Air in Closed Vessels and its Causes. 565 



the same position. The potentiometer reading then gave the 

 final potential of the electrode, and as it had initially been at 

 zero this was a measure of the ionization in the vessel. This 

 method of reading the eleetroscope gave results independent 

 of small changes in the zero and of variations in sensitive- 

 ne>> in different parts of the scale. With the volume of air 

 used — about 400 c.c. — it was found that the electrode was 

 charged to about 04 volt in twenty minutes, an amount 

 which could be measured with a considerable degree of 

 accuracy. As a rule five sets of four readings each were 

 taken with each vessel. During the first, third, and fifth, 

 the vessel was freely exposed to the penetrating radiation; 

 while during the second and fourth it was completely sur- 

 rounded by a lead, or other, screen. Means of each set of 

 readings were taken, and from them the mean leak with and 

 without the screen was calculated. This of course gave the 

 proportionate reduction due to the screen. The vessels were 

 made of lead, tin, iron, aluminium, and zinc. In confirma- 

 tion of Strutt's results, it was found that the ionization varied 

 verv greatly with the material. The following table gives, 

 the average value obtained for the various metals, on a quite 

 arbitrary scale. For purposes of comparison the correspond- 

 ing values obtained by Strutt, reduced to the same arbitrary 

 scale by equating the values for zinc, are placed alongside : — 



Material. Ionization. Strutt's Value. 



Lead D"2 4*2 



Tin 4-7 4-4 to 6'3 



Iron 2*9 — 



Aluminium 2'7 2'7 



Zinc 2-3 2-3 



It will be at once admitted that the two sets of values 

 >how a surprising agreement. This practically disposes of 

 the suggestion that the differences in the ionization in vessels of 

 different materials are due to the presence of radioactive 

 impurities. Probably enough impurities of the kind do play 

 -ome part. For instance, in the case of tin Strutt found a 

 considerable difference between different samples, while two 

 samples of platinum gave 3'8 and 7*4 respectively. A sample 

 of the same metal, tested in the course of tin's research, gave 

 2-4 — a number much smaller than the smaller of the two 

 given by Strutt. That such discrepancies are due to the 

 radioactive impurities -<'em< more than likely, but it seems 

 impossible on tin- hypothesis to account for the agreement 

 shown above. Campbell, working in the Cavendish Labora- 

 tory, has obtained some results, a summary of which he has 

 kindly placed at my disposal. lie obtains an order for the 

 different metal- identical with that given above. McLennan 



