Cooling exhibited by Glass and by Steel. 443 



Discussion. — It is not our object to enter into a full discussion 

 ■of these results here. We merely advert to the fact that 

 whereas density increases continually on passing from hard to 

 soft, resistance passes through a pronounced minimum. 



Whether we regard J as a function of s or of the tempera- 

 ture i° at which the rod was annealed, the relations contained 

 in 'the table lead to an important inference, with an immediate 

 bearing on our present purposes. It appears plainly that the 

 annealing of a rod successively from extreme hard to extreme soft 

 does not exhibit the characteristics of a single and homogeneous 

 phenomenon. This annealing presents two distinct and inde- 

 pendent phases which merge into each other when t° corre- 

 sponds to incipient redness. A graphic representation of J 

 varying with t° shows this transition with singular clearness. 

 Between 2=200° and 2=500°, all the loci pass through two 

 consecutive ci re um flexures in such a way as to change the 

 general and pretty uniform contours of concavity downwards, 

 into convexity downwards for the interval in question. The 

 division of the complete phenomenon into two parts being thus 

 suggested, we find on further inspection that the variations of 

 the electrical constants of temper subside almost completely 

 within the first of these parts (annealing between 0° and 350°) ; 

 whereas in the second part (annealing between 350° and 1000°) 

 they became crowded and complex. It follows plausibly and 

 at once that during the first of the phases under consideration 

 we encounter the subsidence of a mechanical strain. This 

 inference gains much in weight when we find that this strain van- 

 ishes in the way peculiar to phenomena of viscosity, at a slowly 

 decreasing rate, through infinite time.* In the second phase, 

 mechanical and chemical occurrences are superimposed and the 

 above results do not enable us to disentangle them. 



The final important deduction from the data is this, that dur- 

 ing the first phase in which the larger values of resistance are 

 from two to three times the smaller values, the variation of 

 density is slight. In the second phase we encounter large vari- 

 ations of density associated with small variations of resistance. 

 These curious relations characterize the phenomenon as a 

 whole. We shall interpret them partially at least below. 



Density and Strains in Tempered Glass. 



Experimental results. — This result substantiates the evidence 

 in favor of the existence of strain in hard steel, adduced in our 

 earlier papers.f Beyond this it shows almost conclusively 



*In other words the temper strain vanishes just like the simple "drawn" 

 strain imparted by the wire plate, while the wires carrying them are each exposed 

 to the prolonged action of temperature. — Bull. 14, p. 93. 



fBull. 14, pp.88 to 103. 



