1284 
When d makes contact with a, the current flows along c,a,9,7.q, 
coil, p,7,¢ to e; when, on the contrary, dr touches 6, the current 
takes the way d,7,p,coil, ¢,j,g, to e, circulating in opposite direction 
about the cylinder S, so that, the pendulum being in motion, the 
cylinder S is alternately magnetized in one direction and the oppo- 
site one, synchronically with the pendulum. The potential difference 
between c and d, d and e has been chosen in such a way, that 
the cylinder is magnetized to a sufficient degree. 
The torsional motion of the cylinder was observed with telescope 
and scale. 
§ 3. Compensation of the terrestrial field. 
The following considerations show of how much importance this 
is. If the direction of the magnetization deviates 3’ or 4’ from the 
vertical, the horizontal component amounts to one 1000" of the 
vertical magnetization. Now a deviation of this amount may easily 
oceur as the suspension is not accurately centred and the windings 
are not laid quite regularly’). 
If the field of the earth were not compensated and the horizontal 
component of the magnetization were perpendicular to the magnetic 
meridian, the vibrations resulting from the above mentioned cause 
would have an amplitude equal to 180 times that which we wish 
to observe, (for the period used in the experiments). 
We shall now describe in what way a rather satisfactory compen- 
sation of the field can be reached. The compensation problem is the 
same as in the case of a sensitive galvanometer, with this difference, 
that here we have to compensate over the whole length of the 
cylinder with that of the wires p and q in addition, so that the 
terrestrial field has to be compensated in a space of breadth 5 to 
7 mm and length 35 to 40 mm. ; 
In order to make an estimate as to how far this may be done, 
I constructed an apparatus, which made it possible to lower or to 
draw up by a spider-thread a light galvanometer mirror, to which 
two small strong magnets from a spring of Wolfram steel (they 
were about 5 mm long) had been fixed; by this means the mirror 
could be placed at any point of the field that had to be compen- 
sated. The spider-thread was so thin, that the mirror was suspended 
practically without torsion, so that only the terrestrial magnetic 
couple was acting. From the period of oscillation it is easy to deduce, 
in how far the field of the earth is compensated. Starting from a 
1) In a repetition of the experiments more attention than was possible now 
would have to be paid to an accurate construction in these two respects. 
