Diffusion Cells in Ionized Gases. 



503 



between the voltmeter readings with the radium in the two 

 positions gave twice the electromotive force of the diffusion- 

 cell. An insulated switch, not shown in the figure, enabled 

 the connexions to the two sets of plates to be interchanged. 



By means of two switches k 3 and & 4 , actuated by an 

 electromagnet pendulum-bob M, a field of two volts from 

 the battery B 2 could be suddenly applied to the plate A for a 

 definite small interval of time, usually about 0"05 second. 

 During this interval ions of one sign within a fraction of a 

 millimetre of the plate B were swept into the plate, pro- 

 ducing a deflexion of the electrometer. This could be 

 repeated with the ionizing source acting near the other 

 plate, or with the connexions to A and B reversed. The 

 ratio of these deflexions gave the ratio n a /n . All other 

 quantities in equation (2) being known, these measurements 

 enabled a quantitative test of the equation to be made. 



Unfortunately our tube of radium did not produce suffi- 

 ciently strong ionization to permit an accurate measure of 

 the ratio n a /n , the electrometer deflexions being too small in 

 comparison with a somewhat irregular shift due to an un- 

 balanced induction effect while the field was changed by the 

 switches k 3 and & 4 . We were consequently forced to use an 

 X-ray tube as the ionizing source in this part of the expe- 

 riment, and the measurements were subject to the somewhat 

 erratic variations in its action. However, with this arrange- 

 ment, we obtained results of which the following are 

 examples. n and n a are in terms of the electrometer 

 deflexions noted just after the fall of the pendulum M. The 



Table I. 



n . 



n a . 



E (calc). 



E (obs.). 



24 



7 



000497 



00042 



18-5 



5 



000528 



00034 



24 



5 



000633 



0-0067 



electrometer sensitiveness, about 5000 divisions per volt, 

 was such as to make the apparatus very sensitive to dis- 

 turbing influences, so that we feel that our measurements 

 may be taken as a fair support of the theory. Doubtless, by 

 taking greater precautions to insure steady strong ionization 

 and by increasing the distance between the plates A and B 

 so as to increase the amount of recombination, and therefore 

 the ratio n a /n , the experimental test could have been made 

 more accurate. 



