652 STUDIES IN FOSSIL BOTANY 



is not, of course, a chronological one, but it serves to 

 link the stem of Pteridosperms of the Lyginodendreae 

 type with that of true Cordaiteae ; it is at least an inter- 

 esting point that the older Cordaitean stems generally 

 show some trace of the centripetal wood, while it seems 

 to have disappeared in the late Carboniferous or 

 Permian species which Renault investigated. Con- 

 currently with the gradual extinction of the old Crypto- 

 gamic wood, we find on the whole a tendency to 

 greater density of the secondary wood, with a diminu- 

 tion in the width of the medullary rays. The double 

 leaf-trace is a common, if not a constant, character at 

 all stages, the division of the trace extending, on the 

 whole, lower down into the stem in the later forms. 

 Other details, such as the structure of the outer cortex, 

 are also common to many members of the series, from 

 the Devonian Calamopityeae to the Permian Cordaiteae. 

 Without for a moment supposing that we have here 

 the actual course of evolution before us, the series 

 seems to afford as strong a proof as anatomical evidence 

 is capable of affording that the Cordaiteae sprang from 

 a Pteridospermous stock, while the leaf-structure sup- 

 ports this conclusion, the mesarch or exarch foliar 

 bundles of Poroxyleae and Cordaiteae being a distinctly 

 Pteridospermous character. On the evidence of the 

 seed-structure and the anatomy together, the affinity 

 of the Cordaitales with the Pteridosperms seems to be 

 firmly established, though in point of time the connec- 

 tion must lie very far back. 



The various Cycad-like characters which have long 

 been remarked in both vegetative and reproductive organs 

 of the Cordaiteae are doubtless to be explained, not 



