8()5 



± 3° C, caused by Uic variability in elieniical composition as a 

 result of the extraoi'diiuiry volatility of the lilhiimioxidc, which 

 will sublime already at temperatures, much lower than this melting- 

 temperature. Further below we will give the determinations of the 

 meltingpoint of the orthosilicate by means of a closed crucible and 

 a new method of observation ; we shall see, that the true melting- 

 point does not differ appreciably from the here given 'value. 



In tig. I, we have composed these numbei-s graphically in 

 the usual way. At the same time also the results, of our expei'iments, 

 concerning the determination of the field, in which the bisilicate can 

 exist in contact with the liquid, are indicated in this ligure. The 

 last mentioned determinations have given us much tiouble, because 

 great difficulties were connected with the limitation of this very 

 narrow field of stability of the compound. We have only succeeded 

 by numerous quenching-experiments : preparations, heated during a 

 long time on known and constant temperatures, were momentaneously 

 chilled, and investigated in all details by means of the microscope; 

 Only in this way the place of the bisilicate in the series could be 

 fixed with sufficient accuracy ; the expenraents on this subject are 

 described afterwards. 



As a general result, we can thus say, that there are three com- 

 pounds : L/\SiO^,Li^SiO^ and LLSi^Or,; the compound Li^SiO^, 

 proposed by Niggli on a very weak argumentation in no case 

 exists in contact with a molten mass; while the compound L/^ /SV2 O5 

 has evidently no real meltingpoint, but only a transformationtempe- 

 rature at 1032° C, at which it decom|)oses with deposition of some 

 metasilicate, or of a solid solution of a little of the bisilicate in an 

 excess of metasilicate ^). We think we are right in this last view, 

 because the refractive indices of the needles of Ll^ Si 0^. which are 

 deposited at this temperature, have evidently somewhat lower values 

 (1.57 instead of l.o85 etc.), than the pure metasilicate. 



The composition of both the eutectic mixtures E^ and E.^, can be 

 indicated by : 



E, (1022° 0.) : 55 Proc. of Weight S/'JJ, = 37.7 Mol. Proc. 



of Si 0,. 



E, (1027° C): 82,1 Proc. of Weight SiO, = 69.4 Mol. Proc. 



of .SV 0,. 



§ 6. The Utkiuiiimetasiticate crystallizes from the thin molten mass 

 very rapidly ; for this reason we succeeded in obtaining a "glass" 



') P. Niggli, Journ. of' Amer. Chem. Soc. (19t3) ; Zeits. f. anorg. Ghem. 84-, 

 2G3. (1913). In fig. 1 is wrongly written 1034° C, instead of 1032° G 



