But, keenly as he might be impressed with the poetic grandeur of the falls, he could not forget the scientific questions which were ever present to his mind. The gorge of Niagara offered a problem for solution which had for him a special fascination. Not only did it illustrate on a grand scale the potencies of water in rapid motion, but also it furnished data for estimating the period during which this agent had been at work. The gorge has been carved in a plateau of Silurian rock, which terminates, seven miles below the falls, in a precipitous escarpment overhanging Queenstown. There was a time when that gorge did not exist, when the river first took its course along the plateau on its way from Lake Erie, and plunged over the brink of the escarpment. The valley at first was nothing more than a shallow trench excavated in the drift which covers the surface of the country--such an one as may still be seen between Lake Erie and the falls--but the river, slowly and steadily, has cut its way back through the rocky plateau from the first site of the falls near Queenstown to their present position. The upper part of this plateau consists of a thick bed of hard limestone, but beneath this the deposits become softer; and the lowest bed is the most perishable. The water, as it plunges down, undermines the overlying rock. The gorge began at once to be developed, and it has ever since continued to retreat towards Lake Erie. Every year makes some slight change. This becomes more marked when old histories are consulted and old drawings compared with the present aspect of the scene. Father Hennepin's sketch, of which Lyell gives a copy,[96] rude and incorrect as it is, proves beyond all question that the changes in the neighbourhood of Table Rock have been very considerable, for it shows that on this side a third and much narrower cascade fell athwart the general course of the main mass of water. This cascade, by the time of Kalm's[97] visit in 1751, had ceased to be conspicuous, and had quite disappeared before the date of Lyell's visit. The Horseshoe Fall also at the present time is less worthy of the name than it was at that date, for its symmetry has been seriously marred by a deep notch which the northern stream has cut in the more central part of the curve.[98] Careful inquiry convinced Lyell that the slow recession of the falls was an indubitable fact, and that its rate, on an average, was about a foot a year. As the gorge is about seven miles long, this would fix its beginning about 35,000 years ago.[99]