as a Standard of Electromotive Force. 137 



is very important, and goes very far towards helping it to 

 sustain its position as a standard cell against its rival the 

 Clark cell. 



The various coefficients of temperature-change which have 

 been found for Clark cells indicate that it is unsafe to rely 

 upon any cell taken at random, whose history is unknown, 

 having a coefficient of variation accurately the same as that 

 of the cells whose coefficients have been actually determined. 



It would be a great advantage if instrument-makers w r ould 

 construct these cells in the form suggested by Lord Eayleigh 

 — enclosed in a test-tube and having gutta-percha-covered 

 leading-in wires passing through a rubber stopper. The co- 

 efficients of variation would then be easily obtained for any 

 cell, just as those of standard resistance-coils are ascertained 

 and marked on each cell with the range of temperature over 

 which it is applicable. 



As a check on the foregoing experiments on the Daniel], the 

 coefficient of variation of the standard Clark cell used was 

 obtained. Two cells were taken, exactly equal at 18° C, and 

 one of them was immersed for 48 hours in melting ice. The 

 difference on the potentiometer-wdre was then obtained between 

 this cold cell and its fellow kept in w r ater at 18° C. The 

 potentiometer-readings were as follows : — 



i. ii. iii. 



B, cellatO°C. . . 961 971 974'4 



A, cell at 18° C. . . 947 957 960 



14 14 14-4 



From the experiment 1 the reading for the warm cell 

 would have been 947 + 2 / = 949£ a t 15° C. And hence the 

 coefficient of variation is f§ in 949^, or *0819 per cent, in 

 the neighbourhood of 15° C. Lord Rayleigh gives *082 

 as the value for his cells. It is obvious that Clark's cells 

 cannot be used for standardizing galvanometers without a 

 careful determination of their coefficients of variation. 



The normal Daniell cell has a certain advantage in that when 

 null methods for determining electromotive force are employed, 

 its value is independent of temperature throughout a consi- 

 derable range. 



By the employment of the above two instruments— a poten- 

 tiometer as described and a standard normal Daniell carefully 

 constructed wdth all the precautions named — it is very easy to 

 make very accurate measurements of strengths of currents of 

 large magnitude. If a resistance is formed of such character 

 that the current to be measured can be passed through it 



