Theory of Plant Breeding 
another suggested explanation. Probably the original 
seed colour was green, and yellow has been formed 
because the plastids in certain more advanced races have 
stopped at the yellow stage of development, and not 
gone on to the formation of green. This yellow is then 
a progressive step in the evolution of a species, and 
would be, according to Davenport’s view, likely to be 
dominant."* But according to Emerson (U.S.A. Exp. 
St., Nebraska, 1902), one also finds in racial hybrids 
of kidney beans both new colours, and others which 
seem to result from some very distant ancestor (atavistic). 
One very common theory amongst practical breeders, 
viz., that the pollen parent decides the flower colour of its 
descendants, seems not to be substantiated by experiment. 
Chittenden found that of 183 hybrids, 42 were of 
the male colour, 46 of the female; 92 were intermediate, 
and 3 were unlike both ancestors.® But although, as 
the reader will have realised already, Mendel’s results will 
be of great service in practice, they must not be considered 
as all-important. Macfarlane has clearly shown that cer- 
tain hybrids do not follow Mendelian laws, but form a 
series of transitions between the colours of their parents. 
So far as Mendel’s laws bear upon evolution they 
bring out clearly two facts—first, that such characters 
as colour are clearly inherited, which has an important 
bearing upon the inheritance of acquired characters ; 
and, secondly, that it is unlikely that hybridising can 
have played any important part in the formation of new 
species. The tendency in hybrids seems to be for them 
to work back to the original parent types. 
But even these conclusions are somewhat uncertain, 
The genus Iris has three stamens only, and one would 
think that this was an absolutely fixed character. Some 
very distant ancestor no doubt had six stamens, for that 
is almost universal in all allied orders. But Heinricher 
discovered occasional extra stamens in about 18 per 
3° U 
