Gases contained in Small Vessels. 567 
The fact that the ionization in Ni(CO), is not quite so big 
as might be expected from its density (5'1 instead of 5:9), 
may, perhaps, be explained by the circumstance that the 
Ni(CO), was always introduced just before the experiments, 
Fig. 3. (Bulb IT.) 
Nt 
¢ 
a 
LEAK; DIVISIONS PER HOUR. 
° 
a 
DENSITY AlR OF 760 MM. PR. =! 
while the air had always been in the vessel for a considerable 
time. If values for freshly-introduced air are taken, the 
curve for air sinks even below that for Ni(CO),. It was 
not possible to test whether the leak in the latter showed 
the same increase with time as the leak in air, as Ni(CO), 
could not be kept for a long time in the vessel without 
decomposing. 
Conclusions. 
1. The spontaneous ionization undergoes changes due to 
unexplained causes which, however, depend on local 
conditions. 
2. The variations in small vessels are bigger than those in 
large vessels and amount sometimes to as much as 40 
per cent. 
3. The ionization in small vessels of silvered glass is not 
strictly proportional to the pressure; but can be 
represented by a straight line over a considerable 
range, indicating a complex nature of the radiation 
from the walls. 
4, The ionization in nickel-carbonyl is 5:1 times as great 
as that in air, and is thus almost proportional to the 
density. 
These experiments were executed in the Cavendish La- 
boratory. My sincere thanks are due to Professor J. J. 
Thomson for his kind interest and help. I am also indebted 
to Dr. Ludwig Mond for supplying me with the nickel- 
carbonyl. 
