POSITIVE DEDOUBLEMBNT. 87 



of the normal flower. What prove it,, on the other 

 hand, to be a retrogressive, ancestral structure, are 

 firstly, the petaloid and staminodial forms under which 

 the stamens of the outer whorl appear, and secondly, 

 the weak, imperfect (as regards size) development of 

 the median carpels. For this is just the way in which 

 vestigial organs do tend to reappear or to be formed 

 before they finally die out. In another flower one of 

 the stamens of the outer whorl was petaloid and 

 anther-bearing, the other was represented by two fertile 

 ones (showing a further step backwards towards the 

 ancestral condition) ; of the four stamens of the inner 

 whorl the two posterior ones had completely vanished. 

 In the centre were five carpels; beyond these slight 

 proliferation had occurred, as the whorls were further 

 added to by five stamens and two carpels (fig. 966). 



Such cases as those just mentioned represent transi- 

 tions (the normal flower of Gleome spinosa itself forming 

 a link between Cruciferae and Capparidaceae) between 

 that prevailing in the normal flower of the majority of 

 Cruciferae and that of the normal flower of the less 

 frequent genera Megacarpsea and Holargidium in which 

 the flowers may be 10- or 16-androus,* and which thus 

 approximate to, and form links with, the typical 

 CapparidaceEe.t 



Herm aphroditism. 



Female flowers may become once more bisexual by 

 the reappearance of the stamens which had become 

 totally suppressed in the original hermaphrodite flower. 



In Bogonia hermaphrodite flowers were observed 

 due to the production of a group of stamens on one 

 side of the female flower between the two broad petals; 



* Velenovsky observed flowers of the wallflower in which the two stamens 

 of the outer whorl were each represented by a group of three stamens : an 

 outermost and two inner ones. 



t Velenovsky, while allowing that the ancestors of the Ehceadales had a 

 polymerous androecium, regards this condition in Cruciferse as a recently- 

 acquired phenomenon. Celakovsky's view of the matter is, however, here 

 supported. 



