540 



NATURE 



{April ^, 1889 



of intensity." Before deciding the question whether 

 didymium is a homogeneous whole, or whether an argu- 

 ment in favour of its heterogeneity can be based on the 

 fact that the absorption spectra of didymium from different 

 minerals differ inter se, it was necessary to ascertain if 

 the absorption bands seen in its solutions, whatever the 

 thickness of the layer, whether dilute or concentrated, 

 followed the same variations, and also to ascertain the 

 nature of these variations. To contribute to this inquiry 

 I examined the absorption spectrum of a solution of 

 neutral didymium nitrate containing i part by weight 

 of metal in 10 of water, as seen through a series of cells 

 from I mm. to 25 mm. in thickness. For this work I 

 used a new form of binocular spectroscope, fitted with 

 a mechanical tracing arrangement, so that each spectrum 

 can be automatically mapped on paper strips. At the 

 bottom, at 25 mm. thickness, all the known bands are 

 visible, and they become fainter and die out in order, 

 some of them remaining visible almost to the end. For 

 instance, almost as long as the deep line in the blue part 

 of the spectrum at i/A^ 507 can be distinguished, it is 

 possible to see the group of three very narrow ones next to it. 

 Two or three other less characteristic bands can be seen 

 only when there is a very considerable depth of liquid ; 

 thus, the group in the red at about i/V 255 cannot be 

 seen distinctly through less than 20 mm. of this strength 

 of solution. 



Having ascertained in this series how the spectra varied 

 in appearance with different thicknesses of the same 

 solution (strength i of Di in 10 of water), I repeated the 

 experiments, keeping the thickness of layer of solution 

 constant, and diluting the standard solution of didymium 

 so that the rays of light passed through the same 

 quantity of metal as in the former series. The results in 

 each case were practically identical ; the differences being 

 too slight to be detected in my apparatus. The spec- 

 trum exhibited, for instance, by i mm. of the standard 

 solution of didymium is found to be identical with the 

 spectrum shown by the same solution diluted twenty times 

 and viewed through a 20 mm. cell. 



There are at least two points in these researches that I 

 must touch, since they illustrate the necessity of great 

 caution in drawing conclusions from an examination of 

 absorption spectra. Messrs. Paul Kiesewetter and Kriiss 

 {Berichte der deutsch. chem. Gcsellschafi, \o\. xxi., 22,10 ; 

 Chem. Ne7vs, vol. Iviii. pp 75, 91) have recently pub- 

 lished a paper on this subject. They have examined 

 gadolinite, and find that some of the constituents of didy- 

 mium and samarium are absent, notably the group of 

 lines in the green to which I have already referred. In 

 my own laboratory 1 have worked for the last two years 

 alrnost exclusively upon the earths from gadolinite— of 

 which I obtained a large quantity from Fahlun — and there 

 is not the shadow of a doubt that in my gadolinite earths 

 the lines reported absent by Kiesewetter and Kriiss are 

 present in abundance. 



Some hitherto unexplained condition doubtless rendered 

 these lines invisible to Messrs. Kiesewetter and Kriiss— 

 perhaps the presence of some other earths, or some con- 

 dition of concentration or acidity. In the light of this 

 knowledge I do not see how we can take the results of 

 Messrs. Kriiss and Nilson or my own as final. 



Owing to its complicated nature, Kiesewetter and 

 Kriiss consider gadolinite an unfavourable source of didy- 

 mium for these investigations, and recommend that a 

 large quantity of earth from keilhauite should be system- 

 atically worked up, for the reason that keilhauite didymium 

 is more simple in constitution.' 



The Erbium Group. 



It is known that a certain oxide, ten years ago called 

 erbia, and regarded as belonging to a simple elementary 

 body, has been resolved by the investigations of Dela- 



fontaine, Marignac, Soret, Nilson, Clcve, Brauner, and 

 others into at least six distinct earths — three of them, 

 scandia, ytterbia, and terbia, giving no absorption spectra,, 

 whilst others, erbia (new), holmia, and thulia, give 

 absorption spectra. 



The first to announce that erbium was not a simple 

 body was Delafontaine, who in 1878 {Comptes rendus, 

 vol. Ixxxvii. p. 556; Chemical News, vol. xxxviii, p. 202), 

 published an account of philippium, a yellow oxide 

 characterized by a strong band in the violet, X 400 to 405, 

 a broad black absorption band in the indigo-blue, about 

 X 450, two rather fine bands in the green, and one in the 

 red. 



The history of philippium is curious, and I may perhaps 

 be allowed to give it in some detail. A year after Dela- 

 fontaine's discovery, Soret {Comptes rendus, vol. Ixxxix. p. 

 521 ; Chemical Neius, vol. xl. p. 147) published a paper 

 in which he declared that philippia was identical with his 

 earth X. The next month, in a note on erbia, Cleve 

 {Comptes rendus, vol. Ixxxix. p. 708 ; Chemical News, vol. 

 xl. p. 224) said he could not identify .Soret' s X with 

 Delafontaine's philippia, as the latter was characterized 

 by an absorption band in the blue which occupied the 

 same place as one of the erbia bands. In February 

 1880 {Comptes rendus, vol. xc. p. 221 ; Chemical NewSy 

 vol. xli. p. 72), Delafontaine returned to the subject, 

 enumerating ten new earths in gadolinite and samarskite, 

 viz., mosandra, philippia, ytterbia, decipia, scandia, 

 holmia, thulia, samaria, and two others unnamed. He 

 said that the properties of philippia were those of Soret's 

 X and of Cleve's holmia, and proposed that the name- 

 " holmia," being a duplicate name for an already known 

 earth, should be discarded in favour of philippia. In 

 July 1880 {Comptes rendus, vol. xci. p. 328 ; Chemicar 

 News, vol. xlii. p. 185), Cleve repeated his former state- 

 ment that philippia was not the same body as Soret's X 

 or holmia. Delafontaine next withdrew all he had said 

 about the absorption spectrum of philippium, and decided 

 that it had no absorption spectrum {Archives de Geneve, 

 [.3], 999, p. 15). Finally, Koscoe (Journ. Chem. Soc, 

 vol. xii. p. 277), in an elaborate chemical examination of 

 the earth-metals in samarskite. proved that philippia was. 

 a mixture of yttria and terbia. From a prolonged chemical 

 study of these earths I have since come to a similar con- 

 clusion ; but a spectroscopic examination of the earth left 

 on igniting some specially purified crystals of " philippiuni 

 formate " tested in the radiant-matter tube, has shown me 

 that in the separation of Delafontaine's philippium the 

 yttria undergoes a partial fractionation, and three of its 

 components or meta-elements, Gf, GS, and Gi3, are pre- 

 sent in great abundance, while others, Ga and G17, are 

 almost if not quite absent. 



Shortly after the announcement of philippium, Soret 

 {Comptes rendus, vol. Ixxxvi. p. 1062) described an earth 

 which he provisionally called X. This was soon found to- 

 be identical with an earth subsequently discovered by 

 Cleve {Comptes rendus, vol. Ixxxix. p. 479 ; Chemical 

 News, vol. xl. p. 125), and called by him holmia. Soret 

 admitted the identity, and agreed to adopt Cleve's name of 

 holmia. The absorption spectrum of X consists of a very 

 strong band in the extreme red, X 804, two characteristic 

 bands in the orange and green, X 640 and 536, besides- 

 fainter lines in the more refrangible part of the spectrum. 



Simultaneously with the discovery of holmia, Cl^ve 

 announced the existence of a second earth from erbia, 

 which he called thulia. Its absorption spectrum consists 

 of a very strong band in the red, X 684, and one in the 

 blue, X 464-5. 



In 1886 {Comptes rendus, cii. 1003, 1005), Lecoq de 

 Boisbaudran showed by fractional precipitation of Soret's 

 X, and by spectroscopic examination of the simple frac- 

 tions, that this X, or holmium, consisted of at least twO' 

 elements, one of which he named dysprosium, retaining 

 the name of holmium for the residue left after deducting 



