SHUDDEMAGEN. — DEMAGNETIZING FACTORS FOR IRON RODS. 253 



114 cms. and lit = 60 to be always within about 4 per cent of the 

 central induction. These facts justify the use of a secondary coil 

 several cms. in length, provided the in of the rod is not too small. 



The conclusion to be reached from the work on the induction dis- 

 tribution is that for low field-intensities, as well as for high ones, the 

 magnetization of the iron rod is much more nearly uniform than it is 

 in a long interval corresponding to rather high susceptibilities. 



Discussion of Results obtained. 



When we look over the tables we readily see a number of interest- 

 ing things. It is apparent that in general different methods or even 

 different experimental conditions will give different normal curves, 

 and hence different susceptibility curves. A striking result, and one 

 which was obtained entirely unexpectedly, is that in the long solenoid, 

 which was wound on a thick brass tube, the method of reversals agrees 

 very closely indeed with the step-by-step method. This may in fact 

 .turn out to be quite a useful observation, for it points to the proba- 

 bility of getting values for the susceptibility of some kind of iron in 

 the form of a short rod, which conform very closely to the ideal defini- 

 tion of susceptibility, which requires slow, continuous increase of the 

 magnetizing field. Thus by winding our solenoid on very thick brass 

 tubes, a large E. M. F. from a storage battery may be suddenly turned 

 on, without giving almost instantaneously the full value of the magne- 

 tizing field within, on account of the eddy currents in the brass tube 

 acting as a sort of " brake." 



The most important results described in this paper about the de- 

 magnetizing factor N for cylindrical iron rods are the following: 



(1) The demagnetizing factor is not a constant, but shows two 

 opposite curvatures, when plotted as abscissa-differences (H, = NI) on 

 the / vs. Hi plane ; while for the highest values of / it falls to about 

 \ or \ of its value for unsaturated Is. 



(2) For values of B less than 10,000 the N is practically constant. 



(3) Using a solenoid made of wire wound on a non- metallic tube, or 

 a split brass tube, the reversal method gives values for N considerably 

 lower than the step-by-step method. 



(4) If the magnetizing solenoid is wound on a thick brass tube, the 

 reversal and step methods practically agree, and values of N derived 

 from curves taken in this way are regarded as the most desirable for 

 scientific purposes, as they will give most accurate values for the 

 susceptibility or the permeability of the iron. 



(5) The demagnetizing factors are largest for thin rods. The differ- 



