National Institute of Informatics - Digital Silk Road Project
Digital Archive of Toyo Bunko Rare Books

abnormal spellings are collected in Southern Tibet, IX, v, 143 (even that list is incomplete, omitting
for instance « Yerghien » in Astley's map reproduced VII, pl. X).

Yārkānd can neither mean « Extended land » as is said in Hsi-yü t'ung-wên chih (cf. Pa, 141;
Y, I, 188-189), nor « New City » as Cordier preferred to have it. The only plausible explanation
is to see in Yārkānd a Turkish name formed with yār, «cliff», and kānd, «city» (the latter was origi-
nally an Iranian word, kāθ, kānt, kānd, but it passed into Turkish); « Yārkand » is no more a real
hybrid name than Taškānd, our Tashkend (from Turk. taš, « stone », and kānd, « city »), or « Ordu-
kānd » (see « Cascar »).

When Stein published his Ancient Khotan in 1907, he could still say (p. 87) that it was « difficult
to trace back its (i. e. Yārkānd's) name... previous to the Mongol conquest in the thirteenth century ».
From the mention of the name in Kāšyarī (Brockelmann, 244), we are now certain that it existed
at least as early as 1076. Perhaps it was adapted from an earlier, non-Turkish name of the river,
if not of the city. In Gardēzī's Persian text of 1050-1052, which draws from earlier sources, an
itinerary is given from Kāšyar to Khotan, and the text adds that « in the middle [of the country]
between these points flows the ﻯ Yārā River » (cf. Barthold, Otčet o poêzdkê v Srednyuyu Aziyu,
94, 119). This must be the Yārkānd River, and one might think that the yār of Yārkānd was seman-
tically adapted from Yārā. But the reading in Gardēzī's unique Ms. is doubtful; a parallel list,
drawn earlier from the same source, but with forms often more corrupt, occurs in the equally
unique Ms. of the Ḥudūd al-ʿĀlam, and there the name of the river is written ﻯﺮﺑ B.rn.yā (cf. Mi,
93, 260). It must be noticed that the name of Yārkānd is not found in that itinerary, and pro-
bably did not then exist. I shall return to this question farther on.

Cl. Huart has published in JA, 1914, II, 607-627, three documents in Arabic which I had
acquired at Kučā early in 1907. Excavated, I was told, at Yārkānd, they are official deeds for the
transfer of land and are dated respectively 1096, 1112 and 1114. They testify to the rapid
advance of Mussulman culture in the region; some of the witnesses, however, still used Uighur
writing. In the three documents, the name of Yārkānd is written in Arabic ﻛﻨﺪﺎﻳ Yārkāndā,
a form for which I do not know how to account. Fifteen documents belonging to the same find
came into the hands of Sir G. Macartney and now belong to the Indian Government. One of
them, probably of 1101, was published in 1923 by Barthold (BSOS, III, 151-158); there also the
name is written Yārkāndā, but the editor makes no remark on that spelling (neither in that paper,
nor in his later 12 Vorlesungen, p. 133, does Barthold mention Huart's paper of 1914; the date
of 1911 given for the discovery of the documents in an additional note by D. Ross, p. 158, is of
course too late, in view of the year in which my own documents were acquired). Kāndā, kāndāh
means « moat » in Persian, but the word has not passed into Turkish, and Kāšyarī already gives the
modern form Yārkānd at a date somewhat earlier than our three documents. Yārkānd must be
mentioned in the Persian histories of the Mongol period, although I have not been able to find the
name in them hitherto; at any rate, it occurs as Yārkānd in the geographical section of the Nuzhat-
al-Qulūb, completed in 1340 (Le Strange, transl., 251). The Taʾrīḫ-i-Rašīdī of course speaks
very often of Yārkānd (cf. the index of Elias and Ross). Curiously enough, the Persian work
Hāft iqlīm, which copies the Taʾrīḫ-i-Rašīdī, seems to write « Yarkān », without final -d, as is the
case in Kalmuk spelling (cf. the translation by Quatremère, in Not. et Extr., XIV, I, 475-476).

17 A