TRANSACTIONS OF SECTION A. 



151 



bombarding the original electron emitting spot. This bombardment produces 

 a new burst of electrons, and so a self-sustaining current oscillation may be 

 set up, the period of which depends on the applied grid potential. 



The following table shows what frequency and wave-length of oscillation 

 would be expected from the above general theory in the case of a valve with 

 cylindrical gauze grid 6 millimetres in diameter and an axial hot wire. The 

 singly-charged ions of hydrogen and mercury are there worked out for a grid 

 potential of 1 volt. 



Nature of Charged I e/m i u ' n i Wave-length 



Particle j (approximate) (cms/sec) i (cycles/see) (metres) 



Electron 

 Hydrogen Atom. 

 Mercury Atom . 



1-7 



In actual experiments it was found that the arrangement of fig. 2 usually 

 radiated energy at a frequency varying between 7.0 X 10= and 4.0 X id- 

 cycles. The wave-lengths (from which the frequencies were deduced) were 

 measured by means of a heterodyne wave-meter in the vicinity. 



The oscillations may therefore be safely ascribed to mercury vapour. It 

 is to be observed, however, that the above calculated frequencies are based 

 on singly-charged monatomic molecules, and that the frequency 6.6 X 10- 

 cycles corresponds to the monatomic mercury molecule. If polyatomic molecules 

 are involved the frequencies to be expected would be 1 \/2, 1 \/3, iVITetc, times 

 6.6 X 10= cycles. 



In the following table are shown the frequencies to be expected theoretically, 

 and for comparison those actually observed. 



Number of 

 atoms in 

 Molecule 



1 



2 

 3 



4 



Theoretical 

 Frequency 



6.6 X 10- 



4.7 X 10'5 



3.8 X 10* 

 3.3 X 10- 



Observed 

 Frequency 



6-4 X 10- 

 4-5 X 105 

 3-5 X 105 



Oscillations in addition to the above have been detected, believed to be 

 clue to the oxygen and carbon dioxide molecules, but have not yet been 

 investigated in detail. 



Referring back to the formula, it will be seen that the square of the 

 speed of the ions (and therefore the square of the oscillation frequency) should 

 be proportional to the potential applied to the grid. 



This prediction is amply borne out in practice (the frequency of oscilla- 

 tion) - plotted against the grid potential yielding in all cases so far studied 

 an excellent straight line. These lines cut the axis of potential at points 

 determined partly by the position of the emitting spot on the filament and 

 partly by the natural velocity emission of the electrons. This is a point 

 which is being investigated in further detail, particularly from the point of 

 view of getting a more accurate value for the ratio e/m than has hitherto 

 been found possible. 



P 3 



