80 THE ROYAL SOCIETY OF CANADA 



As the value of a is small, slight variations in the shape of the 

 curves have considerable effect on its value. Hence very close agree- 

 ment in the individual values of a could not be expected, and the 

 variations shown in Table IV may be considered as reasonable for this 

 method. 



Some recent determinations of a by other observers over the range 

 0° to 100°C. are given below: 



Eckerlein, P. A., Ann. der Phys., 3, p. 120, 1900 -00362 



Schwarze,W., Ann. der Phys., 11, p. 303, 1903 -00253 



Pauli, E., Ann. der Phys., 23, p. 928, 1907 -00197 



Eucken, A., Phys. Zeits., 12, p. 1101, 1911 -00271 



Mean of these values -00271 



Of these results the most trustworthy is that of Eucken, which 

 agrees with the mean of the four. 



It will be seen that the result of the present determination is in 

 good agreçment with the mean of determinations by other observers. 

 This indicates that the method described in this paper will give 

 satisfactory results if the radiation corrections are relatively small, 

 i.e. if the temperature range is not too large. The range, however, 

 may be between any two temperatures which experimental arrange- 

 ments would permit, so that the temperature variation of thermal 

 conductivity over a wide range of temperature might be found in a 

 series of steps. 



In conclusion, the writer wishes to express his thanks to Dr. L. V. 

 King for placing the facilities of the laboratory at his disposal and for 

 directing this research. He also wishes to express his gratitude to the 

 Honorary Advisory Council for Scientific and Industrial Research 

 of Canada, under whose auspices the earlier part of this work was done. 



McGill University, Montreal, Canada. 



