CHINA, MONGOLIA, AND JAPAN. 31 



hills, occurs in the terrace deposits on their N. W. flank. From this hill we descend 

 into a small valley Avhich empties into that of the Te Hai. In this valley the terrace 

 loam is present to the height of probably not less than 250 feet above the lake. 



From here the road descends to the deep channel cut through the plateau, which 

 connects the great valley of the Te Hai with that of the Sankang Ho. This channel 

 is cut to the bottom of the volcanic mantle, here apparently over 1,000 feet thick, 

 and into the metamorphic rocks on which it lies. 



In this channel we meet with another of those remarkable watersheds of terrace 

 deposit which stretching from wall to wall, slopes on the west toward the Te Hai, 

 and on the east toward the vaUey of the Sankang Ho. The material forming this 

 bar is almost loose sand mixed with fragments from the volcanic and metamorphic 

 rocks, and is but little, if at all, eroded on the western flank, while there are guUies 

 on the eastern in which highly inclined beds of granidite, containing garnets, are 

 exposed. 



At Maanmiau the valley opens to form the broad, swampy plain of Fungching, 

 rising from which are frequent low hillocks of gneiss in strata trending between E. 

 and N. E. Here the high plateau leaves the road; the part that has formed the 

 southern side of the valley since leaving the Te Hai, now trends away to the S. S. W. 

 till the steep face and level outline of its edge are lost in the far distance. On the 

 other side, the part which has formed the northern wall of the valley, continues a 

 few miles farther, and then, before reaching Fungching, bears away to E. N. E. 



Although we have here left the higher plateau, we have not yet reached the south- 

 ern limit of the volcanic formation. At a level of perhaps 1,000 feet below the 

 surface of the higher plateau begins the lower plateau, the flat surface of which is 

 200 or 300 feet above the valley, and extends southward from the very edge of the 

 higher. It consists of the same volcanic formation as the higher table-land of which 

 it was, I think, without doubt, once the continuation, the continuity having been 

 broken by an immense fault — a supposition to which I shall recur further on. 

 . The marshy plain of Fungching is fringed in places with low, flat hills, which owe 

 their form to the terrace deposit of loam, but under this, consist of a bright red, 

 sometimes loose material, apparently a wacke or a product of the decomposition of 

 the volcanic rocks. In this are fragments of a red calcareous mineral, a product 

 of the action of waters on the adjoining rock before or during its alteration. We 

 shaU see a similar mineral fiUing crevices in the volcanic plateau formation. It 

 is perhaps the result of the metamorphic action of mineral springs rising along the 

 great faidt-Hne. 



A few miles beyond Fungching our road rises to the surface of the lower plateau, 

 and we obtain an open vieAv from a ruined part of the Great "Wall. To the north 

 we can see the precipitous edge of the higher table-land stretching far away to the 

 northeast, the break in it formed by the valley of the Kir Noor, and its continuation 

 beyond this toward the Si Ho.^ To the south and east we see the barren crest and 

 peaks of the Barrier range. Between the higher table-land and this sierra is the 

 lower plateau on the southernmost spur of which we are standing. The valley we 



* In Mongol, Djookha Gol. 



