116 PRINCIPLES OF PLANT-TERATOLOGY. 



deceptive data of the development, that petals and 

 stamens in Primulacese constitute a single whorl. 



This is an excellent example of how the ontogeny 

 proceeds in exactly the opposite direction to the 

 phylogeny. For that reason are the ontogenetic facts, 

 in this case as' in so many others, exactly the reverse 

 of serviceable to the comparative morphologist. 

 Teratological facts and comparative data are alone 

 useful. 



In other Gamopetalse, such as Antirrhinum, the 

 same abnormality has been seen. Henslow figures 

 in the dark mullein (Verbascum nigrum), stamens 

 inserted on the receptacle (PI. XXXIX, fig. 10). 



(4) GrYNCEOEUM. 



Apocakpy. — It is generally recognized that the most 

 primitive type of pistil is the apocarpous one, such as 

 occurs in the Ranales, Alismacege, and Rosaceae, in 

 which the individual carpels are separate and distinct 

 structiires. The syncarpous pistil in which the carpels 

 are all laterally united to form a single structure, the 

 compound ovary, is a more recently acquired organ. 



Just as, under the influence of the upset of equi- 

 librium of the plant's constitution, the adelphous 

 androecium becomes^ broken up into its constituent 

 -parts, in the same way may the syncarpous ovary 

 become similarly transformed, and there is no reason 

 for regarding this as other than a reversion, or a 

 tendency to reversion, to an earlier state. 



In the fasciated flower of Scilla nutans above 

 described, which was the equivalent of several flowers, 

 and in which the stamens had mostly become petaloid, 

 the carpels had nearly all become dissociated and were 

 represented by a host of separate leaves, although 

 there was also present an almost perfect trilocular 

 ovary ; these separate carpels bore anthers. 



It is quite common for the ovary of the Cruciferse 

 to become resolved into its two component carpels, as 



