216 Maxwell's Theory of the Viscosity of Solids, 



the effects of mean atmospheric temperature. If, now, the 

 saturated magnet is again thoroughly annealed at 100°, the 

 exceptionally good stability in question is even further en- 

 hanced, because the magnetic configurations unstable as far 

 as 100° have now also been removed. The second magnetic 

 loss has just been stated to be small. Such magnets are no 

 longer influenced by variations of mean atmospheric tempera- 

 ture, nor in general by temperatures appreciably below 100°, 

 since the rods carry the maximum of permanent hardness 

 and the maximum of permanent magnetization corresponding 

 to 100°. This process of consecutive annealing is the one we 

 proposed when the magnets made are to withstand the effects 

 of atmospheric temperature, of percussion, and of secular 

 time* Cf § 20. 



20. In the above I refer to thermal, to carbon, and to mag- 

 netic configurations, using the adjectives merely to designate 

 the cause of the instability under special consideration, 

 whereas the configurations themselves need not necessarily 

 be different, Thus a carbon and a thermal configuration may 

 be one and the same grouping of atoms ; so may also a 

 thermal and a magnetic configuration. I use the latter phrase 

 advisedly, supposing the rod showing residual magnetic 

 induction to consist of configurations in all degrees of mag- 

 netic stability as well as in all degrees of magnetic intensity. 

 Stability and magnetic intensity are the qualities which in 

 the present case correspond to stability and strain respectively 

 in the above configurations. 



Magnetic stability decreases from hard to soft and from 

 soft steel to soft iron ; following, therefore, the inverse order 

 of viscosity. Its character, too, is different from viscosity, 

 the tendency being toward sudden magnetic changes, even, 

 when the cause of such change is superinduced by heat. Cf 



§ 19 - 



The mean magnetic intensity of the configuration must 

 depend on the dimensions of the saturated rod. In the normal 

 case of linear rods, this magnetic intensity increases from 

 hard to soft steelf, and from steel to iron. Hence, from one 

 point of view, carbon configurations interfere with the 

 occurrence of intense magnetic configurations ; from the other 

 point of view, magnetic intensity increases in the direct order 

 of viscosity, or stability of molecular configuration. 



21. Summarizing the results of the above paragraphs, I 

 believe the statement made in § 2 to be fully verified. I have 

 shown that the effect of distributing unstable molecular con- 



* For further details, see Bull. 14, chapter vi. ; or I. c. 

 t S. and B., Wied. Ann. xx. p. 621, 1883. 



