When no more carbonic acid is evolved, another tube from this last flask is connected with the aspirator, the pinch-cock of the pipette open, and air drawn through the apparatus for about half a minute, and thus all the carbonic acid evolved absorbed, or the flasks may be slightly heated. If danger of more carbonic acid being absorbed from the air is feared, and always in very accurate analysis, a potassa tube may be connected to the pipette before drawing the air through. The precipitate formed is allowed to settle, 50 c.c. of the supernatant solution is removed with a pipette and transferred to a beaker; 50 c.c. of decinormal nitric acid and some water is added with sufficient cochineal tincture. It is then titrated back with decinormal soda; from this is now readily deducted the amount of carbonic acid, and from that the MnO_{2}, holding in view that 44 parts of carbonic acid is equivalent to 43.5 of MnO_{2} or 98.87 per cent, and that 1 c.c. of the N/10 baryta solution is equivalent to 0.0022 grm. of CO_{2}. If a carbonate, chloride, or nitrate, be present in the native binoxide, it must be removed with some sulphuric acid. This is afterward neutralized with a little caustic soda. This method yields the following results for its value in amount of manganese to 100: 99.91-99.902-99.895, and can be executed in about twenty minutes. Fifteen determinations can be carried on at once without loss of time, this, however, depending on the operator's skill. I have made many assays, and assays by this method with similarly excellent results. Of the other methods, Bunsen's is acknowledged to be the most accurate, but is, of course, too troublesome to be used in technical work, although it is used in scientific analysis. Ordinary samples are not sufficiently accurate to allow the use of this method. The methods of reducing with iron and titrating this with chromate of potassa, etc., have given a constant average of from 98.60-99.01. These results are fair, but hard to obtain expeditiously. Of the methods of precipitating the compounds of the protoxide and estimating the acid, that of the phosphate is by far the most accurate, titrating with uranium solution; 99.82 is a nearly constant average with me, much depending on the operator's familiarity with the uranium process. The methods of Lenssen, or ferricyanide of potassium method, yields very widely differing results. I have found the figures of Fresenius about the same as my own in this case; that is from 98.00-100.10.