the nouns, have no inflectional or other case signs; in other words, are immutable.* 4th, That the complete form of the pronouns is distinguished by the suffix Ra. Now, every one of these very arbitrary peculiarities belongs to the pronouns in the Gyárúng language not less than in that of Circassia, as the following examples will show; and I should add that by how much the development of this part of speech is anomalous throughout the Tartar or Mongolian tongues, by so much is the instanced coincidence with the Circassian more significant, the anomalous or irregular character of the pronouns of both not sufficing to conceal the coincidence, and therefore doubly illustrating it.

Circassian.-Ab, father. Wara, thou, the full pronoun. Wa, the contracted form, used in composition.

Hence Wáb or Wa-ab, thy father.

Gyárúng.-Pé, father. Nanré, thou, the full pronoun. Na, the contracted form, used in composition. Hence Napé or Na-pé, thy father.

VERBAL USE.

Circassian.—Wará, {a}—kwisloit, thou ridest.

Gyárúng.-Nanré na-syo, thou knowest.

I have changed the Gyárúng verb, because I do not possess the equivalent in that tongue for to ride. It matters not, however, as the sample shows the grammatical form to be absolutely the same in both sentences, just as well as if ride were the verb used in both.

The other rules and examples (scanty, I admit) given by Latham from Rosen may be matched in each instance by

⚫ I have now ascertained that the same principles prevail, with alight variations, in the Hayu, Kuswár, Kiránti, and Limbu languages of the Himalaya, in the U'raon, Ho, Sontál, and Gondi tongues of Tamulian India, and in the Tagala and Malayu languages of the Pelasgian group, though passing out of use in the last-named tongue as in several of the Himalayan tongues. See remarks in the Supplement. I may add that in the Hayu language (of which I have a detailed account nearly completed) the verbs are distinguished into the two classes of transitives and intransitives precisely as in Malay.

Gyárúng rule and sample, as will be seen in the sequel. But there is this difference in respect to the Ra suffix, that it is applied to the first and second pronouns in Circassian, though not to the third; and to the second only in Gyárúng.*

This, however, is in complete conformity with the other and typical Mongolian tongues; for in Mantchú, and in Mongol also, the Ra suffix is found, but attaching only to the third personal; and if we compare the Téré of those tongues † with the Chinese Tá and the Sokpo Thá, we shall perceive the perfect analogy of the suffix throughout these tongues, in spite of its varying applications.

But is there no clue to the irregularities, none to the real force and signification, of this pronominal suffix? Clearly there is; for in the Tibetan language, the word rang, meaning self, and attaching to all the personal pronouns alike, affords us that clue, though the people of Circassia and the Gyárúng, whose common and familiar use of this suffix is so perfectly analogous, seem equally unaware of the fact, and can neither explain the meaning nor the partial application of their suffix, any more than can the Mantchús and Mongols. This I infer from the silence of authors, and should add that the explanations are wholly my own, my Gyárúng interpreter being able only to express very unsophisticated surprise when asked to analyse a word.

But I have not yet done with the analogy of Circassian and Gyárúng pronouns, having still to notice that the third personal in Circassian, which drops the Ra suffix, is not really a personal but a demonstrative, equivalent to ille, iste. Now, the Gyárúng language has a third personal, which the Circassian lacks; but it has also a demonstrative, and that demonstrative is the very

• The first and second pronouns are so nearly alike in Gyárúng (nga, na), that the ré suffix has probably been reserved to the second, in order to difference it more plainly.

+ Recherches sur les langues Tartares, pp. 173, 183. I cannot thus revert to the thoughts of my old antagonist (voce Buddhism) without a fresh tear dropped on the untimely grave of that truly amiable and learned man.

Nga, I, ngarang, I myself, egomet; and so khérang, khórang. Rémusat has sadly confused the Tibetan pronouns, and, as I suspect, those of the other "langues Tartares," though his work be a marvel for the time and circumstances of its publication. Rémusat ut supra, p. 365.

*

same as the Circassian one; that is, ú or w; and this pronoun has, in both tongues alike, a separate, full, and a concrete contracted form. Moreover, in the Gyárúng tongue the forms and uses of this demonstrative afford a perfect elucidation both of its strange metamorphosis (w to t) and of its anomalous suffix (i) in Circassian; for "watú" is the complete separate form in Gyárúng; whilst "wa," the contracted form, alone used in composition, constantly takes í, which is really a genitive sign and recognised as such in Tibetan, but is a mere “particule morte" in Gyárúng as in Circassian. Take the following samples from Gyárúng: Watú, he, iste, ille: Wapé, his father: Womo, his mother: Waimyek, wa-i-myek, his eye (myek, eye): Shaimek, shai-i-mek, leaf of tree (shi, tree, mek, leaf); and then turn to the Circassian samples in Latham, ú-í, he; t-ab, his father;t í-kwisloit, he rides, and you will perceive that (ú being the same with w) the nominal t and the verbal í of Circassian are the secondary or suffix portions of the full Gyárúng pronoun exalted into primaries in order to difference the third person from the second, the second already having the wa or ú (wab, thy father; ú-kwisloit, thou ridest) form. And that such substitution of the secondary for the primary part of a word is no arbitrary assumption of mine, but a regular principle of the Caucasian and of the Mongolian tongues, may be seen by the numerous examples of it occurring in the subjoined list of vocables. The above elucidations of Circassian pronouns for which I am entirely answerable, are so thoroughly in the spirit of Bopp's system that I trust they may find favour

* The change of wa into wo, in wapé and womo, is an instance of that vocalic harmony which these languages so much affect, and which has been erroneously supposed to be peculiar to Turki. We have abundant alliteration both vocalic and consonantal out of, or beyond the Turki branch of, the Mongolian tongues. Shaimek, from shi and mek, has other peculiarities precisely similar to what occur in the Altaic tongues, teste Remusat.

+ In the supplement to this paper will be found an exact and beautiful pendant for this Circassian sample, derived from the Tamulian tongues, the Sontal language having ú and i for the third personal, and these commutable in composition into the conjunct form of tá, precisely as in the Circassian tongue. From the Gondi tongue is there given another example of the commutation of ú to t, so that my exposition from the Gyárúng instance is placed beyond doubt, whilst some fresh and beautiful links are added to the chain of affinities, as to which see prior note.

in his eyes, though I have ventured to demur to his Arianising of the Tartars by too strained applications of that system.

I know not if Rosen at all explains the peculiarities of the pronouns in Circassian, but Latham does not; and it will therefore be felt as a truly interesting circumstance that the explanation just given, like that of the Ra suffix, have been fetched from Lhása and Litháng! The cultivated tongue of Tibet proper continues, it will be seen, to afford the clue to the labyrinth; and that it does so, is surely a strong presumptive proof, as well of its superior antiquity as of its superior completeness. So judging, I cannot moreover doubt that the Circassian preterite sign is the same with the Tibetan preterite sign (chen-tshar), though this be beside the mark of pronominal expositions-and to these I must confine myself, or I shall not know where to stop, so constantly do these Tartarian illustrations of the Caucasian tongue flow in upon me. I am unaware whether the Circassian language is distinguished, like the Gyárúng, by a very ample employment of those prefixes which, as more or less employed, characterise so many of the Mongolian tongues, and which are dropped in composition, like the Ra suffix. Thus, tarti, a cap, in Gyárúng, is compounded of ti the root, and tar* the prefix; but if we join a noun or pronoun to this word the prefix disappears, and "his cap," for example, is wárti, compounded of the wá above mentioned and the radical ti. In like manner taimek, a leaf, when compounded with shí, a tree, drops the tá prefix and becomes shaimek, as túpé, father, becomes ngapé, my father.t Rosen, should this paper fall under his eye, or

• Ta, the common form, becomes tar, differentially as tími, fire; tirmi, man, root mi, used in both senses. In tirmi, tarti, warti, we have the ra particle, which remains in its conjunct form as a medial, whilst the usual prefix ta disappears. The rá, too, would disappear in a compound of roots if not needed to differentials and mark the special sense of such roots, or one of them, or if the root commenced with other than a labial consonant, its prefix being servile.

✦ It has been queried whether the polysynthetic words of the American tongues quoad their principle of construction, as to which there is so much doubt, be not compiled from radical particles only. Judging by the method of forming ordinary compounds in Gyárúng and its allies, I should say, Yes, certainly they are to a great extent, though not exclusively, for the cumulative principle ill brooks control, revelling in reiterations and transpositions of root alike, and of

Latham perhaps, whose quick eye will not fail to catch it, will be able to tell whether the same peculiarity distinguishes the Circassian tongue. For myself I doubt not it will so prove, because the rule for nouns is but another phase of the rule for pronouns.

In the meantime, the striking grammatical analogies* I have pointed out stand in no need of further elucidation, and these analogies, together with the explanation from the Tibetan of the widely-used but heretofore unexplained Ra suffix, constitute in themselves, and as sustaining all those numerous identities of the primitive vocables which have been adverted to, something very like a demonstration of the Mongolidan affinities of the Caucasians, though I would be understood to speak with a due sense of the disqualifications inseparable from my secluded position and want of access to books. I subjoin Latham's sample of the construction of the Circassian language, with its equivalent in Gyárúng.

"I give to my father the horse."

64

In the house are two doors" is, in like manner, "house two doors" in the Circassian and Gyárúng tongues.

its servile adjuncts, though clearly, as to simple compounds, constantly observing the rules of contraction and of substitution noted in the text. In the Gyárúng sentence, Tizécazé papún, he summoned them to feast, the word for to feast shows the root repeated twice, and each time with a separate servile, though we have here only one verb, not two verbs; and in kalarlar, round, still no com. pound, we have the root repeated, but yet with a servile, though only one, being the prefix ka. In such cases that servile is usually omitted, as kaka, sky; pyepye, bird; chacha, hot.

Those analogies might now be largely extended did health and time permit. Take the following instances:-Tam-bus, father; imbas, my father, in Uraon. Sampa, father; ampa, my father, in Kiránti, Ku-kos, child; ing-kos, my child, U'raon. Tam, sam, ku, serviles, replaced by the pronouns; compare Malayan sam-piyan, san-diri, kan-diri, ka-manus, k'anak, &c.

+ Ra suffix subjoined for illustration though not in use with this person. See prior note.