Sec. in HSOAN-TSANG'S NOTICE OF P`I-MO AND MARCO POLO'S PEIN 453
too late did we find out that they knew nothing of such a route as I wished to take. Too timid to admit their ignorance, they thought it safest to guide us further and further south, where at least there was no risk from the dreaded Taklamakan. Thus after leaving on March 23 the left bank of the river at a great bend which the latter makes near the grazing-grounds of Kenk-kiök, we passed into a belt of dunes 20 to 3o ft. high, and further on crossed a big Dawan such as always flanks a riverine area in the desert. After about eight miles we struck the northern edge of a wide area of swampy jungle watered by the stream which flows from the springs and marshes of Shivul, west of the Keriya oasis.
Passing through thickets of Toghraks and luxuriant undergrowth of scrub we reached Terminal
a series of large pools, in which this stream seems to find its end during the winter. Half swamps of
a mile to the south-west of these we came to the edge of a long stretch of boggy marsh, treacherously covered with light sand, through which a safe passage for the ponies could be found only with great difficulty. The ground was quite impracticable for the camels, which accordingly had to be sent northward to turn the obstacle by a détour of many miles. The local knowledge of our guides now quite gave out, and though there were everywhere the tracks of flocks that had grazed here during the winter, we did not succeed in finding a single shepherd to help in guiding us. Fortunately we came at last, after one and a half miles of marsh, upon firm sandy ground near the pasture of Shakün-Öghil. South of this we found the course of the Shivul Darya, which flowed here as a limpid and fairly rapid stream, fully 6 to 8 ft. deep, in a winding but well-defined bed about 15 ft. broad. This helped once more to guide our guides. We followed it upwards to a lake-like depression they called Kazan-köl, and ultimately, after a long and tiring day's march, arrived at the solitary little shrine known as Arish-Mazar (` the Mazar of the Cross-ways'), ensconced in a grove of splendid old Toghraks. A large pond in front of it was said to receive water from another spring-fed stream that traverses the area of sandy jungle between Keriya and Kara-kir Langar. The camels did not arrive until close on midnight, the big bonfires we had kindled en route having helped to guide them.
After the experience gained of the value of our guides, there was no alternative but to March to resign myself to letting them reach again familiar ground near cultivation before striking off Darya. into the desert. Accordingly, on March 24, when the thermometer again registered a minimum temperature close to freezing-point, we made our way south-westwards through a belt of sandy jungle, in which the water of another spring-fed stream, the Kara-kir Darya, finally loses itself. Here, at the grazing ground of Kara-chilan, we got hold of some shepherds, who, after prolonged protestations of complete ignorance of any and every route, were prevailed upon to guide us at least to the northernmost portion of the Domoko oasis, the nearest in the direction I was aiming at.
The track we now followed led through a maze of tamarisk-covered sand-cones, standing Remains of closer together than I had seen them anywhere before on the borders of the true desert. In Ak-raz. their midst, at a point about four miles to the north-west of Kara-chilan, where erosion had created a stretch of ground somewhat more open, we unexpectedly came upon the unmistakable remains of some ancient settlement. The shepherds called the spot Ak-taz, or simply ` Tatilik'. The latter designation, general as it is, seemed appropriate, seeing that the remains consisted mainly of broken bits of old pottery and similar small débris. The fragments were mostly lying on the top of small loess banks rising 8 to 10 ft. above the rest of the ground, and evidently marking the extent of erosion which the soil had undergone where not protected by remains or otherwise. Close to some tamarisk-covered cones the foundations of a few mud