88 W. UPHAM — THE SUCCESSION OF PLEISTOCENE FORMATIONS. 



approximately parallel retreatal moraines, marking stao;es of halt or 

 slight readvance, interrupting the general retreat of the ice-sheet, have 

 been mapped through Minnesota and North Dakota, and partially in 

 southwestern Manitoba. These are the northwestward extension of the 

 moraines mapped in Wisconsin, Illinois, Indiana, and Ohio by Cham- 

 berlin and Leverett ; and the outer two or more of their representatives 

 farther east in New York, northeastern Pennsylvania, northern New 

 Jersey, and the New England states were mapped ten to fifteen years 

 ago by Chamberlin, Lewis, Wright, Cook and Smock, and the present 

 writer. Within the last few years additional moraines farther north in 

 Massachusetts, New Hampshire, and Vermont have been traced by Tarr 

 and Hitchcock. No other problem of our glacial drift can be more in- 

 teresting than the correlation of these moraines across the continent, or 

 a,t least from the Atlantic coast to the northwestern plains bordering the 

 Cordilleran mountain belt, since thereby the successive boundaries of 

 the whole front of the ice-sheet during many stages of its recession are 

 becoming known. 



My attention, however, has been drawn more to the history of the 

 glacial retreat upon the central south to north belt of the Mississippi 

 and Nelson river basins, since this belt includes the area of the glacial 

 lake Agassiz, which was held in the basin of the Red river of the North 

 and of lake Winnipeg by the barrier of the departing ice-sheet. Through 

 my study of this ancient lake I have been led to careful consideration 

 of the retreat of the ice from its outer moraines northward past all the 

 numerous and partl}^ very complex marginal belts of knoUy and hilly 

 dnft accumulated during pauses of its recession contemporaneous with 

 lake Agassiz, until this vast body of water obtained avenues of north- 

 eastward outflow and finally was drained into Hudson bay, excepting 

 its shallow present descendants, lakes Winnipeg, Manitoba, and others, 

 left in the depressions of its bed. This portion of the departure of the 

 ice I find to have been surprisingly rapid. Comparison of the shore 

 erosion and accumulation of beach gravel and sand by the waves of lake 

 Agassiz with those of lake Michigan indicates for the relative durations 

 of these lakes a ratio as 1 : 10 or 1 : 20. Lake Michigan has existed and 

 worn away its shores and built up its beaches and dunes during the 

 postglacial epoch, which appears, from various independent but approx- 

 imately accordant estimates by N. H. Wlnchell, Gilbert, Andrews, and 

 others, to have been some 6,000 to 10,000 years. The existence of lake 

 Agassiz, therefore, with the accumulation of its several associated large 

 moraines, occupied no more than 1,000 years. 



Is this conclusion consistent with all the otherwise known records and 

 inferred conditions of the Ice age upon this central belt of our conti- 



