8 1 
and Affinities of Trap ell a. 
is given a transverse section corresponding to A-B in Fig. 6 b , 
showing the anther after dehiscence. 
The anterior pair are barren and to be regarded as stami- 
nodes. The connective is here 
smaller than in the fertile stamens, 
but still peltate. Its upper edge 
is drawn out into two small pegs, 
to each of which is attached a 
small barren anther-cell (Figs. *]a 
and b). These are slightly diver- 
gent, as in the stamens; evidently 
the staminodes have been derived 
from a pair of stamens quite like 
the fertile ones here. 
Of the median posterior stamen 
no trace remains. 
The stamens in Pedalineae vary considerably from one 
genus to another. Thus in Pretrea I find the anthers parallel 
and dorsifixed, with no conspicuous development of connective; 
in Pedalium the anthers are divergent, and the connective 
produced into a small glandular apiculus ; the same holds for 
Harpagophytum . 
On removing the corolla and cutting away the calyx-limb, 
the free part of the ovary is seen, terminating in the style, 
inserted somewhat anteriorly. The stigma is interesting, 
being cruciform ; in this it deviates from the ordinary Pedali- 
naceous structure which is bi-lamellate, i. e. with equal an- 
terior and posterior lobes 1 . The cruciform stigma of Trapella 
is doubtless derived from this type. The cross-like form is 
due to horizontal lobes standing out right and left (see 
Figs. 8 and 9) at the insertion of the anterior and posterior 
lamellae. These lamellae are not equally developed, but 
the posterior has undergone great reduction 2 (Fig. 9 gives 
1 See my paper, ‘ Ueb. Fortleitung d. Reizes bei reizbaren Narben,’ in Ber. d. 
deut. bot. Ges. 1887. 
2 It is to be noted that although the anterior loculus of the ovary is almost 
obsolete, it is the posterior stigmatic lobe which has undergone reduction. This is 
Fig. 7. — Floral diagram. 
