GENERAL RESULTS— PTEROPSIDA 651 



is in the seeds, which are essentially of the same type 

 in the two groups — in fact, it is impossible at present 

 to point to any constant distinction between them. 

 Of seeds already known in detail, the closest agree- 

 ment is between the Trigonocarpus group and the 

 Cordaiteae (pp. 452 and 546); they have in common 

 the double vascular system, the drupe-like testa, the 

 form of the pollen -chamber, and probably the free 

 nucellus. On the other hand, they differ in the fact that 

 the Trigonocarpeae are radiospermic, the Cordaiteae 

 platyspermic, but this distinction has no general validity, 

 for we have good evidence that bilateral as well as 

 radial seeds occurred among the Pteridosperms (pp. 

 465 and 467). 



RJiabdocarpus, attributed to Poroxylon by Grand'- 

 Eury, only differs in small details from the seeds of 

 Cordaiteae in the narrower sense (p. 5 1 2). We know 

 nothing as yet of the seeds of the Pityeae, the third 

 family making up our class Cordaitales. 



As regards the anatomy, there is a very complete 

 series leading from the stem of the Lyginodendreae to 

 that of the typical Cordaiteae, as described by Renault. 

 In Calamopitys we find the first signs of the dying out 

 of the centripetal wood in the lower part of the leaf- 

 trace bundle (p. 482). In Poroxylon the same condi- 

 tion recurs, with a general structure much like that of 

 Cordaites itself (p. 506). In the Lower Coal-measures 

 of England, stems otherwise indistinguishable from 

 those of Cordaites have distinct remains of centripetal 

 wood in the stem (p. 526). The Pityeae have likewise 

 reduced and modified their primary xylem, perhaps in 

 relation to a change of function (p. 515). The series 



