POLLINATION OF CYANELLA CAPENSIS. 50I 



are almost pendulous due to a sharp bend, which forms at the 

 junction of the petiole and bud; it is in the latter position that 

 the bud expands. I have no record of the length of time elapsing 

 between the first formation of the bud and the opening of the 

 flowers, but in the plants kept under observation it took 12 days, 

 viz., from December 4th to December 15th for very young buds 

 to expand into the adult flower. Buds which have already 

 assumed a position below the horizontal opened within 1-4 days. 

 All the flowers open during the night ; in no single instance did 

 any of the flowers open during the day. I always found buds, 

 which at 4.30 p.m. looked mature, open at 9 a.m. the following 

 morning. In one instance, a bud which was partly open in the 

 morning remained in the same position during the whole day, 

 and only open out completely the following morning. When a 

 bud opens the antherior perianth-segment moves first ; then the 

 two lateral inner segments, followed by the lateral outer seg- 

 ments, and finally the posterior segment. All the segments, with 

 the exception of the anterior one, which hangs down vertically, 

 become reflexed. The anterior anther is parallel with the 

 anterior segment ; i.e., it is also vertical ; the two lateral anthers 

 lie at an angle below the horizontal, and above them are arched 

 the three posterior anthers. In this stage of the flower the style 

 is curved upwards, so placing the stigmas in the same vertical line 

 as the tips of the lateral antihers. The stigmas usually lie mid- 

 way between the two stamens, but in some cases the style is bent 

 either to the left or right, and the stigmas are then immediately 

 under the pore of one of the anthers. The flowers remain in 

 this position for 4-5 days ; in one case seven days elapsed between 

 the opening of the bud and the closing of the flower. Pollina- 

 tion of the stigmas seems to act as a stimulus to the perianth- 

 segments to close, as in flowers which were artificially pollinated 

 on opening, closed within two days ; otiher flowers of the same 

 inflorescence, and which opened at the same time and not pollin- 

 ated, remained open for six days. A minute drop of fluid is 

 exuded from between the stigmas ; in some cases I observed this 

 just after the flower expanded; in other instances not until 2-3 

 days after opening. 



Every bud which expanded in the herbarium eventually 

 became pollinated, and this could only have been due to self- 

 pollination. In some cases pollen was observed on the stigmas 

 before the flowers closed ; in other cases not ; but in all the ovary 

 has swelled and a fruit was in the course of formation. Buds 

 on plants growing in the field were enclosed in muslin bags to 

 prevent the visit of insects, and in every case a mature fruit 

 had formed. Though I have had these plants under observa- 

 tion for some weeks in the field I have failed to notice a single 

 insect visiting them, though several species of bees and flies 

 were common in the neighbourhood. 



The results of my experiments make me believe that there 

 are three stages in the pollination of Cyanella capensis as fol- 

 lows : — 



