How far south the area of glacial accumulation may have extended cannot be definitely ascertained, but doubtless at an early period of the great Ice age the northern portions of the Appalachian range in New York, New England, New Brunswick, and Nova Scotia became themselves centres of dispersion, while only at the height of the period did all their glaciers become confluent, so that there was one continuous ice-sheet. In the western portion of the area covered by the Laurentide Glacier, the depression occupied by the Great Lakes, especially Lakes Michigan and Superior, evidently had a marked influence in directing the flow of ice during the stages which were midway between the culmination of the Ice period and both its beginning and its end. This would follow from the great depth of these lakes, the bottom of Lake Michigan being 286 feet below sea-level, and that of Lake Superior 375 feet, making a total depth of water of about 900 and 1,000 feet respectively. Into these oblong depressions the ice would naturally gravitate until they were filled, and they would become the natural channels of subsequent movement in the direction of their longest diameters, while the great thickness of ice in them would make them the conservative centres of glacial accumulation and action after the ice had begun to retreat. These deductions from the known nature of ice and the known topography of the region are amply sustained by a study of the detailed map showing the glacial geology in the United States. But on this we can represent indeed only the marks left by the ice at various stages of its retreat, since, as already remarked, the marks of each stage of earlier advance would be obliterated by later forward movements. We may presume, however, that in general the marks left by the retreating ice correspond closely with those actually made and obliterated by the advancing movement. From observations upon the glaciers of Switzerland and of Alaska, it is found that neither the advance nor the retreat of these glaciers is constant, but that, in obedience to meteorologic agencies not fully understood, they advance and retreat in alternate periods, at one time receding for a considerable distance, and at other times regaining the lost ground and advancing over the area which has been uncovered by their retreat. "M. Forel reports, from the data which he has collected with much care, that there have been in this century five periods in the Alpine glaciers: of enlargement, from 1800 (?) to 1815; of diminution, from 1815 to 1830; of enlargement, from 1830 to 1845; of diminution, from 1845 to 1875; and of enlargement again, from 1875 onward. He remarks further that these periods correspond with those deduced by Mr. C. Lang for the variations for the precipitations and temperature of the air; and, consequently, that the enlargement of the glaciers has gone forward in the cold and rainy period, and the diminution in the warm and the dry."[BH]