50 MM. Fremy and Feil on the Artificial Production of 



obliquely with respect to their edges. Doubtless they belong 

 to one of the oblique systems- — the oblique prism with rhombic 

 base, or the doubly oblique prism. These crystals gave the 

 following composition : — 



Silica 47-65 



Alumina 5 1*85 



Loss 0-50 



This comes near to the composition of natural kyanite *. 



The action of fluoride of aluminium on boric acid gave a 

 crystallized borate of alumina which corresponds to kyanite. 

 We are at present carrying out a series of trials in which 

 fluoride of aluminium will act upon other mineral acids. 



The important fact of the volatility of the fluoride of alumi- 

 nium, discovered by M. Henri Sainte-Claire Deville, enables 

 us readily to explain the remaining experiments. When a 

 mixture of equal weights of alumina and fluoride of barium, 

 into which has been introduced 2 or 3 per cent, of bichromate 

 of potass, is heated to and maintained at a very high tempera- 

 ture during a long time, a crystallized mass is obtained the 

 study of which is of the greatest interest. If the calcination 

 has been effected in a crucible covered with another (which 

 serves in some sort as a condenser), two sorts of crystals are 

 found in the crucibles : the one sort are long colourless prisms, 

 often several centims. in length, and presenting the aspect of 

 the silvery flowers of antimony ; the others are ruby crystals, 

 remarkable for the regularity of their forms and beautiful 

 rose-colour. 



The long prismatic colourless crystals are formed by a 

 double silicate of baryta and alumina, which present the 

 composition : — 



Silica 34-32 



Baryta 35'04 



Alumina 30'37 



In our essays this double silicate often crystallized in rather 

 short, hard and transparent, clinorhombic prisms which, M. 

 Terreil has ascertained, have the same composition as the long 

 and hollow prismatic needles. 



M. Jannettaz has proved that the long prisms are often con- 

 stituted by four plates with parallel faces, forming the faces of 

 a hollow prism. These plates are very thin ; under the micro- 

 scope they extinguish light ; or rather they let darkness persist 

 between two Nicols, parallel to their mutual intersections ; the 



* The crystals we obtained are very easily produced, but are not of 

 large size ; they may therefore belong to those fibrous varieties of dysthene 

 ■which have been described under the names of Fribolite, Bucholzite, 

 Bamlite, and Sillimanite. 



