Naga ni, Bodo and Garo ang, anga &c, are Tibeto-Ultraindian or Dravirian.

The chief distinction between the Tibeto-Ultraindian and the Draviro-Australian systems consists in the combinations, agglutinations and flexions which are found in the latter. But there is also a real difference in the forms of the roo.8. The proper form of the 1st person in Tibeto-Ultraindian is still nga. This was no doubt the original Indian form also, but from a remote period in the history of Dravirian as an agglutinative formation, modifications of this form have prevailed, the principal being na, ne or en, ing and the contractions e and i. When East Tibetan languages came under the influence of Dravirian phonology similar forms might be produced in them, but in general such forms appear to be of true Dravirian origin. It is not at all probable that so great a transformation as that of nga into i took place in any purely Tibetan language, while the archaic prevalence of e in Dravirian and its original identity with the e of en, eng are certified by numerous facts in different languages. When therefore we find in the obviously compound Naga system, with its flexional Dravirian traits, not only the true Tibetan forms nga "I" and nang "thou" [Gyarung 1st nga, 2nd nan-] but in the plural Ist ni and 2nd ne, and in the possessive 1st i and 2nd ma, there can be no doubt that ni and i are remnants of a Dravirian form of the 1st pronoun similar to the oblique South Dravirian, to the Kol and Limbu, and to the allied forms found in the older or prepositional languages of Ultraindia. Other Gangetico-Ultraindian examples

The comparative table of the Dravirian pronouns will show the great difficulty of distinguishing between the Dravirian and the Tibeto Ultraindian terms. I am by no means satisfied that the classification is correct in all cases. Some of the Himalayan and Ultraindian forms are, in mere phonetic form, as much allied to the southern as to the northern group. The principal facts that have guided me are these. The southern forms of the 1st pronoun in i, e have been produced by the incorporation of the possessive particle i &c. with the pronoun. They are consequently found regularly in the singular. The Tibeto-Ultraindian forms in i, have been produced by the incorporation a Scythic and East Tibetan plural particle, ni, i, (see Horpa) with the pronoun, as is evident from this particle remaining as a postfix in several languages. The Himalayan and the allied Ultraindian forms in i are consequently found regularly in the plural only. Hence I consider the singular ninga Milchanang, inga Milch., Limbu, to be allied to the Dravirian ing, eing, eng, en &c. and not to the plural Garo ning; and the plural ni Serpa, ani Limbu, ain Kiranti, in Murmi to be distinct from the singular aing Ho, ain Binua &c. A few forms in e. obviously Tibeto-Ultraindian (Takpa, Kinawari Tibetan, Mikir) are attributable to the pure y phonetic tendency to replace a by e, found in some of the Tibeto-Ultraindian languages, as is more fully uoticed in the next chapter.

of Dravirian forms occur in the Milchanang and Limbu inga (identical with Kol and Mon-Anam forms), Garo ning and Singpho i (both Pl.), Mikir ne, Nagaung Naga nyi, Khari ni and the Naga forms noticed in a previous page. The Angami a [Manyak also] and the Gangetico-Ultraindian ang (Bodo, Garo), angka Kiranti are probably East Tibetan. The 2nd pronoun is more strongly distinguished in the two systems by its radical vowel, which in Draviro-Australian is i as in Chinese, while in TibetoUltraindian it is a, as in some of the archaic Scythic forms. The Ultraindian members of the Tibeto-Ultraindian family show other Dravirian affinities in their pronominal systems besides the occasional adoption or retention of Indian forms of the roots. Dravirian plurals, possessives and other particles occur in several languages, Bodo, Dhimal, Naga &c (see chap. IV.) Not only the common Dravirian plurals in 1 &c are found, but, as we have seen, the pronominal m.

Amongst the primary affinities of the S. E. Asian languages and Dravirian may be included the plural m and the possessive in i, ni &c. The former is Chinese -mun, -men, -me, -mei, -pei and the latter is Tibetan (yi), Manyak (i), Burman (i), Limbu (in), Bodo and Garo (ni), as well as Scythic, Semitico-Libyan (i) Zimbian (i) &c. The Chinese traits in the Himalayan and Ultraindian languages present great difficulties. Some are of comparatively recent East Tibetan origin and in Ultraindia even more modern. Others appear to belong to a connection as archaic as that between Australian and Chinese roots.

The Draviro-Australian or archaic Indo-Asonesian proninalom system with its numerous distinct elements and combinations, appears to be more ancient or less impaired than most of the systems of other harmonic formations of the Old World. From its general structure it must be considered as cognate with protoScythic or Scythico-American. It is richer than Scythic, which has neither sexual forms nor any plurals save the ordinary generic ones, with the absolute "we" (formed as in Dravirian), although the Scythic power of combining such elements as the formation possesses is similar to the Draviro-Australian, and the position of the subordinate definitives is the same. In some of the [Brahui i.]

Scythic languages a dual is found (ante vol. viii. p. 70),* and as it is preserved in Kol, Australian &c. it was probably common to the archaic Scythic and Indo-Asonesian systems Double plurals occur in Scythic as in Dravirian. The transition forms of Kol and Australian are absent, but the Fin reflexive forms may be considered as analogous remnants of an earlier and richer condition of the Scythic system, when it had departed less from the Semitico-African types on the one side and the American on the other. The Scythic, Caucasian and Africo-Semitic habit of postfixing the pronoun possessively is preserved in the Kol pronominal postfixes to names of kindred and in the ordinary Dravirian persons of the verbs + (ante vol. viii., p. 58). The Scythic postfixed n definitive of the singular is Draviro-Australian. The plural definitives in l, r, are also common to the two formations but the regular m plurals-flexional and postfixual—of Dravirian are not Scythic. In some Ugrian languages the Ist pronoun has then form in the singular and m (the root) in the plural, the former being evidently the definitive postfix left on the elision of the root; and as m does not occur in the plural of the 2nd pronoun it cannot be considered that the Dravirian plural m has any Scythic affinity. The Scythic plural def. k is found in Gond (-k, -nk, g) and Kol (ko). In the Gond pronouns, as in some substantives, it is common and this is also the case in some Ugrian systems (nank Wogul) and in Semitico-Libyan. Combined with the 1, r plural it is found in most of the Dravirian languages (-kal, -gal, -kulu, -kan, &c., so -galai Dhim., -khala,

[It would appear that the dual is not limited to Lap. for according to Castrén it is found in Ostiak and Samoiede also. It is formed by the guttural postfix ga, ka &c., which Castrén derives from ka or ki "also." But is it not identical with the plural guttural particle (ante vol. viii, pp. 56, 70)? "In the Irtishian dialects of the Ostiakian, in Lapponian and Kamassian nouns and adjectives have lost the dual, and pronouns and verbs only have retained it. In the Samoied-Ostiakian it is the pronouns that have lost the dual." Prof. Max Muller in Bunsen's Philosophy of Universal History, vol. ii, 461].

The Asonesian habit of postfixing the pronouns possessively is mainly referable to a Semitico-Libyan source, through Malagasy.

The Dravirian plurals in nar, mar, are probably connected with the Scythic nar, lar (Mongol, Turkish). Those in ra, la are also African, and in Asonesia are thus common to the Dravirian and to the Malagasy derivate formations.

The traces of a labial plural in Scythic are too obscure to be relied on. The Lap. has p, b, as a plural postfix, also dual, and in the 1st pron. -ne, -n is dual. These phonetic affinities with Kol do not appear to me to indicate any glossarial connection. The dual n of the 1st pronoun is evidently the ordinary Ugrian flexion of the pronominal root m, or it is the definitive left as its representative on contraction as in the Hungarian separate forin (en).

--kara Naga, combinations resembling the Samoiede, N. E. Asian and American gada, ganda &c. The Scythic systems in their vocalic flexional plurals and some other traits, are rather IndoEuropean, Semitico-Libyan and Zimbian than Dravirian in their affinities. But we have seen that Dravirian has some traces of vocalic flexion in the change of the agentive a, o of the Ist pronoun to e in the possessive, and in that of i to u in the 2nd.*

The two systems cannot be referred to the same formation, and the affinities, great as they are, must be considered as collateral. They point to a common source, to an archaic postpositional formation at once more crude and more redundant in forms and combinations than Ugrian, Dravirian or even Australian.

The Indo-European system in its possession of a dual number and of sexual definitive postfixes and flexions which extend to the 3rd pronoun, but not to the 1st and 2nd, resembles DraviroAustralian in some of the characters in which it is richer than Scythic. Dravirian in its retention of the sex distinction in the 3rd person of verbs is less abraded than Indo-European. In other respects the latter system is, in its basis form, analogous in roots and structure to the Scythic, although somewhat richer, and has no general affinities with Draviro-Australian save what are observable in Scythic. It is more concreted and flexional than either, although similar flexions and irregularities occur in all three.

The Semitico-Libyan system like the Indo-European, has dual and sexual elements, and in the latter it is richer than either, for it uses them with the 2nd pronoun, and there are even traces of them in the first. The union between the pronominal elements and words used assertively, is more complex than in Indo-European or Scythic, as it has objective or transition forms like DraviroAustralian. The root of the 1st pronoun is Draviro-Australian, but that of the 2nd is not. The postfixed definitive k of the 1st person assimilates the term to the Gond forms in k (nak &c.) The Gond -k although now used in the singular is properly plural and Scythic, while the Semitico-Libyan is generally singular and probably masculine, but in Hottentot it is plural both in the 1st

The Ugro-Fin definitive of the singular changes its vowel to u in the obliqu cases (e. g. mi-na, mi-nu). This may be related to the Dravirian change of the root-vowel i to u in some possessives and plurals. In many of the SemiticoAfrican languages u is plural.

and 2nd pronouns. The fact of both formations having m as a plural, i (variable to e) as a possessive, and u as a plural element can hardly be accidental, but the affinity belongs to the most archaic period in the history of the two formations, like others that will be noticed afterwards. The common radical elements, with the agglutinative and flexional tendencies under which both formations have been developed, have produced several comcidences amongst the various forms which have concreted in both. Thus the possessive i or e represents the 1st pronoun in several Semitico-Libyan languages as a verb postfix or prefix. The Mahrah pl. of the 1st person abu (comp. Hausa mu) is similar to Dravirian forms (abu pl. absolute of Kol, &c.) The Dravirian formation has radical affinities with the archaic ones of S. W. Asia, where it departs from S. E. Asian and Scythic in roots or forms, and although these identical pronominal terms have been independently formed in both formations, the coincidence cannot be considered as purely accidental when it rests on a community of roots and, to a certain extent, of ideologic and phonetic tendeney also.

The Caucasian pronominal systems preserve affinities to those of formations in nearly all the great stages of development. The roots are varied and mixed. The Iron in, an, on sing. of the 1st pronoun and the Kasi Kumuk na are not Scythico-Iranian but Semitico-Libyan, and Draviro-Australian. The plural ma, am, ab is also phonetically, Semitico-Libyan and Dravirian, but it does not occur in the 2nd pron. and is probably Seythic glossarially. The root of the 2nd pronoun di &c. is ultimately a variety of the Chino-Dravirian ni but more immediately connected with Scythic &c. Caucasian has transition forms and attaches the pronoun possessively and assertively to other words, but it wants the complex duals and plurals as well as sex definitives or flexions. The nature and historical import of the affinities between Caucasian and Dravirian are considered elsewhere.

As a definitive postfix the labial is neuter and sometimes feminine in Dravirian as in Sanskrit. In Semitico-Libyan it is plural and masc.

The Semitic plural and dual (Arabic) n definitive may possibly be connected with the dual n of Kol. In Semitico-Libyan languages it is variable to d, l, r, nd, nt, &c. and appears to be radically the same as the Scythic and Draviro-Australian plural element in l. r. In Gangetico-Ultraindian languages li becomes di, ni, nin, ning, &e. Horpa also has ni. In the purer Scythic languages the redu plicated lar, ler, of Turkish becomes nar, ner in Mongol. N forms are also found in Yeniseian (n, ng,) and Yukahiri (l, n,) (ante, vol. viii. p. p. 55 56.)