6o DISCOVERY OF HAN RECORDS CH. LV
But even in the immediate neighbourhood I made an observation which supports that assumption, though at the time I did not realize its true import. On the previous day's reconnaissance to the south-west, towards the next and last tower visible on that side, I had noticed, on the open salt-encrusted steppe intervening, and at a distance of less than a mile from T. xxix., some rough enclosures built of salt-impregnated clay lumps, and in the middle of one a miniature chapel, half-ruined, made up of the same coarse material. Looking back in the light of the indications since gathered, it now appears to me probable that this modest substitute for a shrine, manifestly of recent construction, represents the last lingering trace of the cult which those leaving or regaining the border wall of the Empire once paid to the sanctuary at this ' Gate.'
The westernmost tower extant on this part of the line proved less than two miles distant from T. xxix., and was a solid square mass of stamped clay. The small erosion terrace on which it was built made it a conspicuous landmark on this dismal salt-covered flat. No ancient remains of any sort except broken pottery could be traced near it, and far-advanced erosion had left neither trace of the wall nor any chance for digging. But I visited this tower (T. xxx.) again on my return march to Tun-huang, and then found that the deep-cut cart-track marking the route from the oasis to Hami actually passed between it and the tower T. xxix., which I now believe to have stood by the ancient gate through the wall. Though the view from the terrace of T. xxx. was open, no other ruin could be seen to the west except the tower T. xxiv., which I had already examined on my way north of Shihpan-tung. It just showed its top over a maze of clay terraces to the west. So my survey in this direction was completed.
On March 3oth, when we had a considerable fall of temperature down to thirty degrees of frost Fahrenheit,
with the wind veering round to the west, I took my men back to the ruined watch-station (T. xxvii.) which had yielded the first dated records (Fig. 164). The débris adjoining the tower proved to belong to a room about
IIIF Color HighRes
