UC-NRLF 929 ageoftheYiie-chi 1917 t . MAIN lo-bcythians by ivn-iwfcp LAUFER, Ph.D. Curator, Department of Anthropology, Field Museum of Natural History \ ]ee Announcement on first page inside The Language of the Yiie-chi or Indo-Scythians by . v : BERTHOLD LAUFER, Ph.D. Curator, Department of Anthropology, Field Museum of Natural History See Announcement on first page inside TI fE have imported from Tokyo a font of Chinese-Japanese type, consisting of approximately twenty thousand characters, and are prepared to handle, with the com- petent assistance of a Japanese compositor, the printing of books and other publications requiring the use of these characters. This installation was made at the sugges- tion of Dr. Berthold Laufer, Curator of An- thropology at the Field Museum of Natural History, and the following pages by Dr. Lau- fer show the type as adapted for sinological publications. We solicit especially the printing of books, magazines, and catalogues of libraries, collec- tions, and exhibitions. R. R. Donnelley & Sons Company 731 Plymouth Court Chicago 1 1 A NNO UNCE MENT The Language of the Yiie-chi or Indo-Scythians by BERTHOLD LAUFER, Ph.D. Curator, Department of Anthropology, Field Museum of Natural History CHICAGO R. R. DONNELLEY tf SONS COMPANY 1917 FIFTY COPIES PRINTED FOR THE AUTHOR The Language of the Yiie-chi or Indo-Scythians BY BERTHOLD LAUFER The question of the nationality of the ancient Yiie-Si is still unsettled. It is known that KLAPROTH first classified them with Tibetans, but subsequently became converted to the theory of their Indo-European origin, identifying them with the Goths. 1 The Pan-Turks who have done so much mischief to the history of Central Asia did not fail to claim the Yiie-c"i as their property. 2 This speculation is exploded not only by the very remains of the Yiie-ci language itself, but also by the formal statement of the Chinese annalists to the effect that the Yue-c"i were different from the Hiufi-nu; they belonged to the group of Hu, that is, Iranians. 3 Most writers on the Yiie-ci (and there is a goodly number of them) did not commit themselves to any opinion as to the ethnical position of the tribe. 4 Nationality is based on language: I propose to examine the few remains of the ancient Yue-& language (that is, in times prior to the foundation of the Indo-Scythian empire) preserved to us in the records of the Chinese and to offer some con- clusions with regard to the position of their language. When in A.D. 87 the king of the Yiie-ci asked for a Chinese princess in marriage, he sent as gift to the Emperor Can of the Han dynasty precious jewels and two kinds of animals hitherto unknown to the Chinese, Si ("lion") and fu-pa. 5 It is a common experience that the 1 Tableaux historiques de 1'Asie, pp. 132, 287-289. It is regrettable that F. JUSTI in his history of Iran (Grundriss, Vol. II, p. 489) still speaks of "the Tibetan Yiie-ci or Tochar," and that even to E. H. MINNS (Scythians and Greeks, p. no) they "appear rather to have been nomad Tibetans." Polyandry is not ascribed to the Yiie-ci in any document, as asserted by Minns. The Tibetan hypothesis has been well refuted by O. FRANKE (Zur Kenntnis der Turkvolker und Skythen, pp. 25-27). 2 F. HIRTH, Nachworte, p. 48; H. G. RAWLINSON, Bactria, p. 128; A. STEIN, Khotan, p. 50 ("the Yiie-ci probably spoke a language of the Turkl- Mongolian family"). 3 Hou Han su, Ch. 117, p. 27 b. 4 In regard to the older theories, which are all defective and inaccep table, see E. SPECHT, Journal asiatique, 1883, nov.-dec., p. 320; it is superfluous to discuss these anew in the present state of science. 5 Hou Han su, Ch. 127, translated by E. CHAVANNES, T'oung Pao, 1906, p. 232. 3676^.4. 4 THE kANGTJAGE Otf* THE YUE-CHI OR INDO-SCYTHIANS Chinese, whenever foreign products were brought to them for the first time, adopted together with these their foreign designations. Thus it is in the present case: Si andfu-pa are actual words received from the language of the Yue-Si. 1. flip and subsequently $$, Si, *'i, lion. On a former occasion I remarked that this word originally hailed from some East-Iranian language and was transmitted to China through the medium of the Yue-Ci. 1 This opinion should now be modified by the formula that the word *s'i, si or se, actually represents a Yiie-cl word with the meaning "lion," and that this Yue-ci word is closely related to its Iranian congeners. 2. $ iHfu-pa, *fu-bwa5, fu-bwal, fubal. As is known, this word has been identified by A. v. GuxscHMiD 2 with Greek (3ovf3d\ls or /3ofy3aXis, 3 but he has merely added the Greek word in parenthesis to fu-pa by way of explanation without discussing the philological basis of the case. First of all, it must be stated that *fubal is the Yiie-Si designation of an animal, and that this word may be related to 0ou/3aXts, in the same manner as other words in Indo-European languages. Certainly *fubal is not a Greek loan-word in Yue-5i. Moreover, the animal of the Yiie-ci did not represent the same species as the bubalis of the ancients, as plainly follows from a close comparison of the classical and Chinese traditions. The bubalis of the ancients has been identified with Bubalis mauretanica of northern Africa with long tail and short, lyre- shaped antlers, as well as with other kinds of antelope. 4 Aeschylus 5 is the first author to speak of "the young bubalis serving as food to the lion" (KeovToxoprav @ovpa\iv veairepov). Herodotus (IV, 192) places bubalis among the animals occurring in the Libyan desert, and Polybius (XII, 3, 5) praises their beauty. 6 The point of interest is that in the opinion of the ancients the lion and the bubalis were arch- enemies; they were often represented jointly on engraved gems. 7 This notion of a contest between the two creatures may have been prevalent also in the minds of the Yue-i, and their gift to the Chinese Court 1 T'oung Pao, 1916, p. 81. 2 Geschichte Irans, p. 140. 3 Wrongly written by him /8ou/3a\os (that is, Bos sylvestris, urochs). 4 O. KELLER, Antike Tierwelt, Vol. I, p. 294. 6 Fragm. 322 Nauck. 6 See further ARISTOTLE, Hist. an. (ed. of Aubert and Wimmer, Vol. I, p. 64); AELIAN, Hist, an., xiv, 14; PLINY, vm, 15. 7 IMHOOF-BLUMER and KELLER, Tier- und Pflanzenbilder auf Miinzen und Gemmen, Plate XVII, 43. THE LANGUAGE OF THE YUE-CHI OR INDO-SCYTHIANS 5 savors strongly of a political allegory (the weak swallowed by the powerful). Yet the *fubal of the Yiie-ft was an animal different from the bubalis. Certainly the Yiie-& had not exported it from northern Africa, but it was an antelopine species indigenous in the steppes of Central Asia. According to the Han Annals, the fu-pa occurred in the country Wu-yi-san-li^-AJ^ljgl, 1 and together with lions (or a lion) in A.D. 87 was also sent as tribute from Parthia $ Jj, (*An-sik, Ar-sik), on which occasion it is described as having the shape of a lin ($$> but without antlers. 2 In a late dictionary, the Er ya i of the twelfth cen- tury, the fu-pa is defined as "resembling a stag, and being provided with a long tail and a single horn." The word *fubal, accordingly, was not only Yue-c"i, but also belonged to the speech of the Parthians. Again, the affinity of the Yiie-ci language points to Iran. 3. Another Yiie-ci animal-name is handed down, not in the official annals, but in the Hilan cun ki & ff* ffi written by Kuo $|5 (his personal name is unknown) in the fifth century or earlier. This is the Yue-c"i term for the "ox," transcribed in Chinese by means of the character ]& ki, anciently *g'iep. The text of the Huan cun ki, as far as I know, is not preserved, and we have to rely on extracts from it given by later writers. The fact that the name of the animal was ]& is guaranteed by the very careful work T'ai p'in huan yu ki z published by Yo Si fj| _fe in the latter part of the tenth century. The Chinese ascribe a miracle to this peculiar cattle-breed of the Yiie-ci : three or four pounds of its flesh may be sliced off, yet the wound will heal in the course of a day (in other texts, on the following day) , and the animal is then restored to its normal size. 4 In the most detailed version of the story that I have been able to trace 5 it is added under the name of the Hilan Zuh ki, "Chinese who 1 Regarded by CHAVANNES (T'oung Pao, 1904, p. 555) as a transcription of Alexandria (*U-yir-san-ri) and identified with Strabo's Alexandria in Aria. 2 Hou Han su, Ch. 118, p. 4 (see CHAVANNES, Toung Pao, 1907, p. 177); repeated in Tai p'in huan yu ki, Ch. 184, p. 6b. 3 Ch. 80, p. 7. 4 Cf., for instance, Tai p'in yu Ian, Ch. 900, p. 2b, with the misprint R for 2&- As will be seen from the T'ai p'in huan yu ki (I. c.}, the spurious work Po wu ci erroneously ascribes this ox to the district Yue-sui j$ fl| in Se-c'uan. The Pen ts'ao kan mu (Ch. 51 A, p. -7) cites the "ox of the Yiie-ci" as a special variety, and simply quotes the text of the Huan cun ki, but without giving the native name. Some texts say that the animal occurs in the country of the Ta Yue-ci (Tai p'in yu Ian writes ^ ;) and the western Hu JR j9 (that is, Iranians), others omit the latter term. Sun Yin 5fc US, in his / wu ci jfe tify ;, adopted a humane attitude toward the story, and had the tail ten pounds in weight of the Yue-ci ox cut off and restored. 6 Tu Su tsi ten, sub SJ. Here the name of the animal is given as Jfc, "the ox ki of the day," with reference to its recuperative powers gained in a day. 6 THE LANGUAGE OF THE YUE-CHI OR INDO-SCYTHIANS entered this country [Ta Yue-Si] saw the ox without being aware of the fact that it is there regarded as a precious rarity. The Chinese said that in their country there were silkworms the size of a finger, feeding on the leaves of mulberry-trees and producing silk for the benefit of man, but those foreigners would not believe in the existence of silkworms." It does not require much sagacity to recognize in the transcription *g'iep, g'iev an Indo-European word and in particular one of Iranian characteristics, Avestan gav-, Middle Persian gav, go, New Persian gav, Armenian kov, Sanskrit gdv-. Above all, however, the Yue-ci form agrees closely, also in its vocalism, with Yazgulami yew ("taureau") from *7awa, recently disclosed by R. GAUTmox; 1 Ossetian gawd, and Scythian godi (from *gowdi) . It is a Scytho-Iranian type of word. There can be no doubt, either, that the notion of the decreasing and increasing bull of the Yue-ft answers to a mythical conception of specifically Iranian type, which the sober and prosaic Chinese were of course unable to grasp : it is the waxing and waning of the moon that is symbolized in the image of the bull. 2 Compare Avestan aevoddta and gaotiQra. 4- flfl r *ii$$ hi-hou, *h'iep (or hep)-gou, hiev-gou. Title of the five satraps of the Yue-Si, wrongly read yap-hau by HIRTH S and identified by him with the Turkish title ^ H ye-hu, *yab (dzab, sab)-gu. 4 The two titles, however, have nothing to do with each other. The title hi-hou is 1 Notes sur le Yazgoulami, dialecte iranien des confins du Pamir (Journal asiatique, 1916, mars-avril, p. 264). 2 Cf. for instance, DARMESTETER, Etudes iraniennes, Vol. II, p. 292; L. H. Gray, Spiegel Memorial Volume, pp. 160168; G. HUSING, Iranische Uberlieferung, pp. 2 3~54- The ox appears to have been an important domestic animal among the Yue-ci. According to the T'un tien j|j Jfil, written by Tu Yu /fi Vk from A.D. 766 to 80 1, the Great Yiie-ci availed themselves of four-wheeled carts (unknown to the Chinese), which in proportion to their size were drawn by four, six, or eight oxen. 3 Nachworte, p. 47. The foundation of this reading is the modern Cantonese dialect, but it is erroneous to identify the latter with ancient Chinese (see my Sino- Iranica, No. 1 1). There is no reason to assume that ^ ever had the reading *yap in ancient times; yap is merely a development peculiar to Cantonese. The fan-ts'ie of the character in question is indicated in K'an-hi by f ]& an d & Ri the sound being ]$ that is, h'iep or hep. Moreover, Hirth's identification of the title hi-hou with ye-hu is entirely arbitrary, not being supported by any Chinese text. If the two transcriptions, which phonetically are different, were intended to render the same foreign word, the ancient commentators would certainly not have failed to call attention to it. 4 Regarding the phonology of ye cf . PELLIOT, Bull, de VEcole fran$aise, Vol. IV, pp. 267-269. THE LANGUAGE OF THE YUE-CHI OR INDO-SCYTHIANS 7 applied by the Chinese also to Hiuii-nu, Wu-sun, and Sogdians. There is no reason to assume with the Turkomaniacs that it should be of Turkish origin: the Chinese themselves say nothing to this effect; but if the term is equally found among three Scythian or Turanian groups, compared with a single Turkish tribe, the greater probability is that the title is of Turanian origin and a Turanian loan-word in Hiun-nu. In my opinion the word itself is of Scythian origin, the first element being connected with Armenian $ahap, from Iranian *sarhap, Sahrap (Old Persian xsaQrapdvan, o-arpdTnjs). 1 Again, we observe the peculiar vocalism of Ytie-c'i : the vocalization hiep or hiev, compared with Iranian -hap, corresponds exactly to giev- Iranian gav (No. 3). In regard to the second element gou, I have not yet arrived at a positive conclusion, but will offer merely a suggestion. It is well known that Young-Avestan gava is used as a synonyme of Sogdiana, and that the Pahlavi translation explains this word by da$t ("plain"). DARMESTETER 2 has therefore conceived gava as a noun with the meaning "plain," and compared it with Gothic gawi ("county, country"), Old High German gewi, gouwi, Middle High German gou, gou. The Yue-6i word gou may be related to this Germanic word, and the term hap-gou may signify as much as "county-prefect." 5. From the royal names Kaniska, Huska, Huviska, Vasuska, we may well infer that -ska was an ending peculiar to the language of the I Yue-ci. S. LEVI S has joined to these forms the tribal name Turuska, which in fact is based on the name "Turk," but also serves for the designation of the Kusana or Indo-Scythians. In 1896 I indicated from the Mahavyutpatti the Sanskrit-Tibetan term turuska or turuka for the designation of frankincense (Gummi olibanum or Thus orientate). The Pen ts'ao kan mu gives Sanskrit turuskam as a synonyme of su-ho ("storax"), but evidently a confusion with frankincense has here arisen. 4 Turuska, however, does not mean, as believed by RHYS DAVIDS, "Turkish incense," 5 but "incense of the Indo-Scythians." In the chapter Zui yin f u Jj |g |H of the Sun su it is on record that in A.D. 458 the country Yue-Si ^ j sent as tribute divine incense f$ , which 1 HUBSCHMANN, Armenische Grammatik, Vol. I, p. 208; cf. also the note of ANDREAS in A. Christensen, L'Empire des Sassanides, p. 113. 2 Le Zend-Avesta, Vol. II, p. 7. 3 Journal asiatique, 1897, janv.-fe"vr., p. u. 4 PELLIOT, Toung Pao, 1912, p. 478. 6 See PELLIOT, Journal asiatique, 1914, sept.-oct., p. 418. As to tarukkha, the Pali equivalent for turuska, adopted by Rhys Davids, it should be remarked that A. WEBER (Abh. B. Ak., 1871, p. 85) had already explained turuska from *turukhka. 8 THE LANGUAGE OF THE YUE-CHI OR INDO-SCYTHIANS was examined by the Emperor Hiao Wu ^ jf. It had the appearance of swallow-eggs, and there were three lumps altogether, in size resembling a jujube. The emperor refused to burn it, and had it transferred to the treasury. Subsequently an epidemic broke out in the capital t 'aii-nan. The officials were infected and requested the emperor to burn a lump of the precious incense in order to ward off the pestilence. The emperor then burned it, whereby those sick in the palace were relieved. At a distance of a hundred li around t'an-nan, the odor of this incense was perceptible, and even after nine months, had not yet gone. Hence Chinese writers on incense have established the term "Yue-& incense" ft ^ ^f. 1 It follows from this story that only real frankincense can be involved. The ending -ska certainly is not Turkish, but Scytho-Iranian. 6. The tribal name Yue-Si has been much discussed, but a pho- netically correct restoration has not yet been secured. The name in the writing ft j appears in the first part of the second century B.C., and probably was first committed to writing in the memoranda and documents of General Can K'ien himself. We are confronted, accord- ingly, with the transcription of a foreign name attempted in the early Han period; and, as is well known, we are practically ignorant of what the phonetic condition of the Chinese language was in that era. The philological science of the Chinese permits us to restore the structure of words to the speech of the T'afi period, but beyond this we tread upon unsafe ground. Yet there is hope that the progress of comparative Indo-Chinese philology will also reveal to us some day the sounds of the language of the Han. In view of this state of affairs it behooves us the more to proceed cautiously and to heed all available data in the attempt to reconstruct the name by which the Yue-6i designated their nation. Especially Chinese comment bearing on it must not be taken lightly, as was done by A. v. SxAEL-HoLSTEiN. 2 Unfortunately the l Hian p'u ^ HJ by Hun C'u j* of the Sun period, p. lib (ed. of Tan Sun ts'un $u); Min hian p'u fa ^ fjf, cited in Pien tse lei pien, Ch. 7, p. 6b. 2 The speculations of A. v. STAEL-HOLSTEIN (SPAW, 1914, pp. 643-650 and repeated in JRAS, 1914, p. 754) are entirely inadmissible. He has a rather com- fortable method of discarding any evidence that is opposed to his preconceived theory of the identity of Jf J3Q with his artificial *Kusi, alleged to be the nominative singular of Kusa. In order to suit the purpose of this fantasy, the reading ci jj for Si de- manded by the Chinese philologists must be senselessly sacrificed: it is branded as "unauthoritative," while Wylie is heralded as an "authority," for he consistently transcribes Yue-she. In one of the old dialects, according to Stael-Holstein, ft j was pronounced Gur-si or Kur-si (again on p. 650: Kusi). The authority of Yen Si-ku in matters of Chinese philology is still to be regarded at least as high as that of Wylie and Stael-Holstein. In favor of his theory the data of Chinese history must THE LANGUAGE OF THE YUE-CHI OR INDO-SCYTHIANS 9 ancient commentators, while they give us positive information as to the phonetic value of the second element, fail to enlighten us on the first part. There is, however, a full interpretation of the name in a work of mediaeval date, which has been overlooked by previous writers. Yo Si *H jfe, author of the Tai p'in huan yu ki l in the latter part of the tenth century, explains the pronunciation of ^ j by means of the characters $J ;; that is, *2uk-& or *n'iuk-^i (d'i). This, of course, is striking, since the character ^ was anciently possessed of a final jlgntal, but never of a final guttural. Yo Si adduces no source or authority for his comment, but as he proves himself well informed and appears to have utilized original documents of the T'afi period, he may have derived this suggestion from a T'an source. The idea underlying his explana- tion is that the character is not taken by him as the classifier 73 ('moon'), but as the classifier 130 $J, which is written also with a variant ft (somewhat different in shape from ft 'moon') that appears in combina- tions with this classifier. It is difficult to believe that, if right from the beginning the character ft of Yue-Si should have conveyed the phonetic *2uk or *n'iuk, the ancient commentators should not have drawn attention to this anomaly. The opinion of Yo Si leads us nowhere, but it merits to be kept in mind. The direction of the commentators is that the second element j 'Si should be read ^ ci, 2 and later works have indeed substituted this character for the former. 3 The fact that the verdict of these old philologists is not arbitrary, as arbitrarily asserted by A. v. Stael- Holstein, is plainly to be seen from other names, for instance, the transcription fl jg yen-i, designating the queen of the Hiun-nu, where be discredited and turned upside down. It is perfectly obvious, however, that the two names Yiie-ci and Kusana, both of which were known to the Chinese, are by no means etymologically interrelated, but thoroughly independent. Kusana was known to the Chinese in the form jpf |f *Kwi-san, and they were aware of the fact that *Kwi-san was one of five satrapies or principalities, and that j J if K'iu- tsiu-k'io (Kuzulakadphises; regarding the Chinese transcription see PELLIOT, Journal asiatique, 1914, sept.-oct., p. 401), after the subjection of the four other satrapies, established himself as king of *Kwi-san. Kusana, as pointed out by F. W. THOMAS (JRAS, 1906, p. 203) is not a tribal name, but a family or dynastic title; otherwise we should not have an Indian inscription describing Kaniska as "propagator of the Kushan stock." See also J. MARQUART, Chronologic der alttur- kischen Inschriften, p. 59. 1 Ch. 184, p. 8. 2 Ts'ien Han su, Ch. 61, p. I; Hou Han Su, Ch. 118, p. 5. 8 The writing ft j appears in a document from Niya (CHAVANNES, in Stein, Khotan, p. 540). io THE LANGUAGE OF THE YUE-CHI OR INDO-SCYTHIANS Yen Si-ku clearly formulates the rule that anciently the character j had the sound ti (flf & ^ S)- 1 This verdict cannot be overruled in favor of any hypothesis to be attached to the name Yue-Si. Likewise in the name of the district Jjj$ j, U is to be pronounced &'. The question arises, however, as to how the character ; was articulated in early times. The opinion of KLAPROTH, who adopted the reading Yue-ti, that t may often be replaced by , cannot be set aside so completely, as has been done by O. FRANKED save that Klaproth did not express himself very clearly; he doubtless meant to say that palatal t or tS may develop from dental t; and this, in fact, is a common phenomenon in Indo-Chinese. Moreover, it is justly emphasized by PELLIOT S that the small dash differentiating at present the symbol & tiirom j is a comparatively recent affair, so that formerly the latter character might have been read Si as well as ti. Pelliot is further right in concluding that under the Former Han the character j was sounded with a dental initial more or less palatalized (mouille) , but not with a palatal and still less with a fricative. As Wan C'uii offers the variant J|f Jt Yen-t'i in lieu of ffl j Yen-ci, and as M answers to an ancient *di, there is good reason to assume that also j and likewise ; were at that period articulated d'i, di, or &'. 4 Thus there is also reason to believe that this element -di or -ti was assimilated to the final dental 1 The restoration of the transcription yen- Si presents a complex problem, as the commentators offer various means of reconstructing the prototype. The Si ki so yin states that yen-ci should be read jljf jj *had-di or *hat-ti, which would indeed lead to Turkish xatun (qatun, khaturi). In ancient times this word had several phonetic variants: we have in T'u-kiie *kahatun "fif jf| *jfr (k'o-ho-tun) and *katun ~pf ^ (k'o-tun), in T'u-yu-hun *katsun '$ Jfl (k*o-tsun), and in T'o-pa *kasun "pjf J (k'o-sun). It seems to me that the Hiun-nu word *haddi ( = *haddun) repre- sents the primeval form, and that *katsun and *kasun are subsequent developments. It is difficult to see, however, why the Chinese wrote yen, if the sound phenomenon had was intended; but whatever the basis of this identification may be, there can be no doubt of the existence of the Hiun-nu word *had-di itself. It is obvious that Yen Si-ku visualized a different term of the Hiun-nu language when (Ts'ien Han su, Ch. 94 A, p. 5) he defines the fan-ts'ie of yen by means of j^ ^ (yien). It is singular that K. SHIRATORI (Sprache des Hiung-nu Stammes, p. 4), in dealing with this word, does not heed the Chinese indications, although he quotes them, and identifies the word with Uigur obeli or eoli, which in my opinion is impossible. 2 Zur Kenntnis der Turkvolker, pp. 22-23. The variant Jj does not occur quite so rarely, as assumed by Franke. It is employed in K'an-hi's Dictionary, and may be seen in the Su kien %jg fj edited by Kuo Yun-t'ao Jfl jfc J in 1236 (Ch. 9, p. 3; ed. of Sou San ko ts'un su). At any rate j is merely a graphic, not a phonetic variant. 8 Journal asiatigue, 1912, juillet-aout, p. 169. 4 See also PELLIOT, Bull, de VEcole frangaise, Vol. V, p. 428. THE LANGUAGE OF THE YUE-CHI OR INDO-SCYTHIANS n of the first part, ^ , and that this final may have been not a surd, but a sonant. In the same manner as in Tibetan, it will be shown that also in Chinese the final explosives were originally all sonants (partially also liquids) . The ancient phonetic formation of ^J is somewhat complex and very far from being such a simple affair as get, as confidently asserted by former sinologues, in order to fall into the trap of the Getae and Mas- sagetae. The original initial was not a guttural sonant, but the gut- tural nasal n, which is plainly indicated by thefan-ts'ie & (*ni k'iud) and the sound equivalent (J[, which, in the same manner as ^f , still has in Sino-Annamese the pronunciation nilet. Further, this initial n was palatalized (mouille) and labialized (provided with so-called ho-k'ou fe P); that is, phonetically written, *ii'wiet or n'wied, n'wieS. 1 The initial n seems to have had a tendency to develop into g during the T'an period. In an Uigur Sutra the name Yue-& is said to be tran- scribed Kitsi or Ketsi. 2 Of course, this transcription made after a Chinese mode of articulating the term in the T'aii period (provided the identification were correct, which is doubtful) would have no absolute value for the restoration of the ancient form of the name which makes its debut some eight hundred years earlier. The fact remains that initial n (now y) generally corresponds to g in the transcriptions made in the age of the T'aii; 3 and if the transcription Yiie-ci had originated in that period, we should be perfectly justified in restoring it to a form with initial g. But we must not lose sight of the fact that the transcription was made early in the second century B.C. under the Han; and that the same rules then prevailed as tinder the T'an, no one can affirm. The greater probability is that phonetic conditions were then somewhat different and perhaps more complicated than in mediaeval times. In Iranian and Scythian names we have always to reckon with double 1 In this correct form it is transcribed, for instance, by H. MASPERO, Etudes sur la phonetique historique de la langue annamite, p. 94. See also PELLIOT, Bull, de I'Ecole frangaise, Vol. V, p. 443. The ho-k'ou is still preserved in Fu-kien nwok and Japanese gwatsu. It is only through this ho-k'ou that the utilization of ^f in the transcriptions of Sanskrit for vit and vut becomes intelligible (see, for instance, examples in VOLPICELLI, Prononciation ancienne du chinois, p. 179; and SCHLEGEL, Secret of the Chinese Method, p. 98). Concerning the use of ^j in the transcription of an Iranian word see the writer's Sino-Iranica, No. 18. 2 According to F. W. K. MILLER, Uigurica, p. 15; but according to PELLIOT (Traite" manicheen, p. 29), the Chinese equivalent corresponding to Kitsi is the name g| jf* *Ni-tsin (Yi-tsin). 3 Examples may be seen in CHAVANNES and PELLIOT, Traite" maniche"en, pp. 29, 42. 12 THE LANGUAGE OF THE YUE-CHI OR INDO-SCYTHIANS consonants which it was difficult for Chinese to reproduce. From my Sino-Iranica it will be seen that Chinese initial 5 and may correspond to Iranian xs and xL l A somewhat vacillating initial as that of $ indicates very well that it should answer to a combination of foreign sounds unfamiliar to a Chinese ear. This assumption being made, there are two hypothetical reconstructions possible: *n'wied-di would lead either to *aii'wied-di or to *sgwied-di. The latter is the more probable one, and bears all the characteristics of a Scytho-Iranian name. It comes very near to the Suguda of the Old Persian inscriptions (Avestan i~l f^V Su^ySa), the name of the Sogdoi or Sogdians. I do not mean to say that the two names are physically identical, but only that there is a linguistic relationship between the two. As regards the element di (d'i, or eventually even ti), I hold that it should be conceived as the ending of the plural, and that the plural suffix -di is on a par with the plural-suffixes, -ta of Ossetian, -rat of Scythian, and -t or -y-t of Sogdian and Yagnobi. 2 The identification of the name Yue-c"i with that of the Getae and Massagetae, in my opinion, is out of the question. Not only phonetic, but also geo- graphical and historical reasons run counter to this assumption. 3 Also the identification with 'Icmot 4 must be rejected, likewise any alleged relation of the name to that of the Ye-t'a (*Yep-dal, Ebdal, Abdal) or Ephtalites. The latter, however, are not Huns, as wrongly asserted by SPECHT, but in the same manner as the Yue-c"i, are Indo-Europeans, that is, Scythic Iranians. Likewise so were the ancient Wu-sun, as I hope to demonstrate in a subsequent article. Turkistan, before being settled by Turki, was a country of Iranians. O. FRANKE 5 has justly called attention to the fact that very close relations and intermarriage existed between the Yue-ci and the Sogdians (K'an-kii Jg) ; and the kings of Sogdiana are said to have descended from the Yue-ci and to have gloried in this extraction. It seems to me 1 A similar phenomenon obtains in the Sanskrit transcriptions: for instance, Ki-pin, based on *Ki-spir = Ptolemy's Kasparia. In the same manner, the Sino- Iranian word if| pin ("a fine steel imported from Persia") is based on Iranian *spin (Sariqoli spin, Afgan osplnah or ospanah, Ossetian afseindg, "iron"). 2 Cf . W. MILLER, Sprache der Osseten, p. 42 ; J. MARQUART, Untersuchungen zur Geschichte von Eran, II, pp. 77-79; R. GAUTHIOT, Du pluriel persan en -ha (Mem. Soc. de Linguistique de Paris, Vol. XX, 1916, pp. 74-75). 3 On this point I concur with the opinion of MARQUART (Eransahr, p. 206). 4 ToMASCHEK, Sogdiana, SWA, 1877, p. 159. L.C..P. 67. THE LANGUAGE OF THE YUE-CHI OR INSO-SCYTHIANS 13 that the cause of this mutual good feeling was given in the linguistic relationship of the two peoples. 1 When Sieg and Siegling published their memorable study of one of the Indo-European languages rediscovered from ancient manuscripts of Turkistan, they styled this language Tokharian and further defined it in the very title of their publication as "the language of the Indo- Scythians." 2 It has been recognized long ago that both these designa- tions are hazardous. 3 "Indo-Scythian" is out of the question as a 1 Regarding Indo-Scythian proper names see F. W. THOMAS, JRAS, 1906, pp. 204-216. I do not believe that f|f Sie (the vice-roy of the Yue-ci, vanquished by Pan C'ao in A.D. 90) represents the title sdhi (S. L6vi, Journal asiatique, 1915, janv.-f