318 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
by the greater thickness of the accumulated ice.The result of 
all this is the formation of a rock-basin, the deeper portion of which 
lies towards the upper end, just where the grinding force of the 
glacier is greatest” (’ 95 , 228, 229). 
It seems to me that too great emphasis is here placed on the ero¬ 
sion accomplished near the end of a glacier, as indicated by lakes, 
and not enough upon the deepening of the valleys up-stream from 
terminal lakes, as indicated by hanging valleys. It is also to be 
noted that de Lapparent, Richter, and J. Geikie all describe the 
hanging valleys of Norway as if their preglacial form had not been 
significantly changed, thus failing to bring clearly forward the fact 
that the valleys of today are the ice channels of the past, and that 
the larger and smaller channels must have normally discordant floors 
in a system of glacial drainage, just as they have in a system of river 
drainage, although to a much greater degree. The full analogy 
between ice and water channels, which throws so much light on the 
whole question of glacial erosion, was first clearly set forth by the 
two following observers. 
Gannett on Lake Chelan , 1898. — The most complete statement 
of the general principles involved in the production of hanging 
valleys that I have found in print is in an article on Lake Chelan, in 
the Cascade Mountains of Washington, by Henry Gannett. Chelan 
is a long narrow lake occupying the distal two thirds of the deep U- 
shaped valley of the Steliekin River on the eastern slope of the 
mountains. It was occupied in the glacial period by a heavy ice 
stream, fifty or sixty miles long, and half a mile to a mile broad. 
The rock walls which enclose the valley are strikingly parallel to 
one another, without buttressing spurs; they rise 4000 to 5000 feet 
above the lake waters. Nearlv all the streams which flow into the 
valley tumble over its walls in a series of cascades. “From all 
indications it appears that the ice must have been at least 3000 feet 
deep in this gorge of the Steliekin, since several of the smaller 
branches join the main glacier at that height above its bed.” 
Speaking of these features in a more general way, Gannett says: 
“ A glacier is a river of ice, and it behaves almost precisely as a 
river of water does. Its effects upon its channel are almost pre¬ 
cisely similar to those of a river upon its channel, excepting in the 
fact that all its operations are on a vastly greater scale.A word 
of caution must here be interpolated. The channel of a river, in 
which its water Hows, must not be confused with its valley, which 
it drains. The above comparison refers to the channel of a river 
