Mr. A. A. Campbell Swinton. 

 FIG. 7. 



G' 



that could pass through the small apertures in B, the D'Arsonval galva- 

 nometer would not give satisfactory readings even when a coil with 500 

 turns was used, a reflecting quadrant electrometer was employed 

 instead. The needle of this electrometer was connected to the inner 

 Faraday cylinder F by means of an insulated wire threaded through a 

 lead pipe connected to earth so as to exclude all outside influence, and 

 the electrometer was otherwise connected up according to Mascart's 

 method. The deposition of moisture upon the tube being found to 

 affect th3 results, incandescent electric lamps were arranged so as to 

 keep the whole at a uniform temperature slightly above that of the sur- 

 rounding atmosphere. In the subsequent experiments the anode B was 

 always connected to earth. 



In the following tables each unit of the number given as the de- 

 flection of the electrometer corresponds to a mean pressure of about 

 3 volts, though very probably, owing to the intermittent nature of the 

 discharge, the actual instantaneous value was much higher. In each 

 case the readings of the second series of observations were taken in 



