520 SCIENTIFIC RECORD FOR 1883. 



Now it seems to me that, considering the difference of the ther- 

 mometers and ventilators employed, this difference is easily to be ex- 

 plained, and although still further experiments remain to be made in 

 this direction, they can only in substance confirm this result. 



I will now briefly collect the results of the investigation : 



(1.) The derivation of the psychrometer formula under this assumption 

 of the existence of convection leads to no result,* since the hypothesis 

 that the air flowing past is cooled from t to V by contact with the ther- 

 mometer bulb does not agree with the facts. 



(2.) The derivation of the psychrometer formula of Maxwell and Stefan 

 for absolutely motionless air is perfectly exact for this condition. If 

 we endeavor to introduce into the formula a modification, in order that it 

 may also hold good for moving air, then it undoubtedly loses its precision 

 but does give a very approximately correct expression, that when we 

 consider the sluggishness of the psychrometer in the neighborhood of 

 saturation, reads as follows : 



»-*-*£ ff+ifer] 0-*?+jf4i] 



or if we put v = 0°.5 C. and insert the other numerical values [as given 

 above, assuming r = 0.57 centimeters], we have: 



*-*-pmo«bo|i + vj [<,_,, +p^ki] 



(3.) The term depending on radiation does not disappear even for 

 rapidly-moving air. For absolutely calm air it is, indeed (for bulbs of 

 the radius 0.57 cm ), quite as great as that depending on conduction. 



(4.) For equal wind velocities and barometric pressures the constant 



a is invariable. Assuming equal velocities, it is smaller with lower 



p 

 pressures and most probably in the ratio „ p - . The maximum value of 



a for pressures of about 760 mm results from observations as ^ = 3.0; 

 therefore, in general, 



_ P 



This value, introduced in the above formula, gives 



760 "I T , 0.5 "I 



3.0 PJ H ' ^ {t-t') + l_\ 



1 j =p 1 -¥ 0.000630 

 or for stations at low levels 



i >o=i> l -OOOQ643 (* - n + ^ J^+i ] p 



This simple formula, as has been above shown, should not be made 

 more complicated by giving the factor A some other form, since in no 

 case will a greater accuracy be thereby attained. 



* If we abstain from considering as an important expression the second term within 

 the brackets in the convection formula (D), and seek only to find for m a numerical 

 value that corresponds to the observations, we find wi=78.0 for barometric pressure 

 760 nim , to which (since m indicates the mass of air) it is difficult to attach any intelli- 

 gible idea. • 



