110 PROCEEDINGS OP THE AMERICAN ACADEMY 



The ratio of C for G167 and G1G3 is found in the same way. 



Weight. 



1 Nov. 14 .0310091 



2 Nov. 20 .0309846 



Mean .0309928 



Hence for G163 we have in this way 



C = .057381 C = .056995 m = .000187. 



By direct comparison of No. 6163 with the air thermometer, we 



find the following. 



ight c. m. 



.000239 

 .000166 

 .000226 

 .000155 

 .000071 

 .000115 



000004 .000154 ± .000010 



The values of C agree with each other with great exactness, and 

 the probable error is only ±0.°003 C. at the 40° point. 



The great differences in the values of m, when we estimate exactly 

 what they mean in degrees, also show great exactness in the experi- 

 ments. The mean value of m indicates a difference of only 0°.05 

 between the mercurial and air thermometer at the 20° point, the 

 0° and 40° points coinciding. The probable error of m in degrees is 

 only ±0°.003 C. 



There is one more method of finding m from these experiments ; 

 and that is by comparing the values of C with No. 6167, the glass of 

 6167 being supposed to be the same as that of 6163. 



We have the formula 



0= C (1 + 34.8™). 



Hence 



