As a rule, Professor Lewis and myself found it more difficult to determine with accuracy the exact point to which the ice extended in the axis of these south-flowing valleys than we did upon the highlands upon either side; and, in looking for the positive indications of direct ice-action in these lines of drainage, we were almost always led up the valley to a considerable distance inside of the line. This arose from our inexperience in interpreting the phenomena, or rather from our inattention to the well-known determining facts in the problem. On further reflection it readily appeared that this was as it should be. The ice-front, instead of extending farther down in a narrow valley than on the adjoining highlands (where they are of only moderate elevation) ought not to extend so far, for the subglacial streams would not only wear away the ice of themselves, but would admit the air into the tunnels formed by them so as to melt the masses both from below and from above, and thus cause a recession of the front. If we had understood this principle at the beginning of our survey, it would have saved us much perplexity and trouble. A single further illustration of this point will help to an understanding of many references which will hereafter be made to the water deposits which accumulated in the lines of drainage running southward from the glaciated area. At Warren, Pa., Conewango Creek, which is the outlet from Chautauqua Lake, enters the Alleghany River after flowing for a number of miles in a deep valley with moderate slopes. In ascending the creek from Warren, the gravel terraces, which rise twenty-five or thirty feet above high-water mark, rapidly increase in breadth and height, and the pebbles become more and more coarse. After a certain distance the regular terraces begin to give place to irregular accumulations of gravel in ridges and knobs. In the lower portion of the valley no pebbles could be found which were scratched. Up the valley a few miles pebbles were occasionally discovered which showed some slight indications of having been scratched, but which had been subjected to such an amount of abrasion by water-action as almost to erase the scratches. On reaching Ackley's Station, the stream is found to be cutting through a regular terminal moraine, extending across the valley and full of clearly marked glaciated stones. Above this terminal moraine the terraces and gravel ridges which had characterised the valley below disappear, giving place to long stretches of level and swampy land, which had been subject to overflow.