rges from the carriers, which portions are conveyed to the
inductors. The combs collect the remaining portion of the induced
charge for use as an outer circuit, while the metal rod with its two
brushes neutralizes the plate surface in a line of its diagonal
diameter. When at work it supplies a considerable amount of
electricity. It is self-exciting in ordinary dry atmosphere. It freely
parts with its electricity from either terminal, but when so used the
current frequently changes its direction, hence there is no certainty
that a full charge has been obtained, nor whether the charge is of
positive or negative electricity.

I next come to the type of machine with which I am more closely
associated, and I may preface my remarks by adding that the invention
sprang solely from my experience gained by constantly using and
experimenting with the many electrical machines which I possessed. It
was from these I formed a working hypothesis which led me to make my
first small machine. It excited itself when new with the first
revolution. It so fully satisfied me with its performance that I had
four others made, the first of which I presented to this Institution.
Its construction is of a simple character. The two disks of glass
revolve near to each other and in opposite directions. Each disk
carries metallic sectors; each disk has its two brushes supported by
metal rods, the rods to the two plates forming an angle of 90 deg.
with each other. The external circuit is independent of the brushes,
and is formed by the combs and terminals.

[Illustration: Fig. 10.]

The machine is self-exciting under all conditions of atmosphere, owing
probably to each plate being influenced by and influencing in turn its
neighbor, hence there is the minimum surface for leakage. When
excited, the direction of the current never changes; this circumstance
is due, probably, to the circuit of the metallic sectors and the make
and break contacts always being closed, while the combs and the
external circuit are supplemental, and for external use only. The
quantity of electricity is very large and the potential high. When
suitably arranged, the length of spark produced is equal to nearly the
radius of the disk. I have made them from 2 in. to 7 ft. in diameter,
with equally satisfactory results. The diagram, Fig. 9, shows the
distribution of the electricity upon the plate surfaces when the
machine is fully excited. The inner circle of signs corresponds with