6 REPORT—1843. 
(from the carbon of combination) and silica, and occasionally some oxides 
of combined metals. The residue is filtered and washed, boiled in caustic 
potass, by which the silex and extractive matter are taken up; the graphite 
remains: it is again washed with dilute muriatic acid, then with water, and 
weighed after drying. The difference between this and the total amount of 
carbon given by the combustion is equal to the carbon of combination. 
For the other constituents, after a preliminary qualitative trial, about 120 
grains of the cast iron were dissolved in nitric acid, evaporated to dryness 
with a strong heat, and ignited in a platina crucible with three and a half 
times its weight of carbonate of soda. After cooling, water is poured over it, 
which carries off the excess of alkali and an alkaline phosphate (or sulphate, 
if the iron contained sulphur, which should be ascertained beforehand), leaving 
the peroxide of iron to be separated by filtration. 
The filtered liquor must now be boiled for some time to destroy the man- 
ganesiate of potass in solution and precipitate the manganese, again filtered, 
nitric acid added evaporated to dryness, and silicic acid separated, if any exist, 
on heating with water, after moistening with acid in the usual way. 
Ammonia is now cautiously added, and if the iron contained aluminum, a 
basic phosphate of alumina precipitates, the solution, again filtered, is acidu- 
lated with acetic acid, and the phosphoric acid precipitated by acetate of lead. 
From the phosphate the phosphoric acid cannot be estimated with certainty, 
it was therefore converted into sulphate of lead, and the phosphoric acid got 
from its weight. 
The silex and manganese were always obtained by precipitation from the 
iron, &c. in separate assays. The method with benzoate or succinate of am- 
monia, though inconvenient, is one of the best, where the amount of iron is 
considerable. Liebig’s process of separation by boiling with carbonate of 
barytes succeeds very well and presents no difficulties, but where the amount 
of manganese is so very small in proportion to the iron I preferred the former 
mode. The iron itself, from its inconvenient bulk, was generally estimated 
from the other constituents. 
Separate assays are also best made for sulphur or earthy bases, but as far 
as my observation goes, these are extremely rare in British cast iron of 
commerce. 
296. The usual constituents are carbon, manganese, silicon and phosphoric 
acid, and the metal seems to be an indefinite mixture of carburet, silicuret, 
and phosphuret of iron and manganese, in all cases of gray iron, while the 
perfectly silver-white crystalline cast iron, which contains as much as 5*4 per 
cent. of carbon, does seem to be a definite quadri-carburet. This, however, 
is of little constructive value. 
The observation therefore which has been made, that perfectly definite 
combinations are those least liable to change, and thus that definite metallic 
alloys are those least subject to oxidation, though true, is of no value to us 
here, as no commercial cast iron can be viewed in any other light but that of 
a chance mixture of constituents. 
297. The view already given (1st Rep. 55) of the causes of tubercular or 
local corrosion with concretions of rust, namely, that it is due entirely to want 
of homogeneity of surface, is confirmed by the results of the second immer- 
sion. The surface of all the specimens was necessarily rendered more uniform 
by their first immersion, and accordingly we find much less tubercular cor- 
rosion has taken place on the same specimens in the same water during the 
second than it did during the previous exposure to corrosion. That alkalinity 
of the surrounding fluid is an apparent cause there is no doubt, but it seems 
to act only as an agent in increasing the action of partial voltaic currents on 
a 
