These terminations are radical pronouns, and in the Tungusic dialect attach themselves to nouns as well as to verbs, taking in the former case the character of possessive, in the latter the character of predicative affixes. The Mongolic dialect, in which Castrén observed the same tendency, had advanced another step, for it made also formal distinction between possessive and predicative affixes. These are:

The differences between these two sets appear small, but are characteristic. The possessive affix of the first person singular, for instance, can never be p, because it is connected with the oblique base of the pronoun of the first person, mini, while the p of the predicative affix can only be explained by a reference to the nominative bi.

All this, however, is but a small beginning, particularly if we compare the profusion of grammatical stores which the Turkic languages display. These are next in order. With regard to their system of conjugation, the Turkic dialects can hardly be surpassed. Their verbs are like branches, breaking down beneath the heavy burden of fruit and blossom; and the excellence of the Finnic languages, richer in declension than the Turkic, consists, as far as the verb is concerned, rather in a diminution than increase of forms. Castrén says: "Progrediente in dissertatione apparuit affixa personalia in linguis Burjatica et Tungusica inchoata adhuc esse et quasi nascentia, in Turcicis vero jam formâ uti perfectiore magisque explanatâ, in Finnicis demum et Samojedicis linguis summum evolutionis gradum adepta esse."

The difference between the primary formations of Turanian and the secondary formations of Arian languages may be explained, if we consider that in je vivrai, i. e. ego (aham) vivere (gîv-as-ê, dat. neutr.) habeo (bhâv. ayâ. mi), we have a number of articulated forms,

resolved as it were again into simple matter, while in the Tungusic verb, grammatical form is produced for the first time by the mere connection of material elements.

EIGHTH SECTION.

Evidence of the common Origin of the Turanian Languages
summed up.

Ir after these considerations we look again at the problem of the affinity of the Turanian languages, and compare the evidence brought forward by Gyarmathi, Rask, Schott, and Castrén, with the amount which, from the nature of the case, we have a right to expect, most scholars, I think, will admit, that so far as it can be proved, proof of this affinity has been given. No doubt it may still be more fully confirmed, and many important questions remain for solution. But it may be regarded as no less proved than the affinity of the Indo-European languages was in the days of Sir W. Jones and Frederick Schlegel.

With regard to roots and words, in their primary and secondary meanings, Schott's "Essay on the Altaic Race," making every reasonable allowance for waste, is conclusive as to their natural affinity. Differences, such as exist in Turanian languages, between identical dialects, if spoken in different valleys, we must be prepared to find in cognate idioms, separated so far and so long — by centuries and by continents.

With regard to pronominal roots, Castrén has proved their identity, not only in character but in sound, with such accuracy that more on this point can scarcely be expected.

With regard to grammatical forms, we must consider that nearly the whole grammatical structure of the Turanian languages is built up from pronominal elements, which pervade not only the conjugation but the declension, nay, even the syntax of these dialects. As to the other grammatical elements, postpositions I mean principally, or similar particles, they also exhibit salient coincidences in some points, while their diversity on others does not mean more

than when we see in Italian an ablative formed by da (de a), and in French by de; or where, as in Wallachian, the genitive is formed by a, the accusative by pre (per), the ablative by déla, the dative taking no preposition at all: while further in the same Romanic idiom the article is put behind the substantive, reversing the order of its cognate dialects. Coincidences in these grammatical exponents will have to be mentioned when we point out their similarity with the case-terminations of the Dekhan dialects.

The syntactical character of the Turanian languages is also strongly marked, whether we look at their method of connecting roots and grammatical exponents into words, or words into sentences. In the first case all grammatical exponents must be added to the end of a base: bases tolerate no initial changes or additions. The grammatical terminations, though joined to roots, and this even euphonically, can with few exceptions be separated from the base. They are sometimes written separately, and admit intermediate elements, such as késnek and kes-em-nek. In the second case, as a general rule, the governed or determining always precedes the governing or determined word. Therefore prepositions governing a noun are impossible in Turanian languages. Conjunctions are scarce, the connection of sentences being marked by gerunds, or other verbal forms, with postpositions.

With regard to the phonetic character, the law of the "harmony of vowels" pervading these languages, and manifesting itself most strongly where artificial influences, such as writing, have least interfered, is a family feature not less strongly marked. It can only be compared with the triliteral character of the Semitic, or the peculiar accents and intonations of the so-called monosyllabic languages.*

* That these accents occur in languages more polysyllabic in their structure than either Greek or English, is shown by Hodgson and Robinson. The latter describes four accents in Gangetic and Lohitic dialects:

"These intonations, depending as they do only on a modified action of those parts of the larynx which most immediately affect the voice, are, in general, exceedingly difficult for an European practically to distinguish. On a careful examination, however, it will be found that these tones do not in reality exceed four, and that they are the same as those described by Chinese philologists.

"The first of these may be said to be pronounced naturally, as a middle tone, even and moderate, neither raised nor deepened by any peculiar effort.

Like these numerous accents, the harmony of vowels is such as can hardly be presented accurately in writing; nay, even in speaking it requires a practised ear to distinguish, and a throat still more practised to imitate it. This law exists in the Tungusic, Mongolic, Tataric, and Finnic classes, though it does not influence all their dialects with equal force. Traces of a certain vocalic equilibrium occur, however, also in other classes of the Turanian family, as may be seen from the examples quoted by Mr. Hodgson from the Gyarung dialect. (J. A. S. B. 1853, p. 30.).

With regard to the historical evidence, I need not repeat the leading characteristics common to these nations, so powerfully stated in your Lecture. But I shall conclude with an extract from Abulghasi's History of the Tatars, which has been discussed by Deguignes, Klaproth, Rémusat, Gabelentz, and Schott, and as a tradition is certainly curious, because it shows that even in later times, when Mongolic and Tataric had by mistake become the names of two races, differing in languages, religion, and manners, a feeling prevailed among themselves as to their common descent, which could hardly owe its origin to any preconceived ethnological opinion entertained by Abulghasi, the Khan of Khiva, the descendant of Chinghiskhan, and contemporary of Sanang-Setsen (1664). He relates that all the nations of Central and Northern Asia descended from one ancestor called Turk, who was the son of Japhet, who was the son of Noah. Among his descendants two brothers are mentioned, Mongol and Tatar. It seems probable that Turk, though at Constantinople it has now become a name of abuse, was in truth one of the oldest collective names of the Turanian race. Chinese authors recognised it in the 5th century B. C., when speaking of the Tukiuei, as a branch of the Hiung-nu. The etymology they give is fanciful; for Turk, however it may have. been explained afterwards, whether by the Turks themselves or by

"The second is a strong, rough, and vehement sound, produced by strongly exciting the action of the glottis in emitting it.

"The third tone is formed by raising the action of the glottis, as in forming the second tone, and then somewhat relaxing it, which, while it lengthens the sound, makes it end rather feebly.

"The fourth tone may be characterised as a short, thick, hasty sound, which seems to re-enter the throat, so as at length to be stopped in it." (See J. A. S. B., 1849, p. 192.)

Chinese writers, was originally a corruption of Tûra, Turvasa, Turushka, all names given by the Arians to equestrian Nomads and IndoScythian tribes north of the Himalaya. One of the sons of Feridun, we may further notice, was called Tur; and when the father divided his kingdom between the children, he gave Turán to Tur, Iran to Irij, and Rum and Kháwer to Silim. Irij is killed by his brothers; but the kings of Persia descend alternately from the three brothers, Menúchihr being an Iranian, Afrásiyáb a Turanian, Garshasp a Silimian. The names, therefore, Arian and Turanian, though now confined to scientific use, have yet a history of their own, which in its general bearing answers well with the technical objects for which they are at present employed.

Such is the case for the affinity of the Turanian languages. I have been here able to state the argument only in general: for matters of detail I must refer to Schott, Castrén, Gabelentz and Boehtlingk. To the objections raised by the last-named philologist I have paid particular attention; but although modifying some of the supposed characteristics of the Turanian languages, and recommending caution and more definite argumentation, they cannot be held to invalidate the conclusions arrived at in common by men like Rask, Gabelentz, Schott and Castrén.

If the principles here laid down are considered valid for establishing the relationship of languages, I am inclined to maintain that, similarly with these five classes, Finnic, Samoiedic, Tataric, Mongolic, and Tungusic, the Tamulie, Bhotîya, Taï, and Malay languages also belong to the same Turanian race.