Sec. iii] REMAINS OUTSIDE KHARA-KHOTO 449
of 1920, as to the Hsi-hsia dictionaries and certain identified Buddhist Canonical texts included in
the Petrograd collection.8 Nevertheless it may be of use to record here some brief indications concerning the textual and other remains that the search of the ` waste ' left behind by the fortunate
discoverer of the site allowed us to recover ; for their examination may throw some light on the general character of the deposit and thus help us to define its bearing upon archaeological questions connected with the whole site.
In the first place it is of interest to note that, among the literary remains recovered here, texts in Hsi-hsia language, whether written or printed, vastly preponderate. Leaving aside fragments of small size, the rough inventory prepared when these materials were transmitted to collaborators shows a total of over eleven hundred written, and about three hundred printed, leaves (many, of course, incomplete) in Hsi-hsia language, against fifty-nine and nineteen, respectively, in Chinese. This preponderance of Hsi-hsia texts, probably for the greater part of a Buddhist religious character, contrasts strikingly with the small proportion that Hsi-hsia records bear to Chinese records among the papers recovered from the rubbish-heaps of the town. Assuming that the deposits in both places date approximately from the same period, the conclusion suggests itself that Chinese writing prevailed, for purposes of secular business, even under the Tangut domination, over the cumbrous ` national ' language and script favoured by the ruling dynasty. The great rarity of Tibetan texts from K.K. 11—only thirteen complete folia are recorded in the inventory—is also of interest, when compared with the large number of Tibetan materials from K.K. v. The bilingual leaf, KK. xi. 0234. k, Hsi-hsia with Tibetan, reproduced in Pl. CXXXIV, with a transcript kindly furnished by Dr. Laufer of Tibetan corresponding to Hsi-hsia characters, justifies the hope that the presence of complete texts of this kind among the Petrograd materials will facilitate progress in the study of Hsi-hsia. Uighur script is represented by a single written piece only, while of Brdhcni-Chinese prints we have two specimens (see K.K. I1. 0293. a, Pl. CXXV).
Some quasi-palaeographical interest attaches to the fact that of the remains of Hsi-hsia and Chinese texts, whether written or printed, almost all are of the oblong book form, which, originating from the ` concertina ' arrangement of leaves illustrated by later Chinese manuscripts from the Chien-fo-tung hoard,9 has been in regular use for block-printed literary products in China since the early Sung period. The number of manuscript remains in Hsi-hsia or Chinese in the roll form that prevailed all through Tang times only slightly exceeds two dozen.'a They obviously represent an archaic survival, just as the use of silk as writing material instead of paper, of which a few Hsihsia manuscript pieces furnish examples, may claim descent from very ancient Chinese practice. In conclusion, passing reference may be made to the great quantity of written pieces torn into very small size that were discovered amidst the wreckage. It seemed difficult to believe that their reduction to scraps, often of minute size, could be due solely to careless digging, though of this, too, there was unfortunately evidence in booklets and convolutes that clearly had been cut through by the hoe or pickaxe. Is it possible to assume that those scraps owed their survival to a quasi-religious custom which compelled the preservation of all writing, however much defaced or injured?" The present Chinese practice of carefully collecting all bits of ` waste paper ' from streets, shops, &c., in specially set up boxes, with a view to ceremonial burning would offer a certain analogy.
8 See Ivanov, Monuments de l'écriture Tangout, in J. Asiat., 1920, janvier—mars, pp. 107-9 ; also the articles of M. Ivanov there quoted from the Izvéstiya of the Russian Academy of Sciences, 1909, iii. pp. 1221-33, and 1910, v. pp. 831-6.
9 Cf. Serindia, ii. pp. 765, 802, 920. As in these ` concertina '-shape manuscripts and in Chinese block-printed
books, the reverse of the I-Isi-hsia leaves (see Pl. CXXXVI, CXXXVII), whether written or printed, was left uninscribed. sa For specimens see Pl. CXXXV, CXXXVII.
10 For corresponding remains preserved in little packets among the manuscripts of the Chien-fo-tung hoard, cf. ibid., ii. p. 82o.