magnets at all, but to substitute for these some other force which
should be either absolutely constant, such as the force of gravity, or
at least should, vary only within narrow limits, and this in accordance
with a definite law. This latter condition can be fulfilled by the
employment of electro-magnets.

[Footnote 1: According to recent experiments made by Dr. Hammerl, the
density of current in a copper voltameter should be half an ampere per
square inch of surface.]

[Illustration: FIG 3.]

To imitate with an electro magnet as nearly as possible a permanent
magnet, so that the former can be used to replace the latter, it is
necessary that the magnetism in the iron core should remain constant.
This could, of course, be done by exciting the electro magnet with a
constant current from a separate source. (In a recent note to the Paris
Academy of Science, M.E. Ducretet described a galvanometer with steel
magnet, which is surrounded by an exciting coil. When recalibration
appears necessary, a known standard current from large Daniell cells is
sent through this coil during a certain time, and thus the magnet is
brought back to its original degree of saturation. M. Ducretet also
mentions the use of a soft iron bar instead of a steel magnet, in which
case the current from the Daniell cells must be kept on during the time
an observation is taken.) But such a system would appear to be too
complicated for ready use. Moreover, some sort of indicator would be
required by which we could make sure that the exciting current has the
normal strength.

[Illustration: FIG 4.]