PEIRCE. — MAGNETIC TESTS UPON IRON AND STEEL RINGS. 93 



TABLE V. 



Ring of the Bessemer Steel. {Rectangular Cross Section. Inner 

 Radius, 2a; Outer Radius, 3a.) 



Rule appears to be 2185, and the value of B corresponding to the average 

 value 1.0069 of H is also 2185, so for these dimensions and for this par- 

 ticular excitation, the error represented by B' — B" seems non existent. 



For an excitation great enough to make the value of H at the mean 

 radius 2, a process similar to that just described shows that B' would 

 be 6630 and B" , 6650 ; but if H^ were made 5, the value of B' would 

 be 12560 and B" , 12590. The difference in this instance is less than 

 one quarter of one per cent of either quantity and lies within the limits 

 of error of most magnetic measurements made upon ring specimens. 

 For work that must be very accurate, rings much thinner than this one 

 — in which the ratio of the outer radius to the inner radius is 4/3 — are 

 usually employed, and the error is then practically negligible for almost 

 all excitations. 



If magnetic measurements are to be made upon rings of the dimen- 

 sions sometimes used in practical permeameters, the errors arising from 

 the difference between B' and B" become relatively important as ap- 

 pears from Tables III and V, where the results for two practical cases 

 are given. A comparison of Tables III and IV will show how fast the 

 error decreases when the ring is made thinner. 



My thanks are due to the Trustees of the Bache Fund of the National 

 Academy of Sciences for the loan of apparatus. 



The Jefferson Physical Laboratory, 

 Cambridge, Mass. 



