SEPTEMBER 27, 1918] 
on large iron rods by a method depending on 
the principles of electromagnetic induction; 
the second, on smaller rods of iron, cobalt and 
nickel, by the method of the magnetometer. 
Recently a few preliminary experiments have 
been made on Heusler alloy. 
Some of the essential parts of the apparatus 
used in the first investigation are shown in 
the diagram of Fig. 3. 
Two nearly similar rods of steel shafting A 
and B were mounted with their axes horizontal 
and perpendicular to the magnetic meridian, 
and two similar coils of insulated copper wire 
were mounted about their centers. These coils 
were connected in series with one another and 
with a Grassot fluxmeter, which was the prin- 
cipal measuring instrument, and were oppo- 
sitely wound so that that any variations in the 
intensity of the earth’s magnetic field acting 
in the same way on both rods might produce 
Fig. 4. 
no effect on the fluxmeter. One of the rods, 
which will be called the compensator, as A, re- 
mained at rest; while the other, called the 
rotor, as B, was alternately rotated and 
brought to rest, the change of flux being de- 
termined by the fluxmeter, which, together 
with the.other apparatus, was standardized by 
proper subsidiary experiments. For use in 
these experiments the rods A and B were uni- 
formly wound with solenoids of insulated wire. 
To prevent possible disturbances arising 
from the presence of the earth’s magnetic field, 
the rotor was surrounded by a large electric 
coil which approximately neutralized the 
earth’s intensity in the region occupied by the 
rotor. 
The rotor was directly driven in either di- 
rection at will by an alternating current motor 
SCIENCE 
307 
in part of the work, and by an air turbine in 
the rest. 
In making observations fluxmeter deflections 
were obtained for each of several speeds, first 
with the rotation in one direction and then 
with the rotation in the other direction. 
After making a great variety of tests, and 
after taking many precautions to eliminate 
sources of error, two effects stood out very 
clearly as the result of the observations, instead 
of the one which was looked for. 
If the mean of the two deflections for the 
same speed is plotted against the square of the 
speed, the resulting graph is a straight line as 
shown in Fig. 4. The mean deflection is thus 
proportional to the square of the speed. This 
deflection is due to the increase of the residual 
magnetic flux through the rotor produced by 
its centrifugal expansion during rotation—an 
effect which was not foreseen, and which was 
Fie. 5. 
very puzzling until its explanation became 
apparent. This effect would vanish if the rod 
were completely demagnetized initially. 
If, however, the difference between the two 
deflections for the two directions of rotation, 
instead of the mean deflection, is plotted 
against the speed, and not against the square 
of the speed, a straight line again results, as 
shown in Fig. 5. This is the effect which was 
under investigation. The straight line shows 
that H is proportional to N, as predicted. 
The earlier experiments by this method gave 
for H/N the mean value — 3.6 X 107 em.u.; 
the later and more precise experiments gave 
— 3.1107 em.u., with an experimental error 
for a set of four double deflections equal to 
about 12 per cent. The graph of Fig. 5 is 
drawn for these observations, the dotted 
