1890.] Molecular Theory of Induced Magnetism. 357 



being magnetised, its permeability is not the same as when the wire 

 is completely unloaded and reloaded up to the same load. Expe- 

 rimental results of this kind led me in 1884 to write: "If we apply 

 and remove stress in a wire whose magnetic state is entirely neutral, 

 we cause some kind of molecular displacement in the relation of which to 

 the applied stress there is hysteresis."* The theory now offered shows 

 how this happens. Hence also the remarkable hysteresis which the 

 thermoelectric quality of iron exhibits with regard to cyclic changes 

 of stress, discovered by Cohn, and more fully described in ' Phil. 

 Trans.,' 1886, p. 361. The hysteresis of molecular configuration with 

 respect to stress has been proved to be removable or reducible by 

 vibration. 



From this theoretical explanation of hysteresis in the effects of 

 stress, it at once follows that a cyclic change of stress (provided 

 it be not very small) involves some dissipation of energy in a 

 magnetic metal, whether the piece be magnetised or not. We may 

 expect this dissipation to be most considerable under conditions 

 which make the magnetic hysteresis large. But it will occur even 

 when there is no external trace of magnetism. 



This, of course, implies that in a cyclic process of loading and un- 

 loading, work must be spent. There is no perfect elasticity in a 

 magnetic metal, however slowly the process of straining be per- 

 formed. Under any load there is less strain during application than 

 daring removal. This is borne out by experiments on the extension 

 of iron wires ('Brit. Assoc. Rep.,' 1889, p. 502). 



The same action occurs to a marked degree in torsional strains. In 

 a twisted specimen there will be a tendency on the part of the mole- 

 cular magnets to range themselves along lines agreeing more or less 

 with the direction of maximum contraction. Alternate twisting to 

 opposite sides should therefore cause much molecular swinging 

 through unstable positions, with consequent dissipation of energy, 

 even in a piece which is not magnetised. 



Without going at present into details, it xnay be added that the 

 phenomena of molecular "accommodation" studied by Wiedemann 

 and by H. Tomlinson accord with the theory, and that it seems to 

 lend itself well to explain the very remarkable results which have 

 been obtained by Nagaokaf in experiments with nickel wire under 

 twist or under a combination of pull and twist. It also agrees with 

 what little is known as to the influence that previous magnetisation in 

 one direction has upon subsequent magnetisation in another direc- 

 tion. 



To sum up, I have endeavoured to show 



* 'Phil. Trans.,' 1885, p. 614. 



t 'Journal of the College of Science of the University of Tokio,' vol. 2, 1888. 



