274 ANALYTICAL CHEMISTRY. 



can be precipitated from a solution of its salts by a number of re- 

 agents, which, however, differ widely in sensitiveness. Thus, tartaric 

 acid will cause the formation of a precipitate of potassium bitartrate 

 in a solution containing at least 0.1 per cent, of potassium ; in solu- 

 tions containing a less amount no precipitate is formed. Platinic 

 chloride is somewhat more sensitive than tartaric acid, and sodium 

 cobaltic nitrite, which is still more delicate, causes a precipitate in 

 solutions containing even as little as 0.04 per cent, of potassium. It 

 is evident that by using either one or the other of the three reagents 

 mentioned for the detection of potassium, this metal may or may not 

 be found, according to the quantity present in a solution. The 

 Pharmacopoeia, in directing the use of one of these reagents, limits 

 the amount of a permissible quantity of potassium according to the 

 sensitiveness of the reagent. 



Again, in testing for arsenic, the chemist has his choice between a 

 number of more or less delicate tests. Gutzeit's test is so sensitive 

 that by means of it arsenic can be detected in a solution containing 

 only 0.000001 gramme of arsenous oxide in a cubic centimeter. This 

 test would be, therefore, by far too severe when applied to a number 

 of pharmaceutical preparations, for which reason the Pharmacopoeia 

 directs in many cases the less sensitive tests of Bettendorff or Fleit- 

 mann. 



Quantitative tests for the limit of impurities. While, as above 

 stated, even the qualitative tests are often so made as to be to some 

 extent of a quantitative character, the U. S. P. recommends in many 

 cases methods by which a stated limit of an impurity can be detected 

 without the necessity of determining by quantitative analysis the 

 actual amount of the impurity present. 



Formerly it was, and to some extent it is now, customary to limit 

 the amount of a permissible quantity of an impurity by referring to 

 the intensity of the reaction. In case the impurity was to be detected 

 by precipitation (as, for instance, sulphates or chlorides in potassium 

 nitrate) it was stated that the respective reagents used for the detec- 

 tion (in the case named, barium chloride or silver nitrate) should not 

 produce more than a very slight precipitate, or turbidity, or cloudi- 

 ness, etc. These descriptions are, of course, very indefinite, and the 

 conclusion arrived at depends largely upon the individuality of the 

 observer. 



In order to obviate this uncertainty the U. S. P. has introduced a 

 number of more exact methods. These depend upon the addition of 



