material upward pressure. Therefore the author is of the firm opinion that the failure was not due to the weakness of the structure itself.

If the dam were built upon a solid plate, underneath which were rollers, rolling in a down-stream direction, the conditions which resulted would have been fulfilled exactly; that is, the entire dam would act as a beam uniformly loaded, and if this beam were not sufficiently strong it would fracture near the center and probably at other places, and if there were any cracks due to temperature, it would fail at these cracks, and when final failure took place one might find fresh planes of fracture near each end, providing these were carried sufficiently into the hillsides to prevent going down in a body. This is exactly what took place, and an examination of the rock will indicate that there existed a condition quite analogous to the roller illustration. This rock belongs in the formation, known as "Catskill Sandstone," and at this point the sandstone exists in thin layers, averaging from a few inches in thickness to several feet, but at the bottom of the dam there was probably no very thick stratum. Some of these layers were parted by shaly clay, pervious to water. The upper surface of the rock looked good enough as a bearing surface for the dam, and therefore the builders did not extend the dam down into the rock any material distance. Probably, solid bed-rock was all they were seeking. The consequence was, when the dam was subjected to the full pressure of water, the upper layers of rock upon which the dam was founded slid on the layers below along one or more of these clay partings, and as there was nothing to resist this except a few feet of earth below the dam, the resistance was not sufficient to withstand the 250,000 tons of force, and, therefore, the entire dam with one or more layers of stratification moved downstream, and pushed the earth ahead of it, and crumpled up the stratification in a manner very similar to what is seen at the foot of steep hills in the coal formation due to extensive landslides. (See sketch, Fig. 8.)

The upward pressure of water exerted between these layers of rock, no doubt, aided to facilitate this movement. This theory fully accounts for all the facts, as will no other theory that has been proposed up to date.

As to the question of where the blame lies for this failure the resident engineer, in charge of the work at the time the concrete was started, would probably be the best judge. The cause of the failure of the Austin, Texas, dam was stated by the engineers who made

the investigation, to be due to erosion by constantly falling water against the comparatively soft rock under the down-stream toe, this erosion having gone so far back under the dam as to make the structure unstable. Others have stated that the rock slid on itself causing the dam to fail by sliding in the manner above stated. With this possible exception the cause of the failure of the Austin, Pennsylvania, dam seems to be unique, and, therefore, the judgment of the engineer who was in charge of the work at the time the excavation was completed might not be severely censured in view of the accumulated knowledge and state of the art up to that date. It must be remembered that he was working for a corporation who desired to

build this dam at the lowest possible cost consistent with safety. The engineer saw the surface of the rock as it was exposed to view, and found that it was clean and hard, and certainly possessed sufficient bearing power to withstand the downward pressure. He was also positive that the coefficient of friction between the concrete and this rock would be sufficient to prevent sliding. He had probably sunk test holes enough to indicate that the character of the rock for a considerable distance below the surface was not any more solid or suitable for a foundation. Therefore, he saw no reason to go deeper, or to believe that the dam would overturn or slide on this rock surface. His reasoning up to that point was absolutely sound, and

this would naturally have been sufficient for a great many engineers of standing who desired to save for their employer every possible dollar. However, it is certain that the vital weakness must have been overlooked. The coefficient of friction between hard concrete and sandstone rock, even under the smoothest conditions, is at least 0.70; that is, horizontal, it takes nearly three-fourths as much force to slide a body as the body has weight, but where the concrete is laid on the roughened surface of the rock with indentations, the coefficient of friction would be a great deal more than this, and probably over unity.

This dam weighed about 9400 pounds per lineal inch, and the pressure of water against the dam was about 6300 pounds per lineal inch, so the dam had a factor of safety against sliding of as much as 12, but now, the coefficient of friction between two rock layers separated by a layer of pure clay is only, and when the clay is moist, may be even less, so dividing 9300 by 3, there is found a trifle over 3000 pounds per lineal inch, while the pressure of water was 6000 pounds per lineal inch. It is, however, not likely that the clay was in a pure state, or of a sufficient thickness in any one stratum to reduce the coefficient of friction to .3, but it will be seen from the above calculation that it would not have to be reduced to less than 0.6 to cause sliding. The resistance of the bank of earth below the dam, which bank was probably not more than 6 or 8 feet high, would be trifling and it would not exert more than 300 or 400 pounds per lineal inch against the dam, but there is a matter of probably 2000 or 3000 pounds to resist. Therefore, due to these conditions which are plainly seen in the nearby quarries, sliding in the manner indicated is almost certain to have occurred.

If this, then, was the cause of the accident the remedy would have been, as soon as the structure was seen to be weak, to construct below the dam, a wide trench to a depth of 10 or more feet below the rock surface, and this should have been carried up to 4 or 5 feet above the bottom of the dam so as to form a bulkhead or barrier to prevent sliding. The thickness of this wall should have been sufficient to prevent its breaking off due to shear and beam action necessary to resist the sliding of the dam. A thickness of 15 or 20 feet, and a depth of 10 feet below the surface of the rock, would have been ample for this purpose. Of course, this could be built instead of rectangularly in such a shape as to form a brace or strut against the toe of the dam.

In order to stop the excessive leakage, as well as to prevent pres

sure acting upward against the dam, it would probably have been worth while to sink a deep cut-off trench on the up-stream side of the concrete masonry, and to a depth of 15 or more feet. This trench should have been filled solidly with concrete which should have extended slightly under the dam.

All this could have been obviated, originally, by carrying the dam down 5 or 10 feet deeper.

There is one feature of the construction of the Austin dam that is particularly noticeable and which was mentioned earlier in this paper. The down-stream slope, in all cases, ends at a very short distance below the natural surface of the ground, which would be perfectly justifiable, if the natural ground were solid rock high in resisting power; but, it would certainly not be justifiable when the material is sand, clay and gravel as in this case. An examination of the diagram of the resolution of forces (Fig. 9) shows that the resultant comes within the middle third at all points above the intersection of the slope with the vertical down-stream face; but, below this point the resultant falls considerably beyond the middle third. However, this analysis is not of great importance in view of the fact that the dam did not fail by overturning but by sliding.

As a general proposition, engineers designing masonry dams are not required to give much consideration to the danger of sliding. In the great majority of cases it is necessary, for other reasons, to go to a depth considerably below the rock surface, and this, by proper construction, can be made a bulkhead to resist the sliding of the dam.

From the calculations given above it will be seen that the effect of possible upward pressure was not considered. Neither is it average practice to allow for this upward pressure which, theoretically, might reach a maximum of half the hydrostatic pressure acting upward against the entire base of the dam. The most conservative practice assumes two-thirds of this hydrostatic pressure at the up-stream toe and zero at the down-stream toe. The average pressure would then be one-third of the hydrostatic pressure, and if the forces are as assumed, the resultant of this upward pressure would act at a point distant from the up-stream toe equal to one-third of the base. But as there are a large number of important dams standing safely today, whose section does not take this into account, and which would fail if the upward pressure acted according to this theory, it is evident that it is not by any means a certainty; but, hereafter, no doubt, it will be allowed for in all important structures. The danger of