74 ANATOMY OF THE GYMNOSPERMS 



The Gingkoales and the Conifcrales as a whole exhibit an 

 obviously higher type of development than the preceding group, 

 in consequence of the more pronounced tendency to segrega- 

 tion of the pits, which are now either elliptical or round, and 

 never hexagonal. This distinction is so clearly defined and con- 

 stant as to support the idea, which gains force in other ways, 

 that Cordaites, Araucaria, and Dammara are clearly related 

 members of a principal branch of the original stock, and that 

 they therefore diverge considerably from the particular line 

 of descent within which we find both the Gingkoales and the 

 Coniferales. 



Gingko, being the unique representative of an ancient line, 

 cannot very well be brought into the present discussion very 

 much in detail. On other grounds it is known to be a primitive 

 form representing a group distinctly inferior to the Coniferales, 

 and this view is supported by the disposition of the pits in two 

 series, a character which, if taken alone, would give the genus 

 rank with Torreya taxifolia among the Taxaceae, but when re- 

 garded collectively would place the genus distinctly below the Co- 

 niferales as a whole. This evidence, then, indicates that the 

 Gingkoales must have arisen as a side line at some point inferior 

 to the Coniferales but superior to the Cordaitales. 



In the Taxaceae the bordered pits do not in themselves afford 

 very conclusive evidence as to the relative position of the family. 

 Among the eight investigated species, representative of three 

 genera, only three, and chiefly two, variants occur. Taken alone, 

 the disposition of the pits would lead to no final conclusion, but 

 other factors permit of placing this family in the inferior posi- 

 tion usually assigned to it. In the genus three variants are 

 found, --the one-to-two rows of T. taxifolia, the one row or 

 pairs of T. californica, and the strictly one-seriate form of T. 

 nucifera. In Taxus only two variants appear, --the one row 

 or pairs of T. floridana and the one-seriate disposition as found 

 in the remaining three species. The one representative of Podo- 

 carpus shows but one variant, and that is one-seriate. From 

 this it is obvious that the generic sequence must be in the order 



