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

Plates CLXVI–CLXVIII show, neatly rolled up, after the fashion of papyri, over small sticks of wood
that sometimes had carved or inlaid end knobs. The length of the strips or sheets of wood of which the
rolls were made up varied from about 15 to 20 inches ;² the rolls themselves when complete were
found to extend to considerable lengths. All showed signs of having been much read and handled.
Probably in consequence of this the protecting outer fold, with the silk tape which had served
for tying up the roll, had got torn off very often.³

No attempt
at cata-
loguing pos-
sible. Where the covering folds of the rolls were intact it was easy for Chiang Ssŭ-yeh to read off the
title of the Sūtra, the number of book and chapter, and anything else usually shown there. The
information contained in those titles was of no guidance to me. The fact, however, that the head-
ings of the rolls found in the first bundles were all different disposed of my apprehension that this
great mass of manuscripts might be found to contain mainly an inane repetition of a few identical
texts, after the fashion so widespread in modern Buddhism. At first I caused Chiang to prepare
a rough list of titles ; but as the Tao-shih gradually took more courage and brought out load after
load of manuscript bundles for examination, all attempt even at the roughest cataloguing had to be
abandoned.

Tibetan
texts in
roll form. In this rapid examination of the first bundles Chiang failed to discover any colophons giving
exact dates of the writing. The Tibetan texts, of which some also emerged from these bundles, could
not be expected to help me in approximately determining the terminus a quo for the formation of
the monastic library which was manifestly hidden away in the walled-up chapel. Those found then
were also written in roll form (see specimens Ch. 05, 011, Plate CLXXIII ; Ch. 06, 07, Plate CLXXIV),
though with clearly marked sections, as convenience of reading required in the case of a writing
that ran in horizontal lines. Neither the writing nor the probable contents, evidently portions of
the canonical collections, could furnish chronological clues. But the paper, coarse and of a greyish
tint, looked decidedly later than that of the Chinese Sūtra rolls. There was a presumption for
connecting these Tibetan texts with the period of Tibetan predominance at Tun-huang, which lasted
from about A.D. 759 to A.D. 850. Hence the conclusion suggested itself that the Chinese rolls,
with their superior and manifestly older paper, would prove to belong to the preceding times of
T'ang rule.

Discovery
of texts in
Cursive
Gupta
script. But the first distinct assurance as regards the early origin of portions of the collection here
deposited came when, on the reverse of a Chinese roll on old yellowish paper (now marked Ch. i.
0019, see App. F), incomplete but over 3 feet long, I lighted upon a text written in that Cursive
Gupta script with which manuscript remains from Khotan sites and in the old language of Khotan
had rendered me familiar. Soon there emerged three more fragmentary rolls of a similar kind
(Ch. i. 0021. a, b, c ; App. F), covered on one side or both with writing in the same script. The
evidence seemed clear that at the time when the collection was formed a knowledge of Indian
writing, and probably of Sanskrit, too, still prevailed in the Buddhism of this region. The appear-
ance of Khotanese texts on the reverse of Chinese rolls distinctly pointed to a connexion existing at
the time between the local religious establishments and the places where Buddhism had flourished
in the Tārim Basin. Nor had I long to wait before another bundle yielded a mass of Pōthi leaves,
written in Cursive Gupta script and belonging, as Dr. Hoernle's analysis has since proved, to two