moulded and sand-ground ornamentation of the large fragment of a glass vessel, E. Fort. oo I z (Plate xxxvI), resembles Roman work of the first to fourth centuries A.D. and suggests importation. The abundance of glass at this site, as compared with the Niya ruins, is certainly a notable feature and one for which I am unable at present to suggest a satisfactory explanation. It is all the more noteworthy because we have seen above that the earlier settlement at Endere must belong to a period when the Niya Site was still occupied.
It is left for me to describe the structural remains which on two successive excursions I was able Ruined post
to survey in what may at present be considered the extreme south of the site. The most striking at south end
Y Y P g of site.
among them was the ruin of a fortified post, small but of massive construction, which Mihmân,
my Endere guide, knew by the name of Yalhuz-oi, the isolated structure '. It proved to be
situated just three miles to the south of the Stûpa, amidst closely set tamarisk-cones. These account
for the difficulty Prof. Huntington, its first Western visitor, experienced in finding the ruin.' It was
built in a solid square with walls of stamped clay, about eight feet thick at the base. A projecting
rectangular bastion guarded the single entrance from the south, as seen in the photograph, Fig. 82,
and the plan (Plate 2 t ). Owing to the protection against erosion which was offered by the situation of
the ruin and in particular by the immediate vicinity of the big tamarisk-cone seen on the left in the
photograph, the walls had suffered comparatively little damage and still rose to eighteen feet or
so in most places. On the outside of the walls the outlines left by forms which were used for stamp-
ing the pisé (now known as sighiz in Turki) could still be made out, with a length of about three
and a height of two feet. Just as in the walls of the ancient walled ` town ' near the Stûpa, single
courses of bricks or else shapeless clay lumps intervened horizontally between the stamped clay
layers. But though clearly visible from a distance in the higher portions of the walls these brick
courses were too effaced on the surface for exact measurement in places where access was possible.
The top of the north wall and of the bastion bore in parts scanty remains of a parapet about one foot
thick and apparently built with bricks of the same width.
The interior court, measuring about forty-eight feet square, was completely bare, except for Interior of
accumulations of straw and dung, mainly of horses, left undisturbed just under the lee side of fortified
the east wall. Along this wall the steps leading up to the top from the south-east were still post.
distinguishable. Within the entrance of the court three posts of Toghrak wood, once flanking
the inner gate, still stood upright to a height of about eight feet from the original floor. On clearing
the débris which filled the gate, we came upon some well-carved pieces of timber which the men from
Niya recognized as of mulberry and Elaeagnus wood. So cultivation at the time of this little fort's
construction was proved. But I searched in vain for definite archaeological evidence as to the
period of its occupation. The solidity of the whole structure gave an impression of antiquity which
I felt from the first rather instinctively, without being able to support it. But that this impression
had some basis in fact was revealed to me more than half a year later when I surveyed an ancient
fort of almost the same plan and appearance, T. xiv (Figs. 183, 184), along the Tun-huang Limes,
and ascertained that it marked the position of the Yü-mên Gate of Han times on the early Chinese
route from Tun-huang to Lop-nor.
The structural peculiarity above noticed, of brick courses introduced between the layers of Indications
of stamped clay, might be adduced to support that impression ; for it is seen also in the walls of the of post. date
ancient Sipil' to the north-east of the Stûpa. But I am inclined to attach even more importance to
the fact, noticed below, that some small and much-decayed dwellings found in the close vicinity of
this fortified post proved to have been built with bricks which showed exactly the same measurements
as those of the datable structures E. vi, vu, belonging to the earlier settlement. That the fair
' Cf. Pulse of Asia, p. 215.