234 Prof. J. Patterson on the Ionization in 



required recharging every other day. It was found that 

 about 80 volts gave the maximum sensitiveness. The de- 

 flexions were measured by a telescope and scale about two 

 metres distant. A deflexion of one scale-division was equivalent 

 to 0-000006 volt P. D. between the quadrants, or a P. D. of 

 one volt between quadrants would give a deflexion of 15,000 

 scale-divisions about. 



The efl'ective capacity of the electrometer and its con- 

 nexions depends almost entirely on the charge on the needle, 

 and in the experiments varied from about 400 to 200 cms. 

 two days after charging. Before comparative readings could 

 be taken it v/as thus necessary to determine the capacity for 

 each set of observations. 



The electrometer needle had a very long period, and the air- 

 damping was sufficient to make it dead-beat. The zero-position 

 was not very constant but drifted in one direction. It was 

 impossible to get insulation absolutely free of charge, so if the 

 quadrants were opened they gained a slight charge but always 

 in one direction. It was thus necessary to observe the rate at 

 which the quadrants charged up when the cylinder was main- 

 tained at a constant potential and the quadrants opened. In 

 order that the movement might become steady the time was 

 taken from the instant a certain scale-division crossed the spider 

 line in the telescope, this division was usually about 200 div. 

 from the zero-position. By taking equal intervals of time and 

 observing the deflexions with the cylinder charged alternately 

 positive and negative any extraneous effects would be 

 eliminated. 



To measure the capacity of the electrometer and its con- 

 nexions the number of ions produced per second in the cylinder 

 was taken as constant during the observation, and hence the 

 rate at which the system charged up would be proportional 

 to the capacity. The rate at which the system charged up 

 was determined and then a condenser of known capacity was 

 inserted in parallel and the rate again determined.. If oc and 

 /3 represent the deflexion of the needle in scale-divisions per 

 second without and with the condenser in circuit, C the capacity 

 of the electrometer and its connexions, and Ci the capacity 

 added, then 



from which C can be determined. 



The saturation current between the walls of the cylinder 

 and the electrode was first obtained, and is shown in flg. 2 

 for air that had remained for some time in the cylinder. It 

 will be seen that the air is practically saturated when the 



