________________
SEPTEMBER, 1917] A THIRD JOURNEY OF EXPLORATION IN CENTRAL ASIA
195
of these materials were available in the immediate vicinity, was specially adapted to with stand it. Even where the watch-towers, once massively built in sun-dried bricks or stamped clay, had been under-cut by erosion at the base and been subsequently reduced to shapeless low mounds, difficult to recognize from a distance, the direction of the wall still clearly revealed itself, as it stretched away in a characteristic straight line across wastes of gravel or drift-sand.
The remains proved to have suffered most along that stretch of ground where the Limes, after crossing the Su-lo Ho to its right bank east of An-hsi, ran close to the deep-cut riverbed, and in a due easterly direction. On the bare riverine loess crossed here by the wall the erosive force of the prevailing north-east winds, blowing down with great violence from the gravel plateaus of the Pei-shan, could fully assert itself. But even where all structural features had been completely effaced it was easy for us with the experience gained elsewhere accurately to determine the position of the posts once guarding the border, from the fragments of pottery, coins, metal objects, and other hard débris which could be picked up at these points from the wind-worn surface. It was quite an exciting chase to search for these indications, and my Indian assistants and Turki followers had by now become expert in the game.
Where the Su-lo Ho valley bends sharply southward the line of the Limes was found to turn to the north-east, and to approach closer and closer to the foot of the Pei-shan. The ground crossed by it had remained so far unsurveyed, and the difficulties of our search were much increased by the distances which separated the long-forgotten border from the nearest water. Fortunately the days had now grown longer, and I was able to take out my little detachment of diggers mounted on the big hardy donkeys which abound at the oases of this region. Ample finds of ancient Chinese records on wood, articles of furniture, fragments of arms and implements rewarded the rapid search of the ruined watch-stations. That all these had been left behind by the Chinese troops, who during the first century before and after Christ had guarded this most dismal of frontiers, was made clear on the spot by conclusive archæological evidence. The finds of records still await expert examination by M. Chavannes, my learned Şinologue collaborator at Paris. They may be expected to furnish an important addition to the collection of early Chinese records resulting from my former explorations, which he had published in 1913.
Interesting light was thrown on the climatic conditions prevailing here from early times by the fact that here too the inscribed slips of wood, the “waste paper," to use an anachronism, thrown out of ancient office-rooms, were found often in refuse layers covered by a few inches only of gravel or débris. Their preservation in such conditions presupposes a remarkable dryness of the climate for the last two thousand years. Apart from this and the uniform barrenness, there was considerable variety in the natural features of the ground traversed by this eastern portion of the Limes. Thus all the more opportunity presented itself of observing the remarkable skill and topographical sense with which those old Chinese engineers of Han times had adapted their defensive border-line to different local conditions.
That they were prepared for great and sustained efforts demanding real powers of organisation in the face of formidable natural obstacles was clearly demonstrated when, some 30 miles to the north-east of the little oasis of Ying-p'an ("the garrison ") we found the Limes boldly carried into and through what since ancient times must have been a big area of drift-sand. Where not completely buried by high dunes, the wall built with