CHAPTER XVII. 
THE ESTABLISHMENT OF A FREE-LIVING SPOROPHYTE. 
So far the shoot only of the sporophyte has been the subject of discussion : 
it remains to consider the question how the sporophyte, originally dependent 
upon the parent prothallus, became established as a free-living organism 
on the soil. There will be no two opinions which of the principal 
regions of the independent sporophyte, the shoot or the root, was of 
prior existence: it is a necessary outcome of the evolution of the neutral 
generation as sketched above that the shoot was first established, as a 
body dependent on the gametophyte; it carried out primarily the function 
of spore-production, but ultimately also, as we have seen, that of vegeta- 
tive nutrition. The root is essentially an accessory, which made its 
appearance after those earlier steps were past; it arose from its 
primitive state of dependence to an existence free from the parent 
gametophyte. 
Comparison of living plants indicates, however, a probability that the 
initiation of a root-system followed closely upon the adoption of a free- 
living habit: for roots are present in free-living Pteridophytes with very 
few exceptions, and are, as a rule, formed early in the embryology. It 
seems doubtful, even in the few exceptional cases, whether the rootless 
condition is not due to reduction, rather than representative of a primi- 
tive rootless, but free-living sporophyte. Among the Pteridophytes roots 
are absent in the Psilotaceae, also in certain Hymenophyllaceae, and in 
Salvinia: it seems probable that reduction will correctly account for it in 
such specialised forms as the Hymenophyllaceae; and also in Salvinia, 
with its peculiar floating habit: the question in the Psilotaceae is more 
problematical, and their rootless condition may perhaps have been really 
primitive, though in the absence of any knowledge of their embryos there 
is no clear indication that it was so: moreover, their habit is so peculiar 
as to make any conclusion difficult. Rootless Phanerogams also exist, 
but there is no reason to regard them as other than results of relatively 
recent reduction. Accordingly, it may be concluded that there is little 
