28 PROCEEDINGS OF THE AMERICAN ACADEMY 



of the silver corrected by subtracting from the weight given in 

 column VI. the amount of silver corresponding to the quantity of 

 acid given in column VII. In the same way, the ninth and tenth 

 columns contain respectively the total and the corrected weight of 

 argentic bromide. Hence the weights actually used in the calcula- 

 tion of the results are those recorded in columns V., VIII., and X. 



The discussion of the results is simplified by reducing all the 

 amounts of baric bromide to the basis of 100.000 parts of silver, 

 and the corresponding quantity, 174.080 parts, of argentic bromide. 

 The water of crystallization is included in the following table only 

 because the calculation of Analyses 2, 8, 14, and 18 depends upon 

 the knowledge of its amount. The great variations noticeable in 

 the results for the water of crj^stallization are due to the varying 

 circumstances attendant upon the crystallization, to the fineness of 

 the powder, and to the hygroscoiDic condition of the air at the time 

 of weighing the crystals. Hence for the present purpose it was 

 possible to compare only like samples which had been weighed out 

 under like conditions. Analyses 13 and 15 show that in this way 

 perfect constancy can be reached. This part of the work has of 

 course no other bearing upon the atomic weiglit of barium. 



The first two experiments were merely preliminary, and are not 

 included in the final average. Most of the variations evident in 

 the earlier experiments were undoubtedly due to unfavorable con- 

 ditions existing in the Laboratory during the year 1891-92. In 

 the autumn of the latter year the Laboratory was completely and 

 most admirably remodelled, through the kindness of the Corporation 

 of the University, and the last seven experiments were performed 

 under conditions as favorable as could be desired. 



The presence of any of the most likely metallic impurities — 

 strontium, calcium, potassium, or sodium — would tend to lower 

 the observed values recorded in the third and fourth columns of the 

 Table of Results, and hence the atomic weight of barium. Chlorine 

 would lower and iodine would raise the values given in the third 

 column, but neither would have much effect on those given in the 

 fourth column. The best possible proof of the freedom of the prep- 

 arations from these two impurities, as well as of the purity of the 

 silver, is to be found in the series of results giving the per cent of 

 silver in silver bromide, tabulated in the fifth column of the Table 

 of Results.* The presence of water in the ignited baric bromide 

 would naturally tend to raise the figures given in both the third 



* See also these Proceedings, XXV. 212. 



