64 THE TRUE VALUE OF a OF VAN DER WAALS' 



this is known many obscure physical and chemical phenomena will 

 perhaps become as clear as day. 



On the other hand nearly all of van Laar's values in column 9 

 of Table 2 are seen to be hopelessly wrong. Upon these values he 

 has sought to base such far-reaching conclusions as that sometimes 

 atoms contribute to the cohesion and sometimes they do not, as 

 for example carbon in methane and tin in stannic chloride, which 

 are not supposed by van Laar to play any part in the cohesion. 

 This conclusion is thus shown to be entirely unfounded. Sometimes 

 it is true his values of a happen to come somewhere near the 

 real value of a , but usually they are 10 — 20% low and in some 

 cases, as in hydrogen, they are at the very least 100 °/ too low. 

 There can be no doubt, I think, that the method which van Laar 

 has used is not a reliable method. Most of the arbitrary values he 

 has assumed for the various elements by the addition of which he 

 hopes to compute the square root of a are, therefore, erroneous. 



We may now consider very briefly the seeming exceptions. They 

 are, as is usual Avith exceptions, exceedingly interesting and illumi- 

 nating. Of the 44 substances three, or possibly four, have been 

 found to be, apparently at any rate, marked exceptions to the rule. 

 These substances arc 2 , CO, N 2 and C0 2 . We might add to 

 them S0 2 , since this would be exceptional if the sulphur is hexa- 

 valent, as we have found it to be in other compounds. It is not 

 exceptional, however, if we consider the sulphur here to be tetra- 

 valent, as most chemists believe it to be. In all of these compounds 

 we find fewer valences than we had expected. To make them con- 

 form to the law we have to take in 0., two, instead of four 

 valences; in CO, two in place of four; in C0 2 five in place of 

 eight; and in S0 2 eight instead of ten, if the sulphur is hexavalent. 

 It will be noticed, further, that these substances are all very simple 

 substances containing only two or three atoms to the molecule. This 

 might raise the suspicion that our methods of calculating a in 

 these simple substances were erroneous and that a should really be 

 larger; but that this suspicion is unwarranted is shown by the fact 

 that the equally simple substances hydrogen and nitrogen are cor- 

 rectly determined by these same methods, and also by the fact that 

 in S0 2 , at any rate, the calculation from the surface tension and 

 latent heat shows that the values of a ascribed to S0 2 are correct. 

 There can be no doubt that in C0 2 the value of a is correctly 

 determined. Their exceptional position is not due, therefore, to any 

 mistake in the value of a. It will be noticed also that at least two 

 of these substances, namely S0 2 and C0 2 , are what may be called 



