National Institute of Informatics - Digital Silk Road Project
Digital Archive of Toyo Bunko Rare Books

Sec. Hi]   THE ANCIENT REMAINS OF NAN-HU   619

down for purposes of defence, it appears to me very unlikely that the burial-grounds of the Chinese garrison should have by design been placed to the south, and thus outside the wall, or kuan wai-fou. Taking all these considerations together, I find it easier to accept the local view, which after all Probable

may be based on tradition, and to recognize in that broad gravel embankment not a defensive purpose of

agger but a guiding line. In any case it is clear that the construction of such an embank- 8g ment for close on twelve miles through the desert was a work of considerable magnitude. It seems safe to assume that it would never have been undertaken except at a period when the Nan-hu oasis and its population were far greater than at present, and even then only because the settlement there situated, the ancient frontier station of Yang kuan, was one of particular importance.

Before giving the reasons for this identification of Nan-hu with the ` Yang barrier ', I may Ancient

briefly record here what I observed on a further inspection of the burial-ground passed on my first b ound.

approach to Nan-hu. Besides a widely-scattered series of small circular mounds, rising only a few feet above the level expanse close by a narrow handle-like ridge of gravel, I visited the quadrangular enclosure of which I have already spoken. It contained two tumuli placed side by side near to its north face and both turning their ` handles ' due south towards the gate meant to give access to it. Whereas the lines of heaped-up gravel that marked the enclosing walls were so low as to be barely visible while the sun stood high, the western tumulus showed an annular rim about three feet high with a slight depression in the centre (Fig. 165). The diameter of the whole was about twenty-three feet. The low gravel ridge, forming the ` handle ' southward, had a length of about seventy-five feet, with a width of about two feet where it was widest. In the centre of the north segment of the annular tumulus a small heap of stones lay around and over a lump of red clay, similar in material to the sun-dried bricks of the gate to be described presently. A cutting was made through the low tumulus right down to the natural soil of gravel and hard clay without disclosing any remains whatsoever. I now regret that I did not test the gravel ridge of the ` handle ' in the same way, for, from what my later experience at the Tang burial-grounds of Turfân showed, it is highly probable that we should have struck there the top of the deep-cut trench by which the tomb, carved out from the solid clay at some depth below the tumulus, was approached when the deposition of the body or bodies took place.

The gate' on the south consisted of two fragments of wall about five feet thick, and built of ' Gate ' to

coarse bricks with an average size of nineteen by ten inches and a thickness of four inches. It is closure grave

that what I took for bricks at the time were only fairly uniform pieces of hard clay which had been cut out from some stratified alluvial deposit near the dry river-bed, or obtained in the course of the excavation made for the tomb chambers below the tumuli. The wall to the west of the entrance was badly broken, but the one to the east still rose to about fourteen feet in height. Its length was five and a half feet, and this was continued eastwards for another three and a half feet with a reduced thickness. Close by the north and south faces of this wall was a narrow, terrace-like platform about two and a half feet wide and four feet high. I have already stated the reasons which obliged me to abstain from any attempt to search these burial-places, and without actual excavation their date cannot be definitely established. But comparison of their surface features with those which in 1914 I observed at the cemeteries of the Turfân region makes me inclined to believe that these graves near Nan-hu cannot be later than the Tang period.

4x2