120 PROCEEDINGS OF THE AMERICAN ACADEMY 



more than fourteen or fifteen per cent. Some general reasoning 

 based upon the number of oscillations may be of interest. It must 

 be acknowledged, however, that' this reasoning is open to criticism, 

 although it affords the most plausible explanation. The phenomenon 

 itself is not a doubtful one. 



The time of a double oscillation for the large-sized copper wire was 

 .0000020 sec; for the small copper wire, .0000021 sec. The others 

 as far as could be determined did not differ much from these values, 

 and for this purpose either is sufficiently accurate. Denote by R' the 

 ohmic resistance of the parallel wires to alternating currents of this 

 periodicity ; by R, the resistance to steady currents. 



p^ — = 3,000,000 (practically). 



Taking the cases up in order : 

 (1.) Large copper wire, 



R = 0.285 X 10^ 



and substituting in Lord Rayleigh's formula, R' = y/^plixR, 



i?' = 0.66 X 10^ 



(2.) Large German silver wire, 



i? = 9.2 X 10^ 

 and substituting in the series 



^ ^ V + 12 ~^r ~ i8o E* +•••;' 



R> = 9.2 X 10^ 



(3.) Large iron wire, 



R =2.5 X 10^ 

 and if there is a true time lag, as often stated, such as to prevent 

 action of the magnetic property of the iron, and if on this assumption 

 we make /a = 1, 



R' = 2.78 X 10^ 



(4.) Fine copper. 



R =3.S X 10^ 

 R' =3.5 X 10^ 



(5.) Again, as before, call yu, = 1 in iron, nickel, and steel. The 

 lenjith of these circuits was 7.41 meters, the remaindi^r of the 10.20 

 meters — 2.79 meters — being of copper wire of R' =: 0.94. 



Tlie value of R' in the separate cases, including in each the resist- 

 ance 0.94 of the copper portion, was as follows : — 



