Embryology of certain of the Lower Fishes. 211 



less importance than they appear to be at first sight; for 

 we must believe that in degenerative evolution the char- 

 acters which disappear most easily are those which have 

 been most recently acquired, and those which tend to persist 

 longest are those which are most ancient and most deeply 

 engrained into the constitution of the creature. Degenera- 

 tion will therefore tend to be a process of reversion, and 

 it is hardly profitable to inquire whether the apparently 

 primitive features of, say, such forms as Cyclostomes or 

 Cephalochordata are there because they have been retained 

 throughout, or because they have been reverted to after 

 passing through a more highly specialised condition. 



It is, however, of interest to inquire whether there are 

 really strong grounds for believing that the protocercal 

 Dipnoans of to-day have ever had heterocercal ancestors. 

 In considering any such phylogenetic question, it is essential 

 to consider it from the three points of view of comparative 

 anatomy, embryology, and palaeontology,-^ or at least as 



^ The worker in comparative anatomy or embryology is accustomed to bear 

 constantly in mind the danger of mistaking features which are adaptations, 

 or correlated with adaptive features, for features which are archaic and of 

 phylogenetic importance. In palaeontological work, the risk is perhaps 

 greater of accepting results without due criticism, of imagining that allied 

 forms found in successive formations necessarily form an ancestral series. 

 The expert palaeontologist will doubtless be one of the first to admit how 

 extraordinarily imperfect our knowledge really is of the fauna of any geological 

 age except the present, and how unreliable are such phylogenetic series, 

 unless supported by comparative anatomy or embryology, or preferably by 

 both. The special risks of error in embryological research are keenly realised, 

 particularly by embryologists, who consequently endeavour to take every pre- 

 caution against falling into them. Onetypeof error to which embryology appears 

 not to be liable, is that which has to do with the ordei' in which evolutionary 

 stages occur. If an organism or an organ recapitulates certain phylogenetic 

 stages, say A, B, C, D, they are found to occur in their true phylogenetic order. 

 In palaeontology, however, so imperfect is our knowledge of the geological 

 record, that it may easily happen that in a series of successive deposits 

 we find the same stages in the order D, C, B, A. This may be due either 

 to mere chance, all four stages being equally abundant during each period 

 under consideration, or, on the other hand, it may have been that the group 

 reached its maximum in the earlier period, and evolved into highly specialised 

 forms, intimately adapted to the environmental conditions of the time, Avhich 

 died out as these conditions changed, sooner or later, according as they were 

 more or less specialised. 



