remained alive, and this proceeded to develop that particular part of the
embryo to which it would have given rise if the rest of the egg had not
been killed. There was no regeneration of the part killed, no formation of
a complete embryo. It may be pointed out that segmentation in the insect
egg is peculiar. The nuclei multiplied by segmentation migrate into the
superficial cytoplasm surrounding the yolk, and then this cytoplasm
segments, and each part of the cytoplasm develops into a particular region
of the embryo. This, of course, does not prove that the nuclei or their
chromosomes do not determine the _characters_ of the parts of the embryo
developed, but they show that the parts of the non-nucleated cytoplasm
correspond to particular parts of the embryo. The most important object of
investigation at the present time is to find the origin of these
properties of the chromosomes. We may say, using the word 'determinant' as
a convenient term for that which determines the adult characters, that in
order to explain the origin of species or the origin of adaptations we
must discover the origin of determinants. Mendelism does not throw any
direct light on this question, but it certainly has shown how characters
may be inherited as separate and independent units. When one difference
between two breeds is considered, _e.g._ rose comb and single in fowls,
and individuals are crossed, we have the determinant for rose and the
determinant for single in the same zygote. The result is that rose
develops and single is not apparent. In the next generation rose and
single appear, as at the beginning, in separate individuals. When two or
three or more differences are studied we find that they are usually
inherited separately without connexion with each other, although in some
cases they are connected or coupled. The facts of Mendelism are of great
interest and importance, but we have to consider the general theory based
on them. This theory is that characters are generally separate units which
can exist side by side, but do not mingle, and cannot be divided into
parts. When an apparently single character shows itself double or treble,