THE FLOWER. 



in each cluster with which the stamens cohere. The explanation 

 by chorisis is that each cluster, petal-like body included, is a 

 multiplication of one stamen. The diagram (Fig. 398) accu- 

 rately shows that most of the stamens originate from the outer 

 side of the base of the petal-like portion : this is most naturally 

 explained by median chorisis. The superposition of the clusters 

 to the petals will take the same explanation as that of Rhamnus, 

 Vitis, &c. (Fig. 363.) That the androecium is here composed 

 of the inner circle merely is partly confirmed by the alternation 

 of the carpels with the clusters. According to Duchartre, 1 the 

 development of the androecium in a Mallow indicates a similar 

 structure ; for the whole united mass originates from five protu- 

 berances, one before each forming petal and connected with it, 

 this by collateral chorisis forming a cluster of stamens, and the 

 five clusters coalescing as they develop into a tube of filaments, 

 such as in Fig. 485. Now Hibiscus and its near relatives have 

 a naked tip to the stamen-tube, ending usually in five teeth ; 

 and Sidalcea, as is most strikingly shown in the Californian 

 S. diploscypha, has two series of stamens, the outer (answering 

 to those of Malva and its relatives) in five membranaceous pha- 

 langes, superposed to the petals ; the rather numerous inner 

 series, more or less in phalanges, surmounts an interior filament- 

 tube. Whence it is inferred 

 that these, and the five teeth 

 terminating the column in 

 Hibiscus, represent the in- 

 ner stamineal circle which is 

 wanting in Malva, as it is in 

 Tilia. 2 



381. The case of Paraas- 

 sia would be explained as 

 analogous to that of Tilia, 

 but with the stamen-clusters before the petals wholly sterile, 

 and of fewer divisions, while an inner circle of five stamens 



1 Comptes Rendus, 1844, & Ann. Sci. Nat. ser. 3, iv. 123. Duchartre and 

 others who draw freely upon median chorisis to explain anteposition, and 

 consider that congenital union proves it, take the phalanges in these cases, 

 like the single stamens in Vitis, to be an inner part of the petal itself. But 

 this view appears to have had its day. 



2 Gray, Gen. Illustr. ii. 44, 57, 75-82. The position of the carpels before 

 the petals in Pavonia and Malvaviscus brings the former into symmetrical 

 alternation with such an inner stamen-circle ; but it is not so in Hibiscus, 

 which has the carpels before the sepals. 



FIG. 400. A petal of Parnassia Caroliniana, with a triple gtaminodium before it. 

 FIG. 401. Diagram of the flower of Parnassia Caroliniana. 



