34 Messrs. Wright and Thompson on the Determination of 
Let the E.M.F. of a given cell, taken temporarily as a 
standard, be 1 at temperature t Y (near to 15°), and let the 
E.M.F. of a second cell compared therewith be a x when the 
cell compared is at a temperature t 2 , the standard being still 
at t\. In another experiment, when the standard is at a tem- 
perature t 3 not far from t x , let the E C M.F. of the second cell 
be a 2 , this cell being at the temperature t 4 . Now let x be the 
mean coefficient of variation for 1° between i 2 and t± for the 
second cell, whilst a/ is the analogous coefficient between t l 
and t 3 for the temporary standard. Then, since the E.M.F. of 
the standard at t x is unity, its E.M.F. at t 3 is l — (t 3 — ti)x / ) 
whence the E.M.F. of the second cell at t A is a^ -jl — {t 3 — t x )x'\ . 
The E.M.F. of this second cell at t±, however, is also 
so that 
a^l-it.-t^} =a 2 {l-(t 3 -t i y\ . 
Now, from Clark's experiments and certain preliminary ob- 
servations made by ourselves, it results that x is approximately 
equal to x'; whilst if the temperatures are suitably chosen so 
that the mean of t x and t 3 is sensibly the same as the mean of 
t 2 and t 4 , it must result that the difference between x and x / 
is very small : and, finally, if t x and t 3 differ but little in com- 
parison with the difference between t 2 and t A , any errors in 
the valuation of x 1 will be but small relatively. Hence, taking 
x=x' } it results that 
oi(*4— h)~ <h{h— ti)' 
In order, then, to determine x, it is only necessary to determine 
the relative readings of two cells, first when one is at t x 
and the other at t 2 (say at 15° and 0° respectively), and 
secondly when the first is at t 3 and the second at t± (say at 
14° and 30° respectively), the temperatures being such that 
h + h approximately equals t 2 + t A ( as in the case of the sup- 
posed numbers). 
To carry out this principle two water-jacketed metal cham- 
bers were constructed, furnished with delicate thermometers 
reading to 0°'01 C, and containing respectively the sets of 
cells to be compared, the poles of the cells being connected 
with the mercury-cup arrangement applied to the electro- 
meter by means of covered wires passing through narrow 
glass tubes fixed in the double lids of the chambers, so that 
no conducting contact between the wires themselves or 
between the wires and lids &c. was possible. One of the 
water-jackets was filled with water at near 
