34 



ROYAL SOCIETY OF CANADA 



From our experience with the use of freezing point mixtures made 

 from Montreal water and pure freshly fallen snow or clean river ice 

 for accurate temperature measurements, it seemed impossible that a 

 divergence of as much as 1-1000 of a degree could exist. It is probable 

 that the divergence from absolutely pure water and ice is very much 

 less than this judging by the order of agreement attained in the calibra- 

 tion of very sensitive platinum thermometers. 



Calibrations have been carried out in this laboratory repeatedly 

 using fresh snow and river water, which have agreed to very much 

 better than 1-1000 of a -degree. We considered that a mixture prepared 

 with great care from these materials would not differ from the absolute 

 zero point on the centigrade scale by more than a few ten thousandths 

 of a degree. The primary reason for this investigation was not to study 



<X) 



v^ 



Fig. 1 



the real cause of the gradual increase in the volume of ice, but to 

 study the variation in the density of ice mantles prepared by different 

 means. When we came to work with the ice calorimeter we found the 

 increase in reading was so rapid that we were obliged to undertake a 

 study of this point before going further. It was quite impossible to 

 arrive at any conclusions in the point we had in mind before we had 

 become thoroughly conversant with this peculiar phenomnon. 



