( 807 ) 



2. that plants on fertile soil produce on the whole more male 

 flowers in proportion to the bisexual ones than plants on less fertile 

 soil. 



3. that the male flowers only appear at a stage in which the 

 plant has grown stronger, that they gradually increase in number 

 as the individual grows stronger and gradually decrease in number 

 again when the plant has passed its highest point of development. 



4. that in each umbel as well as in each urabellule which contains 

 both forms of flowers, the male flowers are preferably found in 

 those places which are most favourable with respect to nutrition. 



It is not difiicult to show that observation does not confirm these 

 four points. 



Let us in the first place consider point 4. 



There can be no doubt that (excepting the just mentioned terminal 

 umbellules and terminal flowers) the peripheral umbellules are more 

 fa\ourably placed with regard to nutrition than the more inwardly 

 situated umbellules, and that in each umbellule the flowers at the 

 circumference also occupy a more favourable position than those in 

 the middle. This is seen not oidy by the inner umbellules being 

 less i-ich in flowers but also in the flowers becoming smaller the 

 further they are distant from tlic periphery; often the central flowers 

 do not reach their normal de\elopment or the' setting of the fruit 

 does not take place. We see here the same with the umbels as with 

 long-dra«Ti inflorescences like those of Capsella Bursa pastoris or 

 Pimm sativum, that namely the last-formed flowers, at the top of 

 the inflorescence, no longer reach their normal development on 

 account of insufficient nutrition. Further every umbellule (not only 

 a mixed one but also a purely hermaphrodite one) allows us to 

 notice that the peripheral flowers are ahead of the central ones in 

 their development. 



And now we see with all Unibeliiferae without exception: 

 that the peripheral umbellules retain their bisexual character longest, 

 that the male flowers always occur first at the centre of the umbel, 

 that where the umbellules are mixed, the number of bisexual 

 flowers always decreases from the periphery to the centre, 



that the iimer umbellules often are already entirely male when 

 the outer ones still contain bisexual flowers, and 



that everywhere, except with Oenanthe fistuhsa, Sanicula euvopaea 

 and Astrantia the marginal flowers in the umbellules are bisexual 

 and the central flowers male'). 



1) I thiDk an explanation may be found for the anomalous behaviour of these 

 three genera. 1 cannot dwell on this point, however, in tliis short communica- 



