6 o Bower. — Studies in the Phytogeny of the Filicales. VII. 
so to maturity. But in Cibotium and Dicksonia , though it is marginal 
in origin, it is diverted during development to the lower surface, while the 
upper indusium is more massive than the lower. The inequality is more 
marked in Dennstaedtia and Microlepia , while in Hypolepis the inner 
indusium is absent, or represented only vestigially, as in Polypodium 
punctatum. The upper indusium, however, asserts itself, receives an exten¬ 
sion of vascular tissue from the receptacle, and takes its place in P.puncta¬ 
tum and Monachosomm as an apparent lobe of the lamina. The sorus itself 
then appears to be superficial; but comparison shows that phyletically 
it was marginal, and that the upper indusium, which now appears mar¬ 
ginal, arose originally from the upper surface. Accordingly those last 
named are held as derivative forms. This point comes out still more 
clearly for Davallia , Nephrolepis , and probably also for Oleandra , which 
also will be in this respect far advanced from the original type. 1 
The receptacle , typically marginal in origin, is conical in the rela¬ 
tively primitive forms, such as Thyrsopteris , Dicksonia, and Cibotium , 
where there is a gradate sequence of sporangia. But the gradate order 
is departed from occasionally in Dennstaedtia and Microlepia , while in 
Hypolepis and P. punctatum the receptacle is flattened, and the succes¬ 
sion of sporangia is definitely ‘ mixed ’. 
The same is the case in even more prominent degree in the Daval- 
lioid series, where the mixed sorus is definitely established. 
The sporangium, large and short-stalked in Thyrsopteris , with oblique 
annulus and lateral dehiscence, stamps the fundamental type of the series. 
But Cibotium departs from it at once by lengthening the stalk, 2 while 
in Dennstaedtia and Microlepia this becomes more marked, and the 
annulus is almost completely interrupted at its insertion. 3 In Davallia this 
is fully carried out, and the stalk reduced in thickness and greatly 
lengthened. 
The use of the various criteria thus followed out supports the progres¬ 
sion suggested. The Dicksonioideae are undoubtedly a natural sequence, 
in which the progressive steps can be traced so as to end in types with 
naked superficial sori, as in P. punctatum and Monachosorum , or indusiate 
superficial sori, as in Nephrolepis. In both cases the sori are of the ‘ mixed * 
type with small long-stalked sporangia. But throughout the Dicksonioid 
series the individuality of the sorus is maintained. 
The derivative line of the Pteris series is believed to be related to this 
series, but collateral with it. It is based upon a natural consequence of the 
‘ webbing ’ of the leaf, i. e. the lateral fusion of the originally separate 
Sphenopterid segments to form a flattened leaf-surface, traversed by 
numerous veins. In this condition the sori, still marginal in origin as in the 
1 See Studies, III. Ann. of Bot., vol.xxvii, pp. 461-3, where developmental evidence is given. 
2 Land Flora, Fig. 330. 3 1 . c., Fig. 332 bis. 
