202 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



This, however, is not at all the case, and the actual facts emphasize 

 the fallacy of considering the magnetization in long iron rods, when 

 not completely saturated, as even approximately uniform. As will ap- 

 pear from the results obtained in this investigation, the values of N 

 are not far from being constant below B = 10,000, and they are of the 

 order of magnitude as those found by Du Bois from Ewing's curves, 

 although always somewhat smaller. But let us now find what these 

 iV^-values would be if our various rods were really uniformly magne- 

 tized. In other words, let us find the position of K of the straight line 



3000 



2000 



1500 vn 



<n 



»^ 1000 



500 £ 

 IK 



Centre 22 20 18 16 U 12 10 8 6 4 



Figure 4. 



Lamb's curves showing the change in permeability along an iron rod. The 

 distances along bar are given in inches. 



OK in Figure 3. Our rod has the length L and diameter D, so that 

 uniform magnetization would mean tt {Bl'ifl units of free positive 

 magnetism on one end of the rod and the same number of negative 

 units on the other end. If L is large compared to Z>, we may regard 

 the demagnetizing field-intensity Hi (or NP) at the centre of the rod 



as caused by a single point-pole of strength 27rf — j/ata distance 

 of X/2 units of length from it. Then 



