OF ARTS AND SCIENCES. Ill 



The results for m are then as follows : — 

 From direct comparison of No. G1G7 with the air thermometer .000187 

 From " " " No. 6163 " " " .000154 



From comparison of No. 01 G3 with No. G 1 G7 .000213 



The first and last are undoubtedly the most exact numerically, but 

 they apply to No. G1G7, and are also, especially the first, derived 

 from somewhat higher temperatures than the 20° point, where the 

 correction is the most important. The value of m, as determined in 

 either of these ways, depends upon the determination of a difference of 

 temperature amounting to 0°.30, and hence should be quite exact. 



The value of m, as obtained from the direct comparison of No. G163 

 with the air thermometer, depends upon the determination of a differ- 

 ence of about 0°.05 between the mercurial and the air thermometer. 

 At the same time, the comparison is direct, the temperatures are the 

 same as we wish to use, and the glass is the same. I have combined 

 the results as follows : — 



m from No. G1G7 .000200 



m « 61G3 .000154 



Mean .00018* 



It now remains to deduce from the tables the ratios of the constants 

 for the different thermometers. 



The proper method of forming the equations of condition are as 

 follows, applying the method to the first series : — 



Weight. 



4 21.25 C IU = 115.33 G, — v 



4 255.80 G nl = 422.84 G, — v 



4 341.05 <?,„ = 534.71 G, — r 



5 431.71 C UI = 653.49 G, — v 



where G ul is the constant for No. 61 G6, G t is that for No. 6163, 

 and v is a constant to be eliminated. Dividing by C y , the equations 



Q 



can be solved for -Mr, The following table gives the results. 



* See Appendix to Thermometry, where it is finally thought best to reject 

 the value from No. G1G7 altogether. 



