Chapter XXIV. Note 9 ^{bis}). If there were more such

well-investigated cases of solutions, our knowledge of double salts, solutions, the influence of water, equilibria, isomorphous mixtures, and such-like provinces of chemical relations might be considerably advanced.

[24] The normal ferric salts are decomposed by heat and even by water, forming basic salts, which may be prepared in various ways. Generally ferric hydroxide is dissolved in solutions of ferric nitrate; if it contains a double quantity of iron the basic salt is formed which contains Fe_{2}O_{3} (in the form of hydroxide) + 2Fe_{2}(NO_{3})_{6} = 3Fe_{2}O(NO_{3})_{4}, a salt of the type Fe_{2}OX_{4}. Probably water enters into its composition. With considerable quantities of ferric oxide, insoluble basic salts are obtained containing various amounts of ferric hydroxide. Thus when a solution of the above-mentioned basic acid is boiled, a precipitate is formed containing 4(Fe_{2}O_{3})_{8},2(N_{2}O_{5}),3H_{2}O, which probably contains 2Fe_{2}O_{2}(NO_{3})_{2} + 2Fe_{2}O_{3},3H_{2}O. If a solution of basic nitrate be sealed in a tube and then immersed in boiling water, the colour of the solution changes just in the same way as if a solution of ferric acetate had been employed (Note 22). The solution obtained smells strongly of nitric acid, and on adding a drop of sulphuric or hydrochloric acid the insoluble variety of hydrated ferric oxide is precipitated.

Normal ferric _orthophosphate_ is soluble in sulphuric, hydrochloric, and nitric acids, but insoluble in others, such as, for instance, acetic acid. The composition of this salt in the anhydrous state is FePO_{4}, because in orthophosphoric acid there are three atoms of hydrogen, and iron, in the ferric state, replaces the three atoms of hydrogen. This salt is obtained from ferric acetate, which, with disodium phosphate, forms a _white precipitate_ of FePO_{4}, containing water. If a solution of ferric chloride (yellowish-red colour) be mixed with a solution of sodium acetate in excess, the liquid assumes an intense brown colour which demonstrates the formation of a certain quantity of ferric acetate; then the disodium phosphate directly forms a white gelatinous precipitate of ferric phosphate. By this means the whole of the iron may be precipitated, and the liquid which was brown then becomes colourless. If this normal salt be dissolved in orthophosphoric acid, the crystalline acid salt FeH_{3}(PO_{4})_{2} is formed. If there be an excess of ferric oxide in the solution, the precipitate will consist of the basic salt. If ferric phosphate be dissolved in hydrochloric acid, and ammonia be added, a salt is precipitated on heating which, after continued washing in water and heating (to remove the water), has the composition Fe_{4}P_{2}O_{11}--that is, 2Fe_{2}O_{3},P_{2}O_{5}. In an aqueous condition this salt may be considered as ferric hydroxide, Fe_{2}(OH)_{6}, in which (OH)_{3} is replaced by the equivalent group PO_{4}. Whenever ammonia is added to a solution containing an excess of ferric salt and a certain amount of phosphoric acid, a precipitate is formed containing the whole of the phosphoric acid in the mass of the ferric oxide.

Ferric oxide is characterised as a feeble base, and also by the fact of its forming double salts--for instance, _potassium iron alum_, which has a composition Fe_{2}(SO_{4})_{3},K_{2}SO_{4},24H_{2}O or FeK(SO_{4})_{2},12H_{2}O. It is obtained in the form of almost colourless or light rose-coloured large octahedra of the regular system by simply mixing solutions of potassium sulphate and the ferric sulphate obtained by dissolving ferric oxide in sulphuric acid.

[25] It would seem that all normal ferric salts are colourless, and that the brown colour which is peculiar to the solutions is really due to basic ferric salts. A remarkable example of the apparent change of colour of salts is represented by the ferrous and ferric oxalates. The former in a dry state has a yellow colour, although as a rule the ferrous salts are green, and the latter is colourless or pale green. When the normal ferric salt is dissolved in water it is, like many salts, probably decomposed by the water into acid and basic salts, and the latter communicates a brown colour to the solution. Iron alum is almost colourless, is easily decomposed by water, and is the best proof of our assertion. The study of the phenomena peculiar to ferric nitrate might, in my opinion, give a very useful addition to our knowledge of the aqueous solutions of salts in general.

[25 bis] The reaction FeX_{3} + KI = FeX_{2} + KX + I proceeds comparatively slowly in solutions, is not complete (depends upon the mass), and is reversible. In this connection we may cite the following data from Seubert and Rohrer's (1894) comprehensive researches. The investigations were conducted with solutions containing 1/10 gram--equivalent weights of Fe_{2}(SO_{4})_{3} (_i.e._ containing 20 grams of salt per litre), and a corresponding solution of KI; the amount of iodine liberated being determined (after the addition of starch) by a solution (also 1/10 normal) of Na_{2}S_{2}O_{3} (_see_ Chapter XX., Note 42). The progress of the reaction was expressed by the amount of liberated iodine in percentages of the theoretical amount. For instance, the following amount of iodide of potassium was decomposed when Fe_{2}(SO_{4})_{3} + 2_n_KI was taken:

_n_ = 1 2 3 6 10 20 After 15´ 11·4 26·3 40·6 73·5 91·6 96·0 " 30´ 14·0 35·8 47·8 78·5 94·3 97·4 " 1 hour 19·0 42·7 56·0 84·0 95·7 97·6 " 10 " 32·6 56·0 75·7 93·2 96·5 97·6 " 48 " 39·4 67·7 82·6 93·4 96·6 97·6

Similar results were obtained for FeCl_{3}, but then the amount of iodine liberated was somewhat greater. Similar results were also obtained by increasing the mass of FeX_{3} per KI, and by replacing it by HI (_see_ Chapter XXI., Note 26).

Iron forms one other oxide besides the ferric and ferrous oxides; this contains twice as much oxygen as the former, but is so very unstable that it can neither be obtained in the free state nor as a hydrate. Whenever such conditions of double decomposition occur as should allow of its separation in the free state, it decomposes into oxygen and ferric oxide. It is known in the state of salts, and is only stable in the presence of alkalis, and forms salts with them which have a decidedly alkaline reaction; it is therefore a feebly acid oxide. Thus when small pieces of iron are heated with nitre or potassium chlorate a potassium salt of the composition K_{2}FeO_{4} is formed, and therefore the hydrate corresponding with this salt should have the composition H_{2}FeO_{4}. It is called _ferric acid_. Its anhydride ought to contain FeO_{3} or Fe_{2}O_{6}--twice as much oxygen as ferric oxide. If a solution of potassium ferrate be mixed with acid, the free hydrate ought to be formed, but it immediately decomposes (2K_{2}FeO_{4} + 5H_{2}SO_{4} = 2K_{2}SO_{4} + Fe_{2}(SO_{4})_{3} + 5H_{2}O + O_{3}), oxygen being evolved. If a small quantity of acid be taken, or if a solution of potassium ferrate be heated with solutions of other metallic salts, ferric oxide is separated--for instance:

2CuSO_{4} + 2K_{2}FeO_{4} = 2K_{2}SO_{4} + O_{3} + Fe_{2}O_{3} + 2CuO.

Both these oxides are of course deposited in the form of hydrates. This shows that not only the hydrate H_{2}FeO_{4}, but also the salts of the heavy metals corresponding with this higher oxide of iron, are not formed by reactions of double decomposition. The solution of potassium ferrate naturally acts as a powerful oxidising agent; for instance, it transforms manganous oxide into the dioxide, sulphurous into sulphuric acid, oxalic acid into carbonic anhydride and water, &c.[26]

[26] If chlorine be passed through a strong solution of potassium hydroxide in which hydrated ferric oxide is suspended, the turbid liquid acquires a dark pomegranate-red colour and contains potassium ferrate: 10KHO + Fe_{2}O_{3} + 3Cl_{2} = 2K_{2}FeO_{4} + 6KCl + 5H_{2}O. The chlorine must not be in excess, otherwise the salt is again decomposed, although the mode of decomposition is unknown; however, ferric chloride and potassium chlorate are probably formed. Another way in which the above-described salt is formed is also remarkable; a galvanic current (from 6 Grove elements) is passed through cast-iron and platinum electrodes into a strong solution of potassium hydroxide. The cast-iron electrode is connected with the positive pole, and the platinum electrode is surrounded by a porous earthenware cylinder. Oxygen would be evolved at the cast-iron electrode, but it is used up in oxidation, and a dark solution of potassium ferrate is therefore formed about it. It is remarkable that the cast iron cannot be replaced by wrought iron.

Iron thus combines with oxygen in three proportions: RO, R_{2}O_{3}, and RO_{3}. It might have been expected that there would be intermediate stages RO_{2} (corresponding to pyrites FeS_{2}) and R_{2}O_{5}, but for iron these are unknown.[26 bis] The lower oxide has a distinctly basic character, the higher is feebly acid. The only one which is stable in the free state is ferric oxide, Fe_{2}O_{3}; the suboxide, FeO, absorbs oxygen, and ferric anhydride, FeO_{3}, evolves it. It is also the same for other elements; the character of each is determined by the relative degree of stability of the known oxides. The salts FeX_{2} correspond with the suboxide, the salts FeX_{3} or Fe_{2}X_{6} with the sesquioxide, and FeX_{6} represents those of ferric acid, as its potassium salt is FeO_{2}(OK)_{2}, corresponding with K_{2}SO_{4}, K_{2}MnO_{4}, K_{2}CrO_{4}, &c. Iron therefore forms compounds of the types FeX_{2}, FeX_{3}, and FeX_{6}, but this latter, like the type NX_{5}, does not appear separately, but only when X represents heterogeneous elements or groups; for instance, for nitrogen in the form of NO_{2}(OH), NH_{4}Cl, &c., for iron in the form of FeO_{2}(OK)_{2}. But still the type FeX_{6} exists, and therefore FeX_{2} and FeX_{3} are compounds which, like ammonia, NH_{3}, are capable of further combinations up to FeX_{6}; this is also seen in the property of ferrous and ferric salts of forming compounds with water of crystallisation, besides double and basic salts, whose stability is determined by the quality of the elements included in the types FeX_{2} and FeX_{3}.[26 tri] It is therefore to be expected that there should be complex compounds derived from ferrous and ferric oxides. Amongst these the series of cyanogen compounds is particularly interesting; their formation and character is not only determined by the property which iron possesses of forming complex types, but also by the similar faculty of the cyanogen compounds, which, like nitriles (Chapter IX.), have clearly developed properties of polymerisation and in general of forming complex compounds.[27]

[26 bis] When Mond and his assistants obtained the remarkable volatile compound Ni(CO)_{4} (described later, Chapter XXII.), it was shown subsequently by Mond and Quincke (1891), and also by Berthelot, that iron, under certain conditions, in a stream of carbonic oxide, also volatilises and forms a compound like that given by nickel. Roscoe and Scudder then showed that when water gas is passed through and kept under pressure (8 atmospheres) in iron vessels a portion of the iron volatilises from the sides of the vessel, and that when the gas is burnt it deposits a certain amount of oxides of iron (the same result is obtained with ordinary coal gas which contains a small amount of CO). To obtain the _volatile compound of iron with carbonic oxide_, Mond prepared a finely divided iron by heating the oxalate in a stream of hydrogen, and after cooling it to 80°-45° he passed CO over the powder. The iron then formed (although very slowly) a volatile compound containing Fe(CO)_{5} (as though it answered to a very high type, FeX_{10}), which when cooled condenses into a liquid (slightly coloured, probably owing to incipient decomposition), sp. gr. 1·47, which solidifies at -21°, boils at about 103°, and has a vapour density (about 6·5 with respect to air) corresponding to the above formula; it decomposes at 180°. Water and dilute acids do not act upon it, but it decomposes under the action of light and forms a hard, non-volatile crystalline yellow compound Fe_{2}(CO)_{7} which decomposes at 80° and again forms Fe(CO)_{5}.

[26 tri] When the molecular Fe_{2}Cl_{6} is produced instead of FeCl_{3} this complication of the type also occurs.

[27] Some light may be thrown upon the faculty of Fe of forming various compounds with CN, by the fact that Fe not only combines with carbon but also with nitrogen. _Nitride of iron_ Fe_{2}N was obtained by Fowler by heating finely powdered iron in a stream of NH_{3} at the temperature of melting lead.

_In the cyanogen compounds of iron_, two degrees might be expected: Fe(CN)_{2}, corresponding with ferrous oxide, and Fe(CN)_{3}, corresponding with ferric oxide. There are actually, however, many other known compounds, intermediate and far more complex. They correspond with the double salts so easily formed by metallic cyanides. The two following double salts are particularly well known, very stable, often used, and easily prepared. _Potassium ferrocyanide_ or _yellow prussiate of potash_, a double salt of cyanide of potassium and ferrous cyanide, has the composition FeC_{2}N_{2},4KCN; its crystals contain 3 mol. of water: K_{4}FeC_{6}N_{6},3H_{2}O. The other is _potassium ferricyanide_ or _red prussiate of potash_. It is also known as _Gmelin's salt_, and contains cyanide of potassium with ferric cyanide; its composition is Fe(CN)_{3},3KCN or K_{3}FeC_{6}N_{6}. Its crystals do not contain water. It is obtained from the first by the action of chlorine, which removes one atom of the potassium. A whole series of other ferrocyanic compounds correspond with these ordinary salts.

Before treating of the preparation and properties of these two remarkable and very stable salts, it must be observed that with ordinary reagents neither of them gives the same double decompositions as the other ferrous and ferric salts, and they both present a series of remarkable properties. Thus these salts have a neutral reaction, are unchanged by air, dilute acids, or water, unlike potassium cyanide and even some of its double salts. When solutions of these salts are treated with caustic alkalis, they do not give a precipitate of ferrous or ferric hydroxides, neither are they precipitated by sodium carbonate. This led the earlier investigators to recognise special independent groupings in them. The yellow prussiate was considered to contain the complex radicle FeC_{6}N_{6} combined with potassium, namely with K_{4}, and K_{3} was attributed to the red prussiate. This was confirmed by the fact that whilst in both salts any other metal, even hydrogen, might be substituted for potassium, the iron remained unchangeable, just as nitrogen in cyanogen, ammonium, and nitrates does not enter into double decomposition, being in the state of the complex radicles CN, NH_{4}, NO_{2}. Such a representation is, however, completely superfluous for the explanation of the peculiarities in the reactions of such compounds as double salts. If a magnesium salt which can be precipitated by potassium hydroxide does not form a precipitate in the presence of ammonium chloride, it is very clear that it is owing to the formation of a soluble double salt which is not decomposed by alkalis. And there is no necessity to account for the peculiarity of reaction of a double salt by the formation of a new complex radicle. In the same way also, in the presence of an excess of tartaric acid, cupric salts do not form a precipitate with potassium hydroxide, because a double salt is formed. These peculiarities are more easily understood in the case of cyanogen compounds than in all others, because all cyanogen compounds, as unsaturated compounds, show a marked tendency to complexity. This tendency is satisfied in double salts. The appearance of a peculiar character in double cyanides is the more easily understood since in the case of potassium cyanide itself, and also in hydrocyanic acid, a great many peculiarities have been observed which are not encountered in those haloid compounds, potassium chloride and hydrochloric acid, with which it was usual to compare cyanogen compounds. These peculiarities become more comprehensible on comparing cyanogen compounds with ammonium compounds. Thus in the presence of ammonia the reactions of many compounds change considerably. If in addition to this it is remembered that the presence of many carbon (organic) compounds frequently completely disturbs the reaction of salts, the peculiarities of certain double cyanides will appear still less strange, because they contain carbon. The fact that the presence of carbon or another element in the compound produces a change in the reactions, may be compared to the action of oxygen, which, when entering into a combination, also very materially changes the nature of reactions. Chlorine is not detected by silver nitrate when it is in the form of potassium chlorate, KClO_{3}, as it is detected in potassium chloride, KCl. The iron in ferrous and ferric compounds varies in its reactions. In addition to the above-mentioned facts, consideration ought to be given to the circumstance that the easy mutability of nitric acid undergoes modification in its alkali salts, and in general the properties of a salt often differ much from those of the acid. Every double salt ought to be regarded as a peculiar kind of saline compound: potassium cyanide is, as it were, a basic, and ferrous cyanide an acid, element. They may be unstable in the separate state, but form a stable double compound when combined together; the act of combination disengages the energy of the elements, and they, so to speak, saturate each other. Of course, all this is not a definite explanation, but then the supposition of a special complex radicle can even less be regarded as such.

Potassium ferrocyanide, K_{4}FeC_{6}N_{6}, is very easily formed by mixing solutions of ferrous sulphate and potassium cyanide. First, a white precipitate of ferrous cyanide, FeC_{2}N_{2}, is formed, which becomes blue on exposure to air, but is soluble in an excess of potassium cyanide, forming the ferrocyanide. The same yellow prussiate is obtained on heating animal nitrogenous charcoal or animal matters--such as horn, leather cuttings, &c.--with potassium carbonate in iron vessels,[27 bis] the mass formed being afterwards boiled with water with exposure to air, potassium cyanide first appearing, which gives yellow prussiate. The animal charcoal may be exchanged for wood charcoal, permeated with potassium carbonate and heated in nitrogen or ammonia; the mass thus produced is then boiled in water with ferric oxide.[28] In this manner it is manufactured on the large scale, and is called 'yellow prussiate' ('prussiate de potasse,' Blutlaugensalz).

[27 bis] The sulphur of the animal refuse here forms the compound FeKS_{2}, which by the action of potassium cyanide yields potassium sulphide, thiocyanate, and ferrocyanide.

[28] Potassium ferrocyanide may also be obtained from Prussian blue by boiling with a solution of potassium hydroxide, and from the ferricyanide by the action of alkalis and reducing substances (because the red prussiate is a product of oxidation produced by the action of chlorine: a ferric salt is reduced to a ferrous salt), &c. In many works (especially in Germany and France) yellow prussiate is prepared from the mass, containing oxide of iron, and employed for purifying coal gas (Vol. I., p. 361), which generally contains cyanogen compounds. About 2 p.c. of the nitrogen contained in coal is converted into cyanogen, which forms Prussian blue and thiocyanates in the mass used for purifying the gas. On evaporation the solution yields large yellow crystals containing 3 molecules of water, which is easily expelled by heating above 100°. 100 parts of water at the ordinary temperature are capable of dissolving 25 parts of this salt; its sp. gr. is 1·83. When ignited it forms potassium cyanide and iron carbide, FeC_{2} (Chapter XIII., Note 12). Oxidising substances change it into potassium ferricyanide. With strong sulphuric acid it gives carbonic oxide, and with dilute sulphuric acid, when heated, prussic acid is evolved according to the equation: 2K_{4}FeC_{6}N_{6} + 3H_{2}SO_{4} = K_{2}Fe_{2}C_{6}N_{6} + 3K_{2}SO_{4} + 6HCN; hence in the yellow prussiate K_{2} replaces Fe.

It is easy to substitute other metals for the potassium in the yellow prussiate. The hydrogen salt or hydroferrocyanic acid, H_{4}FeC_{6}N_{6}, is obtained by mixing strong solutions of yellow prussiate and hydrochloric acid. If ether be added and the air excluded, the acid is obtained directly in the form of a white scarcely crystalline precipitate which becomes blue on exposure to air (as ferrous cyanide does from the formation of blue compounds of ferrous and ferric cyanides, and it is on this account used in cotton printing). It is soluble in water and alcohol, but not in ether, has marked acid properties, and decomposes carbonates, which renders it easily possible to prepare ferrocyanides of the metals of the alkalis and alkaline earths; these are readily soluble, have a neutral reaction, and resemble the yellow prussiate. Solutions of these salts form precipitates with the salts of other metals, because the ferrocyanides of the heavy metals are insoluble. Here either the whole of the potassium of the yellow prussiate, or only a part of it, is exchanged for an equivalent quantity of the heavy metal. Thus, when a cupric salt is added to a solution of yellow prussiate, a red precipitate is obtained which still contains half the potassium of the yellow prussiate:

K_{4}FeC_{6}N_{6} + CuSO_{4} = K_{2}CuFeC_{6}N_{6} + K_{2}SO_{4}.

But if the process be reversed (the salt of copper being then in excess) the whole of the potassium will be exchanged for copper, forming a reddish-brown precipitate, Cu_{2}FeC_{6}N_{6},9H_{2}O. This reaction and those similar to it are very sensitive and may be used for detecting metals in solution, more especially as the colour of the precipitate very often shows a marked difference when one metal is exchanged for another. Zinc, cadmium, lead, antimony, tin, silver, cuprous and aurous salts form _white_ precipitates; cupric, uranium, titanium and molybdenum salts _reddish-brown_; those of nickel, cobalt, and chromium, _green_ precipitates; _with ferrous salts_, ferrocyanide forms, as has been already mentioned, a _white_ precipitate--namely, Fe_{2}FeC_{6}N_{6}, or FeC_{2}N_{2}--which turns blue on exposure to air, and with ferric salts a _blue precipitate_ called _Prussian blue_. Here the potassium is replaced by iron, the reaction being expressed thus: 2Fe_{2}Cl_{6} + 3K_{4}FeC_{6}N_{6} = 12KCl + Fe_{4}Fe_{3}C_{18}N_{18}, the latter formula expressing the composition of Prussian blue. It is therefore the compound 4Fe(CN)_{3} + 3Fe(CN)_{2}. The yellow prussiate is prepared in chemical works on a large scale especially for the manufacture of this blue pigment, which is used for dyeing cloth and other fabrics and also as one of the ordinary blue paints. It is insoluble in water, and the stuffs are therefore dyed by first soaking them in a solution of a ferric salt and then in a solution of yellow prussiate. If however an excess of yellow prussiate be present complete substitution between potassium and iron does not occur, and _soluble Prussian blue_ is formed; KFe_{2}(CN)_{6} = KCN,Fe(CN)_{2},Fe(CN)_{3}. This blue salt is colloidal, is soluble in pure water, but insoluble and precipitated when other salts--for instance, potassium or sodium chloride--are present even in small quantities, and is therefore first obtained as a precipitate.[29]

[29] Skraup obtained this salt both from potassium ferrocyanide with ferric chloride and from ferricyanide with ferrous chloride, which evidently shows that it contains iron in both the ferric and ferrous states. With ferrous chloride it forms Prussian blue, and with ferric chloride Turnbull's blue.

Prussian blue was discovered in the beginning of the last century by a Berlin manufacturer, Diesbach. It was then prepared, as it sometimes is also at present, directly from potassium cyanide obtained by heating animal charcoal with potassium carbonate. The mass thus obtained is dissolved in water, alum is added to the solution in order to saturate the free alkali, and then a solution of green vitriol is added which has previously been sufficiently exposed to the air to contain both ferric and ferrous salts. If the solution of potassium cyanide be mixed with a solution containing both salts, Prussian blue will be formed, because it is a compound of ferrous cyanide, FeC_{2}N_{2}, and ferric cyanide, Fe_{2}C_{6}N_{6}. A ferric salt with potassium ferrocyanide forms a blue colour, because ferrous cyanide is obtained from the first salt and ferric cyanide from the second. During the preparation of this compound alkali must be avoided, as otherwise the precipitate would contain oxides of iron. Prussian blue has not a crystalline structure; it forms a blue mass with a copper-red metallic lustre. Both acids and alkalis act on it. The action is at first confined to the ferric salt it contains. Thus alkalis form ferric oxide and ferrocyanide in solution: 2Fe_{2}C_{6}N_{6},3FeC_{2}N_{2} + 12KHO = 2(Fe_{2}O_{3},3H_{2}O) + 3K_{4}FeC_{6}N_{6}. Various ferrocyanides may thus be prepared. Prussian blue is soluble in an aqueous solution of oxalic acid, forming blue ink. In air, when exposed to the action of light, it fades; but in the dark again absorbs oxygen and becomes blue, which fact is also sometimes noticed in blue cloth. An excess of potassium ferrocyanide renders Prussian blue soluble in water, although insoluble in various saline solutions--that is, it converts it into the soluble variety. Strong hydrochloric acid also dissolves Prussian blue.

Potassium ferricyanide, or _red prussiate_ of potash, K_{3}FeC_{6}N_{6}, is called 'Gmelin's salt,' because this savant obtained it by the action of chlorine on a solution of the yellow prussiate: K_{4}FeC_{6}N_{6} + Cl = K_{3}FeC_{6}N_{6} + KCl. The reaction is due to the ferrous salt being changed by the action of the chlorine into a ferric salt. It separates from solutions in anhydrous, well-formed prisms of a red colour, but the solution has an olive colour; 100 parts of water, at 10°, dissolve 37 parts of the salt, and at 100°, 78 parts.[30] The red prussiate gives a blue precipitate with ferrous salts, called _Turnbull's blue_, very much like Prussian blue (and the soluble variety), because it also contains ferrous cyanide and ferric cyanide, although in another proportion, being formed according to the equation: 3FeCl_{2} + 2K_{3}FeC_{6}N_{6} = 6KCl + Fe_{3}Fe_{2}C_{12}N_{12}, or 3FeC_{2}N_{2},Fe_{2}C_{6}N_{6}; in Prussian blue we have Fe_{7}Cy_{18}, and here Fe_{5}Cy_{12}. A ferric salt ought to form ferric cyanide Fe_{2}C_{6}N_{6}, with red prussiate, but ferric cyanide is soluble, and therefore no precipitate is obtained, and the liquid only becomes brown.[31]

[30] An excess of chlorine must not be employed in preparing this compound, otherwise the reaction goes further. It is easy to find out when the action of the chlorine on potassium ferrocyanide must cease; it is only necessary to take a sample of the liquid and add a solution of a ferric salt to it. If a precipitate of Prussian blue is formed, more chlorine must be added, as there is still some undecomposed ferrocyanide, for the ferricyanide does not give a precipitate with ferric salts. Potassium ferricyanide, like the ferrocyanide, easily exchanges its potassium for hydrogen and various metals by double decomposition. With the salts of tin, silver, and mercury it forms yellow precipitates, and with those of uranium, nickel, cobalt, copper, and bismuth brown precipitates. The lead salt under the action of sulphuretted hydrogen forms lead sulphide and a hydrogen salt or acid, H_{3}FeC_{6}N_{6}, corresponding with potassium ferricyanide, which is soluble, crystallises in red needles, and resembles hydroferrocyanic acid, H_{4}FeC_{6}N_{6}. Under the action of reducing agents--for instance, sulphuretted hydrogen, copper--potassium ferricyanide is changed into ferrocyanide, especially in the presence of alkalis, and thus forms a rather energetic _oxidising agent_--capable, for instance, of changing manganous oxide into dioxide, bleaching tissues, &c.

[31] It is important to mention a series of readily crystallisable salts formed by the action of nitric acid on potassium and other ferrocyanides and ferricyanides. These salt contain the elements of nitric oxide, and are therefore called _nitro-(nitroso) ferricyanides_ (_nitroprussides_). Generally a crystalline sodium salt is obtained, Na_{2}FeC_{5}N_{6}O,2H_{2}O. In its composition this salt differs from the red sodium salt, Na_{3}FeC_{6}N_{6}, by the fact that in it one molecule of sodium cyanide, NaCN, is replaced by nitric oxide, NO. In order to prepare it, potassium ferrocyanide in powder is mixed with five-sevenths of its weight of nitric acid diluted with an equal volume of water. The mixture is at first left at the ordinary temperature, and then heated on a water-bath. Here ferricyanide is first of all formed (as shown by the liquid giving a precipitate with ferrous chloride), which then disappears (no precipitate with ferrous chloride), and forms a green precipitate. The liquid, when cooled, deposits crystals of nitre. The liquid is then strained off and mixed with sodium carbonate, boiled, filtered, and evaporated; sodium nitrate and the salt described are deposited in crystals. It separates in prisms of a red colour. Alkalis and salts of the alkaline earths do not give precipitates: they are soluble, but the salts of iron, zinc, copper, and silver form precipitates where sodium is exchanged with these metals. It is remarkable that the sulphides of the alkali metals give with this salt an intense bright purple coloration. This series of compounds was discovered by Gmelin and studied by Playfair and others (1849).

This series to a certain extent resembles the nitro-sulphide series described by Roussin. Here the primary compound consists of black crystals, which are obtained as follows:--Solutions of potassium hydrosulphide and nitrate are mixed, and the mixture is agitated whilst ferric chloride is added, then boiled and filtered; on cooling, _black crystals_ are deposited, having the composition Fe_{6}S_{3}(NO)_{10},H_{2}O (Rosenberg), or, according to Demel, FeNO_{2},NH_{2}S. They have a slightly metallic lustre, and are soluble in water, alcohol, and ether. They absorb the latter as easily as calcium chloride absorbs water. In the presence of alkalis these crystals remain unchanged, but with acids they evolve nitric oxides. There are several compounds which are capable of interchanging, and correspond with Roussin's salt. Here we enter into the series of the nitrogen compounds which have been as yet but little investigated, and will most probably in time form most instructive material for studying the nature of that element. These series of compounds are as unlike the usual saline compounds of inorganic chemistry as are organic hydrocarbons. There is no necessity to describe these series in detail, because their connection with other compounds is not yet clear, and they have not yet any application.

If chlorine and sodium are representatives of independent groups of elements, the same may also be said of iron. Its nearest analogues show, besides a similarity in character, a likeness as regards physical properties and a proximity in atomic weight. Iron occupies a medium position amongst its nearest analogues, both with respect to properties and faculty of forming saline oxides, and also as regards atomic weight. On the one hand, cobalt, 58, and nickel, 59, approach iron, 56; they are metals of a more basic character, they do not form stable acids or higher degrees of oxidation, and are a transition to copper, 63, and zinc, 65. On the other hand, manganese, 55, and chromium, 52, are the nearest to iron; they form both basic and acid oxides, and are a transition to the metals possessing acid properties. In addition to having atomic weights approximately alike, chromium, manganese, iron, cobalt, nickel, and copper have also nearly the same specific gravity, so that the atomic volumes and the molecules of their analogous compounds are also near to one another (see table at the beginning of this volume). Besides this, the likeness between the above-mentioned elements is also seen from the following:

They form suboxides, RO, fairly energetic bases, isomorphous with magnesia--for instance, the salt RSO_{4},7H_{2}O, akin to MgSO_{4},7H_{2}O, and FeSO_{4},7H_{2}O, or to sulphates containing less water; with alkali sulphates all form double salts crystallising with 6H_{2}O; all are capable of forming ammonium salts, &c. The lower oxides, in the cases of nickel and cobalt, are tolerably stable, are not easily oxidised (the nickel compound with more difficulty than cobalt, a transition to copper); with manganese, and especially with chromium, they are more easily oxidised than with iron and pass into higher oxides. They also form oxides of the form R_{2}O_{3}, and with nickel, cobalt, and manganese this oxide is very unstable, and is more easily reduced than ferric oxide; but, in the case of chromium, it is very stable, and forms the ordinary kind of salts. It is isomorphous with ferric oxide, forms alums, is a feeble base, &c. Chromium, manganese, and iron are oxidised by alkali and oxidising agents, forming salts like Na_{2}SO_{4}; but cobalt and nickel are difficult to oxidise; their acids are not known with any certainty, and are, in all probability, still less stable than the ferrates. Cr, Mn and Fe form compounds R_{2}Cl_{6} which are like Fe_{2}Cl_{6} in many respects; in Co this faculty is weaker and in Ni it has almost disappeared. The cyanogen compounds, especially of manganese and cobalt, are very near akin to the corresponding ferrocyanides. The oxides of nickel and cobalt are more easily reduced to metal than those of iron, but those of manganese and chromium are not reduced so easily as iron, and the metals themselves are not easily obtained in a pure state; they are capable of forming varieties resembling cast iron. The metals Cr, Mn, Fe, Co, and Ni have a grey iron colour and are very difficult to melt, but nickel and cobalt can be melted in the reverberatory furnace and are more fusible than iron, whilst chromium is more difficult to melt than platinum (Deville). These metals decompose water, but with greater difficulty as the atomic weight rises, forming a transition to copper, which does not decompose water. All the compounds of these metals have various colours, which are sometimes very bright, especially in the higher stages of oxidation.

These metals of the iron group are often met with together in nature. Manganese nearly everywhere accompanies iron, and iron is always an ingredient in the ores of manganese. Chromium is found principally as chrome ironstone--that is, a peculiar kind of magnetic oxide in which Fe_{2}O_{3} is replaced by Cr_{2}O_{3}.

Nickel and cobalt are as inseparable companions as iron and manganese. The similarity between them even extends to such remote properties as magnetic qualities. In this series of metals we find those which are the most magnetic: iron, cobalt, and nickel. There is even a magnetic oxide among the chromium compounds, such being unknown in the other series. Nickel easily becomes passive in strong nitric acid. It absorbs hydrogen in just the same way as iron. In short, in the series Cr, Mn, Fe, Co, and Ni, there are many points in common although there are many differences, as will be seen still more clearly on becoming acquainted with cobalt and nickel.

In nature _cobalt_ is principally found in combination with arsenic and sulphur. _Cobalt arsenide_, or _cobalt speiss_, CoAs_{2}, is found in brilliant crystals of the regular system, principally in Saxony. _Cobalt glance_, CoAs_{2}CoS_{2}, resembles it very much, and also belongs to the regular system; it is found in Sweden, Norway, and the Caucasus. _Kupfernickel_ is a nickel ore in combination with arsenic, but of a different composition from cobalt arsenide, having the formula NiAs; it is found in Bohemia and Saxony. It has a copper-red colour and is rarely crystalline; it is so called because the miners of Saxony first mistook it for an ore of copper (_Kupfer_), but were unable to extract copper from it. _Nickel glance_, NiS_{2},NiAs_{2}, corresponding with cobalt glance, is also known. Nickel accompanies the ores of cobalt and cobalt those of nickel, so that both metals are found together. The ores of cobalt are worked in the Caucasus in the Government of Elizavetopolsk. Nickel ores containing aqueous hydrated nickel silicate are found in the Ural (Revdansk). Large quantities of a similar ore are exported into Europe from New Caledonia. Both ores contain about 12 per cent. Ni. _Garnierite_, (RO)_{5}(SiO_{2})_{4}1-1/2H_{2}O, where R = Ni and Mg, predominates in the New Caledonian ore. Large deposits of nickel have been discovered in Canada, where the ore (as nickelous pyrites) is free from arsenic. Cobalt is principally worked up into cobalt compounds, but nickel is generally reduced to the metallic state, in which it is now often used for alloys--for instance, for coinage in many European States, and for plating other metals, because it does not oxidise. Cobalt arsenide and cobalt glance are principally used for the preparation of cobalt compounds; they are first sorted by discarding the rocky matter, and then roasted. During this process most of the sulphur and arsenic disappears; the arsenious anhydride volatilises with the sulphurous anhydride and the metal also oxidises.[32] It is a simple matter to obtain nickel and cobalt from their oxides. In order to obtain the latter, solutions of their salts are treated with sodium carbonate and the precipitated carbonates are heated; the suboxides are thus obtained, and these latter are reduced in a stream of hydrogen, or even by heating with ammonium chloride. They easily oxidise when in the state of powder. When the chlorides of nickel and cobalt are heated in a stream of hydrogen, the metal is deposited in brilliant scales. _Nickel is always much more easily and quickly reduced than cobalt._ Nickel melts more easily than cobalt, and this even furnishes a means of testing the heating powers of a reverberatory furnace. Cobalt fuses at a temperature only a little lower than that at which iron does. In general, cobalt is nearer to iron than nickel, nickel being nearer to copper.[32 bis] Both nickel and cobalt have magnetic properties like iron, but Co is less magnetic than Fe, and Ni still less so. The specific gravity of nickel reduced by hydrogen is 9·1 and that of cobalt 8·9. Fused cobalt has a specific gravity of 8·5, the density of ordinary nickel being almost the same. Nickel has a greyish silvery-white colour; it is brilliant and very ductile, so that the finest wire may be easily drawn from it. This wire has a resistance to tension equal to iron wire. The beautiful colour of nickel, and the high polish which it is capable of receiving and retaining, as it does not oxidise, render it a useful metal for many purposes, and in many ways it resembles silver.[32 tri] It is now very common to cover other metals with a layer of nickel (nickel plating). This is done by a process of electro-plating, using a solution of a nickel salt. The colour of cobalt is dark and redder; it is also ductile, and has a greater tensile resistance than iron. Dilute acids act very slowly on nickel and cobalt; nitric acid may be considered as the best solvent for them. The solutions in every case contain salts corresponding with the ferrous salts--that is, the _salts_ CoX_{2}, NiX_{2}, _correspond with the suboxides_ of these metals. These salts in their types are similar to the magnesium salts. The salts of nickel when crystallising with water have a green colour, and form bright green solutions, but in the anhydrous state they most frequently have a yellow colour. The salts of cobalt are generally rose-coloured, and generally blue when in the anhydrous state. Their aqueous solutions are rose-coloured. Cobaltous chloride is easily soluble in alcohol, and forms a solution of an intense blue colour.[33]

[32] The residue from the roasting of cobalt ores is called _zafflor_, and is often met with in commerce. From this the purer compounds of cobalt may be prepared. The ores of nickel are also first roasted, and the oxides dissolved in acid, nickelous salts being then obtained.

The further treatment of cobalt and nickel ores is facilitated if the arsenic can be almost entirely removed, which may be effected by roasting the ore a second time with a small addition of nitre and sodium carbonate; the nitre combines with the arsenic, forming an arsenious salt, which may be extracted with water. The remaining mass is dissolved in hydrochloric acid, mixed with a small quantity of nitric acid. Copper, iron, manganese, nickel, cobalt, &c., pass into solution. By passing hydrogen sulphide through the solution, copper, bismuth, lead, and arsenic are deposited as metallic sulphides; but iron, cobalt, nickel, and manganese remain in solution. If an alkaline solution of bleaching powder be then added to the remaining solution, the whole of the manganese will first be deposited in the form of dioxide, then the cobalt as hydrated cobaltic oxide, and finally the nickel also. It is, however, impossible to rely on this method for effecting a complete separation, the more so since the higher oxides of the three above-mentioned metals have all a black colour; but, after a few trials, it will be easy to find how much bleaching powder is required to precipitate the manganese, and the amount which will precipitate all the cobalt. The manganese may also be separated from cobalt by precipitation from a mixture of the solutions of both metals (in the form of the 'ous' salts) with ammonium sulphide, and then treating the precipitate with acetic acid or dilute hydrochloric acid, in which manganese sulphide is easily soluble and cobalt sulphide almost insoluble. Further particulars relating to the separation of cobalt from nickel may be found in treatises on analytical chemistry. In practice it is usual to rely on the rough method of separation founded on the fact that nickel is more easily reduced and more difficult to oxidise than cobalt. The New Caledonian ore is smelted with CaSO_{4} and CaCO_{3} on coke, and a metallic regulus is obtained containing all the Ni, Fe, and S. This is roasted with SiO_{2}, which converts all the iron into slag, whilst the Ni remains combined with the S; this residue on further roasting gives NiO, which is reduced by the carbon to metallic Ni. The Canadian ore (a pyrites containing 11 p.c. Ni) is frequently treated in America (after a preliminary dressing) by smelting it with Na_{2}SO_{4} and charcoal; the resultant fusible Na_{2}S then dissolves the CuS and FeS_{2}, while the NiS is obtained in a bottom layer (Bartlett and Thomson's process) from which Ni is obtained in the manner described above.

For manufacturing purposes somewhat impure cobalt compounds are frequently used, which are converted into _smalt_. This is glass containing a certain amount of cobalt oxide; the glass acquires a bright blue colour from this addition, so that when powdered it may be used as a blue pigment; it is also unaltered at high temperatures, so that it used to take the place now occupied by Prussian blue, ultramarine, &c. At present smalt is almost exclusively used for colouring glass and china. To prepare smalt, ordinary impure cobalt ore (zaffre) is fused in a crucible with quartz and potassium carbonate. A fused mass of cobalt glass is thus formed, containing silica, cobalt oxide, and potassium oxide, and a metallic mass remains at the bottom of the crucible, containing almost all the other metals, arsenic, nickel, copper, silver, &c. This metallic mass is called _speiss_, and is used as nickel ore for the extraction of nickel. Smalt usually contains 70 p.c. of silica, 20 p.c. of potash and soda, and about 5 to 6 p.c. of cobaltous oxide; the remainder consisting of other metallic oxides.

[32 bis] All we know respecting the relations of Co and Ni to Fe and Cu confirms the fact that Co is more closely related to Fe and Ni to Cu; and as the atomic weight of Fe = 56 and of Cu = 63, then according to the principles of the periodic system it would be expected that the atomic weight of Co would be about 59-60, whilst that of Ni should be greater than that of Co but less than that of Cu, _i.e._ about 50·5-60·5. However, as yet the majority of the determinations of the atomic weights of Co and Ni give a different result and show that a lower atomic weight is obtained for Ni than for Co. Thus K. Winkler (1894) obtained (employing metals deposited electrolytically and determining the amount of iodine which combined with them) Ni = 58·72 and Co = 59·37 (if H = 1 and I = 126·53). In my opinion this should not be regarded as proving that the principles of the periodic system cannot be applied in this instance, nor as a reason for altering the position of these elements in the system (_i.e._ by placing Ni after Fe, and Co next to Cu), because in the first place the figures given by different chemists (for instance, Zimmermann, Krüss, and others) are somewhat divergent, and in the second place the majority of the latest modes of determining the atomic weights of Co and Ni aim at finding what weights of these metals react with known weights of other elements without taking into account the faculty they have of absorbing hydrogen; since this faculty is more developed in Ni than in Co the hydrogen (occluded in Ni) should lower the atomic weight of Ni more than that of Co. On the whole, the question of the atomic weights of Co and Ni cannot yet be considered as decided, notwithstanding the numerous researches which have been made; still there can be no doubt that the atomic weights of these two metals are very nearly equal, and greater than that of Fe, but less than that of Cu. This question is of great interest, not only for completing our knowledge of these metals, but also for perfecting our knowledge of the periodic system of the elements.

[32 tri] For instance, the alkalis may be fused in nickel vessels as well as in silver, because they have no action upon either metal. Nickel, like silver, is not acted upon by dilute acids. Only nitric acid dissolves both metals well. Nickel is harder, and fuses at a higher temperature than silver. For castings, a small quantity of magnesium (0·001 part by weight) is added to nickel to render it more homogeneous (just as aluminium is added to steel). Nickel forms many valuable alloys. Steel containing 3 p.c. Ni is particularly valuable, its limit of elasticity is higher and its hardness is greater; it is used for armour plate and other large pieces. The alloys of nickel, especially with copper and zinc (melchior, _see_ later), aluminium and silver, although used in certain cases, are now replaced by nickel-plated or nickel-deposited goods (deposited by electricity from a solution of the ammonium salts).

[33] The change of colour is dependent in all probability on the combination with water, or according to others on polymeric transformation. It enables a solution of cobalt chloride to be used as sympathetic ink. If something be written with cobalt chloride on white paper, it will be invisible on account of the feeble colour of the solution, and when dry nothing can be distinguished. If, however, the paper be heated before the fire, the rose-coloured salt will be changed into a less hydrous blue salt, and the writing will become quite visible, but fade away when cool.

The change of colour which takes place in solutions of CoCl_{2} under the influence not only of solution in water or alcohol, but also of a change of temperature, is a characteristic of all the halogen salts of cobalt. Crystalline iodide of cobalt, CoI_{2}6H_{2}O, gives a dark red solution between -22° and +20°; above +20° the solution turns brown and passes from olive to green, from +35° to 320° the solution remains green. According to Étard the change of colour is due to the fact that at first the solution contains the hydrate CoI_{2}H_{2}O, and that above 35° it contains CoI_{2}4H_{2}O. These hydrates can be crystallised from the solutions; the former at ordinary temperature and the latter on heating the solution. The intermediate olive colour of the solutions corresponds to the incipient decomposition of the hexahydrated salt and its passage into CoI_{2}4H_{2}O. A solution of the hexahydrated chloride of cobalt, CoCl_{2}6H_{2}O, is rose-coloured between -22° and +25°; but the colour changes starting from +25°, and passes through all the tints between red and blue right up to 50°; a true blue solution is only obtained at 55° and remains up to 300°. This true blue solution contains another hydrate, CoCl_{2}2H_{2}O.

The dependence between the solubility of the iodide and chloride of cobalt and the temperature is expressed by two almost straight lines corresponding to the hexa- and di-hydrates; the passage of the one into the other hydrate being expressed by a curve. The same character of phenomena is seen also in the variation of the vapour tension of solutions of chloride of cobalt with the temperature. We have repeatedly seen that aqueous solutions (for instance, Chapter XXII., Note 23 for Fe_{2}Cl_{6}) deposit different crystallo-hydrates at different temperatures, and that the amount of water in the hydrate decreases as the temperature _t_ rises, so that it is not surprising that CoCl_{2}2H_{2}O (or according to Potilitzin CoCl_{2}H_{2}O) should separate out above 55° and CoCl_{2}6H_{2}O at 25° and below. Nor is it exceptional that the colour of a salt varies according as it contains different amounts of H_{2}O. But in this instance it is characteristic that the change of colour takes place in solution in the presence of an excess of water. This apparently shows that the actual solution may contain either CoCl_{2}6H_{2}O or CoCl_{2}2H_{2}O. And as we know that a solution may contain both metaphosphoric PHO_{3} and orthophosphoric acid H_{3}PO_{4} = HPO_{3} + H_{2}O, as well as certain other anhydrides, the question of the state of substances in solutions becomes still more complicated.

Nickel sulphate crystallises from neutral solutions at a temperature of from 15° to 20° in _rhombic_ crystals containing 7H_{2}O. Its form approaches very closely to that of the salts of zinc and magnesium. The planes of a vertical prism for magnesium salts are inclined at an angle of 90° 30´, for zinc salts at an angle of 91° 7´, and for nickel salts at an angle of 91° 10´. Such is also the form of the zinc and magnesium selenates and chromates. Cobalt sulphate containing 7 molecules of water is deposited in crystals of the _monoclinic_ system, like the corresponding salts of iron and manganese. The angle of a vertical prism for the iron salt = 82° 20´, for cobalt = 82° 22´, and the inclination of the horizontal pinacoid to the vertical prism for the iron salt = 99° 2´, and for the cobalt salt 99° 36´. All the isomorphous mixtures of the salts of magnesium, iron, cobalt, nickel and manganese have the same form if they contain 7 mol. H_{2}O and iron or cobalt predominate, whilst if there is a preponderance of magnesium, zinc, or nickel, the crystals have a rhombic form like magnesium sulphate. Hence these sulphates are _dimorphous_, but for some the one form is more stable and for others the other. Brooke, Moss, Mitscherlich, Rammelsberg, and Marignac have explained these relations. Brooke and Mitscherlich also supposed that NiSO_{4},7H_{2}O is not only capable of assuming these forms, but also that of the _tetragonal_ system, because it is deposited in this form from acid, and especially from slightly-heated solutions (30° to 40°). But Marignac demonstrated that the tetragonal crystals do not contain 7, but 6, molecules of water, NiSO_{4},6H_{2}O. He also observed that a solution evaporated at 50° to 70° deposits monoclinic crystals, but of a different form from ferrous sulphate, FeSO_{4},7H_{2}O--namely, the angle of the prism is 71° 52´, that of the pinacoid 95° 6´. This salt appears to be the same with 6 molecules of water as the tetragonal. Marignac also obtained magnesium and zinc salts with 6 molecules of water by evaporating their solutions at a higher temperature, and these salts were found to be isomorphous with the monoclinic nickel salt. In addition to this it must be observed that the rhombic crystals of nickel sulphate with 7H_{2}O become turbid under the influence of heat and light, lose water, and change into the tetragonal salt. The monoclinic crystals in time also become turbid, and change their structure, so that the tetragonal form of this salt is the most stable. Let us also add that nickel sulphate in all its shapes forms very beautiful emerald green crystals, which, when heated to 230°, assume a dirty greenish-yellow hue and then contain one molecule of water.

Klobb (1891) and Langlot and Lenoir obtained anhydrous CoSO_{4} and NiSO_{4} by igniting the hydrated salt with (NH_{4})_{2}SO_{4} until the ammonium salt had completely volatilised and decomposed.

We may add that when equivalent aqueous solutions of NiX_{2} (green) and CoX_{2} (red) are mixed together they give an almost colourless (grey) solution, in which the green and red colour of the component parts disappears owing to the combination of the complementary colours.

A double salt NiKF_{3} is obtained by heating NiCl_{2} with KFHF in a platinum crucible; KCoF_{3} is formed in a similar manner. The nickel salt occurs in fine green plates, easily soluble in water but scarcely soluble in ethyl and methyl alcohol. They decompose into green oxide of nickel and potassium fluoride when heated in a current of air. The analogous salt of cobalt crystallises in crimson flakes.

If instead of potassium fluoride, CoCl_{2} or NiCl_{2} be fused with ammonium fluoride, they also form double salts with the latter. This gives the possibility of obtaining anhydrous fluorides NiF_{2} and CoF_{2}. Crystalline fluoride of nickel, obtained by heating the amorphous powder formed by decomposing the double ammonium salt in a stream of hydrofluoric acid, occurs in beautiful green prisms, sp. gr. 4·63, which are insoluble in water, alcohol, and ether; sulphuric, hydrochloric, and nitric acids also have no action upon them, even when heated; NiF_{2} is decomposed by steam, with the formation of black oxide, which retains the crystalline structure of the salt. Fluoride of cobalt, obtained as a rose-coloured powder by decomposing the double ammonium salt with the aid of heat in a stream of hydrofluoric acid, fuses into a ruby-coloured mass which bears distinct signs of a crystalline structure; sp. gr. 4·43. The molten salt only volatilises at about 1400°, which forms a clear distinction between CoF_{2} and the volatile NiF_{2}. Hydrochloric, sulphuric, and nitric acids act upon CoF_{2} even in the cold, although slowly, while when heated the reaction proceeds rapidly (Poulenc, 1892).

If a solution of potassium hydroxide be added to a solution of a cobalt salt, a blue precipitate of the basic salt will be formed. If a solution of a cobalt salt be heated almost to the boiling-point, and the solution be then mixed with a boiling solution of an alkali hydroxide, a _pink precipitate of cobaltous hydroxide_, CoH_{2}O_{2}, will be formed. If air be not completely excluded during the precipitation by boiling, the precipitate will also contain brown cobaltic hydroxide formed by the further oxidation of the cobaltous oxide.[34] Under similar circumstances nickel salts form _a green precipitate of nickelous hydroxide_, the formation of which is not hindered by the presence of ammonium salts, but in that case only requires more alkali to completely separate the nickel. The nickelous oxide obtained by heating the hydroxide, or from the carbonate or nitrate, is a grey powder, easily soluble in acids and easily reduced, but the same substance may be obtained in the crystalline form as an ordinary product from the ores; it crystallises in regular octahedra, with a metallic lustre, and is of a grey colour. In this state the nickelous oxide almost resists the action of acids.[34 bis]

[34] Hydrated suboxide of cobalt (de Schulten, 1889) is obtained in the following manner. A solution of 10 grams of CoCl_{2}6H_{2}O in 60 c.c. of water is heated in a flask with 250 grams of caustic potash and a stream of coal gas is passed through the solution. When heated the hydrate of the suboxide of cobalt which separates out, dissolves in the caustic potash and forms a dark blue solution. This solution is allowed to stand for 24 hours in an atmosphere of coal gas (in order to prevent oxidation). The crystalline mass which separates out has a composition Co(OH)_{2}, and to the naked eye appears as a violet powder, which is seen to be crystalline under the microscope. The specific gravity of this hydrate is 3·597 at 15°. It does not undergo change in the air; warm acetic acid dissolves it, but it is insoluble in warm and cold solutions of ammonia and sal-ammoniac.

[34 bis] The following reaction may be added to those of the cobaltous and nickelous salts: potassium cyanide forms a precipitate with cobalt salts which is soluble in an excess of the reagent and forms a green solution. On heating this and adding a certain quantity of acid, a double _cobalt cyanide_ is formed which corresponds with potassium ferricyanide. Its formation is accompanied with the evolution of hydrogen, and is founded upon the property which cobalt has of oxidising in an alkaline solution, the development of which has been observed in such a considerable measure in the cobaltamine salts. The process which goes on here may be expressed by the following equation; CoC_{2}N_{2} + 4KCN first forms CoK_{4}C_{6}N_{6}, which salt with water, H_{2}O, forms potassium hydroxide, KHO, hydrogen, H, and the salt, K_{3}CoC_{6}N_{6}. Here naturally the presence of the acid is indispensable in consequence of its being required to combine with the alkali. From aqueous solutions this salt crystallises in transparent, hexagonal prisms of a yellow colour, easily soluble in water. The reactions of double decomposition, and even the formation of the corresponding acid, are here completely the same as in the case of the ferricyanide. If a nickelous salt be treated in precisely the same manner as that just described for a salt of cobalt, decomposition will occur.

It is interesting to note _the relation_ of the cobaltous and nickelous hydroxides _to ammonia_; aqueous ammonia dissolves the precipitate of cobaltous and nickelous hydroxide. The blue ammoniacal solution of nickel resembles the same solution of cupric oxide, but has a somewhat reddish tint. It is characterised by the fact that it dissolves silk in the same way as the ammoniacal cupric oxide dissolves cellulose. Ammonia likewise dissolves the precipitate of cobaltous hydroxide, forming a brownish liquid, which becomes darker in air and finally assumes a bright red hue, absorbing oxygen. The admixture of ammonium chloride prevents the precipitation of cobalt salts by ammonia; when the ammonia is added, a brown solution is obtained from which, as in the case of the preceding solution, potassium hydroxide does not separate the cobaltous oxide. Peculiar compounds are produced in this solution; they are comparatively stable, containing ammonia and an excess of oxygen; they bear the name cobaltoamine and cobaltiamine salts. They have been principally investigated by Genth, Frémy, Jörgenson and others. Genth found that when a cobalt salt, mixed with an excess of ammonium chloride, is treated with ammonia and exposed to the air, after a certain lapse of time, on adding hydrochloric acid and boiling, a red powder is precipitated and the remaining solution contains an orange salt. The study of these compounds led to the discovery of a whole series of similar salts, some of which correspond with particular higher degrees of oxidation of cobalt, which are described later.[35] Nickel does not possess this property of absorbing the oxygen of the air when in an ammoniacal solution. In order to understand this distinction, and in general the relation of nickel, it is important to observe that cobalt more easily forms a higher degree of oxidation--namely, _sesquioxide of cobalt_, _cobaltic oxide_, Co_{2}O_{3}--than nickel, especially in the presence of hypochlorous acid. If a solution of a cobalt salt be mixed with barium carbonate and an excess of hypochlorous acid be added, or chlorine gas be passed through it, then at the ordinary temperature on shaking, the whole of the cobalt will be separated in the form of black cobaltic oxide: 2CoSO_{4} + ClHO + 2BaCO_{3} = Co_{2}O_{3} + 2BaSO_{4} + HCl + 2CO_{2}. Under these circumstances nickelous oxide does not immediately form black sesquioxide, but after a considerable space of time it also separates in the form of sesquioxide, Ni_{2}O_{3}, but always later than cobalt. This is due to the relative difficulty of further oxidation of the nickelous oxide. It is, however, possible to oxidise it; if, for instance, the hydroxide NiH_{2}O_{2} be shaken in water and chlorine gas be passed through it, then nickel chloride will be formed, which is soluble in water, and insoluble nickelic oxide in the form of a black precipitate: 3NiH_{2}O_{2} + Cl_{2} = NiCl_{2} + Ni_{2}O_{3},3H_{2}O. Nickelic oxide may also be obtained by adding sodium hypochlorite mixed with alkali to a solution of a nickel salt. Nickelic and cobaltic hydrates are black. Nickelic oxide evolves oxygen with all acids, and in consequence of this it is not separated as a precipitate in the presence of acids; thus it evolves chlorine with hydrochloric acid, exactly like manganese dioxide. When nickelic oxide is dissolved in aqueous ammonia it liberates nitrogen, and an ammoniacal solution of nickelous oxide is formed. When heated, nickelic oxide loses oxygen, forming nickelous oxide. Cobaltic oxide, Co_{2}O_{3}, exhibits more stability than nickelic oxide, and shows feeble basic properties; thus it is dissolved in acetic acid without the evolution of oxygen.[35 bis] But ordinary acids, especially on heating, evolve oxygen, forming a solution of a cobaltous salt. The presence of a cobaltic salt in a solution of a cobaltous salt may be detected by the brown colour of the solution and the black precipitate formed by the addition of alkali, and also from the fact that such solutions evolve chlorine when heated with hydrochloric acid. Cobaltic oxide may not only be prepared by the above-mentioned methods, but also by heating cobalt nitrate, after which a steel-coloured mass remains which retains traces of nitric acid, but when heated further to incandescence evolves oxygen, leaving a compound of cobaltic and cobaltous oxides, similar to magnetic ironstone. Cobalt (but not nickel) undoubtedly forms besides Co_{2}O_{3} a _dioxide_ CoO_{2}. This is obtained[36] when the cobaltous oxide is oxidised by iodine or peroxide of barium.[37]

[35] The cobalt salts may be divided into at least the following classes, which repeat themselves for Cr, Ir, Rh (we shall not stop to consider the latter, particularly as they closely resemble the cobalt salts):--

(_a_) _Ammonium cobalt salts_, which are simply direct compounds of the cobaltous salts CoX_{2} with ammonia, similar to various other compounds of the salts of silver, copper, and even calcium and magnesium, with ammonia. They are easily crystallised from an ammoniacal solution, and have a pink colour. Thus, for instance, when cobaltous chloride in solution is mixed with sufficient ammonia to redissolve the precipitate first formed, octahedral crystals are deposited which have a composition CoCl_{2},H_{2}O,6NH_{3}. These salts are nothing else but combinations with ammonia of crystallisation--if it may be so termed--likening them in this way to combinations with water of crystallisation. This similarity is evident both from their composition and from their capability of giving off ammonia at various temperatures. The most important point to observe is that all these salts contain 6 molecules of ammonia to 1 atom of cobalt, and this ammonia is held in fairly stable connection. Water decomposes these salts. (Nickel behaves similarly without forming other compounds corresponding to the true cobaltic.)

(_b_) The solutions of the above-mentioned salts are rendered turbid by the action of the air; they absorb oxygen and become covered with a crust of _oxycobaltamine salts_. The latter are sparingly soluble in aqueous ammonia, have a brown colour, and are characterised by the fact that with warm water _they evolve oxygen_, forming salts of the following category: The nitrate may be taken as an example of this kind of salt; its composition is CoN_{2}O_{7},5NH_{3},H_{2}O. It differs from cobaltous nitrate, Co(NO_{3})_{2}, in containing an extra atom of oxygen--that is, it corresponds with cobalt dioxide, CoO_{2}, in the same way that the first salts correspond with cobaltous oxide; they contain 5, and not 6, molecules of ammonia, as if NH_{3} had been replaced by O, but we shall afterwards meet compounds containing either 5NH_{3} or 6NH_{3} to each atom of cobalt.

(_c_) _The luteocobaltic salts_ are thus called because they have a yellow (luteus) colour. They are obtained from the salts of the first kind by submitting them in dilute solution to the action of the air; in this case salts of the second kind are not formed, because they are decomposed by an excess of water, with the evolution of oxygen and the formation of luteocobaltic salts. By the action of ammonia the salts of the fifth kind (roseocobaltic) are also converted into luteocobaltic salts. These last-named salts generally crystallise readily, and have a yellow colour; they are comparatively much more stable than the preceding ones, and even for a certain time resist the action of boiling water. Boiling aqueous potash liberates ammonia and precipitates hydrated cobaltic oxide, Co_{2}O_{3},3H_{2}O, from them. This shows that the luteocobaltic salts correspond with cobaltic oxide, Co_{2}O_{3}, and those of the second kind with the dioxide. When a solution of luteocobaltic sulphate, Co_{2}(SO_{4})_{3},12NH_{3},4H_{2}O, is treated with baryta, barium sulphate is precipitated, and the solution contains luteocobaltic hydroxide, Co(OH)_{3},6NH_{3}, which is soluble in water, is powerfully alkaline, absorbs the oxygen of the air, and when heated is decomposed with the evolution of ammonia. This compound therefore corresponds to a solution of cobaltic hydroxide in ammonia. The luteocobaltic salts contain 2 atoms of cobalt and 12 molecules of ammonia--that is, 6NH_{3} to each atom of cobalt, like the salts of the first kind. The CoX_{2} salts have a metallic taste, whilst those of luteocobalt and others have a purely saline taste, like the salts of the alkali metals. In the luteo-salts all the X's react (are ionised, as some chemists say) as in ordinary salts--for instance, all the Cl_{2} is precipitated by a solution of AgNO_{3}; all the (SO_{4})_{3} gives a precipitate with BaX_{2}, &c. The double salt formed with PtCl_{4} is composed in the same manner as the potassium salt, K_{2}PtCl_{4} = 2KCl + PtCl_{4}, that is, contains (CoCl_{3},6NH_{3})_{2},3PtCl_{4}, or the amount of chlorine in the PtCl_{4} is double that in the alkaline salt. In the rosepentamine (_e_), and rosetetramine (_f_), salts, also all the X's react or are ionised, but in the (_g_) and (_h_) salts only a portion of the X's react, and they are equal to the (_e_) and (_f_) salts minus water; this means that although the water dissolves them it is not combined with them, as PHO_{3} differs from PH_{3}O_{3}; phenomena of this class correspond exactly to what has been already (Chapter XXI., Note 7) mentioned respecting the green and violet salts of oxide of chromium.

(_d_) _The fuscocobaltic salts._ An ammoniacal solution of cobalt salts acquires a brown colour in the air, due to the formation of these salts. They are also produced by the decomposition of salts of the second kind; they crystallise badly, and are separated from their solutions by addition of alcohol or an excess of ammonia. When boiled they give up the ammonia and cobaltic oxide which they contain. Hydrochloric and nitric acids give a yellow precipitate with these salts, which turns red when boiled, forming salts of the next category. The following is an example of the composition of two of the fuscocobaltic salts, Co_{2}O(SO_{4})_{2},8NH_{3},4H_{2}O and Co_{2}OCl_{4},8NH_{3},3H_{2}O. It is evident that the fuscocobaltic salts are ammoniacal compounds of basic cobaltic salts. The normal cobaltic sulphate ought to have the composition Co_{2}(SO_{4})_{3} = Co_{2}O_{3},3SO_{3}; the simplest basic salts will be Co_{2}O(SO_{4})_{2} = Co_{2}O_{3})2SO_{3}, and Co_{2}O_{2}(SO_{4}) = Co_{2}O_{3},SO_{3}. The fuscocobaltic salts correspond with the first type of basic salts. They are changed (in concentrated solutions) into oxycobaltamine salts by absorption of one atom of oxygen, Co_{2}O_{2}(SO_{4})_{2}. The whole process of oxidation will be as follows: first of all Co_{2}X_{4}, a cobaltous salt, is in the solution (X a univalent haloid, 2 molecules of the salt being taken), then Co_{2}OX_{4}, the basic cobaltic salt (4th series), then Co_{2}O_{2}X_{4}, the salt of the dioxide (2nd series). The series of basic salts with an acid, 2HX, forms water and a normal salt, Co_{2}X_{6} (in 3, 5, 6 series). These salts are combined with various amounts of water and ammonia. Under many conditions the salts of fuscocobalt are easily transformed into salts of the next series. The salts of the series that has just been described contain 4 molecules of ammonia to 1 atom of cobalt.

(_e_) _The roseocobaltic_ (or rosepentamine), CoX_{2}H_{2}O,5NH_{3}, _salts_, like the luteocobaltic, correspond with the normal cobaltic salts, but contain less ammonia, and an extra molecule of water. Thus the sulphate is obtained from cobaltous sulphate dissolved in ammonia and left exposed to the air until transformed into a brown solution of the fuscocobaltic salt; when this is treated with sulphuric acid a crystalline powder of the roseocobaltic salt, Co_{2}(SO_{4})_{3},10NH_{3},5H_{2}O, separates. The formation of this salt is easily understood: cobaltous sulphate in the presence of ammonia absorbs oxygen, and the solution of the fuscocobaltic salt will therefore contain, like cobaltous sulphate, one part of sulphuric acid to every part of cobalt, so that the whole process of formation may be expressed by the equation: 10NH_{3} + 2CoSO_{4} + H_{2}SO_{4} + 4H_{2}O + O = Co_{2}(SO_{4})_{3},10NH_{3},5H_{2}O. This salt forms tetragonal crystals of a red colour, slightly soluble in cold, but readily soluble in warm water. When the sulphate is treated with baryta, roseocobaltic hydroxide is formed in the solution, which absorbs the carbonic anhydride of the air. It is obtained from the next series by the action of alkalis.

(_f_) The _rosetetramine cobaltic salts_ CoCl_{2},2H_{2}O,4NH_{3} were obtained by Jörgenson, and belong to the type of the luteo-salts, only with the substitution of 2NH_{3} for H_{2}O. Like the luteo- and roseo-salts they give double salts with PtCl_{4}, similar to the alkaline double salts, for instance (Co_{2}H_{2}O,4NH_{3})2(SO_{4})_{2}Cl_{2}PtCl_{4}. They are darker in colour than the preceding, but also crystallise well. They are formed by dissolving CoCO_{3} in sulphuric acid (of a given strength), and after NH_{3} and carbonate of ammonium have been added, air is passed through the solution (for oxidation) until the latter turns red. It is then evaporated with lumps of carbonate of ammonium, filtered from the precipitate and crystallised. A salt of the composition Co_{2}(CO_{3})_{2}(SO_{4}),(2H_{2}O,4NH_{3})_{2} is thus obtained, from which the other salts may be easily prepared.

(_g_) The _purpureocobaltic salts_, CoX_{3},5NH_{3}, are also products of the direct oxidation of ammoniacal solutions of cobalt salts. They are easily obtained by heating the roseocobaltic and luteo-salts with strong acids. They are to all effects the same as the roseocobaltic salts, only anhydrous. Thus, for instance, the purpureocobaltic chloride, Co_{2}Cl_{6},10NH_{3}, or CoCl_{3},5NH_{3}, is obtained by boiling the oxycobaltamine salts with ammonia. There is the same distinction between these salts and the preceding ones as between the various compounds of cobaltous chloride with water. In the purpureocobaltic only X_{2} out of the X_{3} react (are ionised). To the rosetetramine salts (_f_) there correspond the _purpureotetramine_ salts, CoX_{3}H_{2}O,4NH_{3}. The corresponding chromium purpureopentamine salt, CrCl_{3},5NH_{3} is obtained with particular ease (Christensen, 1893). Dry anhydrous chromium chloride is treated with anhydrous liquid ammonia in a freezing mixture composed of liquid CO_{2} and chlorine, and after some time the mixture is taken out of the freezing mixture, so that the excess of NH_{3} boils away; the violet crystals then immediately acquire the red colour of the salt, CrCl_{3},5NH_{3}, which is formed. The product is washed with water (to extract the luteo-salt, CrCl_{3},6NH_{3}), which does not dissolve the salt, and it is then recrystallised from a hot solution of hydrochloric acid.

(_h_) The _praseocobaltic salts_, CoX_{3},4NH_{3}, are green, and form, with respect to the rosetetramine salts (_f_), the products of ultimate dehydration (for example, like metaphosphoric acid with respect to orthophosphoric acid, but in dissolving in water they give neither rosetetramine nor tetramine salts. (In my opinion one should expect salts with a still smaller amount of NH_{3}, of the blue colour proper to the low hydrated compounds of cobalt; the green colour of the prazeo-salts already forms a step towards the blue.) Jörgenson obtained salts for ethylene-diamine, N_{2}H_{4}C_{2}H_{4} which replaces 2NH_{3}. After being kept a long time in aqueous solution they give rosetetramine salts, just as metaphosphoric acid gives orthophosphoric acid, while the rosetetramine salts are converted into prazeo-salts by Ag_{2}O and NaHO. Here only one X is ionised out of the X_{3}. There are also basic salts of the same type; but the best known is the chromium salt called the rhodozochromic salt, Cr_{2}(OH)_{3}Cl_{3},6NH_{3},2H_{2}O, which is formed by the prolonged action of water upon the corresponding roseo-salt.

The cobaltamine compounds differ essentially but little from the ammoniacal compounds of other metals. The only difference is that here the cobaltic oxide is obtained from the cobaltous oxide in the presence of ammonia. In any case it is a simpler question than that of the double cyanides. Those forces in virtue of which such a considerable number of ammonia molecules are united with a molecule of a cobalt compound, appertain naturally to the series of those slightly investigated forces which exist even in the highest degrees of combination of the majority of elements. They are the same forces which lead to the formation of compounds containing water of crystallisation, double salts, isomorphous mixtures and complex acids (Chapter XXI., Note 8 bis). The simplest conception, according to my opinion, of cobalt compounds (much more so than by assuming special complex radicles, with Schiff, Weltzien, Claus, and others), may be formed by comparing them with other ammoniacal products. Ammonia, like water, combines in various proportions with a multitude of molecules. Silver chloride and calcium chloride, just like cobalt chloride, absorb ammonia, forming compounds which are sometimes slightly stable, and easily dissociated, sometimes more stable, in exactly the same way as water combines with certain substances, forming fairly stable compounds called hydroxides or hydrates, or less stable compounds which are called compounds with water of crystallisation. Naturally the difference in the properties in both cases depends on the properties of those elements which enter into the composition of the given substance, and on those kinds of affinity towards which chemists have not as yet turned their attention. If boron fluoride, silicon fluoride, &c., combine with hydrofluoric acid, if platinic chloride, and even cadmium chloride, combine with hydrochloric acid, these compounds may be regarded as double salts, because acids are salts of hydrogen. But evidently water and ammonia have the same saline faculty, more especially as they, like haloid acids, contain hydrogen, and are both capable of further combination--for instance, ammonia with hydrochloric acid. Hence it is simpler to compare complex ammoniacal with double salts, hydrates, and similar compounds, but _the ammonio-metallic salts_ present a most complete qualitative and quantitative resemblance to _the hydrated salts of metals_. The composition of the latter is MX_{_n_}_m_H_{2}O, where M = metal, X = the haloid, simple or complex, and _n_ and _m_ the quantities of the haloid and so-called water of crystallisation respectively. The composition of the ammoniacal salts of metals is MX_{_n_}_m_NH_{3}. The water of crystallisation is held by the salt with more or less stability, and some salts even do not retain it at all; some part with water easily when exposed to the air, others when heated, and then with difficulty. In the case of some metals all the salts combine with water, whilst with others only a few, and the water so combined may then be easily disengaged. All this applies equally well to the ammoniacal salts, and therefore the combination of ammonia may be termed _the ammonia of crystallisation_. Just as the water which is combined with a salt is held by it with different degrees of force, so it is with ammonia. In combining with 2NH_{3},PtCl_{2} evolves 31,000 cals.; while CaCl_{2} only evolves 14,000 cals.; and the former compound parts with its NH_{3} (together with HCl in this case) with more difficulty, only above 200°, while the latter disengages ammonia at 180°. ZnCl_{2},2NH_{3} in forming ZnCl_{2},4NH_{3} evolves only 11,000 cals., and splits up again into its components at 80°. The amount of combined ammonia is as variable as the amount of water of crystallisation--for instance, SnI_{4}8NH_{3}, CrCl_{2}8NH_{3}, CrCl_{3}6NH_{3}, CrCl_{3}5NH_{3},PtCl_{2},4NH_{3}, &c. are known. Very often NH_{3} is replaceable by OH_{2} and conversely. A colourless, anhydrous cupric salt--for instance, cupric sulphate--when combined with water forms blue and green salts, and violet when combined with ammonia. If steam be passed through anhydrous copper sulphate the salt absorbs water and becomes heated; if ammonia be substituted for the water the heating becomes much more intense, and the salt breaks up into a fine violet powder. With water CuSO_{4},5H_{2}O is formed, and with ammonia CuSO_{4},5NH_{3}, the number of water and ammonia molecules retained by the salt being the same in each case, and as a proof of this, and that it is not an isolated coincidence, the remarkable fact must be borne in mind that water and ammonia consecutively, molecule for molecule, are capable of supplanting each other, and forming the compounds CuSO_{4},5H_{2}O, CuSO_{4},4H_{2}O,NH_{3}; CuSO_{4},3H_{2}O,2NH_{3}; CuSO_{4},2H_{2}O,3NH_{3}; CuSO_{4},H_{2}O,4NH_{3}, and CuSO_{4},5NH_{3}. The last of these compounds was obtained by Henry Rose, and my experiments have shown that more ammonia than this cannot be retained. By adding to a strong solution of cupric sulphate sufficient ammonia to dissolve the whole of the oxide precipitated, and then adding alcohol, Berzelius obtained the compound CuSO_{4},H_{2}O,4NH_{3}, &c. The law of substitution also assists in rendering these phenomena clearer, because a compound of ammonia with water forms ammonium hydroxide, NH_{4}HO, and therefore these molecules combining with one another may also interchange, as being of equal value. In general, those salts form stable ammoniacal compounds which are capable of forming stable compounds with water of crystallisation; and as ammonia is capable of combining with acids, and as some of the salts formed by slightly energetic bases in their properties more closely resemble acids (that is, salts of hydrogen) than those salts containing more energetic bases, we might expect to find more stable and more easily-formed ammonio-metallic salts with metals and their oxides having weaker basic properties than with those which form energetic bases. This explains why the salts of potassium, barium, &c., do not form ammonio-metallic salts, whilst the salts of silver, copper, zinc, &c., easily form them, and the salts RX_{3} still more easily and with greater stability. This consideration also accounts for the great stability of the ammoniacal compounds of cupric oxide compared with those of silver oxide, since the former is displaced by the latter. It also enables us to see clearly the distinction which exists in the stability of the cobaltamine salts containing salts corresponding with cobaltous oxide, and those corresponding with higher oxides of cobalt, for the latter are weaker bases than cobaltous oxides. _The nature of the forces and quality of the phenomena occurring during the formation of the most stable substances, and of such compounds as crystallisable compounds, are one and the same, although perhaps exhibited in a different degree._ This, in my opinion, may be best confirmed by examining the compounds of carbon, because for this element the nature of the forces acting during the formation of its compounds is well known. Let us take as an example two unstable compounds of carbon. Acetic acid, C_{2}H_{4}O_{2} (specific gravity 1·06), with water forms the hydrate, C_{2}H_{4}O_{2},H_{2}O, denser (1·07) than either of the components, but unstable and easily decomposed, generally simply referred to as a solution. Such also is the crystalline compound of oxalic acid, C_{2}H_{2}O_{4}, with water, C_{2}H_{2}O_{4},2H_{2}O. Their formation might be predicted as starting from the hydrocarbon C_{2}H_{6}, in which, as in any other, the hydrogen may be exchanged for chlorine, the water residue (hydroxyl), &c. The first substitution product with hydroxyl, C_{2}H_{5}(HO), is stable; it can be distilled without alteration, resists a temperature higher than 100°, and then does not give off water. This is ordinary alcohol. The second, C_{2}H_{4}(HO)_{2}, can also be distilled without change, but can be decomposed into water and C_{2}H_{4}O (ethylene oxide or aldehyde); it boils at about 197°, whilst the first hydrate boils at 78°, a difference of about 100°. The compound C_{2}H_{3}(HO)_{3} will be the third product of such substitution; it ought to boil at about 300°, but does not resist this temperature--it decomposes into H_{2}O and C_{2}H_{4}O_{2}, where only one hydroxyl group remains, and the other atom of oxygen is left in the same condition as in ethylene oxide, C_{2}H_{4}O. There is a proof of this. Glycol, C_{2}H_{4}(HO)_{2}, boils at 197°, and forms water and ethylene oxide, which boils at 13° (aldehyde, its isomeride, boils at 21°); therefore the product disengaged by the splitting up of the hydrate boils at 184° lower than the hydrate C_{2}H_{4}(HO)_{2}. Thus the hydrate C_{2}H_{3}(HO)_{3}, which ought to boil at about 300°, splits up in exactly the same way into water and the product C_{2}H_{4}O_{2}, which boils at 117°--that is, nearly 183° lower than the hydrate, C_{2}H_{3}(HO)_{3}. But this hydrate splits up before distillation. The above-mentioned hydrate of acetic acid is such a decomposable hydrate--that is to say, what is called a solution. Still less stability may be expected from the following hydrates. C_{2}H_{2}(HO)_{4} also splits up into water and a hydrate (it contains two hydroxyl groups) called glycolic acid, C_{2}H_{2}O(HO)_{2} = C_{2}H_{4}O_{3}. The next product of substitution will be C_{2}H(HO)_{5}; it splits up into water, H_{2}O, and glyoxylic acid, C_{2}H_{4}O_{4} (three hydroxyl groups). The last hydrate which ought to be obtained from C_{2}H_{6}, and ought to contain C_{2}(HO)_{6}, is the crystalline compound of oxalic acid, C_{2}H_{2}O_{4} (two hydroxyl groups), and water, 2H_{2}O, which has been already mentioned. The hydrate C_{2}(HO)_{6} = C_{2}H_{2}O_{4},2H_{2}O, ought, according to the foregoing reasoning, to boil at about 600° (because the hydrate, C_{2}H_{4}(HO)_{2}, boils at about 200°, and the substitution of 4 hydroxyl groups for 4 atoms of hydrogen will raise the boiling-point 400°). It does not resist this temperature, but at a much lower point splits up into water, 2H_{2}O, and the hydrate C_{2}O_{2}(HO)_{2}, which is also capable of yielding water. Without going into further discussion of this subject, it may be observed that the formation of the hydrates, or compounds with water of crystallisation, of acetic and oxalic acids has thus received an accurate explanation, illustrating the point we desired to prove in affirming that compounds with water of crystallisation are held together by the same forces as those which act in the formation of other complex substances, and that the easy displaceability of the water of crystallisation is only a peculiarity of a local character, and not a radical point of distinction. All the above-mentioned hydrates, C_{2}X_{6}, or products of their destruction, are actually obtained by the oxidation of the first hydrate, C_{2}H_{3}(HO), or common alcohol, by nitric acid (Sokoloff and others). Hence the forces which induce salts to combine with _n_H_{2}O or with NH_{3} are undoubtedly of the same order as the forces which govern the formation of ordinary 'atomic' and saline compounds. (A great impediment in the study of the former was caused by the conviction which reigned in the sixties and seventies, that 'atomic' were essentially different from 'molecular' compounds like crystallohydrates, in which it was assumed that there was a combination of entire molecules, as though without the participation of the atomic forces.) If the bond between chlorine and different metals is not equally strong, so also the bond uniting _n_H_{2}O and _n_NH_{3} is exceeding variable; there is nothing very surprising in this. And in the fact that the combination of different amounts of NH_{3} and H_{2}O alters the capacity of the haloids X of the salts RX_{2} for reaction (for instance, in the luteo-salts all the X_{3}, while in the purpureo, only 2 out of the 3, and in the prazeo-salts only 1 of the 3 X's reacts), we should see in the first place a phenomenon similar to what we met with in Cr_{2}Cl_{6} (Chapter XXI., Note 7 bis), for in both instances the essence of the difference lies in the removal of water; a molecule RCl_{3},6H_{2}O or RCl_{3},6NH_{3} contains the halogen in a perfectly mobile (ionised) state, while in the molecule RCl_{3},5H_{2}O or RCl_{3},5NH_{3} a portion of the halogen has almost lost its faculty for reacting with AgNO_{3}, just as metalepsical chlorine has lost this faculty which is fully developed in the chloranhydride. Until the reason of this difference be clear, we cannot expect that ordinary points of view and generalisation can give a clear answer. However, we may assume that here the explanation lies in the nature and kind of motion of the atoms in the molecules, although as yet it is not clear how. Nevertheless, I think it well to call attention again (Chapter I.) to the fact that the combination of water, and hence, also, of any other element, leads to most diverse consequences; the water in the gelatinous hydrate of alumina or in the decahydrated Glauber salt is very mobile, and easily reacts like water in a free state; but the same water combined with oxide of calcium, or C_{2}H_{4} (for instance, in C_{2}H_{6}O and in C_{4}H_{10}O), or with P_{2}O_{5}, has become quite different, and no longer acts like water in a free state. We see the same phenomenon in many other cases--for example, the chlorine in chlorates no longer gives a precipitate of chloride of silver with AgNO_{3}. Thus, although the instance which is found in the difference between the roseo- and purpureo-salts deserves to be fully studied on account of its simplicity, still it is far from being exceptional, and we cannot expect it to be thoroughly explained unless a mass of similar instances, which are exceedingly common among chemical compounds, be conjointly explained. (Among the researches which add to our knowledge respecting the complex ammoniacal compounds, I think it indispensable to call the reader's attention to Prof. Kournakoff's dissertation 'On complex metallic bases,' 1893.)

Kournakoff (1894) showed that the solubility of the luteo-salt, CoCl_{3},6NH_{3}, at 0° = 4·30 (per 100 of water), at 20° = 7·7, that in passing into the roseo-salt, CoCl_{3}H_{2}O_{5}NH_{3}, the solubility rises considerably, and at 0° = 16·4, and at 20° = about 27, whilst the passage into the purpureo-salt, CoCl_{3},5NH_{3}, is accompanied by a great fall in the solubility, namely, at 0° = 0·23, and at 20° = about 0·5. And as crystallohydrates with a smaller amount of water are usually more soluble than the higher crystallohydrates (Le Chatelier), whilst here we find that the solubility falls (in the purpureo-salt) with a loss of water, that water which is contained in the roseo-salt cannot be compared with the water of crystallisation. Kournakoff, therefore, connects the fall in solubility (in the passage of the roseo- into the purpureo-salts) with the accompanying loss in the reactive capacity of the chlorine.

In conclusion, it may be observed that the elements of the eighth group--that is, the analogues of iron and platinum--according to my opinion, will yield most fruitful results when studied as to combinations with whole molecules, as already shown by the examples of complex ammoniacal, cyanogen, nitro-, and other compounds, which are easily formed in this eighth group, and are remarkable for their stability. This faculty of the elements of the eighth group for forming the complex compounds alluded to, is in all probability connected with the position which the eighth group occupies with regard to the others. Following the seventh, which forms the type RX_{7}, it might be expected to contain the most complex type, RX_{8}. This is met with in OsO_{4}. The other elements of the eighth group, however, only form the lower types RX_{2}, RX_{3}, RX_{4} ... and these accordingly should be expected to aggregate themselves into the higher types, which is accomplished in the formation of the above-mentioned complex compounds.

[35 bis] Marshall (1891) obtained cobaltic sulphate, Co_{2}(SO_{4})_{3},18H_{2}O, by the action of an electric current upon a strong solution of CoSO_{4}.

[36] The action of an alkaline hypochlorite or hypobromite upon a boiling solution of cobaltous salts, according to Schroederer (1889), produces oxides, whose composition varies between Co_{3}O_{5} (Rose's compound) and Co_{2}O_{3}, and also between Co_{5}O_{8} and Co_{12}O_{19}. If caustic potash and then bromine be added to the liquid, only Co_{2}O_{3} is formed. The action of alkaline hypochlorites or hypo-bromites, or of iodine, upon cobaltic salts, gives a highly-coloured precipitate which has a different colour to the hydrate of the oxide Co_{2}(OH)_{6}. According to Carnot the precipitate produced by the hypochlorites has a composition Co_{10}O_{16}, whilst that given by iodine in the presence of an alkali contains a larger amount of oxygen. Fortmann (1891) reinvestigated the composition of the higher oxygen oxide obtained by iodine in the presence of alkali, and found that the greenish precipitate (which disengages oxygen when heated to 100°) corresponds to the formula CoO_{2}. The reaction must be expressed by the equation: CoX_{2} + I_{2} + 4KHO = CoO_{2} + 2KX + 2KI + 2H_{2}O.

[37] Prior to Fortmann, Rousseau (1889) endeavoured to solve the question as to whether CoO_{2} was able to combine with bases. He succeeded in obtaining a barium compound corresponding to this oxide. Fifteen grams of BaCl_{2} or BaBr_{2} are triturated with 5-6 grams of oxide of barium, and the mixture heated to redness in a closed platinum crucible; 1 gram of oxide of cobalt is then gradually added to the fused mass. Each addition of oxide is accompanied by a violent disengagement of oxygen. After a short time, however, the mass fuses quietly, and a salt settles at the bottom of the crucible, which, when freed from the residue, appears as black hexagonal, very brilliant crystals. In dissolving in water this substance evolves chlorine; its composition corresponds to the formula 2(CoO_{2})BaO. If the original mass be heated for a long time (40 hours), the amount of dioxide in the resultant mass decreases. The author obtained a neutral salt having the composition CoO_{2}BaO (this compound = BaO_{2}CoO) by breaking up the mass as it agglomerates together, and bringing the pieces into contact with the more heated surface of the crucible. This salt is formed between the somewhat narrow limits of temperature 1,000°-1,100°; above and below these limits compounds richer or poorer in CoO_{2} are formed. The formation of CoO_{2} by the action of BaO_{2}, and the easy decomposition of CoO_{2} with the evolution of oxygen, give reason for thinking that it belongs to the class of peroxides (like Cr_{2}O_{7}, CaO_{2}, &c.); it is not yet known whether they give peroxide of hydrogen like the true peroxides. The fact that it is obtained by means of iodine (probably through HIO), and its great resemblance to MnO_{2}, leads rather to the supposition that CoO_{2} is a very feeble saline oxide. The form CoO_{2} is repeated in the cobaltic compounds (Note 35), and the existence of CoO_{2} should have long ago been recognised upon this basis.

Nickel alloys possess qualities which render them valuable for technical purposes, the alloy of nickel with iron being particularly remarkable. This alloy is met with in nature as _meteoric iron_. The Pallasoffsky mass of meteoric iron, preserved in the St. Petersburg Academy, fell in Siberia in the last century; it weighs about 15 cwt. and contains 88 p.c. of iron and about 10 p.c. of nickel, with a small admixture of other metals. In the arts _German silver_ is most extensively used; it is an alloy containing nickel, copper, and zinc in various proportions. It generally consists of about 50 parts of copper, 25 parts of zinc, and 25 parts of nickel. This alloy is characterised by its white colour resembling that of silver, and, like this latter metal, it does not rust, and therefore furnishes an excellent substitute for silver in the majority of cases where it is used. Alloys which contain silver in addition to nickel show the properties of silver to a still greater extent. Alloys of nickel are used for currency, and if rich deposits of nickel are discovered a wide field of application lies before it, not only in a pure state (because it is a beautiful metal and does not rust) but also for use in alloys. Steel vessels (pressed or forged out of sheet steel) covered with nickel have such practical merits that their manufacture, which has not long commenced, will most probably be rapidly developed, whilst nickel steel, which exceeds ordinary steel in its tenacity, has already proved its excellent qualities for many purposes (for instance, for armour plate).

Until 1890 no compound of cobalt or nickel was known of sufficient volatility to determine the molecular weights of the compounds of these metals; but in 1890 Mr. L. Mond, in conducting (together with Langer and Quincke) his researches on the action of nickel upon carbonic oxide (Chapter IX., Note 24 bis), observed that nickel gradually volatilises in a stream of carbonic oxide; this only takes place at low temperatures, and is seen by the coloration of the flame of the carbonic oxide. This observation led to the discovery of a remarkable volatile _compound of nickel and carbonic oxide_, having as molecular composition Ni(CO)_{4},[38] as determined by the vapour density and depression of the freezing point. Cobalt and many other metals do not form volatile compounds under these conditions, but iron gives a similar product (Note 26 bis). Ni(CO)_{4} is prepared by taking finely divided Ni (obtained by reducing NiO by heating it in a stream of hydrogen, or by igniting the oxalate NiC_{2}O_{4})[39] and passing (at a temperature below 50°, for even at 60° decomposition may take place and an explosion) a stream of CO over it; the latter carries over the vapour of the compound, which condenses (in a well-cooled receiver) into a perfectly colourless extremely mobile liquid, boiling without decomposition at 43°, and crystallising in needles at -25° (Mond and Nasini, 1891). Liquid Ni(CO)_{4} has a sp. gr. 1·356 at 0°, is insoluble in water, dissolves in alcohol and benzene, and burns with a very smoky flame due to the liberation of Ni. The vapour when passed through a tube heated to 180° and above deposits a brilliant coating of metal, and disengages CO. If the tube be strongly heated the decomposition is accompanied by an explosion. If Ni(CO)_{4} as vapour be passed through a solution of CuCl_{2}, it reduces the latter to metal; it has the same action upon an ammoniacal solution of AgCl, strong nitric acid oxidises Ni(CO)_{4}, dilute solutions of acids have no action; if the vapour be passed through strong sulphuric acid, CO is liberated, chlorine gives NiCl and COCl_{2}; no simple reactions of double decomposition are yet known for Ni(CO)_{4}, however, so that its connection with other carbon compounds is not clear. Probably the formation of this compound could be applied for extracting nickel from its ores.[40]

[38] This compound is known as nickel tetra-carbonyl. It appears to me yet premature to judge of the structure of such an extraordinary compound as Ni(CO)_{4}. It has long been known that potassium combines with CO forming K_{_n_}(CO)_{_n_} (Chapter IX., Note 31), but this substance is apparently saline and non-volatile, and has as little in common with Ni(CO)_{4} as Na_{2}H has with SbH_{3}. However, Berthelot observed that when NiC_{4}O_{4} is kept in air, it oxidises and gives a colourless compound, Ni_{3}C_{2}O_{3},10H_{2}O, having apparently saline properties. We may add that Schützenberger, on reducing NiCl_{2} by heating it in a current of hydrogen, observed that a nickel compound partially volatilises with the HCl and gives metallic nickel when heated again. The platinum compound, PtCl_{2}(CO)_{3} (Chapter XXIII., Note 11), offers the greatest analogy to Ni(CO)_{4}. This compound was obtained as a volatile substance by Schützenberger by moderately heating (to 235°) metallic platinum in a mixture of chlorine and carbonic oxide. If we designate CO by Y, and an atom of chlorine by X, then taking into account that, according to the periodic system, Ni is an analogue of Pt, a certain degree of correspondence is seen in the composition NiY_{4} and PtX_{2}Y_{2}. It would be interesting to compare the reactions of the two compounds.

[39] According to its empirical formula oxalate of nickel also contains nickel and carbonic oxide.

[40] The following are the thermo-chemical data (according to Thomsen, and referred to gram weights expressed by the formula, in large calories or thousand units of heat) for the formation of corresponding compounds of Mn, Fe, Co, Ni, and Cu (+ Aq signifies that the reaction proceeds in an excess of water):

R = Mn Fe Co Ni Cu R + Cl_{2} + Aq 128 100 95 94 63 R + Br_{2} + Aq 106 78 73 72 41 R + I_{2} + Aq 76 48 43 41 32 R + O + H_{2}O 95 68 63 61 38 R + O_{2} + SO_{2} + _n_H_{2}O 193 169 163 163 130 RCl_{2} + Aq +16 18 18 19 11

These examples show that for analogous reactions the amount of heat evolved in passing from Mn to Fe, Co, Ni, and Cu varies in regular sequences as the atomic weight increases. A similar difference is to be found in other groups and series, and proves that thermo-chemical phenomena are subject to the periodic law.