OEIGIN OF GLACIAL GRAYBL COMPLEX. 457 



overflow channels wlifere the snbglacial streams found their tunnels closed 

 and were forced for a time to rise into englacial or superglacial channels. 



3. The heaviest burden that the superglacial hypothesis has to bear is 

 the basal character of almost all the kame and osar drift. By far the 

 greater part of the glacial gravels is stratified and shows no sign of having 

 been deposited on ice, where it would have to fall as the ice melted. It 

 has the appearance of having been deposited on the ground where we now 

 find it, and this is the natural place for snbglacial deposits. It remains to 

 be proved that superglacial streams can cut canyons or enlarge lakes so 

 deep that they penetrate to the bottoms of the ice, except near the ice 

 margin, where crevasses are strongest. This reasoning, however, applies 

 only to deposits within channels in the ice. Beyond the ice front the delta 

 ridges of both kinds of streams would manifestly rest on the till or rock 

 and can be accounted for by either hypothesis. 



4. Some of the marine deltas are very broad at their northern ends, as 

 in the northern part of Alua, where one ends in a broad transverse bar or 

 ridge showing no horizontal assortment of sediments from the center toward 

 the ends. In other places there are two or more short ridges projecting 

 here and there toward the north, as if several streams, not one, had con- 

 tributed to the formation of the delta. But these deltas are a part of an 

 osar system, and except in these places we have no signs of more than a 

 single glacial river. All this is easily accounted for on the snbglacial 

 hypothesis, since those streams can force new channels when their old ones 

 are blocked, or they can rise into englacial or superglacial channels. But 

 how can a single superglacial stream, after cutting a channel down or nearly 

 down to the bottom of the ice, wander into other channels parallel to the old 

 one, all of them also cutting to the bottom of the ice, which was consider- 

 ably below sea level and only a short distance back from the ice front? 

 The supposed superficial streams would sometimes have to cut 100 or more 

 feet beneath sea level, and yet abandon these channels for others, unless 

 we suppose that there were more than one superficial stream tributary to 

 the delta; but elsewhere in the course of the gravel system we have proof 

 of only a single river. The broad osar channels, including the retreatal 

 channels and lakes north of hills, the massive plains or mounds deposited 

 in glacial lakes, also the osar border clay, all point to a rapid enlarg'ement 

 of a glacial-stream channel or a pool when once they became open to the 



