879 LXXX.-Contributions to t h e Chemistry of t h e Cerite Metals. 111. By B. BRAUNER, Ph.D., P.C.S., late Berkeley Fellow of the Owens College, Adjunct and Privatdocent in the Bohemian University, Prague. Introductory Remarks. I N a paper published several years ago (Chem. Xoc. J., Trans., 1882,68, and in extenso, Monatsli. Chem., 1882, 1-60), I endeavoured to deter- mine the position of the cerite metals in the periodic system of elements, and arranged them in the order- CeiV. Djiii-v. 139 141-6 146.6 A little later (Monatsh. Cliern., 1882, 486, and Trans., 1883, 278), I succeeded in proving that the didymium from cerite is a nziz- ture, and that the atomic weight of the true didymium is a t most Di = 145.4. For lanthanum I found a t the same time the number La = 138.28. Working with a more abundant supply of the rare material, Cleve, whilst confirming my number for lanthanum, that is, 138.28 (BuZl.SOC. Chirn., 39, 251, 289), found didymium to be Di = 142.3-142.4. If, as has hitherto been the case, the number found for the atomic weight of cerium by Biihrig (J. pr. Chem. 120, 222), namely, Ce = 141.6, be admitted to represent the truth, the order of the atomic weights of the said elementsin the system would be as follows :- La. Ce. Di. 138.2 141.6 142.3 L-.---JL- ----J Difference.. . . 3.4 0.7 It must be admitted that the lanthanum and didymium preparations with which Cleve determined the atomic weights of these elements were pure and homogeneous, and, as the method used by him does not involve any apparent source of error, we are justified in regarding his numbers as very nearly representing the truth.It is, however, very evident that the difference between the atomic weights of ]antha- niim and cerium (3.4) is much larger than that between cerium and didymium (0.7), and the question therefore arose whether Biihrig’s number (Ce = 141.6) is really the atomic weight of cerium, as is generally believed, especially as Wolf (XiZZ. Anzer. J. [2], 46, 53- 62) has found the atomic weight of thepurest cerium to be Ce = 137,880 BRAUNER : CONTRIBUTIONS TO THE a much lower number than Biihrig’s. tigation is to answer this question. The object of the present inves- Historical Review. Since 1816, the atomic weight of cerium has been determined by many chemists, but the numbers found, especially those which seem to be the most trustworthF, differ more widely from each other than niight be expected.The following is a short review of the determi- iiations made. For details I must refer partly to the origiual papers, 1)artly to the works recently published by G. F. Becker (Constants of LVature: Part IV, Washington, 1880); F. W. Clarke (Constants qf Nature, Part V, Washington, 1882); Lothar Bieyer and Seubert (Die Atomgewichte der Elemente, Lcipzig, 1883) ; and by Ostwald (Lehrbuch der Allgemeiiien Chemie, Leipzig, 1884). My comments on the individual determinations form a special chapter of the present paper. I should state that I call the oxide CezO, and its salts cerous, and the oxide CeOz ceric and not cerium peroxide, as is done by some chemists, as the latter name must be reserved for the oxide Ce03, recently investigated by L.de Boisbaudran (Conzpt. rend., 100, 605), and Clew (Bull. SOC. Chinz., 43, 5 3 ) . The numbers below refer to 0 = 16. The number to be deduced from Hisinger’s (Bchweig., 17, 424 ; Pogg. An?&., 8, 186) experiments, made in 1814-16, is Ce = 137.9. The method is not given, and his cerium mas contaminated with lan- thanum and didymium, the existence of the last, two elements not being known a t that time. Beringer’s (Annalm, 42,134) ceri urn salts were also rose-coloured. The numbers obtained by him were Ce = 142.3 from tlie relation of ceric oxide to silver chloride ; Ce = 142.3 from t1:at of ceric oxide to barium sulphate ; and Ce = 141.6 from the combustion of cerous forniate.Rammelsberg’s (Pogg. Ann., 55, 6 5 ) analysis of anhydrous cerous sulphate, made in 1842, gives Ce = 154.3. Hermann (J. p r . Cliein.. 30, 184) in 1843, determined the relation of anhydrous cerous sulphate to barium sulpbate. His numbers give Ce = 139.4. FromMarignac’s (Ann. Clhn. Phys. [3], 27,209 ; Arch. Sci. Ph. Nut. [ 11, 8, 273) volumetric analysis of cerous sulphate solutions by means of barium chloride, it follows that Ce = 141.8 ; whilst the relation of anhydrous cerous sulphate to bsyium sulphitte, Ce is 141.6. Later on (Ann. Chinz. Phys. [3], 38, 148), the same chemist gave the pre- ference to the number Ce = 137.7, but without giving any experi- meiitd e 1-idence.CHEMISTRS OF TRE CERITE METALS. 881 Jegel, in 1858 (Annalen, 105, 45), deduced the number Ce = 1381, from the combustion of cerous oxalate.An analysis of the sulphate gave Ce = 137.8. I n both cases, the ceric oxide was analyseil iodo- metrically. Lothar Meyer and Seubert calculate, instead of the above num- bers, Ce = 137.4 and 138.3. A combustion of the oxalate, made by Rammelsberg (Pogg. Ann., 108,44) in 1859, gave Ce = 138.1. All the above determinations were made with cerium preparations containing larger or smaller quantities of foreign earths, for the ceric oxide obtained was more or less brown coloured. In 1867, an extended investigation of the subject was undertaken by C. Wolf (Zoc. cit.) in Bunsen’s laboratory, but unfortunately, it remained unfinished in consequence of his premature death, and has, up to the present, never been again taken up.The following details from Wolf’s diary, pub- lished after his death by Genth, may be given here, as being important in relation to the present question. Ceric nitrate prepared from the crude oxides, was decomposed by pouring its aqueous solution into boiling water containing some sul- phuric acid, and the precipitate of basic nitrate and sulphate, N, was converted into cerous sulphate. This was recrystallised at least ten times. Wolf heated the sulphate over a, small flame in a double platinum crucible, in order to determine the amount of water of crys- tallisation (by loss of weight), and from the aqueous solution of the anhydrous sulphnte thus obtained the oxalate was precipitated by a boiling concentrated solution of oxalic acid. This, by careful igni- tion, was converted into ceric oxide.In the filtrate, the sulphuric acid was determined as barium sulphate. The excess of oxygen in Ce02 ( Cez04) over Ce20s was determined iodometrically. From these four data (H,O, Ce02, Ce203, and SOs) the composition of the anhy- drous cerous sulphate was first calculated, and from this the equiva- lent of cerium. As the ceric oxide obtained in the first series of determinations had a, brownish colour, part of the precipitate N was purified by dis- solving and precipitating with boiling water; and in this way the precipitate N, was obtained. The sulphate obtained from it gave a much paler ceric oxide. The precipitates N,, N,,, and Ns were obtained by Wolf in the same way, and the colour of the oxide was in each case paler than in the preceding.Wolf calls the oxide from N7 aZrnost white, and that from Ns white. I t is a very remarkable circumstance that, after every purification, as the oxide became more nearlg white, the equivalent of cerium was fonnd to decrease. VOL. XLVII. 3 P882 BRAUNER : COKTRIBUTIONS TO THE I n Wolf’s original paper the following numbers are calculated for the equivalent of cerium :- I. 11. IV. V. 46-187 45.760 45.699 45.664 This multiplied by 3 would represent the atomic weight as- I. 11. IV. V. 138.561 137.280 137.097 136.992 If, according to Clarke, the atomic weight be calculated from the relation of Ce2(SOJ3 to 2Ce02, as involving the least experimental error, the following numbers will be obtained (for 0 = 16, S = 32.06) :- I. I I. IV. V. Ce = 139.64 138.27 138.04 138.00 From these experiments Wolf concluded that the remarkable dimi- nution of the atomic weight of cerium could not be due to the elimi- nation of didymium only, but that the first portions may have contained a foreign substance.Wing (SiZZ. Amer. J. [2], 49, 358), in 1870, repeated Wolis experi- ments and process of purification, without, however, going further, and hispurest material gave Ce = 137.85. From Wolf’s and Wing’s experiments Clarke calculates as a mean Ce = 138.039. Buhrig’s (Zoc. cit.) experiments made in 1875, with a material free from didymium, in which large quantities of cerons oxalate were analysed by combustion in a current of oxygen, gave a remarkably high number, namely Ce = 141.523 (Clarke). His ceric oxide varied from yellow to salmon colour.Later on I shall return to the objec- tions which he raised against the analysis of cerous sulphate. While my present investigation, which has occupied me for some time past, was in progress, H. Robinson (Proc. Roy. Soc., 37, 150) published his paper on the same subject. The purification of his cerium salts was effected by Gibbs’s method. The cerium solution, free from didymium, finally obtained, was precipitated with oxalic acid, and the air-dried oxaIate was converted into the chloride by heating it in a current of hydrogen chloride. The cerous chloride, free from water and hydrogen chloride, was analysed volumetrically with a silver solution, according to Stas’ method. As a mean of the results of seven experiments (reduced to a vacuum), Robinson found- Ce = 140.2593 (0 = 16) or Ce = 139.9035 (H = 1).This uumber, as I shall show later on, is to be considered exact, forOHEMISTRY OF THE CERITE METALS. 883 in the method used by Robinson, all possible experimental errors are reduced to a minimum ; but I have nevertheless continued my experi- ments for several reasons. Firstly, the method I have used is diffe- rent from that adopted by Robinson, and our experiments therefore mutually control each other. Secondly, Wolf's experiments challenge investigation as regards the homogeneous character of the material, and as Robinson did not touch this question, it was all the more necessary for me to devote my attention to it. Experience has shown, especially in the case of the rare earth metals, that the atomic weights, determined even by the most trustworthy methods, may differ by several units from the truth, if the material used consisted of a mixture of earths.The history of scandium, yttrium, didgmium, and erbium may be quoted as examples. Method of Investigation. I n the operations of dissolvhg, evaporating, and boiling, only such vessels were used as had been treated for some time beforehand with acids. The acids, viz., hydrochloric, nitric, and sulphuric, were distilled shortly before use from a platinum retort, and kept in well-stopped bottles of Bohemian glass. A platinum tube, 60 cm. in length, 2.5 cm. wide, and weighing 600 grams, was used as a condenser, and for tbis 1 am much indebted to the Royal Society for a grant from the Govern- ment Chant Fund. Distilled water was once redistilled by means of the same arrangement, and absolute alcohol was similarly treated, the first and last portions of the distillate being rejected.The filter- paper used was kept in contact with warm hydrochloric acid for some time, and then well washed. In weighing, a fine balance, made by Verbeeck and Peckholdt, of Dresden, was used. In order to prevent the very sensible effect of radiant heat during weighing, the balance was covered on all sides with thick flannel, so that only the scale could be seen. To make the scale more visible, light mag reflected npan it from the sides by mirrors, and a t night the light of distant gas flames was concentrated upon i t by lenses. If the method of vibrations be used in weighing, and the observation of the " point of rest" be often repeated, the weight of a body weighing about 50 grams may be determined accurately wiihin O*OOOr)l to r).OOUO3 gram.The platinum crucibles were weighed in small thin glass bottles, and a similar vessel with platinum was used as counterpoise. As the weight, of substance used was generally about 2 grams, rarely 5 grams, only a few small weights were used: in this way the errors of weighing were reduced to a minimum. The following was the true weight of the pieces used F 3 P Z884 BRAUNER : CONTRIBUTIONS TO THE 2 = 2.00006 1 = 0.99996 1’= 0.99996 l”= 1.00005 0.5 = 0.49978 0.2 = 0.19992 0.1 = 0.10001 0.1’ = 0~10001 0.05 = 0,04996 0.02 = 0.02012 0.01 = 0~01010 0’01’ = 0~01012 rider= 0.Ol.002 CeOz (sp. gr. = 6.7) = 0.000172 gram.Ce2(S04)3 (sp. gr. = 3.9) = 0.000303 gram. In order t o obtain pure cerium preparations, I proceeded as follows : -From 2600 grams of cerite, 1380 grams of crude oxides were ob- tained by the method described in a former paper. After dissolving the oxides in moderately concentrated nitric acid and removing the excess of acid by evaporation, the remaining syrup was dissolved in a little water. On pouring this into a large quantity of pure boiling water, almost the whole of the cerium present (about half of the weight of the crude oxides) was precipitated as basic ceric nitrate. This could be easily and quickly washed on a Bunsen funnel with boiling water containing a little nitric acid. I shall call this first pre- cipitate N. For further purification, instead of using sulphuric acid, as is generally done, nitric acid was employed.This has the great advan- tage over the old method, that the excess of acid can be very easily removed by evaporation from the solution. If the solution of the resulting ceric nitrate, which is now crystalline, be again poured into boiling water, the filtrate will contain, besides the impurities which we wish to remove, much less cerium in solution than when sulphuric acid is used. This almost neutral ceric nitrate dissolves easily in a very large quantity of water, without undergoing decomposition. This method, especially the removal of tlie excess of acid by evapo- ration, enabled me to carry the purification much further than Wolf was able to do. In using nitric acid, the quantity of cerium is dimi- nished far less rapidlj by each precipitation than is the case with sulphuric acid.Whilst Wolf could repeat the process of precipitation only five times, I could obtain, by repeating this tedious process withOHEMISTRY OF THE OERITE METALS. 885 a part of the first precipitate only, the following series of precipitates (N) and corresponding filtrates (F), the last being kept separate :- N N, Nz NS N, N5 Ns N7 N* Nil NlO Nl, F Fi Fz F S Fd F, F, F7 F, F, Fio Fii. Now the question arose which cerium compounds were to be used for the atomio weight determination. After long deliberation, 1 chose the anhydrous cerous snlphate, as-with the exception of cerous chloride, which H. Robinson had succeeded in preparing only after my experiments were in progress-cerium does not form any other compound of definite composition suitable for the purpose.Buhrig rejected cerous sulphate-firstly, because the salt retains free sul- phuric wid from the mother-liquor so energetically that it cannot be fi-eed from i t even by repeated crystallisations ; secondlg, because he could not obtain the anhydrous sulphate, the salt retaining traces of water at a moderate heat, and undergoing partial decomposition when heated to incipient redness ; and thirdly, he remarks, that on preci- pitating the solution of the sulphate with barium chloride and after- wards with oxalic acid, the barium sulphate thrown down cont.ains cerium; whilst the cerium oxalttte, on the other hand, contains barium, but he does not consider that, if the analysis is to be made by precipitation at all, the process may be executed in the inverse order, without fear of committing the above errors.The first source of error was avoided in the following way:- Cerous snlphate, prepared by dissolving baRic cerie nitrate i n dilute sulphuric wid and sulphurous acid, and evaporating the solution in a platinum basin, was heated for some time in a magnesia bath, in order to expel the greater part of the excess of sulphuric acid. The pro- duct was then dissolved in a, small quantity of ice-cold water, the heavy metala (platinum, &c.) precipitated by hydrogen sulphide, and the excess of the latter expelled first, by the use of a water-pump, then by meam of a current of air. I n this way a solut.ion was obtained, from which, on adding three times its volume of absolute alcohol, the whole of the cerium was thrown down in the form of a fine crystalline powder or" the sdt Ce2(S0J3 + 8Hz0.After washing the salt with absolute alcohol, dehydrating at a geut1.e heat, and again precipitating with alcohol, a completely neutral cerous sulphate was obtained. Although free from any excess of sulphuric acid, the salt is not pure, even if it be thrown down from most carefully prepared purified alcohol, i t retains traces of foreign organic matter, probably betaine, and consequently turns yellow or brownish on subsequently heating. I t dissolves in water, also wiih a peculiar feeble empyreumatic odour. To p u ~ f y it, the salt must therefore be once more dissolved in cold water, and after filtering, the solution contained in a beaker is pliinged886 BRAUNER : CONTRIBUTIONS TO THE quickly into boiling water, so as to heat it to 100".If the hot super- saturated solution be now stirred with a glass rod, the salt present is instantly precipitated as a fine crystalline powder of the composi- tion Cez(SOa)3 + 6Hz0. This is collected with the aid of the pump on a platinum cone, and can be at once placed in a bottle, for it dries in a few instants when put on a smooth filter-paper, Originally I had intended to convert the hydrated snlphate by strong calcination into ceric oxide, and to calculate the atomic weight from the relation of the oxide to the hydrated salt, a method which has been used by Nilson and Pettersson (Bey., 13, 1441 and 1453) in determining the atomic weights of beryllium and scandium.Unfor- tunately all cerous sulphates either alter on exposure to the air, or, if they are stable, they seem to include the mother-liquor in small cavities, a property of salts first noticed by Sorby. On heating the neutral solution of the sulphate to 40-50", Marignac's (Zoc. cit.) hydrate, Ce2(S04)3 + 9Hz0, was never obtained, but instead of this, white turbid crystals of the salt with 8H20 always separated out. By spontaneous evaporation in the air at the ordinary temperature, the same salt was obtained in beautiful clear and glistening crystals instead of the salt CeL(S04)3 + 12H20. At loo", the salt with 6Hz0 alone is separated, but it quickly alters in the air. It is possible that the hydrates with 5H,O, 9Hz0, and 12Hz0, which I could not obtain from neutral solutions, crystallise only when free sulphuric acid is present.For these reasons, I was unable to prepare a, cerous sulphate with a definite (theoretical) amount of water of crystallieation which could be used for the atomic weight determination. In only two out of twenty cases in which the water was exactly determined was the theoretical amount of water found, viz., in the clear crystals, Ce2(SOJ3 + 8Hz0, obtained on spontaneous evaporation at the ordi- n a q temperature. As I had to give up the above plan, I tried to prepare the anhy- drous sulphate. It is impossible to obtain it by simply heating the hydrated salt in the air, for at a low temperature the water is not entirely given off, and at about 500" the salt may lose a trace of sulphuric acid, or it absorbs some oxygen and becomes heavier and slightly yellow-colonred by passing partly into ceric salt.But, on following Baubigny's (Compt. rend., 97, 854) example, I found that, at the temperature of boiling sulphur, cerous sulphate entirely loses its water without being decomposed or otherwise altered. Hitherto it has not been pery easy to operate with boiling sulphur, and yet I wanted an arrangement that would allow me to heat the salts for many weeks at 440" without special difficulty. After many trials I succeeded in devising an apparatus for this purpose. A veryCHEMISTRY OF THE CER’CTE METALS. 887 thin glass beaker, 18 cm. long and 7 cm. in diameter, is covered with a, sheet, of thick cardboard, having in its middle a round opening 4 cm.in diameter. To prevent this cardboard from burning, it is soaked repeatedly with alum, soluble glass, or sodium tungstate. Through the opening in the cardboard is passed a test-tube of thin glass, 20 cm. long and 4 cm. wide, the bottom of the tube being 4 cm. above the bottom of the beaker. The beaker is kept suspended in a wire triangle, and its bottom rests on a piece of thin wire gauze. The beaker is first placed on a sand-bath, and about 50 grams of sulphur are fused in it. Then it is placed on the gauze and heated with 2 or 3 Bunsen’s burners, so that the sulphur boils. In this way the whole of the beaker becomes filled with the vapour of boiling sulphur, which condenses on the sides, and flows down again. In order to prevent the upper portion of the beaker from being too strongly heated, the cardboard from carbonisng, and the sulphur-vapour from escaping, it is surrounded with a shorter open glass cylinder (a broad beaker with the bottom cut off) about 12 cm.high and 11 cm. wide, covered at the top by a plate of thin sheet copper (15 cm. square), having in its middle an opening of 7 cm., through which the thin long beaker (the sulphur-bath) passes, so that only its lower two-thirds are heated to the boiling point of sulphur, whilst the upper third, being freely exposed to the air, is prevented from becoming too strongly heated. With such an arrangement, the vapour of boiling sulphur reaches to two-thirds of the height of the bath, and the lower part of the test- tube is surrounded by it.When the cerous sulphate is to be dehydrated, it is placed on a small platinum crucible, and this is suspended by a loop of long thin platinum wire, the upper end of which is bent over the upper edge of t h e test-tube, in order to prevent the crucible from falling into the boiling sulphur, if the test-tube should crack. (Once, before I made this arrangement, the tube cracked, but, although the crucible wa,a plunged for a quarter of an hour in boiling sulphur, it was not altered in appearance or weight after being washed with potash solution.) The sulphur is then heated slowly to its boiling point, and after a while the platinum crucible is surrounded by sulphur-vapour far above its upper edge. Whilst the greater part of the water of crys- tallisation of the sulphate is escaping, the test-tube is kept open until a cold beaker held over no longer shows signs of dew.During the operation, the crucible is covered with a lid having the ear cut o f f , so that there is but very little space between the crucible and the side of the test-tube. When water-vapour ceases to escape, the test-tube is covered with a porcelain crucible lid, and the bath is heated until the weight of the crucible plus sulphate is found to be constant. This is easily effected if after one hour’s heating the salt be stirred with a888 BRAUNER; CONTRIBUTIONS TO THE thick platinum wire and heated for another hour. When the opera- tion is finished, the sulphur which has cooled down so that it no longer takes fire in the air, but is still fused, is poured out into a porcelain basin, for if it is allowed to remain in the beaker i t would certainly crack either as the sulphur cools or on fusing it again.When these precautions are observed, the same quantity of sulphur may be kept boiling in this simple apparatus for half a year or longer, from morning to night, without fear of anything happening t o it. When hydrated cerous sulphate is heated in the above apparatus until the weight is constant, and then allowed to cool in a small desiccator containing phosphorus pentoxide, it is found to be per- fectly anhydrous, without, however, undergoing the slightest decom- position ; the water which escapes from it, and which can be collected on a cool beaker, being neutral and unaffected by barium chloride.I f the anhydrous sulphate be heated to a higher temperature in 8 test-tube, no trace of water is given off, but a mixture of sulphur dioxide and trioxide escapes, so that the decomposition takes place in accordance with the equation Ce20,,3S0, = Ce204 -t 2S03 + SO,. This behaviour affords evidence of the powerful reducing power of cerous oxide, and it will be easily understood that it cannot be pre- pared in the free state. For analysis, the anhydrous sulphate was care- fully heated in a double platinum crucible over the flame of an ordinary burner, until sulphur trioxide fumes were no longer given off. In order to expel the last trace of sulphuric anhydride, I tried heating the salt to a white heat in a Fletcher’s blast furnace with injector, but I had to relinquish this plan, as not only does the cruciblelid becomes welded to the crucible, but the platinum seems to evaporate per- ceptibly at this high temperature; the inner of the two platinum crucibles which is not exposed to the direct flame, losing in a short time as much as 12 mgrms.I f the injector is put in a vertical position, and the air blown in in such a way that the flame, which is generally 30 to 50 cm. long, is converted into a short, hissing, hardly visible one (the flame will be often blown out entirely before the necessary practice is obtained in regulating the air and gas supply), a temperature is obtained in the free air at which an ordinary pretty thick platinum wire fuses instantly. This is certainly the highest temperature obtainable from a mixture of gas and air without a furnace, and I think that this simple arrangement will prove useful to chemists.The cerous sulphate contained in a double platinum crucible loses every trace of its sulphur at this temperature in 10 to 15 minutes, and it will take weeks before the inner crucible loses 0*0001 gram in weight.CHEMISTRY OF THE CERITE METALS. Weight of CeO,. 0 -7717 .7g58 1.03534 1‘1308 889 Loss of Per cent. of (‘Atomic weight.” weight. CeO,. Ce = 0 -5033 o . 5195 ::: z:! :$ :;:} Mean. . 139 -78 0 %7771 60.440 139 ’53 Mean,, 139 .49 0 -7394 60 -44 139 -46) ---------- ----------- Experiments with Mixt uws. Before the above method of operating with boiling sulphur had been worked out, I tried to decide whether the above precipitate Na could be split up into different fractions.For this purpose, one part of it was converted into the neutral sulphate, which was dissolved in water, and then fractionally precipitated with dilute ammonia. I will call the most basic portion remaining in solution A. The precipitate was dissolved in dilube sulphuric acid, and again precipitated. After four precipitations, the least basic portion B was obtained. On heating the hydrated neutral sulphate prepared from the portions A and B high over the flame in a double platinum crucible t o constant weight, Wolf’s “ anhydrous ” sulphate (it is really n o t anhydrous) was obtained, and this was converted into ceric oxide by strong calci- nation. The formulze used for the atomic weight calculation will be given later on. The following results were obtained :- Weight of Ce? (S 0,) 3.These numbers are not absolutely exact, as the water was not entirely expelled, but they may be considered relatively true, for they were determined by the same method. As the “ atomic weight ” was in one case Ce = 139.78, and in the other Ce = 139.49, the question remained open whether the material used was homogeneous. This circumstance suggested an attempt to split up a portion of the precipitate N,, which WAS less pure, but contained only traces of didymium. The basic ceric nitrate from this was converted into cerous sulphate, from which the excess of free acid was removed by alcohol. The aqueous solution of the sulphate was partially precipi- tated by adding to it strong alcohol, drop by drop, and in this way the portions A, B, C, D were obtained.These precipitated sulphates were dried between smooth blotting-paper, and then in a state of fine powder dehydrated in the sulphur-bath and analysed as before, witch the following results :-890 BRAUNER : CONTR1BI;TIONS TO THE Per cent. of “Atomic CeO,. weight .” 60 *591 140 - 34 60 -568 140 -18 60 -545 140 90 i 60.906 142 *65 I ----- Weight of Weight of Loss of Ce,(SdA. I CeO,. 1 weight. A.. . . 1.3601 B . . . . 2.4780 C . , .. 2.3191 D.... 1’5179 0.8241 0 *5360 1 * 5 0 N 0 -9771 1.4041 0 -9150 0’9245 0 -5934 The difference in the percentage of ceric oxide of the single frac- tions is far more striking than in the first case. The oxide obtaiued from fraction D had a peculiar flesh colour mixed with a pale orange, and it is very remarkable that it turned grey under the influence of light, whereas pure ceric oxide is white with a yellowish tint, and does not alter when exposed to light. The portions A, €3, and C were far less orange than I).I f we assume that the loss of weight on ignition is represented by the same equation as in the case of pure cerous sulphate, viz., Ce203,3S03- (3S03-0) = 2Ce02, the “ atomic weight ” of the earth metal in the fraction D would be R = 142.65. It will be seen from the numbers given hereafter, that the numbers found for A, B, and C represent very nearly the true atomic weight of cerium. The high percentage of oxide in fraction D cannot be due to the presence of didymiurn, because anhydrous sulphate of didymium contains only 58.09 per cent. of the oxide, and the per- oxide would be completely decomposed at such a high temperature.Still, it might be due to the pretience of thorium, its anhydrous sul- phate containing 62.42 per cent. of the earth. In order to entirely exclude any thorium possibly pyesent, the following process was used. The earths contained in the filtrates F1, F,, F,, and F4 were preci- pitated with potassium hydroxide, and the precipitate, consisting chiefly of cerons hydroxide, was suspended in strong caustic potash solution, and treated for three days with chlorine, in order to remove the greater part of the lanthanum and didymium present. After thoroughly washing, the precipitated ceric hydroxide was dissolved in nit.ric oxide, and the excess of acid removed by evaporation, when a gelatinous mass of ceric nitrate was obtained, Its Rolution in cold water was poured into boiling water, and so the greater part of the purer cerium salt was thrown down a8 basic nitrate. From the filtrake containing impure cerium salt, the earths were thrown down with ammonia and converted into the sulphates.After dissolving in five parts of ice-cold water and separating the heavy metals with hydrogen sulphide, the solution was heated at 60-70” for some time. At this temperature most of the cerous sulphate separates out, and if any thorium be present, its sulphate will, according to Nilson’s experiments (Bey., 15, SSlU), also crystallise out. The same holdsCKEMISTRY OF THE CERITE METALS. 891 Loss of weight. good as regards lanthanum sulphate. The last mother-liquor contains, besides cerium, some didymium.It was split up into two parts by addiug strong alcohol, and the Zast precipitate (Le., the most soluble salt) was recrystallised from water and used for the atomic weight determination :- Per cent. of R02. ROz. 0 '7190 0 -9299 0 -3034 1 -8649 2 -4310 0 -7863 61 *M 61 *135 61 -414 Mean.. . . 61 '331 ---- 1 -1459 1'5011 0 *dl829 I I I This strikingly high percenta,ge of oxide, if calculated out i n accordance with the above equation, would correspond with an *' atomic weight " of R1ii-'v = 145-72. The material wa5 not entirely free from didymium, and, although it was improbable, for the reasons stated, that its presence could render the percentage of the oxide higher, experiments were under- taken with cerous -sulphate artificially contaminated with a didymium salt, in order to decide the question.The following tive results were obtained :- Sulphate. Oxide. Per cent. of oxide. 0.6941 0.4203 60-55 1,6320 0.9878 60.33 little nega- I could not yet decide the question as to whether the presence of one of the many rare earths, some of which have been but very little studied up to this time, makes the atomic weight of impure cerium higher, neither could I find out the reason of the high numbers obtained. In any case, it is seen from the above experiments, that under certain conditions " cerium '' may consist of a mixture. The nature of this admixture must be ascertained by further experiments ; but before this is done we must become acquainted with the pro- perties of really pure cerium.Experiments with Pure Material. The object of the following series of experiments was to ascertain how far cerium has to be purified in order to furnish a truly homo- geneous product. For the experiments, the following of the above- mentioned precipitates of basic nitrate and the corresponding filtrates were used :-892 2 -2670 2 -1869 2 *5807 2 *0149 1 * 6519 1 -6193 ---- --I_-- ---- BRAUNER: CONTRLBUTIONS TO THE 1,3749 1 *3252 1 -5648 1.2216 1 -0013 0 -9815 NS NB N, N8 Ng NIO NIL FS E', F+J F8 Fg FIO FII. These were found to be entirely free from lanthanum and didymium. It will be seen from an inspection of this series, and a considera- tion of the process of fractionation, that the cerium in the Jiltrates is far more strongly " fractionated " than the precipitates, especially as the quantity of earth in each filtrate is much smaller than that in the corresponding precipitate. For this reason the filtrates alone were used to determine the question of homogeneity, and, as no funda- mental stoechiometric numbers were to be obtained, the following simple process was used for the preparation of cerous sulphate from them:-The filtrates, after adding to them some of the sulphuric acid, and later on some sulphurous acid, were evaporated to dryness in a platinum basin, and from the sulphates obtained in this way the excess of snlphuric acid was driven of€ by heating the residue in a magnesia-bath; it was then dissolved in water, and the sulphate thrown down by alcohol.After repeating this process, the anhydrous sulphate was dissolved in water, and the neutral aulphate separated out by heating the solution at 100".The dehydration in the sulphur- bath and the analysis were carried on in the way described above. The results are given in uncorrected numbers :- I--- I --- 2 *3797 1 -4.424 { 1 1.8258 1 1.1073 .............. Fs F6 + F, .......... The two together.. .. F8 + F g . .......... The two together.. .. F1, + F11 .......... The two together.. .. Per cent. of Ce02. 60.614 60 *648 60 *649 (maximum). 60 -597 (minimum). 60.634 60 -627 ----- 60 -615 60 -612 Mean.. .. 60 -624 The material used for these experiments, although containing no other earth metal .but cerium, had been too much in contact with filter-paper, with glass and porcelain vessels in strongly acid solution, and with air, so that the nunibers were neither corrected, nor were they used for the calculations of the atomic weight.But, as they are for all fractions exactly the mme, showing neither a regular increase nor decrease, and differ from the mean number at the out- side by * 0.025 of a unit in percentage, the conclusion must be drawnCHEMISTRY OF THE CERITE METALS. 893 that the cerium of the filtrates F6 to FI1 is a perfectly homogeneous body. Now it remained only to use the last of all the precipitates, viz., Nil, containing the best purified cerium, for the deJinite atomic weight determination, and to ascertain whether, on comparing it with the filtrates mentioned above, it furnishes the same results. E'or the same purpose, in I and I1 (see table), part of the precipitate N1, was used.The remaining 21 determinations were made with a material obtained from difeyent preparations, in order t o meet, the objection that only one kind of material had been used. For, in such a case, even if the results agreed, they might not be accurate. It is almost unnecessary to remark that the following series of determinations was made with the greatest possible care, and that they involve the correction both of weights and for displaced air. I did not exclude one single experi- ment which I made, not even numbers XIV and XXIIT, although they may be considered a little too low, for the difference of weight to which these low numbers are due, falls within the allowed experi- mental errors. Number of experiment. I.. 11.. 111.. IV.. V..TI.. VII.. VIII.. IX.. X . XI.. XII.. XIII.. XIV.. xv.. XVI.. XVII.. XVIIT., XIX.. xx.. XXI.. XXII.. XXIII.. Total.. . . --- -- -- Cerous sulphate. 2 - 16769 2.43030 2 * 07820 2.21206 1.28448 1.95540 2 * 46486 2 *04181 2.17714 2 -09138 2.21401 2 * 44947 2.22977 2 * 73662 2-62614 1 -67544 1 a 57655 2 * 72882 2 10455 2 * 10735 2 * 43 557 3.01369 4.97694 53.77424 -- --- Ceric oxide. --- 1 * 31296 1 * 47205 1 a 25860 1 *33989 0.77845 1 * 18436 1 - 49290 1'23733 1'31878 1 -26654 1 * 34139 1 * 48367 1 '35073 1 *65699 1 59050 1 - 0 1470 0'95540 1'65266 1 * 27476 1-27698 1-47517 1.82524 3 -01372 32 -57367 Loss by calcination. --- 0.85473 0-95825 0'81960 0.87217 0 - 50603 0.77104 0 * 97196 0 * 80448 0.85836 0 82484 0.87262 0.96580 0.87904 1.07963 1'03564 0 a 66074 0 * 62115 1 * 07626 0.88979 0.83037 0 * 96040 1.96322 21.20057 1 * 18845 -~ Percentage of CeOz.60.5695 60.5707 60.5620 60-5721 60 * 6043 60.5687 60 * 5673 60.5997 60.5739 60.5600 60 5863 60.5'71 1 60-5771 60 -5488 60 -5642 60- 5632 60.6007 60.5600 60.5716 60 * 5965 60- 5678 60 * 5649 60.5537 -- 60.5747 Atomic weight of cerium. 140 - 183 140.191 140 * 128 140-201 140.176 140- 167 140.400 140.215 140' 11 4 140 * 304 140.194 140.237 140.033 (min.) 140 * la 140.137 14Q * 407 140.110 140 * 198 140.377 140 * 170 140*150 14Q * 068 140.433 (mx.) 140 * 2210 For the calculation of the atomic weight Lothnr Meyer's and Seubert's example was followed, viz., the several amounts of sub-894 BRAUNER : CONTRIBUTIONS TO THE stance weighed were added together, and the atomic weight calculated from the totals according to the following formula.(It has been proved mathematically by Ostwald, that such a method of calculation is correct.) Ce2(S04), : 2Ce02 = 53+77424 : 32.57367 grams, when if Cer(SO& = 100, CeOz = 60.5747. grams. Difference of Ce2(S04), and Ce02 (loss of weight) = 21.20057 32.57367 x (3303-0) 2CeO2-40 = Ce. 2 = 2Ce02, I€ we replace all by numbers and take for 0 = 16 and S = 32.06, 2 1.2005 7-- we have (3S03 - 0) = 22418; then we have- 32.57367 x 224.18 = 344.4$20 2Ce = 280.44'JO Ce = 140.2210. 64 21.2005 7 'If calcuIated with hydrogen numbers, 0 = 15.96 and S = 31.98, then- Ce = 139.8707. I have not calculated the atomic weight and its probable error by the method of least squares, because the following different numbers have been Calculated by different chemists (for 0 = 16) as the atomic weight of suIphur on which such a calculation &ould be founded.Stas ........................ 32.074 Ostwald .................... 32.0626 Sebelien" .................. 32.0608 L. Meyerand Seubert ........ 32.0592 Clarke.. .................... 32.058 The extreme difference of these numbers is = 0,016, and the calcu- lation of the probable error would become uncertain. It is, no doubt, a very good principle to determine the atomic weight of an element by several independent methods ; unfortunately i t was impossible in the above case from want of choice of suitable cerium compounds. I have tried to obviate this objection by making a greater number of determinations. On the other hand, my number is severely controlled by the numbers f Sebelien, Beitvage zur Geschichte der Atomgewichte, Braunschweig, 1884, p.155.CHEMISTRY OF THE CERITE METALS. 895 found by Robinson (Zoc. cit.) foy both numbers, determined by widely different methods, are in as good an accordance as can be expected :- For 0 = 16 Ce = 140,2593 140.2210 For 0 = 15.96 Ce = 139.9035 139.8707. Robineon. Brauner. If we compare the above series of determinations with that made for the purpose of investigating the homogeneous nature of cerium (from the filtrates I?, to F,,), both series are seen to agree well, and from this the conclusion must be drawn that the’ pure cerium pre- parations used by me were homogeneous. The simple method which I adopted for the purification of cerium preparations must be considered very good, as from 1380 grams of crude oxides, containing in all 690 grams of ceric oxide, I obtained 720 grams of pure nitrate, representing over 500 grams of ceric oxide.Discussion of Former Determkations of the Atomic Weight of Cerium. It will be almost unnecessary to discuss several of the above quoted determinations, for they were carried out with a material containing other cerite metals. This may be said of the experiments made by Hisinger in 181616, by Beringer in 1842, and by Rammelsberg in the same year. The method followed by Hermann (1843) and by Marignac (1848), vie., precipitation of the suIphuric acid in a solution of cerous sulphate with barium chloride, has been declared by Marignac himself to involve a considerable source of error, as the barium sulphate carries down some cerium with it.Besides, for the reasons given by Buhrig, and quoted above, it is very improbable that the sulphate used by these and other authors (especially by Jegel) had a definite normal composition. Jegel (in 1858) and Rammelsherg (in 1859) determined the atomic weight of cerium by the combustion of cerous oxalate. It has been shown by Nilson (Ber., 15, 2519) in his paper on the atomic weight of thorium, that the method of elementary analysis of an oxalate involves several errors which make it unsuitable for the exact determination of the atomic weight of an element. I shall therefore not enter more fully into an analysis of the papers published up to 1860, but proceed to discuss the work done by Wolf.As regards, first of all, his observation that the atomic weight of cerium diminishes on further purification of the preparations, I think it may be regarded as conJirmed by my own experiments. But this decrease ceases as soon as we get to the precipitate N, (it would cor- respond to Wolf’s NE which, however, he never obtained). Wolf’s ceric oxide was white. The same was the case with my purest ceric oxide, though I should prefer to call it. the palest ‘‘ chamois.”$96 BRAUNER : CONTRIBTJTIOSS TO THE On the other hand, I cannot confirm the very low atomic weight found by Wolf. Firstly, Wolf did not prove that his sulphate bad a definite normal composition. His hydrated salt, as can be concluded from my own experiments on this subject, most probably did not con- tain the theoretical amount of water of crystallisation, and according to Buhrig it must have contained some free sulphuric acid, which can be got rid off only by the process quoted above.I f these two sources of error are not taken into consideration, the atomic weight of cerium will be found lower than it really is. As regards the numbers calculated from the “ anhydrous ” sulphate, they cannot be regarded as exact, for cerous sulphate, heated high over a small flame in a doable plabinum crucible, retains a trace of water, as has been already pointed out by Buhrig, and as may be seen from the following experiment. 3.0283 grams of crystallised cerous sulphate gave, on heating for two hours in a double platinum crucible at a temperature at which the bottom of the outer crucible was red hot for a short time (the tem- perature applied by Wolf was never so high), 2.3343 grams of “an- hydrous ” salt. On heating it in the sulphur-bath at 440°, it lost the last trace of its water, and its weight diminished by 0.0046 gram, namely to 2.3797 grams.As at a high temperature the sulphate yielded 1*4424 gram of Ce02, the atomic weight, calculated from the first number, would be Ce = 139.65 instead of Ce = 140.22. Secondly, I found that, on precipitating a solution of cerous sulphate with boiling oxalic acid solution, a trace of cerium, the oxalate of which is partly soluble in the free sulphuric and oxalic acids, remains in solution even after long standing. 1.3506 gram of anhydrous sulphate, precipitated and ignited by Wolf’s method, gave 0.8173 gram CeOa = 60.314 per cent.Directly analysed, tlhe same sulphate gave 60.60 per cent. of CeO,. This error, although slight, gives a smaller atomic weight. Thirdly, on precipitating the filtrate from the cerous oxalate with barium chloride, a, little more barium sulphate is always obtained than corresponds with the sulphuric acid contained in it, for not only is the trace of cerium which remains in solution carried down with barium sulphate, but also some barium oxalate. This causes the atomic weight of cerium found to be lower. From the above filtrate, after separation of cerous oxalate, 1.6831 gram of BaS04, corresponding with 0.5781 gram, or 42.803 per cent. of SO3, was obtained (theory requires only 42.24 percent). From the relation of 0.8173 CeOz : 0.5781 SO3, the atomic weight calculated is Ce = 137.78 instead of Ce = 140.22. If only one or all of these sources of error are left out of considera- tion, t 7 ~ e atomic weight of cerium found will always be lower than theCHEMISTRY OF THE CERITE METALS. 897 true one. I think, therefore, that I have sufficiently explained +he low numbers found by Wolf. In striking contrast with Wolf’s work is that done by Biihrig, who found the high number Ce = 141.5. Biihrig analysed large quanti- ties of cerous oxalate by combustion (the details,will be found in Clarke’s recalcnlafion), but he made the mistJake of using a material resulting from one preparation only. I do not believe, however, that Buhrig’s oxalate contained an admixture of basic salt: as is sometimes stated. Another cause of error in Biihrig’s work must be looked for in the fact, as Nilson has shown (Zoc. cit.), that the analysis of oxalates by combustion is subject to constant errors which make the atomic weight higher. For example, Cleve, by this method, found the atomic weight of thorium to be Th = 23390 to 233.97, whereas Nilson’s analysis of the sulphate gave Th = 232.40. But, as Buhrig used for combustion large quantities of oxalate- about 10 grams at once-the loss of carbonic anhydride and theplus of water cannot have been the only source of the difference between t’he number obtained and the true one. I am inclined to believe that the reason of Buhrig’s higher number was partly the same which caused me to find the ‘‘ atomic weight” of the impure cerium to be R = 142.65, and even R = 145.72, for Biihrig points out distinctly that his oxide was yellow. Further, as this oxide when strongly heated is converted with loss of weight into one which is of a pale salmon colour, the former may have contained some of the unstable peroxide, t’he admixture of which would cause the atomic weight found to be a little higher. In conclusion, it may be pointed out that, in consequence of the present new determinations, the differences between the atomic weights of lanthanum, cerium, and didymium, elements following each other in the periodic system, harmonise far more than was pre- viously the case. Thus we have, La. Ce. Di. L---L~-~-J 138.2 140.2 142.3 Difference.. . . 2.0 2.1 VOL. XLVI1.