"light," preserves the original of the Bhotian hod in the same sonant form. Ugrian and Turkish retain sonant forms of an ancient root for " river," which has become softened in the prevalent Tatar, Tibetan, Ultraindian and Asonesian glossologies (Comp. Ugr. jug-an, Turk. sug, Turk., Mong. u-sun, chun, Tibeto-Indonesian chang, sung &c.) The Ugrian log, "horse", (also lo) is more sonant than the derivative Himalayo-Burman and Indonesian forms rang, ra &c. In the less emasculated IndoEuropean vocabularies, the sonant forms of the ancient Turanian roots are frequently retained. There can be no doubt that the Chinese and Chino-Ultraindian or Mon-Anam formation was also originally highly sonant, but the strong glossarial affinity of Bhotian to the Ugrian alliance renders it clear that the sonant character of Bhotian was immediately related not only to the archaic Chinese but to the Scythic, and through it, to the archaic Indo-European. It has a greater range of final consonants even than the most consonantal and sonant of the known Chinese dialects, the central and southern. At the period when the Tonic Dictionaries were compiled-the 6th or 7th centuries of the Christian era the phonology of the Kiang provinces was more emasculated than the written Bhotian. The latter probably preserved an example of very archaic Chinese phonologies, anterior it may be to the development of the harmonic phonology and when the mother dialects of Scythic, Indo-European and all of other formations consisted of crude, monosyllabic and tonic roots*.

The Bhotian phonology is much cruder and more archaic than the Scythic or that of any of the other harmonic formations. When the formation separated from the common stock the latter was little in advance of the Chinese, monosyllables and homophons abounded, agglutination was feeble or only beginning to affect the form of vocables, the definitives and other particles were not concreted with substantial words or with each other. The Bhotian phonology contrasts so strongly with the highly harmonic Scythic, that

Since chap. III. was published the Rev. Mr Edgkin in his Grammar of the Shanghai dialect has shown that the sonant tendencies of some of the middle and southern languages are more decided than previous Grammars had led us to believe. In a subsequent section the results of Mr Edgkin's original and important enquiries into the phonologies of the Chinese dialects will be noticed.

it may even be considered doubtful whether the harmonic development had commenced when the mother-dialect of the former was first separated from the proper Chinese. Much of its slight agglutinative and harmonic power has probably been acquired since, and Scythic may have had much influence on its progress. In another place the conclusion was arrived at that the collocation of Scythic was older than its harmonic phonology, and in Bhotion we have a partial example of its pre-harmonic condition.

Its general structure although Scythic when compared with Chinese, Mon-Anam, Asonesian, Semitico-African and Caucasian, is not purely Scythic.

The use of postfixed definitives is an archaic Turanian, Caucasian, Semitico-Libyan and Indo-European trait. The most common Tibetan postfix* ma, pa, va, ba, &c occurs frequently in Ugrian vocabularies, and it is also Semitico-Libyan, Caucasian, Indo-European and Dravirian. In Chinese it is a 3rd pron. The postfixes distinguish Bhotian strongly from Chinese and there can be no hesitation in considering them as of Ugrian affinity. The other Turanian postfixed definitives are na, ni, n, &c; ra, la, ol, el, er, &c; ka, ga, k, &c; s, z, t, d, ch, j; which with the labials comprise the whole range of the Tibetan.

The prefixed consonants of Tibetan b or v, m; h, s, z; l, r, d; g are not prevalent in the Turanian languages, but Hungarian has az, as a separate preposed definitive, and in others vocalic prefixes occur which are probably in many cases contractions. Turkish appears to have prefixual t, d, ch; l, s concreted. The Yeniseian languages will probably prove to be the chief existing link between the proper Scythic and the N. E. Asian and American. In many respects they may be considered as entering with the Samoiedean group into the Ugrian family. But with strong Ugrian affinities they combine independent traits, and others that are N. E. Asian and American. Amongst the latter is the retention of prefixed, along with postfixed definitives, embracing the entire

I give a few examples in which both the root and the postfix are the same in the two families.

Leaf, Bhot. loma, Mordv. lopa;
Finger, Bhot. sormo, Fin sormi;
Rain, Bhot. charba, Sam. serwo.

Scythic and Bhotian range,-má, pa, pi, bi, &c; tá, dá, di, d &c; si, hi, chi, cho &e; al, il, ol, &c; ki, ke, ku, gi, yi, &c; a, u, o, i, e. These definitives are more common as prefixes than as postfixes, and when the habit of the formations which succéed Scythic on the N. E.* and S. W. and have fundamental affinities with it, is considered, no doubt can remain that the distinctively postfixual idiom of Scythic was exceptional in its origin, and was preceded by a condition of the mother-language in which the definitives were current as separate particles, and capable of being preposed as well as postposed according to dialectic taste and fashion. To this proto-Scythic stage of the Mid-Asian formations Bhotian, like Yeniseian, partially adheres. In this respect their form is older than the proper Scythic and more akin to the basis-form of the Caucasian, Semitico-African and other formations that separated from the common stock before the dialect in which Scythic originated had acquired its peculiar postpositional structure. In the use of prefixed definitives as in many other traits the Tibeto-Ultraindian and N. E. Asian families have departed less than the Scythic from the archaic type preserved by Chinese. In Chinese the true definitives precede the words they definet. The full range is also preserved in Chinese, although the definitives are now rarely used save emphatically or as demonstratives. It has ki, ke, chi, che, ti, i, ku, tsze, hi, ho &c; na; and pe, wa. Chinese also uses double demonstratives, or rather the demonstratives followed by the generic definitive or segregative ko, ku,—na ko, che ko, ti ku, i ku, ku ku, In the first stage of an adhesive phonology these would become nako, cheko, tiku, iku, kukų. They are thus the prototypes of the double definitives, prefixes and postfixes found in most of the harmonic formations.

It is obvious that the full forms of the definitives, as in Chinese, must have preceded that in which they lose the vowel and coalesce with the root into one monosyllable. The Bhotian initial consonants were originally separate preposed definitives and they are preserved in the full form as prefixes in other dialects of the

The Aino-Kurilian group has prefixes as well as postfixes-ma, pu, p, f; t, d; so, sa, sha, sh, si, i; no, on, &c; ku, g &c. Yukahiri has also prefixes, but its general habit is postfixual like Scythic.

The Gyarung prefix ki- is the Chinese definitive ki, ke, chi, che. Hence we find coincidences such as kitan Ggyar., chi tun Gyami, egg. Kwan-hwa has the Gyarung vowel tan.

Tibeto-Ultraindian family. In the N. E. Asian, Caucasian, Semitico-African and Asonesian provinces both forms of the prefixes are also found.

In Tibetan the labial definitives are still current in their primary character of substantive words "father", "mother". As a definitive postfix -pá, po has acquired a generic masculine application, and -ma, -mo a feminine, and they are even extended to neuter names. In Scythic both the primary and the sexual significations have been lost. In Draviro-Australian, Indo-European and Semitico-Libyan agglutinated definitives are found retaining a sexual force but with the primary substantial meaning lost. Tibetan here also stands between Chinese and the more agglutinated and concreted formations. In Chinese there are several classes of postposed sexual particles, as in Tibeto-Ultraindian and Dravirian. Thus for human beings Kwan-hwa has nan masc., neu fem.; for the lower animals generally kung m., muf.; for birds heung m., tsze f. As in Bhotian, Indo-European and SemiticoLibyan the idea of gender has been transferred to inanimate things, for which keen m., kwǎn f. and yin m., yang f. are used. In some of the Scythic languages there are traces of a similar attribution of a distinction of sex, energy &c to inanimate objects.

A marked departure not only from the Scythico-Dravirian but from the Chinese collocation occurs in the position of the qualitive, which follows the substantive. This idiom connects TibetoUltraindian with the adjacent Mon-Anam. It is clearly abnormal, because the primary relation of possession and attribution, of which the qualitive is but a variety, is denoted in the Tibeto-Ultraindian languages, as in Chinese and Scythic, by preposing the possessive. Consistently also with the normal structure the adverb precedes the qualitive or verb, and the subject the predicate.

The Bhotian glossary is highly Scythic but in its basis it is independent to a considerable extent and with strong Chinese affinities. The Scythic glossarial basis, in pronouns and many particles and formatives, is so uniform that it may be referred to one mother-dialect. The Bhotian basis is not a modification of this dialect like that of all the Scythic languages. It is a distinct Chino-Scythic sub-formation, and Chinese more than Scythic.

Sec. 2. THE GENERAL CHARACTERS OF THE SI-FAN LANGUAGES AND THEIR RELATION TO BHOTIAN.

Since this paper was written Mr Hodgson has published a series of vocabularies spoken by the tribes occupying the mountainous country between the land of the proper Tibetans or Bhot and that of the proper Chinese. These vocabularies are of remarkable interest. They prove that the Tibeto-Ultraindian formation extends northward, from the most northerly dialects previously included in it [Singpho, Jili] to a point in N. E. Tibet which has not yet been ascertained, but where they appear to be succeeded by Sok or Mongolian tribes identified by Mr Hodgson as the Olet and Kalmak of Remusat and Klaproth. These Mongolians occupy the eastern portion of northern Tibet, the western being in like manner the southern extremity in this quarter of the Turkish province and traversed by tribes called by the Tibetans Hor and considered by Mr Hodgson to be Turkish. These Tatars chiefly roam on the north of the Nyenchhen-thangla range but there are also numerous scattered Horpa and Sokpa in southern Tibet. The new series of Tibeto-Ultraindian vocabularies comprises, 1st the Takpa (of the so-called Towang-Raj west of Kwombo), 2nd the Manyak,* Gyarung+ and Thochu spoken by tribes which occur in this order, between Yunan and Amdo, the latter division of Tibet being occupied by a Si-fan tribe who for the most part speak Bhotian. To these are added the Gyami, a dialect of Chinese, and the Sokpa and Horpa. The last is considered by Mr Hodgson as Turkish, but it appears to be Tibeto-Ultraindian in phonology and glossary. It is a very archaic dialect of Chino-Tibetan, preserving some evidently archaic varieties of the common root now obsolete in Chinese, in its forms intermediate between Bhotian and the East Tibetan dialects but leaning more to the latter than the former, and possessing special affinities with current Chinese and Tatar, from which it may be inferred that Horpa has not only been long conterminous with Scythic languages, but that it was in contact with Si-fan dialects and like the southern Takpa directly acted on by Chinese before the modern expansion of Bhotian to the eastward.

Mr Hodgson describes the physical characters of a Manyak, a native of Rakho, six days south of Tachindo.

↑ Mr Hodgson describes a Gyarung from Tazar, north of Tachindo.