There are in the hotel 29 sets of soil and ventilation riser pipes extending above the roof, and each serving clusters of fixtures in vertical tiers in most or all of the 15 stories. These pipes, all of screwed wrought iron, were tested from time to time as they were extended from the bottom up, and again when completed ready for the connection of the fixtures. The average height of each stack was about 170 feet, and an idea of the number of joints, 1,800 to 3,000 for each set of two pipes, connections and fittings thus tested and retested, may be derived from the statement that nearly every one of the 29 sets of risers served 10 bath and toilet rooms arranged as indicated in Fig. 2, where, however, only the system is shown and the necessary offsets and extra pipes are not indicated. A special trap screw, not shown in the Figure, was put just beyond the vent pipe on the soil-pipe side of the bathtub trap, which was originally designed to be accessible through the overflow connections.

Two important features of the arrangement are the unvarying location of the vent-pipe branches above the overflows, so as to avoid the possibility of their acting as overflows should the waste be obstructed, and the provision of what is substantially a double ventilation of the water-closet trap, which may be effected either through branch A or connected branch I. In testing, the open ends were all closed by screwed caps and air pressure applied through the flexible hose A, Fig. 3, one end being connected by screw coupling to a point near the foot of the soil pipe which was connected to its adjacent vent pipe, or vice versa, and the other end to the discharge of an air pump P, Fig. 3, on which was set a mercurial gauge G. The commercial portable air pumps not proving of sufficiently rapid action, and being considered not adapted to this work, it was found convenient and economical to construct pumps on the spot, which was very easily and simply done, as indicated in Fig. 4. The barrel of the pump was simply a piece of 2-inch brass pipe B, which was so true and straight and of so exact a bore as manufactured as to need no finishing whatever to receive the plunger C, the rod of which D was constructed of a piece of ½-inch gas pipe, and furnished with a gas-pipe crossbar F long enough for four men to grasp. A little piece of ¾- inch iron pipe J served both for suction and delivery, and the inlet and discharge were respectively controlled by the two ordinary check valves E and K, both opening upwards. On the upstroke of the plunger valve K opened and admitted outside air to the partial vacuum formed in the lower part of barrel B. On the downstroke of the plunger this air was compressed, and closing valve K opened valve E and was discharged through it, pipe H and hose A into the closed system of pipes. When plunger C again ascends the pressure in the barrel becomes diminished and the pressure in pipe H closes valve E. Atmospheric pressure opens valve K, more air is drawn into the pump barrel and forced into the pipes, and so on. Just above check valve E there is in the pipe H a stop-cock, which is not shown in Fig. 3.

Figure 4 shows the details of construction of the plunger, etc., and like Figs. 2, 3, and 5, is not drawn to exact scale or dimensions, but is intended merely to show the arrangement, construction, and operation. Rod D passes loosely through cap I, but as the clearance did not allow the air to escape freely enough on the upstrokes, L L were cut in the rod and handle and allowed free discharge as indicated by the arrows. The apparatus was secured by a bottom flange to a 2-inch base plank, 12 inches wide by 6 feet long, on which the men stood to operate it. The entire weight was about 100 pounds, and it proved very efficient and satisfactory. Four of these pumps were made for this job at a nominal cost, as all the materials were taken from ordinary stock, and they required no repairs or alterations.

In the beginning of a test, first one and then two men were required to pump. The average time required to produce a pressure of 10 pounds was 20 minutes. The Board of Health Inspectors specified that a test would be considered satisfactory if a mercury gauge. column 20 inches high did not fall more than one half inch in 20 minutes after the pumping ceased. In the final tests the greatest leakage discovered caused a fall of only one-half of an inch in the 20-inch mercury column in 22 minutes after the pumping ceased, and in many instances there was no appreciable fall whatever. When it is considered how slight an aperture will permit the escape of air and that the pressure is uniform at all points, it will be seen how thorough this test is, and how much more positive it is than the usual water column, especially for the upper part of the system, where the hydrostatic pressure diminishes to zero.

Figure 5 shows the ordinary mercurial gauge G, Fig. 3, used to indicate the test pressures. Air pressure from the pump discharge pipe H is admitted by pipe M to the cast-iron chamber N, in which is an open iron cup O filled with mercury. A plain glass tube Q, open at both ends, dips beneath the surface of the mercury nearly to the bottom of the cistern, and is protected by a brass tube or case R, which is capped at the upper end and screwed to the iron case containing the cup O, the packing S making an airtight joint. The pressure is in the pipe system being communicated by pipe M to chamber N, acts downwards on the surface of the mercury in the cup and forces it up in tube Q, where its height is observed through a slot T T, in case R, and can be marked by the index U, which is a close-fitting sleeve, which can be set at any position to measure variations from. About 100 tests were made by Foreman John Hanson, who devised the pump, and was in charge of the work.