186 PEINCIPLES OF PLANT-TEBATOLOGT. 



median placentas have fclie forked apex which is a 

 prerequisite for anther-formation according to the 

 first method described when treating of that subject. 



The value of these carpellodic petaloid stamens of 

 Primula is further shown by the light which, by means 

 of them, we are able to throw on the morphology of 

 the " free central " placentation of this genus, if we 

 imagine these basal, ventral placentas, which are 

 detached from both midrib and margins of the stamen, 

 to become all five fused laterally, the placentas then 

 covering the upward-growing axis in the centre of all, 

 the " free-central " placenta would be attained. 



By a comparative study Celakovsky has shown 

 that all carpels without exception are ascidia or 

 pitcher-structures, with a ventral portion formed by 

 the infolding and fusion of the basal margins of the 

 dorsal portion, or of those of lateral segments. He 

 has shown, from Payer's data on the development of 

 the carpels in Malachium (Oaryophyllacese), that in 

 this plant the carpel-margins fuse with each other 

 and with the central axis, the loculi assuming gradu- 

 ally the form of deep pockets ; the axis extends some 

 little way above the united carpelrmargins ; two rows 

 of ovules occur, developing basipetally, along the axis 

 in the neighbourhood of the carpel-margins, and also 

 some little distance above them. On account of their 

 position and basipetal development they are ovules and 

 not independent leaves ; hence the whole central axis, 

 not only so far as the upper margins of the carpels, but 

 a short way beyond also, must be clothed with the 

 ventral portion of the carpellary pitcher; and the 

 ventral portion in this case extends at first higher 

 than the dorsal portion. Van Tieghem has shown that 

 the foliar nature of this part of the central axis is also 

 indicated by the reversed orientation of the bundles 

 which belong to the leaf-margins. Celakovsky further 

 points out that we have a right to assume, in those 

 cases where the carpel-margins of a syncarpous 

 ovary do not turn inwards or fuse with the axis, that 



