148 PHENOMENA, ATOMS, AND MOLECULES 



pressure) was thus calculated to be 1.2% at 2300° K., and 44% at 3100° K., 

 and 84% at 3500° K. The heat of formation of the molecules from the 

 atoms was given at 130,000 calories (for 2 g. of hydrogen). 



These results seemed to conflict seriously with the measurements of 

 the specific heat of hydrogen obtained by the explosion method. Pier ^ hau 

 concluded from his experiments that the mean specific heat of hydrogen 

 from room temperature up to 2500° K. could be represented by the 

 equation 



Cv = 4.70 + 0.00045^ 



On the other hand, Bjerrum ''' found that the specific heat of water 

 vapor was abnormally large at temperatures above 2000° K. 



The method adopted by Bjerrum was one in which it was not possible 

 to distinguish between variations in the specific heat of water vapor and 

 of hydrogen. An increase of perhaps 10-20% in the specific heat of 

 hydrogen at 3000° would, therefore, be quite consistent with Bjerrum's 

 experiments. 



A dissociation as high at 44% at 3100° K. would, however, lead to an 

 apparent specific heat of Ho two or three times greater than the usually 

 accepted value. 



One explanation of this discrepancy might be that the velocity of the 

 hydrogen dissociation is so small that during the short time of an explosion, 

 equilibrium was not reached, whereas in the other experiments, the cata- 

 lytic action of the hot tungsten wire caused the dissociation to approach 

 the equilibrium more closely. 



It seems, however, improbable that the velocity of the reaction would 

 be so slow at temperatures as high as 3100° K. Bjerrum's experiments, 

 therefore, seem to indicate that the degree of dissociation is considerably 

 less than the values obtained by the methods described above. 



If we examine the chain of reasoning by which the results were obtained 

 we see that there is one extremely weak link, namely, the method by which 

 the diffusion coefficient D was obtained. 



It is .true that this method would give a fair degree of accuracy if ap- 

 plied to almost any pair of ordinary gases at ordinary temperatures, but a 

 good deal of uncertainty arises when it is applied to a dissociating gas. In 

 fact, under such conditions, the diffusion coefficient might be very much 

 greater than that calculated in the usual way. An example of such a case 

 is already known in the abnormally great mobility of the hydrogen and 

 hydroxyl ions in aqueous solutions. A larger value for the diffusion co- 

 efficient would lead to lower values for the dissociation and might thus 



^ Z. Elektrochem., 15, 536 (1900). 

 'Z. physik. Chem., 79, 513 (1912). 



