of aurin, is not a necessary assumption, but its presence as a sublimed bye-product simply bears witness to the tendency of the first phase of the reaction, which is to the formation of this substance, some of which has escaped the second phase. The first phase can then only be considered as a passing-reaction. The two phases thus shade one into the other, so as to form the full complex reaction.

It is a circumstance worthy of note that the temperature at which the phenyl-oxalic ether melts, and above which it steadily decomposes, is also the maximum limit of temperature to be employed in the method of aurin manufacture, found by experience to yield the best results. Hence this connection of circumstances would seem to bear reasonable witness to the truth of the theory we advance. In any case, pure aurin never having been directly prepared, it is certain that the formation of corallin is not limited to the two phases of the above reaction.

We hope that this research may not be without interest as a contribution to the study of the aurin question.

LII.-On Samarium and its Compounds.

By Prof. P. T. CLEVE, Hon. Member of the Chemical Society.

INTRODUCTION.

IN the year 1878 Delafontaine found, on examining impure didymia, extracted from samarskite, some new absorption-bands, which did not belong to didymium but to an unknown element of higher atomic weight. He named this metal Decipium. Some time later Lecoq de Boisbaudran found also in the fractions, rich in didymia, which were extracted from samarskite, a new oxide, distinguished by its absorption-spectrum and by peculiar spark-lines. The oxide was thrown down with ammonia before that of didymium. He called the metallic radical Samarium, and described more exactly than Delafontaine its absorption-spectrum. In March, 1880, Delafontaine published in the Bibliothèque universelle de Genève a paper "On the Atomic Weight and the Compounds of Decipium." The atomic weight was found to be 114, if the oxide be RO (= 171 if R203). The salts were colourless and the sulphate less soluble than the sulphate of didymium. Shortly afterwards Marignac published his researches on the earths which in the samarskite accompany terbia, and found an oxide, named Yẞ, distinguished by its yellow salts, its spectrum of absorption agree

The

ing with that of samarium. The atomic weight of the metal was found, as a maximum 996, if the oxide be RO (= 1794 if R2Oз), thus differing widely from the figures given by Delafontaine. latter has more recently found that the decipia of 1878 could be split up into an oxide without absorption-spectrum, for whose radical, of the atomic weight 117 (or 171 if R2O) he reserved the name of decipium, and another oxide with the spectrum described by Lecoq de Boisbaudran. He adopted the name of samarium for the metal which has as a maximum the atomic weight 101 (or 151·5 if R2O2).

Method of Extraction.

The raw material, which I used for extracting the oxide of samarium, consisted principally of a mixture of almost all the rare oxides of the earth-metals, which were obtained from Dr. Paijkull, in Stockholm, and said to be derived from orthite found at Arendal in Norway. As the matter, about 10 kilos., contained almost 10-12 per cent. thorina, it may be assumed that a considerable quantity of thorite may have been mistaken for orthite. The mixture of earths was mixed with nitric acid and heated in china basins till it began to give off red vapours. After cooling, the mass was treated with water, which left undissolved a large quantity of basic nitrates, principally of cerium and thorium. These basic nitrates were insoluble in saline liquids, but dissolved in pure water, forming opalescent solutions. They were separated by decantation and filtering, and washed as long as the liquid went clear through the filter. The basic salts were mixed with strong sulphuric acid, and the solution of the sulphates in cold water was precipitated with large quantities of hot water, by which operation most of the cerium was separated as basic sulphate.

The solution was then precipitated with caustic soda and the hydrate, principally of thorium, dissolved in diluted sulphuric acid. On evaporating off the solution, large and bulky masses of thorium sulphate were obtained. The oxides, which remained in the motherliquors, were transformed into nitrates and added to the solution of nitrates obtained by the first operation. These nitrates were eva

porated, and the residuum was heated in platinum basins till it began to give off red vapours. On treating the heated mass with water, a

solution free from cerium and thorium was obtained.

By this treatment also considerable quantities of the nitrates of the more positive metals, yttrium, didymium, &c., were decomposed, and separated together with the thorium and cerium, and to complete the separation, the basic nitrates were once again and carefully treated in the same manner. The solutions free from the oxides of cerium and thorium, or containing only small traces of these oxides, were

precipitated with potassium sulphate with the view of separating the cerium- and yttrium-oxides, of which the former were (incompletely) precipitated, and the latter (incompletely) retained in the solution.

The yttria earths (about 700 grams) were transformed into nitrates and subjected to the usual melting-process, for separation of erbia from yttria, terbia, and didymia, which offer greater resistance to the action of heat. From the basic nitrates first obtained a small quantity (about 1.5 gram) of scandium oxide was separated.)* They contained also a very small quantity of thulium, some erbia and ytterbia, and a considerable portion of holmia. The fractions containing yttria and didymia were saturated with potassium sulphate, and this precipitate was added to the similar precipitate formerly obtained. These difficultly soluble double sulphates were decomposed with boiling potash-ley, and the hydrates dissolved in diluted sulphuric acid were again treated with potassium sulphate. In the solution there then remained considerable quantities of terbia (about 200 grams impure earth). The precipitate was converted into nitrate and its solution evaporated. The residue was then heated to partial decomposition in order to separate the last remaining traces of thorium and cerium.

The soluble nitrates thus obtained, which contained principally didymium and lanthanum, but also samarium, terbium, yttrium, &c., were mixed with impure didymium, extracted from various minerals as cerite, gadolinite, keilhauite, &c., and then subjected to partial precipitation with cold dilute ammonia. The last fractions contained much lanthanum, which was separated partly by the Mosander method, and partly by precipitation with ammonia. No oxide intermediate between that of lanthanum and that of didymium could be obtained, although I sacrificed much time in searching for such an oxide. The first precipitations were repeatedly treated in the same manner with dilute ammonia. In this way I obtained several fractions containing both terbia and didymia (samarium, &c.). They were again treated several times with sulphate of potash.

The slightly soluble double sulphates were now decomposed, and the solution of the hydrates in nitric acid was repeatedly precipitated with ammonia. I obtained finally the following fractions :

1

about 67 grams light-coloured oxides, whose solution was very little coloured, but showed, besides the spectrum of samarium, also the spectrum of didymium. Atomic weight R"" 149-150.

sium.

No scandium could be obtained from the precipitate with sulphate of pɔtas

The sixth fraction was subjected to partial precipitation with ammonia, but the atomic weight of the four fractions thus obtained was still found to be 142.3.

This is a proof of the correctness of my latest determinations of the atomic weight of didymium, which gave the same numbers. No lanthanum was present in these fractions. As the fractions intermediate between the pure or almost pure didymium and the samarium are only trifling, there is very little probability of finding any unknown element intermediate between samarium and didymium.

The 67 grams of earths containing the principal quantity of samarium were now subjected to the same treatment with dilute ammonia, until the absorption-spectrum of didymium was completely eliminated. The oxides freed from didymium were far from pure samarium oxide, and I was therefore obliged to precipitate their solutions with potassium sulphate.

At first I several times used solutions with 2 per cent. oxide, and later with only per cent. Terbia and other oxides (Ya, decipium) were thus eliminated together with much samarium oxide. The treatment with potassium sulphate was continued until the oxide, which remained in the solution, had the same molecular weight as the oxide thrown down, or R′′ = 150.

A considerable quantity of samaria was separated at the same time as terbia and Ya, for which reason I was obliged to subject these oxides to the same treatment, whereby I finally obtained the fractions used for the atomic weight determinations A and B, A being oxide left in the solution, and B precipitated as double sulphate.

I have described this method, as long experience has convinced me that the coarse separation of all the rare earths may be most conveniently effected thereby.

Atomic Weight of Samarium.

The samarium oxide, obtained as above described, was purified by treatment with sulphuretted hydrogen, repeated precipitation of the solution with ammonia, and finally in acid solution with pure oxalic acid. The oxide obtained by calcination of the oxalate was dissolved in pure nitric acid, and by precipitation with freshly distilled ammonia, divided into four fractions, I, II, III, IV. These were transformed into oxalate and oxide. The four fractions thus obtained were subjected to atomic weight determinations in the usual way, by solution in nitric acid, addition of sulphuric acid, and estimation of the sulphates. The fractions A and B were obtained, as stated in the fore

going section, from the samarium oxide which remained in the solution of potassium sulphate.

The atomic weight determinations gave the following results :

As a mean of these six closely agreeing experiments, the atomic weight is 150-021, or in round numbers 150. In the calculation I have supposed the oxide to be Sm2O3, which follows beyond doubt from its chemical properties, that is from the composition of the chloroplatinate, of the samarium and ammonium sulphate, and of the selenite. Whether samarium gives other oxides besides Sm2O, is a point which I have not yet determined. As to its position in the system of Mendelejeff, I think it may most probably be placed in the 8th group, on the 8th line, where there is a gap for a group of unknown elements.

Spectrum of Samarium.

The salts of samarium are distinguished by a peculiar spectrum, composed of several bands, among which four in the blue part of the spectrum are most characteristic. The absorption-spectrum has been examined by Lecoq de Boisbaudran and by Thalén. The latter used a solution of the nitrate prepared by myself.

Both found about the same bands, viz. :

The intensity of the absorption is not very great, so that the absence of these bands in didymium salts cannot be regarded as a sure test for the absence of samarium.