308 M. Bessel's Correction of the Thermometer. 



5 + Q5- (e + <pe) : 180 = (x + f x)—(e+ <p e) : F— 32 

 where 5 and e are the points of boiling and freezing water on 

 the scale, and F the true degree of Fahrenheit's thermometer, 

 answering to the point x of the scale. 



In order to ascertain the correction f x, I have used the 

 following method. At first, a part of the mercury in the tube, 

 equal to about 50° or 60° Fahrenheit, was separated from the 

 rest, either by shaking, or over the flame of a candle ; the lower 

 extremity of this mercury was then moved to every tenth de- 

 gree of the scale, and the corresponding place of the upper 

 extremity noted down. The mercury was then united again, 

 and the process repeated with five or six other columns, always 

 differing from 10° to 20° in length. 



The thermometer of Schafrinsky showed the following de- 

 grees : 



In order to deduce from this table the correction denoted 

 by <px, the lengths of the different columns of mercury may be 

 so assumed as they appear in that part of the tube nearest cor- 

 responding to the scale (which, although not necessary, leads 

 to a more speedy approximation in the calculation) ,- this is the 

 upper part, and accordingly the length of every column was 

 assumed equal to the mean of all readings between 90° and 



210°. 



