OF ARTS AND SCIENCES. 247 



insoluble bases present were necessarily precipitated. It is needless 

 to state that the operations were conducted in platinum vessels. Ne- 

 glecting two determinations which were vitiated by known impurity, 

 the weight of total residue obtained varied from 0.15 to 0.41 milli- 

 gram, while the weight of sodic carbonate taken varied from 1.02 to 

 3.06 grams. The average of twenty determinations showed about 

 thirteen parts of impurity to exist in one hundred thousand parts 

 of the salt. Silica constituted about forty per cent of this impurity. 

 The third specimen did not ditit^r essentially from the second on these 

 averages. Owing to the slight solubility of silica in sodic chloride 

 solutions,* the observed amount of this impurity may be slightly too 

 low, but for the present purpose the correction is unimportant. The 

 filter papers used in this part of the investigation left upon ignition an 

 ash of 0.00004 gram. 



Since the silica and alumina are probably present as sodic silicate 

 and aluminate in the original carbonate, and may be considered as 

 replacing carbon dioxide molecule for molecule, it is evident that the 

 correction to be applied will not equal the whole weight of the residue. 

 Upou this assumption the correction when phenol phthalein is used as 

 an indicator amounts to about half the weight of the impurity, or 

 about 0.007% of the weight of the sodic carbonate. Toward methyl 

 orange aluminic hydroxide is alkaline ; hence in this case the errors 

 due to the two impurities tend to counterbalance each other, and the 

 correction may be omitted. The presence of a small amount of 

 alumina thus explains the slight difference observed later between the 

 results obtained with these two indicators. 



The crystallized sodic carbonate was dried over pure boiled sulphuric 

 acid in a vacuum, and subsequently ignited to a dull red heat in a 

 double crucible over a Berzelius spirit lamp. The latter apparatus 

 was employed in order to avoid the possible introduction of impurity 

 from illuminating gas. It was found that the salt could be ignited to 

 perfectly constant weight at any temperature between dull redness 

 and its fusing point. The light powder contracts and " sinters to- 

 gether " at a low red heat, and then remains essentially unchanged 

 in weight until it melts. The unfused salt was not observed to gain 

 perceptibly in weight upon an hour's exposure to the air of the bal- 

 ance case ; upon the other hand, the fused salt, which of course was 

 never used as a basis of determination, was markedly hygroscopic. 



* Stas, Aronste'm's translation, p. 279. In tliis place the correction is 

 applied to sodic chloride, but not to the nitrate formed from it. 



