Gases contained in Small Vessels. aus yi 
by C. T. R. Wilson*. Most of the experiments were ex- 
ecuted in glass bulbs of various sizes, thinly silvered inter nally, 
some, however, in brass vessels of cylindrical shape. A 
quartz rod, attached by sealing-wax, was substituted; for the 
sulphur bead used by Wilson on account of the action of 
nickel-carbonyl on sulphur, and the supporting rod was kept 
at a high potential by a set of very constant storage-cells. 
An initial potential of 80 volts was used in most experl- 
ments, and the leakage observed over a period varying from 
about 2 to 48 hours. The eyepiece micrometer-scales were 
calibrated by means of Clark cells ; the value of one division 
varied in the different electroscopes between (0°16 and 0-4 of 
a volt ; the capacities were compared with that of a sphere 
of ews radius fF. 
The insulation was tested from time to time, and found to 
be so satisfactory that the leak varied by less than 0°2 of a 
division per hour, whether the supporting rod was to earth 
or to 80 volts ; at low pressures, however, where the leak is 
very small, an equal number of observations was alw ays 
taken either w ay, to eliminate any error due to want of 
insulation. 
iL. Eaperiments with Arr at Atmospheric Pressure. 
As the experiments were always extended over so long a 
range of time that the error of the reading only amounted 
to a few per cent., and as the sensitiveness of ‘the scale and 
the capacity of the leaking system. were practically constant, 
individual readings with the same electroscope ought to have 
agreed to about 5 to 10 per cent., supposing the quantity under 
consideration to be a constant. But it was soon found out 
that ithe accuracy attainable was far less, and that different 
readings sometimes disagreed to 30 per cent. and even more. 
It was therefore concluded that the spontaneous ionization 
undergoes variations, and a great number of experiments 
were undertaken in order to find out the cause of this 
variability. It was thought that light, temperature, moisture, 
and weather-conditions “might account for. the | bseneet 
changes ; a careful examination, however, showed this not to 
be the case. The experiments leading to this conclusion are 
therefore not recorded in detail; but to show the order of 
magnitude of the variations involved, the maximum, mini- 
mum, and mean values of observations: taken wo five 
different vessels during several months are given in Table I. 
* Ibid, 
T Cf. C. T. R. Wilson, Joe. cit. 
