the definitives themselves, and the African affinities are probably owing to derivation from a like source. The African terms present more affinities in roots and combinations than the Scythic, but Scythic has remnants of similar forms.

The African affinities connect the Draviro-Australian quinary or basis system with the most archaic form of the Semitico-African system more closely than with any other. But the former is sim. pler and more primitive than the latter, in which the sibilant series of terms, found in all the Asiatic systems, blends with the labial and preponderates over it. The African systems appear to have been more influenced by the Semitic in its later gradations, and the Semitic by the Scythico-Iranian, than the Draviro-Australian by any foreign systems. The Draviro-Australian would appear to be the most faithful representative now existing of an archaic S. W. Asian system of definitives and numerals. This system is still homogeneous, the labial being the principal current definitive as well as unit. In the Semitico-African systems there are remnants of the labial pronoun, but the sibilant is now the principal one. The labial unit of these systems is hence more closely connected with the existing Draviro-Australian, than with the existing Semitico-Lybian, pronouns.

It is worthy of remark that the Euskarian, which has close affinities with the oldest form of the Semitico-African systems, preserves a labial 1 ba-t, bo-t, 2 bi, and 5 bo-r-t z, bo-st. It is found also in 9, be-dera-tzi, and 10 ha-mar (ante, sec. 5). The Caucasian, like the Scythic, Semitic and Indo-European, is mainly sibilant, but there are some labial remnants, 2 wi-ba Abkhasian (Eusk. bi, African bi-li, bi &c.); 3 ab-al Lesg., (but this is probably a contraction of chab in which the initial is sibilant as in the Georgian sa-mi &c); 4 wor-ts-tcho Georg., mn.uk-ba, boo-gu, ohw-al Lesg., p-shi-ba, p-tle Circ.; 5, wo-chu-si Georg., p-chi Mis.; 6, f-ba Abkh.; 7 s-wi-di, &c. Georg., wer-al Lesg. buor, uor-l, uosh Misj., b-le, Circ., bish-ba Awar.; 8 rwa, ruo, &c. Georg., mitl-go, mek-go, betel-na, beetl-gu, mei-ba &c. Lesg., bar, bar-/ Misj.; 9, b-gu, boro Circ.; 10 wit, with Georg., wez-al Lesg. p-she Circ. In some of these terms, however, the labial is probably prefixual, The labial system would appear to have predominated in S. W. Asia and spread thence to India and Africa before the

Ꮓ

sibilant acquired its present prominence. Both terms may have co-existed as definitives and units in the oldest pronominal and numeral systems, although their relative importance varied in different eras. The acquired sexual application of the two definitives, and the proneness at one time to extend the application ofthe masculine and at another that of the feminine to inanimate substances, would account for this. The later tendency to throw off the distinction of gender, and to retain only the form in most common use, ends in a still greater impoverishment of the original variety of forms and terms. The Draviro-Australian, like the Tibetan and some other Asiatic systems, has no trace of gender in its labial definitive. In the Semitico-Libyan the labial and sibilant appear to have been also originally common, but at an early period the former became masculine and the latter feminine.

The system may be considered as of equal antiquity with a very archaic formation which was diffused on the one side as far as Africa, and on the other over Central and Eastern Asia. Although the system, both in its terms and in the principle of its formation, has affinities with other languages, it cannot be derived as a whole, or even in the bulk of its materials or in the model of its construction, from any other now extant. The affinities, however, point distinctly to S. W. Asia more immediately, and to an epoch anteriour to the diffusion not only of the SemiticoLibyan and Iranian but of the Caucasian systems. It appears to be of the same archaic origin as the basis of these systems themvessel and of the other systems which were dispersed over Asia before the former began to predominate. The Ugro-African affinities of the Dravirian establish this. There is another test of its relative ethnic position. The remotest and least advanced Asiatic and American systems have only terms for 1 and 2, for 1, 2 and 3, or for 1, 2, 3 and 4. This may be said to be the case with that of the Australian formation, the general Dravirian affinities of which are strong. The Australian proves that the primary Ugro. Dravirian formation prevailed in S. W. Asia, including India, at a barbarous epoch, prior to the expansion of the simple numerals 1, 2, 3, into higher binary and ternary terms by combination and acquired flexion, a process which preceded the adoption of the quinary and denary scales in S. W. Asia, as is testified by the

Iranian, Semitic, Caucasian and other Asiatic and African systems retaining terms so formed. The Dravirian numerals belong to the same era of S. W. Asian civilisation that gave birth to these improved systems, and they must therefore have been brought into use in India long subsequent to that period of its history represented by Australian civilisation. The denary system was not imported by the earliest race, whether Negro or Australian, which laid the foundation of the Indian languages, but by a subsequent race from S. W. Asia, whose civilisation was connected with that in which the subsequent Semitic and Iranian diffusions originated. The Dravirian numerals are not derived from any of the leading Asiatic systems, and their connection with these is extremely remote. The Turkish and Ugrian systems are nearer to the Caucasian on the one side and to the more remote N.E. Asiatic on the other, the Iranian is nearer the Semitic, and the African are nearer the Semitic, the Iranian and the Scythic, than the Dravirian is to any of them. The introduction of the denary scale into India is probably connected with the advance into it of one of those Scythoid races of partially Irano-Semitic character, the archaic influence of which on the physical form of the Southern Indians is so observable. The Todas may be nearly pure descendents of the very race which imported the system.

III. MISCELLANEOUS WORDS.

For the miscellaneous glossarial comparisons of the Ultraindian and Indian division of the present enquiry, it will be convenient to take the list of sixty miscellaneous substantives originally com

• Dr Stevenson in his "Collection of words from the Toda language" (Journ. Bombay As. Soc. i, 155, for 1842) gives some foreign affinities. For 1 he adduces the Latin unus, Tungrs. mukom, Koibal unem. 2, Tungus. djuhr, Arm. yergu. 5 Chinese ing. 6 Turkish alti, Yenis. ram, agam. 7 Arm. yeotn. 8 Arm. ut, Lat. octo, Eng. eight, Sansk. ashta. 9 he explains as 1 less 10. 10 he compares with the Tibetan bachu, bet [the true Tib. form is behu in which b is prefixual and unconnected historically with the Drav. labial root, save in so far that both are ultimately the same definitive].

The Rev. Bernhard Schmid, in his "Essay on the Relationship of Languages and Nations" (Madras Journal v, 133) had also previously (1837) given tables in which the Dravirian numerals are compared with a great variety of foreign ones, but his affinities are too indiscriminating. As I had not read this paper when my comp. voc. was printed I give his list (p. 157) of the Toda terms, which contains some variations not found in my voc. 1 odd, corresponding with Dr Stevenson's orr and a contraction of vodda. 2 atu, ait. 3 muthu, mud. 4 nalk, nank. 5 uj. 6 or, od. 7 or, ud. 8 otthu. 9 unboth. 10 potthu, 11 ponnod &c.

Some remarks on Dr. Müller's comparisons of Dravirian with Scythic numerals will be found in another place.]

piled by Mr Brown in twenty two Ultraindian and East Himalayan languages, and to which other Ultraindian and many Gangetic languages have been added by Mr Brown himself, Captain Phayre, Mr Hodgson and others. Mr Hodgson has adopted this list of substantives for his series of comparative vocabularies, adding to it a large number of words of other classes. I have used the vocabularies of the South Indian languages compiled for him by Mr Walter Elliot and others, and which have been already mentioned in another place, but I have also taken words from my own smaller comparative vocabulary of above 300 words in the compilation of which all the vocabularies and dictionaries within my reach have been availed of. It will be borne in mind that the present paper is mainly directed to phonetic and grammatical affinities, and that the vocabulary in question belongs to the glossarial branch of the Asonesian affinities which will be examined separately. I do not of course assume that the absolute glossarial affinities of the Indian and Ultraindian languages will be accurately represented by the results of an examination of Mr Brown's 60 substantives, and of the pronouns, particles and numerals which have been already adverted to. A collection of whole vocabularies will probably greatly diminish the amount of agreement, because most of Mr Brown's words are of classes that are very subject to diffusion and displacement. It is totally deficient in those words expressive of the most generic actions and attributes which appear to me to be more persistent

than other.

The following is Mr Brown's vocabulary. I have added numbers in order to save the repetition of words in some of the comparative lists, given in the next chapter.*

I have only been able to compare about 40 terms in the list with a large range of foreign vocables. Two of them"Name" and " Village" are not included in my own comparative vocabulary, and several of the others, such as Ant, Buffaloe, Elephant, Flower, Goat, Hog, House, Light, Monkey, Musquito, Oil, Plantain, Root, Salt, Skin, Snake, Tiger, Tooth, Yam, are not included in most of the shorter of those vocabularies which have contributed to its compilation. The omission is especially to be regretted in the case of many of the Scythic vocabularies in Klaproil's great collection.

In estimating the per-centage of affinities I have added 40 words of different classes to complete the hundred. A numerical mode of stating the amount of agreement has been adopted because it is the most definite whatever be the extent of the vocabularies collated; but the value of the result varies of course with the kind and number of the words compared, and all deduction from purely glossarial data must be taken in combination with the evidence of other kinds as to the past and present relations of the tribes themselves. The absolute proportions obtainable from a comparison of entire vocabularies will probably differ greatly from those derived from 100 words. But the relative proportions will not be affected in an equal degree by enlarging the basis of comparison. For example the affinity of the South Indian vocabularies with the Gond may prove to be only 25 per cent. But if so that with the Kol will probably be reduced in a proportion not very dissimilar, so that the relative amount of the South Indian affinities of the Kol and the Gond will not be seriously affected.

In tracing the glossarial history of any formation we must begin with the modern changes. For general ethnology also this is the best course, because the only scientific principle that can guide us in our enquiries into pre-historic events is that nations and their