60 SOME PRINCIPLES AND METHODS OF EOCK ANALYSIS, [bull. 176. 
boil the freshly precipitated mixture with a solution of the alkali, on the 
assumption that the titanium oxide is hereby rendered wholly insolu- 
ble and thus separated from the alumina. This, however, is in part 
an error long since pointed out by Gooch, 1 who showed that pure 
titanic oxide is markedly soluble under both conditions of treatment. 
Experiments very recently made by the writer to test the extent of 
this error brought out the following- interesting results. 
When 0.045 gram of titanic oxide was fused by itself with sodium 
hydroxide, the clear aqueous extract of the fusion held 0.0031 Ti0 2 , or 
about 7 per cent, determined colorimetrically. When freshly precipi- 
tated and boiled with the alkali the solubility was less. When fused 
with sodium carbonate but an infinitesimal trace was dissolved, which 
required strong concentration for its detection. When mixed with a 
large excess of alumina and fused with the caustic alkali, the solubility 
was still very marked, though less than when alumina was absent. 
With a large excess of ferric oxide, with or without alumina, no tita- 
nium could be detected in the unconcentrated filtrate. 
It thus appears that fusion with caustic alkali after first removing 
iron involves an error in the gravimetric determination of both alumi- 
num and titanium which does not appear if the iron has not been 
removed. 
DIRECT PRECIPITATION OF ALUMINA. 
A recent and promising method for the " direct determination of 
alumina in presence of iron, manganese, calcium, and magnesium" is 
that of Hess and Campbell, 2 but, as with the methods just considered, 
it involves finally weighing aluminum and phosphorus together, and 
the behavior of titanium has not been investigated. For this latter 
reason the details of the method will not be given. Suffice it to say 
that precipitation of the aluminum and phosphorus is made by phenyl- 
hydrazine, after first neutralizing the (preferably chloride) solution b}^ 
ammonia and reducing iron by a saturated solution of ammonium 
bisulphite. Phenylhydrazine "precipitates aluminum from its solu- 
tions quantitatively as the hydroxide without a trace of the precipi- 
tate being redissolved in excess of the precipitant." 
IX. MANGANESE, NICKEL, COBALT, COPPER, ZINC. 
Ammonia is added to the flask containing manganese, the earths, etc. 
(p. 56), and hydrogen sulphide gas is introduced, whereby manganese, 
nickel, cobalt, copper, zinc, and a small part of the platinum from 
the dish are precipitated. The flask is set aside, corked, for at least 
twelve hours, and preferably twenty-four, or even longer; the pre- 
iProe. Am. Acad. Arts and Sci., Vol. XII, p. 436, 1885; Bull. U. S. Geol. Survey No. 27, pp. 16 and 17. 
2 Jour. Am. Chem. Soc, Vol. XXI, p. 776, 1899; Chemical News, Vol. LXXXI, p. 158, 1900. 
