Sec. ii] KHARA-KHOTO AND ITS REMAINS 437
station close to the old course of the Ümne-gol branch of the river would therefore certainly have recommended itself as saving a great detour.
The ground beyond the ruined fort grew more and more sandy, with small rudimentary Ya.rdangs here and there. At a point nearly two miles from Adana-kôra we first sighted the high walls of Khara-khoto, ` the Black Town ', across a dry river-bed edged on the west by a belt of low tamarisk-cones. It was a striking sight, the most impressive perhaps that I had ever seen on true desert ground, this dead town, with massive walls and bastions for the most part still in fair preservation, rising above the bare gravel flat which stretches towards it from the river bank. A conspicuous Stapa, of distinctly Tibetan appearance, crowning the big bastion of the northwestern corner (Fig. 240) and a row of smaller Stapas on the ground outside the bastion, seemed at first sight to proclaim the predominantly religious character of the site. But that Buddhism had not held exclusive sway among those who knew the town while it was in being, became evident when on our first approach I recognized a Muhammadan tomb or ` Gumbaz ' in a conspicuous domed structure near the south-western corner of the town walls (Fig. 251). Its vaulted interior offered a convenient place for storage, and outside it I pitched my tent, while the interior of the big bastion guarding the western town gate was occupied by my men as an effective shelter from both heat and winds. How much protection from the latter was likely to be needed was sufficiently indicated by the big accumulation of drift-sand which lay heaped up against the walls, especially on the west side (Fig. 243), and the succession of violent gales, almost all from the north-west, that broke upon us with regularity every second day during the time of our stay fully confirmed this anticipation.
There was nothing in the surroundings of the dead town to impair the imposing effect created by the massive strength of the town walls and the utter desolation which reigned within. All round the walls there stretched a bare flat of gravel, supporting only here and there a small stunted tamarisk or a few tufts of thorny scrub. To the west this gravel plain was bounded by the dry river-bed that we had crossed on our approach (Fig. 235), while winding to the south and east we found a much wider branch of the same, fully half a mile across for the most part and forming a big bay to the south as seen in the sketch-plan (Pl. 17). The point where the two beds bifurcated was located about three-quarters of a mile from the south-west corner of the town, and there the steeply cut banks of the more southerly branch showed a depth of about twenty feet. To the north, at a distance of over half a mile, the gravel plain between the two branches was overrun by a big ridge of sand, rising to 40 or 50 feet in height and fixed by tamarisk-cones. Its western portion had completely blocked up the more westerly of the dried-up river-beds. To the question of the connexion of these beds higher up with the Ümne-gol I shall have occasion to refer farther on.
On the very day of our arrival at the site I sent off Afrâz-gul with some camels on a reconnaissance into the desert to the north-east. A Mongol from Ili, whom we had met on our march down the Etsin-gol and who proved less unwilling than the local Torguts to talk of old remains, had stated that he had come upon ruins in that direction which had not been visited by Colonel Kozlov and were apparently unknown to those who had served as his guides. All ponies and camels were then sent back to water and grazing at Dzusulun-tsakha, whence the camels were subsequently to return at regular intervals with water-tanks and leather ` mussucks ' refilled. Two days later, I was able to let Lal Singh start, with the animals brought back by one of these parties, on his survey up the dried-up river branch of Khara-khoto and thence across to the Mörün-gol and the terminal lake-beds. With the few of our own people who remained and the dozen of indolent Mongols as diggers, I myself set to work on the morning of May 27th upon the exploration of the remains found within the town walls and immediately outside them. The task kept us fully occupied