182 PLEISTOCENE OF INDIANA AND MICHIGAN. 



by northward rather than southward drainage. This terrace has no continuation beyond the 

 Kalamazoo morainic system, the country west of the latter being much below its level. In 

 explanation of this high terrace it seems necessary to assume a barrier of some sort below its 

 position. Possibly the morainic accumulations in the valley were sufficient to hold the river 

 up to this level during the development of the terrace. It seems quite as probable, however, 

 that the ice sheet did not entirely melt away until the terrace had been formed. Opposite this 

 terrace the slopes are very irregular, as if ice might have persisted during their development and 

 have thus prevented their smoothing over by the stream. Some evidence of such ice 

 persistence is cited below in the discussion of the Kendall moraine (p. 183). 



INNER BORDER. 

 INTERPRETATION. 



The narrow strip between the inner or western morainic ridge of the Kalamazoo system 

 and the Valparaiso morainic system is by no means easy to interpret. The portion from Paw 

 Paw northeastward to the Kalamazoo Valley is especially complex, and it was here that the 

 writer was led astray (see p. 174) in the interpretation presented in Monograph XXXVIII. 

 The outer part of the Valparaiso system from Paw Paw southwcstward to the State line and 

 beyond is marked by a strong ridge with a regular border, outside of which is a nearly parallel 

 outwash apron that fills much of the space between it and the inner moraine of the Kalamazoo 

 system, from which it is clearly separable. 



In the interval from Paw Paw northward to the Kalamazoo Valley in at least two places 

 the Kalamazoo and Valparaiso systems seem to be connected by weak cross ridges, which 

 suggest that the ice border persisted here in its occupancy of the inner face of the Kalamazoo 

 morainic system, while it withdrew several miles from it hi the district southwest from Paw 

 Paw and began the development of the strong outer ridge of the Valparaiso system. Aside 

 from these two weak cross ridges there is, in the northeastern township of Van Buren County, 

 a conspicuous drift aggregation which is not easy to correlate with neighboring moraines. 

 There are also conspicuous swampy channels which appear to mark courses of glacial drainage. 

 The intervening tracts not embraced in ridges or in channels are largely sandy and plane 

 surfaced, the only till plain noted being in the southern edge of Allegan County. 



An inconspicuous, gently undulating, bowldery strip parts from the Kalamazoo system 

 about 8 miles west of Kalamazoo and passes south westward to South Fork of Paw Paw Kiver 

 at Paw Paw, nearly opposite the terminus of a strong ridge of the Valparaiso system. Its swells 

 are only 10 to 15 feet high and have very gentle slopes, and its weakness is strikingly in contrast 

 both with the strong moraine of the Kalamazoo system from which it separates and with the 

 bulky moraine of the Valparaiso system with which it may connect at Paw Paw. Bowlders 

 are less numerous on it than on the strong moraines just mentioned. Its soil is sandy and no 

 clayey drift is reported in wells having depths of 40 or 50 feet. On the south side of this undu- 

 lating strip is a sandy plain, which may perhaps be an outwash apron for it slopes gently south- 

 ward to the west-flowing portion of the South Fork of Paw Paw River. 



Two correlations are possible at the west end of the undulatory strip, one being found 

 in a chain of knolls and short ridges that leads southwestward past Decatur into the north- 

 western part of Cass County, and the other in a strong morainic ridge of the Valparaiso 

 system that sets in on the west side of South Fork of Paw Paw River just north of Paw Paw. 

 The most prominent knoll hi the first-named chain is about a mile south of Paw Paw and has a 

 height of nearly 150 feet above the village and about 100 feet above the bordering plains. Else- 

 where the knolls are only 10 to 20 feet high and are separated by plane tracts of considerable 

 width. Most of the knolls carry bowlders, but the bordering plains are nearly free from them. 

 It is a question whether this weak and fragmentary chain constitutes an ice-border feature. 

 The knolls may have been formed incidentally in the retreat of the ice from the Kalamazoo to 

 the Valparaiso system and not be definitely related to each other nor to the weak ridge that 

 leads into Paw Paw from the inner ridge of the Kalamazoo system. 



