CONTEMPORARY ADVANCES IN PHYSICS 359 



i.e., when the temperature assumes the value of that constant 6 

 which previously entered into our equations. If T is greater than 6, 

 there should be no residual magnetism. If T is adjusted to be equal 

 to 9 and then reduced gradually to zero absolute, the residual magnet- 

 ization given from the theory — the ordinate of the point where the 

 curve is intersected by the line of slope kT/nM passing through the 

 origin — increases continuously from zero to its limiting value NM, 

 following the curve traced in Fig. 9. That is the central idea of 

 Weiss' theory of ferromagnetism. 



The first of the predictions from the theory which can be put to 

 test is the equality between the temperature at which residual mag- 

 netism disappears — the Curie-point — and the constant 6 in the 

 equation (17) for the paramagnetism of the substance beyond the 

 Curie-point. For nickel, the agreement is good: 633° against 645° 

 absolute. For cobalt and for iron, the first short straight line out 

 of the sets of two and four respectively, which are given for these 

 metals in Fig. 14, is so adjusted that 6 agrees perfectly with the Curie- 

 point; its aptness to the plotted data supports the theory. 



The next question to be asked is whether the curve of Fig. 9 

 corresponds to experience. In analyzing this question, one makes 

 the discomfiting discovery that the quantity which was defined as 

 residual magnetization in the theory cannot be identified with the 

 quantity defined as remanence in describing the experimental hyster- 

 esis-loops. This results from an imperfection, or at least an incom- 

 pleteness, in the theory. There is nothing in it to account for the 

 initial curve; there is nothing to account for the gradual increase 

 in / produced by applying a gradually-increasing field to an initially- 

 demagnetized piece of iron, and in fact there is nothing to account 

 for the existence of demagnetized pieces of iron at all — every block 

 of iron at a temperature below 9 should possess, whenever it is not 

 under the influence of an external field, the residual magnetization 

 calculated from the intersection-point of the curve NML{a) and the 

 line of slope kTjnM which passes through the origin. 



On grasping this situation, one is likely to feel that the theory has 

 collapsed. The situation can be saved, however, by supposing that 

 the "demagnetized" metal subdivides itself into a vast number of 

 little regions, zones, or filaments, each of which possesses the full 

 residual magnetism of the theory, while in direction their magnetic 

 moments are oriented quite at random. It is not possible to identify 

 these with individual crystals, nor with any other discernible granu- 

 lations of the metal. Perhaps they are to be identified with the chains 

 of elementary magnets once postulated by Ewing; it would be grati- 



