242 Dr Dickson on Diplostemonous Flotvers, dc. 



succession upon a cushion-like body which precedes their 

 appearance. Payer has, however, by the most convincing 

 arguments, determined the staminal phalanges of poly- 

 adelphous flowers to be compound stamens, the parts of 

 which, when centrifugally developed, correspond to the 

 basipetally developed leaflets or lobes of ordinary leaves ; so 

 that in these flowers we have no real examples of centrifugal 

 succession of parts ujjort an axis. A curious arrangement, 

 however, is described by Payer as occurring in Opuntia, 

 where vast numbers of stamens are developed in centrifugal 

 succession, and apparently distributed uniformly round the 

 receptacle. We may arrive at a comprehension of this re- 

 markable form, if we direct our attention to certain cases 

 which appear to connect it with more easily intelligible 

 forms. In Bratliys {Hypericum) prolijica, Payer has shown 

 that the staminal cushions (which usually remain distinct 

 in the H3q3ericacese) become fused together, at an early 

 period, into a single nearly uniform annular cushion, ujjon 

 which the stamens make their appearance centrifugally. 

 Again, in Cistus, he has shown that, in the early condition, 

 the centrifugally developed stamens exhibit distinct traces 

 of grouping, although the annular cushion, on which they 

 are developed, is always entire. From Cistus we pass at 

 once to Helianthemwm, where all trace of grouping in the 

 stamens disappears,* presenting us with a condition quite 

 analogous to that in Opuntia. Such a series of forms leaves 

 us in no doubt that in Opuntia, as in SelianthemuAn and 

 Cistus, we have merely an extreme case of the fusion of 

 compound stamens, which differs from that in Bratliys only 

 in being congenital, while in that plant it does not com- 

 mence until a little after the appearance of the staminal 

 cushions. 



In connection with the above, I must not omit allusion 

 to Payer s own determination of the signification of the an- 

 droscium in Cistus ; and as the questions suggested by it 

 have no unimportant bearing upon the subject of this paper, 

 I may be excused the apparent digression of commenting 

 upon it. In this plant he has found that the stamens 



* Organogenie, p. 17, plate iii. fig. 25. 



