98 REPORTS OF inve;stigations and projects. 



zellan Alanufactur suggested the existence of a second compound in this 

 series, but their preparations were heated in the presence of carbon and the 

 temperature measurements made with Se.c:er cones, so that the results are not 

 quite convincing. Although the region in which this supposed compound oc- 

 curred has been examined with the greatest care, neither a maximum nor any 

 new solid phase has been discovered. Between the composition of sillimanite 

 and pure alumina the solid phases are sillimanite and alumina, the latter con- 

 taining a small admixture of silica in solid solution, indicated by its slightly 

 altered properties. The sillimanite-alumina eutectic falls at about 6"] per cent 

 alumina. Between sillimanite and silica the solid phases are tridymite and 

 sillimanite, with a eutectic w-hich is somewhat difficult to locate. With in- 

 creasing quantities of silica, the viscosity of the melt makes such observations 

 very difficult, and all members of the series melt at temperatures too high 

 for determination with the thermocouple. 



The attempt to prepare the two modifications — andalusite and cyanite— 

 which occur in nature with the same composition as sillimanite, has not been 

 successful. After overcoming considerable experimental difficulties, we suc- 

 ceeded in obtaining a small quantity of sillimanite glass, but the recrystalliza- 

 tion of this glass, either at low temperatures, dry, or with lo per cent sodium 

 chloride in bombs, resulted invariably in the formation of sillimanite. At- 

 tempts to produce either of the m.odifications by metathetical reactions have 

 thus far yielded no positive results, though traces of crystals similar to anda- 

 lusite appeared in a few of the preparations. We have been able to show that 

 the change from either of these modifications to sillimanite is not reversible. 

 When heated to temperatures above 1,200° both go over slowly into silliman- 

 ite, the change for small charges of finely powdered material requiring about 

 24 hours (at 1,200°). In bombs at temperatures between 400° and 500° the 

 two minerals are not appreciably affected. It is worth noting that while anda- 

 lusite when heated readily changes into a substance resembling sillimanite, 

 the cyanite appears rather to decompose without yielding any readily identifi- 

 able product. At higher temperatures, approaching the fusion-point of silli- 

 manite, both give very fair sillimanite. 



In the lime-alumina series, both microscopical and thermal examinations 

 have been made of all the compositions with fair success. The existence of 

 three definite maxima with corresponding eutectics has been established — 

 CaO.ALOg, aALOg.sCaO, and ALOs.sCaO. The first and last of these 

 compounds melt above 1,600° and have therefore to be determined with the 

 optical pyrometer. The intermediate compound at 52 per cent is well within 

 the range of the thermocouple. The i : i compound is a very definite, well- 

 crystallized, birefracting compound which has been shown to be without other 

 modifications. The 3 : 5 compound exists in two forms. In its stable con- 

 dition it is isotropic, but it also appears in the birefracting form in monotropic 

 relation to the first. The compound ALOs-sCaO, occurring at about 37.78 per 



