As soon as the flask, B, has become sufficiently cool, the ferrous chloride and hydrochloric acid are introduced through the tube, a (which has been full of water from the first), in the same manner and quantities as in the well-known Tiemann-Schulze method.

The pinch cock at d is then opened, and the apparatus allowed to fill with carbon dioxide.  When the pressure has become sufficient to force the gas through the solution of permanganate, the pinch cock at f is removed.  It should be opened only slightly and with great caution at first, unless one is certain that the pressure is sufficient.  If the pressure is insufficient, the fact will be made apparent by a rise of the permanganate in the small internal tube.

The flow of carbon dioxide is now reduced to a very slow current, or entirely cut off.  The contents of B are slowly heated, until the decomposition of the nitrate is complete and the greater part of the nitric oxide has been expelled, when the apparatus is again closed at f and d, and allowed to cool.  The tube, a, is then washed out, by the introduction through it into B of a few cubic centimeters of strong hydrochloric acid.

The process of filling the apparatus with carbon dioxide, and of heating the contents of B, is repeated.  When it becomes apparent, from the light color of the liquid in B, that all of the nitric oxide has been expelled from it, the current of carbon dioxide is increased and the heating discontinued.  Care must be taken, however, not to admit too strong a current of carbon dioxide, lest some of the nitric oxide should be forced unabsorbed through the permanganate solution.  It is also necessary, for the same reason, to avoid too rapid heating during the decomposition of the nitrate.

When all of the nitric oxide has been forced into the solution of permanganate, the determination is made in the manner already described.

To test the method, nine determinations were made with quantities of pure nitrate of potassium varying from 100 to 200 milligrammes.  The maximum difference between the volumes of permanganate actually used and those calculated was 0.05 c.c., while the main difference was 0.036 c.c.  The measurements of the permanganate were made from a burette which had been carefully calibrated.  We also made a number of determinations, using a solution of manganous sulphate in the place of the oxalic acid.  The advantage of this method lies in the fact that it is not necessary to dissolve the oxide which is precipitated upon the glass within the tubes, E, E, since, in the presence of an excess of permanganate, the reduction by nitric oxide extends only to the formation of MnO_{2}; also in the fact that the solution of manganous sulphate is more stable than that of oxalic acid.  A solution of the sulphate having been once carefully standardized, can be used for a long time to determine the value of permanganate solutions.