Molecular Constitution of Water. 475 



efficient of a l (lS/p 2 ) i is T38. This, then, is decisive, and 

 furnishes our proof that the liquid 1 is dihydrol. 



The exceptional properties of the surface-tension of solu- 

 tions have been discussed in " Molecular Force and the Surface- 

 tension of Solutions'" (Phil. Mag. [5] xl.), and it is now 

 evident that they must be largely due to the difference 

 between the surface-layer and the body of water, and to a 

 possible action of the solute in dissociating some of the tri- 

 hydrol. The solubility of substances in trihydrol may be 

 different from that in water. A re-examination of the surface- 

 tension of solutions would be full of interest. 



5. Latent Heat of Fusion, Specific Heat and Latent Heat 

 of Evaporation. 



According to what precedes, ice is pure trihydrol. The 

 crystallization of water in the hexagonal system is strong 

 confirmation of this ; throughout the enormous variety of 

 forms of ice-flowers the angle of 60° is the dominant factor, 

 and in the theory of halos the ice-crystals of the higher 

 atmosphere appear as necessarily hexagonal prisms. All this 

 points strongly to a decided three-directional symmetry in the 

 molecule of ice to which we shall return in section 8. 



Meanwhile we must regard the latent heat of fusion of ice 

 as no ordinary physical latent heat of fusion, for the melting 

 of ice is accompanied by the conversion of "625 of its trihydrol 

 into dihydrol ; the latent heat of fusion must be mainly a 

 chemical latent heat of dissociation. It is evident that this 

 must be so, because ice in melting contracts by one-ninth ; 

 and if there were no dissociation involved ought to show a latent 

 cold and not a latent heat of fusion. We can calculate what 

 the true physical heat of fusion of solid trihydrol into liquid 

 trihydrol ought to be approximately. In section 3 we took 

 the virial constant for dihydrol to be 1852 X 10 7 with the 

 dyne as unit of force ; and according to the laws of molecular 

 force the value for trihydrol must be nearly the same, as we 

 shall see it to be in a subsequent calculation in this section. 

 If p 2 ' and p 2 are the densities of solid and liquid trihydrol 

 at 0°, then its true heat of fusion in ergs should be l(p 2 —pz), 

 nearly. But p 2 '— P2 has been taken as '0366, and therefore 

 the true heat of fusion in calories of trihydrol expanding on 

 fusion would be 1852 x 10 X -0366/42 = 16 calories. The 

 greater part then of the 80 calories that go to the fusion of a 

 gramme of ice must be used to dissociate *625 gramme of 

 trihydrol into dihydrol and dissolve the remaining *375 gramme 

 in '625 gramme of dihydrol. 



