204 BOTANICAL GAZETTE [MARCH 
from a carpellate plant. The staminate flower has four distinct 
sepals and the normal gynoecium has two stigmas. In some cases 
neatly typical carpellate flowers appear on staminate plants 
(fig. 3), and carpellate plants often bear normal staminate flowers 
toward the end of their life. A common abnormality is the develop- 
ment of a typical stigma or one of varying degrees of perfection 
at the outer end of an anther (figs. 4, 7, 9a, also 6, 8, 11). Occasion- 
ally a stamen grows directly out of the side of an ovulary (fig. 5), 
or the ovulary may have more or less perfectly developed microspo- 
rangia on one side. Fig. 10 represents such a structure from the 
tip of the inflorescence of a carpellate plant. This flower has the 
characteristic staminate perianth, although in other respects it is 
more carpellate in nature. Figs. 6 and 7 represent two types of 
abnormal flowers, both from a carpellate plant. Fig. 6 has one 
good and nearly normal stamen. Fig. 8 is a stamen-carpel complex 
from the center of a staminate flower with four normal stamens, 
taken from a staminate plant. Apparently an attempt was made 
to develop a normal bicarpellary gynoecium. The one side has 
an imperfect ovulary with a normal stigma, while the other has a 
distorted anther with two microsporangia ending in an imperfect 
stigma. Figs. 9 and oa represent a flower from a staminate plant 
with an imperfectly developed gynoecium in the center. The 
separated stamen (fig. 9@) has a well developed stigma, and one | 
other stamen has ‘a rudimentary structure at the tip which shows 
a slight development in the direction of a stigma. Fig. 11 represents 
a staminate flower from a carpellate plant with three stamens on a 
central stalk, the normal position for the gynoecium, two of the 
stamens having rudimentary stigmas. Fig. 12 is an imperfectly 
bisporangiate flower from a staminate plant. As intimated, a 
great number of such variations and abnormalities of every con- 
ceivable diversity appear on both staminate and carpellate plants. 
They are readily intelligible on the theory that they are caused by 
varying and reversible sexual states. Any attempt to bring such 
phenomena within the limits of discrete Mendelian units, segregated 
and combined normally or abnormally during the reduction and 
fertilization stages, would appear extremely absurd to the writer. 
