228 ANCIENT SITES OF THE KHOTAN OASIS [Chap. VIII
continued to flow ever since. Hsüan-tsang tells us that the drum of the dragon (Naga) had long ago disappeared. But there existed at the place where it had been suspended a lake known as ` the drum-lake ', and there were still seen the ruins of the convent which had been built by its side but was then without monks and deserted 22.
Location of Seeing that the Aiding-Kul lies exactly south-east of Ybtkan, and that apart from it there
Hsüan- is no sheet of water in the vicinity of the site of the ancient capital which could be called a
drum-lake' lake or even a natural pond, it is difficult to resist the conclusion that the local name of
convent. Naghara-khanah, ` the drum house,' is connected with the old legend about the miraculous drum
which, as we know from Hsüan-tsang, gave its name to the lake and in all probability also to the ruined convent he mentions by its side. The examination of the mound now known as the ` Naghara-khanah ' leads me to believe that we have in it the remains of the structure which was shown to Hsüan-tsang as a ruined convent far more probably than a fragment of an otherwise wholly untraceable city wall. The mound measures about i 75 feet from east to west, with a maximum width of 72 feet : its approximately level top rises about 27 feet above the adjoining fields. I failed to trace on its slopes any marks of brickwork or stratification, the whole surface showing what looked to me like stamped loess soil. The general appearance of the ` Naghara-khanah' reminded me of the large oblong mound once bearing a Vihara or convent which rises behind the Mauri-Tim Stûpa and which has been fully described above 23. It can scarcely be doubted that the interior of the Mauri-Tim structure consists of nothing but stamped loess ; and that when once the facing of sun-dried bricks, which still shows traces of niches or small cellas arranged in receding terraces, has undergone further decay in the course of centuries, its surface will present the same shapeless look as the Naghara-khanah mound does now. The latter, if our identification is right, must already in the seventh century have been a ruin of considerable antiquity, and its wholly amorphous condition at the present day could thus in no way surprise us.
Connexion It would be of little use to offer conjectures as to the way in which the folklore story of
of Naghâra- the minister's Curtius-like sacrifice and the miraculous drum came to be attached to this ruined khanah with
Hsüan. mound. But it •may be worth noting that, just as all old mounds in this region are now
tsang's popularly supposed to have been constructed as watch-stations by ancient kings, a ruined
legend. mound of such conspicuous height as the Naghara-khanah must have shown before the ground-
level around it rose and erosion reduced its elevation, was likely to exercise popular imagination in old times. The distance from the Naghâra-khanah to the excavated area of Ybtkan is only two miles, and thus the belief in its having served for the location of a great drum intended to warn the capital of an approaching enemy could readily be accounted for.
Continuance I am unable to make any definite suggestion as to the stream which the legend looked
of local upon as the habitation of the Nagini and her self-sacrificing human husband, seeing the great
variations in distance and bearing which the several translations show. Possibly some point higher up on the Yurung-kash might have been thought of as the scene of the minister's sacrifice. That the legend itself was popular can be safely concluded from Hsüan-tsang's detailed mention. I believe that we have visible proof of this in a fresco excavated by me in one of the shrines of Dandan-Uiliq, of which I hope to show in the next chapter that it represents the characteristic figures of the legend 24. At a sacred spot connected with such a popular legend we might expect to see local worship continued to the present day ; and in fact,
22 This purport of the passage, which seems clear enough, 23 See above, pp. 81 sqq.
is brought out better in Rémusat's version (loc. cit., p. 6o) 24 See below, chap. ix. sec. iii.
than in that of Julien which Beal slavishly follows.