ubt the view you state, and which Gulick and Romanes make
much of, that a portion of a species, separated from the main body, will
have a different average of characters, unless they are a local race
which has already been somewhat selected. The large amount of variation,
and the regularity of the curve of variation, whenever about 50 or 100
individuals are measured in the same locality, shows that the bulk of a
species are similar in amount of variation everywhere. But when a
portion of a species begins to be modified in adaptation to new
conditions, distinction of some kind is essential, and therefore any
slight difference would be increased by selection. I see no reason to
believe that species (usually) have been isolated first and modified
afterwards, but rather that new species usually arise from species which
have a wide range, and in different areas need somewhat different
characters and habits. Then _distinctness_ arises both by adaptation and
by development of recognition marks to minimise intercrossing.

I wonder Darwin did not see that if the unknown "constant causes" he
supposes can modify all the individuals of a species, either
indifferently, usefully, or hurtfully, and that these characters so
produced are, as Romanes says, very, very numerous in all species, and
are sometimes the only specific characters, then the Neo-Lamarckians are
quite right in putting Natural Selection as a very secondary and
subordinate influence, since all it has to do is to weed out the hurtful
variations.

Of course, if a species with warning colours were, in part, completely
isolated, and its colours or markings were accidentally different from
the parent form, whatever set of markings and colours it had would be, I
consider, rendered stable for recognition, and also for protection,
since if it varied too much the young birds and other enemies would take
a heavier toll in learning it was uneatable. It might then be said that
the character by which this species differs from the parent species is a
useless character. But surely this is not what is usually meant by a
"useless character." This is highly useful in itself, though the
difference from the other species is not useful. If they were in contact
it would be useful, as a distinction preventing intercrossing, and so
long as they are not brought together we cannot really tell if it is a
species at all, since it might breed freely with the parent form and
thus return back to o