40 PROCEEDIXGS OF SECTION A. 



Then Eotvos announces (Wiedeman 27) the remarkable law, 



for nearly all liquids ; water, the alcohols, and the fatty acids are 

 exceptions at the lower ranges of temperature. 



If now we evaluate the attraction of a meniscus bounded by a 

 hemispherical surface on a column of liquid of unit section, and 

 terminating at a distance e from the surface on the hypothesis 

 of a force Aurjr* between the molecules, we find that 



TT A e 



where p is the average density of the surface film. 



We do not know p ; in order, therefore, to deduce from the 



experimental values of a, tlie relative values of A for a number of 



compounds, we make the reasonable assumption that we cannot 



make any serious error if we assume that in all liquids at their 



ordinary boiling points the ratio of p to p, the density in the 



body of the liquid is the same for all substances. And further, e 



is the distance at which we must suppose a homogeneous meniscus 



and a homogeneous column kept apart, in order that their mutual 



attractions may be the same as that of the corresponding meniscus 



and column composed of discontinuous molecules ; e may be 



expected to be proportioned to the cube root of the molecular 



domain. 



^ L 

 Thus, then, we get a = k A. p''^ m'-^ where ^ is a constant, the 



same for all bodies. 



Applying Eotvos's law we find that — {k A m p) is the same 



for all bodies, hence the parameter A varies inversely as the 

 product of the molecular weight, and the modulus of dilatation, 

 as Mendelejeff proposes to call dpjdt. 



Now, if the pressure is small and constant, the external virial 

 can be neglected in comparison with the internal, and we can 

 write approximately — 



Kinetic energy = internal virial. 



It is shewn in my previous paper (Phil. Mag., July, 1887) that 

 the internal virial of the mutual attractions of the molecules in 

 unit mass of a substance, according to the law of the inverse 



fourth power is, tt A p, log. — , where L is a sensible length such 



as the cube root of the volume of unit mass, and a is a length 

 approximately proportioned to the cube root of the molecular 

 domain (usually called molecular volume.) L/a is so large a 

 number that log. L/a may be regarded as constant v,'ithin the 



