348 Mr. J. H. J. Poole on a Possible Connexion between 



the iron surface. There is no doubt that even with the 

 large field employed (about 1000 volts per cm.) some ions 

 would succeed in moving back against the field, as at air 

 pressure the saturation value of the current is not obtained 

 before ionization by collision sets in (see Hughes. ' Photo- 

 Electricity,' p. 55, and Sir J. J. Thomson's ' Conduction of 

 Electricity through Gases,' p. 268, 2nd edition). We would 

 expect, however, that the number of such ions would not be 

 large owing to the mean free path of an electron being so 

 much greater than that of an ion. On these grounds, we 

 are accordingly led to conclude that the expected change 

 in the ionization current on magnetization would be the 

 resultant of three separate effects. These are : — 



A. An increase in the total number of electrons emitted 



directly, due to the change of orientation of the 

 atoms. 



B. An increase in the number of electrons which escape 



after refraction in the metal, due to a change in the 

 direction. 



0. An increase in the ions which return to the iron 

 surface, due to a change in the direction of emission. 



What Table II. really shows is that this resultant is not 

 large, and as B and C would probably not be large effects, 

 we are, I think, justified in concluding that their difference 

 conld not balance A, unless it was also a small effect. The 

 probability that all three effects must be small is further 

 increased by the figures given in Table II., which show 

 that their resultant is again zero when the field is parallel 

 to the iron surface, a result we wou'id hardly expect if 

 the zero resultant in the first case had been produced by 

 a large difference between B and C balancing out a 

 large effect due to A. It would, accordingly, appear likely 

 that we are justified in concluding, even when we take the 

 effects of possible refraction of electrons in the metal and 

 ionic diffusion into account, that the chance of an atom 

 emitting an electron is not largely affected by its magnetic 

 orientation relative to the incident light. 



In addition to testing an iron surface, illuminated nor- 

 mally, the effect of oblique illumination was tried. N"o 

 effect was, however, detected. Some experiments were also 

 conducted with a bismuth disk, fitted on the end of the 

 pole piece, to see if any change could be detected in diamag- 

 netic substances. The properties of a magnetite crystal and 



