256 PROCEEDINGS OF THE AMERICAN ACADEMY. 



balanced with weights. Upon placing the electro-magnet in its former 

 position and passing the same current through it as before, no perceptible 

 difference in the weight of the iron could be observed. Two repetitions 

 of the experiment gave the same result. Evidently the weight of iron is 

 not influenced by the earth's magnetism to an extent which would 

 necessitate the consideration of the permanent magnetic field even in the 

 most accurate analytical work. 



About ten grams of pure ferric oxide were now counterbalanced and 

 the electro-magnet placed in a horizontal position so that one pole was 

 directly below the weighing bottle and five centimeters distant from it. 

 A much stronger current (0.4 ampere) was passed through the magnet, 

 but no perceptible change in the weight of the oxide could be observed. 

 The same result was obtained when the oxide was replaced by about the 

 same quantity of bromide. When the metallic iron was placed upon the 

 balance pan, an increase in weight of 0.0035 gram was produced by 

 the magnet under the above conditions. The empty balance was not 

 affected in the least by the magnet. It is easy to calculate from the figures 

 given above that the vertical pull of the magnet upon the iron in the 

 new position must be over a thousand times as great as it was in the more 

 distant position. Hence it is not surprising that the magnet produced 

 no change in the first experiment, for a difference in weight of 0.000003 

 gram could not have been detected. 



To sum up the results of the investigation : — 



1. The result for the atomic weight of iron previously obtained by 

 analysis of the oxide, 55.88 (0=16.000), was confirmed. 



2. The specific gravity of ferrous bromide was found to be 4.636, at 

 25° referred to water at 4°. 



3. The effect of the earth's magnetism on the weights of small 

 quantities of the magnetic metals was shown to be negligible in even 

 the most precise atomic weight work, as was to have been expected. 



I wish to express my gratitude to Professor T. W. Richards for many 

 friendly suggestions in the course of the work, especially in connection 

 with the effect of terrestrial magnetism upon the weight of the magnetic 

 metals. I am indebted to the Cyrus M. Warren Fund for Research in 

 Harvard University for much of the necessary apparatus, and to Dr. 

 Wolcott Gibbs for indispensable platinum vessels. 



