358 BELL SYSTEM TECHNICAL JOURNAL 



and in that of liquid oxygen) it must even be supposed antiparallel 

 to the field //; for 6 is negative, and consequently so is n. Necessities 

 such as these make it hard to accommodate the "molecular field" to 

 what is known or conjectured about the interior of solid bodies. 



Since it is necessary to assign several distinct values to the coefficient 

 M in order to explainTthe'behavior of iron over various ranges of 

 temperature, one cannot maintain that the iron atom possesses a 

 constant and characteristic magnetic moment which is the source of 

 ferromagnetism. Any such notion, of course, would have been 

 destroyed by facts already mentioned; but it is useful to know these 

 in addition. Changes in M sometimes coincide with great and striking 

 changes in the condition of the metal; at 920° iron exchanges its 

 body-centered lattice (spacing 2.88A) for a face-centered lattice 

 (spacing 3.60A) which it retains as the temperature rises until 1395° 

 is attained, whereupon it returns to the body-centered-cubic arrange- 

 ment. These alterations in atom-lattice are attended by changes in 

 the physical properties of the metal, so great that three separate 

 "modifications" of iron were distinguished and named before ever 

 the atom-lattices were known or suspected: /3-iron normally existing 

 from the Curie-point to 920°, 7-iron from 920° to 1395°, 5-iron from 

 1395° upward. By certain processes these modifications may be 

 enabled to survive in temperature-ranges not appropriate to them, 

 but that is too long a story for these pages. Changes in M sometimes 

 occur quite unaccompanied, so far as can be made out, by changes 

 in atom-lattice or other physical features. The variation occurring 

 at 828° in iron is of this type, and so is a mysterious change in nickel 

 which in occasional samples brings about values of M near 9, instead 

 of the usual 8 Weiss magnetons. 



We turn to residual magnetism, on its explanation of which every 

 theory of ferromagnetism must stand or fall. It is the supreme 

 merit of the theory of Weiss that residual magnetism figures as a 

 property which substances paramagnetic at high temperatures natu- 

 rally and gradually acquire, when they are cooled below a certain 

 critical point. We shall see this best by returning to Fig. 13. Begin 

 by imagining the line corresponding to a particular pair of values of 

 // and T\ leave T constant, reduce // steadily to zero; the intersection 

 of curve and line slides down the curve, reaching the origin if the slope 

 of the line is greater, stopping short of the origin if the slope of the line 

 is less, than the slope of the tangent to the curve at the origin. 



The slope of the line is kTjnM; the slope of the tangent is NM/3; 

 the critical condition is, that these be equal, and this occurs when 



T = nNMySk = e, 



