position within the box, the latter is closed with a strong lid, and
covered with closely fitting metal sheets, the joints of which are
soldered very carefully. On the top two small holes are drilled,
passing through the metal sheet and the wood, and in these holes two
small glass tubes are inserted and the joints made air-tight. One of
the tubes is connected to a vacuum pump, and the other with a vessel
containing a sufficient quantity of boiled-out oil. The latter tube
has a very small hole at the bottom, and is provided with a stopcock.
When a fairly good vacuum has been obtained, the stopcock is opened
and the oil slowly fed in. Proceeding in this manner, it is impossible
that any big bubbles, which are the principal danger, should remain
between the turns. The air is most completely excluded, probably
better than by boiling out, which, however, when gutta-percha coated
wires are used, is not practicable.

For the primaries I use ordinary line wire with a thick cotton
coating. Strands of very thin insulated wires properly interlaced
would, of course, be the best to employ for the primaries, but they
are not to be had.

In an experimental coil the size of the wires is not of great
importance. In the coil here used the primary is No. 12 and the
secondary No. 24 Brown & Sharpe gauge wire; but the sections may be
varied considerably. It would only imply different adjustments; the
results aimed at would not be materially affected.

I have dwelt at some length upon the various forms of brush discharge
because, in studying them, we not only observe phenomena which please
our eye, but also afford us food for thought, and lead us to
conclusions of practical importance. In the use of alternating