The apparatus which I shall describe in this paper has been in use for some time at the Silver Bullion Assay laboratory of the Pennsylvania Lead Company’s works, and has been found to give good results, and to be simple and convenient.

Steam Bath.—This steam bath is shown by Figs. 1, 2, 3, and 4. It is made of sheet copper, about one-twelfth of an inch thick,

the joints being brazed, and is used for heating the bottles in which the silver samples are dissolved, previous to the fineness determinations by the Gay Lussac volumetric method; for heating the flasks in which the gold parting is made in bullion or other gold assay-buttons ; and for general analytical work.

Fig. 1 shows a plan of the bath. The orifices, marked a, are made of the proper shape to hold the silver assay bottles in an inclined position, the necks resting on the raised flanges, e e. There is described in Percy’s volume on Gold and Silver, Part I, page 289, a bath for heating silver assay bottles. It has this disadvantage, however, that the bottles stand erect, and there is a slight danger that some of the fine spray arising with the evolution of nitrous-oxide during solution of the silver may be projected from the bottle, or upon the stopper neck. When the bottles rest in an inclined position, there can be no loss in this way, though the stopper neck should always be washed down with distilled water before adding the salt solution. The orifices, marked b, are used for heating the small flasks, in which the gold is parted from silver buttons, the necks of the flasks resting on flanges, also marked e e. The ordinary one- ounce flasks we find to be more convenient for gold parting than the special long flasks called parting flasks, especially in transferring the

gold to the small porcelain crucibles in which it is annealed in the muffle furnace. The orifices, marked c, are used for heating beakers, flasks, etc.,—for general analytical work. They are covered by the

usual overlapping spun-copper rings. Covers are also used on the orifices for silver assay bottles and gold parting flasks.

Figs. 2 and 3 show two cross-sections at right angles to each other. The flanges above referred to are shown at e ; f is the inlet for steam ; d is a strengthening diaphragm. In Fig. 4, which represents an end elevation, g is the outlet for condensed water.

Shaking Engine.—Fig. 5 gives an elevation of a small upright engine, which has a rod, c, running through the packing gland, d, in the upper cylinder head. A box, B, is screwed to the rod, c. This box is made to hold nine silver assay bottles. A piece of thick sheet rubber is fastened upon the bottom of the box, and also a sheet to the under side of the cover, f. The cover, f, is clamped down tightly

upon the bottles by the hinged holders, e e, which are held in place by a rubber band. In this way the bottles and their stoppers are held firmly in place, and will not be disturbed by the shaking. This small engine, as arranged for shaking the silver assay bottles, was introduced by Mr. E. F. Eurich, and has proved to be a great convenience in making silver fineness determinations by the Gay Lussac method. The bottles are excluded from the light, and there is no difficulty in obtaining a perfectly clear solution, when the bottles are shaken from one to two minutes. If more convenient, the small engine could be replaced by a crank, run from a line of shafting, but I think the engine to be preferable, and to be a great improvement over hand shaking, or the shaking case, as arranged with springs above and below the case.

We have found that the Gay Lussac method of volumetric assay is the most reliable and convenient for determining the silver fineness of bullion, that contains but a small percentage of copper or other impurities. The Stas pipette is the best form for convenience and accuracy. The arrangements for determining the temperature of the salt solution, when assays are made as described by Sire and others, are not reliable in general work, unless more time and care are used in making corrections than would be needed to make a determination upon fine silver of the value of the salt solution with each set of assays. This latter method is probably the better one. With each set of determinations one should be made upon pure silver, from which the strength—whether above or below the normal—of the salt solution can be determined. All the assays of any set, including that of fine silver, are made under exactly the same conditions, and the results should be accurate if the test silver is pure. There is no difficulty in preparing pure silver, if care is taken to obtain a pure chloride. If silver, 998 or 999 fine, cannot be used to dissolve for precipitation as chloride, it is best to redissolve the silver first reduced from the chloride and reprecipitate. The assays, made by different persons, of silver bullion bars, often do not agree exactly, especially when the bullion contains a considerable percentage of impurities. One cause for the different results is due to the sampling. The bar should always be sampled, when it is possible, as it is poured, a small portion being taken with a sample ladle, as soon before the bar chills as possible, and granulated in water. There will be sometimes a slight difference between the poured sample and one which is cut from the surface of the chilled bar, due to molecular changes in the alloy as the bar cools, especially if the bullion is quite impure and cools slowly. It is evident, I think, that a sample, properly taken when the bullion is poured from the cupel test, or from a crucible, is the sample which will most accurately represent the bar.