Sec. ii] TO P`O-CH`ÊNG-TZO AND SHINDI 729
by a tower, probably built of clay. The opposite slightly lower end bears a platform, measuring about 24 by io feet, built of alternate layers of stamped clay and brushwood about 6 or 8 inches thick, with a thin stratum of reeds above the latter. This recalled the method of construction of which I had seen examples among the ruins of Lou-lan, in towers and foundations of Stûpas, and in circumvallations. The whole wore a distinctly ancient appearance, suggesting the remains of a watch-station or of a rallying-place for a small settlement.5
In order to visit an inscribed rock of which I had heard reports, I then proceeded down the defile. For a mile or so we passed fields on the right bank of the stream which had been roughly prepared for cultivation. For another mile ye met patches of ground on both banks which were clearly cultivable but were overrun by luxuriant scrub ; elms, willows, and wild poplars grew in profusion among it. Lower down, the defile became very confined, between precipitous spurs of rock jutting out from either side, but still afforded good grazing at different points. At one of these, known as Inkur-oiak, an almost vertical cliff, several hundreds of feet high, rises immediately above the left bank of the stream. Its foot for a distance of about 20 yards is covered with a miscellaneous assortment of shallow rock sgraffiti, roughly executed, representing horses, camels, ibex, deer, and in a few places figures of men. Interspersed with these are Cakras, tridents, Svastikas, an eight-petalled lotus, and some figures apparently meant for mystical diagrams. I t was getting too dark for an attempt to secure a photograph ; but I could see that the sgraffiti high up looked much weathered and old, and those lower down more recent. That local worship of some kind attaches to the spot was shown by a flag which a Mongol visitor had set up near it. The gorges furrowing the Shindi-tagh on either side looked extremely rugged and confined, and reminded me of forbidding ravines in Hunza or Chitral. The defile cut by the Shindi stream was said lower down to become impracticable for camels. The precipitous scarps of all these gorges and the absence of detritus bear striking evidence to the power which water erosion has asserted in this outermost range also of the western Kuruk-tagh.
After arranging for the dispatch of fresh instructions to Lai Singh, who was continuing his triangulation under difficulties amongst the much broken ranges to the north-west, I left Shindi on the following morning, March 1st, in order to gain the course of the Kuruk-darya half-way between the sites of Lou-lan and Ying-p`an. In order to secure water while on the way to this destination I was obliged to proceed first to the route connecting Singer with Ying-p`an, and after striking it at Azghan-bulak to follow it down to Toghrak-bulak. The march to the first-named place was long and monotonous, but served to bring out well the typical character of all these latitudinal valleys or rather peneplains of the central Kuruk-tagh. It led all the way along the northern foot of the hill chain that stretches from above Shindi eastwards, the relative height of its flat-topped ridges gradually sinking as it approaches the Ala-tagh near Azghan-bulak. The ground, at first covered with fine alluvium supporting scrub and a few elms by the side of dry flood beds, changed into a stony Sai as the almost imperceptible watershed towards the drainage area of Nanchan-bulak drew nearer. Some miles before this was reached there cropped up along the axis of the wide valley on our left a series of low broken ridges, which continued with interruptions all the way to Azghan-bulak. They clearly represent the last remnants of a completely decayed range parallel to and intermediate between that of the Ala-tagh on the south and the Mohur-shan on the north. A conspicuous peak in the latter rising to over 8,400 feet showed through the dust haze raised by the persistent west wind, and relieved the monotony of the desert landscape. No spring or well is to be found on this ground before reaching Azghan-bulak ; but tamarisk growth and some elms appeared farther east where dry Nullahs coming from the Mohur-
5 The Stùpa symbol shown in the map is a draughtsman's error in place of the symbol for a watch-tower.
Rock carvings, old and new.
March to Azghanbulak.