hillebrand.] NOTES ON ROSCOELITE. 71 
of the V 2 3 . Iii one case, unfortunately, the air could not have been 
fully expelled, for the solution after decomposition was blue instead of 
green and much less oxygen was required in titration than when the 
color was green. 
The contents of the tube, still warm, were poured into fairly hot 
freshly boiled water and titrated rapidly. Iron and vanadium were 
then reduced by H 2 S gas, the latter boiled out in a current of C0 2 , and 
titration repeated on the hot liquid. The solution was then boiled 
with ammonia, the precipitate fused with Na^CO^ leached with water, 
and the residue again fused with Na 2 0O 3 and leached to remove the 
last of the vanadium. This residue was then fused with KHSO4, 1 dis- 
solved in dilute H 2 S0 4 , boiled first with H 2 S and then in a current of 
C0 2 , and the liquid titrated for total iron. The solution held titanium 
which was then estimated colorimetrically. 
The first of the titration results gave the effect of all iron, assuming 
its existence as FeO, and of all vanadium that might exist in a lower 
state of oxidation than V 2 5 . The second gave all iron as FeO and all 
vanadium as V 2 4 . Deduct from both the figure for FeO and the 
remainder gives that for vanadium. In this way two very concordant 
results were obtained for total vanadium as V 2 4 , which were supple- 
mented by tests on portions used for other constituents, but only one 
was obtained for the vanadium as it exists in the mineral, a second 
being vitiated by evident oxidation during decomposition in the tube. 
Asa check, however, a fresh sample of un purified mineral was similarly 
treated, and it was found that fully nine-tenths of the vanadium existed 
as V 2 3 , a result confirming the single test on the purified material 
which showed 93.5 per cent as V 2 3 . It is not impossible that slight 
oxidation had laken place even in these cases, and I feel justified in 
assuming with Genth that the vanadium should be considered wholly 
as V 2 3 . 
In the other portions analyzed the vanadium was likewise titrated in 
V 2 4 condition, but only after separation from iron, titanium, and alumi- 
num by fusion with Na 2 0O 3 , extraction with water and separation of 
dissolved alumina by ammonium carbonate. A second fusion of the 
residue and of the precipitated alumina was necessary in order to 
extract all the vanadium. These numerous manipulations render the 
figures for A1 2 3 perhaps the least trustworthy of all, but the average 
given is probably not far from correct. 
The iron is assumed to be present as FeO j and the titanium to belong 
to a foreign mineral, since a test on unpurified material gave much 
more, namely, 1.50 per cent Ti0 2 , without accompanying increase in 
FeO, which latter observation seems to exclude ilmenite as the source 
of the titanium. 
Both the iron and magnesium are supposed to belong to the roscoe- 
! Any slight trace of vanadium remaining will impart a bright yellow color to the cold KIIS() 4 , 
fusion, a test which proved useful more than once duriug the analysis. 
