46 PROCEEDINGS OF THE AMERICAN ACADEMY 



The temperatures, T, by the air thermometer, were computed as 

 described ia Professor Rowland's paper on the Mechanical Equivalent 

 of Heat.* The numerical value of v was 0.80 cc. ; of F, 20.21 cc. 

 The themometer was first filled with air dried by calcium chloride 

 and concentrated sulphuric acid. The coefficient of expansion of this 

 air was then measured several times. Six independent measurements 

 ffave values of a = 



These results are sufficiently concordant in view of the small capacity 

 of the bulb of the thermometer, but the average is nearly one per cent 

 higher than the proper value, 0.003670. The cause of this deviation 

 is unknown. It is possibly in jjart due to imperfect drying of the air. 

 Its effect on the resulting temperature measurements will be small, 

 however, since a and T are determined by the use of the same mass 

 of air, and the moisture, if any were present, must have been in so 

 small quantity as not sensibly to influence the results through con- 

 densation, etc. If the deviation be due to any undetected constant 

 error, it will produce but little error in the temperatures obtained, 

 since the air thermometer is used under essentially the same condi- 

 tions for the determination of a and of T. 



The vessel for holding the cold bath, or " freezing mixture," was 

 composed of several concentric glass and metal vessels, the spaces 

 between successive vessels being filled either with dry sawdust or 

 with a freezing mixture, or with these alternating. The arrangement 

 employed when the freezing mixture of salt and ice was used con- 

 sisted of five concentric vessels. Between the outer and the second 

 vessels was a space filled with sawdust. Between the second and 

 third was a space of lai-ger capacity than the former, which contained 

 a mixture of salt and ice. Between the third and fourth vessels was 

 another sawdust layer; and between the fourth and inner vessels, a 

 second fieezing mixture of salt and ice ; while the inner vessel was 

 filled with alcohol, in which the bulbs were immersed and which was 

 thoronglily stirred. A suitable cover of several thicknesses of badly 

 conducting material was provided for the whole. When the liquid 

 nitrous oxide was used, the whole space between the third and a 



* These Proceedings, June, 1879, vol. xv. 



