Tricotylous Races Do Not Arise by Selection. 415 
412), successful throughout. Nevertheless an interme- 
diate race has not arisen, either gradually or by means 
of a sudden jump. The progress in the last two years was 
more rapid than before as the result of increased strin- 
gency of selection, without, however, affording any in- 
dication that the mean of 50 /t was likely soon to be 
reached. 
Oenothera Berteriana. Besides the two cultures men- 
tioned which were begun in the first years of my ex- 
periments in selection, I have cultivated yet a third race 
with the same object. This race was one in which the 
intercrossing of the various individuals could always be 
avoided. There is, however, no ground for fear that 
occasional unavoidable crosses in Amarantus and Scro- 
phularia had any considerable effect on the selection pro- 
cess ; for both species must be fertilized almost entirely 
with their own pollen on acount of the great number of 
their flowers which are open at the same time ; and they 
are, when thus fertilized, perfectly fertile as isolated in- 
dividuals show. Moreover what is spoiled by crossing 
is eliminated bv selection. 
ff 
But the evidence is more satisfactory if self-fertiliza- 
tion can be insured. This occurs in Oenothera Berteriana. 
Its flowers form perfectly normal fruits and seeds, when 
the visits of insects are excluded. I enclosed my plants 
in a cage of fine metal gauze. Some years I have fer- 
tilized them artificially in it ; but this is quite superfluous, 
because when the flower withers the stigma bends down- 
wards and thus reaches the pollen. In the two last sum- 
mers the cage was shut from the beginning of the flow- 
ering period until the seed began to ripen. Nevertheless 
they all produced fruits with scarcely an exception. These 
fruits contained an abundance of seed, and a few from 
