SECRETIVE TISSUES. 143 



or else anemophilous. In these cases, in unison with the 

 degradation in size and colour of the corolla, or else its entire 

 loss, the nectaries tend to and g-enei^lly vanish entirely ; as 

 may be seen in Polygomcm aviculare as compared with P. 

 Fagoinjrum and P. Bistorta. 



The simple origin of nectaries, then, accoi'ding to my 

 theory, is that insects, having been attracted to the juicy 

 tissues of flowers, by perpetually withdrawing fluids have 

 thereby kept up a flow of the secretion which has become 

 hereditary, while the irritated spot has developed into a 

 glandular secreting organ.* These spots occur wherever the 

 prevailing insect found it most convenient to search ; hence it 

 is sometimes at one place, sometimes at another, even in 

 closely allied plants. Thus, in Buttercups the stamens and 

 carpels form a compact globe, especially the latter, and defy 

 the penetration of a proboscis. The corolla, however, admits of 

 an entrance of its base. In Atragene alpina the basal portion 

 of the filament forms a nectary (Fig. 44). Comparing these 

 with Caltha, the large carpels of this plant admit the passage 

 of a proboscis between them ; and the nectaries are now 

 developed on the sides of the ovaries, exactly where they 

 would be irritated. 



In Ranunculus cortuscefolius, of the Canary Islands, which 

 has a corolla more than two inches in diameter, the petals 

 are entirely without honey-glands. On the other hand, the 

 carpels are very large and flat, with plenty of space between 

 them. Although I could detect no honey in plants grown at 

 Floore, Weedon, the tissue over the centre of the ovary was 

 modified, and exactly resembled the ordinary tissue o£ a 

 honey-gland. If I am justified in assuming the carpels as 



* It is closely analogous to the action of the pollen-tube, which 

 causes a flow of nutriment to the conducting tissue, only there is a 

 physiological as well as mechanical irritation in that case. 



