Table of Contents

The balance must readily indicate differences of 0.005 per 1,000, or 1/40 milligramme, with a weight of ½ gramme in each pan. The “checks” or “proofs” (vide infra) are weighed first, and their mean excess or deficiency in weight applied as a correction to all the cornets worked with them. This may be done by means of a light rider. The “ weighing-in ” correction is also allowed for, and the report is at once indicated by the marks on the weights without further calculation.

The weights are the ½ gramme, which is stamped “ 1,000,” and decimal subsidiary weights stamped 900, down to 0.5. The stamped numbers denote the number of ½ milligrammes ( “milliemes” ) contained in the weight. Ordinary weights in the gramme system may, of course, be used, each milligramme corresponding to 2 milliemes in the assay system. The report finally made gives the number of parts (in milliemes and tenths) of pure gold in 1,000 parts of the alloy.

The weights usually supplied even by the best makers differ from the standard weights by varying amounts, which sometimes exceed 0.1 millieme. These errors cannot in general be disregarded. In many sets of weights a cumulative error of 0.5 millieme or more may be introduced. It is necessary to allow for the errors, or still better, to adjust the weights by carefully polishing them with rouge, or by finest emery, or by rubbing them on a ground glass plate, if they are too heavy, or by plating them with gold if they are too light. It is not advisable to gild weights unless they consist of platinum. Riders should be used instead of small weights.

**Auxiliary Balance for Weighing Cornets**

An auxiliary balance which gives the approximate weight of the cornets without the use of any weights. The balance has a pan to hold the cornet attached to one end of the beam, and the other end is prolonged as a pointer, the position of which (when it comes to rest), with reference to a curved scale, denotes the percentage of gold present. The cornet is then transferred to the pan of an ordinary balance, and the necessary weights, already determined by the auxiliary balance within narrow limits, are placed on the other pan. For weighing large numbers of cornets of widely differing weights, this balance is a great convenience, saving both time and wear of the balance and weights. It has been in use for some years at the Mint.

**Fire Assay Verification Checks & Proofs**

Since the losses and gains detailed above are dependent on so many conditions, it is always necessary to subject check-pieces of known composition to the same operations as the alloys under examination. The use of checks in the Royal Mint was prescribed by law as early as the fourteenth century. Standard trial plates (916.6 fine) were made and used for this purpose. If such trial plates contain copper, however, it is impossible to guarantee that they shall have absolutely the same composition throughout, so that it is better either to use trial plates consisting entirely of gold alloyed with silver, which are uniform in composition, or else to use pieces of pure gold, a corresponding amount of pure copper being added in order to make the assays absolutely comparable. If 1,000 parts of pure gold were taken originally as a check, and the weight of the resulting cornet is 1,000.3, then 0.3 must be deducted from all the other results. No appreciable error is caused, provided that the standard of the alloys under examination is not much below 1,000. For example, if an alloy is 900 fine, the amount to be deducted may be assumed to be nine- tenths of 0.3 per 1,000 or 0.27, the error in deducting 0.3 being in this case one-tenth of the surcharge. Experiments show that the assumption made above is justified, and that the loss of gold is proportional to the amount of gold present if the composition of the assay pieces is similar. The errors introduced by the use of different weights are usually much greater, unless they are carefully adjusted.

In the foregoing paragraph it was assumed that absolutely pure gold was used for proofs. It is doubtful whether it has ever been obtained. The method of preparation of proof gold in use at the Royal Mint is given below. Since the assay of an alloy only gives the relative fineness of proof gold and the alloy, it follows that, if the proof gold is not quite pure, the amount found in the alloy will be in excess of the truth. If a sample of proof gold is less pure than the finest yet obtained, an allowance is made. Thus, if it is 999.9 fine, a deduction of 0.1 is made from all results of assays checked by it. This deduction is readily proved to be a very close approximation to the correct one.

Lastly, the “weighing-in” correction is applied. If the original weight taken was, say, 1,000.4 (recorded as + 4), it is sufficient to deduct 0.4 from the final weighing. In this correction 0.4 of alloy is reckoned as fine gold, but the error is inappreciable when alloys differing but little from pure gold are under examination. In the case of an alloy 900 fine, nine-tenths of the weighing-in correction must be added to or deducted from the weight of the cornet. If the alloy is 500 fine, one-half of the weighing-in correction must be applied, and in general if the weighing-in correction is x and the weight of the cornet is y, the correction to be applied in weighing out is xy/1000.

**What to do When Silver is Retained in the Cornet**

It has been found at the Royal Mint that after boiling in the first acid the average amount of silver retained is about 2.5 parts per 1,000. The amount left undissolved by the stronger acid varies with the length of time of its action. Under normal conditions with a surcharge of 0.6 to 0.8, the silver left in the cornets is from 1.0 to 1.2. If the acid continues to boil until its constant strength is reached, gold is dissolved to a considerable extent (0.5 per 1,000), while the proportion of silver present suffers little diminution, rarely falling below 0.6. If much gold is dissolved the results are not so uniform. When checks are used perfectly satisfactory results are obtained with surcharges of from + 0.2 to + 1.0 per 1,000 or even more.