CHEMISTRY: W. D. HARKINS 
545 
principle the surface of a metal is not considered as a region of the 
same class as the surface of water or an organic liquid, since the 
characteristics of the two regions are quite unlike. 
Corresponding to the various empirical relations the following 'normal' 
values of the entropy may be given: 
Entropy in ergs 
per degree per 
molecule 
1. Liquid to surf ace 2.96 
2. Liquid to vapor at the special concentration of: 
c = 0.00507 mols per liter 18.8 
c = 0.0127 mols per liter 16.7 
c = 0.0201 mols per liter 15.7 
(These values become less accurate as the concentration of the vapor in- 
creases.) 
3. Solid to vapor at the melting point 21.0 
4. Solid to liquid 9.0 
5. Solids dissociate to 760 mm. vapor pressure 22 . 0 
Of these the first is the most exact, the second holds moderately well 
under the conditions imposed, and the fourth, as might be expected, is 
one of the least accurate. Walden's rule is that the molar heat of 
fusion divided by the temperature is equal to 13.5 calories per degree 
for normal substances, or the molar entropy of fusion has the given 
value. This rule Walden^ found to hold for a large number of organic 
substances. When the data did not correspond with what should be 
obtained according to the rule, Walden assumed that it still remains 
valid, but that the molecular weight is different from that given by the 
formula. However it is evident that this explanation is not sufficient 
to account for all of the deviations which exist. 
At my request Mr. L. E. Roberts has studied practically all of the 
available data on the entropy of melting, and has found the general 
relations which hold. One of the greatest obstacles in this connection 
is that the data are in many cases extremely inaccurate. They indicate 
that the latent heat of melting of a metalHc element which crystallizes 
in the regular system, increases as the melting point rises, and the entropy 
averages about 2.2 calories per gram atom per degree.^ The salts show 
a somewhat similar relation, and at the same time there seems to be a 
general increase in the entropy of fusion with the number of gram 
atoms in the formula weight of the salt. There is a great deal of irregu- 
larity, hut for the halogen salts the entropy is of the order of 2.2 calories 
per degree per gram atom, or about the same value as is found for the 
metals.^ Hydrogen, hydroxides and water of crystalHzation are repre- 
sented by lower values. Thus the entropy of fusion of the hydroxides 
of sodium, potassium, rubidium, and caesium, is about 2.8 calories per 
