438 PROCEEDINGS OF THE AMERICAN ACADEMY. 



of melting of corresponding quantities of several single component sys- 

 tems ; of water, the latent heat of melting of a gram-molecule is 5.8 

 kilojoules ; and of bromine, 5.4 kilojoules ; of iodine, 6.3 ; of phosphorus, 

 0.6 for each gram atom. It is apparent from this comparison that, even 

 taking into account the fairly large molecular weight of sodic bromide 

 (139), the heat of transition compares favorably even with the best single 

 component systems. 



It becomes now a matter of great interest to calculate as nearly as 

 possible the effect of impurity in the sodic bromide upon the transition 

 temperature. Sodic chloride may be taken as an example, being the 

 most probable impurity. Van't Hoff has pointed out, and Lowenherz 2 

 has shown practically, that the transition temperature of a crystallized 

 salt is lowered by the addition of a foreign substance just as a true melt- 

 ing point is lowered. Lowenherz found the molecular lowering to be a 

 constant when the foreign substance added was not an electrolyte, and 

 very nearly a constant in the case of other sodium salts. Undoubtedly 

 the reason for this latter fact is the slight ionization of the added sodium 

 compound in the presence of the large concentration of ionized sodium 

 already dissolved in the saturated solution. Precisely the same condi- 

 tions apply here ; therefore it is safe to assume that sodic chloride, if 

 present in the solution, would act essentially as an undissociated substance. 

 If, now, the phenomenon of transition is viewed simply as the melting 

 of the hydrate, a simple thermodynamic cycle leads to the approximate 



equation A = -jr T — r— for the heat of transition of the crystallized salt, 



in which n and ^represent respectively the numbers of moles of impur- 

 ity and solvent salt, and At the change in temperature of transition 

 caused by the admixture, n being supposed to be very small in propor- 

 tion to N. Substituting the value of A resulting from the calculation 

 given just above and the molecular weights, it is easy to find that the 

 lowering of the transition point produced by one per cent by weight of 

 this impurity will be about a degree. In other words, the lowering of 

 the transition temperature caused by an impurity of sodic chloride in 

 sodic bromide should be in thousandths of a degree approximately equal 

 to the per cent of impurity expressed in thousandths of a per cent. That 

 is, an impurity of, for example, 0.002 per cent would cause a depression 

 of about 0.002°. This, of course, is only a rough calculation for several 

 reasons. In the first place, the temperature of transition is 50.7°, not 



2 Zeit. phys. Chem., 18, 70 (1895). 



