198 
MR. H. L. CALLENDAR ON THE PRACTICAL 
are connected by a thick copper piece. Thick copper wires pass from D and E to 
Cups 1 and 5 ; C, D, and E, are connected to mercury cups d, e, c, into which the 
galvanometer terminals may be dipped. The porcelain tube and the apparatus for 
exhausting 1 and heating it are not shown. The method of in terwin din g the wires is 
roughly indicated by the plain and dotted lines. To observe the resistance of the 
connexions, Cups 4 and 5 are short-circuited, and the resistance B3J4D is measured. 
The copper tails corresponding to Cups 3 and 4 are fused together at their extremities, 
J, where they join the platinum wire; by symmetry, therefore, this resistance repre¬ 
sents that of the connexions, which may be taken to be the same for both coils, and 
is observed on each occasion, since it varies slightly with the temperature. 
To test the insulation, the copper piece connecting 2, 3, is removed, and the resis¬ 
tance between the coils observed as a matter of precaution. The insulation was, 
however, found, in all cases, to be amply sufficient. The more nearly the wires are 
alike, the more accurate the comparison, since all sources of error affect both more 
equally. This is illustrated by the following comparison between two platinum wires, 
one of which was the standard wire, described previously, and the other was sent 
from Messrs. Johnson, Matthey, and Co., in reply to the request for the purest 
possible platinum wire. As was expected, their temperature coefficients nearly 
agreed. The ratio of their resistances was, in consequence, very nearly constant. 
Let R, R/, denote the simultaneous resistances of the wires, and r then- ratio, then we 
have, evidently, 
r_r; r/ 
E 0 R'o EoV r 
Since r is very nearly constant, the term (L — is always very small; consequently, 
V r o)\ 
if r be observed accurately, an approximate value of R will suffice to give the 
differences of the temperature coefficients with all the accuracy of which the 
resistance-boxes are capable. It is evidently unnecessary to correct the value of r for 
changes of temperature in the resistance-boxes. Errors will, however, be introduced 
by any want of uniformity in then’ temperatures, or in the material of which the 
various resistance coils are composed. Precautions were therefore taken to maintain 
the temperature of the boxes as uniform as possible. 
The correction to be applied to the observed value of the ratio r to allow for the 
resistance of the connexions is taken as (r— 1)=;, where 2 is the observed resistance of 
the connexions ; this is sufficiently approximate, since 2 is small. 
The great advantage of the method of ratio is that the observation affects both 
ocils simultaneously ; moreover, when the ratio is nearly constant, the reduction of 
observations is much more simple, and may be effected with sufficient accuracy by the 
aid of a small slide-rule, since it depends on small differences. 
The following Table is given to show the method of recording observations, and 
practically explains itself. The letters M, E, refer to the resistance-boxes used. The 
galvanometer throws are given in scale-divisions, and could be read with great accuracv 
