142 Mr. R. H. M. Bosanquet on 



It still remains to be found which are the most convenient 

 dimensions for the globe and lens, so that the minimum of 

 alteration of position will be requisite to obtain a sharp 

 image during all times of the year ; and also the best method 

 of fixing the paper so that it may be easily changed and the 

 time-scale marked on it. 



XVIII. Permanent Magnets. — I. 

 By R. H. M. Bosanquet, St. John's College, Oxford. 



To the Editors of the Philosophical Magazine and Journal. 



Gentlemen, 



IN some notes on this subject (Phil. Mag. xv. pp. 205, 257, 

 309) I showed that, when a permanent magnet is divided 

 into short lengths, its total moment is considerably diminished. 

 I also endeavoured to treat the theory of the question from 

 what I take to be Faraday's point of view. The measures 

 then executed were only of one element — the moment. The 

 view of the phenomena was therefore insufficient. Having 

 now in my hands processes for determining the magnetic 

 induction in the magnet, I have made determinations of this 

 element, as well as of the moment, in three compound magnets. 

 Further, I have measured in one magnet the number of 

 ampere-turns of a uniformly wound coil necessary to reduce 

 the external action temporarily to zero. These measures 

 form the first part of this communication. I find that the 

 properties of magnets, between the state of thin disks and bars 

 of about five times the length of their diameter, are governed 

 by extremely simple laws. Yet these are of such a nature 

 that I am unable to imagine how the vague generalities 

 about induction, which some consider sufficient to account for 

 the facts I have previously brought forward, can be applied 

 to them with any useful result. 



I then show how the theory I have based on Faraday's 

 views offers an account of the phenomena, and especially of 

 the cases comprised within the above limits. 



The complete determination of the functions involved in the 

 case of bars of greater length will require additional expe- 

 riments. 



N = total number of pieces into which a whole bar is 

 divided. 

 Then, if m, n be two numbers such that wn = N, 



m is the number of separate bars in any combination, 



n the number of pieces in each separate bar of the combi- 

 nation, 



