REPRODUCTION AND DISPERSAL 877 



very different insects in different countries; naturalized flowers seem 

 to be pollinated by the insects of the new country quite as successfully 

 as their congeners are pollinated by the insects of their native country ; 

 insects sometimes are held captive and even are killed by specialized 

 floral structures, as in Asclepias and Hedychium; bees that are unable 

 to reach the honey in long floral spurs frequently bite holes at the side, 

 thus getting the nectar without effecting pollination. 



Probably the chief reason for not holding to natural selection as a 

 factor of prominence in the origin of floral structures is that flowers, 

 though they are the most diversified and specialized of plant organs, 

 probably have played a comparatively minor role in determining the 

 success of plant groups. It is likely that the success of the grasses and 

 the catkin-bearing trees is due less to the floral features above noted 

 than to vegetative reproduction in the former and to the tree habit in 

 the latter. 



Perhaps the best evidence in support of the view that flowers contribute 

 largely to the success of plants is found in the composites, though even 

 here it is likely that vegetative reproduction and the wind dispersal of 

 seeds play a larger part. Even the composite flower owes its advantage 

 not so much to floral specialization as to the massing of inconspicuous 

 and relatively non-specialized flowers into compact heads, which greatly 

 facilitates pollination. Furthermore, it is to be remembered that the 

 composites, forming supposedly the highest of plant groups and certainly 

 the largest in number of species and one of the largest in display of indi- 

 viduals, are notable for their geitonogamy and autogamy, for their rela- 

 tively actinomorphic flowers (the disk flowers being strictly actino- 

 morphic) with their pollen exposed for any insects that may visit them, 

 and for their tendency toward dicliny; it may be significant, also, that 

 the greatest display of parthenogenesis among seed plants is among the 

 composites. 



The preceding paragraphs appear to show that the fundamental pos- 

 tulate of natural selection, namely, that the trend of evolution is along 

 the line of maximum advantage, is untenable, at least so far as flowers 

 are concerned. The evolution of the orchids beyond the point of maxi- 

 mum advantage, the phenomenal success of the groups with generalized 

 flowers, and the probable dominance of the vegetative over the repro- 

 ductive factors in determining success in the majority of groups, all 

 appear to indicate that some other factor than natural selection has 

 determined the diversity of floral structures. Though the theory of 



