SPORE-PRODUCING MEMBERS 633 
At first sight it might seem probable that the non-soral state was 
primitive, since it is seen apparently in such early forms as Botryopéeris 
and AMyriotheca, and in Osmunda. But there are objections to this as a 
generalisation ; for, in the first place, definitely soral types, such as the 
Marattiaceae, are quite as well represented in the primary rocks as any 
non-soral forms: secondly, while Botryopieris itself appears to be non-soral, 
there is a distinct indication of a disposition of the sporangia around a 
central point in Zygopieris (Fig. 272, p. 503), while in Corynepteris there 
are very definite sori: thirdly, there are among living Ferns clear indica- 
tions that the soral may pass into the non-soral state: such a progression is 
suggested in the species of Dzpteris (p. 620), while the condition of Acrost- 
chum and of Platycerium can hardly have been produced in any other way 
than by spreading of the sporangia of some soral type over an enlarged 
surface, as is indeed suggested by such genera as Gymnogramme and 
Hemionitts, etc. This is exactly what would be expected as a consequence of 
indefinite multiplication of parts closely aggregated together, in cases where 
no biological check determined their exact position. Thus it would seem 
probable that the soral state is the original condition and the non-soral 
the derivative, notwithstanding its early appearance. 
But the sorus, whether marginal or superficial, does not always maintain 
its identity, even in those cases where it is habitually circumscribed. In 
many Ferns, and especially in those in which there is an enlarged leaf-area, 
sori may be found of unusual size, elongated, and constricted in the middle ; 
and from these it is a slight step to complete fission, two smaller sori being 
then seated close together (Figs. 281, 310). In such changes from the normal 
as these there lies a capacity for increase in number of sori, and there is 
hardly room for doubt that in such cases as Kaudfussia and Dipteris, as 
well as in many of the broader-leaved Polypodiaceae, where the sori con- 
stitute more than a single row on either side of the midrib, the more complex 
condition has been brought about in this way; in fact, the statement seems 
fully justified that the primitive disposition of the sori was in a single 
marginal or intra-marginal row: all more complex arrangements in Ferns 
are secondary and derivative. 
A further matter for discussion is the position which the sorus holds 
relative to the leaf which bears it. Two positions are common, marginal 
and superficial, the latter almost always on the lower surface of the leaf: 
both of these are of very early occurrence, the superficial being characteristic 
of the Marattiaceae and Gleicheniaceae, and the marginal of the Botryo- 
pterideae and Schizaeaceae, while both types are continued upwards into the 
Gradatae and Mixtae. It would be important to know which of these 
positions was the more primitive in Ferns. Here, again, an indirect 
indication may be obtained by comparison of more recent types: among 
the Leptosporangiate Ferns there is ample evidence to show that the 
marginal sorus has shifted by gradual steps to the lower surface. This 
is clearly proved by comparison within the Dennstaedtia-Davalia series: 
