456 Saunders. — A Reversionary Character 
plan is never realized, two members being always suppressed (Chodat). 
And so G 2 it has remained. Arguments in favour of a 4-membered whorl 
have, it is true, been advanced by Schmitz, 1 but they are urged in favour of 
a schematic rather than an actual ground plan : G 4 stands, not for the 
siliqua as it is, but for an abstraction, the concern of the author being to 
realize a common generalized diagram for all Dicotyledons (!). Somewhat 
later Kernel*, 2 regarding the placental commissures as arising after the 
valves and forming a second inner whorl, definitely ranged himself as 
a supporter of Lindley’s view that they represent whole carpels. But this 
premise was directly opposed to the statements of both Payer and Eichler, 
and current opinion remained unchanged. A distinct advance in knowledge 
is, however, reached by the observations of Klein 3 upon the course of the 
vascular bundles. He concludes that in the case of what he calls £ open ’ 
flowers the course of the four bundles destined for the four longer stamens 
indicates a true diagonal position for these members, and that only in flowers 
with petals clawed to accommodate the nectaries do they approximate 
to the median plane. In accord with the conclusion that the four longer 
stamens represent a true tetramerous whorl, he holds that there are four 
carpels in the gynoecium, two of which form the replum, basing this view 
again on the course of the bundles and on the fact that nowhere else do the 
meeting edges of carpels develop such a considerable amount of fibro- 
vascular tissue as is to be found in the replum of the Cruciferae. Both 
arguments afford strong grounds for belief, but the outline needs to be 
extended and the blanks to be filled in if the picture is to be convincing. 
One feature especially conspicuous in glabrous strains of Matthiola 
incana , R. Br., which has always appeared to me difficult to reconcile with 
the view generally held, is the considerable and often varying width of the 
tract of tissue (commissure) separating the valve boundaries, 4 a difficulty 
which does not become less after examination of the internal structure. It 
was, however, only after investigation of a number of fruits of the excep¬ 
tional type mentioned above (p. 451) that the true nature of this tract was 
apprehended. It then became clear that the accepted formula G 2 did not 
fit all the facts observable in this species. On the other hand, these appear¬ 
ances, and various records by other observers of similar occurrences in other 
Cruciferous genera, could all be brought satisfactorily into line on the 
supposition that the minimum member of carpels in the siliqua is not two but 
four , and that the full number in the exceptional freiits is eight. A con- 
1 Die Familien-Diagramme der Rhoeadinen (Abh. Naturf. Ges. Halle, Bd. xiv, 1878). 
2 Pflanzenleben, ii, 1891, p. 683. 
3 Der Ban der Cruciferenbluthe auf anatomischer Grundlage (Ber. d. deut. Bot. Gesell., xii, 
i894)- 
4 The difficulty of regarding the commissures merely as placentae appears fully as great 
in Sisymbrium pannonicum , Jacq., where the disparity in width of valves and commissures is still 
less. (See Reichenbach, leones, ii, Table LXXIV, Fig. 4406.) 
