cube in capillary crystals, of which I possess a specimen from Tol Carn mine, in recomposed granite, and of the most lively carmine colour. Fig. 32 is an octahedron formed of minute cubes, of which there are several on the same specimen. Fig. 33 shews the passage of the cube into the rhomboidal dodecahedron by the deposition of the cubic facets, progressively diminishing in size, on each face of the cube. This interesting crystal at first excited the suspicion that the cube is the primitive form of the red oxyd, which abated on reflecting that the octahedron will admit a fracture in the direction of its faces, and that the cube will not, as has been already noticed: the direction of the laminæ in both cases is shewn by figs. 24 and 37. These circumstances indeed prove the octahedron to be the form of the primitive crystal.
SECOND MODIFICATION.
This modification arises from a decrease along the edges of the primitive crystal, which replaces each by a plane perpendicular to the axis that passes through the middle of the edges. It shews the passage of the primitive form into the rhomboidal dodecahedron. The angle formed by the meeting of the face P with the plane 2, Fig. 34, is 144d. 44′, 8″ as given by Haüy. The striæ on fig. 37, which shews the passage of the primitive form into the rhomboidal dodecahedron, denote the direction of the laminæ. In fig. 41, which resembles in its general form the crystallization of the oxyd of tin, the decrease on the edges formed by the meeting of the two pyramids is so considerable, as to give the shape of a parallelogram to the four planes which replace the four solid angles, also formed by the meeting of the two pyramids.
