It is supposed by those who have made this branch of geology a study that prior to the glacial period a river flowed down through Lake Superior, which connected with Lake Michigan at a point near its present outlet at Sault Ste. Marie, the channel of the river passing down through what is now the bottom of Lake Michigan, which had an outlet at the head of the lake near Chicago and flowed off into the Mississippi River. All of the lake bottoms of this great chain, with the exception of Lake Erie, are now below sea-level. The reason for this exception will appear further on. Before the ice age there was supposed to be no connection between Lake Michigan and Lake Huron, as there is now, through the Straits of Mackinac.
Another preglacial river had its rise in the region of Lake Huron and flowed through an old river bed extending from the Georgian Bay in a southeasterly direction through the province of Ontario, and emptied into the present Lake Ontario. From Lake Ontario there is an old river bed running through the Valley of the Mohawk which empties into the Hudson at Troy. Neither of these two rivers, having their sources in the north, found an outlet through the present St. Lawrence River. During the time of the glacial period there is evidence that there was more than one center of snow and ice accumulation and each of these great centers probably had several subcenters. This theory has color given to it by the directions of movement shown by the glacial drift.
The rounded appearance of bowlders was caused by the grinding action of the ice. These bowlders, when they were first torn from their rocky beds by the irresistible power of ice pressure, were rough and jagged in shape, the same as any rock would be, torn from a quarry by a blast. They have been smoothed and rounded by rubbing against the moving ice and against each other in the progress of their long journey from their original homes. Where their home was the geologist can immediately tell upon examination. It is only necessary then to examine the bowlders of any particular locality to determine the direction of the ice flow at that point.
There seem to have existed centers of ice accumulation to the north of all of the great lakes. And when they had grown to a sufficient height they joined at their edges, making one grand glacier, the movements of which were the resultant of the combined pressure exerted by these great centers of power, so that all of North America north of the line of the terminal moraine, with the exception of a small area (heretofore noted) chiefly in Wisconsin, became covered with one vast sheet of ice.