730 Dr. S. P. Thompson and Mr. E. W. Moss on the 



the space between any two point-poles the intensity of the 

 magnetic field that is due to these poles, at any point in the 

 line joining them, is expressed by the equation : 



^ 



{a-\-x) 2 (a — a) 2 



where the respective strengths of poles are mj and — m 2 ; a 

 the half of the distance between them, and a the distance of 

 the point in question from the mid-point between them. The 

 value of this expression in no way depends on the material 

 in the space between the poles, whether non-magnetic or 

 magnetic, or actually magnetized in any manner. 



If m Y and -—m 2 are numerically equal, the expression 

 becomes : 



(a 2 -a? 2 ) 2 ' 



At the mid-point, under the same condition, the intensity 

 has the minimum value of 



mm. 



= 2m-7-a 2 . 



If the space between the two point-poles be regarded as 

 occupied by a thin, cylindrical, uniformly-magnetized steel 

 magnet the ends of which constitute the point-poles in 

 question, then these equations will be the expresions for a 

 self-produced magnetic field acting in a direction which 

 opposes the actual magnetism of the magnet, and tending to 

 demagnetize it. Each portion of the filiform magnet will be 

 acted upon by a demagnetizing field, strongest towards the 

 poles, weakest at the middle. The supposed uniform mag- 

 netization of the magnet will of course be unstable. If it 

 were produced, even for a moment, there would at once be a 

 retrocession of a portion of the magnetization from the ends, 

 with a new distribution of the polarity. On the supposition 

 that the middle part of the rod retains still its full flux, the 

 retrocession of the pole would shorten the effective length of 

 the magnet, diminishing the magnetic moment, but increasing 

 any self-demagnetizing internal action. This tendency to 

 produce a retrocession of the pole may operate to different 

 degrees according to whether the bar consist of soft iron, or 

 hard tungsten steel. In either case the retreat of the pole can 

 be only incomplete ; because if we suppose the pole to have 

 actually retreated by any given amount — for example 1 centi- 

 metre — the end piece of that length will now be subjected to 

 the magnetizing action of the rest of the bar, and will be 

 remagnetized up to a certain point, namely, such that the 



