Temperature on Spontaneous Ionization. G17 



Table I. 



i 



V. 



t. 



t . 



Q CxlO 3 . 



'i- 



__T_ 



-Q'/CxlO*.| 



•261 



203 



157 



59 



209 



243 



47^ 



•261 



207 



155 



65 | 



210 



259 



59 I 



•261 



199 



145 



58 y 60 



202 



257 



71 ^60 



•26] 



175 



129 



69 | 



176 



213 



55| 



■261 



171 



140 



47; 



181 



230 



71 ; 



•572 



471 



403 



S3 A 



492 



561 



80 



•572 



472 



4^;8 



41 I 



492 



552 



70| 



•572 



4(H) 



420 



67 } 57 



-^83 



535 



62 K>-"> 



•572 



520 



479 



45 



540 



601. 



65 ! 



•572 



513 



467 



51 



532 



583 



54 i 



8. The equality of Q and — Q' may be tested more directly. 

 The electrode is insulated, the flame applied for a minute 

 and then removed : after the lapse of another minute all the 

 heating changes are over. If Q= — Q' the total leak ^^ill 

 be the same as if no heating had taken place. However, 

 such observations showed that Q' was a little greater numeri- 

 cally than Q. The difference between the two results may 

 arise from the fact, that in the second of the two methods the 

 initial cooling does not take place from a condition of thermal 

 equilibrium. 



Conditions which determine the changes. 



i-. The remainder of the observations were directed to 

 determining the factors which influence the magnitude of Q, 

 which may be taken as a measure of the intensity of the 

 changes. 



Q is independent of the direction of the current through 

 the vessel and of the magnitude of the P.D. between the 

 walls and the electrodes, so long as this is sufficient to send 

 a saturation current through the gas. (The degree of satura- 

 tion was such that an increase of the P.D. from 500 volts, 

 the value usually employed, to 1000 volts produced an in- 

 crease of 5 per cent, in the leak.) If the P. I) is considerably 

 less than that required for saturation the magnitude of the 

 effect diminishes, until, when the vessel is heated with the 

 walls and the electrode at the same potential, no charge is 

 communicated to the latter. 



10. Q does not vary greatly with the material of the 

 walls of the vessel, except in so far as these influence the 

 ionization in the vessel (see § 12) : experiments were made 

 with zinc, tin, copper, and lead walls. It is also nearly 



