280 BOTANICAL GAZETTE [APRIL 
Two later papers by HILpEBRAND (19, 20) deal with trimor- 
phous species of Oxalis. He found in Oxalis Valdiviana (20, p. 43) 
that seeds from any one of the six possible legitimate unions pro- 
duced all three forms, but that the two parent forms greatly pre- 
dominated. He also states that long-styled plants of Oxalis rosea, 
growing by themselves, have always produced long-styled plants 
(‘Jahr aus Jahr ein dieselbe Form entsteht’’). 
In 1864, JoHN Scott published a paper (32) in which he arranges 
all the known species of Primula in four groups: dimorphic, short- 
styled, long-styled, and non-dimorphic (homostyled), and describes 
experiments with 7 dimorphic species. He pollinated each of these 
7 species both legitimately and illegitimately and counted the seeds 
obtained by each method. The result was uniformly that legiti- 
mate unions produced a markedly greater number of seeds than 
illegitimate. 
Fritz Mier (27), in a brief paper dealing with a trimorphous 
species of Pontederia growing in Brazil, mentions the fact that in 
Oxalis Regnelli, another trimorphous species, the seeds of long- 
styled plants, legitimately fertilized with pollen from the longest 
stamens of the mid-styled form, produced plants which belonged 
exclusively to the two parent forms. 
In his Forms of flowers, DARWIN includes all that had been done 
on heterostylous plants up to the time of its publication. He cites 
38 genera known to include heterostyled species. These genera 
are distributed as follows (11, p. 254): Hypericineae 1, Erythroxy- 
leae 2, Geraniaceae 2, Lythraceae 2, Rubiaceae 17, Primulaceae 3, 
Oleaceae 1, Gentianaceae 3, Polemoniaceae 1, Cordieae 1, Bora- 
gineae 1, Verbenaceae 1, Polygoneae 1, Thymeleae 1, and Ponte- 
deriaceae 1. The wide geographical distribution of the genera 
which contain heterostyled species and the fact that the families 
to which they belong are mostly very distinct from one another, 
indicate that heterostyly has arisen independently in several 
phylogenetic lines. 
DaRWIN considers (p. 245) that the morphological differences 
between the forms of a heterostyled species are confined to the 
flower. His observations on this point may be summed up briefly | 
as follows: In the calyx there are no differences. The corolla 
