If, during the evaporation of the filtrate, any flocculent matter separate, it should be removed by filtration and examined before precipitating the magnesia. =39. Variation of Marioni and Fasselli.=—Glaser’s method has been shown to be subject to errors by Marioni and Fasselli[29] in the following respects: 1. The precipitation of a small quantity of calcium phosphate with the ferric and aluminum phosphates. 2. The possible precipitation of basic phosphates if all the iron and alumina are not combined with phosphoric acid in the mineral. 3. The partial solubility of ferric and aluminum phosphates in dilute acetic acid. 4. The decomposition of ferric orthophosphate into soluble acid phosphate and insoluble basic salt by boiling. To avoid these errors the following procedure is proposed: From one to five grams of the phosphate are boiled in a flask for ten minutes with fifteen cubic centimeters of strong hydrochloric acid, and afterwards diluted with a double volume of water. A few crystals of potassium chlorate are added, and several drops of nitric acid, and the liquid boiled to expel chlorin. It is then filtered and washed until the volume of the filtered liquid amounts to 150 cubic centimeters. After cooling, a half gram of ammonium phosphate in solution is added, and two cubic centimeters of glacial acetic acid, followed by dilute ammonia, drop by drop, until a slight precipitate persists on stirring. Again the same quantity of acetic acid is added as above, well shaken, and left for two hours. The precipitate is collected on a filter and washed with a one per cent ammonium phosphate solution. The precipitate is dissolved by a minimum quantity of hydrochloric acid and the solution collected in the same vessel in which the precipitation took place. A second precipitation is conducted just as described above. The precipitate is washed as above described and ignited at a dull red heat. Half the weight obtained represents the ferric oxid and alumina. =40. Method of Ogilvie.=—For the separation of alumina from phosphoric acid Ogilvie recommends that the filtrate from the phosphomolybdate precipitate be neutralized with ammonia, the precipitate thus formed redissolved in nitric acid, again precipitated with ammonia, filtered, ignited, and weighed as aluminum oxid.[30] If iron be present it will, of course, appear in the product. For use in the examination of mineral phosphates the method can not have a wide application without amendment. =41. Method of Krug and McElroy.=—Krug and McElroy show that when sufficient alcohol is added to precipitate all of the calcium sulfate in the Glaser method, it will also cause a precipitation of a considerable quantity of iron, by means of which the calcium sulfate will be colored.[31] The presence of potassium and ammonium salts also affects very notably the precipitation of calcium. The method employed by them, in order to avoid these sources of error, is as follows: