THE ANALYST. 3 THE REDUCING ACTION OF HYDROGEN. 11. The Estimation of Traces of Arsenic by the Marsh-Berzelius Method, and the ‘‘ Insensitiveness ” of Zinc. BY ALFRED C. CHAPMAN F.I.C. and H. D. LAW B,Sc. (Bend at the Meeting December 6 1905.) THE urgent need which arose a few years ago for a method capable of detecting and estimating with accuracy minute traces of arsenic caused numerous investigators to devote their attention to the solution of a problem which was soon found to be one of very considerable difficulty. It is unnecessary for us to refer to the numerous methods which were at various times suggested more especially as the modified Marsh-Berzelius process with which we are particularly concerned in this paper, is now almost universally employed. The energies of analysts were at first devoted to getting reagents of the high degree of purity necessary and as the result of much labour zinc and mineral acids were obtained which were either free from arsenic or contained only such infinitesimal traces that perfectly blank results were given in an apparatus capable of revealing with certainty & mgr.It soon however came to be recognised that freedom from traces of arsenic was not the only qualification which zinc for use in the Marsh-Berzelius flask must possess since many samples of the presumably pure metal were found to exhibit the property of preventing the reduction of more or less of the arsenic added and so were worse than useless. One of the first observations made in connection with this aspect of the subject was that the addition of a few drops of platinum chloride to the contents of the Marsh-Berzelius flask was entirely inadmissible as it had the effect of causing the ‘‘ retention ” of comparatively large quantities of arsenic.One of the authors showed that the same applied in a less marked degree to solutions of copper sulphate. These facts naturally led to an investigation of the effects of depositing traces of certain metals on the surface of ‘‘ sensitive ” zinc and it was found that whilst some metals-e.g., Au and Pt-exerted a marked retentive effect others-e.g. Ag and Cu-exhibited this property in a far less degree. At one time it was thought that the influence exerted by the added metallic impurities was in some way connected with their position in the electro-chemical series but one or two results were opposed to this view and seemed to show that the true explanation was not by any means so simple.As the experimental difficulties were very great and as the work involved the ex-penditure of more time than could be spared the matter was allowed to drop for the moment. One or two chance observations however which seemed to throw a little light on the subject induced the authors to recommence the study of these pheno-mena and it was decided to attack the problem in a systematic manner. TO thi 4 THE ANALYST. end we studied in the first place the effects produced by adding to the Marsh-Berzelius flask containing sensitive zinc solutions of various metallic salts the quantities taken varying from 2 grams to 0.1 gram.The following table in which the numbers represent thousandths of a milligram will show the results obtained : Salt added. Palladium chloride . . . Platinum chloride . . . Niccil sulphaik . . . Cobalt sulphate . Stannous chloride . . . Lead acetate . . Cadmium sulphate . . . 9 . 9 , . Quantity in Grams. 0.1 0.1 0.1 0.1 2.0 2.0 2.0 1.0 2.0 Arsenious Oxide added. 50 500 500 1,000 200 200 10 10 10 Arsenious Oxide found. None Practically none None Practically none 20 20 10 10 10 Arsenious Oxide left in Flask. 50 Practically 500 500 Practically 1,000 180 180 I n addition to the above experiments were conducted with small quantities of salts of metals incapable of being deposited on the zinc.Thus we experimented with the sulphates of potassium aluminium magnesium sodium and ammonium and found that there was either no retentive effect at all or else that this was so slight that it might well have been accounted for by traces of impurity. An inspection of the above table shows that whilst palladium platinum nickel, and cobalt all exert a more or less marked retentive effect cadmium tin and lead appeared to be entirely without such action. The ‘‘ retention ” was clearly there-fore not in any way connected with the position of the added metal in the electro-chemical series and it became necessary to find some other explanation of these apparently anomalous results. The observation made by one of us on the reducing action of the hydrogen in palladium-hydrogen had clearly shown the insufliciency of the conceptions of active or atomic and inactive or molecular hydrogen to explain various reduction phenomena and had rendered it evident that the chemical activity of the hydrogen in many of the reactions studied depended very greatly on the source from which it was obtained.Having this aspect of the question in mind, we turned our attention to the work which had been done by Caspari in this direction. The interesting observations of Thorpe Tafel Sand and Hackford, Trotman and Thomson in regard to the varying reducing efficiency of different hydrogen electrodes all supported this view and appeared to us to furnish an additional reason for an extended and more systematic study of the question. Before proceeding to give an account of our experiments we may perhaps be permitted to refer to the following theoretical considerations : If a rod of zinc be immersed in a dilute solution of a salt of copper the zinc tends to pass into the solution and drive out the copper.On the other hand a strip of copper placed in a solution of a zinc salt endeavours to force out the zinc from its solution and cause it to be deposited in the metallic state and at the same time to form a salt of copper. Whether the zinc or the copper goes into solution depends on whic THE ANALYST. 5 metal exercises the greater force and the same reasoning applies to every cathion pair. In the case we are considering the solution tension of the zinc is about 1.7 volts whilst that of copper is much less and so a deposition of metallic copper might be expected.There is however another important factor to be borne in mind-viz. the supertension. I t would be expected that every metal should be deposited from the solution of its salts when a potential difference is exerted slightly in excess of the particular solution tension of that metal. As a rule however a considerable fGrce (known as supertension) in excess of this has to be applied before the deposition is brought about. Thus in the case of a mixture of zinc and dilute sulphuric acid very little hydrogen is liberated although the solution tension of zinc is about 1.7 volts and that of hydrogen only about 1 volt the zinc in this case exerting a retaining effect on the hydrogen equal to about 0.7 volt. Our experi-ments deal principally with the solution of metals in sulphuric acid which we will now proceed to study in detail as a particular example of a general law.Each metal exerts a supertension with respect to hydrogen the largest in our experiments being zinc and the smallest platinized platinum If now we deposit a particle of platinum on a piece of zinc placed in dilute sulphuric we have obtained an outlet for the hydrogen at a low potential and consequently as the solution tension of the zinc is 1.7 volts and the supertension of the platinum only a very small fraction of a volt the hydrogen is liberated with greatly increased rapidity. The increased velocity of the evolution we may suppose to be directly proportional up to a certain point to the number of molecules present on the surface.The hydrogen obtained from a pure zinc surface will however have properties differing entirely from those in the other case. In the first case where the potential of the hydrogen is high more difficult reductions may be effected than in the other. These con-siderations have been applied to the Marsh-Berzelius test for minute quantities of arsenic. In the first place let us take the simplest case of the solution of pure zinc in sulphuric acid in presence of arsenious oxide. The hydrogen bringing about reduction may be supposed to be in the atomic condition possessing one free bond and a potential of 1.7 volts. This hydrogen must be in solution and the amount present at any instant will be constant and independent of the rate of evolution-of course after the solution in the neighbourhood of the metal has become saturated.On the other hand the arsenious oxide may be supposed to dissociate to a small extent into trivalent ions possessing three free bonds and a certain definite potential, a reduction taking place when these free bonds are satisfied by hydrogen. This may be expressed in the following manner : As,O - 2 A s r + 3 0 A s r +3H 4- ASH,. We have now a number of factors influencing the reaction to take place Sup-pose the active mass of hydrogen is 31~ and that of the free arsenic ions MA^. After the action has been allowed to take place until as much arseniuretted hydrogen has been driven off as it is possible to obtain there will be still left in solution a very minute quantity represented by M A ~ H ~ .We shall have now obtained a state of equili-brium depending on our five factors. MAsH3 must according to the mass law be proportional to the product M,,.MH. There is however another influence at work 6 THE ANALYST. and one overlooked apparently by previous investigators viz. the potential of the various active ions. Thus the factor M~,H must also be proportional to the product of two factors PH.PA~ depending on the potential of the hydrogen and of the arsenious ions respectively or expressed in the form of an equation : or-where k is a constant The arseniuretted hydrogen will also tend to decompose with some constant force which has been included in the factor k. The amount of arseniuretted hydrogen left in solution may be regarded as constant or nearly so at any moment during our experiment.MH and Pa are also constant factors inde-pendent of the condition of the metal and we may therefore write our equation-M A ~ H ~ ,C k ~ I A S * PH, when k is a constant factor. That is to say the active mass of the ASH is propor-tional to the product of the potential of the hydrogen and the active mass of the arsenious ions. Owing to the extreme dilution of the solutions in question this latter also may be considered proportional to the total arsenious oxide present and we may write finally : MA~H~ ,- KA~PH, where As represents the amount of unreduced arsenious oxide and K a constant factor. If now the above equation is disturbed by the addition of As-ie. if the factor As is increased-reduction takes place or the reaction proceeds from right to left and the same result will be reached if PH is made larger.On the other hand if PH is made smaller As must be made larger to preserve the equilibrium. I n other words there is always left in solution after each experiment a certain residuum of unreduced arsenious oxide. This is very small when the factor PH is large as in the case of pure zinc and must we know be far less than 3$o mgr. and probably less than i$c8 mgr. as this quantity can with certainty be detected. On the other hand, when PII is small then the oxide left in solution may be relatively considerable. This is very pronounced in the case of zinc coated with platinum in which case it may be more than 1 mgr. This phenomenon was observed recently by Sand and Hackford in the case of copper spirals in an electrolytic apparatus when it was shown that any quantity of arsenious oxide above T$o mgr.could be detected but nothing smaller. I n the above considerations we have been dealing with arsenite solutions, but very little additional difficulty is presented in the case of arsenates. In the latter case we have two processes taking place simultaneously-first the reduction of arssnate to arsenite and secondly the formation of arseniuretted hydrogen. The first processes must take place with extreme ease as our experiments with the Marsh-Berzelius apparatus show that in both cases equally good arsenic deposits are obtained. It is extremely probable that similar explanations apply to other reduc-tion process and it should be in many cases possible to predict results.Some of these we propose to consider in a future communication. The Grst to draw attention t o the phenomenon of supertension was Caspari who arranged the metals in th THE ANALYST. 7 following order Platinum (black) gold iron platinum (polished) silver nickel, copper palladium cadmium tin lead zinc and mercury the first-mentioned metal having the lowest value whilst mercury possesses this property in the highest degree. According to our view this will also represent the order of efficiency of these metals in any reducing action in which they take part a view of the correctness of which we have obtained ample experimental confirmation. The only exception is palladium, which we have placed along with platinum and gold. Tafel in a paper on the electro-lytic reduction of certain organic substances placed the metals in the following order Platinum silver tin copper mercury zinc iron in which iron is credited with reducing properties equal to those of mercury and lead.This is however, directly opposed to both Caspari’s and our own work and the point was therefore, studied in detail five alloys of iron being prepared arid examined ; but in each case fhe result confirmed our view that iron must be placed among themetals of compara-tively feeble reducing activity. Iron however exhibits a very curious property, which we will proceed to describe. The introduction of a metallic salt into the Marsh apparatus is invariably accompanied by a sudden increased liberation of hydrogen, if the cathion happens to be a metal of low supertension.This however did not, as a rule tcke place in the case of iron salts until after violent agitation of the solu-fion in the apparatus but on the addition of a small quantity of arsenious oxide the action became as brisk as in the case of other metals of low supertension. The explanation appears to be that the layer of hydrogen prevents the iron from coming into contact with the. zinc and does not therefore permit the formation of an actual couple. Violent agitation or the addition of an oxidizing agent however removes this hydrogen and permits the formation of the couple. A similar phenomenon was noticed in the potential experiments described below. Here it was observed that the potential of the zinc did not drop suddenly on the addition of an iron salt as might have been expected but only slowly.When however the surface hydrogen was removed from the zinc by means of a little arsenious oxide the iron fell into line with our other experiments. Similar properties were noticed in the case of other metallic salts but the action was very much less marked. On studying the results of Caspari’s experiments on supertension it at once became clear to us that those metals which appeared to have no effect in causing the retention of arsenic and whose behaviour we had been unable to explain were precisely those possessed of high supertensions and we therefore proceeded to study the question in greater detail with this possible solutionin our minds A b the outset it became evident that the addition of metallic solutions was not the best way in which to make a compara-tive study inasmuch as in many cases the deposited metal separates largely from the zinc and so ceases to be effective.We therefore decided to work with zinc alloys. To this end small quantities of the metals whose effects were to be studied were dloyed in a graphite crucible with highly sensitive zinc and the resulting product granulated in the ordinary way. The numbers given below in the second column, and representing the proportions of the added elements are in most cases only approximate owing to the difficulties inseparable from the preparation of such alloys ; but they of course represent in all cases the maximum quantities of the added impurities 8 THE ANALYST. ALLOYS. Metal added.Nickel I. . Nickel 11. . Nickel 111. . Cobalt . . . . Iron I. . . . 9 ) . Coialt Ii. . . 9 . * * Iron 11. . . I r o l III. . Iron (reduced) 9 9 Iron (wire 9 7 9 7 Copper . . Silver . Platinum Sodium Tin . Cadmium , 9 ’ . 1 - 1 -. . . . . . . . . . . Quantity of Metal added to 100 Grams Zinc. Arsenious Oxide added. 30 20 20 10 30 20 10 30 20 30 20 20 30 10 50 30 ’1 0 30 20 30 20 30 10 10 10 Arsenious Oxide found. 10 5 10 4 6 4 2 20 2 10 5 2 5 2 10 5 1 10 2 12 5 6 5 10 10 Arsenious Oxide left in Flask. 20 15 10 6 24 16 8 10 18 20 15 18 25 8 40 25 9 20 18 18 15 24 5 --At this point we would call attention to the iact that iron exercises a very marked retentive effect as this is opposed to certain published observations.We propose to revert to this a little later. From the above results which entirely confirm those obtained with saline solutions it will be seen that those metals which are known to possess low supertension values (see Caspari’s order above) have in all cases the property of preventing the conversion of arsenious oxide into arsenic hydride whilst others-e.g. cadmium tin and lead-which have high values, exercise no appreciable effect. Very numerous experiments having confirmed the correctness of this explanation of the so-called ‘( insensitiveness ” of zinc it clearly became a matter of interest as well as of practical importance to see whether the results we had obtained could not be applied to the conversion of insensitive zinc into the sensitive metal.Since in a zinc-iron couple the hydrogen may be supposed to be liberated largely from the iron particles it seems reason-able to conclude that other cathions would also be liberated at the same points, or in other words form a coating over them and thus mask their action. To test the correctness of this view we added cadmium sulphate to the contents of the Marsh-Berzelius flask containing highly insensitive zinc and found that under these circumstances even such small quantities of arsenic as & mgm. coul THE ANALYST. 9 be readily and witJh certainty detected. In the following table we give examples of this: Alloy of Zinc with Sron I.. . 9 9 . . Iron 11. . . Iron (reduced) . Silver . Platinum . . . . Cobalt . . Sodium . I . . Nickel I. . . Nickel 11. . . . Iron (wire) . . Copper . . Salt added. Cadmium sulphate ( 2 grams) . 9 9 9 9 9 7 9 9 9 9 9 9 9 9 9 7 9 7 > ? 9 9 9 9 9 9 9 9 9 9 7 7 9 9 9 9 9 9 > ? $ 9 P 9 Arsenious Oxide added. 20 10 10 10 10 I0 10 10 10 10 10 1 0 Arsenions Oxide found. 20 10 10 10 10 10 10 10 10 10 7 7 Similar results were obtained with tin and lead as might have been anticipated In the following table from a consideration of the views we have put forward. experiments with these metals are recorded : Alloy of Zinc with Iron 11. . . Iron (reduced) .Iron (wire) . Silver . . Platinum . . 7 7 . . 9 9 . 7 9 . > > . . Copper . . CoiLlt . . Sodium . . Nickel I. . . Nickel I T . . . . . I . 9 7 . . 9 9 . . ? 9 . . Salt added. Stannous chloride . Lead acetate . Stannous chloride . Lead acetate . Stannous chloride . Lead acetate . Stannous chloride . Lead acetate . Stannous chloride . Lead acetate . Stannous chloride . Lead acetate . Stannous chloride . Lead acetate . Stannous chloride . Lead acetate . 9 9 9 > 9 Y 9 . Arsenious oxide added. 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 Arsenious Oxide found. 10 10 10 10 10 10 10 10 10 10 10 10 10 10 7 7 10 10 On referring to the above numbers it will be seen that some of the results with nickel-zinc alloys are low.This appears to be due to the fact that the nickel exhibits a marked tendency to become separated in the form of minute dark particles from the zinc during the solution of the alloy ; these particles pass into solution in the acid and subsequently become deposited on the cadmium surface 10 THE ANALYST. Three samples of insensitive zinc as supplied by chemical dealers were then experimented with. B,” and “ C,” contained the varying amounts of iron as stated. It may be added in this connection that several samples of highly sensitive zinc were found to contain only infinitesimal traces of iron. These referred to below aa (( A,” (6 7 9 ) c. ( 6 Am)? < < B.” Fe .. 0.09 per cent. 0.24 per cent. 0.66 per cent. The following results were obtained when using these three samples in the Marsh-Berzelius flask : ZINC “ A.” Zinc alone . . . . Zinc + 2 grams cadmium sulphate Zinc alone . . . . Zinc + 2 grams cadmium sulphate . . . 9 , -Zinc alone . . . . Zinc + 2 grams cadmium sulphate . . . 9 9 * * . 9 9 3 Y Y 9 2 Arsenious Oxide added. 3 5 ZINC (‘B;” 5 10 5 10 ZINC ( ( C.” 5 10 5 10 Arsenious Oxide found. 4 5 1 5 5 10 1 2 5 10 Arsenious Oxide left in Flask. 5 Having thus been successful in overcoming the tendency exhibited by some samples of zinc to retain arsenic in the Marsh-Berzelius flask we thought it would be interesting from a theoretical point of view to see whether other ( ( insensitive ” metals capable of being used for the purpose of obtaining hydrogen might not be similarly rendered sensitive.To this end we experimented with two samples of magnesium and the results are given in the following table : MAGNESIUM ( ( A’’ Magnesium alone . . . Magnesium + 0-1 gram nickel sul-phate . . . . Xagnesium + 0.1 gram iron sul-phate . . . . Magnesium + 0.1 gram cadmium sulphate . . . . Magnesium + 0.1 gram zinc sul-phate . .,. . . Arsenious Oxide added. 20 1000 500 20 20 Arsenious Oxide found. 10 5 4 20 20 Arsenious Oxide left in Flask. 10 995 496 - THE ANALYST. 11 MAGNESIUX ‘‘ B.” Magnesium alone . . . . . Magnesium + 2 grams cadmium sulphate . .Magnesium + 2 grams cadmium sulphate . . . . Magnesium + 2 grams stannous chloride . . . . Magnesium + 2 grams lead acetate . . Arsenious Oxide added. 10 10 10 10 Arsenious Oxide found. 4 10 5 10 10 Arsenious Oxide left in Flask. 6 -Similar experiments made with calcium and aluminium failed to give results : in the former case owing to the impossibility of regulating the reaction and in the latter owing to the formation of a white deposit i n the heated arsenic tube. In the next place solutions of arsenic oxide were experimented with in pre-cisely the same manner as the arsenious oxide solutions above mentioned and the results were found to be in all cases strictly comparable as will be seen in the following table : I I Arsenic Oxide Arsenic Oxide (Equivalent to (Equivalent to I Arsenious added Oxide).I Arsenious found Oxide). ~ Zinc + 2-0 grams nickel sulphate . . . Zinc + 2.0 grams cobalt sulphate . Zinc + 0.1 gram platinic chloride . . . Zinc + 2.0 grams cadmium sulphate . 500 500 1000 10 50 50 10 -Arsenic Oxide left in Flask (Equivalent to Arsenious Oxide). 450 450 1000 I Seeing that all the above results are susceptible of explanation on the assump-tion that the potential of the hydrogen is one of the determining factors we felt that the study of this question would not be complete without a repetition of some of the electrolytic work which has already been done. This appeared to us to be the more necessary inasmuch as one or two of our observations are not in agreement with those of previous investigators.The following is a description of the apparatus employed. The cathode compartment consisted of a wide glass tube 8 cm. deep having an internal diameter of 3.5 cm. and provided with an outlet tube a t the side. The top of this compartment was closed with a rubber stopper through which passed a tap-funnel and a short piece of glass tube to support the cathode. Over the bottom was stretched a parchment diaphragm which was held in position by an elastic band. The side tube was connected by meam of a cork with the drying apparatus con-taining the usual lead acetate paper and calcium chloride. The anode consisted of a platinum ring placed in a beaker slightly larger than the cathode compartment. The whole apparatus was placed in a vessel of cold water to keep the solutions cool during each experiment.As it was necessary to change the cathode from time to time this was kept in position by means of a sealing-wax joint. In the preliminar 12 THE ANALYST, work it was found that the best results were obtained with high-current densities and from moderately concentrated solutions of the arsenic compounds. For this reason a cathode having a superficial area of 4 square cm. on both sides was used, and the amount of liquid in the cathode compartment did not exceed 18 C.C. I n every case a current of 6 amperes was used and no particular attention was paid to the voltage as this varied very considerably with different parchment diaphragms. Two sets of experiments were made and no arsenic was added until the apparatus had been in action for ten minutes to test the purity of the reagents.The following results were obtained from arsenite solutions with various cathodes the acid used being in each case 10 per cent. sulphuric acid by volume : Cathode. Lead . . . Cadmium . . . Tin . .<. . . Copper I. . . . Copper 11. . . . Silver . . . Nickel . . Platinum (black) . ,. $ 1 . 9 9 9 . . Iron . . . . Arsenious Oxide added. 10 10 10 100 50 50 40 1,000 10,000 50 Arsenious Oxide found. 10 10 10 10 2 10 20 0 Large 5 Arsenious Oxide left in Flask. ---90 48 40 20 1,000 45 -SimiIar electrolytic experiments to the above were next made using solutions of arsenic oxide and in all cases smaller mirrors were obtained than those yielded by corresponding amounts of arsenic in the arsenious state THE ANALYST.ARSENATES. Arsenious Oxide added. Time of Experiment. 13 Arsenious Oxide Arsenious Oxide found. left in Apparatus. Cadmium . . Lead . . . Nickel . Copper 11. . . ) (black) . . . . Platinum (bright). * . Arsenates added corresponding to Arsenious Oxide. 10 10 70 40 40 10,000 Arsenic found. 7 7 10 10 2 Large Arsenic left in Flask. s 3 60 30 38 The above results with cathodes of platinum lead cadmium tin nickel and copper confirm the view we have expressed above. It will be seen that with the iron electrode we have a considerable retention just as with the solutions and alloys of that metal.We make special reference to this point inasmuch as Tafel and Sand and Hackford working conjointly have credited an iron electrode with a reducing efficiency comparable with that of lead and zinc. When wing copper cathodes the following phenomena were observed which we think worthy of being recorded : The cathodes of this metal invariably became dulled after a few experiments, owing probably to the deposition of metallic arsenic on the surface; but at the same time the electrode became more and more sensitive although no mirror could be obtained in a blank run for half an hour. The following experiments were there-fore made Two copper cathodes were allowed to act in electrolytic cells one con-taining a large excess of arsenious oxide and the other a similar quantity of arsenic acid for one hour when both had become covered with elementary arsenic in a non-reducible form.ARSENIOUS OXIDE. The following experiments were then made : 15 minutes . . 9 ) ,, . . . . 30 20 20 10 2 5 20 18 15 Time of Experiment. 45 minutes . . . . 9 ) 9 9 7 9 7, 9 9 . . . . . . . . AICSENATES. Arsenate added Equiva-lent t o Arsenious Oxide. 30 20 20 20 20 10 Arsenic found. 10 1 2 5 10 7 Arsenic left in Apparatus . 20 19 18 15 10 14 THE ANALYST. In the case of the arsenates it will be seen the electrode had in some of the experiments become as sensitive as cadmium or lead from which it may be con-cluded that the supertension of arsenic is large.An endeavour was made to deposit arsenic chemically on copper for the use of electrodes but in this case a mirror was always obtained in the blank test. The lead copper and nickel electrodes were cut from sheets of the pure electrolytic metals. Iron was deposited on a sheet of the same metal a thickness of at least 1 mm. being necessary otherwise it was all dis-solved during the experiment. The cadmium tin and silver were deposited on platinum foil. It may be remarked that it was found almost impossible to get two electrodes to give the same results although prepared apparently in the same manner. Thus the copper electrodes I. and II. which were cut from the same sheet and cleaned in different ways gave different results. Further the introduction of salts of the metal composing the cathode often interfered seriously with the reaction.In the case of arsenate solutions the formation of the mirror was much slower than with arsenites and for this reason the experiments were continued for three-quarters of an hour in the former case. The results were also invariably lower which is partly accounted for in the following way Arsenic pentoxide forms with water a hydrogen salt (acid) which must be partially dissociated in solutions as dilute as those we have used. The ions thus formed must of course participate in the con-veying of the current from pole to pole the cathion moving to the negative pole, while the anion (AsO,) travels to the anode in the opposite direction and so causes the arsenic to be removed from the sphere of action of the hydrogen.To test this experimentally 0-02 mgr. arsenious oxide was introduced into the cathode compart-ment and the apparatus allowed to run for a quarter of an hour. Ten C.C. were then removed from the anode compartment and the arsenic determined in the Marsh apparatus when a deposit corresponding with 0.005 mgm. was obtained. This experi-ment was repeated with a copper cathode using in this case 0.2 mgr. arsenious oxide. A mirror corresponding to 0.005 mgm. was obtained from the contents of the anode compartment although the solution contained a little copper sulphate. We may add that we also worked with the apparatus of Sand and Hackford but as with our own the results were too low in the case of arsenates. I n what has gone before we have assumed that the presence of any metal capable of producing insensitiveness in zinc is accompanied by a fall in the potential of the hydrogen formed as this appeared not only to explain the results we had obtained but also to be in accord with the published observations of other workers.We felt however that the investigation would not be complete without an attempt to verify this suggestion by direct experiment and we therefore proceeded to deter-mine the potential of dissolving zinc in the presence of various salts. The potentio-meter method was the one employed as it was considered the most sixitable for the purpose and the results are given in tabulated form below. The E.M.F. of the following element was measured mercury-mercurous sulphate N sulphuric acid -zinc.For each experiment a new piece of zinc was used the surface of which was cleaned with strong nitric acid. After having been thoroughly washed with water this was connected with the normal electrode and the potential read off. Thisvaried somewhat and never became quite steady notwithstanding the fact that the aci THE ANALYST. 15 around the zinc was well stirred The salts referred to in column 1 were then added and the sudden changes in potential recorded. These are given in columns 3 and 4 the signs + and - denoting respectively rise or fall in the E.M.F. of the element in question. Salt added. None . . . Cadmium sulphate . . . None . . . Cobalt sulphate . None . . . Platinum chloride . . . None . . . Copper sulphate . , Quantity.-2.0 grams -2.0 grams -0.2 gram -2.0 grams E.M.F. of Elcmen t. 1.398 volts 1.419 volts 1,420 volts 1.400 volts 1.412 volts 1.237 volts 1.398 volts 1.375 volts Change of E.M.F. Remarks. +0*021 volt -0-02 volt --0,175 volt --0,023 volt E.M.F. was falling very slowly but almost steady. Slowly rising. E.M.F. dropping E.M.F. dropping. Very smallquan-tity Codeposited. slowly. Dropping slowly. Dropping rapidly. Dropping slowly. Dropping . In a11 these experiments 600 C.C. of normal sulphuric acid was used and the salts The zinc was never completely covered as the For this reason stronger solutions were employed added were dissolved in similar acid. solutions used were very dilute. in the next series of experiments.Salt added. None . . . Cndrnium sulphate . . . None . . . Iron sulphate . None . . . Cobalt sulphate . None . . . Copper sulphate . . . Quantity. -20 grams -20 grams 20 grams 20 grams --E.X.F. of Element. 1.403 volts 1.372 ,, 1.430 volts 1.343 ,, 1.416 volts 1.320 ,, 1.428 volts 1-146 ,, Change in E.M.F. --0.031 volt --0.087 volt --0.096 volt -- 0.282 volt Remarks. Cadmium sulphate not quite pure At this point 10 grams cadmium sulphate were added although the solution still contained copper sulphate and the E.M.F. first rose to 1.30 volts having in 16 THE ANALYST. creased 0.154 volt and still continued to rise. were added one after the other without changing the solution.I n the next experiment the salts Salt added. None . . . Iron sulphate . Cobalt klphate . . . Copper sulphate . . . Platinum chloride . . . . Quantity. E.M.F. of Element. 1417 volts 1.377 ,, 1.369 ,, 1.331 ,, 1,219 ,, 1.078 ,, Change in E. M. fi’. -- 0.040 volt -0.008 ,, -0038 ,, -0.112 ,, -0.141 , _-0.339 ,, Eteniarks. There can be no doubt that the reducing.efficiency of the hydrogen obtained as the result of the interaction of metals and acids is dependent on a number of factors both chemical and physical and that even those reactions which are apparently the most simple are in reality exceedingly complex. We think however, that the experiments of which an account is given in this paper show very clearly that among those factors the question of ‘L potential ” or 4 ‘ supertension ” plays a very prominent and important part.In conclusion we desire to record our best thanks to Mr. E. R. Saunders for placing at our disposal the delicate apparatus necessary for the determination of the potential changes above referred to, DISCUSSION. The PRESIDENT (Mr. Bevan) having invited discussion, Professor W. A. TILDEN said that he had come to hear this paper quite in the character of a novice for although it was perfectly true that he had occupied himself considerably with the subject three years ago he had not had any opportunities of pursuing it since and he should hesitate to express any very strong views in the presence of so many who were accustomed to apply the Marsh-Berzelius method daily in their work.The authors however had not only given an exposition of the results of their experiments which had led to the very great improvement in the method which they had described but had also furnished theoretical views as to the nature of the changes that took place. With regard to those theoretical views he confessed that he did not feel at that moment as though he could accept them quite without reserve-at any rate without a little time to think about them. These reducing effects bad been spoken of as the effects of the hydrogen but that was a sort of elliptical phrase which no doubt had a good deal behind it. He wasnot quite sure what the authors’view was as to the exact action of the hydrogen in the process. At the same time in every one of these cases there was a couple and it was the couple that was the active agent.Of course it was perfectly true that hydrogen, even in the molecular form had a greater power of reduction than was commonly supposed. He did not know whether everybody was aware that even hydrogen gas was capable of reducing some metals from solution. Solution of silver nitrate fo THE ANALYST* 17 instance could be made to deposit metallic silver by merely bubbling hydrogen gas through it It was probable however that the active hydrogen in the case they were considering was not in the molecular form but was ionized. In the reduction of arsenious acid or of arsenic acid the process necessarily took place in two stages. First there was the removal of the oxygen and there might also be some deposition of elemental arsenic; then came the production of the hydrogen compound.At present he was certainly rather inclined to cling to the idea that the differences which had been observed in the effects of different metals were really not so much attribut-able to the difference of potential in the couples employed as to the peculiar surface action which many of them were capable of exercising. He could not help thinking that some of them simply adhered to the deposited arsenic in some way which had been spoken of as “ surface action,” though nobody knew what that meant ; and that, in consequence of this surface action the arsenic was not affected by the reducing action of the hydrogen and not carried over in the form of a hydride. The reduction of arsenic acid was notoriously difficult but he confessed that he was a little reluctant to accept the explanation which the authors had given.White arsenic dis-solved in water undoubtedly contained an acid-h. it contained hydrogen ions and arsenious ions as well as arsenious oxide. Of course it was perfectly true that the greater part of the arsenious acid was in a state of dissociation but the comparison which the authors instituted in the two formulx? they had given did not to his mind, afford a sufficient explanation of the difference. That arsenious oxide was very much more easily reducible under all circumstances than arsenic acid was an elemental fact which did not admit of further analysis. At any rate if the same experiment were made with arsenious acid instead of with arsenic acid he should rather expect to find that the arsenious ion would in like manner find its way by diffusion into the anode chamber.His experience of the practical application of the Marsh-Berzelius process was 80 limited that he did not think that he could add any-thing to what the authors had said on this head. He confessed however that if he had occasion to use the process he should at once proceed to the use of the electro-lytic method by which in his experience using clean platinum electrodes there was no difficulty in getting the whole of the arsenic out of an arsenious solution in all ordinary cases. What might be the cause of the retention of the arsenic in some cases where there were very complicated mixtures of organic materials was of course a matter about which there might be considerable difference of opinion.When it was necessary to,use zinc the authors had undoubtedly placed in the hands of the analyst a most valuable modification of the method. It was certainly a most remarkable thing that cadmium and tin stood apart as they apparently did from all the other metals which they might be expected to resemble in this respect. Mr. HERBERT JACKSON said that he was sure that the authors to whose paper he had listened with very great interest would not mind his saying that he did not think their explanation included everything. While the experiments and their numerical results were being detailed it had passed through his mind thatJ a closer investigation of the arsenide combination was necessary He believed however, that the authors’ idea with regard to potential was in the main correct in the following sense The arsenic remaining in the flask would if a sufficient time wer 18 THE ANALYST.allowed to elapse be practically all deposited on the zinc. The potential difference between zinc contaminated with iron and zinc arsenide was probably low. I t would be necessary to have a knowledge of the potential difference between pure zinc and arsenic and between zinc and zinc arsenide. Now that potential difference might be regarded-though it was true that the view was a somewhat mechanical one-as representing the difficulty of breaking up the complex molecules into simpler con-stituents; and if a larger potential difference were required before there was any change in the molecular complexes of the metals then he could quite understand that with a high potential difference there would be a greater number of arsenic and zinc molecules in a simple and active condition.He had made an attempt to obtain a distinction between the lines in the spectrum of arsenic (in view of the possibility of the separation of two arsenic spectra) by means of some experiments in vacuum tubes with a high-tension current which might be referred to now because, to his mind there was no necessary distinction in kind between a spark in a vacuum tube an electric arc and a current in an electrolyte. In those experiments arsenic electrodes having been found inconvenient he had used electrodes of zinc alloyed with arsenic in an electrolyte of water vapour at a pressure of about 0.5 mm.with a view to obtaining arsenic hydride. The potential difference under those conditions was apparently too low to give more than the spectrum of water vapour. On intro-ducing a spark-gap however a certain number of arsenic lines came up but only those that would be expected-namely the strong lines of shorter wave-length that were usually the first to appear when any metal was being exposed to electrical oscillation. He had experimented with other alloys similar in character to those with which the authors had worked and his observations-in the case of a lead-arsenic alloy especially-bore out what they had said as to the effect of an increase in potential being more completely to separate the arsenic from combination and to convert it into arsenic hydride.He did not think however that the full explanation with regard to potential had yet been arrived at for the number of possible com-binations was very large and the question of the molecular complexity of the arsenides came in as well. He was sorry that the authors had not been able to deal more fully with aluminium. It was a well-known fact that in presence of aluminium in potash solution the arsenic mirror could be obtained from arsenites but not from arsenates. The potential difference in a zinc-aluminium couple rose very rapidly indeed and if the surface of aluminiuin were examined after the rise of potential it-would be found to be a very finely divided surface which if washed and dried in vacuo and then exposed to the air oxidized very rapidly indicating a finely-divided and active state of the aluminium.He was therefore disappointed that the authors had found aluminium unsuitable owing to its yielding such a peculiar substance and should much like to know what it was. Although aluminium in potash solution would not under ordinary conditions give any arsenic hydride with an arsenate yet with a high-tension current an arsenate could be reduced from such a solution almost as easily as an arsenite. All this went to show very strongly that the question was one of potential difference and that it was necessary to consider the relation of the arsenic to the other metals and also of these metals to hydrogen. The influence of cadmium showed that the arsenic would not come off unless the potential wa THE ANALYST.18 relatively high and a state of high potential was against the existence of polymeric modifications of the metals on their immediate surfaces. In the case of iron this was well shown by the differing views of Caspari who gave it with platinum a very low value and of Tafel who placed it high in the scale. If a low electromotive force were used the iron acted as stated by Caspari whereas when a high electromotive force was used as was the case in Tafel’s experiments the iron probably assumed a, passive state. With regard to platinum he remembered particularly the experiments made by Professor Bloxam on the electrolytic separation of antimony and arsenic in which the mixed antimony and arsenic sulphides were simply electrolyzed in a porous cell with platinum electrodes in dilute sulphuric acid when the hydrogen at the cathode reduced the arsenious sulphide and the mirror was obtained perfectly, whereas the sulphide of antimony was left in suspension and did not show.No explanation however even if correct could add to the great practical value of the authors’ work on this subject and the criticisms he had ventured to make so far from detracting from its value only showed how important it was from a theoretical point of view also. Dr. MORITZ said that the means which the authors had placed in their hands for dealing with the insensitiveness of zinc were of the greatest value for insensitive zinc often caused much inconvenience and delay. He should like to ask-though it was perhaps hardly within the scope of the paper-whether the treatment of the zinc with cadmium mould have the desired effect on iron and copper introduced in minute quantities in the materials to be analysed because the products with which he was most concerned did often contain minute quantities of iron and copper.Mr. BLOUKT said that-viewing the matter not as one of arsenic determination, but as an electrolytic question-the experts appeared to disagree as to whether the evolution of arsenic depended upon some critical difference of voltage between the metals concerned or on the absorption or nd-sorption of the arsenic by the other metals. That might it seemed to him-he made the suggestion for the authors’ consideration-be determined by using electrodes of the metals the behaviour of which it was desired to examine.His point simply was that if it were a fact that the arsenic came away more readily when the hydrogen was liberated at a particular potential that potential could be more accurately determined electrolytically by using a suitable electrode of the metal than by adding a salt of that metal to a mass of zinc and so constructing a composite electrode. Mr. LAW in reply referring to the reducing action of molecular hydrogen said that the potential of the platinum black electrode was just slightly higher (namely, 0.005 volt) than the potential of pure hydrogen so that the reducing action of pure molecular hydrogen might be expected to be almost equal to that of platinum black in the presence of electrolytic hydrogen. There was according to Caspari, a very marked retaining effect on the hydrogen due to surface action and to this retention waa due the phenomenon of supertension.The greater the retaining effect on the hydrogen the higher the supertension was and consequently the more efficient the electrode was as a reducing agent. Arsenious oxide as was well known did not practically speaking form any acid. In these very dilute solutions the proba-bility was that any arsenious acid present dissociated first of all into water an THE ANALYST. As203 and that the arsenious oxide itself dissociated further into arsenic and oxygen. As a matter of fact if an electric current were passed through a very dilute arsenious oxide solution using a copper electrode metallic arsenic was actually deposited on the cathode and did not wander to the anode as it would if the ion were a complex one.Further the conductivity of arsenates and arsenites -or rather of arsenic acid and arsenious acid-was entirely different. For instance in the case of arsenic acid the conductivity was comparatively high. Arsenic acid was a strong aoid in comparison with arsenious acid. The latter was a substance which had practically lost all its acid properties and its conductivity was almost inappreciable. I n addition to this full deposits were obtained from arsenious oxide solutions which would not have been expected where arsenic formed part of a complex anion. He must confess that this paper did not extend quite as far as Mr. Jackson would like it t o go. They did not deal with the exact mechanism of the reaction but it was quite possible that complex arsenic compounds were actually formed and then reduced.Possibly in a future paper they might be able to do something in that direction but at present they had not gone so far. It was however a fact that as the potential increased more and more arsenic was reduced and they had laid down this as one of the factors determining the reduction. They did not say that it was the only factor but they regarded it as an important one. There might of course be others. There might for instance be complex arsenides formed but these they had not attempted to deal with. He was not quite sure as to the voltage used by Tafel in his electrolytic work but Tafel mentioned that a lead electrode gave a voltage of about 4 using a porous pot and a current of 1 ampere.In this paper no details were given of any break or rise. . Mr. JACKSON said that the rise was very slow. Mr. LAW said that Tafel had used a current of from 1 to 3 ampAres. With respect to small traces of iron nothing conclusive could be stated in the present paper but in a future communication full particulars would be given. They had not themselves used electrodes of different metals but several other investigators had done so and in fact Caspari’s table was compiled from results obtained from wires of these pure metals while numerous other investigators had studied the action of such electrodes. For that reason they had not considered it necessary to use various electrodes in that way. They had however used cathodes covered with cadmium but had not measured the potential.Dr. H. SAND who has had the opportunity afforded him of perusing an advanced .proof of the paper has favoured us with the foUowing remarks : I have been very much interested by Messrs. Chapman and Law’s important paper more especially by the experiments on the sensitization of insensitive zinc in the ordinary Marsh-Bereelius test. The fact that the authors obtained different results when using an iron electrode from those recorded in Tafel’s paper as well as in that of Mr. Hackford and myself, is I think very probably to be ascribed to the circumstance that they experimented wiih electrolytically-prepared iron whereas we used thin iron wire such as is employed for the standardization of permanganate solutions. Our experiments (Table IV.of our paper) on solutions containing FeSO, it will be found agree with those o THE ANALYST. 21 Messrs. Chapman and Law. The difference seems the less surprising as electrolytic iron very probably contains hydrogen and is known to differ from ordinary iron n other ways such as in its great resistance to rusting etc. The difference between Messrs. Chapman and Law’s paper and ours on the behaviour of a lead electrode towards arsenate solutions is not very great. The lead they experimented with appears to have required rather longer than ours for the complete reduction of the arsenatc. This being so it does not seem surprising that a8 much of the arsenate should have been lost by diffusion if not through leakage through the partition as by the cause mentioned by the authors. In any case I think the difference would be more marked when experimenting with large than with small quantities. With regard to the authors’ theoretical introduction I should be inclined to the view that the hydrogen before being given off in the form of bubbles is contained in a state of extreme supersaturation in the layer of solution immediately adjoining the electrode. In this case M should not be considered constant but a function of the supertension in such a manner that the supertension would according to our present theoretical views enter into the expression for the equilibrium concentration of the arsenic hydride in an exponential form thus : where P is the supertension and K and K are constants. Personally however I think it corresponds best to our actual knowledge of the matter to say that the rate of evolution of arsenic hydride is an unknown function of the supertension and the concentration of the arsenic compounds this function increasing from zero upwards, as the concentration of the arsenic and the supertension increase from certain definite finite values