450 C. Barus and V. Strouhal — Strain-effect of sudden 



homogeneous glass drop must be heated in order that the 

 observed mean volume of the soft state may be equal to the 

 observed mean volume of the hard state. Then 



; . (< ^^r (8) 



where a is the ordinary coefficient of linear expansion of glass. 

 By inserting the values of Table III into (3) we derive 



£< 900°. 

 In other words the quenched globule has retained the volume 

 which the hot glass possessed at a temperature certainly smaller 

 than 900°. This result taken together with the other is con- 

 clusive.. 



The general result of our investigation on the process of 

 sudden cooling, therefore, points out the fundamental impor- 

 tance of a rigid shell. As the material (glass or steel) cools, 

 contraction as a whole takes place not centripetally but centrif- 

 ugally, i. e., not towards the center of figure but away from it. 

 So long as the interior remains liquid or viscous, the result is 

 simple separation, commencing at points where minute air 

 bubbles may preexist,* rather than at points of the continuity 

 of glass. The final result is a vacuum bubble such as we 

 have observed it. In proportion as the interior becomes rigid, 

 the temper strain appears. Should the glass hold gases in solu- 

 tion, they would tend to escape into the vacua in question. 

 There is probably another reason why in most instances the 

 bubbles cannot be brought completely to vanish. 



Annealing reverses the whole phenomenon. The small 

 changes of density observed in the first phase may be easily 

 accounted for; it is probable that during these stages of anneal- 

 ing the re-arrangement of molecules and disappearance of strain 

 is accompanied by expansion inward, toward the vacua. Hence 

 the incommensurately small variation of mean density which 

 accompanies the great optical effects (glass), and the great 

 electrical effects (steel).f During the second stage of anneal- 

 ing, again, the density-effect is incommensurately large ; for 

 glass it is the expression of a mere collapse due to atmospheric 

 pressure. At all events the density-effect (error) due to bub- 

 bles quite swallows up the density-effect due to strain. Hence 

 to represent the true relation of density and resistance or of 

 density and annealing temperature in case of steel, it may be 

 necessary to lower the part of the curves corresponding to the 

 second place of annealing by amounts equivalent to the bubble 



*P. R. drop No. 1, heated intensely in a blast lamp and cooled in air shows a 

 decrement of density. This is due to the rapid cooling. For when this drop is 

 heated and cooled in the crucible, density is again incremented. The experiment 

 shows the tendency of glass to retain strains. During the heating to 1000° no- 

 expansion of the very small bubbles was observable. 



f The electrical and optical criteria may be considered equally sensitive. 



