It is common to blame irregular assay results upon volatilization and much has been written upon the subject, but there is no real evidence that, in a properly conducted assay, the loss of either gold or silver by volatilization is sufficient to affect appreciably the result, even when arsenic or antimony may be present. Bulk assays of flue dust from assay muffles have been published, but the data given are totally insufficient to even approximate the volatilization from a single assay; and such figures as we have indicate the volatilization to be extremely small. Diligent search in the literature and wide inquiry among assayers and instructors have failed to produce a single case where the litharge volatilized in making an assay has been collected and assayed for gold and silver. Having attempted to do this, with most indifferent success, I am not surprised that this has not been done.

While various textbooks give volatilization as a cause of loss in cupellation, it may safely be said that in the rare cases where this volatilization is sufficient to affect appreciably the result it is due to excessive temperature. Percy says “The loss of silver by volatilization during cupellation is very slight (unless the temperature has been much too high), and may be dis-regarded.” This statement is repeated by the Beringers and Smith, who includes gold with silver. Mitchell says the litharge fume rarely contains over one ten-thousandth of silver. Campredon says that at a proper temperature no silver is lost by volatilization.

Primarily, excessive temperature may be due to too high general temperature in the muffle; but it may often be due to temperature localized within the bead by oxidation of the metals and this may be influenced by other causes than the general temperature. I have shown an extreme case of local temperature, resulting in the burning of zinc with its characteristic flame and variations in bead temperatures with uniform pyrometer readings. Volatilization from a bath of mixed metals is not the simple question of vapor pressures of the contained metals. Based entirely upon vapor pressures, Richards, in speaking of the distillation of zinc from Parkes crust, says “Successively larger amounts of lead, silver, and gold come over the higher the temperature,” but I have shown that more silver goes with the zinc in the early stages of the operation than later.

Volatilization in Cupelling

The possible volatilization of the precious metals in cupelling is a most complex proposition. It may be due to the vapor pressure of the individual metals, the vapor pressure of alloys or possible metallic compounds, of compounds of metals with metalloids or with non-metals, particularly oxygen. Also it is not possible to draw a sharp line between true volatilization due to vapor pressure and the drag of a volatilizing body exerted upon a less volatile associate, against which is the counter-drag, or stabilizing effect, of the non-volatile upon the more volatile, all of which are influenced by concentration and association. The temperature conditions of the cupelling beads vary constantly, which may change any or all of these conditions. The oxygen supply influences them. Even more important, perhaps, is the extent to which the surface of the metal is covered and protected by the litharge formed. At moderate heats much of the button surface is covered by the litharge and the opportunity for direct volatilization of the metals reduced. Intimately connected with this is the question of secondary volatilization, oxide of lead being volatile, oxide of silver may be dragged along, even if it is not co-volatile. No one knows how much, if any, of the lead, or any other metal, present in cupel fumes volatilized before oxidation.

Rose suggests that gold and copper go off as “vaporized alloy” and Mitchell says silver is much more volatile when alloyed with lead. Wright bessemerized silver bullion high in arsenic and the arsenious oxide produced showed about 800 oz. silver per ton. I have roasted rich arsenide ores at a very low temperature and the arsenious oxide given off showed only a few ounces of silver per ton. Mitchell says the litharge fume in cupeling is “produced by the vapor of lead burning in the atmosphere,” but gives no data.

Again other questions of molecular physics are involved. Keller has shown that gold travels into copper on prolonged heating of the alloy below its melting point. So the rate at which the precious metals withdraw into the cupeling bead must affect the drag. Gillett says that on heating brass, loss of zinc depends, in part, “on the rate of diffusion of zinc from the body of the melt into the surface that is losing zinc.”

Loss Through Dusting

Many statements regarding volatilization in assaying are based upon data obtained under conditions entirely different from cupeling and much that has been written fails to distinguish between volatilization and dusting, which is purely mechanical. In practical operations, losses are often ascribed to volatilization that are largely, if not entirely, due to dusting and other mechanical causes.

In modern practical work, there is now probably very little straight cupellation of lead. Where works have lead rich enough to cupel, advantage is generally taken of this to work up other high-grade products, rich ores, slimes, etc. Much of the extremely rich silver ore from Cobalt, Can., has been treated in this way. These additions, particularly very fine slimes, favor dusting and high silver in fume or dust collected from cupeling furnaces does not prove high volatilization. Iles gives many figures on fumes and dusts from various sources. A mixed cupel fume showed 85.90 oz. silver per ton; this might indicate volatilization, but the fact that it was only 72.50 per cent, lead showed dusting, oxide of lead containing 92.83 per cent. lead.

Mr. Dieffenbach, has furnished a month’s record of the results from the Cottrell condensers on the cupel furnaces. The fume collected weighed 24,659 lb. and contained 9863 lb. lead, or approximately 40 per cent., indicating much dust. This is accounted for by the fact that about 70 tons of slimes were treated on the lead bath. The fume contained 5830.11 oz. silver or 0.64 per cent, of the silver charged. In view of so much mechanical dusting of rich stuff, the actual volatilization must have been very small. The slimes charged contained nearly 555,000 oz. silver. In a negative way, the figures show that there was practically no volatilization of gold for, notwithstanding the dusting, only 12.07 oz. gold were found in the fumeout of 16,855 oz. charged, of which over 13,000 oz. were in the slimes.

Condensation of Assay Muffle Fumes

Various persons have reported the finding of gold and silver in the fumes condensed in the flues of muffles, but these generally show only minute quantities of the precious metals and all lack quantitative significance. Smoot says “we assayed the condensed fumes from the muffle ventilation pipe and found only a very small amount of silver and practically no gold” and “this seems to indicate, in a qualitative way, that only a negligible amount of silver is lost in the lead fume.”

The best figures that I know of regarding the condensation of assay muffle fumes have been supplied by Donald M. Liddell. One muffle was used only in pig copper assays by the combination method. It carried very close to 90 oz. per ton and the residue from 1 assay-ton, after burning off the paper, was scorified with 1 assay-ton lead down to about 7 gm., which was cupeled in the same muffle. The muffle had a small forced-draft apparatus at the back and all fume passed through an iron pipe, which formed a fairly efficient condenser. This pipe was cleaned out every 6 mo., during which time about 5500 assays were made, yielding about 175 gm. of litharge. One lot of 175 gm. gave 142.7 mg. silver but not enough gold to weigh. This would roughly mean the volatilization of about 0.026 mg. silver per assay or approximately 0.03 per cent. It would seem fair to divide this between the scorification and cupeling, estimating the cupel volatilization as 0.015 per cent.

Vapor Pressure and Volatility

Johnston has published a most valuable review on the vapor pressure and volatility of several high boiling-point metals, giving many references, particularly of exact scientific investigations. The ultimate object of this paper was to devise a formula for determining the approximate volatility of various metals at different temperatures. The data regarding gold, however, were too scanty for this. There is absolutely no connection between melting point, vapor pressure, and boiling point. Various metals show vapor pressure below their melting point. Tin, with a melting point of 232°, is said to boil at about 2300° while copper, with a melting point of 1083°, is said to boil about 100° above tin.

Volatility of Gold

Varied statements have been made regarding the volatility of gold. Gmelin says “It exhibits a scarcely perceptible volatility at the strongest heat of a blast furnace, in the focus of a burning mirror, and in a flame fed with oxygen gas.” Napier appears to have been the first to make systematic tests and is often quoted, but his figures are small, about 0.1 per cent., and the time of heating long, 3 to 8 hr. Makins is often quoted, but his figures are extremely small even though an extraordinary heat was used in the cupellations; there is no connection with the amount of gold operated on; he simply found 0.087 gr. of gold in 1000 gr. of flue dust. The volatility of gold before the extreme heat of the blowpipe is described in Crookes.

Seamon and Parker heated beads of gold, 1 to 20 mg., varying periods of time, to temperatures from 1000° to 1650°. Losses were observed only twice and were attributed to unobserved errors of manipulation. Under the conditions of a cathode-ray vacuum, on prolonged heating, Krafft and Bergfeld were able to observe volatilization of gold at 1070°. Moissan places the boiling point at 2530°. Mostowitch and Pletneff state that in oxygen, nitrogen, carbon monoxide, and carbon dioxide at 1400°, volatilization is not appreciable but is noticeable at 1250° in hydrogen. Kern and Heimrod made extensive tests on the volatility of gold and silver, obtaining some interesting results, but their conditions were quite different from cupellation. Many hours heating in contact with carbon did not volatilize silver from copper, but did volatilize gold from copper, possibly by the formation of a carbide of gold. Incidentally, their quotation from Napier is faulty. Wurtz gives a review of early information.

T. K. Rose has been a prolific investigator of this subject and his books and papers contain many references. In 1893, he stated that volatilization of gold begins just below 1100° and at 1250° the loss per minute was four times as great as at 1100°; the loss on a charge of 1200 oz. was figured to be 0.02 per cent, per hour;the volatilization in cupellation was insignificant. Later, he says that the loss in an ordinary melting furnace is seldom over 0.01 per cent, on 1200 oz. and is influenced by temperature, metal surface exposed, and time, and is increased by air passing over the metal and by impurities. Still later, he says, “True volatilization of gold is so small as to be negligible at the temperatures of industrial melting furnaces, say 1000° to 1300°. It is difficult to measure with accuracy the infinitesimal amounts volatilized at these temperatures, ” “Even in a strong draft the amount volatilized remains exceedingly small” provided the nature of the gas does not change. Alloys, however, may take up and eject gases with spurting, which may give globules less than 0.001 mm. in diameter, which may be carried away by even a slight draft and are difficult to recover. The first table of results, p. 4, is an excellent illustration of a fact I have often met in the literature and in my own experience with gold; that is, the effect of unrecognized conditions leading to erratic results when expecting regularity.

In a paper on accuracy in assaying, Rose attempted to follow the course of the gold and gives results from which he concludes that while final loss is shown in each case it does not represent the amount volatilized with certainty, errors of five determinations being accumulated here. I would add that he determined the silver and other metals left in his cornets in a lump by a reassay, which must have been subject to all the errors of the original assay.

Hillebrand and Allen give some figures upon volatilization, but they exhibit serious irregularities. On gold alone two cupel recoveries exceeded the total loss and the volatilization figure on No. 4 cupel was more than twice No. 6, two cupels behind. In series B, Table IX, 15 mg. gold, 45 mg. silver, No. 3 showed a higher figure for gold volatilization than No. 6, three cupels behind it in the furnace, and on silver No. 4 showed more than No. 6. While the total losses show considerable regularity from front to back of the furnace, the ratios between volatilization and absorption are erratic. The possibility of losing gold mechanically in decanting and washing was not provided for in the volatilization tests.

Schneider attempted to determine volatilization by cupeling 450 mg. gold, 50 mg. copper, 1200 mg. silver, 4 gm. lead, recupeling the resulting bead with 4 gm. lead and repeating eight times. When properly carried out, this method might give a good approximation, but Schneider generalized as to some of his figures and assumed that no gold went into his nitric acid, although he assayed the ten cupels for gold and silver absorbed.

I have done considerable work in trying to arrive at a figure on volatilization of gold in cupellation by the examination of the products of the operation. No determination of volatilization depending in any way on the weight alone of a bead or cornet can be convincing. The normal bead is never pure precious metal. Cornets always carry silver and copper and often lead. Gold goes into the parting acid and into the cupel. After actually determining these items and balancing the results we get a figure, which might be called volatilization, but it is a difference figure and bears the errors, both plus and minus, of much analytical work; therefore it must be handled with judgment. Sometimes the figure indicates an appreciable loss; sometimes it gives practically a balance, indicating no loss; a gain also has been shown.

All these points are shown in one test. The San Francisco mint prepared an ingot of standard gold, 900 gold 100 copper, for test purposes. It was extremely close to 900 fine in gold. Three sets of nine assays, three rows of three, were run at three temperatures, the regular heat for this work, very low, and very high. The regular heat showed a slight loss of gold. The very low heat showed a gain. The very high heat, which should have favored volatilization of the gold, gave practically a balance, the figures showing a loss of only 0.005 mg. for the set. The figures are as follows:

From these figures it is evident that there is no necessity for running coin gold, and by fair inference fine gold, at such a heat as to favor the volatilising of gold during cupellation. The high loss of weight of the cornets at the very high temperature could not be tolerated as a regular practice and such a temperature would not be used regularly.

Volatility of Silver

Silver is far more volatile than gold. Gmelin states that it boils by the burning mirror when it rises in white fumes and that it volatilizes in an open crucible at incipient white heat, but not if covered with charcoal dust. In a cathode-ray vacuum, Krafft and Bergfeld observed volatilization at 680°. Johnston gives its melting point as 960° and its boiling point as 2090°. He calculates that it has a vapor pressure of 10-3 mm. at 920° and 1 mm. at 1320°. In his atomic-weight work, Stas distilled silver to purify it.

Richards proposed to separate silver from blowpipe beads by volatilization. Over 95 per cent, of the silver was to be removed by heating before a sharp-pointed oxidizing flame to 1100°-1200° estimated and the balance removed at about 1500° estimated at which temperature the gold begins to volatilize. In a subsequent paper, on measuring buttons, he said “It is difficult to drive off all the silver as the last 5 or 10 per cent, volatilize slowly and probably also take a little gold with them,” furnishing an excellent illustration of the drag of a volatilizing metal upon a less volatile associate as well as the stabilizing effect of a less volatile metal upon a volatilizing one.

In a little-known and short-lived journal, Seamon and Parker describe extensive tests on cupellation. As a preliminary, they made many tests on beads of silver 999 fine, of various weights up to 80 mg. for various times up to 4 hr., at various temperatures 900° to 1600°, by simply heating them in cupels. At the conclusion of the heating all the cupels showed a brownish spot below the beads. Several cupels were tested qualitatively for silver, which was always found. They proposed to make quantitative determinations of the silver absorbed, but I have not found such results published. The results are somewhat irregular. In one set of eight, on beads 5.09 to 22.50 mg. at 900° to 1000° for ½ hr., each one showed a loss of 0.02 to 0.07 mg., while another set of six, 20 to 80.06 mg. at the same temperature for 1 hr. showed no losses. They conclude that at 1000° “the loss appears to be mainly due to the oxidation of the silver and absorption of the oxide by the cupel,” but at high temperatures volatilization may become important. Under their conditions 960° to 1020° barely gave feathers in cupellation.

I have done much work in checking up and extending this line of tests. Seamon and Parker appear to consider 999 silver as entirely satisfactory, but it is far from being pure and the impurity might influence the result. I used five grades of silver: proof of the highest purity, parting silver of high grade but not up to proof, ordinary silver reduced from the chloride by zinc, 50 gm. parting silver melted with 50 mg. lead, 50 gm. parting silver melted with 50 mg. copper. Under small beads it was only with the copper sample that the stain was plainly visible. With large beads four of the silvers stained the cupel and sometimes adhered closely to it; parting silver was not tested on large beads. Both large and small beads often showed a white coating. The figures showed irregularities and were larger than Seamon and Parker found, but I was forced to the conclusion that it was futile to expect to obtain exact or even comparative figures and discontinued the tests.

Exposure of molten silver to the air leads to a mixed result. There is oxidation of the silver and maybe volatilization. The oxidation products may be absorbed by the support in part only and part may remain in or on the metal. The loss shown by weighing the silver before and after such exposure cannot lead to definite results on any point.

Relations of Silver and Oxygen

Many and most contradictory statements have been made regarding the relations of silver and oxygen, both chemical and physical. The subject is greatly in need of clarification. It is apparent that they unite in various proportions. While Ag2O loses oxygen at very low temperatures and is said to be entirely decomposed below 350°, pure silver prepared in the ordinary way gave off oxygen at 400° to 500° for 6 hr. at the rate of 57 cc. for a kilo of silver. Silver and oxygen certainly combine at much higher temperatures and separate on lowering the heat. Watts says there is an oxide of silver volatile at high temperatures. Possibly this is based upon Plattner, whose statements are questioned by Percy and were further investigated by Christy. Wartenberg found silver appreciably more volatile in oxygen than in nitrogen and attributes this to the formation of silver oxide at high temperature, which may then possibly exist as a gas owing to its high heat of vaporization. Various writers have stated that sublimed silver often contains oxide of silver.

Troost and Hautfeuille say protoxide of silver exists in the very hot gases. Gmelin says “Silver at a white heat decomposes aqueous vapor taking up oxygen,” but on the next page says “silver does not oxidize in dry or moist air at any temperature below its boiling point.” He also says silver “is less oxidized than platinum by fusion in the air.”

T. K. Rose says “even platinum was much more readily oxidized than silver, ” but in a previous paper on refining by blowing air through molten bullion and fluxing the oxides produced he brought bullion 354 fine in gold and 621 fine in silver up to 885 gold and 103 silver. In comparing his process with cupellation, he says “later when the percentage of silver is high, it is freely oxidized in both processes, and the oxidation is at its maximum when the silver is practically pure.” He says the “whole trend” of his work “is to show that it is silver that carries oxygen to base metals” and does not consider base metals oxygen carriers to silver. In this he appears to be dominated by theoretical considerations regarding normal oxides, Ag2O and PbO for instance, and to overlook the possible interaction of other oxides.

The well known spitting of silver in solidifying has been generally assigned to the evolution of oxygen, but it has also been ascribed to the compression of the solid crust on the molten interior. Having cooled large buttons, 7 to 10 gm., with extreme slowness and having obtained projections in the general form of a large irregular, inverted, hollow cone, in some instances only one cone on a button, I am satisfied that compression contributes to the spitting of solidifying silver. I have also produced spitting by the sudden cooling of a bead by contact with a mass of cold iron. In this connection it is certainly significant that silver melts and solidifies at a lower temperature in oxygen than in a reducing atmosphere. Aside then from the natural cooling and solidifying of the surface first, the interior retains a lower melting point and remains liquid longer than the exterior, which has lost oxygen with the consequent raising of the melting point. The non-spitting of silver containing other metals may be due to changed differentials in the melting points.

Volatility of Lead

Johnston gives the boiling point of lead as 1640° and the vapor pressure as 1 mm. at 960°. Kraft and Bergfeld give 335° as their lowest observation of volatilization of lead in a cathode ray vacuum. Schuller and Kahlbaum, Roth and Siedler distilled lead in vacuo and the latter obtained a crystalline product. Roberts showed that with proper care there was no appreciable volatilization of lead in melting assay samples of base bullion. From the distillation of alloys, Moissan and Watanabe announced that the order of distillation is lead, silver, copper, tin. The litharge cloud that is given off during cupellation is very deceptive. It makes an impressive showing in the muffle but the actual weight of oxide of lead is slight.

More than 50 years ago, Mitchell announced that in the ore-assay cupellation “not more than 2 to 3 per cent, of lead is volatilized;” I have abundantly verified this statement. By weighing the cupels before and after use, I have found that it is quite possible to run an ordinary 20 gm. button with the volatilization of about 0.5 gm. of lead. This, however, involves some risk of freezing and requires much attention. At only a slightly higher temperature, with an increase of the volatilization to 0.75 gm., there is no risk of freezing and the operator may simultaneously look after other things. When the volatilization rises to 1 gm. there is but slight feathering unless special condensing arrangements are used. In one test, three cupels of the front row showed less than 2.75 per cent, volatilized and the fourth froze, while the second row of four showed from 5.5— to 6— per cent, volatilized and there were few feathers. With larger buttons, 50 to 60 gm., seven cupels showed less than 2 per cent, volatilized with a minimum of 1.4 per cent. In the cupellation of bullion, where necessarily a higher temperature is used, the volatilized lead should not exceed 10 per cent.

Most contradictory statements have been made regarding the volatility of oxide of lead. The assayers’ daily experiences disprove the statements of various textbooks as to its volatilizing only at a high temperature. The books appear often to be speaking of boiling rather than simple volatilization. However, the boiling point, 870°, given by Mott is palpably much too low. Doeltz and Grauman say volatilization was slight at 800°, increased rapidly to 900°, when it became liquid, not so rapidly to 950° and was not tested above 1000°. By changing the ratio of surface to mass, they varied the loss from 8.8 to 0.04 per cent, at 900°.

Feathering as an Indication of Temperature Conditions

Feathering is not always a sure indication of the temperature conditions of the bead. With high bead temperature and an abundant air supply the cupel may show good feathers, while the same bead temperature with limited air supply would not. Lack of feathers, particularly on inside cupels, may be due to lack of cooling conditions above the cupel, instead of high bead temperature. In two well-feathered cupels, Hillebrand and Allen found 0.05 mg. loss of gold on 10.67 mg. cupeled, but only 0.03 mg. on 15.56 mg. The cupels immediately behind these showed 0.06 and 0.17 mg. losses, respectively. Lack of feathers may also be due to a leaky muffle.

Effect of Other Metals on Volatilizing of Gold and Silver

When true volatilization occurs in quantity, the volatilizing metal exerts a drag upon its less volatile associates, but the real data upon the subject indicate that in the vast majority of cases this drag is but slight. Bodemann and Kerl say silver is often disposed to volatilization by arsenic, antimony, zinc and lead.

T. K. Rose says that the presence of zinc, arsenic, antimony, and other volatile metals is believed to greatly increase the volatilization of gold, but gives no data. Hellot says an alloy of one part gold and seven parts zinc volatilizes completely at a high heat, but Friedrich investigated the subject and reported fifteen tests, from which he concluded that the loss was mainly due to mechanical action and it is only with rapid volatilization of zinc that the gold volatilization becomes appreciable. Up to 1500°, the loss was so slight as to warrant the conclusion that zinc has no influence in promoting the volatilization of gold below this temperature. I have shown that in distilling the zinc out of Parkes crusts, very little silver goes over in proper work and that volatilizing lead carries but slight amounts of silver. I have also shown that irregular assay results may be due to zinc carrying the precious metals to the surface of the cupel. It may well be that arsenic and antimony cause irregular assays in the same way rather than by volatilization. I have shown, too, that the roasting of rich arsenide ores at a low heat produced arsenious oxide practically free from silver. In some recent tests we have obtained much more concordant results than those reported on arsenical bullion. The assays were run in sets of six, two rows of three cupels with proofs in the center; two sets were run on 5 gm. lead and one on 8 gm. on each bullion. The results were:

The arsenic in No. 1 was high but unknown, in No. 2 it was about 10 fine with, approximately, 90 each of silver and copper.

It is often claimed that tellurium promotes volatilization in cupellation, but T. K. Rose carried on an extensive series of test under many conditions and in only one was gold found in the condensed tellurium. Holloway and Pearse say “loss by direct volatilization is very slight under proper conditions of working” and on cupeling beads containing tellurium with proofs without tellurium in side by side cupels some tellurium beads lost less than the proofs.

In retorting gold amalgam, Rose states that about 1 gr. of gold goes over per pound of mercury. Although outside the scope of this paper it may be mentioned that the recent volumes of Transactions of the American Electrochemical Society contain much interesting information regarding the action of metals, alone or mixed, at the very high temperature of the electric arc; and in 1907, Watts reviewed the work of Moissan.

Determination of Silver by Cupellation

The method I used in following up the course of the gold in a bullion assay is not applicable to silver. The determination of silver by cupellation is quite unreliable and with large amounts the absorption of silver by the cupel is most erratic. On the other hand the determination of the base metals left in small beads is unsatisfactory unless we have an exceedingly large number of identical beads. In another connection, I have shown that the straight assay of a silver bullion showed 987 fine, which corrected for the cupel absorption became 998.5 fine, while cupellation with proof showed 999.3 fine. Liddell gives some very interesting tests on cupellation, but he determined the “fineness” of his beads, and this was subject to the errors of the original assay. Both these tests lack conclusiveness.

L. Campredon’s book says that all the silver lost is absorbed by the cupel, but the tests lack the determination of base metals in the beads. G. Campredon made three tests in duplicate at a high temperature, one set using five times as much copper as silver, but he made no examination of his buttons. His results varied from nothing to 2.10 per cent, on 100 mg. silver. He made two assays of his cupels; one for beads on the surface, which varied from 0.40 to 1.50 per cent., and one for absorption. Various other indirect tests have been made, but all lack important data. Some tests show a gain even when the recoveries of silver were not complete.

It remains then that the only way to determine the volatilization of silver in a cupellation is to collect the litharge volatilized and assay it for silver. Manifestly this is also the only real way to ascertain the volatilization of gold under assaying conditions. I started out to do this without interfering with the normal conditions of cupellation, but found it impossible to do so in an ordinary muffle in constant use for regular work. Various arrangements of temporary baffle plates proved inefficient condensers and in many cases the condensed litharge entered into combination with the plate material. While it is possible by the use of special cooling arrangements to grow a large crop of feathers on the cupel, once the fume got out of the cupel it was practically impossible to stop it. Of course with a small muffle provided with special condensing flues, such as a Cottrell tube, and running a large number of identical assays a good approximation may be obtained, but I did not have these conditions. Again, with the low rate of volatilization of the lead in a proper cupellation of an ordinary 20 to 30 gm. button, the best feather-growing arrangements will yield only insignificant amounts of litharge. Feathers are most deceptive; they bulk large but weigh light.

After various tests I made a very deep cupel with straight sides. In the early tests the condensation of litharge was erratic, varying from 0.2 to 1 gm. and considerable experimenting was required to develop a satisfactory procedure. With 100 gm. lead and 1 gm. silver 0.605 gm. litharge condensed, yielding 0.055 mg. silver; with 5 gm. silver 0.735 gm. litharge gave 0.26 mg. silver; with 133 gm. lead and 6 gm. silver, 0.97 gm. litharge gave 0.26 gm. silver; with 140 gm. lead, 2.11 gm. copper, 8.33 gm. silver, 0.465 gm. litharge gave 0.27 mg. silver. By using large amounts of lead, placing the cupel close to the front of the muffle and filling in around it so as to force all the draft over the cupel, and by protecting the cupel from radiation from the muffle top, I was able to condense a gram or more of litharge for assaying. This is none too much, even when cupeling very rich lead.

A most serious objection to this method is that in time the surface of the bullion sinks out of sight. As one cannot see the end of the cupellation careful watching and much judgment is required in turning off the gas. In ten tests, six buttons weighed more than the silver taken, the maximum being 3.4 gm. on 1 gm. taken and the minimum 0.13 on 5 gm. taken. Of the four losses, the maximum was 0.46 on 5.81 gm. and the minimum 0.04 on 5.045 gm.

In general, 140 gm. of lead were taken, the smallest amount of gold used alone was 140 mg., which would correspond to 20 oz. per ton in a 20- gm. button from an assay-ton fusion. The largest was 1.4 gm., or 200 oz. per ton. The following results were obtained. In the last test only 105 gm. of the lead were oxidized.

A sample of the celebrated Mercur bullion, about 884 fine in gold and less than 8 fine in silver, 1.6 gm., yielded 1.185 gm. litharge containing 0.03 — mg. gold. Samples of miscellaneous bullion, most of them presenting unusual difficulties in ordinary assay work, yielded the following results:

In the last test, a scum appeared early and crystals were slow in forming; probably some of the scum adhered to the feathers recovered.

With 1 per cent., 1.4 gm., gold, and 3 per cent., 4.2 gm., silver, 1.27 gm. feathers showed a very slight trace of gold and 0.21 mg. silver. In one test the lead carried about 1 per cent, arsenic, but the cupel cracked.

With 0.5 per cent, arsenic and 1.4 gm. gold, 1 gm. litharge was collected and the gold obtained was estimated at 0.002 mg.

Three grades of base bullion were tested—good, bad, and very bad for assaying. Using 159.5 gm. of the first gave a button weighing 1.24 gm., 0.762 gm. litharge was collected, which gave a bead of 0.08 mg. The second would not cupel direct, but 20.03 gm. were run with 152.45 gm. proof lead and gave a button weighing 3.475 gm., 0.975 gm. litharge was collected giving a bead of 0.08 mg. In a second test, the figures were bullion 20.62 gm., lead 142.45 gm., button 3.422 gm., litharge 1.322 gm., bead 0.28 mg. Probably the high litharge and bead were due to the early formation of an unabsorbed scum on the cupel, upon which later feathers formed, particles of which adhered to the litharge recovered. The third sample yielded the highest recovery of litharge of any test made, 1.50 gm., due perhaps to the same causes. The figures were, bullion 20.79 gm., lead 139.06 gm., button 3.25 gm., litharge 1.50 gm., bead 0.30 mg. Extreme tests were made on three grades of silver—proof, 900 silver 100 copper, 750 silver 250 copper. Starting with 2 and 3 per cent, actual silver on the 140-gm. lead buttons, the results were:

Discussion

J. W. Richards, Bethlehem, Pa.—The volatilization of the metals in assaying is a special case of differential volatilization and is dependent on the different vapor tensions of the different metals. Combined with this volatilization, however, there is usually the formation of mist, which must also be taken into account and, I think, explains some of the inconsistencies pointed out in the paper.

For instance, Doctor Dewey has shown that more silver goes off with zinc in the early stages of a distillation, in which zinc-silver alloy is distilled. According to the vapor tensions of the zinc and the silver, more should be going over in the later stages. That discrepancy is to be explained by the fact that in the early stages of the distillation there is probably a more rapid formation of zinc vapor, and that it carries away, as it escapes in the form of bubbles, a larger amount of the alloy in the form of mist. Now, the alloy that is being distilled carries alarger percentage of silver than the distillate, or the vapor that is going off; for instance, the vapor going off may contain 2 per cent, of silver while the alloy itself may contain 20 per cent. Therefore, if some of the alloy is carried over by the rapid boiling off of the zinc, that alloy increases the richness, and apparently will increase the amount of silver that goes off. The amount of the volatile metal that goes off as true vapor can be determined exactly from the vapor tensions of the metals. But the amount that goes off as mist or fog, representing alloy carried off mechanically, may contain far different proportions of the precious metal than the vapor.

As far as I know, it is impossible to calculate just how much of a material will be carried off as fog or mist. The amount depends on the rapidity with which the material is boiled; it is like the water that goes over with steam. Perhaps if we analyzed all the conditions and got all the variables, we might commence to find the conditions that regulate the amount of mist or fog formed, but as yet our metallurgical data and physical analyses have not progressed that far.

On page 604, Doctor Dewey makes a statement that needs a little explanation: “There is absolutely no connection between melting point, vapor pressure, and boiling point.” I admit that there is no connection between the melting point and the normal boiling points of the metals; some with a low melting point have a high boiling point and others with a high melting point have a low boiling point; but by boiling point, we mean boiling temperature of 760 mm. pressure, the normal boiling point. The statement that there is no connection between the vapor pressure and the boiling point needs a little amplification. That there is no connection between the melting point of the metal and the normal boiling point, I will admit, but the introduction of the term “vapor pressure” may be misleading.

The next line contains the statement that various metals show vapor pressure below their melting point. I wish to amplify that by stating that all metals show vapor pressure below their melting points. In fact, the question is not whether the metal shows vapor pressure, but simply whether it has been determined. I would even say that at ordinary temperature all metals have vapor pressures. The process of sherardizing by means of zinc dust depends on the vapor pressure of zinc below its melting point.

F. P. Dewey.—I take complete exception to Professor Richards’ closing statement; I do not admit that there is vapor pressure of all metals at all temperatures.