108 



CONDUCTION OF HEAT 



295 



The values of the magnitudes measured at which have 

 been employed for the calculation of these ratios have also 

 been taken from Wiillner's book and are tabulated below. 

 Where Wiillner has given several values I have put 

 down their mean modified by taking K = 1-6027. 



Values at 0. 



The last two rows contain the values t w , 1 G , observed by 

 Winkelmann and Graetz respectively. In the case of 

 three of the gases investigated by both observers, viz. air, 

 hydrogen, and carbonic acid, the theoretically calculated 

 value lies between the two observed values ; we may hence 

 assume that the differences still left may be explained 

 merely by the errors of observation, and that they cannot 

 be charged to a defect in the theory. 



In Wiillner's table there is only mercury vapour which 

 forms any notable exception. This single exception, how- 

 ever, cannot raise any doubt as to the validity of the law 

 when we consider the uncertainty of the basis of the calcu- 

 lation in this case. The specific heat c of the vapour has 

 not been directly measured, but has been theoretically cal- 

 culated by the rule given in 56, from the ratio of the two 

 specific heats and the density deduced from the atomic 

 weight. The coefficient of viscosity ?? has been calculated 

 from the observations of S. Koch, 1 at 273 and 380, and 

 also for the much lower temperature 203, for which 

 Schleiermacher 2 has determined the conductivity f. W T e 

 may assume that such a process cannot give any certain 

 result, especially if the magnitudes used in the calculation 



1 Wi&d. Ann. 1883, xix. p. 857. 2 Ibid. 1889, xxxvi. p. 346. 



