474 
Saunders.—A Reversionary Character 
hitherto unexplained phenomena exhibited by such forms as Ceratocapnos 
( Corydalis) heterocarpa , Ball. (Figs. 52, 53), among the Fumariaceae, and 
Aethionema heterocarpum , J. Gay, and Diptychocarpus stvictus , Trautv., 
among the Cruciferae. These plants are peculiar in that they appear 
habitually to produce two kinds of fruit in the same inflorescence. Those 
first formed are short, thick, indehiscent, composed of four similar (hollow) 
carpels ; those which succeed them are more elongated, slender, dehiscent, 
two of the carpels becoming detached as valves from the other two, which 
have undergone a partial transformation towards the solid type. The 
significance of this heterocarpism is now fully apparent. We have exhibited 
in these plants as a fairly constant feature a transition comparable with that 
observed occasionally in Eschscholzia (see above, p. 471). The process of 
consolidation is taking place before our eyes ! 
It is a striking fact that in these heterocarpic species among the 
Cruciferae and Fumariaceae it is the ancestral type of fruit composed of 
uniform hollow carpels which does not dehisce, whereas the later-evolved 
form in which some carpeL have become solid is able to split. Can it be 
that differentiation into valve and solid carpels provided an easier and more 
reliable method of securing the opening of the fruit than rupture of the fruit 
wall at a spot which, though representing a junction of (valve) carpels, 
perhaps offered insufficient tissue differentiation to secure invariably the 
initiation of a split? Or has this modification of some of the carpels come 
about in the course of, and in association with, evolution from a primitive 
one—or few—seeded fruit in which dehiscence was unnecessary and perhaps 
did not occur, to a many-seeded condition in which it was essential ? These 
questions we can scarcely hope to answer until a more exhaustive study of 
the structure of the gynoecium in these families has been carried out. 
It has now been sufficiently made clear that in certain families of the 
Rhoeadales evolution has been accompanied by processes of reduction and 
consolidation of members of the gynoecium, leading to the production of 
two kinds of carpel, the hollow and the solid, and that in the dry dehiscent 
type of fruit consolidation has sometimes occurred without (apparently) 
a reduction in number, and has led in these cases to the formation of the 
compound (many-carpelled) valve. We can henceforth eliminate from the 
scheme of construction the false partition and the commissural stigma. 
Furthermore, we obtain a rational explanation of the hitherto unexplained 
occurrence of dimorphic fruits which are met with in many types, either as 
a rarity, or so constantly as to have been regarded at the outset as of generic 
value. In them we see the reappearance momentarily or for a longer space 
of some stage of the phylogenetic history. 
