66 DKS. J. A. HARKEi; AND P. CHAPPUIS ON A 



The readings of the mercury thermometers are made by means of a .small vertical 

 reading telescope sliding on the glass plate, and can be made while the stinvr is at 

 work. The space around the inner tank is closed above by a metal lid, pierced with 

 the necessary openings for the axes of the stirrers. 



Observations with Thermometer K.8. 



The thermometer K.8 was compared under these conditions with the four standards 

 at temperatures between 0and 50. These experiments are numbered 4 to 17 in the 

 summary of results for K.8 at the end. To avoid the errors of parallax on the 

 mercury thermometers, ten readings were made with the divisions upwards and ten 

 with the thermometers turned through 180. After each series the zeros of the 

 mercury thermometers were observed in the usual manner with a micrometer 

 telescope. 



Three other observations at higher temperatures (numbered 18 to 20 in the table) 

 were made with the same thermometers in another apparatus, described later when 

 treating of the gas thermometer. In the three experiments all the instruments were 

 used in the vertical position. 



Care was taken to have only a very small emergent column in each case. The 

 bath was filled with oil, and was heated by the vapour of ethyl alcohol boiling under 

 various pressures. 



We also made three measurements below in an apparatus specially constructed 

 for experiments at low temperatures, which has been described in the 'Proces-verbaux 

 des Stances du Comite International,' 1891, page 33. The thermometers plunge into 

 a bath of alcohol cooled by the evaporation of liquid methyl chloride, and stirred 

 continuously by a suitable mechanism. The two mercury standards, Tonnelot ther- 

 mometers Nos. 11,165 and 11,166, which were employed for these observations, have 

 Ijeen studied at the Bureau, and compared directly with the hydrogen thermometer 

 under the same conditions. 



The series we made consists of three experiments numbered 1 to 3 in the Summary. 



Observations ivith Thermometer K.9, 



The later series of comparisons of thermometer K.9 with the mercury standards 

 was made under precisely similar conditions to those described above for K.8, but 

 the number of different points in this case was not so great ; each experiment con- 

 sisted of only ten observations instead of twenty as before. The experiments between 

 and 50 are numbered 1 to 6 in the table. 



