so as to overlap upwards.4 This direction of the oblong scales claims here the more notice because in almost all these Tun-huang paintings the upper or breast portion of the armour is depicted with round-headed scales overlapping each other downwards.
In most of the relievos and paintings of Central-Asia or Graeco-Buddhist art known to me where representations of scale armour are met with, round-headed scales appear in the breast portion and oblong ones are confined to the skirt. But it is significant that among all the lacquered leather scales of the Miran fort not a single one of the round-headed shape was found. This seems distinctly to suggest that the leather mail suits worn by these Tibetan soldiers were made up entirely of oblong scales. We actually have an example of such an arrangement in the old Tibetan suit of mail from Lhassa, with oblong scales overlapping upwards. Similarly, in the small stucco relievo of a mail-clad warrior, Mi. xii. 0015, 0017 (Plate cxxxv), from the
Ming-oi Site of Kara-shahr, all the scales are oblong, though their exact arrangement cannot be macle out. Also in the Lokapâla picture, Ch. lv. 0020, we find oblong scales both above and below the waist, represented here with a clear overlap upwards. In support of the view just mentioned, reference might be made also to the fragment of leather armour, M. 1. vi. ooi, which by its curved shape and unusual thickness suggests use as a shoulder-piece. Another elaborately worked fragment of large size, M. I. viii. 001 (Plate L), seems to have been intended to protect the throat, and may have belonged to a leather helmet ornamented in a style to match the lacquered scale armour. Curiously enough, the representation of a helmet, apparently of scale armour, survives in a fragment of pottery, M. x. 0056 (Plate LI), bearing incised the head of a warrior. Its good drawing contrasts strikingly with the roughness of the make and material of the pottery.
That among the occupants of the fort there must have been, besides the military, a strong clerical element is proved not only by the quantity of written records, but by the finds of wooden and reed pens (M. viii. 0012 ; xxI. 1 ; xxiv. 009 ; xl. 0014 ; )(lii. 003, Plate LI). It is strange to note by comparison the few remains of arms found within and outside the fort. Apart from a fairly well-preserved sling of stout goat's hair, M. I. 0081 (Plate L), they comprise only fragments of arrows (M. I. 0059, Plate LI ; ii. 0026 ; ix. 009). A speciality of the site, no doubt connected with Tibetan usage, are the seals cut into tips of horns, of which three, M. I. vii. 31, 003, 004 (Plate LI), turned up in the same apartment.4a The official seal impressions in red ink found on so many of the Tibetan paper documents must, judging from their shape, have been made with seals of a similar kind. The Tibetan inscription on one of these horn seals, M. 1. vii. 31, should be noticed here, because Dr. Barnett is inclined to read it as containing the ` non-Tibetan syllables ân-iôné' and to believe ` that they possibly represent an attempt to render the Western " Anthony "'. If this interpretation proves well founded, we might well think of the owner of the seal as a Nestorian Christian, settled at Mirân in Tibetan employment. The discovery of fragments of Christian texts at Turfân sites has supplied definite archaeological evidence for what we otherwise knew of the spread of Nestorian Christianity into Eastern Turkestân about this period.6
A variety of implements connected with spinning and weaving, such as M. 1. iv. 009 (Plate LI), 00I2-13 ; vii. 005. a, b (Plate LI); viii. oo8-10 ; xxiv. 005, 0037, prove that the garrison of the fort found time for domestic occupations. The numerous pieces of stout netting found, of which
4 Cf. for such Lokapâla figures in scale armour Pl. LXXXIV, LXXXV, XC. All the ten figures there reproduced, with one exception, have the kilt-like lower portion of their scale armour made up of oblong scales which are represented as overlapping upwards. In Ch. 0010, Pl. LXXXIV, the skirt of the mail suit is shown with scales which have a rounded end and overlap downwards, just as those of the upper or
breast portion of the figure.
4a The seal impressions from M. x. vii. 003, 004 have, through inadvertence, I regret, been reproduced upside down in Pl. L,I.
6 Cf. e.g., Von Le Coq in S.B.A. W., xix, 1908, p. 1202 ; also Ancient Kholan, i. pp. 71 sq., 87.