374 PHILOSOPHICAL TRANSACTIONS. [ANNO 1798. 



sometimes, that the crystallization has not been so perfect as to destroy every 

 appearance of the faces of the primitive rhomboidal parallelopiped; in this case, 

 there remains on the solid angle of 112°, formed by the junction of the new 

 faces with the edges of the prism, a small isosceles triangle, as in fig. 13, which 

 corresponds to those in fig. 5 of the preceding modification. 



The crystals which explained the 2d modification, form also a part of Mr. Gre- 

 ville's collection: one in particular is highly worthy of notice; it is the most per- 

 fect crystal I have ever seen of this substance. The surface of the faces of the 

 prism, though rough, is infinitely less so than that of the others, and much more 

 brilliant. The planes on the ends have the usual polish of crystals; its colour is a 

 pale red, and its transparency may be compared to that of wax. This substance 

 presents a 3d modification, in which the hexaedral prism diminishes in diameter, 

 as is apparent by comparing the diameters of its 2 ends; in some, it appears like 

 a regular hexaedral pyramid truncated, as fig. 15. The crystals of this modifica- 

 tion are usually irregular, and seldom admit of a certain measure of their angles; 

 but among the numerous specimens in Mr. Greville's collection, I have been able 

 to ascertain, in the greater part, that the hexagonal plane at the top forms angles 

 of about 120° with the planes of the pyramid; and the hexagonal plane at the base 

 forms angles of about 78° with the planes of the pyramid. In other instances, 

 the form of the pyramid varies greatly ; in some, the angle at the upper plane was 

 1 10°, and the angle at the base about 70°; in others, the angle at the upper plane 

 was about 100°, and the one at the lower plane about 80°. 



In these 3 varieties, the crystalline laminae can be separated, as in the hexagonal 

 prism, at the 3 solid alternate angles of each end, but in a contrary direction to 

 each other. The planes which appear when the laminae are detached regularly, 

 form solid angles of 22° 34', with the planes of the extremity: this arrangement 

 is analogous to that of the hexaedral prism. The difference of form arises from 

 the crystalline laminae deposited on the planes of the primitive rhomboid, decreas- 

 ing by more than one row of molecules, on the planes of one of the triedral pyra- 

 mids of the rhomboid, and by less than one row, on the planes of its other pyra- 

 mid. This general observation, on the manner in which this primitive crystal of 

 corundum passes to the different varieties just mentioned, is the only one I have 

 established with any great degree of certainty at present. Specimens with perfect 

 crystals, whose angles may be measured with accuracy, will probably arrive from 

 India, and give further demonstration, as to these and other varieties of modifica- 

 tions of corundum. We may conceive, that if, in this modification, the crystalli- 

 zation had ceased before the entire formation of the crystal, there would have re- 

 mained small isosceles triangular planes, on 3 of the alternate solid angles, formed 

 by the junction of the planes on the ends, with the edges of the truncated pyramid. 

 These isosceles triangular planes resemble those we have seen in the first modifica- 

 tion; (fig. 4 and 5), and form, in the same manner, solid angles of 122° 34', 

 with the planes on the ends of the prism. (Fig. l6). 



