1899] ON THE ANATOMY OF FOSSIL PLANTS 367 



occurrence in Lyginodendron is regarded by the authors as merely an 

 occasional variation of no specific value. This point is taken as a peg 

 on which to hang a few general remarks on the indiscriminate use of 

 anatomical characters. They point out that — " The anomaly in 

 question is known to be of very inconstant occurrence at the present 

 day. Both in the genera Tecoma and lodes some specimens show it 

 and others do not, though analogous peculiarities (internal phloem for 

 example), are often characteristic of entire natural orders. Anatomical 

 characters, in fact, like any other characters, are sometimes of 

 great constancy, sometimes highly variable, while the same character 

 which is relatively constant in one family may be most inconstant in 

 other groups." 1 They further point out that — " In the present paper, 

 we have been compelled, in the absence of organs of fructification, to 

 make great use of anatomical characters. We have, however, endea- 

 voured to rely on those which are known to be of great persistency in 

 families which presumably belong to the same cycle of affinities as the 

 plants with which we are dealing." 2 



The leaf-traces, which pass out from the stele of the stem so very 

 gradually as to have led Williamson, at first, to think they were cauline 

 bundles, have been shown to be continuous with the peri-mednllary strands 

 (primary bundles) around the pith, and many sections show these bundles 

 passing through the secondary wood on their way out to the leaves. 



It is a rather remarkable fact that, in all the specimens of 

 Lyginodendron- which have hitherto been examined, no trace has been 

 seen of branching. Of course the evidence is not conclusive, but not a 

 single branching specimen has yet been detected, since all the supposed 

 branches which have been described have turned out to be either 

 petioles or adventitious roots. 



With regard to the petioles of Lyginodendron, Williamson long ago 

 surmised that these were identical with his Raehwpteris aspera, and the 

 discovery of the leaflets in connection with the branched petiole led to 

 the conclusion that the leaf can be referred to Brongniart's form-genus 

 Sphenopteris. We therefore have here the combination of cycad- 

 like stem and distinctly fern-like leaves. But the authors point out 

 that something more is needed to establish filicinean affinities : " The 

 classical case of Stangcria is a sufficient warning against any such hasty 

 inference. It must, however, be remembered that in the foliage of 

 Lyginodendron we have not only fern-like form and venation, but also 

 fern-like structure, whereas in the case of Stangeria a single transverse 

 section of the petiole would be sufficient to prove that the plant is no 

 fern, but a cycad." 3 A striking example this of the importance of a 

 knowledge of the internal structure of plants in these investigations ! 

 The bundles in the petiole are the typical concentric ones of a fern, 

 while the continuations of the same bundles in the stem are collateral, 

 and have a distinctly cycadean character. 



1 Loc. tit. ]>. 722. - Loc. tit. pp. 7^2-723. :; Loc. tit. p. 727. 



