OF ARTS AND SCIENCES. 301 



The centre of the longest armature (150 cm.) advanced 18.25 cm., 

 and the moment became lialf as much again. The centre of the 

 metre armature advanced 14.5 cm., and its moment became a little 

 over quarter as much again. The moment of the short (50 cm.) 

 armature was increased one-sixth. 



From this the distance of the poles of the magnet from the ends can 

 be calculated; and it is about 12 cm. AVhen calculated by the vibra- 

 tions of a comi)ass in two jiositions very near the end, it appeared to 

 be 5 cm. : the latter method is probably, therefore, inexact, and will 

 always give the distance too small. It is only at large distances that 

 the magnetism of one end of a bar can be considered as acting at a 

 point. 



When the magnet was joined to the armature by a short piece of 

 iron shaped like a U, so that it was parallel to the armature, the pole 

 ap2iarently retreated about 7 cm., and the moment decreased about 

 one-tenth. When, however, a second U armature was added to com- 

 plete the circle, the reduction of magnetic moment was scarcely per- 

 ceptible, being less than ^ of the total value. This may possibly be 

 due to poor contact ; but I was unable to obtain any different result, 

 though I measured the distribution in various Avays. Perhaps, when 

 the ai'mature is bent round so near the magnet, the resistance of the 

 air to the line of force is so slight, that the adding of the second arma- 

 ture affects it but little. This, I think, would be the necessary con- 

 sequence of the analogy of magnetic force and electric currents. 



We see, then, that shunting, so to speak, the poles of a magnet with 

 a soft-iron bar, diminishes the magnetic moment about one-sixth. 



It follows, that, if a stronger magnet be shunted by a weaker one, 

 the magnetic moment will be dhuinished in some way ^iroportional to 

 the difference of magnetization. This accounts for the fact that thick 

 magnets are weaker than an equal weight of thin ones : 1st, because it 

 is impossible, owing to the difficulty of temjDering evenly, to obtain 

 the maximum capacity for magnetism in all parts of the bar, and 

 those parts which are weaker diminish the average magnetism per 

 weight, and also diminish the strength of the stronger parts, acting 

 like a shunt ; 2d, because, in the common process of magnetizing a bar 

 by rubbing the surface, the interior is less magnetized than the out- 

 side, and the core acts more or less like a shunt of soft iron. 



If, however, a solid body is magnetized by a force acting on all parts 

 alike, it will not act differently from a bundle of wires of equal length 

 and weight, provided the material be the same. Therefore the bundle 



