the escarpment of the terrace, and wells have been sunk at Beaver, one mile up the stream, that reach river level without finding rock. It may be, however, these wells lie in the course of the buried channel.

Two hundred yards from the mouth of the Beaver we find that stream flowing over the soft shales that overlie the Homewood sandstone. Of the extent of this bench, which here lies at stream level, nothing is known. On the eastward a rock escarpment cuts it off, to the west it is covered with alluvium. Towards the Ohio it is cut off by the buried channel of that river, and up the Beaver it is covered by the impounded waters of that stream. It seems most probable, however, that it is but a small, triangular bench, lying between the buried channels of the Ohio and Beaver rivers and the rock escarpment, here rising on the eastern bank of the Beaver.

On the eastern bank of the Beaver, at Rochester, the rock is seen passing under the higher portion of the terrace. The Rochester terrace is triangular in shape, bounded by the Ohio and Beaver rivers on the two sides, and the bounding hill of the valley on the northeast. The face of the escarpment along the Ohio has been cut by the railroad and no rock is seen. It is impossible therefore to assign any dimensions to the rock shelf seen on the Beaver side.

Passing up the Ohio we find a rock bench showing along the railroad cut below Baden and extending, almost continuously, to Legionville. Here the rock shelf is plainly visible, and even in a cursory examination it is evident that, measuring from the bottom of the buried channel, almost the whole terrace is rock in place.

The terrace at Economy, extending from Legionville to the Allegheny County line, has long been considered entirely gravel. Wells have been sunk at the town of Economy to river level without finding rock, yet a short distance below the town a quarry has been worked, right on the escarpment of the inner terrace. Of the extent of the rock shelf we have yet no means of knowing.

Returning to the Beaver and passing up that stream, we first come to a small rock bench about twenty-five feet above stream level, occurring at the mouth of Brady's run. This stream at its mouth, is now flowing in a rock channel, while a half mile up the stream it has a buried channel fifty feet, or more, in depth. The old channel evidently passed into the Beaver higher up that stream, and the present channel is a post terrace one.

Passing up the Beaver to the upper end of Beaver Falls, we reach a point where the Beaver, passing out of the narrow gorge in which it is confined for some eleven or twelve miles,

flows in a modern channel for about two miles. From records it is known that the buried channel passes to the westward of the present stream in a long curve and is crossed by the present river, nearly at right angles, about one and a half miles below, thence the buried channel, now passing beneath New Brighton, makes a sweep to the right and the present channel passes into it a half mile further on.

The modern channel, from where it leaves the old one, at the upper end of Beaver Falls, flows in a shallow trough, cut in a rock bench left in the original cutting of the inner gorge, and this modern channel represents the post-terrace work of the Beaver. This stretch of the Beaver presents, in many respects, the best place to study the inner rock benches, for a rock escarpment is present for much of the distance. On the New Brighton side, from the point where the new channel. enters the buried one, rock is seen in a low shelf extending under the inner alluvium and reaching to the point where the old channel is crossed by the modern one. Above this point the river runs close to the hill bounding the valley on the east, until Eastdale is reached, at the upper end of the modern channel, where quite a large shelf is seen, extending under the whole of the inner terrace at that point. On the western side of the stream, passing upwards, we find a rock bench, corresponding to the one at New Brighton. It occupies the triangle lying between the present stream, Walnut Bottom Run, and the western bounding hill, having a frontage of about onefourth mile on the Beaver and one-half mile on the Walnut Bottom Run side. About one-half mile above the mouth of Walnut Bottom Run rock is again seen in place, extending continuously to the point of junction with the buried channel. This shelf is nowhere seen on the Walnut Bottom Run side, but is well defined by wells, and is known to extend about one mile in a southwesterly direction, being but a few feet below. the surface until Twelfth Street is reached, from which point it falls rapidly, owing to the proximity of the buried channel.

In the narrow gorge which extends from Beaver Falls to Wampum, it was not expected that any fragments of inner rock benches could be found, but an examination of the gravel deposit at Clark's Run (Homewood) shows that it is resting on a rock bench. In changing the line of the P. & L. E. R. R. it was thrown into the hill, and the necessary cutting disclosed a bench consisting of the soft shales underlying the Homewood sandstone.

Passing out of the narrow gorge at Wampum, we have passed into the glaciated region, and the inner benches are somewhat harder to trace. The character of the inner gorge also changes, becoming wider than at any point below. This

is due to the character of the strata cut, here being much softer than below where the gorge is cut in hard sandstones.

A short distance above Wampum a bench is seen, extending some distance, and, crossing the river to Moravia (east side), a corresponding bench is seen. The river at this point makes a sharp reverse curve, cutting from side to side of the valley, so that these two benches are really one, cut across by the river. The bench at Moravia reaches well to the confluence of the Shenango and Mahoning rivers.

Along the Mahoning river rock in place is frequently seen, and some of the inner terraces are undoubtedly of this character. The same is true of the Shenango valley, but neither of these streams have yet been studied or mapped.

A glance at a map will show we have followed the Ohio river from the Ohio and Pennsylvania line to a point but fifteen miles below the confluence of the Allegheny and Monongahela rivers, a distance of twenty-four miles, and the Beaver river its entire length, a distance of twenty-two miles. The following table gives the terraces on the rivers studied and the occurrences of rock where noted above. The flood plain is not given except where underlaid by rock.

It will be noticed there are twenty-two observed terraces, and of these fourteen are underlaid by rock benches, to a

greater or less extent. The first four, and those of Monaca, Aliquippa and Shannopin are not favorably situated to show the presence of an included rock shelf; no stream cutting them in such a way as to reveal their structure.

It has not been possible in all cases to assign even approximate limits to the observed rock benches. At Raccoon, Beaver, Rochester and Economy it is evident they underlie but a portion of the terrace and there is as yet no data that will enable any estimate of proportion of included rock shelf to entire terrace to be made. At New Brighton the rock extends only under the lower terrace, and whether it underlies more than that portion of the terrace lying between the buried and modern channels of the river has not yet been determined. At Beaver Falls the limits of the rock are well defined, and estimating from the bottom of the buried channel, one-half of the whole terrace would seem to be rock. At Eastdale the rock rises practically to the surface, and fully nine-tenths of this fragment is rock. At Homewood the rock underlies the whole of the gravel deposit and constitutes one-half of the whole (the gravel here rising to the level of the outer rock bench). At Wampum the greater portion of the terrace seems to be rock. The shelf shows for some distance along the river front and rises almost to the level of the terrace, which is also true at Moravia. In the terrace extending from Baden to Legionville the railroad cutting shows rock reaching practically to the surface and constituting at least four-fifths of the entire terrace.

quan

It will thus be seen that the data for forming an estimate is not sufficient to judge accurately, yet it must be evident that the included rock benches form no insignificant portion of the inner terraces, and reduces by that amount the immense tity of alluvium which it has been supposed made up the terraces. From the present data the writer believes the proportion of rock will be found over, rather than under, one-half of the entire volume of the terraces, measured from the bottom of the buried channel.

The attempt to correlate the rock benches has not been entirely satisfactory. Taken as a whole the benches on the Beaver (Moravia, Wampum, Homewood, Eastdale, Beaver Falls, New Brighton) lie at about the same elevation above present stream level (30' to 40'). The benches on the Ohio (Economy, Baden, Rochester, Beaver, Raccoon) lie at a somewhat higher elevation above stream level (50'±), yet it seems they belong to the same series found on the Beaver. The Ohio is flowing over its buried channel in this region and has eroded its post-terrace channel deeper than the Beaver, which has been cutting a new rock channel within about two miles of its mouth, and that through the hard Homewood sandstone.

This view of the benches is in harmony with the facts as shown on the Beaver. The rock benches at Beaver Falls and New Brighton, where the Beaver is cutting its new channel, lie nearer stream level than the benches at Wampum and Moravia, where the Beaver is re-excavating its buried channel. It may also be added that the greatly increased fall of the Beaver in the last four miles of its course where it drops down from the rock channel into its buried one, and thence into the Ohio, is sufficient to compensate for the greater elevation of the benches on the Ohio.

It would seem therefore that sufficient is now known to justify the conclusion that the cutting of the inner gorge is to be subdivided into two periods. (1) The time required to cut from the level of the lowest of the upper rock benches to the level of the inner rock benches, when erosion seems to have approached a base level, and (2) a second period of cutting from the inner rock benches to the bottom of the buried channel. The first period evidently much the longer. The bench at Homewood, for example, is some seventy feet above the bottom of the inner gorge (the bottom of the buried channel), and about one hundred feet below the lowest of the upper benches. The work of erosion in the first period was here through the Homewood sandstone in great measure, while the cutting of the second period was in shales to a great extent; the lower portion of the gorge is also narrower than the upper portion. It seems safe to say, therefore, the first period represents two thirds to three-fourths of the entire work of excavation, and probably a larger proportion in time.

After the excavation of the buried channel, the next chapter of the river history was the silting of the valleys to the top of the inner terraces. It is not purposed in this paper to discuss the inner alluvium, further than to call attention to a mistake regarding it, evidently due to a study of the gravel escarpments, without examination of the structure back from the face. It has been described as quite coarse, ranging up to two feet in diameter, and the impression prevails that it is uniformly of this character, from top to bottom, and from escarpment to the base of the hills bounding the inner gorge.

It seems evident, and generally admitted, that the alluvium should be considered as a whole, from the top of the highest terraces, to the bottom of the buried channel. The records that determined the existence of the buried channel were not sufficient to furnish any knowledge of the character of the filling, and it is only recently any real knowledge has come to light, and rendered it possible to form a correct idea of the filling material and consequently of the stream that deposited it.