The procedure for assay of gold bullion, as described has for its sole object the estimation of the percentage of gold present in the alloy, all other constituents being disregarded. In the first instance, the simple case of the assay of gold alloys containing appreciable quantities of only copper and silver will be dealt with. Refined gold ingots and the alloys used for coinage, and for almost all jewellery, come under this head. The effect of large quantities of other impurities and the precautions thereby rendered necessary will be discussed later.

The method universally employed is that of cupellation and subsequent parting. The gold bullion is cupelled with silver and lead, by which the greater part of the base metals present is removed as oxides dissolved in litharge, and an alloy of gold and silver left on the cupel. This is “ parted ” by nitric acid, which dissolves the silver and leaves the gold unattacked.

In the following the practice at the Mint is described, but the same description would apply, with very slight alterations, to the methods used at other mints and assay offices.

The degree of accuracy now attained in most assay offices reduces the probable error in the report of an assay to about 0.1 per 1,000, but, to prevent the error from rising above this amount, all weighings must be correct to 0.05 per 1,000, which is not always the case in ordinary bullion assays.

The system may be conveniently regarded as comprising six distinct operations:

1. Selection of the sample.
2. Preparation of the assay piece for cupellation.
3. Cupellation.
4. Preparation of the assay piece for parting.
5. Parting and annealing the cornets.
6. The final weighing and reporting.

How to Sample Bullion for Assay

Alloys of gold with either silver or copper or with both are practically uniform in composition if they have been melted and well mixed, and the ingot has not been pickled after casting. In these cases a single outside cut is representative of the composition of the whole of the ingot. The cut must, of course, be clean.

When other metals are present, the solidified ingot is not uniform in composition, and a dip-sample is taken. The metal is melted in a plumbago crucible with a covering of borax, stirred vigorously, and one or more dip-samples taken by an iron ladle, or better, by a plumbago spoon. The sample may also be taken by a plumbago stirrer, the foot of which has a hollow cavity, or by a little plumbago crucible fastened by iron wire to an iron rod. The sample is dipped out with a borax cover, and poured into an iron mould or allowed to cool in the dipper. Granulation by pouring into water may result in partial oxidation of some of the base metals.

Moderately pure tough ingots are sometimes sampled by means of outside cuts, even if there is no certainty that some base metal besides copper is not present. Cuts from a top corner and an opposite bottom corner are then taken, or drillings are taken from the top and bottom of the ingot. The results are not always trustworthy.

The sampling of gold wares is difficult, because the outside is usually finer than the interior, in consequence of the pickling of the wares after manufacture. The sampling of “base bullion” belongs to the metallurgy of lead.

How to Prepare of the Assay Piece for Cupellation

Cupellation

Preparation of the Assay Buttons for Parting

Weighing the Cornets

Surcharge

The gold cornet does not actually contain the whole of the gold present in the original alloy and nothing else. Gold is lost by (a) volatilisation ; (b) absorption by the cupel; (c) solution in the acid. On the other hand, the cornet always retains (1) some silver; (2) occluded gases. The algebraical sum of these losses and gains is called the “surcharge,” since the cornet usually weighs more than the gold originally present in the assay piece; if the reverse is the case, the work is less accurate. The various losses and gains are discussed in detail below.

Losses of Gold

G. H. Makins found gold and silver in the proportion of about 1 to 9 in the dust taken from a flue used only in gold and silver cupellation, but did not attempt to ascertain the percentage loss of gold by volatilisation. He also showed that large amounts of gold were dissolved by nitric acid in the course of assaying, and attributed the dissolution of gold to the presence of nitrous acid, but supposed that it would not be dissolved in the weaker acid, where nitrous acid was formed in larger quantities, owing to the protective action exercised by undissolved silver, which formed the positive element in the gold-silver couple. The fact that gold is actually dissolved in nitric acid, presumably as oxide, and remains dissolved even after dilution with water, was proved.

As the result of a number of experiments made by the author at the Mint, it was found that in the assay of standard gold (916.6 fine) with the ordinary surcharge of 0.4 to 0.8, the loss of gold is about 0.4 per 1,000, of which about 82 per cent, is absorbed by the cupel, 8 per cent, is dissolved in the acid, and the remaining 10 per cent., which is unaccounted for, is probably volatilised. These ratios, however, vary considerably, as a hot fire increases the loss by absorption, while by prolonged boiling in acid, and especially by annealing the fillets at a high temperature, the amount of gold dissolved in the acids is increased.

An increase in the percentage of copper in the assay piece is accompanied by an increase in the loss, as is shown in the following table which gives the relative surcharges obtained in the parting assay of gold-copper alloys of different standards ; it is compiled from a number of results obtained at the Mint:

The third column is obtained from the surcharge and the fineness in gold of the cornets obtained. The amount of lead used was eight times the weight of the alloy in the first three cases, and double this quantity for the last three. This table, as already explained, only gives relative surcharges and losses of gold; the absolute amounts vary with the treatment. The influence of the temperature of the furnace on the surcharge is pointed out. The following table is compiled from the results of experiments:

The loss of gold was found to be 0.645 per 1,000 in series (a), and 0.723 part in series (b).
Rossler has shown that the loss of gold in cupellation increases with the amount of lead used and decreases as the amount of silver is increased.

Effect of Varying the Proportion of Silver to Gold

The views of various authors on the best proportion to be used have been given above. The following are some of the results of a series of experiments made at the Royal Mint in the years 1904-5. In each case the results are the means of a number of closely concordant assays worked together under similar conditions. Except where stated otherwise, the cornets were parted by boiling in nitric acid for two half-hours, the first acid being of SG 1.26, and the second acid of SG 1.32. In all cases 0.5 grm. of gold, 0.045 grm. of copper, and 4 grams. of lead were used:

When the ratio of 2 parts of silver to 1 part of gold is used, it is seen that the amount of silver retained by the cornet is less, the loss of gold greater, and the surcharge less than when more silver is used. If the surcharges are higher, the relation of the results remains approximately the same as shown below:

If less than 2 parts of silver to 1 of gold are used, the surcharge begins to rise again, as follows:

If the ratio of silver to gold is lower than 1.75 to 1, the cornets are not properly parted by boiling in two acids in the ordinary way. By boiling for half an hour each in three acids of sp. gr. 1.26, 1.32, and 1.41 respectively, the following results were obtained:

Among other results, it was found that the uniformity of the assays was greater when the ratio of 2 to 1 was used than in the case of other ratios. This led to the adoption of the ratio for bullion assays at the Mint.

Occluded Gases

Graham proved that cornets retained twice their volume of gases (mainly carbon monoxide) in occlusion after annealing. This amounts to two parts by weight in 10,000, and is reckoned as silver in the preceding paragraph. According to Varrentrapp, the gas retained varies with the temperature at which annealing takes place.

Preparation of Pure Gold

Limits of Accuracy in Gold Bullion Assay

Attention may here be drawn to the errors introduced by the lack of delicacy of the very finest assay balances in ordinary use.

It has elsewhere been shown by the author that by weighing in the way indicated above, errors not greater than 0.15 per 1,000 may be introduced. It is, therefore, clear that this amount represents the limit of accuracy when such balances are used. By weighing correctly to 0.01 per 1,000, however, and performing all other operations with scrupulous care, then in the determination of gold in high-standard alloys of gold and copper, or of gold, silver, and copper, whether pure or contaminated by small quantities of lead, bismuth, zinc, antimony, nickel, and some other elements, the error does not exceed ±0 02 per 1,000, if the mean of three results is taken.

Parting by Sulphuric Acid

The use of sulphuric acid of 66° B. instead of nitric acid for parting is recommended by some assayers on the ground that the losses of gold by dissolution in nitric acid are variable, while sulphuric acid does not dissolve gold. The inconveniences suffered by the use of sulphuric acid are that

1. lead and platinum are left undissolved by it;
2. violent bumping of the liquid occurs during ebullition ; and
3. sulphate of silver is not very soluble in water, and the washing is consequently done with dilute sulphuric acid.

However, it is stated that less silver is left undissolved in the cornets than in the parting by nitric acid if the proportion of gold to silver is between 1 : 2 and 1:3.

Preliminary Assay

If the composition of an alloy is quite unknown, a preliminary assay is necessary in order to determine the right quantities of silver, copper, and lead to be added. This determination may be made by the touchstone, by considerations of the colour and hardness of the alloy, or by cupelling 2 grains of it with 6 grains of silver and 30 grains of lead, and parting the button in a flask. Simple cupellation with lead gives satisfactory results if silver is absent or insignificant in quantity ; according to Fremy this method, in which parting is dispensed with, is accurate to 3 milliemes if carefully performed with proofs.

How to Assay Gold using Cadmium

Balling has shown that cadmium may be substituted for silver in the operation of parting. The ½ gramme of gold alloy is placed in a porcelain crucible in which a little fragment of potassium cyanide has been previously fused in order to protect the metal from the air. Cadmium is then added in the proportion of 2½ to 1 of gold. If silver is present in addition, the combined weight of cadmium and silver must be 2½ times that of the gold. The whole is fused and then cooled and plunged into hot water to clean the button, which is then crushed, parted in nitric acid (specific gravity 1.3), boiled in water for some minutes, dried and weighed. The silver, if any, can be estimated by precipitation as chloride from the acid solution. By this method the losses of gold and silver incidental to cupellation are entirely avoided. A similar method, employing zinc in place of cadmium, had previously been recommended by von Juptner.

Alloys of Gold, Silver, and Copper

These may be assayed by the method just given, the copper being estimated as difference; or the gold may be estimated as usual, and other assay pieces cupelled with enough lead to remove all the copper. The buttons thus obtained contain silver and gold only, and the proportion of silver is found by difference. The method of double cupellation, by which the button of silver and gold is weighed and then subjected to inquartation and parting, is less accurate.

The cupellation designed to remove the copper is made with less lead than the quantities given earlier. If little gold is present, half the amount of lead there given is used, with increasing proportions as the amount of gold present increases. The temperature of cupellation must also be lower than for gold, approximating more to that used for silver. Proofs of similar composition must be used and the operations require much practice before the necessary skill is acquired. If less than 33 per cent, of gold is present originally, the alloy may be parted at once without cupellation and the silver estimated by weighing as chloride.

Effects of the Presence of other Metals on Gold Bullion Assay

The effects on cupellation are the same as those given under ore assay. In general, if a scoria is formed owing to the presence of large quantities of antimony, arsenic, cobalt, nickel, iron, tin, zinc, or aluminium, there is a loss of gold. The alloy should in that case be scorified with lead as a preliminary to cupellation. If mercury is present gold is carried off in the form of spray, and lost as the mercury boils off. The presence of tellurium is sometimes indicated by the formation of numbers of minute beads of precious metal dispersed over the surface of the cupel. Tellurium compounds are best analysed in the wet way.

If members of the platinum group are present they remain undissolved by the parting acid, and hinder the solution of the silver, and the assay is consequently rendered unreliable. The treatment of the alloys is discussed below. The effects of the presence of small quantities of various metals on the surcharge in the ordinary parting assay is shown below. The table is the result of experiments made by the author. The presence of 5 per cent, bismuth does not affect the surcharge. All assay pieces contained 1,000 parts of gold, 2,500 parts of silver, and 91 parts of copper, other metals being added in the proportions indicated.

These differences indicate the necessity of employing special checks containing these metals if such be present in the alloys.

ASSAY OF VARIOUS GOLD ALLOYS

Wet Methods of Assay of Gold Alloys & Compounds

Methods of Bullion Assay

Among methods which have been proposed at various times, and which may still be of service occasionally in particular cases, may be mentioned: (1) The trial by the touchstone (a method formerly more extensively used by jewellers than at present); the assay by means of considerations as to (2) the colour, and (3) the density of alloys; (4) spectroscopic assay. A brief description of each of these methods is appended.

1. Trial by the Touchstone:
   This is the oldest method of assay. It consists in rubbing the gold bullion to be tested on a hard smooth stone, and comparing the appearance and colour of the streak with those made by carefully prepared needles of known composition. The effect of the action of nitric acid and dilute aqua regia on these streaks is also noted. Touchstones usually consist of Lydianstone or of silicified wood, and black or dark green stones are best. Only alloys of gold and copper or of gold and silver can be thus tested. The trial is more sensitive for alloys below 750 fine than for higher standards. The amount of gold in alloys between 700 and 800 fine can be determined correct to 5 parts per 1,000.
2. Colour and Hardness of Alloys
   These properties form a guide to the composition of copper-gold alloys, an increase of copper corresponding to a heightening of the colour and an increase of the hardness as tested by shears or a knife. On heating the alloy to redness in air, the degree of blackening of the surface is a further indication of the percentage composition, if compared with plates of known fineness.
3. Density of Gold-copper Alloys
   The determination of the fineness of these alloys by taking their densities was investigated by Sir Wm. Roberts-Austen at the Royal Mint in 1876. He showed that the densities found by experiments were nearly equal to those obtained by calculation on the assumption that the union of the two metals was accompanied neither by contraction nor expansion. The alloys examined ranged from 860 to 1,000 fine, and were made into discs which were all compressed to the same extent. The conclusion arrived at was that the fineness of large masses of gold can be deduced from their densities correct to 1/10000 part. In the case of individual coins the results are only approximate.
4. Assay by means of the Spectroscope
   This method of determining the composition of gold-copper alloys was investigated by Lockyer and Roberts-Austen. The spectrum of pure gold was shown to be altered by successive additions of copper, and near the English standard (916.6 fine) a difference of 2 or 3 parts in 10,000 in composition could be readily detected, but the amount of metal volatilised is so small that it cannot be made to represent with certainty the average composition of the mass, which is never perfectly homogeneous.

The detection of traces of gold in alloys or ores by means of the spectroscope, though sometimes attempted, is not remarkable for its delicacy or certainty. The method of procedure is to dissolve the auriferous material in aqua regia, evaporate off the nitric acid, and pass induction sparks through the surface film of liquid, when the spectrum shows some narrow bands and some nebulous bands. The latter only are seen if a drop of the solution is placed in a Bunsen flame. The method may sometimes be useful when complex minerals are being examined. Mr. Cupel has shown that 1/4000 of a milligramme of gold will show a spectrum if the spark be passed through a weak solution of the pure metal. But when operating on a slip of alloy formed of

the spectra of copper and silver alone were visible. In an alloy of gold and copper containing from 200 to 250 parts in the 1,000 of the precious metal, the gold spectrum is barely visible. On the other hand, in an alloy of gold and copper containing traces only (.01 per cent.) of the latter, the copper spectrum was distinctly shown. Alloys of gold are rarely so perfectly homogeneous that the particles of metal volatilised and giving the spectrum represent the whole mass.