The fall of the Niagara[52] (Pl. IX) is one of the most remarkable known, both because of its large volume of water and its great descent, between 160 and 170 feet. The rate at which the fall is receding is a matter of interest not only in itself, but because, once determined, it may be made to serve as a unit of measurement for certain important events in geological history. It was formerly conjectured that this fall was receding at the rate of one to three feet per century, but it was not until recent years that its actual rate of recession was approximately fixed. By surveys executed in 1842 and 1890 it has been determined that its average rate of recession between those dates was something like 4½ feet per year, or about 150 times as great as the highest estimate stated above. It is to be noted that this is the _average_ rate of recession, for all parts of the ledge over which the water falls are not receding at the same rate. The point of the “Horseshoe” has, during the same time, gone back at more than twice this rate.[53] [Illustration: +Fig.+ 119.—A group of pot-holes. (Turner, U. S. Geol. Surv.)] Rapids and falls sometimes occasion the development of _pot-holes_ (Fig. 119), a peculiar rather than important erosion feature. The holes are excavated in part by the falling and eddying of silt-charged water, but chiefly by stones which the eddies move. Pot-holes which are not now in immediate association with rapids or falls often point to the former existence of rapids or falls. =Rock terraces.=—The tendency to sapping shown in many waterfalls is also shown in the weathering and erosion of the sides of a valley where a hard layer outcrops above the bottom, and the profile of the side slopes of the valley simulates that of the stream; that is, the slope becomes gentle just above the hard layer, and steep, or even vertical, at and below its outcrop. This is illustrated by Fig. 120, where the hard layer through which the stream has sunk its valley stands out as a rock terrace on either side of the valley. Such terraces are not rare and are popularly believed to be old “water-lines”; that is, to represent the height at which the water once stood. In one sense this interpretation is correct, since a river has stood at all levels between that of the surface in which its valley started, and its present channel, but the shelf of hard rock does not mean that the river, after attaining its present channel, was ever so large as to fill the valley to the level of the terrace. Rock terraces may also result from changes of level. [Illustration: +Fig.+ 120.—Diagram to illustrate the development of rock terraces.]