1911.] PUBLIC DOCUMENT — No. 31. 127 



two-solution processes over the one. In 1905 the results ^ with 

 the Avery-Thatcher and Avery-Haywood modifications were not 

 as satisfactory though the average difference was not excessive. 



In weighing the merits of the Thorn Smith process and various 

 modifications of the Avery-Beans, with apparently little choice 

 as to accuracy, the Avery or Avery-Haywood process, with one 

 titration of a single solution, certainly appeals to chemists in 

 '' control " work from the standpoint of manipulation, possible 

 mechanical losses and time. This does not warrant any less 

 care in conducting the analysis, but, if anything, demands 

 greater attention. The essential features of the Avery-Haywood 

 process have been employed at the Massachusetts station for 

 the work on arsenites, though considerably modified as to detail. 



Practice at Massachusetts Station. 

 Having adopted Thatcher's suggestion ^ as to ratio of sample 

 to acetate solution, 1 to 25, and finding 25 cubic centimeters 

 rather inadequate for proper boiling and agitation, double quan- 

 tity of each is taken. To prevent slight loss of sample in trans- 

 ferring to flask, due to both adhesion and dusting, boats of 

 folded filter paper are employed, and found very serviceable, 

 particularly for Paris green and arsenic for standard solution. 

 After boiling the solution five minutes with acetate, the direc- 

 tions call for the careful addition of concentrated hydrochloric 

 acid until solution is effected. Such a procedure in our hands 

 gave extremely variable results and generally a low test for 

 arsenic. This error necessitated several weeks of experiment- 

 ing, and was found to be due to the addition of concentrated 

 acid, dilute acid (1-3) giving uniform results in practically 

 every instance, and a higher test. Probably this has been one of 

 the sources of trouble v/itli the chemists reporting on association 

 samples by the above process in past years. Neutralizing with 

 sodium carbonate, in dry form or concentrated solution, will 

 introduce an error if added in excess. The use of sodium bicar- 

 bonate is preferable for the purpose as the latter salt does not 

 absorb iodine and eliminates an unnecessary reagent. As con- 



' Proc. Assoc. Off. Agr. Chem., 22, 27 (1905). 

 2 Proc. Assoc. Off. Agr. Chem., 21, 99 (1904). 



