862 



ratnres, the platinum bulb was iulkited to a balloon, however with- 

 out bursting. 



Further we were able to determine tiie refractive indices and the 

 specific gravity of the pure oxide. Under the microscope the com- 

 pound appears as cristalline, irregularly shai)ed scales, with \ery 

 weak birefringence. Often they seem to be wdiolly isotropous, as if 

 they were glass. It may be, that the substance is finally yet of regu- 

 lar symmetry, for the weak bir-efringence often makes the impres- 

 sion of being only localised by tensions in the mass, more than of 

 a real crystallographical anisotropy ; and w^e found neither in any 

 case an interference-image in convergent polarised light, able to 

 prove that the compound belongs to one of the uniaxial classes. 

 Once some trigonally shaped j)lates were observed, which looked 

 like flat tetrahedrons or trigondodecaliedrons ; but it was impossible 

 to prove this view more exactly. The refractive index was found by 

 immersion: yi^= 1,644 ± 0,{)()2 ; so the refraction is relatively high, 

 this giving some evidence of the fact, why a gi'eat many lithiumalumi- 

 nosilicates, which are rich in IaJJ, show higher refractive indices, 

 if they are richer in the oxide. 



The specific gravity was determined by means of a })ycnometer, 

 with ortho-chlorotoluene as a liquid; before it was found, that the 

 oxide does )»ot attack tiiis licpiid in any appreciable way. As the 

 most probable value (from three determinations), we found: d^o=: 

 2,013 ±0,015, at 25°,i C. 



The oxide dissolves slowly into water, wilhout giving a great 

 heat-effect. The solution shows very strong alkaline reaction; it 

 tastes lixivial and at the same time somewhat bitter. With acids no 

 development of CO., was observed; so the product may be declared 

 free from carbonate. 



§ 5. We have now given in the foUovv^ing table the obtained 

 results with mixtures of different composition ; the numbers in the 

 7^^' , 8^'^ , 9'^*^ and 10^'' columns are reduced on the gasthermometerscale 

 of Day and Sosman ^), this being at the moment the most accurate one. 

 The composition of each mixture was determined after each experiment 

 by direct analysis, according to Hili.ebrand's '') indications for the 

 determination of the silicic acid. The thermoelements used, were 

 1 in the furnace, II in the mass; they were read alternately every 

 half minute. Even with masses of hardly 1,5 gram, the heat- 



1) A. L. Day and R. B. Sosman, Carnegie-Publication No. 157 (1912). 



-) W. F. HiLLEBRAND, loCO cUatO. 



