438 WILLIS : MORPHOLOGY OF THE PODOSTEMACEÆ 
ovary is much displaced towards the lower side, a pro 
that may be seen in an incipient stage in Mniopsis, &c,, anv^ 
its ovules are completely absent, while those of the upper 
loculus are reduced in number and increased in size, and 
the placenta is developed mainly at the distal end of the 
ovary. In this form dorsiventrality appears at last in the 
very embryo. 
This remarkably interesting series thus seems to show 
clearly that in the whole of this family dorsiventrality is 
more or less strongly marked, and that its impress is first 
shown in the vegetative organs, i.e.^ in the thallus and 
secondary shoots, followed by the bracts, spathe, and flower, 
while in the latter the stamens first and most commonly 
exhibit it, followed by the stigmas, carpels, interior of the 
ovary, fruit, and finally seed or embryo. Always the 
dorsiventrality is most strongly marked in the vegetative 
organs, and is often present in them to a slight or even great 
degree (Lawia) without necessarily showing in the floral 
organs, while when the latter are exceedingly zygomorphic 
the vegetative organs are usually still more so. 
To put the matter in other words, it would appear that the 
dorsiventrality of the flowers, ivhich is the most important 
morphological character in the classification of this order, 
is a direct result of, or in direct correlation with, that of the 
vegetative organs, being greater the greater the dorsiven- 
trality of the latter. 
This conclusion is confirmed by the argument from the 
ecological relations of the flowers. We have seen above 
that the dorsiventrality of the floral organs appears to be of 
no value to them, but if anything a positive disadvantage, 
and that as it becomes more and more marked the flower 
becomes more and more self-fertilizing by aid of the wind 
or by simple contact, while in Podostemon Barberi it is 
cleistogamic. At the same time, too, the flowers stand erect, 
an unusual feature in such zygomorphic forms . 
It seems, then, not unreasonable to conclude that the 
do)‘siventra,lity of the floral organs has Ijeen, so to speak. 
