438 On Iron and Steel at a Bright-Red Heat. 



torsional stresses, which differ not only in magnitude but 

 even in direction in different annuli of the wire's section 

 when the external stress is removed ; that is to say, if w r e 

 regard the wire in this condition as divided into two parts, 

 a core and an enclosing tube or rind, we may choose the radius 

 of the core so that the internal stresses in the core are opposed 

 to those in the rind. If by any means we can weaken the 

 rigidity or elasticity in one part more quickly than that in the 

 other, it is clear that we should get the permanent strain 

 either increased or diminished accordingly. Here, then, we 

 have another way in which the results recorded in Mr. 

 Tomlinson ; s paper, " On the Effect of Change of Temperature 

 in Twisting or Untwisting Wires " (p. 253), may be regarded. 

 I will not dwell upon these, but will deal with Experiment IV., 

 in which a wire permanently twisted is heated, and shows 

 twist in heating and untwist in cooling. We have then to 

 show that in heating the inner core is heated more quickly 

 than the outer rind, in which case the wire will untwist. 

 This case presents little difficulty, except in that at one point 

 during the heating the core may be cooler than the rind, 

 namely, if the wire shows " darkening/'' with a fall of tem- 

 perature. Next we have to show that at a point in the cooling 

 the outer rind can become hotter than the core. Now in 

 cooling, the rind reaches the temperature at which reglow 

 occurs before the core ; and so it is possible that the rind 

 may become considerably hotter than the core, its rigidity also 

 becoming less, and the wire therefore is untwisted. The 

 suddenness of the twist or untwist depends on the shortness 

 of the period (in both heating and cooling) during which the 

 rigidities of the core and rind differ markedly. At low 

 temperatures the rigidity apparently does not vary much with 

 temperature, or at least there is not much greater variation 

 of rigidity in the permanently strained rind than in the core, 

 and the internal couples are almost balanced. But at a certain 

 temperature the rigidity suddenly falls, as is shown by 

 Experiment I., and the balance of couples in the unequally 

 heated wire is destroyed until the wire is heated so much 

 above this critical temperature that an approximate balance 

 of weaker couples is again established. If the temperature 

 of the wire is kept at about a dull red heat (as seen from 

 outside), it is possible to keep the balance of couples so 

 far disturbed as to get nearly all the permanent twist out of 

 the wire. 



The explanation above suggested is based on (1) the fact 

 that the reglow is connected with a rise of temperature — this 



