756 Bancroft .—^ Xylem Elements of the Pteridophyta. 
of the splits in some cases, when the bars are no longer held together by 
the primary wall; but it seems reasonable to maintain that some disintegra¬ 
tion precedes the separation, which begins before lignification is very far 
advanced. 
It is further suggested that upon the actual amount of breaking down 
of the primary wall between corresponding pairs of bars depends the presence 
or absence of a ‘ split ’ as seen in transverse section. 
If the disintegration is great, the bars will be connected only for a very 
small fraction of their whole width, as in Matonia pectinata (PI. LVI, Fig. 2). 
The corresponding transverse section (Fig. 1) shows the splits wide and clear, 
and it is obvious that in transverse sections it will be comparatively rarely 
that the junction of the bars is cut through—hence the general appearance 
of splits in such a type. If the amount of disintegration is slight, as shown 
in the longitudinal section of Marattia fraxinea (Fig. 4), the connexion 
between the bars extends nearly their whole width, and the prevalence of 
the ‘ non-split 5 appearance as seen in transverse section (Fig. 3) is clearly 
due to the plane of section cutting (in the majority of cases) the regions 
where the bars are joined. The variation in the width of the splits in 
different types may be the result of the different amounts of the raising or 
springing apart of the bars in the pit areas. 
It seems unwise to attempt to classify types under the head of c split ’ 
or ‘ non-split ’, for between the two extremes illustrated by Matonia and 
Marattia there may be any number of cases showing intermediate degrees 
of separation of the thickening bars, representing not only different amounts 
of disintegration of the primary wall, but also different degrees of springing 
apart of the secondary layers, the splits being sometimes wide, as in Matonia 
pectinata (Figs. 1 and 2) and Lygodium scandens (Fig. 11), and sometimes 
narrow, as in Selaginella Wildenovii (Figs. 7 and 8) and Lygodium dichoto - 
mum (Figs. 12 and 13). A transverse section of Aneimiodictyon sp. (petiole) 
showed a type nearest that of Marattia , though not so extreme, while allies 
of this species, Aneimia fraxinifolia (Figs. 16 and 17) and Mohria caffro - 
rum , presented a type almost as extreme as Marattia . Psilotum triquetrum 
(Figs. 9 and 10) and Tmesipteris tannensis (Figs. 14 and 15) showed fairly 
wide splits on the whole, though the relative width of these seemed to vary 
in different parts of the plant. In the species of Osmunda examined, the 
xylem of the rhizome was found to have large splits, while that of the 
petioles exhibited the opposite extreme. This being the case, it is hardly 
surprising that there is little constancy of type among species of the same 
genus or family, though Marattia laxa , M. fraxinea, M. elegans } Angioptems 
evecta , and Danaea sp. (cf. Figs. 3, 4, and 5) were found to approximate to 
the same type. 
Where the connexion between successive pairs of thickening bars is 
narrow it is clear that the mechanical efficiency of the xylem elements will 
