232 The Romance of Wild Flowers 
only occasionally, self-fertilisation may be effected by 
the stamens erecting themselves by the style. Now, 
it is believed that the ancestral Gentian had open 
flowers like this and of the same colour, offering 
honey and pollen to all comers, and getting an 
occasional cross in return for its hospitality. From 
this stage evolution proceeded along two lines, in both 
making the honey less accessible by connecting the 
petals throughout a gradually increasing part of their 
length, and so bringing about the prevailing bell- 
shape, as most befitting the humble-bees who were the 
principal selective agents. Along one line the honey 
got to be hidden in deep narrow passages only 
accessible to those with long tongues, and to secure 
cross-fertilisation the anthers took up their position 
round the style, where they must be crawled over by 
insects that would have honey. And in the variation 
of the flower-colours those that developed any tinge 
of blue—even in spots, as some of the Continental 
forms still possess it—would be selected by preference 
by bees who have special fondness for that colour. 
Our own species, pneumonanthe, is an illustration ; 
its blueness is ‘chiefly within. Then some of these 
got taken in hand by the butterflies, and further 
modified accordingly, the tube being narrowed and 
lengthened, and the hilobed stigma developed into a 
disk partially closmg the mouth of the corolla. 
G. verna and G. nivalis exemplify this type in our 
country. The other series developed along the line 
of excluding unprofitable visitors by hairs on the 
corolla, but narrowing of the corolla had to proceed 
at the same time; finally we find, as in our G. 
campestris and G. amarella, the throat-hairs and the 
