368 Dr. J. R. Ashworth on the Anliysteretic 



method of least squares, I find the following values for 

 the constants I and P for iron : — 



Temperature. 

 16° C. = 289° A. 



933 



P. 



•00132 



R'=P/T. 



4-57 X 10 " 6 



645° C. = 918° A. 



916 



•00383 



447X10" 6 



720° C. = 993° A. 



700 



•00428 



4-31 Xl0 -6 



776° C. = 1049° A. 



511 



•00600 

 Mean ... 



5-72X10" 6 





.. 4-69X10" 6 



If, now, P is divided by the absolute temperature T, the 

 quotient R' is approximately constant until a near approach 

 to the critical point. Thus the constant in Frolich's equation 

 is proportional to the absolute temperature for these curves 

 over a wide range of temperature. The quantity I , however, 

 above 645° C. begins to diminish rather rapidly, a fact which 

 deserves consideration. The values of I calculated from 

 these constants are given in the third column of Table I., 

 and comparison with the observed values allows the accuracy 

 of the equation to be estimated. 



Treating the observations on nickel in the same way, the 

 values for the constants I , P, and R/ are : — 



Temperature. 



Io- 



p. 



B/=P/T. 



21° C. = 294° A. 



359 



•0097 



33'1 X 10 ~ 6 



139° C. =412° A. 



333 



•0109 



26-4X10" 6 



225° C. = 498° A. 



315 



•0130 



26-lXlO -6 



319° C. = 592° A. 



177 



•0154 

 Mean ... 



26-1 xlO" 6 





.. 27-9X10 -6 



R/ is again nearly constant over a considerable range of 

 temperature ; at the same time I diminishes as the tem- 

 perature is raised. The values of I calculated from these 

 constants are given in the third column of Table II. They 

 show larger differences than those for iron, and indicate that 

 the experimental curve is somewhat flatter than the calculated 

 curve. 



(10) As a first approximation the equation to the anhys- 

 teretic isothermals may be written 



H (i 1 -i)= R ' T ' 



T being the absolute temperature. In this form the 



