ON THE MECHANICAL EQUIVALENT OF HEAT. 
399 
packing had been removed from the gland on the cock spindle, while a means of 
catching the whole of the leakage, and at the same time preventing its evaporation, 
had been provided (par. 14). The whole of the leakage was credited with the 
temperature of the water in the discharge pipe, and was weighed with the main 
stream of water which had been caught in the tank. 
Series VII. 
42. These trials were made under similar conditions to those in Series VI. In the 
two last trials, however, viz.. Nos. 39 and 42, some leakage was observed and 
caught from the stuffing-box. 
An approximate estimation of the loss of heat due to this leakage is given in 
Table B, and has been included in the heats givmn in Table A. 
Determination of the Loss of Heat hy Conduction along the Shaft. 
43. In the trials enumerated in Table A, the varying values of the temperature 
gradient, existing in the shaft leaving the brake, might evidently be a cause of com¬ 
paratively large losses of heat which were not eliminated in the differences of heat, 
so far assumed to be equal to the corresponding differences of work. 
It therefore became important to determine, at least approximately, what was the 
loss of heat by conduction along the shaft in each trial. 
I have already said that the temperature of the shaft in the main bearing was 
assumed to be the same as that of the lower brass, while the temperature on leaving 
the brake was similarly taken as that of the stuffing-box cover. 
Unfortunately, before trial No. 21, I had made no record of the temperature of 
the lower brass. 
It was, however, found that in trials Nos. 21 to 41 the mean temperature of the 
lower brass exceeded that of the upper brass by about 7° Fahr. 
Consequently, in Column 6, in the parts of Table A, wdiere no observations had 
been taken, an estimation of the difference of temperature between the stuffing-box 
and the lower brass was made by subtracting seven from the difference occurring in 
Column 5. In this manner the differences entered in brackets were obtained for 
trials Nos. 10 to 20. 
It appears that we have, therefore, 10 determinations, viz., V., VI., VII., VIII., 
IX., X., XI., XII., XIII., and XVIII., in which the differences of heat generated 
require a positive correction on account of the unbalanced conduction along the shaft, 
and four determinations, viz.. Nos. XIV., XV., XVI., and XVII., in which those 
differences require a negative coiTection. 
Assuming, as is very nearly the case, that the losses of heat by radiation are 
eliminated in the differences of the heats, it follows that by taking C = loss of heat 
