so far known specimens of that script and language. A Chinese inscription scratched into the wall of the temple cella recorded the visit of a Chinese administrator and gave a date corresponding to A.D. 719. This together with the deposit of Tibetan texts made it quite certain that the fort must have been occupied in the eighth century A.D., towards the close of which the Tarim basin passed for a time under Tibetan domination.
Now it was curious that Hsüan-tsang, the great Chinese pilgrim who had passed by the same route from Niya to Charchan about A.D. 645, found no inhabited place on his ten days' desert march. But he distinctly mentions in a position exactly corresponding to the Endere site ruins of abandoned settlements which were described to him as `old seats of the Tukhara' famous in Central-Asian history.
Discoveries made on my second visit conclusively proved that we have here a definite historical instance of an old site at one time abandoned to the desert having been reoccupied after the lapse of centuries. A shifting of the low dunes near the fort had exposed much-eroded remains of some ancient dwellings which I had not previously noticed. When I had the consolidated refuse-heaps which had saved them from complete destruction carefully cleared, there came to light some wooden documents in Kharoshthi script which clearly belonged to the early centuries of our era—and thus to the very period of Tukhara, i.e. Indo-Scythian ascendancy.
Further striking evidence of the often-proved accuracy of my old pilgrim guide came to light when I discovered that the rampart of the circular fort, obviously built after his pass-
' age, was in one place actually raised over a bank of refuse
IIIF Color HighRes
