252 Flowers and their Pedigrees. 



on the spike between the scattered pistils. This 

 result is just what you might naturally expect from 

 squeezing a lot of marsh calla blossoms closely 

 together on a spike. Even in the upper half of the 

 spike, the blossoms often keep up some marks of their 

 original bisexual character, for you will occasionally 

 find a few stray green knobs sparsely sprinkled here 

 and there among the golden stamens of the top portion. 

 Nevertheless, we may fairly say that even here a ten- 

 dency towards specialisation has been distinctly set 

 up : the uppermost flowers tend to become almost 

 entirely pollen bearing sacs, and the lowermost 

 flowers tend to become preponderatingly, though 

 not entirely, seed-bearing ovaries. 



Now if we turn from these transitional steps to the 

 completely developed arum, what do we find ? Here, 

 the top of the spike has become absolutely bald and 

 bare of flowers, instead of being covered, as in the 

 Ethiopian lily, with thickly grouped florets up to 

 its very summit ; and at the same time, the actual 

 flowers in the lower portion, instead of running to- 

 gether into an uninterrupted cone, are separated into 

 three distinct groups or bodies. At the bottom of all, 

 as in the .Ethiopian lily, we now get the female 

 flowers alone ; only, instead of being intermixed with 

 stamens, they consist simply of naked ovaries ; the 



