HYDROGEN INTO ATOMS 161 



The previous calculations of the dissociation of hydrogen ^ were based 

 on the use of the equation 



Wx, = SDgiCi (3)' 



as described in Part I. The equation itself rests on a firm foundation, but 

 the assumptions previously made in applying it to experimental results 

 are open to serious criticism. 



In the first place, it was assumed that the diffusion coefficient D could 

 be calculated v^ith sufficient accuracy from some equations of the kinetic 

 theory by the rather arbitrary choice of a value for the mean free path 

 of hydrogen atoms through ordinary molecular hydrogen. 



Secondly, it was assumed that the shape factor S in the above equation 

 had the same value as that obtained from the data on the convection losses 

 in hydrogen at temperatures below that at which dissociation occurred. 



Thirdly, it was assumed that the hydrogen in the immediate vicinity of 

 the tungsten wire was in chemical equilibrium at a temperature correspond- 

 ing to that of the wire. In other words, Ci was taken to be the equilibrium 

 concentration at the temperature of the wire. 



And finally, the value of qi was found by a method which itself in- 

 volved any error occurring in D. 



Each of the four factors in Equation 3 was therefore liable to error. 

 This was realized and pointed out at the time the calculations were made, 

 but there was then no reason to suspect that any of these sources of error 

 was serious enough to greatly affect the results. 



In the present paper we shall endeavor to avoid these arbitrary as- 

 sumptions and, by a careful analysis of the mechanism of the phenomena 

 occurring in and around the wire, we will attempt to estimate and eliminate 

 the errors which would otherwise occur. 



EFFECT OF CONVECTION CURRENTS 



In the experiments with hydrogen at atmospheric pressure the effect 

 of convection currents in heating the upper part of the bulb was very 

 noticeable. But in the bulbs with 200 mm. pressure of hydrogen or less, 

 the upper part of the bulb was not as strongly heated as the lower part. 

 It is, in fact, well known that convection currents decrease rapidly in 

 intensity as the pressure is lowered, also that in hydrogen they are very 

 much less pronounced than in other gases. If one considers the cause of 

 convection currents it is evident that the behavior of hydrogen is just what 

 one should expect. 



'^ Ibid., 34, 860 (1912). This paper will hereafter be referred to as the "paper of 

 1912." 



^ For the derivation of this and some of the following equations, see Part I, or 

 the "paper of 1912." The numbers of the equations are the same as those used in Part I. 



