*■ w
w-wmt:
i^K
481
.■v*Slc -vHlTNEY
CONTRIBUTIONS FROM THE ATHARVA-VEDA
THE LIBRARY
OF
THE UNIVERSITY
OF CALIFORNIA
LOS ANGELES
I!
'J
V>vV
CONTRIBUTIONS
THE ATHARVA-VEDA
TO THE THEORY OF
SANSKRIT VEIBAL ACCENT
WILLIAM D. 'WHITNEY,
PEOFESSOB OF SANSESIT IN TALE COLLEQE.
J/7. ///J?,/ (From the Journal of the American Oriental Society, Vol. V,'^1866.)
' 0 Digitized by tine Internet Arciiive
in 2008 witii funding from
IVIicrosoft Corporation
littp://www.arcliive.org/details/contributionsfroOOwliit
i:K
CONTRIBUTIONS FROM THE ATHARVA-VEDA
THEORY OF SANSKRIT VERBAL ACCENT.
At a former meeting of the Society, I had the honor to lay before it, in connection with a review of a late work by Prof. Bopp, of Berhn, an attempt to state in a new and improved form the rules respecting the accentuation of the finite verb in the Sanskrit sentence.* That such an attempt was called for, will be evident enough to any one who will refer to the statement of these rules which is given in Ben- fey 's larger Sanskrit grammar, f the latest and most elaborate work of its class, and the only one which professes to treat the subject in an exhaustive manner. It is not too much to say that the account of the phenomena of verbal accen- tuation which is there presented is entirely unsatisfactory, or even unintelligible; that it is plainly wanting in true method ; that it is no orderly development from a central principle, subordinating the more particular to the more gen- eral, and giving each special rule its due proportion in the sum of the whole, but, on the contrary, a chaos of rules and exceptions, empirically stated and confusedly thrown to- gether. That this is so, is not so much the fault of Prof. Benfey, as of the Indian grammarians, from whom, and not from the Sanskrit literature itself, he has drawn the materi- als out of which he has constructed his grammar : doubtless his statement is the best that could be derived from such sources ; its imperfections only prove that the native gram-
f Vollstandige Grammatik der Sanskritsprache, § 127, etc. * See this Journal, vol. v, p. 21 3, etc.
1
161159
marians occupy with reference to tliis department of gram- mar the same position as tt) other departments also ; that "while they are laborious and ingenious assemblers and ar- rangers of particular facts, their shallow philosophy, and laboriously unnatural and arbitrary method, render them utterly unreliable guides for us to a true knowledge of the Sanskrit language, since their rules require to be explained, and limited, and re-arranged, by the light of the very facts which the}" attempt to classify and account for. I referred, at the close of my former remarks upon the subject, to this untrustworthiness of those who had been our chief au- thorities with reference to it, and expressed my opinion, that a rational and exhaustive theory of the principles producing the phenomena of verbal accentuation in Sanskrit, could only be arrived at by a careful study of the phenomena themselves, as laid before us in the various accented Vedic texts. I was then already engaged in assembling from the text of the Atharva-Veda all the material which could aid in elucidating the matter, all the passages in wdiich the ac- cent was not determined by, or in accordance with, the most general rules of accentuation, and which accordingly sug- gested more special rules, or appeared to be anomalous and exceptional cases ; and as I have now completed the collec- tion, I take this opportunity of presenting it to the Society, hoping that it will be found not without value as a contribu- tion to the theory of Sanskrit accent. So far as was in my power, I have classified and explained the facts collected, presenting them in connection with the rules which they illustrate, and have thus been compelled to go over in part the same ground w^hich I formerly traversed ; if of a portion of them I am unable to give a satisfactory account, their statement here will at any rate tend to render possible their future explanation, by facilitating their examination by oth- ers, and their farther comparison with kindred facts, to be derived from the other accented texts.
The first and most general rule for the accentuation of the verb in the Sanskrit sentence is this. In a direct or inde- pendent sentence, or clause of a sentence, the finite verb is made enclitic upon any word preceding it which is directly connected with it in construction. It matters not what part of the sentence that word may be which stands before the verb; whether subject or predicate, whether direct or indi-
rect object, or other limiting circumstance, it takes away the accent from the verb itself. Take as instances the fol- lowing clauses.
amhdyo yanty ddhvabhih (i. 4. 1) ; civd' bhava (iii. 28. 3) ;
rlr4iUHt 5T^ FTFT HklMH^ 4:
tdt krnmo hrdhma (iii. 30. 4) ; tdsya bhdjayate 'hd nah (i. 5. 2) ;
abhi krandaprd trdsaya (v. 21.4); d' viro ztrajdyatdm (iii. 23. 2). Even if other unaccented words intervene between the accent and the verb, the effect upon the latter remains the same : thus
mddhund tvd khandmasi (i. 34. 1) ; ndmas te rudra krnmah (xi. 2. 3).
It is well known that, by the operation of this rule, the Sanskrit verb is in a large majority of cases deprived of its accent. Thus verbal forms of the root ^, kar^ which are perhaps found in the Yedic texts with greater frequency and in greater variety than those of any other root, occur in the Atharvan four hundred and ninety-eight times ; but only one hundred and forty-six times do they maintain their own proper accent; in the remaining three hundred and fifty-two instances they are accentless or enclitic.
If, however, the verb stands at the head of the sentence, it cannot, of course, be encliticized, but retains its accent ; thus
^'i m dirjMiHiH^ ^f^TTH 3^Tfrrrt ^n^
dargdyamdydtudhd'ndn{\Y. 20. 6); vrccd'mi (^dtr'dndm bdhU'n (vi. 65. 2).
This is in accordance with Greek usage, by which a word usually enclitic remains orthotone, if it stands first in the sentence. As the Sanskrit has no proclitics, its seutcuces always commence with an accented word,
As regards the -vvorkiug of this rule, it is to be remarked that ill poetry each pdda, or iiitimate subdivision of the verse, is treated as if it constituted an independent clause, and a verb standing at the head of it remains orthotone, even though preceded in another pdda by words directly depend- ent upon it. The following is an instance :
dhdtiir devdsya satyena hrnomi pativedanam (ii, 36, 2).
Other cases are i. 8. 3,' 4 ; 17. 1 ; 31, 1. ii. 9. 4, 6. iii. 10. 12. V. 22. 12, vi. 54. 2 ; 60. 3, etc., etc.
But farther, if the verb is preceded in the sentence or pdda only by a vocative, it retains its accent. The reason of this is sufiiciently obvious. The vocative really forms no part of the sentence to which it is attached ; it is neither subject nor predicate ; it is a mere excrescence, a parenthesis ; it is not, then, so connected in construction with the verb that the latter can be made dependent upon it with respect to accent. We have, accordingly,
si'te vdnddmahe tvd (iii, 17. 8) ; vigve devd vdsavo rdhshate ^mdm (i. 30. 1). It is unnecessary to cite more of the numerous illustrations of this principle which are to be found in the text.
By the first rule, as stated, the verb is made dependent for accent only upon some word construed directly with it. If, then, a sentence be composed of several clauses, a verb standing at the head of any one of them will keep its own accent. Instances are
pd'tu grd'vd pd'tu somo no dhhasah (vi. 3. 2) ; dhd drdtim dvidah syondm, (ii. 10. 7) ;
f%wfe4#r qi^i^^
vicvakarman ndmas te pdhy dsmd'n (equal to pdhi asmd'n) {\i 35. 4).
And even if the object of the verb precede the latter, it does not take away its accent, provided it be also at the same time the object of another verb : thus
ydtudhd'nasya somapa jahi prajd'in ndyasva ca (i. 8. 3). Here the first verb is accented' as standing at the head of the pdda^ the second as commencing a new clause ; the division of the sentence being made between the common object and the latter of the two verbs. A similar case, in which a com- mon subject is regarded as belonging especially to the former of two verbs, and the latter one is accented, is
^rnffrT 5t: gw ^v■^ frt
grrwtu nah subhdgd bodhatu tmdnd (vii. 48. 1).
It is not very often that a division of the sentence into separate clauses thus takes place within the pdda, and that at the same time a verb happens to stand first after the divis- ion. And as the phenomenon is an interesting one, as indi- cating the necessity that the word to whose accent that of the verb is subordinated must be immediately connected in construction with the latter, and not a part of any other clause, I give here a complete list of all the instances of its occurrence found in the Atharvan. They are i. 8. 3 ; 17. 2. ii. 5. 4 {bis) ; 10. 7. iv. 5. 6 ; 11. 12 ; 21. 1. v. 2. 9. vi. 3. 1, 2 ; 4. 2 ; 9. 1 ; 44. 1 ; 77. 1 ; 99. 3 ; 136. 2. vii. 14. 4 ; 48. 1. viii. 1. 12 ; 2.3; 4. 1, 13, 18^ ix. 1. 8 ; 6. 61 ; 10. 6. x. 4. 12 ;;.. yn^ . 8.26. xii.3,31.^;xin,1^0,;"4. 48^55. xvi. 6. 1. xviii. 1. 23. \ [^^"/ xix.45. 5;^9. 6; 58.4. Tliere^ is no case in the text in -^ •'*'•''" which a verb occupying this position is not accented, unless it be the following :
FTHT W iw STTprq" ^^
_ _ — ^
taptS gharmo dukyate vdm isM mddhu (vii. 73. 1). If this is to be translated, as the analogy of the next verse seems to indicate, " the gharma is heated ; honey is poured out to you for food," then the verb needs to be accented ^!^ , duhydte, and the reading should be so amended.
In some of these cases, the accentuation is an important indication of the way in which the structure of the sentence is to be understood.
From tliis list I have omitted, however, all those not in- frequent cases which come under the operation of the famil- iar rule given by the Indian grammarians, that a verb is accented if immediatelj^ preceded by another verb. It is perfectly obvious that such a case is in reality only one com- ing under the general rule for the accenting of a verb at the head of its own clause in the sentence : there can be but one finite verb in a single clause ; if, then, any verb immediately follows another verb, it necessarily occupies the initial posi- tion, and cannot be encliticizcd. Thus, in the sentence
tdsmd arcdma Tirndvdma nishkrUm (vi. 27. 1), the accent of the second verb is in no manner owing to the contiguity of the word which precedes it, but to the fact that it is followed by the only word directly connected with it in construction : it would equally require to be accented if the sentence were thus arranged ;
arcdma tdsmdi krndvdma nishhrtim^
and could be made enclitic only by having its own subject
placed before it ; as
rIFTT WT Pl^Ff '^m^\^
tdsmd arcdma nishkrtim hrnavdma.
We might take one of the sentences previously given, and, by altering a little its arrangement, seem to bring its accentu- ation within the scope of the Indian rule ; as
V(m qriH ^^ ^"rt ^fr ^^^:
grd'vd pdtu pd'tii soviono dhhasah ;
whereas in fact the second Trg, pdtu^ would still continue to retain its accent for the same reason as before, and for no other. Fartlier illustration is unnecessary : it is only to be wondered at that a rule so empirical as that of the Indian grammars should have maintained itself so long in currency, and that the true meaning of the phenomenon should not have been sooner remarked.
But there is another class of cases in the Atharvan, in which the verb retains its accent in virtue of its initial posi- tion, while nevertheless it is only by an arbitrary division of the sentence that it comes to be looked upon and treated as occupying that position. This will be best illustrated by an example:
_ ^ _ _ *^
a! no goshu hhdjatd' " prajd'ydm (vi. 55. 2), "Upon us kine bestow upon us progeny." This is capable of two modes of division ; the comma may be placed either before or after the verb ; we may read " Upon us kine be- stoWj upon us progeny," or " Upon us Sine, bestow upon us progeny." The former is the more natural and easy ; but the latter is not inadmissible, even in the English trans- lation, and is notably easier in the Sanskrit original. In the first case the verb would be enclitic, in the second it would be orthotone ; that in the text it actually does retain its accent shows that the sentence requires to be divided in the second manner. Another example is
jihvd' jyd' bhdvati kulmalam vd'k (v. 18. 8), " Lingua ejus in nervum convertitur in sagittam vox ;" here, too, the verb is accented in virtue of the division "Lingua ejus in nervum, convertitur in sagittam vox." We have, then, the rule, that if the verb be both preceded and followed by either a subject or an object, to each of which it equally in idea belongs, it may be regarded as directly construed with the latter of the two, and may accordingly receive the accent.
Instances coming under the action of this rule are not very rare in the Atharvan. They are* iv. 5. 2 ; 9. 9. v. 18. 8; 27. 6. vi. 55. 2; 92. 3; 106. 1; 107. 1-4. vii. 4. 1. viii. 9. 13 ; 10. 12, 13, 22-29. ix. 5. 37. x. 8. 8. xii. 3. 25, 48. xiii. 1. 19 ; 2. 26 ; 3. 12. xiv. 1. 64. xv. 3. 4, 5, 10 :
* In a few of these passages, viz. viiL 10. 12, 13, 22-29. xv. 4. 1-6, the accent has, owing to a misunderstanding, been omitted from tlie verbs in the publislied text, and requires to be restored, in accordance with the unanimous authority of the manuscripts.
t
8
4. 1-6. xviii. 8. 8. In a few of these passages, however, the accent of the verb admits also of being explained in another manner, as we shall see hereafter.
But on the other hand, it is quite as often the case in sen- tences of this character tliat the more obvious mode of di- vision is followed, so that the verb remains unaccented. Instances are i. 12. 3. iii. 13. 6. viii. 10. 16. ix. 10. 26. xi. 9. 10. XV. 12. 5, 9. xvii. 17. xviii. 2. 7 : 4. 11. xix. 10. 7 ; 36. 2 ; etc., etc.
Before proceeding to take notice of farther instances of verbal accentuation in the independent sentence, which are to be regarded as more special exceptions to the rules already stated, or as isolat(,ed and irregular cases, requiring particu- lar explanation, we will consider the condition of the verb in a dependent clause.
The Sanskrit, like the German, distinguishes in a marked manner its accessorv and dependent from its direct and in- dependent clauses, by its difterent treatment of the verb in the two cases. But while the Grerman removes the verb of the subordinate sentence from its natural position, and places it at the end of the sentence, thus changing, for instance, "Ich habe dem Manne das Buch gegeben" to "Da ich dem Manne das Buch gegeben habe," the Sanskrit, on the other hand, alters in a similar case not the position, but the ac- centuation, of the verb, changing it from enclitic to ortho- tone. We have, accordingly, the folloAving . general rule : the Sanskrit verb retains in a dependent clause its own proper accent; and that, too, even at the cost, in case the verb be one compounded with a preposition, of the accent of the prefixed preposition.
As in German the dependent clause is wont to be intro- duced by some word of such signification as necessarily conditions its dependency, a relative or a subordinating con- junction, sp also in Sanskrit it generally contains some form, of declension or of derivation, from the relative pronominal stem ?r, ya. The phenomenon, indeed, has on this account been always hitherto thus stated : " the verb is accented in a sentence which contains a form of ?i, ya ;" but it is im- possible that we should remain contented with so empirical a rule as this ; we must inquire in virtue of what principle it is that such words have a power to make the verb ortho- tone. And that the principle is indeed what it has above
9
been stated to be may be very satisfactorily sliown. For other particles than those derived from the stem ar, ya, if they have a like meaning, and possess the same power to render the sentence dependent, exercise the same influence upon the verb.
Thus ^rT , ce^ (which the ^ac?a-text divides into ^i^i ca it), which means always distinctly "if," and is accordingly equivalent to uf^, yadi, preserves, as the latter would do, the accent of the verb with which it is connected. Instances of its occurrence are ii. 30. 2. v. 17. 3, 8, etc.
But ^, ca, itself, without always losing its proper signifi- cation "and," ol' meaning distinctly "if," is not very infre- quently made use of to assist in indicating the conditionality of a clause, whose verb is then left orthotone. An instance is
sd cd 'tisrji) juhvyd'n nd cd 'tisrjen ndjvhuydt (xv. 12, 8), "And should he give permission, let him sacrifice; and should he not give permission, let him not sacrifice." Some- times, indeed, the particle almost precisely equals Jri^, yadi^ as in the following passage :
f^FT ^BT^ ^^ ?Tn%Frf ^ ^ f^^if^
Mhste ddattd purusham ydcitd'm ca nd ditsati (xii, 4. 13), " Ungiven she harms a man, if he will not give her when demanded." The conditionality of the clause is the main efficient cause of the accenting of the verb ; whether the par- ticle has a full conditional meaning, or is employed merely as an expedient for facilitating the expression, is a matter of minor consequence. The other Atharvan passages of this character are viii. 10. 31. xi. 3. 28, 29, 32^9a, 55, 56. xii. 4. 1, 16, 19, 25.
Whether a clause in any case, without the presence of a word conditioning or indicating its dependent character, can be in such wise dependent as that its vei'b should be thereby rendered orthotone, is a question for the solution of which the Atharvan hardly presents sutlficietit material. There is but a single passage wlijch seems to speak clearly with reference to this point :
2
10
udahhh yd'caty iid gdyati (ix. 6. 48\
" If he offers water" (the hymn is extolling the merit of hospitable attentions paid to guests), " he sings a sdman'^ (that is, "it is of equal virtue with the religious action of singing a sdman'^). Here the conditionality of the first clause seems to be the sufficient cause of the accent of the verb ; and on the other hand, a comparison of the preceding clauses, as
^^fFT ^#fTTFr
alhl vadati prd stduti,
"If he greets them, he utters praise," would seem to lead us to the recognition of this rule : that in such a situation the verb was left orthotone, except when compounded with a preposition, in which latter case the preposition still retained the accent.* But this single passage is not sufficient to es- tablish a general rule : it is to be hoped that material may be derived from the other accented texts which shall clear up the matter. There is, so far as I have observed, but one other passage in the Atharvan, where it seems necessary to regard a clause as conditional which contains no indicatory particle; viz:
etdd vojyotih pitaras trtVyam pdncdudanam hrahmdne ^ijdm daddii (ix. 5. 11),
" This is (i. e. wins) your third (i. e. highest) brightness, ye Fathers, that one gives to a Brahman a goat with five oda- nasy And here the verb is left unaccented, although not a compounded one. Whether the accentuation in either of these passages is erroneous, or how the seeming discrepancy between them is to be otherwise explained, I must leave an open question, until more light can be thrown upon the sub- ject from other sources.
There is one other passage which might appear to re- quire consideration in this connection :
* See Benfey's Grammar, § 127. 1 (remark), 5, 9 (remark 1), 11, for instances of this difference in accentual usage between the simple and compounded verb.
11
IdJcshma hurva xii mdnyate (xii. 4. 6),
"If he thinks to himself 'I am making a mark;' " jet the evidence to be derived from this is not wholly unambiguous, as it would not be altogether inadmissible to suppose the influence of the relative pronoun with which the verse be- gins to extend itself to this part also.
It is a well known fact that, by Yedic usage, the particle ^, hi^ always accents the verb with which it is construed (as does also its^negative, ^Tf%, nahi). This also I ascribe to the conditional force inherent in it. It is, indeed, originally possessed of no such force ; etymologically, it seems to be merely an asseverative particle, akin with i;, Aa, and 5^, aha. It is, accordingly, in the later language not infrequently employed as an expletive, to fill out the artificial structure of the gloka ; and it is sometimes, even in the Veda, found so used, having a hardly appreciable significance in the sen- tence in which it occurs. But it is ordinarily made use of to accompany and point out a circumstance which is put forward as the ground of, the reason for, the inducement to, some other action ; and by virtue of this usage, it has ac- quired a certain degree of causative or conditional force. The transition of meaning may be illustrated by an exam- ple or two. If we say "Help us, thou art surely mighty," (German "Hilf uus, du bist jamachtig,") there is no distinct subordination of the latter clause to the former, and yet the second clause is evidently alleged as the reason of the first, and it is but a step fiirther to say " Help us, for thou art mighty." The Sanskrit sentence
prd no ava hdlavdn hy dsi,
would ordinarily, and with perfect correctness, be transla- ted as equivalent to the latter form of the phrase ; while it would nevertheless, strictly taken, rather correspond to the former. Indeed, as f|;, hi, is never allowed to stand at the beginning of a sentence in Sanskrit, but must always follow some other word, and as it thus, although not enclitic, holds a subordinate position, it is still more clearly shown
12
to be unequal in force to our conjunction "for," In Ger- man, not even " for" gives to the clause which it introduces a dependent form : we say " Hilf uns, denn du bist raiich- tig:" jet the difference between this and the dependent clause "... well du miichtig bist," "... because thou art mighty," is rather a formal than a logical one. There is a continuous scale of dependency in the phrases "thou art surely mighty," "for thou art mighty," "since thou art mighty," " because that thou art mighty," and while in Ger- man only the last is regarded as dependent, in Sanskrit the first is treated as if equivalent to any of the rest, and its verb is accented, according to the general rule for dependent clauses. We may reverse the order of the clauses in the example we have taken, and write
bdlavda hy dsi prd no ava;
and here too we have the verb accented, as if the transla- tion were "Since thou art mighty, help us," "Da du ja machtig bist, so hilf uns;" while, if closely interpreted, it is rather, "Thou art surely mighty, (then) help us," "Du bist ja machtig, (also) hilf uns ;" the particle hardly exercising a stronger force than to establish the relation of the two clauses as, protasis and ajjodosis.
In almost every instance of the occurrence of f|;, hi, in the Atharvan, it has more or less evidently this semi-condi- tional force. Thus we have
ugrd' hi kanvajdvibhani td'm abhakshi sdhasvatim (ii. 25. 1), " Since it is a fierce destroyer of the kanva^ it, the mighty, I have made use of." And again,
f% FT g^rrt i%5^ f^ Hfrif
VI te^ muncantdm vimuco hi sdnti (vi. 112. 3),
"Let them release him, for they are releasers." It would
* The printed text gives, on the authority of all the manuscripts, te ; but the emendation as above is evidently necessary. In many other cases also, the manuscripts confound te and te.
13
be easy to multiply examples, but it is believed tliat enougli has already been said to establish and illustrate that which it was our purpose to show ; that the particle in question derives its power to render the verb orthotone from the weak causative signification which the usage of the lan- guage has given it.
With regard to the particle ^ net (pac?a-text ^ri^i, na it), usage is divided. It occurs, in connection with a verb, but four times in the Atharvan. In two passages, viz. vi. 50. 1. xiii. 1. 12, it renders the verb orthotone, as if, like the kin- dred particle ^ , cet, already treated of, it had acquired a subordinating force, and were equivalent to Latin ne, Ger- man dass nichtl; in the other two passages, viz. ii. 27. 1. xviii.'2. 58, it leaves the verb enclitic, as if the ^, it, merely strengthened the force of the negative, as should be its most natural effect.
The three particles %rr^, cet, qrr^, net, and f|;, hi, illustrate in an interesting manner each other's history. Neither of them has etymologically any relative or subordinating qual- ity ; they mean originally simply " and surely," " not surely," " surely ;" but each has in the usage of the language devel- oped out of this plain asseverative signification another which gives it the power to render the clause in which it is found dependent; and as "if," "lest," and "since," they make orthotone the verb with which they are construed.
The particle f^ , kirn, nowhere in the Atharvan exercises an influence upon the accent of the verb in its clause, even where, as in vii. 56. 6, 8. viii. 4. 14(?). xviii. 1. 12, 33, it appears to ask a direct question. In v. 11. 5, pdda c, a part of the manuscripts do indeed accent the verb, yet the weight of authority is in favor of the text as printed. When the particle means "what?", "why?", or "how!", as in v. 13. 7. vi. 45. 1. ix. 10. 18, etc., etc., of course no effect upon the accent would be expected from it.
We have thus seen that the direct subordination of one clause of a sentence to another has an effect to render ortho- tone the verb of the subordinated clause. We have also remarked, when treating of the particle f|;, hi, that the sub- ordination does not always require to be absolute, but that a distinctly defined relation of two of the clauses of a sen- tence to one another as protasis and apodosis was sufficient to
i
14
preserve the accent of the verb in the former clause. We have now farther to notice, that this principle has in the usage of the language received a somewhat inorganic exten- sion ; that it has been strctclied to cover cases to which it did not in strictness apply. Such I conceive, namely, to be the explanation of the accent of the verb in a very considerable number of passages, where two clauses stand as correlatives to one another, or even where there is such a parallelism between them that they may be regarded as in a manner correlative. The coordination is treated as if it were a sub- ordination ; the first of the coordinate clauses is looked upon as a protasis^ to wdiich the other constitutes an apodosis, and the verb of the former is allowed to remain orthotone.
Thus, for instance, when ^rjr— =g7?j, anya-anya, "the one ■ — the other," stand opposed to one another, as subject or as object, in two like clauses, the verb of the first clause re- tains its accent. Take as examples
^ FT 5^RT ^^ ^{^'t ^W[ ?T?7T:
ddksham te anyd dvd'tu* vy dnyo vdtu ydd rdpah (iv. 13. 2) ;
ny dnydm cikyur nd ni cikyur anydm (ix. 10. 16). The other instances are vii. 81. 1. ix. 9. 20 ; iOr2B. x. 7. 42. xiii. 2. 11. Also the passage vii. 35. la, b, may properly be regarded as coming under this rule, although only one ^5^, anya^ that of the first clause, is there expressed; the other is contained in idea in the second clause. A.^^' /J|f. c-^^r In two cases, viz. x. 8. 36. xi. 8. 33, we have, instead of J^o,. f.i<>Hi'i ^^^^ ayiya^ ^— '^, eha — e^a, with the same meaning, and .^.vjj-s*t^r>. with a similar effect upon the accent of the verb.
In X. 8. 7, 13. xi. 4. 22, we find a like correlation pro- duced by the use of mr— mr, ardha — ardha, "the one half — the other half."
Bat even where the correlation is less clearly and sharply brought out, if there is nevertheless a distinct antithesis, the same phenomenon of verbal accentuation is not infrequently presented. Thus we have in vi. 11. 3 an antithesis of " else-
* The reading of the printed text is false, and must be amended to agree with this.
15
where" and "here;" in xii. 2. 32, 55, of ''those there" and "these here;" in iv. 5. 7, of "others" and "myself;" in vi. 67. 3, of motion "away" and "hither;" in ix. 10. 9, of "to- day" and "yesterday;" in ix. 10. 23, of an idea and its nega- tion. Moreover, wherever srr— sit, vd — vd, "either — or," are construed in two clauses with two separate verbs, the corre- lation is reo;arded as distinct enouo^h to occasion the accent- ing of the first verb ; the instances are not numerous, but they are all those in which this particle so occurs ; they are V. 1. 7 (where, however, the effect of the relative in the sec- ond clause might possibly be supposed to extend back into the first), viii. 4.v9. In the following passage,
tdsya vd tvdm mdna ichd' sd vd lava (xviii. 1. 16), the accent of the verb is unquestionably due to the same cause, although the sentence is incomplete, a part of the second clause, including its verb, being left to be supplied in idea from the first. More numerous are the cases in which the antithesis of g"— =Er, ca — ca^ " both — and," produces the same efiect : they are ii. 6. 2 ; 13. 3. v. 4. 9 ; 23. 7 (where we have also, as in ix. 10. 23, the antithesis of an idea and its negation), vi. 110. 1. xiii. 1. 34 (ter). xvii. 6. xix. 24. 5, 6. In vii. 5. 5 is a like antithesis of 3rT— 3rr, uta — uta, unless we are rather to suppose the correlative force to lie in the two contrasted instrumentals. The following passage,
striyag ca sdrvdh svdpdya gunac ce 'ndrasakhd cdran (iv. 5. 2), has been included above among the instances of initial ac- centuation, but is perhaps rather to be explained as an anti- thetical sentence of the class here treated of, of which the second member is defective, its verb requiring to be supplied from the first, as in the passage xviii. 1. 16, just now cited. Several other of the passages formerly referred to may also . receive a similar explanation: thus i v. 9. 9 (where si^i^^, „'v)|^AA'^ jambhayat, perhaps requires to be amended to ^^n;u, jmnhhaya)^ V. 27. 6. vi. 107. 1-4. vii. 4. 1. ix. 5. 37. xii. 3. 25 : while vi. 106. 1. viii. 9. 13. xiii. 3. 12. xiv. 1. 64, admit of being looked upon as defective antitheses of the other kinds here treated of.
16
Besides tliese, there are a few passages, composed each of two clauses, in the first of which the verb is left orthotone, where the antithesis is less distinctly marked than in the cases hitherto noted, while nevertheless their accentuation seems to be referable to the same principle. Thej are vi, 82.2; 83.1. ix.5.22; 8.10. xii.3.18. xiii.2. 30 b. xiv.1.13. Had we these passages only, we should not venture to derive from them any such principle ; but, having well established it as a tendency of the language to assume, even on slight occasion, an antithetical relation, and to accent accordingly, we are justified in presuming its extension to these cases also.
We have thus far found all the phenomena of verbal ac- centuation of which we have taken note to be occasioned, more or less regularly and directly, by the working of a single principle ; that, namely, the verb in an independent clause is accented only when occupying the initial position, being otherwise made enclitic upon any member of the same clause by which it is preceded ; while, on the other hand, it maintains in a dependent clause its own proper accent. But there are in the Atharvan a number of instances of accented verbs, which do not seem to fall so clearly within the sphere of action of this principle : either they are the effect of a wholly irregular extension of it beyond its proper limits, or they are due to the operation of some other principle, which needs to be evolved and stated, or they are isolated cases, destitute of all analogies, and on that account of doubtful authenticity. Before we proceed to the consideration of these remaining cases, we must take notice of the condition in which the accentuation of the Atharvan is presented by the manuscripts of the text now extant. The whole text is very much less accurately and correctly constructed than is that of the Rik : there are to be found in it gross blunders, of which the correction is almost at the first sight apparent, and many passages are in a very corrupt state, requiring extensive emendation. But it especially abounds in palpa- ble errors of accentuation : many of these we have even not hesitated to amend in the published edition : thus, words of frequent occurrence have been in an instance or two accen- ted upon the wrong syllable ; nominatives have been erro- neously taken for vocatives, and deprived of their accent, or vocatives have been falsely regarded as nominatives, and
rf.
have received an accent to which they were not entitled ; the true point of division between the two pddas of a hne has been mistaken, and vocatives and verbal forms have been in consequence wrongly accented, or left unaccented, as they were wrongly supposed to stand, or not to stand, at the beginning of the second pdda ; the verb of a clearly dependent clause, even after a form of the relative u, ya, has been left enclitic ; and so on. More than a hundred such cases have been corrected by us in the published text, and not a few which we have left untouched still call for emend- ation : our commentary will, of course, fully explain and account for the alterations we have made in the text offered by the manuscripts, and will point out the places where we suppose that farther alteration is demanded. It may then, of course, not very infrequently be the case, that verbal forms are erroneously accented by the manuscripts ; it would be strange if it were not so, at least in some instances ; yet in so much uncertainty has the subject of verbal accentua- tion hitherto been involved, that we have only very rarely, and in cases which seemed quite clear, ventured to take away from a verb an accent which our authorities gave to it. Sometimes, indeed, we have allowed ourselves even that liberty : I will proceed to give the instances here, in order to avoid the possibility of having omitted from this paper material which ought to be embraced in it. Thus, in iv. 32. 1, we have read
sdha Sjah pushyati vicvam dmcshdk,
while ali the manuscripts give ^: q^fn, ojah pushyati^ be- ^
cause the former reading seems better to suit the sense, and • "' because the Eig-Yeda, in the corresponding passage (x. 83. 1), leaves the verb unaccented. It might, nevertheless, not be impossible to account for the reading as given by the Atharvan manuscripts : if we regard the two words preced- ing the verb as objects of the verb of the preceding pdda^ or, better, if we look upon the word following the verb as a noun constituting an independent object of it, translating "might, strength — he acquires everything in succession," then the verb wovdd be entitled to be accented in virtue of its initial position.
3
18 Again, in iv. 31. 2, stands in the text
agnir iva manijo tvishitah sahasva sendni'r nah sahure
Mud edhi, whereas all the sanhitd manuscripts (excepting one, which is amended to the above reading) give ^pg-, sahasva. The Rik (x. 81. 2) leaves the verb unaccented, which, with the pada manuscript, and the amended sanhitd, seemed to us sufficient authority for the reading which we have adopted. Yet even here I do not regard the accenting of the verb as certainly erroneous : it might be defended by the analogy of vi. 32. 2, and of the other passages cited with the latter above, as an indistinct antithesis.
Another case, iv. 31. 7, is clearer ; we read
pdrdjiidso dpa ni layantdm.,
spite of the authority of the manuscripts, which are unani- mous in favor of yiUc^dm , Idyanidm. Here also the Eik (x. 84, 7) has the former reading, nor does there seem to be any conceivable reason why the verb should be accented, nor, if it were so, could the preceding preposition maintain its ac- cent also, as the manuscripts allow it to do. We have evi- dently a mere blunder of the manuscripts to deal with in this passage.
In iii. 2. 1, all the manuscripts read
^\^ j^^ wm i%^
agnir no dutdh pratyelu vidvd'n,
which we have altered to ti;irirTi prdty etu. The analogy of the first line of the preceding hymn was sufficient authority for the alteration, and it is not easy to see how the manu- scripts should have come to commit the error of accenting the verb here ; unless, possibly, they were led away by the fancied analogy of the last pdda of the second verse in the
19
preceding hymn, where it is in fact the presence of a % hi, that makes the verb orthotone.* In V. 12. 2, the manuscripts have
tdnunapdt pathd rtdsya yd'ndn mddhvd samanjdnt svaddyd sujihva,
while the printed text gives ^st^ttt, svadayd. The latter reading was adopted on the authority of the corresponding passage in the Rik (x. 110. 2), as there seemed to be no rea- son requiring the verb to be accented. Yet here also, it might be possible to defend the reading of the manuscripts : if the accusative in the first ^ao?a be regarded as the object more directly of the participle than of the verb, as would be allowable, the latter might be looked upon as occupying an initial position, and therefore entitled to retain its accent. Again, in vi. 181. 2, the edition has
d'hute sdm iddm namah,
while all the manuscripts agree in reading t^:, ndmah. The propriety of the emendation cannot be questioned: the false reading may have been a mere slip of the pen on the part of the scribe of the original manuscript, or the word may have been mistaken for the frequent noun titt:, ndmah. Another very similar instance is found in xviii. 2. 36,
J[t fTT ^\f^ rVft W^ ^ 7F^{ FTTi
gam tapa md' 'ti tapo dgne md' tanvdrh tdpah: here, too, there seems to be no assignable reason why the last word should be accented : I suspect it to have been taken, by a blunder, for the common noun fw: tdpah^ " pen- ance," and would alter the reading to fr^ ft^;, ianvdm tapah. Once more, in xiv. 1. 16,
tad addhdtdya id viduh,
* But where, by an error of the pre3a,pr(U}f e^w stands, instead ofpratyitu, ivbicb the maauscripts corr«ctly ^v«.
20
all the Athai'van manuscripts give rfl:^:, idviduh: as the
accent of the verb seemed in this passage quite unexplaina- ble, we have not hesitated to amend it to an agrement with the parallel passage of the Rik (x. 85. IG).
These are all the instances in which we have taken away from any verbal form an accent given to it by the manu- scripts ; excepting two, which bear plainly on their face the evidence that they are blunders, being accented upon the wrong syllable. These are i. 24. 1, ?^q- ^^ rupdin cakre, for
which all the manuscripts have ^sm^ cdJcre, while the true accentuation of the form, if accented at all, would be ^wr, calcre; and xii. 4. 28, \su sr^jri^rT, devd' vrgcanti, in place of which the manuscripts unanimously read aidf^, vr'gcantt, although only the accentuation sraf^, vrgcanti, could be tol- erated. It is sufficiently clear that, in both these cases, the errors are due to a slip of the pen of the scribe who copied the original manuscript from which all ours are descended, the mark of the accent being set over the wrong syllable.
In the light of these facts, which indicate clearly what allowance is to be made for inaccuracies and errors in the text, we may now proceed to examine the remaining instan- ces of accented verbs which it presents.
In iii. 23. 5, we have
yds tuhhyam gdm dsac chdm u tdsmdi tvdm bhdva.
The most natural ending to the verse would seem to be
gdm u ydsmdi tvdm Ikdvah,
" and to whom thou in turn mayest be propitious ;" and it is perhaps not impossible that this is felt to be virtually present in the reading as given, and that therefore the im- perative is accented ; yet there is room to suspect ^sr, bhdva, to be a mere slip of the pen for lisr, bhava. Again, in iv. 1. 4c,
mahd'n inaM' dskabhdyad vijdtdh,
21
it is very hard to see why the verb should have in this pdda an accent which it lacks in the preceding one, where the con- struction seems to be the same. Probably we have to amend to ^^wr°, askahhdyad.
In verse 7 of the same hymn, in the last pdda^
havir devo nd ddhhdyat svadhd'vdn^
we seem, indeed, to have an accented verbal form ; but it is only in seeming; for ^[innrT, ddbhdyat, is unquestionably to be amended to -^^yrni, ddhhdya^ dative of 531, dabha ; and the pas- sage means, he "is not for a harming," i. e. "is not one who can be harmed." This construction, frequent enough in the Rik, is quite rare in the Atharvan, and in one or two other instances has been badly blundered over by the establishers of the text. K we had here, it may be remarked, a verbal form requiring accent, it would have to be accented 5>n^fT^, dabhdydt.
Again, in iv. 19. 2,
nd tdtra hhaydm dsti ydtra prdpnoshy oshadhe^ I am inclined to attribute the hardly otherwise explainable accent of the first verb to an original error of transcription, and to amend to ^^ynRri, hhaydm, asti.
Again, in v. 18. 4,
mr vd'i kshatrdm ndyati hdnii vdrcah,
we might suspect ^frf, ndyati, to be an error for ^THrfFT, nayati;
yet it seems better here to assume an antithesis between the two clauses, of force enough to render orthotone the verb of the first.
Again, in vi. 21. 3, we have
vM sthd k&pdr'hhanir dtJw ha heqavdrdhanih.
22
Here, too, it may be made a question whether we are to find a sufficient antithesis to account for the accent of the verb, or whether we are to suppose that the accent-sign has been sHpped away from the ■^j stha, to the succeeding syllable. I incline to prefer the former.
Again, in vi. 32. 1, the printed text gives
antarddve juliutd .w etdt,
but it is by an error of the press, for ^JrU, juhutd', which is
the reading of the manuscripts. But I conceive this to be a reversal of the original error by which the verb got its accent in the manuscripts, as I am unable to find any reason why it should be left orthotone. In the passage vi. 60. 2,
ango nv dryamann asyd' anyd'h sdmanam d'yati,
the pada-text divides the last word ^•stiijfwi, d-dyati, thus
giving the verb an accent. But I do not see how the form,
w^hich is elsewhere always singular, can be borne as a plural ;
it may, perhaps, be amended to ^wmT, that is, mitiijPrfi, d'
ayanti.
A similar case is vi. 131. 3,
tdfas tvdrh punar d'yas\
which the pada-iQ.xX understands to be ^-s^aRrf^i , d-dyasi^
whereas it is rather ^gri^grrf^i , d' ayasi.
Again, in vi. 78. 2,
rayyd' sakdsravarcase 'md'u std'm dnupaJcshitdu, I can discover no ground for preserving to the verb its ac- cent, and believe the accent-sign to have become lost from under it. I would read ^!WJ°, stdm, etc.
28
Again, in vi. 128. 1,
iddm rdshtrdm dsdd itij
we are perhaps to assume that an accent-sign has been omit- ted under the syllable it, ma, the restoration of which would leave the verb unaccented. Again, in viii. 10. 1,
iydm eve 'dam hltavishydti' Hi,
it may be that the last horizontal accent-sign has been slipped away from its place, and that we have to amend to irfsiwmf^ hhavishyatV 'ti.
Had we these two instances only, of clauses cited by means of the particle of quotation jin, iti, we should be inclined to regard them as cases of the accenting of the verb in a de- pendent clause ; since a quoted sentence is in fact a kind of dependent sentence, and is so treated in some languages, being distinguished in German, for instance, by the use of the subjunctive instead of the indicative mood. And per- haps we may be allowed to explain thus the accent of the two clauses under consideration, even though no other analo- gous passages can be adduced to support this explanation. For, of all the numerous cases in the text (more than thirty), where a clause containing a verb is cited by the particle ^f^, iti, these two are the only ones in which the verb re- ceives an accent. Elsewhere, the quotation is made in the form of an independent sentence, just as it would be spoken ; and that, whether it be the direct object of a verb of speak- ing, as in i. 7. 4, or whether it indicate the "reason why," or the " end for which" (which was its use in the two pas- sages last quoted), as in x. 2. 5. Other instances are iv. 17, 4; 20. 6. V. 19. 9 ; 23. 1, etc., etc.
Again, in xiv. 1. 32,
f^^'^^rr: ^^^ oft wffT
mqve devd'h Jcrdnn ihd vo mdndnsi,
I can discover no reason why the verb should be accented,
24
and suspect tlie true reading to be ^f^^, hrann ihd, a sign of accent having been lost in the manuscripts. Again, in xix. 31. 6,
ahdm pafun&'m adhipa' dsdni m&yi pushtdm pushtapdtir dadhdtu, it seems very uncertain whether the antithesis can be re- garded as being distinct enough to warrant the accenting of the verb in the first pdda. And it is moreover to be noticed, that in the nineteenth book of the text the manuscripts are most especially faulty, so that their authority in doubtful and difficult cases is of almost no weight whatever. I have not pretended to give above all the instances in which we have amended in this book the accentuation of verbs : a record of them may be found among the foot-notes to eacb page. We need not, then, hesitate to amend to wrf^, asdni, if it shall seem desirable, in the passage now under consid- eration.
In the passages thus far treated, we have been inclined to suspect an error in the tradition of the text, where the verbal accent has not appeared to be explainable by ordinary rules and analogies. But there are others in which we seem to / ,-; discover irregular and anomalous applications of some of
the rules previously stated ; which we can hardly regard as errors of transcription, but which may possibly be, at least in part, errors of apprehension on the part of those who es- tablished the text. Whether they are to be understood in this way, or whether they are true and faithfully recorded phenomena of the Vedic language, only of a sporadic char- acter, and not reducible to strict rule, may be better deter- mined when we have before us cases of a like character from the other accented texts also.
We have, in i. 20. 1,
asmin yajne maruto mrddtd nah.
Here the verb is accented as immediately following a voca- tive, although the latter does not stand at the head of a pdda^ and has not itself an accent, as ought to be the case, if the verb is to remain orthotone. A similar case is found in i. 32. 1 ;
iddrh jandso viddtha mahdd hrdhma vadishyati.
The reading fsi^ , viddiha^ may be looked upon as somewhat
suspicious here, since the sense requires rather an imperative form than an indicative, and since the Atharvan offers no other instance of a form in the present tense of either mood from this root, as conjugated after the manner of the sixth conjugation-class. But neither consideration is conclusive against the genuineness of the reading, for analogous forms occur in the Ril^, and the substitution in the Veda of indica- tive for imperative is by no means unknown. And the passage is so closely analogous to xx. 127. 1,
iddih jand upa cruta ndrdgahsd stavishyate^
that it seems better to retain the word in question unchanged,
and not to amend it to in^, viddtham, as it would be very
easy to do, making a fair sense. The accenting of the form would be, as in the preceding case, an irregular extension of the rule for accenting after a vocative. We might possibly understand ^, idam^ as a mere exclamation, translating "See here, ye people! hear!" which would account for the accent; but the analogy of ii. 12. 2, jtz^^'- sjwr, iddin devdh
crnuta^ is against it, nor do I know any other instance of such a use of j^tj , idMm.
Again, in i. 30. 1,
f^"^^ cfH^> f#rHHrii^Fn ^TTTFT wtTfr
v'lcve devd vdsavo rdkshate 'mdm iitd' '^diiyd Jdgt'ld yuydm asmin, the accenting of the verb in the second clause may be looked upon as of kindred character with that in the two passages last treated of. Yet the sentence may be also so divided as to make the verb virtually the first word in its clause ; if, namely, we translate " All ye gods, ye Vasus, guard this person ; and ye Adityas likewise, watch ye over him." Again, in xiv. 2. 42, we have
VOL. V. 63
26
yuvdm hrahmcme ^ nicmdnyamdndu hr'hasjmte sdkdm mdrar.ca dattdm.
Hei'e the structure of the sentence appears to be understood as if the Avords between the vocative and the verb in the second pdda were a kind of parenthesis merely, so that the latter is accented as if it immediately followed the former. " Do ye two . . . . O Brhaspati, Indra also along with you, grant."
Ao-ain. in xi. 2. 2,
cune h'oshtre md' gdrirdni hdrtam aliklavehhyah, etc., we have the verb accented, as it seems to me, by an irregu- lar application of the rule allowing the verb to be treated as if directly construed with the following, instead of with the preceding object. Tlie first two words of this passage do indeed stand in the same relation to the verb as the last one, and, so far as they alone were concerned, the verb would be entitled to remain accented : but the introduction of the other two limiting words alters the case, and should render it enclitic again : this, however, appears to have been overlooked, or else deliberately neglected. I do not see any other way of accounting for the accent of the verb here. Again, in xviii. 4. 54, we have
^T HTjft ^ ^ sHIHIJ^HWIHHTPi^ s^T^
urjo hhdgo yd imdmjajd'nd' 'gmd' 'nndndmd'dhipatyamjagd'ma. The meaning and connection of this passage are very ob- scure ; I do not understand them sufhicently to be able to say whether the last verb is correctly accented, as being of the same construction with the first, or whether it should rather be made enclitic, as belonging to an independent clause, or whether its accentuation is to be accounted for in some other manner.
Again, in i. 17. 2c, d, is read
kanishthihd' ca t/islithati lishthdd id dhamdnir mahi'.
27
This seems to be an incomplete construction of the kind noted above, where an antithesis sufficient to accent the verb of the former clause is produced by the particles g-— g-, ca-ca, "both — and." In this case the second clause con- tains, instead of ^, ca, tpt , iV, but the effect may be looked upon as being virtually the same.
A similar case, perhaps, in found in v. 12. 1,
d' ca vdha miiramahac cikitvd'n tvdm dutdh kavir asi prdcetdh. Here the accent^ of the first verb is at any rate assured to it by the fact that the corresponding Rik passage (x. 110. 1) has the same reading. We might possibly conjecture, as the cause of it, such an incomplete antithesis as was sup- ^^ posed in the last case, the completion of the construction being broken off by an anacolouthon. Or we may assign to the particle g-, ca^ such an office as f|;, hi, would fill, if used in place of it (compare vi. 27. 2. viii. 1. 6.); "bring hither, etc. ; [in that case, or if thou so dost] thou art our messen- ger, etc."
Again, in vii. 35. Ic, d,
iddm rdshtrdm piprlii sd'ubhagdya vigva enam dnu madantu devd'h,
the accenting of the verb in the first pdda seems to be the effect of the assumption of an antithesis between the two clauses, which is facilitated, perhaps, by the more distinct antithetical construction of the preceding line of the couplet. In viii. 7. 21, we find
uj jihidhve standyaty abhikrdndaty oshadMh,
while ?rm cF-Tm, abhi krandati, would seem to be the easier
and more natural reading. We cannot well assume here an error of transcription, nor can we plausibly regard the two verbal forms as locatives of the present participle. I do not understand the reason of the accent as it stands. In xi. 9. 9, 11, 25, we have, three times repeated, the words
28
anitt.reshu sanukshdyan^
Avliich can hardly be translated otherwise than "may they show themselves among our enemies ;" so that the accent should be, according to general analogies, Jri»?lTrf<JH, ^dm ikslia- J yan. But there is something especial and unusual about this phrase, inserted each time, as it i», where it seems not par- ticularly in place ; and it may ha\'6 some relation or signifi- cance which I have not discovered. At present I am com- pelled to pass by the accent as problematical.
There are three passages in which the word sjvtsT, habhuva, is accented, at the end of the line, in a manner which is not accounted for by any general rule. They are as follows :
. . . • ^FT m\mt ^i^[7TT^^ ^4^
svdsa r'sMndm hhutakr'tdm hahhu'va (vi. 133. 4) ;
.... WF^4 srt^it wm
mddhumari mddhyam vinidhdvi hahhu'va (viii. 7. 12) j
.... 5^JR^: qff f^i^ ^^
ddahdhacakshuh pari vicvam hahhu'va (xiii. 2. 44). In neither of these cases is the clause a dependent one, or a member of an antithesis, nor am I able to discover any spe- cial ground for the accent of the verbs. It is to be noted, however, that the verbal form here in question stands in the Atharvan very often, indeed, in almost every case in which it occurs at all, at the end of a pdda ; and that in numerous instances (seventeen in all) it receives an accent in that posi- tion ; not without a distinct reason, it is true, in each case, such as is wanting in the three passages now under consid- eration ; yet it may be that the frequent occurrence of that ending led to the transference of its accentuation to these three passages : the tonic cadence was familiar to the ear, and was accordingly intruded upon a few lines to which it did not properly belong. This explanation, however, I do not regard as very satisfactory, especially as there are also in the ^ext nineteen cases of the same word standing unac-
29
cented at the end of a pdda; I only offer it as the most plausible one -which I am able to suggest.
It will be noticed that no other general principle of verbal accentuation than that first enunciated has been regarded as established, or even suggested, by the passages which we have cited. Some, indeed,* have been incliued to assume/^/ that the verb was occasionally suffered to retain its accent when it was sought to give especial force to the expression, or else when a peculiar emphasis, or distinctive stress of voice, was by the sense required to be laid upon the verb itself. But although it seems highly plausible that such causes should sometimes produce such an effect, there is almost no distinct evidence to be derived from the text of the Atharvan that they do produce it. It might not be quite impossible to force such a explanation upon some of the cases which we have looked upon above as problemati- cal, while yet it would be hard to find in them any reason for accenting the verb which would not equally apply to a great many passages in the text which are actually left to be accented according to the general rules. And it is rea- sonable to require that such a principle be established upon the evidence of a sufficient number of unambiguous passages, before we make use of it to explain doubtful and difficult cases.
But there are a few passages in the Atharvan, for whose explanation we are tempted to suppose the existence and efficiency in the language of this principle of energetic or emphatic accentuation. Thus we have, in the first place, four verses, in which the asseverative particles ^r^, aha^ ^^ it^ and f%n=r, hila^ appear to accent the verbs in connection with which they are taken. They are the following:
mdme 'd aha hrdtdv dso mama cittdm upd'yasi (i. e. upa-d'- ayasi) (i. 34, 2) ;
ahdm vaddmi net tvdrh sabhd'ydm aha tvdin vdda (vii. 38. 4) ; * So Benfey, VoUst. Sanskr. Gr., § 129.
30
mdme 'd usas tvdm kev.alo nd' 'nyd'sdm ktridydg cand (vii. 88. 4) ;
o md'm tt Jala tvdm vdndh cd'klid'in mddhwmathn iva (i. 34. 4). With regard to =Er^, aha^ it is to be remarked, that it nowhere else in the Atharvan occurs in such connection as to show whether it possesses a general power to accent the verb. But, of six passages in which it is found in the first Ashtaka of
/;■ ' the Kik, there is but one in which it exercises such a power. As for the first line given, there is room for suspecting an antithesis (certainly not less than in xix. 31. 6, cited above) ; or the partial analogies of iii. 25. 5. vi, 42. 3 may have had some influence upon its accentuation. In the second in- stance, a very slight change of place of the last accent-sign would rob the verb of its accent.* The particle tht^, it^ is of very frequent occurrence in the text, but nowhere else influ- ences the accent of a verb, unless when in composition with g-, ca, and q-, na^ as before explained. And for the third passage also, the analogy of the parallel passage vii. 37. 1 may not have been without effect. The particle f%^, Icila^ ' occurs in two other places in the Atharvan, viz. in iv. 7. 3. xviii. 1. 15, as also in Eik i, 32. 4, without rendering the verb orthotone : I am not able at present to refer to any other passages illustrating its use.
In these four instances, the accent of the verbs certainly is not of the nature of what we call emphasis ; there ex- ists UQ reason why a distinguishing stress of voice should be laid upon them ; in each case, some other word than the verb is the emphatic one. If the verbs are indeed accented in them in virtue of the influence of the asseverative j^arti- cles, it must be as the utterance of the whole clause takes
M. place with so much additional force, that the verb also shares in it, to the extent of having its lost accent restored to it. And yet it would seem as if this effect of increased energy of enunciation would better express itself by laying a stronger stress upon the already accented sylhibles, than
* And this change has actually been made iu the published text.
SI
by giving it to others which Avere not properly entitled to it.
There are, however, two or three passages, in which signs of a real emphasis are perhaps discoverable. Thus, in ii. 7. 4,
drdiir no md' tdrin md' nas idrishur ahhirndtaynh, the second verb may be accented because the difference of its form from that of the first struck the sense, and seemed to call for a special notice. Yet this is quite doubtful, since we have seen hitherto that, in the case of two correlative and contrasted 5?entences, the tendency of the language was to accent the verb of the first, and not of the second. We have, again, in iv. 18. 6, and repeated in v. 31. 11,
_ V _ _ O *N
ydc cahd'ra nd gacd'Jca Jcdrtitm,
"He who hath done, hath not been able to do;" i. e., "He who hath attempted, hath not been able to accomplish." Here we may plausibly suppose the accent laid npon the second verb to be an emphatic one. Once more, in xii. 8. 26, we read
5T^: fTFftFTT 3 ^^ Tof
quddhd'h salt's id! u cumbhanta evd.
In this passage, as in the last but one, the verb is perhaps marked w'ith its independent accent in order to indicate more strongly its distinction from the preceding participle. Whether the evidence of these few passages, themselves in part doubtful, and capable of a different explanation, will be considered of so much weight that we may found upon it the assertion, that the Sanskrit tends to accent the verb in a sentence which is meant to be expressed with pe- culiar energy, or where the sense lays a peculiar force upon it, is very questionable. The existence of such a tendency must reniain doubtful until new support shall be found for it from the other accented texts. It is not unlikely that these will furnish parallel passages which shall explain many of those which have occasioned us difiiculty, by showing them to be referable to new principles, or to new modes of action
S2
of the principles already laid down, which, by the aid of the material furnished by the Atharvan alone, we have not dis- covered.
We have thus passed in review all the eases occurring in the first nineteen books of the Atharvan, in which the ac- cent of ihe verb was determined by other than the most general rules, and which accordingly either threw light upon the theory of verbal accentuation, or required especial treat- ment, as being of an exceptional and anomalous character. I have not included with them the instances derivable from the twentieth and concluding book, because the more proper occasion for presenting these would seem to be a discussion of the accentual rules as illustrated by the Rig- Veda ; since the book in question forms really no part of the Atharvan, and is only a collection of extracts from the Rik.* For the sake of completeness, however, I append here a brief state- ment of the passages in it which are of like character with those already given for the other books.
Instances of a verb accented because standing at the head
of its own clause, the chvision of the sentence taking place
'PU. />■■'-- "^^'ithin the limits of a j^^da^ are xx. 3. 1 ; 8. la, b; 11. 10;
16. lid; 27,2; Sa^ 46.3; 54.1; 65.1; 67.5; 91.12;
92. 8, 16; 95.3; 117. 1; 137.8.
In XX. 16. lie, we have a case of accentuation of the verb regarded as directly construed with the following, in- stead of with the preceding object.
In XX. 20. 6, the particle g-, ca^ indicates the conditionality of the clause, whose verb accordingly remains orthotone. •*' In 113. ij we either have another similar case, or the word
3iTzi, uhhayam^ with which the verse begins, is a general in- troduction to it, and not speciall}^ connected with the follow- ing verb, which is then left accented in virtue of its initial position : as, " Both these two things — let Indra hear our voice .... and let him come hither," etc.
The particle grfsnT, Tcuvit^ accents the verb in xx. 24. 2, 4, the only instances of its occurrence.
* From this statement should be excepted, of course, the few peculiar pas- sages found in connection with those extracts : yet they also were not to be made use of in an investigation like the present ; their accentuation in the manuscripts is too corrupt to be of any authority ; the editors have had to •accent them anew in accordance with rules and analogies elsewhere established.
.h'ii
33
In XX, 70. 6 is a case of an antithesis with orr-on", vdr-vd, which, as in xviii. 1. 16, cited above (p. 15), is incomplete, the second verb being left to be supplied. ,^,
The passage xx. 55. l^ appears to be another instance of ff-'i-fi'/f^ an incompletely stated antithesis, only the former of the two particles g-, ca, being expressed. It may be compared •with i. 17. 2c, d, and v. 12. 1 (cited on pages 26 and 27).
A distinct antithesis is exhibited in xx. 56. 3d; " Whom, on the one hand, wilt thou slay? whom, on the other, set in the midst of wealth ?" and the usual effect of such a con- struction is seen in the accenting of the former verb.
In XX. 16. ll'd,] the former of the two verbs is accented by the action of the same principle. In xx. 8. 1 c ; 89. Sf we ' ,. x. 9^, have two cases closely akin with v. 18. 4 (see above, p. 21), the correctness of the accentuation in which passage may be looked upon as clearly established by their analogy.
In XX. 67. 7d, we have a case of the irregular accenting of a verb after a vocative, in a like situation as in i. 20. 1 (cited above, p. 24).
In the passage xx. 5. 5,
e ^hi "m asyd drdvd piha.
"Come hither now, of this [Soma], run, drink," the intro- duction of 2^, drava, in parenthesis, between fqsr, piha, and its object, has so broken the continuity of the sentence that the latter verb can no longer be made enclitic, but is suffered to retain an independent accent.
It thus appears that in that portion of the Rik text (about a thirteenth part of the whole) of which the concluding book of the Atharvan is composed, there are no phenomena of verbal accentuation inconsistent with the rules which have been given above, nor any that require other principles for their explanation. Whether, in the whole body of the Rik, phenomena of a different character may be found, must re- main to be decided by examination. Considering the greater amount of material which the older Veda presents, as well as the superior accuracy of its text as fixed by tradition, its speedy examination with a view to this subject is greatly to be desired, in order to the full elucidation of the latter.
5
/>
r
V
UNIVERSITY OF tahf^^"^'* '
University o1 California
SOUTHERN ^^^^Z^),:^S^^''^:^^ZSS
^°^ ""^^elur^r-terto the iibrary from whicii it was borrowed.
PAMPHLET BINDER Syracuse, N. Y. Sfeckton, Calif.
PK 481 • V^6i
3 1
58 01239 743
|
1 |
■ 1 -X |
UC SOUTHERN REGIONAL LIBRARY FACILITY
AA 000 356 514 o
University
Southeri
Librarj
^■isj^^^ rt