257 



re-evaporating the filtrate. In one case only was the alumina 

 precipitate stained (by iron), the amount present was, however, 

 far too small to estimate volumetrically. 



The amount of magnesia obtained was unweighable. 



THE ALKALIS. 



Two methods were used. 



(1) That of Lawrence Smith. — About half a gramme of rock 

 was used. A difficulty was here met with as the mass was found 

 to have fused in places to the sides of the tube. On repeating 

 the same difficulty was met. A lower temperature was then 

 tried, and the mass came out readily. This, however, gave a low 

 alkali content. The other method tried was to increase the lime 

 proportion. With twelve parts of calcium carbonate instead of 

 eight the mass came out readily from the crucible. 



(2) Estimation of Alkalis hy the Hydrofluoric Acid Method. — 

 Fearing that the large percentage of potasli might give rise to 

 loss through volatilization in the original fusion, and also as a 

 check, this method was used. It gave exactly the same result as 

 the first method for potassium, but a slightly higher value for 

 sodium. 



Working out the silica and alumina equivalents for lime and 

 alkalis we get 10834 and *1953 respectively. These numbers are 

 equivalent to an excess of -31 per cent, of Si Oo and -46 per cent. 

 AloO.. over the amounts found by analysis. The error is more 

 probably in the lime and alkalis, particularly the latter, as there 

 were not sufficient platinum utensils to carry out all the opera- 

 tions in platinum. 



The formula of the felspar is thus — 



Orog Abg An 4, or nearly Or,; Abo An^^ 

 and its chemical properties place it in the Anorthoclase group. 



