must either suppose that both of these two races, the Arian and the non-Arian, grew up together in India, where we find them in contact from the earliest period, or that one or both of them have immigrated into that country from without. But it seems unlikely that two races whose languages differ so essentially, as those of the Arians and nonArians do, and whose religions also were, no doubt, originally diverse, should have sprung up, and co-existed, in the same country, and under the same climatic infinences. It is much more likely that one or both of them should have been foreign. The fact is that both have probably immigrated into India from the north-west; 15 but the evidence in favour of this supposition is far stronger in the case of the Arian, than in that of the non-Arian tribes. For, besides the proofs derived from the language of the Arians, which clearly connects them with the nations to the west of the Indus, we have the evidence of their complexion, which in the present day is fairer than that of the aborigines, and in earlier times was perhaps still more clearly distinguishable from the dark colour of the latter (see pp. 281, f., 310). But if neither of these two races was indigenous in India, and if they did not at first occupy any portion of that country contemporaneously with each other, which of them is most likely to have been the first possessor? We must, no doubt, conclude that the Dasyus or barbarous races and the Dravidians were the earliest occupants. For, as Lassen observes (see p. 309), we perceive evident traces of the Arians having severed asunder an earlier popula tion, and driven one portion of it towards the northern and another towards the southern hills; and the inhabitants of the Vindhya range, and of the Dekhan, appear always as the weaker and retiring party who were driven back by the Arians. And we cannot ascribe to the non-Arian tribes the power of forcing themselves forward through the midst of an earlier Arian population to the seats which they eventually occupied in the centre and south of the peninsula: for the Arians were from the beginning a more powerful and civilized people than their adversaries, and from a very early period have held them in subjection. It is indeed objected by Mr. Curzon (see above, p. 301), that these rude so-called aboriginal tribes may have been descended from some of the barbaric hordes who under the name of Sakas, Hūņas, etc., are

151 In the App., note O, I shall quote the views of the Rev. Dr. Caldwell and other writers, regarding the origin and relations of the different non-Arian tribes.

mentioned by Sanskrit writers as having invaded India, and some of whom, after their defeat, may have taken refuge in the hills and forests of Hindustan. But I apprehend that this explanation will not meet the facts of the case. We can have no assurance, that such legends as that regarding the Sakas, which is quoted in the first volume of this work (pp. 486, ff.), even if they have any historical foundation, can be referred to any very remote period. For the time at which the Indo-Scythians, who were repelled by Vikramaditya, made themselves masters, and retained possession, of the western frontiers of India, cannot be placed much earlier than the commencement of the Christian era. (See Lassen's Ind. Antiq. vol. ii. 365, ff., 398, 408, 409.) But the traces which we discover in Indian literature of the existence of the Dasyus are (as we have seen from the various Vedic texts cited above) much older than this period.

In conclusion, I return to the point from which I started at the commencement of this volume; and, as the result of the preceding investigations, repeat the following propositions: First, that the Hindus of the superior castes are sprung at least partially from the same race with the Indo-European nations of the west: Secondly, that as the parent race appears to have had its origin in Central Asia, the ancestors of the Indian branch of it could not have been indigenous in Hindustan, but must have immigrated into that country from the north-west.

445

APPENDIX.

NOTE A.-Page 187, last line.

On this subject Professor H. Kern remarks in his recent dissertation, "Indische theorieën over de Standenverdeeling" (Indian Theories on the Division of Classes): "That mention is sometimes made [in the Zendavesta] of three, and at other times of four [classes], proves of itself nothing whatever. The case may once have stood on the same footing as the mention sometimes of three, sometimes of four, Vedas. Here also some have thought to discover a contradiction, and have drawn from it the most adventurous conclusions. When the Hindus speak of the three Vedas, they mean that there is a triple Veda, consisting (1) of recited verses (rich), (2) of verses sung (sāman) and (3) of formulas in prose (yajush), all the three words being comprehended under the name of "mantra." Altogether independent of the three sorts of mantras is the number of the collections of them. Though there were a hundred collections of mantras, the Veda is, and remains, threefold. It happens by accident that the Hindus possess four such collections (and in a certain sense, five), which usually bear the name of the Sanhitas of the Rigveda, Samaveda, Yajurveda (white and black), and Atharvaveda. This does not interfere with the fact that the Atharvaveda is almost entirely a Rigveda, though the larger collection is regarded as the Rigveda in the most eminent sense, whilst the Yajurveda only in part consists of yajush-verses. It does not need to be proved that we must know the principle on which any distribution proceeds before we can deduce any conclusion from numbers." p. 13, f.

NOTE B.-Page 191, line 22.

In his dissertation quoted in the preceding note, pp. 3, f., Professor Kern says of the Atharvaveda: "Sometimes it is difficult to guess what is meant by older and younger. For example, the Atharvaveda is said to be younger than the Rigveda: that has become a sort of article of faith, which some one uninitiated believer receives on the authority of critics; whilst another, again, copies with confidence what has been asserted by the former. Now about half the hymns in the Atharvaveda are, with the exception of an occasional varietas lectionis, the same as in the Rigveda, so that the Atharvaveda cannot be younger than the Rigveda. And it could only be asserted that the remaining portion of the A.V. is of later date, when grounds for this position, derived from language, versification, and style, etc., had been adduced. But, so far as I know, no one has ever even attempted to seek for such grounds. I will show, by a single example, that even in the remaining portion of the A.V. to which I have referred, somewhat may very well be found which, without the least doubt, was known to the Indians in the oldest Vedic period, and even still earlier, though the Rigveda makes no mention of it. In A.V. v. 22, 5, 7, 14, the Bahlīkas (or Balhikas) are named. As Balkh was conterminous with the most ancient abodes of the Arians in India, the Bahlikas cannot possibly have been unknown to the oldest Indians. And yet we find in the Rigveda no traces of these neighbours with whom they were constantly coming into contact, whilst there are such traces in the Atharvaveda."

I am unable to agree with Professor Kern when he alleges that no one has ever attempted to seek for proofs of the posteriority of the A.V. to the R. V. from differences in language, versification, style, etc., between the two, although such proofs have not always been stated in detail. See the remarks quoted from Professor Whitney in p. 190, above. In his Dissertations on the Literature and History of the Veda, p. 12, Prof. Roth writes as follows: "In the pieces which are common to it (the A.V.), with the Rik, it allows itself a great many transpositions and alterations, which further appear to be in most cases of an arbitrary character. In the sections which are peculiar to it, the language approaches to the flowing mode of expression belonging to a later period, though it has the grammatical forms of the older hymns. Between it

and the Rik, there subsists, further, the peculiar relation that the latter too, towards the end (in the last anuvāka of the tenth manḍala), contains a considerable number of sections which bear completely the character of the Atharva-hymns, and are also actually reproduced in the latter. In addition to these general marks of a later origin of this Veda, we find also a number of special characters, of which I here adduce one: The hymns of the Rik celebrate in various ways the deliverances which Indra, the Asvins, and other gods had vouchsafed to the forefathers. The names of the persons so rescued, however, lie beyond the times of the authors themselves, and a Vedic rishi is seldom found to be mentioned. But in the fourth book of the Atharva there occurs, for example, a hymn in which Mitra and Varuna are invoked so to protect the suppliant,-not as they had preserved, for instance, Dadhyach, Rebha, Pedu, and others, but Jamadagni, Vasishtha, Medhātithi, Purumilha, etc., all these being names of men whom the tradition makes to be composers of hymns in the Rigveda. It thus appears to admit of no doubt that the Atharva has not only been collected later than the Rik, but is also of later origin."

In his Dissertation on the A.V.1 pp. 22, ff., the same author writes: "If I have above designated the A.V. as a sort of supplement to the R.V., it is already implied that I regard this collection as later. But it would be a useless undertaking to try to determine its date even approximately, as our information regarding the dates of particular Indian writings is far too uncertain. For the rest, this Veda must, without hesitation, be reckoned as part of the old literature. I shall be able in another place to collect, in the form of a survey, the manifold proofs which may be drawn from the contents of the A.V., to establish the assertions that the greater part of its formulas and hymns are later than the hymns of the R.V., and that this collection has been made subsequently to the other. Here I will confine myself to the single, but quite certain proof, that derived from language.

"With a view to the preparation of the Sanskrit Lexicon, the words of all the Vedic Sanhitās have been completely collected by myself and my co-editor of the A.V., Mr. W. D. Whitney. I can, therefore, state, with tolerable exactness, the number of times that particular words occur in these different Sanhitās. Generally regarded, the 1 Abhandlung über den Atharwaveda, Tübingen, 1856.