24 BULLETIN 1115, U. S. DEPARTMENT OF AGRICULTURE. 



There is a marked difference in the rate of change in the samples 

 attributable to the types of containers. The samples stored in the 

 metal drmns showed practically no change. The change in those 

 stored in the other containers varied from a very slight one, in 

 the case of the tightest hardwood barrels, to a marked one in the 

 case of the more open containers, such as sugar barrels. For the 

 hardwood barrels the average increase in water-soluble arsenic oxid 

 after 12 months was 0.06 per cent (maximum, 0.31 per cent) and 

 in carbon dioxid, 0.85 per cent; at the end of 20 months, the average 

 increase was 0.15 per cent in soluble arsenic oxid (maximum, 0.53 

 per cent) and 1.40 per cent in carbon dioxid. The samples stored 

 in the paper-lined and unlined veneer drums showed little difference. 

 At the end of 12 months the average increase in soluble arsenic 

 oxid was 0.15 per cent (maximum, 0.55 per cent) and in carbon 

 dioxid 1.38 per cent. After 20 months the average increase in 

 •soluble arsenic oxid was 0.30 per cent (maximum, 0.65 per cent) 

 and in carbon dioxid 2.21 per cent. In the lined and unJined sugar 

 barrels the average increase in soluble arsenic oxid after 12 months 

 was 0.31 per cent (maximum, 0.57 per cent) and in carbon dioxid- 

 1.72 per cent. After 20 months the soluble arsenic oxid increase 

 averaged 0.45 per cent (maximum, 0.63 per cent) and the carbon 

 dioxid increase averaged 2.01 per cent. 



The average results showing the behavior of the material stored in 

 the different containers are given in Table 3 and Figures 47, 48, 49, 

 and 50. Each curve represents the average of the products of dif- 

 ferent manufacturers contained in packages of the same type. Four 

 types of containers, metal drum (1), paper-lined sugar barrel (3), 

 hardwood barrel (4), and paper-lined veneer drum (6), are repre- 

 sented under the six brands ^ included in the subdivision tests, and 

 all of these (23 subdivisions) were subsampled on exactly the same 

 dates. Each type of container is treated separately m order to 

 bring out the parallelism between the carbon dioxid and water- 

 soluble arsenic changes, and two graphs, showing all four cui*ves for 

 carbon dioxid and for arsenic oxid, respectively, are presented in 

 Figures 51 and 52. 



8 Only five brands are represented in the ease of the paper-lined barrels. 



