Cooling exhibited by Glass and by Steel. 451 



or fissure discrepancy, or else for like reasons to raise the 

 parts of the curves corresponding to the first phase. If this be 

 done the circumflexures become less pronounced or disappear, 

 and the points for the commercial soft state are more easily 

 referred to the curves to which they belong. If, therefore, 

 allowance be made for the distortion due to internal sensible 

 pores, the relations become more uniform. 



It is interesting to observe that glass retains the volume-ex- 

 pansion corresponding to a temperature somewhat below 900° ; 

 whereas steel retains the volume-expansion of a temperature 

 below 400°. Similarly the strain in steel is very perceptibly 

 affected by annealing temperatures as low as 50°, whereas for 

 glass the perceptible annealing effects are only incipient even 

 at 200°. The amount of strain retained is, ccet. par., not merely 

 a function of the viscosity of the material subjected to quench- 

 ing. It must depend also on the heat conductivity of this 

 substance. For instance, if like figures of glass and of steel be 

 quenched alike, then at the same time and depth the thermal 

 gradient would be much steeper in the case of glass than in the 

 case of steel. Hence, ccet. par., during quenching a rigid 

 shell is possible in the case of glass for higher temperatures of 

 the core than in the case of steel. The sudden contraction of 

 the shell pressure has a marked effect on the melting point or 

 degree of fluidity of the core. But it is best to waive this 

 observation here, for the want of data to interpret it. 



We have finally to consider the bearing of the results of 

 this paper on the structure of steel. The detailed similarity 

 observed in the annealing of glass and of steel suggest the in- 

 ference that the interior of hard steel may be sensibly fissured. 

 Under all circumstances the diagrammatic structure of dense 

 shell and rare core would be inaccurate to the extent in> which 

 these fissures are irregularly distributed. Hence the difficulty 

 of developing the true character of the dependence of the 

 density (d) at any point upon the distance of this point below 

 the surface. We are able to account in part for the quasi-har- 

 monic relations obtained both by Dr. From me* and by our- 

 selves, f While the consecutive shells are being removed by 

 solution, periodic fluctuation of o must result whenever fis- 

 sures are invaded. If, furthermore, we take into consideration 

 that the density effect of even great intensities of temper-strain 

 is small, it appears that the true nature of the strained structure 

 of tempered steel may be beyond the discernment of the den- 

 sity-method of investigation, altogether. The gross variation 

 of density along the radius is a carburation phenomenon. 



Retrospective. 

 In our earlier work on steel we adverted to the singular co- 



* Wied. Ann., viii, p. 356, 1879. f Bull. 35. This volume, p. 386. 



