viles, whereof the first is the emphatic ná above illustrated; and ár, vel rá, vel lá-rá, the so-called plural sign or signs, though in my judgment it is to mistake the true genius and character of these tongues to give to any of their particles, except with extreme reserve, the attributes of strict grammar (declensional marks), or a precise independent signification such as self for ráng in Tibetan. Ráng is a compound of the rá and ang particles. The phases of the latter are á, an, ang, and the reflective or egoistic sense, such as it is (it is most like that of the Sanscrit swa), attaches, not to the compound ráng, but to the simple ang. In Bódó and Gáro and Hayu áng stands for the first personal pronoun; in Limbu and many other allied tongues it is the first possessive, in the form of á. In Tagala and Malayu á and áku represent the first personal, and ang is an articular prefix of the same drift. The first personal is an-ka in Kiranti and a-za in Osetic, prefix in all these instances, in others even of the same tongues it is a suffix; but still, whether attached to pronouns, verbs, or nouns, and whether prefixed or postfixed or standing alone, as root or servile, it is apt to indicate a reflective character. This is the reason why it so constantly marks the possessive case, with or without a preposed particle; but if with one, usually the ná conjunct, which is only one phase, as ang-gé is another phase, of the repetition of itself; and this is also the reason why in so many of these tongues the áng suffix, when appended to verbs and their participles, designates the first person. Thus kazáng, I eat, kazángti, I who eat, I the eater, I eating, from the root zá, zó, in Gyárúng. Piré, give; pi-ráng or pirang-gé or pirang-né, give to me, in Limbú, from the root pi; davo, give, davóng, give to me, in Gyárúng, from the root va, vo. These forms are imperative. The indicative ones are similar, thus piré and dovo mean, you or he (quivis præter meipsum) gives; and piráng, dovong, I myself give, ang-né and ang-gé are equal, and are reiterations of the a, an, or ang particle. Com

* As ang is prefix or suffix, so is any other servile; for instance, the ká of anka, here cited; thus, k' mari, man in Georgian (mari in Suanic), and osurka, maid in Mingrelian (osuri in Lazic). See on to further note.

+ In Sontal, Uraon, Ho, and Hayu, the ang becomes ing, and eng with the very same emphatic reiteration, viz., eng gna and ing ga.

pare ang-gé to me, in Turki and Ouigúr; and máng-gé to me in Ouigúr, with their equivalent má-nán in Osetic. Piré and Piráng show very pointedly that the reflective virtue resides not in the rá particle but in the áng particle. This case also exemplifies their conjunction. Má-náng is the disjunct form; máng, the conjunct; and máng-gé is the same, only more emphatic; máng, to me, máng-gé, to myself; and máng-né and máng-ré are both equivalents and emphasisers merely. So mini is mine; and mining-gé, my own, in Mongol and Mantchú; the nang becoming níng euphonically to harmonise with the mi root. And, by the way, we may here, as in all the other derivatives, note the forthcomingness of the widely prevalent mi root, though obsolete as a nominative in these two tongues, just as it is in the analogous sense of man (ego = homo plur. exem.) in Burmese, wherein, however, we similarly gather it from its derivatives, woman and child, mimma* and sa mi.

I have illustrated the pronominal and verbal uses of the rá particle, as well as explained its relation to rang. Here are some exemplifications of its nominal and other uses. I fear I shall weary the reader, but he must remember that what is true of this particle is true of all the particles; and that, whereas a confined view of the character and functions of this grand element of these tongues has led to very erroneous notions as to their general affinity, so a complete conception of the nature of the particles is the best guide to a just perception of that affinity. For instance, Rosen has dwelt on the unique character of the Circassian pronouns arising in good part out of the operation of the rá particle, and I, following him, have announced with reasonable surprise the fact that the same peculiarities are attached to the Gyárúng pronouns, whereas, in very truth, whatever he or I noticed in this respect as to the pronouns is equally true as to the nouns, adverbs, &c., and that not merely in the languages of the

Compare Esthonian temma, supra, where suffix ma = emphatic na. All these tongues affect illiteration and consonantal as well as vocalic harmony to an extent quite perplexing, since each tongue has its fancies in this respect. Here má is a root.

Circassia and Gyárúng, but in every tongue from Caucasus to the Pacific. Here is the enumeration.

Ma-re, man, Suanic; ma-ri, man,* Georgian; ma-ro, man, Lepcha; mú-rú, man, Sunwár; m-rú, man, Mrú; ilé-ru, before, Turki; uz-ré, upon, Turki; herel-ri, man, Sontál.

Lan-ré, once, Tibetan; kyú-ré, river, Akúsh; thó-ré, tomorrow, Tibetan; wá-ran, rain, Osetic; mu-ran, river, Turki; mai-ran, arm, Mantchú; koöl-ron, child, Mongol; kho-rang, sky, Bódó; chák-reng, hand, Gáró; dí-rang, this, Serpa; dé-ring, to-day, Tibetan; ré-m-bú, man, Limbú; res-ga, where, Tibetan (samples of prefix); ús-rés, man, Gyárúng (sa added); rgu-re, nine, Manyak; ma-r, horse, spoken Chinese; ma-rhi, horse, Sokpa; gá-r, where, Tibetan; gá-rú, where, Tibetan; dé-r and dé-rú, there, Tibetan; ta-r-ti, cap, Gyárúng; ti-r-mi, man, Gyárúng; ok-ur, ox, Magyar; o-zu-r-ka, maid, Mingrelian (ka added, see note); o-sú-ri, maid, Lazic; u-er-ti, boy, Armenian; pu-r-ti, bird, Andi (ti, added, the rati suffix); do-r, stone, Osetic; teng-er, sea, Magyar; sha-r, ox, Mongol; khor, river, Avar; kú-er, hand, Anzúg; ka-r, hand, Tshari; ka-r, hand, Sokpo.

We thus see that the ra particle changes its vowel to the

* I here omit the ka prefix, with full warrant from usage:

See prior note on kmari and klún; ka suffix in ozurka is the same thing and similarly omissible, witness osuri. Here ó is the root, = ú, meaning man, and it also takes the k prefix. Sú is the sa particle harmonised in its vowel to the root. It is a diminutive, so that o-sa, u-sa, or u-a-sa is child, and kusa is equally child. We have kusa and a-sa in Limbú, and u-a-sa in Aver, ú-s in Osetic, ú-as in Wogul, ú-er in Armenian; sa in its capacity of diminutive means woman as well as child when added to any root for man, as ú or mi; and hence Osetic ú-sá womanmi-sa, Newári. Such and so concordant are all the elements. In Armenian uerti, child, erti vel rati being servile, it follows that the ú root for man may express juniors as well as adults, whilst the Gyárúng ús, man, and Osetic ús, woman, prove that the ú root expresses both sexes, meaning man-kind or the species man, and also that sa is not uniformly a diminutive but a synonym. This will be amply proved by and by, when the o-u-w and the sa, si, shi, roots for mankind are arrayed, and then it will be also seen that the name of the Osetic people is derived from two synonyms for man, and that, like tá-tá, or tshe-tshensh, it is = Allemanni. The Caucasian puzzle as to us, ush, ushi, u-as, u-as-sa, u-er, o-su, o-zu, is solved by this explanation, and if we add the Murmi bú root for man (supra), we have the clue to the Caucasian bo-zo, bo-shi, bit-shi, bi-shi, for all which I have numerous Mongolian equivalents, thus po-zo in Pasuko, pu-sa in Karen, bu-cha in Tekpa, bi-sha and bi-shi in Bódó.

utmost (rá, ré, rí, ró, rú), takes the ang or other additional particle (ti, ka, sa), occupies the initial (res-ga), medial (pú-r-ti), or final (ka-r) position, or even both (r gú-re), with reference to the root, and lastly, blends itself with that root, dropping its vowel (gár), or stands apart, retaining its vowel (gá-rú); and all this without change or even modification of the meaning of the word as derived from the root further than a certain emphasising can be so termed, as kho-rang, the sky; ka-r, the hand.

Such elements of speech and all the serviles are essentially alike, can with little propriety be designated by our grammar terms or alleged to be conjugational or declensional marks, except with extreme caution. The essence of a grammatical rule or part of speech is generalisation; the essence of the function of these particles is the very opposite of specialisation; and thus it is that unlimited change of place and change of form belong to the latter, whilst nothing of the sort does or can belong to the former.

Of the habit of applying our grammatical terms to the elements of these tongues in central Asia without any apparent perception of their true character,* as noted in the southeastern islands, I will give a sample from the Altaic group of languages.

The plurals of the Mantchú personal pronouns are thus stated and commented upon.

We.
Bé
Mousé

Ye.

Souwé

They.
Tését.

To this statement of the pronouns it is added that bé, sou wé, and tését constitute the ordinary series; that mousé is a sample of the dualistic form, and that it is regularly derived from mou, I, by the addition of the plural sign sé. Now it is quite true that the existence of a dual or rather of an inclusive * To prove this it suffices to advert to Vater's derivation of the Caucasian kar and kwer, hand, from xep, and Klaproth's of Waran rainm for and Máré

from c. I shall give numerous Tartar equivalents for all three, and thus prove their roots to be respectively ka, wa, and ma, the ra, ré, and ran being serviles, or rather phases of one servile.

plural is one of the characteristics of these tongues, and one that prevails very generally from the Pacific to Caucasus. But how it can be said that in the Mantchú tongue this inclusive plural is formed regularly from the singular mou by means of the plural sign sé, I cannot conceive, since a regular pluralising particle would be uniformly applied and wear one shape, whereas there is here in the three persons of the pronouns no vestige of such attributes in the sé particle. The ordinary “ "we" (bé) has no trace of this or other pluralising suffix; the ordinary "ye" (sou wé) has quite a different augment (wé); and, lastly, the third person shows the sé particle indeed, but with a foreign element or suffixed t (sét). Now surely a grammatical rule must have some identity of character, what it includes must be similar in form and application. But that in the Mantchú pronouns the plurals cannot be said to be regularly formed by the addition of sé, is self-apparent; and if we turn to any collated list of the pronouns of the Altaic tongues generally, we shall immediately perceive the same anomalies prevailing throughout this group of languages, and affecting both the form and the application of all the particles; the áng suffix, for instance, being at once a genitive and a dative sign in a single tongue (sanggé, of thee; manggé, to me, in Ouigúr), and also changing its form entirely in the same case (meaning, of me; sanggé, of thee) in that single tongue. Look again beyond the Altaic group and you will see the same anomalies. Everybody had noticed them in this or that instance, and I have on this account myself demurred to the use of the pronouns at all as a test of ethnic affinity. I am now aware that I was misled by the authority of great names, looking at these particles from a too grammatical point of view. We first make the particles grammatical, and then we declare them to be utterly anomalous; the facts being, that they are not strictly or uniformly grammatical, generally speaking, nor perhaps anywhere so, except as the result of

* This remarkable and arbitrary feature of a dual and two plurals I have already detected in the Kuswar, Hayu, and Kiranti tongues of the Himalaya, and in the Ho, Sontál, and Uraon tongues of Tamulian India. I need hardly add that the same peculiarity belongs to the Tagalan and Alforian languages, as well as the Altaic.