6d or 7d to the _north_. As this was contrary to the _direction_ of
the curvature, if the tail had been curved, it could only arise from a
portion being driven off by the radial stream, or bent towards the plane
of the ecliptic. The curvature observed by others at a later date, was
concave to the south. Towards the middle and close of March, the tail
became straight, and with the above exception, might be considered to
move in the plane of the orbit.

The celebrated comet of Halley, as observed by Dr. Bessel in 1835,
showed that a more or less well-defined tuft of rays emanated from that
part of the nucleus which was turned towards the sun; and the rays being
_bent backward_ formed a part of the tail. The nucleus, with its
emanations, presented the appearance of a burning rocket, the end of
which was turned sideways by the force of the wind. And, Bessel
concludes: "That the cone of light issuing from the comet deviated
considerably both to the right and left of the true direction of the
sun, but that it always returned to that direction, and passed over to
the opposite side; so that the cone of light, and the body of the comet
from whence it emanated, experienced a rotatory, or, rather, a vibrating
motion _in the plane of the orbit_." It is impossible that Bessel should
here mean that this motion was certainly in the plane of the orbit; for
the orbit was then viewed sideways, and he had no means of ascertaining
the fact. His meaning must be that it was apparently in the plane of the
orbit. If a plane be made to pass through the earth, the comet, and the
sun, the tail might be placed in any position in that plane, and yet
appear to be at the intersection of the two; that is, in the plane of
the comet's orbit. The vibration of the tail, in this case, is another
strong proof of the correctness of our theory. To make it more
intelligible, we shall resort to a diagram.

In the following diagram, the comet's orbit, represented by the dotted
line, is drawn on the plane of the ecliptic; it is, therefore, necessary
to bear in mind, that it is tilted up from the line of nodes SN, at an
angle of 17d 45'. The position of the comet, October 9th, is at C,
approaching its perihelion; that of the earth at the same time at T;
while S represents the sun, and SQ the line of equinoxes. Now, from a
cause already explained, the tail always tends to lay behind the comet,
in the direction indicated by the lower tail in the diagram at 1, and,