16 AMERICAN MINERAL WATERS. 



The manganese in about one-half of the samples was determined 

 by the bromin oxidation method, as given in Bureau of Chemistry 

 Bulletin 91, but it was found that with the small amounts of man- 

 ganese usually present in mineral water the method was not entirely 

 satisfactory, results often being discordant, thus necessitating another 

 determination. A colorimetric method described by Hillebrand 1 was 

 finally adopted, with very gratifying results, both because of the 

 delicacy of the reaction and the ease of manipulation. This method 

 depends upon the oxidation of the manganese present to perman- 

 ganic acid, the color produced being compared with a standard solu- 

 tion of potassium permanganate. The method is as follows : 



Make 100 cc of the water (or 1 liter concentrated to this volume) strongly acid with 

 nitric acid, then add a sufficient quantity of a silver nitrate solution (2 grams of silver 

 nitrate to 1 liter) to precipitate all the chlorin; add an additional 10 cc of the silver 

 nitrate solution for each milligram of the manganese expected; agitate the solution 

 to coagulate the silver chlorid, and filter. To the clear solution add approximately 1 

 gram of ammonium persulphate. Place on the steam bath and the pink color due 

 to permanganic acid will soon appear, reaching a maximum in about 20 minutes. 

 Remove the solution from the steam bath, cool, make up to a definite volume, and 

 compare the color in Nessler tubes or in a Schreiner colorimeter with a standard 

 solution of potassium permanganate. Instead of the permanganate, a standard 

 solution of manganese sulphate may be employed, a convenient solution being one 

 containing the equivalent of 2 mgs of manganese to 10 cc. For comparison oxidize 

 10 cc or other convenient aliquot of this solution with the silver salt and ammonium 

 persulphate, parallel with the water under examination, make the final solutions to 

 definite volume, and compare in the usual way. The reaction is an exceedingly 

 delicate one, 0.01 mg of manganese in 100 cc of water being easily determined. No 

 other element usually found in mineral waters seems to interfere with the delicacy 

 of the reaction. 



For the determination of lithium the method devised by Gooch, 

 in which amyl alcohol is used to separate the mixed alkali chlorids, 

 was used in all cases where a weighable amount of lithium was found. 

 This method, which is reported in detail by Haywoocl, 2 has been 

 subjected to additional investigation, and its absolute reliability as a 

 precise method is amply confirmed. A failure to obtain satisfactory 

 results by the method is usually due to magnesium which has not 

 been entirely removed from the mixed chlorids. It is therefore 

 essential that the residue obtained after removing the amyl alcohol, 

 and supposedly lithium sulphate, should be tested for magnesium 

 with a few drops of ammonium hydroxid and ammonium phosphate, 

 allowed to stand 24 hours, filtered, ignited, and weighed. If any 

 magnesium is found it should be calculated to sulphate and sub- 

 tracted from the sulphate residue from the amyl alcohol. A failure 

 to observe this precaution will frequently lead to grave error. 



Because of the very small amount of lithium found in most min- 

 eral waters and the necessarily limited amount of water available 

 for analysis in an investigation of this character, a method which 



i U. S. Geological Survey, Bui. 305. 2 U. S. Dept. Agr., Bureau of Chemistry, Bui. 91. 



