436 GLACIAL GEAYELS OF MAINE. 



to sink to lower level. The suddenness with which the development of the 

 gravels was often arrested and the absence of transition beds laid down in 

 transverse channels or of terraces between the end of the ice and the hills 

 that rose in front of it, may perhaps be best interpreted as proving that the 

 ice sometimes became so greatly shattered near the front that the waters 

 spread ontward and often transversely in a multitude of small delta branches, 

 none of which were large enough to deposit gravels in the short time that 

 elapsed before the ice was all melted in that region. The nature of the 

 development of the glacial sediments during the retreat of the ice down the 

 northern slope and thence back to C (fig. 32, p. 433) would depend on many 

 conditions, and we might expect many difFerent manifestations. One of the 

 critical points, so to speak, is at the northern end of the permanent water 

 trap at C. At the time of diminished flow in the fall and winter the stream 

 would no longer fill its tunnel and more or less sediment would drop where 

 it entered the permanent water trap ABB. Now and then this might result 

 in the channel being clogged during the floods of the succeeding summer, 

 forcing the waters to rise, and causing the formation of an englacial or super- 

 ficial channel and the opening of a lake at the place where the stream rose 

 on the surface or flowed down again after passing the obstruction in the tun- 

 nel. In such a lake broad ridges or an osar-plain might form, or reticulated 

 ridges, but not a delta, unless it was very large compared to the river. Or 

 the broad channel or lake might be extended continuously across the valley, 

 perhaps by the confluence of a number of lakes that originally formed in 

 the course of the channel. When the waters forced a transverse passage 

 north of the hill early enough to permit considerable enlargement and depo- 

 sition of gravels, we have the phenomenon of delta or diverging and trans- 

 verse branches like the intricate reticulations of tlie gravel systems of 

 southwestern Maine. Here the rocks are mostly granitic and the till is very 

 abundant. This must have favored the clogging of the subglacial tunnels 

 and the digression of the streams to new channels that often diverged widely 

 from the original channels. 



We next consider what is conceived to be the probable behavior of a 

 superglacial stream in the same situation, i. e., during the retreat of the ice 

 over a valley situated north of a hill crossed by an osar system, premising 

 that it must be able to deposit stratified osars, and hence that the bottom 

 of its canyon must reach the ground or nearly to it. At first the bed of the 



