Herschel found this rule to be invariable. Thus every time
that during a short interval no star approached in virtue of the
diurnal motion, to place itself in the field of his motionless
telescope, he was accustomed to say to the secretary who assisted
him,--'Prepare to write; nebulae are about to arrive.'"

How does this fact consist with the hypothesis that nebulae are remote
galaxies? If there were but one nebula, it would be a curious
coincidence were this one nebula so placed in the distant regions of
space, as to agree in direction with a starless spot in our own sidereal
system. If there were but two nebulae, and both were so placed, the
coincidence would be excessively strange. What, then, shall we say on
finding that there are thousands of nebulae so placed? Shall we believe
that in thousands of cases these far-removed galaxies happen to agree in
their visible positions with the thin places in our own galaxy? Such a
belief is impossible.

Still more manifest does the impossibility of it become when we consider
the general distribution of nebulae. Besides again showing itself in the
fact that "the poorest regions in stars are near the richest in nebulae,"
the law above specified applies to the heavens as a whole. In that zone
of celestial space where stars are excessively abundant, nebulae are
rare; while in the two opposite celestial spaces that are furthest
removed from this zone, nebulae are abundant. Scarcely any nebulae lie
near the galactic circle (or plane of the Milky Way); and the great
mass of them lie round the galactic poles. Can this also be mere
coincidence? When to the fact that the general mass of nebulae are
antithetical in position to the general mass of stars, we add the fact
that local regions of nebulae are regions where stars are scarce, and the
further fact that single nebulae are habitually found in comparatively
starless spots; does not the proof of a physical connexion become
overwhelming? Should it not require an infinity of evidence to show that
nebulae are not parts of our sidereal system? Let us see whether any such
infinity of evidence is assignable. Let us see whether there is even a
single alleged proof which will bear examination.

"As seen through colossal telescopes," says Humboldt, "the
contemplation of these nebulous masses leads us into regions from
whence a ray of light, according to an assumption not wholly
improbable, requires m