562 Mr. A. Wood on Spontaneous Ionization of 



charging-rod was then brought into contact with the plate,, 

 and the potentiometer adjusted until the leaf was brought 

 back to the position noted. The potentiometer-reading then 

 gave the difference between the initial and final potentials of 

 the plate. This method assumed of course that the potential 

 of the cells remained constant in the interval, an assumption 

 justified by the fact that it was found to alter only by a very 

 small fraction of a volt between Monday morning, when they 

 were first used, and Saturday morning, when they were 

 recharged. The first experiments were performed in a room 

 on the top floor of the building. They were inconclusive, but 

 seemed to indicate a greater screening action with the lead 

 above than with it below. The effect was so small that 

 although a large number of experiments were performed — 

 fifteen — it was thought that it might be due to errors of experi- 

 ment, or even, if real, might be traceable to local conditions. 

 The experiments were therefore repeated in the basement, 

 In this case the effect observed was of the same order, but 

 opposite in direction, the mean of a large number of experi- 

 ments indicating a greater screening action with the screen 

 below. The failure of those experiments to give definite 

 results is not very surprising, when we reflect that the total 

 reduction of ionization due to a plane screen — assuming the 

 radiation to come equally from all directions — is only about 

 l) or 7 per cent. To attain an accuracy sufficient to detect a 

 variation in so small a quantity is extremely difficult in work 

 of the kind. The great obstacle in the way of any attempt 

 at accuracy in those experiments, and in all experiments of 

 the kind, is that the ionization in a vessel is subject to well- 

 marked variations, for which it has hitherto been found 

 impossible to assign a cause. The leak in a closed vessel 

 varies in an irregular manner not only from day to day, but even 

 from hour to hour. During the course of the present investiga- 

 tion, attempts were made to connect those variations with 

 variations in meteorological conditions, such as pressure *, 

 temperature, clearness of sky, and the like, but without 

 success. The changes sometimes amounted to as much as 

 10 per cent.; and were frequently found to accompany sudden 

 changes of weather. For instance, in several cases when a 

 dull clay suddenly became fine, an increase of the leak was 

 noticed, but this rule w r as not universal. That this particular 

 effect may have been largely, if not wholly, due to a tempera- 

 ture effect on the insulation, will be seen from experiments 

 to be subsequently described. 



* Elster and Geitel's experiments have led them to connect the 

 variations with variations of 'pressure. 



