Chemical Attraction between Atoms from Physical Data. 95 

 of the function cf>J — f ) can then in the same way be deter- 

 mined by means of the first or second of the above three 

 equations. If we further suppose that </> 3 {— J may be ex- 

 pressed in the form i—j ; we have 



n-l »-l T n±l 



L = ^)P!(pi 3 -p. 3 ), A=^( r )P 2 (p!~^) 3 , 

 and therefore 



/ n-l n-l\ 



- =Jl 3 »+i ' 



* (Pi-p 2 ) 2 



where P a is a function of n. Applying this equation to the 

 facts, we find that n lies between 7 and 11. This makes the 



value of <pA7Tr) increase with the temperature, in other 



words, the molecular attraction increases with the tempera- 

 ture. But this is highly improbable ; moreover, we will 

 show in a subsequent paper that n cannot be as large as 

 he above value, and it follows therefore that strictly the 

 arbitrary function cannot be expressed as the product of two 

 factors as we have supposed. 



One of the methods used to obtain some information on 

 the law of molecular attraction is based on the change of the 

 coefficient ot viscosity of a gas with change of temperature, 

 making the assumptions that the attraction is independent of 

 the temperature and that the molecules behave simply as 

 centres of force. But this cannot give accurate results, as 

 the attraction depends on the temperature, and moreover the 

 results must be linfluenced by the fact that a molecule has 

 a certain definite volume. If the attraction betwesn two 



molecules is supposed to be given by the expression — , the 



values of n vary between the limits 5 and 12 according to 

 the nature of the substance *. According to the investiga- 

 tions of the writer, tiie law of attraction is the same for all 

 substances. If the attraction is a function of the tempera- 

 ture, then according to the law obtained by the writer the 

 value of n found from viscosity data should depend on the 

 extent the temperatures at which the viscosity measurements 

 have been carried out are removed from the critical tem- 

 peratures. This is probably the explanation why the values 

 of n increase in the order as the critical temperatures of the 

 * Jeans' •' Dynamical Theory of Gases/ p. 2b7. 



