2\Y2 Research Staff of the G. E. C, London, on the 



both curves rise sharply, and just after it the ratio i g \i e is 

 much greater than just before. 



3. It is clear, therefore, that the development of the glow 

 and the beginning of absorption are accompanied by a great 

 increase in the positive ions formed *, The opposite con- 

 elusion suggested by measurements with two electrodes must 

 be wrong, because the positive ions, above the glow potential,, 

 do not reach the cathode and are not represented in i. They 

 either recombine in the gas or they travel to the walls and 

 are neutralized there by electrons from the cathode. 



Two kinds of observation show that at the lower 

 pressures, when the glow potential is sharply defined (less 

 than 0*02 mm.), the second alternative is the more important. 

 If the grid is insulated, an estimate of the potential which it 

 takes up can be made by comparing the values of the glow 

 potential with those when the grid is maintained at different 

 potentials by a battery. It was thus found that it must take 

 up very nearly the potential of the cathode. If i g is 

 measured with the grid at this potential, it is found that 

 nearly all the positive ions are reaching it and that i g is 

 increased but little by making A 7 ' negative. Accordingly, the 

 field with the walls insulated tends to arrange itself so that 

 nearly all the positive ions go to the walls. Again, it has 

 been observed in all vessels with fine wire electrodes that it 

 is only if the pressure is below 0*001 mm., and the positive 

 ionization very small, that the anode can be heated by 

 bombardment. At higher pressures, though the current and 

 energy carried by the discharge may be sufficient to melt the 

 anode, the anode does not become visibly hot ; all the heat 

 is communicated to the walls. When, however, the pressure 

 is reduced so far that there are only just enough positive ions 

 to neutralize the space charge and give the full thermionic 

 emission, the anode can be readily heated. This observation 

 shows that the walls, as well as the anode, must be acting as 

 an electrode for the reception of electrons. But, if the walls 

 are continually receiving electrons, they must either be 

 receiving positive ions or emitting electrons. The second 

 alternative is not likely when the walls are of bare glass, for 

 there does not seem any evidence that insulators can emit 

 slow electrons (delta rays) in sufficient quantity to neutralize 

 the charge they receive from the incident electrons. The 

 first alternative is the more plausible, especially in view of 

 the observations to be described presently, which show what 



* The alternative possibility that the increase of i g above the glow 

 potential is due to photoeL ctric emission from the grid is disproved by 

 all the experiments described below. 



