Further Notes on Seed Structures. 151 



species have been obtained. Both are en do vascular, the inner 

 fibrous flesh being traversed by six bundles in the angles of as 

 many simple ribs in C. dblongum, and four bundles in the 

 angles of the four major or wing ribs which initially alternate 

 with as many minor ribs in the somewhat more complex C. 

 anglo-germanicum. Both species thus agree with Cycadeoidea 

 in size and in being coronate radiospermic ribbed and winged 

 forms with essentially the same wall structure, barring reten- 

 tion of the endosarcal bundles, traces of which are, however, 

 yet likely to be found in the Cycadeoidese. 



Indeed, truly essential testal and apical differences, rather 

 than likenesses, remain to be established now that we know 

 Cycadeoidea to be a characteristically four- and five-ribbed and 

 lobate, winged, tentacled seed which also shows some tendency 

 to alternation of major and minor ribs. 



Evidently Cycadeoidea, though of composite type, has, as 

 befits its Mesozoic extinction, a clear title to admission into the 

 great groups of Paleozoic gymnospermous seed types retaining 

 indisputable evidence of a yet more ancient multiple integu- 

 ment. And this is yet more strongly indicated when we turn 

 to the fuller lobed JVeuropteris heterophylla seeds described as 

 attached to their foliage by Kidston, if, indeed, these are true 

 megaspores. Especially so, if we further consider that most 

 remarkable six-lobed microspore-bearing fruit Codonotheca of 

 quite the same size and outer aspect as the reputed Neuropteris 

 seeds, and on the score of association strongly suspected to be 

 of JSTeuropterid affinity when first described (T). 



Reexamination of this fossil by the writer has so far failed 

 to establish the presence of sporangial structures on the inner 

 surfaces of the campanular lobes where much loose and extra- 

 ordinarily large-sized pollen may be seen, seemingly in quite 

 the position in which it grew.* But while the general appear- 

 ance at first seemed to preclude the drifting in from the out- 

 side of pollen attaching itself to sticky inner surfaces of free 



* These Codonothecan or Neuropterid microspores or prothallate pollen 

 grains are the largest known, even greatly exceeding Stephanospermum, and 

 it remains quite probable that by some modification of Nathorst's collodion 

 method some evidence of extensive prothallial tissue may be found. 



The characteristic make up of the seed wall again appears as in the pre- 

 ceding figure as well as the important additional fact that breaks or gaps in 

 the inner flesh and middle sclerenchyma quite regularly occur near the 

 shoulder and conclusively indicate that these seeds of Cycadeoidea are as 

 truly ribbed tentacular or apically lobate as any members of the Langeno- 

 stoma group. 



[It is probable that these embryos have undergone some shrinkage, though 

 the cell structure is often clear and does not show much evidence of col- 

 lapse. However, various remnants about the sides of the embryos indicate 

 unused parenchyma. The appearance does not entirely reconcile itself with 

 the great size of the proembryo, seen in another species.] 



