It has already been stated that the ice probably dammed the river, and that a lake was formed above Kilbourn City, reaching east to the ice and west over the lowland tributary to the river, the water rising till it found an outlet, perhaps down to the Black river valley.

When the ice retreated, the old valley had been partly filled, and the lowest line of drainage did not everywhere correspond with it. Where the stream follows its old course, it flows through a wide capacious valley, but where it was displaced, it found a new course on the broad flat which bordered its preglacial course. Displacement of the stream occurred in the vicinity of Kilbourn City, and, forced to find a new line of flow west of its former course, the stream has cut a new channel in the sandstone. To this displacement of the river, and its subsequent cutting, we are indebted for the far-famed Dalles of the Wisconsin (p. [69]). But not all the present route of the river through the dalles has been followed throughout the entire postglacial history of the stream. In Fig. [44], the depression a, b, c, was formerly the course of the stream. The present course between d and e is therefore the youngest portion of the valley, and from its lesser width is known as the "narrows." During high water in the spring, the river still sends part of its waters southward by the older and longer route.

The preglacial course of the Wisconsin south of the dalles has never been determined with certainty, but rational conjectures as to its position have been made.

The great gap in the main quartzite range, a part of which is occupied by Devil's lake, was a narrows in a preglacial valley. The only streams in the region sufficiently large to be thought of as competent to produce such a gorge are the Baraboo and the Wisconsin. If the Baraboo was the stream which flowed through this gorge in preglacial time, the comparable narrows in the north quartzite range—the Lower narrows of the Baraboo—is to be accounted for. The stream which occupied one of these gorges probably occupied the other, for they are in every way comparable except in that one has been modified by glacial action, while the other has not.

Fig. 44. -- The Wisconsin valley near Kilbourn City.
[See larger image]

The Baraboo river flows through a gorge—the Upper narrows—in the north quartzite range at Ablemans, nine miles west of Baraboo. This gorge is much narrower than either the Lower narrows or the Devil's lake gorge, suggesting the work of a lesser stream. It seems on the whole probable, as suggested by Irving, [11] that in preglacial time the Wisconsin river flowed south through what is now the Lower narrows of the Baraboo, thence through the Devil's lake gorge to its present valley to the south. If this be true, the Baraboo must at that time have joined this larger stream at some point east of the city of the same name.

The Driftless Area.

Reference has already been made to the fact that the western part of the area here described is driftless, and the line marking the limit of ice advance has been defined. Beyond this line, gravel and sand, carried beyond the ice by water, extends some distance to the west. But a large area in the southwestern part of the state is essentially free from drift, though it is crossed by two belts of valley drift (valley trains) along the Wisconsin and Mississippi rivers.

The "driftless area" includes, besides the southwestern portion of Wisconsin, the adjoining corners of Minnesota, Iowa and Illinois. In the earlier epochs of the glacial period this area was completely surrounded by the ice, but in the last or Wisconsin epoch it was not surrounded, since the lobes did not come together south of it as in earlier times. (Compare Plate [XXXVIII] and Fig. [36].)