Mr Campbell, Discontinuities in Light Emission. 325 



In almost all series there was, in addition to the fluctuations, 

 a slow drift of the spot in one direction or the other. To allow 

 for this drift, it was assumed to be linear so that 



ei=u + vt (13). 



It is easy to see that, if we treat the n observations made as 

 readings from which the constants u and v are to be determined 

 by a least square solution, 6t'^ is the mean of the square of the 

 residuals, so that 



W' = te^ - (?^ _ -^4%; i^ 2^, - s (< . ^4'-(i4). 



n n{n^— 1) ( 2 J 



Any series in which there appeared to be any suspicion of a 

 change in the drift was rejected. 



The arithmetical work in finding 0^'^ Avas heavy, but with 

 practice and the aid of a calculating machine a single series could 

 be solved in 20 minutes. 



§ 18. At the beginning of the work a great deal of time was 

 wasted because a complete theory of the experiment was not 

 known. But once the influence of the various instrumental con- 

 stants upon the fluctuations was discovered, all the experimental 

 difficulties which had been anticipated were quickly overcome. 

 They were such as attend all work upon electrostatic currents and 

 no further reference need be made to them. 



Blank experiments, made without the action of the light at 

 all, showed that, when no light acted, there were no fluctuations 

 whatever. For two hours or more the spot on the scale would 

 keep perfectly steady or move constantly with a very slow drift in 

 one direction. It is certain, therefore, that no trouble need be 

 expected from such sources as changes in the potential of the 

 battery used to send the current through the cells. 



It was also important to show that the high resistance did not 

 introduce disturbing fluctuations, such as would certainly have 

 been introduced by a Bronson air resistance. For this purpose 

 two similar high resistances were used and opposite potentials 

 applied to their ends, so that the currents flowing through them 

 were opposite in direction and equal in magnitude to each other 

 and to the photo-electric current employed. No fluctuations what- 

 ever could be detected, though the steady drift of the spot was 

 greater than in the former case, when practically no current was 

 flowing through the resistance. 



§ 19. But there were two outstanding difficulties. The first 

 of these, the difficulty of determining the capacity, has been noted 

 already. The second was far more serious and has prevented, up 



