Gases contained in Small Vessels. 559 
Every experiment mentioned in this table gives the average 
leak for 6 to 16 hours. The ionization during the day was 
by no means constant, and the readings in “single hoars 
disagreed often even by more than the Above averages ; but 
as an observation over a short range of time is subject to a 
much greater error, only averages are included in the table. 
The numbers in the last two rows of Table I. are given in 
different units (scale-divisions per hour) for each of the 
two cylinders, and cannot therefore be compared. All the 
other numbers, however, represent the number of ions of 
either sign set free per second in each c.cm., calculated from 
the experimental data by taking 3:1x10~-" E.U. as the 
charge on one ion. 
©. T. R. Wilson’s determination of the same quantity is 
39 ions per c.cm. per second*. The value obtained in the 
first series (I.a) is much higher; the bulb was therefore 
resilvered, and the number of ions thereby reduced to the 
smaller value given under I.b and Ic. The first coating 
may have contained some radioactive impurity ; the fact, 
however, affords another instance for the observation of 
Stratt +, that different samples of the same metal can give 
very different values to the spontaneous leak. 
he smaller values I.) and I. ¢, as well as those obtained 
for bulbs IL. and III. are still appreciably bigger than 39 ; 
this may, however, be due to the fact that C. T. R. Wilson 
observed the leak in a vessel of 163 c.cm. volume. If the 
ionization is entirely or partly due to the influence of the 
walls, a bigger production of ions per c.cm. is to be expected 
in smaller vessels. 
It will further be seen from Table I. that the values for 
the ionization in the day and in the night show the same 
variability, and that the means for either are closer to each 
other than one would expect, showing that neither small 
changes of temperature nor light affect the leak. In the 
ease I. b it seems as if the leak in the night was a little too 
high. But this is certainly no genuine effect, as shown by 
I.c¢, which gives observations with exactly the same apparatus 
only in a different room. 
This result suggested the idea that different local conditions 
affect the leak. In order to examine this, two vessels of 
about equal volume (100 ¢.cm.) one of silvered glass and 
one of brass, were at first placed close to each other and, 
* This value has been calculated from C. T. R. Wilson’s observations 
employing the most recent determination of the charge e of an ion by 
H. A. Wilson (Phil. Mag. [6] v. p. 429, 1903); C. T. “R. Wilson takes 
e=6°5 . 10-10 E.U., and dhepalois gets 19 ions per sec. and c.cm. 
t+ ‘Nature,’ Feb. 19, 1903, 
