the bell, the filling is much more difficult, and while the sides of the bell must be closely packed, great care is requisite to keep the powder loose in the centre, or the filter will not run. One point requires mention with regard to neutral hide-powders. If an extract which has been rendered soluble by the addition of alkalis or sulphites (p. 388) be analysed with a perfectly neutral powder, it has been shown by Paessler and Appelius[157] that a part of the tannin combined with the alkali will not be absorbed, while with acid powders, the whole will be estimated. [157] Wissenschaftliche Beilage des ‘Ledermarkt,’ 1901, p. 107. [Illustration: FIG. 67. American Milk-shaker.] _The “shake-method” adopted by the American Association of Official Agricultural Chemists_, possesses some advantages, especially in the analysis of used liquors which, from the acids they contain, are apt to give somewhat too high results by the filter method (see App. B, page 480). It has the further advantage of being much less dependent than the filter-method on the quality of the hide-powder employed. It has therefore been accepted by the International Association as permissive for all tanning materials, and as compulsory for used liquors (see App. A), and must therefore be briefly explained. It can be carried out successfully with somewhat inferior hide-powders to those required for the filter, but generally gives results 1 or 2 per cent. lower in tannins than the latter. A special shaking machine must be employed, capable of thoroughly agitating a mixture of hide-powder and the liquor to be analysed; and if many analyses have to be done, it is convenient that it should be driven by power, as otherwise the work becomes somewhat laborious. A machine called a “milk-shaker,” Fig. 67, employed in the mixing of summer drinks, is generally used. The quantity of powder required for the analyses to be made (about 8 grm. of ordinary air-dried powder for each determination, with say 5 grm. added), is stirred in a large beaker with 25 times its weight of distilled water, and allowed to soak for 24 hours, 1·5 per cent. of chrome-alum previously dissolved in water being added at the beginning of the operation, and 1·5 per cent. more not less than 6 hours before its end. The powder is then washed by squeezing through linen, and the washing is continued till the wash-water no longer gives a precipitate with barium chloride; and is then well squeezed out in linen, preferably with the aid of a press. The damp squeezed powder is now roughly weighed, to determine approximately what quantity it is necessary to take, to give 7·5 grm. of the original dry powder to each estimation (air-dried powder contains about 15 per cent. of moisture), and a portion is accurately weighed in a basin, and dried, first on the water-bath, and then in the drying oven, to determine its moisture by loss. The approximate amount of powder required for each determination--if possible a round number of