58 Bower.—Studies in the Phytogeny of the Fi tic ales. VII. 
before the two lateral straps have themselves disengaged from the axial 
stele. As they pass up into the petiole they maintain their state of two 
equal straps. Such division of the leaf-trace occurs occasionally among the 
Pterids, as in Lonchitis and Pteris ere tic a ; and it is common among the 
Gymnogramminae. It is not a very trustworthy basis for exact comparison. 
There is much sclerenchyma in isolated gritty bands, which makes the 
cutting of sections from mature stems difficult. This again is an unreliable 
character for comparison. 
The sorus has been depicted by Christ. 1 It is without indusium, and 
is borne at or very near to the slightly enlarged end of a vein (Fig. 43). 
The sporangia are almost simultaneous in origin, and the annulus interrupted 
at the stalk. The spores are tetrahedral. These characters, together with 
habit, anatomy, and hairs, justify the position assigned by Diels in proximity 
to Dennstaediia . It is probably a derivative along a line parallel to that of 
Hypolepis and P. punctatum. 
PiiYLETic Conclusions. 
All of the Ferns considered in this memoir have marginal sori, either 
actually so or indirectly by descent. Comparison shows that they illustrate 
in various degrees and in divers ways a tendency of the sorus to shift from 
the actual margin to a superficial position. The most gradual steps in this 
change of position can be traced in nearly related forms amongst them. The 
biological advantage gained by it in the protection of the sorus while young 
needs no insistence. 
The ultimate origin of them all has probably been from some source 
such as is represented by the Schizaeaceae of the present day. Lygodium 
already shows protective growths covering its isolated marginal sporangia, 
which clearly prefigure the double indusium. If we imagine in each indusial 
pocket of Lygodium a plurality of marginal sporangia, the result will be very 
like the first state of the sorus of Thyrsopteris or oiDicksojiiap If a gradate 
sequence of sporangia were produced upon the same receptacle a condition 
would be attained not unlike what is actually seen in later stages of the sorus 
of a Dicksonioid Fern. Thus the Dicksonioid sorus may be traced from 
a Schizaeoid source, while the sporangial structure also offers palpable 
analogies. The anatomy as well as the sporangia of the Dicksonioids 
show some advance on the Schizaeoid type as seen in Lygodium ; but both 
have still dermal hairs and no scales. Thus a comparison suggests 
a relation, though not a near one, of the Dicksonioid F'erns to the 
Schizaeaceae. 
A main line of descent within the Dicksonioid affinity, involving 
progressive modifications, is indicated by the sequence of genera Thyrso- 
2 Land Flora, Figs. 329, 330. 
1 1. c., p. 76. 
