188 PHENOMENA, ATOMS, AND MOLECULES 



As a test of the theory from which (44) was derived, we may now 

 employ this value of B to calculate Wmax. and by Equation 45 to calculate 

 P', using the values of K from Table XV and the values of E given by 

 (54). For these calculations Ta was taken equal to T2, since the pressures 

 at which W^ reaches a maximum are so high that this must be approxi- 

 mately true. Moreover, the value of Wmax. is only dependent on Ta to a 

 very slight degree. 



The calculated values of Wmax. and P' thus obtained are placed in 

 Table XVT side by side with the experimentally found figures. 



The agreement in the values of Wmax. is surprisingly good, especially 

 when it is remembered that the values of K and E, which were used in 

 the calculation, were obtained solely from the experiments at low pressures. 

 The closeness of the agreement over such a wide range of temperature is 

 strong evidence of the correctness of the value of q that was adopted in 

 calculating K. 



The discrepancy at the highest temperatures may be explained, I think, 

 as being due to experimental errors caused by electrical leakage through 

 the gas. This effect was noticed strongly at the highest temperatures in 

 the experiment at 4.4 mm. pressure. In this case the discharge through the 

 gas entirely prevented measurements above about 2800°, since the voltage 

 actually decreased as the current was raised. Since the resistance was 

 observed in all cases it was possible to show that this effect was not very 

 serious in the other experiments. It is possible, however, that this effect in 

 an incipient stage may have slightly increased the values of Wd at the 

 highest temperatures, at pressures below 200 mm. 



In future experiments this effect could be avoided by using wires of 

 larger diameter so that the potential drop along the wire would be in- 

 sufficient to ionize the gas. 



The agreement between the calculated and observed values of P' in 



