DESERTS. 269 
or 3 miles below Damburachi the river flows through a narrow cut, skirting the 
north side of the valley, while the lower terrace rises as a plateau on the south 
with an old channel of the Kizil Su on it. Below there it converges with the present 
flood-plain and just below Pildona, 6 miles above Hawee, it sinks under. From 
there on nearly to Garm, it lies buried in the present flood-plain and the river at 
high water swings against the valley sides it scoured during the third erosion cycle. 
Just above Garm the lower terrace rises out again and the river enters a channel 
that deepens steadily to Obu-garm, the westernmost point of observation, where 
this terrace is about 300 feet high as cut into by the present channel, that of 
our fourth erosion cycle. 
In defining the third stage we inevitably defined the fourth, the channel of 
to-day. But some important facts should yet be noted. The present channel 
is cut in alluvial gravels all the way, excepting for occasional glimpses of bed-rock 
bottom. Obviously, then, the stream had cut down as far as it is now during the 
third erosion cycle, which, 
however, closed with a re- 
filling and valley-widening 
to the third stage. The 
earth-movements which 
wrought its fourth and last 
erosion cycle were of sucha 
nature that about 35 miles 
of the valley, that portion 
between Garmand Pildona, 
was negatively affected, 
that is, aggraded instead of 
corraded. This would 
appear to indicate warping, 
and that idea is reinforced 
by observations on tribu- 
taries. Two large tributa- 
ries converge to Hawee, 
where they debouch in the Fig. 446.—A Galcha of Karategin. 
Kizil Su. One comes in 
from the west, the other from the north. If the great valley suffered a longi- 
tudinal warping, we should expect to find the one from the west rejuvenated, the 
one from the north aggrading in sympathy with the main river. This is precisely 
the condition at Hawee. One more important fact remains: the Kizil Su has 
not yet graded its fourth-cycle channel, excepting over the 35-mile stretch of 
relative depression. The present flood-plain is divided into several long stretches 
where the river splits into a braided stream. Between these the channel narrows 
into short shoots, where it plunges over sills of bed-rock. The fourth erosion 
cycle is, therefore, the result of a warped uplift still in process, or so recently 
completed that corrasion has not yet attained an even grade. 

