COMPOSITION OF N.S.W. LABRADORITE AND TOPAZES. 197 



grams of sodium potassium carbonate and then decomposed. The 

 silicon tetra-fluoride was collected in water after the first few 

 experiments, as it was found that on evaporation with ammonia 

 some of the hydrogen silicon fluoride was converted into silica. 

 The method gave uniform results, which however, were rather 

 low in all cases. 



The cause of the lowness in the results is not apparent. It is 

 probably due to the equation 3 SiF 4 + 4 H 2 = 2 H 2 SiF 6 + H 4 Si0 4 

 not being a quantitative one. Potassium silicon fluoride was 

 found to be slightly soluble in a fifty per cent, solution of alcohol. 

 After filtering off the gelatinous silica, the solution in which the 

 potassium silicon fluoride was precipitated was always made of 

 the same volume and hence the error due to solubility could be 

 allowed for. Potassium chloride was used prepared from different 

 sources but caused no difference in the results. 



The discrepancy was not due to incomplete absorption of silicon 

 tetra-fluoride in the water. A second absorption tube gave no 

 trace of a precipitate. To completely remove the gas from the 

 decomposition flask dry air was passed through for four to five 

 hours. 



The amounts of calcium fluoride used were varied to see if the 

 error was constant or proportional. The percentage of fluorine 

 obtained was increased as more calcium fluoride was taken, but 

 not proportionally, and tended to reach a maximum. For amounts 

 approximately equal the results were constant and hence the error 

 can be corrected from the weight of potassium silicon fluoride 

 obtained. 



Results obtained by the tetra-fluoride method on calcium fluoride, 

 cryolite and apatite. — First series. — The substance was fused with 

 sodium potassium carbonate and then decomposed. The calcium 

 fluoride used in these experiments was from ground crystals, and 

 on testing, by conversion into calcium sulphate, gave 99*6 per cent. 

 of calcium fluoride. Four experiments gave 44*60, 45*97, 46*12, 

 and 46*20 per cent., the theoretical percentage of fluorine being 

 4853 and the average difference 2*82 per cent. 



