BAXTER AND CHAPIN. — ATOMIC WEIGHT OF NEODYMIUM. 227 



unit, the purer fractions were obviously as pure as it is worth while to 

 make them. None of them in all probability were entirely free from 

 praseodymium. In fact, visual spectroscopic examination of both frac- 

 tion 2929 and 2931 with very strong absorption showed them to con- 

 tain perceptible amounts of this impurity. 



It is reasonable to suppose that cerium and lanthanum nitrates, 

 which are more soluble in concentrated nitric acid than praseodymium 

 nitrate, were removed from the neodymium even more completely than 

 praseodymium. Evidence upon this point is more difficult to secure 

 than in the case of praseodymium, on account of the lack of absorption 

 bands of cerium and lanthanum in the visible spectrum. The follow- 

 ing test for cerium was carried out with fraction 2944, which contained 

 0.5 per cent of praseodymium. The hydroxide was precipitated by the 

 addition of an excess of sodium hydroxide, and, after washing by de- 

 cantation, it was treated with saturated chlorine water for some time. 

 A small quantity of undissolved hydroxide was filtered off and dis- 

 solved in nitric acid. To the nearly neutral solution a sufficient amount 

 of a solution of calcium hypochlorite was added to produce a slight per- 

 manent precipitate. This precipitate was pink with no tinge of yellow. 

 The same solution after the addition of a trace of cerium and testing 

 with hypochlorite as before yielded a precipitate distinctly colored with 

 the characteristic yellow of eerie hydroxide. Fraction 2944 was evi- 

 dently nearly if not entirely free from cerium and the less soluble frac- 

 tions of the series must have been purer still. 



The Absorption Spectrum of Aqueous Solutions of 

 Neodymium Salts. 



The absorption spectra of two of the purest fractions, 2929 and 2931, 

 were examined visually with a Hilger wave-length spectroscope. While 

 no very great accuracy is claimed for the measurements, it is probable 

 that the values are correct within one-millionth of a millimeter, for the 

 spectroscope is capable of nearly ten times this accuracy for wave- 

 lengths in the neighborhood of A 500. 



In the first two columns of the following table are given the values for 

 the maxima of absorption observed during the gradual dilution of very 

 concentrated, slightly acid solutions of the nitrates, each value having 

 been obtained in the moat dilute solutions in which the band could be 

 observed. The spectra of the two fractions were identical. 



The band X 580 can be seen only at certain medium concentrations 

 while the band A 547 is visible only with very concentrated solutions. 

 The praseodymium band A 444 also was faintly visible only in the most 



