Civilisation, which occupies the vast space betwixt Persia and China. A discussion of the Aryan Family of Languages naturally led on to that of the Dravidian, which owes its culture to the former. Mixed up with portions of the Dravidian, but linguistically separate, we find the Kolarian. A consideration of the Kolarian naturally leads to the vast Family of the Tibeto-Burman, which again approaches in some particulars, or was formerly deemed to approach, the Dravidian, and is indebted to whatever culture some few members of the Family possess to the Aryan. Like an island in the midst of the great TibetoBurman sea is the tiny Family of Khasi. The TibetoBurman Family is geographically blended with the two other Indo-Chinese Families, the Tai and the Mon-Anam, which, again, with the single exception of the Annamite, owe their culture to the Aryan Family. When I had exhausted them, I found a residuum of the continent of Asia, partly in the kingdom of Siam, partly possessed by the English, and partly independent, occupied by an eighth Family, the Malayan. The same impulse, which compelled me to hunt up the outlying groups of the Tibeto-Burman Family within the kingdom of China, compelled me to follow up the Groups of the great Malayan Family, passing onward "from island unto island at the gateways of the day," until I reached the coast of Africa in Madagascar, and the coast of China in Formosa. I refused to follow out the other branch of the Polynesian Family, and excluded anything with regard to Papuan, except so far as it incidentally affects the Negritos of the Indian Archipelago. Whitmee will tell us all about that in his forthcoming Polynesia Polyglotta.

But what authority is there for this classification? The reply is that there is none. It is obvious, that so vast a subject could only be treated in some order, and a necessity therefore arose to devise some net, which could be thrown over the whole Field, and this net was necessarily made up of the materials already existing on the authority

of the most esteemed scholars. At the best, the present scheme is provisional, and is the one which causes the least amount of difference. The nomenclature of the eight Families was carefully considered, and the reasons for the entry of certain Languages in certain Families was carefully weighed, and are set out in the narrative of each Family and Language. It was obviously necessary to get rid of such vague terms as Turanian and Allo-Phyllian, and such incorrect terms as Tamulic and Scythian, and such an unduly wide term as Non-Aryan. Time will show whether this nomenclature and grouping of Families will be accepted or amended. It has this advantage, that it is tangible, intelligible, and nearly exhaustive.

When the number of Families were settled, the graver question arose as to the number of Languages. The use of a lax phraseology had complicated the subject. The most esteemed authors use the phrases, "language," dialect," "tongue," ""form of speech," without any degree of precision. Different local names were at random applied to the same Language, drawn sometimes from the physical character of the country, such as Pahári, Purbatya, Desi; sometimes from the political divisions of the country, such as Bengáli, Gujaráti; sometimes from the name of the tribe, such as Dogri, Uriya, Chubháli, Gond; sometimes from artificial and historical causes, such as Tamil, Pushtu, Sanskrit, Prakrit, Kawi. It was above all things necessary to use two words only-"Language" and "Dialect"-as major and minor relative terms, and to determine their relation to each other.

Assuming a Language to be the form of speech of a given population, we have to determine what amount of dialectal variation constitutes a Dialect of a Language, and what extreme degree of variation justifies the claim to be a sister-Language, instead of a child-Dialect. That variation may be of three kinds-1. Vocabulary; 2. Grammar; 3. Phonetics. It is obvious that Portuguese and Spanish rank as sister-Languages, while Venetian and

Tuscan are only Dialects of the parent Italian. We shall find, as we proceed, the difficulty, from lack of accurate diagnosis, of deciding whether Punjabi and Nepálese are Dialects of Hindi or sister-Languages. But in dealing with forms of unwritten and uncultivated speech, new difficulties arise; for clusters of clans are found speaking varying, yet obviously kindred, Languages, and it is a great practical difficulty, whether these variations, in the absence. of any superior or literary standard, are to be classed as Languages or Dialects. Again, clusters of tribes are found. bearing one general name; and yet, upon inquiry, it is found that the component members speak totally different and mutually unintelligible Languages. Thus, in the Tibeto-Burman Family we have the Kiranti, of which Bryan Hodgson records no less than seventeen Dialects; under the single head of Naga we have three separate Languages, each of which has Dialects. Where there exists a literary Dialect totally distinct from the common Dialect, as in Tamil, and a poetic Dialect, such as Elu with regard to Sinhalese, and an Archaic Dialect, such as "old Hindui," the Language of Chánd, and "Hindui,” the Language of Tulsee Dass, differing essentially from modern Hindi, they must be noticed. It would be a mistake to suppose, that a sharp line can be drawn as the boundary of a Language. Instances may occur of half a town speaking one Language, and the other half another; but ordinarily there is a gradual shading off of one Language into another, if they are kindred Languages; or, in the case of Languages belonging to different Families, the population of the transition or neutral zone is bilingual. We read of the Balúchi Language becoming more Persianised as it approaches the borders of Persia, and of a mongrel Language spoken in the Upper Godavery District of the Central Provinces, composed of Telugu and Hindi. Occasionally, where a great river separates Religion, Race, and Language, the line may be drawn sharply, as on the Indus betwixt Peshawur and Rawulpindi of the Punjab Province.

As far as possible, no name is entered as a Language, or a Dialect, unless it is represented at least by a wellauthenticated Vocabulary, and unless it can be pointed out with some degree of certainty within what geographical limits it is spoken. In British India we are approaching a certain degree of precision. Wherever a survey has been made, the habitat of the speakers of a Language can be indicated; but on the north-east frontier of Assam and Bengal, on the upper basin of the Irawaddy and Mekong, in the interior of the islands of the Archipelago, there is great uncertainty. Vocabularies are supplied, and no habitat can be pointed out; or the existence of a tribe with a distinct Language is pointed out, and there is no Vocabulary forthcoming. The subject is, therefore, by no means exhausted. Again, as regards classing certain Languages in Families, where there appears to be no ethnical affinity, or even local juxtaposition, a provisional classification has been doubtfully made: thus the Alfurese and Negritos have been, under protest, grouped in the Malayan Family; and with regard to the Languages spoken in certain islands, I have with great hesitation, and contrary to good advice, provisionally grouped them. It is impossible, however, to say what a shipwreck, or a designed deposit of a ship's cargo of slaves, may have in ages past done, with the unintentional result of upsetting linguistic and ethnical theories.

In late years several valuable books of Vocabularies have been published, notably George Campbell's "Languages of India," Dalton's "Ethnology of Bengal," Hodgson's Essays, Hunter's "Non-Aryan Languages," Lewin's "Chittagong Tribes," M'Culloch's "Munipúr Hill Tribes,” Garnier's "Exploration of the Mekong," &c., &c., and some difficulty has been experienced in accounting for all these specimens, and yet it would not be satisfactory to leave any unaccounted for; at least it is admitted, that, in proportion to the number not accounted for, this book is imperfect. The difficulty arises, as regards some of the

greater collections, that they are necessarily compiled from returns made from a considerable number of districts, and by persons not possessed of linguistic or local knowledge, which would enable them to control double entries under different names, or manifestly erroneous entries. No blame whatever is attached to the compiling authority; and possibly any delay, with a view of testing the return, might have jeopardised the publication altogether. A Language-Map and a Classified List have the necessary result of compelling greater accuracy of nomenclature, and a few years later all these difficulties will disappear.

In fixing the boundary of Language-Fields, the Census Reports have supplied authoritative data, and, though perhaps not always strictly correct, at any rate such as can be accepted until corrected. The circumstance that the territory of Native Chiefs is so much intermixed with the Districts of British India, leaves room for great uncertainty, as, for instance, regarding the boundary of the Telugu Language-Field in the Nizam's territory. Beames, Caldwell, John Wilson, and Erskine Perry have contributed to this part of the subject, but it is at once admitted, that it is only a rough approximation. The same remark applies to the estimated Population of each Language-Field. With the exception perhaps of Bengáli, the whole of which Language-Field is included in one Province, and sharply defined by known boundaries, all other entries are mere approximations, even as regards the great Aryan and Dravidian Languages. As regards the Kolarian Languages, I had the advantage of Colonel Dalton's personal superintendence of the entries in the Map, which may be accepted so far as correct. The relative position of the entries of the TibetoBurman Family within the confines of British India may be depended upon, but the entries of the Nepál Group, the Munipúr-Chittagong Group, and Burma Group are only approximate. The same remark applies to the whole of the Tai Family. That portion of the Mon-Anam Family, which falls within the territory of British India and of the