■i^-'S* 
 
 *m^ 
 
 •,'flr-'» 
 
 ^♦•^ 
 
 '%. 
 
 »l 
 
 
 i: 
 
J^HE -^OPP LlBI^I^. 
 
 COLLECTED BY FEANZ BOK 
 
 Professor of Comparative Philology in 
 TJniversity of Berlin. 
 
 Furchased by Cornell University, IS 
 
CORNELL UNIVERSITY LIBRARY 
 
 924 087 972 182 
 
 OLIN LIBRARY ^ CIRCULATION 
 DATE DUE 
 
 tWi-oi 
 
 
 #^n 
 
 
 
 /4 *S^ 
 
 
 
 TiTrc^H 
 
 ra^r 
 
 
 
 Ifla 
 
 IB^ 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 - 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 GAYLORD 
 
 
 
 PRINTED IN U.S.A 
 
Cornell University 
 Library 
 
 The original of this book is in 
 the Cornell University Library. 
 
 There are no known copyright restrictions in 
 the United States on the use of the text. 
 
 http://www.archive.org/details/cu31924087972182 
 
LETTEE 
 
 TO CHEVALIER BUNSEN, 
 
 ^THE CLASSIFICATION OF THE TURANIAN 
 ■\ LANGUAGES. 
 
 BY MAX MiJLLER, M.A., 
 
 TATLOEIAN PKOFBSSOK Or MODEEN EUKOPEAN LANGTJAOES, 
 OXPOKD. 
 
 
 
 ^^<r^ila^^ \%S^^ 
 
..^^m^_ 
 
 
 ;% 
 
 mA 
 
 
 '^y—,;^-^---' 
 
 
 fe^^ r,= .v- 
 
 
 
LETTER 
 
 CLASSIFICATION OF THE TURANIAN LANGUAGES. 
 
 INTRODUCTORY REMARKS. 
 
 " Which of the languages of Hindustan belong to the Avian stock, 
 and which to the Indian family of languages prevalent before the 
 Iranian immigrations, is a disputed point, which we hope will be 
 brought nearer to a settlement by Dr. Mailer's lecture this day." 
 These were the words in which you did me the honour of 
 introducing me to the British Association at Oxford, in 1847, 
 when laying before the Meeting of that Society the results of 
 your researches into the origin and affiliation of the languages 
 of mankind. But the hope you then expressed was not 
 realized ; and I fear that, if you have looked again over my 
 paper as it stands printed in the Transactions of the British 
 Association, it may have disappointed you. The great question, 
 the question of real historical interest, the connexion, namely, 
 of the southern languages of India with any other established 
 family of speech, was in my essay hardly touched on. I had 
 to confine myself there to a vindication of the Arian descent of 
 the northern languages of India — a task which to you may 
 
 B 2 
 
have appeared almost useless, and is indeed of small import- 
 ance if compared with that other problem, the origin of the 
 southern dialects. 
 
 I, therefore, gladly avail myself of your permission ; and, 
 in the space which you kindly allow me in your new work 
 on the philosophy of language, I shall endeavour to state my 
 view of this much controverted question, the origin of the 
 southern dialects of India. It is a question intimately con- 
 nected with some of the greatest problems of comparative phi- 
 lology, and its solution must depend not only on facts, but on 
 the establishment of principles which may or may not be appli- 
 cable to a classificatory study of languages. I fear, how- 
 ever, that at present I shall hardly be able to do justice to 
 a subject so difficult and comprehensive. During the last three 
 years my linguistic studies have necessarily taken a very dif- 
 ferent course, and I have directed my chief attention, away 
 from India and the Southern Peninsula of Asia, to that Western 
 Peninsula of the great Asiatic continent where all the languages, 
 religions, and arts of the old world seem to have been stored 
 up for the present, and formed into what we now call the 
 modern world of Europe. I must, therefore, crave your indul- 
 gence, and that of your readers, for this somewhat hurried com- 
 position ; materials collected several years ago and never 
 intended for publication : and, though I hope I shall be able to 
 defend what I have stated, either as fact or as theory, in the 
 course of this letter, I trust that others, more competent than 
 myself, will take up and will solve a problem that I have ven- 
 tured to state at your instance, and which, up to a certain point, 
 I hope to elucidate. 
 
 That it was not quite unnecessary to establish bej'ond rea- 
 sonable doubt the Sanskrit origin of the Bengali and the other 
 dialects of Northern India, is shown by a remark which a 
 writer of considerable authority on ethnological subjects has 
 since made on that essay. In a work lately published on tlie 
 
Varieties of Man, the author says : " It is not likely any better 
 case will be made out for this (the Sanskrit origin of the 
 Bengdli and its cognate dialects) than the one contained in a 
 Dissertation of Dr. Max Miiller. Yet it is so unsatisfactory 
 that it almost proves the question the other way." Now, it is 
 very true that these languages do contain many features which 
 are apt to deceive us about their real origin and character. There 
 are not only many words of Turanian and Semitic origin which, 
 through channels opened by Mohammedan and Mongolian con- 
 quests, have found their way into these dialects, but there is 
 also a whole layer of aboriginal words, words now belonging to 
 the south of India, but yet of every-day use, in the spoken dia- 
 lects of the north. Some of them have found their way even into 
 the dictionary of the Sanskrit.* Besides, and this is a more im- 
 portant feature, the very grammar of the northern dialects has 
 been infected by the same influences to an extent to which we find 
 analogies only in some of the modern languages of Europe. It is 
 very true that the grammatical system of a language repels foreign 
 intrusions, as every Kving organism repels mechanical influences. 
 But still the grammar of a language may, to a certain extent, ac- 
 commodate itself to the genius of a foreign tongue with which it 
 is brought into constant contact. It may imitate, though it does 
 not adopt or borrow. Instances of this occur in the English of the 
 Norman period f ; and in medieval chronicles we find Latin ter- 
 minations occasionally appended to German words. In Albanian 
 and Bulgarian the peculiarity of placing the article at the end, 
 and not at the beginning of a substantive, was probably bor- 
 rowed from the Wallachian, in which domnul, i. e. dominus ills, 
 
 * See several articles by Dr. Stevenson in the Journals of the Asiatic Societies 
 of Madras and Bombay. 
 
 f See Thommerel, Recherohes sur la Fusion du Franco-Normand et de 
 I'Anglo-Saxon, 1841. In phrases such as " zour honorabile lettres contenand" or 
 " brekawrf the trewis" (Let. de G. Douglas a Richard II., 1385), we actually find 
 a French grammatical termination, though its introduction may have been facili- 
 tated by the similarity of the Anglo-Saxon termination of the gerunds in ende. 
 
 B 3 
 
had become fixed instead of il domnu, i. e. ille dominus. The 
 Latin of Ennius also offers a case in point, and even Cicero uses 
 Greek terminations, not only after Greek, but also after Latin 
 words* W. von Humboldt, in his posthumous work " On the 
 Diversity of the Construction of Human Language, and its 
 Influence on the Intellectual Development of Mankind," speaks 
 of the same thing, where he determines the influence which the 
 sacred language of the Buddhists — the Pali — exercised on the 
 spoken Burmese (p. 380,). But we have, perhaps, the most 
 striking instance in Persian, which, in such forms as "gul-i- 
 keniz," the rose of the maid, " dil-i-men," or " dilem," my heart, 
 has adopted, no doubt after Semitic models, a syntactical prin- 
 ciple not only at variance with, but diametrically opposed to, 
 all Indo-European grammar. On this more hereafter. 
 
 Influences of the same kind are traceable in the northern 
 languages of India; and to a superficial observer they are 
 likely to prove dangerous, and lead to false conclusions or un- 
 founded scepticism. To give an instance: It is a charac- 
 teristic feature of the southern languages of India, that they 
 distinguish the plural from the singular by adding to the 
 
 * The following Greek terminations occur in Latin writers : — 
 First Declension : Nom. e Gen. es Ace. en Abl. e 
 
 es 
 
 .. - 
 
 - „ 
 
 en 
 
 e 
 
 
 Second Declension . Nom. 6s 
 
 Gen. (u) 
 
 Ace. on 
 
 
 
 on 
 
 „ 
 
 — 
 
 „ on 
 
 
 
 us 
 
 „ 
 
 (u) 
 
 „ un 
 
 
 
 OS 
 
 „ 
 
 6 
 
 „ on, 6 
 
 
 
 Third Declension : Nom. — 
 
 Gen 
 
 OS Ace. S 
 
 Dat 
 
 1 
 
 eus 
 
 „ 
 
 eos 
 
 „ eS 
 
 )) 
 
 ei, i 
 
 es 
 
 I, 
 
 eos, us 
 
 „ ea(e?) 
 
 J, 
 
 
 is 
 
 )» 
 
 ios, eos 
 
 , in 
 
 >» 
 
 
 
 ys, y 
 
 „ 
 
 yos 
 
 . yn 
 
 >J 
 
 
 
 
 
 »> 
 
 us 
 
 , 6 
 
 „ 
 
 6 
 
 Plur. Nom. (5s, is 
 
 „ 
 
 on 
 
 , as, is 
 
 )T 
 
 si, sin 
 
 Neutr. e 
 
 „ 
 
 — 
 
 , — 
 
 
 
 See Schneider's Latin Grammar, Berlin, 
 
 1819, vol 
 
 iii. 
 
 
 
noun a suffix expressive of plurality. In order to form the 
 cases of the plural, they affix afterwards the same termina- 
 tions which form the different cases of the singular. This is 
 a grammatical expedient foreign to the Arian languages, even 
 in their secondary stages, though, in itself, it is by no means 
 incompatible with any of the leading features of Arian grammar. 
 In Asamese, " manuh " is man, and without an affix to limit 
 its signification it may be used either for the singular or plural. 
 It may mean man, a man, the man ; men, or the men. 
 
 The Genitive is manuh-or ; 
 Dative, manuh-oloi; 
 Accusative, manuh-ok ; 
 Locative, manuh-ot ; 
 Ablative, manuh-e. 
 
 If we want to express the plural distinctly, we must add bilak, 
 hont, or bur, particles expressive of plurality ; and by affixing 
 the same terminations as in the singular, we get 
 
 Nominative, manuh-bilak ; 
 Genitive, manuh-bilak -or; 
 Dative, manuh-bilak-oloi ; 
 Accusative, manuh bilak- ok ; 
 Locative, manuh- bilak-ot ; 
 Ablative, manuh-bilak-e.* 
 
 We can easily imagine how people speaking the modern 
 Sanskrit dialects, in which the old terminations by which the 
 plural was distinguished from the singular had been worn off 
 almost entirely, should, when again feeling a want to express 
 the idea of plurality more distinctly, have fixed upon a gram- 
 matical expedient which, from their daily intercourse with 
 their aboriginal neighbours, had long been familiar to their 
 
 * See N. Brown's Grammatical Notices of the Asamese Language : Sibsagor, 
 1848. 
 
ear and to their minds. The words which they used as the 
 exponents of plurality were of course taken from the resources 
 of their own language ; but the idea of using such words for 
 such a purpose seems to have been suggested by a xoreign 
 example. 
 
 It was necessary, therefore, to state the case fully, and to 
 prove, once for all, that the Bengali, the Asamese, the language 
 of the Odra, the Hindi and Hindustani, the Mahratti, the lan- 
 guage of Konkana, the Guzerati and Sindhi, the Khasiya or Par- 
 batiya, and the language of Kashmir, are all of Arian descent; 
 that the blood which circulates in their grammar, is Arian blood. 
 If I have succeeded in proving this (and if proved for the Ben- 
 gali, it is proved for all the rest), I consider it established, at 
 the same time, that the other languages of India, spoken princi- 
 pally south of the Krishnd, are of different origin. But beyond 
 this I did not venture to go. My conviction was then, and is 
 now still more strongly, that these souiliem dialects belong to 
 the Turanian family of languages ; that in their dictionary, how- 
 ever, as well as in their grammar, they are largely indebted to 
 their Arian neighbours. But, although I was satisfied myself 
 on this point, I felt at the same time that it involved questions 
 of so great importance that the subject should not be taken up 
 lightly. Nay, I was afraid my advocacy might prejudice the 
 question rather unfavourably, and I thought it ought to be left 
 to persons better qualified than myself to solve this linguistic 
 and ethnological problem. 
 
 Even now, in answer to your kind inquiries, I should rather 
 have adopted the negative method of arguing ; I mean, I should 
 rather have exhausted possibilities, and proved that these same 
 languages cannot be referred to any other race from which, as 
 far as history and geography go, they might possibly have 
 sprung. I might have endeavoured to show they are neither 
 Semitic, nor Chinese, nor Indo-Chinese, nor Malay, nor idioms 
 transplanted from the east coast of Africa. The characteristic 
 
features of all these languages, with the exception, perhaps, of 
 the last, are sufficiently well known to make it possible to prove 
 their absence in the languages of the Dekhan. However, as 
 you wish it, I shall lay my case before you in a more positive 
 form, leaving it to you to judge whether, even in its imperfect 
 state, it deserves the consideration which you were kind enough 
 to accord to it. 
 
10 
 
 FIRST CHAPTER. 
 
 First Section. 
 History of Turanian Philology. 
 
 It is necessary for our purpose, to begin with a general statement 
 on the Turanian family of speech, and to give a short sketch of 
 the progress of Turanian philology. It is a branch of study in- 
 volving problems of the highest importance for the early history 
 of mankind, and which no doubt would have found greater favour in 
 the eyes of comparative philologists, if the number of languages 
 belonging to this family had not been so large as to make an accurate 
 and philological study of the whole stock an impossibility. The 
 maxim, not to write about a language if one cannot write in it, is 
 certainly a most salutary one ; but it must be given up in so com- 
 prehensive a subject as that of Turanian speech in its endless 
 ramifications. In all classificatory sciences the same allowance is 
 made ; and if a comparative anatomist is able to arrange by general 
 characteristics animals of which he has seen but slight sketches, and of 
 which he hardly remembers or can pronounce the names, perhaps it 
 may be possible also to classify the Turanian languages without 
 possessing so familiar a knowledge of them as is required for more 
 special or practical purposes. 
 
 1. GYARMATHI. 
 
 The connexion and family-resemblance of some of the widely 
 separated branches of the Turanian stock, had been discovered and 
 established at a time when the name of the Arian or Indo-European 
 family was still unknown. The close relationship between Hungarian, 
 Finnic, Lapponic, and Esthonic was fully proved by Gyarmathi * in 
 1799; and he quotes one work, published by Sajnovits in 1770 
 
 * Affinltas Linguae Hungaricso cum Linguis Fennicse Originis grammatice de- 
 monstrata. Auctore Samuele Gyarmathi, M- D. : Gottinga;, 1799. 
 
11 
 
 (" Demonstratlo Idioma Hungaricum et Lapponicum idem esse"), 
 and another, published by Hager in 1793, as books of authority in 
 which this point had been established before. If we consider that 
 Gyarmathi's work was written before even the foundation of the 
 science of comparative philology was ' laid, he deserves to occupy 
 a very high rank among the founders of this science. His com- 
 parisons are not mere comparisons of words. In order to establish 
 the common origin of his own language and those of Finland and 
 Lapland, he derives his arguments from their similarity in derivative 
 suffixes, the system of declension and conjugation, the pronouns and 
 their various employments, the postpositions and adverbs, the 
 syntactical rules ; and in the last instance only, as he says, from the 
 " similitudo vocabulorum multorum, quod quidem momentum mihi 
 semper ultimum in istiusmodi disquisitionibus esse solet." Indeed, 
 his parallel columns of grammatical forms from Hungarian, Finnic, 
 Esthonic, and Lapponic can leave no reasonable doubt as to the 
 original identity of these idioms. He rejects, however, distinctly the 
 idea of a similar connexion between these languages and Turkish. 
 The number of words common to both, as collected by Gyarmathi, is 
 considerable ; but, as he could not discover any similarity in their 
 grammatical system, he repudiated the idea of a Finno-Tataric 
 family. A contrary opinion was expressed at the same time by 
 Kollar, who maintained that Turkish and Hungarian agreed in 
 the leading features of their grammar, but denied the similarity 
 of their vocables. It should be mentioned at once that the 
 principal argument which Gyarmathi brings forward against the 
 grammatical aifinity of Hungarian and Turkish, is derived from the 
 pronominal elements, which, he says, differ so much as to exclude for 
 ever the possibility of a common origin. We shall see, feowever 
 that exactly in the pronominal elements the most striking coincidences 
 have since been established. 
 
 2. KLAPROTH, REMUSAT, ARNDT. 
 The first step in advance after Gyarmathi was made by Klaproth*, 
 who proved that the languages of the Caucasus, with the exception 
 
 * Klaproth, Reise in den Kaukasus, 1814. Asia Polyglotta, 1823, p. 133. 
 
12 
 
 of the Ossetic, have a great similarity with the Samoiedie or North 
 Asiatic dialects ; while Ilemusatj though in a different way, con- 
 tributed toward the solution of the same problem by bis " Recherches 
 sur les Langues Tartares " (1820).* 
 
 Eemusat denied the affinity of the Turkish, Mongol, and Mandshu 
 languages. He says ("Eecherches," p. 138.) : " La ressemblance de 
 quelques expressions Turkes, Mongoles, et Mandshoues entre elles ne 
 doit pas faire penser qu'il existe entre les trois langues une analogic 
 essentielle et fondamentale. II y a entre elles plus de differences 
 qu'il n'y en a entre le Russe, I'ltalien, et I'Allemand." This, as is 
 well known now, might be admitted without any prejudice to the 
 question at issue. 
 
 Arndtf, in 1819, tried to prove that the Bask, in the western- 
 most corner of Europe, belonged to the same family with the Finnic 
 and Samoiedie ; nay, that Celtic also clung with some of its roots to 
 the same ancient stratum of speech. 
 
 3. RASK 
 
 The first, however, to trace with a bold hand the broad outlines of 
 Turanian, or, as he called it, Scythian philology, was Rask.J He 
 proved that Finnic had once been spoken in the northern extremities 
 of Europe, and that allied languages extended like a girdle over the 
 north of Asia, Europe, and America. In his inquiries into the origin 
 of the Old Norse, he endeavoured to link the idioms of Asia and 
 America together by means of the Gronland language, which, he 
 maintains, is a scion of the Scythian or Turanian stock, spreading 
 its branches over the north of America, and thus indicatino' the 
 ante-diluvian bridge between the continents of Europe and America. 
 According to Rask, therefore, the Scythian would form a layer of 
 language extending in Asia from the White Sea to the valleys of 
 
 * Abel-Remusat, Recherches sur les Langues Tartares, 1820. 
 
 f Ch. G. von Arndt, tJber den Ursprung der Europaischen Sprachen, pub- 
 lished 1817, and again 1827 ; but written about 1800, during the Russian period 
 of comparatiYe philology. 
 
 J R. K. Rask, Ueber die Thrakische Spraohelasse, 1818 ; R. Rask, Ueber das 
 Alter und die Echtheit der Zend-Sprachc, deutsch von H. von der Hagen 1826. 
 
13 
 
 Caucasus, in America from Gronland southward, and in Europe 
 (as Eask accepts Arndt's views) from Finland as far as Britain, 
 Gaul, and Spain. This original substratum was broken up and over- 
 whelmed first by Celtic inroads, secondly by Gothic, and thirdly by 
 Slavonic immigrations ; so that its traces appear like the peaks of 
 mountains and promontories out of a general inundation. Only on 
 the north of Asia and its central plains, probably the original hive 
 of the Scythic stock, has the race maintained itself in compact 
 masses, and sent forth even in historical times those swarms of 
 soldiers who made the walls of every capital in the Arian world 
 tremble before them. Rask maintains distinctly the affinity of the 
 Finnic and Tataric idioms, and he denies that the coincidences 
 between the two are simply of a lexicographic character. Again, 
 the three races of Tatars, Mongols, and Tungusians, whom even 
 Klaproth, after admitting a connexion between the languages of the 
 Caucasus and Siberia, considered as distinct, are traced back by Eask 
 to one common type of language and grammar. In maintaining the 
 relationship of these and the Finnic races, great stress is laid by him 
 on what were then considered mixed races of Tatar and Finnic 
 descent, — the Woguls, Wotiaks, and Tsheremissians. Eask denies 
 their mixed character ; because, he says, these tribes are peculiarly 
 exclusive in their marriages, and hardly allow members of different 
 tribes to reside among them. Their languages should, therefore 
 (to give Eask's conclusion), be considered, not mixed dialects, but 
 intermediate links in one great chain of speech. 
 
 Eask proposed the following division of the Scythian race : 
 
 1. North Asiatic. 3. Tatar. 
 
 2. North American. 4. Mongol and Tungusian. 
 
 4. SCHOTT, CASTREN. 
 
 Unfortunately, Eask did not live to fill in the grand outlines of 
 this ethnological cartoon. But, as, for his more minute researches 
 into the grammatical growth of the Teutonic languages, he found a 
 worthy successor in Grimm, his attempts to explore the large area of 
 the Scythian world were ably continued by Schott and by Castren. 
 In Germany, Schott's articles kept alive an interest in these re- 
 
 C 
 
14 
 
 searches. In his essay on the Tataric languages (1836) he stated 
 the problem boldly, and in his work on the Altaic or Finno- 
 Tataric race (1849) he has collected all the evidence that could 
 be brought to bear on its solution. But a jiew era in the history of 
 Turanian philology begins with one who, though in delicate health, 
 left his study, travelled for years alone in his sledge through the 
 snowy deserts of Siberia, coasted along the borders of the Polar Sea, 
 lived for whole winters in caves of ice or in the smoky huts of 
 greasy Samoieds, then braved the sand-clouds of Mongolia, passed 
 the Baikal, and returned from the frontiers of China to his duties as 
 Professor at Helsingfors — to die, after he had given to the world 
 but a few specimens of his treasures. This heroic grammarian was 
 Alexander Castren.* The general results at which he arrived, though 
 based on fuller materials and more accurate research, tend on the 
 whole to confirm Rask's views. 
 
 Castren establishes five divisions of the Turanian family, in place 
 of the four given by Eask. Besides, as Castren leaves the North 
 American dialects altogether out of consideration, his researches have 
 really added two new distinctions, the North Asiatic and the Mongol 
 class having each been split by him into two. Thus we have, ac- 
 cording to Castren, the following classes : — 
 
 2. SamoTedic }^°''*'' ^''^*^<= according to Rask. 
 
 3. Turkic Tatar according to Eask. 
 
 5. Tungusic J ^ongol-Tungusic according to Rask. 
 
 In the subdivision also differences occur. The Tshudic class, 
 
 * Castren, Elementa Grammatices Syrjaense. Helsingforsise, 1844. 
 Elementa Grammatices TscheremissEe. Kuopio, 1845. 
 Vom Einfluss des Accents in der Lapplandischen Sprache. 
 Petersburg, 1845. 
 „ Versuch einer Ostjakischen SpracUehre. Petersbm-g, 1849. 
 
 De Affixis Personalibus Linguarum Altaicarum. Helsino-forsiiE 
 
 1850. " 
 
 Reiseerinnerungen aus den Jahren, 1838 — 1844 Petersbure 
 1853. ' ^' 
 
15 
 
 which is the name given by Eask to the Finnic, had been divided by 
 him into I. the Finnic ; II. the Ugric ; III. the Byarmic stock : 
 
 I. The Finnic stock, according to Mask, has Jive branches. 
 
 a. Tsheremissian, d. Lapponian, 
 
 b. Mordvinian, e. Esthonian. 
 
 c. Suomian {i. e. Finnish), 
 
 11. The Ugric, three. 
 a. Hungarian, b. Vogulian, c. Ostiakian. 
 
 III. The Byarmic, three, 
 a. Permian, b. Syrianian, c. Votiakian. 
 
 To this Castren demurs. He insists on separating I. a, and I. b, 
 the Tsheremissian and Mordvinian, and considers that the two (to 
 which he formerly added the Tshuvashian) constitute a new branch. 
 According to Castren, therefore, we get the following stemma of the 
 Finnic stock : 
 
 1. Finnic. 
 I. Ugric. II. Bulgaric. III. Permic. 
 
 a, Hungarian, «. Tsheremissian, 1. Permian, 
 
 b. Vogulian, b. Mordvinian. 2. Syrianian, 
 
 c. Ugro-Ostiakian. 
 
 3. Wotiakian, 
 
 rV. Tshudic. 
 
 1. Lapponian, 
 
 2. Suomian. 
 
 3. Esthonian. 
 
 The second, or Samoiedic class, is divided by Castren into a 
 Northern and an Eastern stock : 
 
 2. Samoibdic. 
 I. The Northern comprises : II. The Eastern comprises : 
 
 a. Yurazian, 
 
 b. Tawgian, 
 
 c. Yeniseian. 
 
 a. Ostiako-Samoiedian, 
 
 b. Kamassian. 
 
 The Turkic or Tataric class, to which Gastrin has devoted less 
 
 c 2 
 
16 
 
 attention, is given here after Beresin. 
 each with a number of branches : 
 
 He establishes three stocks, 
 
 I. Tshagataic 
 (South-East). 
 
 a. Uigurian, 
 
 b. Komanian, 
 
 c. Tshagataian, 
 
 d. Usbekian, 
 
 e. Turkomanian, 
 
 f. Kasanian. 
 
 3. Tataeic. 
 
 II. Tataric 
 (North). 
 
 a. Kirgisian, 
 
 b. Bashkirian, 
 
 c. Nogaian, 
 
 d. Kumian, 
 
 e. Karatshaian, 
 
 f. Karakalpakian, 
 
 g. Meshtsheryakian, 
 h. Siberian (Yakutiaa 
 
 on the Lena). 
 
 ni. Turkish 
 (West). 
 
 a. Derbendian, 
 
 b. Aderbidshanian, 
 
 c. Krimmian, 
 
 d. Anatolian (Asia 
 
 Minor), 
 
 e. Eumelian (Con- 
 
 stantinople). 
 
 The Mongolic class has likewise been divided into three stocks. 
 Castren in his travels came into special contact with the Mon- 
 gols about the Baikal, where he studied the language of the 
 Buriates : 
 
 4. Mongolic. 
 
 I. Eastern 
 Mongols, 
 
 a. Sharra-Mongols, 
 
 b. Khalkhas, 
 
 c. Sharaigol (Tibet). 
 
 a. - 
 
 III. Saikal- 
 Mongols. 
 
 a. Buriates. 
 
 II. Western 
 Mongols (Olot). 
 
 " Kalmiiks, 
 
 Choshot, 
 
 Dsungar, 
 
 Torgod, and 
 . Durbet, 
 
 b. Aimaks (North of Persia), 
 
 c. Tokpas (North-East of Tibet), 
 
 The fifth class, the Tungusic, is principally represented by the 
 Mandshu. This language received its name when it became of poli- 
 tical and literary importance, after the Tungusian conquest of China, 
 
17 
 
 in the 17th century. Tungusian dialects are spoken by the Tsha- 
 pogires and Orotongs in the west, and the Lamutes in the east, of 
 Siberia. Castren studied the dialect of Nyertshinsk. 
 , Thus we have : 
 
 5. TUNGDSIC. 
 
 I. Western. II. Eastern. 
 
 a. Tshapogires, a. Lamutes, 
 
 b. Orotongs, b. Mandshu (in China). 
 
 c. Nyertshintk dialect (Castren). 
 
 Castren, in his dissertation " De Affixis Personalibus Linguarum 
 Altaicarum" (1850), after tracing minutely one of the most cha- 
 racteristic features of Turanian grammar through all the branches of 
 what he calls the Altaic (i. e. Turanian) race, concludes with the 
 following remarks : " What has been brought forward about the 
 origin, the formation, the sound, and the whole character of these 
 personal aflSxes, seems to prove that all the Altaic dialects are more 
 or less related to one another. Some of them are certainly widely 
 distant ; as, for instance, the dialects of the Finnic nations in the west, 
 and of the Mongolic and Tungusic tribes in the east. But their 
 difference is not greater than could easily have originated in the 
 course of a thousand years, and these must have elapsed since the 
 separation of these nations took place. During the same time almost 
 all the Altaic tribes came in contact with foreign nations, and received 
 from them the seeds of their present civilisation. New ideas created 
 new words and new forms — nay, a new principle — in the evolution 
 of these languages. Many things were adopted, many things framed 
 after the type of other tongues. It is the office of comparative philo- 
 logy to find out in every language what owes its origin to a modern 
 evolution. And only after this has been done, will a disquisition 
 on the affinity of languages become safe and profitable. I am fully per- 
 suaded that an intercomparison of the Altaic languages would as yet 
 be premature ; and I have, therefore, in my dissertation attended prin- 
 cipally to the single languages, and only mentioned coincidences 
 in the formation of the personal affixes incidentally. Perhaps it 
 will be my lot at another time to demonstrate the affinity of the 
 Altaic languages in a more convincing manner." 
 
 c 3 
 
18 
 
 We see, in these words, Castren's conviction on the affinity of all 
 the Altaic languages expressed clearly, though with caution and 
 modesty. Another passage in the same dissertation bears on 
 this point. He says : " After studying for a long number of years 
 Finnic, Samoiedic, Turkic, Mongolic, and Tungusic dialects, it 
 seems, as far as I can see from my own researches, that we must not 
 look in them for so close a relationship as that by which the Indo- 
 Germanic languages are held together like so many branches of one 
 and the same stock. But that there exists between them both a 
 formal and a material congruence, particularly between Finnic, 
 Samoiedic, and Turkic, I maintain still, as I stated it some time ago. 
 Whether this congruence is so great as to enable us to trace all 
 these dialects back to one common source, is a question which 
 the next generation may hope to answer. To us it seems that 
 these idioms branch off together, and dissolve themselves into dif- 
 ferent stems or families, but that they still belong to one class or 
 race. Certain it is, that they are more related to one another than to 
 any of the Indo-European languages.'' 
 
 5. VON DER GABELENTZ. 
 Von der G-abelentz has treated the same question in his gramma- 
 tical outlines, and in several articles devoted to Turanian philology 
 published in the " Zeitschrift fiir die Kunde des Morgenlands.'' By a 
 previous study of the Arian languages, Von der G-abelentz was ad- 
 mirably prepared for this larger sphere of linguistic research, and his 
 works give full evidence of his great power of observation and a 
 most comprehensive grasp in arranging. According to his opinion 
 also, the Turanian languages — Tataric, Mongolic, Tungusic, and 
 Finnic— constitute one family. This at least seems to be his last 
 conviction, at which he arrived after a continued study of these 
 idioms ; and it is the more valuable, because in iis earlier works 
 — for instance, in his Mandshu grammar (1832) — he entertained 
 a different view : admitting the striking resemblance between the 
 grammatical and phonetic systems of the Mandshu, Tatai-ic, and 
 Mongolic dialects, but not allowing their affinity. 
 
19 
 
 6. BOEHTLINGK. 
 
 If we may quote Von der Gabelentz as a high authority in favour 
 of the common origin of the Turanian languages, there is another 
 scholar, of no less weight, particularly where questions of gramma- 
 tical detail are concerned, who has lately thrown considerable doubt 
 on this subject ; I mean Professor Boehtlingk, in his work " Uber die 
 Sprache der Jakuten " (1851). It is a work of the most massive 
 industry, and it bids fair to raise the science of Tataric grammar to 
 the level of Greek and Sanskrit philology. It is particularly important 
 for the more special study of the Tataric languages, because, accord- 
 ing to Professor Boehtlingk, the Yakute dialect became separated at 
 a very early time from the still undivided Turko-Tataric speech, and 
 therefore exhibits a most primitive specimen of what he proposes to 
 call, instead of Turkic or Tataric, the Yakuto-Turkic class. An 
 admixture of Mongolian words in Yakutian, and an adoption even of 
 Mongolian grammatical terminations, is explained by a long-continued 
 historical contact between Yakutic and Buriatic tribes. But this 
 work throws also much light on questions of a more general bearing. 
 The Introduction particularly contains most valuable remarks on 
 the true principles which ought to guide us in the classification of 
 languages. Professor Boehtlingk afterwards enters more particularly 
 into the question of the affinity between the Finnic, Samoiedic, and 
 Tataric classes of the Turanian race. On this point he has been 
 engaged in a long controversy with Professor Schott of Berlin, a 
 controversy carried on with an animosity something more than Attic. 
 Professor Boehtlingk stands up for the principle that it is dangerous 
 to write on languages of which we do not possess the most accurate 
 knowledge. Professor Schott, on the contrary, thinks that a limited 
 knowledge is suflacient for settling the general question as to the 
 common origin of languages. No doubt Professor Boehtlingk has 
 proved that several words and forms which Professor Schott supposed 
 to be mutually related are different in origin, and that, with his 
 method, he cannot guard altogether against similar mistakes. In so 
 comprehensive a comparison of the Turanian idioms as Professor 
 Schott undertakes, errors must occur which, in the present state of 
 comparative philology, an Arian scholar can easily avoid in his more 
 limited and more matured researches. 
 
 C 4 
 
20 
 
 No one who has studied in the school of Bopp and Pott would think 
 of comparing avdXoyoc, with German "ahnlich," Persian "behter" with 
 English "better," German "ei," egg, with English " eye ;" or even 
 Greek log, poison, with Ioq, arrow ; Greek vtw and vt'w, Latin " nare" 
 and " nere." In a comparison of Turanian dialects, erroneous com- 
 parisons like these would be more difficult to avoid. Nor would it 
 be possible always, in the present state of Turanian philology, to 
 discover that words so different as " meme " and " semetipsissimus," 
 "larme"and "tear," "redemption" and "ran^on," "age" and "eternity," 
 "cousin" and " sister," were originally identical. There are certainly 
 some very strong points which Professor Boehtlingk has established 
 against Professor Schott; as, for instance, his comparison of the pos- 
 sessive affix lyk (Tataric) and ly (Osmanli) with the Teutonic lich, 
 lik, and ly in " friendly." Yet, after his philological fury is relieved, 
 Professor Boehtlingk never represses a natural impulse of honesty 
 and fairness. He says : " If Professor Schott, in his work on the 
 ' Altaic or Finno-Tataric Languages,' had no other purpose than by a 
 massive collection of words and roots, apparently connected, to make 
 it seem likely that the Ural-Altaic languages stand to each other in a 
 nearer degree of relationship than to other languages, one cannot help 
 admitting that he has gained his point. But, after this is admitted, 
 we must insist all the more strongly, that, before the single classes 
 have been studied more accurately and raised to the standard of 
 comparative grammar, an end should be put to further labours of 
 this kind." 
 
 It is evident from this, that, while Professor Boehtlingk from his 
 point of view considers such preliminary researches as without the 
 pale of science ("unwissenschaftlich "), he forgets that they involve 
 questions of great and pressing importance, and that, on the threshold 
 of every science, attempts of this kind are necessary, nay useful. 
 "Without Frederick Schlegel, we should have had no Bopp and Pott ; 
 without Sir William Jones, no Colebrooke and Wilson. We are but 
 too much inclined, particularly when science becomes a profession, to 
 mistake the means for the object, and to lose sight of those problems 
 to which our professional studies are but subservient. It should be 
 remembered that what is now called comparative philoloo'y is, after 
 all, only a means toward a solution of some of the most important 
 philosophical and historical questions. 
 
21 
 
 However, the great question here before us may be stated in a dif- 
 ferent manner, and the answer that can be given even now will be 
 such as to satisfy all purposes of ethnological research. The first 
 question is this : " Supposing the Finnic, Samoiedic, Tataric, 
 Mongolic, and Tungusic languages had no original affinity, is it 
 possible to account for the coincidences which have already been 
 pointed out between them ?" If not, the next question is: " Supposing 
 they had one and the same source, can we account for the differences 
 such as have been pointed out between them ?" To this latter question, 
 I think, the answer will be in the affirmative, if we consider for a 
 moment the relation between languages such as Portuguese and 
 Sanskrit, and if we take into consideration the peculiar circumstances 
 under which the dialects of the Turanian nations have grown up. 
 It is this latter point which requires a more particular consideration. 
 
 Second Section. 
 
 General Division of Languages into Family, Nomad, and State 
 
 Languages. 
 
 The Turanian languages may be characterised as nomadic, in 
 opposition to the Arian languages, which, in their grammatical and 
 etymological economy, partake of what may be called a political 
 character. A similar idea is expressed etymologically, though perhaps 
 not intentionally, in the very names of Turanian and Arian — the 
 former being derived from a root meaning " to be swift," " to roam 
 about ;" the latter, from a root which is best known to us in the Latin 
 arare, the Greek apow. From this ancient root, AE, we have in San- 
 skrit, arya*, which meant originally a husbandman, a man of the third 
 caste, a Vai«ya ; then took the sense of lord of the soil, " assiduus ; " 
 and lastly, in its derivative form of Arya, became the name of honour 
 in which the Brahmanic Indians delighted as early as the times of 
 the Veda. This climax of meaning may seem peculiar, and peculiar 
 it may perhaps be called if we remember that "peculiaris" is derived 
 from " peculium'' and " pecus," chattel and cattle, and that therefore it 
 means what is proper, right, though it be strange to others. Now it 
 
 * See Pan. iii. 1, 103. 
 
22 
 
 is a well-known fact, — well known, at least, since Wilhelm von 
 Humboldt explained and proved it,— that language is the outward ex- 
 pression of what he calls the spirit or individuality of a nation. Starting 
 from this point of view, and resting on the principles which Humboldt 
 established, I propose to divide languages, according to the same prin- 
 ciples on which we divide the diiferent forms of political societies, into 
 three general classes, into " Family," " Nomad," and " State " lan- 
 guages. These three divisions correspond very nearly with Humboldt's 
 morphological classification, as formularised by Pott, where we find the 
 three classes of " isolating," " agglutinative," and " inflectional" lan- 
 guages. Pott adds a fourth class, which he calls transnormal or 
 incorporative, i. e. the polysynthetic American dialects. Humboldt 
 adds an intermediate class between the monosyllabic and aggluti- 
 native. But there really exists no language which is entirely mono- 
 syllabic, or entirely agglutinative, or entirely inflectional. In most 
 languages, traces can still be discovered which show that every one 
 of these three formative principles has at one time been at work in it, 
 although the general character is sufficiently fixed by the preponde- 
 rating influence of the one or the other. Humboldt, however, con- 
 siders these three classes as perfectly distinct, and denies, or at 
 least does not venture to assert, the possibility of historical transi- 
 tion between them. He establishes in an earlier work the following 
 four principles.* 
 
 " I. Language expresses originally objects only, and leaves the 
 understanding to supply the connecting form. Language endeavours 
 to facilitate this supplementary act by the position of words and by 
 expressions which, though originally indicative of objects and things, 
 may be understood as referring to relation and form. Thus, in the 
 lowest stage, grammatical articulation is represented by phrases and 
 sentences. 
 
 II. These expedients are reduced to a certain regularity ; the 
 position of words becomes fixed ; the words in question lose their 
 independent character, their material sense, often their original 
 sound. Thus, in the second stage, grammatical articulation is 
 
 * Uber das Entstehen der grammatischen Formen und ibren Einfluss auf die 
 Ideenent-vfickelung; 1822. 
 
23 
 
 conveyed by fixed construction, and by words whose meaning is half 
 material, half formal. 
 
 in. The position of words becomes uniform; formal words are 
 brought in contact with material words, and become affixes. Their 
 connexion, however, is not yet inseparable : the sutures are visible ; 
 the whole is an aggregate, but not yet an unity. Thus, grammatical 
 articulation in the third stage is conveyed by what is analogous to 
 form, but not yet formal. 
 
 IV. Formal elements at last prevail. The word becomes one, 
 modified only by a change of inflectional sound, according to its 
 grammatical position. Every word belongs to a category, and has 
 not only a lexicological, but also a grammatical individuality. Words 
 expressive of form have no disturbing secondary meaning, but are 
 pure expressions of relationship. Thus, in the highest stage gram- 
 matical articulation is conveyed by true form, by inflection, by purely 
 grammatical words." 
 
 Third Section. 
 
 Mutual Relation of the three Forms of Language, progressive and 
 
 retrogressive. 
 
 HUMBOLDT, EUNSEN. 
 
 Aptee this lucid statement of the gradual growth of grammatical 
 forms, it is extraordinary that Humboldt should still have doubted 
 a possible historical transition between the different forms. Pro- 
 fessor Boehtlingk's words on this point deserve to be quoted to- 
 gether with Humboldt's. " It is inconceivable," he writes, " how, 
 with such a view on the origin of inflection, any one can doubt for a 
 moment about the possibility of two such languages as Chinese and 
 Sanskrit having the same origin. I say the possibility, not the 
 historical reality, because all attempts at proving such a common 
 origin ought from the very beginning to be stigmatised as vain, 
 futile, and therefore unprofessional." 
 
 With the exception of the last clause, this expresses exactly 
 the point at issue between Humboldt's view and your own convic- 
 
24 
 
 tion on the historical scale of languages enforced in your lecture 
 delivered at Oxford in 1847. Pott also, as Boehtlingk remarks, 
 is on your side, and expressed his opinion in 1836 in the follow- 
 ing words : " It is certainly conceivable that the formation of the 
 Sanskrit language, as it is handed down to us, may have been preceded 
 by a state of the greatest simplicity and entire absence of inflections, 
 such as is exhibited up to the present day by the Chinese and other 
 monosyllabic languages." 
 
 I should say, that, in the same manner as in every body-politic, 
 traces of a former nomadic or even family life can be discovered, we 
 may really discover in all Arian languages traces of a Turanian and 
 Chinese formation through which they had passed. Nay, during 
 periods of anarchy, conquest, and migration, political languages seem 
 to relapse into nomadic unsettledness, and during periods of apathy 
 and stagnation nomadic languages may fall back into a state of 
 Chinese helplessness. But what interests us here is the ascending 
 scale, the primary growth of languages, not their secondary forma- 
 tions and reformations. 
 
 § 1. Character of Family Languages. Chinese. 
 
 In a family, though at first it only lives in and for itself, occa- 
 sional starts of nomadic dispersion must naturally take place; and 
 history again shows us occasionally, in nomadic tribes, incipient traces 
 of a political concentration. The same is the case in language. In 
 Chinese, though it may properly be called the most perfect type ot 
 a family language, we see that the expediency of agglutinative forms 
 began to be felt. This is most palpable in the spoken dialects of China, 
 and in other languages, commonly called monosyllabic. In the 
 Shanghai dialect, wo is to speak, as a verb ; wo-da, a word. JVoda 
 would be the nominative, wodaka the genitive, pela woda the dative, 
 tang woda the ablative.* The characteristic feature, however, 
 which is impressed on the face of the old Chinese language, is just 
 ■what we may observe among ourselves in the conversation of friends 
 accustomed to speak together on familiar subjects. It is a style 
 
 * The Gospel of St. John in the Chinese Language according to the Dialect of 
 Shanghai, by Professor J. Summers, 1853. 
 
25 
 
 of thought and speech, not unusual even now between husband 
 and wife, between mother and daughter. The one generally knows 
 beforehand what the other is going to say, and words are used 
 more to indicate than to describe thought. Long sentences are 
 hardly thought of, because misapprehensions are not possible, and 
 particular intonations, familiar accents, are sufficient to prepare 
 the mind of the hearer for what he has to expect. These intona- 
 tions even have been fixed and preserved in Chinese, though ori- 
 ginally they may have been nothing else than what we may ob- 
 serve in our own parlance, when, for instance, in dictating to a 
 writer, we tell him " Right," or " Write." Sometimes, however, 
 the Chinese, particularly the old Chinese, approaches to a style of 
 speech such as only a solitary thinker could frame in his conversations 
 with himself; a kind of algebraic chain, intelligible to the initiated 
 but not to others. It has been truly said, therefore, that, as a lan- 
 guage, Chinese is admirably fitted for meditation and reflection. It 
 is a language of Brahmanic Munis, but unfit for the forum ; and, 
 though it would convey a false idea to characterise the Chinese as a 
 " parler enfantin," it may truly be compared to the short-hand con- 
 versation of a small and rather monosyllabic family. 
 
 § 2. Character of Nomad Languages. 
 
 The Turanian language goes a step beyond this. It expresses in 
 words, not only ideas, but the relation of ideas. The Turanian 
 life is no longer a family life, or the life of a troglodyte Muni. It is 
 the life of tribes, where the individual and the family are separated 
 only by the floating walls of tents, and in daily intercourse with 
 their clansmen. It is an indispensable requirement in every no- 
 madic language, that it should be intelligible to many, though their 
 intercourse be but scanty. The introduction, therefore, of elements 
 expressing as clearly as possible the grammatical relation of words, 
 the invention of signs, whether natural or conventional, for distin- 
 guishing between nominal and verbal roots, the avoidance of every- 
 thing that might obscure the meaning of words or the intention of 
 their grammatical exponents, distinguishes the Turanian from the 
 Chinese. 
 
26 
 
 § 3. Character of State Languages. 
 
 The difference between the Turanian and the Arian, between the 
 nomadic and the political languages, is not less characteristic. In 
 the Turanian dialects, as long as they remain purely nomadic, the 
 suffixes, whether in themselves intelligible or not, are felt as modify- 
 ing elements, and as distinct from the words to which they are 
 attached or " glued." In the Arian languages, the modifications of 
 sense produced by prefixes and suffixes are perceived ; but the 
 suffixes themselves are no longer felt as the sole cause of these mo- 
 difications. The difference is the same as between a compositor 
 and a reader. The compositor puts the s to the end of a word and 
 looks on the type s in his hand as producing the change of pound 
 into pounds. To the reader the s has no separate existence (except 
 on scientific reflection) ; the whole word expresses to him the modified 
 idea, and in his perception the same change is produced by " penny'' 
 and "pence " as by "pound" and " pounds." 
 
 It is a mistake to imagine that it is a distinguishing mark of the 
 Turanian languages to express the relations of grammar by inde- 
 pendent words. Most of the Turanian suffixes must originally have 
 been independent words ; but the same applies to the Arian and the 
 modern Chinese languages, and, as far as etymological science is con- 
 cerned, more of the Arian than of the Turanian suffixes have as yet 
 been traced back to their original form and independent meaning. 
 Humboldt admits this, and he says that even in Burmese, which is 
 half-brother to Chinese, the case terminations can but rarely be traced 
 back to their original meaning. The sign of the plural "to," for 
 instance, can be explained only if, disregarding the accent, we derive 
 it from "to," to increase, to add. Professor Boethlingk has established 
 the same by abundant evidence. 
 
 FoDETH Section. 
 
 General Features of Nomad or Turanian Languages. 
 
 § 1. Integrity of Roots. 
 
 There has been an instinctive feeling in the Turanian nations, 
 
 which led them to preserve their roots unchanged, although they 
 
27 
 
 allowed them to be surrounded by a large number of prefixes and 
 affixes. The radical and significative portion of their words always 
 stands out in distinct relief, like a living nucleus, and it is never 
 obscured or absorbed, as frequently in the Arian languages. Age, 
 in French, for instance, is eage and edage in Old French ; edage 
 is a corruption corresponding to a Latin cBfaticum ; cetaticum is a 
 derivation of cBtas, cBtas an abbreviation of cevitas, and in cBvum 
 ce only is the radical portion, containing the germ from which all 
 the other words derive their life and meaning. What trace of cs 
 (ahi, al-wv, Sk. ^yus) is there left in age ? Turanian languages 
 cannot afford to retain such words as age in their living dictionaries ; 
 and perhaps, from a linguistic point of view, such words can hardly 
 be considered as an ornament to any language. In the few cases 
 where Turanian civilization has reached the point at which the 
 language of the race becomes the object of philosophical and 
 historical research, in the few cases where we meet with Turanian 
 grammarians, Turanians giving their own thoughts on the pe- 
 culiarities of their own language, the distinctness of the radical 
 elements in every word is generally pointed out by them as a feature 
 which they consider essential to all language, and for the absence of 
 which, in the Arian dialects, they find it difiicult to account. The 
 Bask, which is in this respect the very type and perfection of a Tu- 
 ranian language, has produced several grammarians ; and one of them, 
 Darrigol, dwells very strongly on this point. He says (p. 18.) : 
 
 " Comme c'est un vice dans le langage que les syllabes radicales, 
 sans le concours des inflexions accidentelles, soient souvent impuis- 
 santes pour faire un sens meme generique ; ce serait aussi une autre 
 extremite vicieuse, qu'un mot primitif, par la meme qu'il aurait un 
 sens, fut necessairement determine a un sens speciflque, adjectif, 
 substantif, adverbial, &c. La monosyllable az, par exemple, r^pond 
 h peu pres a I'infinitif wow wV ; je dis a peu pres, parce que le sens 
 qu'elle presente est encore plus vaste et plus indefinie que celui de 
 rinfinitif fran9ais. La monosyllable az est une radicale sur laquelle 
 nous Itablissons naturellement : 
 
 az-te (nourrir), az-cor (nourrissant), 
 
 az-le (nourricier), az-curri (nourriture), 
 
 az-cai (nourrisson), az-i (nourri), &c. " 
 
28 
 
 In Turkish, also, the root is never obscured, though surrounded by 
 a luxuriant growth of conjugational derivatives. We have 
 
 sev-mek, to love, 
 sev-me-mek, not to love, 
 sev-e-me-mek, not to be able to love, 
 sev-dir-mek, to make love (causative), 
 sev-dir-me-mek, not to make love, 
 sev-dir-e-me-mek, not to be able to make love, 
 sev-dir-ish-mek, to make one love one another, 
 sev-dir-ish-me-mek, not to make one love one another, 
 sev-dir-ish-e-me-mek, not to be able to make one love one another. 
 
 In all these forms the radical element " sev" is distinct and pro- 
 minent, and so it is in all Turanian languages ; while in Semitic, and 
 still more in Arian formations, the root may be affected and changed 
 to such an extent that even an experienced scholar has difficulty in 
 disentangling it. 
 
 § 2. Formative Syllables felt as distinctive Elements. 
 
 It is not necessary for the purposes of Turanian grammar, that 
 the suffixes should retain their etymological signification ; but 
 it is essential that they should be felt as distinct from the 
 word to which they are appended. It requires tradition, society, 
 and literature to keep up forms which can no longer be analyzed, 
 and in which the formal elements cannot at once be separated 
 from the base. The Arian verb, for instance, contains many forms 
 where the personal pronoun is no longer felt distinctly. Still 
 tradition, custom, and law keep up the understanding of these veteran 
 words, and make us feel unwilling to part with them. This would 
 be incompatible with the ever-shifting state of a nomadic society and 
 language. No debased coin can there be tolerated, no obscure legend 
 accepted on trust: the metal must be pure, and the legend distinct; 
 that the one may be weighed, and the othei-, if not deciphered, at 
 least recognized as a well-known guarantee. A Turanian might tole- 
 rate the Sanskrit : 
 
29 
 
 as-mi, a-si, as-ti, 's-mas, 's-tha, 's-anti, 
 
 I am, thou art, he is, we are, you are, they are 
 
 or even the Latin : 
 
 's-um, e-s, es-t, 'su-mus, es-tis, 'sunt. 
 
 In these instances, with a few exceptions, root and suffix are as 
 distinguishable as, for instance, in the Tsheremissian : 
 
 ol-am, ol-at, ol-es, ol-na, ol-da, ol-at. 
 Nay, the identity of sound in two such forms as ol-at, thou art, 
 and ol-at, they are, shows the Tsheremissian at a disadvantage if 
 compared with Sanskrit. But a conjugation like the Hindi, 
 
 hun, hai, hai, hain, ho, hain, 
 
 would not be compatible with the genius of the Turanian lan- 
 guages, because it would not answer the requirements of a no- 
 madic life. Turanian dialects exhibit either no terminational distinc- 
 tions at all, as in Mandshu ; or a complete and intelligible system of 
 affixes, as in the spoken dialect of Nyertshinsk, But a state of con- 
 jugation in which the suffix of the first person singular and plural 
 and of the third person plural are the same, where there is no distinc- 
 tion between the second and third person singular, and between the 
 first and third person plural, would necessarily lead to the adoption 
 of new and more expressive forms in a Turanian dialect. New 
 pronouns would have to be used as suffixes, or some other expedient 
 to be resorted to for the same purpose. In the Arian family this 
 confusion of distinctive terminations is most general in, but by no 
 means confined to, the youngest members. In English it is only 
 the second person singular, a form hardly ever used, which has 
 retained its characteristic termination in the imperfect. But even in 
 Anglo-Saxon, instead of the Gothic plural bindam, bindith, bindand, 
 ligamus, ligatis, ligant, we find the second person bindad used equally 
 for the first and third. And in the passive we see the Gothic also 
 equalize the first and third person singular, and the three persons 
 plural, — a proceeding unknown, or at least very rare, in any real 
 Turanian dialect. 
 
 D 
 
30 
 
 § 3. Facility in producing new Forms. 
 
 Hence we may understand how the Turanian languages continue 
 to retain their creative power of producing new grammatical forms. 
 A Turanian, to a certain extent, holds himself responsible for his 
 grammar. Though he does not spontaneously create every gram- 
 matical form as he is using it, still he participates to a certain extent 
 in its formation, inasmuch as he not only forms his words into a 
 sentence, but also his roots and suffixes into words. A language 
 containing this grammatical consciousness may live and grow, and 
 may produce analogous forms, after- discarding forms which had be- 
 come corrupt, dead, and unintelligible. 
 
 Castren, in his dissertation " De Affixis Personalibus " (page 
 13.), bears witness to the fact that, while the literary language of the 
 Mongolians has no pronominal affixes, whether subjective or pre- 
 dicative, this characteristic feature of the Turanian family has but 
 lately broken out in the spoken dialect of the Buriates, and in 
 the Tungusic idiom spoken near Nyertshinsk, in Siberia. We must 
 guard here against a mistake. These primary formations of Tura- 
 nian grammar are different in principle from the secondary or analy- 
 tical formations in the Arian languages which they resemble. The 
 Turanian appends his terminations again and again to verbal or 
 nominal bases, thus forming new grammatical compounds ; while 
 modern Arian dialects retain the corrupt matter of a former orga- 
 nism, and form small sentences by putting explanatory prepositions 
 and pronouns before words worn-out by use. 
 
 If we consider that in Turanian grammar the adoption of the 
 pronominal suffixes, subjective and predicative (as it has taken place 
 but lately in some Tungusic and Mongolic dialects), means really 
 the introduction of a new conjugation and the remodelling of the 
 principal part of declension, we must allow that the Arian languages 
 can show nothing similar to this power, not of renovation only, but 
 of regeneration. 
 
 § 4. Scarcity of irregular Forms. 
 While the Arian languages, compared with the Turanian, are 
 weak on this point, they are, on the other side, strong in what no- 
 
31 
 
 madic races possess hardly at all: — irregular and dialectical 
 forms. To keep up such forms in grammar, language requires 
 tradition and different social elements, which the plains of Central 
 Asia and the taciturnity of Mongolian tribes could not furnish. 
 Without an uninterrupted continuity between successive phases of 
 speech, without a mutual intercourse of dialects, nothing irregular 
 can maintain itself in language. Thus, as most Turanian languages 
 are the languages of the day ; as they are, so to say, in the power 
 of each generation ; as they cannot resist change, cannot preserve 
 what is not continually revived and used, we may understand 
 why they are so extremely regular and monotonous, without any 
 of those strange anomalies which, in the Arian languages, harass 
 the student, but delight the scholar. Professor Boehtlingk's state- 
 ment fully confirms this view. " In the agglutinative languages,'' 
 he says *, " we find that one and the same grammatical relation is 
 always expressed in the same manner, making allowance only for 
 purely euphonic changes, which are regulated by very general laws. 
 In the Indo-Germanic languages, one and the same relation is fre- 
 quently expressed very difierently, varying according to the words or 
 whole classes of words to which they refer. It is impossible there 
 to account for the difference of termination by general euphonic 
 laws. In the Ural-Altaic languages, on the contrary, we have one 
 declension and one conjugation, and only a very small number of 
 irregular forms. In the Indo-Germanic, we meet with several de- 
 clensions and conjugations, and a mass of irregular forms, which all 
 point to a long-continued life, or at least to a life of intense indi- 
 viduality in grammatical formation." 
 
 § 5. Rapid Divergence of Dialects. 
 
 Another feature of the Turanian family of languages, intimately 
 connected with the two which have just been pointed out (their 
 power of renovation, and their regularity of formation), is the 
 great variety of grammatical growth to which the members of this 
 family are liable if once split and separated for any length of 
 time. If a nation retains the consciousness of its grammar, if the 
 
 * Introduction, p. xxiv. 
 D 2 
 
32 
 
 idea which it connects, for instance, with a plural is only that of a 
 noun followed by a syllable indicative of plurality, it is evident that 
 many forms are possible to realize this idea. In Tibetan the plural 
 may be expressed by thamtche (all), tha-ded (each), koun (many), 
 as in Chinese by tchou, ko, tchoung. (Eemusat, " Lang. Tart.," p. 362.) 
 The same applies to several of the modern languages of India ; and 
 in some these plurals of substantives are so clearly felt as compounds, 
 like " animal-mass" or "stone-heap," instead of " animals," " stones," 
 that the verb after them is put in the singular and not in the plural. 
 Nay, even after a suffix expressive of plurality has again been ob- 
 scured, and can no longer be identified with any collective noun, we 
 may still perceive its original nature by seeing that plurals formed on 
 this principle continue to have the verb in the singular. The same 
 applies to the plural of Greek neuters, which were originally collec- 
 tive nouns, i. e. feminines in the singular. If the ablative is expressed 
 by an additional syllable, expressive of removal, distance, or cause, 
 many syllables would equally answer the purpose. Thus we find in 
 Bengali, kartrik, hetuk, pHrvak, diya, rahit, sange, sati, hoite, &c. 
 all used in the sense of the Latin ablative. However, in one and the 
 same clan during one and the same period, one suffix would most 
 likely become popular and be fixed for certain grammatical cate- 
 gories. Thus, out of a large mass of possible formations, a small 
 number only would become customary and technical, so as in the 
 end to lead to a scheme of declension such as we find in political 
 languages. Different hordes, however, as they became separated 
 would feel themselves at liberty to repeat the same process, and 
 might thus fix in their different idioms different phases of gram- 
 matical life, wliich, if confined to one and the same tribe, would have 
 disappeared without leaving any traces. Thus the power of self- 
 conscious renovation which, as confined to one and the same dialect, 
 had only the effect of discarding old and irregular forms, may, if ex- 
 ercised on diverging dialects, produce such a total difference between 
 idioms most closely related, as to make them appear entirely dis- 
 connected. 
 
33 
 
 § 6. Contrast between the Progress and Growth of Turanian and 
 Arian Languages. 
 
 If we try to put the life and growth of such languages clearly 
 before us, we shall find that in a nomadic language the sudden rise 
 of a family or of a small association may produce an effect which, 
 in political languages, can only be produced by the ascendancy of a 
 town or a province, a race or a religious sect. The peculiarities of 
 a family may there change the whole surface of a language, and the 
 accent of a successful Khan may leave its stamp on the gram- 
 mar of his whole tribe. When one of the great Tatar chiefs pro- 
 ceeds on an expedition, he, as Marco Polo tells us in the fourteenth 
 century, puts himself at the head of an army of a hundred thousand 
 horse, and organizes them in the following manner. He appoints an 
 officer to the command of every ten men, and others to command a 
 hundred, a thousand, and ten thousand men respectively. Thus, ten of 
 the officers commanding ten men take their orders from him who com- 
 mands a hundred ; of these, each ten from him who commands a 
 thousand ; and each ten of these latter from him who commands ten 
 thousand. By this arrangement each officer has only to attend to 
 the management of ten men, or ten bodies of men, and the word of 
 command is spread from the Khan to the last common soldiers in a 
 hundred thousand, after passing through not more than four mouths. 
 This is characteristic, linguistically as well as politically. 
 
 In political languages, a change of grammar is generally preceded by 
 a political revolution, by war of races and conquest. Such changes, 
 whether they happen in the steppes of Tatary or in the capitals of 
 Europe, we are accustomed to call the growth of languages, because 
 we generally look only at the surface of languages and are hardly 
 able to discover the continual undergrowth of individual expressions, 
 family words, cockneyisms, provincialisms, and dialects. But lan- 
 guages really cannot be said to grow in the sense of continually 
 advancing and rising. Grammatical forms have no substantive 
 existence {ovaLa). They exist as forms in the speech of nations, and 
 the speech of a nation again has its existence in the speech of 
 individuals. It is, therefore, in the case of phonetic changes only that 
 
 D8 
 
34 
 
 we can speak of one word being changed into another ; but old forms 
 never grow into new ones. Old leaves fall and new leaves ap- 
 pear. Out of many possible forms and varieties some rise to the 
 surface ; while others, which had been classical for a time, are blown 
 away. But the new forms existed long before, and the old forgotten 
 forms may sometimes reappear. When the language of Germany 
 ceased to be Gothic and became High German, it was not because 
 Low German had grown into High German. The people who spoke 
 Gothic had passed away from the literary or political stage of Ger- 
 many; few only lingered behind: large masses of Franks pressed on, 
 and soon the language of the church, of the court, and of the poet 
 was High German and no longer Gothic. But High German existed 
 long before ; just as Italian existed long before Dante, and Italia- 
 nizing forms may be discovered as vulgarisms as early as the time of 
 Cato. 
 
 There are two changes in grammar which must be distinguished. 
 The one is produced suddenly by conquest or migration, and we may 
 call it a dislocation of language. Thus Gothic was dislocated by 
 High German ; and the effects are clearly visible not only in grammar, 
 but also in the regular dislocation (verschiebung) of the phonetical 
 system. The other change is wrought without any violent con- 
 cussion ; as it were, by the wear and tear of a language in its own 
 working. A number of possible analogous forms rise slowly and 
 imperceptibly into existence and use ; individual words or modes of 
 expression become popular and general, and dialects intermix and 
 exchange. This may be called a secondary formation in language. 
 Frequently a dislocation of language brings out more manifestly the 
 accumulated effects of a previous process of secondary formation; 
 because, if the higher ranks of society are broken and literary occu- 
 pations for a time discontinued, the spoken language has an oppor- 
 tunity for throwing off the fetters of literary usage, and legitimizes 
 at once its numerous natural offspring. Arian languages, parti- 
 cularly in modern times, change principally by the former, 
 Turanian by the latter process. 
 
35 
 
 Fifth Section. 
 
 On the Principles of Formation and Derivation in the Turanian 
 
 Languages. 
 
 We have hitherto considered the nomadic state of language in its 
 general effects on grammar. It is necessary now to consider how 
 the same nomadic spirit would act more particularly on the formation 
 of grammatical categories and the derivation of words. 
 
 § 1, Scarcity of Synonymes and Homonymes. 
 
 As most words are originally appellatives or predicates expressive 
 of distinguishing qualities, one object was capable of many names in 
 the ancient languages. In the course of time, however, the greater 
 portion of these synonymes became unintelligible and useless, and 
 they were mostly replaced by one fixed name which might be called 
 the proper name of such objects. The more ancient a language, the 
 richer it is in synonymes. Synonymes, again, if used constantly, 
 naturally give rise to a number of homonymes. If we may call the 
 sun by fifty different names expressive of different qualities, it is 
 clear that some of these names will be applicable to other objects also 
 which happen to possess the same qualities. These different objects 
 would then be called by one and the same name ; they would become 
 homonymes. It is clear that this luxuriant growth of poetical ap- 
 pellatives must lead to confusion ; and it is only in small and compact 
 communities, and by the help of national poetry, epic or sacred, that 
 synonymes and homonymes can be kept up for any length of time. They 
 do exist in the ancient Arian languages, and form a peculiar charm in 
 their poetry ; but even there, even in political languages, they become 
 more and more embarrassing. In the Veda the earth is called " Urvi " 
 (wide), " Prithvi" (broad), " Mahi" (great), and many more names, of 
 which the Nighantu mentions twenty-one. These twenty-one words 
 would be synonymes. But Urvi, again, is not only a name of the 
 earth, but it also means a river. Prithvi or prithivi means not only 
 earth, but sky and dawn. Mahi is used for speech and cow, as well 
 as for earth. Therefore earth, river, sky, dawn, speech, and cow 
 
 D 4 
 
36 
 
 would become homonymes. To the genius of nomadic languages the 
 continuance of such words is utterly repugnant. Most of these 
 old terms, thrown out by language at the first burst of youth- 
 ful poetry, are based on bold metaphors. These metaphors once 
 forgotten, or the meaning of roots from which the words were 
 derived once dimmed and changed, the words themselves become 
 insignificant. This would not matter so much in Arian languages, 
 where people soon learn to look upon nouns as symbolic signs, with- 
 out much reference to their etymological meaning. But in the Tura- 
 nian languages, properly so called, the number of nouns belonging 
 to this class must always be comparatively small. 
 
 § 2. Adjectives, Substantives, and Verbs not always distinct. 
 
 In the Turanian languages many words are still uncertain between 
 substantives, adjectives, and verbs ; that is to say, their radical mean- 
 ing is still so free and general that they can be used as subjects and 
 as predicates, and, therefore, as nouns, adjectives, and verbs. Thus 
 we read in Boehtlingk's Yakute Grammar (§ 238.) : " The sub- 
 stantive is not treated as separate from the adjective, because 
 they are frequently the same.'' If the adjective takes the termina- 
 tions of declension, it becomes a substantive ; as adjective it has no 
 grammatical suffix, but is put before the substantive, as in a Sanskrit 
 compound. For instance, Hungarian .4' szep ^JtVagroA, the beautiful 
 flowers. Here the plural termination {K) is put to the substantive 
 only. But A kesek eletlenek, the knives are blunt. Here the plural 
 is expressed both after the substantive and the predicate. We may 
 compare such phrases as "our knives'' and "the knives are ours ;" 
 but they are different in origin. The same process which in the 
 Turanian languages raises an adjective to a substantive, may also 
 transform it into a verb. In Hungarian, according to Eevay, fagy 
 signifies both "frost" and "it freezes." Lak (now only used ii^ 
 composition) meant " habitation ;" and if followed by a pronoun, it 
 becomes a verb, lak-ik, habitat. "In the infancy of language,'' to quote 
 Kevay's explanation of these forms, " the forms fagy-en, fagy-te, 
 fagy-6, arose from the inartificial annexation of the pronoun the 
 radical having both the force of the noun and of the verb when 
 
37 
 
 predicated of persons : primarily denoting gelu ego, tu, ille, instead of 
 gelu meum, tuum, suum, and then gelasco, gelascis, gelascit. After- 
 wards, by a more perfect formation which is still in use, a distinction 
 was made between them in this way ; namely, iha.t fagy-om, fagy-od, 
 fagy-a or Ja, my cold, thy cold, his cold, lak-om, lak-od, lak-ja, my 
 place, thy place, his place, were employed as nouns, and fagy-ok, 
 fagy-oz, fagy, I freeze, thou freezest, he freezes, lak-om, lah-ol, lak- 
 ik, I dwell, thou dwellest, he dwells, as verbs." The insuificiency 
 of this explanation has been pointed out by Garnett, and we shall 
 have to examine it hereafter; but still Revay's observations are 
 valuable. In Yakutic, "frozen" is ton; but followed by subjective 
 suffixes, it also means " to freeze." Tin, in the same dialect, means 
 breath ; but followed by verbal terminations, it becomes a verb, to 
 breathe. Substantives even which have lost their appellative na- 
 ture, and are real nouns, are verbalized by the mere addition of these 
 subjective suffixes. " Agha" in Yakute means father ; the same word 
 is raised to a verb, " I am father," by simply appending the subjective 
 pronouns, without any intermediate verbal derivative. " Min 
 agha-bin," means I am father ; " an agha-ghin," thou art father ; 
 "kini agha,'' he (is) father. In the same manner the root Sana, 
 which as a root may mean thinking, thought, or thinker, is conjugated 
 sani-bin, I think, sani-gin, thou thinkest, &c. The only difference here 
 consists in the final vowel of the base. Even inflected bases are 
 carried along by the powerful current of verbal formations in these 
 dialects. For instance, "jia," in Yakute, means house; "jiagha," in 
 the house ; hence " kinilar jiaghalar," they are at home (Yakute 
 Grammar, § 419.). In Mandshu, the number of words which have 
 no distinctive termination is considerable, and the same bases may 
 be used there as nouns, verbs, adverbs, and even as particles (Gabe- 
 lentz, p. 19.). In Chinese, owing to the absence of all derivative 
 elements, the identity of verbal and nominal bases is absolute. Not 
 so, however, in the modern Chinese dialects. In the Shanghai dia- 
 lect the use of a noun to express the verbal idea, and vice versa, is 
 rather an exception than a rule. A noun is not transformed into a 
 verb without its proper change of form by suffix, not merely by 
 change of tone, as in the general language of the country. And in 
 like manner the verb does not take the form of the verbal noun, ex- 
 
38 
 
 cept by the addition of a formative particle.* The Bhota and Bho- 
 tanta languages have certain distinctive particles for nouns and verbs; 
 yet many words are still nominal as vyell as verbal. In Burmese, ne 
 means to remain, to last, and the sun ; mai, to be dark, to threaten, 
 and the indigo plant. Humboldt, when speaking of these Burmese 
 roots, says (p. 345.): " They are really Chinese roots, but they show 
 unmistakably an approaching similarity to Sanskrit roots. Frequently 
 these so-called roots have without any change, a nominal meaning, but 
 their verbal meaning shines through more or less distinctly." This 
 similarity with Sanskrit roots may seem a bold assertion ; but traces 
 of the same indeterminate character of bases, nominal and verbal, 
 can really be discovered in Sanskrit, though, of course, on a limited 
 scale. VaA, in Sanskrit, if followed by the case terminations, 
 means speech ; genitive, vaA-as ; dat., vaA-i ; abl., vak-shu. VaA, if 
 followed by subjective suffixes, becomes a verb and means to speak ; 
 va^-mi, vak-shi, vak-ti, I speak, thou speakest, he speaks. In com- 
 position the same word vaA is used almost like an adjective. For 
 instance, kalaha, disturbance ; vak-kalaha, quarrel. The difference 
 between verbal and nominal bases is marked here only by the quantity 
 of the radical vowel. In Latin also the same observation may still be 
 made with regard to voc-s, voc-is, on one side, and voc-o, voc-a-s, 
 voc-a-t, on the other ; only that in voc-a-s and voc-a-t the inter- 
 mediate a indicates the verbal nature of the compound, and thus 
 distinguishes noun and verb. 
 
 § 3. Pronominal Affixes, subjective and predicative. 
 
 Now, it should be observed that, in the Arian languages, where, with 
 few exceptions, the distinction between nominal and verbal bases is 
 drawn most carefully, there was really much less necessity for it, be- 
 cause these languages never employ possessive or predicative suffixes 
 after nominal bases. A base, therefore, if followed by a pronoun, 
 would at once be recognized as a predicate in these languages, and no 
 ambiguity could ever arise, even if the base by itself might mean 
 " speech," " speaker," and " to speak." The compound, i. e. the base, 
 
 • Cf. J. Summers's Translation of St. John, p. vi. 
 
39 
 
 together with the pronoun, would always mean " I speak," and never 
 " my speech." Every base followed by a personal pronoun in the 
 Arian languages is verbal. An ambiguity arises, however, in the 
 Turanian and Semitic languages. Here two sets of pronominal 
 suffixes are used; the one subjective, added to verbs, the other 
 predicative or possessive, added to nouns. I call subjective the 
 pronominal suffixes which Castren calls predicative, and predi- 
 cative those which he calls possessive. The reason for this change 
 of terminology is obvious. A pronoun, if appended to a noun sub- 
 stantive or used as a possessive suffix, is always predicative. This 
 applies to every language without exception. In the Egyptian si-f, 
 his son, si, son, is the subject,/ is the predicate. But if a pronoun 
 is attached to a base really verbal, or if it is used as what is com- 
 monly, but erroneously, called a predicative suffix, the pronoun is 
 always the subject, and the verbal base is the predicate. In the Egyp- 
 tian iri-ef, he does, ef, he, is the subject, which is qualified by iri, 
 doing. We may change the verb, and the subject remains the same ; 
 but the subject shifts as soon as we change the pronoun. 
 
 The Semitic languages also employ their pronominal affixes to 
 mark the persons of the verb : I love, thou lovest, he loves ; and to 
 express the persons of the noun : my house, thy house, his house. 
 The one and the other class of pronominal affixes are attached to the 
 end of words, and in some cases they differ but slightly, or not at all, 
 as in the third person feminine of the singular, which is " ah" both 
 after nouns and verbs. 
 
 The Arian languages, on the contrary, have never possessed more than 
 one set of pronominal affixes, and these are used to mark the persons 
 of the verb. Instead of predicative affixes, they use their genitives, 
 fjLov, aov, or independent possessive adjectives, /jo'e, aoQ. Compounds 
 such as Egyptian si-k, son-thou, i. e. son of thee, thy son ; or Hebrew, 
 lebush-ka, dress (of) thee ; or Hungarian atya-m, atya-d, atya, my, 
 thy, his father, are impossible, nay inconceivable, to an Arian mind. 
 If a compound is to be formed wherein the pronoun is the predicate, 
 the Arian mind is forced to put the pronoun first, and thus we find, 
 indeed, in Sanskrit, but in no other Arian dialect, predicative pro- 
 nominal 'prefixes, such as, mat-putra, tvat-putra, tat-putra, my, thy, 
 his son ; but never predicative affixes. 
 
40 
 
 There is one solitary exception to this general rule, which 
 deserves to be pointed out, the Persian. The Persian is the 
 only instance of an Arian language where in all compounds the 
 predicate can be put first. We say in Persian « puser-i-dost," 
 the son of the friend, which, if expressed as a compound in any 
 other Arian dialect, would have to be expressed by " dost-i- 
 puser.'' The only way to account for this direct violation of the 
 genius of the Arian grammar in Persian is to ascribe it to the in- 
 fluence which the Semitic language and literature exercised on the 
 inhabitants of Persia from the time of Cyrus up to that of Firdusi. 
 If the Persian could once break his mind into the Semitic fashion of 
 placing the subject in a compound first and the predicate last, it was 
 but another step in this direction to do the same where the predicate 
 is a pronoun, and thus we find in Persian a set of predicative affixes 
 attached to nouns in the same manner as in Semitic languages. We 
 say in Persian, not only dil-i-keniz, the heart of the maid, but 
 
 dil-i-men, my heart, 
 dil-i-tu, thy heart, 
 dil-i-o, his heart. 
 
 Here " men," " tu," " o," are the regular personal pronouns. These, 
 however, may be abbreviated again, and in some instances be replaced 
 by distinct pronominal affixes, so as to give 
 
 dil-em, my heart, 
 dil-et, thy heart, 
 dil-esh, his heart. 
 
 Another instance where predicative pronominal affixes seem to occur 
 in an Arian dialect, is an exception only in appearance, for it would 
 be wrong to compare these really anomalous forms with expressions 
 such as we find in the secondary formations of the Arian languages : 
 I mean the Italian " fratelmo," my brother, " patremo," my father. 
 Though "fratelmo" may seem a compound hardly difiering in prin- 
 ciple from the Persian " dilimen," my heart, it is necessary to observe 
 that "fratelmo" is only an abbreviation and corruption of "fratellus 
 meus," or rather of " fratellum meum." Now it is clear that, as soon 
 as two words have once been articulated by indicatory terminations 
 
41 
 
 such as " us," the speaker is at full liberty to place the predicate 
 either before or after the subject. Even if the pronoun is not yet an 
 adjective, agreeing in gender, number, and case with its subject, but 
 is distinguished only by the termination of the genitive, all restric- 
 tions which were felt with regard to the collocation of words in com- 
 pounds, will naturally disappear. Let us only consider what is meant 
 by what we call a genitive, and we shall see that a language which 
 expresses the genitive at all is as free with regard to its collocation 
 as it is with adjectives. 
 
 The genitive in most languages is an adjective, only as yet without 
 terminations to mark case and gender. But the adjective again is 
 generally a derivative where, by means of a pronominal affix, the 
 quality, action, &c. expressed by a noun is grafted on a pronominal 
 subject. In Sanskrit, " dakshind" means the south; and if we add 
 to it the pronominal base "tya" (syas, sya, tyad), we get "dakshi- 
 wa-tyas," he from, of, or in the south, i. e. southern. IIoXic in 
 Greek means city ; and if we add to it the same pronominal deriva- 
 tive, we get 7roX(Vr)c, " urbanus," " civis." Sometimes this pronominal 
 derivative is only a short a ; as Sk. manas, mind, manas-as, what 
 belongs to the mind, Greek ttlixti.q, trust, icicTTi-oe, trusty. The dif- 
 ference between a genitive and an adjective can best be shown in 
 Sanskrit. In Sanskrit the neuter sahas, strength, forms the genitive 
 sahasas. This genitive is the most general predicate, and its ter- 
 mination remains the same, whether the subject to which it refers 
 be in the singular or plural, masculine or feminine, nominative or 
 accusative. We may say " sahasas patis," the lord of power, and 
 " sahasas patim," the lord of power (accus.), the genitive only ex- 
 pressing sahas, power, as a predicate of something. But if we express 
 in sahas-as, not only the predicate, but gender, number, and case, the 
 genitive becomes changed into an adjective ; and instead of saying 
 " sahasas patim," the lord of strength (accus.), we now say " sahasam 
 patim," the lord powerful, both words being in the accusative. The 
 regular genitive of words like iriariQ would be irloTi-oQ (instead of 
 TTtoTEwe) ; and if we make this genitive express gender and case, 
 we get TrioTioQ, la, lov. The usual Sanskrit genitive in "sya" is 
 probably but another form of the pronominal base " tya," which we 
 had in dakshiMa-tya, only that the former cannot be raised to an 
 
42 
 
 adjective, while the latter takes the exponents of gender and case. 
 What we express is nearly the same, whether we say a bird of the 
 water, or an aquatic bird. The adjective aquatic we should express 
 in Sanskrit by ap (water) + tya (aptyas, a, am) ; the genitive, 
 by udaka, water, + sya, " udakasya," of the water. Both forms, 
 genitive as well as adjective, mean originally and etymologically 
 " water-there,'' and " water-there-he, she, it," taking the local adverb 
 " there," as the nearest approach to the radical meaning of the de- 
 monstrative pronoun. Here, then, we clearly see the contrast 
 between Semitic and Arian grammar. In Hebrew we can say 
 first, as it were by one act of intuition, malk-i-zedek, king-justice. 
 In Sanskrit we say dharma-ragra, justice-king. Secondly, we can 
 turn it into a phrase and say in Hebrew, ben oBeor, the son-he 
 Beor, i. e. the son of Beor ; or still more clearly in Ethiopic, anqaz 
 enta samay, "porta ea coeli," anqaz being feminine, and "enta"' 
 being the feminine pronoun. In Sanskrit, on the contrary, we 
 add the pronoun to the predicate, and say ra^a dharma-sya, "the 
 king justice-there," i.e. the king of justice; or we actually form 
 an adjective (and every genitive in Mahratti, for instance, is an 
 adjective distinguishing gender and case), and say "rex Justus," 
 or "regina justa." If a language has once formed genitives and 
 adjectives, it is no longer under the restraint of what we might call 
 the national logic difiering thus in the Semitic and the Arian race. 
 "Without grammatical exponents the Hindu can only say " ragra- 
 putras," king-son, or " tvat-putras,'' thy son. But as soon as we 
 form the genitive, we may say " tava putras," or " putras tava ;" and 
 vnth the adjective, tavakas putras or putras tavakas, or, in Latin, 
 frater meus and mens frater. Phonetic corruption may afterwards 
 reduce the adjective to the state where instead of " meus, mea, 
 meum," for instance, we have only " mo " for all cases and genders. 
 Still " mo," in fratelmo, occupies its place only as a degenerate de- 
 scendant of " meus." It follows the subject as a pronominal adjec- 
 tive, but it does not enter as a predicative pronoun into composition 
 with a substantive, like the Persian dil-em, my heart. 
 
 What has been said with regard to fratelmo applies with equal 
 force to such compounds as " Hotel-Dieu." They may be used to 
 illustrate the Semitic mode of thinking ; but grammatically " Dieu," 
 
43 
 
 in " Hotel-Dieu," is the Romance genitive or casus obliquus, and only 
 as such could it remain in a few expressions without requiring the 
 new sign of the genitive, de. In the Oath of Strassburg we have 
 "pro deo amur,'' "deo"' being the casus obliquus, while in the same 
 document the nominative is " deus." * 
 
 With the exception of Persian, therefore, and after the discovery 
 of the cuneiform inscriptions, we may say, with the exception of 
 modern Persian, no Arian language employs personal pronominal 
 affixes except after verbs. 
 
 § 4. Means of distinguishing nominal and verbal Bases in Turanian 
 
 Languages, 
 
 To avoid the confusion, which would naturally arise if roots can be 
 used nominally and verbally, and if pronouns can be attached to them 
 as subjects and as predicates, languages have at a very early period 
 resorted to various expedients. Instances occur where languages 
 really do not distinguish between asinus ego and asinus mei. For 
 instance, when the definite conjugation is employed in Hungarian, 
 ir-om may mean unguentum mei, or scribo ; lep-em, tegimen mei or 
 tego. In modern Hungarian, eso denotes pluvia, and es-ik, pluit; 
 but in the fifteenth century the simple root es was employed in both 
 senses. There can be little doubt, as Garnett remarks, that at an early 
 period this identity of the verbal root with the noun was a general 
 law of the language. At present the abstract noun in Hungarian 
 commonly differs from the simplest form of the verb by the addition 
 of a syllable, usually as or at : e. gr. ir, scribit ; iras, scriptio ; ir-at, 
 scriptum. In languages without a formal distinction between no- 
 minal and verbal roots, care has generally been taken not to use a 
 root, once sanctioned as nominal, for verbal purposes. Thus it happens 
 that a root is sometimes used in one dialect for verbal, in another for 
 nominal purposes only, but not for both in one and the same dialect. 
 (See Yakute Grammar, § 236. note 71.) The pronominal suffixes 
 might by themselves have served as a guarantee against a confusion 
 of nouns and verbs, if their subjective and predicative forms had been 
 kept sufficiently distinct, because, as a general rule, bases followed 
 
 * Diez, Altromanisclie Sprachdeukmale, 1846 
 
44 
 
 by predicative suffixes would be nominal ; if followed by subjective 
 suffixes, verbal. But to do this was almost impossible, from the very 
 nature of the pronominal suffixes. In some languages they are iden- 
 tically the same, whether used as subjects or as predicates, or, as we 
 should say, as nominatives or as genitives. In the Tungusic class, 
 no distinction exists, so far as the pronominal affixes are concerned, 
 between pay of me, i. e. my fay, and pay I, i. e. I pay. But again, 
 even where there is a formal difference between these two sets of 
 pronominal suffixes, this difference could never be very considerable, 
 because both, after all, must be derived from the same pronouns ; the 
 subject! ves mostly from the nominative, the predicatives from an 
 oblique case. 
 
 Languages, therefore, as soon as they began to care at all for 
 logical distinctness, were obliged to put a stop to the promiscuous 
 use of nominal and verbal bases. They were driven to distinguish 
 in every root the verbal from the nominal pole by some mark more 
 distinct than what was furnished by the slight variations of prono- 
 minal suffixes. In the Turanian family the Yakute language makes a 
 most favourable exception, for in it final letters are in most cases 
 sufficient to mark the verbal or nominal character of a base. In 
 Turkish we can only distinguish by accent between " giizeUm," my 
 handsome one, and " giizelim," I am handsome. 
 
 § 5. Means of distinguishing nominal and verbal Bases in Arian 
 
 Languages. 
 
 In the Arian languages, although none but subjective suffixes were 
 used, it was felt expedient to distinguish a verbal from a nominal base. 
 The most primitive tenses in Sanskrit are the perfect and the aorist. 
 They are formed from the root not burdened as yet by any Vikarawas, 
 i. e. distinguishing verbal marks. The perfect in Sanskrit was origi- 
 nally a present ; it became the perfect, in our sense of the word, only 
 after the introduction of a new special form of the present. Every 
 Sanskrit root, in order to be used for verbal purposes, was originally 
 raised to a perfect ; that is to say, its initial letter was reduplicated. 
 This is as clear in Greek as in Sanskrit, and the number of perfects 
 not restricted as yet to a past tense is considerable in both languages. 
 In Sanskrit we have a root tan, to stretch. If employed for verbal 
 
45 
 
 formations, this root was originallj^ reduplicated and became tatan. 
 To this verbal base subjective pronouns were attached, thus giving 
 tatdn-a, tatan-tha, tatdn-a, I stretch, thou stretchest, he stretches, 
 restricted as yet in time neither to the present nor to the past. In 
 Greek, if we take the root MNA, to remember, we see that, in order 
 to adapt it for verbal employment, it has to be reduplicated first, 
 after which subjective pronominal suffixes are added, and the new 
 compound fiiixvrt-fiai. takes the sense of / remember. 
 
 But although this process of producing verbal bases as distinct 
 from nominal bases was probably one of the most ancient, it was 
 by no means the only one employed, in the Arian languages. Every 
 one of the numerous Vikarawas in Sanskrit, Greek, and Latin is really 
 a derivative element ( a verbal Umadi, as Pawini might call it) put 
 aside for verbal purposes. If we add to the root tan, in Sanskrit, 
 the Vikarawa of the eighth class, we get tan-u, which again, fol- 
 lowed by subjective pronouns, gives us tano-mi, tano-shi, tano-ti, or 
 tan-e, tanu-she, tanu-te, in the sense of I stretch, thou stretchest, he 
 stretches. The same in Greek, where from the root TAN we get 
 not only ravvia, but, by other Vikarawas, rttVu (i. e. revito), riTalvio, 
 &c. These Greek Vikarawas have been exhibited in a most lucid 
 arrangement by Geo. Curtius. (" Bildung der Tempora und Modi, 
 1846.") It was owing to the introduction of these new bases, such as 
 for instance tvtt-t-u) instead of TETv<j>a (i. e. te-tvw-cl), that the old 
 reduplicated forms took the sense of perfects. It was the absence in 
 them of all distinguishing marks which excluded the old reduplicated 
 forms from the present KaTe^oxhv, while most of the Vikarawas, ex- 
 pressing either inchoative activity, or participal quality, or motion, 
 or continuity, were eminently fitted for expressing an action actually 
 present. 
 
 Without entering as yet into the formation of the real preterite of 
 the Arian languages, — I mean what is called the second aorist in 
 Greek, and the multiform preterite in Sanskrit, — it would be of 
 interest to see how other languages gained the same point, — that of 
 forming the first verbal base — which the Arian accomplished by redu- 
 plication of the initial letter. In Chinese, we have no right to expect 
 anything of this kind ; but in the Turanian family, the Yakut has 
 already been mentioned with distinction, in so far as it fixed some 
 
 E 
 
46 
 
 and discountenanced other vowels at the end of verbal bases as a 
 means of distinguishing nominal and verbal radicals. 
 
 § 6. Means of distinguishing nominal and verbal Bases in Semitic 
 
 Languages. 
 
 In Tibetan, and its cognate languages spoken in the Sub-hima- 
 layan districts, many nominal bases become verbal by a mere repe- - 
 tition of the final letter : as nag, black, nag-go, it is black ; sum, 
 three, sum-mo, it is three. The present definite is always formed by 
 reduplication of the final letter, whether consonant or vowel: as 
 jyed, to do, ngajyed-do, I am doing. However distant these dialects 
 may appear from the language of Homer, I am inclined to consider 
 their final reduplication as prompted by the same motive which led 
 the Arians to the reduplication of the initial letter of their roots. 
 The repetition of the whole or part of a root was felt as the most natural 
 expedient to express continuity, activity, or motion ; in fact, to express 
 what Aristotle calls the distinctive point between verb and substan- 
 tive, time.* 
 
 If then the Arian languages, though they used pronominal suffixes 
 after verbal bases only, if the Sub-himalayan languages, though they 
 used hardly any pronominal suffixes (excepting only some more 
 advanced member, like the Naga dialects), were driven to invent 
 distinctions between nominal and verbal bases, much more must 
 this want have been felt by the Semitic nations. With the little 
 difference between their subjective and predicative suffixes, measures 
 of a much more general charactgr were necessary, if confusion was to 
 be avoided. Might not, therefore, the extraordinary idea taken up 
 by the Semitic languages evidently at a very early period, — for it is 
 common to all Semitic tribes, — of reducing all verbal bases alike to 
 a triliteral appearance, be accounted for by the same motive ? It is 
 against the genius of Shem to reduplicate an initial consonant, and 
 there is no real Semitic root beginning with twice the same letter. 
 But the final letter could be reduplicated, and the verbs ghaan-ghain 
 
 * See Aristotle, Poetic, c. 20. De Interpr. v. 2. Lersch, Sprach-Phaosophie der 
 Alten, ii. p. 13. 'Ovoiw. i^iv oZv ^a>v^ a-nfiamiK^, Kmh a-vvSriKTiv Hrev xpimv. 'VrJiM 
 Si ia-ri rh vpoiTar\iJ.(uvov xpi-'o". In German the only -word for verb is " Zeitwort." 
 
47 
 
 show how frequently it was. I do not say that reduplication was the 
 only means of distinguishing verbal and nominal bases in Hebrew. 
 Other expedients were at hand, as various as the Vikarawas of the 
 Arian languages. In the Arian languages these Vikarawas are 
 generally put at the end of a root ; but nasals, and nasals with vowels, 
 are inserted in the middle of roots, in order to transform them into 
 new verbal bases. Thus yu^r, to join, becomes yu-na-gr-mi, I join. The 
 same and many other ways were open to the Semitic dialects. Now it 
 is, I believe, admitted by all Semitic scholars, that the radicals of the 
 Semitic family were originally biliteral ; the point on which they 
 differ is only the method by which triliteral roots can be traced back 
 to their more primitive biliteral state. Fiirst adopts the rather severe 
 process of simply beheading the triliteral roots; Klaproth adopts 
 the other alternative, and proposes to cut off their tails. The best that 
 can be said on the subject was said by Ewald, in 1827. " It is even 
 possible,'' t he says (Grammar, §95.), "to reduce the full-grown 
 triliteral bases to shorter radicals, from which all secondary bases 
 were derived, as their meaning became more and more different. 
 For instance, the triliteral roots, qajaj, qajah, qS/jab, qajar, may all 
 have sprung from the short qaj, to cut. And here it should be 
 observed, that roots, where only the final letter is reduplicated or 
 where a soft consonant has been added, stand nearer to the primitive 
 radical and are more related to one another than those which are 
 distinguished by the addition of a strong consonant. A comparison 
 of such roots, carried out with ingenuity and caution, would lead to 
 many new results ; but it should be remembered that, in etymological 
 researches of this kind, we transcend the limits of the peculiarly 
 Semitic language and grammar.'' 
 
 Now, it is true that, in the present state of Semitic language, all 
 bases, whether verbal or nominal, are alike triliteral, and that there- 
 fore it might seem as if the reason assigned above for the creation 
 of triliteral roots were not commensurate to its effect. But while 
 there is not a single biliteral verbal root in actual use among the 
 Semitic tribes, there still exist some biliteral forms ; and they belong 
 invariably to old nouns, or to still older pronouns. Some of these 
 nouns are without any verbal analogy or etymology. Others are now 
 derived from verba geminantia, hamzata, quiescentia ; but with them, 
 
 E 2 
 
48 
 
 if there is any real ground for derivation, the opposite process would 
 generally be the more natural. No scholar could seriously think of 
 deriving ab, father, from ahah, voluit; hen, son, from banah, sedificavit; 
 kol, all, from kalal, circumdedit. After the Semitic mind had 
 once imbibed the triliteral character of its predicative roots, biliteral 
 roots were eliminated in the most sweeping manner. Even pro- 
 nominal bases were made triliteral, whether by additional syllables, 
 or by changing mere vowels into semivowels. New substantives 
 could, of course, -be formed from verbal roots to any amount; and as 
 these new words were more expressive and intelligible, and could be 
 sufficiently distinguished by peculiar vocalization from the different 
 forms of conjugation, they well nigh supplanted all ancient mono- 
 syllabic nouns. Now, there must have been a reason for this thorough- 
 going change ; and I cannot believe that the first start can be ex- 
 plained simply on phonetic or rhythmical grounds. It is true that 
 peculiar features in a language are sometimes perpetuated which 
 owe their origin to the mere fancies or crotchets of one patriarchal 
 i. e. specific, (etSon-oios) individual. , But in the case before us we 
 may observe analogous tendencies in languages not Semitic in their 
 origin ; and I venture, therefore, to rest my argument for the original 
 verftaZ character of triliteral roots on these four points: — 
 
 I. According to the Semitic system of grammar and orthography, 
 there is now not a single root which is not triliteral. 
 
 II. Nouns and pronouns exist which sometimes in writing, and 
 more frequently in pronunciation, are decidedly biliteral. 
 
 m. Triliteral nouns are mostly secondary verbal formations, and 
 therefore in many cases not absolutely identical in all Semitic dia- 
 lects. They mostly differ in different dialects by verbal derivation 
 and vocalisation. 
 
 IV. In many cases the character of the additional litera tertia, 
 whether initial, medial, or final, is sufficiently marked by this, that 
 it is either a semi-vowel, or nasal, or sibilant, or a reduplicated 
 letter. It frequently varies in different Semitic dialects, while 
 the two radical letters remain the same. I shall give one instance 
 — one not the less instructive because it has been pointed out 
 many times before, and first, I believe, by Klaproth.* If we 
 
 * Prinoipes de I'iltude Comparative des Langues, par le Baron de Merian ; suivis 
 
49 
 
 take the usual Hebrew paradigm qaial, he strikes, it can easily 
 be proved that the ^, as a semi-vowel, is here the litera tertia, and 
 must give way. This leaves us qa<, which in Hebrew shows itself 
 again in g-efel, destruction, and with the change of Ttit into 3ade , 
 as qaj&h, qajaj, qajab, qajar, &c. In Arabic this root has been most 
 prolific. "We get qatta, qa&ba, qaia'ha, qafafa, qaiala, qa^ama, qad'da , 
 qadhdha, qaththa, qajja, qasama, qajaba, qajada, qajara, qajama, 
 qajmala, qajja qajqaja, qajaba, — all in the sense of cutting, striking, 
 killing, dividing, breaking, biting, &c. How true it is, as Ewald 
 remarks, that, by following out etymological researches of this kind, 
 we transcend the limits of language, peculiarly Semitic, is shown by 
 this very instance. The Hungarian Ms, knife, the Mongolic kese , 
 to cut, chasu (tailler), the Turkish kesmek, cutting, the Garo kethali, 
 knife, show us that we are on ground common to the Turanian ; the 
 Sanskrit «as, and Latin caedo, that we are on ground common to the 
 Arian languages. This is by no means a solitary instance where a 
 root, after removing its various increments, or, so to say, divesting it 
 of its national dress, can be reduced to that form in which it may 
 be considered as a radical, common to all human speech. We must 
 not expect to find roots common to Semitic, Arian, and Turanian 
 languages, except those which express the simplest material impres- 
 sions. But roots like LAK, to lick, MAE, to decay, ZAR, to tear, 
 TAR, to transgress, SAR, to go, TAN, to give, &c., may safely be 
 considered as common property. No doubt they approach, in this 
 abstract form, very near to interjections, or mere phonetic imita- 
 tions ; but still there is a well-marked difference between these roots 
 and interjections. An interjection never grows, but is but the mo- 
 mentary outcry of a material impulse ; while a root is the conscious 
 and intentional expression of an impression, remembered and fixed 
 on the human mind. It is owing to this ideal character that a root 
 is capable of entering into the most various processes of assimilation 
 and combination. The root LAK, for instance, in Hebrew has taken 
 the triliteral form laqaq. In Arabic we have : 
 
 la'hiqa, to lick. lasama, to taste. 
 
 la"ha, to speak. laia'ha, to lick. 
 
 d'Observations sur les Raoines des Langaes Semitiques, par M. Klaproth. — Paris, 
 
 1828. 
 
 , E 3 
 
50 
 
 lahata, to exercise the tongue. lasa, to taste, 
 
 lagana, to lick. lahasa, to lick, 
 
 lassa, to lick. lisn, the tongue, 
 
 lasaba, to lick. lasa'ha, to be maligned. 
 
 The same root exists as "lib" in Sanskrit, as Xdxu) in Greek, as 
 " laigon" in Gothic, as "ligh" in Celtic, and in Latin "lingua.'' Again, 
 with the frequent transition of I, the dental semi-vowel, into d, the 
 dental media, we find, corresponding to the Latin lingua, or dingua, 
 the Gothic " tuggo," and the English "tongue.'' That the word " glos- 
 sary" should have grown out of this root LAK, may seem startling; 
 still there is not a link wanting to connect the two words either in 
 their form or in their meaning. Turning to the Turanian languages, 
 we find the Finnic lakkia, to lick, though it may be doubted whether 
 Man^u leke, to polish, Finnic laaha, the same, or Finnic lau, to 
 speak, could safely be referred to the same source. 
 
 § 7. TTie three different Directions of Grammar, Turanian, Semitic, 
 and Arian, represented by the three Sons of Feridun, Tur, Silim, 
 and Irij. 
 
 A. Tur. 
 
 As we have thus been carried back to times when we see the 
 three principal tongues, which we may represent as the three sons 
 of Feridun, as not yet separated, it may be of interest to catch at 
 least one glimpse of them as they are leaving their common home 
 and starting off in different directions. "What they carried away from 
 home were roots and pronouns. Two of them, Silim and Irij, seem 
 both to have held the secret how a root could be divided and 
 changed so that it might be used as a subject or as a predicate. Tur 
 also may have known it ; but he either forgot it, or he did not like 
 to tamper with those sacred relics which he had carried away 
 from his father's house. Under his care they remained the same, 
 without addition or diminution ; and when they had to be used, they 
 were only set and framed like precious jewels, but neither divided 
 nor polished down.* Now there were at least four things which 
 Tur had to express with his roots and pronouns. If he possessed a 
 
 * Conf. pag. 286. 
 
51 
 
 root for cutting, he wanted to say, I cut (present) ; I cut (past) ; 
 cutter, i. e. knife ; and my cutter, i. e. ray knife. These four little 
 phrases were indispensable for him if he wished to get on in the 
 world. As long as he was alone with his family and children, he no 
 doubt could make them understand by some expressive accent when 
 ngo.ta (moi battre) meant " I beat," and when ngo-ta meant " my 
 stick" (moi-baton). What followed would generally remove all un- 
 certainty, if it existed; for ngo.ta.ni, 1-strike-thou (moi battre vous), 
 could only mean " I strike thee." Again, as he could express to-day 
 by " this light," and yesterday by " that light," perhaps his wife and 
 children were not slow in understanding when he said kin-tien 
 ngo.ta, this day I strike, i. e. I strike now (tout a I'heure moi battre) ; 
 or tso-tien ngo.ta, that day I strike, i. e. I struck (jadis moi battre).* 
 All this may seem so natural, as far as construction goes, that at 
 first one hardly discovers any thing peculiar in these different modes 
 of expression. Still, in the construction of these two expressions, 
 ngo.ta, I beat, and ngo-ta, my stick, there is something so individual 
 and peculiar, that neither Silim nor Irij could imitate it. This is 
 the liberty of putting the predicate first in one sentence and last 
 in another. Silim could say ngo.ta, I beat (e'.qfol), but never ngo-ta, 
 my stick. He would have to put the predicate last in both phrases, 
 and say ta-ngo,. stick of me, like fe.q^ol, I-striking. Irij again, at least 
 in his early youth, could say ngo-ta, my stick (mad-damrfa), but never 
 ngo.ta, I-striking. Instead of this he had to say striking-I (tuda.mi). 
 This peculiarity by which Tur put the predicate sometimes first, some- 
 times last, may originally have been involuntary. As his roots were 
 not yet distinguished as nominal and verbal, as subjective and pre- 
 dicative, his ngo.ta, I strike, may' not have been meant for I striking, 
 but, like ngo-ta, my stick, for my-striking. Still we shall see that, 
 among his descendants, even after they had learned to distinguish 
 between nominal and verbal roots, and between subjective and pre- 
 
 * " Qu'un etranger me dise, ' Moi avoir soif, moi vouloir boire, moi desirer man- 
 ger," je comprends ce langage ; mais je ne puis m'empecher de sentir que c'est un 
 langage sans vie, sans nerf, sans liaison. Pourquoi ? Parce que I'ame du disoours, 
 la force unitive, le noeud de la proposition, I'essence du jugement, le verbe en un 
 mot s'y fait desirer, malgre la presence de I'infinitif." — Dissertation Critique, par 
 VAbbi Darrigol, p. 97. 
 
 E 4 
 
52 
 
 dioative pronominal affixes, some retained the power of putting the 
 subject first as well as last ; such as agha-m, father of me, i. e. my 
 father, and sani.bin, knowing-I, i. e. I know. This applies, however, 
 only to the North-Western descendants of Tur ; his other descendants 
 place the predicate first always. 
 
 B. Silim. 
 
 Silim, as we saw, started from home fully aware that his roots 
 might be made to answer two purposes. He therefore divided his 
 roots into simple nouns and fuller verbs ; also, he kept one set of his 
 pronouns, which had already grown and multiplied around him, for 
 his verbs, and another for his nouns. He had only one difficulty, 
 which, with all his acuteness, he could not overcome : he could never 
 think a predicate without first having thought his subject. There- 
 fore he could say wrath (of) God, and wrath (of) me, but not God (s) 
 wrath, and my-wrath. He also could say beating (of or to) me, i. e. 
 I did beat, and I-beatirg, i. e. I beat, but not beating-I, i. e. I beat. 
 The opportunity, however, which he had of forming at least these 
 two verbal compounds, beating (of) me, and I-beating, was not lost 
 by Silim ; and as he found it essential to make his friends understand 
 either that he had paid or that he meant to pay, he took the first 
 form, paying (of) me, i. e. paying (belonging to me, or possessed and 
 had by me), in the sense of the preterite, while the mere assertion of 
 I-paying was left to answer the purpose of a present or a future 
 payment. 
 
 C. Irij. 
 
 The mind of Irij was more comprehensive than that of Silim. He 
 was able to think, as it were by one grasp, ideas such as " gold- 
 piece," " God's love," &c., and he expressed them by a compound 
 word, in which the predicate being second in thought, and therefore 
 more present to his mind, came first in language. Now, as he could 
 say God's love, |U7;rp-o-iro\tc, father-land, Maha-ragra, always putting 
 the predicate first*, he could also say, 1-love, I-wife, but only in the 
 
 » 'linro?r(jTa/ios, which is generally mentioned as an exception, is only a literal 
 translation of an Egyptian word. On the difference between ' k.vip6(pi\os and 'tlKav- 
 Spos, TipiSOeos and 06(iTiyuoj, Awp66ios and QedSapos, see Pott, Personennamen, p. 88. 
 
53 
 
 sense of my love, and my wife, because his first word is always the 
 predicate. South of the " Snows" his descendants retained this 
 manner of expression for many centuries. They said, mat-putra, 
 tvat-putra, asmad-putra, my-son, thy-son, our-son. Their Northern 
 brethren, however, found it more expedient to express the predicative 
 nature of these pronouns more distinctly than could be done by mere 
 position. They therefore formed an independent predicative form, 
 whether genitive or adjective. This they were able to handle with 
 greater freedom, so that they might now say tekvov kfiov as well as 
 ifioii TEKVOV. As to his verbal compounds, Trij had two ways opened 
 before him, only just in the contrary direction to those of Silim. He 
 could say loving -I, i. e. I love ; and he did say so, after his verbal base 
 had been qualified by reduplication or by Vikararaas. This com- 
 pound phrase, however, was a mere predication, and could therefore 
 hardly be restricted to any point in time, whether past, present, or 
 future. It simply asserted a quality or an action. How then could 
 Irij express his preterite ? As he had as yet no auxiliary possessive 
 verb, like the "habere" and "tenere" of his descendants, he could 
 only use his possessive pronouns. But his possessive pronouns he 
 could only use before a verbal base, while he was accustomed to 
 mark all other formal changes at the end of words. Silim, when he 
 found himself in the same dilemma, simply divided his pronouns in 
 two, and put half before and half after the verb.* Irij had to do the 
 same ; but as he was putting his pronoun before the word, trying 
 to pronounce ma-ga, my-going, i. e. I went, the pronouns were so 
 strongly attracted towards the end of the root, that all that remained 
 in the place originally intended for the whole predicative pronoun 
 was not even a distinctive consonant, as in Hebrew, but only a 
 
 * Ewald (§ 152.) explains the fonnation of the Hebrew Aorist in the following 
 manner : — " The prefixes had to be pronounced as short as possible : one conso- 
 nant, not even followed by a vowel, was all that remained of the prefixed pro- 
 noun. This consonant happened to be the same for several persons ; confusion 
 would inevitably have arisen, unless, by a very natural expedient, the pronominal 
 prefix had been divided, so that the characteristic letters only remained as prefix, 
 while the rest were thrown towards the end of the word. The pronoun of the 
 second person sing. fern, being a~tin, atin was divided into at + in. At was 
 shortened into t and prefixed, while in was suffixed, thus giving tLqiel.i(n), 
 thou (woman) killest. 
 
54 
 
 strongly accented vowel, common to all these pronominal prefixes,- 
 and now called the augment ; while the consonants, without their 
 final vowels, were suffixed and placed at the end of the root. Thus, 
 if there was a root lip, to write or paint, it could first be raised to a 
 verbal base by reduplication. This verbal base lilep, writing, fol- 
 lowed by predicative suiRxes, would then give an aoristic compound, 
 lilep-a, writing-I, I write, lilep-itha, writing-thou, thou writest. - If 
 afterwards a new and more actual verbal base was produced by the 
 insertion of a nasal, such as limp, then, by the addition of predicative 
 sufiixes, limpami, limpasi, limpati, might be formed ; and as these 
 forms would express the present act of I am actually writing, the 
 old present lilepa would in time take the sense of a perfect, I have 
 written. The same root lip, however, being used as a subject, and 
 not as a predicate, participating, therefore, more in the nature of a 
 substantive than of an adjective, would, if preceded by possessive 
 pronouns, express my-writing, i. e. writing belonging to me, i. e. I 
 wrote, and thus a-lip-am (instead of ma-lip) would form the simplest 
 and most primitive Arian preterite. 
 
 D. The Descendants of Tur divided according to their Employment 
 of the Pronominal Affixes. 
 
 We have still to see how Tur proceeded in his verbal formations, 
 as it is not likely that he could be satisfied with the Chinese juxta- 
 position of pronouns and words. Some of his descendants in Bhota 
 and Bhotanta introduced formal elements to indicate the predicative 
 or verbal nature of their roots ; they formed their verbal bases, as 
 we saw, by reduplication. They also used formal elements to indi- 
 cate the predicative nature of their pronouns, and thus formed geni- 
 tives, or pronominal adjectives. In Chinese already we have ngo-ti- 
 sin, my heart. In ngo-ti-sin, ti, though originally it may have been 
 a pronoun, cannot be compared with the Hebrew aser, or the 
 Ethiopic za (masc.) and enta (fem.). In the Ethiopic mazmor za 
 Dawith, za is the masculine demonstrative or relative pronoun, re- 
 ferring to mazmor. It means the psalm which (to) David. But the 
 Chinese min-li or min-ti-li expresses not the people which (is) 
 power, i. e. the people of power, but people's power, where people's 
 is the predicate, and therefore to be expressed either as the first 
 
55 
 
 part of a compound or as an adjective. The late B. Garnett, in his 
 valuable treatise on the origin of the Genitive, has not perceived 
 this marked difference between Shem, on one side, and Japhet and 
 Tur, on the other, and has tried to explain the Semitic and Arian geni- 
 tive as the expression of one and the same logical process. In this 
 he could not succeed ; still his essay, like all he has written on com- 
 parative grammar, is very useful and important. 
 
 The Turanians, before! they began to use their pronouns as suffixes , 
 or prefixes, couldv only form these two grammatical propositions — 
 I-going (Bhot. ng4 d6-6), and mei pater (Mandshu, mi-ni ama). But 
 after this period of their grammatical childhood was over, we are 
 able to distinguish three divisions among the descendants of Tur, 
 each marked by the peculiar manner in which they employed their 
 pronominal affixes. 
 
 The first is the Tamulian, where subjective pronouns are always 
 suffixed, and predicative pronouns always prefixed ; where they say, 
 as in Telugu, vaguta.nu, vaguta.vu, vaguta.du, speaking-I, thou, 
 he, for I, thou, he speaks ; and na-tandri, as it were me-pater, i. e. 
 my father. 
 
 The second is the Caucasian, where likewise predicative pronouns 
 are prefixed and subjective pronouns suffixed. For instance, Suanian, 
 s-ab, w-ab, i-ab, my, thy, his father ; and b-chask.a, chask.a, chask.as, 
 I dig, thou diggest, he digs. In the first person of the verb, however, 
 we see the pronoun put twice, prefixed as well as suffixed ; and we also 
 meet with a second verbal formation, where, as far as the very per- 
 plexing changes and additions of the Caucasian verbs allow us to judge, 
 the pronoun was used throughout as a prefix ; I mean such forms as 
 the Lazian ma-zun, ga-zun, a-zun + asere, I ail, thou ailest, he ails. 
 If in this verbal compound, the pronoun was originally and inten- 
 tionally used as a prefix, we must take it as a possessive or predica- 
 tive pronoun, and the tense itself for a preterite. The analogy in the 
 formative process of Sansk. mat-pitar and Suanian s-ab, my father ; 
 Sansk. Khana.ti and Suanian chask.as, he digs; and Sansk. (m) 
 agam.am and Lazian ma.zun, I ail, would then be complete. But 
 whether this is so, or whether the Lazian mazun is altogether an 
 impersonal formation, must remain uncertain until we get more 
 ample information about the living languages of Colchis. 
 
56 
 
 The third division is that of the Altaic Turanians. In them the 
 method of joining roots and pronouns together is most intelligihle 
 and instructive. With the exception of the Samoiedic dialects, we 
 hardly require new materials to enable us to judge of the mechanism 
 of the Altaic suffixes. Castren's work, " De Affixis Personalibus Lin- 
 guarum Altaicarum" (1850), gives all the evidence that is required, 
 carefully collected and arranged. I differ from him in one point 
 only, and one which can easily be settled. All personal suffixes, if 
 attached to nouns, he considers eo ipso as possessive, while all other 
 suffixes are put down by him as predicative. These predicative 
 suffixes, whether used after adverbs (as ende (here) + bi (I) = en- 
 debi ; I am here) or after verbs (as tud.ok, I know), or after verbal 
 adjectives (as sever.im, I love), I call subjective, because they contain 
 always the subject of a logical proposition. This, however, would only 
 be a difference of terminology. But where I really differ from Castren 
 is in what he calls the second set of predicative, i. e. subjective, 
 suffixes. These suffixes, whether they are used to express the preterite 
 tense, or as exponents of transitive or definite verbs, are always (I 
 only except the Samoiedic, of which too little is known to form an 
 opinion) possessive suffixes, or predicative suffixes, in the sense in 
 which I use this word, and they ought to be considered as a second 
 set of possessive suffixes used after verbs, or rather after verbal 
 nouns. In form they agree with the possessive suffixes, wherever 
 these differ from subjective suffixes. 
 
 After this exposition, the mechanism of the Altaic pronouns is as 
 simple, and at the same time as ingenious as can be. The Altaic 
 Turanians differ from their brethren in so far as they put the predi- 
 cative or possessive pronouns after the subject to which they belong. 
 They say, as for instance in Hungarian and Tataric, 
 
 kes-em, 
 
 my knife. 
 
 u"hlu-m, 
 
 my son. 
 
 kes-ed, 
 
 thy knife. 
 
 ^iftligi-ng, 
 
 thy estate. 
 
 kes-e, 
 
 his knife. 
 
 a"hadsh-i, 
 
 his tree ; ana-si, his mother. 
 
 kes-iink, 
 
 our knife. 
 
 u"hlu-muz, 
 
 our son. 
 
 kes-tek, 
 
 your knife. 
 
 jiftligi-ngiz, 
 
 , your estate. 
 
 kes-ok, 
 
 their knife. 
 
 a"ha(;-Uari, 
 
 their trees ; ana-lari, their mothers. 
 
 This applies to all Altaic languages, for not one of them puts pre- 
 dicative suffixes before the word. They agree therefore on this 
 
57 
 
 point with Shem, and differ alike from Japhet and from the other 
 descendants of Tur. For the latter even in their earliest days, though 
 they allowed themselves the liberty of putting the subjective pronoun 
 before the verbal predicate, never ventured to place the predicative 
 pronoun after its nominal subject ; and on the heights of Pamer, as 
 well as in the sub-Himalayan basins of the feeders of the Ganges, 
 they rather formed pronominal genitives and adjectives, which, as in 
 Greek, allowed of a freer construction, but they never pronounced 
 a predicate, even where it was a mere pronoun, after the subject. 
 
 With regard to the subjective pronouns, the Altaic Turanians 
 agree with the rest. Subjective pronouns, without exception, are 
 placed after their predicates, the verbs. Thus we say 
 
 in Turkish ^/aer, to love, in Hungarian, ^/hall, to hear, 
 
 ben se-wesr-im, I love. hallok, I hear. 
 
 sen sewSr-sen, thou lovest. hallasz, thou hearest. 
 
 ol sewer, he loves. hall, he hears. 
 
 biz sewer-iz, we love. hallunk, we hear. 
 
 siz sewSr-siz, you love. hallatok, i/ou hear. 
 
 onlar sewer-ler, they love. hallanak, they hear. 
 
 These forms, with subjective suffixes, invariably express the present ; 
 but they are also put to other uses, which vary according to the 
 genius of different dialects. Before, however, we enter into this, it wiU 
 be necessary to state another general feature of these languages. It 
 is this, that "where they do employ different suffixes for the preterite, 
 these suffixes are always originally possessive or predicative." This 
 is what Professor Boehtlingk remarks, with regard to the Yakut 
 predicative, when he says that the possessive affixes form the 
 (predicative) affixes of the preterite ; as min suoghum, my absence, 
 or I was absent. For instance, 
 
 Tataric. 
 
 Turkish. 
 
 
 ana-m, 
 
 my mother. 
 
 sewer-d-im, 
 
 / loved. 
 
 ana-ng, 
 
 thy „ 
 
 sewer-di-n% 
 
 thou lovedst. 
 
 ana-si, 
 
 his „ 
 
 sewer-di, 
 
 he loved. 
 
 ana-muz, 
 
 our „ 
 
 sew6r-dik, 
 
 we loved. 
 
 ana-ngiz, 
 
 your „ 
 
 sewSr-di-n'iz, 
 
 you loved. 
 
 ana-lari. 
 
 their „ 
 
 sewer-di-ler, 
 
 they loved. 
 
58 
 
 Hungarian. 
 
 var-t-am, / waited (for it), = kesem, my knife. 
 
 vartad, thou waitedst, = kesed, thy „ 
 
 Yarta, he waited, = kese, his „ 
 
 vartuk, we waited, = k^sunk, our „ 
 
 vartatok, you waited, = kestek, your „ 
 
 vartak, they waited, = kesbk, their „ 
 
 In forming these verbal compounds, the Altaic languages felt none 
 of the difficulties which perplexed the Arian in forming their pre- 
 terites. They had already thrown off the spell which bound them in 
 pronouncing the subject before the predicate — that is to say, they 
 had thrown it off where the predicate happened to be a pronoun, 
 though not when it was a noun ; they therefore could express I 
 have loved, by " loving had or possessed of me," or " love belong- 
 ing to me.'' But some of them went beyond this. The Hungarian, 
 for instance, considering that tud.ok, knowing-I, was a phrase in 
 which J (ok) was the subject and knowing (tud) the predicate, very 
 properly refrained from having any object, whether expressed or 
 not, governed by the verb. Even transitive verbs, such as " I ex- 
 pect," were taken as intransitive if followed by pronouns to which 
 they served as predicates. " Varok," where " ok" is the subject, 
 and " var " the verbal adjective, would mean I expecting, I wait. 
 Varom, on the contrary, where " var " is the verbal noun, and " om" 
 the predicative pronoun, would always express I expect something, 
 " var " conveying an action Requiring an object, whether expressed or 
 not. " Olvasok '' would mean I read, i. e. I can read ; but I read Cicero, 
 would be Cicerot olvasom. This gives an entirely new character 
 to the Hungarian verb ; for the Hungarian mind, once accustomed 
 to this distinction, carried it out also through the other tenses ; and 
 while in the present the two sets of pronouns (predicative and sub- 
 jective) naturally offered themselves for these two distinct purposes 
 (transitive and intransitive, determinate or indeterminate), further 
 distinctions were actually introduced into the possessive pronouns, 
 already occupied by the preterites, in order to distinguish in the 
 preterite also between vdrtad, thou expectedest (it), and vartal, thou 
 waitedest. In Ostiakian the possessive pronouns form transitive, 
 the subjective pronouns intransitive verbs, though their difference 
 
59 
 
 is distinctly perceptible only in the second person plural. The dif- 
 ference of the tenses must then be expressed by derivative elements 
 attached to the verbal base. In Lapponian the possessives belong to 
 the preterites, the predicatives to the present. 
 
 Before leaving this subject, which I confess has carried me 
 away beyond the limits it ought to occupy in a general description 
 of the prominent features of Tur, yet in truth of great importance, 
 not only for Turanian grammar, but for grammar in general, I 
 must still mention one fact, to show how the spirit of analogy runs 
 through the whole system of conjugation and declension. We have 
 seen that in Hungarian suffixed pronominal possessives could be 
 used for forming definite verbs. If we knew nothing of the history 
 of that large family of languages to which the Hungarian belongs, 
 and if we only saw, that en varok meant I wait, en varom, I expect 
 (something), we should say, like most Hungarian grammarians, that 
 ok was the exponent of indefiniteness, om of definiteness. Its 
 origin once forgotten, it would become, as it were, the " definite 
 article of the verb." Now what is the origin of the definite article or 
 the definite form in nouns ? Lata in Samoiedic means " board ; '' 
 latada, " the board," and the final da is the possessive suffix of the 
 third person, so that originally it meant, his board. But this has 
 been forgotten, — and if we now want to express his board, we have 
 to say, puda latada, which is really, he-his board-his.* (Castren, 
 De affixis, pag. 11, Syrjasn. Grammar, p. 55.) In Syrjaenian again, 
 what has been taken for the termination of the accusative, is really the 
 possessive pronoun not of the third, but of the first person. Adzja 
 mortas, now means, I see the man, but originally meant I see 
 my man ; and that it was so, we can still see in the second and third 
 person. For while adzya meam mort-as, means " I see my man," 
 I see thy man, would be adzya tead mort-ta ; I take his knife, bosti 
 sya purt-sa. 
 
 * Cf. Castren. Ostiake Grammar §. 61. Boehtlingk, Yakut Grammar, p. 10. 
 
60 
 
 Sixth Section. 
 Etymological Peculiarities of the Turanian Languages. 
 
 But it is time to leave the history of these formal elements, and to 
 proceed to a consideration of the matter of the Turanian languages. 
 I suppose we may carry away with us the conviction that many 
 things in language which now seem formal were originally suh- 
 stautial. 
 
 § 1. Radical Meaning generally discernible. 
 
 We saw ahove how the Turanian roots were kept as integers, i. e. 
 intact and uninjured, though framed, enclosed, and grouped to- 
 gether in various styles, and fitted to express verbs, adjectives, nouns, 
 together with the most abstract and derivative ideas. The etymologi- 
 cal meaning of Turanian words is therefore more palpable than in the 
 Arian languages. Still the dictionary of the Turanians also had gone 
 through many editions before it fell into our hands, and we find in 
 it dead and petrified words just as in their grammar : and many of 
 them more diflicult to decipher and to revive than the pronominal 
 compounds which we examined just now. 
 
 8 2. Scarcity of ancient Words common to all Turanian Languages, 
 and identical in Form and Meaning. 
 
 What are called dead or petrified words are in general the most 
 ancient parts of a language ; they carry us back to that period during 
 which they were young and full of life ; and in cases where a separation 
 of languages took place, they frequently constitute the common heir- 
 loom of difierent dialects, and serve as the strongest indication for 
 determining and settling the exact degree of relationship between - 
 cognate tongues. The general aversion which the Turanian languages 
 have against any thing unintelligible, dead, or corrupt in grammar 
 or dictionary, explains the small amount of these ingredients in 
 most of them. It is well known, for instance, that in the several 
 branches of the Arian family, different degrees of family-life, from 
 
61 
 
 father and son, down to brother-in-law and sister-in-law, have, in many 
 cases, preserved their common Arian name. These words agree, 
 not only in root and meaning, but — and this is important — in their 
 individual derivative suffixes also. The word for father is not only 
 derived, in all the Arian languages, from the same root, pa, to 
 protect, — not only was the meaning of this root raised in the same 
 manner from that of protector to that of father, — but the same 
 derivative suffix also, tar, was preserved by all the descendants of 
 Japhet, thus distinguishing the language of Japhet from the Chinese 
 fu and mu (father and mother), the Tibetan po and mo (male and 
 female), the Subhiraalayan 'ba and md (father and mother), the Bur- 
 mese pha and ami, the Siamese po and me, and from all words similar 
 in sound and meaning, whether in Asia, Europe, or Africa. 
 
 Many derivations from this root pa were possible, such as San- 
 skrit palaka, protector, Vaidik payu, pavan, &c. PS.-tar, there- 
 fore, must be considered entirely as the result of one individual 
 choice. To maintain a word of this kind, even when its origin 
 became dim, not to allow it to be replaced by a new and more in- 
 telligible expression, was possible in an Arian, i. e. a social state of 
 language, not among nomadic tribes, who lived only for the present, 
 little concerned about past or future, without history and without 
 ambition. Thus we find that in the Turanian dialects the number 
 of common words is small. Remusat, in speaking of the Mandshu, 
 says, " Je distingue trois sortes de mots dans la langue Mandchoue : 
 les premiers lui sont communs avec celle des Tongous ; ils expriment 
 des idees simples, ou designent des objets de premiere necessite. 
 Quoiqu'ils soient en assez petit nombre, ils n'en forment pas moins 
 le fond de la langue. Une petite liste de mots essentiels mettra hors 
 de doute I'identite du Mandchou et des diff^rens dialectes des 
 Tongous. La ressemblance d'un petit nombre de mots dans les langues 
 des Mandchous et des Tongous, est d'un tout autre poids pour 
 prouver leur communaute d'origine, que ne pourraient I'etre les 
 differences d'un plus grand nombre d'autres mots, si I'on vouloit en 
 d^duire la consequence opposee." Professor Schott applies the same 
 principle, only on a much larger scale, and for a different purpose : — 
 " We ought not to despair . about the affinity of these four great 
 branches of languages (Tungusic, Mongolic, Turkic, and Finnic)," 
 
 F 
 
62 
 
 he says, " although the words for the most necessary ideas in them 
 are sometimes essentially different. The same remark might be made 
 if we compare languages acknowledged to be sisters, nay, even 
 dialects of the same speech. Tungusic as well as Finnic languages 
 offer the most striking evidence on this point." (page 44.) In a 
 former article Professor Schott had made the same observation with 
 regard to Indo-European languages. There, also, ideas and objects 
 of daily occurrence have sometimes been found under different names 
 in dialects, the close relationship of which cannot be doubted, e. g. 
 
 Sanskrit. 
 
 Greek. 
 
 Latin. 
 
 English. 
 
 1 
 
 2 
 
 s 
 
 4 
 
 putras, 
 
 vt6s. 
 
 filius, 
 
 son. 
 
 1 
 duhitar, 
 
 1 
 
 2 
 
 fiUa, 
 
 daughter. 
 
 1 
 
 2 
 
 1 3 
 
 1 
 
 bhratar, 
 
 ade\(p6Sf 
 
 frater, Sp. hennano, 
 
 brother. 
 
 1 
 
 2 
 
 3 
 
 i 
 
 stri, 
 
 yvi'ii. 
 
 femina, 
 
 woman. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 purusha, 
 
 airlip, 
 
 vir, 
 
 man. 
 
 1 
 
 _ 2 
 
 3 
 
 4 
 
 dyaus, 
 
 ovpav6Sf 
 
 coelum, 
 
 heaven. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 prithivi, 
 
 rfi, 
 
 terra, 
 
 earth. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 Aandra, 
 
 (TeA^VT?, 
 
 luna, 
 
 moon. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 siias. 
 
 (cei^aAlj, 
 
 testa (tete), 
 
 head. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 panis. 
 
 x^ip, 
 
 man us, 
 
 hand. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 vadanam, 
 
 ffrrf/^a, 
 
 OS, 
 
 mouth. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 vrikshas, 
 
 ShSpoy, 
 
 arbor, 
 
 tree. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 pakshi, 
 
 Spvis, 
 
 avis, 
 
 bird. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 pasbanas, 
 
 ireVpo, 
 
 saxum, 
 
 Stone. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 arhas, 
 
 fi|lOJ, 
 
 dignus, 
 
 worth. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 kesas, 
 
 ap(j, 
 
 crinis, 
 
 hair. 
 
 1 
 
 2 
 
 3 
 
 3 
 
 netram, 
 
 iitiea\iJ.6s, 
 
 oculus, 
 
 eye. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 nadi, 
 
 iroTO/i((j, 
 
 fluvius, 
 
 river. 
 
 1 
 
 2 
 
 3 
 
 4 
 
 asrik, 
 
 of/JO, 
 
 sanguis, 
 
 blood. 
 
 In the Semitic family. Professor Schott has pointed out the dif- 
 ference between Hebrew and Arabic words, such as 
 
63 
 
 Hebrew. 
 
 Arabic. 
 
 English. 
 
 yare^a, 
 
 gamar. 
 
 moon. 
 
 bar, 
 
 S'ebel, 
 
 mountain. 
 
 •hez, 
 
 sagar, 
 
 tree. 
 
 dbhen, 
 
 hagar, 
 
 stone. , 
 
 Even in languages whose relation to one another is not that of 
 sister to sister, like Sanskrit, Greek, and Latin, but of daughter to 
 mother, like French and Latin, we find the most common objects 
 expressed by different words. For instance. 
 
 Latin, 
 instead of ignis, 
 - seger, 
 anser, 
 caput, 
 discere, 
 domus, 
 jecur, 
 lapis, 
 
 OS, 
 
 verbum, 
 
 via, 
 cogitar, 
 
 Italian. 
 
 Spanish. 
 
 fuoco, 
 
 fuego. 
 
 malato. 
 
 enfermo. 
 
 oca, 
 
 oca. 
 
 testa, 
 
 testa. 
 
 apprendere, 
 
 aprender. 
 
 casa, 
 
 casa. 
 
 fegato, 
 
 higado. 
 
 pietra, 
 
 piedra. 
 
 bocca, 
 
 boca. 
 
 parola, 
 
 palabra. 
 
 camino, 
 
 camino. 
 
 pensare, 
 
 pensar.* 
 
 French. 
 we find feu, 
 „ malade, 
 „ oie, 
 
 „ apprendre, 
 
 „ maison, 
 
 „ foie, 
 
 „ pierre, 
 
 „ boucbe, 
 
 „ parole, 
 
 „ chemin, 
 
 „ penser, 
 
 It might be objected that in many instances a more careful study 
 of these languages, and particularly of their ancient history and their 
 dialects, would have enabled us to point out corresponding words 
 even where the most usual expressions differ. It might be said that 
 although the usual word for caput be tete in French, still caput 
 could be identified with the French chef, or vice versa, the French 
 tete with Latin testa. Again, it might reasonably be remarked, that 
 in the choice of our words from Sanskrit, Greek, and Latin, we have 
 intentionally omitted synonymes which would establish an agree- 
 ment between these languages. 
 
 If, instead of netram, oculus, we had taken the Sanskrit akshi ; 
 instead of pakshi, avis, the Sanskrit vi ; instead of vadanam, os, as, 
 
 • There are other words in Arian and Turanian languages, which, though 
 they may be considered as common property, have suffered so much from a pro- 
 cess of assimilation and accommodation in each dialect, that though we see a 
 similarity, we hardly are able to recognise identity. I mean such words as 
 wag-tail, which in German has been turned into Bachstelze (hrook-stilter), while 
 in Italian it is translated into coditremola, in French into hoche-queue, in Greek 
 into (retffowvyis and KiWovpos. In Sanskrit the conception is different, for it is 
 called there not " wag- tail " but " lame- walker. " 
 
64 
 
 the Sanskrit and Latin would have agreed. But it was our object to 
 sliow how by the very fact of collateral expressions, or by the under- 
 growth of new popular names, the same diversity which strikes us 
 in closely allied nomadic idioms can be detected, though in a smaller 
 degree, between the members of the Arian family, nay, even be- 
 tween such languages as Italian and Latin. If the sudden irruption 
 of a stream of nomadic tribes over the ruins of the Roman empire 
 could stir up the whole basis of the Latin, and bring out again the 
 long-repressed nomadic tendencies of an Arian language to such an 
 extent as to change the whole surface of its words and its grammar, 
 why should we feel surprised at similar results in languages where 
 no literary or political centralisation has ever checked the super- 
 fetative' tendencies of the human tongue? And further, if in the 
 Arian words we had chosen our instances, not from the leading 
 literary languages, like the Latin of Cicero and the English of 
 Shakspeare, but from provincial dialects, under whose protection 
 the nomadic life of a language continues often unobserved up to the 
 present day, we should have been able to show a still greater ap- 
 proach between Arian fluctuation and Turanian unsettledness. 
 Grimm, when speaking of the earliest periods of the German 
 language, describes this most beautifully.* " The idiom of Nomads," 
 he says, " contains an abundant wealth of manifold expressions for 
 sword and weapons, and for the different stages in the life of their 
 cattle. In a more highly cultivated language, these expressions 
 become burdensome and superfluous. But, in a peasant's mouth, 
 the covering, bearing, calving, falling, and killing of almost every 
 animal, has its own peculiar term, as the sportsman delights in calling 
 the gait and members of game by different names. The eye of these 
 shepherds, who live in the free air, sees farther, their ear hears 
 more sharply, — why should their speech not have gained the same 
 living truth and variety ? " 
 
 § 3. Turanian Numerals. 
 The Turanian Numerals, if considered from this point of view, 
 tend to illustrate and confirm the principles which we before tried 
 
 * History of the German Language, p. 20. 
 
65 
 
 to establish. They do so particularly if contrasted with Arian 
 numerals . The Arian nations, it is well known, have preserved their 
 ancient common numerals as the most precious gifts of their childhood. 
 Even when rust and decay had disfigured and obscui'ed their value 
 and meaning, they were never parted with or replaced by new- 
 coined words. The Turanian languages, though more careful of their 
 numerals than of other words which could be thrown away at ran- 
 dom, and replaced instantaneously, have not been able to preserve in 
 every instance those common terms by which they first counted from 
 one to ten. At first sight, a general similarity between the Turanian 
 numerals is undeniable, unless we extend the limits of chance to an 
 unprecedented extent. But, on closer inspection, it becomes clear 
 that some dialects have lost their ancient numerals altogether, while 
 others have lost them partially, and made good their losses by new- 
 formed words. In some cases, the words particularly for one and 
 two, we may admit the original existence of synonymes, from which 
 each dialect selected its' own pecular terra. The same applies to 
 the Arian languages, for, although a comparison of Sanskrit and 
 Hindustani * numerals would convince every one how faithfully the 
 Arian dialects in general maintain their linguistic conservatism, yet 
 Sanskrit differs with regard to the words for " one," even from 
 its nearest relative, the Zend, and both from Greek and Latin. 
 The same applies to the Latin secundus, Greek devrepog, and San- 
 skrit dvitiya, — nay, perhaps to the Slavonic word for nine, — though 
 here the difference may be explained on phonetic grounds. 
 
 That there are coincidences in the numerals even between Arian 
 and Semitic languages, has frequently been pointed out ; the difficulty 
 has been to explain why these coincidences should be so palpable 
 for six and seven, and hardly perceptible in other cases. But this 
 admits of the same solution as the differences between several 
 Turanian dialects, only on a larger scale. Some numerals were re- 
 tained, and thus account for coincidences ; others were entirely lost, 
 and replaced after the separation of tribes or whole families, such as 
 the Arian and Semitic. In the Brahvi we have, according to Pro- 
 
 * Sanskrit ekadasa = Hind. 'igareA, eleven. 
 dvadasa = „ bareA, twelve. 
 unavinsati = „ 'unis, nineteen. 
 f3 
 
66 
 
 feasor Lassen's researches, a clear case of a language preserving its 
 numerals for one, two, and three, but adopting all the rest from a 
 foreign source. In the Magar language, the numerals from one to 
 five have been preserved, and the rest taken from the Parvatiya. 
 In the languages of the Dekhan, the native numerals and Sanskrit 
 numerals are used promiscuously, which in time may lead to similar 
 results. 
 
 § 4. On Phonetic Corruption. 
 
 The numerals common to several dialects of the Turanian family 
 are also instructive with regard to the extent to which phonetic 
 corruption can be carried in a nomadic state of language. The rea- 
 son why, with numerals and pronouns, the Turanian languages 
 submit to a greater amount of phonetic corruption than they would 
 tolerate in other words, is simply this, that nothing would be more 
 difficult to re-express by any composition or derivation, than the 
 simple ideas embodied in pronouns and numerals. Even where their 
 body is emaciated, and their features distorted, they are retained, 
 because even so more easily recognised by all than newly-invented 
 substitutes would be. In the Turanian numerals, therefore, if 
 compared together, we have what we could not expect to find other- 
 wise in any of these ephemeral languages, — historical deposits of the 
 progress and change of Turanian speech. While in the Arian lan- 
 guages, we may study the changes of letters, by comparing different 
 phases of one and the same dialect, — as Sanskrit and Hindustani, 
 Gothic and English, we must here rest satisfied with comparing 
 diflferent dialects, even though the respective date when each has 
 been fixed may remain indeterminate : we must compare languages 
 which perhaps stand to one another as, for instance, Pali to 
 Italian, — two Arian dialects, which, though distant in time, are 
 so analogous in their phonetic changes, that, if examined on 
 phonetic grounds only, we might take them for twins. The pos- 
 sible phonetic changes in the Turanian dialects, are, of course, to 
 their full extent, not yet determined, though much has been done 
 for this by Professor Sohott. And Professor Boehtlingk, in his 
 Yakut grammar, has succeeded in rt'ducing these phonetic changes 
 
67 
 
 to something like law and order. Sometimes they seem greater than 
 those admissible between Arian languages. Gastrin, in Lis " Dis- 
 sertatio de Aflixis personalibus," considers k = t (p. 43.). He says 
 (p. 49.) that a final t may be softened into a breathing, and this 
 breathing again be hardened into a k. He frequently considers t 
 and n interchangeable (p. 49.), and seems to hold the plural ter- 
 minations, t, k (h), je', san, saSi, la, and ', identical in origin. In 
 his Syriane grammar (§ 26.), he derives jas from as, and com- 
 pares this final s with Lapp, h, and Finn. t. Changes like these 
 may appear fanciful, and, if transitions of gutturals into dentals, 
 aspir£|,tes, and sibilants, were admitted as general principles ap- 
 plicable to every word at random, there would be an end to all 
 scientific etymology. But there is a vast difference between the 
 historical and the unhistorical application of such principles. Ar- 
 menian hayr is the same as Latin pater, not because, as a general 
 principle, p is changeable into h, but because it can be proved by 
 facts to be so in Armenian, where pes (foot) is het ; pHthu (broad) 
 is harth ; panAa, five, is hing ; irvp, fire, is hour. Again, as mater 
 becomes mayr, in Armenian and French, pater in Armenian must, 
 or at least can be hayr. If we know that languages are histori- 
 cally connected, as, for instance, Latin and French, we can state 
 as a fact, that lacryma can be changed into larme. We may even go 
 a step beyond, and say that SaKpv, tear, and larme are all derived 
 from the same root. But if, on the strength of this, we were to 
 assume that Sac could always be changed into lar, and hence identify 
 the Turkish plural lar with the Tibetan plural dag, we should no 
 longer be on historical ground, nor should we be working " in the 
 spirit of Bopp's system."* 
 
 What has been said with regard to the numerals, applies, to a great 
 extent, to the pronouns also. In the Arian languages, we know that 
 the pronouns deviate considerably from the analogy of other nouns. 
 Their terminations are called irregular, and in many cases their origin 
 and meaning cannot be deciphered even by the help of Comparative 
 Philology. The reason is, that in the declension of the pronouns the 
 
 * Cf. Hodgson, Journal of the A. S. B., 1853, p. 31, where what is meant by 
 the " spirit of Bopp's system," refers, I suppose, to Bopp's Comparative Grammar. 
 
 F 4 
 
68 
 
 Arian languages preserved some ancient relics of grammar, while in 
 the declension of nouns the power of analogy tended to eliminate 
 similar husky asperities. The pronouns being used continually, and 
 having less of a material meaning than other nouns, had become 
 fixed, formal, or inorganic, long before the rest of the grammar was 
 consolidated. -Hence, in their further dispersion, the Arian dialects 
 were unable to preserve for the pronouns the same amount of vital 
 growth which in Greece, for instance, formed the common Arian 
 grammar into its Greek type, or which in Germany gave its 
 Teutonic expression. Pronominal forms had arrived at a state of 
 grammatical numbness before the separation of tlie Arian family. 
 Hence, on the one hand, the striking similarity of pronouns in all 
 Arian tongues, and, on the other, their liability to merely phonetic 
 corruption. To this it is owing — to mere awkwardness in pro- 
 nunciation, and not to any regular modification — that Latin ego 
 becomes yo in Spanish, eu in Portuguese, io in Italian, je in French ; 
 and thus also Sk. aham, ego, became finally /in English. Yet even 
 here we can discover rules, or at least broad analogies, according to 
 which certain letters in one language are generally changed into the 
 same letters in another. We find that Sanskrit s becomes Zend h, 
 and Sanskrit h becomes Zend z ; therefore the change of Sk. sahasra, 
 thousand, into Zend hazanra is perfectly regular. According to the 
 same analogy, Sk. aham, I, must in Zend be azem ; and as in Ar- 
 menian this Zend z is frequently represented by s, there is nothing 
 irregular in the Armenian * es, I ; nor shall we be obliged to go to 
 Mongolian dialects in order to explain the Ossetic az, I, whatever 
 Tataric or Tartaric scholars may say to the contrary. 
 
 The Turanian languages, tliough they preserved the vitality of their 
 grammar to a much larger extent than any Arian dialect, yet were 
 unable to avert altogether the same disorganizing influence from 
 their pronouns. Some of their pronominal forms are therefore en- 
 tirely Arian in principle, that is to say, anomalous and unintelligible ; 
 and what has generally been considered (wrongly, as has been shown) 
 a distinguishing feature of Arian grammar, that " by some unknown 
 process forms are evolved from the body of a noun like branches of 
 
 * See Windischmann'e classical Essay " On the Arian Basis of the Armenian." 
 
69 
 
 a tree springing from the stem," would in this case seem to apply 
 with real force to the Turanian languages. If we take Turanian 
 grammar, even in its least developed state, we find, for instance, in 
 Mandshu, forms which, so far as the principle of their formation is 
 concerned, would have to be pi-onounced Arian, according to Schlegel's 
 definition of this term. We find bi, I ; mini, mine ; be, we ; si, thou ; 
 soue, you ; i, he ; fche, they ; that is to say, we find different bases 
 for the same, pronoun, and different forms of the same base produced, 
 not by agglutination, but by what has been called a principle of 
 " inward growth.'' What difference, as far as the principle of de- 
 clension goes, is there between Greek 6 changed in the plural to of, 
 and Mandshu bi, I, changed in the plural to be, we ?* Many similar 
 cases will be seen in an appendix containing a comparative list of 
 pronouns. It is hopeless to attempt to discover in these inorganic 
 forms the elements of agglutination. The same applies to the dis- 
 tinction of gender, which, though in most cases marked by additional 
 syllables, whether nominal or pronominar, is sometimes expressed 
 in such a manner that we can only explain it by ascribing an ex- 
 pressive power to the more or less obscure sound of vowels. Ukko, 
 in Finnic, is an old man ; akka, an old woman (in Canarese, akka, 
 elder sister). In Mandshu, chacha is mas (Mong. acha, Turkish agha, 
 elder brother, uncle) ; cheche, femina. Again, araa in Mandshu is 
 father, eme, mother ; amcha, father-in-law (Mongol, abagha), emche, 
 mother-in-law (Mongol, emeke, grandmother). The same change 
 of vowels expresses in other languages remoteness or proximity, as in 
 Canarese, where "ivanu" is hie, "avanu" ille, and where, according 
 
 ♦ Mr. Hodgson, for instance, analyses the Mandshu tese, they (or, as he writes, 
 te-se-t) into te, he, and se, thou; and he denies that in Mandshu the plural can he 
 formed by an additional se, because it is not always formed so, and because, as he 
 says, a regular pluralizing particle would be uniformly applied and wear one shape. 
 Now, this is not quite true either in Arian or Turanian grammar, and particularly 
 not with regard to pronouns. Sivas in the plural makes sivas ; sarvas makes sarve ; 
 ego makes nos ; Mandshu bi makes be. But in tese, se certainly seems the regular 
 plural termination, only that after nouns it is restricted to words expressive of living 
 beings. Tiras, dchoui, child, makes dchouse, children; wang, ting, uiangsa, kings; 
 morin, horse, morisa, horses. (See Gabelentz, § 24). The se in tese is most likely, 
 therefore, the same se which we 6nd in ese, hi, from ere, hie ; and not the pronoun 
 of the second person glued to that of the first, as Mr. Hodgson supposes. (J. A. 
 S.B.I 853, p. 69. seq.) 
 
70 
 
 to Weigle, there existed formerly a third intermediate pronoun, 
 uvanu. 
 
 What -we have here said proves that in the Turanian languages 
 also, a greater allowance should be made for phonetic influences, 
 whether accidental, as in phonetic corruption, or intentional, as in 
 phonetic distinctions. Though our conviction may be that in an 
 earlier state of language these formal changes also had a material 
 origin, yet their analysis must baffle all ingenuity, and shows the 
 truth of the saying, "Boni grammatici est nonnulla etiam nescire." 
 
 § 5. On scarce Words. 
 
 After considering words which are of daily use an(f frequent oc- 
 currence, and which, therefore, even in so porous a state of society 
 as that of nomadic hordes, have a chance of remaining on the sur- 
 face, we have still for a moment to bring before ourselves the effects 
 which the same state of society would have on words of rare occur- 
 rence. Even at the present day, with all the speaking, preaching, and 
 reading we have to undergo, many men never use half the words 
 which belong to their own language. Writers, again, are so little 
 aware occasionally of the existence of certain words in their own 
 language, that they coin new ones, though there is really no demand 
 for them. If the new, however, become current, the old are melted 
 down altogether, unless preserved in dictionaries, or revived by 
 new editions of old books. But let us think for a moment of all 
 the changes and chances of nomadic tribes, — of the small sphere 
 of ideas and words in which their language moves permanently and 
 continuously, — of the little support which expre.ssions of a higher 
 range, or names of a poetical tinge, though used once or twice by 
 a poet or a king, would receive in Asiatic steppes, where men 
 spend their life between hunting, fighting and eating, and women 
 are kept only for breeding children and feeding cattle ! It is rather 
 surprising, that so many words should have remained for centuries 
 in the sieve of languages like the Mongolians; and we have no right 
 to expect that between tribes separated probably as early as any of 
 the Arian nations, words belonging to the higher ranges of thought 
 should be found to agree entirely. 
 
71 
 
 Seventh Section. 
 On Turanian Languages approaching to an Arian Type. 
 
 § 1. Arian Elements in Hungarian, Turkish, Finnish. 
 
 If the unsettled state of grammar and dictionary in the Turanian 
 languages is the result of that nomadic state of society in which they 
 grew up and live, we should expect that this effect would cease 
 whenever nomadic races enter into a state of political consolidation. 
 This is the case to a certain extent. Wherever there is a written lite- 
 rature and fixed standard of grammar kept up by the higher classes, 
 the Turanian character approaches more and more to an Arian type. 
 For the same reason, we expect a larger number of formal coincidences 
 between Hungarian and Turkish, or between Hungarian and Finnish, 
 than between the Samoieds of the Lake Altin and the Aimaks of Persia. 
 In Turanian languages which have received a literary cultivation, 
 as Finnish, Turkish, and Hungarian, forms occur which are cor- 
 rupted into something very much like inflection : and here the 
 separate stones of the grammatical mosaic can hardly now be 
 taken to pieces. Irregular forms become frequent, and words partake 
 more of a conventional and historical than of an etymological cha- 
 racter. We see Jiere how a Turanian may nearly become an Arian 
 language ; and, in looking at the earliest specim.ens of Arian gram- 
 mar, such as Sanskrit, we may observe in an Arian language traces 
 of an evanescent Turianism. In Sanskrit, although grammatical 
 forms have been regulated and reduced by a sound economy, instances 
 occur of superfluous distinctions, successfully comprehended by the 
 Greek genius within more general categories. In Finnish, for in- 
 stance, every imaginable relation of noun to noun and noun to verb 
 can be expressed by what is called a case termination. We find a 
 different suffix for the objective ease when I beat a child, or when 
 I strike it on a certain part of its body, — resembling thus the Greek 
 genitive and accusative after verbs of a similar meaning. There 
 are no less than fifteen cases in Finnish, and yet no pure accusative ! * 
 
 * Mr. Hodgson makes a similar remark with regard to the verb : "A Tatar,'' 
 he says, (J. A. S. B., 1853, p. 129.) " cannot eiidure that confusion of the preca- 
 tive, optative, and imperative which our imperative mood exhibits. But he 
 
72 
 
 nek- em, to me. 
 en-g-em-et, me(oreng-em). 
 ert-em, on my account, 
 vel-em, with me. 
 
 All these cases are expressed by suffixes, some even by compound 
 suffixes, to exhibit more complicated relations. The following table 
 will give an idea of Hungarian declension : Ms, as we saw before, was 
 knife ; kesem, my knife. This is declined : 
 
 1. Kesem, my knife. en, I. 
 
 2 Kesemnek, of my knife. 
 
 3. Kesemnek, to my knife. 
 
 4. Kesemet, my knife. 
 
 5. Kesemert, on account of my knife. 
 
 6. Kesemmel, with my knife. 
 
 7. Kesemme, toward my knife. 
 
 8. Kesemiil, as my knife. 
 
 9. Kesemkent, like my knife. 
 
 10. Kesembe, into my knife. 
 
 11. Kesemben, inside my knife. 
 
 12. Kesembol, from within my knife. 
 
 13. Kesemre, upon my knife (coming) 
 
 14. Kesemen, on my knife (resting). 
 
 15. Kesemrol, down from my knife. 
 
 16. Kesemhez, toward my knife. 
 
 17. Kesemnel, near my knife. 
 
 18. K^semtol, away from my knife. 
 
 19. Kesemig, as far as my knife. 
 
 It is true that many of these terminations are only postpositions, 
 and might therefore be compared rather with the prepositions 
 than with the case terminations of the Arian languages. Yet the 
 case is somewhat different. The noun, together with these post- 
 positions, forms, in Hungarian, a phonetic unity ; it has but one 
 accent, and the harmony of vowels connects the two still more 
 closely. The real difference is this, that the Arian case terminations 
 can no longer be used separately, while many of these postpositions 
 occur as prepositions also. This may be seen in looking at the de- 
 clension of the personal pronoun in Hungarian, which, therefore, I 
 have put side by side with the nominal paradigm. 
 
 remedies the defect, not by the multiplication of grammatical forms, but by the use 
 of distinct -words, or distinct multiplications of the same -word. Thus, Davo, so- 
 licits, Davong, commands, et sic de cseteris. 
 
 benn-em, in me. 
 bil-em, in me. 
 r-am, upon me. 
 
 rol- am, from me. 
 hozzam, toward me. 
 nal-am, near me. 
 tol- em, away from me. 
 
73 
 
 § 2. Turanian Elements in Sanskrit. 
 
 As we see the tendency of the Arian languages to reduce the 
 variety of their terminations, we may suppose that even the richest 
 grammatical language, the Sanskrit, was, at a period previous to 
 the Vaidik, and beyond our knowledge, richer still. In the dual, for 
 instance, the genitive and ablative might each have had a distinctive 
 form, as in the singular ; and the same power of concentration, ab- 
 straction and method, which made the Greek feel satisfied with two 
 cases in the dual, may have led the Hindu to divest himself of what 
 he began to feel as an "embarras de richesse." After a time, how- 
 ever, this sound economy of the Arian languages seems to lead to 
 an involuntary meagerness. By causes quite unintentional — cor- 
 rupt pronunciation, for instance — cases become identical, and are 
 no longer distinguishable even where their distinction is necessary 
 for logical purposes. A principle reappears then at work in modern 
 languages, which apparently may be called Turanian, — the prin- 
 ciple of periphrastic, or, as it has been called, analytical formation. 
 The phrase " de illo philosopho," the French " du philosophe," 
 instead of "philosopho," is to a certain extent Turanian, though 
 not entirely, because the distinguishing words are put before, not 
 after the word they determine. Its modern contraction again, " du 
 philosophe," is not purely Arian. Du does not stand to le in the 
 same relation as tov to 6. Du, instead of " de illo," is produced by 
 a corruption of words which had before been articulated grammati- 
 cally ; — it is the remnant of a phrase ; while tov* is the corruption 
 of a compound, the component parts of which were pure radicals, not 
 yet determined by grammatical terminations. The same applies to 
 the periphrastic form "faimer-ai," I have to love, which even in 
 its contraction y'aj'merai can only be called quasi-Turanian, because it 
 rests on a diflPerent principle of formation from that which pro- 
 duced ama-bo. There is a distinction between these secondary Arian 
 
 * The Greek Genitive toS stands for roto, toTo for too-io, the Sanskrit tasya. In 
 tasya, fa is the pronominal hase, sya a suffix which forms Genitives and Adjec- 
 tives. Thus Prof. Bopp compares somewhere Srnioa-to, the original form of the 
 Greek genitive, which afterwards hecomes 5^;ioio and Sliiiov, with the Adjective 
 Srifi6atos, 
 
74 
 
 and the primary Turanian formations, as there is also a vast differ- 
 ence between the reduced state of Arian grammar in the middle 
 ages and the undeveloped state of Turanian grammar in the Tun- 
 gusic and Mongolic branches. 
 
 § 3. Ascending Scale in the Turanian Languages. 
 
 There is an ascending scale in the grammatical life of Turanian 
 languages, running nearly parallel with the political and literary posi- 
 tion of these nations. This has been pointed out by Schott and 
 by Castren. The Tungusic branch is the lowest ; its grammar is not 
 much richer than Chinese, and in the structure there is an absence 
 of that architectonic order which in Chinese unites the Cyclopean 
 stones of their language without further cement. This applies, how- 
 ever, principally to the Mandshu ; other Tungusic dialects spoken, 
 not in China, but in the original seats of the Donkis, are said to 
 be richer in form. The Mongolic dialects excel the Tungusic, but, 
 particularly in their written language, the different members of 
 speech are hardly as yet articulated. The spoken idioms of Tungu- 
 sians, as well as Mongolians, are evidently still struggling towards a 
 more organic life. Professor Schott's remark, " that the Turanian 
 verb which in Mandshu and Mongolian seems, as it were, inani- 
 mate, receives its life only in Turkish, by means of a connection of 
 roots and pronouns," requires modification, since Castren brought 
 evidence of an incipient life in the grammar of the Buriats and the 
 dialect of Nyerchinsk. The mere juxtaposition of a pronoun and a 
 root, as we find it in Mandshu : 
 
 bi khoachambi, 1 feed, 
 
 si khoachambi, thou feedest, 
 
 ere niyalma kh6achambi, this man feeds, 
 
 is hardly as yet grammatical. But Castren assures us that instead 
 of the invariable khoachambi through all the persons and numbers, 
 he heard among Tungusic tribes distinctly the following termi- 
 nations : 
 
 Singular. Plural. 
 
 12 3 12 3 
 
 u, f. s. n. wun. sun. 1. 
 
Predicatives. 
 
 
 
 Plural. 
 
 I. 
 
 II. 
 
 bida. 
 
 ta. t. 
 
 Possessives. 
 
 
 manai 
 
 tani. 
 
 75 
 
 These terminations are radical pronouns, and in the Tungusic dia- 
 lect 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 Mongolia dialect, in which Castren ob- 
 served the same tendency, had advanced another step, for it made 
 also formal distinction between possessive and predicative, affixes. 
 These are : 
 
 Singular. 
 I. II. III. I. II. III. 
 
 p, m. s'. c'. 
 
 m. s'. c'. n (ni) manai tani. n (ni.) 
 
 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 no- 
 minative bi. 
 
 All this, however, is but a small beginning, particularly if we 
 compare the profusion of grammatical stores which the Turkic lan- 
 guages 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 lan- 
 guages, richer in declension than the Turkic, consists, as far as 
 the verb is concerned, rather in a diminution than increase of forms. 
 Castren says : " Progrediente in dissertatione apparuit affixa per- 
 sonalia in Unguis Burjatica et Tungusica inchoata adhuc esse et 
 quasi nascentia, in Turcicis vero jam forma uti perfectiore magisque 
 explanata, in Finnicis demum et Samojedicis Unguis summum evolu- 
 tionis 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 jevivrai, i.e. ego (aham) vivere (^iv-as-S, dat, 
 neutr.) habeo (bhav. aya. mi), we have a number of articulated forms, 
 
76 
 
 resolved as it were again into simple matter, while in the Tungusic 
 verb, grammatical form is produced for tlie first time by the mere 
 connection of material elements. 
 
 Eighth Section. 
 
 Evidence of the common Origin of the Turanian Languages 
 summed up. 
 
 If 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 Castren, 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 reason- 
 able 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, Castren has proved in some cases 
 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 con- 
 jugation but the declension, nay, even the syntax of these dialects. 
 As to the other grammatical elements, postpositions I mean princi- 
 pally, or similar particles, they also exhibit salient coincidences in 
 some points, while their diversity on others does not mean more 
 
77 
 
 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 dela, 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 gramma- 
 tical 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 kesnek and kes-em-nek. In the second case, as a general rule, the 
 governed or determining always precedes the governing or deter- 
 . mined word. Therefore prepositions governing a noun are impos- 
 sible in Turanian languages. Conjunctions are scarce, the connec- 
 tion 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 inter- 
 fered, is a family feature not less strongly marked. It can only be 
 compared with the triliteral character of the Semitic, or the pecu- 
 liar 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 de- 
 scribes 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, ex- 
 ceedingly difficult for an European practically to distinguish. On a careful exa- 
 mination, however, it will be found that these tones do not in reality exceed/oar, 
 and that they are the same as those described by Chinese philologists. 
 
 " Th-s 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. 
 
?8 
 
 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 n*ore prac- 
 tised to imitate it. This law exists in the Tungusic, Mongolia, Tataric, 
 and Finnic classes, thougli it does not influence all their dialects with 
 equal force. 'I'races 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 lead- 
 ing 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, 
 Klnproth, Eemusat, Gabelentz, and Schott, and as a tradition is cer- 
 tainly curious, because it shows that even in later times, when Mon- 
 golic and Tataric had by mistake become the names of two races, 
 diiFering 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 Klian 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.) 
 
79 
 
 Chinese writers, was originally a corruption of Tura, Turvasa, Tur- 
 ushka, all names given by the Arians to equestrian Nomads and Indo- 
 Scythian 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 Turan to Tur, Iran to Irij, 
 and Eum and Khawer to Silim. Irij is killed by his brothers ; but 
 the kings of Persia descend alternately from the three brothers, — 
 Menuchihr being an Iranian,. Afrasiyab 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 techiiical 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, Gastrin, 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 recom- 
 mending caution and more definite argumentation, they cannot be 
 held to invalidate the conclusions arrived at in common by men 
 like Rask, Gabelentz, Schott and Gastrin. 
 
 If the principles here laid down are considered valid for esta- 
 blishing the relationship of languages, I am inclined to maintain that, 
 similarly with these five classes, Finnic, Samoiedic, Tataric, Mongolic, 
 and Tungusic, the Tamulic, Bhotiya, Tai, and Malay languages also 
 belong to the same Turanian race. 
 
 a 2 
 
80 
 
 SECOND CHAPTER. 
 
 on the turanian character of the tamulio languages. 
 
 First Section. 
 The Arian Settlers and Aboriginal Races of India. 
 
 The name by which the whole class of the aboriginal languages 
 of India is best known to us, was given it by the Brahmans. 
 " D e k h a n" is a corruption of the Sanskrit "dakshira a," which 
 means " right " (dexter). To the Brahman who, in fixing his posi- 
 tion, always imagined himself looking toward the rise of the sun, 
 whatever lay to the south of his own country, was "dakshiraa" or "to 
 the right." As the frontiers of the Brahmanic settlements were 
 gradually extended, the meaning of Dakshiwa or Dakshiwapatha 
 became more definite, till at last the chain of the Vindhya-mountains 
 was fixed upon as the natural frontier between what the Brahman 
 called his holy-land and the Dekhan. It is now generally admitted 
 that this holy-land of the Brahmans, even within its earliest and 
 narrowest limits, between the Sarasvati and Dnshadvati, was not the 
 birthplace of the sons of Manu. The Arians were strangers in the 
 land of the Indus and Ganges, but no one can now determine the 
 exact spot whence they came and where they had been previously 
 settled. Traditions, current among the Brahmans as to the 
 northern regions, considered the seats of the blessed, may be con- 
 strued into something like a recollection of their northern immigra- 
 tion — holy places along the rivers of Northern India, where even in 
 later times Brahmans went to learn the purest Sanskrit, may mark 
 the stations of their onward course — the principal capitals of their 
 ancient kingdoms may prove the slow but steady progress toward 
 the mouths of the principal rivers of India — but with the sources 
 of those rivers the homes of the Arian strangers vanish from our 
 sight, even after we have reached the highest points of view acces- 
 sible on Indian ground. 
 
81 
 
 The countries which the Bralimans took possession of, or rather 
 over which they gained their priestly ascendancy, were inhabited by 
 races of men, who are sometimes represented to us by the Brahmans 
 as mere monkeys or bush-men, sometimes as uncouth giants, some- 
 times, as in the case of Bribu and Hanuman, as useful allies and 
 faithful servants. In the social scheme of the Brahmans, however, 
 these races could never rise beyond the position of a >Sudra. Excep- 
 tions like that of the Eibhus or Rathakaras, are very scarce and con- 
 fined to the Vaidik age. No Sudra again, as long as Manu's laws 
 prevailed, could ever rise to the dignity of a twice-born man, and 
 though even as a 5'udra, he had caste, yet the distance between him 
 and the poorest Brahman was so wide and unsurmountable in the 
 eyes of both parties, that we can only explain it by a difierence of 
 race, such as we find between the Spaniard and the Negro. 
 
 In ancient times the distinction between the twice-born Arians 
 and the iS'udra was probably a distinction of colour also. The 
 very name of caste in Sanskrit is varwa, colour. Distinctions of 
 colour, however, fade away and sometimes disappear altogether, 
 even in despite of such barriers as the strict "lex connubii," 
 interposed between the difierent ranks of Hindu society. Besides, 
 these laws were not always observed, nor similarly respected in 
 different parts of India. India was conquered and devastated several 
 times — Greeks, Scythians, Arabs and Mongolians, mingled their blood 
 with that of the conquered race, and as the priesthood and their 
 nobility lost strength, it was easier even for the lowest ranks to 
 claim a position, secured not by birth, but by wealth and power. 
 Again, there is that long interval in the history of India, during 
 which caste, at least in its religious sense, was altogether ignored. 
 As long as Buddhism was the state religi-on of a great portion of 
 India, that is to say from the third century before, to perhaps the 
 sixth century after Christ, the different ranks of society could only 
 be held apart by social prejudice and custom, and not by priestly 
 authority. But in spite of all these changes and social commotions, 
 the traveller in India to the present day, though he would look in 
 vain for the distinctive features of a Brahman, a Kshattriya, or a 
 Vai«ya, feels the conviction irresistibly growing upon him, r,s he 
 passes along the streets of cities, or the roads of villages, whether 
 
 G 3 
 
82 
 
 north or south of the Vindhya, that everywhere he is brought in 
 contact with at least two races of man, distinct in mind as well as in 
 body. "No sojourner in India," says Dr. Stevenson, in the Journal of 
 the Bombay Branch, January 1852, " can have paid any attention to 
 the physiognomy of the higher and lower orders of natives without 
 being struck with the remarkable difference that exists in the shape 
 of the head, the build of the body, and the colour of the skin, be- 
 tween the higher and lower castes into which the Hindu population 
 is divided. The high forehead, the stout build, and the light copper 
 colour of the Brahmans and other castes allied to them, appear in 
 strong contrast with the somewhat low and wide heads, slight make, 
 and dark bronze of the low castes." 
 
 The name of "Dekhan languages," to signify the non-Arian 
 dialects of India, is therefore inconvenient in one respect. According 
 to its etymological and geographical meaning, it can only refer to 
 nations and languages to the right of the Vindhya, while we evi- 
 dently want a name sufficiently comprehensive to stand for all ab- 
 original inhabitants of India, wherever they are met with, from the 
 Snows to Cape Comorin. Our highest living authority and best in- 
 formant on the ethnology and phonology of the native races of India, 
 Mr. B. H. Hodgson, of Darjiling, uses "Tamulian" as the general 
 name for all ncn-Arian races. I have adopted this name, though it 
 is not altogether free from objections, because it may be used in 
 three different meanings. Originally it would mean one of the lan- 
 guages in the Dekhan, the Tamil ; secondly, the Dekhan languages 
 in general ; and thirdly, a 1 1 the aboriginal dialects of India. Mr. 
 Hodgson himself uses it in the second and third senses. I should 
 prefer, therefore, as a general name for all the native languages of 
 India, Nishada-languages. Nishada is the oldest name given by 
 the Brahmans to tlieirnon-Arian neighbours. Itmeans Assiduusor 
 Ansassig, and is therefore the most appropriate name for people 
 who occupied the soil of India, before they were dispersed by the 
 Arians. It is tfue the word Nishada does not occur in the Rigveda, 
 but at the time of Yaska, in the fourth century b. c, the "five races," 
 frequently mentioned in the Veda, are always explained as the four 
 castes and the Nishadas. In the Brahmawas also and in the epic 
 poems, the word occurs as a general term together with Mle^Aa. 
 
83 
 
 " Tamulic " might, if this were used, be retained as the general name 
 of the languages now principally spoken south of the Vindhya.' 
 
 Historical Traces of Nishadas, or aboriginal Races in India. 
 
 On the ethnological state of India during the Vaidik periods, it is 
 very difficult to form a correct opinion, because the scanty allusions 
 to this subject which occur in the hymns are at variance with one 
 another in different portions of the Rigveda. It is a fact, that the 
 four castes existed previous to the collection of the Rigveda ; — and 
 
 * The materials which I have used are almost entirely contained in the Jour- 
 nal of the Asiatic Society of Bengal. I subjoin a list of the articles to which I 
 shall have most frequent occasion to refer : 
 
 Vol. 1847. p. 1235. B. H. Hodgson, On the Aborigines of the Sub-Himalayas ; 
 p. 1245. B. H. H., Comparative Vocabulary of the several Languages and Dialects 
 of the Eastern Sub- Himalayas, from the Kali or Ghogra to the Dhansri (Suban- 
 shiri ?). 
 
 Vol. 1848. 1. p. 73. Addenda and Corrigenda of the paper on Aborigines, etc.; 
 p.544. B. H. H. Ethnography and Geography of the Sub-Himalayas. 
 
 Vol. 1848. 2. p. 222. B. H. H. On the Tibetan Type of Mankind ; p. 550. B. 
 H. H. The Aborigines of Central India ; p. 650. B. H. H. On the Chepang and 
 Kusundu Tribes of Nepal. 
 
 Vol. 1849. 1. p. 238. B. H. H. A Brief Note on Indian Ethnology; p. 350. B. 
 H. H. Aborigines of Southern India; p. 451. B. H. H. On the Aborigines of 
 North Eastern India. 
 
 Vol. 1849. 2. p. 702. B. H. H. On the Origin of the Kocch, Bodo and Dhimal 
 Tribes ; p. 761. B. H. H. On the Physical Geography of the Himalayas; p. 967. 
 B. H. H. On the Aborigines of the Eastern Frontier. 
 
 Vol. 1850. 1. p. 309. B. H. H. Aborigines of the North East Frontier ; p. 461. 
 B. H. H. Aborigines of the South. 
 
 While engaged in carrying this Essay through the press, I had the pleasure of 
 making Mr. Hodgson's personal acquaintance in England, and I received at the 
 same time his two important articles pubhshed in the Asiatic Journal of Bengal, 
 1853, Nos. I. and II. 
 
 Besides Mr. Hodgson's articles we find in the same Journal some very useful 
 Essays by W. Robinson. " Notes on the Languages spoken by the various Tribes 
 inhabiting the Valley of Asam and its Mountain Confines," vol. 1849. 1. p. 183. 
 and 310. 
 
 Mr. Walter Elliot's Observations on the language of Goands, published as 
 early as November 1847, in the same Journal, are well known, and have been 
 honoured by a translation by Professor Lassen. 
 
 The Bev. J. Stevenson's articles are principally published in the Journal of the 
 Bombay Branch of the Royal Asiatic Society. 
 
 o 4 
 
84 
 
 therefore previous to any other written authority in India, which 
 might be quoted to disprove their early existence. The hymn in 
 the tenth Mawrfala, where the castes are mentioned with their tech- 
 nical names, though it may have a modern appearance, if compared 
 with other hymns, is still the most ancient authority we can appeal to, 
 and more ancient than any hymn in the other collections, oranyBrah- 
 maraa or Sutra. And further the four social ranks, priests, warriors, 
 house-holders and servants, are clearly distinguishable in many of the 
 hymns of the Rigveda, and in the Brahmawas the (Sudra also is 
 mentioned by name. Though he belongs to a caste, and there- 
 fore has rights as well as duties, he is distinctly called non-Arian, 
 for Aryas, as the iS'atapatha-brahmaMa says, are only Brahmans, 
 Kshattriyas and Vaisyas. In addition to these four castes, who 
 formed the body politic in India as early as the times of Vasish<Aa 
 and Visvamitra, we find in the hymns frequent allusions to the 
 Dasyus. Dasyu means simply enemy, for instance, when Indra is 
 praised because "he destroyed the Dasyus and protected the 
 Arian colour."' The "Dasyus" in the Veda may mean non-Arian 
 races in many hymns ; yet the mere fact of tribes being called 
 enemies of certain kings or priests, can hardly be said to prove their 
 barbarian origin. Vasish^Aa himself, the very type of the Arian 
 Brahman, when in feud with Vi.svamitra, is called not only an enemy, 
 but a "Yatudhana," and other names which in common parlance are 
 only bestowed on barbarian savages and evil spirits. We still have 
 the very hymn in which Vasish^^a deprecates such charges with 
 powerful indignation. He says: 
 
 " If I had worshipped false gods, or if I had called upon the gods 
 in vain — But why art thou angry with me, o Gatavedas? May 
 vain talkers fall into thy destruction." 
 
 " May I die at once, if I be a Yatudhana, or if I hurt the life of 
 any man. But may he be cut off from his ten friends, who falsely 
 called me a Yatudhana.'' 
 
 " He who called me a Yatudhana, or who said I am a bright devil 
 — mav Indra strike him down with his great weapon, may he fall 
 the lowest of all beings." 
 
 In other passages, the word also which I have here translated by 
 devil (rakshas), is clearly applied to barbarous nations. Originally 
 
85 
 
 rakshas meant strong and powerful, but it soon took the sense of 
 giant and barbarian, and in this sense it occurs in the Veda together 
 with Yatudhana. 
 
 Another Vaidik epithet applied as it seems to wild tribes, infesting 
 the seats of the Aryas, is " anagnitra," they who do not keep the flre. 
 Thus we read, " Agni, drive away from us the enemies, — tribes who 
 heep no sacred fires came to attack us. Come again to the earth, 
 sacred god with all the immortals, come to our libation." 
 
 The same races are called " Kravyftd," or flesh-eaters. In a 
 famous hymn of Vasish<Aa we read : " Indra and Soma, burn the 
 Rakshas, destroy them, throw them down, ye two Bulls, the people 
 that grow in darkness. Hew down the madmen, suffocate them, kill 
 them, hurl them away and slay the voracious." 
 
 " Indra and Soma, up together against the cursing demon ! May 
 he burn and hiss like an oblation in the fire ! Put your everlasting 
 hatred upon the villain, who hates the Brahman, who eats flesh, and 
 whose look is abominable. 
 
 " Indra and Soma, hurl the evil-doer into the pit, into unfathomed 
 darkness. May your strength be full of wrath to hold out, that no 
 one may come out again." 
 
 Kravyad, flesh-eater, means people who eat raw meat, Kpco(j>a.yoi, 
 and they are also called amadas, &fj.o(j>ayoi, or raw-eaters, for the cook- 
 ing of meat was a distinguishing feature of civilized nations, and 
 frequently invested with a sacrificial character. Agni, who in the 
 Veda is the type of the sacrifice, and with it of civilization and social 
 virtues, takes an entirely different character in his capacity of 
 "Kravyad," or flesh-eater. He is represented under a form as 
 hideous as the beings he is invoked to devour. He sharpens 
 his two iron-tusks, puts the enemies into his mouth and swallows 
 them. He heats the edges of his shafts, and sends them into the 
 hearts of the Rakshas. He tears their skin, minces their members, 
 and throws them before the wolves to be eaten by them or 
 by the shrieking vultures. These Rakshas are themselves called 
 " aAitas," mad, and " miiradevas," worshippers of mad gods. Nay 
 they are even taunted with eating human flesh, and are called 
 " asutnpas," as enjoying the life of other men. In the Rigveda, we 
 read, " The Yatudhanas who gloat on the bloody flesh of men or 
 
86 
 
 horses, and steal the milk of the cow, o Agni, cut off their heads with 
 thy fiery sword." 
 
 All these epithets seem to apply to hostile, and most likely 
 aboriginal races, but they are too general to allow us the infer- 
 ence of any ethnological conclusions. The Vaidik Rishis certainly 
 distinguish between Arian and non-Arian enemies. The gods are 
 praised for destroying enemies, Arian as well as barbarian (dasa Aa 
 vHtra hatam, aryawi ka), and we frequently find the expression, 
 " Kill our Arian enemies, and the Dasa enemies, yea, kill all our 
 enemies." But there is no allusion to any distinct physical features 
 such as we find in later writings. The only expression that might 
 be interpreted in this way is that of " susipra," as applied to Arian 
 gods. It means " with a beautiful nose." As people are fain to 
 transfer the qualities which they are most proud of in themselves, to 
 their gods, and as they do not become aware of their own good qua- 
 lities except by way of contrast, we might conclude that the beautiful 
 nose of Indra was suggested by the flat-noses of the aboriginal races. 
 Tribes with flat or with even no noses at all, are mentioned by Alex- 
 ander's companions in India, and in the hymns of the Rigveda Manu 
 is said to have conquered Vi-«sipra (Pada-text, visi-«pra), which 
 may be translated by " nose-less." The Dasa or barbarian is also 
 called vrisha^ipra in the Veda, which seems to mean goat or bull- 
 nosed, and the " Anasas" enemies whom Indra killed with his weapon 
 (Rv. V, 29, 10), are probably meant for noseless (a-nasas), not, as the 
 commentator supposes, for faceless (an-asas) people. 
 
 In the BrahmaMas, which represent a new period of Vaidik litera- 
 ture, the Nishadas occur under more distinct features. In the 
 Aitareya-brahmawa, they are once mentioned in the same category 
 with thieves and criminals, who attack men in forests, throw 
 them into wells, and run away with their goods (Nishada va, Se/aga 
 va, papakrito va). 
 
 In some of the later Brahmareas also, the Pan^avin^a, for in- 
 stance, the Nishadas occur, and we there find, that they now live 
 not only in forests but in villages. But there also, they are dis- 
 tinct from the castes as well as from the great mass of the people, 
 the latter, though not under Brahmanic discipline, being yet con- 
 sidered as of Arian origin. This latter class, the Vratyas, are de- 
 
87 
 
 scribed as differing from the Brahmanic laity in laws, customs, and 
 pronunciation, but not in language. They could be readmitted into 
 the Brahmanic community after performing certain rites and pe- 
 nances prescribed by law. Their name is Vratya, but never Nishada 
 In the Taittiriya-brahmawa, we find after the four castes (Brahmaraa, 
 Ra^anya, Vaisya and /Sudra), other names, such as Magadha, Sailusha, 
 Naishada, Vratya, Kaivarta, Kirata, Kkndhlsk, etc., but again no 
 description of their physical peculiarities. 
 
 This is very different in later works. In the Vishwu-pura«a (page 
 100, ed. Wilson), the type of the Nishada is given, — " a being of 
 the complexion of a charred stake, with flattened features, and of 
 dwarfish stature." The inhabitants of the Vindhya mountains are 
 called his descendants. According to the Matsya-puraraa, they were 
 as black as coUyrium. According to the Bhagavata-pura?!a, they had 
 short arms and legs, were black as a crow, with projecting chin, 
 broad and flat nose, red eyes, and tawny hair. The Padma-purawa 
 adds a wide mouth, large ears, and a protuberant belly, and particu- 
 larises their posterity as Kiratas, Bhillas, Bahanakas, Bhramaras, 
 and Pulindas. 
 
 From the most ancient times therefore to the period of the Purawas, 
 we meet everywhere with indications, more or less distinct, of two 
 races brought into contact in the Indian peninsula. A most vivid 
 description of their physical peculiarities at the present time is given 
 by Mr. Hodgson. In one of his articles published in the Journal of 
 the Asiatic Society of Bengal (1849, p. 710), he writes : — 
 
 " A practised eye will distinguish at a glance between the Arian 
 and Tamulian (i. e. Nishada) style of features and form — a practised 
 pen wiU readily make the distinction felt — but to perceive and to make 
 others perceive, by pen or pencil, the physical traits that separate 
 each group or people of Arian or of Tamulian (Nishada) extraction 
 from each other group, would be a task indeed ! In the Arian form 
 there is height, symmetry, lightness and flexibility : in the Arian face 
 an oval contour with ample forehead and moderate jaws and mouth, 
 a round chin, perpendicular with the forehead, a regular set of distinct 
 and fine features ; a well-raised and unexpanded nose, with elliptic 
 nares ; a well-sized and freely opened eye, running directly across the 
 face ; no want of eye-brows, eye-lash, or beard ; and lastly, a clear 
 
88 
 
 brunet complexion ; often not darker than that of the most southern 
 Europeans. 
 
 " In the Tamulian (Nishada) form, on the contrary, there is less 
 height, less symmetry, more dumpiness and flesh : in the Tamulian 
 face, a somewhat lozenge contour caused by the large cheek 
 bones ; less perpendicularity in the features to the front, occasioned 
 not so much by defect of forehead or chin, as by excess of jaws and 
 mouth ; a larger proportion of face to head, and less roundness in 
 the latter ; a broader, flatter face, with features less symmetrical, but 
 perhaps more expression, at least of individuality; a shorter, 
 wider nose, often clubbed at the end and furnished with round nos- 
 trils ; eyes less, and less fully opened, and less evenly crossing the 
 face by their line of aperture ; ears larger ; lips thicker ; beard defi- 
 cient ; colour brunet as in the last, but darker on the whole, and, as in 
 it, various. Such is the general description of the Indian Arians and 
 Turanians." 
 
 In other places Mr. Hodgson undertakes indeed to give some charac- 
 teristic marks by which the principal sub-divisions of thisNon-Arian, 
 or Nishada, stock might be distinguished in different parts of India. 
 But though they would suflice to indicate at once the Nishada in the 
 Dekhan or in the jungles of Gondvan, in the slopes of the Vindhya or 
 in the valleys of the Brahmaputra, in the Tarai or in the G-hats of the 
 Himalaya, from his Arian neighbour, they are hardly sufficient to 
 separate the Tamulian proper from the Kol, the Kol from the Garo, 
 the Garo from the Lepcha, the Lepcha from the Bhotiya. Mr. 
 Hodgson also, admits, in several places, that, on the whole, there is 
 but one stamp impressed on all the Aborigines of India, that will 
 admit of scientific definition. This stamp, he says, is the Mongolian 
 " Look steadfastly at any man of an aboriginal race (an ubiquitarian 
 Dhanger for instance), and say if a Mongol origin is not palpably in- 
 scribed on his face". 
 
89 
 
 Second Section. 
 Ethnology v. Phonology. 
 
 Ethnology, therefore, as a physical science, would hardly bring us 
 beyond a general conviction that India is inhabited by two diiferent 
 races of men. Nor should we, in our phonological studies, either 
 expect or desire more than general hints from physical ethnology, 
 The proper and rational connection between these two sciences 
 is that of mutual advice and suggestion, but nothing more. Much 
 of the confusion of terms and indistinctness of principles, both in 
 ethnology and phonology, are due to the combined study of these 
 heterogeneous sciences. Ethnological race and phonological race are 
 not commensurate, except in ante-historical times, or perhaps at the 
 very dawn of history. With the migrations of tribes, their wars, 
 their colonies, their conquests and alliances, which, if we may judge 
 from the effects, must have been much more violent in the ethnic, 
 than ever in the political periods of history, it is impossible to imagine 
 that race and language should continue to run parallel. The phy- 
 siologist should pursue his own science unconcerned about language. 
 Let him see how far the skulls, or the hair, or the colour, or the skin 
 of different tribes admit of classification ; but to the sound of their 
 words his ear should be as deaf as the ornithologist's to the notes of 
 caged birds. If his Caucasian class includes nations or individuals 
 speaking Arian (Greek), Turanian (Turk), and Semitic (Hebrew) 
 languages, it is not his fault. His system must not be altered in order 
 to suit another system. There is a better solution both for his diffi- 
 culties and for those of the phonologist than mutual compromise. 
 The phonologist should collect his evidence, arrange his classes, 
 divide and combine, as if no Blumenbach had ever looked at skulls, 
 as if no Camper had measured facial angles, as if no Owen had 
 examined the basis of a cranium. His evidence is the evidence of 
 language, and nothing else ; this he must follow, even though it be 
 in the teeth of history, physical or political. Would he scruple to 
 call the language of England Teutonic, and class it with the Low 
 
90 
 
 German dialects, because the physiologist could tell him that the 
 skull, the bodily habitat of such language, is of a Celtic type, or 
 because the genealogist can prove that the arras of the family con- 
 versing in this idiom are of Norman origin ? With the phonologist, 
 English is Teutonic, and nothing but Teutonic, and that because what 
 we may call its soul — the grammar — is Teutonic. Ethnological sug- 
 gestions as to an early substratum of Celtic inhabitants in Britain, 
 or historical information as to a Norman conquest, will always be 
 thankfully received by the phonologist ; but if every record were 
 burnt, and every skull pulverised, the spoken language of the present 
 day alone would enable the phonologist to say that English, as well 
 as Dutch and Frisic, belongs to the Low German branch — this, toge- 
 ther with the High German and Scandinavian, a branch of the Teutonic 
 stock — this, together with the Celtic, Slavonic, Hellenic, Italic, Iranic 
 and Indie, a member of the Arian family. The phonologist can detect 
 by himself the ingredients of Celtic, a large admixture of Norman, a 
 considerable infusion of Latin and even Greek in the English of the 
 present day, although he would gladly admit that it frequently saves 
 him time and trouble, if either historian or physiologist have indicated 
 what residuum lies for analysis in his crucible. The same applies to 
 our case. No physiological or historical evidence was necessary to 
 convince the phonologist that the language of India was not one 
 uniform language. Indeed, this difference was observed even be- 
 fore the difference of race had attracted attention, and ethnology 
 was in this case led, and therefore misled, by phonology. The ethno- 
 logical division of Arian and non-Arian inhabitants of India was at 
 first chiefly based on linguistic evidence. Tribes that spoke Sanskrit 
 dialects were set down as Arian; others speaking a non-Sanskritic 
 tongue were classed as members of the Turanian race. This has led to 
 much confusion and useless discussion. On one hand it was impossible 
 to deny the fact, that in the North of India millions of people speak 
 modern Sanskrit dialects, though their physical type is decidedly 
 Tamulian ; on the other no doubt could exist that many of the Brah- 
 mans of the Dekhan, now speaking Tamulian dialects, were of 
 Arian extraction. The fact ought to have been stated plainly, for it 
 is a fact to which there are analogies all over the world, and which 
 scholars ought to have been familiar with by the knowledge that the 
 
91 
 
 Normans, wlio spoke, every man, a Teutonic dialect, when they took 
 possession of the North of France, spoke a Eomance dialect, every 
 knight and wight, when they conquered England. Attempts have 
 instead been made to prove that Bengali and Hindustani were languages 
 Tamulian in grammar ; or, in an opposite direction, that tribes, like 
 those who now inhabit the valley of Asam and speak Asamese, i. e. 
 a Sanskritic dialect, had Caucasian blood in their veins, and were 
 Caucasians modified and deteriorated by the influence of climate and 
 of diet. But although the majority of people who speak Bengali 
 may be of Tamulian extraction, does it follow th^t the grammar of 
 their language is Tamulian ? Or does it follow that the original 
 inhabitants of Asam were Arians, because the language at present 
 spoken in that country is Sanskritic in its grammar ? In fact, after 
 Asam was brahmanised in language and thought, it was again 
 conquered by the Ahoms.* These overspread and conquered the 
 country, and now constitute a large proportion of the population. 
 Yet scarcely a single term in present use is traceable to the ancient 
 Ahom, a language closely allied to the Shan and Siamese, and now 
 understood only by a few Ahom priests who preserve their old religion. 
 There ought to be no compromise of any sort between ethnological 
 and phonological science. It is only by stating the glaring contradic- 
 tions between the two sciences that truth can be elicited. I feel no 
 doubt that the only natural solution of the problem would have been 
 found and accepted long ago, had it not been for this baneful spirit 
 of accommodation and mutual concessions. Ever since Blumenbach 
 tried to establish his five races of men (Caucasian, Mongolian, Ame- 
 rican, Ethiopian, and Malay), which Cuvier reduced to three 
 (Caucasian, Ethiopian, and Mongolian), while Prichard raised them to 
 seven (Iranian, Turanian, American, Hottentots, Negroes, Papuas, 
 and Alfourous), it was felt that these physiological classifications could 
 not be brought to harmonize with the evidence of language. Blumen- 
 bach's Caucasian race, for instance, was a congeries of at least three 
 phonological races — the Greeks (Arian), Jews (Semitic), and Turks 
 (Turanian). Yet this point was never urged with suflScient strength, 
 
 * Ahom is the same word as Asam. It is said to he the Sanskrit Asama, 
 unequalled, which pronounced according to the Bengali fashion is Asam, accord- 
 ing to native pronunciation Ohom or Ahom. Cf. N. Brown's Grammatical Notices. 
 
92 
 
 till at last Humboldt in his Kosmos (I. 353.) stated it as a plain fact, 
 that, even from a physiological point of view, it is impossible to re- 
 cognise in the groups of Blumenbach any true typical distinction, any 
 general and consistent natural principle. From a physiological point 
 of view, we may speak of varieties of man, — no longer of races. 
 Physiologically the unity of the human species is a fact established as 
 firmly as the unity of any other animal species. So much, then, but 
 no more, the phonologist should learn from the physiologist. He 
 should know that in the present state of physiological science it is 
 impossible to admit more than one beginning of the human race. He 
 should bear in mind that Man is a species, created once, and divided 
 in none of its varieties by specific distinctions; in fact, that the com- 
 mon origin of the Negro and the Greek admits of as little doubt as 
 that of the poodle and the greyhound. No argument, derived from 
 the diversity of language, will shake the physiologist in this convic- 
 tion ; and the phonologist must keep it in view if he wishes to secure 
 his science that honourable place which Humboldt assigned to it, as 
 the connecting link between the physical and intellectual Kosmos. 
 
 The interval between the first beginnings of the natural history of 
 man, and the times to which we can ascend through the evidence of 
 language, may be so great as to make it impossible to gather up the 
 threads of the one, and connect them with those of the other period. 
 It maybe — nay, if we consider the few facts here within reach of even 
 inductive reasoning, most likely it will be — impossible to strengthen 
 the arguments of physical science in favour of a common origin of 
 mankind, by evidence derived from phonological researches ; but it 
 should not be attempted again to disprove the unity of the human 
 race by arguments derived from the apparent diversity of human 
 speech. On one side the phonologist need no longer feel hampered by 
 the classifications of a Blumenbach and a Cuvier*, with regard to 
 
 * Cf. Synopsis of the Physiological Series in the Christ Church Museum, p. 2. 
 Dr. Henry Acland defines the relation of physiology and linguistic ethnology (pho- 
 nology) with exactness and fairness. The crania, he says, will furnish the student 
 with examples of the modification of form of which the human skull is capable. 
 In these forms, sufficient data will not he found for constructing natural groups of 
 the nations ; inasmuch as the researches of ethnologists tend to show, with more 
 and more certainty, that these alliances are to he discovered by linguistic investi- 
 gations alone. But the study of changes which occur in anatomical structure, 
 
93 
 
 his classification of languages ; on the other, he ought to bear in 
 mind, that, if it is impossible to trace the convergence towards one 
 common source of all the dialects of the human species, it will be 
 necessary at least to explain the possibility of their divergence, and 
 to account by analogy for the fact of their apparent diversity. 
 
 Thied Section. 
 
 Subdivision of the Nishdda or aboriginal Languages of India. 
 
 Accepting for our starting point the general distinction between 
 Aryas and Nishadas, which, whether suggested by physical features, 
 or proved by the evidence of grammai", may be considered as an un- 
 disputed fact, we have now to see if all Nishadas are really of one 
 stock; and if so, whether they can be subdivided into distinct 
 groups. 
 
 " The physical aspect of the Nishadas," says Mr. Hodgson, in a 
 passage which just catches my eye, " is of that osculant and vague 
 stamp which indicates rather than proves anything, or rather what 
 it does prove is general, not particular." Their linguistic aspect, 
 however, is more satisfactory, and no doubt the evidence to be de- 
 rived from it will become still more convincing and more distinct if 
 the collections and researches to which Mr. Hodgson has given so 
 powerful a stimulus and so successful an example are continued with 
 an equal zest and in the same spirit. It is, no doubt, a difficult and 
 not always pleasant task to collect words and phrases from the mouths 
 of people whom few would choose for the companions of their studies ; 
 but it is a task that promises to reward most amply the labour ex- 
 pended on it. Mr. Hodgson's plan of inviting cooperation all over 
 India is good ; but I am afraid he will not find that every " collector" 
 is able to collect words or grammars. Mr. Hodgson's instructions 
 also are practical ; but it will require much philological tact, and 
 painstaking scholarship to carry them out successfully. One point, 
 perhaps, ought to be put forward still more prominently. Wherever 
 
 according to modes of origin, of life, of climate, and of society, will remain among 
 the most interesting problems in the natural history of man, and of the animals, 
 the co-tenants of our planet. 
 
 n 
 
94 
 
 it is possible (and it should never be impossible), a grammatical 
 outline of each dialect should be given^ such as can be deduced 
 from a number of phrases written down and compared with one 
 another. Even the largest vocabulary will not make up for the ab- 
 sence of grammatical paradigms. But if time and leisure are wanting 
 for this more tedious task, let the collection of words, at all events, 
 reach the numbers which Mr. Hodgson originally fixed. The small 
 vocabularies which have lately been published, for instance, of the 
 Kole tribes, are not satisfactory, particularly as they involve a great 
 problem. They hardly indicate, still less do they prove, any relation- 
 ship between these dialects and any other. With the exception of 
 Uraon and Rag'mahal, which seem Tamulic in the narrower sense 
 of the word, the otlier lists should certainly be re-made. 
 
 The chief objection to mere lists of words as proofs of the relation- 
 ship of languages is felt where we have to deal with tribes whose 
 previous history we have no means of knowing. It is impossible 
 to say whether words collected among one tribe have been adopted 
 from another ; and even where we know that a language is mixed, 
 we have no means of determining, without the assistance of gram- 
 matical forms, which of the two portions represents the original 
 stock, and which the later additions. If a Brahman came to Europe, 
 and without knowing much of the history and the languages of the 
 continent, collected a number of words in Wales, in London and in 
 Paris, he would no doubt, on his return home, discover a considerable 
 quantity of words identically the same in his Welsh, English and 
 French lists. Or, to take a more extreme case, if he collected words 
 at Bayonne, some from Spaniards, others from Basks, he vc^ould here 
 again find the majority of words, which he is likely to ask for, iden- 
 tically the same in both lists.* The differences in some words he 
 would account for as he accounts in his own country for differences 
 between Bengali and Hindustani, and, on a prima facie evidence, 
 he would feel himself justified in arranging Spanish and Bask as 
 cognate tongues. 
 
 * Bask words taken from Latin or Spanish; gorputz, 1)0(17 ; dempora, times! 
 presuna, person ; arima, soul ; bekatua, sm ; botua, vote ; acceptatcea, to accept ; 
 affligitcca, to afflict ; mendecoste, pentecost ; eliza, church ; aingeru, angel ; ar- 
 rosa, rose; artea, artj arrapostua, answer; azucrea, sugar; donceila, lady. 
 
95 
 
 No doubt there are essential words which one nation very seldom 
 adopts from another, such as pronouns, numerals, prepositions and 
 conjunctions. But these again are generally short words, and very 
 liable to corruption. Now, the chances of accidental coincidences, 
 particularly with short words, are much greater than commonly 
 supposed, and it will be useful to bear this in mind where we have to 
 deal with scanty lists. The rainbow, in Georgian, is Iris. This may 
 or may not have been taken from Greek. But the fingers, in Georgian, 
 are called thithi, in Lapponian tiute, in Syrianian tyute, in Italian diti 
 (i. e. digiti). Here we have a coincidence, the result of mere chance. 
 Compare, besides, Georgian, 
 
 qirili, clamour, and Latin, querela, 
 didi, great, and Lithuanian, didis. 
 qeli, throat, and German, kehle. 
 khata, cat, and Latin, catus. 
 nawi, boat, and Latin, navis. 
 snli, soul, and German, seele. 
 uremi, carriage, and Greek, Sp/na. 
 ghwino, wine, and Latin, vinum. 
 ■wizi, to know, and German, wissen. 
 
 It would be difficult to say, unless we regarded the Georgian as a 
 member of the Arian family, which of these words are taken from 
 Persian, Eussian or Greek, and which are the result of accidental 
 coincidence. But let us take languages between which no inter- 
 course can be imagined, such as Mandshu and the classical languages, 
 and the following list will give an idea how far phonetic coincidences 
 may be produced by chance * : — 
 
 Mandshu.* 
 
 Greek and Latin. 
 
 akha, rain ; 
 
 aqua. 
 
 aniya, year; 
 
 annus. 
 
 toma, grave; 
 
 tumulus. 
 
 ilengu, tongue ; 
 
 lingua 
 
 sengi, blood ; 
 
 sanguis 
 
 cholo, idleness; 
 
 <Txo\-^. 
 
 unun, weight; 
 
 onus. 
 
 koro, care ; 
 
 cura. 
 
 amuran, smitten ; 
 
 amoureux. 
 
 furu, rage ; 
 
 furor. 
 
 ako, not ; 
 
 oIk. 
 
 baru, before ; 
 
 Trpd. 
 
 Cf. Von der Gabelentz, Grammaire Mandchoue. 
 
 H 2 
 
 
96 
 
 Here we have confined ourselves to a collation of the classical lan- 
 guages ; but if we allowed our eyes to wander over the whole surface 
 of spoken languages, if we looked into American, African, Malay, 
 Indo-Chinese and Siberian dictionaries, I believe that there is hardly 
 a word in any language, to which, making the usual allowance for 
 change of form and meaning, some other word might not be found 
 almost identical. I take some instances from Klaproth's Asia 
 Polyglotta: — 
 
 Sun, 
 
 shun, 
 
 Mandshu ; 
 
 Sonne, 
 
 German. 
 
 Moon, 
 
 sara. 
 
 Syriak ; 
 
 sara. 
 
 Calmuck. 
 
 Star, 
 
 choshi, 
 
 Japan ; 
 
 chos. 
 
 Ostiakian. 
 
 "Water, 
 
 don. 
 
 Ossetian ; 
 
 dan. 
 
 New Guinea. 
 
 Mountain, 
 
 oros, 
 
 Greek ; 
 
 ura. 
 
 Tungusian. 
 
 Ear of corn, 
 
 agna. 
 
 Latin ; 
 
 agna, 
 
 Lapponian. 
 
 Nose, 
 
 bini. 
 
 Persian ; 
 
 bi, 
 
 Chinese. 
 
 Ear, 
 
 uhr (1), 
 
 Chinese ; 
 
 ohr. 
 
 German. 
 
 Hand, 
 
 kara, 
 
 Sanskrit ; 
 
 gar, 
 
 Mongolian. 
 
 Cow, 
 
 bo, 
 
 Erse; 
 
 ba. 
 
 Tibetan. 
 
 Dog, 
 
 kyon, 
 
 Greek ; 
 
 kiuan. 
 
 Chinese. 
 
 Blue, 
 
 kyanos. 
 
 Greek ; 
 
 chiuan. 
 
 Chinese. 
 
 Egg, 
 
 eg, 
 
 Jenisseian ; 
 
 egg. 
 
 English. 
 
 All, 
 
 pan. 
 
 Greek ; 
 
 fan. 
 
 Chinese. 
 
 It is true that coincidences of this kind are not likely to deceive us 
 long, because they could never run through tolerably full lists of 
 words taken from languages distant in place and relationship. But 
 where we have to unravel a cluster of languages, confusedly mixed, 
 as, for instance, the Albanian, Wallachian and Bulgarian, on the 
 confines of the Greek, Latin and Slavonic areas ; or the Asamese, 
 Chepang and Eat/mahal on the confines of Sanskrit, Tibetan and 
 Tamulian, it will be necessary to disregard at first all coincidences 
 of words, and look entirely to their grammar. 
 
 In India, after the light that has been thrown on its ethnology by 
 the combined labours of men such as Hodgson, N. Brown, Bronson, 
 Robinson, Stevenson, Elliot, and others, we can clearly distinguish 
 now between at least two classes of Nishadas, the one receding be- 
 fore the stream of Arian civilization across the Vindhya into the 
 Dekhan, the other pouring, at a' time not easily determined, through 
 the valleys of the Himalaya into the north-eastern countries of India. 
 The former class may be called Tamulic, in the narrower sense of 
 the word ; the latter Bhotiya, or Sub- Himalayan. 
 
97 
 
 Fourth Section. 
 
 The JBhotlya Class. 
 
 To begin with the latter, which was recognised by Mr. Hodgson 
 as a distinct class of dialects as early as 1828, there can be no doubt 
 now that it is closely connected with the language of Tibet. Nu- 
 merals, pronouns, and the terminations, or rather postpositions, 
 which occur in these languages, are frequently identically the same 
 as in Tibetan. As far as the evidence of language goes, no doubt 
 can remain on this point. Nor is it difficult to account for it, whether 
 ethnologically, historically, or geographically. 
 
 1. Ethnological Evidence. 
 
 Ethnologically, the Tibetan character is to be read on the face 
 of all these tribes. " Their physiognomy exhibits generally and 
 normally the Scythic or Mongolian type (Blumenbach) of human 
 kind ; but the type is much softened and modified, and even fre- 
 quently passes into a near approach to the full Caucasian dignity 
 and beauty of head and face ; though among the Cis or Trans- 
 Himalayans there is never seen any greater advance toward the 
 Teutonic blond complexion than such as consists in occasional ruddy 
 moustaches and grey eyes among the men, and a good deal of occa- 
 sional bloom upon the cheeks of the children and women. A pure 
 white skin is unknown, and the tint is not much less decided than 
 in the high caste Hindus ; but all are of this pale brown or Isabelline 
 blue in Tibet and the Sub-Himalayas, whilst the many in the plains 
 of India are much darker." (Dec. 1847. ) 
 
 2. Historical Evidence. 
 
 Historically we can never expect much documentary evidence on 
 the past history of nations who had no literature, no alphabet, no monu- 
 ments. But an inference may be drawn, as Mr. Hodgson believes, 
 that these Sub-Himalayan tribes were separated from their Tibetan 
 
 H 3 
 
98 
 
 brethren at least before the introduction of Buddhism from India into 
 Tibet. Indian letters, Indian literature, customs and ideas were carried 
 into Tibet by Buddhist missionaries in the seventh century, and no 
 traces of it are visible in the texture of the Sub-Himalayan dialects. 
 Their own traditions, as Mr. Hodgson affirms, indicate a transit of the 
 Himalaya from thirty-five to forty generations back (1000 to 1300 
 years) ; but their original separation may have taken place long before. 
 Some of these tribes have preserved the same names which they 
 have in the Mahabharata. The position there assigned to the Kiratas 
 and Ki^akas is the same which the Kirantis and KiAakas now hold, 
 and they are no doubt the same people with whom the heroes of the 
 Mahabharata, Arp'una and Bhima, are represented as fighting. This 
 point has been admirably treated by Professor Lassen in his ethno- 
 logical articles in the Zeitschrift fiir die Kunde des Morgenlandes, 
 and again in his Indian Antiquities. It has been proved that the 
 name Kirata was known to the author of the Periplus of the Ery- 
 threan, and to Ptolemy ; and, what is important, this name was known 
 to them east of the mouth of the Ganges and Brahmaputra. The 
 Sabarae of Ptolemy also are as far east as the Ganges, and they have 
 been identified by Lassen with the modern Saur, the ancient ^Savaras, 
 i. e. Mlekhas, names expressive of a pale rather than black colour. 
 The physical description of these tribes, as given by the Greeks, 
 agrees with the low Tibetan type, particularly if the Skiratse of 
 Megasthenes might be identified with the Kiratas. They had flat 
 noses, or, as Megasthenes likes to say, no noses at all. Certain it is 
 that these low Turanian nomad races are mentioned on the frontiers 
 of India so soon as any of the Arian nations come within historical 
 sight. 
 
 In some cases, however, these Sub-Himalayan tribes have preserved 
 a recollection of their former Trans- Himalayan homes — a fact which 
 would seem to point to later immigrations than those which opened the 
 first channel to the Trans-Himalayan population of Northern India. 
 The Limb us for instance, are called Chong by the Lepchas, and the 
 province of Chung in Tibet, south of Lhassa, is said by the Limbus to 
 have been their original country. The Mur mis speak of themselves 
 as having at some remote period crossed the Snows, and they main- 
 tain that they preserved their language and religion (?) unchanged 
 
99 
 
 since their arrival. A Dewan of the Sikkim Raja, who conversed with 
 Mr. A. Campbell, told him that he crossed the original country of the 
 Murmis on his way from Sikkim to Lhassa. (I. A. S. B. 1842, p. 4.) 
 
 3. Geographical Evidence. 
 
 Geographically we must look upon the Himalaya not as an un- 
 broken chain or unsurmountable barrier to separate the high plains 
 of Asia from the basins of the Indus and Ganges, but rather as 
 mountain gates, opening to the bold adventurer a hundred different 
 passes into the gardens of India. Here also we owe much to Hodg- 
 son's genius. His map of the natural divisions of the Himalaya is 
 in truth a grammaire raisonnee of this irregular mountain-utterance. 
 In order to give an idea of its organism in as short a space as pos- 
 sible, we might venture to compare the large mass of mountains 
 between India and Tibet, in the North-East, to a hand with its five 
 fingers expanded towards India. Every interval between two of these 
 fingers marks the basin of one of the four of the principal rivers ot 
 Northern India, and each river draws its feeders east and wes^ from 
 the two ridges by which it is included. The four knuckles would 
 represent the five highest peaks, which are the articulations of five 
 mountain ridges projecting to the plains of India. If we look 
 upon these ridges as the five fingers of a left hand, the knuckles, 
 beginning with that of the little finger, would correspond to the 
 following peaks : — 
 
 1. Chumalari 23929, 27° 52', 89° 18' (^imalhari). 
 
 2. Kangchang, 28176, 27° 42', 88° 10' (KanAin^inga). 
 
 3. Gosain-than, 24700, 28° 20' 86° (Gosvamisthana). 
 
 4. Dhoula-giri, 27600, 29° 10' 83° (Dhavala-giri), 
 
 5. Nandadevi, 25589, 30° 22' 79° 50'. 
 
 Between these five peaks, and included by their rib-like continua- 
 tions, we obtain the following four river-basins : — 
 
 Between 5 and 4, the basin of the Sarayia (Karnali). 
 Between 4 and 3, the basin of the Gandaki. 
 Between 3 and 2, the basin of the Kausiki, 
 Between 2 and 1, the basin of the Tista. 
 
 H 4 
 
100 
 
 All these basins from West to East are successively drained by the 
 Ganges, which takes first the Sarayu, then the Gandaki, Kausiki 
 and Tista, with their respective tributaries. The Ganges itself flows 
 from a basin formed on its eastern side by the Nanda-devi, or the 
 thumb, on the western by the GangSvatari and Yamunavatari ridge 
 (25669, 30° 55', 78° 12'). It has absorbed its western feeder, the 
 Yamuna, before it reaches the Sarayu. The next basin, after the 
 Gangetic, in the West, is that of the (Satadrn, or Sutlej. It is the 
 first river which is absorbed by the Indus. On the east, the next 
 basin, independent of the Ganges, is formed by the Manasa, the 
 first river absorbed by the Brahmaputra. With the iS'atadru, there- 
 fore, in the West, and the Manasa in the East, two new systems 
 begin. The sources of these two rivers, the Indus and Brahma- 
 putra, are on the roof of the same hand, which, by its five, or, 
 if we include the Ganges, its six knuckles, forms the sources of the 
 Gangetic system. The courses of the Indus and Brahmaputra are 
 determined by the northern declivity of the watershed between 
 Tibet and India. They run, the Brahmaputra, alias Hladini or 
 Sanpu, towards the East, swallowing all the waters (Manasa and 
 Subhansri), on the left side of the Chumalari, or the little finger, and 
 disgorging them near the tropic into the Bay of Bengal ; the Indus, 
 towards the West, absorbing the rivers of the Penjab and all the 
 water on the right side of Yamunavatari, and disgorging them near 
 the tropic into the Bay of Cutch. 
 
 The whole length of the Himalaya, from where it is outflanked 
 by the Indus and its tributaries to where it is taken in the rear 
 by the Brahmaputra, is 1800 miles, the mean breadth being ninety 
 miles. Though this proportion would be ill represented by that of 
 the palm to the fingers, there is one feature in the conformation of 
 these mountainous slopes which again it is easy to represent and to 
 remember, by looking at a hand with its fingers slightly inclined. 
 There are three transverse climatic divisions, which Mr. Hodgson 
 established as early as 1847, and which he has since worked out 
 more completely, assisted by Dr. Hooker. Each division takes about 
 thirty miles. The first is called the Upper region from the crest of 
 the snowy range, 16,000 (?) down to 10,000 feet above the sea. The 
 second is the Central region, from 10,000 to 1000 feet above the 
 
101 
 
 sea. The third is the Lower region, extending from 4000 feet to the 
 level of the plains. These three regions, in their gradual decline, 
 correspond in many respects with the arctic, the temperate and 
 tropical zones. The lower region only, which would be the third 
 joint of our fingers, requires and admits a new subdivision, into 
 
 I. The sandstone range (Dhuns or Maris) ; 
 II. The Bhaver or Saul forests (Jhari) ; and 
 III. The Tarai swamps. 
 
 It was necessary to give this short outline, in order to explain the 
 position of the Sub-Himalayan races in their relation to Tibet and 
 India. The Upper region of the Himalayas forms the outskirts of 
 Tibet. This country stretches on a level of about 10,000 feet, 
 towards Bucharia and China, and forms a triangular plateau, having 
 for its longest side the whole chain of the Himalayas. 
 
 The Upper or Cachar region of the Himalaya is therefore natu- 
 rally occupied by the Bhotiyas, who extend along the whole line 
 of the Ghats (mountain passes), and with the name, have retained 
 the lingual and physical characteristics of their tramontane brethren. 
 They may be called" Bhotiyas, or by more special names, Ron gbo, 
 Siena, or Bhotia, Serpa, &c. Their language can be studied by 
 means of the Tibetan proper, and by specimens of the Sarpa dialect. 
 
 The Central or temperate regions are distributed among the several 
 Sub-Himalayan races in the following order. Between the Brahma- 
 putra, or rather its tributary the Subhanshiri and the Chumalari ridge, 
 that is to say, within the aqueous system of the Brahmaputra, we meet 
 with the Mishmis, Bors, and Abors, Akas and Dophlas. The 
 next or Tistean basin is the fatherland of the Dijondmaro (vel 
 Dinjong-maro, man of Dinjong or Sikkim?), and of the Pluh or 
 Lhopa, that is Lepcha and Bhutanese, TheKosean basin is the 
 abode of the Kirantins and Limbus. Between the Kosean and 
 Gandakean basins, we have the high level space of Nepal, peopled by 
 Newars and Murmis. The next or Gandakean basin is the seat 
 of the Sunvars, the Gurungs andMagars. The distribution of 
 these tribes, according to the different river basins, is given entirely 
 on Mr. Hodgson's authority. But as most of them lead a very 
 unsettled life, we must not expect to find their names always confined 
 
102 
 
 to tlie " locale" here assigned to them. Another source of confusion 
 is the variety of synonymous names given to the same tribe by dif- 
 ferent people. The Lepcha, for instance, whom Mr. Hodgson 
 places in the Tistean basin, occupy, according to A, Campbell *, an 
 extent of about 120 miles, bounded on the west by the Tambar 
 branch of the Kuri, and in the east by the mountains of Bhutan. 
 They are found in Nepal, Sikkim, and about fifty miles beyond the 
 Tista. They call themselves Lepcha, but are divided into two races, 
 called R o n g and K h a m b a. The latter state, that about 200 years ago 
 they came together with the first ancestor of the Sikkim Raja, from 
 Kham, a province of China, or rather Tibet, while the Rong have 
 lost all recollection of their Trans-Himalayan origin. In a route 
 from Cathmandu in Nepal to Tazedo, on the Chinese frontier, com- 
 municated by Mr. B. H. Hodgson, Amir, the interpreter, mentions 
 Khambas and Kumis as Bhotiya inhabitants of Pochuzan, close 
 to the frontier of Tibet and China. The Lepcha have an alphabet, 
 whose character Csomo Korosi pronounced to be not-Tibetan. Their 
 religion, however, is Buddhist. 
 
 The Limb us again, whose principal habitat, according to Mr. 
 Hodgson, is the Kosean basin, are found, according to A. Campbell, 
 not only between the Dud-kusi and the Kanki rivers, but, though in 
 smaller numbers, eastwards to the Mechi river, which forms the 
 boundary of Nepal and Sikkim. They exist even in Sikkim, and as 
 far east as the Tista. Their name, Limbu, is said to be a corruption 
 of " Ekthumba," and other tribes, the Kirantis, the Eakas 
 (between Arun and Konki), and Rais are sometimes included in the 
 generic term "Limbu." The Limbus are ranged under two great 
 divisions, viz. Hung and Rai, each subdivided into many clans. 
 Their original religion is neither Buddhist nor Brahmanic. 
 
 The Murmis again, of the Gandakean basin, are said to extend 
 west as far as the Mechi, east through Sikkim as far as the Tista. 
 They are also divided into several clans. 
 
 West of the Gandakean basin, in the basins of the Sarayu and the 
 
 Ganges, the pure Tibetan type ceases, mongrel and mixed races 
 
 occupy the central regions, the north-west parts excepted, where the 
 
 Rongbo or Cis-Nivean Bhotias, the Garhwalis, and the in- 
 
 * Cf. A. Campbell, Journal A. S. B. 1840. p. 379. 
 
103 
 
 habitants of Kan aver and Hungrang are said to be of Tibetan 
 extraction, although their language is no longer Tibetan. The prin- 
 cipal names of these mixed races northward to Gilgit are Khas or 
 Khasias, Kohlis, Garhwalis, Kakkas, Bambas, Gakars, 
 Khatirs, Awans, and Janjuhs. In the central regions we find 
 also the Domes, the Helots of Kumaon. 
 
 In the third or Lower region, the following names occur ; the K o c c h, 
 Bodo, Dhiraal, Mecha, Kichak, Tharu, Denwar, Pallah, 
 Boksar, Hayu (in the central and lower ranges between Arun and 
 Konki), Chepang, Kusunda, Durre, Bramho, and other tribes, 
 who alone can live and breathe the malarious effluvia of these swamps 
 and forests without injury to their health. Some of them, as for 
 instance the Thar us, extend westward as far as the Ganges. 
 
 4. Phonological Evidence. 
 
 It is a point of importance to determine whether these tribes all 
 belong to the Tibetic stock. To judge from their outward appear- 
 ance, particularly in the case of the Chepang, Kusunda, and Haiyus 
 (outcasts in the second and the third regions), they seem to be of 
 Tamulic extraction in the narrower sense of the word, that is to 
 say, resembling in form and colour the aborigines of the plains. 
 But the language of the Chepangs, when carefully collected and 
 collated, proved to be of Tibetic origin. Hodgson found it possible 
 to reconcile the contradiction, and account for the apparent physical 
 differences between Bhotia and Chepang, by the deteriorating in- 
 fluences to which these outcast tribes had been for centuries sub- 
 mitted. As far, therefore, as physical evidence goes, we are free to 
 look upon the darker colour and slender frame of all the tribes in- 
 habiting the malarious region, as marks still reconcilable with their 
 Tibetic origin ; and lingual evidence is certainly in favour of this 
 view. If we had mere lists of words, collected among the inhabitants 
 of the lower region, there might still be a suspicion, that where their 
 words happen to agree with the Tibetan, they were, in many cases, 
 adopted. But the little we know about the grammar of the dialects 
 of the third region, is sufficient to show that they belong to the 
 Bhotiya, and not, as has been supposed, to the Tamulic class. If we 
 
104 
 
 take the numerals of the Bodo and Dhimal, and compare them 
 with Tibetan and Burmese numerals, and afterwards with those 
 of the real Tamulians in the south, their coincidences with the latter 
 are of a very general nature, while they agree with the former in 
 such a manner as to leave no doubt about their common origin. The 
 same applies to the pronoun, although here there is a family likeness 
 even between the Tamulic and Tibetic tongues. This general rela- 
 tionship makes it difficult sometimes to distinguish at once the 
 members of every different branch, all finally referable to one common 
 stock. The Tibetic branch stands to the southern Tamulic in much 
 the same relation as the Mongolic to the Ugric. The roots and words 
 may often be the same, but they differ vastly in the degree of gram- 
 matical perfection attained by each. The Mongolic has no terminations 
 as yet to express the different persons of the verb, neither have the 
 Tibetan and Burmese. Exceptions like the Naga dialect of Namsang, 
 where we find the first signs of a verbal growth, are like the MongoUc 
 dialects lately studied by Castren, where a similar grammatical phe- 
 nomenon was observed. The Tamulic branch, on the contrary, par- 
 ticularly in its leading dialects, has a system of verbal affixes as com- 
 plete as the Ugric. Another distinguishing feature of the Tamulic, 
 consists in the loss, or, at all events, in the absence of the intonations 
 common to the Bhotiya and Chinese branch. Mr. Robinson describes 
 four different accents or intonations prevalent in all the languages 
 spoken by the tribes bordering on the valley of Asam, which includes 
 Mr. Hodgson's Garos, Miris, Abor-Miris, and Kacharis, the last being 
 taken as a general name, and comprehending the Borros (Bodos), 
 Hojai-Kacharis, Kochis (Modai-Kochis, Phul-guriyas, and Hermias), 
 the Mechis, Dhimal, and Rabhas. Mr. Hodgson, in his last articles, 
 thinks that traces of musical intonation can be discovered even in 
 more western dialects. None however exist in the Tamulic lan- 
 guages, and it may be stated as a fact, that of real Tamulic 
 grammar, in the proper sense of the word, no trace has 
 as yet been discovered north of the Ganges. 
 
105 
 
 Trans-Himalayan Dialects. 
 
 Before we proceed to trace the southern ramifications of the Bho- 
 tiya class in India, it may be well to cast a glance on what may be 
 called the Trans-Himalayan dialects of Tibet. 
 
 The Trans-Himalayan members of the Bhotiya class of lan- 
 guages do not properly lie within the limits of Indian phonology. 
 They are mostly, however, dialects of that language which forms the 
 type or norm of the whole Bhotiya class, the literary language of 
 Tibet, and as such they have a certain importance for a study of the 
 whole class. It should be borne in mind, that what we call a literary 
 language, is, after all, only one out of many dialects, which politically 
 may have been more successful than the rest, but which linguistically 
 has no more right to be considered the sole representative of 
 one body of living speech than any other of its dialect. Nay, 
 in many cases, though literary dialects may be richer in words, 
 they have been shown to be more reduced in grammar than their 
 less cultivated sisters ; and comparative philology has elicited more 
 secrets from the lips of vulgar idioms than from classical writings 
 of literary celebrities. Besides, with regard to Tibet, Mr. Hodgson 
 tells us, that what we, after Csoma de Coros, consider as the 
 standard of Tibetan grammar, is positively repudiated by the people 
 of Tibet (1853, p. 125.), so that any new collateral light on this 
 subject will be useful and important. 
 
 Tibet, bounded in the north by the Kuenlun, in the south by 
 the Himalaya, is divided again by a third prominent chain, which 
 Mr. Hodgson calls the " Nyenchhen-thangla." This chain, which is 
 partially indicated by Ritter's Nian tsin tangla, is considerably ex- 
 tended by Hodgson, and forms, according to him, the barrier be- 
 tween the north and south, or between the nomadic and civilised 
 portions of Tibet. Between this range and the northern borders of 
 Tibet, we find three large nomad races, the Horpa in the west, 
 the Sokpa in the east, and the Drokpa in the central portion. 
 The Horpa (Ritter's Khor) reach into Little Bucharia and Songaria, 
 where they call themselves Ighurs; the Sokpa extend as far as 
 
106 
 
 the Kokonur and Tangut, and their country is called Sokyeul.* 
 Besides the Drokpa (Brogpa), remain other nomadic tribes known 
 by the names of Kazzak and Chakpa. The general name of these 
 nomads of Tibet is Horsok, in contradistinction to the settled 
 inhabitants of the southern provinces, who are known by the general 
 name of Bodpa. 
 
 Some of these nomadic tribes coming into frequent contact or 
 collision with the south, speak the pure Tibetan ; others speak dia- 
 lects. Mr. Hodgson gives a specimen of the Horpa in the west, which 
 is a Bhotiya dialect ; while the Sokpa in the east speak a Mongolic 
 idiom. The language of the central Drokpa is not yet known. 
 
 Another tract of language, first explored by Mr. Hodgson, extends 
 from the Sokpa on the north-eastern frontier of Tibet, along the con- 
 fines of Tibet and China, toward the south, as far as Tunan. We 
 have here the Amdoans, the Thochu, Gyarung, and Manyak. The 
 first speak simply Tibetan ; the other three speak dialects first col- 
 lected by Mr. Hodgson. Another language, equally a Tibetan dia- 
 lect, is spoken by the Takpa (Eitter's Gakpo, Gangpo, and Dakpo), 
 not, however, on the eastern frontier of Tibet, but west of Kwombo, 
 in the central province of Tibet. These tribes, with the exception 
 of the last, are known in Chinese by the general name of Si fan, or 
 western aliens. Finally Mr. Hodgson gives us one dialect spoken in 
 the immediate neighbourhood of the Sifan, the Gyami ; and this 
 is no longer Tibetan, but Chinese. 
 
 The information which we possess regarding these languages is 
 as yet extremely scanty, particularly with respect to their grammar. 
 The vocabularies published by Mr. Hodgson are here less trust- 
 worthy than in other tribes. He says so himself, particularly with 
 regard to the Sokpa and Gyami vocables. Still the linguistic evi- 
 dence, incomplete as it is, is sufiicient to warrant the classification 
 of the Sokpa with the Mongolian dialects. The identity in the 
 
 * Of the two lists of words, respectively ascribed to the Sokpa and Horpa, the 
 Sokpa words are Mongolian, the Horpa, Bhotiya. Sok and Sok-bo is the usual 
 Tibetan name for Mongolian tribes ; those who live in Northern Tibet and Tangut, 
 nay all Mongolians between Tibet and the towns of Little Bucharia, call them- 
 selves Slmraigol, and are sometimes called Chor by the Tibetans, Chor being 
 given as a synonyme oi Tata (i. e. Mongol) in the Chinese-Tibetan dictionary of 
 Peking. 
 
107 
 
 numerals is surprising; but in the absence of a complete set of 
 pronouns or pronominal affixes, it is impossible to enter into de- 
 tails. The Horpa language is more ambiguous. By its pronouns 
 and numerals it is Bhotiya, and I have accordingly ranged it with 
 the Trans-Himalayan Bhotiya dialects : forming their most western 
 branch. Mr. Hodgson, however, refers the Horpa to the Turkish 
 family, and he derives his argument " not only from the vocables 
 but from the complex structure of Horpa verbs." The plural 
 termination also, which is riggi in Horpa, sounds like Tataric 
 grammar; for instance, the Horpa gna, I, gnariggi, we; as com- 
 pared with the Tataric ben, I, and bisigi, we. Nay, the Thochu 
 also, with its plural termination lar, the Manyak with its dur, and 
 the Takpa with its r a, have at first sight a Tataric appearance. But 
 we must wait until Mr. Hodgson will give us all his materials, before 
 these Trans-Himalayan dialects can be classified with anything like 
 safety, and I therefore give my own classification only as provisional 
 and open to correction. Of the eastern languages of Tibet, that of 
 the Amdoans is said to be purely Tibetan. The Thochu, Gyarung, 
 and Manyak dialects are also connected with Tibetan ; but again 
 the scantiness of linguistic evidence is such as to make further 
 identifications extremely problematic. The Gyarung, for instance, 
 to judge from occasional instances given by Mr. Hodgson, seems far 
 to surpass the literary Tibetan in grammatical forms. The Gyarung 
 clearly possesses predicative pronominal prefixes, which in the 
 Bhotiya class are confined to some of the Naga dialects. They are 
 used on the principle of composition represented in my table by (3a,, 
 and, therefore, find analogies in the Caucasus, the Dekhan, and in 
 Sanskrit. Mr. Hodgson points out himself the striking similarity 
 between the Circassian and the Gyarung in the use of these prono- 
 minal prefixes, and he contrasts the 
 
 Circassian sara (I), wara (thou), ui (he), 
 
 s-ab (my father), w-ab (thy father), t-ab (his father), 
 with 
 Gyarung nga (I), nanre (thou), watu (he), 
 
 nga-pe (my father), na-pe (thy father), wa-pe (his father). 
 
 Mr. Hodgson maintains that the same principle prevails in the 
 
108 
 
 Hayu, Kuswar, Kiranti, and Limbu languages of the Himalaya, and 
 in the Uraon, Ho, Sontal, and Gondi tongues of Tamulian India! 
 Unfortunately, he has not published his grammatical outlines of 
 these idioms, which no doubt would throw more light on the intri- 
 cate problem of the exact relationship of these tongues than pages 
 and pages of mere vocables. As far as our information of these 
 Indian dialects goes at present, I should feel inclined to doubt any 
 connexion between the Gyarung and such languages as the Ho. 
 There is a grammar, by Philipps, of the Sontal language, but it could 
 not be procured for the present Essay. With regard to the Ho lan- 
 guage, in which, according to Mr. Hodgson, similar possessive prefixes 
 exist, I can only say, that in Tickell's account of this dialect I looked 
 for them in vain. Tickell gives possessive pronouns, but no pos- 
 sessive pronominal prefixes. 
 
 But there are other features in the Gyarung grammar, to which 
 I remember nothing parallel in Tibetan or any other Bhotiya 
 dialects. I give the forms, as well as their explanation, on Mr. 
 Hodgson's authority, who occasionally quotes them in his notes. A 
 verbal root admits of a number of prefixes without any change of 
 meaning. Thus, to go is not only ching, but also y aching, 
 kaching, daching, taching, and naching. These are all used 
 in a present sense. The past is formed by putting t a between the 
 prefix and the root. Thus we get y a-t a-c h i n g, k a-t a-c h i n g, d a-t a- 
 ching, tataching, nataching, allin thesense of "Iwent." Causal 
 verbs are formed by putting sa between the prefix and the root. 
 For instance, zo, to eat; ta-sa-zo, to feed. By using ma instead of 
 the first prefix, we get a negative verb. Thus, ma-ta-ching, I went 
 not ; ma sa zo, I did not feed. Sometimes, we are told, two or three 
 indifierent prefixes may be used, for instance, da-na-ra-gyuk, instead 
 of simple gyuk, to run. The causal form of this would be again 
 da-na-ra-sa-gyuk, to cause to run; and from this again the negative, 
 ma-da-na-ra-sa-gyuk, not to cause to run. This a kind of gram- 
 matical mosaic of which one should hardly have expected a Bhotiya 
 language to be capable. But, on the other hand, it cannot be 
 said to be Turkic ; because there the verbal root always maintains 
 its place at the beginning, and though it allows a number of sufiixes, 
 in some cases even the same as those in Gyarung, at the end of 
 
109 
 
 words, on the contrary, it excludes most rigorously any prefixes. 
 The same applies to Burmese and its cognate tongues. They are full 
 of modifying verbal suffixes ; but the only verbal form which admits 
 of a prefix is the negative, formed by ma. Thus, in Burmese, 
 thwa, to go, forms its causal, thwa-za, cause to go; its past, thwa- 
 bhu-the, I went. Its negative, however, is, ma-thwa-bhu, he 
 goes not. 
 
 A more complete grammatical analysis of the Sifan and Horsok 
 tongues will be invaluable for determining the frontiers between 
 Chinese, Mongolic, Tataric and Bhotiya dialects ; and there is reason 
 to hope that Mr. Hodgson will continue his researches in this direc- 
 tion. With the present evidence we must be satisfied to know that, 
 besides the literary Tibetan, many dialects continue to be spoken, 
 particularly in the north of Tibet, which in their vocables are re- 
 lated to Tibetan, and through it to the Sub-Himalayan idioms. The 
 Sokpa dialect, however, seems to form an exception, for we can 
 hardly be mistaken is treating it as a Mongolic dialect introduced 
 into Tibet from Mongolia by nomadic tribes. 
 
 Fifth Section. 
 
 Further Extension of the Bhotiya Class, and its Subdivision into 
 Sub- Himalayan ( Gangetic) and Lohitic Dialects. 
 
 Aftee this somewhat unsatisfactory survey of the northern mem- 
 bers of the Bhotiya family, wa return to India, to trace there the 
 further spreading of the same speech south of the Himalaya. One 
 imaginary barrier, which seemed to separate the languages of the 
 second and third regions of the Sub-Himalayans, and which consti- 
 tuted the Kocch and Dhimal Tamulian, in contradistinction to the 
 Tibetan immigrants, such as Limbu, Murmi, &c., has already been 
 removed. These two groups of dialects once comprehended by one 
 general title (Bhotiya), it will be easier to advance another step, 
 and to include within the same class, many of the tribes of Asam 
 
 I 
 
110 
 
 and Burmah which have been considered of Tamulic or of Tai 
 origin. If the. Kachari dialects are once admitted to be Bhotiya, 
 and not Tamulic, the Burmese also cannot be liept separate, and 
 with it all its cognate tongues, such as Singpho, Naga, Mikir, 
 Abor, &c., have to follow. They are all non-Tamulic, and non-Tai. 
 They show none of the features which are peculiar either to the 
 cultivated or the uncultivated Tamulic dialects, either to Tamil or 
 Gond ; and where they seem to share in a common expression with 
 the Tamulic or Tai idioms, it must be explained by that more 
 distant relationship which once united all the members of the Tu- 
 ranian family, but which has left such few and solitary traces, that 
 we frequently hesitate whether to treat them as the result of acci- 
 dent or of a primitive community. 
 
 It has been said that in Turanian philology neither numerals 
 nor pronouns were of much weight to prove the relationship of lan- 
 guages, and that therefore the principles of comparative philology 
 which are applicable to the Arian languages would have to be mo- 
 dified in their application to Turanian dialects. This is true only 
 to a certain extent. It is^rue that it would be impossible to prove 
 the common origin of the Tataric and Finnic, for instance, by means 
 of their numerals and pronouns alone. We must admit that the 
 fertility of the Turanian idioms continued after their separation, even 
 with regard to these the most simple parts of speech. We have 
 only to look at the Samoiedic and Mongolic numerals, and compare 
 them with the Finnic and Tataric, in order to appreciate the truth 
 of this remark. But though the numerals in all these languages 
 are less useful for the purposes of generalisation, they are most 
 advantageous for the purposes of subdivison. The Tchuvashian, for 
 instance, formerly considered a branch of the Finnic stock, and 
 arranged together with the Tcheremissian and Mordvinian, shows 
 distinctly by its numerals that it belongs to the Tataric branch, 
 to which it has accordingly been referred. In the same manner a 
 comparative list of numerals is sufficient to show that the Kachari 
 dialects do not, as at first classified, belong to the Tamulic stock, 
 but to the Bhotiya. This is confirmed by an examination of 
 their grammar, even with so slight a sketch as that given by 
 Rohiuaon of the Kachari dialect. Nothing, indeed, shows the 
 
. Ill 
 
 peculiar relation of the TamuHc and Bhotiya languages so dis- 
 tinctly as their numerals. They are instructive in two respects. 
 We may learn from them in how high a degree the different classes 
 of the Turanian family possess that independence, and that power 
 of creating new forms and new words, which in some cases obliterates 
 almost entirely all traces of their common origin. But we learn, at 
 the same time, that in smaller spheres these dialects are as tenacious 
 of their common words as any members of the Arian family. The 
 grounds on which the general relationship of Turanian languages 
 can be proved will always lie in the general principles of their 
 grammar, so different from the grammar of both the Semitic and 
 Arian nations. But the mass of languages which on such grounds 
 would have to be referred to one family, is too unwieldy for any 
 scientific purposes. They require to be divided again, to be classified 
 and arranged so as to form an organic and well articulated whole. 
 It is here that pronouns, numerals, grammatical peculiarities and 
 irregularities assume their highest importance. 
 
 The Bhotiya languages, even after their separation from the Ta- 
 mulic class, require a more accurate subdivision ; but the materials 
 are hardly sufficient as yet to enable us to pronounce definitely on 
 this point. I shall first give a list of all the languages, which, together 
 with the Burmese, must be included within the limits of the Bhotiya 
 class. Afterwards I shall attempt to show that though they all form 
 but one class, in the most general sense, they cannot be treated as 
 such in the same sense in which, for instance, the Tamulic dialects 
 are all but varieties of one common type, 
 
 1. Lohitic Dialects. 
 
 Geographically, the languages which we have here to consider, and 
 which, with the exception of the Burmese, have been all collected from 
 the mouth of uncivilised tribes, might be called Lohitic, in contra- 
 distinction to the former division of Bhotiya languages, which might 
 very properly be designated by the name of Gangetic instead of 
 Sub-Himalayan. Lohita is another name for Brahmaputra in San- 
 skrit; only it is used here in a narrower sense, as the name of the Yarn 
 or Sanpu after its entrance into India. Under this geographical deno- 
 
 I 2 
 
112 
 
 mination, however, the dialects now spoken in tiie third regions of the 
 Himalayas also will be comprised. Although Kocch tribes are scattered 
 at present along the Tistean, and Korean, and Gandakean basins, and 
 included, therefore, within the limits of the Gangetic system, their 
 proper aqueous habitat and source seem to be in the system of the 
 Lohita, on both sides of the valley of Asam ; and there the majority 
 of these tribes yet resides. One stream of Bhotiya population would 
 seem to have reached India by way of Utsang, following the 
 trans-nivean feeders of the Ganges ; the other by way of Kham, 
 following the course of the Lohita and its Indian tributaries, which 
 east of the Chumalhari are kept, by the watershed formed by that 
 ridge from falling into the Ganges. These races extended toward 
 the East into Burmah, and toward the West along the Tarai, 
 Saul, and Mari regions of the Gangetic system. That here their 
 dominion, in times within the reach of historical memory, extended 
 considerably toward the South, is proved by the Kocch-kingdom, 
 which was absorbed by the Company in 1773. Its limits were from 
 25° to 27° North lat., and from 88° to 93^° E. long., Kocch Behar 
 being its metropolis. The Gangetic tribes also, as we saw before, 
 must formei'ly have extended much more to, the South, if we may 
 identify with the present Kirantis the Kirrhadas, whom Ptolemy 
 recognized on the Bay of Bengal. One of the ancient names of the 
 Ganges is Kirati. 
 
 The first language spoken in the valley of the Lohita is the 
 Asamese, a sister-dialect of the Bengali, and therefore of Arian 
 extraction. That it is a sister, and not a daughter, a collateral for- 
 mation, and not a corruption of Bengali, has been proved by the Rev. 
 N. Brown, in his excellent Grammatical Notices on the Asamese 
 Language. 
 
 With the exception of the Asamese, all other dialects spoken east 
 of Bengal constitute a separate class, of which the Burmese is the 
 only language which has been fixed and regulated by literary culti- 
 vation. As a political language, it is now the language of Burmah 
 Proper, and as a medium of political transactions used by the Mons of 
 Pegu and by the Mugs of Aracan (annexed 1825), and wherever the 
 supremacy of the Burmese conquerors was once acknowledo-ed. Pre- 
 vious to the foundation of the Burmese empire the languao-e of the 
 
113 
 
 Myamma* was but one of the innumerable dialects spoken in the 
 peninsula of the Iravati (pronounced Eriwadi), many of which have 
 maintained themselves up to the present day, owing to the peculiar 
 character of the country, wliich in its intricate mountain ranges 
 affords safe refuge to races fond of independence in speech and 
 customs. Even the Burmese, however, though a literary language, 
 is liable to the most violent dialectical corruptions. The word 
 for "say," pronounced rak in Aracan, is sounded like Yet by the 
 Burmese; ri, the word for water in Aracan, sounds ye in Burmah. 
 The Burmese consider an indistinct pronunciation fashionable, and 
 always " chew betel and spices while speaking.'' The changes which 
 Pali words adopted into Burmese have undergone mny give an idea 
 of the ravages to which their own words are liable : — 
 
 Sanskrit vyan^rana, consonant. Burmese, by i. 
 
 Aakra, wheel ; (Pali, iakka) „ j e k and je or tse. 
 
 marga, road; (Pali, maga) „ mag. 
 
 It seems, on the whole, that the pronunciation in Aracan is more 
 correct and distinct than in Burmah Proper. The Rukheng race, as 
 Leyden says ( A.s. Ees., X., 222.), is admitted to be of the same radical 
 stock as the Barmas or Birmans, and is understood to have greatly 
 preceded that nation in civilization. The Barmas, indeed, derive 
 their own origin from the Rukheng, whom they generally deno- 
 minate Barmakyi, or the great Barmas, and they consider the 
 Rukheng the most ancient and original dialect of the Burma lan- 
 guage. It would therefore be of much greater utility to the philo- 
 logist. 
 
 It is impossible to enumerate all the small tribes whose names have 
 been collected by travellers and missionaries. Captain Gordon alone 
 collected not less than twelve dialects in the neighbourhood of Ma- 
 
 * Myamma or Bomma, or Byamma, is the Burmese pronunciation ol 
 Marumma, the national name of the Rukheng race. The Rukheng vel 
 Aracanese are considered as the ancestors of the Burmese. As Aracan is 
 a corruption of Rukheng, Burma is a corruption of Marumma, which again 
 is said to be a corruption of the Sanskrit Mahavarma, the honorary title of 
 Kshatriya races. Mug is a name given to the inhabitants of Rukheng by the 
 Bengalis. Rukheng is originally the name of the country, and derived from 
 the Pall word Rakkhapura (abode of demons) ; the classical name of the 
 country is Dhanya wati. 
 
 I 3 
 
114 
 
 nipura: the Manipuri, Songpu, Kapwi, Koreng, Maram, 
 Champhung, Lahuppa, North-Tangkhul, Central-Tang- 
 khul, South-Tangkhul, Khoibu and Maring dialects, some 
 of them spoken by not more than thirty or forty families, yet so 
 different from the rest as to be unintelligible to the nearest neigh- 
 bours. I shall only endeavour to indicate the localities of those 
 tribes whose languages have been comprehended in the lists at the 
 end of this letter. The principal authorities I follow are again Mr. 
 Hodgson and the Rev. N. Brown. With regard to grammatical 
 questions, Mr. Robinson's articles are of the highest value. Short 
 outlines of grammar, like those given by him, for various tribes in- 
 habiting the valley of Asam and its mountain confines, will be 
 indispensable if we wish to arrive at anything like definite results 
 on the phonology of the country between India and China. 
 
 Tribes which have already been mentioned as inhabiting the 
 malarious districts of the third Himalayan region, and which will 
 have to be included within the Lohitic class of Bhotiya languages, 
 are, starting from Govalpara in Asam, and proceeding as far as Ali- 
 gang in Morang, the Kocch, Bodo, Dhimal, Rablia, Hajong, 
 Kudi, BatarorBor, Kebrat, Pallah, G-angai, Maraha, and 
 Dhanuk. Of most of these tribes we know only the names, but the 
 three first have been made familiar to all ethnologists through a 
 very able treatise by Mr. Hodgson. The kingdom of the Kocch 
 once extended in the West to the Konki, which joins the Ganges 
 near Rag'mahal. Their proper name is said to be KavaAa, prakri- 
 tised into Kocch. They are called Hasa by the Kacharis of Asam, 
 Kamal by the Dhimals, and Kocch by the Mecch. In Asam they 
 are divided into Kamthali and Madai or Shara, and Kolita or 
 K holt a. The mass of the Kocch people have become Mohammedans, 
 and the higher grades Hindiis ; both style themselves Ragrvansi, 
 Few only adhere to the language, creed, and customs of their fore- 
 fathers, so vividly described by Hodgson. The language of the un- 
 converted Kocch has not yet been published. 
 
 llie northern and eastern skirts of the Kocch country are inha- 
 bited by Bodo and Dhimal. The Dhimal are to be found as far 
 west as the Konki ; their numbers are small, and they are generally 
 mixed with the Bodos. These are very numerous, and extend eastward 
 
115 
 
 to the Dhansri (Subanshiri ?), or even beyond, and occupy besides 
 a large proportion of Central and Lower Asam. They are also called 
 Kachdris, and Borros, which probably is the same word asBodos. 
 Their principal locale is said to be Chatgari, where they amount to 
 about 30,000 souls. The whole number is estimated at between 
 150,000 and 200,000, which includes the Mechis ofthe West, andtlie 
 Kacharis ofthe East and South. Sometimes Kachari is used in a 
 more general sense, comprising the Hogai Kacharis of the plains, the 
 Kochis (including Modai Kochis, Phulguriyas and Hermias), the 
 Mechis, Dhimals and Eabhas. Hodgson has given lists of words 
 of the Bodo and Dhimal ; Robinson a grammar of the Kachari 
 dialect, which is the same as Hodgson's Bodo. Robinson's Kachari 
 words were supplied by Captain Gordon. 
 
 If we can trust the traditions of the Kacharis, their ancient name 
 was not Kachari, but Rangtsa, and the country from which they came 
 was situated north-east of Asam. They are said to have conquered 
 the old kingdom of Kamarupa, and to have founded there the royal 
 Dynasty of the " Ha-tsung-tsa." This Ha-tsung-tsa Dynasty was 
 expelled again by the Rap'as of Kocch Behar, and maintained itself 
 in Hirumbha alone to 1130. Now, as in the twelfth century, at 
 the commencement of the Ahora Dynasty in Upper Asam, Kamarup 
 was already in the hands of Kocch-Behar princes, the Kachari 
 Dynasty may have been founded, as the Kachari chiefs assert, about 
 a thousand years ago. Captain Fisher, who collected this information 
 during his residence in Kachar, also asserts that the few remaining 
 traces of the former religion of the Kacharis resemble the system of 
 Confucius more than anything else. Brahmanism wras introduced into 
 Asam in the sixteenth century, but in Kachar Proper, or Hirumbha, 
 its diffusion commenced not more than sixty years ago. 
 
 The Garos also are sometimes classed with the Kacharis, with 
 whom no doubt they are closely allied physically and linguistically. 
 They live, however, in a completely savage state : occupying a trian- 
 gular extent of mountainous country between the left bank of the 
 Lohita and the Khassia Hills. Garo words published by the Rev. N. 
 Brown and Mr. Hodgson, the grammar by Robinson. 
 
 The Changlo, who are only known by Robinson's researches, 
 occupy a portion of the northern frontier^of the valley ofthe Lohita, 
 
 I i 
 
116 
 
 extending from the Binji Duwar to the confines of the Kuriapara 
 Duwar. Their northern limit is unknown. Changlo in their lan- 
 guage means black. Their grammar has been published by Robinson. 
 
 The Mir is seem to have their chief seats in the low hills 
 north of Banokotta and Lukimpur, whence they were pressed into the 
 plains of Asam by their formidable neighbours, the Abors. Gram- 
 mar published by Robinson ; words collected by Robinson, and another 
 list by the Rev. N. Brown of Sibsagor, published by Mr. Hodgson. 
 Robinson's Miri numerals coincide more with Captain Smith's Abor 
 Miri, than with those dictated by a Miri, at Sibsagor, to N. Brown. 
 
 The Mir is are said to resemble the Karens (see page 379.) more 
 than any people in the valley of the Brahmaputra. According to 
 Mr. Cutter, their dress is precisely the Karen. They live in small 
 villages in high raised houses like the Karens, but never stop 
 more than a year in a place. They are scattered along the banks of 
 the river from Bisnath up to Sadiya, and some distance up the 
 Dihing. They speak the language of the Abors, a numerous and 
 powerful race, inhabiting the highest ranges of mountains on the 
 north bank of the Brahmaputra, between Jorhat and Sadiya. 
 
 The Abors or Bo r -Abors occupy an extensive range of moun- 
 tainous country along the southern exposure of the Himalaya chain, 
 reaching, it is said, as far as Tibet and China. They are to be found 
 on each side of the river Sampu. Of them we have only lists 
 of words, published by Robinson, after materials furnished by the 
 Rev. N. Brown ; and by Mr. Hodgson as Abor Miri words, from a 
 Vocabulary by Captain E. F. Smith. The lists do not exactly coin- 
 cide. The Rev. N. Brown gives specimens of a language called Aka, 
 and closely connected with the Abor. The language of the M i s h i m i 
 also, spoken east of the Abor country along the Sampu, has been 
 examined by the same active missionary. It consists of three dialects, 
 and is connected with Abor and Aka languages. 
 
 The territories of the Singphos extend from the Patkoi range to 
 the Lohita, and from the frontier of Asam to the Langtang mountains 
 in the east. Singpho is the language of extensive tribes occupying 
 the northern portions of the Burmese empire. The grammar is pub- 
 lished by Robinson, with the assistance of the Rev. M. Bronson. 
 Lists of words are given by Mr. Hodgson after a vocabulary published 
 
117 
 
 by the same. According to Mr. Eobinson, one-fourth of the vocables 
 in Singpho are allied to the Burmese, and an equal proportion to the 
 dialect of Manipur. As a dialect of the Singpho, the Jili language 
 is mentioned by the Rev. N. Brown. This tribe vras driven out of 
 their seats by the Singpho, and is now nearly extinct. 
 
 The Dophla s inhabit the southern face of the Himalaya from 92° 
 50' to about 94° north latitude, which forms the northern boundary of 
 the valley of Asam, from the Kuriapara Duwar to where the Suban- 
 shiri debouches into the plains. They call themselves Bangni, men. 
 — Grammar and vocabulary published by Robinson. (J. A. S. B. 
 1851. p. 126.) 
 
 The Mikir occupy a tract of hilly country situated within the 
 boundaries of the district of Nowgong (New-village) in central Asam. 
 -Numerous families are scattered in the south of Asam. — Grammar 
 and vocabulary published by Robinson. 
 
 The Naga tribes are most difficult to localize. They are divided 
 into many branches and scattered over a large extent of country. 
 Robinson circumscribes their habitat on the west by the Kopili 
 river, the great southern bend of the Barak and the eastern frontier 
 of Tipperah, in nearly east longitude 93° ; on the north, by the valley 
 of Asam ; on the east and south east, by the hills dividiug Asam from 
 the Bor-Khamti countiy in longitude 97°, and the valley of the Kyen- 
 drens ; and on the south, by an imaginary line nearly corresponding 
 with the 23rd degree of north latitude. Their name "Naga" seems 
 to have been given to them by the Brahmans. Their own name is 
 " Kwaphi." 
 
 The Namsangiya-Nagas occupy the hills near the sources of the 
 Buri Dihing river. Their grammar and vocables have been published 
 by Robinson with the assistance of the Rev. M. Bronson. Of other 
 Naga tribes we have vocabularies only. Two taken down by the Rev. 
 N. Brown from two Nagas at Nowgong, and published by Mr. Hodg- 
 son as Nowgong and TengsaNaga. Four more were published 
 afterwards from the same sources. Of them the Mi than or Tab- 
 lungNagas reside on the hills' east and north of Sibsagor. Their 
 neighbours the Jabokas and Banferas speak nearly similar tongues. 
 The Kharis descend upon the plains near Jorhat. The Angamis 
 occupy the southern end of the Naga country. The Naga tribes are 
 
118 
 
 scattered, but not migratory like the Kacharis and Kukis. The 
 latter, called also Kunjye, are generally reckoned as Nagas, but 
 differ from the Nagas in customs and appearance. The Kukis 
 are not robbers only, but murderers by profession, and they are ac- 
 cused of cannibalism. Another name of the Kukis is Lunkta, i. e. 
 naked, and the name of the Nagas is likewise explained as if derived 
 from the Sanskrit " Nagna," naked. 
 
 The last contributions towards the phonology of these countries 
 consist in vocabularies of languages spoken by the tribes in Arakan 
 (vel Rakheng), collected by Captain Phayre*, and published by 
 Mr. Hodgson. They contain no grammatical outlines. The geogra- 
 phical position of these tribes is here laid down according to Captain 
 Phayre's indications. Ea-khoing-pyi, or the country of Aracan, lies 
 between 20' and 21° 10' N. lat. on the sea-coast; in the interior it 
 extends to about 21° 40'. 
 
 The Khyeng live in the high range of mountains called Yuma, 
 separating Aracan from the valley of the Iravati. They seem to be the 
 same as Dr. Buchanan's Kiayn, only that according to him they 
 term themselves Kolun, but according to Captain Phayre, Shy u. 
 Karieng or Karayn also seem to be mere variants of Khyeng, r 
 and y being frequently interchangeable in these dialects. 
 
 The Karens have lately attracted much attention by their em- 
 bracing Christianity with high zeal and earnestness. The labours of 
 the American Baptist missionaries seem to have met with a success 
 hardly precedented in the annals of missionary enterprise. I take 
 the following notices from an interesting memoir by a Karen mis- 
 sionary, " The Karens, or Memoir of Ko Thah-Byu, Tavoy, 1843." 
 
 The Karens have well-defined traditions of being comparatively 
 recent emigrants in Tavoy. They say, " The elders said, we came 
 down from the upper country. At first we settled on the Attaran ; 
 next we came to Ya ; and finally to Tavoy." Though their dialects, 
 as spoken at Tavoy and Maulmain, differ, the Karens on Belu Island 
 at the mouth of the Salwen, are said to speak precisely the same 
 dialect as that of Tavoy. 
 
 All Karen tribes seem to agree that they have not been long in 
 
 * All Arakanese are termed M u g s by the people in India, although the Mugs 
 are only a very small race in Aracan, and not of pure Myam-ma descent. 
 
119 
 
 Siam, nor are there any Karens in Siam except on the western side 
 of the Meinam. There are but few in Aracan, and these confined 
 to the southern province of Sandoway. The valleys of the Irrawaddy 
 and Salwen bear the most distinct traces of their gradual progress. 
 But beyond this, all is tradition, which points, however, to Tibet as 
 the original home of the Karens. 
 
 I quote a very interesting and important testimony from a work 
 by Mr. Kincaid. The result of all my inquiries, he says, is that 
 Kakhyen is only another name for the Karens. All these moun- 
 tain tribes, through the whole extent of the Shyan country, and 
 further north into Tibet, are called Kakhyens, except in the Hukong 
 valley, between Mogaung and Asam, where they are called Thing 
 bau-Kakyen. The whole mountain country between Mogaung and 
 Cathay is inhabited by the same people. Around the Martaban 
 gulf, and thence inland as far as the Burman population has ever 
 extended, the mountain tribes are called Karens. Between Rangun 
 and Tung-u, and between Tung-u and Ava, they are very nu- 
 merous : also between Tung-u and Monay, a Shyan city, about 
 250 miles east of Ava. There are some tribes scattered along be- 
 tween Burmah and the Shyan states, called Karen-ni, red Karens, 
 and these extend as far as Zimmay. These are less civilized than 
 those who live in the vicinity of Burman cities. Some have erro- 
 neously considered them as belonging to the Shyan family. Their 
 language and everything else pertaining to them is Karen. In ad- 
 dition to this, the south-east part of Tibet is inhabited by Ka-Khyens ; 
 at least I have reason to believe so, as the Shyans, who live in the 
 most northern part of Burmah, and adjoining Tibet, call the country, 
 " the Kakhyen country." This is partly confirmed by Malte Brun, 
 who, arguing from the accounts of Marco Polo, says, " Thus the 
 country of Caride is the south-east point of Tibet, and, perhaps, the 
 country of the nation of the Cariaines, which is spread over Ava. 
 It will be seen, then, that these mountain tribes are scattered over a 
 vast extent of country, and their population is estimated at about five 
 millions." 
 
 " Tin," the Chinese appellation for the Divinity, exists in Karen 
 poetry as the name of a false god, whom they regard as worshipped 
 by a people with whom they were formerly in contact. 
 
120 
 
 The similarity of the Karens and the Miris has been mentioned 
 before, and in a vocabulary of seventy words published to illustrate 
 the language of the Miris and similar tribes, about fifty, with slight 
 modifications, were found in one or other of the Karen dialects. 
 Their country extends from about the 19° to the 20° N. lat. 
 
 The Kami (vol Kimi) and Kumi are two divisions of a race 
 inhabiting the hills along the river called Kuladan (limit of the 
 Kulas or foreigners?) by the Aracanese, Ye-man by the Kamis, 
 and Yan pan by the Kumis. The Aracanese distinguish the two 
 tribes as Awa Kumi and Aphya Kumi. 'I hey are tribes press- 
 ing onward in a south-western direction. They inhabited formerly 
 the seats now held by the Khyeng, and drove the Mm out of the 
 country which they themselves now occupy. The Kami language has 
 been reduced to writing by the Rev. Mr. Stilson, of the American 
 Baptist mission. 
 
 The Mru (vel Myu, vel Tung Mru) driven westward by the 
 Kami, now inhabit the hills on the border between Aracan and Chit- 
 tagong. According to the ''Ra^avansa," tlie history of the Aracanese 
 kings, the Mru were in the country when the Myamma or Burmese 
 entered, and one of their tribe was king of Aracan in the I7th century. 
 
 The Sak (vel Thock) inhabit the eastern branch of the Nauf 
 river, and are called Chain and Chanmas by the Bengalis. 
 
 There is another curious tribe, called Lung-khe, first mentioned 
 by Lieutenant Phayre in his account of Arakan. They live on the 
 upper course of the Kuladan, and generally west of that river. They 
 are sometimes called Bo ung-j u. (Bunzus ?) Both Lungkhes and 
 Boungjus, it seems, were conquered, and are now governed by 
 a third tribe, called Shindus by the Kumis, but by themselves 
 Hling-ju. Their chief, Leng-Kung, who was examined, shortly 
 before be was poisoned, by Lieutenant Phayre, gloried in his descent 
 from that powerful tribe whose seat is N. E. of the Lung-khes, and 
 whose country is fifteen days' journey in extent. He said that the 
 Lung-khe and Shiudu languages are nearly alike. Perhaps we have 
 a specimen of this very language in Captain Tickell's article on the 
 Heuma or Shend u s (J. A. S. B. 1852, p. 207.). Their country is 
 placed by Captain Tickt-ll between lat. 22° and 23° N., and lono-. 
 93° and 94°. Some of the particulars mentioned by Leng-kung of 
 
121 
 
 the Shindus, and by Lebbey of the Shendus, agree. Both bury 
 their dead, while the Kumis burn them. The Shendu, according to 
 Lebbey, make their houses of timber ; the poorer classes only, of 
 bamboo. The Shindus build their houses entirely of plank ; nay, 
 Leng-kung declared that there were no bamboos in the Shindu 
 country, a fact doubted by Captain Phayre. The Shendus and Shindus 
 thatch with grass. Both infest the neighbourhood of Chittagong. The 
 names of their clans, however, differ, and there are other facts men- 
 tioned which make it doubtful whether the Shendus are really iden- 
 tical with the Shindus. The Shendus, for instance, are stated to buy 
 salt ; the Shindus manufacture it from brine-springs in their own 
 country. 
 
 Other tribes are mentioned by Captain Phayre, such as the Daing- 
 nak (speaking a corrupt Bengali), Moung, and Khyau, but no 
 vocabularies for them have as yet been published. The only addition 
 to our knowledge of these dialects is a list of words of the Tung- 
 Iho collected by Dr. Morton, and published by Mr. Hodgson. The 
 Tung-lho live in the Tenasserim provincesj and were recognised 
 at once by Mr. Hodgson as dislocated aborigines driven to the 
 wilds, or as broken and dispersed tribes like the Khyeng, Kami, 
 Kumi, Mru and Sak of Aracan. Leyden also knew that the T a n- 
 engaari, or language of the Tanaserim districts, denominated 
 Tinnaw by the Siamese, is only a rough dialect of the common 
 Burmese. 
 
 2. General Coincidences between the Sub- Himalayan (Gangetic) 
 and Lohitic Divisions of the Bhotiya Branch. 
 
 If it be asked why all these dialects from the Tista to the Iravati 
 are referred to one class, I can here only point to their numerals for 
 an answer, and to the comparative lists of words given in Mr. 
 Hodgson's last articles. The grammatical genius also of these dialects, 
 so far as it can be discovered from the scanty grammatical outlines 
 of a few of them, is uniformly the same. The system of accents 
 or intonations is common to all, and, with the exception of the 
 Naga dialects, none distinguishes the persons of the verb by either 
 affixes or prefixes. In some of these idioms, the meaning of words. 
 
122 
 
 whether nouns or verbs, is modified by additional syllables placed at 
 the end of the substantive or the root, while the persons of the verb 
 have no such distinction. The actual coincidences in the expressions 
 for gender, number, cases, and verbal derivatives are numerous, but 
 it would require too much space here to place them in their proper 
 light. I shall only mention a few which are, of course, restricted to 
 those dialects of which Eobinson has given us grammatical outlines. 
 But, judging from the general similarity of words between all, 
 I believe that the same remarks will be found to apply to the 
 other dialects belonging to this class of which as yet we possess 
 vocabularies only. 
 
 Gender in all Lohitic and Sub-Himalayan (Gangetic) dialects, with 
 the exception of foreign terms, is expressed by separate words, and 
 is always restricted to the expression of natural sex. The usual terms 
 for male and female are pa and ma (nya), though other expressions 
 occur, and in some dialects pa and ma are restricted to a certain 
 class of animals. For instance : — 
 
 Tibetan : 
 
 phag, pig; 
 
 phag-p 0, male ; 
 
 phag-ma, female. 
 
 Chaaglo : 
 
 kurta, horse ; 
 
 kurta-p ho, „ ; 
 
 kurta-mo, „ 
 
 Garo : 
 
 achak, dog ; 
 
 achak-bip ha, „ ; 
 
 achak-bima, „ 
 
 Kaehari : 
 
 ; 
 
 bi-pha, father ; 
 
 bi-ma, mother. 
 
 Miri: 
 
 eki, dog ; 
 
 ki-baka, male; 
 
 ki-neka, female. 
 
 Singpho : 
 
 gui, dog; 
 
 gui-1 a s a, „ ; 
 
 gui-numsa. „ 
 
 Naga: 
 
 hui, dog ; 
 
 hu-pong, „ ; 
 
 hu-n y o n g, „ 
 
 Mikir: 
 
 kipi, a monkey ; 
 
 kipi-alo, „ ; 
 
 kipi-ape, 
 
 Burmese : 
 
 ngan, a goose ; 
 
 ngan-hpa, „ ; 
 
 ngan-ma, 
 
 I have given the list complete, in order to show the rule as well as 
 the exceptions. It is also necessary to remark, that, with the excep- 
 tion of the Tibetan, where it is said that po and ma may be pre- 
 fixed as well as affixed, all other dialects invariably place these 
 words expressive of gender at the end. 
 
 Another general feature of these dialects consists in forming the 
 plural by an affix expressive of plurality. Now, as the idea of plu- 
 rality admits of a much larger number of expressions than that of 
 gender, it is natural that there should be a greater variety in the 
 plural affixes of the same idioms. Two, the Singpho and Mikir, 
 are said to exclude all expression of plurality ; the Singpho, for 
 instance, can only say apanang (many) arleng (man), in order 
 
123 
 
 to express the plural. It should be observed, however, that both 
 Singpho and Mikir have plural affixes for their pronouns. Now, 
 with regard to the syllables employed to express plurality, we find 
 dag in Tibetan, thamcha in Changlo, and ki-ding in Miri. 
 Tibetan and Changlo both employ nam ; and Garo and Miri co- 
 incide in another plural affix, rang and arang. ' This rang seems 
 to be the same as the Aracanese ro ; and as the Burmese cannot 
 pronounce the r, their do may come from the same source, though 
 generally a Rukheng r corresponds to a Burmese y. 
 
 With regard to the case-terminations, or rather the post-positions 
 fixed upon to express the relation of noun to noun, it is extremely 
 difficult to institute comparisons. There are no fixed cases, such as 
 genitive, dative, ablative, and the same post-position has adopted a 
 variety of meanings in various dialects. As by in English may 
 mean near to or by means of, and therefore correspond to a 
 locative and to an instrumental, the same particle in these Lohitic 
 dialects is sometimes made to serve opposite purposes. Still even 
 here coincidences are not wanting if we examine carefully the 
 paradigms given by Robinson. What is called, for instance, the in- 
 strumental, is expressed, in Tibetan by kyi, 
 
 Changlo by gyi, 
 
 Garo by chi, 
 
 Miri by koki, 
 
 Burmese by si.* 
 
 With respect to adjectives, it may be observed that in all these 
 dialects, with the exception of the Burmese, they are either usually, 
 or in some, invariably, placed after the word they serve to modify. 
 
 * I cannot help thinking that in some cases these post-positions, although they 
 form part of the grammar, are words taken from Sanskrit or Asamese. In Asa- 
 mese, p a ra, a Sanskrit word, is used as an ablative post-position. In Mikir, the 
 sign of the ablative is para; in Kachari, phrai ; in Burmese, phraen. Now, 
 that in Burmese Sanskrit post-positions are used becomes almost palpable in the 
 case of kraun. Kraun is in Burmese the sign of the instrumental and the 
 ablative. Both might be expressed by the Sanskrit karana, cause, means ; and in 
 Asamese, karane is actually used as an instrumental post-position. But the same 
 kraun in Burmese is used as sign of the infinitive, and there again the Sanskrit 
 karana, which originally means " doing," would be in its proper place. Coinci- 
 dences like these can hardly be accidental. 
 
124 
 
 Tibetan : 
 
 mi ngam, 
 
 man bad, 
 
 (optional^. 
 
 Changlo : 
 
 aba changlo, 
 
 crow black, 
 
 (general). 
 
 Garo : 
 
 mande nama, 
 
 man good, 
 
 (always). 
 
 Kachari : 
 
 manse gaham, 
 
 man good, 
 
 (optional). 
 
 Miri : 
 
 amie a'ida, 
 
 man good, 
 
 (always). 
 
 SiBgpho . 
 
 singpho kunkan, 
 
 man idle. 
 
 (always). 
 
 Naga : 
 
 asan koa, 
 
 milk good. 
 
 (always). 
 
 Mikir : 
 
 aso kangtuk, 
 
 boy fat. 
 
 (always). 
 
 3. Distinctions between Sub-Himalayan {Gangetic) and Lohitic 
 
 Dialects. 
 
 But thougli it must be- admitted that these Lohitic dialects form 
 one class, it might be asked on what grounds they are to be 
 separated from the Gangetic dialects on the north-west, and from 
 the Ta'i dialects on the south-east. With regard to the latter, the 
 answer will be given when we come to the Ta'i dialects. As to the 
 distinction here proposed between the Gangetic and Lohitic dialects, 
 I admit that it is more or less provisional, and that in any case I 
 should always look on these two streams of language as sprung 
 from the same source. Both have a right to the general title of 
 Bhotiya, and their connection with the language now spoken in Tibet 
 is transparent throughout. But these two streams must have reached 
 India at different times, and having been exposed for centuries 
 to various influences, may thus have grown into two separate dia- 
 lects like the Ionian and jEolian dialects of Greece. I said before 
 that words can only indicate general relationship, and that for the 
 purposes of systematic arrangement we must rely on pronouns, 
 numerals, and grammatical features. Unfortunately, we have no 
 grammatical outlines of any Gangetic dialect, and we are obliged, 
 therefore, to fall back on numerals and pronouns. Now, with regard 
 to the pronouns, there is one well-marked feature to keep the Gangetic 
 apart from the Lohitic class. The pronoun of the second person in all 
 the seventeen Lohitic dialects of which specimens are available begins 
 with n. In the Gangetic dialects the initial letter is an aspirated 
 guttural, with the single exception of the Magar, and perhaps the 
 Murmi. A feature of this kind, where we can hardly suppose a merely 
 phonetic corruption as the cause of difference between the pronoun 
 
125 
 
 of the Gangetic and Lohitio dialects, would be sufficient in Arian 
 philology to shake our confidence as to the common origin of these 
 two classes of speech. This need not be, where we have to deal 
 with Tui'anian dialects. For in them even the pronouns, in other 
 respects the most abstract and therefore most firmly settled parts of 
 speech, are afiected by that lingual superfetation which likes to 
 express by two and more words the different aspects even of the 
 most simple ideas. Some Turanian languages revel in a variety 
 of pronominal forms shading them according to the light in which 
 they look upon the person addressed, or wish to be looked at by him. 
 Some of these pronouns are only ceremonial expressions ; others, 
 however, have become real pronominal bases. In the case before us, 
 we find that the Chinese, which is a model of pronominal politeness, 
 possesses two real bases for the pronoun of the second person, one 
 consisting of the dental nasal, the other of the aspirated guttural : 
 the former, ni, being used in the Kuanhoa; the latter, ghou, in the 
 ancient Kuwen. 
 
 The numerals in the Lohitic and Gangetic dialects do not disguise 
 their common origin. Their pronunciation no doubt varies con- 
 siderably, but it never exceeds the bounds of analogy. In both 
 classes, but more particularly in the Lohitic, the numerals have 
 been encumbered with prefixes and affixes, which sometimes distort 
 the features of the original numerals to an extent that renders their 
 appearance entirely different. Still here also, analogy helps us to 
 separate what is additional from the primitive root. These various 
 changes, whether produced by corrupt pronunciation or by addi- 
 tional syllables known as " generic particles," make it impossible to 
 discover any broad features by which a set of Gangetic could at once 
 be distinguished from a set of Lohitic numerals. If the Tibetan 
 and Burmese numerals could be used as types and representatives 
 of the two classes, Gangetic and Lohitic, it would be easy to point 
 out characteristic distinctions between the two. But if we look at 
 the variations to which both the Tibetan and Burmese numerals 
 are liable, in the endless ramifications of their living progeny, or 
 rather of their kin, what seemed at first characteristic and distin- 
 guishing marks of the two, disappear again before the general like- 
 ness of the whole family. I have no doubt, however, that a more 
 
 K 
 
126 
 
 intimate acquaintance with the grammar of the literary language of 
 Tibet and Burmah wiU enable other scholars to discover the distin- 
 guishing features of these two languages in those ruder dialects 
 also which are spoken between India and China, and which, though 
 they may ultimately flow from the same fountain-head, have under- 
 gone considerable modification in their respective courses. 
 
 Sixth Section. 
 Tal Branch. 
 
 § 1 . Survey of Tat Languages. 
 
 There is another class of languages spoken on the confines of 
 Eastern India and Western China, which might be passed over as 
 foreign to our present inquiry, if some of the dialects belonging to it 
 had not been mixed up with the Lohitic class. In so vast a subject 
 as that of the Turanian family of speech, we must guard most 
 carefully against confusion, which wiU necessarily arise unless we 
 succeed in subdividing this large domain of philology. Now, with 
 regard to the Tai languages, it can not be denied that by their 
 roots they cling to the same soil from which the Bhotiya, or the 
 Gangetic and Lohitic, dialects sprang. But as we distinguish in 
 a tree between its roots, its trunk, its stems and branches, we 
 must try to do the same for language. No scholar would compare 
 Sanskrit and Italian, Celtic and Hindustani, although ultimately 
 they can be traced back to the same origin. StiU less could 
 the Naga dialects be classed with Khamti, as Mr. Robinson pro- 
 poses. For if it be a peculiar feature in the Tai languages, that 
 they are monosyllabic and destitute of inflections, surely the Naga 
 dialects are the very last to be brought under the same category. 
 Mr. Hodgson, therefore, was right when, in his lists of words, he kept 
 the Khamti distinct from the Naga dialects, and his correspondent, 
 
127 
 
 the Eev. N. Brown, declares emphatically that he does not even 
 believe in any very close radical connection between Khamti and 
 Burmese. " This affinity," he says, " seems always to have been taken 
 for granted, as a matter of course, but without any just ground. It 
 is true there are a considerable number of Burman words in the 
 Khamti, but they bear the marks of recent introduction, and are not 
 to be found in the old Ahom, the parent Shyan, nor in the Siamese, 
 with which the Ahom was nearly, if not exactly, identical." 
 
 The languages which belong to the Tax class, and interest us 
 with reference to their Lohitic neighbours, are the Siamese, Ahom 
 (Shyan), Laos, Khamti, and Kassia. The Siamese language 
 was formerly, and is still called Say am a phasa, or the Sayam lan- 
 guage, say am being, according to Bishop Pallegoix, the same as the 
 Sanskrit «yama, brownish, which ia said to be the original name of 
 Siam. Shyan seems to be a corruption of this name. But the same 
 language is more frequently called phasa thai, which means 
 " language of Free-men," a name which the Siamese assumed after 
 shaking off the yoke of Kamboja. Hence the whole class of these 
 dialects has received the general title of Tai or Taic languages. 
 This title may seem not very appropriate, considering its original 
 meaning. But as it has already become a usual term, it may be 
 retained for the present. Most of these languages have alphabets of 
 their own. There is a Khamti and Shyan alphabet, both derived 
 from the Burmese; and a Laos alphabet, derived from the same 
 source, but better adapted to the wants of the language ; and the 
 Siamese alphabet, also related, but more distantly, to the Burmese. 
 A comparative table of these alphabets, promised by Mr. Robinson, 
 has not yet been published. 
 
 The Eev. N. Brown has first drawn attention to the curious 
 contrast between the Naga and the Tai dialects. While the former 
 exhibit an extraordinary exempMcation of the manner in which an 
 unwritten language may be broken up even upon a small extent of 
 territory, the great Tai family offers a not less striking instance of 
 the preservation of language, in almost its original integrity and 
 purity, through many centuries, and in spite of a vast territorial 
 diffusion. For from Bankok to Sadiya, along the Menam, Salwen, 
 Irawadi, and Kyendwen rivers, up to the sources of the Irawadi, 
 
 K 2 
 
128 
 
 through 14 degrees of latitude, there is but one language, not- 
 withstanding the diversity of governments under which the speakers 
 live. 
 
 The discrepancies between the Klamti and Siamese, spoken at the 
 two extremities of this linguistic area, seem to be very trifling. Nine- 
 tenths of the fundamental words, according to the Rev. N. Brown's 
 calculations, are the same in these dialects, with the exception of 
 slight variations in pronunciation. 
 
 The localities of the Tai languages mentioned above are known with 
 tolerable accuracy. While the Lohitic languages seem to spread in a 
 southern direction, the Siamese dialects have at present rather a ten- 
 dency toward the north. The Siamese conquered Asam. The proper 
 name of these conquerors was Shy an* (or Shan), but the conquered 
 nations gave them the name of Ahom (the Sanskrit asama), which in 
 Asamese means "unequalled." What is called the Ahom language is 
 now nearly extinct. Though the present Ahoms of Asam, the de- 
 scendants of the conquerors, still form one of the largest portions of 
 its population, they have relinquished their language and their 
 religion for that of the Hindus. The Ahom is now understood only 
 by a few Ahom priests who still preserve their old religion. f 
 
 The Khamti is the most northern branch of the Tai family. Its 
 position among Lohitic dialects becomes intelligible if we suppose that 
 the Khamtis were driven northward by the same impulse which 
 brought the Siamese as conquerors into Asam. Though separated from 
 the Ahom, it is only through it that the Khamti can be historically 
 linked to the Siamese, to which no doubt it belongs linguistically. 
 
 The same applies to the Kassia (or Khyi) language, which is 
 spoken in the mountain territory surrounded in the north by the 
 valley of Asam ; in the west by the Garo hills ; in the south by the 
 district of Sylhet ; and in the east by Kachar. Of the Kassia we 
 have a grammatical sketch by Mr. Robinson. Captain Fisher, in his 
 Memoir of Sylhet, says that their language exhibits no affinity with 
 any of the languages of the neighbourhood, but that a people re- 
 sembling the Khyi in some particulars formerly occupied a position 
 on the south bank of the Brahmaputra, at Measpara, where they were 
 
 ♦ N. Brown's Grammatical Notices, xxtI. ^f Ibid., p. iv. 
 
129 
 
 called Mek, and that these came originally from the frontiers of 
 Butan and Nepal. 
 
 As the conquerors of Asam were Siamese, we may understand 
 how the ancient Ahom came to be so closely allied to the Shy an or 
 Siamese. According to the Rev. N. Brown, the Ahom was nearly, 
 if not exactly, identical with the Siamese. Grammarians distinguish 
 between a vulgar, a high, and a sacred dialect of the Siamese. The 
 vulgar dialect admits but few foreign words, from Chinese, Malay, Lao, 
 and Eamboja sources. The high and sacred dialects are full of 
 Sanskrit and Pali words, changed according to the genius of the 
 Siamese tongue. It is extraordinary how, after this exposition, 
 Bishop Pallegoix, the author of a Thai grammar, could maintain 
 that the Thai and Lao languages derived their origin from two 
 Brahmans who founded the town of Sangkhalok in the year 150 
 Phra Khodom, i. e. 393 b. c. The language of Siam is surrounded 
 by five or six different classes of language. In the west it is included 
 by the Burmese and cognate Lohitic dialects. At the northern 
 point its frontier line touches the Arian territory, the Asamese. 
 Hence, turning round toward the east, it comes in contact with 
 Gangetic dialects, while the whole eastern frontier is formed by 
 the Chinese and Cochinchinese languages. The most southern 
 limit of the Siamese touches the realm of the Malayan speech, while 
 the language of Pegu, the Mon*, the south-eastern neighbour 
 of the Siamese, is still of unknown origin. 
 
 The name of Laos on our maps, in the interior of the country 
 between the Menam and Mekhong rivers, indicates the locality of the 
 Lao language. The language is only known by collections of words, 
 which are very like the Siamese. It is a language rich in literature, 
 and a country full of relics of a former civilisation and Buddhist 
 hierarchy. 
 
 * A list of words given by Dr. Morton, of the Mon or Talien, shows coinci- 
 dences ■withKamboja words from the Mekhong river. The Burmese and Siamese 
 both affirm that the Mon language has no affinity with their own speech. 
 
 K 3 
 
130 
 
 § 2. Relation of the Tai to the Lohitic Languages, and their 
 Connection with the Bhotiya Class and Chinese, 
 
 We have now to consider some of the grammatical features which 
 are peculiar to these languages, and by which they are held together 
 as a class, and kept distinct from their Lohitic neighbours. The 
 materials available for this purpose are small There are lists of 
 words of all of these dialects ; but it is only of the Khamti, the 
 Kassia, and Siamese, that we possess grammatical outlines, and those 
 again but very slight. 
 
 If there are languages which can properly be characterised as 
 monosyllabic, this title may be applied to the Tai languages. 
 Certainly tlie Bhotiya dialects, whether Gangetic or Lohitic, cannot 
 lay claim to this title, though it is usually bestowed on them. If 
 from a Burmese root or a base, "kung," which means good, or to 
 be good, we may derive a-kung, good, kung-khyeng, goodness, 
 and kung-than, good — we cannot call a language like this mono- 
 syllabic. If monosyllabic means only a language which by a more 
 or less difficult analysis can be reduced to monosyllabic elements, 
 then Sanskrit is monosyllabic also. But if it means a language in 
 which the speaker feels eveiy syllable as a distinct sound, expressive 
 of a distinct meaning, then I doubt whether even the Chinese can be 
 called entirely monosyllabic. No languages, however, come nearer, 
 or, I should say, no dialects are less removed, from the Chinese status 
 of grammar than the Tai languages. 
 
 The system of musical accents or intonations, though it exists in 
 the Gangetic and Lohitic* dialects also, is said to be much more 
 
 * Mr. Hodgson (1853. p. 128) says, the principle of the tonic or accentual va- 
 riant has most erroneously been supposed to be exclusively Chinese and Indo- 
 Chinese, whereas it prevails far and wide, only more or less developed ; most, 
 where the servile particles and so-called silent letters are least in use ; least, 
 where they are most in use ; so that the differential and equivalent function of all 
 three peculiarities — that is, of empty words, of silent letters, and of tones 
 — is placed in a clear light. — The language of Nepal Proper is remarkable for its 
 numerous tones and its scanty serviles, whether literal or syllabic. According to 
 the Rev. N. Brown, Chinese distinguishes eight, the Tai languages five or six, the 
 Karen five or six, the Burmese three accents. In the modern Chinese, as, for in- 
 stance, in the dialect of Shanghai, eight tone-accents are observed. 
 
13] 
 
 marked in the Tai class. This would be so of necessity, because 
 these dialects abound in sounds organically the same, but expressing 
 ideas totally different. In Khamti, for instance, 
 
 m a with the rising tone signifies a dog, 
 » ma with the falling tone signifies to come, 
 
 ma with an abrupt termination signifies a horse. 
 
 In Siamese, khai, as it is pronounced with different intonations, 
 may mean, who ?, egg, fever, to open, rough, camp, to sell. The 
 number of accents in Siamese is fixed at five, tonus rectus,* 
 circumflexus, demissus, gravis, and altus. 
 
 This system of accents, however, by no means excludes the 
 possibility of composition. We are apt to imagine that as long as 
 every syllable has an accent of its own it remains independent, and 
 does not enter into composition. This is true with regard to our 
 accent, which is of a logical or etymological nature; but it does 
 not apply to accents like those in Chinese and Siamese. The 
 Chinese themselves distinguish between full words (shi tse) and 
 empty words (hiu tse). These empty words, although they have 
 an accent, have no independent meaning of their own, but determine 
 and modify the meaning of other words. The same applies to 
 Chinese compounds. Here also, two words form but one logical 
 idea. C6 "gin is not tongue + man, but man of the tongue, i. e. an 
 interpreter. "Gi tse is not sun + son, but the son of the sun, i. e. day. 
 Discarding compound and polysyllabic words, which the Siamese 
 have borrowed from the vulgar Sanskrit, the Pali, or the sacred 
 language of the Buddhists, we find even Siamese words joined toge- 
 ther to express one idea. For instance; nam chai, water (of the) 
 heart, i. e. will; kan suk, opus belli, i.e. beUatio. Abstract words 
 
 * The accents are thus represented by Pallegoix in musical notation : 
 Kectus Circiunflexus Demissus Gravis Altus 
 
 Kong Kong Kong K9ng Kong 
 
 . Leydeu compares the modulation of these accents with the chanting of the 
 Samaveda in India. 
 
 K 4: 
 
132 
 
 are formed by prefixes expressive of " heart," " matter," etc. ; and in 
 a similar manner substitutes are found to express approximately 
 number, gender, and case. However, Siamese, no doubt, may be 
 called monosyllabic in the only sense in which any human language 
 can be so, in so far as it shows very few traces of compounds in 
 which one part has entirely lost its original form and meaning. In 
 this sense the Tai languages are monosyllabic, and the Lohitic 
 are not. 
 
 There are other fundamental principles by which the gramma- 
 tical system of the Tai differs from the Lohitic dialects. We saw 
 that in the Lohitic dialects all words expressive of case, number, 
 and gender were put at the end of words. It is just the contrary in 
 the Tai languages, so far as known to us. Gender only may form 
 an exception, because it may be expressed by an adjective, and 
 the adjective in the Tai, as well as in the Lohitic dialects, follows 
 the substantive. Thus ma in Khamti is horse ; ma-thuk, a stallion, 
 ma-me, a mare ; miau, a cat ; miau-thuk, and miau-me, a male and 
 female cat. In Kassia, however, gender is expressed by prepositive 
 particles ; for instance, u- tang a, husband, ka-tanga, wife ; u-kapa, 
 father, ka-kami, mother. — Number, where it is expressed at all, is 
 expressed by prepositive words. In Kassia the plural is expressed 
 by the preposition ki; for instance, sing, u-mon ; plural, kimon. 
 In Siamese the plural is expressed by a prefix, meaning many. 
 Bishop Pallegoix, in his Thai grammar, gives a complete paradigm of 
 a declension in Siamese. In it all cases, with the exception of the 
 vocative, are expressed by prepositions. It is the same in Robinson's 
 Kassia and Khamti grammars, and we find there that all other local, 
 or temporal, or causal relations, which in the Lohitic dialects are 
 invariably expressed by postpositions, are here rendered, by a large 
 array of prepositions. 
 
 A grammatical feature like this marks the family-likeness of the 
 Tai languages better than even the striking similarity of their 
 numerals ; and it establishes the more distant degree of relation- 
 ship between the Tai and and Lohitic dialects, indicated, though 
 less distinctly, by the variations to be observed with regard to the 
 numerals in each class. Numerals as well, if not more even than 
 other words, are exposed to phonetic accident and fluctuation which. 
 
133 
 
 in the absence of historical documents in these Eastern dialects, will 
 hardly ever be reduced to the same rules as the phonetic changes of 
 the Arian languages. We have only to compare lists of words col- 
 lected among identically the same tribes by different persons, in order 
 to convince ourselves what vague and unmanageable materials we 
 have to deal with. In Siamese, r and 1 are said to be pronounced 
 like n in the close of a syllable ; ma and ba, ty a and chga, are often 
 of difficult distinction when pronounced, as are ya and ja, kye and 
 chye. We saw before how mangled an appearance Sanskrit words 
 have if adopted in Burmese. The same applies to Sanskrit words in 
 Siamese. We should hardly recognise in the adventures of Pram 
 and his brother Pra-lak, and in their wars with Totsa-kan, who 
 carried offNang Seda, the stories of Rama, Lakshmawa, Da«a- 
 kantha and Sita. Nor would the Buddhists of India easily dis- 
 cover their Buddha Siddartha, and Suddhodana in theBugda, 
 Theik-dhat, and Sugdo of the Siamese. It is on account of the 
 phonetic vagueness of these monosyllables, and also on account of 
 the strange corruptions to which words taken down auricularly 
 are exposed, that I abstain from giving long comparative lists. 
 There can be no doubt that many identifications of Gangetic, 
 Lohitic, and Ta'i words, given by Buchanan, by Leyden, and more 
 recently and completely by Mr. Hodgson, are true. But still they are 
 only persuasive, not convincing. Is it possible, in the present state 
 of our knowledge, to discover the foreign words adopted in one or all 
 of these dialects, if we consider the great changes to which, as we 
 have seen, these foreign words are liable? And still more . diflicult 
 it is to say, in each case, whether a Siamese word which we compare 
 with a Burmese may not have been taken simply as a foreign word 
 by one of the two languages, instead of belonging to that common 
 Turanian stock of words from which all these dialects originally 
 descended. 
 
 It will be an interesting problem, to be solved hereafter, how far 
 the Chinese contains, in its most ancient and best authenticated 
 form, the radical elements from which the Bhotiya, as well as the 
 Ta'i languages, branched off at different periods. Unfortunately, 
 Chinese civilisation has so powerfully reacted on these languages, 
 in periods within reach of history, that it will always be extremely 
 
134 
 
 difficult to ascend on safe ground to ante-historical times, in which 
 alone a radical community between these idioms and the Chinese 
 could have existed. The numerals hold out a strong hope that 
 the problem will be solved in the affirmative ; the pronouns also 
 contain similar indications. But principles will first have to be 
 established by which we can tell foreign and adopted from natural 
 and common words in these dialects. We can tell in French whether 
 a word was taken from Italian, or whether both French and Italian 
 derived it from Latin, The same will have to be done for the 
 Ta'i, LohitiCj and Gangetic dialects, in their relation to Chinese. 
 The use of prepositions in the Ta'i declension, and jo£ postpositions 
 in the Lohitic declension, are unmistakeable signs of different stages 
 of grammatical growth respectively attained. They are features as 
 distinct as the use of prepositions and articles in the modern Romanic 
 languages contrasted with the final terminations in Sanskrit or Greek. 
 Grammatical features of this kind must serve as landmarks in the 
 linguistic survey of these countries, and as eras in the historical ar- 
 rangement of their growth and diffusion. 
 
 I subjoin a few instances of similar words in the Chinese, Burmese, 
 and Tibetan, in order to show that they deserve attention, though I 
 quite agree with what Schleiermacher says of them in his Grammaire 
 Barmane, " de telles comparaisons de mots monosyllabiques isoles 
 presentent toujours beaucoup de vague." The question is whether, 
 according to the nature of the case, we have a right to expect more 
 definite proofs. 
 
 Chinese, neng 
 Burmese, nhain 
 
 
 posse. 
 
 Chinese, chi 
 Burmese, si 
 
 
 scire. 
 
 Chinese, cha; Canton, 
 Burmese, sat 
 
 chat "1 
 
 occidere. 
 
 Chinese, eh& 
 Burmese, soey 
 
 
 removere. 
 
 Chinese, nieou 
 Burmese, nuah 
 
 
 bos. 
 
 Chinese, ko 
 Burmese, khu 
 
 
 affixum numeraJe, aliquis. 
 
 Chinese, tchi, ti 
 Burmese, si 
 
 
 particula possessiva ; genitive. 
 
133 
 
 Many more words might be added, particularly if we compare the 
 Chinese spoken nearest to India, — I mean the Chinese of Canton, — 
 with Siamese, Burmese, and the spoken Tibetan. But who could say, 
 in each individual case, whether the following words, for instance, 
 are ancient common words in Burmese, Siamese, and Chinese, or 
 whether the one language borrowed them from the other ? 
 
 I. 
 
 II. 
 
 III. 
 
 IV. 
 
 V. 
 
 Chinese. 
 
 Burmese. 
 
 Siamese. Tibetan, spoken. 
 
 English. 
 
 ngW, 
 
 ngU, 
 
 — 
 
 gna, 
 
 fish. 
 
 too, 
 
 — 
 
 m. 
 
 — 
 
 fire. 
 
 syuk, 
 
 kyiowk, 
 
 — 
 
 — 
 
 stone. 
 
 ogann, 
 
 ngwe, 
 
 ngun, 
 
 ngui, 
 
 silver. 
 
 thong. 
 
 — 
 
 thong kham. 
 
 jhang chep, 
 
 brass. 
 
 ya (evening), 
 
 nya. 
 
 — 
 
 — 
 
 night. 
 
 chew, 
 
 — 
 
 chaw. 
 
 — 
 
 morning. 
 
 yin, 
 
 — 
 
 khon. 
 
 — 
 
 man. 
 
 fu (hu), 
 
 pha. 
 
 pho, 
 
 pha. 
 
 father. 
 
 mu, 
 
 ma, 
 
 me, 
 
 ama, 
 
 mother. 
 
 akko, 
 
 ako, 
 
 — 
 
 — 
 
 elder brother. 
 
 amui. 
 
 umma, 
 
 
 
 
 
 elder sister. 
 
 thow. 
 
 — 
 
 hua. 
 
 go, 
 
 head. 
 
 ngi, 
 
 na, 
 
 — 
 
 na (Changlo), ear. 
 
 li, 
 
 — 
 
 Un, 
 
 li (Chango), 
 
 tongue. 
 
 khuok, 
 
 khye. 
 
 — 
 
 kang pa, 
 
 foot. 
 
 thii, 
 
 — 
 
 (phung) thong. 
 
 phou, 
 
 belly. 
 
 kie, 
 
 
 
 kin, 
 
 ja, 
 
 to eat. 
 
 teko, 
 
 di hu, 
 
 — 
 
 di, 
 
 this. 
 
 koko, 
 
 ho hu, 
 
 — 
 
 kho (he), 
 
 that. 
 
 For the most striking coincidences I should always point to the 
 numerals. Their similarity cannot be denied, and their common 
 origin can hardly be doubted. To suppose that the Tai dialects 
 borrowed their numerals from the Chinese, the Gangetic from the 
 Tai, and the Lohitic from the Gangetic dialects, would be in the 
 teeth of all phonological analogy, and I doubt not, that in Turanian 
 as in Arian philology, the numerals will turn out after all to be the 
 first indisputable ground for establishing the claims to a common 
 descent for the languages of India extra Gangem. 
 
136 
 
 Bhotiya and Tai Numerals. 
 
 There are two things which we must bear in mind in looking 
 at the compartive table of the Tai and Bhotiya numerals, including 
 under the latter term the Gangetic and Lohitic dialects. 
 
 It is true that, with the exeption of the Tai idioms, which agree 
 better in this respect than any two dialects of Greece or Italy, they 
 do not offer at first sight very striking coincidences. But we must re- 
 member, first, by whom these numerals were collected ; secondly, by 
 whom they are used. They were collected by men, not always 
 familiar with the sounds which they tried to write down and 
 translate into the Roman alphabet. It is very difficult to catch the 
 mere sound of a language, if we do not understand its meaning ; and 
 if two travellers in foreign countries endeavour to write down the 
 words uttered by the same individual, their phonetic sketches are 
 likely to vary as ' much as two portraits taken under different effects 
 of light. 
 
 In two lists of the Kuki numerals, we find katka=keaka; nika= 
 panika ; like = taj rungaka=nga ; ruka=koo ; and in all these 
 cases, I believe, the same sound was meant to be represented. Ka 
 is one of those generic numeral affixes which attaches itself to all 
 the Kuki numerals, but may be suppressed if other words follow. 
 The word for "one," I suppose, therefore, to be kat ; but t before 
 k becomes obscure, — as, for instance, in the Burmese name of a 
 magistrate, which Judson writes Sit-kai; Cox, chikoy; Symes, 
 chekey ; and it is the same in octo and otto. Hence katka=keaka. 
 In panika=nika, pa can be proved to be an usual numeral prefix. 
 In lika=ta, we have one of the frequent instances where the strong 
 aspirated 1 has been mistaken for a t, while in other cases the 
 same sound is represented by an rh, or d, or zh. In Burmese, for 
 instance, thyo and r h o, pronounced she, are all the same word for to 
 wash. In runga=nga, the r u is simply the guttural arsis which natu- 
 rally precedes the deep nga, but was not meant to be sounded sepa- 
 rately. The same applies to r u k a vice k o o. The greatest difficulty 
 consists in catching the sound of final letters. In Burmese, as Schleier- 
 macher (§ 31.) says, " on est souvent hors d'etat de distinguersi c'est 
 un p ou un t qu'on entend prononcer ;" and the same grammarian 
 
137 
 
 writes (§ 26.) with regard to the final k, that a k is sounded like 
 a k or at. We need not be surprised, therefore, if the Limbuword 
 for one is written t e e k by Campbell, and t h i t by Mr. Hodgson. 
 On the other hand^ we must also take into account the status of the 
 people who speak these languages. Some are brought in contact 
 with civilisation, whether English, Chinese, Hindu, or Burmese, and 
 in their commercial transactions have to use foreign terms. Thus it 
 happens that tribes who possess numerals of their own, prefer to 
 show their knowledge of the Hindustani numerals, when they are 
 questioned on the subject by an English official. 
 
 Among themselves, also, the most powerful and governing race 
 probably exercises a certain influence on the languages of subject 
 tribes, and in their bartering transactions Burmese numerals, for 
 instance, would form a common medium between distant hordes sub- 
 ject to the Burmese government. Again, many of these tribes 
 migrate, some are conquered and carried off into slavery. The 
 language of the conquerors has to be learnt by the conquered, who 
 again in turn conquer their conquerors and retaliate on them. 
 
 Lastly, the numerals in all these languages are known to have 
 this peculiarity, that they change according to the object numbered. 
 In Burmese n h i t is two ; but 
 
 two men, is lu (man) nhit-yauk, 
 two fowls, is kyet (fowl) nhit-gaung, 
 two pagodas, is tsadi (pagoda) nhit-chu. 
 
 In Kachari these determinative syllables are prefixed. When nu- 
 merals are applied to human beings, the particle s a is prefixed to 
 the numerals ; when applied to other animals, m a ; to inanimate 
 objects, thai'; to trees, p h a n g ; to articles enumerated by pieces, 
 gang. For instance, m a n s e sa nai, twomenjburma mabre, 
 four goats; phitai thai ro, six fruits. 
 
 In Mikir, again (where the simple numerals only go as far as six, 
 thorchi, 7, being thorok 6+ichi 1 ; merkep, 8 = 10—2; chirkep 9 
 = 10—1), the word bang is prefixed when individuals are enume- 
 rated ; jon when inferior animals ; hong and pap when inanimate 
 obj ects. For instance, atebanghini, two brothers ; jon phongo 
 achorong, five cows, i.e. piece-five cow. The same applies to 
 
138 
 
 Khamti and Siamese, and to the Malay, where these generic particles 
 are put at the end of the numerals. In Malay, for instance, ekor 
 means tail, and is used as a generic exponent for cattle. Sa=one, 
 kerra=monkey ; one monkey=sa §kor kerra. Lima=5, kuda, 
 horse; five horses=kuda lima ekor. Now, some of these suffixes 
 and prefixes, if used frequently, and particularly if deprived of their 
 original meaning, coalesce with the numerals. In this manner we 
 must account for long being added to all Dhimal numerals, for 
 s h and s h i at the end of the Limbu numerals, for z h o in Chepang. 
 In Miri, k o is affixed and a prefixed. The Shendus prefix m e, 
 the Gyarung k a ; the Manyak affix b i, the Gyami k u. In other 
 cases we find some numerals with, others without suflixes ; sometimes 
 we are told that this suffix must be dropped if a substantive follows, 
 sometimes that another suffix must take its place. All these changes 
 are based on one and the same principle of determinative syllables, 
 which in monosyllabic languages are for the speaker what deter- 
 minatives are for the writer in Egyptian and partly in Chinese.* 
 
 After these preliminary remarks, we may venture to trace some of 
 these numerals to their original form and, common type. We shall 
 meet with extreme cases, such as, for instance, Naga vanram, three, 
 being the same as Miri aomko, three. But we know that v a or 
 van in Naga is a prefix, as well as a in Miri; and we also know 
 that k o in Miri is an affix. This leaves us ram=om. Now, the 
 r in vanram or varam stands between two vowels, where, as, 
 for instance, in the name of the Burmese capital. Am aapuya, i.e. 
 Amarapura, it is dropped in pronunciation. Hence ram and om, 
 both preceded by a vowel, are the same, and so are their secondary 
 forms, van ram and aomko. 
 
 We may at once proceed to a consideration of the other words for 
 three, and take " two " afterwards ; because " three " is common 
 to the whole class, while " two " will help us to distinguish the Tai 
 from the Bhotiya class. The original type of " three " in these lan- 
 guages was "Sam." Sum, som, san, sun, son, sang, sung, song, are 
 simply varieties of the same sound. This brings together Chinese, all 
 the Ta'i (exc. Kassia), eleven Gangetic dialects, and one Lohitic. 
 Making allowance for the evanescent final nasal (as in Uraon and 
 • See Humboldt's Complete "Works, vi. p. 402. 
 
139 
 
 Urao), we may include Horpa and Angdmi Naga ; and considering the 
 difficulty of distinguishing between the four names of the frontier river 
 between the English provinces and Burmah, (it is written San-luen, 
 Salwen, Saluoen and Thalueyn,) we may also include the words with 
 initial th (hts, shy). Thus we embrace the Shan, Burmese, Bodo, 
 Khyeng, Mru, Sak, Tunglhu, and with t=th, the Kumi turn. In 
 composition we have still to make allowance for s being changed 
 into r, r into 1, or its being dropped altogether. The phonetic pro- 
 cess of these changes is well known from Sanskrit and other dialects. 
 We now take the prefixes : 
 
 With ka, we have, Gyar ung, k a-s a m. Mikir, k a-t h a m. 
 Kami, k a-t u n. 
 
 gi, Garo, gi-tham. 
 
 a, Dophla, a am. Abor, angom (i.e. a-ong). 
 Naga, azam and asam. 
 
 ma, Singpho, m a s u m. Shendu, m e - 1 h a s. 
 
 van, Naga, van ram. 
 
 With suffixes we have — 
 
 ku: Gyami, san ku. Kuki, tumka. 
 
 ri : Thochu, kshi ri. 
 
 bi : Mangak, si bi. 
 
 shi : Limbu, syum sh. 
 
 ya: Kiranti, sum ya. 
 
 zho : Chepang, sum zho. 
 
 lang: Dhimal, sum lang. 
 
 With suffixes and prefixes — 
 
 a-ko: Miri, a|um|ko. aomko. 
 a-a : S. Miri, a|um|a. 
 
 Thus, out of fifty-four dialects, there is but one, the Kassia lai, 
 which resists classification, though here the Naga lem might serve 
 as a link. Even if the value of these comparisons could be tested 
 simply by phonetic similarity, if the common origin of these nu- 
 merals was simply a question of phonetic possibility, I should think 
 even then the chain of changes which connects the Chinese word for 
 three with all the rest less complicated by far than that by which 
 Professor Bopp has tried to connect the Caucasian semi, sumi, 
 sami, j urn, with Sanskrit tray as. 
 
140 
 
 I do not attempt a comparison of the words for one, for reasons 
 stated above, though the number of independent radicals to express 
 one, is not so great as might be expected. The same applies to the 
 words for two. Here also the power of forming new words, such 
 as couple, pair, brace, has not been carried to a great extent; and 
 in its limited operation it becomes useful and instructive, because 
 it assists us in establishing lines of demarcation between the Tai, and 
 the other dialects. The type for two was NY A, with a decided 
 tendency, however, towards the vowel i. The softened or mouille 
 sound of ny, is sometimes expressed by gn, sometimes by ng; nh 
 also seems intended to indicate the same sound. In other dialects 
 this mouille letter becomes evanescent, and the simple nasal only re- 
 mains. It may seem doubtful whether instead of NYA, the original 
 form might not have been NT AT. Several forms occur with a final 
 t. But as there are other forms with a final s, it seems more plau- 
 sible to look on both t and s as additional letters, though their 
 occurrence in distant dialects may be taken as a sign that this ad- 
 ditional consonant dates from an early period. After these remarks, 
 seventeen dialects can be traced back to ny a or nyat, without affix or 
 prefix. The afiixes and prefixes used in the other dialects are the 
 same as for " three." The only new ones are kching in Changlo, 
 and pan in Khyeng. Anomalous forms are Kiranti, has at, Naga, ih, 
 Miri, pre. It is important as a distinction that the Tai dialects 
 have their own word for " two," which is the same as " three," only 
 with a final guttural nasal, sang, instead of the labial nasal in sam 
 (three). We may conclude from this that the separation of the Tai dia- 
 lects from the common stock took place previous to the separation of 
 Tibetan and Burmese, A third base for "two" is ar in Gyami and 
 Kassia; it may be connected with the Chinese eul. 
 
 The original base of "four" was " Chi," a sound which in these 
 dialects is sometimes palatal, sometimes lingual, sometimes sub- 
 dental, accordingly as it is produced by bringing the tongue in contact 
 with the palate, the root of the teeth, the gum, or the teeth, without, 
 however, allowing it to become a pure dental or a pure guttural sound. 
 From chi this sound may pass to zhi, and from zliiit may run into 
 a soft s. It may also approach a lingual d, and then merge into the 
 lingual r and I. Similar changes have been pointed out before. Al- 
 though, therefore, the written words for " four " vary considerably in 
 
141 
 
 appearance, yet it is frequently the same sound which was meant to be 
 represented, only that it is a vague sound, and a sound for which the 
 European alphabets have no distinct sign. Chinese and Tai have a 
 decided s, followed by i, with the exception of the Kassia, which has 
 sau instead of si. 
 
 Zh occurs five times, or eight times if we include the soft initial d, 
 the Sanskrit d (^). The hi, which is the Sanskrit dh (^), and which 
 is also written Ih or /, occurs eleven times, including the cases where 
 it is joined by common prefixes. But there is one prefix which 
 occurs so frequently that it seems to be more than a prefix. It may 
 have been by itself a word for " four," which, as is usual in mono- 
 syllabic languages, was joined to the other word for "four," in order 
 to make the intended meaning more apparent.* This word is p i, and 
 it occurs by itself as the word for four in Newar, pi, Changlo, phi, 
 Miri, a-pi-ko, and Abor, a-pi. Together with Ihi or rhi, Ave find 
 it in Takpa, p|li, Garung, p|li, Magar, bu |li, Murmi, b|li, Lepcha, 
 pha|li, Chepang, pjloi|zho, Mikir, phili, Dophla, ajp|li, Mga, 
 pha|le, phajli, pi|li, Kumi, palu, Shendu, pu|lli, and in Bodo, 
 b|re, G-aro, b|ri, Naga, pa|zr, and Sak, p|ri. What raises a doubt, 
 however, as to the origin and meaning of the initial labial sound in 
 these words, is that the same labial prefix occurs sometimes before 
 " five" in the same dialects which add it to " four," though they do not 
 use it before any other numerals. In two cases "five" is expressed 
 by the single labial, b a, m e-p a. In Tibetan, as pointed out by 
 Mr. Hodgson (1853, p. 59), zhi (four) is written bzhi. 
 
 "Five" in Tai is HA, except the Kassia san. In the other 
 Bhotiya dialects the original base of " five " was GN A. This may 
 be contained in the Chinese ung, for GNA can also be represented 
 by NGA. In the Bhotiya class GNA is so little disguised, that no 
 explanation is required. We might naturally expect na and ga, in- 
 stead of n g a ; but we also find more violent changes, such as g w a 
 and wa; and still more anomalous, ma in pu-ma. Among the 
 prefixes the constant labial has been already mentioned. This is 
 sometimes followed by 1, so as to render the origin of forms such as 
 p i-1 i-n g - k o rather problematical. 
 
 * An analogous feature of the Malay languages is pointed out in Mr. Crawfurd's 
 Malay Grammar, page 81. 
 
142 
 
 Six" in the Tai languages is RUK, HUK, and in Chinese LU 
 and LOK. Now it will easily be seen that the various forms of 
 one and the same numeral in the Tali languages are nothing but 
 phonetic varieties of the same word. Nor is it difficult to account 
 for the transition of r into 1, or for the omission or addition of a 
 final k. (See page 396.) With all these allowances, however, we do 
 not obtain a base which will explain all the corruptions of the Bho- 
 tiya words for " six." A base which would account for most, would 
 be RHU, only that we should have to admit two prefixes, t and k. 
 The simple RHU would account for forms like ru and ro, and also 
 for dhu and thu, if these two are meant as lingual aspirates. The 
 prefix t would then explain forms like ta ru, tau, tarok, thorok, 
 and soru ; and the prefix k would account for the rest, such 
 as kro, krukzho, khyauk (ray in Burmese). Still even thus 
 a residuum remains, which it is better not to attempt to analyse 
 until we receive more accurate lists than those which we have at 
 present, where, as for instance in Garo, "six'' is given in one list as 
 krok, in another as dok. 
 
 For tht same i^eason I abstain at present from tracing the re- 
 maining numerals from seven to ten back to their original types 
 Their general likeness leaves no doubt that they also proceeded 
 from one common source. We find, in spite of occasional deviations, 
 a sufficient number of almost identical words for seven, eight, nine, 
 and ten in the most distant members of the Bhotiya family to be sa- 
 tisfied as to their common origin. We must make allowance, however, 
 for this, that some languages express seven by 6+1, as for instance 
 thorchi, which stands for thorok-|-chi. Again, eight is ex- 
 pressed by a compound 10 — 2, and nine by 10 — 1 ; for instance, in 
 Mikir, where 10 is kip, 9 chirk ep, 8 nirkep. Yet, with all these 
 exceptions, anomalies, and corruptions, this one important fact re- 
 mains established, that the Bhotiya and Tai" members of the 
 Turanian family show in their numerals their former unity and con- 
 tinuity as distinctly as the languages of Arian origin. 
 
143 
 
 Seventh Section. 
 The Malay Languages. 
 
 In the first part of this Letter I endeavoured to show that, where 
 the means are wanting to enable us to trace the genealogical connec- 
 tion of large groups of languages, it is yet possible to classify them on 
 grounds merely morphological. In a nomadic state of language words 
 are liable to such rapid changes, and those arising not from phonetical 
 corruption, but from actual loss and a continued reproduction of 
 words, that, after some generations, one language may be split into two 
 dialects, in which the most common objects are expressed by different 
 terms. Nomadic languages shed their words almost in every century ; 
 while political languages keep their plumage for thousands of years. 
 It would be hopeless, therefore, to attempt to test the relationship of 
 nomadic dialects by the same agencies that briqg out the affinities of 
 political languages ; nor would it be right to deny their proper weight 
 to coincidences in the leading principles of grammatical formation, 
 which, like a natural instinct, may live on where all external signs 
 of relationship are obliterated. "A language," as Humboldt says, 
 " cannot be looked upon as a mere aggregate of words. Every 
 language is a system by which the mind embodies an idea in audible 
 expression. It is the business of the philologist to discover the key to 
 this system. It will then appear that races not only express their 
 ideas in the same manner, but follow the same path in their forms 
 of speech." 
 
 § 1. Formal Coincidences between the Malay and Tdi Lfinguages, 
 
 It is from this point of view that the Tai and Malay languages may 
 be ranged together, as coinciding most strikingly in some of the most 
 characteristic features of their grammar. It is not the geographical 
 proximity of the Malays and Siamese races which suggests this idea. 
 The settlement of the Malays on the continent of Malacca is generally 
 considered as of modern date. Nor is it a mere community of words 
 which led to this supposition. Here again, unless our comparison 
 
 L. 2 
 
144 
 
 extended over the whole dictionary of these two races, it would be 
 impossible to say whether the Siamese had borrowed from the Malays, 
 or the Malays from the Siamese. Nor do I wish to prove that Siamese 
 and Malay are lineal descendants of the same parent. But there exists 
 in their grammar an instinct so peculiar and constant, that it is in vain 
 to try to account for it without the admission, that before the dispersion 
 of the descendants of Tur, the nomads of the Pacific received their 
 first grammatical impressions together with the rest of the Turanian 
 family ; that after their first separation they continued for a long time 
 together with that branch of the southern Turanian division which 
 occupies the valley of the' Brahmaputra and extends to the peninsula 
 of Malacca ; while all that seems to be Arian in their grammar and 
 dictionary, and has been used by Bopp to prove the original connection 
 of the Malay and Arian languages, was simply imported during a later 
 political and religious intercourse between the Arian colonists of India 
 and the Turanian inhabitants of the Indian Archipelago. 
 
 The following are some of the leading features which the Malay 
 share in common with the Tai languages. For the Malay I refer to 
 Ml". Crawfurd's Grammar ; for the Tai, to the Grammatical Outlines 
 of Mr. Eobinson, chiefly taken from the Kliamti. " This," to quote 
 Mr. Robinson's words, " in common with the Siamese, Laos, Shyan, 
 and Ahom, is only a dialect of the language usually known as the Tai; 
 a language more or less prevalent through all that wide tract of 
 country extending from Siam to the valley of the Brahmaputra. In 
 a language so extensive in its use, it might be conjectured that local 
 peculiarities would have given rise to a great diversity of dialects, 
 so that the Khamti and Siamese, spoken at the extremities, would have 
 presented but few links of connection. On the contrary, however, 
 we find that the discrepancies between the two are very trifling." 
 This shows that the Tai must once have passed through a period 
 of literary cultivation and grammatical concentration, and that the 
 Tai dialects spoken at the present day are but varieties of one com- 
 mon type. 
 
 Mr. Brown's investigations led hira to the conclusion, that upwards 
 of nine tenths of the fundamental words are the same in Siamese 
 and Khamti, with the exception of a few slight variations of pronun- 
 ciation. These variations are mostly confined to a few letters : viz. 
 
145 
 
 ch, which the northern tribes change to ts; d, for which they use 
 1 or n ; r, which becomes h ; and u a, which is changed for long o. 
 
 " Different systems of writing," as Mr. Robinson remarks, " have 
 been introduced to express the sounds of the different dialects of the 
 Ta'i. The Khamti and Shyan alphabets are derived from the Bur- 
 mese ; the Laos is nearly related to Burmese, but more complete and 
 better adapted to the wants of the language than the Shyan ; whilst 
 the Siamese character bears only a remote resemblance to Burmese." 
 
 1. In Khamti, inflections are unknown, and the accidents of case, 
 mood, and tense are expressed by means of particles. 
 
 In Malay, there are no inflections to express gender, number, 
 person, time, or mood. 
 
 2. Words or particles which serve as the exponents of these gram- 
 matical relations, and which in the Ural-Altaic languages are always 
 placed after the root, may, both in Malay and Ta'i, be used as pre- 
 positions as well as postpositions. 
 
 3. The relation of the genitive may be expressed by mere juxta- 
 position. But while in Chinese the first word is understood to be in 
 the genitive, the governed word, or what we call the genitive, stands 
 last both in Malay and Tai. 
 
 Examples: Tai: Hang, a tail; pa, a fish, 
 hang pa, a fish's tail. 
 
 Malay: Tuwan, master ', amba, slave. 
 
 Tuwan amba, the master of the slave ; 
 amba tuwan, the slave of the master. 
 
 4. The accusative takes no preposition in Khamti and Malay; in tlie 
 latter, pada (to) may be added. 
 
 The accusative follows the verb in Siamese and Malay. 
 
 5. The other cases, if cases they can be called, are formed in Ta'i 
 and Malay by prepositions. 
 
 a. Dative in Tai: Hang. 
 
 Ex. Hang man hau da, Give to him. 
 
 a. Dative in Malay: ka, kapada. 
 
 Ex. Maka kata raja kapada estrina. The king said to his 
 spouse (stri). 
 
 h 3 
 
146 
 
 b. Ablative in Tai: luk, from. 
 
 in Malay: dari, ka-luwar, from. 
 
 c. Locative in Tai : ti, in. 
 
 in Malay : di, in. 
 Ex. Di nagri Tringanu, The country of Tringganu. 
 
 6. The Khamti noun admits of no plural. The singular, where 
 necessary, may be expressed by the addition of the numeral o n e in 
 Siamese and Malay. Plurality in Malay is expressed by an adjective 
 having this sense: as, banak, many; sagala and sakalian, all; and 
 the numerals. In Khamti, in those instances where the noun does not 
 express a collective or plural idea, a numeral added to it renders the 
 expression sufficiently intelligible. 
 
 7. Gender is not expressed ; but to indicate the difference of sex in 
 the inferior animals, the term thuk is used to denote the male, and 
 me the female in Khamti. 
 
 Ex. A horse : Masc. ma thuk ; Fem. ma me. 
 A cat : Masc. miau thuk ; Fem. miau me. 
 
 In Malay, a male of the lower animals is expressed by the 
 adjective jantan ; a female, by batina. 
 
 Ex. A horse : Masc. kuda jantan ; Fem. kuda batina. 
 A cat : Masc. kuching jantan ; Fem. kuching batina. 
 
 For individuals of the human family different words are used in 
 Tai and Malay : sau and ying, in the former ; laki and param- 
 piian or estri (Sanskrit), in the latter. 
 
 Ex. Tai : Luk sau, son ; luk ying, daughter. 
 
 Malay : Anak laki, son ; anak parampiian, daughter. 
 
 Tai : Pi sau, brother ; pi ying, sister. 
 
 Malay : Saudara laki, brother ; saudara parampiian, sister. 
 Exceptions where distinct words are used, are, 
 
 Malay : Pa, father ; ma, mother. 
 Tai : Po, father ; me, mother. 
 
 8. In Tai, the adjective follows the substantive. 
 Ex. Kun ni, a good man. 
 
 Ma m"a-ni, a bad (not-good) dog. 
 
147 
 
 In Malay, an adjective, by its form, is not distinguishable from the 
 noun ; for the same word is often either, according to position. It js 
 its place, following the noun, which marks the word as expressing 
 quality. 
 
 Ex. Putish kayin, the whiteness of cloth. 
 Kayin putish, white cloth. 
 
 9. In forming the comparative degree of the adjective in Khamti, 
 the word leu, beyond, is added to it in its positive form. 
 
 In Malay, the comparative is expressed by daripada; dari mean- 
 ing from, p^da, with. 
 
 ±yx. Jihamti: Yau leu hiin, great beyond the house; or, Noi sung 
 
 3 4 1 2 3 4 
 
 leu peun, a mountain high beyond all ; i. e. the highest mountain. 
 
 12 3 12 3 
 
 Malay: Bayik daripada samuwana, good from with all; or, Labih 
 bayik daripada samuvana, very good from with all, i. e. best. 
 
 10. In the numerals, no coincidences can be pointed out between 
 Malay and Tai ; and the great similarity of the Tai numerals with 
 Chinese, makes it not improbable they were actually adopted from 
 Chinese. But the Malay numerals participate in a feature pecu- 
 liarly Turanian, that of forming the words for eight and nine, by 
 10 — 2 and 10 — 1. Dulapan, eight, contains du, 2; sambilan, 
 nine, contains sa, 1; as determinative elements. If sal ap an is used 
 for eight, this can only be explained as a mistake ; and in the Sunda 
 dialect of Java, salapan has retained its original meaning of nine, 
 d a 1 a p a n, of eight. Different etymologies have been giv«n of 1 a p a n, 
 and bilan, which may be seen in Humboldt, Bopp, and Buchanan ; 
 but they do not affect our argument as to the Turanian character of 
 these formations. 
 
 11. Another feature deserves to be pointed out with regard to the 
 numerals, connecting the Tai languages most closely with the Malay. 
 Mr. Crawfurd describes it in the following manner : 
 
 In the enumeration of certain objects, the Malay has a peculiar 
 idiom, which, as far as I know, does not exist in any other language 
 of the Archipelago. It is of the same nature as the word " head," as 
 we use it in the tale of cattle; or "sail," in the enumeration of ships ; 
 but in Malay, it extends to many familiar objects. 
 
148 
 
 Alai, of which the original meaning has not been ascertained, 13 
 applied to such tenuous objects, as leaves, grasses, hairs, and feathers. 
 
 Batang, meaning stem or trunk, to trees, logs, spars, spears, and 
 javelins. 
 
 Bantak, of which the meaning has not been ascertained, to such 
 objects as rings. 
 
 Bidang, which means spreading or spacious, to mats, carpets, 
 thatch, sails, skins, and hides. 
 
 B i j i, seed, to corn, seeds, stones, pebbles, gems, eggs, the eyes of 
 animals, lamps, and candlesticks. 
 
 Bilah, which means a pale or stake, to cutting instruments, as 
 knives, daggers, and swords. 
 
 Butir, a grain, to pepper, beads, cushions, pillows, and, strangely 
 enough, to brooks and rivers. 
 
 Buwah, fruit, to fruit, loaves, cakes, mountains, countries, lakes, 
 boats and ships, houses, palaces and temples. 
 
 Ekor, tail, to beasts, birds, fishes, and reptiles. 
 
 Kayu, which means wood, to any object roUed up, as a piece of 
 cloth. 
 
 Keping, a sheet, to any foliacious object, as a sheet of paper. 
 
 Orang, man or person, to human beings 
 
 Puchuk, which means literally top, to cannon and small-arms, to 
 candles and torches, and to letters or missives. 
 
 Rawan, which is literally gristle or cartilage, to all descriptions 
 of cordage. 
 
 Ex. Ada saorang saudagar kapada sabuwah nagri. There was one- 
 man merchant, in one fruit (of a) country. 
 
 Bad'il limapuluh puchuk, dan pad-ang lima ratus bilah. Fire-arms, 
 fifty pieces, and sword five-hundred stake. 
 
 Let us now compare the Khamti. Here, according to Mr. Ro- 
 binson, numeral affixes, or as they have sometimes been called, ge- 
 neric particles, are in common use. These particles are affixed to 
 numeral adjectives, and serve to point out the genus to which the 
 preceding substantive belongs. 
 
 T o is the numeral affix applied to animals. When the number to 
 be expressed is one, the generic particle precedes the numeral ; in 
 every other case it follows : — 
 
149 
 
 -,21s 3 4 12346 
 
 -fca;. Pe nan luk on yang song-to. That goat has two kids. 
 Bai is the numeral affix applied to such nouns as leaf, paper, 
 umbrella. 
 
 Nue is applied to things round. 
 
 Thep, and phen, to flat substances. 
 
 Phiin, to pieces of cloth. 
 
 Sen, to things having length. 
 
 Ho, to bundles, packets, and the like. 
 
 Sem, to sticks, posts, spears. 
 
 Khot, to ropes, and such articles as can be coiled up. 
 
 Ban, to villages, hamlets, and towns. 
 
 The use of these numeral affixes is evidently based on a pe- 
 culiarity of conception, remarkable as any in the grammar of 
 nations. The nations who employ these generic exponents, were in- 
 capable of conceiving quantity in the abstract ; a defect in their 
 logical powers more suggestive to the ethnologist than any peculiarity 
 in the anatomical structure of their skull. We find the same ge- 
 neric particles in Burmese, where, as in Malay and Tai, they are 
 placed after the numerals; while, in Kachari and Mikir, they are 
 placed before. In its most developed state, we find the same 
 custom in Chinese. There also the numerical exponent stands after 
 the numeral and before the substantive, except in accounts, when, as 
 in Burmese, the noun is put first, then the numeral, and last the 
 generic term. These generic terms were collected by P. Basilius in 
 his Dictionnaire (p. 933), and by Morrison in his Chinese Gram- 
 mar (pp. 37 — 59), and alphabetically arranged by Endlicher, in his 
 Chinese Grammar. Humboldt discusses them in his work on the 
 Kavi language (p. 428). Besides the Chinese, the Tai, Burmese, 
 and Malay languages, the Mexican also employs similar generic 
 exponents. 
 
 12. The pronouns in Khamti are the only words which have a 
 separate form for the plural : 
 
 Kau, I, becomes Hau, we. 
 
 Mail, thou, „ Mail su, you. 
 
 Man, he, ,, Man khau, they. 
 
150 
 
 In Malay, the pronouns afford the 6nly instance of a distinction 
 of number which exists in the language : 
 
 Ku, I, becomes Ki-ta, we. 
 
 Mu, thou, „ Ka-mi, you. 
 
 Na, he, „ Marika, they. 
 
 The coincidence of the radical portion of the pronouns, particularly 
 in the first and second persons, requires no comment. 
 
 13. Malay and Siamese are both distinguished by an extraordinary 
 number of pronouns of the first and second persons, which are used 
 according to the rank which the speaker or the person addressed is 
 supposed to hold. They are in reality not pronouns, but substan- 
 tives, meaning servant, lord, etc. 
 
 14. The relative pronoun in Khamti is Yang; in Malay, Yang. 
 
 15. The Malay has possessive pronominal suffixes, which are 
 really the personal pronouns in their shortest forms, appended to 
 nouns. 
 
 For instance : arta-ku, my property, 
 
 arta-mu, thy property, 
 arta-na, her property. 
 
 The same phrases can be formed in Khamti, where 
 mii, is hand, man, he ; and 
 mil man, his hand. 
 
 16. Demonstrative pronouns : 
 
 Malay : Ini, this. Khamti : Annai, this. 
 
 Itu or nun, that. Annan, that. 
 
 17. Interrogative pronouns : 
 
 Malay : Apa, who ? Khamti : Phaii, who ? 
 
 18. The verb in Malay and Tai is so simple that we can hardly 
 expect many coincidences between these two languages, which might 
 not be pointed out in other dialects standing on the same low level 
 of grammatical development. A few, however, may here be men- 
 tioned. There are no terminations in either, to express the persons 
 of the verb ; the root remains the same whatever pronoun may pre- 
 cede it ; and even the particles used to indicate the past or future, 
 the transitive or intransitive, the potential or subjunctive mood, 
 
151 
 
 are frequently omitted, particularly in conversation. "In Malay," as 
 Mr. Crawfurd says, " we can fancy a period in the liistory of the 
 language, in which these particles may not have been used at all, 
 the single radicals being found sufficient for all the ideas of a rude 
 people and an uncouth tongue." During this state of language the 
 absence of all formal distinction between nouns and verbs, would 
 naturally bring out that other feature of Turanian grammar to which 
 we have frequently alluded before, and which Mr. Crawfurd describes 
 in the following terms: — "With the exception of some pronouns, 
 nouns representing material objects, the prepositions which stand for 
 the cases of languages of complex structure, and a few conjunctions 
 and adverbs, any part of speech may, by the application of in- 
 separable particles, be converted into a verb. Thus the nouns ati, 
 the heart, tuwan, master, prang, war ; the adjectives, bayik, 
 good, batul, straight, putih, white ; the pronouns, aku, I, and di-ri, 
 self; the prepositions ad*ap, before, balakang, behind, ampir, 
 near ; and the adverbs lakas and sigr a, quickly ; are all convertible 
 into verbs by the application of certain inseparable particles." 
 
 19. Three tenses can be traced in Khamti ; the present, the past, and 
 the future. In the present we have the verb in its complete state ; in 
 the past, a particle is added, denoting completion or fulfilment ; and 
 another particle^ expressing will or determination, makes the future : 
 
 I!x. Kau kin, I eat. 
 
 Kau kin yau, I have eaten. 
 Kau ta kin, I shall eat. 
 
 In Malay, time is often left to be inferred ; but when it becomes 
 necessary to state present time, such adverbs as sakarang, now, or 
 the verb ada*, to be, are employed. A preterite or past time is 
 expressed by the adddition of talah, past, sudah, enough, abis, 
 ended, 1 a 1 u, gone. Future time is expressed by the verbs m a u, to will, 
 and an dak, to desire ; and by the preposition akan, to, for instance, 
 
 Aku kan, I eat. 
 
 Aku kan lalu, I have eaten. 
 
 Aku akan kan, I shall eat. 
 
 * In Siamese, the verb to be, ay u, is used to form the present. 
 
152 
 
 20. A potential mood is commonly expressed in Khamti by pa, 
 can : 
 
 Ex, Kau kin pa, I can eat, I could eat. 
 
 In Malay, a potential mood is expressed by the verbs bulih and 
 dapat, used as auxiliaries ; and which, literally translated, mean to 
 can or be able, and to get or find ; but which have the English sense 
 of can or may : 
 
 Ex. Aku kan dapat, I can eat. 
 
 21. Nouns are derived from verbs in Malay, by applying to a 
 radical the affix a n, or the prefix p a, or both together : 
 
 Ex. Dagang, to trade ; pa-dagang, a trader, 
 suruh, to order ; suruh-an, a messenger, 
 asap, to smoke; par-asap-an, a censer. 
 
 In Khamti the participial form is denoted by the particle an, put 
 before the verb : 
 
 Ex. An-kin, eating. 
 
 A language which shares so many grammatical principles in 
 common with Khamti and Siamese, and differs from Sanskrit on 
 every essential point of grammar, can no longer be counted as a de- 
 graded member of the Arian family, however great the authority of 
 him who first endeavoured to link Sanskrit and Malay together. 
 Without entering into the question of the spreading of the Malay 
 dialects, and the connection of the Malay and the other Polynesian 
 idioms, we may safely assert that the grammatical fibres of the Tai 
 and the Malay languages hold closely together ; and that the Malays, 
 whatever their later wanderings may have been, must, in their first 
 state, be traced to the Continent of Asia, and to the same home 
 from which the inhabitants of the whole Eastern peninsula proceeded 
 southward in times unreached by history or by tradition. If the Malay 
 is thus secured to the Turanian family, the whole question of its 
 connection with the Polynesian languages will have to be viewed in 
 a nevf light, and the conflicting opinions of Humboldt and Crawfurd 
 may receive a Solution consistent both with that fundamental unity 
 which struck the comprehensive genius of the former, and the startling 
 discrepancy of local varieties that attracted the notice of so patient a 
 collector and so careful an investigator as Mr. Crawfurd. 
 
153 
 
 § 2. Humboldt's and Crawfurd's Views on the Languages of Poly- 
 nesia and the Indian Archipelago. 
 
 The following extracts from Humboldt's and Crawfurd's works 
 will show the present state of this disputed question. I have availed 
 myself of an excellent article " On the conflicting Views of European 
 Scholars, as to the Eaces inhabiting Polynesia and the Indian 
 Archipelago," by the Honourable Sir Erskine Perry, in which the 
 evidence on both sides is summed up with fairness and lucidity. Sir 
 Erskine gives the passages from Humboldt in so masterly a trans- 
 lation, that we hope he may soon publish a more complete translation, 
 which, he tells us, he has prepared for educational purposes. Hum- 
 boldt's view on the Malay language, as given in his posthumous work, 
 " On the Varieties of Human Language and its Influence on the 
 Mental Development of Mankind," may be stated in his own words, 
 as follows : 
 
 " The races of Malay origin, with respect to locality, government, 
 history, and, above all, language, are perhaps more singularly con- 
 nected with races of different cultivation than any other people in 
 the world. They inhabit only islands and island groups, but these 
 extend over so wide a ran^e as to afford unmistakeable testimony to 
 their early acquaintance with navigation. Their settlement on the 
 Continent at Malacca scarcely deserves to be mentioned here, as it is 
 of modern date, and proceeded from Sumatra, and that on the coasts 
 of the China Sea, and of the Gulf of Slam, at Champa, was a still 
 later occurrence. "With these exceptions, we are unable to trace, 
 with any certainty, even in the most remote history, the existence of 
 Malays on the mainland. If from these races we separate those who 
 in a strict sense deserve the name of Malays, and who, according to 
 undeniable grammatical researches, speak closely allied tongues, easily 
 intelligible to one another, we shall find them settled (only mentioning 
 those points where the inquiry into languages has had sufficient ma- 
 terials to work on) in the Philippines,— where the language is to be 
 found in the richest development of forms, and -in its most original 
 condition, — in Java, Sumatra, Malacca, and Madagascar. A large 
 number of words, however, of unquestionable relationship, and even 
 
154 
 
 the names of a considerable number of islands, betoken that the 
 islands in the neighbourhood of the above localities are peopled by a 
 similar race, and that even the more strictly so-called Malay lan- 
 guage extends itself over all that portion of the South Pacific which 
 reaches from the Philippines southerly to the West Coast of New- 
 Guinea, and, more westerly, to the chain of islands which joins the 
 eastern point of Java, and runs up between Java and Sumatra to 
 the Straits of Malacca. It is a matter for regret that the large 
 islands of Borneo and Celebes, to which probably all that has been 
 said above may apply, have not yet had their languages sufficiently 
 examined to allow of any conclusion being drawn on grammatical 
 grounds. 
 
 " To the eastward of the zone here drawn of the pure Malay lan- 
 guage, from New Zealand to Easter Island, thence northerly to the 
 Sandwich Islands, and then back again westwards to the Philippines, a 
 race of islanders is to be found, who display most unquestionable traces 
 of an old connection in blood with the Malays. This is proved by 
 the number of similar words, and essential coincidences of physical 
 structure, in the languages whose grammar we know intimately, such 
 as those of New Zealand, Tahiti, the Sandwich Islands, and Tongu. 
 A like similarity is to be found in manners and customs, especially 
 where pure Malay customs are recognisable, unadulterated by Indian 
 usages. Whether the races to the north-west in this part of the 
 Pacific belong wholly or in part to the latter division, or to the Malays 
 in the strict sense ; or whether they form a connecting link between 
 the two, cannot yet be decided with our present materials, as even the 
 researches which have been set on foot with respect to the language 
 of the Mariana Group have not yet been made public. The whole 
 of these races possess social institutions sufficiently complicated to 
 make it improper to exclude them wholly from the class of civilized 
 nations. They have a well-established, and by no means simple 
 system of government, of religious doctrines, and of usages, and 
 some of them possess a species of spiritual government ; they display 
 skill in various arts, and are bold and experienced seamen. We find 
 amongst them in several spots the remains of a sacred language, un- 
 intelligible even to themselves ; and their custom of recalling formally 
 obsolete expressions into life on certain occasions, speaks not only to 
 
155 
 
 the richness, age, and depth of the language, but also to their powers 
 of observation as to the effect of time in modifying circumstances. 
 With all this they allowed, and still partly allow, barbarous prac- 
 tices inconsistent with civilisation.* They appear never to have 
 acquired the art of writing, and, consequently, are deprived of all 
 that literature which is founded upon it, although they are by no 
 means wanting in fanciful legends, impressive eloquence, and poetry 
 in defined rhythmical cadence. Their languages, however, have not 
 sprung out of any corruption or change of the Malay tongue of the 
 narrower zone, but we may rather trace in them an uniform and 
 original condition of the latter. 
 
 " Along with the race thus described in the two divisions of the 
 Great Southern Archipelago, we meet, on some of the islands, with 
 people who, from their appearance, must be attributed to a wholly 
 different stock. Both the Malays in the stricter sense, and the more 
 eastern inhabitants of the South Sea, belong without doubt to the 
 same human family, and they form, if one makes an accurate division 
 by colours, the class passing from the light brown into white. The 
 races of whom we are now speaking approximate, by their black 
 skin, occasionally by their woolly frizzled hair, and by their peculiar 
 features and build, to the African Negro, although, according to the, 
 most trustworthy evidence, they are nevertheless essentially different, 
 and can by no means be considered as the same race. Writers on 
 these countries, in order to distinguish them from Negroes, call them 
 either Negritoes or Austral-Negroes, and but few of them exist. 
 Both in the islands inhabited by the Malay races, and in the Philip- 
 pines, they usually occupy the middle of the island, and inaccessible 
 hills, to which they appear to have been gradually driven by the 
 more numerous and powerful white race. We must carefully, how-, 
 ever, distinguish them from the Haraforas, or Alfuris, the Turajos of 
 Celebes, who are to be found in Borneo, Celebes, the Moluccas, 
 Mindenao, and some other islands. These latter appear to have been 
 driven out in a similar manner by their neighbours, but belong to 
 the light brown race ; and Marsden attributes their disappearance 
 
 * Mr. Crawfurd mentions a somewhat cultivated race in Samatra, well ac- 
 quainted with letters, who appear to be the only literary cannibals recorded in 
 history. — E. P. 
 
156 
 
 from the coast to Mahomedan persecution. In wildness they ap- 
 proximate to the black race, and they constitute a population of 
 uniformly low development. Other islands, amongst which are some 
 large ones, like New Guinea, New Britain, New Zealand, and some 
 of the Hebrides, contain these Negro races only, and the inhabitants 
 of the large continents of New Holland and Van Dieman's Land, so 
 far as there has been hitherto opportunity of becoming acquainted 
 with them, belong to the same race. But although this race in all 
 the localities here indicated displays general marks of similarity and 
 relationship, it is by no means thoroughly established how far es- 
 sential differences of race exist among them, for their language has 
 not yet been investigated so as to satisfy the exigencies of a thorough 
 grammatical inquiry. We have only the materials collected by the 
 Missionary Trelkeld as to one race in New South Wales, by which 
 we are enabled to form any judgment as to its organic and gram- 
 matical structure. The race everywhere distinguishes itself by a 
 greater wildness and barbarism than appears in the lighter races ; 
 and the differences herein relate solely to their greater or less in- 
 tercourse with the latter. The inhabitants of New Holland and 
 Van Dieman's Land appear to stand on the lowest grade of civiliza- 
 tion which has ever yet been occupied by mankind. It is a remark- 
 able phenomenon to meet, even on the peninsula of Malacca, the 
 light and dark races in contact with one another ; for the Semangs, 
 who occupy part of the mountain range of that country, are by most 
 unquestionable testimony, a woolly-haired Negrito race. As this 
 is the only point of the mainland of Asia where the fact occurs, it 
 is unquestionable that immigration must have taken place here at a 
 comparatively recent period. Among the lighter races, also, as the 
 Malay expression orang henna (men of the country) appears to 
 prove, more than one immigration seems to have occurred. Both 
 occurrences only show, therefore, that the same kind of connection 
 between countries at different periods brings about similar historical 
 facts, and, consequently, to this extent there is nothing remarkable 
 in them. In reference to the state of culture of the different races of 
 mankind in this Archipelago, however, any explanation by means of 
 colonization becomes deceptive. To enterprising nations, the sea 
 offers rather a means of easy connection than of distinct separation, 
 
157 
 
 and the general diffusion of bold active seamen, like the Malays, ex- 
 plains itself in this way, by short trips from island to island, some- 
 times intentionally made, and sometimes by their being driven away 
 through the violence of the prevailing winds; for activity, expert- 
 ness, and knowledge of sea-craft, are not characteristics of the proper 
 Malay only, but are to be found amongst the whole of the light brown 
 race. I need only mention here the Bugis of Celebes, and the South 
 Sea Islanders. But if this description of the Negritoes, and of their 
 diffusion from New Holland to the Philippines, and from New Guinea 
 to the Andamanns, is correct, these races must hava deteriorated 
 more than is usually supposed from a more civilized condition, and 
 have become wild. Their present condition rather favours the 
 hypothesis, which is not in itself improbable, of revolutions of 
 nature, old traditions of which still exist in Java, by which a popu- 
 lous continent became broken up into the present island groups. 
 Men, like ruins, might, so far as mankind could survive such con- 
 vulsions, have remained on the scattered island tops. Both of these 
 explanations, perhaps, if united, so as to consider the dislocation by 
 the powers of nature as occurring during a lapse of centuries, and dis- 
 tinguished from the connection through human colonization, might 
 perhaps afford us some sort of account of the various races which now 
 appear. 
 
 " Tanna, one of the Hebrides, but a word of Malay origin^ New 
 Caledonia, Timor, Ende, and some other islands, possess a population 
 which is left doubtful after inquiry whether we are to reckon it, with 
 Crawfurd, as a third race, or, with Marsden, as a mixture of the two 
 others ; for the inhabitants, in their physical ma,ke, wooUiness of hair, 
 and colour of skin, occupy a middle place between the light brown 
 and black races. If, at the same time, a similar affirmation can be 
 made a? to their language, this circumstance would tell authoritatively 
 for their being a mixed race. There still remains an important 
 question, but one very difficult to decide from the materials at hand, 
 viz., how far older and more intimate mixtures of the white and 
 black races have occurred in these countries, and how far gradual 
 changes may thereupon have ensued in language, and even in colour 
 and growth of hair, the wooUiness of which, moreover, in some 
 localities, is cultivated as an ornament. To judge correctly of the 
 
 M 
 
158 
 
 Negro races in their pure form, we must always commence with the 
 inhabitants of the Great Southern Continent, as between these and 
 the brown races no direct contact is conceivable, and according to 
 their present condition it is difficult to suppose any kind even of in- 
 direct connection. The remarkable fact still remains, that many 
 words in the languages of these races, although we certainly possess 
 only a few of them, bear an evident likeness to the words of the 
 South Sea Islands. 
 
 " Amid these geographical relations, in some instances amounting 
 to close neighbourhood, certain Malay races adopted Indian civiliza- 
 tion to such an extent that perhaps no similar example is to be found 
 of a nation undergoing such a complete infusion of the national 
 spirit of another race, without losing its own independence. The 
 phenomenon as a whole is very intelligible. A large part of the 
 Archipelago, and the most attractive from its climate and fertility, 
 lay at a very short distance from the great continent of India — 
 opportunities and points of contact were consequently abundant. But 
 where such occurred, the preponderating influence of a civilization so 
 ancient, and so diffused through every branch of human activity as 
 the Hindu, could not fail to attract towards it other nations of 
 active and impressionable temperaments. This was rather a moral 
 than a political revolution. We recognise it in its consequences in 
 the Hindu elements, whieh undeniably present themselves to view 
 in a certain -range of Malay races; but how did this mixture arise? 
 On this pointy even amongst the Malays, as we shall see, nothing but 
 obscure and doubtful traditions exist. If inroads of powerful races 
 and extensive conquests had produced this state of things, clearer 
 traces of such political events would have been preserved. Intel- 
 lectual and moral causes work, like nature herself, in silence, and their 
 operation is similar to the growth of a seed, eluding observation. 
 The modus operandi in which Hinduism struck root amongst the 
 Malay races, proves that as a mental spring of action it excited the 
 imagination, and became powerful through tiie impressions of wonder 
 which it produced in races susceptible of culture. In India itself, so 
 far as I know, we find no mention of the South-eastern Archipelago in 
 Hindu history or literature. Even if Lanka were perhaps considered 
 to extend further than the limits of Ceylon, this was only dark and 
 
159 
 
 uncertain surmising, or mere poetical license. From the Archipelago 
 itself, on the other hand, as we may well conceive, nothing proceeded 
 which coirld have any influence on the mainland. It was India that 
 exerted a substantial influence, and perhaps even by colonization, 
 which was not intended to keep the mother country in view as a 
 home, or to preserve relations with it. Reasons for establishing set- 
 tlements might be various. How far Buddhist persecutioa might 
 have co-operated, I shall have to discuss hereafter. 
 
 " But to explain properly the mixture of Malay and Hindu ele- 
 ments, and the influence of India on the whole of the Indian Archi- 
 pelago, we must discriminate between its different modes of opera- 
 tion, and thereby commence with that which, early as it may have 
 began, has continued to the latest times, and consequently has left 
 the clearest and most indelible traces. It is not only the influence 
 of a spoken foreign language which in this case, as in all mixtures 
 of nations, operates powerfully, but also the whole of the mental 
 culture which springs out of it. This phenomenon is unquestion- 
 ably apparent in the introduction of Indian language, literature, 
 myths, and religious philosophy into Java. The whole purport of 
 the following work is to discuss this question, but principally with 
 reference to language, — I therefore must content myself here with 
 this mere allusion. This species of influence affected only the Indian 
 Archipelago, properly so called, and the Malay zone in its stricter 
 sense ; but possibly not even the whole of the latter, and certainly 
 not to an equal extent. The focus was so undoubtedly Java, that 
 we may reasonably doubt whether that island was not the immediate 
 source from which it extended itself over the rest of the Archipelago. 
 Independent of Java, we find, however, distinct and complete proofs 
 of Indian civilisation amongst the proper Malays and Bugis of 
 Celebes. A true literature, from the essential elements of the form- 
 ation of language, is only capable of existing contemporaneously 
 with a written character which is in daily use. It is an important 
 fact, therefore, for the mental development ot the South-eastern 
 Archipelago, that just that portion of the island group which has 
 been designated as strictly Malay possesses an alphabetic character. 
 A distinction not to be overlooked, however, here occurs. The al- 
 phabetic character in this part of the world is Indian. This arose 
 
 M 2 
 
160 
 
 naturally from the intellectual relations of these countries, and is 
 visible in most of their alphabets, with the exception, perhaps, of the 
 Bugis, in the similarity of the letters, not to mention their arrange- 
 ment to designate sounds, which undoubtedly does not furnish any 
 decisive proof, as it might have been adopted subsequently to a 
 foreign alphabet. Nevertheless, a complete similarity, with merely 
 an adaptation to the simpler phonetic system of the indigenous 
 tongues, occurs only in Java, and perhaps at Sumatra. The cha- 
 racter of the Tagalis and of the Bugis is so different, that it may be 
 regarded as an example of alphabetic invention. In Madagascar the 
 Arabic character has planted itself, as the Indian has done in the 
 centre of the Archipelago. At what period this occurred is uncer- 
 tain. And there does not appear to be any trace of an original 
 character which it displaced. The use. of the Arabic character 
 amongst the Malays proper decides nothing as to their intellectual 
 relations, which we are now discussing, for it is notoriously a modern 
 introduction. I have already mentioned the total want of all writing 
 in the South Sea Islands, and amongst the woolly-haired races. 
 The traces of Hinduism which we have here in sight are so distinct 
 that we may recognise them everywhere without difficulty, and we 
 can distinguish them as foreign elements. No true intermixture or 
 amalgamation is here discernible, but a mere mosaic union of foreign 
 and native. So far as relates to manners and customs, we may clearly 
 recognise in Indian antiquity the foreign words in the Sanskrit de- 
 scended to us, and which have not entirely lost their grammatical 
 forms : we may even discover the laws which governed the trans- 
 plantation of foreign elements of speech into a native soil. This is 
 the foundation of the cultivated and poetic language of Java, and is 
 closely connected with the introduction of literature and religion. 
 All that has been said above undoubtedly has not operated with the 
 language of the people, and still less can it be affirmed, that merely 
 because Indian words are to be found in it, they were introduced in 
 a similar manner. In thus tracing minutely the operations of the 
 different modes of Indian influence, two deeply-seated questions 
 arise, suggested by actual phenomena, but which are extremely dif- 
 ficult to answer accurately, viz. whether the whole of the ci\ ilisation 
 of the Archipelago is traceable to an Indian origin ; and whether, 
 
161 
 
 from a period anterior to the rise of literature, and to the last and most 
 complete development of the language, any connection existed between 
 the Malay and Sanskrit languages, which is still capable of being 
 traced in the social elements of speech ? 
 
 "I am inclined to answer the former of these questions in the 
 negative. It appears to me to be made out that the brown race had 
 an original civilization of their own. It is still to be found in the 
 Eastern portion, and is not altogether unrecognizable in Java. It 
 may, indeed, be said, that the population of the Archipelago prin- 
 cipally issued from its centre, where the influence of India was most 
 powerful, and extended itself thence towards the east and west, so that 
 the distinct Hindu element becomes more diluted at each extremity. 
 This proposition, however, is supported, less by any distinct simi- 
 larity than by remarkable coincidences in manners, which have 
 nothing specially Indian to distinguish them, amongst the races of 
 the central and eastern parts of the Archipelago. One sees also no 
 reason why we should deny to a race like the Malay a self-developed 
 civilization, in whatever subsequent direction the march of population, 
 and their gradual culture may have been. A proof is even afforded 
 by the readiness of the different tribes belonging to the race to adopt 
 the Hinduism imported among them, and, still further, by the manner 
 in which they still retain the indigenous element, and scarcely ever 
 allow its peculiar form to merge in the Indian. The contrary would 
 have happened if these races had been wild, uncultivated savages, 
 when Indian colonization first came in contact with them. "When I 
 speak here of Hindus, I of course only mean people speaking the 
 Sanskrit language, and not the inhabitants of the continent of India 
 generally. How far the one race came in contact with, and was, 
 perhaps, driven out by the other, I do not now enter upon, as my 
 purpose is only to show the different elements of civilization by which 
 the Malay'races were influenced. 
 
 " The second question, which alone relates to language, must, I 
 conceive, be answered in the affirmative. In this respect the limits 
 of Hindu influence have a wider range. Without mentioning the 
 Tagali, which contains a tolerable number of Sanskrit words, with 
 completely different meanings, there are to be found, even in the 
 languages of Madagascar and the South Sea Islands, both words and 
 
 M 3 
 
162 
 
 Bounds belonging to the Sanskrit, and in such an elementary part of 
 speech as the pronoun ; and even the modes of change of sound, 
 which may be looked upon as a good comparative test of the period 
 of introduction, are different in the languages of the narrower Malay 
 zone, in which, as in the Javanese, it is notorious that the influences 
 of Hindu literature and language displayed themselves at a much 
 later period. It becomes, therefore, a matter of great difficulty to ex- 
 plain this phenomenon, and to ascertain what reciprocal operation 
 these two great families of languages have on one another. At the 
 end of this essay I will return to the subject, as it is sufficient for me 
 here to call attention to the influence of Sanskrit on the Malay lan- 
 guages, which appears to be distinct from the subsequently intro- 
 duced mental cultivation and literature, and to belong to a much 
 earlier period, and to different connections between the two races. 
 I shall subsequently touch on the languages of the Negro race, but 
 must make the preliminary remark now, that if in some of these 
 tongues, as in the Papuan of New Guinea, for example, similarities 
 with Sanskrit words are to be found, this does not at all prove any 
 immediate connection between India and those islands, as such common 
 words might have been introduced through the commerce of the 
 Malays, just as we see now with Arabic terms. 
 
 " On seeking, therefore, to take a general view of the state of the 
 civilization of the great Archipelago, we find the Malay populations to 
 be hemmed in, as it were, between influences and characteristics 
 which are strongly contrasted. On the same islands and island groups, 
 which still contain races on the lowest level of civilization, or where 
 at all events such tribes once existed, we find a very ancient state of 
 culture, which had borne choice fruits, and which, derived from India, 
 had become indigenous. Tlie Malay races have appropriated this cul- 
 ture, in nearly all its parts, to themselves. Herein they may be per- 
 ceived to be connected in race to the inhabitants of the South Sea 
 Islands, who, compared to them, may be looked on as savages ; and it 
 is even doubtful whether their language is altogether strange to the 
 Negro races. The South Sea Islanders have kept themselves distinct 
 from those rude races by institutions peculiar to themselves, and by a 
 language which in their present form is quite their own. The popu- 
 lation of the Great Archipelago, which, according to our present 
 
163 
 
 knowledge, cannot bo traced to the continent of Asia, is found in 
 places where all foreign influence must be left out of consideration, 
 in a most rude and savage state, or on the lowest step of civilization. 
 This is especially true if we regard only the Negro races and the 
 South Sea Islanders, and exclude the Malay races, strictly so called, 
 although no very sufficient ground presents itself for ascribing to 
 these races a much higher station in civilization before Indian in- 
 fluences had operated upon them. We still find, even with the 
 Battas of Sumatra, whose myths and religion display unmistakeable 
 traces of Hindu influence, the barbarous custom of cannibalism on 
 certain occasions. The Great Archipelago, however, gxtends itself 
 along the whole coast-line of Asia, and connects itself with both its 
 extremities, stopped only by Africa on one side, and America on the 
 other. Its centre lies at a considerable distance, so far as navigation 
 is concerned, from the nearest point of the continent of Asia. At 
 different times, therefore, it has been acted upon from the three great 
 focuses of the earliest awakening of the human mind amongst man- 
 kind — China, India, and the seat of the Semitic races. It has felt 
 the different influences of all of them at proportiona^^ely remote 
 periods. To its earlier progress India alone contributed anything of 
 importance ; Arabia nothing, even if we except Madagascar ; and 
 China just as little of importance, notwithstanding its early settle- 
 ments." 
 
 While, as these extracts show, the philosophical mind of Hum- 
 boldt was always turned toward the problem of the unity of lan- 
 guage, and bent on the discovery of the few remaining threads that 
 would hold the vast tissue of the Polynesian dialects together, Mr. 
 Crawfurd approaches the subject from the very opposite point, as a 
 careful observer, awake to all that is 'peculiar in each dialect, and 
 anxious rather to distinguish than to combine. Nothing is more use- 
 ful to the progress of scientific discovery than the cooperation of 
 men following principles so antagonistic. They mutually check 
 and correct one another. While Humboldt thought already of 
 linking the whole Polynesian family with the Arian through the me- 
 dium of Sanskrit, Mr. Crawfurd shows that the Polynesian dialects 
 themselves have not yet been definitely traced to one common source. 
 But the disparity of dialects which rivets Mr. Crawfurd's eye, dis- 
 
 M 4 
 
164 
 
 appears in great part under the comprehensive grasp of a scholar like 
 Humboldt. Their methods, though different, will tend in the end to 
 elicit the fact, that languages apparently unconnected in grammar and 
 dictionary, can yet be reclaimed and comprehended under one common 
 name by the discovery of a few characteristic features, which it would 
 be impossible to consider as the result of mere accident. 
 
 Mr. Crawfurd holds, 1st, that there is no foundation for the pre- 
 valent idea that, Negroes excepted, all the descriptions of men in the 
 limits above described belong to the same race ; on the contrary, there 
 are several races. 
 
 2nd. He also contends that many of the nations belonging to the 
 same race, for example, the Malays and Javanese, speak distinct 
 languages. 
 
 3rd. He holds that the black race, the Austral-Negroes, or Negri- 
 toes, are not identical, and that their languages, like their races, are 
 also distinct. 
 
 4th. He admits that the Polynesians speak one very largely diffused 
 language, with dialectic differences, but maintains that it is quite 
 distinct from Malay. 
 
 As to identity of words being an indication of relationship between 
 languages, Mr. Crawfurd denies it. " In the Malay and Polynesian 
 languages," he says, " well sounding foreign words very readily gain 
 admission. Instead of words expressing simple ideas being excluded, 
 I should, on the whole, owing to the familiar and frequent use of the 
 ideas, consider them the most amenable to adoption of any class of 
 words whatsoever. Accordingly, such words will be found to have 
 supplanted native terms altogether, or to be used as familiar terms 
 along with them. Thus, to give some examples in Malay : the most 
 familiar words for the head, the shoulder, the face, a limb, a hair, a 
 pile,, brother, house, elephant, the day, to speak, to talk, are all 
 Sanskrit. 
 
 "In Javanese, we have from the same Sanskrit: the head, the 
 shoulders, the throat, the hand, the face, father, brother, son, daughter 
 woman, house, buffalo, elephant, with synonymes for the dog and 
 hog, the sun, the moon, the sea, and a mountain. 
 
 " In the language of Bali, the name for sun in most familiar use is 
 
165 
 
 Sanskrit, and a word of the same language is the only one in use for 
 the numeral ten. 
 
 " It is on the same principle that I account for the existence of a 
 similar class of Malayan words in the Tagala* of the Philippines, 
 although the whole number of Malayan words does not exceed one 
 fiftieth part of the language. 
 
 '' In the Maori or New Zealand, the words forehead, sky, great, 
 stone, point, to drink, to die, are Malay or Javanese ; yet of these 
 two tongues there are not a hundred words in the whole language. 
 
 " As to the personal pronouns, which have often been referred to as 
 evidence of a common tongue, in as far as concerns the language 
 under examination, they are certainly the most interchangeable of 
 words, and cannot possibly be received as evidence. Some of them, 
 for example, are found in the Polynesian dialects, where, in a voca- 
 bulary of five thousand words, a hundred Malayan terras do not 
 exist. 
 
 " The numerals must surely be considered as out of the category 
 of early invented words, for they imply a very considerable social 
 advancement, and seem to be just the class of words most likely to 
 be adopted by any savages of tolerable natural capacity. The Aus- 
 tralians are not savages of such capacity, and although with the op- 
 portunity of borrowing the Malayan numerals, they have not done 
 so, and in their own languages count only as far as ' two.' " 
 
 All these principles thus laid down by Mr. Crawfurd, are, of course, 
 liable to considerable limitation, according to the language and people 
 which form the subject of our researches ; yet, as a general thesis, it 
 must no doubt be admitted that mere similarity of words does not 
 prove the common origin of languages. It follows, on the other 
 hand, that mere dissimilarity of words does not prove the absence of 
 
 3^ In a Tagala Dictionary of 16,482 words, published by Father Juan de Nouda, 
 Mr. Crawfurd discovers not more than 
 
 Malay and Javanese words - - 399 
 
 Sanskrit - - - 33 
 
 Arabic - - - - 7 
 
 Persian - - - - 2 
 
 Telinga - - ^ 
 
 442 
 
166 
 
 an original connection of languages. As these points has been dis- 
 cussed before, we shall proceed at once to give what Mr. Crawfurd 
 considers as the safest test of a common origin of languages. He 
 says, " The words which appear to me most fit to test the unity of 
 languages are those indispensable to their structure, which con- 
 stitute, as it were, their framework, and without which they cannot 
 be spoken or written. These are the prepositions, which represent 
 the cases of language of complex structure, and the auxiliaries, which 
 represent times and moods. If a sentence can be constructed by 
 words of the same origin in two or more languages, such languages 
 may be safely considered as sister-tongues, — to be, in fact, dialects, or 
 to have sprung from the same root. In applying this test, it is not 
 necessary that the sentence so constructed should be grammatical, or 
 that the parties speaking sister-tongues should be intelligible to each 
 other. The languages of the south of Europe can be written with 
 words common to them all, derived from the Latin without the assist- 
 ance of any of the foreign words which all of them contain. The 
 common stock, therefore, from which they are derived is Latin, and 
 they are sister-tongues. English can be written with great ease with 
 words entirely Anglo-Saxon, and without any French words, although 
 French forms a sixth part of the whole body of its words, but no 
 sentence can be constructed consisting of French words only." 
 
 So far as this is meant as an acknowledgment that grammatical 
 elements are the only safe basis for a classification of languages, 
 nothing could be said against it. But first of all, languages do borrow 
 even prepositions and conjunctions. In Turkish, every preposition 
 in the true sense of the word, I mean every preposition standing 
 before the noun which it governs, is Persian, Turkish* prepositions 
 being always placed after the noun. Many conjunctions in Turkish 
 are of Persian and Arabic origin. f Secondly, sentences can be 
 constructed in English, consisting of French, .i e. Latin, words only. 
 If I say " avarice produces misery," every word is Romanic, but it 
 does not follow that, therefore, English is a Romanic language. In 
 fact, the single letter s, used as the exponent of the third person sin- 
 
 * Cf. Redhouse, Grammaire de la langue Ottomaue, § 994. 
 t Cf. Redhouse, Grammaire, § 999. seq. 
 
167 
 
 gular, is sufficient to stamp the language in which such a sentence 
 can be framed as non-Romanic. Nothing, therefore, but grammatical 
 forms can settle the relationship of languages definitely, and even gram- 
 matical forms have occasionally been transferred from one language 
 into another. But in no instance has an entire grammatical system, 
 a complete set of terminations of declension or conjugation, been ap- 
 propriated by a foreign tongue, and where these terminations coincide 
 as a whole, we may be sure that we have to deal with cognate idioms. 
 Next to the evidence of grammatical terminations, come pronouns, 
 then numerals ; then conjunctions and prepositions ; and, lastly, words 
 expressive of the simplest ideas and the most common objects of every 
 day's life. There are instances where even such words as father and 
 mother, brother and sister, have been replaced by foreign appellations, 
 or by words newly formed in members of the same family of lan- 
 guages. But, on the whole, owing to the familiar and frequent use 
 of these words, people are unwilling to part with them and afraid to 
 replace them by foreign terms not intelligible at first to the whole 
 community. The Saxons learned to use many foreign words, yet 
 their household words remained on the whole Saxon. So did their 
 numerals without exception, so did their pronouns, and so did in the 
 highest degree their grammatical terminations. 
 
 But although we cannot agree with the somewhat too general prin- 
 ciples by which Mr. Crawfurd tests the relationship of languages, we 
 shall give the results to which his method has led him with regard 
 to the Polynesian languages. 
 
 " Applying this test to the Malayan languages, it will be found," 
 Mr. Crawfurd maintains, " that a sentence of Malay can be constructed 
 without the assistance of Javanese words, or of Javanese without 
 the assistance of Malay words. Of course, either of these two lan- 
 guages can be written or spoken without the least difficulty, without 
 a word of Sanskrit or Arabic. The Malay and Javanese, then, al- 
 though a large proportion of their words be in common, are distinct 
 languages, and as to their Sanskrit and Arabic element, they are ex- 
 trinsic and unessential. When the test is applied to the Polynesian 
 languages, we find an opposite result. A sentence in the Maori and 
 Tahitian can be written in words common to both, and without the 
 help of one word of the Malayan which they contain, just as a sen- 
 
168 
 
 tence of Welsh or Irish can be constructed without the help of Latin, 
 although of this language they contain at least as large a proportion 
 of words as Maori or Tahitian do of Malayan. The Maori and 
 Tahitian are therefore essentially the same language, and their 
 Malayan ingredient is extrinsic." 
 
 The Malayan races, according to Mr. Crawfurd, have diffused 
 themselves, and the civilization which they attained by self-derived 
 culture, from two distinct and independent centres. " The Malay- 
 speaking Malays from the rich table-lands of the interior of Sumatra, 
 — Sumatra, which, from its physical gifts, and large proportion of 
 coast line abutting on placid seas, would be at once seized on by the 
 geographer as a focus of civilization. And the Javanese-speaking 
 Malays from Java, an island not less richly endowed in physical 
 advantages." 
 
 Into the question of the common origin of the language of the 
 Malay, the inhabitants of the more eastern islands of the South Sea, 
 the Negritoes, and Haraforas, we cannot enter at present, though we 
 believe that Humboldt's work has laid open so many traces of re- 
 lationship, that even after his theory with regard to a distant con- 
 nection of the Malay with Sanskrit and the other Arian languages 
 is dropped, much remains to encourage the comparative philologist 
 to work that mine of philological research which the genius of 
 Humboldt has opened, but not yet exhausted. 
 
 The formal coincidences between the Malay and Tai grammar 
 here pointed out for the first time, furnish a link between Asia and 
 Polynesia, which, even by itself, is strong enough to hold two of 
 the mightiest chains of languages together ; the Nomads of the sea, 
 extending from the east coast of Africa to the west coast of Ame- 
 rica ; the Nomads of the Continent swarming from the south-east to 
 the north-west of Asia. But further researches will strengthen this 
 link, and add new traces of their common origin, though we have 
 hardly a right to expect many, considering that we have to deal with 
 languages, in which grammatical elements, are, as it were, at the mercy 
 of every speaker, in which roots are of the vaguest character, and 
 can, by means of accents and determinate syllables, be made to express 
 every conceivable shade- of meaning — languages which had received 
 no individual impress before their first separation, and have grown up 
 
169 
 
 since under the guidance of but few logical or grammatical principles, 
 so as to make us sometimes doubt whether we should call them works 
 of art or products of nature, or mere conglomerates of an irrational 
 chance. While in political languages, comparative philology has to es- 
 tablish a principle by which to account for coincidences such as Asmi, 
 I am, of the Veda, and Esmi, I am, used by the Lithuanian peasant of 
 the present day, a principle must be found in nomadic dialects to ac- 
 count for differences such as we find between Mandshu and Finnish, 
 Chinese and Tibetan, the Tai and Malay languages. These differences 
 must be explained by analogies to be derived from American, Indo-Chi- 
 nese, or Siberian idioms, where we still meet with tribes who, after a 
 short separation, have become unintelligible to one another, and where 
 hut few traces remain in their idioms to enable the philologist to dis- 
 cover the common basis whence all proceeded. Unless such principles 
 can be established, all attempts to prove the common origin of nomadic 
 languages will fail. To transfer the rules of Arian or Semitic phi- 
 lology to this vast field of linguistic research, would betray an utter 
 ignorance of the nature of language ; it would be, as it has been well 
 expressed, like cutting stones with razors. To consider the few re- 
 maining coincidences between such idioms as the result of accident, 
 would be a view incompatible with the philosophy of language, which 
 allows indeed casual parallelisms between dialects no longer connected 
 by any ties of relationship, but distinguishes carefully between these, 
 the result of mere accident, and other congruences, which, though 
 few in number and small in extent, could not, like the segments of a 
 circle, coincide without the admission of a common centre whence all 
 proceeded, and from which their various distances must be measured. 
 
 EicJHTH Section. 
 
 Tamulic Class. 
 
 S 1. Early Traces of the Tamulic Nishadas. 
 
 We now return through the valley of Asam to where the Brahma- 
 putra joins his sister, the Ganga. It is here, on the coast of Bengal 
 that we meet with the first historical traces of the Tamulic languages. 
 
170 
 
 Where the Gomati falls into the Brahmaputra, stood formerly the king- 
 dom of Tripura. As Garurfa, the bird of Vishwu, is praised as the de- 
 vourer of the Kiratas, /Siva's triumph over Tripura is the continual 
 theme of the worshippers of Siva. In either fable we may recognise 
 the signs of Arian conquest over Nishada races. In the north-west, 
 the Nishada had been driven into the deserts of Sugdh and Merwar 
 as early as the times of the Veda, where the Sarasvati marks the "iron 
 gate " between Aryas and Nishadas. In the south-east we see the 
 cities of the Nishada burnt by Siva, ; in the north-east the Kiratas are 
 eaten by the bird of Vishwu ; and in the south Ravawa is punished 
 and destroyed by Eama, the hero of the Ramayawa. Now Ptolemy 
 knew a royal city beyond the Ganges, and he gives its name as Tri- 
 lingaor Triglyphon. The former name is clearly Sanskrit, and 
 it is the same which the Brahmans gave to the Andhras, or the 
 most northern branch of the Tamulians, on the east coast of the 
 Dekhan ; it is Trilinga, the modern Telugu. But such could 
 not have been the name of these people, or of their capital before the 
 Brahmanic conquest. Linga may have been Arian or not, but tri 
 (three) is certainly Arian alone ; and if Trilinga was a Sanskrit 
 translation of a Tamulian word, we should in its original form 
 expect, instead of the Arian tri, the corresponding Telugu numeral, 
 which is mod&i and morfaga. Now we read in Pliny (H. N. vi. 
 21.), " Insula in Gauge est magnse amplitudinis gentem continens 
 unam, Modogalingam nomine. Ultra siti sunt Modub» (Mutiba, 
 Ait. Brah.), Molindse (Pulinda, Ait. Brah.), Uberae (Savara Ait. 
 Brah. or Sauvira ?), cum oppido ejusdem nominis magnifico ; Gal- 
 modroesi, Preti, Calissse (Kalinga ?), Sasuri, Passalae, ColubEe (Kau- 
 lubha, Lassen, ii. 206.), Orxulse, Abali, Taluktas. Rex horum. pe- 
 ditum L. M., equitum IV. M., elephantorum CCCC in armis habet. 
 Validior deinde gens Andarse (Andhra, Ait. Brah.), plurimis vicis, 
 XXX. oppidis, quae muris turribusque muniuntur," etc. Should not 
 this one nation, inhabiting what is called a large island in the 
 Ganges, and having the name of Mo dog a ling a, be again Trilinga, 
 or at least the people of the Trilingas, i. e. Telugus ? There is a 
 difficulty about the "insula." But whatever was meant by it, 
 certain it is that, in Pliny's time, a national name, Modogalinga, 
 was known near the mouth of the Ganges, and in the immediate neigh- 
 
171 
 
 bourhood of the Andhra, or the Northern Telugus. The names of 
 the neighbouring nations, also, such as Mutibas and Pulindas, are 
 known in the Aitareya Brahmana as outcast nations on the frontiers 
 of Aryavarta, and as the cursed descendants of the sons of Vi«va- 
 mitra.* There is another word, either Tamulian, or at least a Tamu- 
 lian corruption of a Sanskrit original, which proves the presence of 
 the Tamulians in the Dekhan as early as the time of Solomon. That 
 the Ophir of the Bible is Abhira or the country near the mouth of the 
 Indus, is firmly established by the fact that some of the articles which 
 Solomon received from Ophir are indigenous to India, and in some 
 cases found in no other country, for instance, sandal-wood, algumim, 
 Sk. valgu; and that the names of these articles, which are foreign, or 
 non-Semitic, can be explained by a reference to Sanskrit. Lassen's 
 "Indian Antiquities" (i. 538.) leaves no doubt on this point. Now 
 sandal-wood is obtained on the coast of Malabar, and therefore points to 
 the Dekhan. But the Plebrew name of peacocks, tuki-im, is simply 
 Malabar, vfhere togei is the word for peacock. This is again derived 
 or corrupted from the Sk. sikhin, but its occurrence in the book of 
 Kings under its Tamulian garb, shows that at a very early period the 
 eastern coast of the Dekhan belonged to the Tamulian Nishadas. That 
 the Brahmans had driven them back beyond the mouths of the Ganges 
 and Lohita before the Christian era, is proved by the Sanskrit names 
 of these localities at the time when they were collected by the Greeks, 
 and it is highly significant that even the southernmost promontory 
 of the Dekhan, Cape Comorin, was known to Ptolemy and the author 
 
 * These outcast nations are not fixed in their localities like the Arians. We 
 find Pulindas where the Aravali mountains join the Pariyatra, near Guzerate, 
 and again, where the Vindhya bends toward Pataliputra, between the Keimur 
 and Korair mountains we find Pulindas. Why should the Greeks not have 
 heard their name near the mouths of the Ganges and the Lohita ? We recognise 
 Andhras and Trilingas to where the Arian Oriyas drove them from the coast. 
 Why should the Triglypton of the Greeks, which is as near a translation of Tri- 
 linga as possible, be too far east for a Telugu kingdom in those early times? The 
 other etymology of Trilinga, which Ellis gives, and Lassen adopts, seems much 
 too Brahmanic, if Trilinga is an old national name. And that it was so, we can- 
 not doubt ; for Ptolemy knew it not only as Triglypton, but as Trilingon also. 
 If, therefore, to repeat, Pliny knew a nation in the immediate neighbourhood 
 of the Andhras by the name of Modogalinga, whatever he meant by " insula," it 
 fixes the Telugu name and the Telugu language near the mouths of the Ganges 
 and the Lohita at about the beginning of the Christian era. 
 
172 
 
 of the Periplus, not by a Tamulic name, but as Kofiapta, anpov koX 
 TToXtc ; this Ko/ictpia being, as Prof. Wilson has shown, the Sanskrit 
 name Kumaii, the Virgin, the wife of Siva. 
 
 § 2. Geographical Distribution of the Tamulic Nishddas. 
 
 In the Tamulic languages, therefore, we may recognise the most an- 
 cient dialect spoken in India ; and spoken there long before the arrival 
 of the Arians, which, however, cannot be placed after 1500 b. c. 
 There is no reason to suppose that the peninsula south of the Vin- 
 dhya mountains had not been occupied by these Nishadas until they 
 felt the pressure of the immigrating Arians in the north. Nishadas 
 were probably spread over the whole of India. They seem to have 
 had kingdoms and capitals in the most favoured spots of the 
 country, and the resistance they offered to the Arian gods shows 
 that they could not have been without a certain amount of civilisa- 
 tion. This stratum of native population was broken by the Arians, 
 absorbed in the centre, scattered towards west and east, and violently 
 pressed together in the south. Everything agrees with this supposi- 
 tion. We find the Dekhan occupied entirely by aboriginal races, 
 with only a small and late sprinkling of Brahmanic blood. Civilisa- 
 tion there is Brahmanic, and the native languages are fuU of 
 Sanskrit vocables ; but the grammar has resisted, and language has 
 thus retained its independence. In the west there are traces of 
 Nishadas from the Sarasvati and Drishadvati down to the mouth of 
 the Indus. Lassen's map, where the Nishada races are marked with 
 blue, exhibits the whole as clearly as possible. Where the Sarasvati 
 disappears before it could join the Indus, we have in the earliest 
 times traces of the Nishadas; for the Sarasvati, the sacred frontier 
 river of the sons of Manu, was fabled to disappear in the desert, 
 that the Nishadas might not see it. Along the Indus, Arian civilisa- 
 tion has made but little progress ; and whatever was done there, 
 belongs to the Vaidik times more than to the later periods of Indian 
 history. In later times, the Ganges and its tributaries carried off 
 with them toward the south-east the whole stream of Arian immi- 
 grants. When afterwards no longer the Sarasvati and Sindhu, but 
 the Yamuna became the frontier stream of Arian conquest, there 
 again, on the south-western limits of the Gangetic system in the 
 
173 
 
 valleys tributary to the Yamuna and Ganga, we find the seats of the 
 NishMas in the impermeable fastnesses of the Dandaka forest. 
 Even north of the Ganga a Nishada king, a vassal of the kings of 
 Oude, is known in the Edmayana. Still later, when the Vindhya 
 also was crossed by the Aryas, and the Narmada and Tapati stood 
 in the place of the Ganga and Yamuna, on the Satpura mountains, 
 or still more south, in the Raivata chain, we again find the re- 
 treating masses of the Nishadas, together with lUeMas, and also in 
 the immediate neighbourhood of the Kolagiri, or the Kole mountains. 
 
 Under Greek names again, we may recognise the former more 
 northern stations of these Tamulians. Where the Sarasvati is sup- 
 posed to have joined the Indus in ancient times as the sixth river of the 
 Penjab, south of Bhavalpur, the Sudrtk population is marked by the 
 name of Sydri in Ptolemy. Another well-known name of the lowest 
 tribes was ^andala. This also seems to have been originally an ethnic 
 name, for the KavSaXm were known to Ptolemy, together with the 
 Bhills, south of the Narmada, that is to say, in the Satpura moun- 
 tains — the very places where the Bhills, the Phylittse of Ptolemy, have 
 maintained themselves to the present day. The Paharias of Eajmahal 
 and Bhagalpur, have kept their homes in the Par^vanatha hills, 
 formerly the seat of the Pundras, and to the present day, a Paharia 
 of Eajmahal can converse with the Bhills and Gonds on the frontier 
 of Berar (Lassen, i. 368.) ; thus proving that they are all the scat- 
 tered fugitives of one and the same conquered army. 
 
 "We may look on the watershed between the Ganges and the rivers 
 of the Dekhan, as the broad line where Arian civilisation made front 
 and halted. This Vindhya chain, however, is not to be regarded as a 
 straight line running from Guzerate to Orissa, but rather as a winding 
 mountain-enclosure, which beginning nearly at Delhi, runs towards the 
 south, as the Aravali chain — a screen against the sands of Merwar. 
 Near Guzerate, it turns eastward, or rather is continued in this new 
 direction by the Pariyatra range, which connects the Aravali with 
 the Vindhya proper. The Vindhya then runs in a decided angle 
 towards the north, and closes the basin of the Ganges near Patali- 
 putra. These two lines, one drawn from Delhi to the Pariyatra, the 
 other from the Pariyatra to Patna, enclose a territory which sends all 
 its waters westward and N.-westward into the Ganges, or rather 
 
 N 
 
174 
 
 into its right arm, the Yamuna. For one southern stream only, the 
 Soaa, reaches the Gangetic basin after the junction of Ganges and 
 Jumna. The waters which run from the southern declivities of this 
 mountain range towards the Dekhan, are divided, nearly in the centre, 
 by the Rikshavat (Uxentus). From the Rikshavta, two ranges run to 
 the west, first, the Satpura range, which forms the southern shore of 
 theNarmada, the northern being the Vindhya, and secondly the Eai- 
 vatan mountains, which collect the southern feeders of the Tapati, its 
 northern feeders being supplied by the southern side of the Satpura 
 range. Toward the east, or rather north-east, the Eikshavat is con- 
 tinued by several mountain clusters, which stand like buttresses 
 to support the back of the Vindhya. They are chiefly known as 
 the Amarakawfaka, the Korair, the Bikeri, Malagiri, Gumagha^fe, 
 Par«vanatha or Eajmahal mountains. These stem off the waters 
 from the Ganges, and send them into the Mahanadi and Godavari. 
 
 The absence of Sanskrit names and Arian cities within the whole 
 territory which has for its base a line not very distant from the bed 
 of the Yamuna, and which is enclosed by an arch formed by the 
 Aravali, Pariyatra, Vindhya and Eajmahal mountains, shows that 
 the main army did not press strongly on this position. Detached 
 forces penetrated beyond, but principally along the coasts, not in the 
 centre of the country. The Vindhya mountains, through which the 
 southern feeders of the Yamuna and Ganges break their course, 
 offered a safe retreat for races who disliked the contact of Arian 
 society. All along the Vindhya, therefore, we find in ancient times 
 from west to east, Sydri, Abhiri, Phylittffi, Kandali, MoUndae, 
 and Sabarse ; in modern times, Minas, Meras, Chitas, Ahirs, 
 Koles, Bhills, Khonds, Gonds and Sourahs. Their strongholds are 
 regions composed of lofty and rugged mountains, impenetrable forests, 
 swampy woodlands and arid wastes, interspersed with extensive 
 tracts of open and productive plains, but possessing a climate in many 
 districts highly pestilential, like the Terais in the Subhimalayas. 
 Even now but little is known of these tribes, and their languages 
 have hardly been explored. But wherever attention has been paid 
 to any dialects, they betray a decided relationship with Tamu- 
 lic. Ellis was the first to point out that the idiom of the moun- 
 taineers of Eajmahal, close on the Ganges, if not of the same 
 
175 
 
 radical derivation, abounds in terms common to Tamil and Telugu. 
 The Eajmahali words collected by the Rev. M. Hurder, at Bhagalpur, 
 leave no doubt as to the correctness of this supposition ; and the 
 Uraon words collected by Col. Ouseley, exhibit nearly the same 
 dialect, though I do not know its proper habitat. It must be con- 
 siderably south-east of Rajmahal, almost within the Gond territory. 
 The language of the Gonds, which occupies so large a space comprised 
 between the Vindhya range on the north, the eastern chain of Ghats, 
 and a line connecting these, drawn from the mouth of the Godavary 
 to the centre of the valley of the Narmada — was first suspected of a 
 Tamulian origin in 1842, by M. Loesch, a German missionary. The 
 same subject was alluded to by Mr. D. F. McLeod, in 1844, and 
 the first list of words was published by Dr. Manger in the Journal of 
 A. S. B., in March, 1847. I have not seen either of these articles, 
 but the results to be obtained from them, were published in a highly 
 interesting essay by Mr. W. Elliot November, 1847. The Gond 
 dialects will henceforth be classed together with the Tamulic 
 languages. 
 
 §3. Separate Class of Munda Dialects. 
 
 It has commonly been supposed that the chain of these uncultivated 
 Tamulic dialects could be traced across the Dekhan without interrup- 
 tion from the Rikshavat mountains to Pariyatra, the connecting links 
 being furnished by the idioms of the Koles and B hills. Of the 
 original Bhill dialects no specimens have as yet been published, so 
 far as I am aware ; nay, it seems as if the Bhills bad adopted the 
 language of their conquerors to an extent obliterating all traces of 
 their original speech. Some of the Kole dialects have been collected 
 by Captain Haughton But in the lists printed by Mr. Hodgson, I 
 observe an agreement between Rajmahal, Uraon, and Gondi words, 
 and, so far as words are concerned, I should say that the dialects 
 spoken by the Rajmahal-Koles, and the Uraons, are of the same 
 family as the Gond, and, therefore, of Tamulic origin. But this 
 cannot be said of the Sinhbhum, Sontal, Bhumij, and Mundala 
 Koles, though Mr. Hodgson inclines to believe that all these dia- 
 lects belong to the same class. He says, " the affinities of these 
 
 N 2 
 
176 
 
 tongues are very striking, so much so, that the five first (Sinh- 
 bhum, Sontal, Bhumij, Uraon, and Mundala Kole) may be safely 
 denominated dialects of the great Kole language ; and through the 
 Uraon speech we trace without difficulty the further connection of 
 the language of the Koles with that of the hill men of the Eajmahal 
 and Bhaugalpur ranges. Nor are there wanting obvious links be- 
 tween the several tongues above enumerated and that of the Gonds." 
 Here I must diifer from Mr. Hodgson, although I confess the mate- 
 rials hardly suffice for arguing the point satisfactorily. But 
 taking his own lists of words, I can see indeed many coincidences 
 between Uraon, Rajmahali, and Gondi on one side, and Sinhbhum, 
 Sontal, Bhumij, and- Mundala words on the other, but none whatever 
 between these two classes. I, therefore, suppose that in the dialects 
 of the last four tribes, we have traces of a language spoken in India 
 before the Tamulian conquest, and I feel confirmed in this suppo- 
 sition by finding that these dialects are the same as the Ho, on 
 which we possess a most interesting memoir published by Lieutenant 
 Tickell, in the Journal of the Asiatic Society of Bengal, 1840. The 
 four Kole dialects and the Ho are spoken in the same locality with 
 the other Kole dialects belonging to the Tamulic family. But the 
 numerals, the pronouns, and the grammatical system (known of this 
 dialect only), differ strongly from the Tamulic ; though at the same 
 time, they do not show any traces of relationship with either Arian, 
 or Bhotiya, or Tai languages. The race by which these dialects are 
 used may have merged into the Tamulic in places where both have 
 been living together for some time. Both are, therefore, promis- 
 cuously called Koles. But historically as well as physiologically there 
 is sufficient evidence to show that two different races, the Tamulic and 
 an earlier race, came in contact in these regions, whither both fled 
 before the approach of a new civilisation. 
 
 Traces of this earlier race, as distinct from the Tamulic Koles, 
 have been found by Lieutenant Tickell from the jungles of Ramgurh 
 (near Hazaribaugh) to the south and southward, along Moherbunj, 
 Keonjur, Gangpur, down to the confines of Buna Nagpur. Here 
 they are distinguished from the Gonds by the name of "Kirkee."' 
 Their colonies, as described by the Gonds, are insulated, semi- 
 barbarous, and confined to the wildest parts of that country. The 
 
177 
 
 country lying north and north-east of Gondwana, and west of 
 Gangpur, and south of Surgujia, is in all probability inhabited by 
 the main stock whence these small settlements have wandered 
 The inhabitants of Chota Nagpur are also called Koles, but Lieutenant 
 Tickell describes them as a totally distinct race, differing from the 
 Hos not only in language, but in manners and origin. The Koles 
 of Chota Nagpur are properly named Ora6us (the same as Hodgson's 
 Uraons). They are of the same lineage as the Gonds and other 
 uncivilised Tamulian races. The Uraons still remember their former 
 habitation west of the Sone, and they ascribe their transmigration 
 across that river towards the south-east to the inroads of Hindus 
 from the vicinity of the Ganges. Now the Uraons found other 
 people already in possession of the country into which they were 
 migrating. These people, called themselves " Munda," which as an 
 old ethnic name, I have adopted for the common appellation of the 
 aboriginal Koles. Kole is too general a name, because it is applied 
 promiscuously to uncivilised races, and has become the English term 
 for porters (Coolee, or Kholee, or Kuli's) all over India. It is said 
 that the Mundas and Uraons lived peaceably together until the 
 Brahmans reached their country. The political and religious op- 
 pression exercised by the Brahmans, drove the Mundas from their 
 country ; a great portion traversing the hills and forests of Koehang, 
 passed out eastward into the open tract now called Singbhum and 
 the Kolehan. 
 
 Here they found a people called Bhuians, with whom they shared 
 the country. ' A race of Bengali Brahmans, called Sarawaks, who 
 endeavoured to establish their supremacy on the Kolehan also, was 
 driven back by the Hos and Bhuians in common. But as the Kolehan 
 was lying in the route of hosts of pilgrims from Patna and Benares to 
 Jaggernath, other adventurers tried to possess, themselves of these 
 fertile tracts of country. In this some Marwari Rajputs succeeded 
 at last, and after the total discomfiture of the Bhuians, the Marwari 
 Singbhunsis and the Hos divided the country, the Hos withdrawing 
 from the rich open plains, now called Singbhum, into the country 
 now called Hodesum or Kolehan. It is of these Mundas that Lieu- 
 tenant Tickell has given so accurate an account. He maintains that 
 they are related with the Mundas of Chutia Nagpur, although many 
 
 N 3 
 
178 
 
 of tliese form part of the good-tempered, but ugly-figured Dhangurs 
 seen in Calcutta. The Mundas of Hodesum, on the contrary, have 
 preserved their race in greater purity, and are described as men emi- 
 nently handsome, " with figures like the Apollo Belvedere." The 
 Mundas of Hodesum shave the hair off the forehead, and wear it tied 
 behind, a custom which may account for their name " Munda," i. e. one 
 who shaves his head. The other Mundas wear their locks dishevelled, 
 or clubbed at the top of the head, transfixed with a long pin or comb, 
 by which they are at once distinguished. For further distinguishing 
 marks between the Hos and Uraons, or as they are also called, the 
 Lurka-Kolea and Koles, I must refer to the two first portions of 
 Lieutenant Tickell's Memoir. My own reasons for distinguishing 
 between the Ho language and Ihe Tamulian dialects are principally 
 derived from grammatical sources. The dictionaries of the Munda 
 and Tamulian languages differ more than could be the case with 
 cognate dialects. But it is again fi-om the numerals and pronouns 
 only that we can derive a full conviction on this point. Lastly, we 
 have declension and conjugation, which, as we may trust Lieutenant 
 Tickell's account, are decisive as to the non-Tamulic character of 
 these dialects. 
 
 § 4. Languages belonging to the Tamulic Branch. 
 
 This broken line of Tamulic outposts once crossed, we are in 
 the midst of the Tamulic division of languages. The encroachment 
 of Arian Dialects, as the Mahratti and Konkani on the western, 
 and the Oriya on the eastern coast, show the course which Arian 
 civilisation, hemmed in in the middle by the Vindhya range, took on 
 each side of the Dekhan. The Mahratta conquest, which belongs to a 
 much later period, secured to the dominion of Arian speech a large 
 portion of country east of the Ghats, and surrounded by the Eik- 
 shavat, Satpura, and Eaivata mountains. After this, we have, on the 
 coast and in the interior of the Dekhan, the following Tamulic 
 dialects: — First, Tuluva; then Malayalam, on the Malabar coast 
 as far as Cape Comorin ; then, in the whole tract from the western 
 Ghats to the eastern coast, Tamil ; and, after this, until we reach 
 again the territory of the Gond, the Telugu. The central portion, 
 
179 
 
 between the Tuluva in the west and the Telugu in the east, belongs 
 to the Karnata or Canarese language. Besides these large 
 lingual provinces, smaller districts must be assigned to several popular 
 dialects. In the Nilghiri hills, for instance, we find the To das, 
 whose language becomes the more interesting the less it lias been 
 affected by any influence of Sanskrit grammar and literature. In the 
 Tulu country the Koragas and Malekudias speak similar lan- 
 guages, as has been stated by Weigle in a very interesting article on 
 the Canarese language, published in the Journal of the German 
 Oriental Society. Other dialects of the same kind are mentioned by 
 Professor Lassen (Indian Antiquities, i. 364.) 
 
 The geographical position of these languages and dialects, however, 
 is so well known, and has been so frequently discussed in the works 
 of Ellis, Wilkes, Wilson, and Lassen, that little of importance could 
 here be added. 
 
 §5. Character of the Tamulic Class of Languages. 
 
 Nor is it necessary to prove that all these languages and dialects 
 are held together by the ties of a close lingual relationship, for this is 
 one of the few facts in comparative philology, which, after it had 
 once been stated, has never been called into question. Only one re- 
 mark remains to be made on this point, and it is more of an historical 
 than philological nature. It has been pointed out once before, that 
 we have no right to suppose the Tamulic Nishadas to have been 
 mere barbarians when they first came in conflict with their Arian 
 conquerors ; that, on the contrary, the destruction of their cities, and 
 even the character of their leaders as represented, in the most hostile 
 spirit, in Sanskrit poetry*, give evidence of a former civilisation 
 crushed and scattered by the superior power of the Aryas. This 
 view is confirmed by the close relationship which unites these 
 numerous languages spoken over a surface as large as that of tht 
 Romanic dialects in- Europe. There is a certain kind of similarity 
 between languages, which can only be explained on the supposition 
 
 • Havana, the Kakshasa king of Lanka, conquered ty Rama in the Dekhan, 
 is a god worshipped by Sub-Himalayan tribes, as for instance, the Hay us. 
 J. A. S. B. 1840, p. 611 
 
 N 4 
 
180 
 
 that they had once a common historical centre. Thus, even if we 
 had no knowledge of the former existence of Latin as a political 
 language, the resemblance of the Romanic dialects would force us 
 to admit a political concentration of language previous to the time 
 when this fixed and settled speech became broken up into various 
 dialects. This resemblance between secondary dialects is a different 
 one from that which may be observed between primary dialects, 
 isuch as precede the formation of every political language. These 
 primary dialects are earlier than the Koivr), just as mountain streams 
 are earlier than rivers ; the secondary dialects, on the contrary, are 
 later, just as channels are later than lakes. Among savage 
 tribes, where these primary dialects have never been called together 
 into a literary system, we find, as in America, Africa, and Cochin- 
 China, that there exists so perplexing a, variety of idioms, that the 
 inhabitants of neighbouring villages are unintelligible to one another ; 
 and, in the absence of all checks on the caprices and peculiarities of 
 individuals, old forms are changed and new forms iatroduced by 
 every individual with such recklessness as to obscure for ever the 
 traces of a primitive community of speech. In the history of the 
 Arian family we can distinguish between several lingual centralisa- 
 tions. After one dialect has attracted or absorbed the floating elements 
 of other popular dialects, and been raised to the dignity of a classical 
 language, we see it again diverge into new branches. Latin first 
 absorbs all the idioms of Italy, and after it has become the language 
 of the then civilised world, it is broken up in turn into many dialects. 
 If this political centralisation of Italy had not intervened, and if no 
 Roman empire had brought the provinces of Italy under one common 
 sway, the dialects of the Umbrians in the north of Italy would have 
 developed themselves and become so different from that of the 
 Sabines in Lower Italy as to appear to us a totally different language, 
 differing from the Oscan at least as much as Greek from Slavonic. 
 But these two dialects, the Oscan and Umbrian, were themselves 
 political and literary languages, not to be compared with the unsettled 
 idioms of savage tribes such as we find in America. If, then, we 
 imagine a state of things where the different provinces, nay the 
 towns and villages in the separate valleys of Italy, had each retained 
 its lingual independence, each continuing to use its local dialect for 
 
181 
 
 centuries, without any political intercourse, or common literature, 
 political, religious, or legal, we should then find, as we do in 
 South America, almost as many distinct languages as there are settle- 
 ments. Kircher fixes the number of languages known to be spoken 
 in South America toward the end of the seventeenth century, at 
 five hundred ; and in most cases the people who speak such idioms 
 are said to be unintelligible to their nearest neighbours. Here, 
 therefore, in America, we should say, that the immense diversity of 
 dialects shows the absence of a previous political centralisation. 
 Now, on exactly the same ground, it follows that in the Dekhan the 
 great similarity between the different Tamulic dialects can only 
 be the result of a former period in the history of the Tamulian 
 speech, during which its character became fixed, grammatically and 
 etymologically. Such a process we can only ascribe to the influence 
 of a more comprehensive civilisation, and a more extended political 
 and literary intercourse than is generally ascribed to the aboriginal 
 inhabitants of India. The Tamulic dialects agree not only in roots, 
 not only in pronouns and numerals, but in derivative words which 
 must have been known to all before they began to diverge and 
 grow into new dialects. Perhaps it will be possible to fix on one 
 of these, dialects as the eldest of the Tamulic sisters, and derive 
 from it some of these words which are common to all. But even 
 then our conclusion would be the same; for the adoption of words 
 from one dialect into another necessitates equally the admission of a 
 political and literary intercourse, which can only take place during 
 a period of advancing civilisation. 
 
 Another reason for supposing the Tamulic languages considerably 
 advanced in their literary capabilities before their struggle with the 
 Sanskrit began, may be discovered in their successful resistance 
 against the introduction of Sanskrit elements into their grammar 
 Although the dictionary of the Tamulic languages is as full of 
 Sanskrit words as English is of Norman, yet the Tamulians did not 
 give up their grammatical independence. And even the words which 
 were adopted from Sanskrit had to submit to the genius of these 
 dialects. With the exception of those adopted ready made, and 
 simply transferred from Sanskrit, as Latin expressions are in English, 
 the majority of Sanskrit terms in the Tamulic dialects has been 
 
182 
 
 changed to such an extent, that it is difficult sometimes to discoTer 
 their foreign origin. Words simply taken from Sanskrit are, according 
 to Ellis, called " Tatsama," i. e. equal to Sanskrit. For instance, 
 sarapadu in Telugu is the Sanskrit s am pad (fortune). 
 
 Appa in Telugu is the Sanskrit apah (water). 
 Payasu in Telugu is the Sanskrit payas (milk). 
 
 Words adopted from Sanskrit with considerable phonetic changes 
 are called " Tadbhava, " i. e. produced from Sanskrit. Some of the 
 changes which these words have undergone must be ascribed to the 
 spoken or vulgar Sanskrit, for they depend on the same rules by 
 which Sanskrit words are modified in the Prakrit dialects. These, 
 the spoken or vulgar dialects of the Sanskrit, would be the most 
 natural channels through which Sanskrit words could have reached 
 the Tamulians. And as in French we find frequently the same 
 Latin word under two different forms, of which the one (as for 
 instance " redemption") might be called a tatsama, the other (as for 
 instance "ranjon") tadbhava, instances occur in the Tamulic lan- 
 guages where the same Sanskrit word has been adopted under two 
 different forms. Parva in Sanskrit means a knot or joint, and, 
 with particular reference to the moon, it means the day of the full 
 and new moon. As these were festival days, pabba and habba in 
 Canarese* mean a "festival." But in the learned language of the 
 Brahmans par van came also to signify a chapter or book, and in this 
 sense it is used in Canarese parva, section of a book. Instances 
 where Tadbhava words in Telugu seem to have passed through 
 Prakrit channels are the following : — 
 
 Sanskrit. Prakrit, f Telugu. 
 
 brahma bamha (VararuAi, v. 47.) bomma (Brahma). 
 
 brahma>ias bambhadu (Abhira) bapadu (a Brahman). 
 
 ' Cf. Weigle, Journal of the German Oriental Society, II. 265. 
 
 f The Prakrit forms are given on the authority of Ellis, in his introduction to 
 Campbell's Teloogoo grammar. Ellis must have availed himself, however, of 
 other sources besides VararuAi. Where his forms agree with Vararu/Si I have 
 added a reference to the excellent edition of this grammarian by my friend Mr. 
 CoweU, at Oxford. Where they differ, or where they do not occur at all in Va- 
 raruAi, Ellis may have followed HemaAandra, or other authorities, as he was too 
 accurate a scholar to have formed them merely on general analogy. 
 
183 
 
 Sanskrit. 
 
 Prakrit. 
 
 Telugu. 
 
 dvipas 
 
 divo (Maharashft-i) 
 
 divi (island). 
 
 ya^as 
 
 ^-aso (Varar, II. 31.) 
 
 asamu (glory). 
 
 pra%w3,tam 
 
 pae?iwwa(?am ( 5auraseni) pannidama (promise). 
 
 nedish^Aam 
 
 nedistam (Magadhi) 
 
 nestamu (friendship). 
 
 rlma 
 
 lama (ibid.') 
 
 lema (woman). 
 
 trilingas 
 
 tilingo (Varar, vi. 56.) 
 
 telugu or tenugu. 
 
 svar/zam 
 
 sannam (Paisa^i) 
 
 senna (gold). 
 
 suvarwam 
 
 pauno (Kulica-pai«aAi) 
 
 ponnu (gold). 
 
 It will appear, even from this short list, that some phonetic changes, 
 generally ascribed to the influence of the Telugu, can be traced back 
 to Prakritic corruptions, but that, at the same time, the Telugu went 
 beyond the limits of the Prakrit. 
 
 Sanskrit words form so large a portion of the Tamulic dictionaries, 
 that they are no longer considered as foreign words. Foreign words, 
 according to Telugu grammarians, are called " Anyade«iya," i. e. of 
 another country ; and the Appakaviyam explains their origin in the 
 following manner : — " The natives of Andhra (i. e. Telugus) having 
 resided in various countries using Telugu terms conjointly with those 
 of other countries, these have become Andhra terms of foreign 
 origin." 
 
 What remains, after subtracting all these extraneous ingredients, 
 is called Desya, i.e. native words. Thus it is said, in a stanza 
 of the Adharvana Vyakarana with regard to Telugu: — "All the 
 words which are in use among the several races who are aborigines 
 of the country of Andhra' which are perfectly clear and free from 
 all obscurity; these shine forth to the world as the pure native 
 speech of Andhra (suddha . andhra . desyam)." There is only one 
 more distinction made, between what are called native and vulgar 
 words. The latter are termed gramya, i.e. belonging to villages, and 
 explained by the Appakaviyam as follows : — " Such Telugu words as 
 are commonly used by rustic folk are known as gramyam; these 
 lose some of their regular letters, and are not found in poetry, unless, 
 as in abusive language, the use of them cannot be avoided." 
 
 If we now look at the grammar of the Tamulic languages, we 
 shall find at once that we have before us a system of declension and 
 
184 
 
 conjugation much more developed than in the Bhotiya dialects. The 
 forms are more settled according to general grammatical categories ; 
 and although the cases, as in all Turanian languages, are formed 
 by postpositions and are, therefore, liable to great variety, yet 
 there exists a formal distinction between the casus rectus and 
 obliquus. This base of the casus obliquus and the terminations of 
 the cases, when brought in contact, are liable to phonetic changes 
 similar to the changes of Sandhi in Sanskrit and other Arian 
 languages, and both coalesce into one grammatical whole. This gives 
 rise, as in some of the more advanced members of the Finnic and 
 Tataric branches, to some real grammatical cases, which become 
 technical, and are used in preference to mere compounds : particularly 
 in the modern and spoken dialects, where the number of independent 
 postpositions expressive of case is much smaller than in the ancient 
 languages. A still greater advance toward grammatical forms is 
 made in the conjugation. Here we find moods and tenses formed by 
 the addition of letters and syllables which by themselves have no 
 more meaning than any termination in Greek or Latin. The persons 
 are expressed by pronominal terminations, and these terminations 
 vary according to the tenses, in the same manner as in Greek 
 and Latin. A grammar like this could only be the grammar of a 
 civilised people. It shows signs of wear and tear, and in what it 
 has retained as well as in what it has given up, we can discern the 
 working of a spirit of wise economy. 
 
 Ninth Suction. 
 Comparison of the Tamulic and Ugric Languages, 
 
 If, therefore, we look for analogies to the Tamulic grammar in 
 other branches of the Turanian family we should naturally take 
 those which, like the Tamulic, have reached a certain degree of 
 grammatical perfection. This grammatical perfection, as was stated 
 before, consists first in the production of those formal elements 
 which are wanting altogether in family languages, such as Chinese, 
 and which are extremely scarce as yet in the lower Nomad lan- 
 guages, as in the Tungusic or in some of the Gangetic, the Lohitic, 
 
185 
 
 and Tai class. Secondly, in the. reduction of these formal ele- 
 ments to certain limits ; in the introduction of distinct grammatical 
 categories ; and in the suppression of many artifices which at first 
 suggest themselves as means of expressing all the minutije of the 
 most complex relations, but which, in the progress of the intellect, are 
 found not only useless, but cumbersome, for the practical purposes of 
 speech. We should, therefore, naturally look to the Tataric or Ugric, 
 and not to the Tungusic or Mongolic branches, if we expected to 
 find a similarity between the grammar of the Tamulic and that of 
 any other branch of the Turanian family. 
 
 But there are other indications, which lead us in the same di- 
 rection. 
 
 Though it is generally admitted that most members of the Turanian 
 family separated before their numerals had become fixed and un- 
 changeable, and although, at first sight, we discover hardly any traces 
 of similarity in the numerals of languages so nearly allied as Turkish 
 and Hungarian, it is the duty of the comparative philologist to search 
 for points where any two branches of this prolific family may have 
 preserved faint indications of their former unity. As the Finns are 
 the most northern, and the Tamulians the most southern colonies 
 of this Asiatic race, both were probably the last to separate from 
 their common stock. Both, also, have been removed for many 
 centuries from contact with the ever floating and changing population 
 of Central Asia, and thus may each have preserved the impress of 
 the language as it was spoken by the remaining nucleus of the 
 Turanians long after the separation of the Tai, Malay, and Bhotiya 
 branches in the South, and of the Tungusic, Mongolic, and Tataric 
 branches in the North had taken place. 
 
 Now, if we compare the Ugric and Tamulic Numerals, they cer- 
 tainly do not seem to offer much encouragement. The words for 
 one, two, and three, are evidently derived from more than one 
 root in the Tamulic, as well as in the Ural-Altaic languages. These 
 three first Numerals, however, are liable to change and fluctuation 
 in languages the common origin of which admits of no doubt. They 
 are, so to say, the most concrete Numerals, expressive of more than 
 abstract quantity, and therefore capable of being rendered in various 
 manners. Thus one has two roots in most European languages ; 
 
186 
 
 Sk. ekas and prathamas; the former expressing singleness, the 
 latter priority. "Two," also, can be expressed by duo and am bo, 
 by pair, couple, twin, and the like. One, two, three, are words, 
 and not only numerals ; they are declinable, therefore, in languages 
 where, as for instance in Latin, the other numerals are so no longer. 
 This shows their vitality and concreteness, or, if I may so say, their 
 uninterrupted self-consciousness. Now, as we have frequently seen 
 before, words which continue to be understood by the genius of a 
 language are more liable to organic change and natural variation 
 than others whose sound and meaning must simply be taken for 
 granted. It is possible, therefore, that the three first numerals may 
 differ, owing to that power of renovation and reproduction inherent 
 in Turanian languages, while the rest may yet have been preserved, 
 or at least have been exposed only to that influence of phonetic cor- 
 ruption to which such mummified words are most exposed. 
 
 But although it would be in vain to attempt to re-establish the 
 original root from which all the names for " o n e " in the Ugric and the 
 Tamulic languages could be derived, we need not shut our eyes to 
 some cases where one or the other base for one, two and three, occurs 
 north and south of the Himalaya. 
 
 The most general base for one in the Ural-Altaic languages is 
 AKAT, which reminds us forcibly of the Hebrew ekhad, the Pehlevi 
 achad, and even the Sanskrit ekas. Professor Schott traces this 
 base in the Lapponic akt and akta, the Teheremiss ikta; in the 
 Finnish yht and yksi, changed by the Esthonians into tits. In the 
 Ostiakian there remains but ot, from which the Syrianian otik may 
 be derived. Other Ostiakian forms are i t and i. The Tcheremissians 
 have, besides the full forms ikta, and iktat, a shorter form ik, and 
 the same abbreviation has taken place in the ok of the Syrianes, and 
 the ak of the Voguls. The egy of the Magyars, and the vaike of 
 the Mordvines require no explanation, the addition of an initial v 
 being of frequent occurrence; n9r can there be any doubt that vai 
 and va, which equally occur in the dialect of the Mordvines, are but 
 phonetical varieties of the same type. Instead of an additional v, 
 which we find in the Mordvinian, the Mongolian adds an initial n, and 
 forms nige, one. This nige may be said to stand for an original 
 jige, as several words in the Finnic languages show an interchange 
 
187 
 
 of j and n at the beginning of words. In the Tungusic languages the 
 form jeg does really exist, and is used to form the word for nine, 
 as " ten minus one." 
 
 Now, in the Tamulic class we find at least the Telugu oka, which 
 might be compared. But, going back to the most ancient repre- 
 sentative of Turanian speech, we can point to the Gyami iku, the 
 Chinese 'i and y ut. The Mongolic forms in n (nige) find analogies 
 in the Tai niing and liing; and in numerous members of the Bhotiya 
 family the combination of guttural and dental may be traced as having, 
 in various combinations, the power of one. 1 only mention the Naga 
 akhet, and khatu, the Kuki katka, the Miri ako, as types from 
 which many more of these Gangetic and Lohitic numerals can be 
 derived. That all can not, is hardly an objection, if we consider that 
 the Turkish also shows in its bir (one) a base independent of the 
 old AKAT ; and that a third radical for one must be admitted to 
 exist in the Tungusic u m, which explains the Manj u e m u, and several 
 cognate expressions in Mongolic and Tcheremissian, where on, in, 
 and en occur with the general meaning of unity or separation. This 
 n forms again a chief element in the Tamulic names for one. 
 
 Some of the changes by which A K A T is reduced to i may seem 
 violent, but they are so not in theory, but in reality. If we find that 
 languages so closely connected as Mordvinian and Tcheremissian allow 
 themselves forms like iktat, ikta, ik, vaike and va, and that 
 even in the same language such variants as 6t, i, and ja occur, 
 all that we can do is to state the fact in order to show that the 
 Finnic yksi and the Hungarian egy need not be considered as 
 words difierent in their origin. Besides, although the rules afiecting 
 the interchange of letters have not yet been brought to that degree 
 of completeness and certainty which in the Arian languages makes it 
 easy to prove with full evidence the common origin of such words as 
 Sanskrit A H A M and English I, yet general analogies have here 
 been discovered, and in following Professor Schott through his ad- 
 mirable analysis of the Turanian Numerals we are never left without 
 a precedent for the changes which he wishes us to admit. 
 
 The base which in a former paragraph was obtained as the most 
 likely source of all Bhotiya words for two, NYA, seems at first to 
 stand without any corresponding forms in the Ural-Altaic languages. 
 
188 
 
 We shall find, however, the most luculent proofs that in the earliest 
 state of the Ural-Altaic languages NYA was the etymon of 
 "two," and that it was supplanted by a secondary form (AKAT, 
 preceded by a guttural), in a manner that reminds us of the relation 
 between Sanskrit "tur" in turya, the third, and ^atur, four. This 
 new base for two K + AKAT, is liable to the same modifications as 
 AKAT, and hence it is sufficient to point out the correspondence 
 between Finnish ytsi, one, andkaksi, two. 
 
 Esthoni'an iits, one, and kats, two. 
 
 Lapponian akt, one, and kvekt, two. 
 
 Lapponian oft, one, and guoft, two, Mordvin, kavto. 
 
 Vogulian ok, one, and Syrian, kik, two. 
 
 Tcherem. iktat,one, and k ok tat, two. 
 
 Ostiak. ot, one, kat, two, Vogulian, kit, Magyar ket. 
 The Turkish also, which has simply iki, for two, shows traces of an 
 original initial guttural, which, as in many other words, was dropped 
 in the progress of this language. The Turkish word for twenty, there- 
 fore, is not only igirmi, butyigirmi; and y i g i r being afterwards 
 contracted into jir and sir, explains the Tchuwashian sirim, and 
 Yakute svirba, twenty. 
 
 These forms have no analogies south of the Himalaya. The only 
 approach to the Tamulic radical for two, which, in contra-distinction 
 of on, the term for one, seems to have been er, may be discovered in 
 Mongolic and Tungusic dialects. The Mongolic has the initial k, and 
 it forms its words for two, as kuyar, and contracts it into kur, in 
 kur-in, twenty. This kur exists in Tungusic as j ur ; in Mandshu 
 as j ue, 2. In the Mandshu or- in, twenty, the initial k has been lost 
 altogether, as before in the Turkish igir-mi. An inter-comparison, 
 however, of the" Mongolic and Ugric words for two shows that the 
 characteristic and significative power lies in this k, while yar, in 
 Mongolic, and kta, in the Tchudic numerals, are secondary elements. 
 This is stiU further confirmed by a reference to their terms for twenty, 
 when, as in Syrianian ky-f, Mordvinian ko-ms, Vogulian ku-s, 
 Ostiakian chu-s, and Hungarian hu-sz, the simple guttural ex- 
 presses the value of two. In Tcheremissian kok-lu, the full word for 
 two has been employed ; and the same applies to the Tungusic 
 jur-men, the Mongolic chur-in, and the Turkish yigir-mi. 
 
189 
 
 The words for three which had preserved so many traces of a 
 common origin in the Chinese and Bhotiya languages allow of hardly 
 any inter-comparison, when we look to the Tamulic and Ural-Altaic 
 branches. In the latter, the primitive base of three might be repre- 
 sented as KK, with a tendency to add a final labial b or m. With 
 this base we can explain the Mongolic gur-b-an, the Magyar har- 
 om and charm, the Vogulian kor-om, and the Ostiakian kol-ym. 
 Again, the Lapponian, Esthonian, and Finnish k.olm. The Syrianian 
 kuyim leads naturally to the Tcheremissian kum. Professor Schott 
 connects Tataric forms like ol-tuf, thirty, with the Ugric kol, 
 appealing to the frequent loss of an initial guttural in Tataric. The 
 Tungusic el-an also would thus be accounted for. The final 1, 
 which corresponds to an r, and which in Ostiakian is represented by d 
 (chudem) and dl (kodlem and cholym), may become a palatal sound ; 
 because 1, taking a mouille pronunciation, has the same influence on 
 a preceding t asiorj in "nation," and in this manner ol or odl (ori- 
 ginally kodl or kor) may emerge again as the Tataric uch and Uts. 
 Thus the Tungusic el-an, three, and got -in, thirty, would descend 
 from the same root, as well as the Mongolic gur-ban, three, and 
 g uch -in, thirty. G uch would explain the Tataric uch, three; and 
 Tataric ol-tuz, thirty, would receive its solution from the Tungusic 
 el-an, three. 
 
 Although, we have seen, before, that an initial k, before it is lost 
 altogether, may take the sound of ch, j, sh, and s, and although the 
 Tcheremissian k u m, three, has been traced back to KR-M, it would 
 hardly be possible to take our stand on these secondary forms, and 
 to compare them with the Bhotiya base, SAM. The Tamulic words 
 also for " three " must be left unexplained, in the present state of our 
 knowledge, as the phonetic changes which are sanctioned by these 
 languages have not hitherto been explored with sufficient accuracy. 
 
 We must now compare the numerals from four to seven, which 
 alone can be considered as the common property of the Ugric and 
 Turanic races. Before their first separation these races did not count 
 beyond seven ; and it is, therefore, one of the most characteristic 
 features of the two classes of the Turanian family, that their words for 
 eight and nine are compounds, expressing 10 — 2, and 10 — 1, like 
 
 O 
 
190 
 
 the Latin duodeviginti and undeviginti. Some tribes of the Bodo 
 never count beyond seven at the present day.* 
 
 The simplest form for "four" in the northern division of the Tu- 
 ranian family is found in the Tcheremissian nil. The base from 
 which all other words for " four " were derived may, indeed, be repre- 
 sented by N I L or N A L. This explains the Mordvinian nile, and 
 nilen, the Vogulian nila, and Ostiakian ni'il. The final I of nil 
 was liable to a mouille pronunciation, which is naturally expressed 
 by the palatal semivowel y. This explains the Finnish nelja,the 
 Lapponian nielj, the Syrianian njolj. Now, we saw before that 
 a final I, particularly one that is liable to this palatal softening, is 
 interchangeable with dl and d (as kodlem, chudem, and kolym); and 
 this must account for the Ostiakian njedla, and njeda. The 
 Hungarian negy is pronounced nedj, and this, therefore, merely a 
 phonetic variety of n j e d 1. 
 
 The coincidences between these and the Tamulic words for " four " 
 need no explanation. Tamil, Malabar, Gond, and Tuluva, have 
 simply the same word, nalu ; and the Canarese nalku and Telugu 
 nalugu are less violent deviations than the Hungarian negy. 
 
 Professor Schott goes even beyond this, and considers the Mon- 
 golic, Tungusic, and Tataric words for four as derived from the 
 same radical. Supposing this radical to have been nalk, he allows 
 a transition of n into d (as in Sanskrit navan, Lithuanian devyni). 
 He then explains the Tungusic diig-iin (Mandshu, duin), four, as a 
 variety of diilg-iin; and, by substituting different final letters, he 
 arrives even at the Turkish diirt, four, and the Mongolic diirb-en. 
 The transition of a final Ij into the palatal ch being established before, 
 he likewise explains the Mongolic diichindiich-in, forty, as analo- 
 gous with Turkish uch, three, instead of 'ulj. These combinations 
 must rest on the authority of one who is, no doubt, better acquainted 
 with the possible changes of Turanian words than any scholar in 
 Europe. 
 
 " Five," if reduced to its radical elements in the Northern or Ural- 
 Altaic division, is VI T. This coincides with the Lapponian vit ; 
 and the Syrianian vitj, Mordvinian v ate, Ostiakian vet, are easily 
 
 • Journal of the Asiatic Society of Bengal, 1849, p. 720. 
 
191 
 
 traced to the same source. The loss of an initial v requires no expla- 
 nation in Arian or Turanian dialects. Hence Hungarian ot, Vogulian 
 at, may be reconciled with the same root. A transition of t into s also 
 has occurred more than once, and is confirmed here by the Finnish 
 viisi, Esthonian wiis, and Tcheremissian vis. As in the mouth of a 
 Basque, vivere is bibere, the Turkish besh also may enter into the 
 same category. A Turkish sh is represented in Tchuvashian by 1 ; 
 hence pil-ik also has probably passed through the forms of vit, 
 bit, vis, and besh. 
 
 In the Tamulic class I consider the Canarese and Telugu ayidu as 
 mere amplifications of ed, a form not far distant of the Ostiakian vet 
 and uet. The Tuluva ayinu, Tamil anju, Malabar inthu do not 
 differ so much as to warrant the admission of a different radical. 
 
 " Six " was expressed, according to Professor Schott, by a modifi- 
 cation of "three." Analogies exist in the Japan numerals mitsu, 
 three, and mutsu, six; and, again, fitotsu, one, futatsu, two; 
 jotsu, four, and jatsu, eight. Now, as the radical of three was 
 KE or KL, changeable into KD, KDI, and KDj, Professor Schott 
 maintains that this was raised to six by the addition of a final t, 
 which t absorbs, in most cases, the final semivowel L or R, of the 
 radical KR. In this manner he derives Ostiakian kut, Vo- 
 gulian and Lappon. kot, Tcheremissian kut, Mordvinian koto, from- 
 a presupposed kurt or kutt, and by a transition of this final t into s, 
 he accounts for Finnish kuusi, and Esthonian kuus. More diflScult 
 is the Turkish alty. As, however, in uch, three, the Turkish alone 
 had sacrificed the initial k, we are justified in allowing the same pro- 
 cess in what is only a modification of the radical three. We should then 
 arrive at AR or AL, and the additional t of the six would give us the 
 Turkish alty. If this last process is admitted, it need hardly be 
 pointed out that an opening is gained for the Tamulic forms, which 
 all point to A R as their common source. 
 
 " Seven," in its most abstract form, might be rendered by SAT. 
 With this the Vogulian sat is identical; the Yakute sett a nearly so; 
 and the assibilation of the final t would account for Mordvinian and 
 Syrianian si s-im, Esthonian seitse, Finnish seitse-ma. That a 
 final s may be pronounced like a palatal, we saw before, and by this 
 change we arrive at the Tchuvashian sichche, but we require the 
 
 o 2 
 
192 
 
 same admission for an initial s, in order to explain the Lapponian 
 Chech. The initial letter alone has become palatal in the Turkish 
 yedi, and in the Hungarian there remains but an initial h in het, 
 seven. 
 
 Allowing the total loss of this h, we may compare the Tamil 
 ezhu, and Telugu edu, of which elu in Canarese and Malabar, and 
 al in Tuluva, are natural varieties. 
 
 The admission of SAT as a radical for seven, does not exclude 
 the possibility that this SAT may be itself but a secondary form. 
 For, although SAT suffices to explain most of the Turanian 
 numerals, it does not explain such forms as the Ostiakian sabet 
 (tabet, tlabet); and !frofessor Schott points out that, as in the 
 Arian family septem has taken a secondary form set, which 
 would suffice as the radical of French sept (pronounced set) Italian 
 sette, Spanish siete, SAT also may be but a secondary radical as 
 compared with SAB AT. This would be a most extraordinary dis- 
 covery, for it would actually restore the word for seven to so primi- 
 tive a state, that not only the Turanian, but the Arian and Semitic 
 languages might, in this case, be traced back to the very cradle of 
 human speech. 
 
 Eight intlgricis expressed by 1 — 2; nine by 10 — 1. The Syria- 
 nian kokjaamys, 8, is derived, according to Sjogren, from kok, 2, 
 and jaamys, the elative of j a am. It means two taken out often. 
 Although j aam, 10, in Syrianian is now represented by the (Russian?) 
 das, it has been preserved in the Ostiakian jon.* Okmys, 9, 
 according to Castren, is derived from otik and kym (10). The 
 elative of kym would be kymys, which, together with otik, is con- 
 tracted into okmys. In the same manner Sjogren derives the Finnish 
 kahdeksan, 8, and yhde-ksan, 9, from kahde, 2, andyhde, 1, fol- 
 lowed by ksan, which again is explained as the Ablative in san, of 
 kym, lO.f I do not hesitate, therefore, to propose the same ex- 
 planation for the Tamulic words for 8 and 9. 
 
 • Another Finnic scholar, Dr. Europseus, derives kokjaamys from kjam, 
 a variety of the modern kamen or kjemen, 10 ; kammen, in Finnic, meaning 
 "hand." See Schott, p. 27. 
 
 t Professor Schott's derivation will be given hereafter. Dr. Europffius divides 
 kah-deksan, yh-deksan, and endeavours to establish deksan as one of the most 
 primitive words for "finger "and ten. 
 
193 
 
 Two in Canarese is er|adu ; ten hat-tu ; eight entu. 
 Two in Tamil is ir|andu; ten pat-ta; eight etu. 
 One in Canarese is on|du; ten hat-tu ; nine ombhattu. 
 One in Malabar ison|du; ten pat-thu; nine on-pathu. 
 
 The euphonic laws of the Tamulic languages have been too little 
 explored to enable us to explain the contractions which have taken 
 place in these compounds. But that they are compounds, and com- 
 pounds formed on the same principles as those in Ugric, is palpable. 
 Even if the elisions are without analogy, it would still be possible to 
 go back for an explanation of these words to an earlier state of lan- 
 guage., in which one and two were on and ar, and in which ten was 
 tu instead of hat-tu, pat-tu, ba-da, &c. Indeed, I believe that 
 wherever 8 and 9 have an identical element in their names, and 
 where this element has any similarity with the names of ten, we 
 may safely apply the same principle of formation which Sj ogren and 
 Castren have established for the Ugric. In Mandshu, for instance, 
 we find juan for 10, the same root we met before in Ugric. Now, 
 jue in Mandshu is 2, and jakon is 8 ; emu is 1, and onyan is 9. 
 
 Humboldt discovered a similar process for expressing eight and 
 nine in the Malay languages. 
 
 Professor Schott has treated this question in the most comprehen- . 
 sive manner in an Essay " On the Numerals in the Tschudic Class of 
 Languages.'' I received one copy of it in time to avail myself of his 
 suggestions while my own Essay was partly in print j and I subjoin 
 the following abstract, containing all the evidences that can be 
 brought to bear on this interesting feature of the Numerals of the 
 Turanian family. 
 
 The first root for Ten, in the Tchudic languages, is T-S, or D-S. 
 It occurs in the Syrianian DAS, 10; SIZIM-DAS, 70; KOK- 
 JAMYS-DAS, 80; OKMYS-DAS, 90 ; and in the Hungarian 
 TIZ, 10. The same root, only contracted, appears in the Hunga- 
 rian HAR-MIN-CZ, 30, instead of HARMIN-TIZ; and in 
 H U-S Z, instead of H U - T I Z, 20. The Ostiakian C H U S, 20, Vogu- 
 lian HU-S, and Syrianian KY-S, 20, are too like the Hungarian 
 HU-SZ to admit of a different etymology. 
 
 The Turks used the same root, in "thirty," which is OL-TUZ, 
 and OTUZ; in Yakut OTUT; in Tchuvash VU-TUR. 
 
 o 3 
 
194 
 
 The Hungarian NY OL-CZ, 8, and KILEN-CZ, show C Z= 
 TIZ as the root for ten. N YOL was originally a name of "two; " 
 but the root N Y A, well known in Chinese and Bhotiya dialects, as 
 the exponent of two, was used by other Turanian tribes as a Dual 
 to express four. Thus it became fixed as " four " in the Altaic (and 
 Tamulic) languages, while as "two" it was replaced by new words. 
 
 In the Ostiakian dialects, for instance, "four" is expressed by 
 NJETLA, NJEDLA, NJETA, NJEDA, NJET, and 
 NJEL. 
 
 Eight is expressed by NJIGEDL A -CH, NIDA, NIT, and 
 NJIL. 
 
 In NJIGEDLA-CH, the CH must be taken as the exponent of 
 ten, probably an abbreviation of the Finnish ksan, used for the same 
 purpose. In the other forms this final ch has been lost, as its etymo- 
 logical importance ceased to be felt. 
 
 Prof. Schott admits the possibility that the two roots for ten, T-S, 
 and KSAN, were originally identical. He traces ksan as ten in 
 
 Finnish, kahde-ksan, 8, and yhde-ksan, 9. 
 Mordvinian, kav-kso, 8, and vaj-kse, 9. 
 Tcheremiss. kanda-chse, 8, and ende-chse, 9. 
 Lapponian, kak-tse, 8, and ak-tse, 9. 
 
 The original form might have been T S A N, interchangeable with 
 KSAN, which Prof. Schott considers as a full root for ten, while 
 Sjogren takes KSAN as an ablative in san from kym, ten. 
 
 What is important is the establishment of N YA in the Ural- 
 Altaic languages with the meaning of " two," a meaning which no 
 doubt it had previous to that of "four" (a dual of two, like ash^au 
 in Sanskrit, eight, a dual of four). NY A lost this signification of two 
 afterwards altogether, in the Ural-Altaic branch, but it must still 
 have possessed it at the time when these Ural-Altaic dialects formed 
 their words for eight and nine. Other traces of n i in the sense of 
 two, are the Ostiakian NIT SOT, which means eighty, i.e., 
 20 — 100; and also eight hundred, i.e. 8 x 100; nit being, in the latter 
 case, the usual word for eight, a corruption of NJIGEDLA-CH. 
 In Vogulian NJOL-SAT is eighty, i.e. 20—100. In Mongo- 
 lian eight is NAIM AN, i. e. 2 — 10; and even the Tungusian six, 
 
195 
 
 NJUGUN is explained as 2x3. In Vogulian NJALA-LU is 
 eight, i. e. 10—2. 
 
 To return to the Hungarian KILEN-CZ. This is explained 
 by Prof. Schott as a composition of cz, ten, andkilen, an adverb, 
 meaning " without," or "minus." The "one" which ought to have 
 been added has been dropped, as in the Ostiakian, where, in the 
 dialect below Surgut, nine is expressed byiirchjeung, while above 
 Surgut it is'ej erek jong, one without ten. 
 
 In Turkish we saw the root T-S used before in o-tuz, thirty. 
 Prof. Schott detects the same root, which has been identified with 
 K-S, in the Turkish SE-KIZ, eight, and D 0-KUZ, nine. In both 
 words, however, he supposes that K has been lost at the end of the 
 words of one and two ; for, according to his statement, se in sekiz 
 stands for jak, the Mandshu name for two as preserved in j a k-on, 
 eight, i.e. 10 — 2, while do is traced back to tok, and this to okt, 
 one. 
 
 Another root for 'ten,' is found in the Lapponian LOKKE, and 
 LOGE; which in Vogulian became LAG A and LAVA; finally 
 L6U, and in Tcherem. LU. From this we have the Vogulian 
 njala-lu, 8, i. e. 10 — 2; and anta-lu, 9, i.e. 10 — 1; one, anta, 
 being the same as the Tcheremissian ende in ende-chse, 10 — 1 
 =9. In the Tcheremissian, LU occurs in Kum-lu, 30 ; in Kok-la, 
 20 ; in viz-lu and viz-le, SO. In Turkish the same root is traced in 
 el-li, fifty, and allig ; where al would be an assimilated form of at, 
 Vogulian at, five. 
 
 A third root for " ten " is one of a very pliant nature if we 
 accept Professor Schott's identifications. The Turkish N, the 
 Ostiakian AN, and their derivatives; the Mongolian AN, in dal- 
 an, 70, yer-en, 90; the Mandshu IN, in orin, 20, and the mere 
 I in dech-i, 40, are all traced back to this root. The same root is 
 pointed out in Mongolian j is-un, i. e. 10—1 ; and Tungusian j ag-in, 
 
 9, i.e. 10 1. Likewise in the Tungusian word for eight, jak-un, 
 
 i, e. 10—2. 
 
 In Tchuvashian, ten is VONNA, and VAN. The same root 
 is found again in Hungarian hat-van, 60, andhet-ven, 70; both 
 
 varieties of the same word. 
 
 o 4 
 
196 
 
 The Ostiakian JANG and JONG, and the Samoiedian JU are 
 likewise referred to this radical. 
 
 TheTungusicMEN, 10, and MEE, ingur-mer, 20; theVogulian 
 MEN, in nali-men, 40; the Mongolian MAN, in nai-man, 8, i. e. 
 10 — 2; the Vogulian PEN in at-pen, 50; the Syrianian MYN in 
 nelja-min, 40; and MYZ in ko-myz, 30; the Turkish MYSH in 
 alt-mysh, 60; the Syrianian MYS, in ja-mys, 8, i. e. 10—2, and 
 ok-mys, 9, i. e. 10 — 1 ; and finally the Tchuvash MIL J. in sit milj, 
 70, are also brought under the same category. 
 
 MEN is again supposed to have been changed into MA in Turkish 
 jer-ma, 20; Yakut, siir-ma; Tchuvash, sir-im; and Osmanli 
 yigir-mi. 
 
 A new change takes place in the Tungusic dialects. Here we find 
 this root for ten, as J AN and JUAN, as JAR and JU. Thus j u r- 
 jar, 20, would stand instead of jur-men. 
 
 A fourth and fifth root for ten are added by Prof. Schott ; the one 
 being the Mordvinian KAMEN (gamen, kam and ka) ; Esthonian 
 KUMME; Finnish KYMMENE; the other the Mongolio 
 ARBAN. 
 
 If in the Ural-Altaic brancli " ten " is expressed by five diflferent 
 roots, we need not wonder that the Tamulic branch also has fixed 
 upon its own root for ten, which is PAT. 
 
 The pronouns of the Ugric and Tamulian languages show but 
 faint traces of relationship. The characteristic letters of the personal 
 pronouns in the Ugric branch are M, T, S, for the three persons : 
 identical with those of the Arian languages. Whether this coin- 
 cidence between the Ugric and Arian pronouns should be considered 
 as the result of primary connection, or as a mere phonetic accident, 
 depends on the view which we entertain of the origin of language 
 in general. Certain it is that the coincidence between the Lapponic 
 pronouns 
 
 Mon, Todn, Sodn, 
 
 and the Swedish M in. Din, Sin, can no longer be explained by 
 supposing that the Lapps borrowed these pronouns from their neigh- 
 bours, the Swedes * ; for the same pronominal bases exist in Ugric 
 
 * Gyarmathi (p. 17) considers this Swedish origin of the Lapponian pronouns as 
 
197 
 
 dialects, which have never been in contact with Swedish. Besides, 
 supposing for a moment that pronouns could thus be imported, no 
 one would lightly admit that the terminations of the verbs also had 
 been transferred from the same source; and that the Tcheremissians, 
 for instance, had not distinguished the three persons of the verb, 
 lodajw, lod&f, lodas, I read, thou readest, he reads, until they 
 received the materials for these verbal forms from Teutonic sources. 
 I believe that the similarity of the pronouns in Ugric and Sanskrit 
 has an historical meaning, but that its explanation must be sought in 
 earlier times than the Finnic migration toward the north of Europe. 
 How early some of the Turanian pronouns began to lose their pri- 
 mitive character may be seen in the Scythic Inscriptions at Behistun, 
 so ably decyphered by "Westergaard and Norris. In one of the last 
 numbers of the Journal of the Asiatic Society, the latter ingenious 
 and patient scholar gives us the following pronouns as the result of 
 his investigations. 
 
 Hu (I), Ni (thou), Tufri (he); Niku (we), — (you), Appi 
 (they). Here the pronoun of the first person shows the same base 
 as the Arian aham, ego, Guzerati. hun. M, as the exponent of the 
 first person, shows itself in the possessive mi, mini; thus bearing 
 witness to the existence of the two bases of the pronoun of the first 
 person, which we find in all the Arian and in some Turanian dialects. 
 As in the Arian dialects the guttural base properly belongs to the 
 casus rectus, the labial base to the casus obliquus; we find in Scythic, 
 also, Hu for the nominative, and Mi for the possessive. 
 
 But although the pronouns in different branches of the Turanian 
 family have diverged so much from their original type as to render 
 a phonetic restitution of the Ugric and Tamulic pronouns extremely 
 hazardous, we may yet point out as a coincidence the absence in both 
 
 a fact. He says, " verum equidem est, quod Pronomina personalia Lapponum men, 
 tod n, sod n, a Suecorum min,din, sin, descendisse videntur." Castren (De AflSxis, 
 p. 63) admits the same supposition, not indeed for tlie first and second person, but 
 for the Finnic pronoun of the third, han. I give his own words : " Quod denique 
 ad tertias persona: pronomen han attinet, tanta est ejus cum prisco septentrionali 
 pronomine hann, Svetice han, similitude, ut videatur nohis Sjogren summo jure 
 originem peregrinam ei trlbuere." The same grammarian goes still further, and 
 derives the termination of the Syrianian passive sya from the Russian Cjli for 
 instance, ystysya, lam sent. 
 
198 
 
 of the relative pronoun. The Tamulic laoguages have no relative 
 pronouns, and in Turkish the relative pronoun is evidently borrowed 
 from Persian. 
 
 There are other parts of grammar, however, which offer more 
 positive evidence, and have preserved a common type with so much 
 tenacity, that, although the people who speak these dialects are sepa- 
 rated by the whole continent of Asia, we can still discover that they 
 once resided in close proximity, and received the first impressions of 
 their grammatical system, as it were, in the same school. 
 
 As it would be impossible to go here through all the chapters of 
 the Tamulic and Ugric grammar, and as there are many points in 
 each of later growth and secondary importance, I shall only dwell on 
 those features which have been pointed out by Tamulic scholars as 
 essential in Tamulic grammar, and shall endeavour to show their 
 equivalents in the different dialects of the Ugric and Tataric speech. 
 I take the characteristic features of Tamulic grammar as collected by 
 Ellis in his Introduction to Campbell's Teloogoo Grammar, and from 
 Weigle's excellent sketch of Canarese grammar. 
 
 I. " Eoots in Canarese,'' as Weigle says, " are monosyllabic, bisyl- 
 labic, and trisyllabic. The latter can generally be reduced to a more 
 simple form." 
 
 " Ugric as well as Tataric roots are originally monosyllabic, but 
 bisyllabic and trisyllabic exist, which generally, but not always, can 
 be reduced to a monosyllabic form." SeeBoethlingk,Yakute Grammar, 
 § 442.; Gastrin, Ostiake Grammar, § 96.; Tsherem. Grammar, § 8. 
 " Ceterum voces polysyUabae a primitivis bisyllabis plerumque de- 
 rivantur." 
 
 II. " Some Tamulic roots are also used as nouns, or become nouns 
 by slight modifications ; or, as Rhenius expresses it, verbal forms may 
 be declined, and nouns be conjugated in Tamil." The same feature in 
 the Ugric languages has been discussed before, page 296. ; see also 
 Boehtlingk, Yak. Gr. §§ 235., 339., and note 71. 
 
 III. In order to avoid ambiguity, different dialects sanction either 
 the verbal or the nominal character of a root. Thus, it frequently 
 happens that in one dialect a root is verbal only, in another nominal 
 only. 
 
 "In Tamil accarei occurs only as a substantive; for instance, 
 
199 
 
 yenac accareiyillei, it is not a want to me, i.e. I do not want. 
 In Canarese accariy is common only as the root of a verb; acca- 
 riyadu, to be desired." (See Ellis, Introduction.) The same habit, 
 with regard to Ugric and Tataric dialects, has been discussed before, 
 page 303., and by Boehtlingk, in his Yakute Grammar. 
 
 IV. " Particles in the Tamulic languages show more or less clearly 
 their origin from simple nouns." — "The postpositions of the Ugric 
 languages do not constitute a separate part of speech, for with few 
 exceptions they are real nouns. Adverbs, like postpositions, are 
 derived from nouns by d^erent inflections." See Gastrin, Ostiake 
 Grammar, §§ 127. 129., Yakute Grammar, § 402. 
 
 V. " Compound nouns are comparatively scarce in the Tamulic 
 branch; they occur in the more ancient dialects as imitations of 
 Sanskrit compounds." (Weigle.) "In the Ural-Altaic languages 
 the scantiness of compound words has led several scholars to deny 
 even the possibility of real composition in this family of languages. 
 This point has been discussed, and particularly with regard to the 
 Finnic languages, by Boehtlingk, p. xxxi. Kellgren, p. 31. The power 
 of forming compound words, though not used extensively, exists, 
 however, both in Tamulic and Ugric. For instance, Syrianian 
 ydzyd-tos'a, longa barba ornatus; ydzyd-koka, longis pedibus 
 pr^ditus ; kos-soja, sicca manu. — Canarese, davare gawwu, 
 lotos-eyed. In Syrianian Castren speaks of " mcfny " compound words. 
 § 42. Gr. Syrian. 
 
 VI. " Canarese adjectives may either be placed before the nouns 
 which they determine ; in this case they have no inflections : or if they 
 are used as substantives, they are joined with the pronoun of the third 
 person, and then declined in a manner which reminds us of the 
 strong declension in German." (Weigle.) "Adjectiva Tsheremissa 
 declinari quidem possunt similiter atque substantiva ; quum vero 
 attributa substantivorum sunt, non declinantur. Ex. j azo, magnus ; 
 jazovyla, magni ; jazo edemvyla, magni homines." See Gr. 
 Tsherem. § 15. ; Gr. Syrian. § 73. 
 
 VII. The Tamulic languages have no distinct forms to express the 
 comparative and superlative. The same deficiency exists in the lower 
 branches of the Ural-Altaic family, but has been remedied in its more 
 developed members. In Yakut the absence of the degrees of com- 
 
200 
 
 parison is quoted by Boehtlingk as a " logical characteristic of this 
 primitive Turkic idiom." 
 
 VIII. Gender in the Tamulic languages is distinguished only by 
 means of pronouns ; and that only in the third person. The third person 
 of the verb, being formed by pronominal affixes, has three forms to 
 distinguish the three genders. Adjectives are not subject to any 
 change to denote the incidents of gender, number, or case; nor are 
 the distinctions of gender denoted in primitive nouns by any distinct 
 forms of termination. The pronouns therefore vary, not according 
 to the grammatical gender of the nounq^to which they refer, but 
 according to the natural sex of the objects expressed by the noun. 
 The Tamulic languages admit a "sublime gender" and an "inferior 
 gender." All rational beings belong to the former class ; while the 
 latter comprises the whole of the irrational creation, whether animate 
 or inanimate. For the singular the sublime gender is subdivided into 
 masculine and feminine. 
 
 The Finnic languages have not even these remnants of gram- 
 matical gender. The pronoun of the third person is the same, whether 
 applied to male, female, or inanimate subjects ; so is the third person 
 of the verb. " Omnes omnino linguae Finnicse originis carent genere." 
 The difference, therefore, between the Ugric and Tamulic languages 
 is only this, that the latter have three pronouns of the third person, 
 while the Ugric have but one. In other respects grammatical gender 
 is ignored by both. 
 
 IX. The plural in Canarese is expressed by the termination ar, 
 whether the noun implies a male or a female object. In Gond the 
 plural is formed by nk ; in Telugu by lu (ru) ; in Brahvi by k and t. 
 The termination gal, which is used for nouns expressive of inanimate 
 objects, has been called a neuter termination ; but in reality it is 
 only a secondary affix, expressive of abstract totality. Dr. Stevenson 
 considers the Tamil gaZ, Canarese gaZu, and Telugu lu, to be ab- 
 breviations of the Sanskrit sakala, which in Tamil becomes sagala, 
 in Marathi sagaZe. 
 
 The old Ugric termination of the plural is as, or, if we consider 
 a merely as a connecting vowel, s. This exists in the Syrianian j as 
 (as). In Lapponian the s becomes h, in Finnish t, which exists in the 
 Ostiakian et. According to Castren, the original form of the plural 
 
201 
 
 was t. This is changed in Hungarian and Lapponic into k, in Kamas- 
 sian intoje, san, san; in Samo. Ostiakian into la. In other Samoiedic 
 dialects it is elided, or leaves only a final aspiration. This simple 
 termination has frequently been replaced by secondary forms, such as 
 the Tsheremissian vyla. These plural terminations in the Tataric, 
 Mongolic, and Tungusic languages are, as Castren says, " proprije in- 
 dolis et recentioris ut videtur originis." In Turkish the original sign 
 of the plural was t or s ; this s in Osmanli became k; but ler is 
 now used as a secondary formation of the plural in Turkish. 
 
 While the Tamulic had retained the distinction of sex in the pro- 
 noun of the third person, which the Ugric has lost, the Ugric in some 
 of its dialects (Samoiedian, Lapponian, and Ostiakian) has preserved 
 traces of a dual, which has disappeared in all Tamulic dialects. In 
 Ostiakian the termination of the dual is kan, xan, gan ; in Yur. ha', 
 g', k' ; in Taw. and Kamass. gai ; in Samoied-Ostiakian ga, ka. In 
 the Irtishian dialects of the Ostiakian, in Lapponian and Kamassian, 
 however, nouns and adjectives have lost the dual, and pronouns and 
 verbs only have retained it. In the Samoied-Ostiakian, it is the pro- 
 nouns that have lost the dual. Gastrin derives the termination of the 
 dual from a particle ka or ki, which means also ; as in weliki, fra- 
 terque. 
 
 X. In Canarese there is a third termination of the plural andir. 
 This is used only after nouns which express relationship. Weigle sup- 
 poses that it was originally an honorific particle, though he admits that 
 " this cannot be proved." Gyarmathi writes, " Habent autem tarn 
 Hunirari quam Lappones prseter plurj^lem hunc alium adhuc numerum 
 pluralem, qui non in omnibus observatur vocibus, sed tantum in 
 nominibus cognationis (L. tyah; H. mek). Significat vero is, non 
 personas pluralitatem, sed consortium aut sodalitium cum ilia persona 
 junctum. Duplicem hunc pluralem, Hungari possessivis tantum 
 nominibus tribuunt, Lappones vero nominibus cognationis simplici- 
 bus." The nature of this Lapponic plural will perhaps serve to 
 explain the original meaning of the Canarese andir. 
 
 XL In the Tamulic as well as the Ugric languages, the declension 
 of the plural is the same as in the singular. The same terminations 
 which in the singular are added to the base, are in the plural added 
 to the base after it has received the nota pluralis. A Turkish noun, 
 
202 
 
 after it has taken ler as the exponent of plurality, is considered as 
 a singular so far as case terminations are concerned (ler, ler | in, 
 ler|e, ler|i, ler|den.) The same in Hungarian. After k has been 
 added to the base of the noun, no further distinction is made between 
 the cases of plural and singular. This is a great advantage in Tura- 
 nian grammar, if compared with the Arian system of declension. 
 The same simplicity and lucidity distinguish the Tamulic declension 
 where, after gal is adiled, the plural is the same as the singular. 
 The same system has been imitated by the Bengali and other Sans- 
 kritic dialects. The sign of the plural in Bengali (dig') has been 
 explained by Dr. Stevenson as an abbreviation of the Sanskrit a d i k a 
 (adi); a derivation which, though not yet confirmed by historical 
 evidence, is much more probable than one proposed by myself in a 
 former essay. In Asamese the signs of the plural are bilak, hont, 
 and bur. The only irregularities which occur apply to the nomin- 
 ative, where in some dialects the old plural in r or n is maintained. 
 Occasionally, also, the contact of the terminations with the sign of 
 the plural gives rise to phonetic changes. 
 
 XII. But, while nothing can be more regular and intelligible than 
 this Turanian process of distinguishing the plural from the singular, 
 plurals occur particularly in the pronouns which seem not to be 
 formed by external addition, but (to adopt a favourite expression of 
 Arian grammarians) to have been produced by some unknown pro- 
 cess from the body of the noun. This applies particularly to the pro- 
 noun of the first and second person. The change of Hungarian me (I), 
 te (thou), into mi (we), ti (you) ; of Syrianian, me, te, into mi, ti; 
 of Mordvin. mon, ton, son, intomin, tin, sin; of Lapponian mon, 
 ton, son, into mi, ti, si; of Finnish mina, sin a, ban, into me, te, 
 he; of Tsheremissian min, tin, into ma, ta; and of several other 
 languages, which may be seen in the comparative table of pronouns, is 
 certainly not based on agglutinative principles. Whether we have a 
 right to assume that these forms were therefore produced by an 
 internal revolution, an idea of which no clear conception can be 
 formed, remains to be proved. But if such changes as Sanskrit yas, 
 and Greek og, becoming in the plural ye, o'l, are considered pecu- 
 liarly Arian, the above-mentioned Turanian forms will serve to 
 show that they are not so. And it should be remembered that similar 
 
203 
 
 forms exist even in the lowest and least developed of the Turanian 
 languages, as, for instance, in the Ta'i. The Kassia pronouns, nga, 
 I, p ha, thou, become ngi, we, and phi, you. 
 
 In the Tamulic languages the plural of pronouns exhibits the same 
 exception. The Canarese nan, nin, tan (I, thou, himself), form their 
 plurals not by an additional ar, but as navu, nivu, tavu. In old 
 Canarese the plurals are nam and tarn, while the plural of the 
 second person is formed by means of the usual plural sign r; nir, 
 you. In several cases it is clear that the Turanian languages used 
 a different base in the plural from that used in the singular. This 
 is intelligible ; but about the process which raised nga into ngi, or 
 m e into m i, we know as little as about the growth of the Sanskrit 
 yas into ye. Whether we explain the change of ya into ye by an 
 additional i (ai = e"), or whether we look upon e as an evolution of a, 
 in either case we assume facts which we do not know, and never can 
 know, either by means of analogy or induction. But if afterwards 
 we base further conclusions on grounds so hypothetical, if we classify 
 languages according to what we thus assume, to have been their 
 principle of formation, we really are trying to stand on our own 
 shoulders, and lose entirely sight of the necessary limits of our 
 knowledge.* 
 
 XIII. It is owing to the influence of Sanskrit grammarians, as 
 Weigle says, that in early times the number of case terminations in the 
 Tamulic languages has been fixed at eight. Most of them are particles 
 attached to the noun and there is no doubt that the whole declension 
 could be reduced to one casus rectus and one casus obli quus. 
 The ancient dialects are richer in these case-particles, which express 
 more delicate shades of meaning, so that even a larger number of cases 
 might here be admitted than is usually found in grammars. It is more 
 practical, however, to consider these particles as separate syllables. 
 The same opinion is expressed by Dr. Stevenson. He writes, — " Twice 
 seven cases might easily be made out in the Dekhan dialects." 
 
 Exactly the same applies to the Ugric languages. I quote 
 Castren(Gra. Smyr. § 24.) : "Omnes omninolinguasFinnicse originis 
 
 • See some excellent remarks on a similar point in Boehtlingk's Yakute Gram- 
 mar, p. iii. 
 
204 
 
 varietate casuum abundant. Casibus non solum indicant actionem, 
 quae notio in lingua Syriaena inest Nominativo, Genitive, Dativo, 
 Accusative, Infinitive, Ablative II., Instructive, atque statum, 
 casibus Essiyo, Factivo, et Caritivo expressuna, sed etiam varias 
 loci relationes quse in aliis Unguis prsepositienibus reddi sclent, 
 at in Finnicis ipsaque Sjriaena casibus AUativo et Illativo, Adessivo 
 et Inessivo, Ablative I., Elative, Consecutive et Prosecutive." 
 
 The distinction which Dr. Stevenson tries to establish between a 
 post-position and the sign of a case, that the one is by itself sig- 
 nificant, while the other is not, is true in the abstract, but not always 
 in reality. Many post-positions in Tamulic and Ugric are no longer 
 intelligible as independent words, though they clearly have descended 
 from nominal or pronominal bases. 
 
 XIV. There are, however, some terminations in Tamulic as well 
 as in Ugric dialects, which, as they express the most general gram- 
 matical categories, have become fixed and technical. These, in either 
 branch, have a claim to a higher antiquity than other terminations or 
 affixes whose origin is more palpable. "With regard to these primitive 
 terminations, attempts have been made to identify the corresponding 
 forms in Ugric, Tataric, and Tamulic languages. Dr. Stevenson 
 compares — 
 
 (1.) The Tamil Accusative in ai, (Malay e) with the Turkish 
 Dative i. 
 
 (2.) The Tamil Dative ku, Canarese ge, Telugu ki, ku, ko, 
 Malayalim ka, with the Dhimal (Bhotiya) khe, the Tibetan gya, 
 the Tataric ga. 
 
 (3.) The Genitives (or Adjectives) in n, such as Canarese an a, in a, 
 Tamil in, Telugu ni, Gond na, with the Turkish in, Lappenian en, 
 Finnish n, Mordvinian en. In Tchuvashian we have from man, I, 
 manyng, mens, man-yng-yng, mei; again, man-yng-ki, meu's, 6 
 £^dc; man yng ki nyng, mei. 
 
 XV. The Ugric languages have two classes of pest-positions, simple 
 and compound. In Finnish, for instance, the simple Partitivus is 
 formed by ta, the lUativua by s. Both together form the Presecu- 
 tivus tse ; as karhu-ts6, passing along the bear. The same in Canarese, 
 we meet with compound cases, such as maneyellinda. Locative and 
 Instrumental, " from within the house." 
 
205 
 
 XVI. With regard to the personal pronouns, the admission of their 
 apparent difference in Tamulic and Ugric has already been made. 
 Instead of the characteristic letters m, t, s, we find n, n, t. The older 
 form of the Canarese pronoun of the first person, yan, instead of 
 nan (Malay alim, gnan,Uraon en), might indeed be reconciled with the 
 subjective base of the first person in some of the Ugric dialects ; and 
 the initial n of the second be derived from a t, as in Syrianian, 
 Ugro-Ostiakian and Samoiedic dialects the original t of the second 
 person has been supplanted by an n.* But as all intermediate links 
 are lost (except Uraon, asu, you), such comparisons would only 
 show the phonetic possibility, not the historical reality of the common 
 origin of the pronouns in Ugric and Tamulic. 
 
 XVn. In the Ugric and Tamulic languages the pronouns form 
 their plural by a modification of the base, not as in substantives, by 
 the addition of a suffix expressive of plurality. 
 
 Syranian : 
 
 Me, I, and Te, thou, Sya, he, become in the plural 
 Mi, we, and Ti, you, Nya, they, 
 
 while the common termination of the plural is j as. 
 In old Canarese, 
 
 Nan, I, Nin, Thou, Tan, ipse, become 
 Namf, We, (Nim, You), Tarn, ipsi. 
 
 XVIII. Besides the usual personal pronouns, most Turanian lan- 
 guages have produced a large number of polite or conversational 
 pronouns, such as " Servant," " Elder Brother," " Sister," " Block- 
 head," &c. Their number becomes smaller with the progress of 
 civilisation and literary culture. Hence but few traces of them 
 remain in the Tamulic, and hardly any in the Ugric branch. 
 
 XIX. The coincidences. between the numerals in the Tamulic and 
 the other branches of the Turanian family have been discussed before. 
 Besides the agreement in several radicals, it was shown there that 
 the Tamulic shared in the thoroughly Turanian feature that " seven," 
 
 * Cf. Castren, De affixis, p. 71 ; also p. 66. 
 
 t The modern plurals are, navu, nivu, tavu, showing the same transition of m 
 into V -which we find in the termination of the future, which is m in old, v in 
 modern Canarese. 
 
 P 
 
206 
 
 is the last common numeral, the words for "eight" and "nine" 
 being formed by means of substraction from ten (10-2, 10-1). 
 
 XX. "With regard to the verb, we have first to point out in 
 Tamulic the double system of personal terminations, one for the 
 present, the other for the past. The origin of these two classes of 
 terminations has been discussed in the first part of this letter, and 
 we need only add here, that in Tamulic also the shorter terminations 
 belong to the past, the fuller to the present. 
 
 XXI. The radical termination of the present in Tamulic, which is 
 p in old Canarese and Tulu, and utt in modern Canarese, kir in 
 Tamil, kindr in old Tamil, must most likely be considered as a 
 participial sufiSx, like the termination er of the .present in Turkish. 
 The coincidence between the Canarese utt, and the termination of 
 the present participle utt'a, is sufiicient to allow this hypothesis. 
 The termination of the preterite is actually the same in Turkish and 
 Canarese, d, for which in old Canarese we find i, the terminations of 
 the past participle in Canarese being likewise i and du. 
 
 XXII. The infinitive in Canarese was originally al or a 1 u, its modern 
 form ad or adu. The latter termination has been recognised by 
 Weigle as the pronoun of the third person, adu. In Syrianian the 
 participle is formed by ysj, ys being the pronoun of the third person 
 preserved in the possessive suffix ys, as purt-ys, his knife. Another 
 form of the infinitive is vawa or una, and this reminds us of the 
 Syrianian infinitive in yny. 
 
 XXIII. Canarese has no passive form, but expresses this form of 
 thought periphrastically. For instance, " he eats a beating," instead 
 of " he is beaten ;" " he falls a choosing," instead of " he is chosen." 
 Similar contrivances are known from Chinese, Tibetan, and other 
 languages which have not yet left the first stage of materialism in 
 their grammatical growth, and from others, like Bengali, which have 
 relapsed into that state after having passed through the highest 
 dev lopment of grammatical forms. In Chinese* they use kian, to 
 see ; for instance, pa 6, to protect ; kian pa 6, to be protected. 
 
 Another passive auxiliary in Chinese is p e i, to receive ; for instance, 
 k'ian-ts'e, to punish; pei tcao-ti'ng k'ian-ts'e, to be punished 
 by the Emperor, i. e. to receive Emperor-punishment. 
 * Endlicher, Chinese Grammar, § 230. 
 
207 
 
 A third root is k'i, to eat. For instance, t^,, to beat ; k'i th, to be 
 beaten. 
 
 In Kachari, a Bhotiya dialect, tbe passive voice is usually formed 
 by means of an auxiliary verb, signifying to be, to eat, to exist, added 
 to the root of the primary verb. Thus, from b u, strike, and j a, eat, 
 we have — 
 
 Present tense, A'ngbujadang, lam struck. 
 Imperfect, A'ngbujdbai, Iwas struck. 
 Perfect, A'ng buja dangman, I have been struck, or I 
 
 have eaten a beating. 
 
 In Bengali I remember to have met with similar expressions, khai, 
 to eat, being used as the auxiliary of the passive. But though I 
 cannot refer to a Bengali authority, a reference to the spoken dialects 
 of Germany would suffice to prove that languages, after producing 
 the most abundant grammatical organisation, fall back again upon 
 these simple and childish expressions. As in Chinese, we may say 
 in German, Schlage besehn, to see blows, Priigel kosten, to 
 taste a beating, in the sense of to be beaten. 
 
 In Syrianian no passive exists, except that, on the authority of 
 Castr&, we must admit a passive borrowed from Kussian. I subjoin 
 the ipsissima verba: "Passivi finis est — sja, l.-cja (Russ. CJI), qui 
 adjungitur secundae personas imperativi. Ut forma passiva e lingua 
 Eussica orta est, ita ssepe vi verbi reflexivi utitur, quae vis participio 
 semper inest. Quare passivum etiam per verbum auxiliare redditur." 
 
 This would show Syrianian at a great disadvantage if compared 
 with Tamulic dialects. Both were deprived of a passive, both were 
 brought in contact with languages, Sanskrit and Russian, possessing 
 a passive form. But while the Tamulic languages supplied their 
 deficiency by an ingenious application of their own resources, the 
 Syranian stooped to borrow a grammatical form from its more power- . 
 ful neighbour — a grammatical depravity almost without a parallel in 
 the whole history of human speech. Other Ugric languages possess a 
 passive. For in Mordvinian, although the participle is used in an 
 active and passive sense, the terminations van, vat, vi, have always 
 a passive power. 
 
 XXIV. The coincidence between the Tamulic and Tataric Ian- 
 
208 
 
 o-uages with regard to a negative conjugation has attracted the atten- 
 tion of several writers. 
 
 As to the Finnic dialects, which we have chosen as the most appro- 
 priate for the purpose of comparison with the Tamulic, they share in 
 the same grammatical feature: "Conjugatio negativa omnibus Fin- 
 nicis Unguis propria." (Castren, Grammat. Syriaena, § 66.) 
 
 The negative conjugation in Bengali and Mahratti is perhaps an 
 imitation of Tamulic, but formed in a different manner. 
 
 XXV. A causal form is produced in Canarese by appending isu 
 to verbal bases. In old Canarese this isu is represented byichu, 
 in Tamil by ka. The same derivative is employed to form deno- 
 minative verbs, and is of frequent occurrence at the end of foreign 
 words thus verbalised in Canarese. It then corresponds to the 
 termination ize in English, iren in German. 
 
 In the Ugrie and Turkic languages causal and denominative forms 
 are so frequent that they are mentioned as a characteristic feature of 
 this class of dialects. The suffixes, however, by which this modifi- 
 cation is expressed vary even in Ugric and Turkic. Causatives in 
 Finnish are formed by tan, in Lapponian by tam, in Syrianian by ta. 
 The Turkic dialects show a final r in tar and dar. Neither do 
 the terminations of verbs derived from nouns offer any coincidences, 
 and it is only the frequency of both these verbal forms which consti- 
 tutes a congruence between Tamulic and Ugric dialects. Literal 
 coincidences between the verbal derivatives used by the Turanians 
 North and South of the Himalaya, might indeed be pointed out, but 
 they would be of little weight unless the genesis of both could be 
 made out at the same time, thus establishing, not an accidental 
 similarity of sound, but a real identity of origin. Inchoative 
 verbs, which are a class of denominative verbs, are formed in the 
 Turkic branch by a final guttural. This might provoke a comparison 
 with the Tamil ka. But in the Turkic branch * this guttural can be 
 traced back to an original palatal vowel, while in Canarese no light has 
 yet been thrown on the analysis of this termination. The same 
 remark applies to the Hungarian derivative it, by which denomi- 
 native verbs are formed. 
 
 XXVI. The auxiliary verb "to be," in the Tamulic languages, has 
 
 * Cf. Boehtlingk, Yakut Grammar, § 493. 
 
209 
 
 likewise attracted attention by its great similiarity with Turkish. 
 There are two bases for this verb in Canarese, ir and u II. In Turkish, 
 one of the radicals of the auxiliary verb is ol, which is shared in 
 common by Turkic and Finnic dialects. It is the Syrianian viili, I 
 was; the Tcheremissian olam, I am. Its radical is originally a 
 pronominal base, and in the same manner the Ostiakian tajem, I 
 am, is derived from the pronominal root ta, that. 
 
 XXVII. Before we leave this comparison of the leading grammatical 
 features of the Tamulic and Ugric languages, it will be necessary to ex- 
 hibit at least a few traits of their syntactical similarity. The arrange- 
 ment of words and sentences might perhaps appear so entirely a matter 
 of individual choice and taste, that we could hardly expect coincidences 
 between nations who, so far as history and tradition can reach back, 
 have always been distinct in their language and nationality. Yet 
 there are no doubt laws, powerful as any in the realm of nature, 
 which make it impossible for certain languages to place their words 
 in the same succession as those of other dialects. No Semitic mind 
 can realize the idea of "ox-tail;" no Arian mind can break itself 
 into the conception of " tail-ox." The following will show how far 
 this influence extends, and how important an argument it is in favour 
 of or against the long-continued community of nations. 
 
 The syntactical characteristics of the Tamulic family are taken 
 from Ehenius * ; those of the Tataric languages from Schott.f 
 
 Tamulic. Tataric. 
 
 1. As to the position of the 1. Every word which determines, 
 parts of a sentence, the sub- and so far as it determines, 
 
 ject always precedes the finite another word, takes prece- 
 
 verb, and the latter always dence of the latter without 
 
 concludes the sentence. All exception. The object pre- 
 
 other words which depend cedes the verb, because the 
 
 upon these principal parts verb is determined by its 
 
 precede them respectively ; object, inasmuch as it indi- 
 
 so that the most important vidualises the action of the 
 
 of the dependent words is verb, 
 
 placed nearest to its prin- 
 
 * Tamil Grammar, p. 117. t Essay on the Tataric Languages, p. 3. 
 
 p 3 
 
210 
 
 Tamulic. 
 cipalj and the least important 
 farthest from it. 
 
 2. The adjective always precedes 
 
 the substantive ; as, "good- 
 father." 
 
 3. The noun precedes its govern- 
 
 ing participle or preposition ; 
 as if "father-loving," " fa- 
 ther-from." 
 
 Tataric. 
 
 4. The adverb precedes the verb ; 
 
 as, "I shall much love." 
 
 5. The infinitive precedes the 
 
 governing verb ; as, " to 
 eat go." 
 
 6. The negative branch of a sen- 
 
 tence precedes the affirma- 
 tive. 
 
 7. The number precedes that 
 
 vp-hich is numbered. 
 
 8. The genitive precedes the 
 
 governing noun ; such as. 
 
 The adjective 
 substantive. 
 
 precedes the 
 
 3. The object precedes the verb ; 
 
 what depends on a preposi- 
 tion precedes the preposition 
 (i. e. post -position). The 
 post -position is originally 
 a substantive standing to the 
 noun in the relation of a 
 genitive. 
 
 4. Jfhe adverb precedes the verb. 
 
 5. This would be included under 
 
 No. 3. 
 
 6. A relative sentence comes he- 
 
 fore the relative upon which 
 it depends. 
 
 7. See No. 2., and add the pos- 
 
 sessive pronominal adjective 
 preceding the noun. 
 
 8. The genitive precedes that 
 
 which governs it. 
 
 " king's palace." 
 
 From these general remarks it is evident that the order of the parts 
 of speech in Tamil is opposite to that in English, so that the Euro- 
 pean student has to effect an entire change in the arrangement of his 
 ideas. 
 
 After enumerating the organic and fundamental coincidences which 
 affect the formative principles of these two extreme members of the 
 Turanian family, we need not dwell much longer on smaller traits of 
 similarity. Yet, as in a picture a single line may often help to bring 
 out a likeness which did not strike the eye before, one feature may at 
 
211 
 
 least be mentioned, which, though in itself of little significance, is 
 yet of interest to those who are fond of watching the wonderful 
 instinct of language in its various manifestations. 
 
 The Canarese * possess, for the expression of collective ideas, a 
 large number of what are called " pair-words," or " double words." 
 They resemble the English "topsy-turvy," "chit-chat," &c. In most 
 of them the principle of alliteration has been observed, and many 
 obsolete words have been preserved in these compounds only. It is 
 curious that, as in German many expressions of this kind have 
 been kept in legal documents, the Canarese law, anterior to the 
 Mohammedan conquest, teems with the same class of compounds. In 
 some cases the Canarese simply repeats the same word, changing the 
 first syllable into gi, in order to give it a collective or more compre- 
 hensive meaning. A Brahman says that he has to perform snana 
 gin a, which means bathing (snana) and similar ceremonies connected 
 with it. Niru is water; niru giru, water and similar things. 
 Ata. is play; ata gita,, play and other amusements. Matu is 
 speech; matu gitu, speeches and the rest. Arasu, king; arasu 
 girasu, the king and other magistrates. Not only Sanskrit words, 
 but even foreign terms taken from English, have to submit to this 
 process, and a Canarese cook, who has to prepare the dessert, speaks 
 of it as "cake glke. 
 
 Gyarmathi describes the same peculiarity in the language of the 
 Hungarians and Laplanders. Both, he says, delight in forming such 
 expressions as 
 Lapponian — Pekkest pekkai. Hungarian — Diribrol darabra, de 
 
 frusto in frustum. 
 „ Jepestjapai. „ Eszendorol eszendore, 
 
 de anno in annum. 
 „ Katest katei „ Kezrol kezre, de manu 
 
 in manum. 
 „ Orron orroje. Orokkon orokke, in 
 
 seternum. 
 „ Lakkas laka. „ Idebb idebb, non pro- 
 
 cul. 
 
 * Weigle, On Canarese Language and Literature, p. 276. 
 p 4 
 
212 
 
 Lapponian — Pako lako. Hungarian — Pelda beszed, ada- 
 
 gium. 
 
 - la Malay, again, the same feature is most prominent. It exists 
 there, as in Canarese and the TJgric languages, not only in isolated 
 cases, or, as in German, in obsolete words and expressions, but as a 
 grammatical principle applied in various manners, — all showing 
 that plastic power of language, which is able to express the intel- 
 lectual and merely formal by the material, and which in the Arian 
 languages also has left the traces of its former existence in such 
 forms as the Intensive, Desiderative, and similar grammatical deriva- 
 tions. 
 
 In Malay* a word is sometimes simply repeated, as mata-mata, 
 a scout. 
 
 When, however, an inseparable prefix is annexed to a radical, this 
 prefix is usually omitted in the second member of the reduplication, 
 as barlari-lari, to run on; barturut-turut, consecutively. 
 
 When the word is a verb having a reciprocal sense, the particle 
 is annexed to the second member of the reduplicated word, and not 
 to the first, as bunoh-mambunoh, to slaughter frequently and 
 mutually. • 
 
 Sometimes, the reduplicated word is a primitive of which the 
 etymology cannot be traced, as an tar- an tar, a rammer; rama- 
 r a m a, a butterfly. 
 
 More frequently, the etymology can be traced, although the deri- 
 vation is often whimsical. From api, fire; api-api, a firefly. 
 Prom anak, young; anak-anakan, a puppet. From kera, to 
 think; kera-kera, to conjecture. 
 
 Adverbs are frequently formed by the redupUcation of other words, 
 as from kunung, sudden; kunung-kunung, suddenly. From 
 churi, to steal; churi-churi, stealthily. With this compare 
 Italian poco poco. 
 
 Often the reduplication of an adjective makes only an intensitive, 
 asbasar-basar, very great; manis-manis, very sweet. 
 
 The mere love of alliteration has contributed to multiply these 
 reduplicatives. Thus gilang-g^milang, effulgent. So laki, 
 
 * Crawfurd, Malay Grammar, p. 57. 
 
213 
 
 a man, is most generally written and pronounced laki-Iaki, and 
 this by abbreviation becomes lalaki, man. Similar abbreviated re- 
 duplicated words are, lalaba, a spider, instead of laba-laba; 
 papuwah, frizzly, instead of puwah-puwah. This is one of the 
 many cases where in a Turanian language we can watch the process 
 of which in Arian dialects we see but the result. What better ex- 
 planation can be given of intensive or frequentative verbs, such as 
 yay&A, to implore, from yak, to ask, in Sanskrit, than lalaki, 
 man, instead of laki-laki? 
 
 What, then, it may be asked, is the difference between such forms 
 as pointed out in Nomadic and Political languages ? It is this, that 
 Nomadic languages retain the consciousness of this process, and 
 therefore can apply it to any word, though it has never been applied 
 to it before. They know that lalaki is laki-laki; they still use 
 both; while, to a Hindu, yayaA was as little a repetition of yaA, as 
 waiirdWii) and SaiSr'XKd) were to a Greek, gurgulio and gingrio to 
 a Koman.* 
 
 CONCLUSION. 
 
 THE POSSIBILITY OF A COMMON ORIGIN OF LANGUAGE. 
 
 Hebe I must close for the present this communication on the Ta- 
 mulic languages, and their claims to be considered a branch of that 
 vast family of speech which should be called Nomadic or Turanian, in 
 contradistinction to the two political concentrations of human speech,, 
 the Arian and Semitic. I doubt not that the results at which I have 
 arrived, and the method by which I have been guided, will be ques- 
 tioned on ethnological as well as philological grounds. To classify 
 languages as such, regardless of the physiological characteristics of 
 the races by whom they are spoken, will appear presumptuous in the 
 eyes of the Ethnologist, while to me it seems to hold out the only 
 hope of settling eventually the conflicting claims of Ethnology and 
 Phonology. What we are accustomed to call "race,"t and what, as 
 
 * See Bopp's Comparative Grammar, § 753. 
 
 t If " race " is derived, not from " radix '' as was hitherto supposed, but from 
 the Old High-German reiza, line, lineage, it might be retained as a technical term. 
 
214 
 
 Humboldt has shown, should more properly be called " variety," 
 may date from a period in the history of the world anterior to any 
 division of language. Or, on the other hand, its first effects may 
 have been felt long after the confusion of speech had led to the dis- 
 persion of mankind. In either case the classification of language 
 could not be expected to coincide with the classification of the varie- 
 ties of man. Only on the supposition that the first divergence of 
 race took place contemporaneously with the first divulsion of lan- 
 guage, could a coincidence between ethnological and phonological 
 classes be reasonably anticipated, though even then the mysterious 
 intervals of so many centuries between this first parting and the later 
 meeting again of the world's inhabitants through war, conquests, and 
 migrations, would be sufficient to account for any disturbance that 
 may be now observed in the parallel progress, ramification, and 
 intertwining of race and speech. 
 
 Physiological Ethnology has accounted for the varieties of the 
 human race, and removed the barriers which formerly prevented us 
 from viewing all mankind as the members of one family, the off- 
 spring of one parent. The problem of the varieties of language is 
 more difficult and has still to be solved, as we must include in our 
 survey the nations of America and Africa. But over the languages 
 of the primitive Asiatic Continent of Asia and Europe a new light 
 begins to dawn, which, in spite of perplexing appearances, reveals 
 more and more clearly the possibility of their common origin. 
 
 In order to perceive this, and to command this wide view, we must 
 put aside the microscope through which we examine the organism 
 and the ramifications of so small and modern a cluster of dialects as 
 the Arian and Semitic. Different subjects require different methods, 
 and because the method of Bopp and Grimm has been found appli- 
 cable to an analysis of Arian speech, it does not follow that the same 
 would lead to satisfactory results in higher and more comprehensive 
 branches of linguistic study. We must open our eyes, and ask our- 
 selves what, according to the nature of the case, we can expect to 
 scan and to comprehend, even from that distant point of view, which 
 we necessarily occupy in looking toward the primordial epochs of the 
 history of language. The millions of people who speak and have 
 spoken for centuries from Ceylon to Iceland the innumerable dialects 
 
215 
 
 of Sanskrit, Persian, Gallic, Teutonic, Sclavonic, Italic, and Greek, 
 shrink here together into one small point, and are represented, as it 
 were, by one patriarchal individual, the first Arian, the ancestor of 
 the Arian race. For on all these languages, from Sanskrit to Eng- 
 lish, there is one common stamp — a stamp of definite individuality — 
 inexplicable if viewed as a product of nature, and intelligible only as 
 the work of one creative genius. Sanskrit, Persian, Greek, Latin, 
 Sclavonic, Teutonic, and Celtic, are simply continuations of one common 
 spring of language, as much as Spanish and Portuguese, French and 
 Proven9al, Italian and Wallachian, are all but Latin under different 
 aspects. The differences between languages, as distant geographically, 
 chronologically, and grammatically, as Sanskrit and English, vanish ; 
 and all that remains in this comprehensive view is, that one system 
 of grammar, and that patrimony of common roots, which we call 
 Arian, in opposition to Semitic. No new root has been added, no 
 new grammatical form been produced in any of the Arian provinces 
 or dependencies, of which the elements were not present at the first 
 foundation' of this mighty empire of speech. 
 
 The Semitic languages also are all varieties of one form of speech. 
 Though we do not know that primitive language from which the 
 Semitic dialects diverged, yet we know that at one time one such 
 language must have existed. In it all the peculiarities which now 
 distinguish the three branches of Semitic were not yet developed, 
 but they must have existed potentially. We cannot account for the 
 coincidences between the language of Mohammed and Moses without 
 the admission that, before the separate existence of the oldest Hebrew 
 and the earliest Arabic, there was a real language to which Hebrew and 
 Arabic stand as French and Italian stand to Latin. 
 
 The Semitic, therefore, and the Arian languages must be viewed as 
 two individuals, or as the manifestations and works of two individuals 
 which it is impossible to derive from one another. They differ in all 
 that is formal, following sometimes opposite directions in the first 
 principles of grammatical combination. They differ even in their 
 radical elements, inasmuch as each adopted its own process of deter- 
 mining roots by reduplication of final or initial letters, or by distinct 
 additional elements. They differ again in the meaning of roots, be- 
 cause it was a matter of individual choice what power should 
 
216 
 
 become fixed and technical in radicals, which, according to their Very 
 nature, must originally have possessed an indefinite applicability. 
 
 But, though in physical Ethnology we cannot derive the Negro from 
 the Malay or the Malay from the Negro type, we may look upon each as 
 a modification of a common and more general type. The same applies 
 to the types of language. We cannot derive Hebrew from Sanskrit, 
 or Sanskrit from Hebrew, but we can well understand how both may 
 have proceeded from one common source. They are both channels 
 supplied from one river, and they carry, though not always on their 
 surface, floating materials of language which challenge comparison, 
 and have already yielded satisfactory results to careful analysers. It 
 is true, if there were any strong arguments against the common origin 
 of these two channels of speech, the coincidences between them, 
 hitherto pointed out, would perhaps not suffice to silence them. 
 But, unshackled as we are by any contrary evidence, and encou- 
 raged as we must feel by the success of physical research, there is 
 even now sufficient evidence with regard to a radical community 
 between Arian and Semitic dialects, to enable us to say that their 
 common origin is not only possible, but, as far as linguistic evidence 
 goes, probable ;_ while to derive the Semitic from the Arian, or the 
 Arian from the Semitic type, may henceforth be declared a gram- 
 matical impossibility. 
 
 Before we allow our eyes to swerve to still more distant regions, 
 we must confront those uncounted dialects of Asia and Europe, whose 
 grammar does not run in either an Arian or Semitic channel. They 
 share in none of the features which distinguish the Arian and the 
 Semitic types, and the first point which we can establish with regard 
 to them is, that at no time, after the first separation of the Arian and 
 Semitic types, can they have formed part of these two historical 
 developments of language. Nothing of what is traditional, petrified, 
 or individual in either Semitic or Arian grammar, can be discovered 
 in any of the other dialects of the Asiatic continent. General 
 features common to Arian, Semitic, and Turanian languages, can only 
 be ascribed to the very earliest period of Asiatic speech. 
 
 Thus the Turanian dialects share one thing in common, they 
 
 all represent a state of language before its individualisation by the 
 Arian and Semitic types. But these Turanian languages cannot be 
 
217 
 
 considered as standing to each other in the same relation as Hebrew 
 and Arabic, Sanskrit and Greek. In smaller spheres, similar families, 
 like the Arian or Semitic, can be established within the Turanian king- 
 dom. The Tamulic dialects, for instance, are held together by the 
 same close ties of relationship as Greek and Latin, Hebrew and 
 Arabic. They necessitate the admission of a common parent, of a 
 long continued grammatical concentration preceding their gradual 
 dispersion. The same applies to the different branches, which have 
 been called Taic, Bhotiya, Malaic, Mongolic, Tungusic, Tataric, and 
 Finnic. The languages belonging to each of these branches, point to 
 so many parent-languages, whence they proceeded, and which they 
 represent under different aspects. But these branches themselves 
 must be viewed as separate in their beginnings, neither of them 
 being subordinate to any other, neither of them parent or offspring, but 
 all springing side by side from the same soil, though with different 
 powers of growth, and under circumstances more or less favourable 
 to their grammatical organisation. 
 
 Nor can these Turanian stems be considered as standing to one 
 another in the same relation as Semitic to Arian. The separation of 
 these two dialects and their independent growth is the result of an 
 individual act, unaccountable in its nature and origin, like every- 
 thing individual, while the separation and divergence of the Turanian 
 languages can be explained as the result of a gradual, natural, and 
 simple process, which, out of many things that were possible in the 
 mechanical combinations of roots, fixed a certain number of real forms 
 which, under geographical and political influences, became consolidated 
 into national idioms. As in the formation of political societies, we do 
 not require the admission of any powerful individual mind to account 
 for the presence of governed and governing classes, or of laws against 
 theft and murder, but can explain these as the necessary result of 
 social agglutination, we see nothing in the organisation of the Tura- 
 nian languages that betrays the influence of some individual poetical 
 genius, as the framer of peculiar laws, or the author of certain gram- 
 matical principles. In the Semitic and Arian languages, on the 
 contrary, we find institutions, laws, and agreements, which, like the 
 laws of inheritance and succession at Rome or in India, show the 
 stamp of an individual will impressed on the previous traditions 
 
218 
 
 of scattered tribes. It is possible that the Semitic and Arjan 
 languages also passed through a stage of mechanical crystallisation, 
 or uncontrolled conglomeration of grammatical elements ; but they 
 left it, and entered into a new phase of growth and decay, and that 
 through the agency of one creative genius grasping the floating 
 elements of speech, and preventing by his fiat their further atomical 
 concretion. It is after this had taken place, that the real life of 
 Arian and Semitic language begins, and all Arian and Semitic dialects 
 which we know are the descendants of these two languages, already 
 individualised to the highest degree. 
 
 In the Turanian group this individual element is wanting. Hence 
 the different branches, the Tungusic, Mongolic, Turkic, and Finnic 
 in the North, the Taic, Malaic, Bhotiya, and Tamulic in the South, 
 are deficient in that family likeness which is peculiar to the offspring 
 of the same individual. They are radii diverging from a common 
 centre, not children of a common parent. This explains their simi- 
 larity as well as their differences. They share much in common, and 
 show that before their divergency a certain nucleus of language was 
 formed, in which some parts of language, the first to crystallise and 
 the most difficult to be analysed, had become fixed and stationary. 
 Numerals, pronouns, and some of the simplest applied verbal roots 
 belong to this class of words. But even these parts of speech had not 
 yet grown into a system, before the Turanian camp broke up, and 
 hence were not retained as a whole. We may even distinguish two 
 such nuclei of Turanian speech, a Northern and a Southern ; and we 
 may trace both back to a still higher point where their repective pe- 
 culiarities are merged again into one common current. Here, where 
 the differences between the Turanian languages cease, the first sta- 
 mina of the Arian and Semitic languages also would be found to con- 
 verge toward the same centre of life. Eadicals, applied to certain defi- 
 nite but material meanings in common by all Turanian dialects, belong 
 to this primitive era, and some of them can even now be proved the 
 common property of the Turanian, the Semitic, and Arian branches. 
 And here the last question presents itself, which Comparative 
 Phonology has to answer. Does this common ground, where the 
 differences of Arian, Semitic, and Turanian dialects are neutralised, 
 correspond with that stage in the growth of language, where the 
 
219 
 
 vital powers of the Chinese were arrested, or is there still an interval, 
 not bridged over by any traditions of language, between this one 
 patriarchal utterance, and the common inheritance of the " three sons 
 of Feridun ? " Some few roots that could claim this primeval origin 
 have been pointed out. Their number will never be very great; 
 and their sound and meaning will always have, as Schleiermacher 
 remarked, " quelque chose de vague." But could this be otherwise ? 
 Suppose we actually found a number of definite roots, with secondary 
 and tertiary letters, and with complicated significations, in this common 
 treasury of all the languages of Asia. Should we be able to explain 
 such a fact ? Would it not invalidate all arguments, and entirely de- 
 stroy all conclusions to which a careful study of the broken traditions 
 . of mankind has led us? Such roots cannot, have not, and will not 
 be found. But if the view here proposed on the origin and growth 
 of language — a view according with all the evidence which the 
 documents of the various dialects of Asia and Europe supply — 
 be accepted, these vague, effaced, and fragmentary roots rise into im- 
 portance, because confirming, though not proving, our anticipations, 
 like the segments of a circle whose centre we have guessed. 
 
 As to the formal elements, or the grammatical growth of language, 
 no difllculty exists in considering the grammatical system of Sanskrit, 
 the most perfect of the Arian dialects, as the natural development of 
 Chinese — an admission made even by those who are most opposed to 
 the generalisations in the science of languages. 
 
 These two points, therefore, Comparative Philology has gained : — 
 
 I. Nothing necessitates the admission of different 
 independent beginnings for the material elements of 
 the Turanian, Semitic, and Arian branches of speech, 
 — nay, it is possible even now to point out radicals 
 which, under«various changes and disguises, have been 
 current in these three branches ever since their first 
 separation. 
 
 II. Nothing necessitates the admission of different 
 beginnings for the formal elements of the Turanian, 
 Semitic, and Arian branches of speech — and though 
 it is impossible to derive the Arian system of gram- 
 mar from the Semitic, or the Semitic from the 
 
220 
 
 Turanian, we can perfectly understand how, either 
 through individual influences, or by the wear and 
 tear of grammar in its own continuous working, the 
 different systems of grammar of Asia and Europe 
 may have been produced. 
 
 If we translate these grammatical conclusions into historical 
 language, we arrive at the following facts : — 
 
 The first migration from tlie common centre of mankind proceeded 
 eastward, where the Asiatic language was arrested at the first stage of 
 its growth, and where the Chinese, as a broken link, presents to the 
 present day a reflection of the earliest consolidation of human speech. 
 
 The second dispersion was that of the Turanian tribes. Lan- 
 guage had slowly advanced, and formed certain deposits of nume- 
 rical, pronominal, and verbal roots, before the Turanians separated and 
 spread with their dialects to all the corners of the earth. Gram- 
 matical growth had commenced, and an abundance of forms had been 
 thrown out from which all took what seemed useful and necessary to 
 them according to their different tastes and characters. Certain 
 grammatical and syntactical principles also had been deeply impressed 
 upon the mind of the Turanian colonists before they started, and 
 these impart to their languages a similarity, even where the material 
 elements of the single dialects have since been changed and replaced. 
 
 We must admit two directions for the migrations of the Turanians, 
 as indicated by their languages — a northern and a southern. 
 
 The Northern Division comprehends the Tungusic, Mon- 
 golic, Tataric, and Finnic branches. 
 
 The Southern Division comprehends the Taic, Malai'c 
 Bhotiya(Gangetic andLohitic), and Tamulic branches. 
 
 These two divisions had not arrived at any social or political consoli- 
 dation before they were broken up respectively into different colonies. 
 They probably had no laws, no popular poetry or sacred songs which 
 might have served as a common standard. They broke up carrying 
 away each a portion of their common language — and hence their 
 similarity ; but they possessed as yet nothing traditional, nothing like 
 a common inheritance in language or thought, — and hence their 
 differences. 
 
 In following the indications of the gradual advance which the 
 
221 
 
 ascending scale in the grammatical growth of these different branches 
 holds out to usj we should be led to suppose that the first migration 
 in the south was that of the people speaking Tai dialects, who settled 
 along the rivers Meikong, Meinam, Irawaddi, and Brahmaputra. 
 
 In the north the first migration was that of the Tungusic tribes, 
 following the course of the rivers Amur and Lena. 
 
 Both are conterminous with China, and their languages have scarcely 
 left the Chinese stage. 
 
 The second migration is that of the Malaic tribes in the south, 
 who followed the same direction as the Tai tribes, but, finding the 
 land occupied, pushed onward to the islands and the sea. 
 
 In the north the second migration would be that of the Mongolic 
 races, pressing on the Tungusic races, their predecessors ; and then 
 spreading westward along the chain of the Altai mountains. 
 
 Both nations are characterised by a spirit of enterprise, which 
 on the sea made them feared as pirates, in the desert as robbers. 
 Their languages are more adapted for stern and short command, 
 than for persuasive discussion and argument. 
 
 The third migration in the south tended toward Bhota or Tibet 
 and the frontiers of India. The Kamboja peninsula and the coast 
 being occupied, these tribes chose the high plateau, north of India, 
 and in later times poured into India through the mountain passes 
 of the Himalaya. Their language, particularly where it has received 
 literary cultivation, is capable of expressing abstract reasoning, but 
 is liable to lose itself in artificial complications and polysynthetic 
 confusion. 
 
 The same applies to the third migration in the north. The 
 Turkish tribes, finding all the intermediate country taken pos- 
 session of, proceeded westward to the Ural and the frontier of 
 Europe. Their language, particularly in Turkish, arrived at so high 
 a degree of formal perfection as to make it almost inconvenient for 
 the purposes of common conversation. 
 
 The last colony in the south was the Tamulic, in the north the 
 Finnic — both at an early period advanced to a high degree of 
 civilisation, of which we find the traces even now in the wise 
 economy of their languages, and in the few remains of their early 
 institutions and literature. Both were crushed by the later con- 
 
222 
 
 quests of Arian nations ; so that in the south we have but vague 
 traditions of their former state, and even these perverted by the 
 jealousy of their Brahmanic conquerors ; while in the fens of Finland 
 oral tradition has handed down to us not only the names of these 
 ancient heroes, but the very songs which celebrated their deeds. 
 
 If we adopt this view of the gradual spreading of the Turanian 
 branches, we have to suppose that each successive migration, finding 
 the nearest ground occupied, pushed forward to more distant quarters. 
 This seems the more natural supposition ; for if we inverted the 
 historical order, and looked upon the last migration as the first, we should 
 have to account for the retrograde movement in the grammatical forma- 
 tion of the four southern and northern dialects. Finnic would then 
 represent the earliest state of Turanian grammar, while the Tungusic 
 would correspond to the latest, — a view which might be defended in 
 the later history of Arian languages, but is untenable in Turanian 
 philology. With the former view, the different degrees of gram- 
 matical perfection, and the respective geographical distance of each 
 branch from China, would closely correspond with the historical 
 separation and individualisation of each Turanian branch. 
 
 Besides these northern and southern radii of Turanian speech, 
 there are still several sporadic clusters of dialects, equally belonging 
 to the Turanian stage of language, but left to themselves, as it were, 
 and lost in impervious mountains and deserts. In their seclusion, 
 and debarred from the severe attrition which every dialept ex- 
 periences in intercourse with other languages, they have each pro- 
 duced the utmost variety of grammatical forms, and revel in a 
 luxuriance of verbal distinctions which small and secluded tribes 
 alone are able to indulge in. 
 
 These are the aboriginal languages spoken in the impenetrable 
 valleys of the Caucasus ; the Basque in the Pyrenees, and on the 
 very edge of Europe, and the Samo'iedic in the still less accessible 
 Tundras of the north of Siberia. 
 
 In these secluded dialects, the peculiarities of individuals may 
 gain an influence which changes the whole surface of grammar and 
 dictionary. Turanian languages, particularly, are so pliant that they 
 lend themselves to endless combinations and complexities, unless a 
 national literature or a frequent intercourse with other tribes act as 
 
223 
 
 safeguards against dialectical schism. Tribes who have no litera- 
 ture and no sort of intellectual occupation, seem occasionally to take 
 a delight in working their language to the utmost limits of gram- 
 matical expansion. The American dialects are a well-known in- 
 stance : and the greater the seclusion of a tribe, the more amazing 
 this rank vegetation of their grammar. We can at present hardly 
 form a correct idea with what feeling a savage nation looks upon its 
 language ; whether, it may be, as a plaything, a kind of intellectual 
 amusement, a maze in which the mind likes to lose and to find itself. 
 But the result is the same everywhere. If the work of agglutination 
 has once commenced, and there is nothing like literature or society to 
 keep it within limits, two villages, separated only for a few gene- 
 rations, will become mutually unintelligible. This takes place in 
 America, as well as on the borders of China and India ; and in the 
 North of Asia, Messerschmidt relates, that the Ostiakes, though 
 really speaking the same language every where, have produced so 
 many words and forms peculiar to each tribe, that even within the 
 limits of twelve or twenty German miles, conversation between them 
 becomes extremely difficult. It must be remembered also, that the dic- 
 tionary of these languages is small if compared with a Latin or Greek 
 Thesaurus. The conversation of nomadic tribes moves within a narrow 
 circle, and with the great facility of forming new words at random, 
 and the great inducement that a solitary life holds out to invent for 
 the objects which form the world of a shepherd or huntsman, new 
 appellations, — half poetical, perhaps, or satirical, — we can under- 
 stand how, after a few generations, the dictionary of a nomadic 
 tribe may have gone, as it were, through more than one edition. 
 
 There are still a few languages which for the present must remain 
 unclassed, because the means are wanting for subjecting them to a 
 grammatical analysis. Such are the languages of Korea, of the Ko- 
 riiiks, Kam^adales, and of Japan. Their number is small, and in 
 them also some traces of a common origin with the Turanian lan- 
 guages have, it is probable, survived, and await the discovery of phi- 
 lological research. 
 
 Other branches of Turanian dialects may have existed in Asia and 
 Europe during times of which we have no records, and previous to 
 the first immigration of Arian and Semitic races. Wherever these 
 
 Q 2 
 
224 
 
 two races arrive, they find the land occupied by barbarians, repre- 
 sented as giants or evil spirits, and speaking languages unintelligible 
 to the new arrivers. They were exterminated, and their languages 
 silenced for ever. Here the links may have been broken and lost which 
 once united the language of Asia and Europe with the scattered dia- 
 lects of Africa and America. An extension of the Turanian family 
 to these two continents has been hinted at by several scholars. The 
 Greenland language has been pointed out as showing a transition of 
 Turanian into American dialects, and the researches of physical 
 science have clearly indicated the islands east of Siberia, as the only 
 bridge on which the seeds of Asia could have been carried to the 
 New World. As to African dialects, all is still conjecture, except 
 this, that, besides the Semitic type of some African languages, such 
 as the Galla, spoken north of the equator, there is another gram- 
 matical character impressed on other idioms, as, for instance, the 
 Hottentot, which, by its mechanical perfection and artificial compli- 
 cation, invites a comparison with the grammatical system of the 
 descendants of Tur.* 
 
 What was the state of the Arian and Semitic dialects during this 
 early period of ethnic migration and struggle we do not know. 
 Their history begins only when they cease to belong to the chaotic 
 mass of Turanian Nomads. They appear at once on the stage of 
 history, fully clad in their own armour, the enemies of the barbarians, 
 the worshippers of brighter gods, and with a language which has left 
 for ever the tumult of a Turanian arena. They are Arians, or She- 
 mites, inasmuch as they are no longer Turanians ; and though their 
 antecedent growth must have passed through a Turanian phase, this 
 is overcome when they appear as the heralds of a new era in the his- 
 tory of man. It is only after having conquered in themselves Tura- 
 nianism, in every sense of the word, that they advance through Asia 
 and Europe as the conquerors of the descendants of Tur. This 
 battle is not yet ended ; and the largest share of the earth still be- 
 longs to its earlier occupants. The Ai-ian and Semitic languages 
 occupy but four peninsulas of the primeval continent, — India, 
 Arabia, Asia Minor, and Europe ; all the rest belongs to the fomily 
 
 '^ See Boyce's Kaffir Grammar, Introduction, page ix. 
 
225 
 
 of Tur. But the countries reclaimed by Shem and Japhet mark the 
 high road of civilisation, and comprehend the stage on which the 
 drama of ancient and modern history has been acted. 
 
 Shem was in advance of Japhet ; and his first colonies represent a 
 stage of language not yet decidedly Semitic, not yet "freed from all 
 Turanian influences, and, hence, less distant also from the stream of 
 Arian speech. These were the colonists of Africa, who have fallen 
 back into nomadic habits, but whose language is still the language 
 of the people in Marocco, Algiers, Tunis, Tripoli, and Fez, wherever 
 it has not been supplanted by the tongue of the conquering Arabs. 
 A second colony, not yet decidedly Semitic, but, owing to political 
 influences, more settled in its grammatical system, took its abode in 
 Egypt. A third made its idiom the language of Babylonia and 
 Assyria. 
 
 These three early colonies exhibit the Semitic in its struggle 
 towards grammatical form and consistency ; and the individuality of 
 Shem has not yet in them obscured those traces of a common past 
 which enable us to connect the radical elements of the Semitic with 
 the Turanian, and through it with the Arian family. 
 
 After these three colonies, the limits of the Semitic speech were 
 drawn more closely together, and the three later branches, the Arabic, 
 Aramaic, and Hebrew, stand before us as cognate descendants of one 
 parent, who has left to each the sharp and decided features of his own 
 expression. 
 
 The Arian family has had but one generation of dialects. There 
 was a time when the ancestors of this race formed one family^ in the 
 proper sense of the word. Their language was then the idiom of 
 a hamlet, as Latin was at one time spoken by the few adventurers 
 who built their cottages on the hills of the Tiber. Without some 
 such previous concentration, as it is impossible to account for the 
 perpetuation of the most minute and fanciful forms in the Roman 
 dialects of modern Europe, it would be in vain to account for the 
 coincidences between the Arian dialects of the ancient world. The 
 Arian language, which grew, or became nationalised, into Sanscrit, 
 Persian, Greek, Latin, Teutonic, Slavonic, and Celtic, must have 
 been a language richer perhaps than any of its descendants, but a 
 language with such settled principles, and such intense individuality 
 
 Q 3 
 
226 
 
 in grammar and dictionary, that the national, or, as we may here call 
 it, the individual character of its descendants, though widely diiferent 
 as the meditative Hindu and active Greek, could never obliterate or 
 efface the stamp of their common parent. 
 
 And if now we gaze from our native shores over that vast ocean of 
 human speech, with its waves rolling on from continent to conti- 
 nent, rising under the fresh breezes of the morning of history, and 
 slowly heaving in our own more sultry atmosphere, — with sails 
 gliding over its surface and many an oar ploughing through its 
 surf, and the flags of all nations waving joyously together, — with its 
 rocks and wrecks, its storms and battles, yet reflecting serenely all 
 that is beneath, and above, and around it, — if we gaze, and hearken to 
 the strange sounds rushing past our ears in unbroken strains, it 
 seems no longer a wild tumult, or avripiOfiov yiXaa-fia, but we feel as 
 if placed within some ancient cathedral, listening to a chorus of in- 
 numerable voices ; and the more intensely we listen, the more all 
 discords melt away into higher harmonies, till at last we hear but 
 one majestic trichord, or a mighty unison, as at the end of a sacred 
 symphony. 
 
 Such visions will float through the study of the grammarian, and 
 in the midst of toilsome researches his heart will suddenly beat, as he 
 feels the conviction growing upon him that men are brethren in the 
 simplest sense of the word — the children of the same father — what- 
 ever their country, their colour, their language, and their faith. 
 
 MAX MiJLLER. 
 
 Note. — Circumstances over which I had no control made it impossible to carry 
 out a uniform system of transcription in the letter on the Turanian Language and 
 in the Tables appended to it. 
 
227 
 
 The Languages of Asia and Europe arranged according to their 
 Grammatical Principles. 
 
 
 LIVING LANGUAGES. 
 
 
 
 
 ja 
 
 
 
 
 u .„„„.,„ 
 
 
 
 O J3 
 
 <»:« 
 
 
 
 nnic. 
 Samoiedic, Caucasi 
 
 a (Gangetic and Lo 
 amulic. 
 
 Africa, N.W 
 
 Egypt. 
 
 Babylon. 
 
 m of Arabia. 
 
 , Aram. 
 
 , Palestin 
 
 the Indie bra 
 Iranic 
 Celtic 
 Italic 
 Hellenic 
 Windic 
 Teutonic 
 
 
 
 ^ o ■■ -t<- 
 
 o ;g o s s = ^ ^ " 
 
 s :: a 
 
 .2 " "^ .2 
 
 
 
 .■M^4 ^^ 
 
 :^ g :; srs 
 
 
 
 ^ &- 0) CS O 0? 
 
 l—j +J 1— ^ 
 
 
 
 "S -S rf3 CQ o ?i -=) 
 ,; Os -tm .. J3 ^ t*. 
 
 1 . .^ , . .g . . . . . . 
 
 
 
 •SScuOm ^S®o 
 
 +i 
 
 
 4^ 
 
 
 
 &^ 'v, .2 ra t^ ^ q- .2 
 
 iz; 
 
 
 iz; 
 
 .1 < 
 
 
 
 f the Tur 
 ion of the 
 ntration o 
 oncentrat 
 ;red langu 
 fthe Taj- 
 ion of the 
 ntration o 
 Joncentrat 
 
 Semitic Nucleus. Ariau Nucleus. 
 
 POLITICAL 
 STAGE. 
 
 1 
 
 AMALGAMATION. 
 
 
 ation 
 centrat 
 -Conce 
 
 (Scatte 
 ation c 
 centrat 
 
 -Conce 
 
 r 
 
 i 
 
 
 
 O 
 
 ^g 
 
 
 
 
 
 
 
 O 
 
 So 
 
 
 go 
 
 
 
 
 
 f3 
 
 13 
 
 
 
 c 
 
 
 
 
 
 
 
 O 
 
 
 
 
 
 
 
 
 
 "S 
 
 U 
 
 
 p 
 
 
 
 1 
 
 
 
 
 
 c 
 
 Northern Southern 
 
 
 
 0) 
 
 Branch. Branch. 
 
 
 NOMADIC 
 STAGE. 
 
 
 AGGLUTINATION. 
 
 FAMII-Y 
 STAGE. 
 
 JUXTAPOSITION. 
 
 ANTE- 
 DILUVIAN. 
 
 ROOTS. 
 
 Q * 
 
229 
 
 FIRST APPENDIX. 
 
 COMPARATIVE TABLE OF SUBJECTIVE AND 
 PREDICATIVE COMPOUNDS. 
 
 Explanation of Letters. 
 
 Capitals are used to represent Verbal bases. 
 Small Letters to represent Nominal bases. 
 Greek Letters to represent Pronouns. 
 
 A. a. a. to represent a word in the Nominative, or as Subject. 
 
 B. b. p. to represent a word in the Casus obliquus, or as Predicate. 
 
 For instance : 
 
 a. b.=Nominal base as subject, followed by Nominal base as 
 predicate: Hotel-Dieu. 
 
 a. /3.= Nominal base as subject, followed by Pronoun as predi- 
 cate: Hebr. Bl-i, G-od (of) I, i.e. my God. {Different 
 from fratelmo, i. e. fratellus mens.) 
 
 a. B.=Nominal base as subject, followed by Verbal base as predi- 
 cate. (Possible only if the verbal predicative base becomes 
 an adjective.) 
 
 The sign - is used after nominal bases. 
 
 The sign . is used after verbal bases. 
 
230 
 
 1.1 
 
 CfUM. 
 
 Egypt 
 a.b. 
 
 si-Hes, son {of) Isis. 
 
 uakh en nub, chain 
 {which or where) 
 gold,\.e. chainofgold. 
 
 suten-ter, king {who or 
 as) father. 
 
 Tieter'T[n\t.goddessif{iho 
 or as) mother, cf. 
 qtbeen-nmther . 
 
 I. 2. a. j8. 
 
 si-k, son {of} thee. 
 
 Bi-f, son {of) him. 
 
 set-ten daughter {of) 
 
 them. 
 (Coptic, No.) 
 
 1.3. a. B. 
 
 neter-naa, God-great. 
 
 cf. neter-mut, goddess- 
 mother, (a.b.) 
 
 II. 1. A. b. 
 
 Never. 
 
 II. 2. A. j3. 
 
 iri. en. a. 
 
 iri. en. ek. 
 
 iri. en. ef. 
 doing where I, i.e. 
 doing of me, i .e. I did, 
 thou didst, he did. 
 
 II. 3. A. B. 
 
 Never. 
 
 III. 1. a. b. 
 
 Never. 
 
 IIL2. 
 
 a. ^. 
 Never. 
 
 m. 3. «. B. 
 
 No. 
 (Exc. Cop t ic, L'l iii, 
 ek iri, efiri, I, thou, 
 hcjnakcs. cf. B«.) 
 
 Shem. 
 
 Palestine. 
 I. 1. a. b. 
 
 debar-melek, word {of a) 
 
 king, 
 malki-zedek, king {of) 
 
 justice. 
 ben o Be'or, son he {of) 
 
 Bear. 
 
 shir (asber le) Shelomoh , 
 
 the song {which to) 
 
 Solomon. 
 Syr. nausad sivao^chest 
 
 {where) silver. 
 Ethiop. wald a MSry- 
 
 am, son who {to) Mary. 
 Ethiop. maztnor za 
 
 Dawith, psalm that {of) 
 
 David (za=he, psalm 
 
 being a masc.) 
 anqaz enta samfiy, the 
 
 gate {which) heaven 
 
 (enta=she, gate being 
 
 a fern.) 
 
 I. 2. a. |3. 
 
 lebush-i, dress {qf) me. 
 
 Iebush-ka,(irf55 (of) thee. 
 
 lebush-ah, dress (of) her. 
 (cf. labsli-ah, she dress- 
 es). 
 
 I. 3. a. B. 
 
 d^m-Tikqi,blood-innorent. 
 
 IT. 1. A. b. 
 
 Nevbr. 
 
 n. 2. A. 0. 
 
 qathal.ti,qathal.td,qathl. 
 kh, hilling {to) m.e. 
 thee, her, i.e. / killed, 
 thou killcdst, she killed. 
 
 II. 3. A. B. 
 
 Never. 
 
 III. 1. a. b. 
 
 Never. 
 
 III. 2. u. a 
 
 Never. 
 
 III. 3. a. B. 
 
 (flt l'."). ni.qthol, ti.qttiol. 
 nah, yi.qthi.u, «ft' kift- 
 '"J?. 1/ou killing {/cm, ), 
 the*/ /ciltijtg. 
 
 China. 
 
 I 1. a. b. 
 
 I. 2. a. ;3. 
 
 I. 3. a. B. 
 
 n. 1. A. b. 
 
 Never. 
 
 II. 2. A. /3. 
 
 I. 3. A. B. 
 
 Never. 
 
 m. 1. a. b. 
 
 Never. 
 
 III. 2. a b. 
 
 Never. 
 
 III. 3. «. B. 
 
 No 
 (ngd la, I strike.) 
 (ni ta, thou strikc{st)). 
 
 N I A N. 
 
 India extra Gangem 
 1. 1, a. b. 
 
 No. 
 
 Exc. T ai dial ects. 
 Khamti, hang-pa, tail 
 
 (of) .fish, a fishes tail ; 
 
 pa=j?5A. 
 K a s s i a, ka.reng- 
 
 u.\i\a.n$, horn {of) goat. 
 Siamese, kua-khort, 
 
 head {of) man. 
 Malay, kapala-oraDg, 
 
 head {of) man. 
 A n a m, chua-nya.maff^r 
 
 {of) house. 
 
 I. 2. i*. 3. 
 
 Exc.K h am ti,etc.,mii- 
 man, hand {of) him, 
 i.e. his hand. 
 
 I. 3. a. B. 
 
 No? 
 Exc. Kham ti, etc., 
 
 kuo-ni, man good. 
 
 kun-mani, rrutn bad. 
 Miri, dinie-^ida, man 
 
 good. 
 Garo, mande-namja ni, 
 
 Tnjn had o/ (GenlL). 
 
 n. 1. A. b. 
 
 Never. 
 
 n. 2. A. 0. 
 
 No. 
 Exc. Naga dialects. 
 
 II. 3. A. B, 
 
 Never. 
 
 m. 1. a. b. 
 
 Never. 
 
 in. 2. a. $. 
 
 Never. 
 
 III. 3. a. B. 
 
 No. 
 
231 
 
 1.1 
 
 T 
 
 Cavcasms. 
 a, b. 
 
 No. 
 
 I 1. a. b. 
 
 Dehhan. 
 
 1. 2. a. ft 
 
 1. 3. a. B. 
 
 No? 
 Abchasian and 
 Tsherkessian, 
 aphshitB-abz^-khwa, 
 fish good, plW. =good 
 Jishes- 
 
 n. 1. A. b. 
 
 Never. 
 
 n.2. A. ft 
 
 No. 
 
 n. 3. A. B. 
 
 Never. 
 
 m. 1. u. b. 
 
 Never. 
 
 III. 2. o. ft 
 
 Never. 
 
 III. 3. a. B. 
 
 No. (cr. "But). 
 
 1. 2. a. ft 
 
 No. 
 ? Son tal. apa-t, 
 father ? 
 
 I. 3. a. B. 
 
 No? 
 
 n. 1. A. b. 
 
 Never. 
 
 n. 2. A. ft 
 
 No. 
 
 U. 3. A. B. 
 
 Never. 
 
 in. 1. a. b. 
 
 Never. 
 
 m. 2. a, )3. 
 
 Never. 
 
 III. 3. «. B. 
 
 No. 
 
 1. 1. a. b. 
 
 N. 
 
 Altai. 
 
 1. 2. a. ^. 
 
 Lapp.atzya-m, atzya-d, 
 
 atzya-B, Twy, thy, his 
 
 father. 
 Hung. atya-m,atya-d.at- 
 
 tya, my, thy, his father. 
 Y a k u t e, agha-m, agha- 
 
 n, agha-ta, my^ thy, his 
 
 father. 
 I. 3. a. B. 
 
 No? 
 
 n.i. A.b. 
 
 Never. 
 
 II. 2. A. $. 
 
 Hung. Transit, definite, 
 hall.ora, hall.od, hall. 
 ja, hearing (to) tne, i.e. 
 / heard {it), thou, he 
 heard. 
 
 Ibid. . Preterite indef. 
 var t .am, var t.al, var t, 
 waiting (to) me ; I, 
 thou., he waited. 
 
 Yakut e. Perfect, sanS- 
 tSm, sa^at.en.sanM.a, 
 thin/itTig (to) me; I, 
 tkoUf he thought. 
 
 n.3. A.B. 
 
 Never. 
 
 in. 1. a. b. 
 
 Never. 
 
 III. 2. a. B. 
 
 Never. 
 
 III. 3. u. B. 
 
 No. 
 
 Japhet. 
 
 In do-European. 
 1. ji. b. 
 
 No. 
 
 Exo. Pehlevi, kup i 
 Fars, mountain (there) 
 Persia, i.e. mountain 
 of P. 
 
 P a r s 1, qarI-Gar6thman, 
 the splendour (of) Gc- 
 rothman. 
 
 Parsi, vinasn i kasm, 
 the si0ht (where) the eye, 
 
 Persian, puser i dost, 
 the son (where) the 
 friend, the friend's son. 
 
 Afghan, Sard£ran da 
 Candahar, the Sar. 
 dars (they) Candahar, 
 i.e. of Candahar. 
 
 Cf. P e h 1 e V i . Z e n d, 
 gSum yim Sugdh(j- 
 sa yaoem, regionem 
 (quam) Sugdhee-situfn 
 hahentem. 
 
 I 2. a, )8. 
 
 No. 
 Exc. Persian, din-era 
 my religion i din-esh, 
 thy religion ,■ but not 
 in Parsi, exc. after 
 prepositions, as az-ash, 
 from him.. 
 
 1. 3. a. B. 
 
 No? 
 S a n 8 k. pita-maha, fa- 
 ther-grand, i.e. grand- 
 father. 
 
 n. 
 
 Never. 
 
 n. 2. A. ft. 
 
 IL3. A.B. 
 
 Never. 
 
 in. 1 a. b. 
 
 Never. 
 
 ni. 2. u. ft 
 
 Never. 
 
 m. 3 a. B. 
 
 No. 
 
232 
 
 Cham. 
 
 Egypt. 
 IV. 1. b. a. 
 
 No. 
 
 IV. 2. b. a. 
 
 Never. 
 
 IV. 3. b. A. 
 
 Nevek. 
 
 V. 1. B. a. 
 
 V. 2. B. a. 
 
 Present : 
 iri. a.do-I, I do. 
 iri. ek. thou does!, 
 m.ei.hedoes. 
 
 (Coptic. No.) 
 
 V. 3. B. A. 
 
 Neveb. 
 
 VI. 1. j8. a. 
 No. 
 
 Shem. 
 
 Palestine. 
 XV. 1. b. a. 
 
 No. 
 
 IV. 2. b. a. 
 
 Never. 
 
 IV. 3. b. A. 
 
 Never. 
 
 V. 1. B. a. 
 
 V. 2. B. a. 
 
 No. 
 
 V. 3. B. A. 
 
 Never. 
 
 VI. 1. ft a. 
 
 No. 
 
 VI. 2. A a. 
 
 NEVER: 
 
 VI. 3. ft A. 
 
 No. 
 
 VI. 2. ft a. 
 
 Never. 
 
 VI. 3. ft A. 
 
 No. 
 
 T TJ B Jl 
 
 ChiTta. 
 IV. 1. b. a. 
 1 . min-li, people's power 
 
 mln-ti li, 
 power. 
 
 IV. 2. b. a. 
 
 Never. 
 
 IV. 3. b. A. 
 
 Never. 
 
 v. 1. B. a. 
 
 pe-m^ white horse. 
 
 V. 2. B. a. 
 
 No. 
 
 V. 3. B. A. 
 
 Never. 
 
 VI. 1. $. a. 
 
 ngo-sin l-heart, i.e. mi/ 
 
 2. ngo-ti sin, mitie heart. 
 
 VI. 2. j3. a. 
 
 Never. 
 
 VI. 3. /3. A. 
 
 No. 
 
 India extra Gangem. 
 IV. 1. b. &. 
 1. Changlo, kurta-bi, 
 
 horse's leg. 
 Burmese, lu-khaung, 
 man's head. 
 
 2. Genitive AdjRctives : 
 Singpho, kansu-nd 
 
 rung, a cow's horn (a 
 bovinu cornu). 
 G a r o, ambal ni kethali, 
 
 a wooden knife. 
 
 IV. 2. b. o. 
 
 Never. 
 
 IV. 3. b. A. 
 
 Never. 
 
 V. 1. B. ». 
 
 B h o t. khang-zang, 
 
 house-good, a good 
 house. 
 zang-mi, good man. 
 
 V. 2. B. a. 
 
 No. 
 Exc. NAga. Present, 
 thier.ang, thien.o, 
 
 thie .a, put-I, i.e. / 
 put, thou pattest, he 
 pitts. 
 
 Preterite, thient.ak, 
 thien t.o, thien t.a, 7, 
 thou, he did put. 
 
 V. 3. B. A. 
 
 Never. 
 
 VLl. jS.a. 
 
 No. 
 
 Exc. Naga, i-lih, my 
 
 /.ite. 
 
 The same in Gya- 
 
 rung and Kiranti. 
 
 2. Genitive Adjectives : 
 
 Kachar i, ang-ni 
 
 ndwa, ?nei nomen. 
 N 4 g a, irang lah, mine 
 kite. 
 
 VI. 2. /3. a. 
 
 Never. 
 
 VI. 3. $. A. 
 
 No. 
 
233 
 
 Caucasus. 
 IV. I. b. a. 
 
 Suauian, mare-shiar, 
 
 man*s haTids, 
 Abchlasian, aph- 
 
 wizba-ala, girl's dog. 
 
 2. Genitive Adjectives. 
 
 IV. 2. b. a. 
 
 Never 
 
 rV.3.b. A. 
 
 Neveb. 
 
 V. 1. B. a. 
 
 Suanian, 6ilader-dc<is, 
 daily bread. 
 
 Dekhan. 
 IV. 1. b. a, 
 
 T am. vldu-kkatavu, 
 
 •hoitse-door, 
 kaUvari, stone-road. 
 
 2. Genitive- Adjectives. 
 
 IV. 2. b. a. 
 
 Never. 
 
 IV. 3. b. A. 
 
 Never. 
 
 V. 1, B. u. 
 
 Altai. 
 
 rv. 1. b. 
 
 Lapp, 
 
 It. 
 
 V.2.B. 
 
 "Ba. Lj 
 
 b ch, 
 
 chask. 
 
 chask 
 
 digly 
 
 thou, 
 
 hCfViS 
 
 a. 
 
 izian, Present, 
 isk.a, chask.a, 
 
 as, b chask.at, 
 •at, ' chask.an, I 
 
 i.e. f dig; dig 
 i.e. thou diggest ; 
 , you, they dig. 
 
 V. 3. B. A. 
 
 Never. 
 
 VI. 1. )3. a. 
 
 Suanian. s-ab, w-ab, 
 i-ab, h-ab, sh-ab, r-ab, 
 my, thy, his, our, your, 
 their father. 
 
 2. Genitive Adjectives ; 
 Lazian, shkimi ili, my 
 
 spear. 
 Suanian, mishcwa mu, 
 
 tneus pater. 
 
 VI. 2. j3. o. 
 
 Never. 
 
 VI. 3. i3. A. 
 
 Lazian, ma ma-zun, si 
 ga-zun, himuz a-zun+ 
 asere, my-ailing, i.e. / 
 ailed ; tkou^ he ailed. 
 
 V. 2. B. a. 
 
 Telugu, vaguta-nn, 
 vaguta-vu, vaguta-dn, 
 vaguta-mu, vaguta-ru, 
 vjiguta-iu, apeaking-I, 
 i.e. / speak ; thou, he, 
 we, you, they speak. 
 
 V. 3. B. A. 
 
 Never. 
 
 pana-kritjem, 
 tooth-ache. 
 
 Hung, fog-fajas, tooth- 
 ache. 
 
 Yakut e, kS^-u^ug ar, 
 winters end in, at the 
 end qf winter. 
 
 tas. kharakli, stone-eye 
 {spectacles). 
 
 2. Genitive-Adjectives : 
 
 Mandshu, irgeni amo, 
 populipater. 
 
 IV. 2. b. a. 
 
 Never. 
 
 IV. 3. b. A, 
 
 Never. 
 
 V. 1. B. a. 
 
 L a p p o n. denkewes- 
 almats, fat man. 
 
 Genit. denkewes-al- 
 mats a, fat man of. 
 Hung, kover -ember, 
 fat Tuan. 
 
 Genit. kover-embere, 
 fat man^s. 
 Turk, altyn-zengirira, 
 my golden chain. 
 
 Gen. altyn-zengiri- 
 mira, of my golden chain. 
 
 V. 2. B. a. 
 
 Hung. Present intrans. 
 lialKuk, halt.asz, hall, 
 hearing-Ij i.e. / am 
 hearing j thou, he 
 hears. 
 
 Yakute, ^ani.bin, ^anl. 
 gin, sanar, knowing-I, 
 i.e. I know i thou, he 
 knows. 
 
 V. 3. B. A. 
 
 Never. 
 
 VI. 1. j3. a. 
 
 Telugu, na-tandri, my 
 
 father. 
 U r a o n, im-baa, m. 
 
 father. 
 
 2. Genitive Adjectives. 
 
 VL 2. 0. o. 
 
 Never. 
 
 VI. 3. ^. A. 
 No. 
 
 Japhet. 
 
 Indo-European . 
 IV. 1. b. a. 
 S ns k. rSga-purusfaa 
 king's man. 
 
 2. Genitive-Adjectives : 
 Hind. KudA-kA beta, 
 God's son {divinus 
 Jilius). Kudakl mS, 
 God''s Tnolher {divina 
 mater). 
 Latin, Dei filius, Alius 
 Dei. 
 
 rv. 2. b. o. 
 
 Never. 
 
 rv. 3. b. A. 
 
 Never. 
 
 V. I. B. a. 
 
 S a n s k r t. mahd-deva, 
 Great-god. (of. mah4- 
 
 mas, magnify vie.) 
 en. maba-dev. 
 
 devasya. 
 
 VI. 1 jS. a. 
 
 No. 
 
 ((3«j3) Syr. tead-mort 
 
 ta, thine m,an of thee. 
 
 2. Genitive Adjeotivps : 
 Maudshu, mi-ni amo, 
 met pater. 
 miningge, meus. 
 
 VL 2. iS. a. 
 
 Never. 
 
 VI. 3. yS. A. 
 
 N 
 
 Gen. 
 
 Great-god's., 
 Greek, t^xi-xiraiv. 
 
 Anglo-Sax. sin-greoe, 
 
 Ever-green. 
 Ohg. ain-fluot, great 
 
 flood, Deluge^ 
 
 V. 2. B. a. 
 
 S n s k. Perfect. Atm.(-f 4) 
 dad.^, dadi.ahe, dad.e, 
 dadi.mahe, dadi.dhve, 
 dadi.re. 
 
 Present, dad.e,dat.se, 
 dat.te, dad.mahe, dad. 
 dhve, dada.te, taking-I, 
 i.e. / take, and {con- 
 tinue to) take. 
 
 Greek, Perf.'Pa.ss. SeS*. 
 
 fAeti, diho.tr6ti,^tio.Tce.t. 
 
 Fres. Pass. mo.fMtii 
 
 V. 3. B. A. 
 
 Never. 
 
 VL 1. j3. a. 
 
 Snsk. jnat-putra, tvat- 
 putra, tat-putra, my, 
 thy, his son. 
 
 2. Genitive Adjectives : 
 
 Asm4-kam pitd or dsm^- 
 
 kah pita, our father. 
 
 VI. 2. /3. o. 
 
 Never. 
 
 VI. 3. /3. A. 
 
 Snsk. /3A;3. Preterite: 
 [m] alip.am,[s]a lip. as, 
 [t]alip.at, my writing, 
 i.e. I wrote, thou, he 
 wrott.} 
 
234 
 
 SECOND 
 
 COMPARATIVE TABLE OP PERSONAL PRONOUNS, 
 
 ATTACHED TO 
 
 1. EGYPTIAN. 
 
 Nominative, a 
 Status subjectivus. b 
 
 [present] 
 Casus obliqutts. c 
 Status preEdicativus. d 
 
 1. cum nom. [possessive] 
 
 2. cum verb, [preterite] 
 
 2. HEBREW. 
 
 Nominative, a 
 Status subjectivus. b 
 
 [present] 
 Casus obliquus c 
 Status pnsdicativus. 
 
 1. cum nom. [possessive] 
 
 2. cum verb, [preterite] 
 
 3. SANSKRIT. 
 
 Nominative, a 
 Status subjectivus. b 
 [present] 
 
 Casus obliquus. c 
 
 Status prcBdicativus. d 
 
 1. cum nom. [possessive] 
 
 2. cum verbis, [preterite] 
 
 A. GREEK. 
 
 Nominative, a 
 Status subjectivus. b 
 
 [present] 
 Casus obliquus. c 
 Status prcedicativus . d 
 
 cum verbis [preterite] 
 
 5. LATIN. 
 
 Nominative, a 
 Status subjectivus. b 
 
 [present] 
 Casus obliquus- c 
 Status priBdicativus. d 
 
 cum verbis [preterite] 
 
 6. GOTHIC. 
 
 Nominative, a 
 Status subjectivus. b 
 
 [present] 
 Castis obliquus. 
 Status pnsdicativus. 
 
 cum verbis [preterite] deest. 
 
 First Person, 
 
 nuk. anuk. 
 
 Va. (the same as d) 
 
 n.a(n.u) (where 7, i.e. of me) 
 
 Oa-(u) (Me same as b) 
 v'en.a. (en.u.) the same as c. 
 
 dnoki. ani. 
 
 Vti. (Ethlop. ku.) 
 
 ah am. 
 Vmi. 
 
 ("mama, (mdmakas.) 
 (me. (madlyas.) 
 
 mad □. 
 
 tyai. 
 
 a/nou. (ifjcis.) 
 
 %°; 
 
 mei.(meus,) 
 Vm. 
 
 meiiia. (meins.) 
 
 Note. — Q, Nominal base. 
 Singular. 
 Second Person. I Third Person. 
 
 entek. 
 Vek. 
 
 Dek. 
 Ven.ek. 
 
 euta. 
 Vet. 
 
 net 
 
 \/en.et. 
 
 atdh. 
 tiV- 
 
 
 tvam. 
 
 at(ati). 
 tiv*!. 
 
 Dek. 
 Vt.Cka.) 
 
 tava. (tSvakas.) 
 te. (tvadlyas.) 
 
 tvad □- 
 Vs. 
 
 irou(tr6s.) 
 -Vs- 
 
 tu. 
 
 tui, (tuus.) 
 
 Vs. 
 
 thu. 
 
 theina. (theins.) 
 
 entuf. 
 Vet 
 
 entes. 
 \/es. 
 
 Qef.C-se.) nes. 
 •x/en-ef. Ven-es. 
 
 hu. 
 
 Do. 
 V. 
 
 hi. 
 
 tiV. 
 
 Qah. 
 V4h. 
 
 svayam. sa, sa, tad. 
 Vti. 
 
 sve. (svas.svakas.) 
 tasya. (tadiyas. ) 
 
 svan, tadn. 
 -Vt- 
 
 on. (r^fiff.) 
 -V(t.) 
 
 ipse, hie, haec, hoc. 
 Vt. 
 
 tui. (suus.) bujus. 
 
 silba. la, s6, tliata. 
 Vth. 
 
 seina. (seins.) this. 
 
235 
 
 APPENDIX. 
 
 AND OF PRONOMINAL PEEFIXES AND AFFIXES 
 NOUNS AND VEEBS. 
 
 V^, Verbal base. 
 
 First Person. 
 
 x/ea. 
 
 (n.en.) 
 
 Den. 
 V en. en. 
 
 anakhnu. (nakhnq.) 
 
 Qenu. 
 
 vayam. 
 -Vmas. 
 
 asmdkam. (dsm^kas.) 
 nas. (BBmadtyas.) 
 
 asmad □■ 
 
 
 nos. 
 Vmus. 
 
 nostri. (noster.) 
 
 vei». 
 
 unaara. (unsar.) 
 
 PlTJRAI/. 
 
 Second Person. 
 M. p. 
 
 emtuten. entuten. 
 Vten. 
 
 □'ten. 
 \/en.ten. 
 
 atem. 
 
 ateii. (atenah.) 
 
 □ kem. Qken. 
 
 V tem(Eth.kemmu)\/ten. 
 
 yGyam. 
 Vvas. 
 
 yushmdkatn.Cyaushni^kas.J 
 vas. (yushmadlyas.) 
 
 yushmad □• 
 
 VtS. 
 
 vos. 
 Vtie. 
 
 ve8tri.(vester.) 
 
 jus. 
 v'th. 
 
 izvara. (izvar.) 
 
 1 hird Person. 
 
 M. F. 
 
 entesen. sen. 
 
 Vsen. 
 
 □ sen. 
 V en. sen. 
 
 hem(hemah.) hen(henah.) 
 yiV"J. tiVn^h. 
 
 □ dm. nan. 
 
 svayam-te, tSs, Uni. 
 Vnti. 
 
 sve. (svas. svakas.) 
 teshfim. (tadlyas.) 
 
 svan- 
 
 <r<pus- 
 
 tadD. 
 -Vn. 
 
 0i, et,t, Toe. 
 
 
 ipsi. 
 V^nt. 
 
 ipsorum. 
 Vnt 
 
 silbai. 
 Vud. 
 
 hi, hse, hse. 
 hot urn. 
 
 thai, th&s, tha. 
 
 Sign of 
 Genitive §• 
 
 ne;theEgi/p- 
 tianpron, 
 
 demonst, and 
 relat-s Of. 
 
 Plural, 
 
 -s.-kas, 
 -iyas. 
 
236 
 
 7. CHINESE (Kuanhoa). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c. 
 Status prtsdicativus. d. 
 
 8. GYAMI. 
 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prcEdicativtts. d 
 
 9. TAI (Siamese). 
 Nominative, a. 
 Status subjeclivus. b 
 Casus obliquus. c 
 Status preedicativus. d 
 
 10. TAI (Laos). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 SiatTis prtBdicativus. d 
 
 11. TAI(Ahom). 
 Nominative, a 
 Status subjectivus. b. 
 Casus obliquus. c 
 Status prtBdicativus. d 
 
 12. TAI (Khamti). 
 Nominative, a 
 Status subjectivus. b. 
 Casus obliquus. c 
 Status preedicativus. d 
 
 13. TAI (Kassia). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status preedicativus. d 
 
 14. CHINESE (Kuwen). 
 Nominative, a 
 Status subjectivus. b 
 Casus obhquus. c 
 Status prmdicativus. d 
 
 16. TRANS-HIMALAYAN 
 (Tibetan spoken). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prtsdicativus. d 
 
 16. TRANS-HIMALAYAN 
 (Horpa). 
 Nominative, a 
 Status subjectivus. b 
 CosMi obliquus. c 
 Status prcBdicativus. d 
 
 JPzV^i Person, 
 
 ngo. (tsa.) 
 ngoV- 
 ngo-ti. 
 ngoQ. 
 
 gno. 
 
 gnov'- 
 
 gno-ti. 
 
 gnoD- 
 
 kha. 
 khaV- 
 khang-kha. 
 □ khang-kha. 
 
 ong (ku) 
 
 kau. 
 kauV- 
 
 kau. , 
 kau\/. 
 kau. 
 Gkau. 
 
 hang-man, ffii'Z (o/) Aer. 
 
 hang-pa, Zai/ {of) fish. 
 
 nga 
 ngav'. 
 jong-nga. 
 Djong-nga. 
 
 ukapa jongngi^aiAer o/w4-. 
 
 kakarteng jong umon, i^e 
 ncEWze o/^Ae ?naM. 
 
 ngo. 
 ngoV- 
 ngo-tui. 
 ngoQ. 
 
 gnyA. 
 giiya^. 
 gna. 
 gnaQ. 
 
 gna. 
 gna- 5. 
 
 Singular. 
 
 Second Person. 
 
 ni-ti. 
 
 niQ. 
 
 nly/. 
 ni-ti. 
 niD- 
 
 tua. (miing.) 
 tuaV- 
 khang-tua. 
 □ khang-tua. 
 
 mo. 
 mov'- 
 
 mau. 
 maiiv'- 
 mail. 
 Dniaii. 
 
 me. (pha.) 
 meV. 
 jong-me. 
 n Jong-me. 
 
 gliou. 
 g'houV. 
 g'hou-tci. 
 g'houD- 
 
 khye. 
 khye\/. 
 khe-yi. 
 khe-yiQ- 
 
 Third . 
 
 t'a. 
 t'aV- 
 t'a-ti. 
 t*aD- 
 
 tha. 
 
 thaV- 
 
 tha-ti. 
 
 than- 
 
 khon. (man.) 
 khony*. 
 khang-khon. 
 Qkhang-khon 
 
 tan. (man) 
 
 heu. 
 heuy'. 
 
 man. 
 manV- 
 man. 
 Qman. 
 
 jong-u. jong-kfi. 
 OJong-u. Qjong-kfi. 
 
 khi. 
 khiv*. 
 khi-tci. 
 khiQ- 
 
 khu. 
 khuv'- 
 kho-yi. 
 kho-yi □ 
 
 jya. 
 jya-a. 
 
237 
 
 First Person. 
 
 ngo-men. (tsa-men.) 
 
 ngo-meiiv'- 
 
 ngo-men-ti. 
 
 ngo-men n- 
 
 gno-me. 
 
 gno-me\/' 
 
 gno-me-ti. 
 
 gno-meCl- 
 
 rau. 
 
 raiLv^. 
 
 khang-rau. 
 
 Qkhang-rau. 
 
 rau. 
 rauV* 
 
 hau. 
 
 hauV* 
 
 hau. 
 
 Dhau. 
 
 ngi. 
 
 ngiv*. . 
 Jong-ngi. 
 Djong-ngi. 
 
 ngo-shu. 
 ngo-8hiiv/. 
 ngo-shu- tci. 
 ngo-shu D- 
 
 gnan-jo. 
 
 gnan-jo\/. 
 
 gnan-jo-yi. 
 
 gnan-jo-yi D- 
 
 gna-ni (gna-riggi) 
 gna-a-rigya. 
 
 Plural. 
 
 Second Person, 
 
 ni-men. 
 ni-menv''. 
 nl-men.ti. 
 ni-men □. 
 
 nl-me. 
 
 ni-meV- 
 
 ni-me-ti. 
 
 ni-meQ. 
 
 EU. 
 
 suV* 
 khang-6u. 
 
 O khang-8u. 
 
 khau. 
 khaiiv*. 
 
 mau-su 
 mau-suV* 
 man-su. 
 G mau-su. 
 
 phi. 
 phiV. 
 jong-phi. 
 Djong-phi. 
 
 g'hou-shu. 
 
 g'hou-shuv'. 
 
 g'hou-shu-tci. 
 
 g'hou-shuQ. 
 
 khen-jo. 
 
 khen-joV- 
 
 khen-jo-yi. 
 
 khen-jo-yiO' 
 
 ni-ni (riggi). 
 nii-rigya. 
 
 Third Person. 
 
 t'a-men. 
 t*a-menV. 
 
 t'a-men-ti. 
 t*a-men □. 
 
 tha-rae. 
 
 tha-meV. 
 
 tha-me-ti. 
 
 tha-me □. 
 
 khau-arai. 
 khau-arai\/. 
 khang -khau-arai. 
 D khang- khau-arai. 
 
 khreu. 
 khreuv*. 
 
 man-khau. 
 man-khauy'. 
 man-khau. 
 Qraan-khau. 
 
 ki. 
 
 kiV. 
 jong-ki. 
 
 Qjong-ki. 
 
 khi-shu. 
 khi-shuV- 
 khi-shu-tci. 
 khi-shu D- 
 
 kon-jo. 
 
 kon-joV- 
 
 khon-^o-yi. 
 
 khon-jo-ylQ. 
 
 ji-ni (ji-riggi). 
 ja-a-rigya. 
 
 Sign of 
 Genitive Sf 
 Adjective. 
 
 khang-. 
 
 jong-. 
 
 Elongation. 
 
 Plural. 
 
 -men. 
 (pel.) 
 (mei.) 
 
 BU. 
 
 -khau. 
 
 -shu. 
 
 (tchai.) 
 
 (teng.) 
 
 -JO. 
 
 (nam.) 
 (dag.) 
 (chag.) 
 
 -ni. 
 -rlggi. 
 
238 
 
 17. TRANS-HIMAI.AYAN 
 
 (Thochu). 
 Nominative, a 
 Status s//bjectivus. b 
 Casus ohhquus. c 
 Status -prmdicativus. d 
 
 18. TRANS-HIMALAYAN 
 
 (Gyarung). 
 Nominative, a 
 Status suhjectivvs. b 
 Casus obliquus. c. 
 Status prcedicativtts. d 
 
 19. TRANS-HIMALAYAN 
 
 (Manyak). 
 Nominative, a 
 Status objectives, b 
 Casus obliquus. c 
 Status prtEdicativus. d 
 
 20. TRANS- HIMALAYAN 
 
 (Takpa). 
 Nominative, a 
 Status subjectivus, b. 
 C(MM* obliquus. c 
 Status prtsdicativus. d 
 
 BHOTIYA. 
 
 21. SUB-HIMALAYAN 
 (Kenaverl-). 
 Nominative, a 
 Status subjectiuvs. b 
 Casus obliquus. c 
 Status prcedicativus. d 
 
 22. SUB-HIMALAYAN 
 
 (Serpa). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c. 
 Status prcsdicativus. d 
 
 23. SUB-HIMALAYAN 
 
 (Sunwar). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c. 
 Statics pra;dicativU3- (1 
 
 24. SUB^HIMALAYAN 
 
 (Gurung). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prtsdicativus. d 
 
 25. SUB-HIMALAYAN 
 
 (Magar). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prcsdicativus. d 
 
 2fi. SUB-HIMALAYAN 
 (Newar). 
 Nominative, a 
 Status subjectivus. b 
 
 CastM ohliqmis. c 
 Status prcedicativus. d 
 
 i'Vrs^ Person, 
 chi. (ka.) 
 ka-kchi. 
 
 gna. 
 
 gnaQ. 
 
 gne. (nye.) 
 gne-ku. 
 
 gna. (gnarung). 
 gnay'iing. 
 gnaring-i. 
 gnarung D' 
 
 gna. 
 
 gnaV. 
 
 gna-ti. 
 
 go. 
 gov- 
 a-ke. 
 a-ke D- 
 
 gna. 
 gnav/- 
 gna.la. 
 gna- la D- 
 
 gna. 
 gnaV- 
 gnou. 
 gnou Q. 
 
 
 jimho 
 jimhoD- 
 
 ji-gu 
 ji-guQ. 
 
 Singular. 
 
 Second Person. 
 kwa. 
 kwe-kchi. 
 
 nan re. 
 
 naV- (J. A. S. B., 
 1853, p. 29.) 
 
 niQ. 
 
 keot. (kherung.) 
 keotv/ uk. 
 keot-ki. 
 kherung Q. 
 
 kbyo. 
 
 khyo\A 
 
 khyo-tt 
 
 gai. 
 gaiV- 
 i-ke. 
 l-keD- 
 
 ken. 
 keny'. 
 ken-la. 
 ken-la □• 
 
 nang. 
 nangV. 
 nuwo. 
 nuwoQ- 
 
 chha. 
 chhaV- 
 
 chhang-gu. 
 chhang-guQ- 
 
 Third Person. 
 
 tha-cha. 
 
 tha-kchi. (kwana- 
 kchl ) 
 
 gnapos. (watu.) 
 
 waQ. 
 
 thi. 
 thi-i. 
 
 pe. 
 pe-ku. 
 
 phai ; te ; khong. 
 phaiV ung. 
 te. 
 phai Q. 
 
 khwo. 
 
 khwov*. 
 
 khwo-ti. 
 
 hari. 
 hariv'. 
 harea.ke. 
 harea-ke Q. 
 
 thi. 
 thV. 
 thi-la. 
 thi-la Q. ^ 
 
 hos. 
 hosv*- 
 hochu. 
 hochuD* 
 
 wo. 
 
 waya-gu. 
 waya-gu Q- 
 
239 
 
 First Person. 
 chi-ki. (cha-klar.) 
 chUkuk. 
 
 a-dur. 
 a-dur-i 
 
 gna-ra. 
 gna-rarku. 
 
 net. (gna-tamshe.) 
 nety ung. 
 gnaring-i, 
 gnaring Q, 
 
 m-rang. 
 
 ni-rangV- 
 
 ni-ra-ti. 
 
 go-vki. 
 go-vki\/. 
 go-ain-ke._ 
 go-ain-ke Q. 
 
 gm-mo. 
 gni-moy'. 
 gni-mo-Io. 
 gni-mo-lo D' 
 
 kan-kurik. 
 kan-kurikV- 
 kan-kurik- um. 
 kan-kurik-um D- 
 
 Ji-pmg. 
 |i-pingV. 
 
 ji-ping-gu. 
 ji-piiig-gu □. 
 
 Plural. 
 Second Person. 
 
 kwen[-ko (bwani-klar). 
 kwani-kuk. 
 
 Third Person. 
 
 tha-ko, (tha-klar.) 
 tha-kuk. 
 
 non-dur. 
 non-dur-i. 
 
 i-ra. 
 i-ra-ku. 
 
 keo2huk.(kherung-tamshe.) 
 keozhdkv' ung. 
 khering-i. 
 kherungQ. 
 
 khyo-rang. 
 
 khyo-rangV- 
 
 khye-ra-ti. 
 
 thi-dur. 
 thi-dur-i. 
 
 pe-ra. 
 pe-ra-ku. 
 
 Sign of 
 Genitive Sf 
 Adjective, 
 
 -kchi. -uk. 
 
 -ku. 
 
 gai-vki. 
 gai-vkiV. 
 gai-ain-ke. 
 gai-ain-ke Q* 
 
 ken-mo. 
 ken-moy'. 
 keme-mo-lo ? 
 keme-mo-lo Q. 
 
 nang-kurik. 
 nang-kurikv'- 
 nang-kurik- um. 
 nang-kurik-um Q. 
 
 chha-ping. 
 chha-ping\/, 
 
 chha-ping-gu, 
 chha-plng-gu Q. 
 
 wateshe. Cphai-tamshe.) 
 
 wateshey' ung. 
 
 te. 
 
 ■^atesheO 
 
 khwo-rang. 
 
 khwo-rangV. 
 
 khwo-ra-ti. 
 
 hare-vki, 
 hare-vkiV- 
 hari-ain-ke. 
 hari-aiu-ke Q- 
 
 tiii-rao. 
 thi-moV. 
 tha-me-la. 
 tha-me-la Q. 
 
 hos-kurlk. 
 hos-kiirikV- 
 a-kurik-um. 
 a-kurik-um Q. 
 
 wo ping, 
 wo 'pvng\/. 
 
 I wo-ping-gu. 
 I wo-ping-gu Q 
 
 Plural. 
 
 -ki. 
 -klar. 
 
 •ki.-i. 
 
 -ke. 
 
 tamshe« 
 
 -vki. 
 
 E 2 
 
240 
 
 27. SUB-HIMALAYAN 
 (Murnii). 
 Nominative, a 
 Status subjeclivus. b 
 Casus obliquus. c 
 Status pr^dicativus. d 
 
 28. SUB-HTMALAYAN 
 (Limbu). 
 Nominative, a 
 Status subjectivus. b 
 Cas7is obliquus. c 
 Status pr^dicativus- d 
 
 29. SUB-HIMALAYAN 
 
 (Kiranti). 
 Notninative. a 
 Status subjectivus. b 
 Casus obliquus. c. 
 Status prcedicaiivus. (i 
 
 30. SUB-HIMALAYAN 
 (Lepcha). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prcsdicaiivus. d 
 
 31. SUB-HIMALAYAN 
 (Bhutanes«). 
 . Notninative. ^ 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prtBdicativus. d 
 
 32. LOHITIC 
 
 * (Burmese spoken). 
 Nominative, a 
 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prcBdicativus. d 
 
 33. LOHITIC 
 
 (Dhimal). 
 Nominative, a 
 Status su'ijectivus. b 
 Casus obliquus. c 
 Status prccdicatiazis. d 
 
 34. LOHITIC 
 
 (Kachari-Bodo). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquns. c 
 Status presdicativus. d 
 
 35. LOHITIC 
 
 (Garo). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status pr(Edicativu$. d 
 
 i^iVs^ Person. 
 
 gnaV- 
 gna-la. 
 
 gna-laC- 
 
 inga. (eruga. C) 
 
 ingaV- 
 
 inga-in. 
 
 inga-inQ. (also as abbrevi- 
 ated prefix, J. A. S. B., 
 1853, p, 28.) 
 
 anka. 
 
 anka V . 
 
 angko. 
 
 angkoQ- (also as abbrevi- 
 ated prefix, J- A. S. B., 
 1853, pp. 28. 32. am-pa, 
 my father.) 
 
 go. 
 goV- 
 kaseusa. 
 kaseusa D* 
 
 gna. 
 gna*/- 
 gne-yi. 
 gne-yi Q^ 
 
 nga {superior). 
 
 ngaV- 
 
 nga-i (ngaha). 
 
 ngai Q. 
 
 ka. 
 
 kaV. 
 ka-ng. 
 ka-ng Q. 
 
 ang. 
 ang»/. 
 ang-ni. 
 ang-ni n* 
 
 ang (aoga. Rob ) 
 angV. 
 ang-ni. 
 ang-ni □■ 
 
 Singular. 
 Second Person. 
 
 'a.W. 
 ai-la. 
 'ai-laD- 
 
 khene. 
 kheneV- 
 khene-in. 
 khene-in Q. 
 
 khana. 
 khanav'- 
 amko. 
 amko □■ 
 
 hau. 
 hauv*. 
 hadosa. 
 hadosa □• 
 
 chhu. 
 chhu\^. 
 chhe-gii 
 chhe-giQ 
 
 meng (e^c/); men 
 {inferior) \ then. 
 the sajne. 
 men-i- 
 the same. 
 
 nang. 
 n ang-ni. 
 
 nang. (naa. R.) 
 nang-ni. 
 
 Third Person. 
 
 the 
 theV. 
 the-la. 
 the- la Q. 
 
 kbune. 
 khane\/. 
 khune-in. 
 khune-in Q. 
 
 moko. 
 mokov'. 
 moso. 
 moso Q. 
 
 he. 
 hV. 
 heu-sa. 
 heu-sa □• 
 
 kho, 
 kh5V. 
 kheu-gi. 
 kneu gi Q* 
 
 thu ; i (is) ; 
 (hie). 
 
 wa. 
 wa-ng. 
 
 bi. 
 
 bi-ni. 
 
 u. (aa. R.) 
 u-ni. (ua-ni. R.) 
 
241 
 
 jPirst Person. 
 
 gna-ni. 
 gaa-nj\/. 
 iu-na. 
 iii-na Q. 
 
 ani-ge. 
 ani-gev*. 
 aui-gen-in. 
 aui-gen-iu Q. 
 
 aiika-n. 
 aiika-nV> 
 ainko. 
 iiinko Q. 
 
 ka-yu (ki) 
 ka-yuV. 
 ka-yu pong-sa. 
 ka-yu pong-sa Q. 
 
 gna-cha. 
 giia-chaV* 
 gna-che-gi. 
 gua-che-gi Q. 
 
 nga-do. 
 nga-do-i. 
 
 kj -el. 
 ki-ug. 
 
 jang (jaiig-phur). 
 jaug-ni. 
 
 niQg. (chinga. R.) ; 
 ning-ni. (ching-ui. R.) 
 
 Plural. 
 Second Person. 
 
 ai-ni. 
 al-iiiy. 
 an-na. 
 an-na □ . 
 
 khen-ih*. 
 khen-ihV- 
 khen-ih-in. 
 khen-ih-in □• 
 
 khana-nln. 
 khana-ninv'. 
 am 110. 
 amno Q. 
 
 ha-yu. 
 ha-yuy/. 
 ha-yu i»ong-sa. 
 ha-yu pong-sa Q 
 
 kha-cha. 
 kha-chaV- 
 kheu che-gi. 
 kheu che-gi Q. 
 
 men-do. 
 men-do -i. 
 
 ny-el. 
 ni-ng. 
 
 nang-chur. 
 nang-chur-ni. 
 
 nanok. (na-si-mong. R.) 
 
 nanok-ni. (na-si-moiig-ni. 
 R.) 
 
 Third Person. 
 
 the-ni. 
 the-niv*. 
 then-iia. 
 then-naQ. 
 
 khun-chi. 
 khun-chlV- 
 khun-chi-in. 
 khun-chi-in □. 
 
 moko-chi. 
 moko-chiv'. 
 myaucho tnoyoso. 
 myauchu Q. 
 
 ho-yu. 
 ho-yuv'. 
 ho-yu pong-sa. 
 ho-yu pong-sa Q. 
 
 khoiig. 
 khongV. 
 khong-gi. 
 khong-gin. 
 
 thu-do. 
 thu-do-i. 
 
 ub-al. 
 ub-al-ko. 
 
 bi-chur. 
 bi-chur-ni. 
 
 wor,ok. (ua-madang. R.) 
 
 woiiok-ni. (ua-madang-ni. 
 R.) 
 
 I Sign of 
 Genitive ^ 
 Adjective. 
 
 -la. na. 
 
 R 3 
 
242 
 
 36. tOHITIC 
 
 (Changlo). 
 iTominative. a 
 Status subjectivus. b 
 Casus obliguus. c 
 Status pTtSdicativUS . d 
 
 37. LOHITIC 
 
 (Mikir), 
 Nominative, a 
 Status subjectivus. b 
 Casus obhquus. c 
 Status prcsdicativus, d 
 
 38. LOHITIC 
 
 (Dophlas). 
 NominaiivF . a 
 Statics subjectivus. b 
 Casus obliguus. c 
 Status pradicaiivus. d 
 
 39. LOHITIC 
 
 (Abor-Miri). 
 Miri (Robertsonj. 
 
 Nominative, a 
 Status subjectivus. b 
 Casus obliguus. c 
 Status subjectivus. d 
 
 40. LOHITIC 
 
 (Sibsagor-Miri). 
 Abor (Robertson). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliguus. c. 
 Status prcedicativus. d 
 
 41. LOHinC 
 
 (Singpho)/ 
 Nominative, a 
 Status subjectivus. b 
 Casus obliguus. c 
 Status pradicativus. d 
 
 42. LOHITIC 
 
 (Mithan-Naga, &c.) 
 Nominative, a 
 Status subjectivus. i> 
 Casus obliguus. c 
 Status pi (Edicutivus. d 
 
 43. LOHITIC 
 
 C Namsang-Ndga) . 
 Nominative, a 
 Status subjectivus. b 
 Casus obliguus. c 
 
 Status pr^dicativus. 6 
 
 44. LOHITIC 
 
 (Khyeng). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquns. c 
 Status prcEdicativus. d 
 
 J^iV.?^ Person. 
 
 jang. 
 jangv*. 
 jang-ga. 
 jang-ga □• 
 
 ne. 
 
 ne\/. 
 ne-ne. 
 
 ne-D-- (oi-ali, fiiVd'j nes?; 
 ne-sal, m^' work.) 
 
 ngo. 
 
 ngov'- 
 
 ngo-g. 
 
 ngo-g □. 
 
 ngoV. 
 
 ngo-ke. fngog. R.) 
 
 ngo-ke Q. 
 
 ngo. 
 n2;o\/. 
 ngo-kkef. 
 ngo-kke Q 
 
 ngai. 
 ngaiV- 
 nge-na. 
 ng^-na □- 
 
 ku (tail ; ni ; a) 
 kukiilie (tesei ; n>). 
 
 nga. 
 
 Vang. 
 
 nga-nang (irang). 
 
 kyi. 
 ki~ko. 
 
 SiKGULAH. 
 Second Person. Third Person. 
 
 nan . 
 nang-ga. 
 
 nang. 
 nang-iie. 
 
 no-g. 
 
 no (na). 
 
 ao-kke. (no-g. R.) 
 
 i-kke 
 
 nang (ni). 
 na-na. 
 
 nang (no), 
 nang. 
 
 nang. 
 v'o. 
 
 nang-nang • (ma- 
 rang). 
 
 maQ. 
 
 nang. 
 nang-ko. 
 
 dan. 
 dan-i 
 
 ma. 
 
 ma-£ 
 
 bu (bil). no. 
 bu-kke. (bii-g. R.) 
 
 ba. (bii. R.) 
 bQ-kke. 
 
 khi. 
 khi-na. 
 
 mih. (taupa ; pan ; 
 me.) 
 
 ate. 
 
 -v/a. (e)* 
 ati-eng (a-rang). 
 
 aQ. 
 
 ni. 
 ui-ko. 
 
243 
 
 First Person, 
 jang-tharache. 
 jang-thamche-ga. 
 
 a-H. 
 
 ngo-lu, 
 n«o-lu-g. 
 
 ngo-lu. 
 
 11 go. lu-ke. {ugo-lu-g. R. ) 
 
 n go- sin. 
 ngo-lii-kke. 
 
 a-kau (a-we)* 
 
 ni-ma. 
 ni-ma-nang. 
 
 kjn-ni. 
 ki-ni-ko, 
 
 Plukal. 
 Second Person. 
 nan-tharache. 
 nan-thamche-ga. 
 
 no-Iu. 
 no-lu-! 
 
 no-lu. 
 
 no-lu-ke. (no-lu-g. R.J 
 
 ao-lu-sin. 
 no-lu-kke. 
 
 nUtheng. 
 
 nl-khala (notoleli). 
 
 ne-ma. 
 ne-ma-naiig 
 
 nang-ni. 
 nang-ni-ko. 
 
 Third Person. 
 
 dan-thamche. 
 dan-thamche-ga. 
 
 ma-Iu. 
 ma-lu-g. 
 
 bii-Iu. 
 
 bu-lii-ke. (bii-lu-g. R.) 
 
 ii-Ilu. bii-lii. 
 bu-lu-kke. 
 
 tung-khala (tothete). 
 
 se-ning. 
 se-ning-nang. 
 
 ni-di (ni-li). 
 ni-di-ko. 
 
 Sign of 
 Genitive Sf 
 Adjective. 
 
 -nang. 
 -rang. 
 
 R 4 
 
244 
 
 45. 
 
 LOHITIC 
 
 (Kami). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus- c 
 Siaitis prisdicativus. d. 
 
 46. LOHITIC 
 
 (Tunglhu). 
 Nominative, a 
 Status subjectivus. h 
 Casus obliquus. c 
 Status prcedicaiivus. d 
 
 47. MUNDA. 
 
 Ho. (Tickell. A. S/B.ix.) 
 Nominative, a 
 Status subjectivus. b 
 Ca5U5 obliquus. c 
 Status prtEdicativus. d 
 
 48. MUNDA 
 (Sinhbhum-Kol). 
 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prcedicaiivus. d 
 
 49. MUNDA 
 
 (Soutal-Kol). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prcedicativus. d 
 
 .50 MUNDA 
 
 {Bliumij-Kol). 
 Nominatiiie. a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status preBdicativus. d 
 
 61. MUNDA 
 
 (Mundala-Kol). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status pra;dicdtivus , d 
 
 62. TAMULIC 
 
 (Canarese). 
 Nominative, a 
 Status subjectivus. b 
 
 Casus obliquus. c. 
 Status prcedicativus. d 
 
 53. TAMULIC 
 (Tamil). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 
 Status prardicaiivus. d 
 
 i^i>»i Person. 
 
 ka-i. 
 ka-i-un. 
 
 4iiig (ing). ^ 
 
 aing\/ ; or, aingvaing. 
 
 aingia (ing-aj. 
 
 aingia □ ; or abbreviated 
 
 prefix? (J. A. S. 13., 1853, 
 
 p. 28.) 
 
 aing. 
 iyan. 
 
 inge. 
 ingrea. 
 
 jhatana. 
 
 n5nu (ySn, y6n). 
 
 \/ene. Present. 
 
 Venu. Preter., FuL^' Neg 
 
 Venu. Second future. 
 
 nanna. 
 
 nSnD- 
 
 iiAn (yAn). 
 -v/en. 
 en-adu (fic?, 
 
 en D- 
 
 eiin-udeiya 
 
 Singular. 
 Second Person. 
 
 nan. 
 nan-un. 
 
 um. 
 umm-j 
 
 um. 
 umraa. 
 
 umge 
 ami. 
 
 am. 
 umma. 
 
 am. 
 am-atana. 
 
 nil! (ni). 
 
 VI. 
 
 Vi-, Sye. 
 niniia. 
 nln. Q 
 
 nl (un). 
 A/Ay. 
 I un-adu (nin-adu). 
 
 un a- 
 
 Third Person, 
 
 hana-i 
 hana-i un. 
 
 ayl(ayo). (ni, 
 
 this). 
 ay -a. 
 
 unea. 
 
 Qt. apa-»t, his fa- 
 ther. (J. A. S. B. 
 1853, p. 75.) 
 
 mm. 
 anner-dtana. 
 
 avanu, ava/u, adu. 
 
 V'ane. \/kl^. -y^'ade. 
 Vanu. \/a/u. \/Uu. 
 V^anu. Va/u. v^itu. 
 avana. ava/a. adera. 
 
 Ivan. ival. 
 \/kTi. Va'.. 
 ivan-adu 
 
 idu. 
 ■y/adu. 
 
245 
 
 First Person. 
 
 kfuchi. 
 ka-chl-un. 
 
 alle. 
 all^.a. 
 
 alle-a. 
 
 abusaban. 
 
 allege, 
 ahu-atana. 
 
 xikvu (ndm, fern). 
 
 ySvu. 
 Vevu. 
 nSmma. 
 nam Q. 
 
 Ddm fn^n-gal). 
 iiam-adu (en-gal), 
 nam □• 
 
 Plural. 
 
 Second Person. 
 nan-chi. 
 nan-chi-un. 
 
 appe. 
 app6-a. 
 
 appe-a, 
 
 inkoghi. 
 api-atana. 
 
 nivu(nlr, Ir). 
 
 \/lri. 
 
 nlmma. 
 
 nlixiD- 
 
 nir (nin-gal). 
 Vir-gU. 
 um-adu (un-gal). 
 
 um □- 
 
 Third Person. 
 hun-na (hani-chi). 
 hani-clii-un. 
 
 a-kd. 
 ako-a. 
 
 an-ko. 
 anko-atana. 
 
 V^re. 
 
 Vave. 
 
 v/ani. 
 
 V^vu. 
 
 V^l'U. 
 
 -y/avu. 
 
 a vara. 
 
 avugala 
 
 ivar (ivar-gal) ivei-gal. 
 V^ar-gal. {m. f.) \/:ina.. (n). 
 ivar-adu (ivar-gal-udeiya). 
 
 Sign of 
 Genitive ^ 
 Adjective. 
 
 -iidu. {id.) 
 udeiya. 
 (udfi, pro- 
 prium.) 
 
246 
 
 54. TAMULIC 
 
 (Telugu). 
 NoTnmative.'A 
 Status subjectivus. b 
 Casus ohliquus. c 
 Status prtedicatzvus. d 
 
 55. TAMULIC 
 
 (Malabar). 
 Nominative, a. 
 Status subjectivus. b 
 Casus obliquus. c 
 
 Status prcedicativus. d 
 
 56. TAMULIC 
 
 (Malayalim). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status praaiicativus. d 
 
 57. TAMULIC 
 
 (Gond). 
 Nominative, a 
 
 Status subjectivus. b 
 Casvs obliquus. c 
 Status prcedicativus. d 
 
 58. TAMULIC 
 
 (Brahvl?). 
 Nominative, a 
 Status subjectivus- b 
 Casus obliquus. c 
 Status prttdicativus. d 
 
 59. TAMULIC 
 (Curgi and Todava). 
 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. c 
 Status prcedicativus. d 
 
 60. TAMULIC 
 
 fUr4on-Kol). 
 NoTninative . a 
 S'atus subjectivus. b 
 Casus obliquus. c 
 Status prcsdicativus. d. 
 
 61. TAMULIC 
 
 RajinahSli-Kol). 
 Nominative, a 
 Status subjectivus. b 
 Casus subjectivus. c 
 Status prcEdicalivus . d 
 
 62. UGRIC 
 
 (Finnisli). 
 Nominative, a 
 
 Status subject/vu-:. b 
 Casus obliquus. c 
 Status prcedicativus. d 
 
 i^'/r^i Person. 
 
 nenu. 
 \/nu. 
 
 nd-yokka. 
 
 nan 
 
 nan (yan). 
 enn-udeyathu, en-athu. 
 
 gndn. 
 gaAn\/. 
 en -re. 
 
 ini-kuIIaQ. 
 Dative-j-uila. 
 
 nauna (nak). 
 
 y'an. 
 
 no-wa. 
 
 no-waQ. 
 
 Vt (ta), V^ (va). 
 
 kand, 
 
 kan^ D- 
 
 nan, one. (T) 
 en-na, en-na. (T) 
 
 enan. 
 
 en-ghi. 
 
 ira Q. (im-bas, mt/ father, 
 
 J. A. S. B. 1853, p. 32. ; 
 
 ing-kos, 7nt/ child). 
 
 ong-ki. 
 
 /mte.\ 
 la I ma. l 
 
 Vma. / 
 
 y/w (ni). 
 Dni. 
 
 Singular. 
 
 Second Person. 
 nlvu. 
 
 nl-yukka. 
 
 nlQ. 
 
 ni (nir). 
 
 umm-udiathu, um- 
 athu. 
 
 nl. 
 
 nlV- 
 nin-re. 
 
 nani-kiilla □. 
 
 ni-wa. 
 
 nl. 
 
 \/s (sa). 
 
 nin. ni. (T) 
 nin-na. nin-na.(T) 
 
 men. 
 nien-ghi. 
 
 nm. 
 Ing-ki. 
 
 sinae 
 
 v/t(s). 
 
 a si. 
 
 ii) 
 
 Third Person. 
 
 vadu. adi, 
 Vdu. V^i. 
 vkm. d^ni. 
 
 avan. aval, ahthu. 
 avan-udeyathu. 
 
 avan. aval. ada. 
 a van -re. 
 avanna+uHaQ , 
 
 \/ur. 
 wunna. 
 
 oi:od);CdAdad):eCe( 
 
 v/k. 
 
 o-aa. d^-nS. e-nd. 
 
 av. ad. (T) 
 
 ava-na. ada-na. (T 
 
 asan. 
 as-ghi. 
 
 ath. 
 ahi-ki. 
 
 ban (se ?) 
 \/.(hn).[v^l)i,^wi]. 
 Dnsa. Qia. 
 
247 
 
 First Person, 
 memu. 
 mi-yokka. 
 
 man. 
 
 nan->gal (ndm). 
 
 en-gal-udeyathUf em- 
 athu. 
 
 gnan-gal (n&m). 
 
 nan-galude (namm- 
 
 ude). 
 gnangal-kulla Q. 
 
 raak(wak). 
 mar, b^ore verbs. 
 
 iiiow-an (wo-man) 
 
 nan. 
 
 v'n in, na). 
 
 nanS. 
 
 eng.wom.(T) 
 en-gal-e. emmadd.{T) 
 
 nam (om). 
 nam-ki (emki). 
 
 me (met). 
 
 Vmme. 
 
 Qmme. 
 
 Plural. 
 
 Second Person. 
 
 mJru. 
 
 ml-yokka, 
 miQ. 
 
 nin-gal. 
 un-gal-udpyathu. um-athu. 
 
 nin-gal. 
 
 niu-ga de (uin-gal-ude). 
 
 ningal-kulla □• 
 
 ima-t (me-kum) 
 
 imar, before verbs. 
 v/ir. 
 mi- wan. 
 
 nunt. 
 Ve (ri;. 
 numS. 
 
 niDg. nimmai 
 nin-gal-e. nimma. 
 
 asu. 
 ass-ghi. 
 
 nma. 
 nim-ki. 
 
 te (tet). 
 
 Vtte. 
 
 ' O nne. 
 
 Third Person. 
 
 vkrn, 
 \/r\i. 
 vAri. 
 
 avi. 
 Vvi . 
 
 aver-kal, (atei), 
 one (?) 
 
 ava-r. ava. 
 
 ava-ru-de ava-yu-de. 
 
 •v/urg. 
 wurran. 
 
 ofk; ddfk"; efk. 
 
 V (re). 
 
 oitk. daftl efta. 
 
 aaabar (awar). 
 asaberi-kl. 
 
 he (het) (ne ?) 
 V[wat]t. (ht). 
 n nsa. Q Ba. 
 
 Sign of 
 Genitive^ 
 Adjective. 
 
 -di, 
 -yokka. 
 
 -athu. 
 'udeyathu. 
 
 -na. 
 •an. 
 
 -ghi. 
 
248 
 
 63. UGRIC 
 
 First Person 
 
 (Lappoiiian: Norwegian). 
 
 
 Nominative, a 
 
 mon. 
 
 Stalus subjectivus. b 
 
 -m. 
 
 Casus obliquus. c 
 
 
 Status preEdicativus. d 
 
 -m. 
 
 1. cum nom. 
 
 -m. 
 
 2. cum verbis [preterite] 
 
 
 64. UGRIC. 
 
 (Syrianian). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus. C 
 Status prcedicativus. d 
 
 65. UGRIC 
 
 (Tsheremissian). 
 Nominative, a 
 Status subjectivus. b 
 Casus obliquus- (geniiive.) c 
 Status pn^dicativus. d 
 
 66. UGRIC 
 (Ostiakian: Irtishian). 
 
 Nominative, a 
 Status subj:ctivus. b 
 Casus obliquus. (locat.) c 
 Status pra;dicativus. d 
 
 1. cum nom. 
 
 2. cum verbis transitivis 
 
 €7. UGRIC 
 
 (Hungarian). 
 Nominative, a 
 Status subjectivus. b 
 
 1. intransitive. 
 
 2. passive. 
 
 Casus obliquus. (accusat.) c 
 Status pnedicativus. d 
 
 1. cum nom. 
 
 2. cum verbis transitivis 
 
 68. UGRIC 
 
 (Mo.-dvinian). 
 Nominative. .i 
 
 Status subjectivus. (indefinite.) b 
 Casus obliquus. c 
 Status prcedicniivus. d 
 
 1. cum nom. 
 
 2. cum verbis. 
 
 69. SAMOIEDIC 
 
 (Dialects). 
 
 Nominative, a 
 
 Characteristic consonants of mis- 
 cellaneous pronominal suffixes. 
 
 me. 
 
 me-nam. 
 Dm- 
 
 mm. 
 \/xn.. 
 niin-in. 
 
 Dra- 
 
 ma. 
 
 •Jrti. 
 mana. 
 
 Dm- 
 Vm. 
 
 en (ma) 
 
 Vm. 
 engem-et. 
 
 Dm. 
 Vm. 
 
 mon. 
 Vn. 
 
 Dm (n). 
 
 Vraak. (1 + 2) me-^tu. 
 
 Vm am. (1+3) me+ille. 
 
 Vm isk. (1+2) me+vos. 
 
 O. man (mat). 
 J. man'. 
 K. man. 
 Ja. mod'l. 
 T. maiman. 
 
 [ '", b, p, 
 
 SlNGUI-AH. 
 
 Second Person. 
 
 ton. 
 -k. 
 
 -t. 
 -k. 
 
 te. 
 
 Vn- •/■ 
 te-nad. 
 
 □ d. 
 
 tin. 
 
 vt. 
 
 tin-in. 
 
 nt. 
 
 neii. 
 \/n. 
 nenna. 
 
 Dn. 
 
 \/sz, Vl- 
 tiged-et. 
 
 Dd- 
 Vd. 
 
 ton. 
 Vt. (k). 
 
 Dt. 
 
 Vt an. (2+ 1 ) 
 
 U-i-ego. 
 v/nz at. (3+2) 
 
 itlc+le 
 Vd ez. (2 pi. +3) 
 
 vobis+id. 
 
 tan. 
 pudar. 
 tan (than), 
 tod'i 
 lannan. 
 
 t,t',d,d',r, 1, Ir, n. 
 
 Third Person. 
 
 sya. 
 
 V... 
 
 sy-lan. 
 
 Q s. (\/s cum verbo 
 definito.) 
 
 tida. 
 
 Vs, Vz'e, V- 
 
 tida-u. 
 
 Dz'e. 
 
 tea. 
 
 Vd- (t). 
 teu na. 
 
 Dt. 
 Vt- 
 
 V-A/n. 
 
 ^/k. 
 
 dt-et. 
 
 DJa. 
 V'ja. 
 
 son. 
 \/s (zo). 
 
 Q zo (nzo). 
 y'a. 
 
 s/ze (3) id. 
 
 Vnk. (3+2 plur.) 
 id+vos. 
 
 tam, tap. 
 pu da. 
 di. 
 
 ni toda. 
 se te. 
 
 t, t%d, d', r. 
 
249 
 
 First Person. 
 
 mi. 
 
 -pa. 
 
 -mek. 
 -mek. 
 
 ml. 
 
 mi-an. 
 □ nu m. 
 
 ma. 
 
 Vna. 
 
 nia-mnan. 
 
 men. 
 meniia, 
 
 nu. 
 
 nfl. 
 
 Vnk. 
 i^/nk. 
 mink-et. 
 
 nnk. 
 v'juk. 
 
 m!n. 
 ■V^nok. 
 
 □ mok (nok). 
 
 v'm isk (1 +2) nos+tu. 
 V'mla(l+3)>K)S+ffl<. 
 
 me (ml). 
 
 man'a. 
 
 mi. 
 
 mo'di. 
 
 men. 
 
 PlCRAL. 
 
 Second Person. 
 
 \. 
 
 ■baettet (ppet). 
 
 -dek. 
 dek. 
 
 ti. 
 
 Vn ny d. 
 
 ti-au. 
 
 nnyd. 
 
 ta. 
 Vda. 
 t^mdan. 
 Dda. 
 
 nen. 
 
 V-Ja (ta). 
 neiia(?) 
 
 P den (ten). 
 Vden. 
 
 Vtok. 
 Vtok. 
 titek.et. 
 
 Otok. 
 y'jatok. 
 
 tin. 
 
 Vdo (nk). 
 
 Qnk. 
 
 Vd ezr. (?) 
 
 te. 
 
 pudara. 
 
 s'i. 
 
 to'di. 
 
 ten. 
 
 Third Person. 
 
 nya. 
 
 vnys. y's ny s. 
 
 ny-lan. 
 
 D ny s. 
 
 Vs't.Vt. [be. be-»]. 
 nina-n. 
 
 Qs't. 
 
 teg. 
 teg na. 
 
 V^nak, 
 Vnak. 
 dk-et. 
 
 DJok. 
 Vjak. 
 
 sin. 
 
 Vt, ^81(2). 
 
 nst. 
 
 Sign of 
 Genitive ^ 
 Adjective. 
 
 Plural. 
 
 -k (t). 
 
 -k (t). 
 
 Vnze. 
 
 -k(t). 
 
250 
 
 
 
 SlNGUlAR. 
 
 
 70. TATARIC 
 
 i^iV^i Person. 
 
 Second Person. 
 
 Third Person. 
 
 (Castren). 
 
 
 
 
 Nominative, a 
 
 ben, men, min. 
 
 sen, sin, an. 
 
 ol, o, kini. 
 
 Status subjectivus. b 
 
 V'man (pan-bao). 
 
 Vsan. 
 
 ^/.(sin, in imperat). 
 
 cum verbis. 
 
 
 
 
 Casus obliquus. c 
 
 ben-i, etc. mcz. 
 ben-inki, etc. mewj. 
 
 
 
 Status prcEdicativus. d 
 
 
 
 
 1. cum nomin. 
 
 Dm- 
 
 Qfi. 
 
 Q i (in). D si (sin). 
 
 2, cum verbis [preterite] 
 
 Vm. 
 
 Vfi. 
 
 V- 
 
 71. TATARIC 
 
 
 
 
 (Yakute: Boehtling ). 
 
 
 
 
 Nominative, a 
 
 min. 
 
 an. 
 
 kini, bu {hie), «1 
 {iUe). 
 
 Status subjectivus^ b 
 
 Vben (pen, men). 
 
 Vghen (gen, ken). 
 
 y.- 
 
 Casus obliquus. c 
 
 miy-iana {meus). 
 
 ay-iana {tuus). 
 
 kin-iana (suus), 
 mane {hunc), 
 onu iillum). 
 
 Status prcEdicativus . d 
 
 
 
 
 1. cum nom, [possessive] 
 
 □ m. 
 
 DH. 
 
 Dta(ten).Oa(en). 
 
 2. cum verbis [perfect] 
 
 V'-m. 
 
 Vn. 
 
 Va. 
 
 72. TATARIC 
 
 
 
 
 (Osmanli). 
 
 
 
 
 Nominative, a 
 
 ben. 
 
 sen. 
 
 ol (0), bu (flic). 
 
 Status subjeqfivus, b 
 
 Vim. 
 
 Vsen. 
 
 V"* 
 
 Casus obliquus. c 
 
 benim. (genitive.) 
 
 senin. 
 
 onnn 
 
 Status prtedicativus. d 
 
 
 
 
 1. cum nom. [possessive] 
 
 2. cum verbis, [preterite] 
 
 Dm. 
 Vm. 
 
 On. 
 Vn- 
 
 ni(in). D si (sin). 
 
 73. MONGOLIC 
 
 
 
 
 (Buriatian). 
 
 
 
 
 Nominative, a 
 
 bi. 
 
 s'i, lyi. 
 
 Vs', Vc 
 
 s'in, c^n, f s'ini. 
 
 ene. 
 
 Status subjectivus. b 
 
 Vp, ^/m. 
 
 Casus obliquus. c 
 
 min ((mini, mei). 
 
 eneni. 
 
 
 J mni. 
 
 tin. ) sni. 
 
 
 
 IT' 
 
 jsn. 
 
 
 Status prtBdicativus. d 
 
 Dm. 
 
 (s. 
 
 Qs'. DC. 
 
 nn(Oni). 
 
 74. MONGOLIC 
 
 
 
 
 (Sokpa). 
 
 
 
 
 Nominative, a 
 
 mi (bi, abu). 
 
 cha. 
 
 til a. 
 
 S^d^Ms subjectivus. b 
 
 
 
 
 Casus obliquus. c 
 
 mini. 
 
 chini. 
 
 thani. 
 
 5/a/Mj preedicativus. d 
 
 
 
 
 75. TUNGUSIC 
 
 
 
 
 (Mandshu). 
 
 
 
 
 Nominative, a 
 
 bi. 
 
 si 
 
 j^ 
 
 Status subjectivus. b 
 
 biV- 
 
 siV- 
 
 V- 
 
 Cff«K5 obliquus, c 
 S/a<MS prcEdicativus. d 
 
 fmin-i (or mi-ni ?) »k;. 
 Imin-ingge (meus). 
 mini □■ 
 
 r sin-i. 
 \ sin-ingge. 
 
 fini. 
 
 I in-ingge. 
 
 76. TUNGUSIC 
 
 
 
 
 (Nyertshinsk). 
 
 
 
 
 Nominative, a 
 
 bi. 
 
 si, s'i. 
 Vsi. Vs. 
 sin. 
 
 (i) nunan. 
 (v'n). 
 vin. 
 Qn. 
 
 5ia/tts subjectivus. b 
 
 Vbi, Vu, Vf. 
 
 Ca5MS obliquus. c 
 
 min. 
 
 StofK« prcBdicativus. d 
 
 (Dbi,nu,DO. 
 
 (Dsi.Ds). 
 
251 
 
 First Person. 
 
 bis, bister, bisigi. 
 Vbis (pis, mis). 
 
 □ bis (pis, mis). 
 Vbis (pis, mis). 
 
 bisigi. 
 
 yTjet (pet, met). 
 bi«-iana. 
 
 □ bet. 
 ylaet. 
 
 biz (bizler). 
 
 ^iz. 
 
 bizum (bizlerin). 
 
 □ -miB. 
 
 bida. 
 
 Vbida(bda, mda). 
 manai. 
 
 D manai (Dmnai). 
 
 be (miise) 
 
 be\/. 
 
 fmen-i. 
 
 I men-ingge. 
 
 Plural. 
 Second Person. 
 
 sis, sislur, siler, asigl. 
 Vsis (ser). 
 
 □ nis. 
 Vnis. 
 
 'isigi. 
 
 Vghet (get, ket). 
 as-iana. 
 
 □ ghet. 
 y.get. 
 
 siz (sizler). 
 
 \/siz. 
 
 sizin- 
 
 □ -nis. 
 vTiiz. 
 
 ta. 
 
 Vta, \/t. 
 tanai. 
 
 Qtaiiai. 
 
 chini ? 
 
 sue. 
 Buev'- 
 
 f suen-i. 
 
 ( suen-ingge. 
 
 bu. 
 
 s'i. 
 
 (\/W\itL.) 
 
 (Vsun). 
 
 mun. 
 
 sun. 
 
 Qwun. 
 
 nsun. 
 
 Third Person. 
 
 onlar, onnar, kinilar. 
 Vlar, Vsinler, in imper. 
 
 □ lary. 
 Vlar. 
 
 kinilar (balar (^0. 
 
 V^lar. 
 kin-ian-nar-a. 
 
 □ lara. 
 \/-lara. 
 
 onlar, bunlar. 
 Vlpr. ^ 
 onlarun- 
 
 Sign of I 
 Genitive^ Plural 
 Adjective. 
 
 □leri(Ierin).Oi(in)-n8i(sin). 
 Vler. 
 
 V- 
 
 □ u(Dni). 
 
 thani ? 
 
 tcbe. 
 
 tche V- 
 
 f tchen-i. 
 
 I tchen-ingge. 
 
 (c'e) nunar, 
 (Vtin) VI- 
 e'en. 
 
 □ tin. 
 
 -sa, ta, 
 
 ri. (sei, 
 
 urse, 
 
 jergi, 
 
 turoen, 
 
 gemu.) 
 
252 
 
 THIRD 
 
 COMPARATIVE TABLE OF THE NUMERALS 
 
 
 I. 
 
 II. 
 
 III. 
 
 IV. 
 
 1. Chinese 
 
 M 
 
 eul 
 
 san 
 
 ES6 
 
 2. Gyami - 
 
 iku 
 
 liangku (ar) 
 
 sanku 
 
 siku 
 
 3. Kong-Chinese - 
 
 Canton (spoken). 
 
 yut 
 
 ni 
 
 sam 
 
 si 
 
 4, Tai Siamese 
 
 nflng 
 
 song 
 
 sam 
 
 si 
 
 5. Tai.— Ahom 
 
 ling 
 
 sang 
 
 sam 
 
 lii 
 
 6. Tai.— Loos - 
 
 niing 
 
 song 
 
 sam 
 
 a 
 
 7. Tii.—Khamti 
 
 niing 
 
 song 
 
 sam 
 
 si 
 
 8. Tai Kassial 
 
 wci 
 
 ar 
 
 lai 
 
 sau 
 
 9. Tai.— SAuK 
 
 Tenasserim. 
 
 nein 
 
 htsong 
 
 htsan 
 
 htse 
 
 wuku 
 ung 
 
 VI. 
 
 lu 
 
 leuku 
 
 lok 
 
 ha 
 
 hok 
 
 ha 
 
 ruk 
 
 ha 
 
 hok 
 
 ha 
 
 hok 
 
 san 
 
 hinriu 
 
 ha 
 
 hoht 
 
 10. Trans-Himalayan. — Tibetan chil! 
 
 (spoken.) 
 
 11. Trans-Himalayan. —/forpa ra 
 
 N.W. Tibet. Bucharia. 
 
 12. Trans-Himalayan.— T^ocAu-Stftin ari 
 
 N. E. Tibet China. 
 
 13. Trans-Himala — Gyarung-Sifan kati 
 
 N.E.Tibet. China. 
 
 H. Trans-Himala Manyak-Sifan tabi 
 
 N. E. Tibet. China. 
 
 15. Trans-Himalayan.— ZTaArprt 
 
 West of Kwombo. 
 
 16. Su B- Himalayan.— .fff mi weri 
 
 Setlej basin. 
 
 17. Sub-Himalayan. — -Sar/Jd - 
 
 West of Gandakfean basin. 
 
 18. Sob-Himalavan — Sunwar 
 
 Gandakgan basin. 
 
 19. Sub-Himalayan Gurung 
 
 Gandakean basin. 
 
 thi 
 
 kanes 
 
 nyi 
 nishi 
 
 sum 
 
 ihyi 
 
 gna 
 
 thu 
 
 6U 
 
 hla 
 
 gwe 
 
 chho 
 
 kshiri 
 
 gzhare 
 
 ware 
 
 khatare 
 
 kasam 
 
 kadi 
 
 kunggno 
 
 kutok 
 
 sibi 
 
 rebi 
 
 gnabi 
 
 trubi 
 
 sum 
 
 pli 
 
 liagne 
 
 kro 
 
 sum 
 
 rhi 
 
 gna 
 
 tuk 
 
 sum 
 
 2hyi 
 
 gna 
 
 tuk 
 
 sang 
 
 le 
 
 gno 
 
 ruk 
 
 song 
 
 pli 
 
 gna 
 
 tu 
 
253 
 
 APPENDIX. 
 
 IN NINETY-SEVEN LANGUAGES. 
 
 VII, 
 
 VIII. 
 
 IX. 
 
 X. 
 
 1. tsi 
 
 pa 
 
 kieu 
 
 ehi (fiun) 
 
 S. chhiku 
 
 paku 
 
 chyuku 
 
 ishsa 
 
 3. chhat 
 
 pat 
 
 kau 
 
 sap 
 
 4. cbet 
 
 pet 
 
 kau 
 
 Bip 
 
 6. chit 
 
 pet 
 
 kau 
 
 sip 
 
 6. tset 
 
 pet 
 
 kau 
 
 sip 
 
 7. tset 
 
 pet 
 
 kau 
 
 sip 
 
 6. hinian 
 
 prah 
 
 kandai 
 
 shi pon 
 
 9. tsit 
 
 tet 
 
 kaut 
 
 tsit 
 
 kad wei kad ar 
 
 XX. 
 
 c. 
 
 M. 
 
 eul shi 
 
 pe 
 
 tsian 
 
 airsa 
 
 ipe 
 
 - 
 
 — 
 
 pak 
 
 chin 
 
 ye flip 
 
 voi 
 
 phan 
 
 sau 
 
 pak 
 
 — 
 
 sau nung 
 
 hoi 
 
 — 
 
 eau 
 
 pak 
 
 heug 
 
 ar phon 
 
 shi spah 
 
 ehi hajar 
 
 htsong 
 
 hpat 
 
 . 
 
 10. dun gye guh 
 
 11. zne rhieg go 
 li, stare khrare rgure 
 
 13. kushnei oryet 
 
 14. skwibi tibi 
 
 15. nia 
 
 16. dun 
 
 17. dyun 
 
 18. chani 
 
 19. nil 
 
 gyet 
 gya 
 gye 
 yoh 
 pre 
 
 chuh chub ehi ■ chu nyi nyi.chu gyathamba tong 
 
 (thamba) 
 
 6ga 
 hadure 
 
 kungga sih 
 gubi chechibi 
 
 dugu 
 
 gu 
 guh 
 guh 
 kub 
 
 pchi 
 
 naska rhya 
 
 gninaso akshi 
 
 kinnis si parye 
 
 nachabi teje 
 khaU _ 
 
 chuthamba chuchik chuni nishu gyathamba tong 
 
 chuh — — nyi shu gya — 
 
 sa shi — — khatka swaika •» 
 
 chuh — — kuti — — ? 
 S 
 
254 
 
 
 I. 
 
 II. 
 
 III. 
 
 IV, 
 
 V 
 
 VI. 
 
 SUB-HlM4L4YAN — Magar - 
 Gandakean basin. 
 
 kat 
 
 nis 
 
 song 
 
 buli 
 
 banga 
 
 - 
 
 SUD-HlMALiYAN — Newdr - 
 
 Between Gand. & Kosean bas. 
 
 cbhi 
 
 nl 
 
 son 
 
 pi 
 
 gna 
 
 khu 
 
 Sub-Himalayan.— jT/Mrmi' 
 
 Between Gand. and Kosean bas 
 
 ghrik 
 
 gni 
 
 6om 
 
 bD 
 
 gna 
 
 dhu 
 
 SUB-HlMALAVAN.— ZimW 
 
 thit 
 
 nyetsh 
 
 syum sh 
 
 li sh 
 
 gnash 
 
 tuksb 
 
 Kosean basin. 
 
 SuB-HiMALA'YAN. — Kirantt 
 Kosean basin. 
 
 Sub-Himalayan. — Lepcha 
 Tisht^an basin 
 
 Sub- Himalayan, — Bkutanese 
 Manasfean basin. (?) 
 
 Sub- Himalayan. — Chepang 
 Nepal -Terai. 
 
 ektai ? hasat ? samya laya 
 
 kat nyet sam phaH 
 
 chi nyi sum zhi 
 
 yazho nhizho sumzho ploizho 
 
 gnaya tukya 
 
 pha gnon tarok 
 
 gna dhu 
 
 pumazho krukzho 
 
 28. JjOUMIC. — Burmese 
 
 Burmah and Arakan. 
 
 29. LOHiTic Bhim/il 
 
 Between Konki and Dhorla 
 
 30. LoHiTic. — Kackari-Bodo - 
 Migrat. 88° to 93^0 and 25° to 27°. 
 
 31. I.QHITIC. — Gaj'O - - , 
 90° to 91° E. long. 25° to 26° N. lat. 
 
 32. LoHiTic. — Changlo 
 
 91° to 92° E. long. 
 
 33. LoHiTic — Mikir 
 
 Nowgong 
 
 34 LoeiTic. — Dophla 
 
 92° 50' to 97° N. lat. 
 
 35. ItOHnic. —Miri 
 
 94° to 97° E. long, (?) 
 
 36. LoHiTic— .45or-M/n 
 
 37. LOHITIC JbOT' 
 
 97° to 99° East long. 
 58. LoHiTic Sibsagor-Miri 
 
 39. LoiiiTic. - 
 
 27° to 28° North lat. 
 
 40.homTic.—Naga Tribes 
 
 93° to 97° E. long. 23° N. lat. 
 (Mithan). E. of Sibsagor. 
 
 41. LoHiTic — Naga Tribes 
 
 (Namsang), 
 
 42, LoniTic. — Nagii tribes 
 
 Nowgong. 
 
 tit (ta) 
 
 Tihit 
 
 thong 
 
 le 
 
 nga 
 
 khyauk 
 
 e-long 
 
 nhe-long 
 
 sum lang 
 
 dia long 
 
 nalong 
 
 tulong 
 
 che 
 
 nai (gni) 
 
 tham 
 
 bre 
 
 ba 
 
 ro 
 
 sha 
 
 t. 
 
 gini 
 
 gi thara 
 
 bri 
 
 bonga 
 
 dok (krok> 
 
 thur 
 
 ngik ching 
 
 sam 
 
 phi 
 
 nga 
 
 khung 
 
 ichi 
 
 hini 
 
 ba tham 
 
 phi li 
 
 phong 
 
 thorok 
 
 aken 
 
 ani 
 
 a am 
 
 apH 
 
 ango 
 
 akple 
 
 ako 
 
 a ni ko 
 
 a am ko 
 
 a pi ko 
 
 a ngo ko 
 
 a keng ko 
 
 ako 
 
 aniko 
 
 aomko 
 
 apiko 
 
 pilingoko 
 
 akeko 
 
 ako 
 
 ani 
 
 angom 
 
 api 
 
 pilango 
 
 akyc 
 
 atero 
 
 ngoye 
 
 auma 
 
 apic 
 
 iingo 
 
 akunge 
 
 aima 
 
 nkhong 
 
 ma sum 
 
 meli 
 
 manga 
 
 kru 
 
 anyi 
 
 1 zam all 
 
 vanthe vanyi van ram beli 
 
 kalang anna a&am pazr 
 
 aga arok 
 
 banga irok 
 
 pungu tarok 
 
so. 
 
 VII. 
 
 VIII. 
 
 IX. 
 
 X. 
 
 21. 
 
 nhe 
 
 chya 
 
 gun 
 
 sanho 
 
 22. 
 
 nis 
 
 pre 
 
 kuh 
 
 chiwai 
 (kun) 
 
 23. 
 
 nush 
 
 yetsh 
 
 phang 8h 
 
 thi bong 
 
 94. 
 
 bhagya 
 
 reya 
 
 phangya 
 
 kip 
 
 25. 
 
 ka kyok 
 
 kakeu 
 
 ka kyot 
 
 kaU 
 
 26. 
 
 dun 
 
 gyt> 
 
 g" 
 
 cha tham 
 
 255 
 
 XI. 
 
 katlp 
 
 — sang sanho gun sanho — 
 
 (nie) (satchi) 
 
 — nhi 6hu bokal gna — 
 
 — ni bong thi bunggip — 
 
 nyct'tp kha kat kha pha 
 gnon 
 
 — nyi sho kh6 gna 
 
 (khcchik) 
 
 27. chanazho prapzho takuzho gyibzho 
 
 28. 
 
 khunnhit 
 
 shyit 
 
 29. 
 
 nliilong 
 
 yelong 
 
 30. 
 
 Slii 
 
 jat 
 
 31. 
 
 sni 
 
 chet 
 
 32. 
 
 2um 
 
 yen 
 
 33. 
 
 thorchi 
 
 nirkep 
 (10.2) 
 
 31. 
 
 kanag 
 
 plag nag 
 
 35. 
 
 ki nit ko 
 
 pi ni ko 
 
 36. 
 
 kunitko 
 
 punitko 
 
 37. 
 
 konange 
 
 pini 
 
 38. 
 
 kiinnide 
 
 pinye 
 
 39. 
 
 sinit 
 
 matsat 
 
 40. 
 
 anath 
 
 achet 
 
 11. 
 
 ingit 
 
 isat 
 
 42. 
 
 tanet 
 
 tc 
 
 ko she she tit 
 
 kuhalong telong — 
 
 chku ji — 
 
 shku skang chish^ 
 
 gu se (song) song thur 
 
 chirkep kep kepaichi 
 (10.1) 
 
 kayo rang rang la 
 akin 
 
 ko nang u ying ko u ying ko 
 
 ko a ko 
 
 ko nangko uyingko — 
 
 kinide iiyinge — 
 
 konange u y i n ge ^ 
 
 tsekhu si si ai 
 
 aku ban — 
 
 she nhit nhit she taya 
 — elong bisa na bisa 
 
 — 
 
 chokai ba 
 (bisha che) 
 
 bisha ba — 
 
 chigini 
 
 chiskang 
 (rung) 
 
 rung bonga — 
 
 song nyik 
 
 khai thur 
 
 khenga nisi nik ting 
 
 ching 
 
 
 dang khaJse 
 
 kepahini 
 
 kepa kep 
 (ingkol) 
 
 phar — 
 
 rang la 
 
 rang chang 
 
 — — 
 
 ani 
 
 
 
 u ying ko 
 
 u ying an 
 
 _ _ 
 
 aniko 
 
 iko 
 
 
 — irlingko 
 
 uying uying 
 
 anyiko uyingko 
 
 si nkhong 
 
 khun 
 cha 
 
 ikhu 
 
 taku 
 
 hing 
 
 ichi vantlie ichi vanyi ruaknyi chathe cha ichi 
 
 s 2 
 
43. LoHiTic— ^a^a tribes 
 Tengsa. 
 
 41. LoHiTic. — Naga tribes 
 
 Tablung, N. of Sibsagor. 
 
 45. LoBiTic. — Nagtt tribes 
 
 Khari. Jorhat. 
 
 46. LoHiTic Naga tribes 
 
 Angami, South. 
 
 47. IjO^Ync.—Kuki 
 
 N. E. of Chittagong. 
 
 48. LoHiTic. — Khyeng {Shyu) 
 
 19° to 21° N. lat. Arakan. 
 
 49. LOHITIC— ^flTK* 
 
 Kuladan R. Arakan. 
 
 50. LoHiTic — Kumi 
 
 Kuladan R. Arakan. 
 
 51. LoHiTic Shendus 
 
 22° to 230, and 93^ to 940. 
 
 52. LoHiTic — Mru 
 
 Arakan. Chittagong. 
 
 53. LoHiTic. — Sak 
 
 Nauf River, East. 
 
 54. LoHiTiG — Tunglhu 
 
 Tenasserim, 
 
 1. II. III. 
 
 khatu annat asam 
 
 lem 
 
 256 
 
 IV, 
 
 phale 
 
 pili 
 asam phali 
 
 sii deh 
 
 katka nika tumka lika 
 
 nhat pan nhi thum Ihi 
 
 ka tun ma li 
 
 cha ih 
 
 akhet anne 
 po kane 
 
 ha ni 
 
 ha nhu 
 
 mekha » meny 
 loung pre 
 su war nein 
 
 tum 
 
 pa lu 
 
 V. VI. 
 
 phungu thelok 
 
 nga vok 
 
 phanga tarok 
 
 pangu soru 
 
 rungaka ruka 
 
 nghau sauk 
 
 »ang nga ta u 
 
 pan 
 
 ta ru 
 
 me thao me pulll mepa me churru 
 
 shun ta li ta nga ta ru 
 
 thin pri nga khyouk 
 
 thung lit ngat ther 
 
 55. MuNDA. — Ho 
 
 Kolehan. 
 
 56. MuNDA. — Sinhbhum Kol 
 
 Chyebossa. 
 
 57. MuNDA. — Sontal 
 
 Chyebossa. 
 
 58. MuNDA — Bhumij 
 
 Chyebossa. 
 
 59. MuNDA. — Mundala 
 
 Chota Nagpur. 
 
 miad barria 
 
 mi barria 
 
 midh barria 
 
 moy barria 
 
 mia "baria 
 
 appia upunia moya turuia 
 
 apia upunia moya turia 
 
 pi a ponia monego- turm 
 
 tang 
 
 apia upunia monaya turuya 
 
 apia upnia moria turia 
 
 60. TAMULJC. — Canarcse 
 
 ondu 
 
 erarfu 
 
 muru 
 
 nalku 
 
 ayidu 
 
 aru 
 
 61. TAMVhic. — Tama 
 
 onru 
 
 iranrfn 
 
 munru 
 
 nalu 
 
 anju 
 
 aru 
 
 62. TiMULic— TV^w^ 
 
 oka 
 
 renrfu 
 
 murfu 
 
 nalugu 
 
 ayidu 
 
 aru 
 
 63. Tamulic. — Jl/ti/flft«r 
 
 ondu 
 
 irandu 
 
 mnndu 
 
 nalu 
 
 inthu 
 
 aru 
 
 64. Tamulic. — Malatjalam 
 
 onna 
 
 renrfa 
 
 munnar 
 
 nala 
 
 anja 
 
 ara 
 
 Q5. Tamulic— Gond 
 
 uiidi 
 
 ranu 
 
 munu 
 
 nalu 
 
 aaiyan 
 
 sarong 
 
 66. Tamulic. —5ra//&( 
 
 asit 
 
 irat 
 
 muoit 
 
 Sk. char 
 
 panj 
 
 shash 
 
 67. Tamulic. — Tuluva 
 
 onji 
 
 erarf 
 
 muji 
 
 nalu 
 
 ayinu 
 
 aji 
 

 
 
 
 
 257 
 
 
 VII. 
 
 VlII, 
 
 IX. 
 
 X. 
 
 XI. 
 
 43. 
 
 thanyet 
 
 thesep 
 
 thaku 
 
 thelu 
 
 - 
 
 U. 
 
 nith 
 
 thath 
 
 thu 
 
 pan 
 
 - 
 
 45. 
 
 tani 
 
 sachet 
 
 teku 
 
 tarah 
 
 - 
 
 IS. 
 
 thene 
 
 thctha 
 
 thaku 
 
 kurr 
 
 - 
 
 7. 
 
 sarika 
 
 riktka 
 
 koka 
 
 sumka 
 
 _ 
 
 ♦ 
 
 
 
 
 
 
 48. 
 
 she 
 
 eat 
 
 ko 
 
 ha 
 
 - 
 
 49. 
 
 sari 
 
 ka ya 
 
 tako 
 
 haeuh 
 
 - 
 
 50. 
 
 sam 
 
 taya 
 
 ta kau 
 
 bau 
 
 - 
 
 51. 
 
 me sharri 
 
 me charla 
 
 me chuku 
 
 me hra 
 
 hlekha 
 
 52. 
 
 ra nhit 
 
 ri yat 
 
 taku 
 
 ha 
 
 - 
 
 S3. 
 
 thani 
 
 a tseit 
 
 tafu 
 
 sisu 
 
 - 
 
 51. 
 
 nwot 
 
 that 
 
 kut 
 
 tahti 
 
 _ 
 
 XX, 
 
 C. 
 
 machi 
 
 mesung 
 phungu 
 
 - 
 
 - 
 
 makhi 
 
 rukra 
 
 maku 
 
 kre 
 
 M. 
 
 — kur klaat ■» 
 
 — ku suh ta ra — 
 
 — a pum re chum wari — 
 hie ny meku ya kha sho kha 
 
 — pi ra mi — — 
 
 — hun taya -. 
 
 — he taloyeu — 
 
 55. aya irilia 
 
 56. iya irlia 
 57< iair iral 
 
 58. Sk. sath ath 
 
 59. Sk. sath ath 
 
 noko 
 
 gel 
 
 gelea 
 gel 
 das 
 dasgo 
 
 gelmiad gelbarria hissi 
 — — hissi 
 
 mi sow 
 moy hissi 
 monay hisii 
 
 60. elu (yelu) entu ombhattu hattu (pattu) 10+1 10+2 pat (ippatu) nuru 
 
 61. ezhu 
 
 ettu 
 
 onbailu 
 
 patta 
 
 — 
 
 62. eda 
 
 enimidi 
 
 tommidi 
 
 padi 
 
 — 
 
 63. elu 
 
 ettu 
 
 onpathu 
 
 pat thu 
 
 - 
 
 64. ezha 
 
 e«a 
 
 ombada 
 
 patta 
 
 — 
 
 65. yenu 
 
 anamur 
 
 urmah 
 
 pada 
 
 - 
 
 6S. haft 
 
 hasht 
 
 nuh 
 
 dah 
 
 yazda 
 
 67. al 
 
 ename 
 
 orambo 
 
 pattu 
 
 s 3 
 
 irupadu nuru 
 
 iruvai nuru 
 
 irupathu nuru (vanda) 
 
 Iruvada nura 
 
 bi&a nur 
 
 bifit Bad 
 
 irvo nuru 
 
258 
 
 68. Tamulic. — Toduva - 
 
 69. Tamulic. —Z/raon-fto/ 
 
 70. Uo-Ric. -^ Finnish 
 
 71. Ugric. — Esthonian - 
 
 72. Vo.'Bac. — Lapponian - 
 
 73. Ughic. — Syrianian - 
 
 74. Ugric. — Tsheremissian 
 
 75. Ugeic. — Mordvinian - 
 
 76. Vgric. ~ Ostiakian - 
 
 77. Vaaic. — Hungarian - 
 
 78. Uqric. — Fogulian - 
 
 79. Samoiedian- - - - 
 
 80. TdTARic. _ Yakut 
 
 81. Tataric. — Uigur 
 
 82. TkTKViic.— Tshuvash- 
 
 83. Tataric. — O^maTi/i 
 
 84. MoNGOLic— 0/0* 
 
 85. MoNGOLic — Sokpa 
 
 N.E.? Tibet 
 
 86. MoNGOLic. — j^i'maA 
 
 87. TuNGUSic — il/antisAu 
 
 88. Caucasc. — Z-flzian 
 
 89. Caucasic. — Suanian 
 
 90. Caucasic. — Mingrelian 
 
 91. Caucasic — Gijorg-ifftt 
 
 92. Caucasic. — -f46c/ifl«rtn 
 
 I. 
 
 TI. 
 
 III. 
 
 IV. 
 
 V. 
 
 VI. 
 
 won 
 
 edd 
 
 minn 
 
 nonk 
 
 yajj 
 
 orr 
 
 unta 
 
 enotan 
 
 manotan 
 
 nakhotan 
 
 Sk. panje 
 
 se 
 
 yksi 
 
 kaksi 
 
 kolme 
 
 nelja 
 
 viisi 
 
 kuusi 
 
 (iks (ut8) kaks (kats) 
 
 kolm 
 
 nelU 
 
 wiis 
 
 kuus 
 
 akt (oft) 
 
 kvekte 
 
 kolm 
 
 nielj 
 
 Tit, 
 
 kot 
 
 otik 
 
 kyk 
 
 kujim 
 
 njolj 
 
 vit 
 
 kvait 
 
 ik 
 
 kok 
 
 kum 
 
 nil 
 
 vis 
 
 kut 
 
 viiike (va 
 
 ,) kavto 
 
 kolmo 
 
 nile 
 
 Tate 
 
 koto 
 
 it(i.ja) 
 
 kat 
 
 cbudem 
 
 njeda 
 
 vet 
 
 chut 
 
 egy 
 
 kettS 
 
 harom 
 
 negy 
 
 ot 
 
 hat 
 
 akva (va; 
 
 ) kit 
 
 korom 
 
 nila 
 
 at 
 
 kot 
 
 op 
 
 side (siri) 
 
 nar 
 
 tet 
 
 samlik 
 
 mat 
 
 bir 
 
 ikki 
 
 UB 
 
 tuort 
 
 bias 
 
 alta 
 
 bit 
 
 iki 
 
 utsh 
 
 tort 
 
 bish 
 
 alty 
 
 per 
 
 ikke 
 
 wisse 
 
 dwata 
 
 pilik 
 
 olta 
 
 bir 
 
 iki 
 
 uc' 
 
 dort 
 
 besh 
 
 alti 
 
 nike 
 
 khoyor 
 
 gurban 
 
 durbun 
 
 tabun 
 
 zurgan 
 
 nege 
 
 hoyur 
 
 korba 
 
 tirba 
 
 thaba 
 
 chorka 
 
 nikka 
 
 koyar 
 
 ghorban 
 
 dorban 
 
 tabun 
 
 jolan 
 
 emu 
 
 jue 
 
 ilan 
 
 duin 
 
 sunja 
 
 ningun 
 
 ar 
 
 zur 
 
 jum 
 
 otch 
 
 chut 
 
 ash 
 
 eshchu 
 
 ieru 
 
 semi 
 
 wooshthch 
 
 wochushth 
 
 uskhwa 
 
 arti 
 
 shiri 
 
 6umi 
 
 otchi 
 
 ■ chuthi 
 
 apchshui 
 
 crthi 
 
 ori 
 
 sami 
 
 othchi 
 
 chut hi 
 
 ekhwssi 
 
 aka 
 
 wiba 
 
 cliiba 
 
 plishiba 
 
 chuba 
 
 fba 
 
VIII. 
 
 IX. 
 
 X. 
 
 XI. 
 
 ett 
 
 onbod 
 
 pott 
 
 - 
 
 ate 
 
 no 
 
 dan 
 
 
 VII. 
 
 70. seitsemiin kahdeksan ybdeksan kyrame yksitoista 
 (10.2) (10.1) nen kymmenta 
 
 XX. 
 
 ivvod 
 
 C. 
 onnur 
 
 kaksi eata 
 
 kymmcnta 
 
 71. seitse 
 
 kattesa 
 
 Uttesa 
 
 kUmine 
 
 73. Sec 
 
 kaktse 
 
 aktse 
 
 lokke 
 
 73. sizim 
 
 kokjaamyg 
 (10-2) 
 
 okmyB 
 (10.1) 
 
 das 
 
 74. Sim 
 
 kandaxse 
 (10.2): 
 
 endexse 
 (10.1) 
 
 lu 
 
 75. tabet 
 (sabet) 
 
 nlda (nit) 
 
 arjon 
 
 jon 
 
 76. het 
 
 nyolcz 
 (10.2) 
 
 kilencz 
 (10.1) 
 
 tiz 
 
 77. siu 
 
 sindet 
 
 chasawat 
 
 luzeyu 
 
 78. sisem 
 
 kavkso 
 
 TOikse 
 
 kamen 
 
 79. satta 
 
 aghes 
 
 tOghUB' 
 
 uon 
 
 80. yidi 
 
 aekiz 
 
 toghuz 
 
 on ■ 
 
 81. eat 
 
 njalalu 
 
 antalu 
 
 lava 
 
 SK sicce 
 
 takkyr 
 
 tuhur 
 
 wonna 
 
 83. yedi 
 
 sekiz 
 
 dokuz 
 
 on 
 
 81. dolon - 
 
 naiman 
 
 yese 
 
 arban 
 
 S5. tolo 
 
 dasotik dafikyk kyzj sjo 
 
 luat iktat luat kok- koklu syde 
 tat 
 
 j a chat jon katchat chus sot 
 jon 
 
 tiz enegy tizen ketto husz szaz 
 
 uon bir uon ikki «Urba bus 
 
 onbir oaikki igirmi yus 
 
 au kulplu — kit kuiplu kus 
 
 woni per won ikke sirim sije 
 
 onbir oniki y igirmi yuz 
 
 arban 
 nike 
 
 arban khorin 
 
 khoyor 
 
 chovo 
 
 86, 
 
 jurghan 
 
 - 
 
 — 
 
 87. 
 
 nadan 
 
 jokon 
 (10.2) 
 
 onyun 
 (10.1) 
 
 88. 
 
 shkit 
 
 ovro 
 
 cchoro 
 
 89. 
 
 ishkwid 
 
 ara 
 
 cchara 
 
 90. 
 
 Bhqwithi 
 
 ruo 
 
 cchoro 
 
 91. 
 
 Ehwidi 
 
 rwa 
 
 zehra 
 
 92. 
 
 biahba 
 
 aaba 
 
 shba 
 
 Juan ]uan emu juan jue orin tanggo 
 
 wit witwar witzur oc osh 
 
 iehsth ieshth eshchu ieshth ieru ieruieshth ashir 
 
 withi — -~ etshi oshi 
 
 athi — — ozi assi 
 
 thwaba shweiza ihwewa eshwa shke 
 s 4 
 
 tubat 
 
 tiisacja 
 
 (sjurs) 
 
 tesensa 
 ming 
 
 pin 
 
 bing 
 
 shilia 
 athaa 
 
260 
 
 I. II. HI. IV. V. VI. 
 
 93. Bask ...-•. bat bi hiiur lanr bortj sei 
 
 M. CoFTio .... . va enous somnt ftov (ftu) tiv (tu) sov 
 
 93. Hebhew 6khad shnayiro 6hlofih4h arbighSh khamlshSh shishAh 
 
 96. Pehlevi (Coins) - - - achad larein talata atba khomasha shate 
 
 97, SiNiawT ..... elsas dviu trayas iatviras panfe ibai 
 
261 
 
 vn. VIII. IX. X. XI. XII. XX. c. 
 
 93. zazpi zortzi bcderatzi hamar hamelza hamabi hogoi ehun 
 
 M. 
 nulliun 
 
 sho 
 
 94. sashf flhmun psis ni6(; metva metsnous ^uot she 
 
 95. shibgh&h shmon&h tishgh^h ghasSr&h akhadghds&r 8hnighdsS.r ghesrim in&^h§ i^lSph 
 
 96. fiheba tomena tisha ashra yaj deh duajdeh vUt sat — 
 
 97. sapta *^ ash/au nava da^a ekdda^a dvada^a vin^ati satam fahasrsm 
 
ANALYTICAL INDEX. 
 
 Page 
 
 Introductory Eemarks ----- 3 
 
 First Chapter : The Results of Turanian Philology : 
 First Section : History of Turanian Philology : 
 
 § 1. Gyarmathi - - - - - 10 
 
 § 2. Klaproth, Eemusat, Arndt - - - 11 
 
 §3. Rask .--.-. 12 
 §4. Schott, Castren - - - - - 13 
 
 § 5. Von der G-abelentz - - - - 18 
 
 § 6. BoehtUngk - - - - - 19 
 
 Second Section: General Division of Languages into 
 
 Family, Nomad, and State Languages - 21 
 
 Third Section : Mutual Relation of the three Forms of 
 Language, Progression and Retrogression : 
 Humboldt, Bunsen - - - - 23 
 
 § 1. Character of Family Languages - - - 24 
 
 § 2. Character of Nomad Languages - - - 25 
 
 § 3. Character of State Languages - - - 26 
 
 Fourth Section : General Features of Nomad or Turanian 
 Languages : 
 §1. Integrity of Roots - - - - 26 
 
 § 2. Formative Syllables felt as distinctive Elements - 28 
 § 3. Facility in producing new Forms - - 30 
 
 § 4. Scarcity of irregular Forms - - 30 
 
 §5. Rapid Divergence of Dialects - - - 31 
 
 §6. Contrast between the Progress and Growth of 
 
 Turanian and Arian Languages - - 33 
 
 Fifth Section : On the Principles of Formation and Deriva- 
 tion in Turanian Languages : 
 § 1. Scarcity of Synonymes and Homonymes - - 35 
 
 § 2. Adjectives, Substantives and Verbs, not always 
 
 distinct - - - " - 36 
 
264 
 
 Page 
 
 § 3. Pronominal Affixes, Subjective and Predicative - 38 
 
 § 4. Means of distinguishing Nominal and Verbal 
 
 Bases in Turanian Languages - - - 43 
 
 § 5. Means of distinguishing Nominal and Verbal 
 
 Bases in Arian Languages - - - 44 
 
 § 6. Means of distinguishing Nominal and Verbal 
 
 Bases in Semitic Languages - - - 46 
 
 § 7. The three different Directions of Grammar, Tura- 
 nian, Semitic and Arian, represented by the 
 three Sons of Feridun, Tur, Silim, and Irij - 50 
 
 A. Tur - - - - 50 
 
 B. Silim- - - - - - 52 
 
 C. Wj 52 
 
 D. The Descendants of Tur, divided according to 
 
 their Employment of the Pronominal Affixes 54 
 
 Sixth Section : Etymological Peculiarities of the Turanian 
 Languages : 
 § 1. Radical meaning generally discernible - 60 
 
 § 2. Scarcity of ancient Words common to all Tura- 
 nian Languages, and identical in Form and 
 Meaning - - - - - 60 
 
 § 3. Turanian Numerals - - - - 64 
 
 § 4. On Phonetic Corruption - - - - 66 
 
 § 5. On scarce Words - - - 70 
 
 Seventh Section .- On Turanian Languages approaching to 
 Arian Type : 
 § 1. Arian Elements in Hungarian, Turkish, Finnish - 71 
 § 2. Turanian Elements in Sanskrit - - - 73 
 
 § 3. Ascending Scale in the Turanian Languages - 74 
 Eighth Section: Evidence of the common Origin of the 
 
 Turanian Languages summed up - - - 76 
 
 Second Chapter : The Turanian Character of the Tamulic 
 Languages : 
 First Section : The Arian Settlers and Aboriginal Races 
 
 of India . - - - . 80 
 
 Coroll. : Historical Traces of Nishadas, or Aboriginal 
 
 Races of India - - - - - 83 
 
 Second Section : Ethnology versus Phonology - - 89 
 
265 
 
 Page 
 
 Third Section : Subdivision of the Nishada, or Aboriginal 
 
 Languages of India - - - 93 
 
 Fourth Section : The Bhotiya Class : 
 
 § 1. Ethnological Evidence - - - - 97 
 
 § 2. Historical Evidence - - - - 97 
 
 § 3. Geographical Evidence - - - - 99 
 
 § 4. Phonological Evidence - - - - 103 
 
 Coroll. : Trans-Himalayan Dialects - - - 105 
 
 Fifth Section: Further Extension of the Bhotiya Class, 
 and its Subdivision into Sub-Himalayan (Gan- 
 getic) and Lohitic Dialects ... 109 
 §1. Lohitic Dialects - - - - -111 
 
 § 2. General Coincidences between the Sub-Himalayan 
 (Gangetic) and Lohitic Divisions of the Bhotiya 
 Branch - - - - 121 
 
 § 3. Distinctions between Sub-Himalayan (Gangetic) 
 
 and Lohitic Dialects .... 124 
 
 Sixth Section : Tai Class : 
 
 §1. Survey of Tai Languages ... i26 
 
 § 2. Relation of the Tai to the Lohitic Languages, and 
 their Connection with the Bhotiya Class and 
 Chinese - - - - - 130 
 
 § 3. Bhotiya and Tai Numerals - - 136 
 
 Seventh Section : The Malay Languages : 
 
 § 1. Formal Coincidences between the Malay and Tai 
 
 Languages - - - - 143 
 
 § 2. Humboldt's and Crawfurd's Views on the Lan- 
 guages of Polynesia and the Indian Archipelago 153 
 
 Eighth Section : Tamulic Class : 
 
 § 1. Early traces of the Tamulic Nishadas - - 169 
 
 § 2. Geographical Distribution of the Tamulic Nishadas 172 
 
 §3. Separate Class of Munda Dialects - - 175 
 
 § 4. Languages belonging to the Tamulic Class - 178 
 
 § 5. Character of the Tamulic Class of Languages - 179 
 
 Ninth Section: Comparison of the Tamulic and Ugric 
 
 Languages - - - " - 184 
 
266 
 
 Page 
 
 213 
 
 Conclusion : The possibility of a common Origin of Lan- 
 guage -...-. 
 
 Appendix I. Comparative Table of Subjective and Predi- 
 cative Pronouns ----- 229 
 
 Appendix II. Comparative Table of Personal Pronouns 
 and of Pronominal Prefixes and Affixes attached to 
 Nouns and Verbs ----- 234 
 
 Appendix III. Comparative Table of the Numerals in 
 
 Ninety-seven Languages - - . - 252 
 
 THE END. 
 
 LoNDuN ; 
 
 A a^,l li.A.Si.oTi'iMi.u 
 Ncw-Ktnjct-Sqa.ii-L\ 
 

 .j^^f^f^: w-^ 
 
 m 
 
 wi-^ 
 
 tk 
 
 
 
 f^ i 
 
 
 
 ii'