158 PHENOMENA, ATOMS, AND MOLECULES 



In the previous paper, S was calculated from results obtained with gases 

 other than hydrogen. This method of calculation of S, although giving an 

 approximation good enough for most purposes, may be open to criticism. 

 In the present case, therefore, it has been thought better to determine S 

 directly from the experimental results of Tables I and II. 



The values of cp2 — qpi were taken from the data given in the previous 

 paper,^^ in which qp was calculated from the heat conductivity of hydrogen 

 as determined by Eucken^^ and from the measurements of the viscosity of 

 hydrogen by Fisher. ^*^ 



By trial a value of S was found, which, when multiplied by cp2 — (pi, 

 would give the best agreement with the values of W observed at tempera- 

 tures so low that Wx» was negligible. At higher temperatures, Wd was 

 obtained by subtracting S((p2 — cpi) from W. The logarithum of W^ was 

 plotted against i/T from the data at rather low temperatures (1700- 

 2500° K.). It is only in this range of temperatures that the calculated 

 values of Wd are materially afifected by errors in the value S. By trial a 

 value of S was finally chosen which would give most nearly a straight line 

 relation between log Wd and i/T. 



These values of S have been used to calculate Wd and the results are 

 given in Table III. 



The results in Table III were obtained from the data in Table I and II, 

 merely by subtracting the product of S (given at the head of each column 

 of Tal)le III) by cp2 — <^i (given in second column of Table III). 



The data on the heat losses from tungsten wires in nitrogen given in 

 Table I, was used in a similar way to calculate W — Wp and the results 

 are given in Table IV. The values of (po — qpi in this table are calculated 

 from the viscosity and heat conductivity of nitrogen. ^'^ 



A comparison of Tables III and IV shows clearly the very great differ- 

 ence between hydrogen and nitrogen. With nitrogen the heat loss in- 

 creases with the temperature very nearly proportionally to qp2 — qpi. The 

 differences observed and tabulated are not greater than the possible ex- 

 perimental errors. In the case of hydrogen, up to temperatures of 1500- 

 1700° K., the results are similar to those of nitrogen; that is, there is no 

 evidence that the differences between W and Wp are anything more than 

 experimental errors. But at very high temperatures Wd increases very 

 rapidly, and becomes 50 or 100 times as large as the possible experimental 

 error. 



In the second part of this paper, these data will be used to calculate the 



^* Physic. Rev., I. c. 

 ^^ Physik. Z., 12, iioi (191 1 ). 

 "^^ Physic. Rev., 24, 385 (1907). 

 ^' See Physic. Rev., I. c. 



