98 Annals of the Soutli African Museum. 



occasionally bifurcate. The lamina is curved so as to present a 

 convex surface ; it extends distally into the sandstone, so that the 

 apparently triangular apex does not represent the actual termination 

 but merely the boundary of the exposed part of the specimen. This 

 leaf may be compared also with Euryi^hyllum, as figured by Feist- 

 mantel from India. 



Plate XIII., fig. 2 (XXI.). 



Part of a leaf 12'5 cm. long and 4 cm. broad. The venation is not 

 very clearly shown, but the specimen is of interest on account of its 

 striking resemblance to a leaf of Cordaites. 



Other si^ecimens : Fragments occur on specimens xx., xxi., xxii., 

 xxiv. One clearly preserved impression (xxi. 6) represents a broad 

 linear leaf like that shown in pi. xiii., fig. 2, but with a naiTow basal 

 termination. 



Plate XIII., fig. 3 (XXV.) (1 nat. size). 



A torn leaf 10 cm. long, showing the original form of the lamina. 

 The slightly spreading veins appear to be rather farther apart than 

 those in the specimen shown in fig. 4, but this need not necessarily 

 indicate a specific difference, as some at least of the finer ridges 

 in fig. 4 may represent hypodermal strands of mechanical tissue such 

 as are known to occur in the leaves of Cordaites. 



Plate X., fig. 5 (XX.). 



Part of a leaf showing the veins more clearly. The narrow end of 

 the fragment does not represent the actual apex, as the lamina 

 extends further into the rock. The veins are numerous and more 

 oblique to the margin than in the other figured specimens. 



PLANT OF DOUBTFUL POSITION. 

 A. — Ehizobie or Eggt. ' 



Plate XII., fig. 2 (XX.). 



This imperfect specimen, represented natural size in the drawing, 

 may be a rhizome giving off numerous branched roots, but it is 

 impossible to speak with confidence as to its precise nature. It is 

 perhaps not improbable that the fossil may be part of the rhizome, 



