198 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



In Figure 3 let P and Q be two points on the / vs. H' curve for ttti, 

 where Q, has the ordinate of the point of inflection Qq, and P is any- 

 other point of the magnetization curve. Now suppose the rod were 

 magnetized by an infinite H' to the maximum I^, so that all the 

 Tra^/oo li^ss ^^6 straight and enter and leave the rod at the squared-ofi" 

 ends {a being the radius of the rod). In this case the distribution of 

 magnetism which we may consider the cause of the demagnetizing force 

 Hi, or AZT, is wholly superficial, and as far away from the secondary 

 coil, where / is measured, as possible, and it has a perfectly definite 

 value AZToo , say, which we lay off on the / vs. {H'—H) plane, getting 

 the point K, and we draw the line OK. We see now that if, as we in- 



FlGURE 3. 



Diagram illustrating magnetization and back-shearing curves. 



crease / from zero to I^ by continually increasing H', the lines of mag- 

 netization were always straight, then the demagnetizing force would 

 always be proportional to /, no matter what the susceptibility might be, 

 and the i\r-curve would be the straight line OK. Another case where 

 the iV-curve would be a straight line OKx would be realized if the sus- 

 ceptibility were a constant for all values of / from to I^. In this 

 case no volume density would appear by magnetization, and any two 

 fields Hi and H^, giving separately the surface densities of magnetism 

 o-i and 0-2, could be superposed, so that a magnetizing field Hi + H^' 

 would give the superficial distribution o-i + o-j. This last supposition 

 would result in there being no limit to the intensity of magnetization. 

 As a matter of fact the / is uniform only for an infinite H'. At the 

 point P, HP is not the origin, more or less lines of induction will leave 



