172 Perkins — Velocity of the Propagation of Magnetism. 



mined directly from the leakage curve, assuming it to be fairly 



well represented by the equation /3 — log ~ / x. The average 



value was *16, which gives a curve fitting the original fairly 

 well near its middle point. This agrees quite closely with 

 Oberbech's value for steel ; the difference in the position of the 

 decimal point is due to his choice of the meter as the unit of 

 length instead of the centimeter. 



The calculation of C m is more difficult, particularly as Dr. 

 Zennech is not quite explicit in explaining Y x ; but I venture 

 to suggest the following method as applied to a point four inches 

 from the center of the exciting coil. The difference of poten- 

 tial of the two faces of a coil of n turns carrying a current I is 



as determined by the amount of work done in carrying a 



unit pole from one face around outside the coil to the other. 

 This work would be increased if the medium were of greater 

 permeability than air ; moreover, the work done would be half 

 the total if the unit pole were carried only from one face to 

 the far end of the bar, where the ilux is almost zero. The fall 

 of potential w r ould then vary as Q along the bar. From these 

 considerations I calculated Y x by the following equation : 



4irnl. ' Q x 



V v = IX —pr- = 27, 



10 ^ 2Q ' 



where n = 4077 (No. turns in coil), 



I = '015 amperes, 



II = 2-95, 



and ~ = ^tq, as may be seen from the leakage curve.* The 



value of fM was found by carefully measuring the resistance of 

 the exciting coil and the currents flowing in it under an alternat- 

 ing pressure of 120 volts, when the steel core and an air core 

 were successively used. The impedance in one case was 7998 

 ohms, in the other 2727. Knowing the resistance and the 

 frequency, it was easy to calculate the coefficient of self-induction 



for iron (L t ) and for air (L a ), and fx = —• 



Referring to equation (2) we still have ~- to determine, which 



can be done readily by differentiating Q = Q e ~$ x and substi- 

 tuting the values of Q , j3 and x. {3 has been determined, x = 



* These and other values were taken from the original curves as plotted in 

 the laboratory. Those prepared for photographic reproduction give the gen- 

 eral form correctly but are not strictly accurate copies. 



