Sec. ii] EXCAVATION OF THE ENDERE TEMPLE 425
and from the indications furnished by the pagination figures, I concluded at the time that we
possessed in these leaves the greatest portion of the MS. Dr. Hoernle's exact analysis 2 has
fully confirmed this conclusion, and has made it probable that all the 46 leaves of which the
MS. when entire appears to have consisted are represented among the recovered pieces. Three
of the folia are complete, their right and left halves having been pieced together with certainty
(for a specimen see PI. CI X). The full size of the leaves must have been about 14 by 3 in.,
the text being written in five lines on each side in clear upright Gupta characters, which
Dr. Hoernle assigns to the seventh or eighth century. The text is a Buddhist canonical work
of the Dhârani class. From the uniformly faded and perished state of the lowest leaves in
E. i. 5 (left halves) and E. i. 40 (right halves) it appears very probable that the Pôthi in its
entire state must have lain for a long time before the left and right portions got separated.
Possibly the break occurred in course of the excavation made into the central base ; but there
can be no doubt that this burrowing had caused the dispersion.
The experience gained at the Dandan-Uiliq ruins suggested from the first that the MS. Detached
leaves found had originally served as votive offerings. Proof for this surmise was soon forth- as votive leaves
coming in plenty. Close to the north foot of the central base there turned up two closely- offerings.
packed rolls of paper, which might have fallen from the pedestal of the image above. One
contained the fragmentary leaf E. i. 6, about 9 by 3 in., written in slanting Central-Asian Brâhmi
and a non-Sanskritic language, which P1. CIX reproduces. The other roll (E. i. 7), which was
secured with a paper strip still closely wound round the centre, and which could be opened
only in the British Museum, resolved itself into four large folia measuring i 8 by 3â in., written
in bold upright Gupta characters (see Plate CXI). As Dr. Hoernle's examination has shown s,
they must have been taken from an extensive work composed in the non-Sanskritic language
which he tentatively distinguishes as proto-Tibetan' 4. The frequent occurrence of the Sanskrit
term hkaisajya suggests that its subject was medical or magical. Among the remaining finds
in Brähmi there were three small pieces of leaves (E. i. 9, 33), which lay on different parts of
the projecting moulding at the foot of the central base facing to the north, west, and south-
west, respectively. As two of these pieces must have belonged to the same leaf, their position
made it at once clear that it was torn up purposely in order to furnish fragments for deposition
before different images.
This conclusion forced itself upon me in a still more striking way when all round the foot Finds of
of the central base, as well as on the floor near the pedestals of the images in the north-west, Tibetan
south-west, and south-east corners, there turned up besides a large and well-preserved sheet of
Tibetan writing (E. i. i t ; see PI. CXVIII) and a number of miscellaneous Tibetan fragments
(E. i. 15, 19, 20, 25, 31 ; see Pl. CXVIII), which, by the uniformity of the clear, well-formed writing
and of the paper, I could easily recognize on the spot as portions of an identical Pôthi. The
widely different positions in which these fragmentary pieces were recovered will be found noted
in the list. Some picked up underneath the walls or on the floor might have been blown away
from their original place of deposition ; but most of them had evidently been retained by the
accumulation of drift-sand in the place where the last owner of the MS. had intended to pro-
pitiate with them the various divinities. The fragments recovered, 27 in all, have been proved Tibetan
by Dr. Barnett to have originally formed part of a large Pôthi, about 171 in. long and 2; in.1i °Q~~~ba-
broad, containing the canonical text of the S'dlistamba-siclra embodied in the collection of the sùlra.
2 See his note xxiii. below, p. 439. ' Comp. regarding this language, represented also among
s See Dr. Hoernle's note xxvi. below, p. 440. the Dandan-Uiliq finds, above, pp. 271 sq.