210 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



But as even for the high value ff = 30, B is somewhere near 15,000, 

 we see that the error introduced by neglecting the H in the brackets is 

 but 1 part in 500, which is much less than the experimental errors. 

 Therefore, since the ballistic throw is proportional to B, it is very 



O 10 16 



Figure 8. [Table IV.] 



Reversal magnetization curvfes for a Bessemer soft steel rod of diameter 

 b.4763 cm. 



much more convenient to construct the iNT-curves from the formula 



H=H'-KB/i7r. 



Experimental Results for Demagnetizing Factors. 



Let us now tabulate the actual values obtained for the end correc- 

 tions, or demagnetizing factbrs iV, of a number of rods of Bessemer 

 steel (copper coated), which is a very homogeneous soft iron. Later on 

 we shall see just how these values were determined, and give the 



