446 PROCEEDINGS OF THE AMERICAN ACADEMY. 



ing the salt in an agate mortar, as had been our practice in the case of 

 sodic sulphate, we crystallized it by rapid cooling. It is necessary to 

 have the salt finely divided by some means in order to stir it successfully. 

 As far as we could discover, the transition temperature was not affected 

 by the size of the solid crystals. 



The ice used in these experiments was a commercial ice of very good 

 purity. Before crushing it was well washed, and thereupon introduced 

 into very carefully purified, boiled, and fully cooled water. Under these 

 circumstances very little ice melted, and accordingly the water around it 

 was very pure. Moreover, the total residue proceeding from a very 

 considerable quantity of this ice alone amounted to only one ten- 

 thousandth of one per cent. Even if the impurity had consisted of sodic 

 chloride or some other substance this small amount could have caused 

 no appreciable effect on the freezing-point ; and practically we found 

 that according to this method of procedure the commercial ice sur- 

 rounded by pure water gave results exactly identical with the purest ice 

 we could make. 



As was to be expected, the preliminary experiments gave results 

 somewhat too low, because of traces of chloride ; for example, salt II, 

 when crystallized once, gave a transition temperature of 50.661° ; when 

 twice recrystallized, 50.670°; when three times recrystallized, 50.671°. 

 It was now almost pure, since the probable value of the true transition 

 temperature is 50 674°. Other specimens gave preliminary values 

 ranging from 50.663° to 50.673°. On the other hand, the salt labelled 

 III, which was known to be of greater purity, showed no difference 

 in the transition temperature between the first and the seventh succes- 

 sive crystallization, giving the constant value 50.674. Having thus both 

 analytical and physico-chemical evidence of the purity of our material, 

 it was therefore safe to make a final series of determinations with all 

 possible care. The data of these are given below. 



Final Determinations of Transition Temperature. 



Three thermometers were used for the final work, as has already been 

 said. The work with the Tonnelot thermometer was carried out with 

 salt III, which had been recrystallized seven times. It is a pleasure to 

 thank Mr. J. B. Churchill for his assistance in carefully recrystallizing 

 the salt and in aiding in the temperature measurement in this case. 

 The data of the measurement are given below. The salt used for the 

 determinations made with the two Baudin thermometers was that 



