444 Gooch and Eddy — Separation of Magnesium. 



Art. XLYII. — The Separation of 3Iag?iesium from the Alka- 

 lies by Alcoholic Ammonium Carbonate : by F. A. Goooh 

 and Ernest A. Eddy. 



[Contributions from the Kent Ckemioal Laboratory of Yale Univ. — elxxiv.] 



Nearly fifty years ago Schaffgotsch* brought out the 

 ammonium carbonate method for the separation of magnesium 

 from the alkalies. According to this process, the very concen- 

 trated solution of the sulphates, nitrates or chlorides of 

 magnesium, sodium, and potassium, is treated with a concen- 

 trated solution of ammonium carbonate. The voluminous 

 precipitate which first falls is acted upon by an excess of the pre- 

 cipitant, sometimes dissolving completely, and crystalline am- 

 monium magnesium, MgC0 3 .(NH 4 ) 2 C0 3 .4H„0, is soon formed. 

 After standing twent} T -four hours the precipitate is filtered off, 

 washed with the concentrated ammoniacal solution of ammo- 

 nium carbonate, dried and strongly ignited. In the absence 

 of salts of potassium, the residue is weighed at once as mag- 

 nesium oxide, and from the filtrate sodium salts are recovered 

 by evaporation. When a salt of potassium is originally present, 

 with or without a salt of sodium, the ignited magnesium oxide 

 is to be washed out and again ignited before weighing, and the 

 washings are to be added to the filtrate containing the greater 

 part of the alkalies. 



When Schaffgotsch published this methodf it was customary 

 to deal in analysis with larger amounts of material than at 

 present and the requirements as to results were not so exacting. 

 In eight complete determinations, upon which Schaffgotsch 

 rested, weights of salts taken ranged between 1*75 grm. and 

 3*69 grm., and the magnesium oxide taken approximated 06 

 grm. ; while the weight of magnesium oxide found varied from 

 0*0007 grm. to - 0082 grm., with an average error amounting 

 to 0*6 per cent. 



Though these results do not meet the modern requirements 

 of a good analytical method, the process has been recently 

 recommended and applied in rock analysis by Wuelfing,'j: and 

 mentioned by Hillebrand.§ It has seemed to us desirable, 

 therefore, that the method should be again submitted to care- 

 ful testing, and to that end the following work was undertaken. 



*Ann. Phys., civ, .482, 1858. 



f A similar method was published by H. Weber, Jahresb. Client., 1858, 

 p. 606. 

 JBer. Dtsch. Gesellsch., xxxii, 2214. 1899. 

 §XJ. S. Geol. Sur. Bulletin, 305, 149. 



