One hundred cubic centimeters, equivalent to one gram of the substance, in a nitric acid solution, are placed in a half liter flask and a solution of ammonium molybdate added until all the phosphoric acid has been precipitated. The addition of ammonium nitrate will hasten the separation of the ammonium phosphomolybdate. The liquid should be allowed to stand for twelve hours. The flask is then filled to the mark, the contents well shaken, filtered through a dry filter, and duplicate samples of 200 cubic centimeters each of the filtrate taken for analysis. A small quantity of ammonium nitrate is dissolved in the liquid, and ammonia cautiously added, keeping the solution as cool as possible. The iron and alumina are precipitated as hydroxids. The mixed hydroxids are collected on a filter, washed with water, the filtrate and washings being collected in a beaker. The precipitate should be dissolved with a small quantity of a solution of ammonium nitrate and nitric acid, again precipitated with ammonia, filtered, washed, ignited, and weighed. This treatment is for the purpose of excluding all possibility of error from the presence of molybdic anhydrid. After weighing, the mixed oxids should be fused with sodium bisulfate, the magma dissolved in water, and the iron determined volumetrically with potassium permanganate after reduction to the ferrous state. McElroy has further shown by experiments in this laboratory that even the molybdate method of separating the iron and alumina from phosphoric acid with the improvements as first suggested by Krug and himself, may not always give reliable results.[32] In a solution containing ferrous iron equivalent to 56.4 milligrams of ferric oxid, were placed enough of a solution of sodium phosphate to correspond to 100 milligrams of phosphorus pentoxid. The precipitate was dissolved by adding nitric acid, oxidized with bromin water, and the phosphoric acid thrown out with ammonium molybdate. The precipitate was washed with weak nitric acid and the combined filtrate and washings neutralized with ammonia. The resultant precipitate was dissolved in a solution of ammonium nitrate and nitric acid, filtered, and again precipitated with ammonia. In two instances the quantities of material recovered after ignition were 56.9 and 57.3 milligrams, respectively, instead of the theoretical amount, _viz._, 56.4 milligrams.