490 STUDIES IN FOSSIL BOTANY 



arc of secondary wood on the outer side of the primary 

 strand ; the spiral elements lie near the limit of primary 

 and secondary wood, so it appears that the former was 

 wholly or mainly centripetal in its development. 



The parenchymatous cortex, which contains secretory 

 sacs, or canals, is coated externally by periderm ; this 

 may account for the absence of any hypodermal fibres, 

 which had very probably been lost by exfoliation. 



The branches, at their base, had a normal vascular 

 ring, but as they became free, gradually assumed the 

 complex structure of the main stem. 1 



The stems just described, Cycadoxylon robustum, 

 C. Fremyi, and Ptychoxylon Levyi, may be grouped 

 under the family-name of Cycadoxyleae, a designation 

 already used by M. Renault, but in a somewhat more 

 extended sense. 2 The interpretation of their structure, 

 which, in agreement with the views of Professor Seward, 3 

 I desire to suggest, is that they may have been derived 

 from some form resembling Lyginodendron oldkamium, 

 from which they have deviated in two principal respects. 

 On the one hand, they have gradually lost the primary 

 centripetal wood of the vascular bundles in the stem. 

 In Cycadoxylon robustum, this tissue has not been 

 recognised with certainty, and must at most have been 

 relatively unimportant ; in C. Fremyi it was certainly 

 on the verge of extinction ; in Ptychoxylon it had 



1 For a full description of Cycadoxylon and Ptychoxylon, see Renault, 

 Bassin Houiller et Permien d'Autun et a'Epinac, Flore fossile, Part ii. pp. 

 307-21, 1896. 



2 M. Renault includes Medullosa and Colpoxylon in the Cycadoxyleae. 

 On the view here taken, these two genera belong to a different, though 

 related, line of descent, as has already been shown. 



3 " Contribution to our Knowledge of Lyginodendron" Annals of 

 Botany, vol. xi. 1897. 



