429 
Cast Iron, Steel, and Malleable Iron, 
diminished, having only lost somewhat of their brightness ; 
the black spots sometimes disappear entirely. 
1*72 grains of the remaining gray powder, separated from 
the filter and ignited, lost 0*203, and seemed in appearance to 
have undergone no alteration except having become a little 
more bulky and flocky. 
Hydrochloric acid extracted oxide of iron, 0*036. 
1*32 grains of this residuum deprived of oxide of iron was 
mixed with five times its weight of carbonate of soda and ig- 
nited : after cooling, the lid of the crucible was found covered 
with green drops of manganate of soda, and on the bottom 
was the fluxed mass crystallized, perfectly white on the bor- 
ders, but having in the middle a dirty yellowish spot. 
I separated in the common way 
Silica 0*963 
Iron and Manganese 0*076 
Loss (of Carbon) 0*281 
1*320 
The same result was obtained, when, instead of melting the 
remaining gray powder with alkali, it was boiled in solution of 
caustic potash and strongly concentrated ley ; the silicon is 
taken up by the potash and iron, and manganese, mixed with 
some carbonaceous matter, is left in blackish-green bulky 
flocks on the filter, oxidizing very rapidly. 
35 grains of the same metal analysed in the common way 
gave 
Silicon 1*702530 = 4*86430 
Aluminum 0*352583 = 1’00738 
Manganese 0*262960 = 0*75130 
Phosphorus .... 0*189000 = 0*54000 
Sulphur 0*062110 = 0*17740 
Carbon 1*183000 = 3*38000 
Iron 31*152607 = 89*00740 
Loss 00*095210 = 0*27222 
35*000000 100*00000" 
The white transparent nodules are undoubtedly silica; 
before the blow-pipe they melt with soda into a transparent 
globule, separating at the same time black scales of graphite; 
melted with microcosmic salt, a skeleton of silica is separated 
as usual. After subtracting silica, nothing remains but a 
small quantity of iron and manganese, and the loss as already 
shown is = 0*281, which amounts for 35 grains of iron to 
1*17 ; the actual quantity of carbon was, as we see, 1*183, and 
therefore we may safely assume the loss to be carbon. The gra- 
phite scales, freed by means of solution of caustic soda from 
