The broad channel now followed by the Minnesota River is analogous to that across the Maumee-Wabash divide, although the lake from which the river flowed to cut the channel has shrunk away so far as to withdraw its waters beyond our northern boundary. The former occupation of the Minnesota channel by a large river was first pointed out by General G. K. Warren, in the Annual Report of the Chief of Engineers, 1868, page 307; and a fuller account of it was published in the American Journal of Science for December, 1878. Warren looked to a northward elevation of the land as a reason for the former southward direction of drainage, as Gilbert had done in Ohio; and this view generally obtained until the region was carefully studied out by Warren Upham, whose reports are found in the annual volumes of the Minnesota Geological Survey, and by whom a special account of the valley is given in the Proceedings of the American Association for 1883. Here the name of river Warren is proposed for the ancient stream by which this great trough was excavated: in earlier papers Upham had given the name of Agassiz to the lake from which the river issued. Further account of the ancient lake and river is found in the first two volumes of the final report on the Geology of Minnesota, now in progress.
The same observer has described a southwest overflow from Lake Superior, when the greater part of its basin was presumably occupied by retreating ice, and its waters rose about five hundred feet above their present level. The overflow took place across the pass between the valleys of the Bois Brulé and the St. Croix Rivers in northwestern Wisconsin; the channel across the pass being about a thousand feet wide and nearly a hundred feet deep, although its depth is now somewhat decreased by a marshy filling from which the headwaters of the Bois Brulé run back to the present lake (Geological and Natural History Survey of Minnesota, ii, 1888, p. 642). The lower course of the St. Croix follows a "great valley," whose fuller history is deferred to later volumes of the Minnesota Survey; its lakelike expansion above its confluence with the Mississippi is said to be due to alluvial obstruction by tributaries (ibid., pp. 377, 643). These studies by Upham in the Northwest were only continuations of the work that he had begun in New Hampshire several years before, where he recognized the shore lines of a small glacial lake in the southern part of the north-sloping Contoocook Valley, with an overflow to the southeast at Greenfield, N. H. (Geological Survey of New Hampshire, iii, 1878, pp. 116-119). A few years ago, in company with Mr. C. L. Whittle, I traced out a number of deltas and shore lines on the slopes of this picturesque valley, but have not since then been able to complete the attractive study of mapping and restoring the old lake.
