HOW THE TESTS WERE CONDUCTED. 21 



sharpening of the metal point requires much the less atten- 

 tion. 



The driving mechanism for the work referred to here has in 

 110 case been any form of reducing-wheel. 



GENERAL METHOD OF CARRYING ON THE 

 FEED-WATER TEST. 



The testing apparatus being in readiness, and the engine 

 working with the desired load, the height of water in the 

 gauge glasses is observed, the time taken, and the position of 

 the water in the reservoir of the weighing apparatus observed. 

 Thereafter all the water fed is weighed. At the expiration of 

 the time determined upon, the water in the gauge glasses and 

 in the lower reservoir is brought to the starting-point, and the 

 exact time observed. During the progress of the test indicator 

 diagrams are taken every thirty minutes, and sometimes every 

 twenty minutes, and at the same time the gauges are observed 

 and the number of revolutions per minute counted. If the 

 steam is superheated, the temperature of the steam is observed ; 

 and if calorimeter tests are made, these are either continuous 

 or made at convenient intervals. Where special accuracy is 

 required, the atmospheric pressure is determined by observa- 

 tion of a barometer at some time during the progress of the 

 test. For this, it is sufficient for all practical purposes to rely 

 upon the record of the United States signal service at the 

 nearest station. When the test is made in a factory running 

 ten hours per day, say five hours in the forenoon and five hours 

 in the afternoon, the record in some instances embraces the 

 whole period from the time the engine starts until the time of 

 stopping. In that case the initial and final readings of the 

 water glasses are taken just before the engine starts, and just 

 after it stops. The duration is taken from the time the engine 

 attains its working speed till the time the throttle valve is 

 closed ; and no further account is taken of the power devel- 

 oped while the engine is reaching its speed after first 

 starting. In that case, the first set of diagrams is taken not 

 less than five minutes after the load is put on ; and the as- 



