436 PROCEEDINGS OF THE AMERICAN ACADEMY. 



It is well known that the presence of impurities affects greatly the 

 transition temperature of a crystallized salt, just as it affects the melting 

 point of a pure substance. Obviously, great purity is demanded if an 

 accurate temperature is to be attained ; hence, from the practical point 

 of view, those salts will be most desirable whose purity is most easily 

 attained, other conditions being equal. Thus the turning point of a 

 research upon this subject is the purification of the materials, just as it 

 is in any other accurate chemical investigation. It is therefore necessary 

 that such a research be carried on in a chemical, and not in a physical, 

 laboratory. 



As has been said, the ease of purification of the salt chosen is one of 

 the important criteria determining the choice, but this is by no means the 

 only criterion. It is important also that the salt should have as large a 

 heat of transition as possible in order that the accidental addition of out- 

 side heat or impurity shall affect the transition temperature as little as 

 possible. Moreover, it is desirable that the salt should not possess the 

 property of forming many hydrates, because there is in this case danger 

 of confusion, and of consequent inexact results. It is also important, 

 although less essential, that the change of volume during the transition 

 should be slight, in order that change of barometric pressure should cause 

 but little difference in the transition temperature. As a rule, however, 

 this last condition is not a very rigorous one, because, even in a case 

 where the change of volume is as much as in melting ice, the changes 

 ordinarily occurring in barometric pressure are barely perceptible on the 

 most accurate thermometers. 



From the point of view of the phase rule the transition temperature 

 of a crystallized salt is a quadruple point, involving the coexistence of 

 four phases. As a matter of convenience, however, it is more satisfac- 

 tory to determine the temperature of another point, almost identical, 

 whose fixity is determined, in the presence of the two components anhy- 

 drous salt and water, by the coexistence of three phases, — hydrated salt, 

 dehydrated salt, and solution, under a definite pressure, — that of the 

 atmosphere plus that of the small layer of liquid above the thermometer 

 bulb. These four conditions (three phases and the pressure) are in this 

 case the n + 2 conditions which determine invariance in the system com- 

 posed of the n = 2 components. Theoretically, as long as a trace of 

 each of these jthases is present the temperature will be constant, but 

 practically it is well to have a fair proportion of each in order to estab- 

 lish the equilibrium as promptly as possible ; supercooling or superheat- 

 ing are thus most satisfactorily avoided. 



