[ 482 ] 



XL VII. On the It elation between Illumination and Electrical 

 Conductivity in Selenium. By A. 0. Rankine, D.Sc, 

 Professor of Physics in the Imperial College of Science and 

 Technology* . 



IT appears to have been generally accepted by those who 

 have worked experimentally with selenium that the 

 ultimate increase of conductivity due to illumination is 

 proportional to the square root of" the incident light intensity. 

 Fournier d'Albef quotes the work of Rosse, Adams, Berndt, 

 and Minchin in support of this law, and his own experiments 

 have led him to the same conclusion. The present author, 

 as a result of the experiments about to be described, has 

 found that for several selenium cells made by Fournier d'Albe 

 himself the ultimate increase of conductivity is much more 

 nearly proportional to the fourth root of the light intensity. 

 This law has been found to hold over a wide range of inten- 

 sities, and, without exception, for all the cells tested. 



The discrepancy between these results and those by earlier 

 observers may possibly be accounted for by differences in 

 interpretation of what the increased conductivity actually is. 

 Fournier d'Albe, for example, in the paper already alluded to, 

 defines it as being measured by the " recovery " after the ces- 

 sation of illumination — i. e., the ultimate reduction of current 

 in the selenium cell on constant voltage. But he does not mea- 

 sure this recovery directly, probably because of the long time 

 it requires to become complete. Instead, the procedure is to 

 take early ordinatesof the recovery-curve (current-time), and 

 to deduce what the total recovery would be on the assumption 

 that the curve is hyperbolic. JSTo doubt this method results 

 in a saving of time, and for many purposes it may be suffi- 

 ciently accurate. But it suffers from the defect of being 

 based on an assumption which is by no means exactly true. 

 The recovery-curve is not a true hyperbola, and it is therefore 

 unjustifiable to estimate the true total recovery in the way 

 described. Nor is it necessary, even from the point of view 

 of time-saving. The experiments of the present author 

 indicate that, in spite of the reputation selenium cells com- 

 monly have of being subject to unaccountable variations, 

 they are very reliable^ within certain definite limits, and the 



* Communicated by the Author. 

 t Proc. Eoy. Soc. A. vol. Ixxxix. p. 75 (1913). 



+ This applies, at any rate, to the particular type of cell used in these 

 experiments. 



