268 The Land of Deep Corrosions 
marks the junction between the rain-belt and the arid region 
in this valley, I watched a curious spectacle. At seven 
o'clock in the morning there was blue sky overhead extending 
northwards as far as the eye could reach, but the clouds: 
were gathering in the south and away down the valley 
it was snowing heavily on the mountains, their summits 
being completely concealed. A furious wind was raging up 
the valley at the time. Very slowly the clouds sailed up to 
the assault, and an hour later there were broken masses of 
cloud just south of us. By ten o'clock there were puffs 
of cumulus almost overhead; but struggle as they would 
they were unable to cross what seemed a physical barrier 
forcibly keeping them back. To the north the sky was 
blue and cloudless, and it was mid-day when a few scattered 
rain-drops fell at Yang-tsa. In the afternoon I walked some 
miles down the river, finding it heavily clouded over and 
a continuous drizzle falling, but the sky was still blue in the 
north though a few puffs of cloud had at last succeeded 
in crossing the dividing line, and were rapidly dwindling 
even as they triumphed. By mid-afternoon more clouds 
had forced the barrier and were concentrating themselves 
over the mountains on either side of the valley, but blue 
sky still maintained the position intact above the river. 
This phenomenon served as an excellent illustration of 
the part played by these parallel ridges in determining 
the rainfall in the valleys to north and south of the rain- 
screen. 
I have already alluded to the great altitude of the snow- 
line east of the Mekong, and also to the very considerable 
elevation of the peaks and passes on that watershed. There 
can be no doubt that, did the Mekong-Yang-tze ridge at 
present receive as heavy a rainfall as does the Mekong- 
Salween ridge, there would be snow-fields and glaciers on 
the former where none now exist. This is an important 
consideration, because after many climbs amongst these 
mountains I became convinced that there had actually been 
glaciers in many of the valleys. The valleys which descend 
immediately from the watershed are in the first place all 
hanging valleys, and at their upper extremities is always 
to be found in place of a gradual ascent growing steeper 
and steeper towards the cirque, a peculiar ‘tread and riser’ 
