100 



PROCEEDINGS OF THE AMERICAN ACADEMY 



about 1 in 1,000, assuming that the temperature t was the same at the 

 determination of the zero point, the 40° point, and the 100° point. 

 It seldom varied much. 



The coefficient of expansion of the glass influences the result very 

 slightly, especially if we know the difference of the mean coefficients 

 between 0° and 100°, and say —10° and +10°. This difference I at 

 first determined from Regnault's tables, but afterwards made a deter- 

 mination of it, and have applied the correction.* 



The table given by Regnault is for one specimen of glass only; and 

 I sought to better it by taking the expansion at 100° from the mean 

 of the five specimens given by Regnault on p. 231 of the first volume 

 of his Relation des Experiences, and reducing the numbers on 

 page 237 in the same proportion. I thus found the values given in 

 the second column of the following table. 



TABLE VIII. — Coefficient of Expansion of the Glass of the Air 

 Thermometer, according to the Air Thermometer. 



The second column contains the values which I have used, and one 

 of the last three columns contains my experimental results, the last 

 being probably the best. The errors by the use of the second column 

 compared with the last are as follows : — 



■ro^ from using b m — b i0 = .0000008 instead of .0000011 ; 



tgW from usin S *ioo 

 or, ^ 6XJ for both together. 



= .0000264 instead of .0000287 ; 



* This was determined by means of a large weight thermometer in which 

 the mercury had been carefully boiled. The glass was from the same tube as 

 that of the air thermometer, and they were cut from it within a few inches of 

 each other. 



t Relation des Experiences, i. 328. 



t Pogg. Ann., cxxiii. 135. § Experimental Pliysik, Wiillner, i. 67. 



