placed on the new map. The fifth survey was made in the spring of 1905, by the United States Geological Survey and the State Engineer of New York, the work being done by W, Carvel Hall, and his report follows this paper.

The crest lines determined by the five successive surveys are platted together on Pl. II, and their examination demonstrates clearly the gradual retreat of the crest of the Horseshoe Fall. Each mapped crest line is, on the whole, farther upstream than its predecessor, and their interspaces are roughly comparable with the time intervals between the making of the surveys; but each of these statements requires qualification. The region of maximum retreat has shifted from one part to another of the crest during the period of observation, so that in any one part the rate of retreat has been irregular; and when the chart is closely scrutinized it is found that the different lines overlap one another at various points, so that if all of them were rigidly accurate their record would show that the crest line had in places advanced downstream, instead of retreating. In the report of the last survey it is suggested that some of these discrepancies may be explained by an actual sliding forward of upper layers of limestone before they toppled over the brink, but the greater discrepancies can not be explained in this way, and the discrepancies as a whole are unquestionably due to errors in the topographic work, chiefly thru failure to identify points previously sighted when intersecting bearings were taken. Fortunately, they are not of such character or extent as to impair the general conclusions to be drawn from the work; but they serve to caution the student against any overrefinement in the discussion of results.

The Erie canal is supplied with water from the Niagara river at Buffalo, the Welland canal is supplied from Lake Erie, and the Chicago Drainage canal draws water from Lake Michigan. All the water thus diverted is withdrawn from the cataract. So also is water diverted from the river above the falls for factory purposes and for use in the generation of electricity. In recent years the diversion for electric power has rapidly increased, and existing charters authorize so large a draft upon the river that it has come to be recognized that the scenic value of the cataract is in

c Seventh Ann. Rept. Comrs. State Res. Niagara for fiscal year October 1, 1889, to September 30, 1890, Albany, 1891.

peril. A vigorous protest has been made by lovers of natural beauty, and negotiations are in progress for an international agreement to check and regulate the economic exploitation of the river. Whatever the outcome of these negotiations, there is no reason to expect that the natural flow of the river will be restored, and it is believed that from this time onward the natural conditions will be so far interfered with as to modify the rate of recession. As the geologist is primarily interested in the natural rate of recession, the present time is opportune for a summing up of the data. In fact, the survey of 1905 was ordered in view of the change of conditions from natural to artificial.a

THE HORSESHOE FALL.

The Horseshoe Fall is at the head of the gorge. From its edges the walls of the gorge run northeastward approximately parallel. The American Fall is at the side of the gorge, 2,500 feet from its head, and is separated from the Horseshoe Fall by Goat Island. A few hundred years ago the two falls were together, the position of the united cataract being somewhere in the neighborhood of the present American Fall. The subsequent retreat of the Horseshoe Fall has had the effect of lengthening the gorge, but the American Fall has not in the same time made an alcove in the side of the gorge. With reference therefore to the question of the age of the gorge, it is the Horseshoe Fall whose rate of recession is important.

The chief data for the estimation of the rate of recession are the maps of 1842 and 1905, the time interval being sixty-three years. The outlines from those maps are shown in fig. 4. These data, like other statistical data, can be discussed in a variety of ways and made to yield widely divergent results a fact sufficiently illustrated by earlier estimates of the rate of recession based on comparisons of the map of 1842 with that of 1875, 1886, or 1890. The following paragraphs therefore set forth somewhat fully the

a Since this paper was written it has come to my knowledge that a resurvey of the Niagara River is being made by the United States Lake Survey, the field work for the crest of the falls having been done in the summer of 1906. This will afford an additional datum on the rate of recession, but is not likely to affect the computation to a material extent. The addition of one year to the period of observation will probably be offset by changes occurring within that year. Inspecting the Horseshoe curve in August, 1906, I was confident that a salient near the angle of the curve, which was recorded by the surveys of 1890 and 1905, did not then exist. G. K. G.

HORSESHOE FALL IN 1827. Copy of sketch by Capt. Basil Hall, made with camera lucida, from veranda of Forsyth's Hotel.

BULLETIN NO. 306

methods here used, with the principal considerations on which they are based.

In the lengthening of the gorge the river does its principal work in that part of the Horseshoe curve where the current is deepest. The agitation of the plunging water is there so powerful as to roll about the fallen blocks of limestone, using them as tools to grind the shale, and at the same time breaking them up and eventually washing them downstream. The scour maintains a deep hollow beneath this part of the fall, a hollow whose depth is greater than the height of the fall. (Fig. 1, p. 41.) At the sides of the channel, especially near the right bank, where the sheet of falling water is comparatively thin, the fallen blocks are not cleared away, but cumber the base of the cliff. (Pl. X, p. 61.) As the cataract redeep channel, or elongated pool, in

treats it leaves behind it a

When the shale is ex

which the current is slow. Below the cataract the gorge is widened at top by the falling away of its banks. posed to the air it becomes subject to frost action, and for a time the limestone ledge above continues to be undermined, but a practical limit is reached as soon as the talus of fallen material covers the slopes of shale, and thereafter the change is exceedingly slow. The real lengthening of the gorge is along that portion of the Horseshoe where the sheet of falling water is heavy enough to clear away the debris and maintain a deep pool. The retreat of the cliff on either side of this portion is secondary, and appears to have little or no bearing on the question of the rate at which the gorge is growing longer. I have therefore restricted attention to the central part of the Horseshoe curve.

As the two crest lines compared are irregular in outline, a certain confusion arises if the recession of different parts is considered separately. At one place the recession seems to have one direction, at another place to have another direction, and various complications ensue when attempt is made to combine measurements made in different directions. In view of this difficulty it has appeared to me both convenient and legtimate to assume some one direction as the general direction of recession and at all points measure the amount of recession on lines parallel to that direction. From an inspection of the crest lines as wholes and in their relation to each other I have inferred such a general direction of re