248 SlNO-lRANICA 



?). Persians $& $r IK designate them a-yile-hun fruits." 1 For the same 

 period we have the testimony of the Arabic merchant Soleiman, who 

 wrote in A.D. 851, to the effect that pistachios grow in China. 2 



As shown by the two forms, a-yue of the Yu yan tsa tsu and a-yue-hun 

 of the Pen ts'ao $i i and Hai yao pen ts*ao, the fuller form must repre- 

 sent a compound consisting of the elements a-yue and hun. In order to 

 understand the transcription a-yue, consideration of the following facts 

 is necessary. 



The Old-Iranian word for the walnut has not been handed down to 

 us, but there is good evidence to prompt the conclusion that it must 

 have been of the type *agOza or *afigOza. On the one hand, we have 

 Armenian engoiz, Ossetic angoza or anguz, and Hebrew egoz; 3 on the 

 other hand, we meet in Yidgha, a Hindu-Kush language, the form 

 oguzo, as compared with New Persian koz and goz.* The signification 

 of this word is "nut" in general, and " walnut" in particular. Further, 

 there is in Sanskrit the Iranian loan-word dkhota, aksoja, or aksoda, 

 which must have been borrowed at an early date, as, in the last-named 

 form, the word occurs twice in the Bower Manuscript. 5 It has survived 

 in Hindustani as axrdt or dkrot. The actual existence of an East- 

 Iranian form with the ancient initial a- is guaranteed by the Chinese 

 transcription a-yue; for a-yiie M H answers to an ancient *a-nwie5 

 (nw'e5) or *a-gwie5, a-gwu5; 6 and this, in my opinion, is intended to 

 represent the Iranian word for "nut" with initial a-, mentioned above; 

 that is, *angwiz, afigwOz, agOz. 



Chinese hun answers to an ancient *7wun or wun. In regard 

 to this Iranian word, the following information may be helpful. E. 



1 If it is correct that the transcription a-yue-hun was already contained in the 

 Nan lou ki (which it is impossible to prove, as we do not possess the text of this 

 work), the transcription must have been based on an original prototype of early 

 Sasanian times or on an early Middle- Persian form. This, in fact, is confirmed by 

 the very character of the Sino-Iranian word, which has preserved the initial a-, 

 while this one became lost in New Persian. It may hence be inferred that Li Sun's 

 information is correct, and that the transcription a-yue-hun may really have been 

 contained in the Nan cou ki, and would accordingly be pre-T'an. 



2 M. REINAUD, Relation des voyages faits par les Arabes et les Persans dans 

 1'Inde et a la Chine, Vol. I, p. 22. 



3 Whether Georgian nigozi and the local name Nlyovfa of Ptolemy (W. 

 TOMASCHEK, Pamirdialekte, Sitzber. Wiener Akad., 1880, p. 790) belong here, I do 

 not feel certain. Cf. HUBSCHMANN, Armenische Grammatik, p. 393. 



4 In regard to the elision of initial a in New Persian, see HUBSCHMANN, Persische 

 Studien, p. 120. 



6 HOERNLE'S edition, pp. 32, 90, 121. 



6 Regarding the phonetic value of ^ , see the detailed study of PELLIOT (Bull. 

 de VEcole fran$aise, Vol. V, p. 443) and the writer's Language of the Yue-chi or 

 Indo-Scythians. 



