90 ROYAL SOCIETY OF CANADA 
ing the experiment, and as no such action was exhibited when the two 
compartments were not united, the result pointed to the existence of an 
electric conduction along the surface of the glass joining the two parts 
of the tube. 
In order to test this the electrometer was simply connected to the 
wax a’ instead of to the electrode C. On then exciting the tube the 
same charging action was obtained. 
It was therefore clear that, although the sealing wax was a good 
insulator for small voltages such as those used in testing the insulation 
of the electrode C when no discharge was passing in the tube, it was 
not sufficient to cut off conduction along the glass when the tube was 
excited by the induction coil. 
To overcome this difficulty part of the glass tube Z, F, was removed, 
and replaced by one of brass held in position by sealing wax. With this 
tube joined to earth the experiments just described were repeated, and 
not the slightest trace of any electric action was observed in connection 
with the electrode C. ; 
The effect of reducing the pressure in the lower as well as in the 
upper chamber was then investigated. The two taps # and I’ were kept 
open and both chambers exhausted together until any desired pressure 
was reached. The tap / was then closed and the exhaustion in the 
upper chamber completed. 
In this way a wide range of pressures was examined and in no case 
did the electrometer give any indication that the electrode C either 
gained or lost a charge when the tube was excited. In all these tests 
the tap F was kept closed when the discharge was passing in the tube 
in order to prevent any possible conduction in the gas from one chamber 
to the other. This was found to be especially necessary at very low pres- 
sure when A and H were selected as the discharging electrodes. Other- 
wise the well known phenomenon illustrated by Hittorf’s experiments 
occurred, the discharge preferring to take the longer path round by the 
tube H, F', rather than the short one across the bulb of the tube from 
the cylinder D to the dise A. 
In all the experiments with this form of tube hitherto described 
the end of the cylinder D consisted of a dise of aluminium d ‘04 milli- 
metres thick. This dise was now replaced by one only ‘904 millimetres 
in thickness and both chambers were as highly exhausted as possible. 
The tap F was then closed and the tube excited. Different methods of 
connecting the coil to the tube were tried but in every case the cylinder . 
again effectively screened the electrode C, and no charging was ob- 
served. Charges given independently to the electrode C were also main- 
| 
| 
) 
| 
| 

