204 Love and Sineal : 



tures both above and below zero. Thus Regnault's seventh 

 series was taken with the thermometer covered with cotton 

 muslin and iced. In order to represent his observations with 

 reasonable accuracy, he found it necessary to use two values of 

 A over different ranges — viz., 0.00075 up to 70 per cent., and 

 0.0013 above 70 per cent, humidity. Ekholm finds that the 

 single formula — 



A =^0.932/ - 0.000496 B(^/ - /'; 



satisfactorily agrees with the whole set of observations, the 

 values of the tAvo constants ?/ and A having been determined 

 by least squares. If, however, Regnault had correctly inter- 

 preted his observations by using ice-pressures instead of water- 

 pressures for f, the superiority of the Ekholm formula would 

 not be so striking. The latter employs his own revised tables 

 of vapour-pressures throughout. 



Svensson's observations range from about — 3° to -f- 26^^ in 

 dry temperature, and from 5 per cent, to 95.5 per cent, 

 in humidity. For the temperatures above zero, the formula 

 takes the value — 



.v = 0.9737 / - 0.000590 B(/ - /' ). 



Assuming the value of q to be the same whether the material 

 — in this case linen — be covered with water or ice, and reduc- 

 ing the value of A in the theoretical ratio 600 : 680 of the 

 latent heats, the formula appropriate to the lower temperatures 

 becomes 



a: = 0.9737/ - 0.000526 Br/-/';. 



This is shown to be in good agreement with the observations, 

 thus apparently confirming the suppositions made. 



The same value of q is found to be suitable to Regnault's 

 fourth series of observations, temperatures ranging between 7° 

 and 30"", and a slightly different value, 0.968, to tlie nintli 

 series, which Regnault regarded as taken in specially un- 

 favourable circumstances. 



The argument then rests on the two facts that the modi- 

 fied formula provides a better agreement between the calcu- 

 lated and observed values than the old one ; and, secondly, 

 that the value of ?? which does this is always a proper fraction. 



