CAVITIES IN MINERALS. 379 



in the cavity, or the spaces A m ra, C o p, either as filled 

 with solid matter, or as corners into which the expanding 

 fluid would not penetrate. The regular curvature, how- 

 ever, of the boundary line m n o p, and other facts, 

 rendered these suppositions untenable. 



This difficulty was at last entirely removed by the 

 discovery of a cavity of the form shown in the annexed 

 figure, where A, B, and C are three portions of the 

 expansible fluid separated by the interposition of the 

 second fluid D E F. The first portion A of the expansible 

 fluid had four vacuities V, X, Y, Z, while the other two 

 portions B, C had no vacuity. In order to determine if 

 the vacuities of the portions B, C had passed over to A, I 



Fig. 78. 



took an accurate drawing of the appearances at a tempera- 

 ture of 50, as shown in the figure, and I watched the 

 changes which took place in raising the temperature to 

 83. The portion A gradually expanded itself till it 

 filled up all the four vacuities V, X, Y, and Z ; but as 

 the portions B, C had no vacuities, they could expand 

 themselves only by pushing back the supposed second 

 fluid D E F. This effect actually took place. The 

 dense fluid quitted the side of the cavity at F. The two 

 portions B, C of the expansible fluid instantly united, and 

 the dense fluid having retreated to the limit m n n o, its 

 other limit advanced to p q r, thus proving it to be a real 

 fluid. This experiment, which I have often shown to 

 others, involves one of those rare combinations of circum- 



