Sec. ii] OLD REMAINS NEAR AN-HSI & HSÜAN-TSANG`S YÜ-MEN KUAN io95
Limes wall, and at what might be called the regulation distance from the ancient watch-tower to the west, was probably of ancient origin was an inference which readily suggested itself. That local
tradition of some kind had influenced the selection of this exact spot where the road crossed the line of the Limes, for what Chinese convention treats as the mark of a district headquarters boundary,
seemed also likely enough. But only on my return here on the occasion of my visit to An-hsi in
April, 1914, did I notice that close to the tower there stood also a modest little shrine, evidently owing its existence not to any requirement of official etiquette, but to a motive far more effective—
tenacity of local worship. As explained above,ß my explorations of 1907 and 1914 have shown me again and again that wherever roads or tracks still frequented pass through the line of the ancient Great Wall some mark of continued local worship almost invariably survives. So I see no reason to doubt that the same was the case here, too, and that it was the pious tradition clinging to this spot which led to its being also chosen for the ceremonies of official welcome and farewell.
Beyond to the east the trace of the line of wall has completely disappeared on the soft scrubby soil. But at a distance of about a mile from the road I found what perhaps is the last remnant of
a badly decayed watch-tower in the shape of a small but solid clay mound, some 8 feet high and 5
feet in diameter. There were no indications here to guide me in a search for . the further continuation of the Limes. But on June 24, when marching to Ch`iao-tzû, I came upon its line again some 7
miles away to the east-south-east, as will be related below, not far from the gravel glacis of the foot-hills.e"
Why the ancient border wall should have been erected on a line which left the arable land around the present An-hsi and lower down outside its protection is a question which cannot be definitely
answered. Possibly the Su-lo Ho then followed a course which lay further to the south and by its
inundations rendered it advisable to keep the Limes line some way off on slightly higher ground. In any case the same observation applies to the whole of the Limes between An-hsi and Tun-huang.
Its line keeps well away from the river along the section T. xxvi-T. xxxv which has been described above,' and my explorations of 1914 have proved the same also with regard to the remaining portion of the wall that links it up with the traces of the Limes near An-hsi.
Apart from these remains of the Limes, the ruins of a walled town about a mile and a half to the south-south-east of the present An-hsi were of archaeological interest, notwithstanding their
relatively recent date. According to local information, which En T`ai-chin, the district magistrate,
confirmed, this earlier site of An-hsi town had been deserted towards the close of the eighteenth or the beginning of the nineteenth century after repeated destruction by fire. The interior proved
to be absolutely clear of structural remains, just like the barren waste of fine gravel in which the
enclosing walls rise. These form a square of about 600 yards and are solidly built of stamped clay, with a thickness of about 15 feet (Fig. 2 39). There was nothing to indicate that they were of con-
siderable age, and in any case they are likely to have been kept in tolerable repair until the town was
abandoned. The more striking was. the extraordinary effect which wind-erosion has produced upon them. The east face, and to a somewhat lesser extent also the west, has been breached at short
intervals by deep fissures due solely to the scouring of wind-driven sand. These breaches reached down in many places to within 5-6 feet of the ground, and at the north-east corner, shown by Fig. 239, had been carried down so low that the wall there has been razed off altogether.
It was easy to study here in full clearness the results which the action of the powerful winds, rarely ceasing at An-hsi for more than a few days, is capable of achieving where there is nothing to
stop or reduce the force of their corrosive weapon, the drift-sand. It was obviously the prevalent east or east-north-east direction of these winds which accounted for the cuttings being always broader on the east than on the west face of the wall, thus presenting the appearance of trumpet-shaped
s See above, p. 602. ca See below, pp. 1099, 1039 sq: 7 Cf. above, pp. 604 sqq.; Map. No. 8z. A, B. 3.