from the RIVER DEE. 259 
red oxide, when raifed to the ftate of white oxide, increafes ex- 
aétly one-third of its weight. It was the knowledge of thefe 
facts, that led me to the preceding numbers. And I think they 
may be ufed, till fome more direct experiment lead us to pre- 
cife conclufions. — 
REpD oxide being the only ftate in which this metal has yet 
occurred feparate, we may conclude that it combines, in this 
ftate, with metallic oxides, and that the titanium in iron-fand, 
is moft probably in this ftate. But white oxide, diminifhed by 
one-fourth, gives us the equivalent quantity of red oxide. On 
that fuppofition, the titanium prefent, before the analyfis, in 
the 100 grains of ore, weighed 9.5 grains. 
THE appearance of the arfenic furprifed me a good deal, as 
it was altogether unexpected. I am difpofed to afcribe it to 
fome particles of arfenic pyrites which might have been acci- 
dentally prefent. This conjecture will appear the more pro- 
bable, when we reflect, that arfenic pyrites very frequently ac- 
companies iron-fand. Before the microfcope, the iron-fand ap- 
pears to contain fome white fhining particles, which, probably, 
are arfenic pyrites. 
Tue fmall quantity of filica and alumina, I afcribe, without 
hefitation, to grains of quartz and felfpar, which had adhered 
to the iron-fand, and been analyfed along with it. Some fuch 
grains were actually obferved and feparated. But others, pro- 
bably, efcaped detection. 
12. Ir thefe fuppofitions be admitted as well founded, the 
iron-fand was compofed of 
Protoxide of iron, 85.3 
Red oxide of titanium, 9.5 
Arfenic, - - I.0 
Silica and alumina, - 1.5 
Lofs, - - 27 
100.0 
Vou. VI.—P. II. Kk The 
