244 Research Staff of the Gr. E. C, London, on the 



It will be seen that all values now lie between 1/9 and 

 2*5, and that, in general, it is the values for which the re- 

 combination correction is the greatest which are the highest. 

 But even when this correction, which is doubtless excessive, 

 is applied, the values still lie, on the whole, above 2. This 

 circumstance leads us to consider one more source of error, 

 which was mentioned in par. 5, but has been left out of 

 account since. Even if there is no recombination, i g must 

 always be slightly less than i p , because some of the positive 

 ions must go to the cathode and not to the grid. The 

 number of positive ions lost in this way is not easy to deter- 

 mine. An attempt was made by measuring, at different 

 pressures and values of V, the relation between i e and 

 temperature of the filament, deduced from its resistance. If 

 the resistance is a true measure of the temperature, and 

 if the saturated thermionic emission is determined by the 

 temperature alone, then the difference between i e when gas 

 is present and i e for the same resistance in a high vacuum 

 should be equal to the current carried to the cathode by the 

 positive ions. But the measurements are not capable of 

 much accuracy, because the thermionic emission varies so 

 rapidly with the resistance ; moreover, it is not certain that 

 the assumptions made are true. The results indicated, as 

 might be expected, that the current conveyed to the cathode 

 by the positive ions is always less than that carried by the 

 electrons emitted, but that the two currents might be of 

 the same order of magnitude. If the current due to the 

 positive ions were equal to that carried by the electrons, 

 the correction to be applied to th<\ values of a in different 

 experiments would vary between 20 per cent, and 60 per 

 cent.; in view of the uncertainty of the data no attempt has 

 been made to apply it systematically, but it appears possible 

 that this error might account for the outstanding discrepancy. 



14. The general conclusions that we would draw from 

 these experiments are two. First, that there is considerable 

 evidence that the production of C0 2 from CO by the dis- 

 charge does take place in the manner suggested, and that 

 the reaction always occurs at a collision of a positive CO ion 

 with a neutral CO molecule. Second, that the "attempt to 

 find a simple relation between the rate of any chemical 

 action caused by the discharge through a gas and the current 

 through that gas is likely to succeed only if the conditions 

 are very carefully chosen. Many experiments have been 

 described by many workers which appear to exhibit Faraday's 

 law of electrolysis in gases; if the reaction is one which 

 place in the gas, and not at the electrodes, this appearance 



