THE FOSSIL OSMUNDACE^. 655 



trace as a whole becomes wider. While these changes are going on the single proto- 

 xylem group spreads itself out over the inner margin of the island or of the bay (figs. 

 22 and 24) and eventually divides into two separate groups (figs. 25-27). The actual 

 division is usually delayed until the island has opened out into a bay, although it may 

 take place before this happens. A few of the centripetal tracheae immediately in 

 contact with the actual protoxylem elements may still persist even after the bay has 

 opened out, but these soon disappear and the protoxylems become truly endarch 

 (fig. 28). After this stage is reached the xylem strand of the leaf-trace gradually takes 

 up the form of a low arch with slightly incurved ends (fig. 29), and the protoxylems 

 divide again until four or five distinct groups are found projecting from the adaxial 

 concavity of the arch (figs. 28 and 29). All these changes take place while the leaf- 

 trace is still in the inner cortex and its outline as a whole is still oblong or elliptic. 

 Just as it is about to enter the sclerotic outer cortex the endodermis and ground-tissue 

 invades the concavity of the bay and the trace becomes reniform (fig. 29). Once in 

 the sclerotic cortex it rapidly takes up the horseshoe-shape outline characteristic of the 

 Osmundacese, and the protoxylems increase to ten or more (fig. l). 



The great majority of the leaf- traces conform exactly with the above description. 

 Three traces, however, were found in which the single protoxylem group was markedly 

 excentric in position (figs. 30 and 31) ; this, however, seems to be of no particular im- 

 portance, and the traces opened out in the usual way (fig. 32). A much more inter- 

 esting variation was exhibited by four other traces, which may lead to conclusions of 

 some importance. In these the single mesarch protoxylem had divided into two separate 

 and distinct groups before any parenchyma whatever had appeared in the xylem of the 

 trace (figs. 33-35). Indeed, in one case the protoxylem was already double before the 

 xylem of the leaf-trace was yet free from the endodermis of the stem stele (fig. 35). 

 The other cases occurred in traces that were well out in the inner cortex. The two 

 protoxylem groups may have arisen by the early division of a single protoxylem, or 

 they may run down independently into the xylem of the stem. Our specimens give no 

 means of settling this point, nor do they give any information as to the behaviour of 

 these leaf-traces in their further course out into the petiole. 



The Structure of the Petiole. 



The leaf- trace is accompanied out through the sclerotic cortex of the stem by a 

 sheath of thin- walled parenchyma which has mostly become decayed (fig. 36, i. par.) with 

 the exception of a number of cells forming an irregular and much interrupted zone in 

 close proximity to the trace (figs. 36 and 38, res. I.). These cells only differ from the rest of 

 the parenchyma by their more resistant walls and by the amount of black carbonaceous 

 matter they contain, indicating that they possessed dense and specialised contents. 

 Similar elements also surround the leaf-trace while it is still in the inner cortex (fig. 1). 

 In this region, at any rate, they are not sclerotic, although they may become so in the 



