GEITONOGAMY. 323 



branches diverge, not in a liorizontal but in a vertical plane, the consequence being 

 that one of the stylar branches comes in contact with the clumps of pollen just 

 pushed out from a neighbouring higher flower. Of course there exist among Com- 

 posite flowers quite a number of forms intermediate between those here described, 

 all of which promote geitonogamy, but we cannot enter upon them further. 



Among Umbelliferous plants, as in Compositse, tlie numerous small flowers are 

 crowded so closely together that the stigmas and pollen of neighbouring flowers can 

 easily touch and combine, and a glance at these inflorescences is enough to suggest 

 the possibility of geitonogamy. The conjecture is confirmed on a closer scrutiny, for 

 the Umbelliferae, as a matter of fact, exhibit an even greater variety of contrivances 

 for geitonogamy than do the Composite. The most important of these will now 

 be detailed. First, let us examine the group which is typified by the genera 

 Eryngium and Hacquetia. In these the flowers are crowded together in capitulate 

 masses surrounded by broad conspicuous bracts. They are all hermaphrodite and 

 protogynous. The stamens with closed anthers are bent inwards in each flower 

 like hooks, and the petals are still unfolded, but the sticky, shining stigma on the 

 top of its long style already projects some distance out of the bud. At this stage 

 the stigma can only be pollinated with pollen from other plants, indeed, from other 

 species. Later, the stamen-filaments elongate and straighten, whilst the anthers 

 dehisce, and pollen appears through the clefts. This pollen comes into contact 

 either at once or very soon with the still receptive stigmas; for the long styles have 

 meanwhile inclined more to the "side, so that their stigmas are placed in such a 

 position with regard to neighbouring flowers that either they brush against the 

 pollen-covered anthers, or else are pollinated with the crumbling pollen which falls 

 from these anthers. 



The genera Sanicula, Astrantia, and Laserjntium, diflTer somewhat from this 

 group of Umbelliferae. The chief modification is that in the species of these three 

 genera staminate as well as hermaphrodite flowers occur. In Sanicula each umbel 

 consists of 3 true hermaphrodite flowers in the centre, and 8-10 staminate flowers 

 arranged round them like a wreath. The hermaphrodite flowers are protogynous 

 and the flrst to develop, so that at the beginning of flowering the stigmas can only 

 be fertilized with the pollen from plants of other species. The stamens straighten 

 afterwards and project like the style far out of the flowers. But the anthers and 

 stigmas of the same flowers do not touch, since the styles stand up erect, whilst the 

 stamens have an oblique direction. A crossing soon takes place, however, between 

 the hermaphrodite flowers and the adjoining staminate flowers, in the following 

 way. The stamens of the hermaphrodite flowers wither and fall off, and the 

 stigmas of these flowers become divergent, curving slightly outwai-ds, so that their 

 still receptive surfaces are brought into the surrounding circle of stamens. The 

 anthers of the male flowers have meanwhile dehisced and are shedding abundant 

 pollen. The stigmas are necessai-ily pollinated either by the direct contact of 

 stigmas and anthers, or, by the falling of the pollen from these anthers. The 

 arrangement of the flowers in Astrantia has already been described on p. 296; it 



