638 GENERAL COMPARISON OF THE FILICALES 
so as to follow roughly the probable phyletic sequence, the massive 
sporangium of the Marattiaceae has its archesporium deeply sunk: the 
walls all cut at right angles, and since the outer surface is but slightly 
convex, the walls are almost parallel, and the archesporial cell approximately 
cubical (Fig. 349 g). The segmentation in the Osmundaceae is variable, 
and it has been observed to be so even in sporangia on the same plant 
in Zodea barbara. In some cases the archesporium is still square-based, 
and may be square in its transverse section: but as the outer surface 
becomes early convex, the lateral walls converge, and the archesporium 
has the form of a four-sided truncated pyramid (Fig. 349 /). In other 
sporangia of the same plant the lateral walls limiting the archesporium 
converge more strongly, the outer surface being more convex, and one of 
them inserts itself upon another: consequently the archesporium takes the 
form of a three-sided pyramid (Fig. 349 ¢). It is to be noted that in 
this figure the wall («, x) is inserted on an inner periclinal; but in 
Fig. 349 @, which represents the segmentation in Schizaea, or in Tricho- 
manes or Thyrsopieris, the wall (x, x) cuts another anticlinal. This marks 
another step in attenuation of the sporangium, though only a slight one, 
and in other essentials the segmentation is as in the simplest of the 
sporangia of Osmunda or Todea. Figs. 349 6, ¢ show the segmentation 
seen in various Gradatae: (c) corresponds to the condition of <Adsophila 
and Cyathea, and (8) is a slight variant upon it which is sometimes found: 
it is seen also in Ceratopterts. In the Polypodiaceae, however, where the 
sporangium may often be long-stalked, the wall cut by the wall (x, x) may 
be no longer inclined, but transverse. From this series of diagrams it is 
seen how gradual are the steps from the segmentation typical of the 
Eusporangiate Fern to that of the most advanced Leptosporangiate. The 
unity of the scheme cannot naturally be divided by any distinction of 
origin from a single cell or from more. The difference of type thus gently 
graded over is an index of the progressive attenuation of the sporangium 
seen in descent, and it will be shown to go along with progressive reduction 
of the individual productiveness. 
Closely related to the segmentation of the young sporangium is the 
structure of its sfa/e when mature. Putting synangia aside, the stalk 
varies from the short massive type of Angiopierts, through various types 
such as the Osmundaceae, Gleicheniaceae, Schizaeaceae, and Hymeno- 
phyllaceae with relatively thick stalks, to the Polypodiaceae, where the 
stalk is commonly attenuated and long. It may, in extreme cases, be 
reduced to a unicellular filament, as in Scolopendrium. These steps again 
show a general parallelism with the individual spore-output, the thickness 
of the stalk being roughly proportional to the stream of nourishment 
required. 
But it is the sporangia! head, with its relatively thin wall surrounding 
the cavity filled with spores, that is the most distinctive part, and as the 
character of the opening mechanism, or annulus, has been made the chief 
