The boiler had one defect, seen in the front view, cross-section. A straight passage 2 inches wide was given to the gases between each pair of tubes. The boilers having all had a preliminary trial during the first week, I observed the vapor arising from the exposed surface of the water in the tank, and that this unmeasured loss of heat differed considerably in the different boilers, and was enormously greatest on the trial of the Allen boiler. I said nothing, but went down early on next Monday morning and on my way bought a common tin cup about 3 inches deep and 4 inches in diameter, and secured it in one corner of the tank, immersed to a quarter of an inch below its rim, and filled even full of water. This was completed before the arrival of the Committee, and was at once approved by them. I made it my business every day to note the fall of the water level by evaporation from this cup. On the trial of the Allen boiler only the water in the cup was all evaporated, and I had to fill it again. The temperature of the water in the cup was always 8 degrees below that of the surrounding water. It was thus obvious that the evaporation from the tank was greater than the fall of the level in the cup would indicate. The Committee considered that this should be increased as the tension of the vapors. The result was that the report contained the following item: Units of heat carried away by evaporation at the surface of the tank: Root boiler 721,390.8 units Allen boiler 1,178,404.5 „ Phleger boiler 378,371 „ Lowe boiler 692,055 „ Blanchard boiler 268,707 „