BRUSH— GENERATION OF HEAT IN STEEL. 357 



vacuum jars. Both the steel and the water were usually brought 

 to the same room temperature before being placed in the calorimeter. 

 Fifty-five scale divisions indicate a temperature difference of i° C. 



The curve of normal cooling runs out of the figure at the upper 

 left hand corner, and is easily distinguished from the others. This 

 curve was obtained from a quantity of untreated steel equal in 

 weight to " specimen B," and warmed a few degrees above room 

 temperature before being placed in the calorimeter. It shows the 

 normal loss of heat due to imperfect thermal insulation alone, and 

 is the basis of comparison for all the other curves. Obviously this 

 curve may be plotted further to the right or left without impairing 

 its validity ; and it may be plotted to intersect any of the other curves 

 at any desired point, to facilitate study of the other curve at and 

 near the intersection. For my own convenience I have constructed 

 a metal template of the normal cooling curve, and find it most use- 

 ful. Of course it is necessary that the base of the template be always 

 kept coincident with the base line of the curve sheet. 



The curve of "first hardening" shows the spontaneous genera- 

 tion of heat which followed the first quenching at about 'JJ'J^ , the 

 temperature of complete loss of magnetic susceptibility, after which 

 the scleroscope hardness must have been about 74. 



The curve of second hardening, indicated by " 2/j," shows con- 

 siderably greater generation of heat. Quenching temperature and 

 hardness were not observed ; but it is known that the quenching 

 temperature was much higher than 'jyy'' ■ 



The three curves thus far discussed were shown in the " second 

 paper " already referred to, and the other curves here shown were 

 subsequently plotted on the original curve sheet. 



The third curve showing spontaneous generation of heat is indi- 

 cated by " 3^/," meaning third quenching (not hardening). To make 

 it clear that heat was generated in this case I have drawn the curve 

 of normal cooling in a position for easy comparison (the upper 

 dotted line). The "35" curve was described in the second paper, 

 but not plotted. The quenching temperature in this case must have 

 been slightly below 660°, and hardness only about 2)7 • 



" Specimen B," left in the annealed condition at the close of 



