HYDROGEN INTO ATOMS 169 



above for h is in as close agreement with the actual bulb diameter (7 cm.) 

 as the experimental conditions would seem to warrant. 



Nitrogen. — Similar calculations from the data given in Table I, for a 

 filament at a temperature of 2400° in nitrogen yield the results shown in 

 Table VI. The values of To in Col. 3 were obtained from a formula similar 

 to (11) in which, however, the coefficient 49.6 was replaced by 166, corre- 

 sponding to the different molecular weight and specific heat of nitrogen. 

 The columns of Table VI correspond exactly to those of Table V. 



It is seen by the variation of h that the effect of convection is much 

 more marked than with hydrogen and persists at pressures even as low as 

 50 mm. 



TABLE VI 



Wire in Nitrogen 



The accommodation coefficient is much larger than that of hydrogen, 

 but is again less than that given by Knudsen (for air) at room temperature. 



The agreement between the values of To calculated by the two inde- 

 pendent methods, in both Tables V and VI, is sufficiently good to show 

 that the general theory of heat conduction and convection used in making 

 the calculations can safely be applied to the case of fine tungsten wires in 

 hydrogen and nitrogen at all pressures. Evidently the heat conductivities 

 of these gases, within the ranges of temperature considered above, vary 

 with the temperature in the way that is to be expected from their behavior 

 at ordinary temperatures. 



CONCENTRATION DROP AT THE SURFACE OF THE WIRE 



We have seen that the temperature of the layer of gas next to the wire 

 may be very much lower than that of the wire itself. To assume that 

 these two temperatures are equal, which amounts to assuming temperature 

 equilibrium at the surface of the wire, would lead to entirely erroneous 

 results. 



The diffusion of one gas through another is a phenomenon closely 

 related to heat conduction. In the case of the evaporation of a solid 



