CHOKISIS OK DEDUPLICATION. 



245 



next within, the lateral and exterior petals ; those alternate and 

 within these, the inner circle of petals ; and alternate with these are 

 the anthers of the two stamen-clusters. The centre is occupied by 

 a section of the pistil, which consists of two united. The three sta- 

 mens are lightly connected in Dicentra (Fig. 371) ; but in Corydalis 

 and Adlumia there is only one strap-shaped filament on each side, 

 which is three-forked at the tip, each fork bearing an anther (Fig. 

 374). We have a similar case in some Hypericums and in Elodea 

 (Fig. 375), except that, while the floral envelopes are in fives, the 

 circles within tliem are commonly in 

 threes. The three members of the 

 androeciura are normally placed, alter- 

 nating with the three members of the 

 gynascium within, and also with three 

 glands, which probably replace another 

 circle of stamens. Now each real 

 stamen is here multiplied into three, 

 united below ; so that the whole compound body may be viewed as 

 homologous with a compound trifoKolate leaf (289). If this be so, 

 then each cluster of numerous stamens in the common St. Johns- 

 wort may be regarded as answering to one stamen greatly multiplied 

 in the same way, and as analogous to a sessile decompound leaf. 

 And the same may be said of each stamen-cluster in the Linden 

 (Fig. 383). The actual development of the cluster, from a protu- 

 berance which in the forming flower-bud occupies the place of a 

 single stamen, has been traced by Duchatre, Payer, &c. in this and 

 other cases. 



457. Thus far we are sustained by a clear analogy in the organs 

 of vegetation. As the leaf frequently develops in the 

 form of a lobed, divided, or compound leaf, — that is, as 

 a cluster of partially or completely distinct organs from 

 a common base, — so may the stamen, or even the pistil, 

 become compound as it grows, and give rise to a clus- 

 ter, instead of completing its growth as a solitary organ : 

 and it appears that the organogeny is strikingly sim- 

 ilar in the two cases. Nor is it very unusual for petals to become 

 divided or deeply lobed in the same manner ; as, for example, those 



no. 375. Diagram (croas-section) of'a flower of Elodea Virginica. 376. One of the three 

 Btamen-clusters, coasisting of a trebled stameD, enlarged. 

 FIG. 377. A petal of Mignonette, enlarged. 



21* 



