PHYSICS: S. J. BARNETT 
179 
the effect converse to mine, viz., rotation by magnetization, which had 
been predicted and looked for by O. W. Richardson in 1907; and de 
Haas^ has recently continued this work in a somewhat different manner. 
Both investigations are indirect but excellent confirmations of my own 
earlier work. This work has also been confirmed by further experi- 
ments of my own of somewhat increased precision described before the 
American Physical Society in April, 1915. ^ 
In the last year, with financial aid received from the University 
through the interest of the dean of the graduate school. Prof. W. Mc- 
Pherson, and with the help of Mrs. Barnett, I have extended the in- 
vestigation to other specimens of iron and to cobalt and nickel. In 
all the earlier work the method of electromagnetic induction was used; 
this later work has been done by an entirely different method, viz., 
that of the magnetometer. 
The magnetometer was an astatic instrument, and each rod under 
experiment, or rotor, about 30.5 cm. in length and from 2.3 cm. to 3.2 
cm. in diameter, was mounted with its axis horizontal and normal to 
the magnetic meridian in the equatorial position of Gauss, which of- 
fered important advantages for this work. Calibrations were made 
by means of solenoids wound permanently on the rotors, and sub- 
sidiary solenoids wound on wooden cores. 
To avoid magnetic and mechanical disturbances as much as possible, 
nearly all of the observations were made after one o'clock at night. 
This precaution, together with the use of true, carefully adjusted, and 
frequently oiled bearings, heavy mountings of bronze and concrete, and 
a special method of driving, eliminated mechanical disturbances very 
largely. 
Disturbances due to variations of the earth's intensity were greatly 
reduced by mounting a compensating rod of the same substance and 
nearly the same size as the rotor in approximately the same position 
with respect to the upper magnetometer magnet as that occupied by 
the rotor with respect to the lower magnet. 
Possible. errors due to eddy currents in the rotor and to minute shifts 
of the rotor's axis in altitude or azimuth were avoided by compensating 
accurately the earth's intensity in the region occupied by the rotor by 
means of a very large coil traversed by a steady electric current. Much 
greater variations of the compensating current on both sides of the correct 
value than the maximum allowed in the rotation experiments produced no 
appreciable effect on the results obtained with some of the rotors and not. 
more than small effects with the others. The magnetometer magnets, 
control magnet, compensating rod, and a small electric coil in series with. 
