OF ARTS AND SCIENCES. 



119 



mometers between the two, and so had probably used up the first 

 stock of glass. And even glass of the same lot differs, especially as 

 Regnault has shown that the method of working it before the blow- 

 pipe affects it very greatly. 



It is very easy to test whether the calorimeter thermometers are of 

 the same glass as any of the others, by testing whether they agree 

 with No. 61 63 throughout the whole range of 40°. The difference 

 in the values of m for the two kinds of glass will then be about 

 .003 of the difference between them at 20°, the and 40° points 

 agreeing. The only difficulty is in calibrating or reading the 100° 

 thermometers accurately enough. 



The Baudin thermometers were very well calibrated, and were 

 graduated to -fo° C, and so were best adapted to this kind of work. 

 Hence I have constructed the following tables, making the and 40° 

 points agree. 



TABLE XXIV. — Comparison of 6163 and the Baudin Standards. 



Taking the average of the two, it would seem that No. 6163 stood 

 about .015 higher than the mean of 7334 and 7316 at the 20° point, 

 or 6163 has a higher value of m by .000045 than the others. 



These differ about .1 7 from the air thermometer at 40°, which gives 

 the value of m about .000104. Whence m for 6163 is .00015, as we 

 have found before by direct comparison with the air thermometer. 



I am inclined to think that the former value, .00018, is too large, 

 and to take .00015, which is the value found by direct comparison, as 

 the true value. As the change, however, only makes at most a differ- 

 ence of 0°.01 at any one point, and as I have already used the previous 

 value in all calculations, I have not thought it worth while to go over 

 all my work again, but will refer to the matter again in the final 

 results, and then reduce the final results to this value. 



* A correction of 0°.01 was made to the zero points of these thermometers 

 on account of the pressure of the water. 



