The Origin of Chrysanthemum Segctwn Plenum. 177 
explanation suggested by the double curve has thus been 
fully substantiated by the result of selection. On the 
other hand it is perfectly plain that the dimorphic curve 
is not simply the sum of the two monomorphic ones. 
The mixed assemblage does not simply contain the two 
mixed races, either in equal parts or in any other pro- 
portion. It cannot be synthesized from its components. 
This is proved by two circumstances : on the one hand 
by those parts of the curve that lie outside the maximum 
ordinates, on the other by the middle part. The two 
component curves begin at 7 and end at 28 (32) and 
their sum should do so too. But the curve of the mixed 
race is limited by 11 and 23. This is seen more clearly 
by looking at the ordinates 12 and 22, since there are 
far too few individuals in these in Fig. 30. Thus we 
see that the limits of the curves are, so to speak, "drawn 
in" in the mixture. On the contrary the individuals are 
heaped up between the two apices. Moreover in this 
part there is a secondary maximum. This is seen at 17, 
but in the commercial mixture of 1895 falls on 16 1 ac- 
cording to the figures given above (see p. 172). 
We come now to the double race. It is a well-known 
saying amongst horticulturists, that any one who wishes 
to obtain novelties must be eagerly on the lookout for 
small differences (See Vol. I. Part I, p. 185, and this 
volume, 2, p. 9). If these deviations are not cases of 
fluctuating- variabilitv but strike the eve by the fact that 
fJ * m ^ 
they are much rarer than these, it is probable that they 
are the external manifestations of semi-latent characters. 
If this is actually the case it is further probable that the 
character can be brought by isolation and selection to 
1 16 (= 3 -f- 5 -j- 8) is one of the subsidiary numbers in LUD- 
WIG'S law. The question arises whether by the crossing- of two pure 
races these subsidiary numbers may arise elsewhere also. 
