302 PROCEEDINGS: BOSTON SOCIETY NATURAL HISTORY. 
explained the breaches as marking the former northwest paths of 
transverse members of the Is&re system, from which they have been 
diverted by the subsequent growth of the main longitudinal valley — 
that northeast-southwest part known as the Gr6sivaudan— above 
Grenoble. His discussion of the problem takes, however, no account 
of modifications of valley depth by glacial erosion; and as this must 
have been considerable (for the valleys hereabouts have superb 
basal cliffs, as appears in the valley of the Romanche by Bourg 
d’Oisans in Plate 3, Fig. B , after a photograph by Neurdin Fr&res 
of Paris), it may well be that the rearrangement of river courses 
in this interesting region is not altogether the work of river action. 
Similarly, the various modifications of the Rhine system in eastern 
Switzerland, explained by Heim as the work of streams alone, may 
come to be at least in part referred to ice erosion. 
It may be further supposed that if the preglacial valleys were 
so arranged that a glacial distributary found a shorter and steeper 
course to the piedmont plain or to the sea than that followed by the 
master glacier, the distributary might under a long enduring glaci¬ 
ation become the main line of glacial discharge; and if so, it could 
be eroded to a greater depth than the former master valley at the 
point of divergence. In such a case, the postglacial river drainage 
would differ significantly from the preglacial. There is reason for 
believing that examples of this kind are to be found in Norway, the 
evidence of which will soon be published in an essay by Barrett 
( 1900 ). The diversion of the head of a stream in the Sierra Costa 
of northwestern California to a deeper-lying valley through a 
gorge cut by a glacial distributary has lately been described by 
Ilershey ( 1900 , 47). 
The Depth of Mature Glacial Channels. — The depth with 
respect to sea-level to which the channels of a glacial system may be 
eroded when the graded condition is reached, is a subject of special 
interest. For many miles along the lower course of a branchless 
trunk glacier, its volume is lessened by melting and evaporation, 
and at its end the ice volume is reduced to zero; slow ice motion 
being progressively replaced by rapid water motion. In such a case 
the law of continuity does not demand that the ice velocity shall be 
inversely proportional to the area of the cross section, as is the case 
in the normal river (where it is assumed that there is no loss by 
evaporation). Indeed, in the lower trunk of a mature glacier, it 
may well be that the velocity of ice movement is in a rough way 
directly proportional to cross-section area. This appears to be veri- 
