Electric Discharge in rarefied Hydrogen and Oxygen. 309 



down a potential fall of 5 volts. Now, the experiments 

 recorded in this paper and previous papers prove that the 

 breaking up of a molecule into atoms needs the expenditure 

 of less work than the breaking o£ it up into ions, for the ions 

 are generated by the collisions of the moving ions with the 

 molecules of the gas, and at least 5 times as many molecules 

 of water were always formed as pairs of ions. Hence, on 

 these grounds we can fix 5 x 10"" 12 ergs, which is practically 

 the same figure as before, as a superior limit to the work 

 involved in separating the atoms of a molecule of oxygen. 



Very little Ozone present. 



Nothing has so far been said about the formation of ozone 

 during the discharge, although it is known that that gas is 

 always present. Some of the atoms dissociated by the 

 colliding ions group themselves into molecules of ozone, but 

 the number of these molecules can be safely neglected 

 beside the number of molecules of water which are formed. 

 For it has been repeatedly verified in the course of these 

 experiments that the same results are obtained whether one 

 uses freshly prepared hydrogen and oxygen or those gases 

 after they have been already used till more than half has been 

 made to combine into water. Now r , if any appreciable quantity 

 of ozone were formed, the error involved in the calculation 

 by neglecting it could not in such circumstances escape 

 notice. 



Possible influence of the form of Apparatus. 



A few of the observations recorded in this paper were 

 made under conditions similar, in respect of pressure and 

 distance between electrodes, to observations contained in my 

 List paper on this subject. The latter were approximately 

 covered by the formula 



w pD 3X, 



W =%-+ 800 ^ ame 110tatl011 )> 



and with the present apparatus W = 3Ap/AQ. It is therefore 

 possible to make some comparisons between the two sets of 

 observations. In the case of the smallest distance "25 cm. 

 the agreement is good enough, but in the other cases 

 discrepancies occur which cannot be accounted for by errors 

 of observation, since they reach in two cases about 30 per 

 cent. Now I have shown that these chemical effects are not 

 dependent upon the nature of the electrodes, so that one is 

 forced to conclude that the form of the apparatus plays a 



