326 Shepherd, Rankin, Wright — Binary Systems of 



MgO.Al,0 3 : Artificial spinel, MgO 38-32, Al 9 t 71-68.— 

 Crystals of this compound were obtained by direct crystalliza- 

 tion from the pure melt in the electric arc furnace. The 

 resulting crystals were minute but sharply developed octahe- 

 drons, clear and transparent, colorless and splendent in luster. 

 In the aggregate they occur frequently with approximately 

 parallel orientation in rows and clusters, not unlike skeleton 

 salt crystals in appearance. — Evidence of twinning after the 

 usual spinel law was sought for but without decisive results, 

 chiefly because of the fineness of the material. Cleavage if 

 present is imperfect and not well marked in the powdered 

 material. The hardness is about 8. The refractive index, 

 n = 1/723 db *002 (determined by immersion method), is slightly 

 higher than that of spinel ordinarily, although' spinels of even 

 higher refractive index have been observed. Under the 

 microscope the grains and crystals are isotropic and without 

 abnormal interference phenomena. 



Evidence of slight solid solution, both with alumina and 

 magnesia, was indicated by the slightly lower refractive index 

 of the spinel crystals from preparations adjacent to the true 

 compound in composition (A1 2 3 75, MgO 25 and Ai 2 3 60, 

 MgO 40). Satisfactory tests for homogeneity of adjacent 

 preparations were, however, not possible because of the 

 presence of small quantities of a birefracting substance of 

 refractive index about 1*66 and medium birefringence, but too 

 fine for definite identification. This substance occurred in 

 different members of the alumina-magnesia series and together 

 with both periclase and spinel, and is therefore possibly due to 

 impurity from the carbon in which the preparations were 

 melted. 



Solid solution over a long range in the above series does 

 not exist. This is evident not only from the thermal work 

 but also from the optical tests for homogeneity and the deter- 

 mination of the optical constants especially of refractive indices 

 of the components of preparations intermediate in composition 

 between the compounds. The refractive indices of the com- 

 pound 5Ca0.3Al 2 3 appear slightly higher in preparations on 

 either side of the true compound, and this might be taken to 

 indicate very limited solid solution, but the observed differ- 

 ences are practically within the limit of possible observational 

 error, and too much stress cannot be placed on the evidence. 

 The same holds true for the optically positive compound 

 3Ca0.5Al 2 3 , and also the optically negative form of the same 

 composition, for which the refractive index 7 appeared very 

 slightly lower in the preparation OaO 25*78, A1 2 3 74'22, 

 than in the true compound, but the difference was within 

 the possible error limit and the evidence is not definite. The 



