168 PHENOMENA, ATOMS, AND MOLECULES 



At the higher pressures the effects of convection currents cannot be 

 neglected, so that Equation 12 would not be expected to hold accurately 

 under these conditions. However, even in the case of convection from 

 small wires in the open air we may profitably look upon the phenomena 

 as consisting essentially in conduction through a film of stationary gas 

 of a certain thickness. If we adopt this view-point, Equation 12 applies 

 even at the higher pressures, but h then no longer represents the diameter 

 of the bulb, but becomes the effective diameter of the film of gas through 

 which conduction occurs. To determine the value of h we need only sub- 

 stitute the values of Wc, d, X and cp^ — qpi in Equation 12 and solve for h. 

 For this purpose the values of qp^ — qpi were calculated from Ta, for the 

 pressures of 760 to 100 mm., and are given in Table V (Col. 5) in brackets. 

 These then served to calculate h from (12). At the lower pressures (50- 

 10 mm.) it was then assumed that convection currents were absent and 

 that h was therefore independent of the pressure and equal to the value 

 3.7 cm., as found at 100 mm. Below 10 mm. h was placed equal to the 

 diameter of the bulb, for in these experiments a straight wire in a cylin- 

 drical bulb was used. 



The values of (pa — cpi at the lower pressures were obtained from (12) 

 by substituting in it Wo, h, d and l. as given in the table. The results are 

 given in Col. 5 (without brackets) and were used to calculate the corre- 

 sponding values of T^ given in the seventh column. 



By comparison of Cols. 3 and 7, it is seen that there is reasonably good 

 agreement between the values of Ta calculated by these entirely inde- 

 pendent methods. The fact that the temperature drop at the surface of the 

 wire (T2 — Ta) is so large, even at the higher pressures, is the reason 

 for the variation of the shape factor (given in Table III at the heads of the 

 columns). 



The accommodation coefficient a is found above to be equal to 0.19. 

 This is considerably lower than the value 0.36 found by Knudsen for 

 hydrogen in contact with polished platinum at approximately roorn tem- 

 perature. 



The results given in Col. 6 of Table V show that the effective thickness 

 of the layer of "stationary" gas around the wire increases as the pressure 

 decreases from 750 to 100 mm., owing to convection currents. At lower 

 pressures, however, the results indicate that there is no further change in h. 

 If the lamps used in these experiments had contained a long, straight 

 filament in the axis of a cylindrical bulb, then b should be equal to the 

 diameter of the bulb. Actually, however, the filament was a short loop 

 only about 3 cm. long. The resulting divergence of the lines of heat flow 

 would tend to increase the shape factor, which would be equivalent to a 

 decrease in diameter of a cylindrical bulb. Hence the value of 3.7 cm. found 



