482 Boodle.—Anatomy of die Hymenophyllaceae. 
type resembles T. reniforme , the more bulky forms with a 
solid stele have died out. 
Or 2. That in Trichomanes the structure of T. reniforme is 
primitive and that specialization went on in two directions, 
namely reduction of conducting elements, leading to col¬ 
lateral structure in T. mnscoides , and increase in conducting 
elements, leading to types with a large stele, e. g. T. radicans , 
T. scandens , &c.; while in Hymenophyllnm , starting with 
a structure like that of T. reniforme , as found in the larger 
forms, specialization took place only in the direction of re¬ 
duction to the sub-collateral type. 
Or 3. That the collateral type is primitive (as held by 
Prantl), and specialization involved complication and increase 
in the size of the stele in both genera 1 . 
Of these, the second seems in some ways the most probable. 
The type of structure found in T. reniforme and in the 
similar species of Hymenophyllum is evidently specialized. The 
somewhat dorsiventral differentiation of the xylem, with a 
lower band developed early for the attachment of roots, is an 
adaptation to the creeping habit. The forms with upright 
stems do not appear to form a series by themselves, but 
species nearly allied to one another may differ in that the 
stem is upright in one and creeping in another, and this 
may occur in different sections of the genus 2 . Among the 
dimorphic forms of Trichomanes , T. spicatum and T. hetero- 
phyllum differ in this way. 
Though the Hymenophyllaceae may have been derived 
originally from ancestors with upright stems, it is quite possible 
that all the living members of the order may have been 
evolved from forms with a creeping rhizome, some species 
retaining this character, and others having changed to upright 
habit, others possibly having progressed from creeping to up¬ 
right habit and back again 3 . When the upright habit leads, 
as it usually does in this order, to short internodes and 
a slower-growing stem, the protoxylem will be liable to 
1 Prantl, 1 . c., p. 59, &c. 2 Prantl, 1 . c., p. 58. 
3 This is also Prantl’s view. 
