of G 



ns 



the temperature of the room was to show that the temperature 

 was the same as for /<//<//"</' /<. 



In the former paper the results were reduced so as to show to 

 what power () of the absolute temperature the viscosity was propor- 

 tional. 



Since practically only two points on the temperature curve were 

 examined, the numbers obtained were of course of no avail to deter- 

 mine whether or no any power of the temperature was adequate to 

 represent the complete curve. The question of the dependence of 

 viscosity upon temperature has been studied by Sutherland,* on t he- 

 basis of a theoretical argument which, if not absolutely rigorous, is 

 still entitled to considerable weight. He deduces from a special form of 

 the kinetic theory as the function of temperature to which the 

 viscosity is proportional 



c, being some constant proper to the particular gas. The simple law 

 (ft, appropriate to " hard spheres," here appears as the limiting form 

 when 6 is very great. In this case, the collisions are sensibly unin- 

 fluenced by the molecular forces which may act at distances exceeding 

 that of impact. When, on the other hand, the temperature and the 

 molecular velocities are lower, the mutual attraction of molecules 

 which pass near one another increases the number of collisions, much 

 as if the diameter of the spheres was increased. Sutherland finds a 



D, .......... 5895-0 



D, .......... 5889-0 



Dj .......... 5875 -9 



D b .......... 5849 -6, 



HO that the above-mentioned intervals would be as 19*1 : 26*3. [June 23. Subse- 

 quent observations with the aid of a scale showed that the intervals above spoken 

 of were as 20 : 21. According to this the wave-length of the line seen, and sup- 

 posed to correspond to D ( , would be about 5855 on Rowland's scale, where I), 

 58962, D, = 5890-2, D 3 = 5876'0.] I may record that the refractivity of the 

 gas now under discussion is 0*132 relatively to air. 

 ' Phil. Mag.,' vol. 36 (1893), p. 507. 



