Sec. iv] FROM RÖSHAN TO DARWAZ 885
glaciers. Some three miles below this junction the bottom of the valley becomes very confined and the track, in places only a foot wide, along the steep slopes of detritus present great difficulty. But after passing two small patches of cultivation we suddenly emerged into the Bartang valley amidst the small terraced meadows and fields of Khaizhéz hamlet (about 6,800 feet elevation), where we halted.
The two days' journey which brought me from Khaizhéz down to Kala-i-Wamar on the Oxus covered but a small portion of that main valley of Rôshan which I had first sighted high up at Saunab. But it sufficed to impress me with the exceptional difficulties of traffic offered by the tortuous gorges in which the Bartang river has cut its way down to the Oxus. It also showed why Riishan has always been the least accessible of all the valleys descending from the Pamirs, and why its people and ways have shown themselves most retentive of their old inheritance. The line of progress through these narrow deep-cut gorges (Figs. 421, 422, 424), between towering mountain masses wildly serrated above and very steep at their foot, was more troublesome even than the glacier pass by which we had approached them. Having crossed from Khaizhéz to the right bank of the river on a raft of goatskins we were immediately faced by a steep climb to the Khaipast spur. With the river racing past it in cataracts it serves as a formidable ` Darband' or chiusa, and remains of stone breastworks showed that it was defended on occasion.
A couple of miles beyond there followed a succession of trying climbs up and down precipitous rock faces (Fig. 421), where the track leads along narrow ledges or is represented only by footholds a few inches wide. Fortunately it was possible for a few of us to avoid some of the worst of these awrinz by the use of small goatskin rafts (Fig. 422), where the absence of dangerous cataracts allowed of their employment. Guided by dexterous swimmers, they let us glide down the tossing river, forgetful of all fatigue, in scenery of impressive grandeur. Boldly serrated snowy peaks showed again and again above the high frowning rock walls, which, as they rapidly passed in succession, ever seemed to close in upon us. Meanwhile the baggage was being carried in safety by sure-footed R6shanis past precipices as bad in places as any I remembered in Hunza.
The uniform grimness of this mountain scenery was relieved by the pleasant contrast offered by the hamlets nestling here and there at the mouth of ravines and half hidden amidst fine walnut and other fruit trees. The dwellings I was able to examine at Rit and again at Paghû (Fig. 423), where we broke our journey, seen from outside were indeed unpretending rubble-built hovels. But the interior, smoke-begrimed as it was, invariably showed arrangements indicative of rude comfort and interesting as obviously derived from antiquity. For there was the closest resemblance in essential features, such as the ground-plan of the living hall with its skylight ceiling and sitting platforms, to the internal architecture with which I had been made familiar by residences on ancient sites explored in the Taklamakàn and by others still occupied by the living in Hindukush valleys to the south.' Alpine seclusion seemed to have kept this small corner of the world almost untouched by the change of ages, and I felt inclined to wonder whether it would have presented a very different picture to some Bactrian Greek or Indo-Scythian visitor of old.
The same impression was conveyed by the physical character of the men whom I met on my way down the valley (Fig. 443) or was subsequently able to examine anthropometrically at Kala-iWamar (Fig. 444). Clean of limb and made wiry by constant movement over such impossible tracks, they all showed clear-cut features, generally light-coloured eyes and often faces of almost classical regularity. The conclusion suggested itself to me at the time that among the Iranian-speaking hillmen of the valleys that I traversed in this region it was the population of Rôshan that has preserved the Homo Alpinus type in its greatest purity, and Mr. Joyce's expert analysis,
1 For the plan and description of such a Rôshâri dwelling, see below, ii. p. 887.