MAGNET STEELS AND PERMANENT MAGNETS 



389 



After this empirical correlation was obtained, it was suggested that 

 Brem/Br VS. L^HcjD^lfr would be preferable as a choice of variables, 

 from theoretical considerations based upon the assumption of uniform 

 magnetization in the magnets. Although it can be shown that this 

 assumption is not fulfilled in the actual case of straight bar magnets, 

 this method of plotting gives as good a correlation as is shown in Fig. 

 5. Such a result might be expected, since the values of Br commonly 



a. CD 



Fig. 5— The same data as in Fig. 4. Showing the correlation between Brem/Br and 

 (LID)ylHc for bars of any kind of material, magnetic properties, length or cross- 

 section. 



encountered range from 9,000 gauss to 11,000 gauss, thus differing 

 from 10,000 gauss by 10 per cent at most. The square root of Br, 

 therefore, differs from 100 by not moj;e than 5 p^ cent in most cases, 

 whence it follows that dividing L^HcjD by ^Br shifts the relative 

 position of the abscissas of the points in Fig. 5 by not more than 5 per 

 cent in most cases. Fig. 6 illustrates the results obtained by the 



second method of plotting. 



Dimensional considerations favor the use of the quantity L^JHc/ 



