THE EVOLUTION OF STRUCTURES 159 



nuclei became comparatively small. It is to be remem- 

 bered that in free nuclear division in female gameto- 

 phytes and embryos the mechanism for the formation 

 of walls is present. The nuclear divisions simply occur 

 in such rapid succession that the mechanism does not 

 get into operation. 



Although my series of stages in the embryogeny of 

 Cycas, Encephalartos, and Macrozamia are not yet quite 

 complete, it seems safe to say that at the close of the 

 free nuclear period normal cell walls are formed through- 

 out the egg, just as in the well-known Ginkgo, or maiden- 

 hair tree. In Stangeria and Dioon evanescent cell walls 

 appear throughout the egg but almost immediately dis- 

 appear, leaving the nuclei free again, except at the base, 

 where the embryo proper is to be organized. In Zamia 

 and perhaps others there is not even an evanescent 

 formation of walls in the main body of the egg, the only 

 walls coming late and at the base. I have no doubt that 

 three stages in the phylogeny of the cycad embryo are 

 represented in these three illustrations, which may be 

 typified by Cycas, Dioon, and Zamia, with Cycas as the 

 most primitive and Zamia as the most modified type of 

 embryogeny. 



Tracing the evolution of embryogeny is not a simple 

 matter. As far as the free nuclear period is concerned, 

 we beHeve that it has become more and more prolonged 

 as the egg has increased in size, reaching its maximum 

 in eggs like that of Dioon edule, where the egg is usually 

 an eighth of an inch in length, and occasionally reaches 

 a length of more than three-sixteenths of an inch. Here 

 there may be a thousand free nuclei before any walls 

 appear. In Stangeria there are about five hundred 



