298 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
(Wallace, 96 , 176) and in the lower Gasternthal near its junction 
with the Kandertlial; in both these cases the basins have been 
aggraded and the sills have been trenched by the postglacial streams. 
In the lower Gasternthal the height and steepness of the rocky 
sill, when approached from up-stream, is astonishing; its contrast to 
the basin that it encloses is difficult enough to explain even for those 
who are willing to accept strong glacial erosion. It should, how¬ 
ever, be noted that river channels also are deeper in the weaker 
rocks up-stream from a hard rock sill; if the river volume should 
greatly decrease, a small lake would remain above the sill, drained 
by a slender stream cutting a gorge through the sill. 
If an initial depression occurred on the path of the glacier, so 
deep that the motion of the ice through it was much retarded, an 
ice-lake would gather in it. Then the waste dragged into the basin 
from up-stream might accumulate upon its floor until the depth of 
the basin was sufficiently decreased and the velocity of the ice 
through it sufficiently increased to bring about a balance between 
ability to do work and work to be done. Here the maturely graded 
condition of the ice stream would have been attained by aggrading 
its bed, instead of degrading it; this being again closely analogous 
to the case of a river, which aggrades initial depressions and degrades 
initial elevations in producing its maturely graded course. 
Water streams subdivide toward the headwaters into a great 
number of very fine rills, each of which may retrogressively cut its 
own ravine in a steep surface, not cloaked by waste. But the 
branches of a glacial drainage system are much more clumsy, and 
the channels that they cut back into the upland or mountain mass 
are round-headed or amphitheatre-like; but the beds of the branch¬ 
ing glaciers cannot be cut as deep as the bed of the large glacial 
channel into which they flow: thus corries, perched on the side- 
walls of large valleys, may be produced in increasing number and 
strength as glacial maturity approaches, and in decreasing strength 
and number as maturity passes into old age. As maturity ap¬ 
proaches, the glacial system will include not only those branches 
that are consequent upon the initial form, but certain others which 
have come into existence by the headward erosion of their nev6 
reservoirs following the guidance of weak structures; thus a ma¬ 
turely developed glacial drainage system may have its subsequent 
as well as its consequent branches. It is entirely conceivable, as has 
been suggested by Meunier, that one ice stream may capture the 
upper part of another. The conditions most favorable for such a 
