184 Union of Hydrogen and Oxygen at Low Pressures, 



the electrodes, break up into new ions those molecules of the 

 gas which they strike under certain conditions. To do this, 

 the velocity of the ion at impact must exceed a certain 

 magnitude ; and, possibly also, the position in the molecule 

 of the point of impact must satisfy some condition. 



Now this reproduction of ions by collision is probably con- 

 fined, at least approximately, to the region of the cathode 

 fall, close to the cathode, where the electric force is intense 

 enough to secure to a sufficient number even of positive ions 

 an adequate velocity of impact (see Townsend, Phil. Mag. 

 Dec. 1904, p. 751). 



Probably, then, nearly all the negative ions which are 

 formed by collision cross approximately the whole distance 

 D between the plates, and this must be accurately true of 

 any other negative ions in the current, for they are discharged 

 from the cathode. 



The second term of our last equation, which is, numerically, 

 of much the most importance, is proportional to the electrical 

 work done in separating the pair of ions and in bringing them 

 to the electrodes. This work is accounted for by the forma- 

 tion of new ions and by the energy imparted to those mole- 

 cules with which the pair collide, for the energy imparted by 

 impact to the electrodes is probably a negligible fraction 

 of it. 



Possibly, then, the second term chiefly represents the 

 activity of dissociation within the region of the cathode fall, 

 and due, partly or wholly, to the positive ions. 



In any case, however, if we assume that the atomic 

 dissociation which must precede the formation of water is 

 effected by the motion of ions, it is evidently easier for a 

 colliding ion to dissociate a molecule into atoms than to 

 decompose it into ions ; for, as we have seen, from 6 to 9 

 times as many molecules of water as pairs of ions were 

 always formed. 



The curve here obtained does not agree with the curve 

 given in my former paper (Phil. Mag. Feb. 1904). It should 

 be mentioned that the pressures there given are all 3" 5 per 

 cent, too small, owing to an error in standardizing the 

 McLeod gauge. The distance between the electrodes was 

 •97 cm. Therefore, to compare that curve with the one here 

 given, the abscissas and ordinates of the former curve must 

 be increased by 6*5 per cent. In addition to this, since the 

 present apparatus had just twice the capacity of the former 

 one, the ordinates of the former curve, thus slightly corrected, 

 must be halved. 



A comparison made in this manner shows a marked 



