i>.'>4 P. W. Bridgman — Failure of Cavities in Crystals 



om> of the cylinders, which was in form an elongated 

 ellipse with pointed ends, extended entirely across the 

 cylinder. The outer surface of this piece had in places 

 been displaced bodily into the cavity. In the other piece, 

 however, the erosion was much less extensive, and the 

 outer surface had received no permanent change. In this 

 second piece the most extensive erosion was not at the 

 mouth of the hole but was nearer the bottom, showing 

 that the flaking-off is not an end effect. This remark is 

 confirmed by many other observations ; the flaking-off has 

 no relation to the end effect, and may take place at any 

 point on the interior surface. 



The second specimen of tourmaline was entirely 

 opaque, so that no intimation could be obtained as to its 

 internal structure; there were no flaws apparent in the 

 original piece. The tests on this specimen were espe- 

 cially designed to find the effect of the element of time. 

 It was exposed to 5100 kg. for 28y 2 hours, to 6630 for 46U< 

 hours, to 8170 for 38y 2 hours, to 9500 for 38y 2 hours, and 

 to 12,200 for 4014 hours. This specimen was much 

 stronger than the first. Up to and including 9570 there 

 was practically no effect. The edges of the mouths of the 

 holes splintered a little, as they always do, because of 

 imperfect register of the two parts, and there were a few 

 barely perceptible flakes from the inside. At 12,200, 

 however, the failure of the interior was complete, as it 

 had been for the other specimen. There was no flow or 

 permanent alteration of the outside. The average density 

 of the sand with which the eroded cavity was packed was 

 215, against 309 for the original crystal. The inter- 

 stitial space in the sand was therefore about 30%. 



The change in internal dimensions of the cavity, found 

 from measurements on the solder plugs, was at the rate 

 of 0-65% per 12,000 kg. for the first, and 0-74% for the sec- 

 ond specimen. Using Voigt's values for the elastic con- 

 stants of tourmaline (see Love, page 157) the theoretical 

 value is. about 0-90%. Tourmaline varies greatly in its 

 properties ; the discrepancy may therefore be due to dif- 

 ference in the specimens. Here again, as for quartz, the 

 relation between distortion and stress remained linear 

 within the errors of measurement, which were large. 



Calcite. — The interest in this material lies in its extra- 

 ordinarily easily developed planes of cleavage, whereas 

 both quartz and tourmaline show practically no such 



