﻿444 Prof. Lyle and Mr. Baldwin on Propagation of 



the retardation itself a maximum at distances between 40 

 and 50 cms. from the starting-point, depending upon fre- 

 quency and initial flux. After this the retardation diminishes, 

 that is, the phase of the flux advances for the remainder of 

 the length of the rod, and near the end the rate of advance 

 ( — d9/dx) is greatly increased, this latter being due to the 

 end effect. 



From this it would seem that the deduction by Oberbeck 

 and others of a " velocity of magnetization " from the rate of 

 retardation (dO/dx) of the phase of the flux close to the 

 source by means of the formula 



2tt 



dx 

 can hardly be legitimate. 



For if so, we should have to allow that the " velocity " is 

 infinite where the phase is stationary and d6/dx = 0, and 

 negative beyond this point, where the phase advances and 

 dSjdx is negative. 



It is interesting to note that the points at which maximum 

 phase retardation and maximum leakage coefficient occur 

 are always situated very near each other. 



7. It was thought that perhaps for the same frequency 

 the phenomena at points in the rod beyond x might depend 

 in great part only on the flax at x ; and if that were so, by 

 making the initial flux as small as that at which maximum 

 retardation occurred in any one of the preceding series, we 

 should get an advance of phase right from the start at the 

 magnetizing coil. To test this a rod of J in. Lowmoor iron 

 12 i't. long, taken from the same stock as specimen A, was 

 used *. 



The phase of the flux-wave at any point was taken as 

 given directly by the divided-circle readings of the wave- 

 tracer when the ordinate was zero, and the amplitude was 

 obtained by applying the proper factor to the maximum 

 ordinate of the same flux- wave. This was sufficiently 

 accurate for the purpose in hand. The results obtained are 

 exhibited graphically in figs. 3, 4, and 5. The initial values 

 of the fluxes used were 69, 163, 288, 687, 1732, 2994, and 

 4100, while the period was *052 sec. (q.p.) in all cases. In 

 fig. 3 the logarithms (nat.) of the fluxes are plotted as ordinates 

 against the corresponding distances from the origin as 

 abscissae, the points pertaining to any one initial flux being 



* When we thought of this, test specimen A had been cut up for the 



purpose of investigating- change of length, 



