904 Proceedings of Royal Society of Edinburgh. [sess. 
drawn through by a water-pump. The rate of leak of the 
electroscope was read by a low-power microscope provided with 
a micrometer scale, and was taken — 1°, when the flame was 
not lit and the field not on ; 2°, when the flame was lit and 
the plate c d to earth ; 3°, when the flame was lit and the 
field on. 
When the flame was first lit, the leaf of the electroscope was 
somewhat unsteady. This unsteadiness was due to convection 
currents of air in the electroscope case ; but after an interval 
of fifteen or twenty minutes the leaf became steady and readings 
could be taken. The air in the electroscope case was kept dry 
by means of calcium chloride. 
In order to be certain that the ions from the flame really reached 
the enclosure B in an uncombined state, the aluminium foil was 
removed and the rate of leak of the electroscope was taken — 1°, 
with the flame unlit; 2°, with the flame lit; -the plate cd being 
to earth in both cases. The result was that the rate of leak due 
to the ions was from 50 to 80 times the natural rate of leak, 
showing that a sufficient supply of ions was present. 
The following are two sets of observations of the kind taken in 
the course of the experiments. In all cases given, the leaf of the 
electroscope was charged to a potential of 360 volts. 
1°. Flame unlit. 
(Plate cd to Earth.) 
Time in Minutes. 
Reading. 
Leak per 5 Minutes. 
0 
33*3 
5 
35-4 
2*1 
10 
37*7 
2*3 
15 
39*8 
2*1 
20 
41*9 
2*1 
Average leak per 5 minutes, 2*15. 
