358 AUTOGAMY. 



but a process of involution like that exhibited in Allionia ensues, and as soon as 

 autogamy has been initiated the limb of the perianth folds up, turns pulpy, and 

 then forms a kind of stopper above the knot of twisted filaments and style. The 

 flowers of the Purslane (Portulaca oleracea) differ from those of Commelyna, Allio- 

 nia, and Mirabilis in having five stigmas which are like delicate feathers in form, 

 and are spread out in a star in the middle of the erect flower-cup. The stamens 

 project obliquely from the receptacle, and are arranged in a circle round the stigma; 

 but when the flower opens first, there are little spaces between anthers and stig- 

 mas, and this prevents a spontaneous transference of pollen to the stigmas. After 

 the lapse of a few hours the petals, which in the sunshine are expanded in the form 

 of a cup, draw together, and the flower begins to close up; all the five feathery 

 stigmas bend over to the same side and gradually coil up into spirals. The thread-like 

 stamens also undergo inflection, at first into semicircles, and subsequently into 

 spirals, and the pollen-coated anthers are in consequence pressed against the stig- 

 mas. At this period, in the Purslane as in the Marvel of Peru and other plants 

 whose flowers are ephemeral, the petals may be seen in a pulpy condition covering 

 over the knot of tangled filaments. 



As has been said before, this form of autogamy occurs chieflj^ in flowers which 

 last only a single day. Where the whole period during which the flower is open is 

 but a few hours the movements in question may all be followed by the observer. 

 In the few species, which resemble the foregoing in respect of autogamy, but differ 

 from them in that their flowers remain open two or three days, or even longer, 

 these movements of inflection and torsion take place much more slowly. Thrifts 

 (Armeria alpina, A. vulgaris, &c.) display in the middle of each of their cup- 

 shaped flowers five stigmas disposed in the same manner as those of Purslane- 

 flowers. The stigmas in this case, however, are not feathery, but in the form of 

 slender cylinders covered with short, closely -packed papillae, which give them a 

 velvety appearance. The stamens are adnate to the short corolla-tube, and rise 

 up in front of the petals holding their anthers between the rays of the stigma. 

 Notwithstanding the proximity of the anthers to the stigmas, neither in the first 

 nor in the second stage of the flower's development is any pollen transferred, with- 

 out extraneous aid, to the receptive stigmatic tissue. At first the stamens are so 

 placed as to have their anthers brushed b}^ insects visiting the flower, whilst the five 

 stigmas are still erect. A little later the anthers and stigmas change places as in 

 so many other cases; the stamens stand up and bring the anthers together nearer 

 the middle of the flower, whilst the stigmas diverge from one another, and place 

 themselves close to the way leading to the honey. Attention has been so often 

 directed to the connection between an interchange of position of this kind and 

 the accomplishment of cross-fertilization that it is needless to repeat the facts of 

 the case. Supposing, however, that insects do not visit the flower, and that, in 

 consequence, heterogamy fails, the styles wind themselves up spirally, and move at 

 the same time towards the middle of the flower, where they become entangled with 

 the filaments, which have likewise undergone spiral torsion. In these circum- 



