Prof. R. Bunsen's Spectral- Analytical Researches. 537 



stance, which has been already purified by several precipitations 

 with sodium thiosulphate, pure thorium oxide may be obtained. 

 When tested, as chloride, in the flame- and spark- spectrum, 

 thorium affords no characteristic lines. 



The discrepancies in the estimations of the atomic weight of 

 thorium by different observers are probably to be accounted for 

 by the presence of considerable impurities in the compound 

 supposed to be thorium oxide. 



For the preparation of pure compounds of yttrium and erbium 

 I may refer to the method published by Bahr and myself*. 

 The chlorides employed in the production of spectra were pre- 

 pared from the material with which the atomic weight of yttrium 

 was determined to be 3085, and that of erbium to be 56*3. As 

 proof of the complete separation of the two earths, it may be 

 mentioned that the w T hole of the chlorides obtained from the 

 fractional precipitates by oxalic acid yielded spectra in which 

 the number and relative intensity of the individual lines were 

 identical. The same thing was true with regard to the frac- 

 tionally precipitated sulphates, and the last mother liquor 

 remaining from these precipitations. In these observations I 

 have looked in vain for traces of a third earth (terbia), which 

 has been supposed by many chemists to exist in Gadolinite and 

 to yield absorption-bands. 



Yttrium and erbium chlorides give no flame-spectra; while 

 the spark- spectrum of the former chloride is very rich in 

 lines, Plate V. no. 1 2 e ; that of the latter is shown in Plate 

 V. no. 11 e. Erbium chloride, however, is much more easily 

 recognized by the absorption-spectrum of its solution, Plate 

 V. no. 11 a. Erbium oxide belongs to the small group of 

 substances which yield a discontinuous spectrum when heated 

 in the solid state; the clear lines no. 11/ correspond to the 

 dark bands of the absorption-spectrum. This oxide may be 

 thus easily and clearly distinguished from all other oxides. 



Spectral analysis affords the easiest and most reliable means 

 for recognizing the minerals which contain cerium and yttrium. 

 The method to be adopted will be seen from the following 

 examples. 



1. Ceritefrom the Bastn'ds Quarry. — A few centigrammes, eva- 

 porated with hydrochloric acid and taken up with the same 

 acid and water, gave a concentrated solution which behaved as 

 follows : — The light passing through it showed the characteristic 

 absorption-spectrum of didymium, especially the bands 55 and 

 75 (no. 15 a, Plate V.). As no trace appeared of the charac- 

 teristic absorption-bands of erbium, 35 (11«), the absence of 

 erbium in this cerite was proved. The carbon points saturated 

 * Liebig's Ann, vol. cxxxvii. p. 1. 



