languages having the same roots, probably belong in part to much later periods, and while some are doubtless of Arian origin in the trans-Indus ages of that formation, others, it is reasonable to conceive, must be of Dravirian origin. If, as appears to result from ethnic evidence of all kinds, the Dravirian formation preceded the Indo-European in eastern Irania, it is very improbable that no native terms were adopted by the intrusive Arian vocabularies. It is equally improbable that in Northern India, where the ancient formation has never been wholly eliminated, Sanskrit did not receive other additions from the vocabularies of the subject and partially helotised tribes.*

When we compare the various forms of roots common to the eastern Indo-European languages-those of Irania and India-with the Draviro-Asonesian, we frequently find that several of the archaic insular forms, Australian, remote Papuanesian &c., are identical with Irano-Indian forms. It is sometimes erroneously assumed that roots common to Sanskrit and Zend with the spoken Medo-Persian and Indian languages are necessarily original in the former and derivative in the latter, and that all the variations from the Sanskrit or Zend forms are corruptions of them. There is no reason to believe that in archaic times one Indo-European nation, speaking one dialect, was ever so civilised, populous and powerful as to occupy all Ira. nia. The Sanskrit-speaking tribe, when it first comes into the dawning light of history, is found restricted to a petty district in N. W. India, and it never succeeded in imposing one dialect even on the basin of the Ganges. The present vocabularies prove that dialects preserving Dravirian ingredients of different kinds have always existed in this province. The living vocabularies of Irania afford similar evidence, for they possess roots that are not Sanskrit or Zend, in common with Indian and Asonesian languages, and varieties of Sanskrit roots which have an equally wide dissemination. A large proportion of these vocables probably existed in different Iranian dialects not only contemporaneously with Sanskrit

See the remarks on this subject in the Introductory Chap. of this Part (ante vol. vi. p. p. 686-8). Dr Stevenson, in a paper which had not reached me when these remarks were written, has alluded to the additions which Sanskrit may more recently that is since it ceased to be a spoken language-have received from native words introduced into the language by provincial writers, and then adopted by lexicogra-phers. Journ. Bombay As. Soc. vol. iv. p. 119.

and Zend during the period of their predominance, but throughout the earlier ages of the formation. Those that are most widely dispersed in the Irano-Indian and Draviro-Australian languages, and those that are found not only in Australian and other archaic Asonesian vocabularies but in Caucasian, Ugrian, western IndoEuropean and African, render it certain that, even in the crude proto-Arian stage of the Indo-European formation, various dialects existed. In this stage the formation approximated to the DraviroAustralian in its general character, and when it is found that dialectic varieties of a common root are also common to existing Irano-Indian and Draviro-Australian vocabularies, it results that there was a period when the external limits of the two formations were not so far sundered as Ireland and Australia, and when the line of mutual contact was further west than the basin of the Ganges. The dialectic varieties were produced not only before the Iranian formation began to spread to the shores of the German Ocean but before the Dravirian began to move eastward on its route to the Indo-Pacific islands. If they belong to the earliest dialects of the Dravirian formation, they must have existed before the Iranian formation took its distinctive shape. It is probable that they belong to the proto-Scythic basis of both formations. They establish an early and close connection between them, and render it probable that they were at one time contemporaneous in Irania.

The further our comparative glossology advances the more minute and accurate will be our classification of the root varieties common to the two families. But until the vocabularies have been carefully compared not only with each other but with those of all the other families of language, their full historical import will remain concealed. While many of the common Irano-Dravirian roots may, by the structure of the vocables in which they occur or by their distribution, be referred to Ugrian or other families, and some to more modern sources, others appear to be entirely pre-Scythic, in other words they are older than the Dravirian and Iranian formations, and older than the Scythic or proto-Scythic formational basis itself. The form of the pure root in such instances is referable to a monosyllabic condition of the family, not only because it is free from any adventitious characters derived

from the phonetic and structural habits of other harmonic families, but because it is extant in these or in the monosyllabic family in a similar form, although in the former it may be concreted with a definitive. The investigation of the proper glossarial history of the formation as such, commences with the separation of this basis portion of the vocabulary from that which has been since acquired. In the Dravirian formation this appears to be less difficult than in the Indo-European. Its basis is closer to the monosyllabic stage. The basis of the latter is Scythic to a large extent.

The other foreign Asiatic affinities of the Dravirian vocabularies must in general be either of similar origin to the common Sanskritic, that is, derived from languages that intruded into India from Irania prior to the Sanskrit era, or they must belong to the preIndian era of the Draviro-Australian formation, and have accompanied it in its first advance across the Indus. This does not exclude the derivation of a certain portion from visitors by sea, and from any alien northern and eastern tribes that may have bordered the Dravirian province before the Tibeto-Ultraindians crossed the mountains. There is no evidence of the existence of such tribes, or of the Dravirian having been preceded in India by any other formation from which words having extra-Indian affinities could have been borrowed.

The affinities of the vocabularies are much more numerous with other foreign languages than with the Tibeto-Ultraindian. They are very various, and those with remote languages-as the Caucasian and North Asiatic-are so abundant and direct, that they afford similar evidence of the long independence and the archaic position of the mother-formation to that which we have found in an examination of the more generic words and particles.

From the time that diffusive nations of higher civilization than the original Indo-Australian existed to the west of the Indus, a flow of foreign words into the Indian vocabularies similar to the comparatively recent Arian current, must have been going on, age after age, and millenium after millenium. Each foreign, mixed or native tribe that spread such words by its migrations and conquests, would become the cause of further movements and diffusions. The Dravirian terms relating to arts and usages appertaining to a higher civilisation than that of the Australians, Sĭmangs

and Andaman islanders, if compared with those of the other languages of the Old World, will probably enable us to ascertain with what races the Indians were most intimately connected prior to the intrusion of the Arians. So far as I have hitherto been able to carry such a comparison, the result is strongly in favour of a great influence having been exerted on the vocabularies of India during pre-Brahminic ages, by Iranian, Semitic, Caucasian and Scythic nations, or by nations of one or more of these races whose vocabularies had borrowed from those of the other races. It is not intended to assert that a Semitic or even a Scythic formation prevailed over Irania as far as the Indus, prior to the Indo-European. That must depend on other than merely glossarial considerations. Whether or not the formation of East Irania remained Dravirian, more or less modified by Scythic influence, until it was displaced by Arian, does not affect the conclusion that, from this province, words of a more western and northern derivation, were transmitted by its tribes to India, during the great interval between the Australian and the Arian epochs. There is no ground to believe that the Caucasian tribes were ever themselves nomadic and diffusive-although other tribes of the same family were-or that purely Semitic tribes speaking purely Semitic languages were ever durably established as far to the eastward as the Indus. The more important modifications which the Dravirian formation has undergone since the Australian era are not of a Caucasian or Semitic character, but, of a Scythic and Scythico-Iranian. Whatever changes the vocabularies of eastern Irania underwent, and however much its tribes were modified physically and in civilisation, the linguistic basis would appear to have remained faithful to the Scythico-Dravirian type. The probability

therefore is that the Dravirian vocabularies derived those Western and Asiatic terms of art and civilisation, which are posterior in origin to the Australian era, mainly from Scythic, Scythico-Iranian and Iranian tribes, that successively dominated in the basin of the Indus. This is far from excluding Semitic influence, direct or transmitted, for most of the eastern branches of the Iranian race, particularly the tribes near the Indus as the Afghans and Beluchis, are physically highly Semitoid.

The first class of N. W. vocabularies after the Sanskritic, with

which the Dravirian fall to be compared, are the remaining IndoEuropean, and particularly the various Medo-Persian. Hitherto the glossarial study of the Indo-European family has been chiefly directed to the vocables and roots common to Sanskrit with the other languages of the formation, so that materials are not yet prepared for an ethnic comparison of the Indo-European roots in the mass with those of other formations. As necessarily happens in an ancient, very widely extended, and much divided family, the roots of any one language, such as the extreme eastern-Sanskrit-form but a small portion of the variety now possessed by the family as a whole. Besides the more modern acquisitions of each vocabulary, there can be no doubt that, as a general rule admitting of exceptions, each large group received most of its peculiar roots from the prior languages of the province in which it prevails, or of those provinces through which the tribes which established it advanced from the original Indo-European seat to the lands where they were found at the dawn of history, and that the radical differences in the glossaries are, in great measure, to be so accounted for. Thus while the Arians, moving eastward into the Dravirian province, would have their vocabularies more or less Dravirianised, the ancient Medo-Persian tribes moving on the Caucasian and Semitic provinces, would have their vocabularies affected by those of the native tribes amongst whom they penetrated. Those hordes which passed through the variable Scythic region or continued to occupy portions of it, would, in many cases, receive fresh accessions of Scythic words. Those which moved north westward would probably receive Fino-Ugrian accessions, while those which went westward through Asia Minor would, for a time, be subjected to influences similar to those which have for a longer period operated on the Medo-Persian. In Europe the pioneer migratory tribes must have come in contact not only with Scythic in the north, but with Euskarian, and probably other Scythico-Libyan languages, in the south. Hence probably it is that the glossarial divergency of the Celtic, the Skipetarian, the Russian, the Armenian and the Sanskrit, is greater than that which divides many languages of entirely distinct formations.

The ethnology of S. W. Asia cannot be well understood until the vocabularies of all the races who occupy it have been carefully compared. A comprehensive comparison of this kind must