4-i 0. T. Sherman — A Study of Thermometers. 



time \ We have observed the corrections to the following 

 treated thermometers on Feb. 16th, March 9th and 22d, and 

 April 15th. The record is given in the adjoining table : 





Feb. 16. 



Y. O. 81 (Cent.). 

 March 9. March 22. 



April 15. 



Feb. 16- 

 April 15. 



0° 



— 6°-6 



-6 2 '6 — 6°-5 



-6 C 5 ) 





100' 



— 5 - 85 



-5-78 -5-76 



-5-68 )■ 



-0-16 



200° 



-6-3 



Y. 0. 89 (Cent). 



-6-1 \ 





0° 



-5°-5 



— 5°"5 — 5 0, 5 



-5°-5 ) 





100° 



— 4-1 



-4-0 —4-1 



-4-0 ]■ 



-o-i 



200° 



-5-3 



Y. O. 90 (Cent.). 



-5-1 J 





0° 



— 8°-9 



_8°-9 -8°-6 



-8°-5 ) 





100° 



-6-8 



-7-0 -70 



-6-9 j. 



-0'06 



200° 



-6-3 



Y. 0. 20967 (Fahr.). 



— 6-4 ) 





0° 



-16°-8 



-16°-9 — l6 D -9 



-17°-01 





212° 



-18-6 



— 18-4 -18-6 



-18-5 1 





387° 



-25-2 



-25-2 



+ 0-17 



420° 



-27-8 







—28*2 j 





On all of these instruments the closeness of the graduation 

 renders an error of observation of a tenth not improbable, so 

 that, with one exception, there is no difference which seems 

 worthy of remark. 



These observations indicate that after treatment the ther- 

 mometer is as serviceable as a measure of temperature ranging 

 from 0° to 300° C. as the standard to which we are accustomed 

 is for the range 0° to 100°. In the curves representing the 

 movement of the zero, the record of April 15th is represented 

 by a dotted line, that of Feb. 16th by the full line. The former 

 are slightly more curved than the latter. Again, in the final 

 column of the preceding table are given the mean differences 

 between the corrections due to Feb. 16th and April 15th. Both 

 of these differences we would interpret as small effects still oc- 

 curring in the bulb, such as occur in every new thermometer, 

 rather than as evidence that the instrument does not repeat 

 itself. 



It is of interest to ask what is the nature of the change which 

 has been effected in the glass. If we compare the errors before 

 and after treatment, we obtain the following differences: 



Y. O. 81. Y. 0, 89. Y. 0. 90. 



0° 4°-0 4°-9 7°-5 



100° 2-8 2-9 6-6 



200° 4-6 4-0 4-4 



The differences for Y. O. 20967, upon which the points of com- 

 parison are more frequent, are given in the adjoining curve. 

 These differences indicate a change in the coefficient of ex- 



