DAVIS: GLACIAL EROSION. 
293 
and the branch channels may be discordant at their junction, and 
this discordance will increase with the difference in volume of trunk 
and branch stream. Truly, the discordance of stream beds is sel¬ 
dom noted, because the beds are hidden b} 7 the streams; but if a 
river system were laid dry, we may be assured that the beds of the 
smaller tributaries would open in the banks of the main river a 
number of feet from its bottom. In the case of the Mississippi, the 
discordance might easily measure fifty or more feet. 
All this applies equally to glacial streams. The surface of a trib¬ 
utary glacier is adjusted to the surface of the trunk glacier that it 
joins ; but the depth of the beds may be very different. As long 
as the glaciers occupy their channels, the discordance of their beds 
may not be often considered, but when a climatic change causes 
the glaciers to melt away, their channels are called “ valleys,” and 
the discordance of main and lateral glaciated “valley floors” is 
taken as an abnormal feature. In reality the discordance is per¬ 
fectly normal to the peculiar system of ice drainage by which it was 
produced, however discordant it may be to the system of water 
drainage now in possession of the valleys. Let us compare the 
maturely developed channels of rivers and glaciers. 
Channels of Mature Hirers and Glaciers. — A river flows rap¬ 
idly ; and the cross-section of its channel is but a small fraction of 
the cross-section of its valley. The river channel is U-shaped, very 
broad compared to its depth, while the valley sides flare open, 
Y-like, above the river banks. The water surface ' slopes steadily 
down-stream, but the channel bed has many small inequalities in 
the form of bars and basins, and the water in the bottom of the 
basins must ascend a little to get out of them. If the river should 
dry away, the deeper parts of the bed would be occupied by pools 
of standing water, while the bars would show lines suggestive of 
flowing water. The banks of the river channel are smoothly worn 
in nearly horizontal lines, parallel to the flow of the river current, 
while the sloping sides of a river valley are buttressed with spurs 
and scored by the down-hill ravines of descending streams. At the 
junction of trunk and branch streams, a moderate discordance in the 
level of the channel beds is to be expected; but this is seldom con¬ 
sidered, because the channels are usually occupied by water and the 
beds are hidden. 
A glacier moves slowty, and the cross-section of its channel may 
be a considerable part of the cross-section of the valley that it 
drains. Forel estimates that the glacier of the Rhone, even where 
