Residual discharge 391 



In forming a theory of the magnetization of iron, Poisson was led to the 

 hypothesis that the magnetic fluids are free to move within certain small portions 

 of the iron, which he calls magnetic molecules, but that they cannot pass from 

 one molecule to another, and he calculates the result orfthe supposition that 

 these molecules are spherical, and that their distances from each other are large 

 compared with their radii. 



When Faraday had afterwards rediscovered the properties of dielectrics, 

 Mossotti, noticing the analogy between these properties and those of magnetic 

 substances, constructed a mathematical theory of dielectrics, by taking Poisson's 

 memoir and substituting electrical terms for magnetic, and Italian for French, 

 throughout. 



A theory of this kind is capable of accounting for the specific inductive 

 capacity being greater than unity, without introducing conductivity through 

 portions of the substance of sensible size. 



Another phenomenon which we have to account for is that of the residual 

 charge of condensers, and what Faraday called electric absorption. The only 

 notice which Cavendish has left us of a phenomenon of this kind is that recorded 

 in Arts. 522, 523, in which it appeared "that a Florence flask contained more 

 electricity when it continued charged a good while than when charged and dis- 

 charged immediately." 



To illustrate this phenomenon, I gave in "Electricity and Magnetism," Art. 

 328, a theory of a dielectric composed of strata of different dielectric and con- 

 ducting properties*. 



Professor Rowland has since shownf that phenomena of the same kind 

 would be observed if the medium consisted of small portions of different kinds 

 well mingled together, though the individual portions may be too small to be 

 observed separately. 



It follows from the property of electric absorption that in experiments to 

 determine the specific inductive capacity of a substance, the result depends on 

 the time during which the substance is electrified. Hence most of those who 

 have attempted to determine the value of this quantity for glass have obtained 

 results so inconsistent with each other as to be of no use. It is absolutely 

 necessary, in working with glass, to perform the experiment as quickly as 

 possible. 



Cavendish does not give the exact duration of one of his "trials," but each 

 trial probably took less than two or three seconds. His results are therefore 

 comparable with those recently obtained by HopkinsonJ, who effected the 

 different operations by hand. 



* [See J. Hopkinson's collected Scientific Papers for theory and experiment.] 

 f American Journal of Mathematics, No. I, 1878, p. 53. 



J Proceedings of the Royal Society, June 14, 1877; Phil. Trans. 1878, Part r, p. 17. 

 [Reprinted in collected Scientific Papers.} 



