Upham.] 
230 
[Feb. 18, 
Each of these four classes of lakes presents interesting ques- 
tions as to the causes and conditions by which their basins were 
formed 1 ; but we shall only investigate these questions for the 
lakes enclosed by modified drift, with which the others are here 
brought into comparison for the purpose of more clearly discrim 
nating and defining the characters of the class under considera- 
tion. Deep lakes of small area, as Walden and Cochituate, bor- 
dered by steep banks and adjacent plains of water-deposited 
gravel and sand, present first the difficult question how these 
beds could have been prevented from being spread where the lake 
basin now exists, as well as on the adjoining areas. What kept 
the gravel and sand from being carried into the hollow and filling 
it? We shall answer, from study of the final melting and de- 
parture of the ice, that large island-like ice-masses, remaining for 
some time after the recession of the main glacial boundary, occu- 
pied the places of these lakes and became surrounded by the beds 
of modified drift, so that when the ice-masses disappeared they 
left bowl-shaped or irregular basins filled with water. 
Next comes the inquiry, How could the gravel and sand be 
supplied in such abundance and be deposited so fast as to form 
the enclosing plains or terraces before the ice-mass occupying the 
lake basin was melted ? The reply must be that the lower por- 
tion of the ice-sheet contained plentiful drift upon many areas, 
and that the glacial melting was so rapid as to set free a great 
amount of this drift and to produce, with the aid of accompany- 
ing rains, broad drainage systems of rills, brooks, and rivers, 
by which gravel, sand, and fine silt were gathered from the drift 
exposed on the ice and swept forward to their place of deposition 
beyond the ice-border. 
Some writers on this subject hold that much of the drainage 
bringing these beds of modified drift was subglacial, flowing be- 
neath the ice and there forming in tunnels the long gravel and 
sand ridges called eskers or osars ; and certain features in the 
stratification of the sand plains formed along the ice-front are 
ascribed to the upflowing currents of subglacial streams. My 
studies, however, lead me to think that streams under the ice- 
sheet were exceptional and rare during its rapid final melting; 
1 See “On the Classification of Lake Basins,” by William Morris Davis, Proceedings 
of this Society, vol, xxi, 1882, pp. 315-381. 
