IF. T. Scliallev—Dumortievite. 221 



silica separated and refused with sodium carbonate. The silica 

 was filtered off and weighed and treated with HF, which left 

 some residue. The alumina (plus iron and titanium) was pre- 

 cipitated three times to be sure of removing all of the sodium 

 salts. It was ignited and weighed with the silica residue, fused 

 with sodium bisulphate, some silica recovered, the iron reduced 

 and determined by titration and the titanium determined colori- 

 metrically. The presence of titanium was suspected from the 

 color of the pleochroism of the mineral. The boric acid was 

 determined by the Gooch method, using all of the known pre- 

 cautions. The mineral was twice fused with sodium carbonate 

 and the boric acid was finally weighed as lime borate. The 

 water was collected in a calcium chloride tube, the mineral 

 being heated in a Gooch tubulated crucible in the usual manner. 

 A blank determination was run before and after each water 

 determination and a small correction applied. All possible 

 precautions were taken throughout the analysis, which was 

 made in duplicate. The results are : 



1 2 Ayerage. 



Si0 2 28-58 28-78 28*68 



A1 2 3 63-31 63-30 63-31 



Ti 2 3 1-49 1-40 1-45 



Fe 2 3 -21 -25 -23 



H 2 1-53 1-51 1-52 



BO„ 5-21 5-53 5-37 



100-33 100-77 100-56 



The titanium is regarded as present as Ti 2 3 , replacing the 

 alumina. 



Combining the alumina, titanium and iron, the following 

 ratios are obtained : 



Si0 2 5-94 or 6 



AL,0 3 8-00 8 



B,0 3 1-06 1 



H" 2 -96 1 



The formula for clumortierite then is 



8Al 2 3 .lB 2 3 .lH 2 0.6Si0 2 . 



It has not been proven that either boric acid and alumina or 

 boric acid and hydroxyl may mutually replace each other in 

 minerals as fluorine and hydroxyl are known to do. There is 

 then no reason that the writer can see why the alumina, boric 

 acid and water in dumortierite should not be present in fixed 

 quantities ; that the variations shown in analyses are not due to 

 " isomorphous replacements" but to inaccuracy of analyses or 

 impure material. 



Having established the above formula from the analysis, let 

 us see how close the other analyses conform to the formula. 



Am. Jour. Sci.— Fourth Series, Vol. XIX, No. 111. — March, 1905. 

 15 



