262 GENERA INCERTAE SEDIS [ch. 



sori, which are covered by a cowl-hke sheath. The lobe undergoes an early 

 branching, each branch ending in a sorus. Thus each sorus represents an 

 abbreviated leaf-segment resulting from a branching of the second order. 

 Each is protected by its own indusium, but both are covered also by the 

 cowl, which springs from the upper lobe. The terminal position is in fact 

 the same as in Salvinia (Fig. 754, iii, iv). 



The sporangia of the Salviniaceae have no annulus : the number of spore- 

 mother-cells as shown by the microsporangia is 16, but in the megasporangia 

 it is only 8, while only a single megaspore comes to maturity (see Vol. I, 

 p. 267). 



From such facts it is difficult to attach the Salviniaceae definitely to any 

 Family of living Ferns. It seems inevitable that they sprang from a type 

 that was homosporous, but the absence of an annulus removes one of the 

 most reliable features for comparison, while the segmentation of the sporan- 

 gium and the exact number^ of the spore-mother-cells give general rather 

 than precise indications of relationship. The basipetal succession of the 

 sporangia upon the elongated receptacle, and the basal indusium, together 

 with the distal position of the sori upon the leaf-segments, constitute Ae 

 best grounds for comparison. These all point towards the gradate Marginales 

 in general, and if to any one Family of them, probably to the Hymenophyl- 

 laceae. Provisionally this relation may be regarded as a possibility: but in 

 the absence of forms connecting these highly specialised aquatics with other 

 Filical types the question of a more exact relation must be left open. 



Finally, Seward remarks that "there is no evidence contributed by fossil 

 records which indicates a high antiquity for the Hydropterideae. It is unsafe 

 to base any conclusion on the absence of undoubted Palaeozoic representa- 

 tives of this group; but the almost complete absence of records in pre-Tertiary 

 strata is a fact which may be allowed some weight in regard to the possibh 

 evolution of the heterosporous Filicales at a comparatively late period in 

 the earth's history" {Fossil Plants, No\. II, p. 477). 



The examples of Ferns of uncertain affinity, brought under discussion in 

 this Chapter, may serve to indicate some of the difficulties that beset any 

 attempt at a complete grouping of this complicated Class. The list oi'' genera 

 incertae sedis" might be greatly extended: the result would still be the 

 same, viz., while determining with reasonable probability the relationship 

 of some, to leave others definitely unattached. But the difficulties thus ap- 

 parent in Ferns are probably less grave than those encountered in a similar 

 attempt for other Classes of Plants. These habitually offer less distinctive 

 features for comparison than do the Ferns, and they are still more deficient in 

 palaeontological history. The chief source of uncertainty in the construction 



