118 



PROCEEDINGS OF THE AMERICAN ACADEMY 



TABLE XXIII. — Eighth Series. 



From these tables we would draw the inference that No. 6163 

 represents the air thermometer with considerable accuracy. At the 

 same time, both tables would give a smaller value of m than I have 

 used, and not very far from the value found before by direct compari- 

 son, namely, .00015. 



The difference from using in = .00018 would be a little over 

 0°.01 C. at the 20° point. 



All the other thermometers stand above the air thermometer, 

 between and 100°, by amounts ranging between about 0°.05 and 

 0°.35 C, none standing below. Indeed, no table has ever been 

 published showing any thermometer standing below the air ther- 

 mometer between and 100°. By inference from experiments above 

 100° on crystal glass by Regnault, thermometers of this glass should 

 stand below, but it never seems to have been proved by direct experi- 

 ment. The Fastre thermometers are probably made of this glass, 

 and my Baudins certainly contain lead ; and yet these stand above, 

 though only to a small amount, in the case of the Fastre's. 



The Geissler still seems to retain its pre-eminence as having the 

 greatest error of the lot. 



The Baudin thermometers agree well together, but are evidently 

 made from another lot of glass from the No. 6107 used before. These 

 last two depart less from the air thermometer. The explanation is 

 plain, as Baudin had manufactured more than one thousand ther- 



* See note on preceding page. 



