I IO 
A. G. Tansley. 
progress from the protostelic through the solenostelic to the dictyo- 
stelic condition, as well as from the sorus with relatively few large 
sporangia, developed simultaneously, and each containing many 
spores, to the sorus with many small sporangia, developed 
successively, and each containing few spores. But we do not get, 
nor should we expect, an exact parallelism in the evolution of these 
two sets of organs. The factors that influence the one set do not 
directly influence the other, and consequently we find such anomalies 
as the genus Lindsay a, with a vascular cylinder not far removed from 
the protostelic type, among the Mixtm, and forms like Aneimia and 
Mohria, with dictyostelic vascular cylinders, among the Simplices. 
There is, I think, no doubt that in most cases the soral and 
sporangial characters are a better guide to the natural affinities of 
a fern than is the general construction of the vascular system. 
Thus for instance within the protostelic genus Gleichenia we have a 
solenostelic species, G. pectinata ; within the very natural family 
Schizmaceae we have protostelic, solenostelic and dictyostelic forms. 
On the other hand anatomical characters such as those under 
consideration have undoubtedly remained relatively constant for 
long periods through considerable cycles of affinity, though they 
may change within quite narrow groups, sometimes obviously in 
response to changes in the habit of the plant. The determination 
of the exact relations of anatomical characters to changes of habit, 
and the resulting separation of the relatively variable from the 
relatively permanent features of internal structure is the most 
interesting and important task which confronts the modern com¬ 
parative anatomist. It is, however, a task with which comparatively 
little progress has yet been made, and till we know more in this 
direction no general authoritative statement on the relation of 
anatomy to taxonomy is possible. 
We may, however, say this much. The internal structure of a 
plant has a semi-independent course of evolution of its own, which 
can undoubtedly be traced. The right course from our present 
standpoint is to follow where this seems to lead, whether the 
changes we have to pursue have taken place rapidly in response to 
changes of habit, or slowly in response to other factors which are 
still largely obscure. 
The Hymcnophyllaceae are a very well-marked family of mainly 
tropical ferns, whose predominant vegetative characteristic is the 
“filmy” habit, i.e„ the possession of leaf-blades without a lacunar 
