THE EXPERIMENTAL STUDY OF GENETIC RELATIONSHIPS 1 45 
which go to show that the chromosomes are themselves the mechanism 
by which MendeHan characters are transmitted. It is only necessary 
to call to mind Gregory's discovery that the Mendelian factors are 
doubled in tetraploid plants of Primula. If the mechanism is altered, 
the genetic qualities of the plant must of necessity be altered. For 
this reason I am disposed to lay especial emphasis upon those mutations 
in which a cytological change in the cell can be demonstrated, not, 
however, without stating my belief that mutations for which there is 
no visible basis are quite as independent of ancestral hybridization as 
the gi gas -muta.tions. In certain groups, as the Coniferae for example, 
there is great uniformity in the chromosome number, showing that 
evolution has taken place through invisible modifications of the germ 
plasm. Probably most of the mutations which take place in any group 
of plants, even Oenothera, have the same chromosome number as the 
parent. 
My Oenothera cultures of the last three years have given many 
mutations, aside from the gigas forms, which confirm in all essential 
points de Vries's experience with Oe. Larmarckiana. Some of them I 
regard as progressive, although it has not been possible yet to demon- 
strate that they are dominant in a Mendelian sense. In fact, Mendel- 
ian inheritance seems to play so small a part in Oenothera that in 
general we cannot expect to apply the test of dominance in judging of 
the progressiveness of mutations. We must consider a mutation as 
progressive when it shows characters which are not present in the 
parent. 
It has been shown by de Vries that reciprocal hybrids between 
Oenothera species are frequently very unlike one another. Both de 
Vries and Davis have encountered cases in which the hybrids are 
strongly patroclinic. I have just the opposite experience with some 
of my interspecific hybrids, which are strikingly matroclinic. It is 
clear that in this genus reciprocal hybrids may be either matroclinic 
or patroclinic. In either case it is impossible to say that the characters 
of one parent or the other are dominant in the ordinary sense. With 
this explanation I shall proceed to a very brief discussion of two of my 
new mutations. 
In Oenothera pratincola the seedling leaves are ovate. Seven 
different strains of this species have given rise to a mutation with 
round seedling leaves, which I have called mut. nummularia. In 
three strains the mutation has appeared in three successive genera- 
