which is nearest to the refractory incandescent body--should be cut
out diagonally, as it often happens that, in consequence of the
intense heat, this corner turns toward the inside and comes very near
to, or in contact with, the wire, or filament, supporting the
refractory body. The greater part of the energy supplied to the bulb
is then used up in heating the metal tube, and the bulb is rendered
useless for the purpose. The aluminium sheet should project above the
glass stem more or less--one inch or so--or else, if the glass be too
close to the incandescing body, it may be strongly heated and become
more or less conducting, whereupon it may be ruptured, or may, by its
conductivity, establish a good electrical connection between the metal
tube and the leading-in wire, in which case, again, most of the energy
will be lost in heating the former. Perhaps the best way is to make
the top of the glass tube, for about an inch, of a much smaller
diameter. To still further reduce the danger arising from the heating
of the glass stem, and also with the view of preventing an electrical
connection between the metal tube and the electrode, I preferably wrap
the stem with several layers of thin mica, which extends at least as
far as the metal tube. In some bulbs I have also used an outside
insulating cover.

The preceding remarks are only made to aid the experimenter in the
first trials, for the difficulties which he encounters he may soon
find means to overcome in his own way.

To illustrate the effect of the screen, and the advantage of using it,
I have here two bulbs of the same size, with their stems, leading-in
wires and incandescent lamp filaments tied to the latter, as nearly
alike as possible. The stem of one bulb is provided with an aluminium
tube, the stem of the other has none. Originally the two bulbs were