2 Messrs. Thorpe and Bucker's Remarks on 



IV. Poggendorff's correction. 



V. Comparison of the mercurial with the air-thermometer. 

 VI. Effect of compression. 



In the present paper we confine our observations to the 

 second and third of these points; in a subsequent communica- 

 tion we may have something to remark on the others. 



The Exposure Correction. 



An account is given of a very lengthy series of experiments 

 made to determine the correction due to the difference between 

 the temperatures of the mercury in the bulb and in the stem 

 of a thermometer when the latter is not immersed in the source 

 of heat. This has usually been expressed as follows : — If N 

 is the number of exposed scale-divisions, T— t the difference 

 between the readings of the principal thermometer and of a 

 second thermometer attached halfway up the exposed column, 

 and if z/is the correction (of course expressed in scale-divisions), 

 then 



y= -0001545 (T-0 N - 

 The number '0001545 is the coefficient of the apparent expan- 

 sion of mercury in glass ; but there is no doubt (as Holtzmann, 

 Mousson, and Wiillner have long since pointed out) that this 

 makes the value of y too great. 



Dr. Mills deduces from his experiments a corrected formula 

 of the form 



y=(* + /3JSr)(T-0N. 



The expression a + fiN, for shortness, he indicates by #. We 

 have here two points to notice. 



Dr. Mills finds that x has different values for different ther- 

 mometers, and then says (on p. 571): — " The values of y agree 

 [?'. e. in the case of the undernoted thermometers] , first when 

 N has the following values: — 



Thermometer. N. 



2 0-0 



3 166-0 



4 278-6 



6 123-2" 



It is evident that y is here a misprint for x, as y must 

 vanish when N = 0, and can vanish for no other positive value 

 of N. Hence it is impossible for the correction for a thermo- 

 meter with an exposed column 166 divisions long to be equal 

 to that of another when no part of the column is exposed ; for 

 the former cannot (unless T = t), and the latter must, be zero. 



For equal values of T— t, the correction for thermometers 2 



