PsycliTometric Formula. 207 



of a permanent nature, remaining- more or less equal in all 

 observations. Hence we have a means of directly testing the 

 theory. If it be true, the observed value of /' will depend on 

 the material used, and will always be too high. This will neces- 

 sitate a correction to /' before inserting it in the ordinary 

 psychrometrical tables, which are calculated for the standard 

 case of 77 = 1. A short provisional table is given by Ekholm 

 as an example of the corrections Avhich would have to be 

 applied. If, therefore, several wet-bulb thermometers covered 

 with different materials be simultaneously observed, their read- 

 ings should differ by amounts deducible from the values of 77 

 for each. As a rough indication of the extent of the differences 

 to be expected at ordinary temperatures, consider the formula 

 in the approximate form 



a: - 7;/ - h(^-/'J. 

 Using differential notation, suppose rj to change by drj, and 

 therefore /' by d^' and / by d/, while x and / remain unaltered. 



Then 



= jdi] + y]dr -f \ dt' 

 or, putting j; = 1 iti the second term as a close approximation^ 



— fd^ = df+^dt' 

 For linen and cotton, di) is given by Ekholm as 0.056 : near 

 0° C. we have df = J dt' very nearly, whence dt' = 0.3F 

 neglecting sign. Simihirly for linen and silk, dq is given 

 as 0.034 : in this way we can draw up a table such as the fol- 

 lowing : — 



t' a})out 0" - df = 



6° - 



12° - 



18° - 



Such differences are, of 



exist. 



Thermometep-scpeen and Preliminary Work. 



In carrying out this comparison it was thought advisable 

 to restrict ourselves to natural atmospheric conditions, since 

 they are the ones which are of practical importance. To examine 



