below Concrete; F. Well, 3 feet deeper than the Tank; H. Automatic Valves; 7. Concrete of Limestone, Sand, and Cement; K. Cement of Sand Plaster; L. Roofing Felt and Asphalt; M. Brick laid in Asphalt ; A. Buttresses.

At the weak points in the bottom we removed the brick and roofing-felt covering, cut down a little into the concrete, and laid down, in Roman cement, two courses of 2-inch oak plank, the top course at right angle to the course beneath, covered these with pitch, relaid the brick pavement, and after giving the entire brick wall from top to bottom two coats of Roman cement plaster, we were again ready to fill up with water.

The pumps were set six feet above the floor, and the water pumped from beneath the bottom was discharged into the tank; and when the height of water reached the pumps they were disconnected, their suctions plugged, and left in position. The tank was then filled from the water works to a height of 13 feet, and left standing four days, when it appeared to be practically tight. Another 12 feet in height of water was then admitted; with this additional pressure the tank leaked some, but so little that we concluded to put it into use, which we did about September 1st of last year, or almost one year after the completion of the holder.

It would be impossible for you to have a full conception of the numerous difficulties encountered, even after you have heard this brief description of our troubles. Our most serious difficulties occurred during our last effort to repair tank, caused by the pitch that had been used in putting down the asphalt bottom. Sometimes whole days and nights would be spent in trying to get our pumps to work. A pump would be taken all apart and thoroughly cleaned, put together, set to work, and in ten minutes time would be so clogged with pitch that it would cease to take water. The packing alone for two small pumps would, for days at a time, cost us $2.50 to $3 per day. Hemp packing would be cut and torn to pieces in a few moments, and gum would both cut and dissolve.

I have told you of some of the difficulties encountered, and you can doubtless see where mistakes were made without my enumerating them. There were three mistakes made, however, that will not be apparent to you without further information.

First, the concrete was not sufficiently rammed. Do and say what we would, we could not get the sub-contractor to ram the material as the specifications required. We were compelled, owing to the sliding tendency of the earth walls, to cover the

bottom with concrete as fast as a few feet were graded. In this way the south and southwest sections, after the concrete was put on, received the travel of wagons and carts passing around the cone in removing the earth from other portions of the pit. That part of the bottom was good, compact, and water-tight. Other portions that were not well rammed would, when the plaster was removed, let the water run through like a sieve.

Another mistake I made was in letting the contractor persuade me that more of the wall ought to be grouted than was provided for in the specifications, and less of it made with slush joints. Grouting does not make as good work as slushing, though I know the custom is to grout most of the wall. The excess of water greatly weakens the tenacity of the work; and when the excess is absorbed by the brick there must of necessity be unfilled spaces left in the wall.

The third, and perhaps the most serious, mistake I made was in using Louisville cement. As this cement could be had at onethird the cost of English, Portland or Roman cement, we were tempted to use it, thinking the walls could be made heavy enough to stand, and still make a saving in the cost of the work. The samples of cement submitted with bids for doing the work were good, but that used in the wall and bottom was irregular in quality; and it is not possible to test each barrel as the work progresses, as time alone can determine the properties of a cement. In this particular, at least, I advise you not to follow my economical practice.

Discussion.

MR. GILBERT-Is Mr. McMillin familiar with the operation of the system by which his city is supplied with water?

MR. MCMILLIN—Yes, sir.

MR. GILBERT—It would seem to me as though you penetrated the same strata of porous material which gave them so much trouble in getting a water supply.

MR. MCMILLIN-Our city of Columbus is supplied with water under the Holly plan, and the water works are located on the banks of the river. There is from 10 to 20 feet of a gravel

deposit over the bottom of the river; and the water supply is obtained by tunneling through this gravel. A great many tunnels have been driven through the gravel bed. The water percotunnels, and the result is a good, clear, and The only objection to it is that it is rather limey. That gravel deposit is not identical with the one encountered by us. It is a more recent deposit-it is a river deposit, pure and simple:

lates through those cold water supply.

MR. GREENOUGH-I think Mr. McMillin is to be thanked for bringing to our notice the history of his misfortunes. As a general rule the Association's members hear only of the successes which one of their number may achieve; but it appears to me that it is sometimes equally as instructive and valuable to learn of the failures of others. Seemingly, one likes to hear of the misfortunes of other people, because then he can take some comfort in thinking that he is not alone in encountering trouble. As a supplement to the history of his misfortunes, the account of which has afforded us so much pleasure and interest, I may say that some years ago, in the building of a holder, we had in Boston a somewhat similar experience to that narrated by Mr. McMillin. We had water within three feet of the surface of the ground all the time, so that the pumps had to be kept going continually. A consequence was that when the tank was filled with water it was found to leak-the pumps went down below the wall. We pumped out the water from the tank (there was perhaps a foot in depth left around the edge of the ring), but could not then find the source of leak. It was not until all the water had been taken out that we were able to discover the existence of a little crack, under the wall, about large enough to insert your finger. I mention this to illustrate the great danger of pumping too fast while excavating for a tank. If there is any sand that can by any possibility get into the pump, you may pump out a little from under the concrete bottom; then when you put the weight of the water on it, down it goes, and you have a leak.

MR. MCMILLIN-I think that we escaped the difficulty you speak of by having a drain pipe about three feet inside the wall. This drain pipe was laid with a tight joint; and it was along the line of drain pipe that we took the sand out.

MR. LITTLEHALES—In my experience it is far more satisfactory to make an annular ring rather than remove the entire body of earth. If you leave a central core you have something left to shore up against. Even if the earth is of a shifting character, or alternation of clay and gravel, you have something to hold it there in position. My experience is that the cost of removing the core is more than compensated in the additional security thereby gained.

MR. MCMILLIN-I think that is true; and if I were going to place another gasholder along side of this I should adopt that plan. We did not anticipate any trouble. We had excavated to the same depth for a tank but two hundred feet away from the one under discussion, and had no trouble at all.

Mr. RamsdelL-I wish to say a word with regard to the matter of cement. I think what Mr. McMillin names as his third mistake was the most important one of all in respect to the construction of that tank. In 1881 we put up a new holder. Our soil at Vincennes, Indiana, is a difficult one on which to build a tank. We have no clay whatever; it is all gravel and sand. We found, if we were to sink the well to the depth originally intended, we would strike quicksand. Accordingly we stopped at a point two feet above the original plan. I figured in like manner with Mr. McMillin in regard to the cost of the cement; and I did not find that Louisville cement would be much cheaper. The Louisville brand is, of course, a great deal lower in price per barrel; but when it is remembered that the Portland barrels contain a greater weight, and that it is so much stronger, there is really very little difference in the respective amounts required for the work in hand. For instance, in the mortar for the brickwork we could use four parts of sand to one of Portland cement; while, with the Louisville sort, we could only use two parts sand to one part cement. And so also with the concrete. We have a concrete bottom built in three layers of four inches each, thus making solid depth of one foot. In that work we used a proportion of seven or eight parts broken stone to one of cement. The whole tank is tight. On the morning after we finished building it we found the water in the tank had lowered about two feet. I was very much alarmed until I discovered the lowering was due to the absorption of the water by the porous bricks. I think, in the long run-regarding