from Ordinary Materials. 535 



could not be prevented nor their abolition secured by any of 

 the numerous measures that were tried. They were indu- 

 bitably changes in the ionization due to the walls, for they 

 affected the leak at small volumes in a greater ratio than the 

 leak at large volumes, but their cause has remained a 

 mystery. 



It might be thought that jumps of this kind would render 

 impossible the plotting of such curves as have been drawn ; 

 but fortunately they were well marked — there was never any 

 doubt when one had occurred. By patiently continuing 

 observations, a period between two jumps could always be 

 found sufficiently long to enable a complete series of obser- 

 vations to be made throughout the whole range of the sliding 



o ft o 



side. But if the series had been made in a different interval, 

 the constants of the curve — but not, so far as can be ascer- 

 tained, its form — would have been markedly different. 



Dr. Jaffe has recently published in this Magazine * a paper 

 in which he has given some account of similar changes which 

 he observed. Our experiments were in progress at the same 

 time and in the same laboratory ; and it is worthy of note 

 that while changes of type (1) occurred in general with both 

 of us at the same time in the same sense, no correlation could 

 be established between changes of type (2). 



The vessels which I used were not perfectly air-tight, but 

 it seems improbable that this fact is the cause of the changes, 

 in view of the impossibility of restoring the leak to its former 

 value by blowing in fresh air from outside the room. 



§ 6. We svill now consider the meaning of the curves of 

 fig. 1, Plate VII. The difference between the curved portion 

 near the origin and the straight line into which it develops sug- 

 gests at once that two different agents are at work ; and we 

 can easily see that the form of the curve? can be explained on the 

 assumption that the ionization is caused by an easily absorb- 

 able radiation proceeding from the walls acting in conjunction 

 with a much more penetrating radiation. For this penetrating 

 radiation, not being perceptibly decreased in intensity by 

 passing through the air of the vessel, will cause ionization 

 proportional to the volume ; the part of the curve due to this 

 cause will be a straight line inclined to the axis. On the 

 other hand, the ionization caused by easily absorbable radia- 

 tion coming from the wall> will increase with the distance 

 apart of the sliding sides, only so long as a layer of air of a 

 thickness equal to that distance is insufficient to absorb all 

 the radiation. When all the radiation is absorbed in passing 

 from one side of the box to the other, an increase in the 

 * Jaffe, Phil. Mag. Oct. 1904. 



