The only condition in the problem about which there can be much chance of question relates to the constancy of the volume of water flowing in the Niagara channel. Mr. Gilbert had suggested that, as a consequence of the subsidence connected with the closing portions of the Glacial period, the water of the Great Lakes may have been largely diverted from its present outlet in Niagara River and turned northeastward, through Georgian Bay, French River, and Lake Nipissing, into a tributary of the Ottawa River, and so carried into the St. Lawrence below Lake Ontario. Of this theory there is also much direct evidence. A well-defined shore line of rounded pebbles extends, at an elevation of about fifty feet, across the col from Lake Nipissing to the head-waters of the Mattawa, a tributary of the Ottawa; while at the junction with the Ottawa there is an enormous delta terrace of boulders, forming a bar across the main stream just such as would result from Mr. Gilbert's supposed outlet. But this outlet was doubtless limited to a comparatively few centuries, and Dr. Robert Bell thinks the evidence still inconclusive.[EB] [Footnote EB: See Bul. Geol. Soc. Am., vol. iv, pp. 423-427, vol. v, pp. 620-626.] A second noteworthy glacial chronometer is found in the gorge of the Mississippi River, extending from the Falls of St. Anthony, at Minneapolis, to its junction with the preglacial trough of the old Mississippi, at Fort Snelling, a distance likewise of about seven miles. Above Fort Snelling the preglacial gorge is occupied by the Minnesota River, and, as we have before stated, extends to the very sources of this river, and is continuous with the southern portion of the valley of the trough of the Red River of the North. Before the Glacial period the drainage of the present basin of the upper Mississippi joined this main preglacial valley, not at Fort Snelling, but some little distance above, as shown upon our map.[EC] This part of the preglacial gorge became partially filled up with glacial deposits, but it can be still traced by the lakelets occupying portions of the old depression, and by the records of wells which have been sunk along the line. When the ice-front had receded beyond the site of Minneapolis, the only line of drainage left open for the water was along the course of the present gorge from Minneapolis to Fort Snelling. [Footnote EC: See above, p. 209.] Here, as at Niagara, the problem is comparatively simple. The upper strata of rock consist of hard limestone, which is underlaid by a soft sandstone, which, like the underlying shale at Niagara, is eroded faster than the upper strata, and so a perpendicular fall is maintained. The strata are so uniform in texture and thickness that, with the present amount of water in the river, the rate of recession of the falls must have been, from the beginning, very constant. If, therefore, the rate can be determined, the problem can be solved with a good degree of confidence.