THE SHAPES OF FLOWERS. 169 



result if the several sides of the flower did not afford it equally 

 free access. In like manner, flowers which are distributed 

 over a plant in such ways that their discs open out on 

 planes of all directions and inclinations, will have no tend- 

 ency to lose their radial symmetry; since, on the average, 

 no part of the periphery is differently related to insect- 

 agency from any other part. But flowers so fixed as to 

 open out sideways in tolerably-constant attitudes, have 

 their petals differently related to insect-agency. A bee or 

 butterfly coming to a laterally-growing flower, does not settle 

 on it in one way as readily as in another; but almost of 

 necessity settles with the axis of its body inclined upwards 

 towards the stem of the plant. Hence the side-petals of a 

 flower so fixed, habitually stand to the alighting insect in 

 relations different from those in which the upper and lower 

 petals stand; and the upper and lower petals differ from one 

 another in their relations to it. If, then, there so arises an 

 habitual attitude of the insect towards the petals, there is 

 likely to be some arrangement of the petals that will be 

 most convenient to the insect — will most facilitate its entrance 

 into the flower. Thus we see in many cases, that a long 

 undermost petal or lip, by enabling the insect to settle in 

 such way as to bring its head opposite to the opening of the 

 tube, aids its fertilizing agency. But whatever be the special 

 modifications of the corolla which facilitate the actions of 

 the particular insects concerned, all of them will conduce to 

 bilateral symmetry; since they will be alike for the two sides 

 but unlike for the top and bottom. And now we 



are prepared for understanding the exceptions. Flowers 

 growing sideways can become thus adapted by survival of 

 the fittest, only if they are of such sizes and structures that 

 insect-agency can affect them in the way described. But 

 in the plants named above, this condition is not fulfilled. A 

 Hollyhock-flower is so open, as well as so large, that its petals 

 are n$t in any appreciable degree differently related to the 

 insects which visit it. On the other hand, the flower of the 



