148 THE THEORY OF MUTATION 
from the parental type, yet their first appearance has 
nothing to do with normal or continuous variability. 
6. The mutations take place indefinitely, showing 
no special tendency in any particular direction. 
7. The tendency to mutate recurs periodically. But, 
as was previously stated, there is no direct evidence 
of this last supposition. 
In addition to what has already been said with 
reference to the method of origin of garden varieties in 
general, de Vries has described a number of special 
phenomena regarding the behaviour of garden varieties 
of plants, some of which are of considerable interest. 
Taken together, the facts substantiate to a great extent 
the view that selection does not of itself lead to the 
production of specific characters. But de Vries also 
introduces certain new conceptions which require to 
be briefly described on account of their great general 
interest to practical breeders and gardeners. They 
consist in the idea of races existing intermediate be- 
tween a species and a complete variety or sub-type of 
it. Such between-races are of two kinds, of which it is 
unusual to find both in the case of the same species ; 
moreover, either of them may occur even when the 
complete variety is quite unknown. In the case of a 
haljf-race a small percentage only of seedlings is found 
to produce plants which show the racial character, the 
remainder being of the original specific type; and even 
if the racial type is selected for several generations, the 
percentage of plants of this type which is produced 
does not notably increase. A mid-race, on the other 
hand, can readily be improved by selection, and when 
