﻿Waves of Magnetic Flux along Iron Wires. 443 



where the leakage coefficients \x= —d(logf^)/d/v are plotted 

 against the distances from the origin at which they occur, 

 as well as against log/i, the logarithm of the corresponding- 

 flux. Thus in all cases, including the statical, as we move 

 away from the magnetizing coil, \i first increases, attains a 

 maximum, and then diminishes until near the end of the rod 

 when it again increases. For the same initial flux both the 

 points at which max. X x occurs and the corresponding fluxes 

 are different for different frequencies. 



Figs. 1 and 2 show the effects of the end of the rod on 

 the characteristics of the fluxes in its neighbourhood. Thus 

 in fig. 1 the downward turn of the flux and phase curves 

 due to the end effect can be recognized 30 cm. from the end 

 of the rod, and the corresponding rise in the \ curves 

 in fig. 2 is very marked. The end effect always manifests 

 itself in this way, and in very long rods this change in the 

 direction of curvature does not appear. 



From the X l5 log f\ curves in fig. 2 we see that for the 

 same value of f u Xj increases with the frequency. In 

 the same set it will be noticed that one curve 1 can be drawn 

 (q.p.) through the two series of points for \ x when T = - 053 

 arising from two different initial fluxes. This would su a 

 that X x is independent of x and is a function of f\ only. In 

 the sequel it will be seen that X t does depend on x as well as 

 on/i, and the coincidence of the above two curves is due to 

 the fact that with the initial values used (Tables I. and II.) 

 there is never much difference between the abscissa? at which 

 equal fluxes occur in the two series. 



Returning to fig. 1 we see how, for different frequencies, 

 the phase of the first harmonic of the flux at any point x 

 lags behind that of the initial central flux at <r = 0. 



Thus as the flux moves away from the magnetizing coil its 

 phase is at first retarded at a fairly regular but diminishing 

 rate for any one frequency, while for the higher frequencies 

 the rate of retardation cW/dx is higher. To this part of the 

 phenomenon the observations of Oberbeck, Zenneck, and 

 Perkins have been limited, and following them we would 

 deduce (see § 1) from our results the following mean values 

 for the " velocity of magnetization " over the first 30 cms. 

 of the rod in question : — 



3380 cm./sec. when T = '053 sec. 

 4790 cm.'/sec. „ T = '032 „ 

 6230 cm./sec. „ T='0216 „ 



Continuing along the rod, however, we see that the space- 

 rate of retardation (dd/dx) of the phase becomes zero, and 



