pattern reproducing very closely, both in its design and in its rich yellow and blue colours, that of a small piece of silk fabric that I well remembered among the relics from the refuse heaps of the watch station T. xv. a of the Tun-huang Limes. These relics are proved by numerous dated Chinese documents to belong to the period extending from the middle of the first century B. c. to about the first third of the second century A. D.4 Thus one of the very first finds at L.C. furnished me at once with a definite indication that the remains preserved in the burials of the site went back to the Han period.
My hopes were abundantly fulfilled as soon as the arrival of my diggers made it possible to start the clearing of the graves, first where they lay near the edge of the Mesa top and then about its centre. The importance of the antiquarian treasure that I had come upon became apparent almost at once ; but what impressed me even more at the time was the quite bewildering confusion in which it presented itself. Instead of regular burials with coffins and human bodies more or less recognizable, such as the first rapid inspection of the half-eroded remains had led me to look for, there emerged from the grave-pits a mass of detached human bones mixed up in utter disorder with fragments of boards once evidently belonging to coffins ; with objects of personal use, such as decorated bronze mirrors, wooden combs, & c., deposited with the dead ; wooden eating-trays, jugs, &c., used for sepulchral offerings ; wooden models of arms, and, above all, with rags of every sort comprising a wonderful variety of fabrics. Among these rags were beautifully woven and coloured silks, often showing rich polychrome designs ; fragments of delicate enbroidery and tapestry ; torn pieces of fine woollen pile carpets, by the side of numerous coarse fabrics in wool, felts and what appeared to be cotton. That all these materials were of Chinese origin or had been in Chinese use there could be no doubt, and finds of Chinese records on paper and wood confirmed it.
It soon became evident, from the way in which rags of various fabrics were often found stuck to the same bones, that these were remnants of garments which after long wear by the living had finally been employed to wrap tightly the dead bodies. Tattered and torn as all the fabrics were, yet many of them were excellently preserved, notwithstanding their dirt and sand-encrusted condition. In strange contrast with this was the fact that not a single intact skeleton was found, and that all such human bones as were not still protected by swathing plainly showed marks of weathering and corrosion. This was obviously the result of a prolonged exposure which these human remains must have undergone before their final deposit in the pits that I was clearing. Thus a variety of indications soon led me to realize that the contents of these pits must have been collected, before the final abandonment of the Chinese station of Lou-lan, from older graves which wind-erosion, decay, or some similar cause had exposed or was threatening.
Before I refer to the observations made at other sites which have since fully confirmed this conclusion, or to the valuable archaeological indications which may be deduced from it as regards the dating of the relics here recovered, it will be convenient to note a few further details about the place where they were found. As the plan, PI. 12, shows, the disposition of the grave-pits on the flat top of the Mesa was quite irregular. While a few, as already mentioned, lay close to the longitudinal edges and had in consequence become partially exposed on the outside, the rest were grouped together more or less closely near the centre. The pits were all roughly rectangular in outline and cut into the hard salt-impregnated clay of the Mesa to a depth of 5 to 6 feet. But their surface dimensions, where intact, varied greatly, from about 4o square feet in the case of pit v to
g Cf. Serindia, ii. pp. 700, 781 ; iv. Pl. LV, for T. xv. a. 002. a. As the several refuse heaps of T. xv. a. i, ii and iii date from three different and well-defined periods, I must specially regret now the absence of a more exact record as to the