Viscosity of Solids and its Physical Verification. 213 



resin*. Reckoned from the observed volume-increase f due to 

 quenching, the stress-intensity corresponding to the observed 

 strain may be estimated at 10 10 dynes per square centiin. in 

 steel and 10 9 dynes per square centim. in glass. It is thus of 

 the order of the respective tenacities of steel and of glass. In 

 view of the fact that the viscosity of hard steel is not above 

 that of glass i, exceptionally great strain-intensity would not 

 be permanently retained. Hence the secular changes of glass- 

 hard steel. At this point, moreover, the function of carbon 

 appears. Sudden cooling from red heat induces iron and 

 carbon to remain in the combined state, in a way favourable 

 to the observed dilatation. Throughout the process of cooling, 

 carbon and iron at any place within the metal are united in 

 conformity with the given degree of carburation, and with the 

 intensity of strain there experienced. In the cold metal, at 

 the given place, strain is to a certain extent permanent, and 

 independent of the strain of the surrounding medium of steel§. 

 Hence if, by gradual secular annealing of massive glass-hard 

 steel, a sufficient number of carbon configurations are broken, 

 stress may increase to an intensity sufficient to rupture the 

 metal explosively. 



In our earlier papers on this subject Dr. Strouhal and I 

 were much puzzled to know whether the temper-strain, and 

 in general the phenomena of annealing, were to be interpreted 

 physically or chemically ; whether annealing was a case of 

 viscous subsidence of the temper-strain, or a case of mere 

 chemical decomposition. In the light of the present advanced 

 conceptions this distinction is superfluous. It makes no dif- 

 ference whether the configuration breaks up into parts chemi- 

 cally different (as carbon and iron, say, in steel), or into parts 

 chemically, though not structurally, identical (as in homoge- 

 neous metals) . Viscosity is conditioned by the degree of in- 

 stability. Again,, it is clear that the principles which account 

 for the subsidence of the mechanical strain will also account 

 at once for such chemical decomposition as is here in question; 

 the difference of the two cases being vested in mere details of 

 molecular mechanism. 



16. However complex the nature of the temper-strain in 

 steel may be, the behaviour of hard steel, when subjected 



* Marangoni, N. Cim. [3] v. p. 116, 1879 (Kupert's drops of resin) ; 

 De Luynes, C. R. lxxvi. p. 346, 1873, or Phil. Mag. [4] xlv. p. 464, 

 1873 (Rupert's drops of glass). 



t Am. Journ. xxxi. pp. 441, 443; xxxii. p. 191, 1886; xxxiii. p. 33 

 1887 ; Bull U. S. G. S., no. 27, pp. 30 to 50, 1880. 



X Am. Journ. xxxiii. p. 30, 18b7. 



§ Bulletin U. S. G. S., no. 35, p. 42, 1886. Structure studied by the 

 density method, shells being consecutively removed by galvanic solution. 



