40 



PROCEEDINGS OP THE AMERICAN ACADEMY 



the Baudiii thermometers, both from its quite considerable calibration 

 error, which is irregular between the 95° and 100° points, and from 

 the lack of the calibration chamber at the top, as well as from the 

 general form of the tube and bulb. The Kew register was also 

 obtained, but was not used, as it was given only to 0°.l, i. e. half- 

 divisions of the scale, and corresponds neither with the result of the 

 comparison between Kew Standard No. 104, as given below, with the 

 ascertained calibration error, nor with the air thermometer error. 

 The pair of Fastre tubes, kindly lent for the comparison by Di\ Wol- 

 cott Gibbs of Harvard University, were of special interest, as being 

 probably nearly identical with the crystal glass used by Regnault and 

 calibrated by the method employed by him, Fastre being the maker of 

 many of Regnault's thermometers. 



The chief special points of interest in the results of these compari- 

 sons are indicated by Prof. Rowland at the place cited, and I will 

 allude here only to the fact that all of the mercurial thermometers 

 used in this study appear to read higher than the air thermometer 

 below 100°. 



TABLE X. 



* The original readings in ice were 58.08 and 58.45, to which .15 was added 

 to allow for pressure of water in the comparator. Tliis, of course, gives tlie 

 same final result as if .15 were subtracted from each of the other temperatures. 

 No correction was made to tlie others. 



t Probably some error of reading. 



