Introduction 187 



rigid phonetic principles is the essential point, and means much more 

 than any haphazardly made guesses at identification. Thus Mu-lu 

 /fc }*&, name of a city on the eastern frontier of An-si (Parthia), 1 has 

 been identified with Mouru (Muru, Merw) of the Avesta. 2 Whether 

 this is historically correct, I do not wish to discuss here; from an his- 

 torical viewpoint the identification may be correct, but from a phonetic 

 viewpoint it is not acceptable, for Mu-lu corresponds to ancient *Muk- 

 luk, Mug-ruk, Bug-luk, Bug-rug, to be restored perhaps to *Bux-rux. 8 

 The scarcity of linguistic material on the Iranian side has imposed 

 certain restrictions: names for Iranian plants, one of the chief subjects 

 of this study, have been handed down to us to a very moderate extent, 

 so that in many cases no identification can be attempted. I hope, 

 however, that Iranian scholars will appreciate the philological con- 

 tributions of the Chinese to Iranian and particularly Middle-Persian 

 lexicography, for in almost every instance it is possible to restore with 

 a very high degree of certainty the primeval Iranian forms from which 

 the Chinese transcriptions were accurately made. The Chinese scholars 

 had developed a rational method and a fixed system in reproducing 

 words of foreign languages, in the study of which, as is well known, 

 they took a profound interest; and from day to day, as our experience 

 widens, we have occasion to admire the soundness, solidity, and con- 

 sistency of this system. The same laws of transcription worked out 

 for Sanskrit, Malayan, Turkish, Mongol, and Tibetan, hold good also 

 for Iranian. I have only to ask Iranian scholars to have confidence in 

 our method, which has successfully stood many tests. I am convinced 

 that this plea is unnecessary for the savants of France, who are the 



is, Dik-lat, Dik-rat), which has passed into Greek Tlypijs and TLypis and Elamite 

 Ti-ig-ra (A. Meillet, Grammaire du vieux perse, p. 72). It will thus be seen that 

 the Chinese transcription *Dak-rat corresponds to Babylonian Dik-rat, save the 

 vowel of the first element, which cannot yet be explained, but which will surely be 

 traced some day to an Iranian dialect. — The T'ai p'in hwan yii ki (Ch. 185, p. 19) 

 gives four geographical names of Persia, which have not yet been indicated. The 

 first of these is the name of a city in the form ^ §| J§ Ho-p'o-kie, *Hat(r, 1)- 

 bwa-g'iat. The first two elements *Har-bwa correspond to Old Persian Haraiva 

 (Babylonian Hariva), Avestan Haraeva, Pahlavi *Harew, Armenian Hrew, — the 

 modern Herat. The third element appears to contain a word with the meaning 

 "city." The same character is used in $§j ^ J§'J Kie-li-pie, *G'iat-li-b'iet, name of a 

 pass in the north-eastern part of Persia; here *g'iat, *g'iar, seems to represent 

 Sogdian yr, *yara ("mountain"). Fan-tou ^ or ^ {ftj (Ts'ien Han $u, Ch. 96 a), 

 anciently *Pan-tav, *Par-tav, corresponds exactly to Old Persian Parflava, Middle 

 Persian Parflu. 



1 Hou Han Su, Ch. 116, p. 8 b. 



* Hirth, China and the Roman Orient, p. 143. 



• Cf. also the observation of E. H. Parker (Imp. and As. Quarterly Review, 

 !903. p. 154)1 who noticed the phonetic difficulty in the proposed identification. 



