CYCADACE.E. 17 



particular specimen showed a lamina on either side of the basal 

 part of the rachis, having the appearance of several pinnae fused 

 together laterally, the position of each segment being indicated 

 by a strong vein. 



In every classification which is based on artificial characters, 

 and which gives us provisional genera, there must necessarily 

 be inconsistencies, and in all probability plants possessing no close 

 relationship will often be included in the same genus. Among 

 fossil ferns this is especially the case ; as regards cycads, although 

 not perhaps to an equal extent, there are the same difficulties 

 to be encountered owing to the isolated and fragmentary nature 

 of the specimens on which determinations are based. It may, 

 perhaps, be possible to add to the convenience of classification, 

 or to minimise the danger of conveying wrong impressions by 

 ill-chosen names, by adopting some more admittedly provisional 

 classification than is at present employed. An attempt to modify 

 our present system, which is too often inadequate and unsatis- 

 factory, will be more appropriately undertaken after the "Wealden 

 and Jurassic genera have been subjected to a detailed treatment. 

 For the present, attention may be drawn to some of the obstacles 

 in the way of accurate determination of fossil fronds. 



As regards the manner of attachment of pinnae to the rachis : 

 among recent genera there are some in which the pinnae are 

 readily detached from the rachis by a well-marked line of articu- 

 lation ; e.g. in such forms as Zamia furfuracea, Ait., with broad 

 oval pinnae, and other species of the same genus. In species of 

 Encephalartos, Ceratozamia, Dioon, etc., there are distinct and 

 sharply defined scars left on the axis of the frond on the fall 

 of the pinnae ; in others, again, the pinnae are persistent. Among 

 fossil forms, the rachis scars and detached pinnae with clearly cut 

 bases evidently point to a deciduous habit ; but it is often a matter 

 of great difficulty to decide definitely as to the existence of such 

 a character, a,nd it is quite unsafe to trust to a feature of this 

 kind as an essential character in generic classification. It is by 

 no means easy in some cases to distinguish the true auriculate 

 base of a pinna, from a cordate form produced by the crushing 

 and flattening of a thick and leathery segment. Eornemann has 

 called attention to this possible source of error, and points to 

 the absence of any true auriculate base in the pinnae of recent 

 fronds. In examining herbaria specimens of some Encephalartos 



