The superheating of the steam for our own engine was not affected by the connection of our steam-pipe with those of the other engines. The explanation of this phenomenon seemed to be that as our boiler generated far more steam than our own engines required, the current was always _from_ our pipe into the connected pipes. [Illustration: Diagram from Allen Engine, back end of cylinder, at Fair of American Institute, 1870.] I was here first made alive to the enormous waste of steam in the feed-pumps, a separate one for every boiler, including our own. In these the steam has to follow full stroke, at a pressure sufficient, on the larger area of the steam piston, to overcome the pressure in the boiler. Moreover, the extreme heat interval between the temperatures of the entering and the exhaust steam and the slow motion, permitting the walls of cylinder, heads and piston to be cooled very deeply by the exhaust, produces the condensation of probably from five to ten times as much steam as is usefully employed, differing according to the rate of piston motion. I began to rather admire the practice of the English, who knew nothing about boiler feed-pumps, except those on the engine, and I certainly wonder that the genius did not arise long before he did, who first thought of exhausting the feed-pump into the feed-water under atmospheric pressure only, so returning to the boiler all the heat received in the pump that is not converted into the work of overcoming the boiler pressure and the atmospheric resistance or lost in external radiation. The above diagram represents the performance of this engine in its regular work. It shows distinctly the compression curve, the points of cut-off and release, and the back pressure required to expel the exhaust. It will be seen that the expansion fell to 5 pounds below the atmosphere. I have added to it a line representing the waste room in ports and clearance, and the theoretical expansion curve plotted according to the law of Mariotte, showing the expansion terminating 2.5 pounds above this curve, from the re-evaporation already noted and the heat abandoned by the steam as the pressure fell.