328 Carbonates of alumina, glucina, iron, chromium, etc. 



reduced to protoxyd, which does not drive out carbonic acid 

 from carbonate of' soda. Tiie residue in the bulb was washed 

 somewhat, ignited, evaporated to dryness with hydrochloric acid 

 and clilorid of arnrnoniurn, and ignited in a stream of hydrogen. 

 After thorough washing, it was again ignited in hydrogen and 

 weighed as UO. This process is, according to Rose,* the best 

 method for separating uranium from the alkalies. 



The three following analyses were of precipitates from the 

 nitrate. The first two were dried : the last was not. 



3U2( 



The five next analyses were of precipitates from the sulphate. 

 The last was precipitated by pouring the sulphate of uranium 

 into the carbonate of soda: the others, as usual, by pouring the 

 carbonate of soda into the uranium solution. 



Notwithstanding the tolerably close agreement of the above 

 results with the formulas just given, I am strongly inclined to 

 think that these formulas do not express the true composition ot 

 the precipitates. A portion of one of the precipitates, after very 

 protracted washing with cold water, still contained alkah, show- 

 ing the presence of uranate of soda.f This result might indeed 

 be inferred, almost with certainty, from the very strong affinity 

 of sesquioxyd of uranium for the alkalies4 On account of this 

 impurity, as well as the ease with which all these carbonates ap- 

 pear to lose carbonic acid, the observed percentage of carbom|. 

 acid should be less than the calculated. Instead of this, it is i» 

 nearly every case larger. I suspect, therefore, that the precipi- 

 tates in question are mixtures of a less basic carbonate ^Mt 

 uranate of soda. More of the latter appears to be formed m tti _ 

 precipitate from the sulphate of uranium than in that from tne 

 nitrate. Why this should be I cannot explain. What the true 

 composition of the carbonate of uranium would be, if it cou 



