316 Intelligence and Miscellaneous Articles. 



so far as to the apparent reversal of the poles, whereas after the 

 cessation of the current the bar is found to be still magnetized in 

 the original direction. 



The following fact, which I have observed, appears to me still 

 more curious. I take a bundle formed by the combination of four 

 square bars of the same length. It is tempered hard and magnet- 

 ized immediately ; its magnetic moment is measured ; lastly it is 

 taken to pieces, and the magnetic moment measured of each bar 

 separately. It is found that the sum of these moments is very 

 considerably greater than the magnetic moment of the bundle. If 

 the bars be united in twos, the sum of the moments of the pairs is 

 intermediate between those of the whole bundle and of the sepa- 

 rate bars. When, finally, the bundle is reconstructed, the magnetic 

 moment also returns to its former value. 



In this experiment the bundle, which, innocent of all anterior 

 magnetization, has only once undergone the action of the magnet- 

 izing spiral, is in an absolutely normal condition at the moment of 

 the first separation ; and no new force appears to intervene, to 

 which one might attribute the observed augmentation of the per- 

 manent magnetism. It is true that, in separating the bars, we 

 suppress their reciprocal reaction ; and we know that in each of 

 them it acted in the opposite direction to that of the permanent 

 magnetism ; but this suppression can only have effect on the tem- 

 porary magnetism. Thus, even in a normal bar, a certain degree 

 of permanent magnetism is found to be superposed to a temporary 

 magnetism of the opposite direction. 



It would therefore be very natural to recur to an old hypothesis, 

 according to which the condition, whatever it may be, which corre- 

 sponds to the conservation of a certain permanent magnetism, is 

 communicated, in the conversion into steel or in the tempering, 

 only to a certain number of molecules, the others retaining their 

 former properties. If we remark (1) that the laws of the tempo- 

 rary magnetism of steel appear to be identical with those of induced 

 magnetism in soft iron, (2) that the development of permanent 

 magnetism is eminently variable from one sort of iron or steel to 

 another, and for one and the same sort according to sometimes in- 

 significant physical conditions, we shall be led to examine more 

 closely than has been done yet the consequences of this hypothesis. 



Let us consider a cylinder of elemental dimensions, but of very 

 great length relatively to its diameter. Let us suppose the two 

 kinds of magnetic elements scattered at random, but in a determi- 

 nate proportion, in all parts of the cylinder, and a magnetic force 

 F acting in the direction of the axis. If the molecules devoid of 

 coercive power existed alone, the cylinder would take a magnetic 

 moment IcFAv, — Av representing the volume of the cylinder, and 

 Tc a coefficient which depends on the density of the molecules. In 

 the same way the molecules endowed with coercive power, if they 

 were alone, would take a magnetic moment qFAv. 



If we suppose the coefficients of induction h and q constant 

 (which is sensibly true for small values of the inductive forces), and 

 designate by c a coefficient dependent on the grouping of the mag- 



