I accordingly started on New Year's Day, 1908, to the south-west with light baggage and a small posse of labourers and guides. All Korla was interested in what, no doubt, was locally understood to be a real ` treasure quest '. In spite of the bitter winter cold and expected hardships, men in numbers offered themselves as labourers, a novel experience on such occasions. Two short marches brought us first to the extreme south-west fringe of Korla cultivation beyond Tazken, and then across fine grazing land and luxuriant riverine jungle to the Konche-darya. We crossed the hard-frozen river below the Konche-mazar where it fills an unfordable bed over 3o yards wide (Map No. 45. D. 3), and then moved up the belt of riverine jungle accompanying the bed of the Charchakdarya, which in certain years carries flood-water coming from the Kucha side.
On this march to the well and shepherd-station of Döru-sukte (Map No. 45. c. 3) I first noticed traces of a form of cultivation which is widely practised along the lower courses of the Tarim, the Inchike, and Konche rivers, and is not without antiquarian interest. Stretches of open level ground which are inundated after several years' interval by a particularly high flood can be sown in the following spring and, as they retain sufficient moisture from that big watering, yield then an abundant harvest. But only in very rare instances can cultivation be continued for a second year by means of small irrigation cuts from the river-bed. Cultivation carried on in this fashion occasionally leads to transient occupation, and objects left behind at the end of it may, when turning up as ` finds' centuries later, give rise to quite erroneous conclusions as to the former existence of permanent settlements on such ground.
It was in the desert belt south between the Charchak and Inchike river-beds that Musa. I.Iaji had, as he declared, seen his ruined site. The short expedition which I made into this wholly unsurveyed area, and which was facilitated by the ice found in a newly formed lateral lakelet of the Inchike-darya known as Jigda-sala (Map No. 45. B. 4), proved very instructive geographically. It showed me in typical form the constant changes brought about on this ground by shifting river-courses and the concomitant struggle of the vegetation belts which they produce with the drift-sand, ever close at hand to follow up local desiccation. But after several days' search in the desert Musa liaji had to confess his inability to find the old town ' which he still firmly believed that he had seen and approached. Fortunately I had taken care from the start to have him accompanied by level-headed Darôghas from Korla. It was due to their careful search and topographical sense that on a patch of bare clay steppe surrounded by lines of dunes, 8-10 feet high and strewn with dead Toghraks and tamarisks, a small ruined circumvallation was ultimately located (Map No. 45. c. 3).
It was a circular rampart of earth, about 18o yards in circumference, about 3o feet thick at its base and rising with its narrow top to about 12 feet above the present ground-level. A few pieces of coarse but hard pottery lay near what appeared to have been the entrance. No other marks of occupation were found on the surface ; but it must be remembered that there were no signs of wind-erosion either, which alone could display such relics on ground of this kind. That the enclosure was meant to serve as a place of safety or as a watch-post is certain, and its size and shape recalled Merdek-tim.10 But there was nothing to help towards determining its age. On ground subject to moisture through riverine changes at recurring periods no ancient refuse and the like could possibly survive. Musa. I.Iaji stoutly denied that this ruined enclosure was his ` old town '—but anyhow he could show us no other. Going about 21 miles north, a dry river-bed lined with dead Toghraks was crossed, and beyond this, after another 1; miles and not far from the edge of the living jungle belt, I was shown a small ruined ` Gumbaz', about 8 yards square, built of clay lumps. It • manifestly marked a Muhammadan grave, and did not look of great age. I may note here in passing that
Start for desert site.
Intermittent riverine cultivation.
Search of desert belt south of Charchak river.
Remains of circular enclosure.