Magnetic Induction in a Long Iron Bar. 



269 



was likewise read off' curve R, and is given in Table II. 

 under B/,; it is the induction that would have been produced 

 by the mean value of H had the induction been uniform all 

 along the bar. 



The values of fju at successive points were likewise read off 

 from curve, and these are shown in curves T (fig. 8). 



Table II. contains also the ratios of some of the more 

 important quantities. 



Fig. 8. 



2000 w 



Centre 22 20 18 16 ) 4 12 10 8 6 4 2 end 

 Distances along Bar in inches. 



On comparing P and R (fig. 5) it will be noticed that the dif- 

 ferences in the abscissas (giving the " demagnetizing force " at 

 the centre) are far from proportional to B, until we get to 

 regions where the induction through the bar is so high as to be 

 practically constant all alono- (as shown in Q. IX., fig. 4) ; below 

 such values the demagnetizing force is considerably larger than 

 the amount that is usually taken for the length-ratio of 250 to 1 

 that was used here. Possibly some of this difference is due to 

 imperfect annealing ; but great care was taken, and curves P 

 and R agree very well in those parts w here from Q. IX. we 

 see the induction is nearly constant along the bar. Further- 

 more, we should expect that with the great variations shown 

 in set Q the leakage-paths would demand very varying pro- 

 portions of the impressed magnetomotive force, instead of a 

 constant portion, as is the usual assumption. In fact, we 

 cannot possibly tell what value of B to take in the equation 

 H = A,B for demagnetizing force, since in the cylindrical bar 

 (unlike the ellipsoidal) the leakage-flux is not independent of 

 the induction. The ratio of H TO to H^ attains a minimum 



