148 Mr. K. H. M. Bosanquet on 



The moments of B and C were determined by suspending 

 the combinations in a tray in the torsion-balance. 



I have hitherto assumed that the resultant magnetism of a 

 disk, or short cylinder, acts, as it is commonly supposed to do, 

 at the terminal surfaces. It is now clear that the foci, as I 

 define them, are at a distance greater than the length of the 

 short cylinder. This distance tends to a finite limit for thin 

 disks. The limiting focal distance appears to be about three 

 fourths of the radius. 



This idea is new to me, though it is an obvious deduction 

 from the general appearance of the lines of force. It will 

 probably also hold for electromagnets. 



By the foregoing investigation we have reduced the moment 

 of the magnet to that of a series of infinitely thin disks, 

 which correspond to the magnetic shells of the amperian theory. 

 In the case of bar A, for example, the total moment of the 

 whole magnet is more than twenty times as great as the 

 limiting disk-moment. It now appeared to be of interest to 

 see what would be the number of current-turns which would 

 be required to demagnetize the magnet temporarily. It was 

 to be expected that the moment of the current-turns required 

 would be equal to the limiting disk-moment. 



The magnet was enclosed in a coil of 255 turns having 

 exactly its own length, and was placed on the deflection-scale 

 near the needle so that it produced a deflection of about 20°. 

 A current was then sent through the coil, so as to reduce the 

 deflection to zero. The first time this is done, it always 

 reduces the permanent magnetism a little. The value em- 

 ployed must therefore be ascertained before all the other 

 measurements are made. 



Ampere-turns which demagnetize A = 430. 



Ampere-turns X area of magnet - 9Q 



Limiting disk-moment of A . . . = 150. 



It has been mentioned that the distribution of the magnetism 

 in A was far from uniform. It seems probable that the uni- 

 form coil acts with an advantage over the irregular distribu- 

 tion in the magnet. And if this be the case, we must suppose 

 that the actual disk-moment of 150 is balanced by the effect 

 of a uniform coil corresponding to 129. In fact it is known 

 that a difference of potential localized at a point in a bar, as 



