206 Mr. E. V. Appleton and Dr. B. van der Pol on 



for the difference between the oscillation characteristics in 

 the region mentioned. Further evidence that this is the 

 correct explanation was obtained by measuring the maxi- 

 mum values reached for both positive and negative values 



of v g . Since * — - is negligible for all negative values of v g 



r 



and appreciable for positive values when v ff ^v ao -\-v a , we 

 should expect these maximum values to be different 

 when v g is large. The circuit used to test this point is 

 shown in fig. 4. 



Eur. 4. 



. Illjlll-^^ijj 



A diode vacuum-tube D with a reversed battery B 2 in the 

 anode circuit was connected to the grid and filament of 

 the oscillating triode T. The voltage of the battery B 2 was 

 increased until the reading of the sensitive galvanometer Gr 

 was almost zero. The maximum positive potential reached 

 by the grid with respect to the filament was then given 

 approximately by the value of B 2 . By reversing the switch S 

 the maximum positive potential reached by the filament 

 with respect to the grid could be also measured. These 

 potentials were found to be different, two typical values for 

 a case of the circuit shown in fig. 4 being + 64 volts and 

 -80 volts* 



It is more difficult to gauge the effect of (fr), but the 

 following considerations show that for the range of fre- 

 quencies we have been considering so far (400 to 700 J 

 per sec.) the thermal changes in the filament would not 

 be large enough to produce the effect obtained. Very simple 

 experiments carried out with static voltages (see fig. 5) 

 showed the marked difference in the temperature of an 



* Further support for this explanation was obtained by investigating 

 the phase relation between v g and v a with a cathode ray oscillograph. 

 The simple 180° phase relation was found to be disturbed for large- 

 positive values of v q . 



