SANGER AND BLACK. — QUANTITATIVE DETERMINATION OF ARSENIC. 317 



chloride acted more slowly than with arsenic alone, giving a slight 

 brownish red at first and finally a somewhat lighter purple than the 

 pure arsenic color. There was apparently little increase in evolution 

 of phosphine in the presence of the arsenic, and the arsenic compound 

 in the mixed band was not appreciably obscured. 



Stibine. Solutions containing 30 mmg. arsenious oxide and 70 mmg. 

 antimonious oxide were reduced together for thirty minutes. The re- 

 sulting band was pale red in color and over twice as long as the band 

 from 30 mmg. of arsenious oxide. Hydrochloric acid gave a color not 

 essentially different, which faded on drying to a rather dirty brownish 

 red. The evolution of the two hydrides was apparently more rapid 

 than either alone, and the mixed baud was longer than from either 

 amount. 



It is evident from the above results that if we have with the arsenic 

 an amount of hydrogen sulphide even below that required to give a 

 band of the same length as the arsenic, the latter will be so altered as 

 to make its quantitative estimation impossible and its detection doubt- 

 ful. But, as unavoidable amounts of hydrogen sulphide would be held 

 back completely b)^ lead acetate paper, we should have no difficulty in 

 estimating the arsenic if the solution had not been properly oxidized 

 before testing. Even if the solution contains considerable reducible 

 sulphur, the lead acetate paper will protect the mercuric chloride 

 strip. 



We have also little to fear from phosphine, since we should not put 

 a solution into the reduction bottle until the phosphorus had been 

 oxidized as completely as possible. Accidental amounts of phosphine 

 would not affect the quantitative estimation of the arsenic. We have 

 not thought it necessary, for this reason, to verify the statement of 

 Dowzard ^'^ that phosphine is held back by cuprous chloride in hydro- 

 chloric acid solution, nor have we sought any other reagent which 

 could be adapted to this purpose under the conditions of our method. 



In the presence of stibine arsenic may be qualitatively recognized, 

 but not quantitatively determined, when the amount of antimony is 

 enough to give, if alone, an ammonia band equal to that of the arsenic. 

 But we should not test a solution without getting rid of any antimony 

 it might contain, and the methods for that purpose are satisfactory. 

 Slight traces of antimony would not affect the determination. 



If the arsenic is accompanied by any two or all three of the sub- 

 stances in question, cases which we think would seldom arise, their in- 



3* Loc. cit. 



