Willis .— The Origin of the Tristichaceae and Podostemaceae . 303 
which water plants would find it very difficult to retain a foothold in 
the rapidly moving water. Reaches such as these do not, whether in 
temperate or tropical countries, offer satisfactory places for growth to any 
water plants, and it is extremely rare to find such growing in them. In any 
case, such plants would not be likely to have creeping roots developing 
secondary shoots ; rather they would probably have deeply growing roots 
to try to get a foothold. 
Assuming, however, the improbable thing that such plants did exist, 
it is very remarkable that they have died out and left no trace, for even if 
we allow Natural Selection a large part in evolution, there can be no com¬ 
petition between them and the Podostemaceae on the rocks, and there 
is now nothing in the intermediate reaches. Their habit would probably 
be quite different from that of the Podostemaceae, and to adopt the mode of 
life that characterizes the latter they would have to undergo great changes 
in their morphology. Land plants, on the other hand, would require a great 
change in their anatomy. The question is a difficult one, but it seems 
to me that a change in morphology, such as would likely be required here, 
from deeply rooting plants to creeping-rooted plants with secondary shoots 
upon the roots, would be a larger change than the loss of the rigid anatomy, 
which we know may disappear to some extent in a shoot that happens 
to live in water. Taking this together with the absence of any water plants 
in the reaches, I think we may pretty safely say that the ancestor did not 
live there and send seeds with the current to stick upon the rocks, while it 
would find it almost impossible to creep on to the rocks on account of 
the rush of water and sand. 
We are thus reduced to the first of our suppositions, that the immediate 
ancestors of these plants were land plants, a supposition which from any 
point of view is the most reasonable. It is necessary to suppose that these 
plants grew on the bank of the river at the rapids, on account of the difficulty 
about the seeds which we have already considered, but the rapids where 
these plants grow are always surrounded by vegetation. 
Land plants living on the edge of the stream could very well put out 
feelers, so to speak, in the form of adventitious roots, on which, as happens 
in many families, secondary shoots might arise. These shoots might very 
easily from the first, as is evident from the behaviour of Littorella and other 
amphibious plants, be able to go through life in water, and as the water 
is sufficiently aerated would never need to develop any large intercellular 
spaces. Thus the only adaptation that they would require for a start would 
be to be able to hold firmly to the rock. This could be easily enough 
accomplished by the development of numerous root-hairs. 
The important point here is that such ‘ experimental ’ shoots might 
form year after year in considerable numbers without in any way endanger¬ 
ing the existence of the parent form, until at last the necessary ‘ adaptive ’ 
