THE LOU-LAN SITE [Chap. XI
the corner,xbeing dovetailed. In front of the south-east beam there lay, stretched out on the eroded slope, a portion of the timber frame of the wall which it had once borne, just as it probably may have been thrown down by the wind after centuries of exposure. The photograph, Fig. 104, shows it in the foreground. The beam which had carried the roofing, r7-} feet long, was still intact, with the dowels into which posts had fitted. The two posts at the eastern end of the beam were still fitted into their dowels and held together in their original position by horizontal joining pieces (see Fig. 104 and elevation of frame drawn to scale in Plate 27). Two dowels, on either side of the centre of the roof beam and 5' 8" apart, indicated the position of two jambs once flanking an entrance, and as the fine and well-preserved beam, L.B. H. 0037 (Plate XXXI), with mitred ends and floral scroll carving on its face, has a length exactly fitting this width of entrance, it is highly probable that it once served as the lintel over this very doorway.
I was unable to trace any indication of the architectural features of the interior of the shrine, a circumstance which the disturbed condition of the remains suffices to explain.' It may, however, be surmised that the elaborate lathe-turned balusters, L.B. II. oolo, 0038-45 (Plate xxxIlI); and the rails which they formerly supported, L.B. iI. 0046-5i (Plate XXXIII), had once served to enclose some portion of the space within the shrine, and probably that containing objects of worship. That there myst have been images in the shrine may be considered as certain, however scanty the relics which prove it. They consist of the arm of a wooden,figure, carved in the round or high relievo, L.B. ii. 0052 (Plate xxxIv), and two small pieces of stucco relievo, of which one, L.B. u. 6, probably formed part of the head-dress of a large stucco image, while the other, L.B. Ir. 7, is likely to have belonged to the flame border of a large halo or vesica in relievo as found so often behind Buddhist images of Tang times, from. Khotan to Tun-huang. The surface of the wooden arm has suffered so much from exposure that it cannot be determined whether it was originally painted or coated with stucco. But if a conjecture may be hazarded, the clenched hand seems curiously suggestive of a pose common in those figures of Lokapâlas, or ` Guardian Kings of the regions ', which in the stucco image groups of the ` Thousand Buddhas ' of Tun-huang are so frequently associated with a Buddha or Bodhisattva.5
If there were any need of demonstrating the fact that the ruin L.B. II is that of a Buddhist shrine, it would suffice to refer in proof of it to the two finials in wood, L.B. H. 0033-34 (Plate XXXII), representing Stupas. About twenty and thirty inches in height respectively, they accurately reprodtice the characteristic structural features of Stûpas, and may appositely be compared with the small Stûpa model, N. v. xvi. ooi (see above, pp. 246 sq.), found at the Niya Site. They show a square base at the bottom and above it a circular drum, surmounted by a dome which carries a square member spreading out at the top, such as is regularly seen in well-preserved Gandhara Stiipas and also in the small stone-carved model, Yo. 00I2I (Plate vi). The whole is crowned by a series of umbrellas' like those which Indian Stûpas invariably once carried, whether in wood or stone. The mortice found in the base for attachment suggests that these little Stûpa representations served an architectural purpose as finials. But their exact position cannot be -proved, and there is the neatly-carved little finial, L.B. ir. 009 (Plate xxxv), still retaining traces of
presence of four massive foundation beams (ibid. p. 645). Pl. 76 illustrates the men's modus operandi in searching the remains. [For the use of Dr. Hedin's sketch-plan of The north-westernmost village of Loû-lan', ibid. PI. 75, it should be noted that by some inadvertence the north-south bearings in it have been reversed.]
See below, Figs. 207, 227, 228.