Oisr THE MECHANICAL EQIJIYALENT OP HEAT. 
387 
"WeigEt number. 
True weight. 
1 
25-00 
o 
25-02 
3 
25-03 
4 
25-02 
0 
25-01 
6 
24-99 
7 
25-02 
8 
25-02 
9 
25-03 
10 
25-00 
11 
25-04 
Hanger 
24-99 
i And a lead balance weight to be re- 
ferred to later, 
which weighed 13-98 
lbs. instead of 13’97 lbs. as assumed 
On the 17th of January, 1896, a set of four of these 25-lb. weights, at that time 
all supposed accurate, were used as a standard 100 lbs., by which a series of correc¬ 
tions to the 100-lb. scale of the weighing machine were obtained. These corrections 
have been used throughout the investigation, and are given below 
Reading . 
300 
400 
500 
600 
700 
800 
900 
1000 
1100 
1200 
1300 
Correction 
0-4 
• • 
• • 
-0-12 
-0-42 
-05 
-0-65 
-1-12 
-1-22 
-1-31 
-1-78 
Rider weio-hts Numbers 2 and 3 were at the same time made correct on their whole 
range. 
In June another comparison was made, and the set of four weights, Numbers 2, 8, 
9, and 10 were found to give substantially the same list of corrections as previously 
obtained. 
The complete set of weights were then again weighed on the weighing machine, 
using the list of corrections given, together with the true value of the standard 
100 lbs. The result was a verification of the list of their values already given. 
The maximum error that might possibly be produced by using the weights on the 
brake in specially arranged groups was found to be— 
In a pair of trials carrying moments of 1200 and 600 ft.-lbs. respectively, 
— 0‘037 per cent, or -f 0'043 per cent., and in a pair of trials run with moments of 
1200 and 400 ft.-lbs. respectively, — 0'025 per cent, or -f- 0 03 per cent. 
The value of the equivalent obtained from a set of six trials in which the weights 
had been specially arranged to eliminate the above possible error entirely, gave a 
result which did not differ at all from that previously obtained, and it may therefore 
3 D 2 
