DISTRIBUTION OF MAGNETISM IN MAGNETS. 645 



retardation is due solely to induced currents, or if magnetisation 

 does really take an appreciable time to establish itself. MM. Bichat 

 and Blondlot have observed* that by passing the oscillating dis- 

 charge of a Leyden jar through a coil surrounding a piece of flint 

 glass, the oscillations of the plane of polarization coincide to within 

 o '00003 f a second with the oscillations of the needle; even if 

 this retardation exists in the case of diamagnetism, it does not 

 attain this value. 



1211. DISTRIBUTION OF MAGNETISM IN MAGNETS. We know 

 (315) that the internal action of a magnet is equivalent to that 

 of a fictive layer formed of two equal and opposite masses dis- 

 tributed on its surface, or more generally on a closed surface S 

 which surrounds it.f We may then propose to determine either 

 the fictive density of the magnetism on this surface, and the pro- 

 blem is then definite, or the real distribution of the internal 

 magnetisations, if we make a hypothesis as to the mode of dis- 

 tribution. 



We have again seen (498) that the external action of a magnet 

 is equivalent to that of a system of superficial currents that is to 

 say, of magnetic shells. We may then replace the magnet by a 

 body of the same form with a solenoidal or with a lamellar mag- 

 netisation. 



The methods of Coulomb, by the oscillations of a magnetic 

 needle near a magnet (414) or by a torsion balance (415), give 

 the normal component of the force, if we take care that the mag- 

 netism of the needles is not altered by the field and does not 

 react on the magnet itself. 



The use of soft iron (416), by oscillations or by the test-contact, 

 enables us to make experiments near the surface ; but it does not 

 give more certain results, because we must assume further that the 

 coefficient of magnetisation is independent of the magnetising force, 

 which is far from being the case. 



A more correct method would consist in applying to the sur- 

 face S a small circuit of surface S', which, by analogy with the 

 proof plane, may be called proof ring^ and to measure the charge 

 induced when it is carried to a great distance. We should then 

 have the total flow of force in the surface S', and consequently 

 the mean value of the normal component. We may in like manner 

 place the circuit S' in any direction in the field of the magnet, 



* BICHAT and BLONDLOT. Comptes rendus, Vol. xciv., p. 1590. 1882. 

 t POISSON. Ann. dc Chim. et de Phys. [2], Vol. xxv., p. 127. 1824. 



