taking. In addition, the maintenance of the dam in effective condition to face the summer floods necessitated the annual employment on repairs of large contingents of men, considerably in excess of the number which Domoko alone could furnish.
By these efforts Domoko was assured its former supply of kara-su (` black water ') or water from springs, which everywhere in these oases along the southern edge of the desert is indispensable for irrigation during the months preceding the summer floods. In addition a fresh and constant supply was forthcoming from the springs which appeared at the head of the newly formed Yâr. This was turned to use by the formation of the Malak-âlagan colony. The steady growth of the latter was attributed to the water of these springs having remained uniformly ample. Owing to this fact the new settlement was declared to be less dependent than Domoko itself upon the varying amount of ak-su, or summer floods. It was interesting to note the uniform assertion that the volume of ` Kara-su ' water available for the canals of Domoko had not been reduced by the formation of the new springs. This was fully borne out by the occupation of ` new land ' towards the western bank of the Domoko-yâr, already noted, and by the vigorous reclamation of former desert waste which on my subsequent visit in March, 1908, I was to find proceeding in the direction of Gulakhma. The obvious explanation was that the new springs were draining strata saturated with subsoil water far lower than those which feed the sources of the Domoko stream some eight or nine miles higher up on the ` Sai '.
The observations I was thus able to gather at the great dam of Domoko help to bring out facts which offer a wider interest both to the geographer and the historical student. I t is clear that the opening of the Malak-âlagan colony was the direct result of a movement by which the Domoko stream had endeavoured to carry its water once more towards the old village site abandoned about A. D. 1840. But for the timely construction of the dam practically all the available water would have flowed into the Yâr, and the canals irrigating the present village lands would have run dry. In that case, it is safe to assume that the settlement would have been shifted back again to the site of ` Old Domoko'. Of this I had direct proof on my subsequent visit in 1908 when I found the old village lands gradually approached again by the surplus water of Malak-alagan which was being brought northward along the old canal alignment, still traceable, even without any aid of water from the Domoko stream.
In the fate of abandonment which threatened the extant oasis and which was averted only by an engineering feat on a scale unusual for these parts, lies a clear proof that changes in the cultivated area of this region may take place on ground peculiarly situated through physical causes which have nothing to do with desiccation and a consequent diminution of the water-supply. However potent a factor desiccation has been in determining the economic conditions of the Târim Basin and the historical development dependent upon them, this instance must warn us against necessarily attributing to its action every deserted site which archaeological inquiry may there reveal.
Equally instructive is the lesson we may draw from it as to the influence which a quasi-historical factor, the assertion of human energy, must have in respect of such changes. Had it not been for the effective administration introduced on the Chinese reconquest of the province after the disastrous upheaval of the Muhammadan rebellion, and for the economical development which it has fostered, the damming up of the Domoko-yar would certainly have proved too great a task to be attempted with local resources. The villagers of Domoko would have been left to face their calamity as best they could, and would probably have adopted the remedy indicated by local tradition which tells of repeated shifts of cultivation backwards and forwards.6 In the same way
5 Cf. Ancient Khotan, i. p. 459.