HISTORY OF THE COASTAL GRAVELS. 405 



tical differences between the southern terminations of the gravels left by so 

 many glacial rivers, having so many topographical relations and scattered 

 over so wide an area, we are driven to seek for agencies capable of acting 

 horizontally over the whole coast simultaneously. In view of the great 

 differences between the glacial rivers, also of the lobal ice front during the 

 retreat, it seems improbable that there was any agency capable of thus 

 widely acting so nearly parallel to the surface of the sea except the sea 

 itself 



What conditions of the ice-sheet independent of the sea would tend to 

 produce such a development as is shown by the coastal glacial gravels! 



Little or no pei'manent accumulation of sediments can be made within 

 small subglacial or englacial tunnels — small, that is, compared to the flow 

 of waters — because of the great velocity of the streams during the summer 

 floods. And such they remain while the ice is deep and the rate of flow 

 rapid. Before the streams have time greatly to enlarge their channels new 

 ice advances from the area of accumulation and the ice containing the tun- 

 nels already somewhat enlai'ged has reached the front, where it is melted or 

 discharged as bergs. Under these conditions the rapid subglacial streams 

 transport almost all their sediments to near the front of the ice and deposit 

 them as marginal kames or beyond the ice as overwash aprons. In this 

 they are often assisted by the pushing of the ice. On rather steep down 

 slopes, especially where the waters are collected in the lower parts of val- 

 leys, these conditions prevail throughout the whole time of the retreat, 

 until the glacier becomes too small to support large streams. Thus all the 

 larger glaciated valleys of the Rocky Mountains contain retreatal plains 

 of frontal gravels up to about 5 miles from their sources. The gravels of 

 the Androscoggin River from Bethel to Grorham, New Hampshire, and 

 also those of the upper Carrabassett Valley are probably in part of this 

 character. 



But when an ice-sheet covers a variety of "surface, such as plains and 

 gentle down slopes, and especially adverse slopes, or when there is a sub- 

 sidence greater toward the source than at the distal extremity, the glacial 

 streams become able greatly to enlarg-e their channels as the ice grows 

 thinner and by degrees stagnant, and are no longer able to keep them free 

 from sediment. The process of sedimentation begins at favorable places in 

 the channels, such as local enlargements, or where obstructions rise on the 



