CONTEMPORARY ADVANCES IN PHYSICS 243 



within each domain increases, and this increase requires additional 

 heat over and above that which goes to augment the kinetic energy 

 of the atoms. The specific heat of iron (as of its congeners) is greater 

 than it would be, but for this effect; the excess may be computed from 

 the foregoing theory as function of temperature, and the computed 

 values agree with the data to an extent which speaks very strongly 

 for the theory. (The like is the case with a paramagnetic body 

 exposed to a magnetic field; and as a result, such a body will grow 

 cooler when the applied field is withdrawn, the kinetic energy of the 

 atoms being levied upon when the dis-alignment occurs. The effect 

 is imperceptible in usual circumstances, but with such substances as 

 iron-ammonium alum at liquid-helium temperatures, it becomes so 

 strong that the lowest temperatures ever achieved have been attained 

 by making use of it.) There are the " magnetostrictive effects," 

 arising because, w^hen the atoms of the domains change their orienta- 

 tion, the metal as a whole is strained. It follows that there are 

 interrelations between magnetization, strain, and stress; and anyone 

 remembering even a little of the mathematical theory of elasticity 

 with its moduli and its stress-strain tensors will readily believe that 

 the theory of these interrelations is marvelously complicated. As 

 one sensational example of the consequences, I cite the fact that when 

 a certain permalloy is exposed to a field of, say, one half of one gauss, 

 its magnetization ranges between a few per cent and nearly one hun- 

 dred per cent of saturation, according to the strength of the tensile 

 stress applied to it. The many processes of the metallurgical arts 

 have often vast effects upon the magnetic properties of the ferro- 

 magnetic metals exposed to them: some are due to the changes in the 

 elasticity and hence in the magnetostrictive effects, some to the 

 changes in the chemical constitution (e.g. in the proportion of im- 

 purities), some to the changes in phase (of alloys) which these processes 

 entail ; but it would be risky to affirm that they have all been traced 

 to one or another of these causes. The finer details in the shape of 

 the I-vs-II curve for ferromagnetics remain to be explained, and to 

 account for one of them it seems to be thought necessary to assume 

 that the domains may gain or lose in size at one another's expense; 

 it is too bad that this impairs the concept of the domain as an immu- 

 table super-atom. I leave without overmuch regret this infinitely 

 detailed and complicated topic, to conclude by brief allusions to the 

 spinning electron and to diamagnetism. 



Hitherto in these pages I have let it be inferred that when we obtain 

 the magnetic moment of the atom of some element or the molecules of 

 some compound by magnetic experiments upon the substance, it 



