458 BOTANICAL GAZETTE [December 



Discussion 



The presence in the bast of Gnetum of companion cells which 

 are in most respects of the angiospermic type is at first sight 

 another striking indication of real relationship between Gnetum 

 and angiosperms. They are to be compared with the vessels 

 of the wood, broad rays, general habit, style (4), absence of arche- 

 gonia, free-nucleate embryo sac, endosperm formation (4), and 

 other reproductive characters. In regard to all of these points 

 Gnetum is angiospermic. 



The study of the development of the companion cells, however, 

 shows that the resemblance does not necessarily indicate genetic 

 relationship. Whereas the companion cells of angiosperms are 

 formed from the same row of cambial cells as are the sieve tubes, 

 and each one is contiguous to a sieve tube in such a row, those of 

 Gnetum are formed in rows quite separate from the sieve tubes 

 and are the products of different cambial cells. Of course, it is 

 possible that the companion cells of Gnetum are really genetically 

 related to those of angiosperms, and that, after originating in 

 Gnetum in the method described, their formation has later been 

 taken over in angiosperms by the same cambial cells that form sieve 

 tubes. On the other hand, it seems more logical to conclude that 

 we are dealing with a case of parallel evolution, just as I have 

 shown to be true with regard to the vessels of the wood (5). The 

 vessel of Gnetum with a single large perforation in the end wall 

 is almost identical with that of many angiosperms. Nevertheless, 

 it has been evolved in an entirely different way. The perforation 

 of the angiospermic porous vessel has resulted from the breaking 

 down and disappearance of scalariform bars, whereas that of 

 Gnetum has resulted from the enlargement of typical, circular, 

 haphazardly arranged, bordered pits accompanied by the dis- 

 appearance of the middle lamellae, the enlargement proceeding 

 until th^ intervening portions of the vessel wall have vanished. 

 In the case of the companion cell we seem to have a close parallel 

 to that of the vessel. While the completed structure is similar to 

 that of angiosperms, the course of its development is quite different. 



If two such striking points of resemblance as vessels in the 

 wood and companion cells in the bast are really results of inde- 



