of you in Germany, as he hopes to do, it will be his endeavor to extend to you the same kindness which you have shown to him here. He thanks you for your kindness. (Applause.)

THE PRESIDENT-We are all very glad to meet Mr. Buhe, and he is very welcome to take part with us in our proceedings. The Association took a recess until 2 o'clock P. M.

FIRST DAY-AFTERNOON SESSION.

The Association reconvened at 2 P. M., and the President introduced Mr. B. E. Chollar, of St. Louis, Mo., who read a paper on

A FOUNDATION THAT FAILED, AND THE LESSON IT TAUGHT.

The literature of engineering, on the stability of natural earth for the support of foundations, is neither extensive nor explicit. Beyond the statement that such-and-such kinds of earth ought reasonably be able to sustain such-and-such stresses per foot, under such-and-such conditions of moisture, etc., and an admonition to the engineer to make a very careful examination of a subfoundation before making a guess at what pressure it should safely carry, not much more is to be found in the books.

The existence of any important relationship between foundation area and linear dimensions seems to have received little thought; in other words, a foot is a foot, and a ton is a ton, and a foot should support more or less tons, according to the nature of the earth, but independent of the size or shape of the structure to be erected.

If we take a bar of iron, I inch square and 4 feet long, and stand it upright on the ground, the pressure at the surface will be at the rate of 1 ton per square foot. There will be an angle of repose, or rupturing angle, below. If we take this angle at 261⁄2°, and neglect friction, the diameter of the compressed section will increase by I unit for each unit of depth, and the stress at 1 inch down will be one-quarter of that at the surface, or only 500 pounds per foot, while at the depth of 1 foot the stress will be at the rate of only 14 pounds per foot.

If we start out with I ton upon a square foot of surface, the stress at 1 inch in depth will be at the rate of 1,800 pounds per foot, or at I foot, 500 pounds, while we will have to go down 12 feet, instead of 12 inches, to reduce the stress to 14 pounds per foot, as in the preceding case.

If, now, we load a large area, say 100 feet square, with weight of I ton, per square foot, under the same conditions of friction and angle of repose, we shall have quite a different condition of affairs.

In this case the total load at the surface will be 10,000 tons, but at I foot below it will be 9,800, or within 2 per cent. of the load at the surface. At the depth of 5 feet the area supporting the pressure will be increased only by 1,000 square feet, and the rate of stress reduced only 10 per cent., or 1,800 pounds per foot.

A failure on my part to comprehend the enormous differences in the rates of decrease of underground stresses, under various conditions of shape, independent of weight, was the cause of unequal settlement and considerable damage to a gasholder tank at St. Louis, in 1896.

The foundation was 128 feet in diameter, and the tank to be supported 31 feet deep, making a total surface pressure of a little more than I ton per foot. As the ground had been previously tested at a pressure of 4 tons per foot, and had maintained that pressure for a period of two months without showing signs of weakness, it was concluded, therefore, that if it would stand 4 tons per foot, I ton would be an easy load; but such, however, did not prove to be true, for before the tank was three-quarters filled with water, one side of the foundation had settled 6 inches, and continued to settle until it reached its limit, about 81⁄2 inches. By this time the top of the column at the point of greatest settlement was nearly 2 feet out of plumb. Fortunately, the distortion of the tank and guide frame resulted in no permanent injury, for it had been specified that the angle iron at the bottom of the tank should project outward instead of inward, and that the bases of the columns should rest directly on the angle iron. The lower ends of the columns, being thus securely fastened, could not move outwardly, and, resting on the bottom of the tank itself, would settle equally with it. Hence there could be no shearing strain on the rivets fastening the columns to the tank.

After the tank had finished settling, the water was taken out, the tank was raised by means of jacks, and the spaces between the foundation and the bottom of the tank filled up with sand. A row of sheet piling was then driven entirely around the foundation, and as near to it as possible. Some slight adjustments to the guide frame were made, after which the holder was put into service, and it has continued in action until the present time. The foundation is now only about 1 inch out of level.

A remarkable similarity will be seen to exist between the stresses under this foundation, and those affected by other forms of radiant energy, such as light and heat.

If we could have stood our gasholder on a point, the stress below would have decreased in proportion to the square of the distance down, but as it rested, not on a point, but upon 15,000 square feet of area, hence, at 10 feet below upon 17,000 feet, at 20 feet the supporting area was less than 20,000 feet, so we see that it makes a great difference whether we calculate our energy as coming from what may be called a point or from a surface.

Light emitted from a point may be considered to decrease in proportion to the square of the distance, but diffused light, or that from a surface, never. A simple experiment will prove this. One has only to look at a lighted surface through a narrow tube, say I inch in diameter and 2 feet long. The small area seen through the tube will appear much brighter than the rest. The reason is that the eye that will not look through the tube gets only one ray from each point of the surface, or an extremely small proportion of the light emitted from the whole, the other rays being sent in other directions, while the eye that looks through the tube gets a large proportion of rays nearly parallel, which do not decrease rapidly with the distance.

The lesson is that big foundations push down harder than small ones, and that lights which have appreciable size decrease at all conceivable rates, except in proportion to the square of the dis

tance.

DISCUSSION.

THE PRESIDENT-This paper is upon an engineering subject in which we are all interested. Has any member any different opinions to express, or has any one had a different experience to that of Mr. Chollar?

MR. CHOLLAR-I would add, as to this particular holder, that after we drove the sheet piling as close to the foundation as the railing at the top would permit, and after we had filled the tank with water, the outward pressure below was so great that you could put your finger between the top of the piles and the earth next to the foundation. This showed there was a great deal of strain, even after the sheet piling was put down. It caused no injury, however. The point that those building gasholders had better look out for is the liability to get a foundation, which is softer on one side than on the other.

MR. MILLER Did you notice any tendency of the ground to bulge up outside of the foundation? Or did it all seem to be one way?

MR. CHOLLAR-There was just the downward thrust.

MR. MILLER-Was that down at the river front, at Station A? MR. CHOLLAR-Yes.

MR. CARPENTER-As to one point Mr. Chollar made-speaking of the angle iron being turned out.

MR. CHOLLAR-I mean the angle iron at the bottom.

MR. CARPENTER-The column supports came down and rested on that angle iron?

MR. CHOLLAR-Yes.

MR. CARPENTER-From your paper I got the impression that your idea was that this fact saved the tank, and helped to relieve the strain. I cannot quite see how this could be.

MR. CHOLLAR-The bottom extended out, and the angle iron also extends outward instead of inward, as is the usual practice. Then the columns come down and rest on the angle irons. Now, if the bottom of the tank should settle, the columns would go right down with it, and it would be impossible to have any shearing strain where the columns were attached to the tank, for all would go together.

MR. CARPENTER-The greatest strain would naturally come on the diameter perpendicular to the radius of settlement.

MR. CHOLLAR--What I refer to is only where the columns were joined to the tank there could not be any shearing strain. We had already estimated upon a plan which provided for fastening the columns directly to the tank, and without resting on the bottom.

MR. CARPENTER-It would be a simple matter to put in the necessary rivets to carry the shear.

MR. CHOLLAR-In the one case the whole column would rest on the rivets to sustain the weight; in the other case on the bottom. MR. CARPENTER-But the whole weight of the column would be comparatively small.

MR. CHOLLAR-In the one case you would have that weight in shear on the rivets, and in the other case you would not. The tank would have gone over if the columns had not rested on the shell instead of the bottom.

MR. CARPENTER-I do not see how that affected it at the point of greatest strain due to the unequal settling. I understand that it settled on one side more than on the other?

MR. CHOLLAR-Yes; about 30° of the circumference.

MR. CARPENTER-Naturally that would tend to start the tank out of round, and the pressure of the water would tend to balance this strain.

MR. CHOLLAR—The pressure of the water after the start would tend to throw it further out of round.

MR. CARPENTER-If the tank were oblong, the tendency of the water pressure would be to make it round.

MR. CHOLLAR-Not while settling on one side. Then the pressure would tend to make it go further out of round.

MR. CARPENTER-If you take a sphere how would it be?

MR. CHOLLAR—If you take a soft felt hat, fill it with water, and then tilt it to one side, you will see whether it will come back round. The New York gasholder did not get round. The strain came on the diametrical section opposite the settlement.

MR. CARPENTER-The strain would come on the diametrical section opposite the settlement; whichever the case, the strain would go to the same point, and would throw the maximum load on that section, which would be enormously greater than the other strain.

MR. CHOLLAR-I mention that because the strain there was maintained by the diagonal braces throughout the frame.

MR. CARPENTER-It seems to me the strain caused by that unequal settlement would come on the diametrical section at right angles to that settlement, and would be far greater and more dangerous than any local shearing strain at points of attachment of the column supports.

MR. CHOLLAR-I do not doubt that; but if the columns had been riveted to the tank, and supported by the shell, and had not rested on the bottom, they would have certainly been sheared on the rivets; but in this case it was not so, for the columns rested on the bottom. If the tank had a tendency to tilt over, and the columns were fastened to the tank, the resting point of the columns would be in motion, and the whole thing in an unstable equilibrium.

MR. CARPENTER─But so long as the column supports terminate near the bottom of the tank shell, I do not think it would make much difference.

MR. CHOLLAR-But it is not good practice. The best way is to put it on the bottom, and not on a separate foundation.

On motion of Mr. Wheeler, a vote of thanks was passed to Mr. Chollar for his interesting paper.