390 



BELL SYSTEM TECHNICAL JOURNAL 



D^Br, however, since this quantity, like Brem/Br, is dimensionless, 

 whereas L^Hc/D is not. For this reason it is felt that the method 

 of plotting shown in Fig. 6 is to be preferred to that of Fig. 5 although 

 the correlation obtained in Fig. 5 appears to be as good as that shown 

 in Fig. 6. Further, the necessity for injecting the variable of tem- 

 perature into the picture is done away with. 



Inasmuch as the data from a large number of bars of widely different 

 compositions, magnetic properties, dimensions and heat treatment are 



3 , 4 



D V B 



Fig 6.— The data of Fig. 4 with BremlB, plotted against LyjHc/DyjBr 



all found to cluster quite closely along a single curve, it is felt that it is 

 legitimate to use the curve as a basis for some generalizations, even 

 though the curve was arrived at largely by empirical processes. 



The type^of curve that is obtained when BremlB r is plotted against 

 LyHc/D-^Br indicates that other factors being constant, the remanence 

 of a magnet is roughly proportional to ^[B'r and, for large values of 

 LylHc/DyJBr, in other words, for large values of dimension ratio or 

 coercive force, the remanence is practically equal to B^ This fact is 

 of interest from a design standpoint. 



