the attendants of the shrines, as well as the pilgrims, would necessarily make upon them for whatever in the way of wood and other useful materials had remained in them.
However this may be, it must be considered as certain that the abandonment of the settlement was a gradual one, and in no way connected with any sudden physical catastrophe such as some European travellers have been only too ready to assume, on account of popular legends they had heard about the so-called " ancient cities " of the Taklamakan. The Sodom and Gomorrah legends related all over Eastern Turkestan about " old towns " suddenly submerged under the sand-dunes, are more ancient than the ruins of Dandan-Uiliq themselves. Hiuen-Tsiang had already heard them more or less in the same form in which they are now current, as is seen from the story of the town of ` Ho-lo-lo-kia,' which we shall have occasion to refer to in connection with the site of Pi-mo. These legends undoubtedly are interesting as folk-lore. But where we have such plain archwological evidence to the contrary as the examination of the Dandan-Uiliq ruins, and in fact of every other ancient site in this region has supplied to me, scientific inquiry need have no concern with them.
My detailed survey of Dandan-Uiliq, together with other observations of a. semi-topographical, semi-antiquarian nature which gradually accumulated during my explorations at this and other sites, make it very probable that the lands of Dandan-Uiliq were irrigated from an extension of the canals which, down to a much later date, brought the water of the streams of Chira, Domoko and Gulakhma to the desert area due south of the ruins. The débris-covered site of Uzun-Tati, which I discovered there amidst the sand-dunes, is identical with the ' Pi-mo ' of Hiuen-Tsiang, Marco Polo's ' Pein,' and can be proved by unquestionable evidence to have been occupied for at least five centuries longer than DandanUiliq. A number of historical as well as topographical considerations, for a detailed discussion of which I must refer to my scientific publication, point to the conclusion that the successive abandonment of both Dandan-Uiliq and ` Pi-mo ' was due to the
IIIFカラー高解像度
