42 MORPHOLOGY OF GYMNOSPERMS 



Paleozoic seeds. — Our knowledge of the internal structure of 

 seeds known positively to be those of Cycadofilicales is restricted to 

 the Lagenostoma group; but there is much larger knowledge of 

 detached paleozoic seeds, some of which may well belong to Cycado- 

 filicales. At least it is not out of place to consider the general charac- 

 ters of paleozoic seeds. Numerous detached seeds occur in the 

 Carboniferous, and they have naturally been referred to gymnosperms. 

 Although called seeds, they have been preserved evidently at the stage 

 just preceding or soon after fertilization, and are ovules in the modern 

 sense. Especially in the Permian beds of France are these seeds 

 remarkably well preserved, having been elaborately described by 

 Brongniart (6) and Renault (i6). Seeds from the English beds, 

 usually less well preserved, have been described by Williamson 

 (2, 3). The whole subject, however, has lately been reopened by 

 Oliver (38), in connection with the recent rapid development of our 

 knowledge of gymnosperms. 



Brongniart for convenience referred all paleozoic seeds to two 

 categories, which Oliver has named radiosperms and plaiysperms, 

 the seeds of the former being circular in transverse section or radially 

 symmetrical, those of the latter flattened or bilateral. Some of the 

 platysperms have been referred definitely to the Cordaitales, and the 

 same connection has been inferred for the rest of them; but the seeds 

 of Aneimites and of Pecopteris, described above, suggest that this 

 reference has been too sweeping. The radiosperms have had no 

 such definite assignment, until recently some of them have been found 

 to belong to CycadofiUcales. It is evident that Brongniart's two 

 types hold no relation to the two paleozoic groups of gymnosperms. 



There are certain general features belonging to some of the paleo- 

 zoic radiosperms that deserve special mention in connection with their 

 possible relation to Cycadofilicales. For this purpose the seed genus 

 Stephanospermum, formerly referred to Cordaitales but now probably 

 to be assigned to the MeduUosineae, may be used, both because it 

 is the simplest and because it has been described in great detail by 

 Oliver (47). The seed of S. akenioides (fig. 45) is cylindrical and 

 pointed, the testa being differentiated apparently into an outer stony 

 layer and an inner fleshy layer.' The nucellus is entirely free from 



1 An outer fleshy layer might be expected, but either it has not been preserved 

 or has not been recognized. 



