again the inference perfectly
corresponds with the facts. While all the planets go round the sun from
west to east, comets as often go round the sun from east to west as from
west to east. Of 262 comets recorded since 1680, 130 are direct, and 132
are retrograde. This equality is what the law of probabilities would
indicate.

Then, in the fourth place, the physical constitution of comets accords
with the hypothesis.[15] The ability of nebulous matter to concentrate
into a concrete form, depends on its mass. To bring its ultimate atoms
into that proximity requisite for chemical union--requisite, that is,
for the production of denser matter--their repulsion must be overcome.
The only force antagonistic to their repulsion, is their mutual
gravitation. That their mutual gravitation may generate a pressure and
temperature of sufficient intensity, there must be an enormous
accumulation of them; and even then the approximation can slowly go on
only as fast as the evolved heat escapes. But where the quantity of
atoms is small, and therefore the force of mutual gravitation small,
there will be nothing to coerce the atoms into union. Whence we infer
that these detached fragments of nebulous matter will continue in
their original state. Non-periodic comets seem to do so.

We have already seen that this view of the origin of comets harmonizes
with the characters of their orbits; but the evidence hence derived is
much stronger than was indicated. The great majority of cometary orbits
are classed as parabolic; and it is ordinarily inferred that they are
visitors from remote space, and will never return. But are they rightly
classed as parabolic? Observations on a comet moving in an extremely
eccentric ellipse, which are possible only when it is comparatively near
perihelion, must fail to distinguish its orbit from a parabola.
Evidently, then, it is not safe to class it as a parabola because of
inability to detect the elements of an ellipse. But if extreme
eccentricity of an orbit necessitates such inability, it seems quite
possible that comets have no other orbits than elliptic ones. Though
five or six are said to be hyperbolic, yet, as I learn from one who has
paid special attention to comets, "no such orbit has, I believe, been
computed for a well-observed comet." Hence the probability that all the
orbits are ellipses is overwhelming. Ellipses and hyperbolas have
countless varieties of forms, but there is only one form of parabola