May 15, 1903.J 



SCIENCE. 



salts of neodidymium (143). The absorp- 

 tion spectra of these two are complement- 

 ary. The next year Crookes eliminated 

 band after band of the didymium until 

 only ^ 443 remained. Kriiss and Nilson 

 and Kieserwetter and Kriiss prepared didy- 

 mium from several sources, fractioned the 

 preparations and arrived at a similar con- 

 clusion. It is now known, as shown above, 

 that erbium has been resolved into seven 

 other well-characterized elements, viz., be- 

 sides erbium (166.3), scandium (44.1), 

 yttrium (89), terbium (160), ytterbium 

 (173), thulium (170.7), holmium (162) 

 and dysprosium (?). After the elimina- 

 tion of samarium, didymium shows at least 

 nine distinct absorption bands: ^ 728.3, 

 679.4, 579.2 to 575.4 (which is easily re- 

 solved into two), 521.5, 512.2, 482, 469, 

 445.1 and 444.7 (443) (ultra-violet and 

 infra-red not considered). In short, these 

 tAvo elements neo- and praseodidymium 

 consist of at least nine elements. The full 

 conclusion of Kriiss and Nilson may be 

 stated in their own words: ' Nach obigen 

 Auseinandersetzungen batten wir an Stelle 

 des Erbiums, Holmiums, Thuliums, Di- 

 dyms, und Samariums die Existenz von 

 mehr als zwanzig Elementen anzunehman. ' 

 "While the acceptance of such conclusions 

 without (luestion would be wholly unsci- 

 entific, we must carefully consider the gen- 

 eral idea involved, and the investigations 

 upon which the conclusions were founded 

 and the investigations carried out subse- 

 (luenlly to test them. Absorption bands 

 deteimined under variable conditions are 

 not to be accepted as essential character- 

 istics of new elementary substances, as they 

 have been shown to vary with the salts 

 used. Sorby and Liveing have shown that 

 the character of the solvents and traces of 

 impuj'ities bear importantly upon the in- 

 tensity of the absorption bands. Lawrence 

 Smith a2id de Boisbaudran and latterly 



Dennis and Chamot have called attention 

 to variations in the absorption spectrum 

 of didymium, when nitric acid is present. 

 Becquerel showed there were variations in 

 the spectra of crystalline compounds of 

 the same element. Very recently Muth- 

 inann and Stiitzel have shown that, if a sub- 

 stance be regarded undecomposable, its ab- 

 sorption spectrum varies considerably with 

 dilution and amount of free acid present. 

 Demarcay has urged the necessity of giving 

 the thickness of the medium used, with 

 a statement of its strength. 



C. M. Thompson reported that didymium 

 salts from various sources showed no ma- 

 terial differences in absoi^ption spectra. 

 Schottlander remarks, however, that the 

 material used contained several oxides, giv- 

 ing absorption bands, so the intensity of 

 certain bands of a particular element may 

 have been increased by the superposition of 

 bands of other elements. 



Crookes and Dennis independently made 

 the extremely interesting observation that 

 the heavy orange bands (575-579), which 

 were resolved by Auer, were not altered in 

 their fractions when the remaining lines 

 had undegone some changes, hence the 

 former stated that probably 'didymium 

 will be found to split up in more than one 

 direction, according to the method adopted. ' 

 The work of Dennis on the relative inten- 

 sities of the bands observed, by varying 

 the procedure of fractionation, is in direct 

 accord with the observations of previous 

 investigators as to the compound nature 

 of neo- and praseodidymium. 



Von Scheele (1898) carried out a series 

 of investigations looking toward the proof 

 of the elementary character of praseo- 

 didymium. Bettendorff, by a spectro- 

 scopic examination of the mother liquors 

 obtained by the Welsbach method, con- 

 firmed the observations of Kriiss and Nil- 

 son, especially with regard to the absorp- 



