Penfield and Foote — Composition of Tourmaline. 105 



but its application to an insoluble silicate, especially one con- 

 taining fluorine, needed careful study. Mixtures were accord- 

 ingly made of silicates to which weighed quantities of borax 

 and fluorite were added and the following conditions were 

 determined, which yielded the most accurate boron determina- 

 tions. The mineral was fused with from four to five parts of 

 sodium carbonate, the fusion extracted with water, and, with- 

 out filtering, an excess of ammonium carbonate was added. 

 The insoluble residue and the precipitate were filtered off, the 

 filtrate concentrated, acidified slightly with nitric acid and dis- 

 tilled with methyl alcohol. The residue from the sodium car- 

 bonate fusion together with the precipitate produced by the 

 ammonium carbonate was fused again with sodium carbonate, 

 treated with water, filtered and distilled with methyl alcohol 

 after acidifying with nitric acid. About one half of one per 

 cent of boric oxide was obtained from the second treatment, 

 and in no case did we succeed in obtaining an exact determi- 

 nation of the boric oxide without repeating the fusion. The 

 weighed mixture of calcium borate and oxide was in all cases 

 found to contain a small amount of fluorine. It was there- 

 fore dissolved in hydrochloric acid, and part of the lime 

 together with calcium fluoride was precipitated with sodium 

 carbonate. The precipitate was ignited, treated with acetic 

 acid and the resulting calcium fluoride weighed. The amount 

 of fluorine thus found never amounted to over 0*20 per cent. 



Fluorine was determined by the modified Berzelius method 

 described by Penfield and Minor.* 



Water was determined by fusing the mineral with sodium 

 carbonate in a combustion tube and collecting the water in a 

 weighed tube containing sulphuric acid, a method which has 

 been thoroughly tested and is known to give reliable results, f 



For the determination of the bases the mineral was decom- 

 posed by fusion with sodium carbonate and the silica separated 

 as usual. It was found by experiment that the amount of 

 silica volatilized by the small amount of fluorine in the min- 

 eral could practically be neglected. In one variety of tourma- 

 line two determinations of silica made by the Berzelius method 

 of fusing the mineral and a weighed amount of silica with 

 sodium carbonate, and separating the silica with ammonium 

 carbonate and an ammoniacal solution of zinc oxide, gave 36*69 

 and 36*76 per cent, while determinations by the ordinary 

 method gave 36*75 and 36*73 per cent. A similar conclusion, 

 that when the amount of fluorine is small it is not necessary to 

 separate the silica by the Berzelius method, was also reached 

 by Riggs. 



*This Journal, III, xlvii. 1891. p. 387. 

 f This Journal, III, xlviii, 1894, p. 31. 



