180 
PHYSICS: 5. /. BARNETT 
the large coil and mounted near the upper magnetometer magnet to 
make the sensibility approximately independent of the compensating 
current, produced in the region occupied by the rotor intensities so 
small that their effect was negligible. 
Rotation observations were made at equal intervals in sets of four as 
follows: The rotor was first driven (by means of an alternating current 
motor) at given speed in one direction and the magnetometer scale read; 
then the motor was reversed and the scale again read ; then the readings 
were repeated in inverse order, all for the same speed. The double 
deflection obtained by subtracting the mean of the second and third 
readings from the mean of the first and fourth was the quantity sought. 
This process eliminated the difficulties due to the presence of residual 
magnetization of the rotor, the error due to magnetometer drift, and 
other possible errors. Error due to torsion of the rotor was found to 
be negligible by reversing some of the rotors in their bearings. 
With nickel and cobalt observations were made at three speeds, and 
H/n was found to be independent of the speed within the limits of the 
experimental error, as had been found in the earlier experiments with 
iron. 
Since mechanical disturbances were almost wholly absent and the 
magnetic disturbances became relatively less important with increased 
speed, the observations at the highest of the three speeds were more 
precise than the others. Table 1 contains the approximate results of 
the observations on four rods at the highest speeds. A few observa- 
tions, consistent with the others, on a fifth rod, of soft iron, in poor 
condition are not included in the table. 
TABLE 1. 
ROTOR 
MEAN SPEED 
NO. OF SETS 
10^ x~ 
AVERAGE 
DEPARTtJRE 
FROM MEAN 
Steel (larger) 
47 r. p. s. 
21 
5.2 e. m. u. 
1.2 
Steel (smaller) . 
44 
21 
5.2 
0.5 
Cobalt 
45 
79 
6.1 
0.8 
Nickel 
45 
37 
6.1 
0.9 
The mean value of H/n is in all cases less in magnitude than the 
standard value of 4:irm/e = —7.1 X 10"^ e.m.u. for electrons in slow 
motion, as was the case in the earlier experiments; but the experimental 
errors, on account of the great difficulties involved, are such that im- 
portance cannot in my opinion be attached to the discrepancy. The 
investigation must rather be taken as confirming equation (1) and the 
assumption that only electrons are in orbital revolution in all the sub- 
stances investigated. 
