320 Andrews and Farr — Determination of Arsenic. 



solution by dissolving 9*900 grams of the oxide in sodium 

 hydroxide solution, saturating with carbon dioxide and then 

 making up to one liter. The amounts of arsenic which appear 

 in column 1 of the table were obtained by measuring the 

 appropriate volume of one of these solutions. 



All volumetric apparatus employed was, of course, carefully 

 standardized. The volumes of the standard solutions, as 

 given in columns 4 and 5, are corrected for temperature and 

 titre. 



A somewhat similar process to that described in the present 

 paper has been published by Engel and Bernard.* These 

 authors reduced the arsenic with a mixture of hypophosphor- 

 ous and hydrochloric acids and titrated the precipitated metal 

 in essentially the same manner as that adopted by us. Their 

 test -analyses are very good on larger amounts of arsenic. 

 They publish none for quantities less than 54 milligrams. 



The reaction is very much slower than it is with stannous 

 chloride. This necessitates allowing the mixture to stand for 

 twelve hours and then boiling, a proceeding which may well 

 result in loss of arsenious chloride. 



It is far less easy to obtain hypophosphorous acid free from 

 traces of arsenic than it is stannous chloride. The most 

 important advantage which the use of the tin salt presents is 

 probably to be found in the broad range of its applicability. 

 In almost all of the salts which one ordinarily desires to sub- 

 ject to an arsenic determination, no other operation is required. 

 Even compounds of lead, bismuth or antimony need no pre- 

 liminary separation. We have found that titanous chloride 

 (TiCl 3 ) may be substituted for the tin salt, but without advan- 

 tage so far as known. It is probable that the lower chlorides 

 of chromium, molybdenum, or vanadium would answer the 

 same purpose, but they hold out no promise of superiority. 



It is extremely likely that for the determination of fainter 

 arsenical mirrors, obtained by the Marsh method, the iodo- 

 metric titration will be found useful. A mirror of 0*01 mg. 

 can scarcely be weighed satisfactorily, even with the best 

 balance, because the weight of the tube when subjected to the 

 action of the necessary reagents might change by an amount 

 considerably in excess of the weight of the mirror. But it 

 could be dissolved in millinormal iodine or starch iodidef and 

 the excess titrated back by arsenite or thiosulphate of the 

 same normality. For smaller mirrors than this, optical methods 

 of comparison will still have to be used. 



Laboratories of the Mallinckrabt Chemical Works, 

 St. Louis, Jan. 23, 1909. 



* Comptes rendus, cxxii, 390, 1896. 



f Compare Zeitschr. f. anorg. Chem., xxvi, 180, 1901. 



