98 RESEARCHES UPON ATOMIC WEIGHTS. 



The platinum-plated brass weights were twice carefully standardized to hun- 

 dredths of a milligram, and the two sets of corrections agreed to within two 

 hundredths of a milligram in every case. Two short-armed Troemner balances, 

 sensitive to 0.02 mg. with a load of 50 gm. were used in the work. All weighings 

 were made by substitution. In the case of the Gooch crucible a platinum tare, 

 and in the case of the porcelain crucibles, porcelain tares were used. 



In the preceding table are given the results of all the syntheses which were 

 completed without accident. 



The results tabulated in Series I were undoubtedly less accurate than those of 

 Series II, since it was in the earlier analyses that experience in handling silver 

 iodide was gained. Hence these analyses are collected by themselves. In 

 Series II the rejection of analysis 12 seems justifiable, since two other analyses 

 made with the same material yielded results in close agreement and con- 

 siderably higher than that of analysis 12. 



From an examination of this table two important deductions may be drawn. 

 First, there can be little doubt of the identity of the different samples of iodine; 

 for the extreme averages of the results from the various samples differ by only 

 seven thousandths of a unit. The variation in the values from the analysis of 

 Sample II, fraction i, is undoubtedly due in the case of analysis 8 to lack of 

 experience in manipulating silver iodide, for this analysis was the first one to be 

 completed in the second series, and in the case of analysis 9 to the small quan- 

 tity of available material. One may conclude without hesitation that, in the 

 material examined in this investigation at least, no important quantity of a 

 new halogen existed. 



The second important deduction is that the atomic weight of iodine is evi- 

 dently very slightly greater than 126.913; for most of the experimental errors, 

 such as loss of silver iodide, or loss of iodine by the silver iodide, would have 

 caused the result to be too low. 



