If a trace of protoxide of iron remains in the solution another cubic centimeter of nitric acid ought to be added and the boiling continued so long as any reddish-yellow vapors are visible, more hydrochloric acid also being added to keep the solution from being dried up. The process is continued in this way until two tests have given no reaction of protoxide of iron, when the solution is diluted with water; but no dilution should take place until the oxidation is complete, because in the course of it the solution ought to be kept as concentrated as possible. Silica, and graphite when it is present, need not be removed by filtration, if it is not intended to estimate them, or there be no fear that the graphite is accompanied by any humous substance, or that any oily, viscous compound has been deposited on the sides of the beaker. In the last mentioned case the solution should be transferred into another beaker, and filtered, if graphite be present. When the solution is evaporated to dryness, the remainder has five cubic centimeters hydrochloric acid added to it, and the liquid is then brought to boiling in order that the perchloride of manganese possibly formed during the evaporation to dryness may be reduced to protochloride, after which the solution is diluted with water till it measures about 100 cubic centimeters. To this is now added in small portions and with constant stirring as much of a saturated solution of bicarbonate of soda (thirteen parts water dissolve one part salt), that all the iron is precipitated, after which, when the escape of carbonic acid has ceased, the solution is diluted with water till it measures 200 cubic centimeters and is then ready for titration.
A large excess of bicarbonate ought to be avoided, because in a solution of pure protochloride of manganese it renders the liquid milky and turbid; the addition of more water, however, makes it clear. The solution of bicarbonate must be free from organic substances which may tend to remove the color of the chameleon solution. To ascertain this, the latter is added to the former drop by drop so long as the color is removed.
If it be desired to estimate the silica in the same test, the iron, as when it is analyzed for silica, may be also dissolved in sulphuric acid, and afterward oxidized with nitric acid, after which the solution is boiled to near dryness, so that the organic substances are completely destroyed. In order afterward, to drive off the nitric acid and get the manganese with certainty reduced to protoxide, the solution is boiled with a little hydrochloric acid. In this way the solution goes on rapidly and conveniently, but the titration takes longer time than when the iron is dissolved in hydrochloric acid, because the iron precipitate is more voluminous, and, in consequence, longer in being deposited. To diminish this inconvenience the solution ought to be made larger. In such a case the rule for dissolving is, one gramme iron (more if the content of silica is small) is dissolved in a mixture of two cubic centimeters sulphuric acid of 1.83 specific gravity and twelve cubic centimeters of water in the way described above, and boiled until salt of iron begins to be deposited on the bottom of the beaker. Five cubic centimeters hydrochloric acid are now added, and the solution tested with red prussiate of potash for protoxide of iron, and the boiling continued till near dryness, when all the nitric acid is commonly driven off. Should nitrous acid still show itself, some more hydrochloric acid is added and the boiling continued.