the Form of Free Triode Vibrations, 215 



voltage cycle, but that during the positive half cycle the 

 maximum value of grid current is reached in two quite 

 definite stages. An analysis of the characteristic surface 

 •of the grid current (i g = /(v a , v g )) which was obtained sepa- 

 rately for the same tube showed that in the first stage the 

 grid current consists of electrons emitted by the filament 

 which are deviated from the main anode stream. When the 

 grid reaches higher positive values of potential, however, a 

 second increase in the value of the grid current is obtained 

 which is due to electrons arriving at the grid but originating 

 not at the filament but at the anode, which under the action 

 of the electronic bombardment from the cathode emits 

 secondary electrons at relatively low potentials (of the order 

 of 40 volts). This loss of secondary electrons at the anode is 

 the principal cause of the sudden fall of the anode current 

 observed during the saturation phase. It thus seems clear 

 that secondary emission plays quite an important part in the 

 action of a triode even at comparatively low anode potentials. 



The oscillograms (PI. IX. A, B, (J, D, E, and F) were 

 taken immediately after one another, all electrical constants, 

 with the exception of M, being left unaltered. Bringing the 

 coil L 2 of fig. 9, together with the iron core, nearer the 

 middle of 1^ resulted, as is shown by the oscillograms, in a 

 decrease of the frequency. The frequency for the cases 

 D, E, and F, as measured with a tuning-fork, was 16'3 per 

 second. 



The straight line on the oscillograms indicates the steady 

 anode current value when the triode is not generating (mutual 

 inductance zero). 



At this very low frequency the cooling which the filament 

 undergoes, due to the "evaporation " of the electrons, is very 

 conspicuous. During the saturation phase (see PL IX. F) 

 the saturation current is seen every time to decrease, thus 

 indicating a fall of temperature during the phase of satura- 

 tion and an increase of temperature during the phase of zero 

 current. These experiments therefore confirm the theory 

 outlined above. 



This fall of temperature due to the loss of heat conse- 

 quent on electronic evaporation, which was first observed 

 with these oscillograms, was afterwards confirmed by a 

 separate experiment. 



Again the string galvanometer was used and the emission 

 current, which is a very sensitive indicator of filament tem- 

 perature, was observed immediately after the anode circuit 

 was closed. An " oscillogram >3 of the emission current (grid 

 and anode being connected) taken in this way is shown in 



