220 PLEISTOCENE OF INDIANA AND MICHIGAN. 



to be barely high enough to have been connected with the drainage to the Kankakee, for the 

 swamp which leads southward from Paw Paw to Dowagiac is only about 750 feet in its highest 

 part and has been filled to some extent by peaty accumulations. 



If, as seems probable, the ice at the time the Kendall moraine was forming extended to the 

 edge of the Kalamazoo system from Kalamazoo River northward, its drainage in the district 

 north of the Kalamazoo is likely to have been in the same direction as it was during the devel- 

 opment of the Kalamazoo system or southward past the site of the city of Kalamazoo to St. 

 Joseph River. 



As the ice shrank back from the inner border of the Kalamazoo system in the district north 

 of Kalamazoo River it formed an extensive gravel plain thickly set with basins. This plain 

 slopes eastward in the region west of Gun Lake and Gun River, but farther north in the reentrant 

 angle between the moraines of the Saginaw and Lake Michigan lobes it slopes southward. 

 Gravel appears at two distinct levels, a part of the outwash apron having been cut away to a 

 depth of 30 or 40 feet by glacial drainage and a part remaining at the original level of filling. 

 This lower plain has one arm extending northwestward from Gun Lake to a strong ridge of the 

 Valparaiso system near Wayland and another arm extending northeastward to the moraine of 

 the Saginaw lobe at Middleville. From the junction of these two arms the drainage is south- 

 ward along the Gun River marsh to the Kalamazoo and thence along the marsh which con- 

 nected the Kalamazoo with the Paw Paw along the eastern side of the Kendall moraine. The 

 features suggest that at the time this lower plain was being developed the ice had withdrawn 

 sufficiently from the lower course of Paw Paw River to permit drainage down its valley to the 

 vicinity of Benton Harbor. At the time the higher or main outwash apron was being developed 

 the drainage seems likely to have been southward to South Bend along the lines followed by 

 the portions of the glacial drainage south of the Kalamazoo. 



When the reentrant between the Lake Michigan and Saginaw lobes stood just north of 

 Grand Rapids an outwash apron was developed in front of the Saginaw lobe on the east side 

 of Grand River at an altitude of about 700 feet. An outwash apron along the front of the 

 Lake Michigan lobe, south of Grand Rapids, in Wyoming and Byron townships, Kent County, 

 has a similar altitude, being just below 700 feet, as shown by the Grand Rapids topographic 

 sheet. The water evidently found escape southward between the two lobes past Grand 

 Rapids, Carlisle, and Ross to Rabbit River near Dorr. The course beyond Dorr is less easy 

 to interpret. Two courses are possible, and their relation to the drainage southward from 

 Grand Rapids depends on the amount of recession that had been made by the Lake Michigan 

 lobe at that time. If the ice still covered the strong moraine north of the city of Allegan 

 and also the moraine southwest of that city the drainage is likely to have followed the line of 

 a swampy depression utilized by the Lake Shore & Michigan Southern Railway from Dorr 

 to Allegan, and then to have passed up the Kalamazoo Valley from Allegan to Otsego and 

 turned southward into the Paw Paw drainage ; to have taken this course, however, the waters 

 must have passed over divides that rise a few feet above the 700-foot contour, or higher than 

 the source of the glacial drainage near Grand Rapids. If, on the other hand, the ice had with- 

 drawn from the prominent morainic tracts near Allegan the glacial drainage may have passed 

 southwestward between the ice border and these moraines into a local lake in the lower part of 

 the Kalamazoo Valley. The drainage from Dorr into this lake came eventually if not at the 

 beginning of the drainage southward from Grand Rapids. The indefiniteness of the correla- 

 tions in different parts of the Valparaiso morainic system renders questions of this sort per- 

 plexing, and the full solution must depend upon the fuller correlations along the lines of the 

 several members of the system. 



The drainage from the reentrant between the Lake Michigan and Saginaw lobes in south- 

 eastern Newaygo, southwestern Mecosta, and western Montcalm counties is likely to have 

 passed southward through the valley of Rapid River to its junction with the Grand River near 

 Grand Rapids, and possibly it continued southward to Dorr along the line just discussed into 

 Rabbit River. An outwash apron, developed on the east border of the Lake Michigan lobe in 



