404 
PROFESSOR O. REYNOLDS AND MR. W. H. MOORBY 
54. This last lot of trials 
(778-14-777-85) 
afforded no explanation of the small difference 
~ 0'3 ft.-lb. nearly, 
which occurred between the results give by the 1200-600 ft.-lbs. determination and 
the 1200-400 ft.-lbs. determination respectively. 
The difference, of course, may be due to terminal errors, which, I think, have been 
mainly resjjonsible throughout for the small discrepancies found to occur between 
individual determinations. It is more likely, however, that the small quantity of 
water dealt with in the 400 ft.-lbs. trials, and the consequent greater effect of the 
oscillations of the brake on the mean moment, may have introduced some error into 
these lightly-loaded trials. Further, some slight bias may have been given to the 
Series, Nos. VIIT. and IX., by the long rest caused by the Easter Vacation, between 
trials 47 and 48. 
55. In the annexed table I give the mean value of the work done and of the heat 
generated in the heavy and light jacketed trials respectively, against which no 
known sensible error can be placed. 
Ti’ials Numbers. 
Mean woi’k 
per trial. 
Mean heat 
per trial. 
Heavy trials : 
(13, 17, 18, 19, 20, 35, 36, 37, 38, 39, 
134,337,403 
172,685 
46, 47, 48, 49, 50, 72, 73, 74, 75, 76, 
and 77) 
Light trials : 
(14, 15, 16, 21, 22, 23, 33, 34, 40, 41, 
61,355,503 
78,867 
42, 43, 44, 45, 51, 52, 53, 69, 70, 71, 
: 78, 79 and 80) 
DiS'erences .... 
72,981,900 
93,818 
and, dividing the mean difference of work by the mean difference of heat we have 
K = 777-91. 
This mean value of K deduced from the experiments requires correcting on a few 
counts, which are due to the method of working. These will be dealt with later. 
56. The table given below illustrates the almost perfect manner in which losses of 
heat were eliminated on the mean result, by the method adopted throughout the 
investigation of always working on the differences of the quantities of work done 
and heat generated in a pair of trials. 
