rollers through projections, J. The two systems are made to approach
or recede from each other, in order to form the arc, by means of a
regulating screw, H.

At present, the lighting of these lamps is effected by means of this
screw, H, but Mr. Solignac is now constructing a model in which the
lighting will be performed automatically by means of a solenoid that
will react upon a carbon lighter, as in several already well known
systems.

[Illustration: Fig. 1]

If the preceding description has been well-understood, it will be seen
that the carbons are arrested in their movement toward each other only
by the glass rods, G, abutting against L; but, as the stops, L, are not
far from the arc, and as the heat to which they are exposed is so much
the greater in proportion as the incandescent part of the carbons is
nearer them, it results that for a certain elongation of the arc the
temperature becomes sufficient to soften the glass of the rods, G, G,
so that they bend as shown at O (Fig. 3), and allow the carbons to move
onward until the heat has sufficiently diminished to prevent any further
softening of the glass. In measure as the wearing away progresses, the
preceding effects are reproduced; and, as these are produced in an
imperceptible and continuous manner, there is perceived no jumping nor
inconstancy in the light of the arc. Under such conditions, then, the
regulation of the arc is effected under the very influence of the
effect produced; and not under that of an action of a different nature
(electro-magnetism), as happens in other regulators. It is certain that
this idea is new and original, and the results that we have witnessed
from it have been very satisfactory. There is but one regulation to