41 6 PROCEEDINGS OF THE AMERICAN ACADEMY § 1 6 



weight are known, and also the quantivalence of the vari- 

 ous atoms which compose the molecule. 



To present the results of the theor}' in the clearest form, 

 and in the smallest possible compass to compare them with 

 those of observation, Table XVIII. has been constructed; 

 the old-fashioned names of the substances, which are ex- 

 clusively liquids, are placed in the first column; the most 

 modern s3'mbols follow, and to still further identify the 

 liquids, the density, boiling point and melting point are 

 added, mostly from the determinations of Kopp and 

 Pierre. The value of the Principal Argument, /' -r- /, is 

 then tabulated in three columns, calculated respectively 

 from the expansion, the critical temperature and the latent 

 heat.* The last column contains references to the tables 

 or sections where the liquid in question has received a 

 special examination, so that the comparison may easily be 

 made. 



The truth of the theory is illustrated by the e^eneral 

 agreement of the values of the Principal Argument calcu- 

 lated from different data; its practical use is limited only 

 by the accuracy w^ith which, from the most probable value 

 of the Argument, the various constants may be derived. 



§ 1 6. The treatment of the specific heat of gases, the 

 tension of vapors and the expansion of liquids subject to 

 a change of state involves the use of the theory of proba- 

 bility, which is foreign to the purpose of this paper. It 

 seemed, however, desirable to show that there was nothing 

 in these phenomena necessarily inconsistent with the law 

 which we have assigned to the variation of the force of 

 cohesion. 



The similarit}' between a velocity in the kinetic theory 

 and an accidental error in the theory of probability is seen 

 from various considerations. Knowing onl}', in any spe- 



* Calling unity the value of A'. 



