Observations on the Hygrometer. ^9 



B (and t) being constant. This limitation respecting the tem- 

 perature of the air, which is marked above, by being enclosed 

 within brackets, is the only modification proposed. The neces- 

 sity for its introduction is grounded upon the observation, that 

 the ratio of D to ft, -—ft", seemed evidently, though slowly, to 

 increase with t or J]. 



Having been unable theoretically to determine the law ac- 

 cording to which the ratio of D to /;' —fi', increased with the 

 temperature of the air, it was assumed that 



or yj' — C7i -f'^' ^^) 



and the constant quantity C was found by means of the first of 

 Dr Anderson's experiments- 



Although this formula did not correspond in a satisfactory 

 manner with experiment, yet the relation which was accidental- 

 ly observed to subsist between its results and those of equation 

 (2), suggested what appears to be either accurately, or very 

 nearly, the law according to which the ratio of D to ^ — Jt' 

 varies. 



Having found the values of b and c, in formulas (2) and (5), 

 by means of Dr Anderson's experiment, it was remarked that 

 both equations erred when t was either much greater or less 

 than 67.2, and, what was of most importance, that the errors, 

 which then were nearly equal, were always of opposite kinds. 

 The evident inference was, that the arithmetical mean between 

 the corresponding sides of the two equations, that is, the formula 



would represent very nearly the manner in which the ratio of 

 D to/t' — Jl' varies with the temperature of the air. The writer 

 has accordingly found that, under a considerable range in the 

 values of t and D, the differences between the results of this 

 equation and experiment, are not greater than may be ascribed 

 to errors of observation. 



The values of b and c will afterwards be introduced into the 

 above formula, along with the correction for pressure ; and, in 

 the first of the accompanying tables, the results of equations (1), 

 (3), (4), and (6), will be found compared with obsenation in 

 fifteen instances. It may be remarked, that, in all the calcula- 



