put within it, which however had no metallic connexion with that in the
quill, except through the magnet itself and the copper rod (fig. 34.). The
wires A and B from the galvanometer were dipped into these two portions of
mercury; any current through them could, therefore, only pass down the
magnet towards its equatorial parts, and then up the copper rod; or vice
versa.

226. When thus arranged and rotated screw fashion, the marked end of the
galvanometer needle went west, indicating that there was a current through
the instrument from A to B and consequently from B through the magnet and
copper rod to A (fig. 34.).

227. The magnet was then put into a jar of mercury (fig. 35.) as before
(219.); the wire A left in contact with the copper axis, but the wire B
dipped in the mercury of the jar, and therefore in metallic communication
with the equatorial parts of the magnet instead of its polar extremity. On
revolving the magnet screw fashion, the galvanometer needle was deflected
in the same direction as before, but far more powerfully. Yet it is evident
that the parts of the magnet from the equator to the pole were out of the
electric circuit.

228. Then the wire A was connected with the mercury on the extremity of the
magnet, the wire B still remaining in contact with that in the jar (fig.
36.), so that the copper axis was altogether out of the circuit. The magnet
was again revolved screw fashion, and again caused the same deflection of
the needle, the current being as strong as it was in the last trial (227.),
and much stronger than at first (226.).

229. Hence it is evident that there is no discharge of the current at the
centre of the magnet, for the current, now freely evolved, is up through
the magnet; but in the first experiment (226.) it was down. In fact, at
that time, it was only the part of the moving metal equal to a little disc
extending from the end of the wire B in the mercury to the wire A that was
efficient, i.e. moving with a different angular velocity to the rest of the
circuit (258.); and for that portion the direction of the current is
consistent with the other results.

230. In the two after experiments, the _lateral_ parts of the magnet or of
the copper rod are those which move relative to the other parts of the
circuit, i.e. the galvanometer wires; and being more extensive,
intersecting more curves, or moving with more velocity, produce the greater
effect. For the discal part, the direction of