538 Sino-Iranica 



nian, loan-word from Persian, aprUum)\ hence Arabic ibarisam or 

 ibrlsam; Pamir dialects warSum, war$um, Sugni wrelom, etc.; Afghan 

 writs' am. 1 Certain it is that we have here a type not related to any 

 Chinese word for "silk." In this connection I wish to register my utter 

 disbelief in the traditional opinion, inaugurated by Klaproth, that 

 Greek ser (" silk- worm " ; hence Seres, Serica) should be connected with 

 Mongol Urgek and Manchu sirge ("silk"), the latter with Chinese se 

 M. 2 My reasons for rejecting this theory may be stated as briefly as 

 possible. I do not see how a Greek word can be explained from Mongol 

 or Manchu, — languages which we merely know in their most recent 

 forms, Mongol from the thirteenth and Manchu from the sixteenth 

 century. Neither the Greek nor the Mongol-Manchu word can be 

 correlated with Chinese se. The latter was never provided with a final 

 consonant. Klaproth resorted to the hypothesis that in ancient dialects 

 of China along the borders of the empire a final r might (pent- tire) have 

 existed. This, however, was assuredly not the case. We know that the 

 termination V jrcl, so frequently associated with nouns in Pekingese, is 

 of comparatively recent origin, and not older than the Yuan period 

 (thirteenth century) ; the beginnings of this usage may go back to the 

 end of the twelfth or even to the ninth century. 3 At any rate, it did not 

 exist in ancient times when the Greek ser came into being. Moreover, 

 this suffix V is not used arbitrarily : it joins certain words, while others 

 take the suffix tse ■?", and others again do not allow any suffix. The 

 word se, however, has never been amalgamated with V. In all probabil- 

 ity, its ancient phonetic value was *si, sa. It is thus phonetically im- 

 possible to derive from it the Mongol-Manchu word or Korean sir, 

 added by Abel-R6musat. I do not deny that this series may have its 



root in a Chinese word, but its parentage cannot be traced to se. I do 



L, 



1 Hubschmann, Arm. Gram., p. 107; Horn, Neupers. Etymologie, No. 65. 

 The derivation from Sanskrit ksautna is surely wrong. Bulgar ibriUm, Rumanian 

 ibrisin, are likewise connected with the Iranian series. 



2 Cf. Klaproth, Conjecture sur l'origine du nom de la soie chez les anciens 

 (Journal asiatique, Vol. I, 1822, pp. 243-245, with additions by Abel-Remusat, 

 245-247); Asia polyglotta, p. 341; and M£moires relatifs a TAsie, Vol. Ill, p. 264. 

 Klaproth's opinion has been generally, but thoughtlessly, accepted (Hirth, op. 

 tit., p, 217; F. v. Richthofen, China, Vol. I, p. 443; Schrader, Reallexikon, p. 757). 

 Pelliot (T'oung Pao, 191 2, p. 741), I believe, was the first to point out that Chinese 

 se was never possessed of a final consonant. 



3 See my note in T'oung Pao, 1916, p. 77; and H. Maspero, Sur quelques textes 

 anciens de chinois parle\ p. 12. Maspero encountered the word mao'r (" cat ") in a text 

 of the ninth century. It hardly makes any great difference whether we conceive V 

 as a diminutive or as a suffix. Originally it may have had the force of a diminutive, 

 and have gradually developed into a pure suffix. Cf. also P. Schmidt, K istorii 

 kitaiskago razgovornago yazyka, in Sbornik stat'ei professorov, p. 19 (Vladivostok, 

 1917). 



