238 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



iron cylinder of m = 300, as being unwarranted. He determines K by 

 botb the ballistic and magnetometric methods for a rod of m = 300 

 by back-shearing the ballistic curve into the normal curve, using 

 N'^^i = 0.0005, according to Du Bois, thus finding the N to be 0.0028 

 for the magnetometric method. He would, therefore, correct Mann's 



Figure 19. 



Back-shearing curves for Ewing's soft iron wire of diameter 0.158 cm. 

 mined from results found in the present paper. 



Deter- 



values of N by adding 0.0020 to each N throughout. Now it seems 

 to us quite clear, as remarked somewhere in the earlier part of this 

 paper, that we have no right to assume that the normal / vs. H curve, 

 as obtained ballistically, should be even approximately the same as the 

 Mean / vs. Mean H curve of the magnetometric method. This as- 

 sumption is rendered particularly doubtful when we see the very wide 



