XXXIX] COMPARISON 73 



(Vol. II, p. 280) that a relation exists between Plagiogyria and Osimmda, so 

 that the comparison of Sir W. Hooker of Llavea with Osmunda now receives 

 a tardy though indirect corroboration. These genera appear to form a nexus 

 of Ferns that suggests an origin of a primitive Gymnogrammoid type, with 

 superficial sori throughout, quite independent of the Dicksonioid origin of 

 the Pteroids, in which the phyletic origin of the sorus has been shown to be 

 marginal. 



The relatively primitive character of those Gymnogrammoids above 

 named appears in features that are distributed rather arbitrarily among them. 

 The shoots are creeping or ascending: mostly with scales on the rhizomes 

 and leaf-bases {Llavea, Cryptogramme, Ceratopteris), but sometimes with 

 hairs only {Jajuesoiiia). The vascular system of the adult is based upon a 

 type of solenostele which is not far removed from a typical one in Llavea 

 and Jamesoiiia, though in Ceratopteris there is an elaborate dictyostele, with 

 medullary strands also in the sappy axis. The leaf-trace of Llavea and of 

 Ciyptogrannnc comes off as a single strap, but in the latter it shows an in- 

 dication of a median xylic break that probably corresponds to the median 

 fission seen in Jamesonia and in Gymnogramme. But Ceratopteris h^s a highly 

 disintegrated leaf-trace, with medullary complications: this accords further 

 with its reticulate venation, while the others have forked veins with free 

 vein-endings. Thus anatomically Llavea, Cryptogravime ^.nA Jamesonia, and 

 in particular Llavea, are relatively primitive, while Ceratopteris is advanced. 

 In respect of dermal appendages /(tw^^j-c'w/rt is peculiar in this whole affinity 

 in having hairs only, thus comparing with /'/^^/c'^rM with which anatomically 

 both Llavea and Jamesonia have much in common. 



It is in respect of the sori and sporangia that the most important comparisons 

 appear. In all these Ferns various ages of the sporangia are intermixed, and 

 they share the absence of any true indusium, while their sporangia, seated 

 individually rather than grouped in definite sori, are distributed along the 

 veins, and are large and pear-shaped or approximately spherical, but with 

 short stalks. In Ceratopteris a section of the stalk may present a rosette of five 

 or six cells, comparable with that of Plagiogyria (Vol. I, Fig. 243, vi), and with 

 other Ferns that are relatively primitive. But in Llavea d^nd Jamesonia there 

 is the usual three-rowed stalk. Cryptogramvie cinspa, however, gives varying 

 results between a three-rowed stalk and one with six rows. The most striking 

 sporangial feature is the inconstancy of the annulus, which all these genera 

 share. This is most marked in Ceratopteris, where normally the annulus con- 

 sists of an unusually large number of indurated cells. It is as a rule interrupted 

 at the insertion of the stalk, but Cryptogramine gives indications of obliquity 

 of position, even with continuity of the series of cells past the stalk-insertion. 

 In all the four genera there is this inconstancy of structure of the annulus, it 

 is not merely exceptional : \x\ Jajuesonia Thompson found 150 irregular out of 



