171 



seen in a style consecutively immersed in absolute alcohol and 

 turpentine. Now the pollen when deposited is always at a good 

 distance from the stigma proper but as to the stigraatic line 

 sometimes so close by that the slightest movement of an insect 

 can cause pollinisation. This circumstance together with the 

 presence of numerous ripening seeds and fruits led me to the 

 supposition of selfpoUinisation with or without the aid of in- 

 sects. When in Java (Buitenzorg) I studied the said process in 

 Canna Annaei during several weeks ') but failed, on account of 

 the premature fall of the fruits, to reach a definite conclusion. 

 My last notes, however, point to self fertilisation. KRaNZLiN (XII) 

 in his well known monography has given a survey on this subject. 

 After this expatiation we will return to the rudimentary 

 styles .s' and s' which as we know appear at the left side of 

 our figures. Just behind s' we see a filament bearing the only 

 anther of a Canna-flower. The fact that both are placed on 

 the same radius together with the ala y, attracts attention 

 (fig. 4 and fig. 14). It points to the fact that the stamen be- 

 longs to the outer whorl of stamiuodes and consequently has 

 from a genetic point of view nothing to do with ,v with which, 

 however, it forms one whole. This conception is confirmed on 

 studying transverse sections as represented in fig. 7, in which 

 m indicates the stamen or more accurately its vascular bundle, 

 and also when carefully removing the ala y hj the fact that 

 the anther, whether sessile or stalked, firmly sticks to it and 

 easily gives up its weaker connection with x. When doing this 

 we see at the same time a stout vascular bundle in a\ which 

 may be looked upon as one of its principal ribs. Fig. 15 shows 

 in which way the connectiou with y in its higher portions 

 becomes weaker and weaker to end finally in a separation 

 (from /) and a lasting cohesion with ,v. That the rudimentary 

 style s' in front of the fertile stamen even after its release 

 does not part company is a striking proof of the latter's anti- 

 sepalous origin. 



■]) Probably a hybrid of C. glauca L. and C. nepalensis Wall. (XII, p. 25). 



