Table of Contents
In looking about for some means of determining the relative affinities of the metals for cyanide solutions, I long ago came to the conclusion that the determination of the relative electromotive forces of the metals in solutions of different strengths was the simplest, readiest, and most certain that could be selected. For, properly considered, it shows the actual tendency of the metal to go into solution. My first experiments were made in this direction. I made at that time a large number of preliminary determinations, the results of which were presented in a lecture given, before the California Academy of Sciences, in San Francisco. At that time, the curves shown in Fig. 5 were projected on the screen by a stereopticon before an audience of 300 persons.
The results of these experiments have filled me with constant surprise, when I have noticed what apparently slight causes were capable of making great changes in the electromotive force of the same metal. The great delicacy of the method proved to be the chief source of difficulty in its application, while at the same time it reported faithfully the facts as they exist in nature.
Two methods have been used in these determinations, the first being what I have, for brevity, called the “ Deflection ” method, and the other the “ Compensation ” or “ Zero ” method of Poggendorf.
In each case an electrolytic cell is constructed with two electrodes, each immersed in a separate solution. One, consisting of the metal to be tested, was held in the points of a platinum-tipped pair of forceps, electrically connected with a galvanometer, and was immersed in a vessel containing the cyanide solution of the given strength. The other was in all cases the “normal” electrode of Prof. Ostwald, consisting of mercury, electrically connected with the galvanometer by means of a glass-coated platinum wire. The surface of the mercury is covered with a layer of mercurous chloride, a couple of inches thick; and a solution of chloride of potassium of one gramme-molecule M/1 (in this case also a normal solution).
The two vessels containing the electrodes are connected, as shown in Fig. 3, by means of the tube C and the siphon D, the latter being filled with M/1 KCl solution, like that in the normal electrode. I have added a small tube E, ordinarily closed with a cork, for the purpose of displacing at intervals the solution in the siphon D with fresh KCl solution, to avoid the diffusion of the cyanide solution through the latter hack into the normal electrode. For the same reason the position of the normal electrode is ordinarily a little higher than that shown in the figure, so that any accidental action of the siphon shall be rather away from the normal electrode than into it.
The purpose of the normal electrode of Ostwald is to have a non-polarizable electrode in a solution of known strength and electromotive force. This is fixed at —0.560 volts. That is, in the case of the normal electrode, the quicksilver ions tend to precipitate themselves on the surface of the mercury, and the solution is therefore negative to the metal by 0.560 volts. That is, the positive current tends to flow through the solution to the mercury, which becomes positively electrified, while the solution itself becomes negatively electrified.
Now, if we neglect the slight electromotive force due to the contact of the two solutions, the resulting electromotive force of the combined cell is the algebraic sum of the electromotive forces active at the two electrodes. Hence, if we subtract 0.560 from the EME of the cell, we have the EMF of the metal under consideration. The algebraic sign indicates the direction of the positive current.
Procedure of the Deflection Method
This method is much the most convenient for such investigations, particularly in the first roughing-out of a large amount of material. With proper precautions, it gives results not less reliable than those of the zero method; and it has the great advantage over the latter that the rapid changes of electromotive force may be followed almost as they occur.
The method is illustrated in Fig. 4. B is the cell containing the cyanide solution and the metal M to be tested; NE is Ostwald’s normal electrode; R is a resistance which varied in the tests from 30,000 to 200,000 ohms; G is a Wiedemann reflecting galvanometer; K, a make-and break-circuit key; and C, a commutator.
The galvanometer was calibrated by replacing the cells B and NE with a Latimer-Clark cell, prepared according to the directions of Ostwald, and noting the deflection produced by its voltage through the given resistance of 30,000 to 200,000 ohms. The voltage was taken as EMF = 1.438 — 0.001 x (t° — 15° C.) volts.
Most of the concentrations of potassium cyanide were M/1 (one gramme-molecule, 65 grammes per liter, or 6.5 per cent.), or fractional multiples of this in tenths. Thus the series used, was frequently M/1, M/10, M/100, M/1000, M/10,000, M/100,000, M/1000,000.
As there is no little difference in the methods of notation in use, and much resulting confusion, the following methods of notation will always be used in this paper: We shall follow the motion of the positive ions through the solution, and the mode of notation will depend entirely on that. When the motion of the positive ion in the solution is from the metal to the solution, the metal is said to be electropositive, because it gives up positive ions to the solution and causes the solution to become positively electrified, itself becoming at the same time negatively electrified. Such, for instance, is the case of zinc in a solution of zinc sulphate. When, on the other hand, the solution gives up positive ions to the metal immersed in it, as is the case with copper in a solution of copper sulphate, the metal is said to be electronegative, for it causes the solution in which it is immersed to become electronegative, itself, at the same time, becoming positively electrified. The quicksilver in Ostwald’s normal electrode is another example. The + or — sign, then, here indicates the direction of ionic motion, and simply shows whether the given positive ions tend to flow away from the metal into the solution or towards the metal from the solution. That is, whether the “ solution-pressure ” of the metal is greater or less than the “ osmotic pressure ” of the ions in solution.
Now, when an electropositive and an electronegative metal are coupled, the direction of flow of the ions of both through the solution is the same, and the electromotive force of the combination is the arithmetical, sum of those of the ingredients. When two electropositive or two electronegative metals are coupled, the ions tend to flow through the solution in opposite directions; hence, the electromotive force of the combination is equal to the arithmetical difference between the separate electromotive forces, the direction of motion, and hence the sign, being that of the greater.
In combinations in which the Ostwald normal electrode is one member, we know the amount and direction of one electromotive force; and hence, when we measure that of the combination, it is easy to calculate that of the other (neglecting the slight electromotive force due to the contact of the solutions).
Thus, if, against the normal electrode, aluminum in a M/1 solution of KCy gives an EMF = + 1.55 volts,—that is, if the current flows from the aluminum to the mercury, the same as in the case of mercury,—it follows that the EMF of the aluminum in M/1 KCy will be + 1.55 — 0.560 = + 0.99 volts.
Again, if a strip of amalgamated zinc under similar circumstances gives a voltage of + 1.49 volts, the EMF of amalgamated zinc in a M/1 KCy solution will be + 1.49 — 0.560 = + 0.93 volts.
In making the determinations, it must be evident from the formula that, if there are few ions of the given metal present in the solution at the start, the introduction of a very few more will make great changes in the value of the EMF.
For in log. P/p it must be evident that, as P is constant (for a given temperature), the value will depend entirely on p; and the smaller p is, the greater will be the effect due to slight changes in p. Hence, it will be impossible to get constant values for the EMF, unless the value of p is nearly constant; that is, when the solution is saturated with ions at the given temperature. That is the case with the normal electrode, where the mercury lies in a saturated solution of mercurous chloride. The mercury is thus in equilibrium with its ions, and a constant EMF results.
To get perfectly constant results with cyanide solutions, it would be necessary to have the solution saturated with the cyanide of the metal in question. But while this would give us a very satisfactory electromotive series, it would not give us a measure of the action of the unsaturated cyanide solution, just as it acts on the ores. We must, therefore, be content with results that are not entirely concordant, and take the best of a large number of determinations.
The strips used were always freshly burnished with sandpaper, cooled, and touched to a grounded platinum wire to discharge any electricity with which they might have been charged in burnishing.
Preliminary Results with the Deflection Method
The following preliminary results were obtained:
with some of the common metals and minerals. The metals were good commercial articles, such as are in use in the arts, except in the case of gold, silver and quicksilver, which were chemically pure. In the case of some of the minerals, such as zinc blende, stibnite, etc., the electrical resistance was probably so high in comparison with that of the intercalated resistance that the results may be somewhat low.
Nevertheless, they give at once some important relations which must exist whenever the cyanide process is applied in the treatment of ores.
The electromotive forces of the metals and minerals marked with an asterisk in the above table have been plotted in Fig. 5. The Y axis shows the potential in volts, the X axis the concentration in gramme-molecules and also in percentage of KCy.
It will be noticed that in most cases the curves approximate quite closely to the logarithmic curve which theory would give (see Fig. 2), supposing the osmotic pressure of the metallic ions present to be inversely proportional to the concentration of the free potassium cyanide present; but they have different origins.
It will be noticed that the electromotive force of commercial sheet-zinc is increased by amalgamation, probably, by reducing local action with some of its impurities, by which some of the current produced is short-circuited. In all the other experiments, amalgamation reduces the electromotive force of the combination.
With some substances, particularly aluminum, copper, iron, platinum and gas-carbon, it was very difficult to get concordant results; with aluminum and copper this seemed to be due to a tendency to form an insoluble film on the surface of the metal, which put a stop to further action. With copper and iron it was also possibly due to a tendency of the metals to a change of valency, which is accompanied by a change in the electrical state. With platinum and gas-carbon, it was not improbably due to a varying content of absorbed gas.
In testing the minerals, it was in all cases difficult to get a complete electrical contact between the tips of the platinum forceps and the rough surface of the mineral fragment, so that the results are only provisional, particularly as the resistance in some of these cases was very high. Nevertheless, the results are very interesting. They show, for instance, that not all copper minerals have a strong action on the current. Pure chalcopyrite, for instance, has hardly more action than pure pyrite, while bornite and copper-glance have a very decided tendency to go into solution. Cuprite is also apparently very little acted on, though this may be due to its high resistance rather than to a lack of tendency to dissolve. The soluble salts and minerals of copper could not be tested in this manner, owing to their non-conductivity.
It is plain, however, that pure chalcopyrite, galena, argentite, magnetopyrite, fahlore, arsenopyrite, blende, boulangerite, bournonite, ruby silver-ore, stephanite and stibnite, when free from their oxidation-products, are apparently very little acted on by cyanide solutions.
It is also plain that a particle of metallic gold, in contact with a particle of pyrite, forms a galvanic couple in M/1 or 6.5 per cent. KCy solution, equal to + 0.65 volts: in M/10 or 0.65 per cent. KCy solution, + 0.65 volts, and in M/100 or 0.065 per cent. KCy solution, + 0.57 volts. With zinc under the same circumstances (if we take for the M/1 KCy solution the figures for amalgamated zinc), taking the zinc as the more electropositive metal, and subtracting the potential of gold, we have differences of + 0.56 volts, + 0.54 volts, and + 0.50 volts. In short, these figures would measure the tendency of the zinc to dissolve, or of the gold to precipitate in KCy solutions of these strengths.
According to these figures, the precipitating power of the zinc seems to hold up quite well for the dilute solutions. The actual failure to precipitate the gold, sometimes met with in dilute solutions, is no doubt due to films of cyanide or hydrate of zinc, which form incrustations on the surface of the zinc and thus prevent contact. The fact that the use of a small amount of fresh cyanide or of caustic potash in the zinc-boxes starts precipitation again, seems to favor this explanation.
This method is shown in outline in Fig. 6. NE is the Ostwald normal electrode. B is the cell containing the cyanide solution in which, as before, is immersed the metal M to be tested. At G is a galvanometer. At R is a resistance, graduated, in my experiments, into 10,000 parts. A storage-battery of two volts and the combination-cell NE-B are so connected that their positive poles are both connected at the same end of the resistance R. The negative pole of the storage-battery is attached to the other end of the resistance R, so that the whole current of the storage-battery discharges constantly through R. The latter should be great enough to avoid heating, and to maintain a constant potential between the ends of R. The other terminal of the combination (the negative pole) is then moved along the resistance R till some distance, a, is reached at which the EMF force of NE-B is exactly balanced by the EMF force of the storage-battery for that fraction of R represented by a. In this case there is no deflection of the galvanometer; at other points the galvanometer will be deflected either to the right or left, according as too much or too little EMF is used to balance NE-B. The EMF of the storage-battery, is, of course, first calibrated by comparing it with a standard Latimer-Clark cell, placed where NE-B is.
This method of determining the EMF of a cell is deservedly considered one of the most reliable. With non-polarizing cells, it certainly leaves nothing to be desired. But, in investigations of this kind with cells that are easily polarized, accurate results are obtained only by a long number of very tedious approximations, which render the work almost interminable. For it is, of course, impossible to hit the right balance at first; and, if the connection is made at any point except the right one, the
metallic electrode will receive either a positive or negative charge from the storage-battery, and a true reading will be thus made impossible. It is necessary to change the entire solution in B, put in new electrodes at M, drive out the diffused cyanide solution from NE, and so on, till these operations have been repeated perhaps a dozen times. If this is not done, the results are very unreliable. With the deflection method, on the other hand, the observations may be made very rapidly, and though there is a tendency for the readings to be a little low unless they are quickly made, still, with a high intercalated resistance, and a delicate reflecting galvanometer, this method seems to be reliable for these quickly polarizing electrodes.
As I have already stated, and as was first pointed out by Ostwald, strictly concordant results are possible only when the electrode is surrounded with a medium already saturated with its ions.
I had intended to verify the results in Table I. with the zero method before publication, but although I had all the apparatus set up for over two years, ready to begin at any time, I was prevented by the constant pressure of routine-work from touching it, till shortly before the time set for the San Francisco meeting of the Institute. Meantime Prof. A. von Oettingen, professor of physics in the University of Leipzig, read a very valuable paper on this subject before the Chemical and Metallurgical Society of South Africa. In this paper he gives the results of a large number of determinations which he made of the electromotive force of metals in cyanide solutions by means of Poggendorfs compensation method, or, as I shall call it for brevity, the zero method.
Professor von Oettingen’s results are given in Table II.
The above results were all obtained by the Poggendorff compensation or zero method, the Lippman capillary electrometer being used as an indicator instead of a galvanometer.
Prof, von Oettingen says of these results:
“The two figures in each column refer to the first and last observations on each metal, the intermediate values being omitted. The time occupied by the change is very variable ; Cu, for instance, took an hour. When no changes occur, this is indicated by a constant. The changes of potential are not always in the same direction ; sometimes decreasing, sometimes increasing. But the direction of the changes in any given metal is always the same.”
It will be noticed on examining the table that this last sentence is not correct (unless there should be a typographical error in his table). For gold, silver, copper, cobalt, ferric oxide and lead peroxide, the highest value for the same metal is sometimes the first and sometimes the second value. In the case of copper in M/100 KCy the results jump from —0.212 volts to + 0.380 volts—a difference of 0.592 volts. I shall speak of the probable cause of these differences later.
In order to make more clear the meaning of Prof, von Oettingen’s results, I have plotted them in Fig. 7 as mine are plotted in Fig. 5. In the figures, x is made to mark the molecular concentration, M/1, M/10, M/100, M/1000 ; the y axis shows the potential in volts. The designation Zinc 1 means that this was
the first value obtained with zinc, the designation Zinc 2, the final value, etc. It will be noticed that sometimes the first value is higher than the second and sometimes vice versa; but the results are not consistent throughout, sometimes crossing each other.
The first curves of each metal, except mercury, approximately follow the logarithmic law (on the assumption that the number of metal ions is inversely proportional to the potassium cyanide concentration). Evidently the curves will cross the X axis at different points, and not usually at a molecular concentration M = 1, unless it should accidentally happen that P/p = 1 for M = 1. The second curves of zinc, copper, gold and silver, also approximately follow it. But the second curves of mercury, cobalt, nickel and iron depart considerably from it. It is possible that these departures are due to polarization effects, as already explained. The irregularities are much more marked than with the deflection-method. With that method, provided a sufficiently large resistance is used, the first deflection is the greatest, and is taken as the reading nearest to the truth. The deflection then gradually falls (often
quite rapidly, if there is a formation of gas on the face of the electrode); but the electromotive force never rises unless the first effect of the current is to produce a film of gas or insoluble cyanide which puts a stop to the current, either by setting up an opposing EMF or by preventing or reducing contact by its resistance. In this case, shaking the solution or jarring the
electrode usually gives an increase of the EMF by destroying the film in part; but, if the metallic surface is untarnished to begin with, the EMF rarely rises again to its first value.
New Method of Plotting Results
The method of plotting results hitherto used, while it shows very well the near approach of the curve to the true logarithmic curve, has the disadvantage that only three or four values for the tenth ratio can be plotted. If, however, instead of making x = P/p, as we have done, we let x= log. P/p, and plot the curve y = 0.058 log. P/p volts, the curve becomes a straight line passing through the origin at 0. For x = 0, y = 0.
This curve is plotted in Fig. 8 for values of x = log. P/p from + 13 to — 12, which gives voltages from + 0.755 to — 0.696, and the table shows values from x = log. P/p = minus infinity to 40. It shows what an enormous change in the value P/p is necessary to produce a very moderate change in the voltage. Thus, to produce a change of 2.32 volts, a change in the ratio P/p = 10 40 (or ten to the fortieth power) is necessary.
In our experiments, of course, we do not know the value of P/p, but as a first approximation we may assume it inversely proportional to the molecular concentration M x 10 n.
On the axis of x is plotted the logarithm of the molecular concentration expressed in the powers of 10. Thus: x = log. M=log. 10 ±n. The y axis gives the EMF in volts. For comparison the theoretic formula of Hernst is also given.
If we plot Prof, von Oettingen’s results, as in Fig. 9, on this plan, they become at once more intelligible. We see at once that all the curves do not remain straight lines. The zinc follows along very nearly in the theoretic straight line. The copper starts well, but soon falls quite rapidly, due probably to increasing dissociations. The gold and silver approximate fairly well, also; but the rest depart from it considerably.
Comparison of the Deflection and Zero Methods
Since the results of Prof. von Oettingen were published, I have thought best to try the zero method as well as the deflection method, and to compare the results so far obtained with each other. I have also decided to plot the results by the same method as shown in Fig. 8, as it enables us to compare the results over a wider range of dilution than the former method of tabulation would cover.
After the foregoing description and discussion of the various methods employed in this investigation, the reader will be able to study intelligently the tabulated results of the tests herein-after stated.
Evaluating Results of Laboratory Experiments
The following tables show, for the several metals tested, my own results, obtained at different times and by different methods, as well as those of Prof, von Oettingen. In every case, each observation was made independently, without regard to the ultimate result of its reduction. But the later readings are more reliable than the earlier, because a certain knack in catching the needle at its maximum position, before the voltage begins to fall, was acquired during the work. The tables give the readings as reduced from the actual observations, without attempted correction; but when any anomaly rendered the observation uncertain, this is indicated by a (?). Such was the case particularly in the readings with distilled water (M divided by ∞ ), which were very uncertain, especially for easily oxidizable metals like zinc and iron.
Experiments with Commercial Sheet-Zinc
Experiments were made at different times, with both the zero and the deflection method, on the ordinary commercial sheet-zinc, such as is actually used in making zinc-shavings for precipitating gold. The results are given in Table III., and those of Prof, von Oettingen have been introduced into the same table, for comparison.
These results are plotted together for comparison in Fig. 10. It will be evident that from M/1 to M/100, or from 6.5 to 0.065 percent., the curve nearly follows the theoretic straight line.
Curves a, f and g appear to follow it to M/1000 or 0.0065 per cent., but for more dilute solutions beyond that point the curve approximates a horizontal straight line. This, according to the Nernst theory, would mean that the number of zinc ions in such solutions remains nearly constant. In spite of all the irregularities in the curves, the point — 3 or M/1000 or 0.0065 per cent.
KCy is evidently a critical or inflection-point in the curve.
The results obtained with high dilutions of cyanide and with distilled water were very uncertain, probably because of the formation of insoluble films of oxide of zinc and occluded hydrogen, which prevented the accurate reading of the needle.
In my results with the deflection method, I have always taken the highest reliable reading as the most probable result. It was often quite difficult to make sure of the proper reading, as a slight insoluble film of cyanide of copper formed almost instantly, and this lowered the potential almost before a reading could be taken. On agitating the copper, so as to bring it into contact with fresh solution, the potential would gradually rise to a maximum, after which, on being left at rest, it would again fall off more gradually. It is possible, also, that the tendency of copper to form cupric, as well as cuprous cyanide, may in part explain the discordant results, such, for instance, as that obtained by Prof, von Oettingen with M/100 KCy solution. He says in a footnote concerning this case, “ On shaking, the potential suddenly rises from — 0.212 to + 0.380.”
The results contained in Table IV. have all been plotted in Fig. 11. The mean results of these curves show a tendency to follow the course of a straight line from M/1 down to M/1000 ; or perhaps to M/10,000; that is, from 6.5 down to 0.00065 percent., when it breaks off sharply and runs along flat again, just as the zinc-curve did.
On plotting the gold-curves, as has been done in Fig. 12, it is evident that the gold follows the logarithmic law fairly well
as far as M/100 or 0.065 per cent. KCy. A considerable fall of potential occurs, according to my experiments, between M/100 and M/1000, or 0.0065 per cent. KCy, indicating an increase of osmotic pressure, probably due to an increasing dissociation of the potassium aurous cyanide. This point seems again a critical point in the curve, which, beyond it, runs off more flatly, indicating an approach to a constant osmotic pressure of the gold ions.
Electromotive Force of Gold in KCl and KHO.
In order to bring out the effect of the potassium cyanide in reducing the osmotic pressure of the gold ions in the solution (according to the Nernst theory), I append the following experiments on the electromotive force of gold in solutions of potassium chloride and potassium hydrate. These results are given in Table VI., and are plotted in Fig. 13. It is evident that there is a very much smaller electromotive force in each of these cases. It is particularly low in the case of potassium chloride. According to the Nernst theory, the solution-pres-
sure of the gold is the same in each of these solutions; that is, the pressure with which the gold tends to go into solution is exactly the same (at a given temperature), whether the gold is immersed in either potassium cyanide, potassium chloride or potassium hydrate. But the number of gold ions in each solution, and hence the resulting osmotic pressure, is very different. According to this theory, it is least in potassium cyanide, much greater in potassium hydrate, and greatest of all in potassium chloride. Consequently, the EMF varies inversely as p, according to the ratio log. P/p.
The curves in both cases run rather flat, indicating an approach to a constant osmotic pressure for high dilutions.
The normal electrode, checked on M/1 KCl, showed -0.560, as it should do.
As a check on the foregoing results, I am able to quote the observations of an independent observer, Brandenburg. He conducted a number of experiments with mercury in various depolarizing solutions. Instead, however, of using Ostwald’s normal electrode, be used as one electrode mercury covered with sulphate of mercury (instead of the chloride used in Ostwald’s). This electrode was then connected, by means of a siphon containing a neutral salt in solution, with a vessel containing mercury covered with the various solutions to be experimented on.
The solutions he experimented on to find their ion destroying-power, or their power to form complex ions with mercury, were : potassium sulphide, potassium cyanide, potassium sulphocyanate, sodium hyposulphite, potassium ferrocyanide and potassium chloride. As he did not use the same strengths that I have found most convenient, I have had to plot his results, reduce them to zero potential, and interpolate the results for the strengths I have used. The results so obtained are compared
with the results obtained by Professor Oettingen and myself with the normal electrode. The results obtained by us for potassium cyanide are higher than Brandenburg’s, but show the curves to be of the same general nature. They are shown in Fig. 17.
In order to bring out more clearly the nature of the relations existing between EMF of the different metals, I have combined, from the plotted curves of each metal, what appear to be the most probable values for each metal. These results are contained in Table XI.
These results have been plotted in Fig. 19. These curves all show critical points at either log. M = — 2, — 3 or— 4. Most of them show the greatest amount of inflection at log. M = — 3. In fact, most of them seem to change in character at this point. According to the Nernst-Ostwald theory, this would be explained by the assumption that below say M/1000 KCy, the dissociation of the complex ion containing the metal in point is practically complete, so that the osmotic pressure p of the given metallic ions in the dilute solutions becomes practically constant below this point, so that as the ratio P/p is nearly constant, so its
logarithm, and hence the voltage, becomes also nearly constant, as is shown in the figure.
The curves for lead and iron are very remarkable; at first quite low, they maintain themselves at a higher level than either of the other metals except zinc. This is explainable on the supposition that the values of P for lead and iron are for these metals rather low, but that the values of p reach a nearly constant value sooner than for the other metals, so that the resulting curves flatten earlier.
These curves also show a number of remarkable crossings.
Copper, which starts at a voltage slightly less than that of zinc, rapidly falls off, crosses the curve of lead a little below log. M = — 2.5, and that of iron a little before log. M = — 3.5, and then remains permanently below these metals. The gold-
curve crosses the curves of mercury, silver and iron at just about log. M = — 2.5. Gold and silver both cross mercury again at about log. M = — 3.5. Gold finally crosses mercury again at a point beyond log. M = — 6, and remains permanently below it after that.
It will be observed that the metals change their sequence from that of zinc, copper, gold, silver, lead, mercury, iron, which they possess in a M/1, or 6.5 per cent. KCy solution, to the order zinc, lead, iron, copper, silver, mercury, gold, in distilled water, which is the usual electrochemical series in acid solutions quoted by Wilson except that iron is placed above lead. The determination of iron in my experiments was not entirely satisfactory, by reason, apparently, of the formation of films; and the results are probably too low. Water, also, appears to act like a weak alkali.
All the metals show a critical point somewhere between log. M = —3 and —4, at which dilution they seem to change from the voltage due to the cyanide solution to that which they ordinarily possess.
From a study of these curves there seems to be little support for the so-called “ selective affinity ” of dilute cyanide solutions for gold and silver, except in the case of copper down to log. M = —4, or 0.00065 per cent. KCy. In the case of zinc, lead, iron and mercury the strong solutions give a better relative voltage in favor of the gold than do the dilute cyanide solutions. But in the case of copper, there seems to be a distinct advantage in favor of the gold in dilute solutions down to 0.00065 per cent. Then the curves widen again. These facts will appear from the following table taken from the figure:
It should be remarked that if we had an independent method of determining the number of metallic ions in cyanide solutions, and were thus able to plot the EMF in terms of the actual ionic concentration instead of the molecular concentration, we should probably reach a more perfect agreement with the logarithmic law than in the curves here shown. Nevertheless, even as it is, a general agreement is certainly evident.