CONTEMPORARY ADVANCES IN PHYSICS 313 



annealing,* followed by a gradual cooling, obliterates the traces of 

 earlier experiences; and consequently a sample of unknown (or 

 known) antecedents can be restored, by putting it through this process, 

 to a standardized initial state. 



Imagine, then, a sample of nickel which since its latest rejuvenation 

 by annealing has undergone a recorded set of experiences; for instance, 

 that it has been "stretched almost to the point of rupture, bent into 

 a circle, and allowed to restraighten itself" (I quote an actual case 

 investigated by R. Forrer). It might now be thought that, so long 

 as the greatest care is exercised to avoid subjecting the metal to new 

 stresses, concussions or heatings, the I-vs.-H curve would be fixed 

 for good. Not so! for in order to determine the I-vs.-H curve, the 

 metal must be magnetized; and magnetization, like stress and heating, 

 is one of the events that leave an imprint, one of the experiences 

 which the metal remembers. If two I-vs.-H curves are measured in 

 succession, the second is generally not the same as the first; during 

 the process of ascertaining the first, the material was changed into a 

 new one. To predict or classify an I-vs.-H curve, one must not 

 only know the composition of the substance, not only have records 

 of its entire mechanical and thermal history since it was last rendered 

 forgetful of its past by annealing, but also have the protocol of all its 

 magnetizings since the last occasion when it was "completely de- 

 magnetized" — whether by the annealing which effaced all the mem- 

 ories, or by the gentler process f prescribed by Ewing which cancels 

 the imprints of past magnetizations without destroying those of past 

 stresses and heatings. 



To make some choice among this staggering mass of data, it is 

 suitable to concentrate one's attention on two, or rather on one and 

 a group, of the infinite multitude of curves. The first of the chosen 

 curves is obtained by applying to a sample which is freshly demagne- 

 tized a magnetizihg field He which is increased by consecutive small 

 steps, and measuring the field of the magnet after, or (by the method 

 of the loop) the increase of the induction in the magnet during, each 

 of these steps. From either of these sets of data, after making the 

 allowances and the reductions indicated in the first section of this 

 article, one may determine the I-vs.-H curve for steadily-increasing 

 magnetizing fields applied to a piece of metal initially demagnetized. + 



* I use the word "anneal" to denote a long-continued maintenance at a high 

 temperature, irrespective of the rate of cooling thereafter. 



t By applying an alternating magnetic field of which the amplitude is at first 

 greater than any field which has been applied to the magnet, and thence diminishes 

 gradually to zero. However, the effect of this process is not quite thoroughgoing. 



X There is a risk that the increase in magnetization at a certain step may be so 

 great that, when due allowance is made for the demagnetizing effect of the magnet 

 upon itself, it will be found that H has actually decreased in spite of the increase 

 of H»- 



