IONIZATION POTENTIALS IN DISSOCIATED GASES. 219 



could be connected to the quadrant electrometer, which measured both 

 the current due to ionization of the gas between NN and PP and the 

 photoelectric current from PP or W due to the radiation set up by- 

 electron bombardment of the gas. Those effects, setting in at critical 

 potentials which were due to radiation, were relatively more pro- 

 nounced when the plate PP was joined to the electrometer. Effects 

 of ionization were relatively more marked when the wire IV was used, 

 since this wire could collect all the positive ions formed, but would 

 receive but little of the radiation. 



The grid GG was used to vary the proportion of atomic hydrogen 

 in the neighborhood of the net A'^A''. This was done by heating it to a 

 high temperature by an electric current, if atomic hydrogen was 

 desired. The relative proportion of atomic to molecular hydrogen 

 was never large, but was sufficient for our purpose. Those critical 

 potentials which were relatively more marked when the grid GG was 

 hot were ascribed to atomic hydrogen and the others to molecular 

 hydrogen. 



By thus varying the sensitiveness to detection of radiation, and 

 by varying the amount of atomic hydrogen present, it was possible 

 definitely to determine the nature and origin of the effect setting in 

 at each of the critical potentials of hydrogen. 



The critical potentials are shown by Fig. 5, taken with the plate 

 PP joined to the electrometer, and with the grid GG hot. Similar 

 curves with the wire W, or with the grid cold, show fewer " breaks," 

 as expected. The interpretation of these critical potentials is indi- 

 cated by the " ratio " curves of Fig. 6. These curves show ratios of 

 electrometer deflections under the various conditions, / (on)// 

 (off) means the ratio of the deflections using the wire JV (relatively 

 sensitive to effects of ionization) with the grid current on (hot) and 

 off (cold) respectively. R (on)// (on) means the ratio of the de- 

 flections with the plate (relatively sensitive to radiation) to those with 

 the wire (relatively sensitive to ionization), the grid current being on 

 in both cases. When these curves are interpreted in the light of the 

 discussion above, and when proper allowance is made for the effect of 

 the initial velocities with which electrons are emitted from the fila- 

 ment (which introduces a small correction for weak effects and a 



