262 P. W. Hriihjnnin— Failure of Cavities in Crystals 



nution in the size of the bubble, indicating that there had 

 been no permanent change of dimensions. A disturbing 

 question naturally arises here ; did not the liquid in the 

 cavity afford important support to the walls from the 

 inside, after the external pressure had slightly decreased 

 the size of the cavity? But such cavities are usually 

 filled with C0 2 , and in this case the support would be 

 quite inappreciable, because of the high compressibility of 

 this gas. However, even if the cavity had been initially 

 entirely filled with water, a simple computation shows 

 that the maximum internal pressure would have been of 

 the order of 1000 kg., which may be neglected in compari- 

 son with 18,000. 



Some parts of the original quartz crystals remote from 

 the negative crystals were smoky in appearance, and 

 under the microscope, before the application of pressure, 

 were seen to contain minute bubbles of quite irregular 

 shape. In these regions a positive effect was produced 

 by pressure, the appearance being that the interior of the 

 bubbles had been eroded and tightly packed with sand, 

 exactly as had the larger cavities in the cylindrical speci- 

 mens. The same effect had been previously observed in 

 a small region in the end of the second quartz specimen. 



The conclusion must not be drawn that the flaking-off 

 process is an end-effect connected with the unavoidable 

 surface of separation, for this flaking-off was repeatedly 

 observed at all points of the cavity, irrespective of the 

 distance from the mouth. For instance, the fine splinters 

 referred to in the experiment on andesite were distributed 

 quite uniformly over the entire length of the brass wire. 

 The cracks, however, which frequently appear, and are 

 usually especially prominent near the surface of separa- 

 tion, may in the majority of cases be intimately con- 

 nected with imperfect joining. 



With respect to flaking-off, the conclusion seems to be 

 that the state of polish of the surface is an important 

 factor; if the polish is complete down to molecular 

 dimensions, as it was for the negative crystals, the ten- 

 dency to flake off is much less than if there are grosser 

 irregularities in the surface structure. The polish must 

 be very complete indeed to produce an appreciable effect ; 

 no artificial polish that could be applied to the interior of 

 the specimens above had any effect. The first specimens 

 were left rough after drilling the hole, but the later ones 

 were polished. 



