varied approximately as the pressure of the air between the pressures examined, viz. 43 mms. and 743 mms. of mercury. These results point to the conclusion that, in a good vacuum, a charged body would lose its charge extremely slowly. This is in agreement with an observation of Crookes, who found that a pair of gold-leaves retained their charge for several months in a high vacuum.
Wilson[1] at a later date investigated the leakage for different gases. The results are included in the following table, where the ionization produced in air is taken as unity:
+-++-+
| Gas | Relative |(Relative ionization)/(density)|
| |ionization| |
+-++-+
|Air | 1·00 | 1·00 |
|Hydrogen | 0·184 | 2·7 |
|Carbon dioxide | 1·69 | 1·10 |
|Sulphur dioxide| 2·64 | 1·21 |
|Chloroform | 4·7 | 1·09 |
+-++-+
With the exception of hydrogen, the ionization produced in different gases is approximately proportional to their density. The relative ionization is very similar to that observed by Strutt (section 45) for gases exposed to the influence of the α and β rays from radio-active substances, and points to the conclusion that the ionization observed may be due either to a radiation from the walls of the vessel or from external sources.
Jaffé[2] has made a careful examination of the natural ionization in the very heavy gas nickel-carbonyl, Ni(CO)_{4}, in a small silvered glass vessel. The ionization of this gas was 5·1 times that of air at normal pressure while its density is 5·9 times that of air. The leak of the electroscope was nearly proportional to the pressures except at low pressure, when the leak was somewhat greater than would be expected if the pressure law held. The fact that a gas of such high density and complicated structure behaves like the simpler and lighter gases is a strong indication that the ionization itself is due to a radiation from the walls of the vessel and not to a spontaneous ionization of the gas.
