May, 1923] 
SHULL-SPECIES CONCEPT 
22 3 
what doubtless seem to the taxonomist rather shocking liberties with the 
taxonomic species. 
A simple example from my own work will illustrate how taxonomic 
distinctions may become useless to the geneticist. For a number of years 
I bred the dioecious forms of Lychnis (Melandrium) quite extensively, 
growing many hundreds of pedigreed families from seeds secured from 
various parts of Europe and America. I found that nearly every lot of 
seeds produced a progeny perceptibly different from those produced from 
other lots of seeds; but the differences, which were in some cases sharp and 
easily stated in precise terms, were in other cases difficult of exact descrip¬ 
tion. All these forms—of whatever name and wherever collected—bred 
together without any diminution in fertility—in fact, usually with increased 
fecundity; and, although there were numerous differentiating hereditary 
characters, these were not grouped in the wild forms in such manner as to 
make it profitable or convenient to attempt to keep track of specific dis¬ 
tinctions, as such, in the experimental garden. In my published accounts 
of these experiments the Linnaean designation, Lychnis dioica, was the 
only one which could be usefully employed, the English forms L. diurna 
and L. vespertina and the German forms Melandrium album and M. rubrum 
being incapable of maintenance. 
Since the condition which here made the minor distinctions between 
taxonomic species useless was the perfect freedom with which all forms 
interbred, and the independence with which the hereditary characteristics 
were distributed, Koelreuter’s criterion of specific difference receives support, 
and it might be assumed that under experimental conditions Koelreuter’s 
method of delimiting species on the basis of compatibility or incompatibility 
—fertility vs. sterility—would have general genetical validity, except in 
the cases already mentioned in which vegetative or parthenogenetic methods 
of breeding occur; but any attempt to make extensive use of this idea 
promptly meets difficulties of most formidable dimensions on account of 
the numerous kinds and degrees of incompatibility and sterility which are 
met with. 
Sterility may result from the presence of a single unit factor, of no more 
fundamental nature than the factor which changes a white flower to a 
colored one, or vice versa , in any one of a large number of species which 
might be mentioned. In such a case only a single gene differentiates the 
two forms, all the rest of the complex organization of the genotype being 
identical. On the other hand, incompatibility and sterility may result 
from general dissimilarities in organization, involving, conceivably, in¬ 
numerable details of genotypic structure. I may cite several examples 
from my experiments with Bursa to illustrate the difficulties attendant 
upon any attempt to utilize incompatibilities and sterilities as general criteria 
of specific differences. 
For fifteen years I have been working extensively wdth the common 
