may have remained under Tibetan influence somewhat longer. That the great mass of the Tibetan records found at Mazar-tagh, on wood and paper, must be assigned to this period cannot be subject to doubt ; for the inference drawn from their language and contents is fully confirmed by the evidence of the coin finds already set forth above, as well as by the exact date furnished by one of the Chinese documents which will be presently mentioned. It is equally certain that the prevailing character of the contents is that of short military reports, requisitions, statements about arms, supplies, and the like, as might be expected at an advanced post controlling an important route. But the inventory prepared by Dr. A. H. Francke from which this general information is derived is not at present within my reach, and, as his published notes on the Tibetan documents of my collection 7 do not distinguish between records from Mazar-tagh and Miran, I must leave the examination of any data of antiquarian or topographical interest to be gathered from them until later.
Neither Tibetan nor preceding Chinese political control is likely to have interfered with the use of the Khotanese language for purposes of local administration and personal communication in the Khotan region. This observation resulted already from previous finds of documents in Brâhmi script (Cursive Gupta) and Khotanese language at Dandan-oilik and elsewhere, and the considerable number of similar records found at Mazar-tagh, over seventy in all, has fully confirmed it.8 They are mostly on paper, but a few are on wood, and some in complete preservation. On certain of the Brahmi paper documents red seal impressions can be traced, such as are frequent on Tibetan papers both from this site and the fort of Miran. A number of bilingual records, with Khotanese text on one side and Tibetan or Chinese writing on the other,° deserve special notice. They attest the need which must have made itself felt in administrative routine for the concurrent use of the local language along with that of the power in military and political control.
Here it may conveniently be mentioned that the fragmentary papers found in the Mazartagh refuse-heaps include also two in Uigur and another small piece, M. Tagh. a. 0048, showing a script which seemed to me to be derived from Aramaic and possibly Early Sogdian. If the much-effaced writing on the fragment of a wooden tablet, M. Tagh. a. III. oo61, is really Kharosthi, as I thought at the time, it must obviously go back to a very early period of the occupation of the site. On the other hand, a paper fragment bearing what seemed to me a line of very cursive Arabic writing need not necessarily take us lower down than the Tibetan period ; for we know that the Tibetans were in contact with the Arabs west of the Pamirs early in the eighth century, and that Arabs from Western Turkestan actually found their way right into China by A. D. 757.10
Notwithstanding the relative insignificance of their number, the Chinese documents from the rubbish-heaps, all on paper, are of particular value on account of the antiquarian information which they furnish, and which M. Chavannes' learning and minute care have fortunately rendered accessible. It is solely on the translation and comments of that lamented great scholar that the following observations are based.11 In the first place should be mentioned the well-preserved document, Doc. No. 974 (Plate XXXVI), which bears a full date of A. D. 786. It is an official certificate—whether a clean copy or a draft is not certain—issued by the chancellery of a high Chinese dignitary bearing the title of Tu fu shill. Owing to the uncertain reading of a character, repeatedly recurring
' See Notes on Sir Aurel Skin's collection of Tibetan in Chavannes, Documents, PI. XXXII, No. 963. For similar
documents from Chinese Turkestan, J.R.A.S., z 9 z 4, pp. 37 bilingual pieces from Balawaste and Mazâr-toghrak, see ibid.,
sqq. ; also above, pp. 467 sq. ; below, Appendix G. Pl. XXxVII, Nos. 977, 981, 982.
8 For specimens of such records, see Pl. CLI; cf. also 10 Cf. Chavannes, Turcs occid., pp. 290 sq., 299; Ancient
below, Appendix F. Khotan, i. pp. 62 sq.
9 See e.g. M. Tagh. c. 0020, Pl. CLI ; M. Tagh. b. 002, 11 Cf. Chavannes, Documents, pp. 201-17.