formula Ni (CO),, viz., 34:33 and 34-26 per cent. of nickel, the formula requiring 34.28. The vapour density determined by Victor Meyer's method at 50° was found equal to 6·01; the formula Ni (CO), requires 5·89.

The compound is chemically very inactive; generally speaking, it only reacts with substances having a considerable affinity for nickel, such as the halogens, sulphur, oxygen, and oxidising substances, which combine with the nickel and liberate carbonic oxide. Chlorine and bromine when used in excess also enter into combination with the carbonic oxide. Sulphur in the dry state forms a sulphide of nickel corresponding to the formula Ni,S,, and dissolved in bi-sulphide of carbon it forms a sulphide containing more sulphur, but of varying composition. Selenium acts similarly but very slowly. Tellurium shows hardly any action. Metals (even potassium) are not acted upon.

Alkalies and acids (even strong hydrochloric acid) produce no change except they are oxidising agents, such as nitric acid and aqua regia. With metallic salts no reaction is obtained unless they have oxidising properties as hypochlorites, which form a higher oxide of nickel, or which are capable of giving off sulphur, such as hyposulphites and bisulphites.

The author has tried in vain to substitute other bivalent groups for the carbonic oxide in this compound, or to introduce the carbonic oxide by means of this compound into organic substances. Experiments in this direction have covered a very wide range and have included, amongst others, the following: hydroxylamine hydrochloride, phenylhydrazin hydrochloride hydroxylamine, dichloracetic acid, tetrabromphenolbromide, ethylinechloride, and aceto-acetic-ether, but in no single instance was the desired result obtained.

On exposure to moist or dry air a flocculent substance, which varies in colour from a light green to a dark brown, is very slowly formed. This substance dissolves completely in dilute acids with evolution of carbonic acid; numerous analyses have not led to a definite proportion between Ni and CO, in this compound. On heating it to dull red heat it turns black. Professor Berthelot, in a paper recently communicated to the French Academy of Sciences, assumes that this black colour is produced by the separation of carbon, and bases upon this an argument that the compound is of a complex composition, and that the nickel carbon oxide, on exposure to air, behaves like a real compound radical analogous to organo-metallic radicals. As, however, the black substance so obtained dissolves in dilute acids without leaving any residue, and as an exactly similar black substance is obtained by heating precipitated nickel carbonate, this argument does not seem to be conclusive, since Professor Berthelot has not substantiated so important a conclusion by a complete analysis of the black substance.

Professor Berthelot describes in the same paper a very beautiful blue compound obtained by treating nickel carbon oxide with nitric oxide. Unfortunately he does not publish an analysis of this beautiful substance either, so that until he has done so we are unable to judge of its bearing on the constitution of nickel carbon oxide.

With a view to elucidate this constitution the author has, in conjunction with Professor R. Nasini, of Rome, studied the physical properties of the liquid, more especially its refraction and dispersion. The details of this investigation have been communicated to the Accademia dei Lincei at Rome, and have also been published in the Journal für physikalische Chemie.'

The author has determined the freezing-point of a dilute solution in benzole containing 4.8991 per cent., and has found the coefficient of diminution 2776, corresponding to a molecular weight of 176.5; while nickel carbon oxide requires 170.6. The mean cubical coefficient of expansion between 0° and 36° C. is equal to 001853, which is one of the highest coefficients of expansion yet found for any liquid, and is only slightly exceeded by ethylic ether, ethyl chloride, and silicium tetrachloride. The indices of refraction and the dispersion for the lines a, ẞ, and y of hydrogen, and for the lines of lithium, sodium, and thallium have been determined at three different temperatures, and are found to be very high. The dispersion is about the same as carbon disulphide. The refraction varies very much with the temperature, the amount of variation being very nearly equal to that of carbon disulphide. The index of refraction for the D line at 10° C. is 1-45843.

According to Gladstone's formula this leads to the specific refraction of .3437 and the molecular refraction of 58-63. Under the supposition that the group CO had the same value in this compound which results from the sum of the atomic refraction of carbon and that of the divalent oxygen molecule in organic compounds, which is the more probable, as the group CO shows very nearly the same molecular refraction in compounds of the most different constitution, such as oxalic acid, ketones, and carbonyldichloride, the atomic refraction of nickel would come out equal to 25 02. This figure is very much higher, nearly two and a half times as high as it is in nickel salts, in which it has been found by Gladstone to be about 10; and about four times as high as the atomic refraction of metallic nickel as determined by Kundt and Dubois and Rubens, viz., about 6.

This difference of the atomic refraction of nickel in this compound and in its ordinary combinations is by far greater than that found in any other element. According to the generally accepted view, such differences are due to the element possessing a large number of valencies, and are proportional to the number of valencies of each compound. Nickel is generally bivalent. Its very high atomic refraction in nickel carbon oxide would thus lead to the conclusion that in this compound the nickel exercises a considerably higher valency than two, and that it has probably reached its maximum of saturation foreseen by Mendeléeff, who placed this metal in the eighth group of his Periodic System, to be equal to eight; so that the constitution of our compound would be a simple combination of one octovalent equivalent of nickel with four bivalent equivalents of carbonic oxide, or that of nickel tetracarbonyl.

All that we definitely know of the chemical properties of the compound is in accord with this view of its constitution.

A determination of the magnetic rotary power of the compound kindly made by Dr. W. H. Perkin has shown this to be quite as exceptional as its refraction, and, with the exception of phosphorus, greater than any substance he has yet examined.

Professor Quincke, of Heidelberg, has had the kindness to investigate the magnetic properties of the liquid. He found the constant of diamagnetism, at 16° C., k=-3131 x 10-10 for magnetic fields of 6,000 to 14,000 C.G.S. units. This is nearly the same as the constant for ethylic ether -3.218 × 10-10.

The liquid is an exceptionally bad conductor of electricity. Up to 40 volts no current was observed to pass, the electrodes of 1 sq. cm. area being 1 cm. apart.

The highly interesting properties of nickel carbon oxide naturally led the author to try whether he could not obtain similar compounds of other metals. It seemed a foregone conclusion that cobalt, in every respect so much like nickel, must give an analogous compound. It seemed probable that other metals of the eighth group and those standing near to nickel in other groups would also combine with carbonic oxide. A large number of elements were tried, including osmium, palladium, ruthenium, rhodium, iridium, and manganese by acting upon them in the finely divided state with carbonic oxide gas over a wide range of temperature. The author tried it by double decomposition with numerous compounds, including zinc ethyl and mercury methyl, but, with one sole and single exception, without success.

This sole exception is iron. This metal, too, had for a long time given negative results; but by preparing it at the lowest possible temperature by reduction of the oxalate in a current of hydrogen, and by acting upon this at about 80° C. with a very slow current of very pure carbonic oxide, the author succeeded at last, in conjunction with Dr. F. Quincke, in obtaining evidence that a volatile compound of this element with carbonic oxide exists. The gas obtained imparted a yellow tinge to a Bunsen flame and yielded slight metallic mirrors composed of pure iron. The quantity of the iron compound in the gas was, however, extremely small. By passing the gas through heavy tar oils, in which the compound is soluble, but from which it cannot be separated by fractionation, as on heating it decomposes the solution into iron and carbonic oxide before it volatilises, and by determining the iron and carbonic oxide so obtained, it was ascertained, as far as the very small quantities of the substance available would

allow, that it contained iron and carbonic oxide in the proportion of 1 equivalent of iron to 4.126 of carbonic oxide, or very nearly 1 to 4.

Since these results were communicated to the Chemical Society (June 18, 1891) the author has continued the study of this body, in collaboration with Dr. Carl Langer, and has obtained it as an amber-coloured liquid, which, on standing, deposits tabular crystals of a darker colour, and solidifies entirely below -21° C. to a mass of needle-shaped crystals. It boils at 102° C., but leaves a small quantity of green-coloured oil behind.

Several analyses and vapour density determinations have been made, but it is not yet certain whether a pure substance is in hand or a mixture of several iron carbonyls. The author hopes to be able very shortly to publish a full account of this interesting substance, which differs considerably in its chemical behaviour from nickel carbon oxide.

The fact that under ordinary circumstances nickel alone is acted on when a mixture of this metal with any other metallic or mineral substances is treated by carbonic oxide gas led the author to institute experiments to ascertain whether it would not be possible by means of carbonic oxide to extract nickel direct from its ores, and such metallurgical products as nickel speiss and nickel matte. As the nickel is volatilised at the ordinary temperature in the form of a vapour disseminated through other gases from which it can be deposited without first condensing the nickel compound by simply heating these gases to the moderate temperature of 200° C., as it is thus obtained in the form of bright coherent masses of great purity, as the carbonic oxide used is completely liberated and can be employed over and over again, and as small quantities of the poisonous nickel compound which may escape decomposition would thus never leave the closed apparatus in which the process would be carried out, it seemed probable that such a process might be capable of industrial application, and might prove more economical than the very complicated operations metallurgists have now to resort to to produce tolerably pure nickel.

Experiments carried out under the author's instructions by Dr. Langer with a great variety of nickel ores from all parts of the world, containing from 4 to 40 per cent. of nickel, as well as a number of samples of nickel speiss and nickel matte, have proved that as long as the nickel is combined with arsenic or sulphur the process is entirely successful on a laboratory scale. In the majority of cases the nickel has been extracted almost completely in three to four days.

Such ores or matte or speiss have in the first instance to be calcined, so as to convert the nickel completely into oxide. The mass is then reduced in a current of hydrogenous gases, in practice water-gas, at a temperature of 450° C. It is cooled down to ordinary temperature and treated with carbonic oxide in a suitable apparatus. For this purpose any good apparatus for treating solids by gases, of which a great number are in common use, will answer. Methodical apparatus moving the reduced ore in a direction opposite to the current of carbonic oxide, at the same time exposing fresh surfaces, facilitates the operation. After a certain time the action of the carbonic oxide upon the nickel becomes sluggish. The mass is then heated to about 350° C. in a current of carbonic oxide, which regenerates the activity of the nickel. This may be done in the same apparatus, but it is preferable to use a separate apparatus connected with the first, and from which it is returned to the first by mechanical means, so that each apparatus can be kept at the same temperature. The carbonic oxide gas can be employed dilute, as it is obtained from gas-producers; but since it is continuously recovered, a purer gas such as can be cheaply prepared by passing carbonic acid through incandescent coke is more advantageous, as it extracts the nickel more quickly and requires smaller apparatus. The gas charged with the nickel compound leaving the apparatus is passed through tubes or chambers heated to about 200° C., in which the nickel is deposited. The gas leaving these tubes is returned to the first apparatus and circulates continuously. From time to time the nickel is removed from the tubes in which it has been deposited. To facilitate this operation thin nickel sheets bent to fit the tubes are inserted, on which the nickel deposits, and which are easily taken out. The metal so obtained is almost chemically pure;

only very rarely in the case of certain ores it is slightly contaminated with iron. Its density is equal to that of ordinary sheet nickel. Its mechanical properties still await investigation.

As the nickel is deposited in perfectly coherent films upon heated surfaces exposed to the gas containing the nickel carbon oxide, the author finds it possible to produce direct from such gas articles of solid nickel or goods plated with nickel resembling in every way those obtained by galvanic deposition of metals, and reproducing with the same exactitude and fineness any design upon such articles.

This result can also be obtained by immersing heated articles in a solution of nickel carbon oxide in such solvents as benzole, petroleum, tar oils, &c., or by applying such solution to the heated articles with a brush or otherwise.

These processes open up a wide perspective of useful application, considering the many valuable properties of nickel, especially its power of resisting atmospheric and other chemical influences.

4. On the Electrical Evaporation of Metals and Alloys.1
By W. CROOKES, F.R.S.

5. On the Cause of Imperfections in the Surface of Rolled Copper Alloys. By T. TURNER, A.R.S.M.

In those rolled copper alloys which are of a yellow colour it is common to find surface stains of a copper colour. These stains render the rolled metal unfit for many purposes. The cause of these stains has been much discussed, but no definite evidence as to their origin has been forthcoming. Among the supposed causes have been overheating during annealing, sulphur in the fuel used, the presence of soot or ashes, irregularity in the alloy, and the use of an iron stirringrod. The author has conducted a number of experiments, and concludes that none of these causes is responsible for the production of the stains observed, but that the stains are merely on the surface, and are produced by dirt in some form or other. The use of wash-water containing chlorides, after pickling, is in the author's opinion the chief cause of the imperfections (see Trans. Birmingham Phil. Soc., May 1891).

SATURDAY, AUGUST 22.

The Section did not meet.

MONDAY, AUGUST 24.

The following Papers were read :—

1. Certain Pyrometric Measurements and Methods of Recording them. By Professor W. C. ROBERTS-AUSTEN, C.B., F.R.S.

2. On the Existence of a Compound in Alloys of Gold and Tin.
By A. P. LAURIE, M.A.

The alloys are prepared by melting the metals in a clay pipe and drawing into the stem. They are then used in place of zinc in a voltaic cell, consisting of

1 See Proceedings of the Royal Society, 1891.

stannic chloride, gold chloride, and gold. The E.M.F. is found to rise abruptly on passing from alloys containing 36 per cent. to those containing 38 per cent. of tin, thus indicating the existence of a compound of the formula AuSn. This agrees with the maximum point in Matthiessen's conductivity curve for these alloys. Some preliminary experiments with gold_aluminium alloys show that there is a rise in E.M.F. on passing over Professor Roberts-Austen's new purple alloy. This method not only indicates the existence of a compound, but also enables us to calculate approximately the heat of formation of the compound from the rise in E.M.F. in passing from the alloys below to those above the compound.

3. On the Relation between the Composition of a Double Salt and the Composition and Temperature of the Solution in which it is formed. By A. VERNON HARCOURT, F.R.S., and F. W. HUMPHERY, of Christ Church, Orford.

The particular double salt of which the authors have prepared and analysed some seventy specimens is ferrous and ammonium chloride. Probably many other double salts, if similarly examined, would show similar variations. When ferrous chloride and ammonium chloride are dissolved together in warm water, and the saturated liquid is allowed to cool, white crystals are deposited, whose composition varies with the proportion of the two salts in solution and with the temperature at which the crystals are formed. Since the proportion of the two chlorides in the liquid and in the crystals which form in the liquid is very different, the composition of the liquid changes continually during crystallisation, and no two portions of the double salt are formed under exactly the same conditions. In the authors' later experiments the variation from the beginning to the end of a crystallisation was reduced to about 2 per cent. by taking a crop of crystals weighing only eight or ten grams from a liquid containing some hundreds of grams of each chloride. By the use of a water-bath, in which the flask holding the solution was plunged, warmed by a gas-burner which was governed by a thermostat in the liquid, the temperature was kept at that point at which crystallisation began within 0.02 C.

The general relations observed are as follows:-The proportion of ammonium chloride in the salt increases with that in the solution nearly in direct ratio for a given temperature, the ratio varying from about thrice the proportion of ammonium chloride in the salt that there is in the solution, at the highest temperature, up to nearly six times the proportion at the lowest. The same solution yields crystals containing more ferrous chloride at a higher temperature and less at a lower. Solutions containing two or more molecules of ammonium chloride to one molecule of ferrous chloride yield well-formed crystals, white and transparent, having sides of one millimetre or more in length. When the salt contains twenty or thirty molecules of ammonium chloride to one molecule of ferrous chloride the crystals are as large as when the proportion of ammonium chloride is smaller. With less than two molecules of ammonium chloride the crystals are apt to be very small and to resemble those of ammonium chloride in form though not in composition.

In most cases the composition of the crystals can be represented by a formula Fe Cl2, n H1NCl where n is integral. Salts have been analysed with closely concordant results in which n has the following values:-7, 11, 12, 13, 15, 16, 17, 18, 19, 20, 23. But in some cases, where equal or greater care has been taken that the conditions may not change during crystallisation, the values of n are not integral. Also the double salt is hydrated; the determination of the amount of water presents greater difficulties than the rest of the analysis, a temperature of about 200° C. being necessary to expel the whole; the results thus obtained, which generally agree with the estimation of water by difference, frequently correspond to 2.5 molecules of water for each molecule of ferrous chloride. Perhaps, therefore, the formula of the double salts should be multiplied by two, or some larger even number.

The whole of the results obtained can be grouped together in a Table, in which each salt is represented by the value of n and is assigned a position showing (1) the composition, (2) the temperature, of the solution in which it was formed.