Genth.] 3(38 [Sep. 19, 



repeatedly examined with a powerful magnifying glass, until the material 

 for analysis was as clean as it was possible to obtain it. This plan was 

 always strictly adhered to, except in a few cases where it was an impos- 

 sibility to obtain perfectly pure material, but where an approximation 

 was still desirable to identify the species. 



The finely powdered minerals were then dried under an exsiccator, over 

 sulphuric acid. As very few of the species analyzed could be decom- 

 posed by acids, they had to be fused with sodic carbonate to render them 

 soluble. For the alcalies, the mineral was in a few cases decomposed 

 by fiuohydric acid, generally, however, by J. L. Smith's very accurate 

 and convenient method. 



The mass of the sodium fusion was then completely disintegrated by 

 water, and the whole dissolved in very dilute chlorhydric acid. If the 

 decomposition of the silicates was complete and no appreciable quantify , 

 of corundum was mechanically mixed with the mineral, the solution in di- 

 lute acid was perfect, and rarely only a few flakes of silicic acid separated, 

 which were easily distinguished from the heavy, hard, sandy corundum. 

 Both (if both were present) were filtered, washed, dried, ignited and 

 weighed ; the silicic acid extracted by a dilute solution of potassic 

 hydrate. After washing out with boiling water, to which towards the 

 end of the washing, a few drops of chlorhydric acid was added, the 

 corundum was weighed, then rendered soluble by fusion with potassic 

 hydric sulphate, from the solution of which the pure alumina was pre- 

 cipitated as usual. 



From the fusion for the alcali determination, the residue after extrac- 

 tion by boiling water was treated in a similar manner. The fusion with 

 sodic carbonate acts more powerfully upon corundum, but unfortunately 

 even the fusion of corundum with calcic carbonate and amnionic cblo- 

 ride, renders such a quantity of corundum soluble, that this method can 

 not be used for a quantitative determination of the same. 



The analyses were then conducted as usual, the alumina was separated 

 together with ferric oxide by amnionic hydrate and boiling of the am- 

 monical magma, until it had a neutral or even acid reaction. After 

 filtering and washing they were re-dissolved and re-precipitated, and 

 this operation generally repeated three or four times. The ferric oxide 

 in the alumina was determined after the conversion of both into sul- 

 phates and its reduction by zinc, by titration with potassic permanga- 

 nate. 



The amount of ferrous oxide was always ascertained by dissolving the 

 minerals in a sealed tube under pressure with dilute sulphuric acid, and 

 subsequent determination with potassic permanganate. 



Some apparently quite pure spinels left when treated in this manner a 

 small quantity of a sandy powder, which proved to be corundum. 

 I had hoped that by this method an accurate determination of corundum, 

 mechanically mixed with other minerals, could be obtained, but I found 

 that dilute sulphuric acid under such circumstances dissolves corundum, 



