48 TO THE DEPENDENCE ON INSECTS. [CHAP 



(Fig. 40) are nearly twice as large as those of G. pyre- 

 naicum (Fig. 49 b\ which again are much larger than 

 those of G. molle (Fig. 49 c\ while those of G.pusillum 

 (Fig. 49 d] are still smaller. These differences of 

 size appear to be connected with other remarkable 

 differences between these species. Fig. 41, as already 

 mentioned, represents a flower of G. pratenze when 

 first opened. Five of the stamens have raised them- 

 selves and stand upright, and surround the still 

 immature pistil. When they have 

 shed their pollen they sink back and 

 shrivel up, when the other five raise 

 themselves. At a later stage these 

 in their turn fall back and shrivel up, 

 but the stigma does not become ma- 

 ture (Fig. 42), until all the stamens 

 have shed their pollen. Under these 

 circumstances G. pratense has lost 

 the power of self-fertilisation, and is 

 absolutely dependent on the visits of 



FIG. 49. Corolla of a, 

 Geranium pratense i _ . . . 



t, G. tyrenaicum; c , G. pyrenatcum (Fig. AQ &) is also pro- 



G. molle; d, G. f>u- . 



siiium. terandrous ; but while in G. pratense 



the pistil is not mature until the sta- 

 mens have shed all their pollen and fallen back, in 

 G.pyrenaicum the second series of stamens are still 

 upright when the stigmatic lobes unfurl ; the flower is 

 consequently less absolutely dependent on insects, and 

 we see that the corolla is much smaller. 



In the third species, G. molle (Fig. 49 c), the pistil 

 matures before the second series of stamens, and the 

 corolla is still smaller ; while in G.pusillum (Fig. 49 d) 



