[BARNES A LUCAS] GROWTH OF ICE CRYSTAL 33 



The form of calorimeter we employed is the usual one, and shown 

 in Fig. 1, The bulb was filled with boiled distilled water, and sub- 

 sequently closed with pure freshly distilled mercury. The expansion 

 of the mercury took place along a horizontal capillary tube fitted 

 to a three-way tap, so that the thread could be brought to any con- 

 venient point on an attached scale. 



The capillary tube was 35 cm. long and had a volume of .0006619 

 c.c. per millimetre. The height of the capillary tube above the water 

 was 35 cm. Thus the water in the calorimeter was under a pressure 

 of little less than one-half an atmosphere which lowered the freezing 

 point appreciably below the surrounding mixture. Even allowing a 

 temperature in the mixture from impurity of as much as 1-1000 of a 

 degree below freezing, the water in the calorimeter would still lie over 

 2-1000 lower than that. It seemed impossible to conceive of any pro- 

 gressive freezing going on in the ice mantle from the fact that the 

 ice mixture was below the water in the calorimeter. 



A large copper vessel was used to hold the freezing-point mixture, 

 which was very carefully scoured and was perfectly clean. 



The calorimeter could be completely buried up to the three-way 

 tap in the mixture. The top and sides of the vessel were lagged heavily 

 with thick felt and cotton wool. The whole apparatus was kept in a 

 room at about 10° C. so as to preserve the mixture as long as possible. 

 As the observations extended over several days the mixture was repaired 

 at intervals of every few days with fresh snow or cracked ice. Four 

 sets of experiments were made, which agreed very well. In one set, 

 clean freshly fallen snow was selected and moistened with river water, 

 and in the other sets clean cracked ice was used and moistened with 

 distilled water or river water. The ice mantle was prepared in two 

 cases by pouring into the inside tube a little liquid air after bringing 

 the water in the calorimeter to the freezing point. In two cases 

 a mild refrigerant was used in the shape of a salt mixture. 



In all four sets a rapid increase of reading set in which persisted 

 for the entire time of observations. 



One of the sets obtained with a mantle frozen by means of salt 

 and snow comes between the first and third set where liquid air was 

 used. The fourth set was the same as the second, and did not extend 

 more than a few hours, and is not therefore recorded here. 



First Experiment. 



A mantle about 8 m.m. in thickness was formed with liquid air and 

 the calorimeter was then packed in a mixture of pure snow and water. 



