Ascending the east bluff a little east of the point where the drift ridge drops off the west bluff, the ridge is again found (at i) in characteristic development. For some distance it is located at the edge of the precipitous south face of the bluff. Farther on it bears to the north, and soon crosses a col (j) in the ridge, building it up many feet above the level of the bed-rock. From this point eastward for about three miles the marginal ridge is clearly defined, the slopes about equal on either side, and the crest as nearly even as the topography of the underlying surface permits. The topographic relations in this part of the course are shown in Fig. 40. At k, this marginal ridge attains its maximum elevation, 1,620 feet. At this great elevation, the ridge turns sharply to the northwest at an angle of more than 90°. Following this direction for little more than half a mile, it turns to the west. At some points in this vicinity the ridge assumes the normal morainic habit, but this is true for short distances only. Farther west, at l, it turns abruptly to the northeast and is sharply defined. It here loops about a narrow area less than sixty rods wide, and over half a mile in length, the sharpest loop in its whole course. The driftless tract enclosed by the arms of this loop is lower than the drift ridge on either hand. The ice on either side would need to have advanced no more than thirty rods to have covered the whole of it. From the minor loop just mentioned, the marginal ridge is continued westward, being well developed for about a mile and a half. At this point the moraine swings south to the north end of Devil's lake, loses the unique marginal ridge which has characterized its outer edge across the quartzite range for so many miles, and assumes the topography normal to terminal moraines. At no other point in the United States, so far as known to the writers, is there so sharply marked a marginal ridge associated with the terminal moraine, for so long a distance. From Plate II it will be seen that the moraine as a whole makes a great loop to the eastward in crossing the quartzite range. From the detailed description just given of the course of the marginal ridge, it will be seen that it has three distinct loops; one on the Devil's nose (west of g, Plate XXXVII); one on the main ridge (west of k) and a minor one on the north side of the last (southwest of m). The first and third are but minor irregularities on the sides of the great loop, the head of which is at k.