West. — A Contribution to the Study of the Marattiaceae, 405 
less upright caudex of these species rests upon the soil (often consisting 
of loose forest humus), the greater part of its weight is supported by 
the roots, which apparently function in much the same way as the so- 
called prop roots of certain mangrove plants, e. g. Rhizophora. In this 
connexion it should be remembered that the stem of all known genera of 
Marattiaceae is characterized by the complete absence of sclerenchyma. 1 
On the other hand, the main roots of Danaea alata , which is a much 
less bulky plant with a horizontal rhizome (PI. XXI, Fig. 2), differ from those 
described above in the much feebler development of the peripheral zone of 
stereome (cf. Text-fig. 33, A, with 33, b). Whereas in the roots of Danaea 
alata this zone is frequently only a single layer of cells in thickness, in the 
roots of Danaea nodosa and of Danaea simplicifolia as many as six layers 
were observed. A corresponding difference was noticed by Campbell (20, 
p. 179) between the roots of Danaea jamaicensis (= dorsiventral species) 
and of Danaea elliptica (= radial species) respectively. 
If we except the ‘ stone ’-cells which occur sparingly in the cortex of 
roots of Angiopteris , this peripheral zone of stereome is entirely wanting 
from the main roots of all the other known genera of Marattiaceae (Text- 
fig. 33, c), the adult plants of which possess either a strictly dorsiventral 
rhizome (e. g. Kaidfussia ) or a massive conical stem (e. g. Angiopteris , 
Macroglossum , and Marattia). In both cases the entire weight of the plant 
is borne by the substratum, the stem therefore requiring anchorage only ; 
this need is well supplied by the main roots, which act as stays, their 
anatomical structure being well fitted for withstanding a pulling strain. 
Discussion. 
Two main points of theoretical interest emerge from the statements 
found in the preceding pages. Firstly, the question as to whether the 
radial or the dorsiventral type of symmetry is primitive in the Marattiaceae, 
and secondly, the question as to whether the results derived from a 
comparative study of the morphological, anatomical, and histological 
characters in this group of Ferns give any clear indication of the probable 
affinities of this family. 
We will now briefly consider the first question. According to Campbell 
(20), the organization of the embryonic plant (excepting Kaidfussia ) is 
from the very beginning of its development strictly radial, the vertical 
growth of the young sporeling being initiated by the primary segmentation 
(or basal) wall of the. fertilized ovum ; this wall in all genera of Marattia- 
ceae is always transverse, with the result that the first leaf, or cotyledon, 
arising from the half of the embryo which is turned away from the 
1 ‘ Stone ’-cells are sometimes found near the periphery of the cortex of the rhizome of Danaea 
alata. 
