EQUATION AND THE NATURE" OF COHESION. 



59 



9 are the values given by van Laar. A comparison between colonm 3, 

 where a is computed by the formula 6.5 P e Vf, with the gravita- 

 tional formula in column 8 shows the remarkable agreement between 

 them. There can be no doubt that if these values of a are correct 

 then the law is sustained that cohesion is a function of the mole- 

 cular weight and the number of valences. 



In order to make this remarkable agreement at once apparent 

 to the eye 1 have plotted in Figure 1 the logarithm of the product 

 of the molecular weight by the number of valences on the abscissa 

 against the logarithm of a as given in column 3 of Table 2 as 

 ordinate. I have included also for the sake. of completeness two or 

 three substances in which a is determined solely from the surface 

 tension. Such substances are bromine, diisoamyl, cymene and mesi- 

 tylene; in all other cases except CS 2 the values of a are those in 

 column 3. The actual values of the logarithms used in the plot 

 are given in table 3. I have used in these computations of log 

 (3 f X Val) always the usual valence numbers. Carbon is always 

 tetravalent except in CO, hydrogen univalent, chlorine trivalent, fluorine 

 monovalent, sulphur tetravalent in 80 2 but hexavalent elsewhere, 

 oxygen always divalent, tin tetravalent, and nitrogen monovalent. 

 It will be seen at a glance that with the exception of CO, 2 , C0 2 

 and N 2 the values fall right along a straight line from hydrogen 

 to diisoamyl. There has been no juggling with the valences at all. 

 The valences used have been those ascribed to the elements by 

 chemists. Figure 1 should be perfectly convincing to all that the 

 law I have discovered showing the dependence of cohesion on gra- 

 vitation and the valence electrons is a true law. I have included 

 in this list not only all the normal substances in which the critical 

 data have been accurately determined but even some substances of 

 which the critical data are still somewhat uncertain. 



TABLE 3. !) (p. 60) 

 Values of Log a and Log (31 X Valence) used in plotting figure 1, 



H 2 .. 



N 2 .. 

 2 .. 

 CO. . 

 CB,. 

 N 2 0. 

 C0 2 . 



Log a 



(Log MX Val) 



ValXM 



11.486 



0.6021 



2X2 



12.253 



1.7482 



2 X 28 



12.258 



2.1072 



4 X 32 



12.267 



2.04922 



4 X 28 



12.465 



2.1072 



8X16 



12.690 



2.2455 



4 X 44 



12.640 



2.5465 



8X44 



