282 BOTANICAL GAZETTE [APRIL 
distinctly useful or harmful to the organism, but not before the 
structure is present. 
DaRWIN recognized such linkage, calling it correlative varia- 
bility. He accounted for many useless characters by assuming their 
linkage with other factors which are sufficiently useful to have a 
survival value. It has already been shown, however, that there is 
great difficulty in definitely pointing out that certain characters 
are actually acted upon in the environment by natural selection 
on a consistent basis. In developmental selection there is no doubt 
about a definite selection taking place, and the experimental 
evidence that certain adult sporophyte characters are linked with 
the gametophytic or embryonic factors having definite values in 
selection is also available. 
This will also explain the production of overdeveloped structures, 
or those having no advantage to the organism. Suppose the 
factors determining the length or shape of the spines in the ends of 
some pine cone scales, or other useless details of the plant, should 
become linked with the genes producing either the most vigorous 
pollen tubes or embryos, there is no doubt that the selection for the 
pollen tubes or the embryos would result in the selection of the other 
characters in the same linkage. The spines would then be selected 
in or out, as the case may be, by developmental selection. Details 
of variations or mutations, be they ever so small, may either tend 
to disappear or become fixed or overdeveloped, as they are affected 
by this internal selective mechanism. Such phenomena of deter- 
minate variations and evolution in definite directions have long 
been recognized, and have been attributed to a principle called 
orthogenesis. The mechanism of developmental selection described 
here looks very promising as a means of ssabaelea for many 
or thogenetic phenomena. 
It is probably needless to point out that many characters would 
not be linked with these factors involved in developmental selection 
at all. These would remain unaffected by the developmental selec- 
tion processes. Among these most of the heritable characters that 
have been studied by the Neo-Mendelians are to be found. The 
genes of these characters segregate independently of the factors 
affected by this selective mechanism, and yield Mendelian ratios in 
the expected proportions. Among those which should probably be 
