i82 MORPHOLOGY OF GYMNOSPERMS 



that the permo-carboniferous Poroxylon illustrates what must be 

 regarded as a stage theoretically intermediate between the two groups. 

 The general habit and the leaves are those of Gordaitales in all the 

 essential features. In the vascular cylinder of the stem, the secondary 

 wood is that of Cordaitales, but the primary xylem is mesarch as in 

 Lyginodendron. The leaf traces of Poroxylon, leaving the central 

 cylinder in pairs and finally becoming exarch in the leaves, are com- 

 mon to both Cycadofilicales (notably in Medullosa) and Cordaitales. 

 Poroxylon is too recent to serve any phylogenetic purpose except to 

 illustrate an intermediate stage that has persisted. It will be remem- 

 bered, however, that almost identical forms have been discovered 

 (i8) in the Lower Coal-measures of England; and similar forms of 

 still greater age have been recognized. In fact, among the araucari- 

 oxylons (presumably the wood of Cordaitales) of the early Carbon- 

 iferous there occur stems with mesarch primary xylem; and Scott 

 (9, 10) has arranged a series of such stems connecting the Lygino- 

 dendron type with the Cordaites type, some of which cannot be referred 

 to either group without the foliage. 



In view of the great antiquity of Cordaitales and Cycadofilicales, 

 it is perhaps safer to conclude that the similarities and gradations 

 described above are such as might be expected in two phyla diverging 

 from a common stock and living under the same conditions. Whether 

 this common stock is to be regarded as still more ancient Cycado- 

 filicales or not, is a question that probably never will be answered. 

 In any event, it seems clear that the Cycadofilicales, as we know them, 

 retained more of the features of this ancient stock than' did the Cor- 

 daitales; and that long after the Cordaitales became distinct, there 

 began to arise from the Cycadofilicales two divergent branches, the 

 Bennettitales and the Cycadales. At about the same time, the great 

 phylum Coniferales probably began to emerge from the Cordaitales, 

 and perhaps also the less conspicuous phylum Ginkgoales. 



It will be useful to summarize the important resemblances of Cor- 

 daitales to the living groups Cycadales, Coniferales, and Ginkgoales, 

 in order that their significance may be pointed out. This is especially 

 necessary since there is a tendency to use similar characters at random 

 as indicating immediate relationship. For example, the characters 

 common to Cordaitales and Cycadales seem to include those of higher 



