CONTRIBUTIONS FROM THE CHEMICAL LABORATORY 

 OF HARVARD COLLEGE. 



A REVISION OF THE ATOMIC WEIGHT OF IRON. — 



PRELIMINARY PAPER 



By Theodore William Richards and Gregory Paul Baxter. 



Received and presented December 13, 1899. 



It is rather surprising that the atomic weight of a metal so important 

 as iron should have received so little recent attention us it has. Berze- 

 lius* was among the first to determine the quantity under discussion ; and 

 in order to accomplish the determination he converted Swedish piano- 

 wire into ferric oxide. For many years his very low result, 54.3, was 

 almost unchallenged, when in L843 Wackenroder called attention to the 

 fact that in some early analyses of ferric oxide Stromeyer and he had 

 found the suhstance to contain oidy 30.1 per cent of oxygen, correspond- 

 ing to an atomic weight of nearly 56. This led immediately to several 

 investigations by Svanberg and Norlin, Berzelius, and Erdmann and 

 Marchand, whose individual results upon the same ratio varied from 55.81 

 to 5G.23, showing without a doubt that the low value was incorrect, and 

 settling the question with all the accuracy necessary in that day. Rivot 

 in 1850 published two poor analyses which shed no further light upon 

 the subject; and the more recent work of Dumas, who analyzed the 

 chlorides of iron in 1859, was no more accurate than his similar work 

 on other metals. Since that time no work on the atomic weight 

 worthy of notice has appeared, so that our present knowledge depends 

 upon data obtained over fifty years ago. In the intervening time Stas 

 has taught the world accuracy in chemical analysis, and countless com- 

 plications and sources of inaccuracy have been discovered which then 

 were unsuspected. The variations of nearly 0.8 per cent in the best de- 

 terminations (all depending upon the composition of ferric oxide), point 

 with emphasis toward the necessity of a modern determination. Such 



* The references to all the sources of information are to be found in the critical 

 discussion at the close of this paper. 



