Gases contained in Small Vessels. 561 
be observed, as bigger irregular variations were superposed 
and obscured the periodicity. In Table IIT. only one typical 
example is quoted, where obviously some disturbance took 
place. Such disturbances, however, and the whole resem- 
blance in the behaviour of two vessels close to each other 
seem to be due to some local uncontrolled circumstances, as 
the agreement disappears when the electroscopes are far 
apart. 
Tasxe III. 
| Leak in the Leak in 
_ silvered the 
glass vessel. | brass vessel. | 
Range of time over whici the 
average is taken. 
| Jan. 21, 6 p.w.—Jan. 22, 9 a.m.... 76 76 
Jan. 22.10 a.m.— ,, LPM. a eck 66 
dan. 22, 1Ppma.— ,, 3 PM: s 8-4 : 76 
| Jan. 22,3 P.M.— ,, 5 P.M. i 64 | 6:3 
Jan. 22, 6 p.w.—Jan:; 23, 9 a.M. ... 8:0 8°3 
As the electroscope-method does not permit of measure- 
ments of the amount of ionization at a certain moment, but 
only of the average over a considerable time, it seemed advis- 
able to carry out some experiments by the electrometer- 
method, to see whether it would be possible to detect well 
marked changes also in this case. The spontaneous leak was 
therefore measured in a brass vessel of about 3 litres, and in 
a tin vessel of about 10 litres capacity, during several months, 
by aid of a sensitive Dolezalek electrometer. 
The values thus observed were found to be much more 
constant than those mentioned above ; but there were certainly 
variations which exceeded the limit of error and could not be 
otherwise explained than by changes of the spontaneous ioni- 
zation itself. The greatest differences between two sets of 
readings were 16 per cent. in the case of brass, and 15 per 
cent. in the case of tin; the greatest deviation from the 
mean of all observations did not exceed 10 per cent. in either 
case. 
’ The number of ions produced in the c.cm. per second 
varied in the brass vessel between 29 and 35, and in the tin 
vessel between 42 and 50. These absolute values are not 
very accurate, as the capacities were not very carefully de- 
termined and are uncertain to at least 10 per cent.; but the 
relative measurements in the same vessel are certainly exact 
to 5-7 per cent. 
It followed from these experiments, fo the variations in 
Phil. Mag. 8. 6. Vol. 8. No. 46. Oct. 1904. 2Q 
