verses into decads runs through this grand division. It is a noteworthy fact, moreover, that the material of division II. appears distinctly to form a collection by itself in the Pāippalāda recension, being massed in books xvi. and xvii. The Vulgate books viii.-xi. are mostly in Pāipp. xvi. and the Vulgate book xii. is mostly in Pāipp. xvii. This is readily seen from the table on p. 1022.⌋

Their hieratic character: mingled prose passages.—More important differential features are the following. In the first place, if it be admitted that the first division is in very large measure of popular origin (p. cxlvii), the second, as contrasted therewith, is palpably of hieratic origin: witness the hymns that accompany, with tedious prolixity, the offering of a goat and five rice-dishes (ix. 5) or of a cow and a hundred rice-dishes (x. 9); the extollation of the virā́j (viii. 9), of the cow (x. 10), of the rice-dish and the prāṇá and the Vedic student (xi. 3-5) and the úcchiṣṭa (xi. 7); the hymn about the cow as belonging exclusively to the Brahmans (xii. 4); the prevalence of "mystic" hymns (cf. viii. 9; ix. 9-10; x. 7-8; xi. 8); the priestly riddles or brahmodyas (cf. x. 2, especially verses 20-25); and the taking over of long continuous passages from the Rig-Veda, as at ix. 9—10. In no less striking contrast with division I., in the second place, is the presence, in every book of division II., of an extensive passage of prose (viii. 10; ix. 6, 7; x. 5; xi. 3; xii. 5). This prose is in style and content much like that of the Brāhmaṇas, and is made up of what are called (save in the case of x. 5) 'periods' or paryāyas: see pages cxxxiii and 472. It is evident that we are here in a sphere of thought decidedly different from that of division I.⌋

Table of verse-totals for the hymns of division II.—The following table may be worth the space it takes, as giving perhaps a better idea of the make-up of the division than does the table on p. cxliv. Opposite each of the five prose paryāya-hymns is put a P, and opposite the hymn x. 5 (partly prose) is put a p. Disregarding the hymns thus marked, the verse-numbers are confined, for books viii.-xi., within the range of variation from 21 to 44, and from 53 to 63 for book xii.

 Hymn in viii. in ix. in x. in xi. in xii. 1 has 21 24 32 37 63 verses 2 28 25 33 31 55 3 26 31 25 56 P 60 4 25 24 26 26 53 5 22 38 50 P 26 73 P 6 26 62 P 35 23 7 28 26 P 44 27 8 24 22 44 34 9 26 22 27 26 10 33 P 28 34 27 ⌋

General make-up of the material of this division.—Whereas division I. contains a hymn or hymns of every number of verses from one to eighteen and none of more, division II. consists wholly of hymns of more than twenty verses, and contains all the hymns of that length occurring in books i.-xviii. except such as belong of right (that is, by virtue of their subject) to the third division.[28] The forty-five hymns of this division have been grouped into books with very evident reference to length and number, as shown by the table just given: the five longest have been put together to form the last or twelfth book, while each of the four preceding books contains an even quarter of the preceding forty or just ten hymns. Disregarding ix. 6 and xi. 3 (paryāya-hymns), books viii.-xi. contain all the hymns of from 21-50 verses to be found in the first two grand divisions, and book xii. contains all of more than 50 in the same divisions. Anything more definite than this can hardly be said respecting the arrangement of the several books within the second division. From the tables it appears that no such reference to the length of the hymns has been had in division II. as was had in division I. None of the books viii.-xii. is without one of the longer, formular, and mainly non-metrical pieces (marked with P or p in the table above); and this fact may point to an inclination on the part of the text-makers to scatter those prose portions as much as possible among the poetical ones.

Order of books within the division: negative or insignificant conclusions.—If we consider, first, the amount of text in pages[29] for each book, namely 22, 21, 27, 25, 22, the series appears to have no connection with the order of the books; on the contrary, the books are, on the whole, remarkable for their approximate equality of length. The case is similar, secondly, with the hymn-totals of the Bombay edition, 15, 15, 10, 12, and 11. Thirdly, the verse-totals for each of the five books, according to the numeration of the Berlin edition, are 259, 302, 350, 313, and 304 (see above, p. cxliv), a sequence in which we can trace no orderly progression. On the other hand, fourthly, if we take the verse-totals of the Bombay edition, to wit, 293, 313, 350, 367, and 304,[30] we see that the first four books, viii.-xi., are indeed arranged, like books i.-vi. (p. clii), on a continuously ascending arithmetical scale. Furthermore and fifthly, if, for the verse-totals of each of the five books, we make the (very easy) substitution of the average verse-totals of the hymns of each book, we obtain again a series, to wit, 29.3, 31.3, 35.0, 36.7, and 60.8, which progresses constantly in one direction, namely upward, and through all the five books.⌋

Arrangement of the hymns within any given book of this division.—⌊From the table on p. clvi it would appear that the individual hymns are not disposed within the book with any reference to length. It may, however, be by design rather than accident that the only hymn with the smallest number of verses in this division is put at the beginning, and that the longest is put last.⌋ The arrangement in this division, like that in the first, shows no signs of a systematic reference to the subjects treated of, although (as in division I.: p. civ, top), in more than one instance, two hymns of kindred character are placed together: thus viii. 1 & 2; 3 & 4; 9 & 10; ix. 4 & 5; 9 & 10; x. 7 & 8; 9 & 10; xi. 9 & 10; xii. 4 & 5.

Possible reference to this division in hymn xix. 23.—Such reference, I suspect, must be sought in verse 18, if anywhere, and in the two words mahat-kāṇḍā́ya svā́hā, 'to the division of great [hymns], hail!' See p. 931 ¶7, and the note to vs. 18.⌋

⌊Postscript.—Such was my view when writing the introduction to xix. 23. Even then, however, I stated (p. 932, line 12) that verses 21 and 22 were not accounted for. Meantime, a new observation bears upon vs. 21.⌋

⌊Immediately after the passage referred to at p. cxlviii, foot-note, the Major Anukr., at the beginning of its treatment of book viii., proceeds: 'Now are set forth the seers and divinities and meters of the mantras of the sense-hymns of the kṣudra-kāṇḍas (? or -kāṇḍa?). To the end of the eleventh kāṇḍa, the sense-hymn is the norm.' etc. atha kṣudra-kāṇdā- ’rthasūkta-mantrāṇām ṛṣi-dāivata-chandāṅsy ucyante. tato yāvad ekādaçakāṇḍā-’ntam arthasūkta-prakṛtis tāvad vihāya paryāyān virāḍ vā (viii. 10) prabhṛtīn iti etc. What pertinence the word kṣudra may have as applied to books viii.-xi. I cannot divine; but it can hardly be an accident that the very same word is used in the phrase of homage to parts of the AV. at xix. 22. 6 and 23. 21, kṣudrébhyaḥ svā́hā, and that this phrase is followed in h. 22 and in the comm's text of h. 23, by the words paryāyikébhyaḥ svā́hā, that is, by an allusion to the paryāyas, just as in the text of the Anukr. Apart from vss. 16-18 of xix. 23, vss. 1-20 refer most clearly to the first grand division; and vss. 23-28 refer just as clearly to the third. The allusion to the second ought therefore certainly to come in between vs. 20 and vs. 23, that is it ought to be found in vss. 21 and 22. We have just given reason for supposing that vs. 21 contains the expected allusion. The meaning of ekānṛcébhyaḥ of vs. 22 is as obscure as is the pertinence of kṣudrébhyaḥ; probably ekānṛcébhyaḥ is a corrupt reading. If I am right as to vs. 21, the mystery of vs. 18 becomes only deeper.⌋

Third grand division (books xiii.-xviii.): books characterized by unity of subject.—The remaining six books constitute each a whole by itself and appear to have been on that account kept undivided by the arrangers and placed in a body together at the end of the collection. The books in which the unity of subject is most clearly apparent are xiv. (the wedding verses), xviii. (the funeral verses), and xv. (extollation of the Vrātya). ⌊The unity of books xiii. and xvii., although less striking, is yet sufficiently evident, the one consisting of hymns to the Sun as The Ruddy One or Rohita, and the other being a prayer to the Sun as identified with Indra and with Vishṇu. In book xvi., the unity of subject is not apparent,[31] although it seems to consist in large measure (see p. 792) of "Prayers[32] against the terror by night."⌋ Book xvi. is not so long that we might not have thought it possible that it should be included as a paryāya-sūkta in one of the books of the second division; and book xvii., too, is so brief that it might well enough have been a hymn in a book.

⌊Hindu tradition assigns at least four of the books of this division each to a single seer; the whole matter is more fully set forth at p. 1038. However much or little value we may attach to these ascriptions of quasi-authorship, they are certainly of some significance as a clear mark of differentiation between this division and the other two.⌋

Division III. represented in Pāippalāda by a single book, book xviii.—An item of evidence important in its relation to the Vulgate division III. as a separate unity is afforded by the treatment of that division in the Kashmirian recension: the Vulgate books xiii.-xviii., namely, are all grouped by the makers of the Pāippalāda text into a single book, book xviii., and appear there either in extenso or else by representative citations. The relations of the Vulgate division to the Pāipp. book are set forth in detail at p. 1014, which see.⌋

Names of these books as given by hymn xix. 23.—It is a most significant fact, and one entirely in harmony with the classification of books xiii.-xviii. on the basis of unity of subject, that they should be mentioned in hymn xix. 23 by what appear to be their recognized names. It is therefore here proper to rehearse those names as given in verses 23-28 of the hymn (see pages 931, ¶5, and 933). They are: for book xiii., 'the ruddy ones,' róhitebhyas, plural; for xiv., 'the two Sūryās,' sūryā́bhyām, or the two [anuvākas] of the book beginning with the hymn of Sūryā's wedding; for xv., 'the two [anuvākas] about the vrā́tya,' vrātyā́bhyām (accent!); for xvi., 'the two [anuvākas] of Prajāpati,' prājāpatyā́bhyām; for xvii., 'the Viṣāsahi,' singular; and for xviii., 'the auspicious ones,' man̄galikébhyas, euphemism for the inauspicious funeral verses.⌋

Order of books within the division.—The verse-totals for the books are, by the Berlin numeration, 188, 139, 141, 93, 30, and 283, and, by the Bombay numeration, 188, 139, 220, 103, 30, and 283 (above, p. cxxxvii). But for the disturbing influence of the numerous brief paryāya-verses of book xv. upon the third member of these series, they would both coincide in their general aspect with the series based on the amount of text in pages of the Berlin edition, namely, 13, 12, 10, 5, 3, and 21 (as given above, p. cxliii). From the last series, it appears that these books, except the last, are arranged in a descending scale of length, therein differing from divisions I. and II. in which the scale was an ascending one. In all three divisions, the final book is an exceptional one: in the first, it is a scanty lot of after-gleanings; in the second, it contains the five longest hymns, each about twice as long as the average of the four books preceding; and in the third, again, it contains very long hymns, which are, moreover, an extensive and peculiar collection of verses, unified indeed (like those of book xiv.) in large measure by the ritual uses to which they are put, but on the whole quite different in origin and character from most of the rest (see the introductions to the hymns of book xviii.).⌋

Table of verse-totals for the hymns of division III.—The following table is made like that on p. clvi, and may give a better idea of the make-up of the division than does the one on p. cxlv. That seems to me wrong, because it follows the Berlin edition in treating the 18 individual paryāyas of book xv. and the 9 of book xvi. each as one hymn (see p. cxxxvi), and in having to recognize accordingly hymns of 3 verses, of 4 and 5 and so on, in this division. We certainly must recognize some larger unity than the paryāya in books xv. and xvi.; and, whether that unity be the book or the anuvāka, in either case we avoid the necessity of recognizing any hymns with a verse-total of less than 20 in this division (see table 3, second form, p. cxlv). Assuming that xv. and xvi. make each two hymns, the table is as follows:

 Hymn in xiii. in xiv. in xv. in xvi. in xvii. in xviii. 1 has 60 64 50 P 32 P 30 61 verses 2 46 75 91 P 61 P 60 3 26 73 4 56 P 89

The scale of hymn-totals for each book is thus 4, 2, 2, 2, 1, and 4; and it then appears that all the books of the division except the last are arranged on a descending scale, the three books of two hymns each being arranged among themselves on a descending scale of amount of text.⌋

Order of hymns within any given book of this division.—As to this, questions can hardly be raised; or, if raised, they resolve themselves into questions in general concerning the hymn-divisions of books xiii.-xviii. and their value.⌋

The hymn-divisions of books xiii.-xviii. and their value.—In these books the whole matter of hymn-division seems to be secondary and of little critical value or significance (cf. p. cxxxi).—First, as to the metrical books (xiv., xviii., xiii., xvii.: that is, all but the two paryāya-books xv. and xvi.). In them, the hymn-division is, as in book xii. of division II., coincident with the anuvāka-division. Book xiv. is divided into two hymns by both editions, not without the support of the mss.; but the Major Anukr. seems rather to indicate that the book should not be divided (for details, see pages 738-9): the hymn-division is here at any rate questionable. Book xviii., properly speaking, is not a book of hymns at all, but rather a book of verses. The Pañcapaṭalikā says that these verses are 'disposed' (vihitās) in four anuvākas (see p. 814, ¶5, and note the word paraḥ, masculine): from which we may infer that the anuvāka-division is of considerable antiquity; but the significance of the coincident hymn-division is minimized by the facts that a ritual sequence runs over the division-line between hymns 1 and 2 (see p. 814, ¶6, and p. 827, ¶2) and that the division between hymns 3 and 4 ought to come just before 3. 73 (and not just after: see p. 848, ¶8). Even with book xiii. the case is essentially not very different: see the discussions in Deussen's Geschichte, i. 1. 215-230. Book xvii. consists of a single anuvāka (it is the only book of which this is true: p. 805); and although in the colophons the mss. apply both designations, anuvāka and artha-sūkta, to its 30 verses (which the mss. divide into decads), it is truly only one hymn.⌋

⌊The paryāya-books, books xv. and xvi. remain. These, as appears from the tables on pages 771 and 793, consist each of two anuvākas with 7 and 11 and with 4 and 5 paryāyas respectively. When writing the introductions to those books, I had not seriously considered the proper grouping of the paryāyas (cf. p. 770, lines 29-30). The discussion at p. cxxx, above, seems now to make it probable that the paryāya-groups should be assumed, as everywhere else from book xii.-xviii., to be conterminous with the anuvākas. The bearing of this assumption on the method of citation is treated at p. cxxxvi, above. The effect of this assumption upon the summations is shown in table 3, second form, p. cxlv, and in the table on p. clx.⌋

1. ⌊M. is the ms. listed by Aufrecht, in his Catalogue of the Bodleian Sanskrit Manuscripts, p. 392 b, as No. 80 of the Codices Milliani.⌋
2. ⌊The printer's copy of this paragraph in Whitney's handwriting says clearly "second volume"; but the original description of the mss. (made by him probably in 1853) says clearly "first volume": I feel sure that the original is right and have altered the proof to correspond therewith.⌋
3. ⌊Listed by Aufrecht, p. 385 b, as Nos. 499 and 500 of the Codices Wilsoniani.⌋
4. ⌊While reading proof, I see that Weber had made the same observation in 1862, Ind. Stud. v. 78. Moreover, the fact that çaṁ no figures as opening stanza of AV. in the GB. at i. 29 is now used (1904) by Caland, WZKM. xviii. 193, to support his view that the GB. attaches itself to the Pāipp. recension.⌋
5. ⌊The date quoted at top of page 1 is not quite correct. Whitney spent from March 19 to May 10 (1853) at Paris, May 12 to June 1 at Oxford, and June 1 to July 22 at London.⌋
6. ⌊On this topic, Whitney left only rough notes, a dozen lines or so: cf. p. xxix.⌋
7. ⌊For this chapter, pages cxxiii to cxxvi, the draft left by Whitney was too meagre and unfinished to be printed. I have rewritten and elaborated it, using freely his own statements and language as given in his notes to the Prātīçākyas.
8. Cf. p. 832, ¶ 4, below.
9. Nearly all the mss. and SPP. violate it at xi. 1. 22.
10. ⌊This part of the statement is subject, for books xiii.-xviii., to the modification implied in the preceding paragraph.⌋
11. As to what this grouping should be, see the discussion at p. cxxx, near end.
12. ⌊I beg the reader to compare my remarks on the Method of Citation in the preface to the Karpūramañjarī, pages xv-xvi. For citations of the Māhārāṣṭrī or verse passages, the exponent is a letter; for Çāurasenī or prose, it is a figure.⌋
13. ⌊Another and wholly different matter is the norm assumed for the verse-totals of the individual hymns of each book (see p. cxlviii): thus book i. is the book of four-versed hymns.⌋
14. The mss. read: ’ntyānu-, with double sandhi; -saç for -çaç, with confusion of sibilants; saṁkhyā (but one has indeed -yām); and adhikānim-, with omission of a needed twin consonant (cf. p. 832). As to the use of kṛ, cf. below, p. 52 end, and p. 186, ¶3.
15. ⌊There are five verses which, although occurring in our xix., are yet cited by Kāuç. in full, as if they did not belong to the Atharvan text recognized by Kāuç. Moreover, there are cited by Kāuç. six pratīkas which, although answering to six hymns (between 51 and 68) of our xix., may yet for the most part be regarded as kalpajā mantrās. For a detailed discussion of the matter, see pages 896-7.⌋
16. ⌊This statement is true without modification, if we treat books xv. and xvi. each as two hymns or paryāya-groups in the manner explained and reasoned at p. cxxx, and implied in the second form of table 3, p. cxlv: cf. p. cxxxvii, line 13.⌋
17. ⌊And so would hymns 22 and 23, if judged by their actual length.⌋
18. ⌊That books i.-vii. are distinctly recognized as a separate unity by the Major Anukr. appears also from the fact that for the right or wrong study of its first five paṭalas (in which books i.-vii. are treated), special blessings or curses are promised in a passage at the beginning of the sixth. The fact was noted by Weber, Verzeichniss, vol. 2, p. 79; and the passage was printed by him on p. 81.⌋
19. ⌊At i. 1, and also at v. 9 and 10 (these two are prose pieces), the treatise states the number when it is normal. This is not unnatural at i. 1, the beginning; and considering the prevailing departure from the norm in book v., it is not surprising there. On the other hand, the omissions at iv. 36 and vi. 121 are probably by inadvertence.⌋
20. ⌊This is the true number. The number 26, given at p. cxliv in table 1, rests on the actual hymn-divisions of the Berlin text. On account of the discordance, the 30 hymns may here be named: 1, 6. 1-2, 6. 3-4, 13, 18, 22, 25, 29, 40-42, 47-49, 52, 54. 2 with 55. 1, 57-58, 61, 64, 68. 1-2, 72. 1-2, 75, 76. 5-6, 78, 108, 112-114, 116. (They are very conveniently shown in the table, p. 1021.) Note on the other hand the silence of the Anukr. as to our 45, 54. 1, 68. 3, and 72. 3. Its silence means that our 45. 1 (seer, Praskaṇva) and 45. 2 (Atharvan) and 54. 1 (Brahman) form three one-versed hymns, a fact which is borne out by the ascriptions of quasi-authorship; and that 68. 3 and 72. 3 form two more.⌋
21. ⌊in the Kuntāpakhila there are two hymns of 20.⌋
22. ⌊For the productions of modern hymnology, one hardly errs in regarding three verses as the standard minimum length, a length convenient for use, whether in reading or singing, and for remembering. A two-versed hymn is too short for a dignified unity. Possibly similar considerations may have had validity with the ancient textmakers.⌋
23. ⌊If asked to discriminate between the books of that nucleus, I should put books vi. and i. and ii. first (cf. p. cliii, ¶3); at all events, book v. stands in marked contrast with those three.⌋
24. ⌊Whether this amount is judged by verse-totals or by pages, the order is the same.⌋
25. ⌊That the two orders, based on the one and the other determinant, should agree throughout books i.-v. is no doubt partly fortuitous; but it is not very strange. The variation in the number of hymns for each book (35, 36, 31, 40, 31) is confined to narrow limits; and if, as is probable, the departures from the norm were originally fewer and smaller than now, the verse-totals for each book would come nearer to being precise multiples of those ascending norms
26. ⌊Cf. p. 281, ¶2⌋
27. ⌊A very great part of the data necessary for the conduct of such an inquiry may be found already conveniently assembled in this work in Whitney's critical notes; for, although scattered through these notes, they may yet be said to be "assembled" in one work, and more "conveniently" than ever before. The investigation is likely to yield results of interest and value.⌋
28. ⌊See the tables, pages cxliv-cxlv.—Book xix. contains two hymns, mostly prose, of which the subdivisions number 21 and 30 (of. p. cxlvii); and among the Kuntāpa-hymns are three of 20 or more verses.⌋
29. ⌊As printed in the Berlin edition (see above, p. cxiiii). From a nāgarī ms. written in a hand of uniform size, I might obtain different and interpretable data.⌋
30. ⌊This series differs from the Berlin sequence by a plus of 34 and 11 and 54 in the first and second and fourth members respectively: see p. cxxxvii, and cf. pages 516, 546, 632.⌋
31. ⌊In one of the old drafts of a part of his introductory matter, Whitney says: Until we understand the character of the ceremonies in connection with which book xvi. was used, it may not be easy to discover a particular concinnity in it. With reference to that remark, I have said, at p. 792: The study of the ritual applications of the book distinctly fails, in my opinion, to reveal any pervading concinnity of purpose or of use.⌋
32. ⌊Perhaps, using a Pāli term, we may designate book xvi. as a Paritta.⌋