ON THE MECHANICAL EQUIVALENT OF HEAT. 
403 
decided to vary the trials somewhat, in order to determine if any errors had been 
overlooked. For this purpose I made two fresh series of six trials each, the light 
trials carrying’ a moment of 400 ft.-lbs. only, none of the other conditions being altered 
in any way. 
50. The full reduction of these Series (Nos. VIII. and IX.) is shown in the two 
Tables E and F. 
As before, three trials were run on each day, but the last trial, on April 1, was not 
finished on account of an accident preventing me getting the correct weight of the 
water discharged by the brake. There are, consequently, only eleven trials in the 
tables. The radiation constant for these trials worked out to 8T6. 
The mean value of K, given by the whole eleven trials, was 778T4, which is lower 
than the two means for the separate series in Ttible F, on account of the inclusion of 
the light trial No. 45, which does not appear in Table F. 
This new value of K, viz 778T4, did not agree so closely with the former one of 
777-85 as we had hoped, and, after reducing the last two series of trials, I devoted 
all my time to the checking of the whole of the apparatus anew. 
It was a consecjuence of this stringent supervision of every separate part that the 
small errors in the 25-lb. weights, already noticed, were discovered (par. 27). 
51. Calculation showed that this error might account for the discrepancy observed, 
and so it was decided to run a fresh series of trials with the weights so arranged that 
no error could appear on their account. 
In order to have no known outstanding errors whatever, I made a small rectangular 
trough, fitted with a drain-pipe, by means of which all leakage from the stuffing-box 
was caught. 
52. A series of fifteen trials, numbered 54 to 68 inclusive, was accordingly made, 
beginning on June 29, 1896. Owing, no doubt, to the long rest which the apparatus 
had had since Easter, a number of accidents were met with which completely spoiled 
the whole series. 
The lagging of the brake was very damp when the series was begun, and, on 
account of the bursting of the various rubber-pipe connections, it did not thoroughly 
dry during the whole course of this series of trials. 
For these reasons the results are not tabulated. 
53. After remedying all the defects which had developed in the previous weeks 
running I made two fresh series of six trials each between July 7 and 10 inclusive. 
No further accidents occurred and the results were in every wmy satisfactory. 
These are shown in Tables G and H. 
The radiation constant worked out at II = 7-98. 
The mean value of K, given by the two series, was 
K = 777-85, 
which happens to be exactly the same as obtained previously from Series II., III., 
VI., and VII. 
