362 DR D. H. SCOTT ON THE 



interesting to find that, while the bundles approaching their exit are so distinctly mesarch 

 in structure, those lower down in their course tend to lose their centripetal wood, so 

 that, in C. beinertiana, most of them become actually endarch. These facts may in- 

 dicate that the Calamopitys group had advanced rather further towards the usual stem- 

 structure of Gymnosperms than was the case with Lyginodendron or even Poroxylon. 

 Another point is the narrow-rayed secondary wood, quite Cordaitean or Araucarian in 

 structure, co-existing with primary characters pointing to the Cycadofllices. The width 

 of the rays, however, varies a good deal within the genus, and sometimes even within a 

 single species (C. annularis) and is clearly a character on which too much stress has 

 been laid by palseobotanists. On the whole, Calamopitys may be regarded as decidedly 

 the most primitive of the three groups dealt with. 



The Pitys stems are known to have belonged to tall branching trees.* We know of 

 no Cycads or Cycadofllices with at all a similar habit, nor is there any evidence that the 

 Coniferse existed at so early a period. The only known family to which these trees 

 could be referred is that of the Cordaitese, leaves of which have been found at a similar 

 horizon.t The species of Pitys differ from stems, traced with certainty to true 

 Cordaitese, in the broad medullary rays, the non-discoid pith (for the slight approach 

 to discoid structure which they exhibit is of doubtful value), and in the presence of the 

 primary xylem-strands. On the whole, I am disposed to regard the genus Pitys as a 

 primitive member of the Cordaitean family, retaining some of the characters of an 

 earlier stock. The mesarch xylem-strands, in spite of their reduced size, and the 

 peculiarities of their arrangement, are evidently comparable to those of Lyginodendron 

 or Calamopitys. Thus the Pitys trees appear to afford a new link, so far as stem- 

 structure is concerned, between the Cycadofllices of the family Lyginodendrese and 

 the true Cordaitese. Such a connection was already indicated by the structure of 

 Poroxylon, but that genus, from its later horizon (Permian), has a less direct bearing 

 on the question of the origin of the Cordaitese. 



Lastly, Dadoxylon Spenceri, with its dense wood and double leaf-traces, appears to 

 stand near the typical Cordaitese (though its pith seems not to have been discoid), and 

 also, as Williamson pointed out, strongly suggests the recent genus Ginkgo, which itself 

 may have Cordaitean affinities.} The primary xylem-strands of this species are small 

 in size and few in number, but though so much reduced, have essentially the same 

 structure as in Lyginodendron. The fossil indicates that in the period of the Upper 

 Carboniferous deposits, stems which in other respects had attained a typically 

 Gymnospermous character, had not quite lost the primitive form of wood, which we can 

 trace back, through the Cycadofllices, to the Ferns. 



* The trunk of the Craigleith tree (P. Withami) found in 1830 was 47 feet in length, and at the top still had a 

 diameter of about 1 \ feet. Witham, Internal Structure, p. 29. 



+ Kidston, " On the various Divisions of the British Carboniferous Kocks," Proc. Boy. Phys. Soc. Edin. } 1894, 

 p. 255. 



% Seward and Gowan, "The Maidenhair Tree, Ginkgo biloba," Ann of Bot., vol. xiv., 1900, pp. 137 and 146, 



