516 SCIENTIFIC RECORD FOR 1883. 



falls tbe further hypothesis that as much vapor is developed as is nec- 

 essary to the saturation of the air at t' degrees. Both of these condi- 

 tions are only approximately fulfilled. It is, therefore, certain that the 

 convection theory affords only an approximate formula. 



Therefore, we abide by the hypothesis of perfectly quiescent air; for 

 which case the Maxwell Stefan formula (F or Gr) holds good with per- 

 fect exactness. If for this hypothesis we compute the factor. 



assuming spherical thermometer bulbs, we obtain for the first part 

 J=- =0.000630: and if we put K=0.000097, according to Stefan, and 

 take r=0.57 centimeters, whence 



Er=0.000055 

 the second part becomes 



^^=0.000630 



whence A=0.001260. 



In fact, Eegnault found A =0.001280 from observation in perfectly 

 quiet air in a small closed room (Pogg. Anwaten/Lxxxvin, p. 428), and 

 Sworikin (Wild's Bepertorium, vn, No. 8, p. 17) finds from 0.001100 to 

 0.001500 for quiet air, therefore again on the average 0.001300. 



But in fact we almost never have to do with perfectly quiet air, and 

 even because of the cooling of the air close to the wet thermometer a 

 convection is always present there. If one would obtain results to any 

 extent consistent with themselves he must, since convection is unavoid- 

 able, provide for one that will assure a permanently uniform movement 

 of the air. This is attained by means of the ventilation-psychrometer. 



If, therefore, we begin with assuming a ventilated psychrometer, and 

 therefore strong convection, we have no theoretical formula for this case. 

 We can, however, attempt to adapt the formula for quiescent air to the 

 case of ventilation. 



Stefan (Zeit. f. Met., xvi, p. 181) has remarked that in the formula 

 [G] for quiescent air it can be assumed that the introduction of convec- 

 tion simply causes K and D to increase uniformly; therefore, =- will re- 

 main very nearly constant. This being assumed, it is plain that for 

 increasing values of D the second part ~—- of the factowwill therefore 



Xjlpa 



become smaller. Herein we find a reason why the theoretical value of 

 the whole factor A is not attained in the comparisons of the ventilation 

 psychrometer. In fact the value empirically determined for a barometric 

 pressure of 760 mm is almost exactly 0.000800 on the average of the various 



