92 Dr. Balth. van der Pol on the Production of 



the wire connecting the battery B x (in our case 80 volts) 

 with the condenser (J can be varied within wide limits without 

 influencing the action of the generator appreciably. The 

 same is true for the length of the wires connecting the 

 heating battery B 2 with the filament. 



The frequency and amplitude of the generated current are, 

 however, greatly influenced by the form and length of the 

 wires PBO and GAF. The simplest way perhaps to start 

 the valve oscillating is as follows: — 



Both PBO and GAF are bent open so as each to be 

 approximately arcular. The filament being heated, the dis- 

 tance of the plates of the condenser C is gradually reduced 

 from say a few centimetres. During this gradual increase 

 of capacity a point will in general be reached where the 

 mean current in A T suddenly alters. This is an easy way of 

 making sure that the valve is oscillating. In our experiments, 

 e. g., the mean plate current jumped from 5 milliamperes to 

 12-15 milliamperes when the plates of were made to 

 be 1 centimetre apart. Varying the capacity C can be 

 interpreted as varying at the same time both the capacity in 

 the plate circuit and the coupling coefficient, the capacity 

 between grid and filament being constant, and once for all 

 given by the dimensions of the ionic tube. There is further 

 a certain range of values of over which the Generator will 

 oscillate, just as in the ordinary valve circuits the mutual 

 inductance between grid and plate can be altered over a 

 certain range without the tube ceasing oscillating. In our 

 ■case, how-ever, not only is the intensity of the oscillations 

 altered when the distance between the plates of C is varied, 

 but the wave-length emitted as well. With a small capacity 

 C the wave-length is smaller than with a big capacity. This 

 shows that C does not act merely as a path for the high- 

 frequency oscillations, but has its share in determining the 

 wave-length. The wave-length is further determined by the 

 self-inducfanees of both PBO and GAF. For when one of 

 these self-inductances is diminished, by bending the wire 

 more into the shape of a single loop instead of a circle, the 

 wave-length will be decreased. This is equally true for the 

 grid inductance as for the plate inductance. 



However, in diminishing the inductances of these circuits 

 a state will soon be reached w 7 here no value of the capacity C 

 can be found which enables the system to oscillate. It seems, 

 as in the case of the arc generator, that a minimum self- 

 inductance exists for which a particular tube can oscillate, so 

 that each valve would have a definite upper limit of frequencies 

 at which it can generate oscillations. Experiments showed 



