The Theory of Evolution 219 
polyhedron may roll over on to another face, which would corre- 
spond to the change of a species into a new one. Should the 
polyhedron be so constructed that one of the new faces is much 
more stable, or easily reached, we can understand why in some 
cases only a single mutation is likely to occur when the balance is 
upset. Such would be the case with the turnspit type of dog, 
the Japanese peacock, the merino sheep, the short-horned cattle. 
If, on the other hand, one new face is as likely as the other to 
give a stable condition, results like those in the primroses would 
seem to be produced. We might even go farther and conclude 
that this has been the case with the two hundred elementary 
species of Draba verna, with the many elementary species of the 
wild apple, and possibly with the races of domesticated pigeons. 
The analogy need not be pushed too far, nor be taken for more 
than it is worth. 
De Vries attempts to draw a sharp line between elementary 
species and varieties, and while the distinction is useful, yct, on 
the whole, it seems rather arbitrary. A variety is a type in 
which one of the parent-species characters has disappeared — 
not necessarily absolutely, for it may remain only latent in the 
germ-cells, but it never develops or only under unusual condi- 
tions. Asan example, an albino mouse is a variety of the house 
mouse. It is characterized by the loss of a character — the 
pigment. Nevertheless as we have seen in the preceding chap- 
ters the character that disappears may really be only latent in 
the germ-cells. 
It is evident that it may often be convenient to distinguish 
between new types in which a new character or combination of 
characters has appeared, and new types distinguished only 
by the loss of a character. The distinction may seem to be one 
of only minor importance were it not that it has been held to 
separate new types that obey quite different laws of heredity. 
It is sometimes said that varieties, in de Vries’s sense, follow 
Mendel’s law, while elementary species do not. De Vries has 
indicated how he supposes this difference might come about. 
1 Hybridization often brings such latent characters once more to the front. 
