208 Prof. Millikan and Mr. Winchester on Influence of 



negatively charged air-molecules on the outside, and on the 

 side o£ the metal of the positive induced " images r of the 

 negative charges on these air-molecules. This double-layer 

 would, presumably, according to this view, be superposed 

 upon the double-layer due to the contact electromotive force 

 between oxygen and the metal. In order to account for the 

 recovery phenomena, it would be necessary to assume that this 

 double-layer, which is due to absorption of electrons by the 

 air, disappears in the course of a few hours, provided new 

 electrons are not being continually discharged. If this is 

 the complete explanation of fatigue effects, they should 

 obviously not be observable at all in a vacuum. 



Laden burg, however, found strong fatigue effects in a high 

 vacuum in the case of all the metals except aluminium, the 

 discharge rate sometimes falling off to less than one-half its 

 initial value. He refers these effects to the roughening of 

 the surface caused by the incidence of the light upon it. 

 Presumably, therefore, his fatigued surfaces did not recover, 

 although he makes no mention of any experiments designed 

 to test this point. On account of these fatigue effects he 

 was unable to obtain any constancy in the rate of discharge 

 from any particular kind of surface except aluminium. It 

 must be borne in mind, however, that his surfaces were 

 polished with oil and leather, and are therefore quite likely 

 to have possessed surface-films having properties quite 

 different from the properties of the metal itself. 



At any rate, with surfaces prepared in the manner described 

 above we found that all the metals experimented upon gave 

 remarkably definite and constant discharge rates. Tables II. 

 and III. show how T accurately results could be duplicated at 

 different times. The tables show no evidence at all of fatigue 

 effects of any kind. In order, however, to obtain this 

 constancy, it was our practice to allow about five-minute 

 intervals between successive readings on a given metal. If a 

 particular disk was illuminated continuously for three or four 

 minutes and a reading taken immediately thereafter, or if a 

 large number of readings were taken in rapid succession, the 

 results became irregular and the values of the discharge rates 

 were in general smaller than those found in Table II. This 

 was at first thought to indicate fatigue effects which recovered 

 rapidly, but it now seems more probable that these results 

 were due to a charging of the inner wall of the glass tube 

 because of the passage of some of the discharged electrons 

 throuoli the meshes of the wire gauze. Such a charge would 

 probably take some minutes to leak off to earth in spite of 

 the fact that the wire gauze was in quite intimate contact 



