126 PROTECTION OF POLLEN. 



clear nights must inevitably rest on the open flowers. The immediate wetting 

 •of the entire flower is, however, prevented by peculiarities in the form of the limb. 

 The anthers close with great celerity upon the deposition of the drops, the expla- 

 nation being that the perianth-lobes are connected with the anthers standing in 

 front of them by a bunch of twisted hairs which not only are themselves peculiarly 

 susceptible of being wetted, but conduct the water to the anthers and so cause the 

 anther-walls to close. 



A characteristic manner of protecting the pollen by means of the anther- 

 walls after the pollen has been set free, and when it is ready to be carried away 

 by insects, may be observed in several Composites (e.g. Onopordon, Centaurea). 

 There is no material difference between these plants and the other Composites 

 discussed on p. 114 in respect of the structure of the tube of syngenesious anthers, 

 the discharge of the pollen into that tube, or the structure of the stj'le and its 

 situation inside the anther- tube; but an essential distinction exists in the fact that 

 the pollen is conveyed to the mouth of the tube not through the elongation of the 

 style but the contraction of the filiform supports of the anther-cylinder. These 

 filaments in Onopordon and Centaurea contract in response to mechanical stimuli, 

 and in shortening they pull down the anther-tube with them. The top of the style 

 thereupon becomes visible, for the style is sheathed in the tube, and does not 

 shorten when the filaments do so nor change its position. The pollen resting on 

 the st34e is consequently exposed, and appears in the form of a pulverulent 

 mass on the top of the stjde surmounting the anthers. If the mechanical stimu- 

 lation of the filaments is due to the hovering of an insect about the capitulum, 

 the pollen is no sooner exposed than it is brushed off" by the insect, and the entire 

 contrivance is obviously so devised that the same insects as cause (by the touch 

 of their legs or probosces) the contraction of the filaments, the retraction of tlie 

 anther-tube, and the exposure of the pollen may be themselves loaded with the 

 pollen. Up to the moment of the insect's visit, however, the pollen is hidden in 

 the sheath formed by the anthers, and this position is of advantage to it inasmuch 

 AS it is there sheltered from rain and dew. The Composites in question have 

 their capitula erect. The capitula of Onopordon include neither movable ligulate 

 •ray-florets nor radiating bracts capable of closing. Centaurea has trumpet-shaped 

 marginal florets, but they do not possess the power of arching over and protecting 

 the tubular florets of the centre. The stalks of the capitula become neither 

 pendent nor nodding in wet weather. In short, the pollen of these particular 

 Composites is destitute of any of the various means of protection which are present 

 in other genera of the same family and which have just been discussed. But 

 instead, the anther-tube itself undertakes the task of sheltering the pollen after 

 the latter is liberated until the moment when the insects which are to carry it away 

 alight upon the flowers. 



We need only notice incidentally that extrorse anthers, which turn their 

 recesses filled with coherent masses of pollen towards the earth and their backs 

 to the sky are also to a certain extent protected against wet. A more impor- 



