According to Rosenbusch,[174] the calculated specific gravity as made above is not wholly reliable on account of the contraction which takes place. An empirical process is rather to be commended which consists in introducing a fragment of mineral of known or desired specific gravity and then adding water drop by drop until the fragment remains suspended in the mixture. Should too much water be added, the necessary increase in density can be secured by adding a little of the strong solution. =269. Method of Packard.=—A separatory funnel, according to Packard,[175] may be safely used to hold the solution while separation is going on. As the lighter minerals form the bulk of soils, the heavier constituting only a small percentage, it is well to use a wide funnel holding as much as one-half liter for quantitative separations, because a large quantity of soil, say 100 grams, is necessary from which to recover the small quantity of heavy particles satisfactorily. The soil is introduced into the solution contained in the funnel, agitated, stirred with a glass rod, and allowed to stand some time. This operation may be repeated as often as desired. Separation is not absolute by this operation, the heavy and light particles being sometimes so united that they sink or float together according as one or the other preponderates. There are also particles having so nearly the same specific gravity as the solution that they remain indifferent to its action in any position. After separation has been effected, the heavy portion is drawn off through the stop-cock of the funnel and the lighter is skimmed off the top. Both must be thoroughly washed from the adhering heavy solution for further examination with the microscope, and by chemical, microchemical, and blow-pipe tests. One who has familiarized himself with the appearance of minerals in minute fragments under the microscope, in ordinary and polarized light, will be able to determine some minerals in that way. But for certain identification it is necessary to ascertain their optical properties as is done in the case of the minerals in thin sections of rocks. _Illustration._—The following example from the work of Packard will serve to illustrate the results of separating a soil by the specific gravity method: