72 Mr. P. W. Bridgman on Breaking Tests under 



cylinder in a state of ease, and since the diagram show- 

 plainly that the beginning of yield is not sharply defined, 

 there is here no possibility of a natural initial yield-point to 

 which a criterion might bo applied. 



So far this collapsing test has been discussed only in its 

 application to ductile materials like copper or steel. The 

 behaviour of a brittle materia] like glass under the same 

 conditions is strikingly different. Cylinders of glass made 

 of heavy capillary tubing sealed at both ends, and hollow 

 spheres with thick walls, have been subjected to hydrostatic 

 pressure up to 24,000 atmos. No permanent measurable 

 change is produced, there being neither crushing nor 

 alteration in the dimensions of measurable amount. It 

 should be said, however, that there must be some slight 

 amount of flow and of interior adjustment to the pressure, 

 although too small to measure, because several of the glass 

 cylinders have broken spontaneously several weeks or even 

 months after the release of pressure. Others have been 

 kept without fracture for a couple of years. One would 

 expect that under the conditions of this test the glas> 

 would be crushed. This is the case if the cylinder has not 

 been caYefully annealed or if it is geometrically imperfect ; 

 the material may then be reduced to an almost im- 

 palpable powder. This complete destruction of the minutest 

 fragments of the glass apparently is because the wave of 

 expansion, travelling through the mass after the break has 

 started, at any point is of such intensity that every minute 

 portion is reduced to powder by its own inertia. 



For a brittle substance like glass which shows no flow we 

 may calculate the distribution of stress by the ordinary 

 theory of elasticity. It is thus found that the maximum 

 stress difference will occur at the inner surface of the 

 cylinder, and for a thick cylinder will be equal in amount to 

 i lie external hydrostatic pressure: We haw already seen 

 that if the maximum stress difference criterion were valid, 

 the greatest Btress difference that the material could support 

 would be equal to its tensile strength. Therefore since the 

 tensile strength of glass is seldom as high as 7000 lbs/in 1 , 

 we should have had crushing under the conditions of the 

 test at a hydrostatic pressure no higher than 7000 lb- in'-', it 

 (ho maximum stress difference theory were valid Eor brittle 



substances. I > 1 1 1 fifty times this value has been reached 



without rupture. 



The result of this experiment with the glass seems even 

 more unexpected than that with the ductile materials. 

 With the metals we have at least the rearrangement of the 



