Adaptation in the Tristichacece and Podostemacece. 549 



ordinary meaning of the term. When we come to consider what this really 

 means, we are met by great difficulties. 



There is one, and apparently only one, or at most two, permanent 

 deflecting factors, that act now, and have always acted since the evolution of 

 the families began, when their first ancestors took to life in running water. 

 These are, first and foremost, their plagiotropic method of growth, forced 

 upon them by the fact that they live only upon an unyielding substratum ; 

 they have not, and can never have had, primary roots going downwards into 

 the rock, and are thus, one might almost say, cut in half, or deprived of 

 one-half of their polarity. The other factor is the longitudinal strain of the 

 water, which to a large extent acts in the same general direction. 



Now there is also one, and apparently only one, or at most two, permanent 

 results visible in the evolution of the members of these families, viz., the 

 plagiotropic or dorsiventral habit both of the growing organs and the 

 structure of the flower, which is by far the most striking feature in them ; 

 added to this is the considerable elongation and division of the leaves (or 

 root-thalli) in some of the forms. Their differentiation into genera and 

 species depends to a large extent upon these two features, which are merely 

 expressions of increasing dorsiventrality, and increasing size and division of 

 leaves (or root-thalli). 



Now, when one sees in a family where the complexity of the ordinary 

 evolution process is so much simplified as we have seen it to be here, one 

 principal and one subsidiary deflecting factor, and one principal and one 

 subsidiary result, it is very difficult to avoid the conclusion that these are 

 cause and effect. But it must be again insisted upon, and clearly understood, 

 that there has been no selection of advantageous variations. All the forms, 

 whether slightly or highly dorsiventral, can live in practically all the 

 localities affected by the families, so that the increased dorsiventrality of 

 the more modified forms is of no advantage to its possessors. The least 

 modified forms are the most common and widespread (and this by the way 

 appears to be the general rule in other families, as I have already pointed 

 out), while the much modified ones are local. 



This being so, we are almost forced to believe that a definite deflecting 

 factor can make evolution go more or less in a definite direction, without 

 regard to any advantages to be gained by it, though any modification that 

 was seriously disadvantageous would be weeded out by natural selection. 



But now comes a serious difficulty. If we have to admit that 

 evolution can take place without natural selection, and that there is no 

 selection of advantageous variations, why do we (so far as present evidence 

 goes) not get growing together all the stages in the evolution of the very 



