doms of the Yellow River it was much nearer to the North Eastern and Eastern tribes of Tibet than to those of Ultraindia. I have, in another place, suggested that a special connection in race exists between the Bhotians and the Chinese. The Tibetan civilisation, at all events, is of Chinese origin, and amongst the Chinese acquisitions are included the numerals. The early and wide spread of these numerals over Tibet is proved by their presence in the Ultraindian and Gangetic languages in forms allied to the Bhotian but distinct from them, and obviously very ancient. Some are also closer to the Chinese. The Bhotian term for 7, is not Chinese at all, but Mongolian, Tungusian &c and it has not found its way across the Himalayas. I infer from these facts that the Chinese numerals were bestowed, at a very remote period, on all the tribes of Tibet, and that the Tibeto-Ultraindian and Himalayan forms in general were directly received not from the West Tibetan nation that eventually became predominant, but from the eastern tribes, an inference that is in strict accordance with the other facts from which the East Tibetan relationship of the Gangetico-Ultraindian tribes and languages has been deduced. The numerals of the North Ultraindian languages thus tend to prove that the influence of the Chinese civilisation first reached Ultraindia from Eastern Tibet, using that term in an ethnic sense, so as to embrace those tribes allied in race and language to the Si-fan who are scattered over the western borders of China. At a much later period the Lau appear to have received Chinese numerals and spread them over Ultraindia as far as their range extends. Some of their terms are peculiar, the remnants probably of a native or pre-Chinese system. The Chinese terms in Lau are directly derived from Chinese, and not from an intermediate Tibetan or Tibeto-Ultraindian source.

The question whether the Chinese numerals were current in Ultraindia and the Gangetic basin before the Arian era appears to resolve itself into the more general one respecting the period when the eastern Tibetans crossed the Himalayas into Ultraindia, for there is no reason to think that the numerals were not imported with the other glossarial possessions of the race. The mode in which they are partially blended with nearly all the Mon-Anam systems in the most remote and sequestered parts of Ultraindia

and its islands, appears to prove that they were slowly disseminated along with the other Tibeto-Burman words of which a sprinkling is found in the purer Mon-Anam vocabularies. In the Himalayas the fragments of the older numeral systems have the same character as the Tibeto-Ultraindian. They are Tibeto-Chinese in some of the peculiar Ultraindian forms, with traces of the more ancient Mon-Anam terms. The inference from all the data is that the Burmah-Himalayan tribes carried the Tibeto-Ultraindian numerals with them in their progress up the Gangetic basin and into that of the Indus, and that the Chinese terms were consequently used in northern India before the Arians introduced theirs.

The principal remnants of a pre-Chinese or non-Chinese system in the Burmah-Himalayan numerals are those contained in the terms for 7 and 8. Some of the other terms are also not Chinese, either in a Chinese or Tibetan form.

The Chino-Tibetan terms are, in a large number of the cis-Himalayan languages, curiously blended with older ones. In some cases the ancient binary and quinary principles have been retained, while the trans-Himalayan terms have been partially adopted. In others both systems and both sets of terms are intermixed. There are even languages in which the Dravirian, Mon-Anam and Tibeto-Ultraindian formations have each assisted with numeral roots or modes of combining them. Lastly the Sanskrit and the modern derivative systems of India have here and there contributed a numeral.

Several of the Ultraindian and Himalayan systems take postfixes, e. g. chi, shi or sh Limb.; zho, Chepang; ya, Kiranti (Vindyan -ia); long, Dhimal; ke, Abor Miri; ka, Kuki; ka or kar, Bongju; bo, ple Karen (2 dialects). The N. Ultraindian have also prefixes as with other words,-ta, tha, pha, pe, pi, va, ba, pa; a; i Naga &c; a- Dophla; ga, gi, Garo, ka Mikir (2), Lepcha ka, kha (7, 8, 9, 10, &c). Khyeng has pa- as in Naga. It appears also in the Kuki and Bongju 2, with the postf., pa-ni-ka ; pe-na-kar, and in some of the Himalayan terms, e. g. 4, Lepcha pha-li, Mag. buli, Murm. bli, Gur. pli, contracted in New. to pi; 5, Lepch. pha-gnom, Mag. banga, affording an unequivocal proof of the western influence of the N. Ultraindian formation. In

Singpho it takes the form ma, (3 masum, 4 meli, 5 manga, 8 makat). The prefix in the terms for 4 may be exceptional.

In written Tibetan the terms for 1, 2, 3, 4, 7, 9 and 10 have the prefixed consonants g, b, or d and in some cases it will be found that these have been preserved in cis-Himalayan vocabularies.

The publication of Mr Hodgson's Si-fan vocabularies since the preceding remarks were written now enables me to trace the exceptional Tibeto-Ultraindian numerals to their sources in Eastern Tibet.

The Tibeto-Ultraindian numerals are fully discussed in Appendix C, and I shall here confine myself to some remarks on the connection of the Tibeto-Chinese with the other Asiatic systems and on the distribution of the different varieties in the GangeticoUltraindian province. I begin with the Tibeto-Chinese.

All the Tibetan numerals are Chinese with the exception of 7 and 8, which are quinary and denary. The Bhotian 7, as above remarked, is a foreign engraftment and probably not ancient, as it has made less progress even amongst the Himalayan dialects than other Bhotian vocables and Bhotian varieties of Tibetan vocables.

The formation of 7 from 2 (5, 2) and of 8 from 2 (4 dual, or 10—2, generally the latter) is a common archaic Aso-African idiom,— N. E. Asian, Scythic, Zimbian &c. Hence the prevalent Tibetan terms are normal, and the Chinese exceptional, if the latter be really substantive words. A comparison of the different numeral elements with those of other Mid and North Asiatic systems will throw some light on this.

1. CHINESE, chit, yit, it, i', ih, chek, cha', ja'; (Gyami i).

TIBETAN, gchik, chik Bhot., kati Gyarung, tabi Manyak, (che and chi in 10) ra Horpa, ari Thochu. The Hok-kien chit, Tiechieu chek, (Quang-tung yit) preserve the ancient Chinese form, of which the Kwan-hwa i' is a contraction. The Tibetan and Bhotian forms have not been derived from the Kwan-hwa but from the archaic chit. The common dental and sibilant def. which passes into the palatal, guttural &c; and the full Tibeto-Chinese form is double as in the unit of Yeniseian, chus, khus, huch, hautu, and Kamschatkan, dis, tash (also ta). The Ugrian ik, it, yksi has the k, t, final element, and the slender vowel also connects it with the

Chino-Tibetan. In some of the higher Scythic numerals the unit is preserved in broad archaic forms similar to the N. E. Asian, chut, kut, kuus &c while others have the slender Chinese and Ugrian forms. Thus in 2 Ugrian has kyk, kit &c, Samoiede sit, side &c, Turkish iki (for sik as in 7). The Chinese unit may be compared with the 3rd pron. and demonstratives ki or i, ti, tsz, che, chi, chit, and with the segregative chik. The broad form of Manyak ta is probably an archaic Chinese form, a varying to i in the pronominal system of Chinese, Tibetan and Ultraindian. The Hailam ja' is a current Chinese form in a. The Horpa ra is an example of that common change of t, s to r in the Scythic and Tibeto-Ultraindian phonologies on which I have before remarked. A similar variation takes place in some of the forms of 4. The Thochu a is a contraction of the Manyak-Horpa form. In 2 and 3 the Thochu forms also correspond with the Manyak and not with the intermediate Gyarung, which with the Bhotian forms have a closer resemblance to the current Chinese in its oldest forms. The Thochu and Manyak are probably representatives of more archaic Chinese forms, the dialects which possessed them in China being now obsolete.

2. CHIN. urh, ir, il, li, liang, ni', ji, gi, no; (Gyami liang, ar), TIB. gnyis, nyi Bhot., kanes Gyar., nge Hor., ngari Thoch.. nabi Many. There is little difference between the Kwan-hwa and the other Chinese forms. The Bhotian nyi resembles the Shanghai ni. The liquid definitive is current in Chinese as a demonstrative na "that", and is found in most Aso-African formations. It is a very common element in the numeral 2, but it appears to be archaically a mere variety of t, s in the N. and Mid-Asiatic definitive and numeral systems. In the Samoiede si-ri, si-ti, si-t, Mongolian ko-ir, cho-yur, Tungusian ju-r, dzu-r, Caucasian zu-r, shi-ri, o-ri, ie-ru the final t, s, becomes r as in the Turkish bir for bis. But the Chinese li is probably radically identical with the the first element si and not with the second. The def. appears in the same r form in Dravirian, the archaic connection of the pronouns of which with the Chino-Tibetan has been elsewhere indicated. Dravirian ira-ndu, era-d, ira-t &c, 2. It is also singly or in combination the prevalent Semitico-African root for 2, and a common Aso-African dual and plural particle. The Chinese

forms appear to be connected with the Scythic. From the interchange of k, ch, j, t, and s, and of s, r, 1, n, in the Scythic numeral and definitive systems it is not probable that there is any radical distinction between the forms above given and the Ugrian and Turkish kyk, kok, kit, iki &c. The existence of the r form in the S. E. branches of Tatar (Mongolian, Tungusian) and in the adjacent Chinese, indicates an archaic prevalence of the Samoiede variety in this region and the Chinese may possibly be a contraction of sil, sir. The Tibeto-Ultraindian nyi, ni &c is evidently from the Chinese li, ni, and not a direct derivative of any of the Scythic forms. The final s of Bhotian and Gyarung may have been archaic Chinese, but it is more probably a Tibetan augment. The Thochu and Manyak nga, na are probably archaic Chinese forms, Chinese having na as a demonstrative and no as one of the varieties of the numeral.

3. CHIN. san, sang, sam, sa, ta, (Gyami san, sang).

TIB. kasam Gyar., gsum, sum Bhot., su Hor.; kshiri Thochu, sibi Many. The root appears to be the sibilant def., and as in the binary basis of other systems the same as that used for 1. The broad vowel now distinguishes the form from that used for 1. In the Chinese pronominal system the same definitive occurs as a third pron. in the forms tha, ta " he &c," as a relative in the form so and as an interrogative in the form shu, shui. Similar forms with variations of the vowel (thi, ti, si &c) are current in the Tibeto-Ultraindian pronominal systems. Although the vowel is a in all the Chinese varieties it does not follow that the Tibetan su and si, shi are merely local variations of an original sa, for similar forms may have been current in the archaic Chinesc numeral as in the pronominal system. The Manyak and Thochu si, shi are probably obsolete Chinese forms. From the occurrence of -m in one of the least emasculated of the Chinese dialects (Kwang-tung) and in Gyarung and Bhotian it was probably the original form of the final. If the vocable be native, m must be considered radical, as in other Chinese monosyllabic roots having final -m in the ancient phonology. But the analogies between the Chinese numerals and the archaic N. and Mid. Asian and the irregular character of the Chinese system suggest the cuquiry whether sam may not be a derivative from a