THE ANGIOSPERMAE 



1 105 



an uncompromising adherence to any one of the three theories has led to 

 some remarkably twisted interpretations of difficult cases. On the other 

 hand evidence has been produced, especially from floral anatomy, giving 

 support to each of the theories respectively. Considering this and consider- 

 ing also the great plasticity of floral organs and the vast range of variation 

 in known floral structure, it does no violence to the probabilities if we con- 

 clude that epigyny has been reached by various means and that no uniform 

 explanation can be devised, though modern evidence favours the view that 

 the receptacular type of inferior ovary, which truly involves an invaginated 

 axis, is at least uncommon. 



We have referred above to the fusion of floral parts as a common pheno- 

 menon. Less common but yet not rare is the opposite event of the splitting 

 of an organ originally single. This was first recognized by Payer in his 

 great work on the ontogeny of flowers (1857) which provides a wealth of 

 information on the subject. He called it dedoublement, but the term in 

 general use is Eichler's chorisis. No genuine distinction can be drawn 

 between the branching of an organ, which partially divides it, e.g., the 



stamens of Tilia, and congenital chorisis, in which two 

 rudiments appear in a position where only one would 

 be expected either from the general symmetry of 

 the flower, or from comparison with allied types 

 (Fig. 1076). It is, however, possible to distinguish a 

 simple splitting, such as that of the stamens of 



Fig. 1076. — Androe- 

 cium of Vella 

 showing chorisis 

 of stamens. {After 

 Velenovsky.) 



Fig. 1077. — Adoxa moschatellina. Flower show- 

 ing chorisis of stamens. Each filament bears 

 only a half-anther. Gvnoccium omitted. 



Adoxa (Fig. 1077) or Carpiruis, where each portion carries only two 

 pollen loculi and is plainly the half of a complete anther, and a true 

 chorisis from which two, or sometimes more than two, apparently complete 

 members results — a species of twinning — even although the twins may re- 



