gathered round his court from all the surrounding countries; but the language of all these people was different. From their being collected, however, trafficking together, and talking with each other, a camp (Urdu) language became established. . . . At length, the Urdu language, being gradually polished, attained such a degree of refinement that the speech of no city can vie with it."

But it is only in the Urdu dialect, which is used by the Mahomedans and by those Hindus in the north-western provinces of India who have learnt the Persian language, that Persian and Arabic words are extensively employed. The words derived from those sources which exist in the Bengali, Hindi, Mahratti, Guzarati, and other NorthIndian dialects, in the form in which they are generally spoken by the Hindus, are considerably fewer in number. By far the larger portion of words in those tongues are (as has been already said) either (1) pure Sanskrit, or (2) corrupt Sanskrit, or (3) words which can neither be traced in Sanskrit books nor yet are derived from Persian or Arabic, and which may therefore be regarded either as indigenous (i.e. derived from non-Aryan tribes), or colloquial vocables of Aryan origin.

Several interesting questions arise here; as First, how far back can we trace the existing vernacular dialects, Bengali, Hindi, Mahrattī, Guzarati, etc., in the form in which they are now spoken? Secondly, what has been the process of their formation? and, Thirdly, from what source have they derived those words which are not discoverable in Sanskrit, as it has been preserved to us in written records?

The question regarding the antiquity of the existing vernaculars is one which I am not prepared to answer with any precision. Professor Lassen (Institutiones Linguæ Pracriticæ, pp. 59 f.) thinks they have existed since, at least, 1000 a.d. I translate his remarks on the two classes of dialects derived from Sanskrit: "To close this disquisition, I therefore remark that there are two families of corrupted Sanskrit, one more ancient, and not completely broken down, to which belong the Pali and the dramatic dialects; and a second of more recent origin, and diffused in our own day over the provinces of India, which differs more widely from its parent. The former set are genuine daughters of the Sanskrit; the latter grand-daughters, al

though it is to some extent doubtful whether these are the daughters of the former, or of their sister dialects. As regards antiquity, the former family are proved by the history of Buddhism, and of the Indian drama, to have come into existence prior to our era; and it may be shown by probable proofs that the latter arose before 1000 A.D. The discussion of the latter question is, however, foreign to our purpose."

Mr. Beames claims for the modern vernacular dialects a high antiquity, and regards them as springing from an ancient Āryan language, which included elements not discoverable in the classical Sanskrit. His observations are as follows:

"I would here further observe that the written Sanskrit has unfortunately attracted the attention of scholars too exclusively. No one who lives long in India can escape having the conviction forced on him that the written language is quite inadequate to account for many forms and facts observable in the modern dialects. These dialects assert for themselves a high antiquity, and are derived, one cannot doubt, from an ancient Aryan speech, which is as imperfectly represented in Sanskrit as the speech of the Italian peasantry of their day was represented by Cicero or Virgil. The process of selection which led the polished Roman to use only stately and euphonious words-a process which is abundantly exemplified in the pages of modern English writers-was doubtless at work among the ancient Brahmins; and the fact that the cognate Indo-Germanic languages preserve words not found in Sanskrit, but which can be matched from the stores of humble and obscure Hindi or Bengali dialects, is another proof of this fact. The line taken by Professor Lassen, in his valuable Prakrit Grammar, of treating all Prakrit words as necessarily modifications of Sanskrit words, is one which he has borrowed whole from Vararuchi and Hemachandra, and, however excusable in those ancient commentators, seems unworthy of an age of critical research."

It is not, however, necessary for my purpose that I should decide, even approximately, the question of the antiquity of the modern vernaculars. It will be sufficient if I can show that they have becn derived by a gradual process of change from other provincial dialects which preceded them; and which, in their turn, have sprung from the Sanskrit, at some stage of its development.

There is no difficulty in conceiving that the Indian vernacular dialects should have undergone great modifications in a long course of ages. The mere fact above adverted to, which every one recognizes, of their having at a particular assignable date admitted into their vocabulary a large influx of Persian and Arabic words, is sufficient to render it probable that they may have formerly experienced other mutations of various kinds.

The circumstance, too, that the people who inhabit the different provinces of northern India make use of different, but kindred, provincial dialects, Bengali, Hindi, Mahratti, etc., which must, for the most part, at a period more or less remote, have sprung from some common source,' is a proof of the tendency to change which is inherent in all spoken language. For as the inhabitants of all these provinces profess, with some modifications, the same creed, receive the same religious books, and are divided into the same or similar castes, and for these and other reasons appear to be descended, though perhaps not exclusively, from one common stock, it is highly probable that their common ancestors must, at one time, have employed one and the same language: and that that language has in process of time undergone various provincial modifications, out of which the several modern vernaculars have been gradually formed.

We shall also see, a little further on, that the differences between the North-Indian dialects (the old Mahārāshṭrī, Saurasenī, etc.) which preceded the modern vernacular tongues, were few and unimportant; whereas the modern vernacular tongues, Bengali, Hindi, Mahrattī, and Guzarati, differ widely from each other in their forms of inflection and conjugation. This greater divergence between the modern than we find to have existed between the earlier dialects, evinces clearly the tendency to continual alteration, which I have remarked as a characteristic of language in general.

3 Mr. Beames says (Jour. Roy. As. Soc. for 1868, p. 498): "It is, however, clear that each dialect of Hindi has had an independent existence for centuries, and I think an independent origin." This, however, can of course apply only to forms, not to the words which the dialects, whether Hindi or other, have in common; and which in many cases are diversely modified from the Sanskrit original. And although some of the grammatical forms may be original or invented, and not modified from those of any pre-existing Aryan language, there must be other forms which are merely modifications or developments.

I shall first of all state briefly the facts by which it is proved that the modern vernaculars are not, comparatively speaking, of any high antiquity, but have arisen out of earlier provincial dialects: and then proceed to establish these facts more in detail.

First. In extant Buddhist histories, such as the Lalita Vistara composed in Sanskrit, numerous verses, styled Gāthās, are interspersed, the language of which differs from pure Sanskrit, by the forms of inflection being varied or mutilated. This popularized Sanskrit, or something akin to it, appears to have been at one time the spoken language of India; or, at least, this Gāthā dialect exhibits some specimens of that ancient spoken language, and exemplifies the process by which the ancient Sanskrit, itself at one time a spoken language, became gradually corrupted.

Second. It has been discovered that many inscriptions are extant, engraven on rocks in different parts of India, bearing date apparently between two and three hundred years anterior to the Christian era, in which a language differing both from Sanskrit and the modern vernaculars is used.

Third. There are extant in other countries, such as Ceylon and Burmah, very ancient Buddhist books written in a language called Pāli or Magadhī, which also is different from the modern vernaculars, as well as from Sanskrit, while it closely resembles the language of the rock inscriptions just alluded to.

Fourth. In ancient Indian dramas such as the Mrichhakaṭī, Sakuntala, etc., while kings and Brahmans are made to speak Sanskrit, various forms of speech called Prakrit and Apabhranśa are employed for the inferior castes and for women, which in like manner, differ both from Sanskrit and from the existing vernacular tongues.

The four foregoing classes of language have a more or less close affinity to each other; and from the use made of the last three in particular, viz., that used in the rock inscriptions, that found in the Pālī Buddhistical writings, and those employed in the dramas, it is impossible to doubt that either they, or forms of speech closely connected with them, were formerly current, during a long course of centuries, as the actual vernaculars of the periods when they were employed for literary, political, and religious purposes.

But while we thus discover that Pālī and different forms of Prākṛit, such as have been described, were employed in former times, we can find no traces of the modern vernacular dialects, Hindi, Bengālī, or Mahratti, etc., in their present shape, in the ancient records of that same period; and we must therefore of necessity conclude that these modern vernaculars did not at that time exist, but have been subsequently developed out of the above-mentioned Prakrit languages or other pre-existent forms of speech; in other words, that the former vernaculars (or Prakrits) have been gradually altered until they have assumed the form of the modern Hindi, Bengālī, Mahrattī, etc.

As regards the second question started in p. 7, the process by which the modern vernaculars arose out of the earlier modifications of Sanskrit, viz., the manner in which the grammatical forms of the latter, i.e. the Prakrits, were broken down into those of the former, it is not necessary that I should enter into any detailed investigation, although some insight into the process will be afforded by the Comparative Tables which will be given further on. It is sufficient to know that by a particular operation of the general laws of linguistic change, the more recent forms of speech have naturally grown out of the older.

I shall now proceed to supply a more detailed account of those forms of vernacular speech already alluded to, which appear to have preceded the existing varieties, and which are now obsolete. In carrying out this design, it will be advisable to begin with those dialects which seem to be the most recently formed and employed of the four Indian classes of speech which have been before alluded to, viz., first, that found in the Buddhist Gāthās; secondly, that used in the rock inscriptions; thirdly, the Pali; and fourthly, the dramatic Prākṛits. The last-named class appearing to be the most recent, I shall first subject it to examination, and then proceed to the others.

SECT. II.-The Prakrit Dialects employed in the Dramas.

With the view of ascertaining the relation in which the Prakrit languages stand to the modern vernaculars of northern India, I have gone cursorily over several of the dramas in which they are employed, such as the Mrichhakaṭī, attributed to King Sudraka, and the Vikra