392 
PROFESSOR 0. REYNOLDS AND MR. W. H. MOORBY 
lutions per ioinute requii’ed as the ordinate in the curve of water ecpoivalents, was 
not reliable to one or two revolutions, and, therefore, unless a large difference of 
speed was indicated between the commencement and end of a tidal, this difference 
was altogether ignored, and the rise in temperature was multiplied by the constant 
corresponding to any particular load at 300 revolutions to obtain the terminal 
correction. 
The speed gauge required a negative correction of 11 at 300 revolutions, and, 
consequently, the curves give 57’6 and 54’6 as the water equivalent of the bi-ake 
when loaded with 1200 and 600 ft.-lbs., re.spectivel}’. 
By interpolation from the above values 53‘6 was obtained and used as the water 
equivalent in trials carrying a moment of 400 ft.-lbs. 
Loss of Water hy LLrajOoration and Leakage from the Discharge Pipe and Tanh .— 
(Bart I., par 37.) 
31. In order to test the general efficiency of the discharge pipe as a convever of 
the water used, it was disconnected in June, 1896, from the brake, and the circu¬ 
lating pumji was arranged to pump the water out of the tank and through the 
discharge pipe, which emptied itself again into the tank by means of the tipping 
switch. 
The stream of water was regulated so as to correspond exactly w^ith the quantities 
passed in trials carrying loads of 400, 600, and 1200 ft.-lbs. In a period of 
62 minutes it was found that in each of these cases the loss approximated very closely 
to a quarter of a pound of water when its temperature was between 90° and 100°. 
Since this loss was the same in all the trials it has not been thouo-bt necessarv to 
make a correction rectifying the heats on this account, for it would be completely 
eliminated in the differences of heat used in the calculation of the A’alues of K 
given in the tables, if the interval of temperature through which the water was 
raised in the brake was the same in corresponding light and heavy trials. 
When, however, I examined the results after the final reduction had been made, 
I found that the mean temperature of supply in the light trials was 0'7° lower than 
that in the heavy trials. 
Consequently the mean difference of heat would require a slight correction, which, 
however, is less than — 0-000002 relatively to the whole. This, being quite outside 
our limits of accuracy, has been ignored. 
The Main Lxperiments. 
32. In December, 1895, the apparatus, though not yet quite complete, was in a 
sufficiently advanced state to make it possible to commence the main K experiments. 
The observations were taken and reduced in every experiment in substantially the 
