THE GRAPTOLITES AND CORALS 287 



graptus and afterwards a Tetragraptus stage ; and he imay 

 ask whether this does not contradict all that has been 

 deduced from other animal phyla as to the relation of 

 ontogeny to phylogeny. 



To this it must be replied, in the first place, that in the 

 graptolites we know very little of actual ontogeny. .The 

 development of the sicula is ontogeny ; all that follows 

 is not the development of an individual, but of a colony, 

 by budding astogeny. However, students of other 

 colonial animals fas the Bryozoa) have shown grounds 

 for the view that astogeny is also, like ontogeny, a re- 

 capitulation of phylogeny. Apparently, then, the grapto- 

 lites contradict this conclusion. But we must not stop 

 short at the Tremadoc graptolites in tracing their 

 descent. It is inconceivable that these complexly 

 branched forms should have been derived abruptly from 

 non-colonial ancestors. They must have had inter- 

 mediate ancestors (quite unknown) in which the branch- 

 ing gradually became more and more complex. The 

 known graptolites of this family must illustrate the cata 

 genetic stages of branching from an acme in the 

 Tremadoc epoch. 



However the extent of the " genera " so far mentioned 

 may have to be altered, they appear to form a natural 

 family the Dichograptidcv, of which the five series men- 

 tioned would be sub-families. The family is distinguished 

 by the simple structure of its thecae, and by the fact that 

 the first-formed theca? always grow downwards from the 

 sicula. It ranges from Tremadocian to Llandeilian. 

 Another " genus '' which should find a place here is 



