118 



where a is a constant (for a definite substance^. TliU dependence 

 of 0^ on die denslti/ quite cn/rees with that, wliicli in Siippl. N°. 30a 

 was derived for the molectdar translatorj/ motions from the hypo- 

 theses assumed there, ef equation (18/;) of liiat paper. 



This residt can be interpreted as indicating, that the proportionality 

 factor in the rehxtion 



c ^ ?<,V-2, 

 (cf. Siippl. W. 32r^ § 2), in which c represents the vek^cit}- of the 

 "rotational waves" coiisickered in ihe paper mentioned, is independent 

 not only of the tem|)eralnre but also of the density, as according 

 to Sn[>pl. N". 30^ equation (7) is the case for the corresponding 

 "li'anslational waves". 



In Fig. 2 the points indicated l)y small circles represent the values 

 of 0^ derived from the observations as a function of o. The curve 



gives r7f>"/3 , where a is 



chosen so as to obtain agree- 

 ment for the higher values of 

 {J. This agreement is in fact 

 very good for (>^1, as results 

 fi'om the fact that the two 

 curves do not intersect here at 

 a definite value of u, but coin- 

 cide over a certain range of 

 densities. 



For values of q smaller than 

 1 a deviation begins to show^ 

 itself; this deviation at first 

 increases regularly in proceeding to lower values of ^. 



0,i; 



Fig. 2. 



