THE STRUCTURE OF FLOWERS. 59 



observes very minute petals or none. Moreover, they 

 may have no honey-glands at all. Very often the flower- 

 buds do not open, and yet they can produce plenty of 

 seed. 



In looking for correlations between flowers and 

 insects generally, one observes that nearly all regular 

 flowers are either terminal, i.e. at the ends of flower- 

 ing stems, or, at least, they are so situated that insects 

 can extract the honey from any point of the cir- 

 cumference. 



On the other hand, irregular flowers are mostly so 

 placed upon the stems that they can only be approached 

 from one side, i.e. in front. 



In addition to this fact one finds a great number of 

 features all in correlation to the presence of insects. 

 Thus, e.g., in SaVvia (Fig. 15, II.) there is the strengthen- 

 ing of the calyx to support the slender tube of the 

 corolla, since the front petal is greatly enlarged to 

 carry the weight of the insect ; while the stamens 

 and style are modified and the honey-gland developed, 

 so that all are in adaptation to the insect-visitor. 



If no petal happens to be in front, then the stamens 

 and style are " declinate," i.e. inclined downwards and 

 then upwards, tlius forming a convenient and sufficient 

 resting-place instead, as in Felargo'nium (Fig. 12). 



In regular flowers, as of the group Disciflorce, the 

 circular, fleshy, honey-secreting disc, as in the wilde 

 Kastanien tree, or the five separate honey-glands in 



