as our scanty historical records allow us to judge, to have always figured as a chiefship allied to, yet distinct from, Chitral. It certainly has enjoyed this political separation under the Khushwakt family which, itself a branch of the dynasty established in Chitral since the seventeenth century, has intermittently either asserted its virtual independence from Chitral or even managed to extend its hold over it.2 The extension of Khushwakt rule over Yasin, and at times even Gilgit, which came to a close only in our own time, probably facilitated this independence by increasing the otherwise slender resources of Mastûj.
But just as the Khushwakt extension eastwards is accounted for by the comparatively easy line of communication connecting Mastûj and Yasin across the Shandur Pass (only 12,25o feet above sea level), so the division which history shows between Chitral and Mastûj, the homogeneous character of the respective populations and rulers notwithstanding, finds its natural explanation in geographical features. The open and fertile portion of the Upper Yârkhûn Valley which extends from Mastûj to a point above Shuyist, a distance of some sixty miles, and the equally fertile though narrower side-valley of Laspur descending from the Shandur Pass, are effectively separated from Chitral proper, and to a somewhat less extent, also, from the side-valleys which the river of Drasan drains, by a succession of difficult gorges in the Yârkhûn Valley where strong defensive positions abound. It is probably only the predominance of a strong outside rule which could assure the permanent union of .Chitral and Mastttj, and even now when there is such a suzerain power controlling all the Hindukush valleys, Mastûj enjoys independence under a separate chief though included with Chitral in the same Political Agency.
This rapid glance at the recent history of Mastûj has seemed to me necessary for a proper understanding of the only early notice of this territory which I have so far been able to trace. It is furnished by a passage of the Tang Annals on which I have already commented in my Ancient Khotan.3 A notice of the Tang shu of which M. Chavannes has been the first to give
a full and exact rendering, states that ` Clzii-wei it is also called Shang-mi C ; its capital
is in the town of A-shê yü-shih-to ; it is situated amidst the great snowy mountains, north of the river of Fo-1ü. This country is cold ; it produces the five cereals, the vine and pomegranate. During the winter people live in caves. The inhabitants of this kingdom have always assisted the Little P`o-lü in spying out the " Middle Kingdom" (China).' 4
By the river of Po-1u can be meant only the river of Gilgit, the territory of which, along with Yasin, bears in the Chinese Annals the designation of ` Little P`o-lii'.L A glance at the map shows that the territory of Chü-wei or Shang-mi, with its capital ` situated amidst the great snowy mountains' to the north of this river, corresponds exactly to the present Mastûj and the uppermost Yârkhûn Valley. The whole of this valley above Mastûj lies due north of the main feeder of the Gilgit River which, rising in the Shandur Lake, flows past Ghizar and Gupis eastwards.
The description which the Tang notice gives of the country as cold is correct ; the latest account of Mastûj, too, notes : ' The climate in winter is severe, owing to the cold winds which
2 For some account of the constant struggles between the Khushwakt and Katar branches, cf. Biddulph, Hindoo Koosh, pp. 252 sqq. The main facts seem to be rightly summed up thus: `Consanguinity did not prevent constant wars between the rulers of Chitral and Yasin' (i. e. Mastûj which held Yasin).... The Khush Wakté seem to have shown the greater warlike skill, but this advantage was balanced by the superior wealth and population of Chitral'; ibid., p. 152. About 1790 the Khushwakt Khairullah appears as `supreme Bâdshâh', holding Chitral as well as Mastaj and Yasin ; cf.
Raverty, Notes on Afghânisian, pp. 254, 258, 262 ; but subordinate ` Bâdshâhs' are also named in different tracts. Aman-ul-mulk's supremacy over the whole region from Yasin to Chitral, 1880-92, also proved short-lived ; see Imperial Gazetteer, 2908, x. pp. 302 sq.
3 See Anisent Kholan, i. p. 15, note 31.
' Cf. Chavannes, Turcs occid., p. 129, note 2.
6 Cf. Ancient Kholan, i. pp. 6 sq. ; Chavannes, Turcs occid., p. 250, note 2, with modification indicated in Notes Addit., p. 43, note 2.