Estimation of Minute Quantities of Arsenic in Copper. 295 



ponding to the allotropic conditions of the arsenic. The 

 exigency compels, moreover, the substitution of hydrochloric 

 acid for the sulphuric acid usually employed in the reduc- 

 tion flask ; but, though the opinion is current that hydrochlo- 

 ric acid introduces difficulties in the Marsh test, we have been 

 unable to discover any evidence of the formation of a zinc mir- 

 ror in the ignition tube or to note other unfavorable action due 

 to the use of pure hydrochloric acid. It is, of course, obvious 

 that the hydrochloric acid used must be actually free from 

 arsenic (as was ours) and not merely nominally so, as is often 

 the case with the so-called arsenic-free hydrochloric acid of 

 commerce. 



The copper for our test experiments was prepared free from 

 arsenic by electrolizing in ammoniacal solution the purest copper 

 sulphate obtainable and stopping the deposition before the 

 solution had become exhausted. In this manner we were able 

 to procure copper in which we failed to detect arsenic. This 

 copper was dissolved in nitric acid, arsenic in the higher condi- 

 tion of oxidation was added, and the process of the separation 

 of the arsenic from the copper and conversion to the mirror 

 earned out in the manner described. The results obtained are 

 recorded in the accompanying table. 







Mirror estimated (by 









comparison with 





Copper taken, 



Arsenic taken, 



standard mirror) 



Error, in 



in grams. 



in milligrams. 



in milligrams. 



milligrams. 



none 



none 



none 



none 



0-7 



none 



none 



none 



0-5 



O'OOo 



0-003 



0-002 — 



0-5 



o-oii 



0-013 



0-002 + 



0-35 



0-020 



0-015 



0-005 — 



0-3 



0-030 



030 



none 



0-43 



0-040 



0-035 



0-005 — 



0-44 



0-050 



0-040 



o-oio— 



It is plain from these results that the method is capable of 

 detecting sharply minute amounts of arsenic in copper and of 

 effecting the estimation of quantities less than 0*05 mg. with 

 some approximation to accuracy. There is, as Sanger has 

 pointed out, a good deal of variation even in standard mirrors 

 made with all possible care and precaution, and in the estima- 

 tion of mirrors containing as much as O05 mg. of arsenic the 

 uncertainty of comparison as well as the actual variation of the 

 mirror is considerable. 



When a sample of copper is under test which may contain 

 more than 0-05 mg. of arsenic it is desirable to introduce into 

 the reduction-flask the measured solution containing the 

 arsenic gradually and in definite portions, and to judge by the 



