The physical heterogeneity may be seen at all exposures, and is illustrated in Plate [XXXV]. The larger stones of the drift are usually of some hard variety of rock. Near the Baraboo ranges, the local quartzite often predominates among the bowlders, and since such bowlders have not been carried far, they are often little worn. Away from the ranges, the bowlders are generally of some crystalline rock, such as granite and diabase. Bowlders of these sorts of rock are from a much more distant source, and are usually well worn.

In general the till of any locality is made up largely of material derived from the formations close at hand. This fact seems to afford sufficient warrant for the conclusion that a considerable amount of deposition must have gone on beneath the ice during its movement, even back from its margin. To take a concrete illustration, it would seem that the drift of southeastern Wisconsin should have had a larger contribution than it has of material derived from Canadian territory, if material once taken up by the ice was all or chiefly carried down to its thinned edge before deposition. The fact that so little of the drift came from these distant sources would seem to prove that a large part of the material moved by the ice, is moved a relatively short distance only. The ice must be conceived of as continually depositing parts of its load, and parts which it has carried but a short distance, as it takes up new material from the territory newly invaded.

In keeping with the character of till in general, that about Devil's lake was derived largely from the sandstone, limestone and quartzite of the immediate vicinity, while a much smaller part of it came from more distant sources. This is especially noticeable in the fine material, which is made up mostly of the comminuted products of the local rock.

Topography.—The topography of the ground moraine is in general the topography already described (p. [85]) in considering the modification of preglacial topography effected by ice deposition. As left by the ice, its surface was undulating. The undulations did not take the form of hills and ridges with intervening valleys, but of swells and depressions standing in no orderly relationship to one another. Undrained depressions are found in the ground moraine, but they are, as a rule, broader and shallower than the "kettles" common to terminal moraines.

It is in the broad, shallow depressions of the ground moraine that many of the lakes and more of the marshes of southeastern Wisconsin are located.

The rolling, undulating topography characteristic of ground moraines is well shown about the City of Baraboo and between that point and the lake, and at many less easily designated points about Merrimac.

In thickness the ground moraine reaches at least 160 feet, though its average is much less—too little to obliterate the greater topographic features of the rock beneath. It is, however, responsible for many of the details of the surface.

Terminal Moraines.

The marginal portion of the ice sheet was more heavily loaded—certainly more heavily loaded relative to its thickness—than any other. Toward its margin the thinned ice was constantly losing its transportive power, and at its edge this power was altogether gone. Since the ice was continually bringing drift down to this position and leaving it there, the rate of drift accumulation must have been greater, on the average, beneath the edge of the ice than elsewhere.

Whenever, at any stage in its history, the edge of the ice remained essentially constant in position for a long period of time, the corresponding submarginal accumulation of drift was great, and when the ice melted, the former site of the stationary edge would be marked by a broad ridge or belt of drift, thicker than that on either side. Such thickened belts of drift are terminal moraines. It will be seen that a terminal moraine does not necessarily mark the terminus of the ice at the time of its greatest advance, but rather its terminus at any time when its edge was stationary or nearly so.