Sec. ii] HISTORICAL SITES IN WAKHAN 7i
cessively upwards 4 feet, 4 feet, and 32 feet. The construction in stories, the surmounting dome, and the character of the rough but solid masonry at once recalled to my mind the closely corresponding features of the little Stûpa of Thal, the last Buddhist remains I had seen in 19oo south of the Hindukush when approaching the Pamirs through Hunza." There it was an unmistakable Stûpa base, io feet square in ground-plan, which showed these successive stories and cornices with heights of 3 feet 9 inches, 3 feet 9 inches, and 3 feet zo inches respectively. This similarity of constructive features distinctly suggests pre-Muhammadan origin for the Karwan-balasi ruin, and this assumption finds further support in the horizontal construction of the interior dome and of the arch above the narrow entrance (see Fig. 4). Misled by the present local tradition and the apparent orientation towards the Qibla, which the south-west bearing of the cella wall facing the entrance suggested, I had been inclined at first to look upon the ruined cella as an adaptation to Muhammadan use of an ancient architectural model of Buddhism.'$ But further consideration makes it appear to me far more probable that the ruin goes back to pre-Muhammadan times, and had originally served as a small Buddhist Vihâra.
The very orientation, in fact, on which my first conjecture relied, seems to exclude the use of the structure as a Muhammadan tomb ; for within the narrow space its interior affords it would be impossible to give to a buried body the position, with the feet laid due south, which orthodox Muhammadan custom demands. This consideration must be held in all probability to account for the fact, subsequently tested at several Muhammadan burial-grounds of old date in the Tarim Basin, such as at the ruins near the Charchan and Inchike Rivers," that the ruined tombs examined were invariably orientated towards the cardinal points. Yet, after the many examples recorded both in Ancient Khotan and in the present work of earlier local worship continued with due adaptation into Muhammadan times,20 there can be nothing to cause surprise in the interpretation which later tradition placed upon the ruin as a Muhammadan tomb. Local worship was bound to linger on in the case of a ruin which, however small its size, was conspicuous in a region so devoid of permanent structures as the Pamirs, and given its appearance, so like the ` Gumbaz ' or dome customary in Central Asia, especially with the Kirghiz, no explanation could be more appropriate than that it was a tomb. Local tradition was not likely to trouble itself much about the difference in orientation, and for the strangely small interior of the supposed burial-chamber it could readily find an explanation by representing it as that of a child, a fact which accounts for the name Karwan-balasi, ` the son of the caravan[-leader]'.
If we assume the ruin to be that of a small Vihâra, or chapel, intended to shelter some sacred Buddhist image, there is no difficulty whatever in accounting for its architectural features. The ground-plan and elevation agree well with what M. Foucher's lucid analysis has proved to be the typical construction of Buddhist Vihâras in Gandhâra and elsewhere on the north-western confines of India.21 The interior dome and superimposed cupola are there regular features, and even analogies to the threefold division of the outer walls, with cornices, can be found in various Vihâra models represented in relievo sculptures.22 Nor does the narrow space of the cella interior present anything peculiar, seeing that these Buddhist Viharas in India were very often, if not ordinarily, intended for the reception of a single image only, and that at the great Takht-i-Bahi convent even the largest of
'7 See Ancient Khotan, i. p. 20, Fig. 4.
18 See Desert Cathay, i. p. 77.
19 See below.
40 Cf. for references Ancient Khotan, i. p. 6r r, s. v. local worship ; above, p. 41, and below, sec. iii.
41 Cf. Foucher, L' Art du Gandhdra, i. pp. Ito sqq.
22 Cf. ibid., particularly Figs. 41, 42. The two lower