50 THE STRUCTURE OF FLOWERS. 



at tlie apex, and so form a tube which collects the pollen 

 shed iuto it by the five free anthers, which are included 

 "within this corolla-tube (Fig. 9). 

 They thus form the " cylinder " for 

 the " piston " action of the pistil which 

 continues to grow, and so sweeps out 

 the pollen beyond the extremity of 

 the tube, just as it does from the 

 syngenesious anthers of the Com- 

 positce and Lobelia. The five portions 

 of the corolla thus cohering by con- 

 tact subsequently become more or less 

 free. 



The rationale of Cohesion lies in 

 its adaptation to insect agency, and 



Fig. 9. — rhi/teuma (a.{ter Miiller). . -,. . ^ ^ . i- 



implies a greater degree or specializa- 

 tion than when the parts of the whorls are free. Thus in 

 Thalamiflorce, of such an order as Banunculacece with regular 

 flowers and with all the parts of the perianth whorls free, the 

 flowers are usually visited by a much greater number and 

 variety of insects than are those of orders of Corolliflorce. For 

 example, Miiller records sixty-two species of insects as seen by 

 him to visit Ranunculus acris ; whereas the humble-bee alone 

 enters the gamopetalous tube of the Foxglove. This adapta- 

 tion of form to insect visitors will be better appreciated when 

 we come to discuss that principle of Variation, w^hich so 

 powerfully affects floral structure. 



It occasionally happens that parts normally united become 

 free : the process is called " dialysis," and may be regarded 

 as a reversion to an ancestral free condition. Fig. 10 repre- 

 sents a flower of Mimidus in this condition. The rationale 

 of cohesion in the sepals, petals, and stamens, I regard as the 

 immediate result of hypertrophy set up by insect agency, 



