184 Standardizing of Mercurial Thermometers. 



register about 7°*7, 12°*5, and 17°*7 C. respectively; and in 

 the two comparisons of 63616, 5°*5 and 17°*3 0. These 

 repeated standardizations give the means of ascertaining the 

 experimental error involved in the process. 

 The results obtained were as follows : — 



With 62839, deg . deg . 



A. An interval of 53 2811* = an interval of 4-5302 of 55084, 



B. „ „ 51-9524 = „ „ 4-4260 „ „ 

 0. „ „ 53-7366 = „ „ 4-5819 „ „ 



From which was deduced, that the average value of 1 degree 

 (L e. 10 millim.) of 62839 is 



A. . . . 0° 0601967 C. 



B. . . . 0°-0602480C. 



C. . . . 0°*0o03066C. 



With 63616, defr deg> 



A. An interval of 55*1059 = an interval of 5*0039 of 55084, 



B. „ „ 53-1093= „ „ 4-8184 „ „ 



From which, 1 degree of this thermometer represents 



A. . . . 0°-0641605C. 



B. . . . 0°-0641059C. 



Now the extreme difference in these numbers in the case 

 of the first instrument is o, 0001099 C, while the average 

 difference between any pair of observations is o, 0000733 C; 

 in the case of the second instrument this difference is some- 

 what less, namely 0*0000546 C. This latter number I take 

 to represent more nearly the error which would be likely to 

 be obtained with any ordinary thermometer ; for, owing to 

 the large size of the bulb of No. 62839, the readings were 

 subject to a source of error which was not thoroughly under- 

 stood and guarded against at the time when the comparisons 

 were made. 



This difference of 0°'0000546 C. in the results would be 

 caused by a total error in the readings of 55084 of 0*0043 deg., 

 or in those of 63616 of 0*056 deg., or by an error in those of 

 both thermometers of 0*004 deg, each. As any such pair of 

 comparisons involves four readings of each instrument, it 

 follows that any temperature can be read with an error 

 of 0*001 deg., or 001 millim. (remembering that each 

 such reading is really the mean of twelve observations). 



# One of these arbitrary degrees, with thermometers 62839 and 63616 

 measures 10 millim., with 55084 12 millim. All the numbers given here 

 are fully corrected. 



