28 GEOLOGICALRESEARCHESIN 



S. E., we enter another valley opening in the S. W. Frequent fragments of a cal- 

 careous deposit strewed over the surface indicate the action of mineral springs. 



Gradually ascending this valley, which, as well as that of the Si Ho, is occupied by 

 a deposit of loam, probably contemporaneous with the terrace loam of the Yang Ho, 

 we reach a point where this loam deposit, by forming a bar across the valley, causes 

 a low watershed, on one side of which any drainage there may be flows north to the 

 Si Ho, and on the other south to the undrained lake Chaganoussu. 



We shall see that this remarkable occurrence of alluvial Avatersheds stretching 

 across valleys is intimately connected with the formation of the undrained lakes of 

 this portion of Mongolia, having its origin in a former system of great inland lakes, 

 and its continuance in the dryness of the climate. 



The grassy valley of Chaganoussu has two other openings through the plateau, 

 one on the east connecting it with the Si Ho valley, and another on the Avest leading 

 to the Kir Noor. Both of these are crossed by bars covered by the terrace loam, if 

 not entirely formed by it. Our road, after skirting the shallow pond of Chaganoussu 

 enters the vaUey leading to the southwest, and passing the dried up bed of the Ho- 

 yur Noor descends through a narrow defile till it emerges into the great depression 

 of the Kir Noor. 



From the Si Ho to this point the rocks, both of the adjoining plateau and of the 

 exposed parts of the valley bottom, belong throughout to the volcanic formation. 



From the edge of the plateau, near Avhere the road enters the Kir Noor valley, a 

 view of the whole of this ancient lake-bed is spread out beneath us. It is a large 

 plain about 15 miles broad, its longer axis trending about N. N. W. On both sides 

 the lofty and bold plateau edge is seen stretching away to N. N. W. and S. S. E., as 

 far as the eye can reach, without meeting to inclose the valley. 



Away to the southwest of us a distant x^ortion of the plain covered with a dazzling 

 white efflorescence marks the position of the Kir Noor of a few years since. From 

 this, the most depressed part of the plain, the surface rises toward every point 

 of the compass. Far away to the north a bar of the lake deposit seems to stretch 

 from wall to wall of the valley, Avhile in the south this is certainly the case. Over 

 this southern alluvial bar the peaks of the Barrier range are seen in the distance. 



To the N. N. W. a distant peak, capped with snow (April 18th), is visible rismg 

 above the level line of the table-land. 



The edge of the plateau on both sides of the valley, wherever I visited it, consists 

 of the volcanic formation, from the summit to under the lake deposits, but the pre- 

 sence on the surface of the latter of granite detritus indicates the presence of the 

 older rocks at no great distance. 



East of the Mongol village of Hoyurbaishin, a gully exposes a section of the 

 plain deposit near where this abuts against the edge of the plateau. The deposit 

 is stratified, and its beds have the same dip as the surface of the plain. It consists 

 of coarse sandstones and fine conglomerates, formed from the detritus of the neigh- 

 boring volcanic rocks and cemented by a calcareous mineral, the product, perhaps, 

 of springs, Avhich enveloping each grain or pebble Avith concentric layers produces a 

 hard rock. The only trees seen in the valley of the Kirnoor Avere tAvo old ones 



