288 BOTANICAL GAZETTE [october 



J 



'f 



t 



lar region unseptate and containing free eggs; G. Ulay G. ovalifolium, 

 and G, verrucosum with an embryo sac containing free eggs. Cupres- 

 sus Goveniana is probably near the Tumboa level. 



It seems that during the time the female gametophyte is passing 

 from the scattered archegonium condition either to the archegonium 

 complex or to the free-egg condition there will necessarily be a reduc- 

 tion in the number of male nuclei, because of the impossibility of 

 more than one functioning as we now understand the meaning of the 

 term function. When either the complex or the partially free-egg stage 

 is reached, there can be no further reduction in the number of male j 



cells remaining. From this it follows that the attainment of an advanced 

 stage by one structure may result in the preservation of primitive con- 

 ditions in another structure intimately connected with the highly 

 specialized one. Research in forms showing some highly specialized 

 structures is likely to uncover primitive conditions, which we are now 

 too prone to dismiss lightly as "interesting abnormalities/' but which 

 are in reality stages marking progress. 



The division of the fertilized egg to produce eight proembryonal 

 nuclei is a common habit of gymnosperms. The failure of three to 



F 



six of these free cells in £. trijiirca is suggestive of the next stage, the 

 direct production of a single embryo from the fertihzed egg, which is 

 characteristic of Tumboa, Gnetum, and angiosperms. That the 

 old habit of producing several embryos is occasionally retained by 

 Gnetum is shown by an observation by Lotsy, who says that occasion- 

 ally, as the suspensor is pushing downward, branching of the embryo 

 may occur. 



The sporophyte characters of the Gnetales are so well known that 

 it is needless to discuss them. 



A comparison of all the characters of Gnetales seems to show that 

 we have a fairly consistent group, the most primitive being Ephedra, 

 the most advanced Gnetum, with Tumboa between, but perhaps 

 nearer to Gnetum than to Ephedra. That the three are phylogeneti- 

 cally related seems certain. 



There is no proof that the Gnetales have been derived from or are 

 directly related to any living group. They have many characters of 

 Coniferales, and if related to Coniferales at all it must be to the 

 Taxaceae with which thev have manv things in common. Gnetales 



» 



