420 Tricotylous Races. 
selection of the extreme variants of ordinary fluctuating 
variability. Even the doctrine of unilateral increase in 
variability as a result of selection is of no help in this 
case (see 2, p. 9), for selection could hardly operate 
so rapidly as to produce its whole effect in a single gen- 
eration. The old saying of gardeners that the first con- 
dition necessary for the production of a novelty is to 
possess it already, also applies to these purely experi- 
mental races (Vol. I, p. 185 and elsewhere). If the tri- 
cotylous race does not already exist it cannot, at present 
at any rate, be either isolated or bred. 
A high percentage in tricotyls is seldom found in 
wild species. The highest value I have yet found oc- 
curred in Linaria vulgaris in the spring of 1894 in a sow- 
ing of the seed of a hemipeloric plant of the race that 
I was cultivating at that time (see p. 211). There were 
59 tricotyls amongst the 425 seedlings, i. e., 14%. Amongst 
commercial seed the prospect of obtaining intermediate 
races seems to me to be the greatest, as I have already 
stated, in those sorts which are cultivated on a large scale 
in the field or in the garden. It is much smaller in those 
varieties of flowers which are only grown on a small 
scale every year. Moreover it seems obvious that cul- 
tivation on a large scale should favor the origin of new 
races. 
If the intermediate race, which is being sought for, 
exists in some sample of seed, we may expect to find 
mean, bettei or inferior representatives of it. If the 
former is the case the mean character of the race, that 
is about 50 to 60% of tricotyly is attained at once, and 
this occurs in the majority of cases, as might be expected 
and as the table at the conclusion of this section will 
show (p. 439). Individuals with a higher productive 
