126 Mr. H. Nagaoka on the Effects of Torsion and 



a particular load. For loads greater than this particular load 

 the negative magnetization decreases, and at last completely 

 vanishes. Thus, for stress W = 1770 kilogs., there is no 

 opposite magnetization therein agreeing with the curves for 

 loads smaller than that for which the negative magnetization 

 first appears. But there is the difference that the curve is a 

 single loop in which cf^jdr changes sign during negative 

 twisting and untwisting. 



Taking into account all the series of experiments, we see 

 that the greatest valae in each series of the ratio +!j/~!j 

 tends to increase as the field is increased. The strength of 

 field in which these last experiments were tried seems to be 

 about the critical value for which we can get the two 

 transitions of magnetization — namely the change of sign for 

 particular load and the vanishing of this change for a higher 

 load. 



The peculiarities which have just been the subject of dis- 

 cussion do not, however, persist at all strengths of fields. 

 At still higher fields a different order of things comes in. 

 Take, for example, the next series of experiments with a field 

 of 8-06, as shown in figs. 18, 19, 20, 21, 22. First of all, for 

 the lower tensions, the two-looped curve is more symmetrical 

 than it can be obtained at lower fields (see fig. 18). In fig. 

 19, for a tension of 782, a symmetry begins to show itself; 

 but the diminution, instead of taking place in the left-hand or 

 negative loop, takes place in the right-hand or positive loop. 

 Figs. 20, 21 show the gradual vanishing away of this right- 

 hand loop as the load is increased. Also, exactly as in the 

 former sets of experiments, the curve dips below the zero- 

 magnetization line, as the right-hand portion loses its loop- 

 character. Now if we were to co^npare fig. 21 with figs. 17, 

 12, and 8, 2, and take no account of the intermediate links 

 of development, we should at once regard them as being of 

 essentially opposite character. In the four earlier cases 

 positive twist increased the magnetization and negative twist 

 diminished it ; but in the present case the effects are exactly 

 opposite. In the same way fig. 20 presents features quite 

 opposite to those presented by figs. 1, 7, 11, and 16. How- 

 ever, just as in the former sets of experiments, the twist at 

 which d^/dr vanishes was shown to shift gradually to the 

 negative side as the load was increased ; so in the present 

 case, the twist at which d'j,/dT vanishes shifts gradually to the 

 right as the load is increased. These very interesting reversals 

 of effects clearly depend on the strength of the field. Then, 

 again, the particular strength of field at which the reversal of 

 these effects begins to show itself seems to m connected with 

 the fact already discussed, that for a particular field the ratio 



