492 STUDIES IN FOSSIL BOTANY 



comparatively unimportant peculiarity, had in the Cyca- 

 doxyleae grown to be a constant and marked character. 

 We regard this structure as a new. formation, due to the 

 spreading inwards of cambial divisions through the leaf- 

 trace gaps, as is clearly shown in Lyginodendron itself, 

 and in Ptychoxylon. There is no reason to suppose 

 that the anomalous centripetal wood thus formed had 

 anything to do with the primary and normal centripetal 

 wood of the more primitive forms : the example of 

 Lyginodendron oldhamium shows that the two structures 

 were quite independent. Neither does it seem practi- 

 cable to derive the structure of Cycadoxyleae from a 

 system of Medullosean steles. Scattered medullary 

 bundles are a characteristic anatomical feature in the 

 recent genera Encephalartos and Macrozamia} but con- 

 tinuous zones or extensive arcs of inverted wood and 

 bast do not appear to be known in the pith of undoubted 

 Cycads. Such anomalies, however, as shown by similar 

 cases in many recent Dicotyledons, are extremely 

 variable, and the peculiarity in question cannot hinder 

 us from recognising the essentially Cycadean character 

 of the stems of Cycadoxyleae. 



It must always be remembered that the Cycadophyta 

 were once an extensive and dominant class of plants, 

 and that the few, which have survived to our own time, 

 only give us a very imperfect and partial idea of the 

 range of organisation which the group once exhibited 

 (see Chap. XIII.). 



It is interesting to find that the same Permian beds 

 of Autun, which have yielded the stems of Cycadoxyleae, 



1 See Worsdell, "Anatomy of the stem of Macrozamia," etc., Annals 

 of Botany, vol. x. 1896. "The Structure and Origin of the Cycadaceae," 

 ibid. vol. xx. 1906. 



