The deposits made by the water at the time of the maximum extension of the ice and during its final retreat, were never disturbed by subsequent glacier action. So far as not destroyed by subsequent erosion, they still retain the form and structure which they had at the outset. Such drift deposits, because they lie at the surface, and because they are more or less distinct topographically as well as structurally, are better known than the stratified drift of other stages of an ice sheet's history. Of stratified drift made during the maximum extension of the ice, and during its final retreat, there are several types.

A. At the edge of ice, on land.—If the subglacial streams flowed under "head," the pressure was relieved when they escaped from the ice. With this relief, there was diminution of velocity. With the diminution of velocity, deposition of load would be likely to take place. Since these changes would be likely to occur at the immediate edge of the ice, one class of stratified drift deposits would be made in this position, in immediate contact with the edge of the ice, and their form would be influenced by it. At the stationary margin of an ice sheet, therefore, at the time of its maximum advance, ice and water must have co-operated to bring into existence considerable quantities of stratified drift.

The edge of the ice was probably ragged, as the ends of glaciers are today, and as the waters issued from beneath it, they must frequently have left considerable quantities of such debris as they were carrying, against its irregular margin, and in its re-entrant angles and marginal crevasses. When the ice against which this debris was first lodged melted, the marginal accumulations of gravel and sand often assumed the form of kames. A typical kame is a hill, hillock, or less commonly a short ridge of stratified drift; but several or many are often associated, giving rise to groups and areas of kames. Kames are often associated with terminal moraines, a relation which emphasizes the fact of their marginal origin.

So far as the superficial streams which flowed to the edge of the ice carried debris, this was subject to deposition as the streams descended from the ice. Such drift would tend to increase the body of marginal stratified drift from subglacial sources.

Marginal accumulations of stratified drift, made by the co-operation of running water and ice, must have had their most extensive development, other things being equal, where the margin of the ice was longest in one position, and where the streams were heavily loaded. The deposits made by water at the edge of the ice differ from those of the next class—made beyond the edge of the ice—in that they were influenced in their disposition and present topography, by the presence of ice.

In the Devil's lake region isolated and well-defined kames are not of common occurrence. There are, however, at many points hills which have something of a kame-like character. There is such a hill a mile southeast of the Court house at Baraboo, at the point marked p, Plate [XXXVII]. In this hill there are good exposures which show its structure. There are many hillocks of a general kame-like habit associated with the terminal moraine south of the main quartzite range, and north of the Wisconsin river. Many of them occur somewhat within the terminal moraine a few miles northwest of Merrimac.

B. Beyond the edge of the ice, on land.—As the waters escaping from the ice flowed farther, deposits of stratified drift were made quite beyond the edge of the ice. The forms assumed by such deposits are various, and depended on various conditions. Where the waters issuing from the edge of the ice found themselves concentrated in valleys, and where they possessed sufficient load, and not too great velocity, they aggraded the valleys through which they flowed, developing fluvial plains of gravel and sand, which often extended far beyond the ice. Such fluvial plains of gravel and sand constitute the valley trains which extend beyond the unstratified glacial drift in many of the valleys of the United States. They are found especially in the valleys leading out from the stouter terminal moraines of late glacial age. From these moraines, the more extensive valley trains take their origin, thus emphasizing the fact that they are deposits made by water beyond a stationary ice margin. Valley trains have all the characteristics of alluvial plains built by rapid waters carrying heavy loads of detritus. Now and then their surfaces present slight variations from planeness, but they are minor. Like all plains of similar origin they decline gradually, and with diminishing gradient, down stream. They are of coarser material near their sources, and of finer material farther away. Valley trains constitute a distinct topographic as well as genetic type.

A perfect example of a valley train does not occur within the region here discussed. There is such a train starting at the moraine where it crosses the Wisconsin river above Prairie du Sac, and extending down that valley to the Mississippi, but at its head this valley train is wide and has the appearance of an overwash plain, rather than a valley train. Farther from the moraine, however, it narrows, and assumes the normal characteristics of a valley train. It is the gravel and sand of this formation which underlies Sauk Prairie, and its topographic continuation to the westward.

Where the subglacial streams did not follow subglacial valleys, they did not always find valleys when they issued from the ice. Under such circumstances, each heavily loaded stream coming out from beneath the ice must have tended to develop a plain of stratified material near its point of issue—a sort of alluvial fan. Where several such streams came out from beneath the ice near one another, their several plains, or fans, were likely to become continuous by lateral growth. Such border plains of stratified drift differ from valley trains particularly (1) in being much less elongate in the direction of drainage; (2) in being much more extended parallel to the margin of the ice; and (3) in not being confined to valleys. Such plains stood an especially good chance of development where the edge of the ice remained constant for a considerable period of time, for it was under such conditions that the issuing waters had opportunity to do much work. Thus arose the type of stratified drift variously known as overwash plains, outwash plains, morainic plains, and morainic aprons. These plains sometimes skirt the moraine for many miles at a stretch.

Overwash plains may sometimes depart from planeness by taking on some measure of undulation, of the sag and swell (kame) type, especially near their moraine edges. The same is often true of the heads of valley trains. The heads of valley trains and the inner edges of overwash plains, it is to be noted, occupy the general position in which kames are likely to be formed, and the undulations which often affect these parts of the trains and plains, respectively, are probably to be attributed to the influence of the ice itself. Valley trains and overwash plains, therefore, at their upper ends and edges respectively, may take on some of the features of kames. Indeed, either may head in a kame area.