Niagara is divided by Goat Island into the Horseshoe Falls and the American Falls. The former is supplied by the main current of the river, and from the semicircular sweep of its rim a sheet of water in places at least fifteen or twenty feet deep plunges into a pool a little less than two hundred feet in depth. Here the force of the falling water is sufficient to move about the fallen blocks of limestone and use them in the excavation of the shale of the bed. At the American Falls the lesser branch of the river, which flows along the American side of Goat Island, pours over the side of the gorge and breaks upon a high talus of limestone blocks which its smaller volume of water is unable to grind to pieces and remove. A series of surveys have determined that from 1842 to 1890 the Horseshoe Falls retreated at the rate of 2.18 feet per year, while the American Falls retreated at the rate of 0.64 feet in the same period. We cannot doubt that the same agency which is now lengthening the gorge at this rapid rate has cut it back its entire length of seven miles. While Niagara Falls have been cutting back a gorge seven miles long and from two hundred to three hundred feet deep, the river above the Falls has eroded its bed scarcely below the level of the upland on which it flows. Like all streams which are the outlets of lakes, the Niagara flows out of Lake Erie clear of sediment, as from a settling basin, and carries no tools with which to abrade its bed. We may infer from this instance how slight is the erosive power of clear water on hard rock. [Illustration: Fig. 39. Longitudinal Section of Niagara Gorge Black, water; _F_, falls; _R_, rapids; _W_, whirlpool; _E_, escarpment; _N_, north; _S_, south] Assuming that the rate of recession of the combined volumes of the American and Horseshoe Falls was three feet a year below Goat Island, and _assuming that this rate has been uniform in the past_, how long is it since the Niagara River fell over the edge of the escarpment where now is the mouth of the present gorge? The profile of the bed of the Niagara along the gorge (Fig. 39) shows alternating deeps and shallows which cannot be accounted for, except in a single instance, by the relative hardness of the rocks of the river bed. The deeps do not exceed that at the foot of the Horseshoe Falls at the present time. When the gorge was being cut along the shallows, how did the Falls compare in excavating power, in force, and volume with the Niagara of to-day? How did the rate of recession at those times compare with the present rate? Is the assumption made above that the rate of recession has been uniform correct? The first stretch of shallows below the Falls causes a tumultuous rapid impossible to sound. Its depth has been estimated at thirty-five feet. From what data could such an estimate be made? Suggest a reason why the Horseshoe Falls are convex upstream.