136 THE STRUCTURE OF FLOWERS. 



Why are tlie three anterior petals of Tropceolum fringed, but 

 the two posterior, which stand a long way behind, not so ? 

 Why are hairs produced on the anterior side of a Honey- 

 suckle and Veronica, but all ronnd the mouth of the regular 

 Geniiana ? And many other questions of a like sort might 

 be raised. If we watch the habits of insects with their 

 tongues, we may easily see how they irritate the various parts 

 by licking them, not solely where the honey is secreted, but 

 the filaments, etc. Thus Miiller often watched Rhingia 

 rostrata licking the staminal hairs of Verhascum phceniceum, 

 and in many cases the hairs on the filaments offer a foothold 

 to the insects while visiting the flowers, as in species of 

 Mullein ; such hairs, if my theory be true, being the actual 

 result of the insects clutching the filaments or rubbing them 

 with their claws. In Gentaurea, the epidermal cells of the 

 filaments have produced projecting processes just where the 

 proboscis rubs against them when searching for honey in 

 the little cup (see Fig 11, p 60), from the middle of which 

 the style issues, as shown by the direction of the arrow. 



These filaments also exhibit their extreme irritability by 

 contracting, and so assisting in the " piston action " by 

 dragging the anther-cylinder downwards over the style. 



While recognizing the coincidence between the localiza- 

 tion of outgrowths, enations, trichomes, etc., and the position 

 of the parts of insects in contact with flowers when searching 

 for honey, one must not forget that a great number occur 

 where such contacts do not take place. Hence we must look 

 for other possible causes for their origin as well. One of the 

 commonest forms of trichomes is glandular hairs, and, as Dr. 

 Kerner has pointed out, when they occur on sepals, pedicels, 

 etc., they form admirable barriers to the approach of ants 

 and other creeping insects, which might rifle the flower and 

 yet not fertilise it. We must be on our guard, however, in 



