OF ARTS AND SCIENCES. 263 



(12) 0.3357 gr. gave 0.1213 gr. C 2 3 = 36.16 per cent. 



(13) 0.3856 gr. " 0.1393 gr. " =36.13 



(14) 0.G538 gr. " 0.2906 gr. R,s0 8 = 44.45 " 



(15) 0.7074 gr. " 0.3145 gr. " = 44.46 " 



The analyses leave no doubt whatever as to the necessity of carefully 

 mixing the precipitated oxalates. They also show that in determina- 

 tions of atomic mass by the sulphate method both oxalates and oxides 

 should be well ground up to secure homogeneity. It has been shown 

 that the oxalate and sulphate methods executed with proper care give 

 equally accurate results. Each has its advantages, and in this respi Cl 

 there is but little choice. As the study of the rarer earths as now 

 conducted usually depends more or less upon the properties of the 

 double salts which they form with potassic and sodic sulphates, it may 

 be well to call attention to facts not, I believe, noticed in printed 

 papers, though doubtless recognized. The first is that the earths in the 

 double sulphates may be converted directly into oxalates by boiling the 

 sulphates with chlorhydric and oxalic acids, and then diluting with 

 much water. The second fact is that oxalates obtained in this way. 

 as from any alkaline solutions, should always be converted into oxides 

 by ignition. These should then be dissolved in chlorhydric acid, again 

 precipitated by oxalic acid, and the oxalates thoroughly washed. 



In determining atomic masses by either the sulphate or oxalate 

 method, the assumption is tacitly made that all the oxides taken for 

 analysis are of the type R 2 3 . This is not true when cerium is pres- 

 ent, as in that case a portion at least of this metal is present as Ce0 2 

 after ignition. Higher oxides than R 2 3 are also present to some 

 extent, at least when praseodymia and neodymia are mixed with 

 ceria, or even when this last oxide is not present. Ceric oxide is 

 not reduced by a full red heat to cerous oxide, or even by a cur- 

 rent of hydrogen, at least in a crucible. The error committed is not. 

 however, large when we consider the cerium as CeJ). instead of 

 Ce 2 4 . and is still less in the cases of the other oxides. When great 

 accuracy is necessary, it may be well to remove the four cerite oxide, 

 by means of potassic or sodic sulphate before determining the atomic 



mass. 



In a number of analyses I determined the percentage of oxide, by 

 simply igniting the oxalate with a weighed quantity of pure sodic 

 tungstate. This method gives very accurate results, but, as pure 

 sodic tungstate must always be specially prepared in the labora- 

 tory, is not to be greatly recommended. Ceric oxide is not reduced 



