Many precipitates occlude potash and hold it so firmly that it cannot be washed out with hot water although the potash compounds present in the precipitate are perfectly soluble. It appears to be a kind of molecular adhesion. Barium sulfate has this property of attaching potash molecules in a high degree, and ferric and aluminic compounds only to a slightly less extent. To reduce the losses, consequent on the conditions just mentioned, to a minimum, the sulfuric acid and earthy bases should be very slowly precipitated, with violent agitation, at a boiling temperature. Another source of loss in the platinum method arises from the use of a solution of ammonium chlorid for washing the potassium platinochlorid precipitate. There is danger here, not only of the solution of the impurities present in the precipitate, but also of a double decomposition by means of which some ammonium may be substituted for the potassium in the washed product. In the official method, moreover, there is danger of securing a final precipitate which may contain traces of calcium and magnesium sulfates when these bodies are abundantly present in the sample taken for analysis. The careful analyst must guard against these sources of error, but it is probably true that he will never secure a practically chemically pure precipitate of potassium platinochlorid when working on the mixed fertilizers found in commerce. =266. Effect of Concentration on the Accuracy of Potash Analysis.=—Winton has also studied the sources of error in the determination of potash as platinochlorid, especially with reference to the effect of the concentration of the solution at the time of precipitation.[215] He finds that the method of precipitating in concentrated solutions and drying the potassium platinochlorid at 130°, depends for its accuracy upon the mutual compensation of three errors; _viz._, (1) to the solubility of the potassium salt in eighty per cent alcohol, (2) to the presence of water in the crystals which is not driven off at 130°, and (3) the use of a factor based on the wrong atomic weight of platinum. He finds, further, that the error due to the presence of water occluded in the crystals can be reduced to a minimum, and the process of drying greatly simplified, by adding the solution of platinum chlorid to the potash solution in a dilute condition, not exceeding one per cent in strength. The potassium platinochlorid thus produced can be very effectively dried at 100°. The error due to the solubility of the salt in eighty per cent alcohol can also be greatly reduced by using ninety-five per cent alcohol. The error due to the wrong factor, based on the old atomic weight of platinum, _viz._, 0.3056, can be corrected by using the factor based on the recently determined atomic weight of platinum, _viz._, 195, which is 0.30688.