134 THE VASCULAR SYSTEM OF THE AXIS [CH. 



endodermis without phloem (Fig. 126). The occurrence of the internal phloem 

 is local in the region of a bifurcation, where a "ramulargap" is more or less 

 completely formed. The detail of the gap varies within the genus. In 0. 

 daytoniana there are no such gaps at all : in O. regalis and T. barbara the 

 gaps extend to the xylem only: but in O. cinnamomea the gaps vary, being 

 in some cases complete, the outer endodermis connecting at the gap with 

 the inner, so that there is free communication between the cortex and the 

 pith: the phloem also extends from the outside of the xylem through the 

 xylic gap, and is continuous as a sheet of internal phloem for some distance 

 down and up the stem. Here again the presence of the inner endodermis, 

 and of the inner phloem as well, may find its elucidation in terms of local 

 change of destination of the procambial cells, there being no evidence of any 

 intrusive flow of tissues from without to explain it. 



Such local change may be regarded as a still more advanced stage similar 

 to, though more far-reaching than, that seen in Platyzovia and Scliizaea. 

 Where the complete ramular gap exists the conducting tract is delimited 

 by endodermis on both sides, while the ventilating systems of pith and 

 cortex are placed in communication, a state not arrived at in any other 

 living Osmundaceae than in Osmunda cinnamomea. 



In the Cretaceous fossil Osmundites skiegatensis ventilation throughout 

 the stem was still more fully achieved than in the above cases. The stock of 

 this plant must have been much larger than that of any living member of 

 the family. It had a stele 24 mm. in diameter, with a very wide pith sur- 

 rounded by a ring of about 50 separate vascular strands. Each leaf-trace at 

 its departure interrupts the continuity not only of the xylem but of the whole 

 vascular ring. Through each gap the pith becomes perfectly continuous with 

 the cortex. Phloem lies internally to each xylem-mass, and it is continuous 

 through each leaf-gap with the external phloem. No layer resembling endo- 

 dermis can be distinguished in this fossil, so that it is practically impossible 

 to set a definite limit to the stele. Nor is there to be seen any endodermis 

 round the leaf-trace. This large-sized Osmundaceous Fern thus exceeded 

 any living Osmundaceae not only in size, but also in the complexity of its 

 stelar structure. It had actually taken the step to a sort of dictyostely. 



This is more clearly shown in the still larger Paraguayan fossil stem, of 

 late but uncertain horizon, Osmicndites Carnieri, Schuster {Ber. d. D. Bot. 

 Ges. xxix, p. 534; Kidston and Gwynne-Vaughan, Trans. Roy. Soc. Edin. 

 Vol. 1, Part II, p. 476). Here the stele is greatly dilated, being 33 mm. in 

 diameter, that is, about one-third the diameter of the stem. It shows 33 

 distinct xylem-strands of unusual depth. The stele is surrounded by a line 

 of delimitation, believed to be an endodermis, which is interrupted by the 

 departure of each leaf-trace. This leaves a wide leaf-gap (Fig. 127). Thus 

 the vascular ring is divided up into separate meristeles, each containing a 



