14 PROCEEDINGS OP THE BOSTON MEETING 



a narrow gorge over 100 feet deep. Just to the south of it runs another gully not 

 very much shallower while on the trap range than the Douglass-Houghton, though 

 it has less water and smaller drainage area. Below the falls it conies quite close 

 to the Douglass-Houghton ravine, as the figure shows, and at a point, A, on the 

 side of the deep ravine corresponding to the bottom of the shallower, we find 

 springs where the two enter the porous sandstone. In other words, the bottom of 

 the upper valley becomes leaky and has let out the water into the lower. The 

 shallow ravine swings away again from the Douglass-Houghton and grows deeper, 

 but does not carry water in ordinary weather, and erosion in it must be very slow. 

 The point where the springs occur is liable to landslips. Several have occurred 

 since I first saw^ the place, nearly ten years ago. The softening action of the water 

 renders them especially frequent, and in this particular place the work of geologists 

 anxious to study the contact may have accelerated matters, so that I should not be 

 surprised to see the day when the southern gully. will appear a direct tributary of 

 the main ravine. 



Though the action above described might more properly be designated valley 

 capture than river capture, I have no quarrel with nomenclature, but wish to call 

 attention to the probability that in many cases there is an intermediate stage in 

 which a valley is undrained and generally dry for a time between its transference 

 from one stream to another. This is the more likely to be true the more porous 

 the valley bottom is. This remark suggests one case of especial interest where this 

 method of rearrangement of the drainage occurs. The streams draining the ice 

 front carried away vast quantities of sand and gravel to be deposited in their val- 

 leys ; hence in them the amount of porous subsurface drainage through the porous 

 beds of the valley is likely to be very large. The contribution of water from the 

 melting ice has been abstracted, and henc^e such old channels are now occupied by 

 streams much too small for them. At times the amount of surface drainage is so 

 slight that there are merely swamps or here and there pools and lakes in the sand 

 plains, with no apparent outlet, in spots where here and there the subsurface cur- 

 rent comes to the daylight. Such lakes are typically known as " crooked lakes" 

 in our Michigan nomenclature. The above state of affairs I suppose to be widely 

 prevalent. In Michigan, over the lower peninsula, it is also true that the ice 

 retired first from the higher parts of the rock surface and extended in three lobes 

 with two cusps between. The three lobes occupied lake Michigan, lake Huron, 

 and Saginaw bay. Streams like the Tittabawassee, the headwaters of the Cass, 

 and Huron rivers, and Dowagiac creek are, as Doctor Gordon has remarked, espe- 

 cially liable to the kind of capture that I have described from a host of vigorous- 

 streams which work up directly from the lake and have only to cut through interven- 

 ing unconsolidated moraines, which often seem to have overriden gravels connected 

 with the gravels of these glacial drainage valleys. When they get fairly gnawing 

 at the moraine nuuierous springs appear about their headwaters, and they are 

 helped by flowing wells put down by farmei's. By the time they have got into the 

 old drainage valley, considerably before they have abolished the divide between 

 them and the old glacial channel, the subsurface drainage into them is so ample 

 that the old valley is so dry as to be overgrown with verdure. An excellent cross- 

 section may be studied at Rose City (township 24 north, range 3 east). At Rose 

 City itself we are near the headwaters of a stream which flows fairly direct to lake 

 Huron. The region is full of springs and of artesian wells which have 10 to 20 feet 



