6so STUDIES IN FOSSIL BOTANY 



extends to the Bennettiteae of the Mesozoic. On the 

 whole of the evidence available it appears likely that 

 some at least of the Cycadophyta may have been 

 derived from plants resembling the Medulloseae 

 (Neuropterideae) in certain characters, though probably 

 with monostelic rather than polystelic structure of the 

 stem. 



Leaving the Cycadophyta for the moment, we have 

 now to consider the relation of the Pteridosperms to 

 the Cordaitales, the characteristic Gymnosperms of the 

 Palaeozoic. In some respects two groups of plants 

 could scarcely appear more different than the Fern-like 

 Spermophyta and the typical Cordaiteae. In habit there 

 is no resemblance, the Cordaitean trees having externally 

 much more in common with Araucarian Conifers than 

 with any of the known Pteridosperms (pp. 520, 522). 

 The specialised cones or catkins of the Cordaiteae, 

 grouped in inflorescences, are wholly remote from the 

 compound, scarcely differentiated sporophylls, springing 

 from the main stem, which bore the seeds and pollen- 

 sacs in the Pteridosperms. The Cordaiteae are alto- 

 gether on a far higher level of organisation, and deserve 

 the place among true Gymnosperms which has always 

 been assigned to them. But, though so advanced in 

 structure, the family appears to be as ancient as any 

 land-plants with which we are acquainted — a striking 

 proof of the immense extent of evolution which had 

 run its course before our palseontological record begins. 



There are, however, clear proofs of affinity between 

 Pteridosperms and Cordaitales, great as the differences 

 between them appear. The strongest mark of affinity 



