muffle again closed, and the assay allowed to "drive." The mouth of the muffle is now again opened, with the exception of a low piece of coal or small piece of iron, the cupels are drawn forward towards the mouth of the muffle, and the temperature is lowered by partly closing the draught of the furnace, or also by cooling with a small cooling-iron (Fig. 48, p. 74), until a small ring of litharge and some plumose litharge (Federglätte) appear. The assays are now gradually pushed back, and the temperature is raised by opening the draught of the furnace, so that the assay may "brighten" sufficiently hot, during which the ring of litharge will disappear, but the plumose litharge (Federglätte) remain. The crucibles are now drawn forward towards the mouth of the muffle, and allowed gradually to cool off to prevent "spitting." When sufficiently cool they are taken from the muffle, and the buttons are removed by means of a pair of pincers and brushed. In successful assays the surface of the button is smooth, with a silvery lustre on the top, and a dull silver-white color on the bottom. If the operation has been conducted at too low a temperature, the surface is dull, and has a bluish tint, and the bottom is covered with a yellowish or greenish coating of lead oxide. If the temperature has been too high, the button is dull in some places, very lustrous in others, the surface is sunken, it is liable to spit, exhibits rootlets, adheres stronger to the cupel, and is porous toward the bottom. The buttons are then weighed, and, in assays of top and bottom samples, either the average percentage or the lowest percentage is given. The loss from absorption by the cupel (Kapellenzug) is added.

Bars with over 980 thousandths of silver show no difference, if the work has been carefully done. To 725 thousandths they show a difference of to 3 thousandths; from 720 to 710 thousandths, again, no

difference, or only an infinitely small one (certain chemical combinations seem to be formed at this percentage); but the greatest differences occur at 400 to 200 thousandths fineness. Very considerable differences may occur if the bars or buttons have been badly fused. The silver button contains about 2 thousandths of lead.

Correction Table for the Absorption by the Cupel, determined by the French Commission on Coinage and Medals.

In Freiberg somewhat different results have been obtained. With refined silver the loss by absorption by the cupel was found to be 0.0015 to 0.002, and in alloys of medium richness the loss was greater than that stated in the table; for instance, with 750 thousandths and 16 weights of lead, the loss was 5.55 thousandths, but with 11 weights of lead it accorded with the table, 4.52 thousandths. According to Plattner, fine silver with 5 times the quantity of lead frequently gives a loss up to 0.009, refined silver with 937 thousandths and 5 times the quantity of lead, 0.0042 to 0.0059; refined silver with 687 to 750 thousandths and 14 times the quantity of lead, 0.0073 to 0.0083.

B. Wet assays.

They are used for refined silver and coin alloys of cop

per and silver. Compared with the fire assay, they allow of an accurate determination of the degree of richness to within 0.5, and even to 0.1 thousandths. They are more frequently volumetric than gravimetric assays.

1. Volumetric assays.

a. Gay-Lussac's method with sodium chloride.'-This method is based upon the precipitation of silver from a nitric acid solution by means of a standard solution of sodium chloride. For this purpose a normal solution of common salt is required, 100 cubic centimeters (6.1 cubic inches) of which will precipitate 1 gramme (15.43 grains) of chemically pure silver. There is further required a decinormal solution of common salt, 10 times weaker than the first, and a decinormal solution of silver, consisting of a solution of silver in nitric acid, containing 1 milligramme (0.0154 grain) of silver in 1 cubic centimeter (0.061 cubic inch) of solution.

Preparation of the assay solution.-The degree of richness of the silver is approximately determined by a preliminary assay, the fine assay (p. 147) being generally chosen for the purpose. 4 to 6 thousandths parts the amount of silver found by this assay, are added to the result. It is generally preferred to assume the degree of richness a few thousandths higher than is actually the case, and to base the calculation for the quantity of assay sample required upon this, as, to effect the more rapid settling of the silver chloride, it is preferable to add, during the titration, a few thousandths from the decinormal solution of salt than to be obliged to add from the decinormal solution of silver. The quantity of

Gay-Lussac, Vollst. Unterricht über das Verfahren, Silber auf nassem Wege zu probiren, Braunschweig, 1833; Mulder, Silberprobirmethode, Leipzig, 1859; Muspratt's Chem., Bd. vi. p. 477; Bolley, Handb. der techn.chem. Untersuchung, 5 Aufl. pp. 52, 332; Dingler, cxci. 172.

Fig. 54.

alloy containing 1 gramme (15.43 grains) of silver which is to be taken is then calculated (for instance, if the preliminary assay gives a percentage of 897 thousandths. then 1.115 grammes of alloy containing 1.000 gramme of silver should be taken, 1000: 897x: 1000). The sample in the form of shavings or granules is placed in a numbered flask, together with 6 to 7 cubic centimeters (0.36 to 0.42 cubic inch) of nitric acid free from chloride, and dissolved, either on a water or sand bath. The flasks in which the samples are dissolved are from 10 to 15 centimeters (3.93 to 5.9 inches) high, and 5 to 5 centimeters (1.96 to 2.16 inches) wide. If several assays are to be made, it is advisable to dip the flasks, which are arranged upon a stand (Fig. 54), into hot water. (A black residue may be gold or sulphide of silver; should the latter be the case, some concentrated nitric acid is added and the fluid heated, or sulphuric acid used.) The nitrous acid formed is then driven out of the flask by means of a small bellows with curved extremity, and the contents of the flask is treated with the normal solution. But as the influence of the temperature upon the volume of the normal solution of common salt must be taken into consideration, its titer must always be determined on the same day the assays are to be made, with 1 gramme (15.43 grains) of pure silver + 1 to 2 cubic centimeters. (0.061 to 0.12 cubic inch) decinormal solution of silver, in order to be able, for the above mentioned reason, to use decinormal solution of salt for the final titration. The silver solution is then titrated by placing the glass flask in the metal cylinder C (Fig. 55) standing upon sliding carriage B (Sire's apparatus'). The glass cock e (a pinch-cock may be used instead) is then opened, and,

1 B. u. h. Ztg. 1873, p. 189.

the

accompanied by the admission of air through a, the normal solution of sodium chloride flows from the vessel A

through h, the thermometer tube b, and the rubber tube d, into the burette e. It ascends in this, and a small quantity reaches the saucer g through the orifice f. The cock c is now closed (h and e may be also directly connected by a rubber tube provided with a clip), and the pipette e, which is now filled, will contain exactly 100 cubic centimeters (6.1 cubic inches) of liquid. The index finger of the left hand is now placed upon the mouth f of the pipette, the rubber tube d is detached from the lower end of the pipette e, and the sliding carriage B, upon which stands the metal cylinder C containing the flask with the solution of silver, is pushed underneath the discharge orifice of the pipette. The index finger is now