Atavistic Races. 481 
the several generations of this race with a true syncotyl- 
ous race (see the, table on page 470), we only find in most 
of them an oscillation within much wider limits. 
The hemi-syncotyls are, therefore, only variants of 
the syncotylous race. 
12. ATAVISTIC RACES. 
Although the atavistic specimens of Helianthus an- 
nuus syncotyleus are exactly similar to the normal seed- 
lings of the ordinary sunflower, they nevertheless belong 
to the syncotylous race and do not depart from it. With 
regard to their visible characters they are aberrant forms; 
with regard to their progeny, however, simply extreme 
minus variants ; but the latter is true only on the average, 
and not for the particular atavists whose values often 
approach those of the best syncotyls, and not infrequently 
exceed the mean of the race. 
If the intermediate race were not pure we should, of 
course, still be able to select the corresponding half race 
from it. But after it has once been purified by selection, 
this is no longer possible. A syncotylous intermediate 
race can no more give rise to a half race than a tricotylous 
half race can be transformed by selection into an inter- 
mediate race (see Amarantus and Scrophularia, pp. 398 
and 407). In the years 1890, 1891, 1892 and 1894, I 
made extensive cultures of atavists derived from my 
syncotylous race and also from my hemi-syncotylous 
race, and determined their hereditary values for one or 
two generations. In these experiments curves were ob- 
tained which were not asymmetrical, with a peak at one 
end, but were flattened curves, extending over the whole 
length of the base line. Thus, for instance, I obtained 
from a dozen atavists from the seeds of the seed-parent 
