522 



Mr. E. 0. C. Baly on the 



But this relation does not hold in the case of air, as the 

 following results show : — 



Temperature. 



r' 

 r 



Temperature. 



r' 

 r 



o 







, 



77-9 



•2416 



86-4 



•3141 



78-2 



•2495 



88-2 



•3274 



78-5 



•2582 



89-2 



•3406 



799 



•2710 



90-1 



•3592 



81-4 



•2828 



90-5 



•3725 



83-2 



1 



•2934 



90-7 



•3806 



It will be seen that — , instead of being constant, has a 



steadily increasing value with rise of temperature. 



Lehfeldt, in his paper on " The Properties of a Mixture 

 of Liquids"*, showed that in Brown's case the logarithms 

 of the ratios corresponding in the liquid and vapour phases 

 had a linear relation ; that is to sav 



uucu l? iu a<« v 



log r / = a-\-b log r. 



Inasmuch as this relation held so well for Brown's results, 

 the logarithms of the corresponding ratios with air were 

 plotted, and, as shown in fig. 3, a linear relation was 

 found to exist in this case. The logarithms of 100 r and 

 100 r f were plotted for greater convenience. The points 

 shown through which the curve is drawn were found for 

 each half degree rise of temperature. The general accuracy 

 of the results is shown by this curve, the departure of the 

 points from the straight line being hardly appreciable. The 

 large curves of percentages of oxygen against temperature 

 were smoothed to a slight extent by reference to the logarithmic 

 curve, though this made no change except to remove two 

 slight irregularities which had not before been noticed. 



The values of the constants a and b in the equation 

 log r' = a 4- b log r were first calculated from two pairs of 

 readings from the smoothed logarithmic linear curve. As, 

 however, these did not agree very well, 22 readings were 

 made, and the most probable value calculated by the method 

 of least squares, which gave the results 



a= -2097. 

 6 = 1-06737. 



* Phil. Mag. [o] xl. p. 397. 



