XII LEFTOSPORAXGIAT.E HETEROSPORE.-E 413 



separated by distinct intervals or internodes, and apparently 

 all of the dorsal segments do not give rise to leaves, but just 

 what the relation is between the nodes and internodes was not 

 determined. The roots arise in strictly acropetal order from 

 the ventral segments, but their number does not seem to be 

 constant. In Pilidai'ia Americana the number of roots con- 

 siderably exceeds that of the leaves, as it does in the young 

 sporophyte of P. globulifera. 



The single axial vascular bundle is truly cauline, and 

 extends considerably beyond the base of the youngest leaf. 

 The later leaves in Pzhilaria, both in their growth and complete 

 structure, correspond to the primary ones. They grow for a 

 time from a three-sided apical cell, in which respect they differ 

 from Marsilia. The development of the leaf of the latter has 

 been carefully studied by Hanstein in M. Driimviondii, and M. 

 vestita corresponds exactly with that species. A section of the 

 very young leaf (Fig. 215, C) parallel with the surface shows 

 a large two-sided apical cell. The leaf- rudiment assumes a 

 somewhat spatulate form, and on either side a projecting lobe 

 is formed, the rudiment of one of the lateral segments of the 

 leaf. The apical cell is now divided by a median wall, after 

 which periclinal walls are formed, and from this time the growth 

 of the leaf can no longer be traced to a single initial cell. The 

 first longitudinal wall in the apical cell establishes the two 

 terminal lobes, which at first are not separated (Fig. 215, D). 

 The establishment of the veins follows exactly as in Ferns with 

 a similar venation, and is strictly dichotomous. The stem 

 branches freely in both genera, and the branches arise close to 

 the apex, and below a young leaf somewhat as in Azolla. 



The roots correspond exactly with those of the higher 

 homosporous Ferns. The segmentation of the apical cell 

 follows the same order as in the Polypodiaceae. Goebel's figure 

 of J/, salvatrix^ differs somewhat from the account given more 

 recently by Andrews " for M. quadrifolia. The latter observer 

 states that there are no periclinal walls in the root-cap segments, 

 which remain throughout one-layered, and that the separation 

 of the plerome takes place earlier than Goebel indicates. Van 

 Tieghem's ^ account of the root of J/. Drummondii confirms 

 Andrews' observations upon M. quadrifolia. The bundle of the 

 root is diarch, as in the Polypodiaceae, and the lateral roots arise 

 ^ Goebel (10), p. 238. - Andrews (i). ^ Van Tieghem (5), p. 535. 



