246 PROCEEDINGS OF THE AMERICAN ACADEMY 



drate, nor, after neutralization, with potassic sulphocyanide.* Sodic 

 sulphate and chloride are even more readily separated than traces of 

 iron. 



Three separate preparations of sodic carbonate were used in the 

 course of the work. For the first series of comparatively crude ex- 

 periments the solution of ordinary " chemically pure" material, which 

 almost invariably contains a perceptible amount of finely divided solid 

 matter, was after filtration twice recrystallized in a platinum dish. It 

 was subsequently found advantageous to prevent the clogging of the 

 filter by two or three preliminary recrystallizations, in the course 

 of which most of the solid matter finds its way into the mother 

 liquor. 



The second sample of sodic carbonate was recrystallized six times 

 after filtration. The vessels, rod, and reverse filter were all of plat- 

 inum, and water was used which had been distilled in a platinum 

 retort. 



Besides all these precautions, several additional ones were taken 

 in the preparation of the purest sample of sodic carbonate designed 

 for the third series. The water used had been distilled four times : 

 first alone, next over potassic permanganate, then over acid potassic 

 sulphate, and finally in a platinum still without the addition of foreign 

 matter. Every precaution was taken to exclude dust and acid fumes, 

 and the sodic carbonate was recrystallized ten times after filtration. 

 The salt prepared in this way gives an absolutely odorless warm con- 

 centrated solution. The second and third samples gave essentially 

 identical results. Each of the three [)reparations was pure snow- 

 white, whether fused or unfused ; each gave a perfectly clear and 

 colorless solution with water ; and not even the first save the least 

 test for hydrochloric or sulphuric acid. 



On the other hand, as Stas has already indicated, the most elaborate 

 precautions are unable to free sodic carbonate wholly from traces of 

 silica and a basic oxide, which is probably alumina. The former 

 impurity was determined in the usual manner. The alumina and 

 silica were determined together in new portions of the salt by exact 

 neutralization witii hydrochloric or sulphuric acids, ignition of the 

 evaporated product, and weighing of the insoluble residue. Phenol 

 phthalein was used to determine the neutral point, and hence the 



* Stas appears once or twice to liave had more ilitticulty in accomplishing 

 this result. At other times his experience seems to have coincided with tliat 

 described above. Compare Aronstein's translation (1867), pp. 112, 270, with 

 p. 275. 



