38G 
PROFESSOR 0. REYNOLDS AND MR. W. H. MOORBT 
in the steam generated in the copper boiler previously alluded to. Similar procedure 
gave iu this case a mean rise of 0'0066° per inch rise of pressure. 
After applying corrections (to be dealt with later—par. 62), rectifjdng the ther¬ 
mometric indications on this account, I think that no error of greater magnitude than 
0'01° can have existed in the calculated mean rise of temperature in any trial. 
On 180° this gives accuracy of 1 part in 18,000. 
26. In addition to the thermometers just dealt with, three others were used, on 
the readings of which depended the additive corrections to the heat already referred 
to. One of these indicated the atmospheric temperature, while two others were 
placed one on the stuffing-box and the other on the shaft bearing. 
On the differences of heat which were used as the divisors in the determination of 
the equivalent from each pair of trials, these corrections all became extremely small 
cjuantities, and therefore it was of no im23ortance that small errors should exist in 
these thermometers. Their scales were therefore never calibrated. Still another 
thermometer was used to determine the temperature of the stream of water entering 
the tank. As it was only necessary to keep this temperature in each jiair of trials 
at the same level, errors in this thermometer were neo-lio-ible. 
O O 
Weighing Machine and 25-Ib. Weights used on the Brahe. —(Part L, par. 40.) 
27. The absolute value of the unit used in the graduation of the lever of the 
weighing machine was a matter of indifference, but it was of vital importance that 
the same unit should be used for the weighing machine and for the 25db. weio-hts 
used on the brake. 
A set of iron weights were, however, sent down to the Manchester Town Hall, and 
there compared with the Board of Trade standards. 
The comparison of the 25-lb. weights with our standard 25 lbs. was one of the 
first things undertaken in the course of the investigation. This was done by first 
balancing the standard placed on the jDlatform of a small weighing machine in the 
laboratory b}^ adjustment of the rider weights on the lever of the machine. The 
standard was then removed, and one of the 25-lb. weights substituted, a balance 
being made by adding to or drilling out some of the lead inserted in the weight. 
This adjustment was accepted as jDerfectly satisfactory till towards the close of the 
experiments, when a small difference in the value of the equivalent as derived from 
trials in which different numbers of the weights were used, seemed to suggest an 
error in the weights themselves. 
Accordingly, on the 9th June, 1896, I again compared the weights with the 
standard on a temporary balance, consisting of a simple lever with three knife-edges 
in a straight line, with the following result :— 
