Washington — Kaersutite from Linosa and Greenland. 199 



tions, we have great confidence in its correctness. It would 

 seem to be highly probable, therefore, that in Lorenzen's 

 analysis part of the Ti0 2 was reckoned as A1 2 3 ; and this is 

 the more likely as titanium was determined by him by precipi- 

 tation with sodium thiosnlphate, a method which is known to 

 be very uncertain and apt to give either too high or too low 

 results, depending on the amount of acid in the solution and 

 other conditions. 



In its general features this analysis much resembles that of 

 the Linosa hornblende, especially in the amounts of silica, 

 alumina, magnesia, lime, and alkalies, as well as in the very 

 high titanium dioxide. The only prominent difference is 

 found in the oxides of iron. The sum of these is considerably 

 higher in the Linosa mineral and the molecular amounts are 

 equal, while in the Kaersut mineral ferrous oxide is largely 

 in excess of ferric, the amount of which is very low. 



Interpretation of the Analyses. 



The interpretation of the analyses of these two hornblendes 

 in terms of the molecular constitution is rendered subject to 

 grave uncertainty through the presence of the very large 

 amounts of titanium. The uncertainty arises from the fact that 

 this element may be present either as Ti 2 3 or as Ti0 2 or as both 

 oxides together. Potassium permanganate oxidizes Ti 2 3 to 

 Ti0 2 just as it does FeO to Fe 2 3 , so that if the lower oxide of 

 titanium is present it will appear in the ordinary coarse of 

 analysis as FeO, the apparent amount of which would thus be 

 too high, and that of Fe 2 3 would be correspondingly low, 

 while the Ti 2 3 would be determined colorimetrically or gravi- 

 metrically as Ti0 2 . 



If all four oxides are or may be present simultaneously, the 

 analytical problem becomes complex and somewhat difficult. 

 A promising line of attack is being developed by G-ooch and 

 Xewto-n,* depending on the selective oxidation of the Ti 2 3 

 by cupric salts, bismuth oxide, or ferric sulphate, which have 

 no effect on the ferrous oxide. While the results recorded are 

 excellent and show the possibility of very exact estimation 

 under the conditions observed, yet it is uncertain if the methods 

 are applicable, at least without modification, to the analysis of 

 silicate rocks and minerals, owing to their insolubility except 

 in hot hydrofluoric acid and the very ready oxidizability of 

 the hot solutions of Ti 2 3 and FeO so obtained. It may be 

 suggested that the addition of cupric sulphate to the mixture 

 of hydrofluoric and sulphuric acids employed in determining 



* Gooch and Xewton, this Journal, xxiii, 1907, p. 365 ; H. D. Newton, 

 this Journal, xxv, 1908, pp. 130 and 343. See also G. Gallo. Chem. Zeitnng, 

 1907, p. 399, and A. Cathrein, Zeitschr. Kryst., vi, 1882, p. 248. 



