Alumina with Silica, Lime and Magnesia. „ 305 



attacked by the reagent, we have not jet been successful in 

 our attempts to prepare andalusite and cyanite. Attempts to 

 produce reactions between the hydrated oxide of silicon, i. e., 

 the various hypothetical silicic acids, and alumina, hydrated 

 or dry, gave no positive results. Certain reactions involving 

 the attack upon anorthite and other minerals by A1 2 (S0 4 ) 3 or 

 A1C1 3 , have yielded but little promise of success. A1C1 3 acting 

 on anorthite gave small spherulites of which the determinable 

 properties agree with andalusite, but these crystals were too 

 small for positive identification. We did not obtain positive 

 results by the action of fluxes on the oxides. 



Incidentally, we observed the formation of meionite by 

 crystallizing glass of that composition in a bomb with 10 per 

 cent JS"aCl solution, and grossularite by the reaction of A1C1 3 

 on calcium orthosilicate. 



Sillimanite glass. — Sillimanite melts sharply to a very thin 

 liquid which crystallizes with great rapidity. Dropping the 

 crucible of molten sillimanite in water does not cool even a 

 small charge rapidly enough to prevent 

 crystallization. We used the system 

 shown in fig. 5. A is an iridium 

 crucible with a 3 mm hole in the bot- 

 tom. It is supported in the furnace 

 tube, E, on a magnesia ring, B, which 

 is in turn supported by small mag- 

 nesia rods, C, and the ring, D. The 

 melted charge drops from the tip of 

 the cone directly into the clish of 

 water, F. In this way we obtained 

 a number of globules of glass, with 

 an index of refraction of about 1*625 

 and a density of 2*51, much lower 

 than sillimanite. When crystallized, 

 either by heat at 1200° or in bombs 

 containing 10 per cent NaCl solution 

 at 350 c -400°, the glass yielded only 

 sillimanite. It seems reasonably cer- 

 tain that sillimanite is the stable form 

 and that andalusite and cyanite are 

 formed by crystallization from solu- 

 tion at low temperatures. 



Lime-Alumina. — It is much more 

 satisfactory to deal with systems in 

 which at least a part of the liquidus 

 can be definitely located with the 

 thermoelement. From about 15 per cent to 70 per cent of 

 alumina, this series can be melted in platinum and all of the 



