Vedic Hymns, Part I (SBE32): Preface to the First Edition: Part 11

---

Let us now turn back for one moment to look at the slaughter which has been committed! Is there one single rule of prosody that has been spared? Is there one single short syllable that must always remain short, or a long syllable that must always remain long? If all restrictions of prosody are thus removed, our metres, no doubt, become perfectly regular. But it should be remembered that these metrical rules, for which all this carnage has been committed, are not founded upon any a priori principles, but deduced by ancient or modern metricians from those very hymns which seem so constantly to violate

them. Neither ancient nor modern metricians had, as far as we know, any evidence to go upon besides the hymns of the Rig-veda; and the philosophical speculations as to the origin of metres in which some of them indulge, and from which they would fain derive some of their unbending rules, are, as need hardly be said, of no consequence whatever. I cannot understand what definite idea even modern writers connect with such statements as that, for instance, the Trishtubh metre sprang from the Gagatî metre, that the eleven syllables of the former are an abbreviation of the twelve syllables of the latter. Surely, metres are not made artificially, and by addition or subtraction. Metres have a natural origin in the rhythmic sentiment of different people, and they become artificial and arithmetical in the same way as language with its innate principles of law and analogy becomes in course of time grammatical and artificial. To derive one metre from another is like deriving a genitive from a nominative, which we may do indeed for grammatical purposes, but which no one would venture to do who is at all acquainted with the natural and independent production of grammatical forms. Were we to arrange the Trishtubh and Gagatî metres in chronological order, I should decidedly place the Trishtubh first, for we see, as it were before our eyes, how sometimes one foot, sometimes two and three feet in a Trishtubh verse admit an additional syllable at the end, particularly in set phrases which would not submit to a Trishtubh ending. The phrase sam no bhava dvipade sam katushpade is evidently a solemn phrase, and we see it brought in without hesitation, even though every other line of the same strophe or hymn is Trishtubh, i. e. hendecasyllabic, not dodecasyllabic. See, for instance, VI, 74, 1; VII, 54, 1; X, 85, 44; 165, 1. However, I maintain by no means that this was the actual origin of Gagatî metres; I only refer to it in order to show the groundlessness of metrical theories which represent the component elements, a foot of one or two or four syllables as given first, and as afterwards compounded into systems of two, three or four such feet, and who therefore would wish us to look upon the hendecasyllabic Trishtubh as originally a dodecasyllabic Gagatî, only

deprived of its tail. If my explanation of the name of Trishtubh, i. e. Three-step, is right, its origin must be ascribed to a far more natural process than that of artificial amputation. It was to accompany a choros, i. e. a dance, which after advancing freely for eight steps in one direction, turned back (vritta) with three steps, the second of which was strongly marked, and would therefore, whether in song or recitation, be naturally accompanied by a long syllable. It certainly is so in the vast majority of Trishtubhs which have been handed down to us. But if among these verses we find a small number in which this simple and palpable rhythm is violated, and which nevertheless were preserved from the first in that imperfect form, although the temptation to set them right must have been as great to the ancient as it has proved to be to the modern students of the Veda, are we to say that nearly all, if not all, the rules that determine the length and shortness of syllables, and which alone give character to every verse, are to be suspended? Or, ought we not rather to consider, whether the ancient choregic poets may not have indulged occasionally in an irregular movement? We see that this was so with regard to Gâyatrî verses. We see the greater freedom of the first and second pâdas occasionally extend to the third; and it will be impossible, without intolerable violence, to remove all the varieties of the last pâda of a Gâyatrî of which I have given examples above, pages civ seqq.

It is, of course, impossible to give here all the evidence that might be brought forward in support of similar freedom in Trishtubh verses, and I admit that the number of real varieties with them is smaller than with the Gâyatrîs. In order to make the evidence which I have to bring forward in support of these varieties as unassailable as possible, I have excluded nearly every pâda that occurs only in the first, second, or third line of a strophe, and have restricted myself, with few exceptions, and those chiefly referring to pâdas that had been quoted by other scholars in support of their own theories, to the final pâdas of Trishtubh verses. Yet even with this limited evidence, I think I shall be able to establish at least three

varieties of Trishtubh. Preserving the same classification which I adopted before for the Gâyatrîs, so as to include the important eighth syllable of the Trishtubh, which does not properly belong to the vritta, I maintain that class 4.  ̆ ̆ ̄ ̄, class 5.  ̄ ̄ ̄ ̄, and class 8.  ̄ ̆ ̆ ̄ must be recognised as legitimate endings in the hymns of the Veda, and that by recognising them we are relieved from nearly all, if not all, the more violent prosodial licences which Professor Kuhn felt himself obliged to admit in his theory of Vedic metres.

§4.  ̆ ̆ ̄ ̄.

The verses which fall under § 4 are so numerous that after those of the first Mandala, mentioned above, they need not be given here in full. They are simply cases where the eighth syllable is not lengthened, and they cannot be supposed to run counter to any rule of the Prâtisâkhya, for the simple reason that the Prâtisâkhya never gave such a rule as that the eighth syllable must be lengthened, if the ninth is short. Examples will be found in the final pâda of Trishtubhs: II, 30, 6; III, 36, 4; 53, 15; 54, 12; IV, 1, 16; 2, 7; 9; 11; 4, 12; 6, 1; 2; 4; 7, 7; 11, 5; 17, 3; 23, 6; 24, 2; 27, 1; 28, 5; 55, 5; 57, 2; V, 1, 2; VI, 17, 10; 21, 8; 23, 7; 25, 5; 29, 6; 33, 1; 62, 1; 63, 7; VII, 21, 5; 28, 3; 42, 4; 56, 15; 60, 10; 84, 2; 92, 4; VIII, 1, 33; 96, 9; IX, 92, 5; X, 61, 12; 13; 74, 3; 117, 7.

In support of § 5.  ̄ ̄ ̄ ̄, the number of cases is smaller, but it should be remembered that it might be considerably increased if I had not restricted myself to the final pâda of each Trishtubh, while the first, second, and third pâdas would have yielded a much larger harvest:

§5.  ̄ ̄ ̄ ̄.

I, 89, 9. mâ no madhyâ rîrishatâ̄yūr gāntōh.

I, 92, 6. supratîkâ saumanasâ̄yâ̄gî̄gāh.

I, 114, 5; 117, 2; 122, 1; 122, 8; 186, 3; II, 4, 2; III, 49, 2; IV, 3, 9; 26, 6; V, 41, 14; VI, 25, 2; 66, 11; VII, 8, 6; 28, 4; 68, 1; 71, 2; 78, 1; 93, 7^b; IX, 90, 4; X, 11, 8.

I do not wish to deny that in several of these lines it would be possible to remove the long syllable from the ninth place by conjectural emendation. Instead of â´yur in I, 89, 9, we might read â´yu; in I, 92, 6, we might drop the augment of agîgar; in II, 4, 2, we might admit synizesis in ă͡rătir, and then read gî̄ră-āsvāh, as in I, 141, 12. In VI, 25, 2, after eliding the a of ava, we might read d̆â̄sî̄h. But even if, in addition to all this, we were to admit the possible suppression of final m in asmabhyam, mahyam, and in the accusative singular, or the suppression of s in the nominative singular, both of which would be extreme measures, we should still have a number of cases which could not be righted without even more violent remedies. Why then should we not rather admit the occasional appearance of a metrical variation which certainly has a powerful precedent in the dispondeus of Gâyatrîs? I am not now acquainted with the last results of metrical criticism in Virgil, but, unless some new theories now prevail, I well recollect that spondaic hexameters, though small in number, much smaller than in the Veda, were recognised by the best scholars, and no emendations attempted to remove them. If then in Virgil we read,

why not follow the authority of the best MSS. and the tradition of the Prâtisâkhyas and admit a dispondeus at the end of a Trishtubh rather than suspend, in order to meet this single difficulty, some of the most fundamental rules of prosody?

I now proceed to give a more numerous list of Traishtubha pâdas ending in a choriambus,  ̄ ̆ ̆ ̄, again confining myself, with few exceptions, to final pâdas:

§ 8.  ̄ ̆ ̆ ̄

I, 62, 3. sam usriyâbhir vâvasāntă nărāh.

I, 103, 4. yad dha sûnuh sravase nâ̄mă dădhē.

I, 121, 9; 122, 10^b; 173, 8; 186, 2; II, 4, 3; 19, 1; 33, 14; IV, 1, 19^c a; 25, 4; 39, 2; V, 30, 12; 41, 4; 41, 15;

§ 2.  ̆ ̆ ̆ ̄.

Trishtubh verses, the final pâda of which ends in  ̆ ̆ ̆ ̄, I should generally prefer to treat as ending in a Gâgata pâda, in which this ending is more legitimate. Thus I should propose to scan:

I, 122, 11. prăsāstăyē măhĭn̆â̄ ̍ răthăvătē.

III, 20, 5. văsû̄n rūdrâ̄n̐ â̄dīty̆â̄n̐ ̍ ĭhă hŭvē.

V, 2, 1. pŭrāh pāsyāntĭ nĭhĭtām ̍ (tăm) ărătaū.

VI, 13, 5. văyō vrĭkâ̄y̆â̆răyē ̍ găsŭrăyē.

§ 1.  ̆ ̄ ̆ ̄.

I should propose the same medela for some final pâdas of Trishtubhs apparently ending in  ̆ ̄ ̆ ̄. We might indeed, as has been suggested, treat these verses as single instances of that peculiar metre which we saw carried out in the whole of hymn X, 77, but at the end of a verse the admission

of an occasional Gâgata pâda is more in accordance with the habit of the Vedic poets. Thus I should scan:

V, 33, 4. vrĭshâ̄ sămātsŭ d̄â̆sās̍yă nâ̄mă kīt a.

V, 41, 5^b. râ̄yă ĕs̄hē§6.  ̆ ̄ ̄ ̄.

Final pâdas of Trishtubhs ending in  ̆ ̄ ̄ ̄ are very scarce. In VI, 1, 4,

it would be very easy to read bhadrâyâm te samdrishtau rănăyāntā; and in X, 74, 2,

we may either recognise a Gâgata pâda, or read

which would agree with the metre of hymn X, 77.

§ 7.  ̄ ̄ ̆ ̄

Pâdas ending in  ̄ ̄ ̆ ̄ do not occur as final in any Traishtubha hymn, but as many Gâgata pâdas occur in the body of Traishtubha hymns, we have to scan them as dodecasyllabic:

I, 63, 4^a. tv̆ām hā tyăd īndr̆ă kōd̆î̄h săkhâ̄.

IV, 26, 6^b. părâ̄vătāh săkŭnō māndr̆ām mădām.

The adjective pâvaka which frequently occurs at the end of final and internal pâdas of Trishtubh hymns has always

to be scanned p̆â̄văkā. Cf. IV, 51, 2; VI, 5, 2; 10, 4; 51, 3; VII, 3, 1; 9; 9, 1^b; 56, 12; X, 46, 7^b.

I must reserve what I have to say about other metres of the Veda for another opportunity, but I cannot leave this subject without referring once more to a metrical licence which has been strongly advocated by Professor Kuhn and others, and by the admission of which there is no doubt that many difficulties might be removed, I mean the occasional omission of a final m and s, and the subsequent contraction of the final and initial vowels. The arguments that have been brought forward in support of this are very powerful. There is the general argument that final s and m are liable to be dropt in other Aryan languages, and particularly for metrical purposes. There is the stronger argument that in some cases final s and m in Sanskrit may or may not be omitted, even apart from any metrical stress. In Sanskrit we find that the demonstrative pronoun sas appears most frequently as sa (sa dadâti), and if followed by liquid vowels, it may coalesce with them even in later Sanskrit. Thus we see saisha for sa esha, sendrah for sa indrah sanctioned for metrical purposes even by Pânini, VI, 1, 134. We might refer also to feminines which have s in the nominative singular after bases in û, but drop it after bases in î. We find in the Samhitâ text, V, 7, 8, svádhitîva, instead of svádhitih-iva in the Pada text, sanctioned by the Prâtisâkhya 259; likewise IX, 61, 10, Samhitâ, bhû´my â´ dade, instead of Pada, bhû´mih â´ dade. But before we draw any general conclusions from such instances, we should consider whether they do not admit of a grammatical instead of a metrical explanation. The nominative singular of the demonstrative pronoun was sa before it was sas; by the side of bhû´mih we have a secondary form bhû´mî; and we may conclude from svádhitî-vân, I, 88, 2, that the Vedic poets knew of a form svádhitî, by the side of svádhitih.

As to the suppression of final m, however, we see it admitted by the best authorities, or we see at least alternate forms with or without m, in túbhya, which occurs

frequently instead of túbhyam a, and twice, at least, without apparently any metrical reason b. We find asmâ´ka instead of asmâ´kam (I, 173, 10), yushmâ´ka instead of yushmâ´kam (VII, 59, 9-10), yágadhva instead of yágadhvam (VIII, 2, 37) sanctioned both by the Samhitâ and Pada texts c.

If then we have such precedents, it may well be asked why we should hesitate to adopt the same expedient, the omission of final m and s, whenever the Vedic metres seem to require it. Professor Bollensen's remark, that Vedic verses cannot be treated to all the licences of Latin scanning d, is hardly a sufficient answer; and he himself, though under a slightly different form, would admit as much, if not more, than has been admitted on this point by Professors Kuhn and Roth. On a priori grounds I should by no means feel opposed to the admission of a possible elision of final s or m, or even n; and my only doubt is whether it is really necessary for the proper scanning of Vedic metres.

My own opinion has always been, that if we admit on a larger scale what in single words can hardly be doubted by anybody, viz. the pronunciation of two syllables as one, we need not fall back on the elision of final consonants in order to arrive at a proper scanning of Vedic metres. On this point I shall have to say a few words in conclusion, because I shall frequently avail myself of this licence, for the purpose of righting apparently corrupt verses in the hymns of the Rig-veda; and I feel bound to explain, once for all, why I avail myself of it in preference to other emendations which have been proposed by scholars such as Professors Benfey, Kuhn, Roth, Bollensen, and others.

The merit of having first pointed out-some cases where

two syllables must be treated as one, belongs, I believe, to Professor Bollensen in his article, 'Zur Herstellung des Veda,' published in Benfey's Orient and Occident, vol. ii, p. 461. He proposed, for instance, to write hyânâ´ instead of hiyânâ´, IX, 13, 6; dhyânó instead of dhiyânó, VIII, 49, 5; sáhyase instead of sáhîyase, I. 71, 4; yânó instead of iyânó, VIII, 50, 5, &c. The actual alteration of these words seems to me unnecessary; nor should we think of resorting to such violent measures in Greek where, as far as metrical purposes are concerned, two vowels have not unfrequently to be treated as one.

That iva counts in many passages as one syllable is admitted by everybody. The only point on which I differ is that I do not see why iva, when monosyllabic, should be changed to va, instead of being pronounced quickly, or, to adopt the terminology of Greek grammarians, by synizesis a. Synizesis is well explained by Greek scholars as a quick pronunciation of two vowels so that neither should be lost, and as different thereby from synalœphe, which means the contraction of two vowels into one b. This synizesis is by no means restricted to iva and a few other words, but seems to me a very frequent expedient resorted to by the ancient Rishis.

Originally it may have arisen from the fact that language allows in many cases alternate forms of one or two syllables. As in Greek we have double forms like ἀλεγεινός and ἀλγεινός, γαλακτοφάγος and γλακτοφάγος, πετηνός and πτηνός, πυκινός and πυκνός c, and as in Latin we have the shortening

or suppression of vowels carried out on the largest scale a. we find in Sanskrit, too, such double forms as prithvî or prithivî, adhi and dhi, api and pi, ava and va. The occurrence of such forms which have nothing to do with metrical considerations, but are perfectly legitimate from a grammatical point of view, would encourage a tendency to treat two syllables—and particularly two short syllables—as one, whenever an occasion arose. There are, besides, in the Vedic Sanskrit a number of forms where, as we saw, a long syllable has to be pronounced as two. In some of these cases this pronunciation is legitimate, i. e. it preserves an original dissyllabic form which in course of time had become monosyllabic. In other cases the same process takes place through a mistaken sense of analogy, where we cannot prove that an original dissyllabic form had any existence even in a prehistoric state of language. The occurrence of a number of such alternate forms would naturally leave a general impression in the minds of poets that two short syllables and one long syllable were under certain circumstances interchangeable. So considerable a number of words in which a long syllable has to be pronounced as two syllables has been collected by Professors Kuhn, Bollensen, and others, that no doubt can remain on this subject. Vedic poets, being allowed to change a semivowel into a vowel, were free to say nâ̄sātyâ̄ and nâ̄sā̆ty̆â̄, VIII, 5, 32; prĭthīvyâ̄s and prĭthī̆vy̆â̄h; pītrōh and pī̆tr̆ōh, I, 31, 4. They could separate compound words, and pronounce ghri̅tâ̄nnāh or ghrĭtă-ānnāh, VII, 3, 1. They could insert a kind of shewa or svarabhakti in words like sâ̄mnē, or sâ̄mn̆ē, VIII, 6, 47; dhâ̄mnē or dhâ̄mn̆ē, VIII, 92, 25; ărâ̄vnāh and ărâ̄vn̆āh, IX, 63, 5. They might vary between pâ̄ntĭ and p̆â̄ntĭ, I, 41, 2; yâ̄thănă and y̆â̄thănă, I, 39, 3; nĭdhâ̄tōh and nĭdhâ̄t̆ōh, I, 41, 9; trēdhâ̄ and trēdh̆â̄, I, 34, 8; dēvâ̄h and dēv̆â̄h (besides devâsah), I, 23, 24; rōdăsî̄ and r̆ōdăsî̄, I, 33, 9; 59, 4; 64, 9; and r̄ŏdāsyōh, I, 33, 5; 59, 2; 117, 10;

a. Need we wonder then if we find that, on the other hand, they allowed themselves to pronounce prĭthĭvî̄ as prĭ͡thĭvî̄, I, 191, 6; VII, 34, 7; 99, 3; dhri̅shnăvă as dhri̅shn̆͡avă, V, 52, 14; sŭvâ̄nā as sŭ͡vâ̄nā? There is no reason why we should change the spelling of sŭvâ̄nă into svâ̄nā. The metre itself tells us at once where suvâna is to be pronounced as two or as three syllables. Nor is it possible to believe that those who first handed down and afterwards wrote down the text of the Vedic hymns, should have been ignorant of that freedom of pronunciation. Why, there is not one single passage in the whole of the ninth Mandala, where, as far as I know, suvâna should not be pronounced as dissyllabic, i. e. as sŭ͡vâ̄nā; and to suppose that the scholars of India did not know how that superfluous syllable should be removed, is really taking too low an estimate of men like Vyâli or Saunaka.

But if we once admit that in these cases two syllables separated by a single consonant were pronounced as one and were metrically counted as one, we can hardly resist the evidence in favour of a similar pronunciation in a large number of other words, and we shall find that by the admission of this rapid pronunciation, or of what in Plautus we should call irrational vowels, many verses assume at once their regular form without the necessity of admitting the suppression of final s, m, n, or the introduction of other prosodial licences. To my mind the most convincing passages are those where, as in the Atyashti and similar hymns, a poet repeats the same phrase twice, altering only one or two words, but without endeavouring to avoid an excess of syllables which, to our mind, unless we resort to synizesis, would completely destroy the uniformity of the metre. Thus we read:

I, 133, 6. ăpû̄rŭshāghnō͡

Here noūgrâ̄ ĭ͡shănāntă bhūrvănī,
ăpâ̄m ĭshāntă bhūrvănī.