Instead of using a linen diaphragm the tube is greatly improved, as suggested by Knorr, by sealing into the end of the tube while hot a perforated platinum disk. Before using, the tube is dipped into a vessel containing some suspended asbestos felt and by aspiration a thin felt of asbestos is formed over the outer surface of the platinum disk. By inverting the tube the water which has entered during aspiration is removed. The tube thus prepared is dipped into the boiling solution in the test tube above described and aspiration continued until a drop of the liquor has entered the tube. It is then removed from the boiling solution, the asbestos felt wiped off with a clean towel, and the drop of liquor in the tube blown through the openings in the platinum disk and brought into contact with a drop of potassium ferrocyanid in the usual way. In this way a drop of the liquor is secured without any danger of a reoxidation of the copper which may sometimes take place on cooling. [Illustration: FIGURE 42. APPARATUS FOR THE VOLUMETRIC ESTIMATION OF REDUCING SUGARS.] The careful analyst by working in this way with the volumetric method is able to secure highly accurate results. The apparatus used is shown in the accompanying illustration. =119. Suppression of the Error Caused by the Action of the Alkali on Reducing Sugars.=—Three methods are proposed by Gaud for correcting or suppressing the error due to the action of the alkali upon reducing sugars. In the first place, the common method followed may be employed, depending upon the use of an alkaline copper solution of known composition and the employment of a reducing sugar solution of a strength varying between one-half and one per cent. The error which is introduced into such a reaction is a constant one and the solution having been tested once for all against pure sugar is capable of giving fairly accurate results. In the second place, a table may be constructed in which the error is determined for sugar solutions for varying strengths, _viz._, from one-tenth of one per cent to ten per cent. If _y_ represent the error and _x_ the exact percentage of reducing sugar present then the correction may be made by the following formula; _y_ = -0.00004801_x_ + 0.02876359_x_². In order to use this formula in practice the percentage of reducing sugar obtained by the actual analysis must be introduced and may be represented by θ. The formula for correction then becomes 0.02876_x_² - 1.000048_x_ + θ = 0; whence the value of _x_ is easily computed.