Disappearance of Gas in the Electric Discharge. 241 



ratio igj'ie were measured for very small currents (about a 

 microampere), it was found to be proportional to the pressure 

 over the relevant range— in fact, if p is expressed in mm., 

 •,/i = 10-0 p with V = 300 volts. (Above 200 volts the ratio 

 is almost independent of V.) But if i were increased to 

 (say) 10 microamps., lower values of the ratio were obtained, 

 especially at the lower voltages ; the greater was i e , the less 

 was the ratio. 



This observation shows clearly that there is considerable 

 recombination, and suggests a way of correcting for it. If 

 we take as ip, in place of the measured ig, the value calculated 

 from the measured i e by the relation, valid for small currents, 



: gji e =10p J then we might hope to obtain the true value of i p 

 for the calculation of a. Unfortunately, the application of 

 this correction leads always to values of a less than 2; and 

 further consideration shows that this result might be antici- 

 pated. For now we are including in i p ions which recombine 

 before they collide with a neutral molecule, and therefore add 

 nothing to —p. The current which we should expect to give 

 the value 2 for a is that representing the ions which collide 

 with a neutral molecule before reaching the electrodes, 

 whether or no their positive charges are subsequently neu- 

 tralized in the gas ; it seems impossible to deduce that 

 current from the measurements that were possible. The low 

 " corrected " values are also doubtless due partially to the 

 cause discussed in the next paragraph. 



11. It remains to explain the low values obtained with 

 small currents. They are due to the reversal of the reaction 

 by the action of the discharge on the C0 2 condensed on the 

 walls, which are serving as anode. For if, after some gas 

 lias been absorbed, the remainder is pumped away and a 

 pure electron discharge passed, there is a rapid evolution of 

 gas. No measurements of its rate were made, for it is not 

 likely that the evolution would be the same under such 

 conditions as when the discharge was passing through gas at 

 a pressure of 0'2 mm., but in order of magnitude it was quite 

 sufficient to account for the low values. This result might 

 have been anticipated from the conclusion stated already 

 (but reached subsequent to the experiments on vessel 2) that 

 the reaction under consideration is reversible in the discharge. 

 And here, too, may be found the explanation of the second 

 anomaly of par. 8, namely, that the absorption of gas falls 

 off too rapidly at low pressures. Unless the C0 2 formed is 

 removed with great rapidity, some will adhere to the walls 

 of the vessel and will not be removed when the pressure is 

 reduced; it is, however, easily removed by heating the walls. 



