154 
H. H. BARTLETT 
certain negative conclusions which may be helpful may be made. In 
the first place, we must conclude that the degree of sterility which 
follows hybridization can not be used as an index of relationship. 
The swarm of 200 elementary species which are included under 
Erophila verna all differ from one another in relatively trivial characters. 
Rosen has shown that some of the hybrids among them are fully sterile 
and only one pair of the species which he tried gave fully fertile 
hybrids. Yet we can not doubt that all are closely related genetically. 
In the Onagra group of Oenothera there are many species which cross 
more readily with Oenothera Lamarckiana than the latter does with its 
mutation Oe. gigas. On the whole, we may say that among closely 
related forms neither interspecific sterility nor the lack of it is a true 
guide to the degree of relationship. ' 
The origin of a form by hybridization should not be inferred from 
likenesses to both of the supposed parents, nor should a high degree of 
sterility be interpreted as a sign of hybridization. On the other hand, 
true interspecific hybrids in some cases show almost no influence of 
one or the other parent, and are as fertile as, or more fertile than, either 
parent. 
The extreme morphological dissimilarity between some mutations 
and their parent species must teach us that little reliance can be 
placed upon the guesses of systematists regarding relationships in 
polymorphic groups. The herbarium botanist should fully realize the 
fact that his schemes of classification, in Ruhus and Crataegus for 
example, are probably entirely artificial and do not represent natural 
relationships at all. Very important systematic characters may 
originate repeatedly and independently in unrelated lines of descent. 
There is no test of what constitutes a species, except that it shall 
reproduce itself from generation to generation. Systematists should 
have a pragmatic attitude in describing species. Subdivision should 
extend as far as any one finds necessary. The geneticist needs to have 
definite designations for much smaller groups than the ecologist or 
morphologist is likely to be interested in. The makers of manuals 
should therefore endeavor to produce books which will supply the 
needs of either class of workers without misrepresenting facts. If it is 
necessary to have a more simple treatment of Crataegus than that of 
Sargent, for example, it may be done in such a way as not to discredit 
a large amount of careful work. The synonymy which is so conspicu- 
ous a feature of systematic work should be given a different significance. 
