814 NILSON AND PETTERSSON NEW CHLORIDES OF INDIUM, LXVII1.-On two N e w Chlorides of Indium and on the Tapour-densities of Indium Gallium Iron and Chromium. By L. F. NILSON and OTTO PETTERSSON. THE decisive results for the trivalency of aluminium which we obtained in our researches on alumiuium chloride (Uej'wrsiyt af K. Wetenskaps Akad. fiirliandl. 1887 No. 8 and Zeitschrift f. physikul. C'hern. 1887 4.59) rendered it most desirable that the earlier wark on the molecular weights of the chlorides of other elements of the same group should be revised. At the conclusion of our communication, therefore we invited those chemists who had previously worked in this direction to undertake this revision. In reply we received a letter from Lecoq de Boisbaudran expressing the wish that we would undertake the work on gallium and in the most obliging manner sending us a supply of pure material.Clemens Winkler wrote to the same effect in respect of indium and gallium and ac the same time placed a t our disposal a quantity of these rare elements. With these materials and those already in our possession we have undertaken the investigation. We would however observe that the amount of gallium in our possession notwitlistanding the liberality of tne above-mentioned chemists was insuflicient to enable us to investigate the gallium chlorides as coruyletely as we could have wished. V. Mepr who had already determined the vapour-density of the chlorides of iron reserved to himself the right of revisiiig the density of ferric chloride and has recently published his results (Grunewal AND THE VAPOUR-DENSITIES OF INDIUM ETC.815 and V. Meyer Ber. 1888 21 687) but as he proposed to us that we should definitely ascertain t,he molecular weight of ferrous chloride, we were prompted to undertake some experiments ou the latter. It also seemed of interest to include the chromous and chromic chlorides in our researches especially as these compounds so far as we are aware have not hitherto been examined in this respect. As will be seen from the following account several of our experi-ments were carried out in porcelain vessels at a temperature which was not accurately determined. The reason of this was that our platinum apparatus towards the conclusion of our work must have become slightly defective the leak only becoming apparent at an intense heat and under pressure; the level of the mercury in the measuring tube or of the liquid in the manometer* no longer remain-ing constant but slowly and gradually changing.As it had to be sent to Paris for repair we could not delay the completion of our experiments until it was returned; and from this time forward all the experiments were conducted in porcelain vessels the temperature being approximately estimated from our knowledge of the heat which the furnace was capable of giving. Some of the estimations were made in the vapour of boiling mercury sulphur or stannous chloride contained in tubes of hard Thiiringen glass. By means of a Muencke five-light burner it was easy to maintain the vapour of the mercury or sulphnr a t the desired height if the tube were surrounded with asbestos cardboard.Stan-nous chloride on the contrary required strongly heating in a blowpipe flame; the bottom of the tube in fact appeared to be at a low red heat alhhough Griinewald and V. Meyer (Ber. 1888 21 22 and Zeitschrijt f. physilcul. Chem. 2 184) state that i t boils constantly a t 6U6" a teniperature which the Thiiringen glass stands well. If the substance contained in the bath is not allowed to cool or solidify, several experinieDts can be made one after another without any danger of breaking the tube. A t the conclusion of the series of experiments the substance is of courde poured out of the bath while it is still liquid. I. Chlorides of Indium. Hitherto the only known chloride of indium is that whose composi-tion is represented by the formula In,Cl or InC13.When the metal * For the various details of the method of experimenting and the mode of cal-culating the results we must refer to our former paper on vapour-density determi-nations (J. pr. Chent. 1886 C23 33 l) and merely add here that the platinum apparatus we employed in the experiments in question had a capacity of 118 -19 C.C. for the cylinder which was 24.Q mm. high and 25 mm. in chameter and 7.48 c.c. for the neck which was 300 mm. high and about 5.6 mm. in diameter. That employed for the chromium chloride experiments had a cylinder of a capacity of 116.66 C.C 816 KILSOX AXD PETTERSSON KEW CHLORIDES OF INDIUM, is heated in chlorine the latter being in excess this chloride is obtained in beautiful dazzling white plates which a t a temperature approaching redness volatilise without previously undergoing fusion.Already in 1867 Clemens Winkler (J. pr. Chem. 102 96) in his preliminary investigation of the indium compounds noticed that the metal when exposed to a curi*ent of chlorine melts at first to a brown mass but he did not further investigate this substance. He however, expressed the opinion that it might possibly be a lower chloride of indium corresponding with the suboxide which he had obtained as a black pyrophoric powder on heating indium oxide at 300" in h j drogen. We have found however that indium in the stale of vapour gives three different well-characterised chlorides and =ow proceed to describe their mode of preparation properties and composition.Indium TrichZoi-ide InCl,.-Metallic indium is not; changed by exposure at the ordinary temperature to dry hydrogen chloride free from air but if heated in a current of this gas the hitherto unknown dichloride InC& (see p. SlS) is formed. If this is gently heated in a stream of chlorine it is easily converted into the trichloride having all the properties ascribed to it by Clemens Winkler. It is well known that in the literature of the subject only one vapour-density determination of this chloride occ'irs namely that made by V. and C. Moyer (Be]-. 1879 12 S l l ) in an appropriate furfiace using it tube of hard glass coated with clay and a t'emperature approaching a bright red heat ; the detisity found was 7.87 instead of 7.584 for TnC1,.According to the authors the formula of the chloride is thereby confirmed and the t'rivalency of the eiement established. They describe their investigation as follows :-" Iiidium chloride is not volatile in the vapour of diphenyl perchloride which boils far ahove 440° and sublimes but slowly in boiling phosphorus penta-sulphide (530"). At a low red heat however it sublimes easily but not very rapidly and it is only a t a bright red heat that it passes iiito the state of a normal gas. The supposition that indium chloride in the state of vapour is a mixture of C1 and an unstable In2C14 is improbable as indium does not form any derivatives corresponding with the ferrous compounds but only one series like aluminium. No free chlorine could be detected in the apparatus after the experiment, the chloride being found quite unchanged in beautiful lustrous crystals." As already pointed out however iudium forms two new chlorides, not a dichloride only but also a monochloride both of which a1.e btablo in the gaseous state and as the existence of these then unknown chlorides renders futile the remark of the learned author ASD THE VA4POVR-DESSIlIES OF IXDIUM ETC.817 Rbove quoted we felt it to be necessary to submit the indium trichloride to a renewed investigation. In order to obtain material for the vnpour-density determin a t' ion 8 quantity of indium dichloride sufficient for each experiment was first prepared by heating a weighed quantity of the metal in gaseous hydrogen chloride in a narrow drawn-out tube (see p.SlS) the hydrogen evolved being collected as usual in a Schiff's apparatus. The dichloride thus formed wag converted into the trichloride by heating it gently i n dry chloriiie free from air and then sublimed into an adjacent portion of the tube in a stream of pure carbon dioxide, so as to remove any adhering free chlorine. A platinurn tube about 20-25 mm. long was then inserted in the end of the tube which must be drawn out as shown in the figure. The little platinum tube a thus and c. forms a connection between the ends of the two tubes b By continued and careful turning of the platinum in the drawn-out openings of the glass tubes it may without difficulty be made to fit quite air-tight. If the glass tubes are carefully drawn out perfectly cylindrical and the openings but slightly conical the platinum tube remains quite firm even when the chloride is sublimed into it ; this is effected in a stream of carbon dioxide and as the platinum readily cools the chloride although so volatile condenses almost entirely in the metal tube.As soon as this is cold the chloride is turned into a little cup by means of a pair of tongs. The empty platinum tube is then weighed and also the little cup filled with chloride and inclosed in watch-glasses; in this way the weight of substance to be employed is known which in our experiments corresponded very nearly with the weighed quantity of metal taken. The calculated vapour-density' of the chloride InC1 is 7.548. Experiment 1 (see Table I p. 818) in sulphur vapour gave only a minimal vaporisation not measurable volumetrically.At the conclu-sion of the experiment howLver we observed on the sides of the glass tube several beautifully iridescent well-formed crystals of the chloride as thin six-sided plates. We can therefore confirm the statement of V. and C. Meyer that * The Meyer-Seubert atomic weights are used in all the calculations in this memoir 818 NILSON AND PETTERSSON NEW CHLORIDES OF INDIUM, gas die-placed c.c. at & and 760 -Expt. 1 2 4 5 -apour-density found. TABLE I.- Vapour-density of I n d i u m Trichloride. 6.022 7.021 6'447 7.110 Temperature determina-8.156 '7'391 6.716 6.234 Vol. of gas ex-pelled. C.C. --92 -984 98 -165 - -tion. Temp. of the mea siw-ing tube -15 -8' 14 *O 14 -8 13 *1 -Calcu-lated tempera-ture of expt.-440°1 6062 850 1048 .100-1200: Vapour-density deter-mination. Weight Of chloride in grams. -0 *0635 0 -0671 0 0560 0.0573 I VOl. of I mm. I I- -- I -Remarke. --Minimal VB-porisation. Slow vapor-isation. Rapid volati-iisation. 1 In sulphur vapour. I n stannous chloride vapour. 3 The exact tem-perature could not be determined on account of a defect in the platinum apparatus, but it was evidently higher than in Experiment 4. at 440" indium trichloride does not appreciably vaporise and that at % temperature which according to our experiments lies between 606" and 850" its vnpour-density corresponds with the value 7.548 calculated from the formula InCI,.I n boiling stannous chloride its vapour-density is somewhat higher ; as however it volatilises but slowly at this temperature it is improbable that there is any lower temperature at which an indium chloride of vapour-density 15.168, corresponding with the formula In,C16 can actually exist in the gaseous state. A t temperatures above 85@" indium trichloride a s the values we found show undergoes a progressive dissociation probably into lower chlorine compounds and free chlorine. Indium Dichloride InCl,.-If indium is heated to its fusion point in a current of dry hydrogen chloride free from air it yields indium dichloride with evolution of hydrogen ; at first indeed a reddish-brown liquid is formed containing indium monochloride (p. 820) but this as soon as the h-jdrogen chloride is in excess gradually becomes lighter and lighter and ultimately of a pure amber colour.It then consists of the pure dichloride and on cooling solidifies to a white radiated crystalline mass which if strongly heated volatilises and is deposited close to the flame in colourless needles. Any adhering hydrogen chloride may be removed by distillation in a current of dry carbon dioxide free from air. On exposure to moist air i t deliquesces AND THE VAPOUR-DENSITIES OF INDIUM ETC. 819 hnt it remains unchanged in dry air. Water decomposes the chloride with a reddish sheen (from monochloride formed as an intermediate product ?) into trichloride which remains dissolved and metallic iiidium which is deposited as a grey spongy mass ; the latter acquires metallic lust,re when conso1idat)ed to a lump by means of a glass rod.The dichloride therefore cannot exist in aqueoas solution but is decomposed with formation of the trichloride ; the reaction being-3InC1 = 2InC1 + I n . The amount of metal we obtained on treating the dichloride with water is somewhat less than that required by the equation which is doubtless to be ascribed to the formation of some oxychloride from the oxygen present. The composition of the new chloride is shown by the following syntheses and analysis. Synthesis I.-0.0573 gram of metallic indium treated as above gave the following quantities of hydrogen at 0" and 760 mm. :-Found. Cdculated. r- 7 C.C. gntm- C.C. gram. & Hydrogen. . . 11.04 0.000988 11.28 0.001010 From this weighed quantity of metallic indium 0.0904 gram of indium dichloride was sublimed into the platinum cup and used for Experi-ment 3 the calcula,ted yield being 0.0929 gram InCl supposing none of the chloride to have been lost.Xynthesis 11.-0.0966 gram of metallic indium gave-Found. Calculated. 7 I-- r-; C.C. gram. C.C. gram. Hydrogen. . . 18.60 0.001665 19-02 0.001703 Synthesis IIL -0.1203 gram of metallic indium gave 0.1940 gram of dichloride which was used in the following analysis; the calculated jield should have been 0.1964 gram. AnaZysis.-O.19%0 gram of indium dichloride was treated with boiled water. The solution yielded 0.3013 gram of silver chloride, corresponding with 0.07451 gram of chlorine; and 0.1008 gram of indium oxide corresponding with 0.0832 gram of indium.The metallic indium which separated weighed 0.0374 gram. Expressed centesi-rnally the resalts are 820 XILEON AKD PETTERSSON ISEW CHLORIDES OF INDIUX, Found. Calculated. Residual metallic indium 19.28) 62.1G 20:5Y} 61.58 Indium as oxide in soliltion 42.88 41 05 Chlovine . . . . . . . . . . . . . . . . 38.42 100.57 100*00 38.41 ~-TABLE 11. - Vapour-density of Indium Dichloride. Weigl't of chloride Expt. 1 2 3 -Vol. of gas dis- apour-dellsity Temperature determina-Temp. of the measur-ing tube. tion. Calcu-lat ed tempera-ture of expt. --Vol. of gas ex-pcllecl. C.C. -96 * 440 100 -608 -in and 760 1 mm. grams. found. Vapo iir- d en si t y deter-mi natio n.Remarks. --In platinum } cylinder. I n porcelain. The calculated vapour-density of the chloride InClz = 6.362. Indium dichloride which is incapable of existing in aqueous solution is a very stable compound a t high temperatures. At 9>8", the vapour-density foiind is somewhat higher than is required by the formula InCl, but a t higher temperatures it is quite normal. Indium MonochZoride InC1.-The circumstance that metallic indium when gently heated in gaseous hydrogen chloride the metal being in excess forms a dark-red liquid led us to believe in the existence of a lower chloride probably identical with the product noticed by Clemens Winkler on treating the metal with chlorine but which he did not examine. In order to obt,ain this in a state of purity a weighed quantity of indium was converted into the dichloride which was then distilled in a current of carbon dioxide into a neighbouring portion of the glass tube where a weighed quantity of the metal somewhat larger than the first had been already placed; the small tube was then fused off on both sides of it.On heating this tube in the naked flame reaction set in the liquid becoming blood-red and forming numerous drops like bromine which adhered to the glms. On cooling these solidified to vitreous or radiated masses aomewhat resembling haematite in appearance. The slight excess of metallic indium employed remained as a small button after the chloride had been volatiLsed in a curren AND THE VAPOUR-DESSITIES OF INDIUN ETC. 821 Weight of chloride in grams.of carbon dioxide as will be seen in the synthesis described here-after. I n thin layers the new chloride when melted is a liquid of a besutiful red in thicker layers it is almost black It soon attracts moisture from t-he air deliquesces and decomposes gradually into indium trichloride and metallic indium becoming grey in con-sequence. The monochloride is at once decomposed by water into the trichloride and metal the reaction doubtless taking place according to the following equation :-3InC1 = TnC1 + In,, the same remarks applying as in the case of the similar decom. position of indium dichloride. Synthesis.-O.llOO gram of metallic indium was heated in gaseous hydrogen chloride to convert it into the dichloride. When this had acted on 0.1160 gram of indium in an atmosphere of carbon dioxide in a sealed tube the red chloride formed was distilled off leaving a metallic button which weighed 0*0050 gram instead of 0.0060 gram.Now 0-2200 gram of indium should yield 0.2886 gram of indium mono-chloride; and in fact 0.2741 gram of this chloride was obtained The difference represents the loss occasioned by the distillation. ArzaZysis.-O*l656 gram of indium monochloride after decom-position with boiled water gave i n the solution 0.1606 gram of silver chloride corresponding with 0.0397 gram of chlorine and 0.0575 gram of indium oxide corresponding with 0.04746 gram of indium. The metallic indium which separated weighed 0.0190 gram. Expressed centesimally the results are-50’82} 76.23 23-98 23.77 100.31 100~00 Residual metallic indium.. 47.72 } ,6.33 Indium as oxide i n solution 28.62 25-41 Chlorine . . . . . . . . . . . . . . . . __.- -Of gas Vapour-$:p:$’ density and 760 mm. found‘ TABLE 111.- Vapour-density of Indium Monochloride. Experi-ment. 1 2 3 Temperatiire employed. w - 7 1300-1400” 1100-1150 1200-1 so3 Vapour-density determination. 0 -0660 9 ‘224 5 ‘534 0.0536 1 7.826 1 5.296 0.0549 7 -894 5 *377 11 In porcelain. VOL. LIII. 3 822 NILSON AND PZTTERSSON NEW CHLORIDES OF INDIUM, The calculated vapour-density of the chloride InCl = 5.140. The substance employed in Experiment 1 was not prepared in a, sealed tube in the manner above described but by the repeated dis-tillation of the dichloride over metallic indium in a current of carbon dioxide.This sample consequently in a11 probability contained some unchanged dichloride which accounts f o r the vapour-density found being somewhat too high. The vapour-density determinations given above place it oeyond doubt that the moiiochloride dichloride and trichloride of indium exist as definite compounds. The manner in which they are decom-posed by water is also of especial interest’ for with the exception of the lower chloride of gold there is scarcely an instance known in which a chloride is decomposed by water with separation of the metal and formation of a higher chloride. This behaviour of indium monochloride and dichloride shows us that indium in its solubIe compound is decidedly and exclusively a trivalent element although i t is capable of existing in combination at a high temperature as a univalent or bivalent element.Chlorides of Gal lizcm. Lecoq de Boisbaudran (Wurtz Dictionn,aire de Gkirnie Article ‘‘ Gallium ” by Lecoq de Boisbaudran) the discoverer of this element, has pointed out the existence of two different chlorides of gallium. He is inclined to regard the higher chloride as gallium hexachloride, GaR,Cls and the lower as gallium dichloride GaCI, in accordance with the views held at the time of his discovery as to the constitution of the iron chlorides and of aluminium chloride. The vapour-density determinations which he made by Dumas’ method with the higher chloride seemed also to confirm the composition assigned to it. After the introduction of the air displacement method however Friedel obtained values which indicated that this formula should be halved.13.4 at 247’7 ‘O’O ” 350 \boiling point 215-220”. Lecoq de Boisbaudran. Friedel. { 8.5 , 350 J 7% , 440 6.6 , 440 The calculated vapour-density for Ga$ls = 12.16 that for GaC& = 6-08. In order to place us in a position to decide on the true formula of this chloride Lecoq de Boisbaudran placed at our disposal the necessary material. As the vapour-density of the lower galliu AND THE VAPOUR-DENSITIES OF INDLUM ETC. 823 Experi-went. chloride is quite unknown we have also instituted a couple of experi-ments with it. Hydrogen. Found. I Calculated. Weight of metal. Gallium Trickloride GaC&.-Like aluminium bnt unlike indium, metallic gallium yields the trichloride when it is heated in gaseous hydrogen chloride.The reaction takes place at a gentle heat ; but at the ordinary temperature the metal remains quite unchanged if the gas is dry and free from air. If warmed a highly refractive liquid is formed with evolution of hydrogen and if more strongly heated this volatilises and is deposited in the tube close to the flame in long needle-shaped crystals the metal is completely converted into the chloride and leaves no residue. As soon however as the product is distilled in a current of carbon dioxide to free it from adhering hydrogen chloride a trace of a brownish-yellow slightly volatile residue is noticed in the tube ; this instantly becomes colour-less in gaseous hydrogen chloride. This appearance we believe can scarcely be ascribed to impurity in the metal employed but should more probably be regarded as due to a lower chloride corresponding with indium monochloride.Lecoy de Boisbaudran on treating gallium dichloride with much water obtained a brown subtance which i f left under water evolved gas slowly and rapidly if dissolved in hydrochloric acid. This com-pound which was not analysed he regarded as i n all probability a lower oxide of gallium. We are of opinion however that it was perhaps the monochloride of gallium which might be formed in the f'il lo wing way-2GaC1 = GaCIR + GaCI, and may be identical with the brownish-yellow substance which is formed in small quantity in the preparation of gallium dichloride (p. 824). Its true nature however must remain an open question until larger quantities have been obtained.Synthesis of Ga'lium Tric1doride.-In two experiments metallic gallium gave the following results when heated in dry hydrogen chloride free from air :-1 0 -0236 gram. 2 I 0.0220 ,, I Hydrogen at 0" and 760 mm. 11 *57 C.C. 0 *001036 gram. 11 '30 C.C. 0*001012 gram. 10.86 , 0.000973 , 10.54 , 0*000944 ,, I- I.--'-- I -82-1 NILSON AND PETTERSSON NEW CHLORIDES OF INDIUM, TABLE IV.- Vapour-densit!y of Gallium Trichloride. Vapour-density found. Vrzpour-density determination. Remarks. --- I I in 1 2 8 4 350" 0 *0543 4.40 0.0521 606 0 '0562 1000-1100 0.0567 I I Vol. of gas displaced, C.O. at 0" and 760 mm. 4 948 6 -5% 7.074 8 *457 -I 8 *846 6.118 6 *144 5 -185 In mercury vrtpour.Insulphur ,, In stannous chloride vrzpour. In furnace with platinum vessel. The vapour-density calculated on the formula GaCIB is = 6.081. If our determinations are compared with the earlier ones it is instructive to observe that gallium trichloride like beryllium chloride, aluminium chloride &c. has a vapour-density higher than the normal at the lowest temperature employed; but even at 40" it has the normal value and retains it also a t 606" in the vapour of stannous chloride ; at still higher temperatures it suffers dissociation like many other trichlorides. The difference in the values obtained by Lecoq de Boisbandran and Friedel at one and the same temperahre must be set down to the different methods employed by the two investi-gators.GaZZium DichZoride GaCl,.-According to Lecoq de Boisbaudran this compound may be obtained by heating the metal in chlorine, taking care to keep the gallium in excess. We preferred however to prepare it from weighed quantities of the metal in exactly the same way as indium monochloride. We allowed 0.0352 gram of gallium to act on the trichloride formed from 0.01 70 gram of gallium a quantity more than sufiicient to convert it into dichloride or even into monochloride. After long-continued heating and complet,ion of the reaction we opened the tube and distilled out the colourless chloride in a stream of carbon dioxide a button of gallium being left which weighd 0.0225 gram. The trichloride consequently had taken u p 0.0127 gram of gallium, or 0.0042 gram more of the metal than corresponds with the formula QaC1,.We noticed however not only that several exceedingly minute globules of gallium remained in the tube and could not be extracted but also that the tube was covered with a thin coating of the brownish-yellow substance previously mentioned when speakin AND THE VAPOUR-DENSITIES OF INDIUM ETC. 825 Weight of chloride in grams. of the trichloride ; this we are inclined to regard as a rather unstable gallium monochloride as on exposure to an atmosphere of hydrogen chloride or chlorine it instantaneously becomes white and is con-verted into colourless drops which solidify to a crystalline mass on cooling volatilise when heated and in short exhibit all the pro-perties of gallium trichloride.Gallium dichioride when melted forms a limpid refractive liquid which on cooling solidifies and becomes crystalline ; it distils when heated and is deposited on the tube near the flame in a solid state. Its vapour like that of the trichloride fumes in the air on account of its attraction for the moisture in it. E$:$r Va,pour-c.c. at oo density and 760 mm. found' TABLE V.-Vapour-density of Gallium Diddoride. Experi-ment. -1 2 Temperature employed. looo-llooo 1300-1eo0 Vapour-density determination. -I--)-0 -0402 4 -823 0.8464 1::t:o" I 3'568 Remarks. --} In porcelain. The theoretical vapour-density of the chioride GaCI2 is = 4.859. Gallium dichloride does not appear to be so stable as indium dichloride at a high temperature.Experiment 2 shows that i t is decomposed at a full white heat which is most probably to be attri-buted to the formation of a gallium monochloride and free chlorine. According to our experience dissociation of a chloride with increase of temperature only takes place when a lower chloride exists which is more stable at the higher temperature. For example in our experiments with beryllium chloride indium monochloride ferrous chloride &c. we always obtained values corresponding with those calculated however high the temperature might be; whilst all the chlorides we have examined of which lower chlorine compounds exist gave at the highest temperatures values considerably lower than the theoretical. Remarks on the Palency of the Elements of the Aluminium Group.The elements of the third group whose chlorides we have already examined give the following chlorine compound8 : 826 NILSON AXD PETTERSSON NEW CHLORIDES OF INDIUX, Gallium . . Indium . . . . . Monochloride. Not known for cer-InCl reddish-yellow, when fusedretidish-brown tainty - I -Aluminium I Wanting . . . . Dichloride. -I_-Wanting . ,. . . . GaCl, colourless fusi-ble crystalline InCl, colourless when melted amber-yellow Trichloride. --AlCl, colourless in-fusible a t the ordi-nary pressure crys-talline. GaCI, colourless fusi-ble crystalline. InCl, colourlesa in-fusible crystalline plates. It is noteworthy that aluminium when treated with hydrogen chloride displaces 3 atoms of hydrogen indium 2 atoms and thallium (according to Lepsius! Ber.1888 21 556) 1 atom. In this group also there is a decided tendency to form a larger number of chlorine compounds as the atomic weight increases. Aluminium forms but 1 chloride whilst gallium yields 2 indium 3 and thallium 4 chlorides TlCl T1,C13 T1C12 and TlCl,. The question which is so interesting theoretically as t o whether aluminium behaves towards chlorine exclusively as a triad has already been touched npoii by some French experimentalists. I n researches on the action of aluminium trichloride on metallic aluminium a t various temperatures high and low instituted by Troost and Hautefeuille (Conzpt. rend. 1885 100,1221) Hautefeuille and Perrey (ibid. 1885 100 1220) and Friedel and Roux (ibid. 1885, 100 1191) it has been shown that a grey coating is formed on the metal or a lustrous deposit on the sides of the tube.The origin of these substances however must be assigned to the presence of traces of silicon as silicon could be detected in all or a t least in most instances. Friedel and ROUX indeed state that when metallic aluminium is acted on by silicon chloride a t a red heat it yields aluminium chloride with separation of silicon. I n order to settle this question we prepared aluminium triohloride from a weighed quantity of aluminium by heating it in gaseous hydrogen chloride and then allowed i t to act on a weighed quantity of very thin aluminium foil in an atmosphere of carbon dioxide in ft sealed tube which was heated over a bare flame. By long-continued action of the chloride which under the high pressure fused to a limpid liquid the aluminium foil did indeed become superficially of a greyish colour b u t i t was otherwise unchanged.I f one end of the short tube was now heated the chloride condensed in the other colder end in beautiful well-defined six-sided tables identical in appearance with the crystals of indium trichloride. At the close of the experiment whendl the chloride had sublimed to the other en AND THE VAPOUR-DENSITIES OF INDIUM ETC. 827 of the tube the Iat-ter was opened and the residual aluminium foil with adhering traces of aluminium chloride taken out and thrown into water when after a momentary evolution of gas it completely regained its metallic lustre ; on drying and weighing i t was found to have diminished in weight by a few tenths of a milligram only.This inconsiderable loss is moreover easily explained by the action of traces of silicon chloride on the aluminium foil forming aluminium chloride and silicide of aluminium which gave the grey appearance to the foil. There is no reason to believe therefore in the existence of any chloride of aluminium lower than AlCI,; Friedel and Rfoux indeed (Zoc. cif.) are inclined to regard the amount of chlorine in the above-mentioned grey coating as evidence that a second chloride of aluminium may be found but as Hautefeuille and Perrey have shown that various oxychlorides of aluminium are formed when a mixture of chloride of aluminium and oxygen is passed over the heated metal the low chlorine found may be due to oxygen not having been completely excluded in Friedel and ROUX’S experi-ments.Moreover the vapour-density of a1 uminium trichloride as determined by us a t various tempera-tures (Zoc. nit. see also the remarks under GaZZiim Dichloride p. 825) is evidence of the non-existence of a lower chloride of this element. IV. Ferrous Chloride. V. Meyer (Ber. 1879 12 1193 and 1884 17 1335) has made vapour-density determinations of ferrous chloride under various condi-tions both in an atmosphere of nitrogen and also in hydrogen chloride. I n the last series of experiments he obtained the numbers 6-38 and 6.67 which are almost intermediate between those required by the formulae Fe2C14 and FeC12 namely 8.750 and 4.3Y5. The substance evaporates rapidly a t a yellow heat and the results of the two experiments agree very well.The values found therefore seemed to the experimenter to indicate that a t low temperatures the molecule of ferrous chloride had the formula Fe2C14 and that as the temperature rose it decomposed into FeC1 ; the temperature employed however, was not high enough to enable him to obtain values corresponding with the latter. V. Meyer considered it necessary that the question should be further investigated using platinum vessels especially as the experiments with beryllium chloride have brought to light the disturbing influence involved in the use of glass or porcelain vessels ; he has himself stated however that the glass tube in which the ferrous chloride was sublimed was not a t all acted on.As a t V. Meyer’s suggestion me have undertaken to renew th 828 NILSON AND PETTERSSON NEW CHLORIDE8 OF INDITJM, Experi-ment. investigation of this question we deemed it before all things necessary to determine whether there was really a definite limit to the vapour-density of ferroua chloride a t high temperatures. I n the two experi-ments described below made under somewhat different conditions we obtained results which not only agreed with one another but are in accordance with the values reckoned for the formula FeC12 to the second decimal place and we therefore regard the question of the constitution of Perrous chloride as resolved in so far as that the vapour of this substance at lower temperatures has a complex constitution as shown by V. Meyer's determinations and by analogy with what has been found to be the case with almost all the chlorides we have investigated chloride of beryllium and chloride of aluminium for example whilst at a white heat this decomposes, yielding a molecule of the normal composition FeCI2.In our vapour-density determinations we used vessels of Rayeux porcelain as like V. Meyer we found that the vapour of the chloride during its preparation and sublimation did not act on glass. After the experiment we found that the substance dissolved in water as ferrous chloride without leaving any residue. The substance was prepared by strongly heating soft iron wire in a platinum tube in a current of gaseous hydrogen chloride; the ferrous chloride formed was deposited i n the fore part of the platinum tlibe close to the flame as a radiated crystalline mass of the colour of siderite and was as usual enclosed in a platinum cup for the experiment.Remarks. Temperature employed. Weight of QO1' Of gas Vapour-chloride :!:f2:t density i n grams. 1 and 760 mm. found. I TABLE VI.- Vapour-density of Ferrous Chloride. I Vapour-density determination. 1300-14.00" I 0.0428 7.611 I 1400-1500 1 0.0388 1 6.922 1 $': ~}lnporcelain' The theoretical vapour-density required by the formula FeCI is 4.375. V. Chlorides of Chronzium. As is known chromium forms two compounds with chlorine to which the composition represented by the formula Cr,Cl and CrCl ASD THE VAPOUR-DENSITIES OF INDIUN ETC. 829 has hitherto been assigned as in the case of the Corresponding iron chlorides.The vapour-density of neither of these has been deter-mined and i t seemed to us a matter of no small interest to examine these along with the chlorides of indium gallium and iron. Chromiunz Trichloride CrCl,.-At the last '' Naturforscherver-sarnmlnng" in Wiesbaden we noticed the splendid exhibit of Herr H. Deibel of various chemical preparations amongst which was some resublimed chromium chloride in beautiful crystalline plates. This was obtained from him for the purpose of making the vapour-density determinations and i t gives us pleasure to testify to the purity of the specimen as established by the reduction experiment described on p. 830. The chromium chloride from exposure to the air contained a little hygroscopic moisture but this was removed by gently heating it in dry carbon dioxide free from air before weighing it out for the vapour-density determination.When the platinum cylinder was used in the determination the chromium trichloride was placed in a small platinum cup; in the experiments with the porcelain vessel a piece of a pipe stem was employed which 'was made into a convenient shape before ignition in the blowpipe flame, the one end being closed by a kaolin stopper. TABLE VIL- Vapour-densit y of Chromium Trich loride. -Expt. 1 2 3 4 5 6 7 -Temperature determination. Vol. of gas dis-placed. -95 -802 98 -888 98.824 101 *420 -- - -Tempe-sature of measur-.ng tube 9 *lo -13 -6 9.8 14 *6 --I Calculated tempera-ture. --1065' 1191 1277 1347 L100-120( 1250-135( 13 5 O-140( Vapour-density determinatior Weight ride in grams.of chlo-Vol. of gas displaced, C.C. at 0" 8 760mm 0 9864 0 -0859 0 *0882 0.0'791 0 -0578 0 *0498 0 *0638 10 -890 12 -049 12 * 581 12 -670 7 -890 '7 '440 10.770 Vapour density found. -6 *l35 5 -517 5 *421 4 -827 5 -670 5 -177 4 -580 Remarks. -I n platinum, volatilisa-tion slow. In plati-num voln-tilisation normal. I n porce-lain. I The theoretical vapour-density of the chloride CrCZ is 5.478. Chromium chloride therefore according to the determination made at 1200-1300" has a vapour-density corresponding exactly with the formula of a trichloride. As vaporisation takes place but slowly at VOL.LILI. 3 830 T\;ILSON AKD PETTERSSON NEW CHLORIDES OF I?;L)IUJI. Experi-ment. 1065" when the value found is somewhat too high there is probably no lower temperature a t which a ohloi-ide of the hitherto accepted formula Cr2Cl and vapour-density 10.956 can exist in the gaseous state. At temperatures above 1300" the chloride gave numbers somewhat too low which was to be expected from the dissociation of the tricldoridc into free chlorine and the dichloride which is stable a t high temperatures. Temperature employed. Weight of VO1' Of gas Vapour- Remarks' chlorine :!:'$ density and 760 mm. found* in grams. --- ____-_____)_I______ Chromium Dichloride CrCI,.-This compound was pepared by the reduction of the trichloride with pure and dry hydrogen a t a, temperature so low that the glass tube was not visibly red hot.Reduction Experiment.-O.2927 gram of chromium trichloride gave 0.2274 gram of dichloride a quantity which corresponds accurately with the theoretical. The dichloride employed in the following vapour-density deter-mination was prepared separately for each experiment by introducing a weighed quantity of dry chromium trichloride into a porcelain or platinum cup and reducing it in a stream of hydrogen until hydrogen chloride could no longer be detected in the gas issuing from the mercury seal of the apparatus ; the cup was then intrcduced as quickly as possible into a pair of watch-glasses and weighed. In this way a residue was obtained which always corresponded very accurately with the theoretical. The dichloride prepared in the manner described was white with a greyish shade and volatilised with greater di5culty than any of the metallic chlorides we have examined. It vaporises rather slowly even in the most intense heat of the furnace we could obtain using a Rohrbeck-Luhnie 16-fold blowpipe lamp worked with gas and air a t 2 atmospheres pressure. TABLE VIII.-~apo.zLr-densit~ of Chromium Dichloride. Vapour-density determination. 1 1300-1400° 0.0545 5 -403 7 -800 2 1 1400-1500 1 0*0561 1 5.960 I 7.278 i}hporcelain. 3 1500-1600 0'0351 6 *847 6 -224 The theoretical vapour-density of the chloride CrCl is 4.256. Thus we have found the vapour-density of chromous chloride to b ON SOME DERIVATIVES OF ANTHRAQVINONE. 831 considerably higher than that required by the formula CrCI, but as the three experiments me made show a constant decrease of density with increase of temperature there can be no doubt that chromous chloride in respect of vapour-density is strictly analogous to ferrous chloride except that it volatilises at a much higher temperature and scarcely becomes completely gaseous even at the highest heat attainable