V. G. Ageenkov.

On the Method of Assaying Gold Alloys by Inquartation with Cadmium

(From the Metallurgical Laboratory of the First Tomsk Technological Institute).

With the present revival of the gold-mining industry, there is a noticeable increase in the movement of gold from the bowels of the earth, through the hands of private persons, enterprises, and institutions, into the state fund and, as an inevitable consequence of passing through these stages, a revival of commercial operations with it: purchases and sales, deliveries and receipts, and so forth. And since operations of this kind each time require precise assaying, the question of assaying gold-bearing alloys attracts attention. Whereas in former times it was solved simply, without any complications, now, under the exceptional conditions of the country’s ruined economy, this question may encounter serious difficulties.

The only firmly established method used for this purpose is the classical method of inquartation with silver, followed by cupellation, separation with nitric acid, and weighing of the gold. In simplicity of execution and in the attainable accuracy of determination this method leaves nothing to be desired, and it has long been used both by gold-alloy laboratories and mints, and in general by all assay institutions. However, since it requires special apparatus for cupellation and mechanical treatment of the beads, this method, under our present conditions, may prove inapplicable, for in many cases it is not possible to equip a laboratory with a muffle furnace, a supply of muffles and cupels, and rolls for rolling the beads. This applies in particular to newly arising or revived gold-mining enterprises, for which accounting for gold is necessary and which do not have old laboratory equipment.

In addition, cupellation also has certain disadvantages: it is a lengthy and expensive operation. One determination with 3–4 charges takes, counting from the moment the charges are taken, 5–5½ hours; of these, no less than 2 hours are spent on work at the muffle, since the mere preparation of the furnace (heating the muffle), not including the cupellation itself and unloading the furnace, already takes about 1½ hours. This disadvantage, however, is absent when several alloys are assayed simultaneously, since then the time required for work at the muffle is distributed over a large number of tests, resulting in a reduction of the time spent on each determination. The use of platinum baskets during parting further reduces the time spent on one determination and thus the duration is appreciable only in isolated, non-systematic assaying. The same applies to fuel consumption and wear of the muffle¹: while negligible in mass assaying, they become a noticeable item of expense in isolated assaying.

What has been said is significant for our gold enterprises, which at present have a semi-artisanal character, since they require precisely periodic, isolated assaying by as cheap and rapid a method as possible, even if not with the degree of accuracy necessary in commercial operations with gold.

¹ For one determination, 1½–2 poods of coke are consumed. The service life of a muffle is limited and depends on the number of firings of the furnace.

Thus, the method of cupellation in many cases, especially at the above-mentioned enterprises, may prove inaccessible, and the latter must either refrain from accounting for gold or have the assaying done elsewhere, which, however, is not always possible owing to the remoteness of the mines and plants from the centers where gold-smelting laboratories are available.

But this method is by no means the only way of determining gold in alloys, and the difficulty of carrying it out in practice naturally suggests performing the assay by another available method, one that does not require any special expensive apparatus and whose execution would be accessible under modern conditions.

One such method is the method of quartation with cadmium, proposed by C. A. M. Balling,¹) which, along with the cupellation method, is mentioned in all the major courses on the Assaying Art and the Metallurgy of Gold, such as: B. Kerl (Met. Probirkunst), C. Schiffner (Einf. in die Probierkunde), T. K. Rose (The Met. of Gold), Weil (L’or), etc.

In Russian literature it was reviewed by Tomashevsky,²) who confines himself to a quotation from D. Levat (L’industrie aurifère, 1905) concerning the essence of the method and to a comparison of the time required for parting a cadmium assay with that for a cupellation assay, noting, not quite justly, that the cadmium assay does not shorten the time needed for assaying.

In brief, the essence of this method, according to the descriptions of the authors cited, is as follows. 0.25–0.5 g of the material to be assayed is fused, under a cover of molten potassium cyanide, with cadmium taken in such an amount that the sum of the silver and copper present in the alloy, together with the cadmium, exceeds the amount of gold in it by 2½ times. The resulting bead is parted 2–3 times in nitric acid, washed, dried, and the gold is weighed. Fusion with cadmium proceeds very easily and rapidly; as for parting, it is carried out in exactly the same way as the parting of beads from cupellation, the only difference being the boiling time.

All operations of the method are simple, are easily performed on an ordinary gas burner or on a kerosene stove, and require no special equipment—and in this respect the method is very attractive. But the question arises: to what extent does it meet its purpose with respect to the accuracy of the determination?

On this question the literature contains rather contradictory and often vague statements. Thus, according to D. Levat, as is evident from Tomashevsky’s note,³) the cadmium assay is only approximate, attaining an accuracy of “the third or fourth decimal place.” B. Kerl⁴) indicates the same accuracy. These statements are insufficient and not very definite: they do not include precise limits of the possible error of determination (the interval from the third to the fourth place is too large) and refer to absolute accuracy, i.e., to the weight of the gold obtained, whereas the more interesting relative accuracy, expressed in percent or in fineness, remains undefined, since it depends on the size of the initial sample.

To form a clearer idea of the degree of accuracy of the method, let us turn to the original works, which include two papers by Balling, a paper by Joptner, and a paper by Kraus. In his first paper⁵) Balling gives data from a series of experiments carried out with three alloys, in the performance of which, for treating the bead of metals (alloy + cadmium), he used

¹) Ueber die Methode der goldscheidung durch Quartation mit Zink, Oest. Ztschr. f. Berg u. Httnw., 1879, No. 50, p. 597.
²) Gorn. Zhurn., 1911, vol. I, p. 36.
³) Ibid. Levat’s original work could not be obtained locally.
⁴) Metall. Probirkunst, 1882, 386.
⁵) Oest. Ztschr. f. Berg u. Httnw., 1879, No. 50.

triple boiling in nitric acid: first in acid of specific gravity 1.2 for about 1 hour, and then twice in acid of specific gravity 1.3 for 10 minutes each. As results they give the figures: 1) the weight of the alloy submitted for assaying and 2) the weight of gold obtained in the assay. In order to make it possible to compare these results, I determined, by simple calculations, the fineness of gold in alloy A (see Table I) and converted all the weights of gold obtained by the cadmium method likewise into finenesses (thousandths)¹). The calculated figures are placed in columns I and IV of the following Table I, alongside Balling’s data; the figures of column V constitute the difference between the cadmium assay of the alloy (column IV) and the true one (column I).

Table I.

From the table it is evident that the fluctuations in the degree of accuracy of the cadmium assay reach considerable magnitudes in both directions, amounting to as much as 12 fineness units; experiment 8 gives an exceptional error of more than 100 fineness units. Of course, such accuracy is wholly insufficient in assaying gold alloys.

¹) The calculations were made as follows. For example, the alloy for experiment A was prepared by fusing 0.711 g Au with 0.717 g Ag; consequently the fineness of this alloy will be:

\[ \frac{0.711 \times 1000}{0.711 + 0.717} = 497.9; \]

in a charge of 0.119 g of this alloy, the cadmium assay found 0.058 g of gold, or

\[ \frac{0.058 \times 1000}{0.119} = 487.4 \]

fineness units of gold.

But Balling, as may be seen from his work, used in his experiments ordinary analytical balances with an accuracy not exceeding 0.1 mg, which, for the small charges he employed, could not give a sufficient degree of precision; and the diversity of the charges does not allow a reliable comparison of the results, since the given accuracy of weighing affects the result of the determination differently for different charges. Moreover, Balling, limiting himself to the investigation only of pure gold and two binary alloys (Au—Ag and Au—Cu) of one specified fineness each, leaves aside the determination of gold in ternary alloys (Au—Ag—Cu), i.e., the influence of the simultaneous presence of Cu and Ag on the accuracy of the method; yet this is precisely what is of greatest practical interest. Consequently, this work of Balling’s does not give an adequate conception of the method.

Data of a different character are given by Jüptner,¹ who, recommending the method of quartation with cadmium as entirely accurate, gives the following results of his experiments (Table II).

Table II.

Here the error in the determination already does not exceed 0.9 assays, and, although such a degree of accuracy is incomparable with the accuracy of assay by cupellation, it is nevertheless incomparable also with the accuracy obtained by Balling. However, these data too are not sufficiently reliable to be used as a basis, since Jüptner gives absolutely no indication of the working conditions under which he carried out the determinations that gave such a degree of accuracy; and the alloys he investigated are not diverse. As we see, the experiments described by Balling and Jüptner are of a preliminary character. The most thorough and valuable investigation devoted to the question of interest to us is the work of Fr. Kraus: “Bestimmung von Gold und Silber in Legirungen nach vorgegangener Quartation mit Cadmium”². In this investigation Kraus first subjected a series of alloys to assay in parallel by the cupellation method and by the method of quartation with cadmium, applying in the latter the method of dissolving the metallic bead proposed by Balling, i.e. threefold

¹ Oest Ztschr. f. Berg u Httnw., 1880, No. 14
² Dingl. Polit. Journal, Bd 236, p. 323.

boiling with nitric acid, as indicated above. The results of these experiments are given in the following Table III.

Table III.

The table shows that the cadmium assay gives, in all cases, higher results, considerably exceeding the error permissible in assaying at mints (according to Kraus, equal to 0.2 assay units). Having convinced himself that these excess weights of gold arise from nitrates of foreign metals retained in the pores of the bead, Kraus then carried out a series of experiments with the method of treating the bead altered as follows: instead of the first

acid of specific gravity 1.2; he used acid of the same concentration as in the second case, i.e. specific gravity 1.3, and, instead of a third treatment in acid, employed boiling in distilled water for 5 minutes. To the latter circumstance—the additional treatment of the bead with water—he attached substantial and important significance, which he confirmed by the results obtained, brought together in the following Table IV.

Table IV.

Taken for the experiment: Au, Ag, Cu, Cd—in units of gold assay weight, i.e. 0.5 g = 1000 parts.

Thus, additional treatment of the bead by boiling in water, facilitating the removal of nitrates, is indeed an improvement, since it enabled Kraus to attain greater accuracy in the determination.

We shall note, however, that in the last series of experiments Kraus did not use ready-made alloys, but prepared them each time by fusing specified quantities of metals, weighing in total 0.5 g, with cadmium directly into a bead, which was then subjected to further treatment. Considering the method of inquartation with cadmium to possess great advantages over ordinary cupellation, and believing that he had arrived at correct results deciding the question of the accuracy of the method, Kraus nevertheless remarks that it would be desirable for the latter to be “vergleichweise und näher geprüft”—comparatively and more closely investigated—in various assay institutions conducting routine determination of gold-bearing alloys. Unfortunately, Kraus himself did not carry out such a comparative investigation of alloys (with cupellation) using the method he proposed for treating the metallic bead.

The results obtained by Kraus were checked by Balling, who in his second work¹ investigated a new series of alloys—various combinations of Au, Ag, Cu with platinum. The chief aim of this work was to determine the influence of platinum on the determination of gold by the cadmium method, which in itself is not without interest; but it is also interesting in that the experiments were carried out using Kraus’s proposed method of treating the metallic bead.

¹ Oest. Ztchr. f. Berg u. Httnw., 1881, No. 4, p. 53.

The results of this work, presented in Table V below, at first sight show a quite satisfactory accuracy of determination; but upon detailed consideration it turns out that substantial errors were made in Balling’s calculations. He prepared the alloys for his experiments as follows: he accurately weighed arbitrary quantities of the components of the future material in various combinations, fused them, and then calculated, from the initial weights, the percentage composition of the alloy obtained. From the series of materials thus prepared he took samples and determined in them the gold content by the cadmium assay and by calculation, subsequently comparing these figures. On checking, it turned out that the gold content in most of the samples had been calculated incorrectly. Thus, for example, in the 3rd experiment Balling gives the gold content as 0.1850 g in a sample of 0.2229 g of alloy, containing 11.215% Cu and 0.852% Pt, whereas in fact the gold in it should be:

\[ \frac{0.2229 \cdot [100 - (11.215 + 0.852)]}{100} = 0.1960 \text{ g.} \]

By the cadmium assay, 0.1850 g of gold was determined in this same sample; consequently, the difference is \(0.1850 - 0.1960 = -0.0110\) g, or in assay parts:

\[ \frac{-0.0110 \times 1000}{0.2229} = -49.3, \]

whereas according to Balling this difference \(= 0\).

The figures corrected in this way are given in the same table, in column 4.

Table V.

These data allowed Balling to conclude that platinum, being removed without difficulty by nitric acid together with cadmium and, especially, with silver, does not noticeably affect the accuracy of determinations of gold by the cadmium method, if it enters into

alloys in an amount not exceeding 1%. However, as is seen from Table V, the corrected figures greatly change the picture, and it becomes difficult to say what, in fact, is the cause of such variability in the results: the presence of platinum, the shortcomings of the method itself, or something else¹.

The above-mentioned investigations by Balling, Jöptner, and Kraus exhaust the experimental works devoted to the cadmium assay. Their data on the degree of accuracy of the latter, as we see, are poorly defined and insufficiently concordant. They agree, perhaps, only in that they speak, in general, of a greater or lesser degree of inaccuracy of the method. As for the limits of this inaccuracy, they remain unclear; for although Kraus’s investigations yield a regular series of figures, Balling’s second work does not confirm them. At the same time, in the most recent courses on assaying, as indicated above, the accuracy of the cadmium assay is characterized as approximate, reaching the 3rd or 4th decimal place², which is more in agreement with Balling’s data than with Kraus’s.

Thus, the literature sources do not provide the analyst wishing to use the cadmium method for assaying alloys with a firm basis on which he could rely without a preliminary verification of the limits of the permissible error of determination.

With the aim of carrying out such a verification and elucidating the possibility of practical implementation of Balling’s method, the present work was undertaken.

It was first proposed to assay by the cadmium method a series of artificially prepared alloys with different gold contents and, having established the degree of accuracy, then to verify it on industrial alloys. In this connection, as a basic condition, the following provision was adopted: to compare the results obtained with the assay by ordinary cupellation. The following considerations speak in favor of this condition:

1) determination of gold by cupellation is very accurate—no lower than 0.2 fineness units³,

2) in view of the liquation and oxidation of copper during preparation of the alloy, its fineness, calculated from the initial weights of the components, rarely coincides with the true one; therefore comparison of the results of the cadmium assay with such a calculated fineness cannot be reliable,

and 3) although Kraus’s procedure—namely, weighing the constituent elements separately for each experiment, and comparing the weight of the bead obtained as a result of the experiment with the weight of the gold that went into the charge—is attractive, it requires absolutely pure gold and, moreover, there is no assurance that the behavior of the components during treatment of the bead will be the same as when assaying a finished alloy; therefore such a procedure also cannot be applied in checking the results of the cadmium assay.

To carry out the proposed plan, from the various gold residues available in the laboratory, a first series of initial alloys was prepared; for this purpose these residues were first fused in a crucible furnace in a graphite crucible, and the bead obtained was forged and rolled into a plate for convenience in taking portions from it. The material was then subjected to assay by cupellation, which showed the following composition:

\[ \begin{aligned} \mathrm{Au} &— 961.2 \text{ fineness units}\\ \mathrm{Ag} &— \phantom{0}21.8 \quad »\\ \text{Ligature} &— \phantom{0}17.0 \quad »\\ &\overline{\phantom{—}1000.0 \quad »} \end{aligned} \]

A certain quantity of this material served as one of the alloys investigated, and from the remaining portion, by alloying with pure silver and electrolytic copper, other initial alloys were prepared, with the calculation that they should contain different gradations of gold content; in all cases the addition of copper and silver was made in the ratio \(2:1\). The beads obtained after fusion were rolled into plates, from which shavings were cut. These materials were then examined by parallel assaying by the cupellation method and by Balling’s method.

Assaying by cupellation was in all cases carried out under the following conditions.

1) On bead balances, 3 portions were weighed out for each alloy, 250 mg each, of which 2 served as control assays;

2) quartering with silver was carried out in the ratio:

\[ \mathrm{Ag}:\mathrm{Au} = 2.0\text{—}2.2:1; \]

with such a ratio the greatest accuracy of determination is attained, as is clearly shown by the following data of Rose¹, which are the result of a large number of uniform experiments:

3) The addition of assay lead was made according to Kandelhardt’s table, partly in the form of foil, in which the weighed portion of the alloy was wrapped, and partly in the form of small balls.

4) Cupellation was carried out in cupels of the firm Morgan, Battersea.

5) The cornets were boiled according to the directions of T. K. Rose²: first the cornet was boiled for 25 min. in 25 cm³ of previously heated nitric acid of specific gravity 1.2, the silver nitrate was decanted and the small flask was washed with hot water; then 20 cm³ of hot nitric acid of specific gravity 1.3 was added, a piece of pumice washed with water was introduced, and boiling was continued for another 15 minutes, after which the acid was decanted and the cornet was washed by decantation 3 times with hot water and 2 times with cold; the subsequent manipulations are the usual ones.

6) The weighing of the gold was carried out on bead balances with an accuracy of up to \(\pm 0.02\) mg; moreover, in the event that at least two cornets gave completely coincident weights, the assay of the alloy was calculated from these latter, as usually happened.

The alloys assayed by cupellation were then subjected to testing by Balling’s method, carried out in the following manner. For treating the bead, the procedure indicated by Schiffner³ was adopted as the newest: double boiling in nitric acid—the first in acid of specific gravity 1.2 for 60 minutes and the second in acid of specific gravity 1.3 for 10 minutes, and then boiling in distilled water for 5 minutes. The subsequent washing of the bead, transfer to a crucible, ignition, etc., were carried out as in the cupellation assay. For the assay, 2 portions of the alloy of 250 mg each were taken, and each was treated separately⁴. The addition of cadmium was \(2\frac{1}{2}\) times the weight of the gold, after subtracting the Cu and Ag present in the alloy. With this quantity of cadmium the metallic bead, according to the indications of all authors, possesses the greatest ability to resist disintegration during boiling.

To avoid losses of gold during alloying with cadmium, which could occur from incomplete alloying of the material with it, and mechanical losses when transferring the charge into the crucible with potassium cyanide, this operation was carried out as follows. From cadmium, which very readily lends itself to mechanical working, “foil” was prepared by rolling; its thickness was regulated so that the amount of cadmium required for one assay had the form of a plate about 3 cm square. The sheet thus obtained was folded into a so-called “pocket” (as is done in making a paper filter); the charge was placed in it and wrapped into a “packet,” which was then lowered into potassium cyanide previously melted in a crucible.

The weighing of the gold here too was carried out on the same bead balances as in the cupellation assay; in the event of a considerable disagreement between the two results, a third charge was taken for checking.

The results of the experiments made under these conditions with the first series of alloys mentioned above are presented in the following Table VI.

Table VI.

The difference in the determinations by the cadmium method and by cupellation, as we see, proved to be considerable. But since the experiments were carried out with gold that was known to be contaminated and that had gone into the preparation of the original alloys, it may be thought that the cause of the large difference lies in the fact that one of the contaminating elements—apparently lead—though present perhaps in an insignificant amount, nevertheless strongly promoted the retention of foreign elements in the gold bead. This supposition was justified by the following abnormalities observed in carrying out the experiments. The alloys investigated, when rolled into a plate, even after annealing, showed considerable brittleness, not excluding the original alloy No. 1, despite its high fineness. At the moment when the packet of the charge was lowered into the molten potassium cyanide, the alloying of the metals was accompanied by a bluish flash.

by a flame and by separation from the liquid bead of a considerable amount of a flocculent precipitate, not alloying with the latter, black in color.^1) Further, on boiling in acids and, especially, in water, the beads readily crumbled, often into a fine powder, which on further ignition had not the slightest tendency to sinter into a single whole. The color of such a powder, even after ignition, remained brown, owing to the presence of oxides formed during ignition \((\mathrm{CdO}, \mathrm{CuO}, \mathrm{PbO})\).

That foreign elements were in fact retained in the gold bead, forming the so-called “sweep,” was established in the following way. Three gold beads (after weighing) were dissolved in a small quantity of aqua regia; the solution was evaporated twice with \(\mathrm{HCl}\) and diluted with water, whereupon a precipitate of \(\mathrm{AgCl}\) fell out—proving the presence of silver. After settling, the \(\mathrm{AgCl}\) was filtered off; the filtrate was evaporated to a small volume; the gold was precipitated with hydrogen sulfide, filtered off, and in the filtrate \(\mathrm{Cu}\) and \(\mathrm{Cd}\) were determined—the former by reaction with ammonia, the latter by precipitation with hydrogen sulfide from a cyanide complex.

On the basis of Balling’s statement that, with a larger quantity of added cadmium, the separation of gold from foreign metals proceeds better,^2) in order to determine the possibility of reducing the sweep, a series of experiments was carried out with a \(3 \tfrac{1}{2}\)-fold quantity of cadmium. The results of this series of experiments are shown in the same Table VI. They appear to confirm this statement, but here a new circumstance appeared—the inevitability of mechanical losses, because in these cases the bead crumbled into fine dust, the collection of which is difficult. Assaying the beads of this series of experiments for sweep showed the presence of the latter and, apparently, in the same quantity as in the preceding case; consequently, the negative “differences” of Table VI were for the most part due precisely to mechanical losses of gold,^3) and only to a lesser extent to the influence of the quantity of cadmium. To prove this proposition, it was necessary during parting to preserve the bead in an intact form. Preliminary experiments established that, when boiling in an air bath, this can be done without difficulty, since the little flask in it is heated uniformly over its entire surface and therefore, during boiling, no shocks occur caused by local overheating; the data for one of such experiments with alloy No. 4 are shown in Table VI—they confirm the above assumption.

Experiments with this series of alloys were left at that; they served as preliminary ones. Subsequently it was decided to carry out a series of similar experiments under exactly the same conditions, but with alloys prepared from chemically pure metals.

For this purpose, chemically pure gold was prepared from the pieces that had accumulated in the laboratory from parting, according to Krüss’s method,^4) with alteration of certain details in the separation from silver and, chiefly, with alteration of the method of precipitating the gold, which was carried out by saturating a pure solution of auric chloride with hydrogen sulfide. The spongy gold, after precipitation, drying, and ignition, went directly to the preparation of alloys.

Chemically pure silver was prepared from the chloride by twofold reduction of the metal with soda; after the secondary reduction the silver was cast into a mold in the form of a plate and then rolled into a thin strip, from which “quartation pieces” were cut.

Copper used was electrolytic, of Kahlbaum’s grade.

^1) Consisting, apparently, chiefly of \(\mathrm{CdO}\) and carbon. By reaction with \(\mathrm{SnCl_2}\), gold was not detected in this precipitate.

^2) Oest. Ztschr. f. Berg u. Httnw, 1879, No. 50.

^3) Incidentally, let us note that the negative differences in Balling’s data are probably due to the same cause; Balling himself refrains from explaining this phenomenon.

^4) Liebigs Annalen, 273—238, (1887), p. 43.

From the purified materials thus obtained, 5 alloys of different gold concentrations were prepared by melting definite quantities of the metals in a fireclay crucible under charcoal on a Meyer burner with blast; the addition of copper and silver was made, as in the preceding series of alloys, in the ratio \(2:1\). After melting, the buttons were cleaned of charcoal, rolled into plates, and cut into shavings. The low-standard alloys Nos. 8 and 9, before the last operation, were first cleaned by boiling in dilute sulfuric acid from the thin film of copper oxide that formed during annealing, which proved necessary for rolling these materials because of their hardness¹. The other alloys did not require annealing, and the surface of the plates was sufficiently clean.

In the investigation of these alloys, experiments were also carried out with a \(3 \tfrac{1}{2}\)-fold quantity of added cadmium; in these cases the boiling of the buttons was performed in an air bath. In addition, in order to determine whether the boiling time could be shortened, further experiments were made with a duration of boiling in the first acid of 40–45 minutes.

The results obtained for this series of experiments are summarized in the following Table VII, in which asterisks mark experiments with a shortened duration of boiling.

Table VII.

Further, under entirely analogous conditions, investigations of the same kind were carried out with four industrial alloys. The latter were remelted in the usual way in a graphite crucible, cast into ingots, from which turnings for assaying were taken on a drilling machine. The following results were obtained (Table VIII).

Table VIII.

In carrying out the experiments presented in Tables VII and VIII, the phenomena that had occurred in the investigation of the contaminated alloys described above (p. 98) were not observed: with sufficient attention it was considerably easier to prevent the destruction of the bead (even with a 3½-fold quantity of cadmium), and if it did fall apart, it did so into large particles, which were easily collected into a bead and, upon subsequent annealing, just as easily sintered into a single whole, like the cornets in cupellation, acquiring the color of gold. A flash with a bluish flame during alloying with cadmium was observed in these experiments as well, but the separation from the bead of a flocculent precipitate was here insignificant.

The results obtained (Tables VII and VIII) indicate the degree of accuracy of Balling’s method as not exceeding 2–3 assays and, with a 3½-fold addition of cadmium and treatment of the bead in the first acid for not less than 60 minutes, as reaching 1.0 assay, which differs little from Kraus’s data, who in analogous experiments obtained a value of about 4 assays (Table III). But Kraus, attributing such a small degree of accuracy to the retention of nitric-acid cadmium in the pores of the bead, showed in the second series of his experiments that to some extent this phenomenon can be avoided if an additional treatment of the bead with water is applied, and that then a degree of accuracy not lower than 0.6 assay can be achieved. In all my experiments, additional boiling of the beads in water was included as an obligatory condition and, nevertheless, Kraus’s last figure was not confirmed by them.

Consideration of Tables VII and VIII shows that Balling’s method always gives overweight values and that its inaccuracy is due exclusively to “clogging—

metals,” for negative “differences” were obtained only when the gold bead crumbled to a powdery state and, consequently, mechanical losses were unavoidable. The same may also be seen from Kraus’s data: in almost all cases he obtained gains, not excluding the second series of experiments (Table IV).

Such an impediment, as is known, is also unavoidable in the cupellation method and is fairly considerable. On p. 97 we saw that it amounts to about 0.7 assay units, and often, under normal working conditions, from 1.0 to 1.2 assay units of silver remain in the cupel.² But in cupellation there are also losses of gold through volatilization, absorption by the cupel, and dissolution in acid, and these losses, by a fortunate accident, are precisely compensated by the impediment to such an extent that the determination of gold by cupellation is a very accurate method.

In most cases no similar compensation occurs in the cadmium assay, for the only possible source of gold losses—dissolution during boiling—is evidently insufficiently large; and, owing to the inconvenient form of the bead for dissolving metals (spherical), this dissolution is hindered and the impediment reaches greater dimensions than in cupellation, while it is rarely constant even under completely analogous working conditions, depending on certain chance circumstances. This is indicated by the fact that gold beads from one and the same alloy, assayed in the same way, nevertheless rarely coincide in weight. Only under certain conditions does this compensation sometimes occur also in the cadmium assay, since some experiments, as may be seen from Tables III, IV, VII, and VIII, give a difference equal to 0. And, evidently, there are factors favoring the realization of such compensation. In previous experiments it was shown, for example, that a larger amount of added cadmium and a longer duration of boiling are among such factors, for they promote a reduction of the impediment. Such factors may also be the intensity of boiling, the concentration of the acid, etc.

Turning to the conditions under which Kraus’s second series of experiments was carried out, we note that here, as the first acid, as mentioned above, Kraus used acid of specific gravity 1.3 at once; and the accuracy of determination he obtained in these experiments, about 0.2–0.6 assay units, which was not confirmed by my experiments, in which acid of specific gravity 1.2 was used, may be ascribed not only to the additional treatment of the bead with water, as Kraus asserts, but also to the use, during boiling, of a more concentrated first acid. It may be supposed that the latter, owing to its more energetic dissolving action with respect to the foreign metals, and also to the gold itself—especially in the presence of an abundant quantity of oxides of nitrogen,³ liberated at the beginning of the action of the acid on the bead—is one of the above-mentioned factors. In order to verify this supposition, several experiments were carried out in which the boiling of the bead was performed twice in nitric acid of specific gravity 1.3—in the first instance for 60 min. and in the second for 10 min.—and once with water for 5 min. The same alloys as in the preceding experiments served as the materials for these experiments. The following results were obtained (Table IX).

¹ The presence of an impediment from Ag, Cu, and Cd was in this case also actually established by treating 28 gold beads in the manner indicated on p. 99.

² T. K. Rose, The Met. of Gold, 1906, p. 485.

³ Oxides of nitrogen form nitrous acid, which, according to Rose (The Met. of Gold, p. 484), dissolves gold if there is no silver in the alloy; and since, in quartation with cadmium, the bead contains little silver, dissolution of gold in this case, in the presence of oxides of nitrogen, may occur.

Table IX.

In carrying out the experiments it was observed that the dissolution of foreign metals in the first acid proceeded more vigorously than in the preceding experiments, and without detriment to the integrity of the gold bead; on the contrary, with such treatment the latter could more easily be preserved in an undissolved form without special precautions, as Kraus also indicates in the article cited above.

As regards the degree of accuracy, as we see, it rose to 0.7–1.0 trial; but, taking into account that here the weighed portions were 0.25 g, it may be assumed that with larger weighed portions the accuracy of the method may rise to 0.5 trial, as Kraus indicates by the data of the second series of his experiments (Table IV), for which he took weighed portions of 0.5 g.

Consequently, the method of treating the metallic bead used in the last experiments may be considered the most correct, giving the best results. And the degree of accuracy of Balling’s method with this manner of working may be characterized as reaching 1.0 trial, always on the side of overweight, if the gold bead has retained its form. Only with more careful and prolonged treatment and with alloy weighed portions of 0.5 g can an accuracy of determination of 0.5 trial be attained.

I shall note, incidentally, that an instructive example of what the cadmium method can give in assaying contaminated alloys is furnished by the figures of Table VI.

The practical performance of this method in assaying gold alloys with the indicated degree of accuracy of 0.5–1.0 trial, worked out in agreement with the indications in the literature and on the basis of my own experience, is as follows.

A weighed portion of 500 mg of alloy, wrapped, as described above, in cadmium foil taken in approximately 2½ times the quantity of the weight of gold contained in the weighed portion, is lowered into a bath of 3–4 g of potassium cyanide, melted over a gas, alcohol, or kerosene flame in a porcelain crucible (capacity about 10 cm³). At first the packet is held on the surface of the KCN; then the cadmium begins to melt and the weighed portion sinks to the bottom, where the metals finally alloy into a liquid, mobile bead. The alloying lasts no more than ½ minute, after which the crucible is removed from the burner, allowed to cool completely, and then immersed in a beaker with hot water, in which the potassium cyanide rapidly dissolves, releasing the metallic bead¹. The latter is removed from the beaker with tweezers and washed with water. At this time about 25 cm³ of nitric acid of specific gravity 1.3 is placed in a dissolving flask² and heated with a gas or kerosene flame on an asbestos gauze. When vapors of nitric acid appear from the neck of the flask, the bead of metals is introduced into it with tweezers, without interrupting the heating. Violent boiling immediately begins, with abundant evolution of nitrogen oxides; the latter

after a few minutes it disappears, and the boiling becomes calm and even; it is continued for not less than 1 hour. After this, the nitrates are removed from the flask by decantation, the latter is washed with hot distilled water, 15–20 cm³ of nitric acid, sp. gr. 1.3, heated almost to boiling, is added, and it is again boiled for not less than 10 minutes. Then the acid is removed, the flask is again washed with hot water, 15–20 cm³ of distilled water heated to boiling is added, and it is boiled for another 5 minutes. Next the water is poured off, the flask is washed twice with hot water and once with cold. Having poured off the last washing water, the small flask is filled to the top with distilled water and inverted into a porcelain crucible. Then, as usual, the gold is dried, ignited, and weighed.

If boiling in the second acid begins with difficulty, showing signs of possible jolting, then, in order to avoid destruction of the bead, a piece of pumice, chamotte, or, as Kraus recommends, a piece of charcoal boiled in water is lowered into the flask. Often the bead can be kept undestroyed even without these means, especially when an air bath is used during parting. In the event that the bead is destroyed to a dust-like state, the assay must be regarded as lost; this, however, with sufficient care, occurs very rarely.

The number of alloy samples weighed out for checking should be not less than three.

When weighing out the required quantity of cadmium, it is necessary to know beforehand at least approximately the fineness of the alloy; but if it is not known, this is not necessary, as is clear from the following. Table VII shows that a 3½-fold quantity of cadmium relative to the weight of gold not only does not reduce the degree of accuracy of the method, but even increases it; therefore, if the addition of cadmium is within the limits of 2½ to 3½ times the amount of gold, one may count on obtaining an accuracy of determination not lower than 0.5–1.0 fineness. On the basis of these considerations the following small table has been compiled, by which, in assaying, one may determine the cadmium addition, taking a multiple of the alloy sample, being guided in this by an estimate of the fineness of the alloy by eye, according to the color of the ingot or filings:

i.e., if the fineness of the alloy, by color, is determined, for example, at about 800, then for a sample of 500 mg, cadmium must be taken as \(0.5 \times 2 = 1.0\) g. The cadmium addition taken according to this table is always within the limits of 2½ to 3½ times the amount by weight of gold, and the indicated gradations of fineness, with some experience, are easily distinguished by eye. At the same time, only more attentive and careful parting of the bead is necessary, in order to avoid its destruction. Moreover, the use of an air bath during parting¹ can guarantee the integrity of the bead.

One determination by Balling’s method, with 3 weighed samples, counting from the moment of weighing out the latter, takes 2½–3 hours of time, of which 1 hour of parting in the first acid requires no attention and may be regarded as free. As we see, the saving of time, in comparison with cupellation assaying, amounts to not less than 2–2½ hours and is obtained at the expense of the heavy work at the muffle, requiring skill, and the mechanical treatment of the beads.

¹ In the absence of one, an air bath can easily be made from several pieces of asbestos board.

The method does not require any special laboratory equipment: ordinary porcelain crucibles (whose service life is unlimited), burners, flasks, and the like—apparatus available to any laboratory—make it possible to carry out assaying by this method without special expense, and this is its chief advantage.

As for materials and fuel, their consumption in assaying by the cadmium method is negligible: 10–12 g of potassium cyanide, 2–4 g of metallic cadmium, the usual amount of nitric acid, and a small quantity of fuel for operating the burners for about \(1 \tfrac{1}{2}\) hours—this is all that is required for one determination. This consumption increases in proportion to the number of samples.

It is worth noting that if, in assaying alloys, determination of the ligature is also required, it can easily and quickly be carried out in the method of quartation with cadmium concurrently with the determination of gold (with a degree of accuracy sufficient for this method). For this purpose the acids decanted after boiling and the wash waters are collected in an Erlenmeyer flask, and the silver in them is determined by titration with ammonium thiocyanate according to Volhard¹. As Kraus² has shown, the silver determination thereby attains an accuracy of not less than 2.0 assay units. The ligature can be calculated by difference. Thus, from this side as well, Balling’s method presents no difficulties.

On the basis of all the foregoing, the following conclusion may be drawn. The determination of gold in alloys by the method of quartation with cadmium, giving a degree of accuracy not exceeding 0.5–1.0 assay units and only rarely giving coincident results for one and the same material, cannot serve for purposes of exact assaying. For the latter, the method of quartation with silver remains the only and irreplaceable one. In those cases, however, where the indicated degree of accuracy in the determination of gold is permissible, the method has many indisputable advantages over cupellation, both in the simplicity and uncomplicated nature of the operations and in the accessibility and low cost of the laboratory equipment, and—when assaying single samples—in the saving of time, labor, and materials.

For purposes of accounting for gold at our gold-mining enterprises, where assaying is performed on single samples and an accuracy of 0.5–1.0 assay units is entirely sufficient, Balling’s method can be successfully applied, especially under present economic conditions, in which it may often prove to be the only practically possible one.

In conclusion, I consider it my pleasant duty to express my profound gratitude to Prof. V. Ya. Mostovich, whose direct supervision, advice, and instructions contributed to the completion of the present work.

1. Volhard’s method gives an accuracy in determining silver in alloys of about \(+4.0\) assay units, but owing to losses in collecting the wash waters and the retention of Ag in the bead, some compensation in the direction of better accuracy occurs here. Obviously, an exact method, the determination of silver according to Gay-Lussac, is not applicable in this case. ↩↩↩↩↩

2. Dingl. Pol. Journal, Bd. 236, p. 326. ↩↩↩↩

3. Einf. in die Probierkunde, 1912, p. 108. ↩↩

4. Kraus treated up to 2 beads in one flask. ↩