THE SUCCESSION OF ANIMAL FORMS 1 89 



insect have to be worked out by a series of concurrent develop- 

 ments so complex and absolutely incalculable in the aggregate, 

 that the cycles and epicycles of the Ptolemaic astronomy were 

 child's play in comparison, we need not wonder that the com- 

 mon sense of mankind revolts against such fancies, and that we 

 are accused of attempting to construct the universe by methods 

 that would baffle Omnipotence itself, because they are simply 

 absurd. In this aspect of them, indeed, such speculations are 

 necessarily futile, because no mind can grasp all the com- 

 plexities of even any one case, and it is useless to follow out an 

 imaginary line of development which unexplained facts must 

 contradict at every step. This is also, no doubt, the reason 

 why all recent attempts at constructing " Phylogenies " are so 

 changeable, and why no two experts can agree about almost 

 any of them. 



A second aspect in which such speculations are too partial, 

 is in the unwarranted use which they make of analogy. It is 

 not unusual to find such analogies as that between the em- 

 bryonic development of the individual animal and the succes- 

 sion of animals in geological time placed on a level with that 

 reasoning from analogy by which geologists apply modern 

 causes to explain geological formations. No claim could be 

 more unfounded. When the geologist studies ancient lime- 

 stones built up of the remains of corals, and then applies the 

 phenomena of modern coral reefs to explain their origin, he 

 brings the latter to bear on the former by an analogy which in- 

 cludes not merely the apparent results, but the causes at work, 

 and the conditions of their action, and it is on this that the 

 validity of his comparison depends, in so far as it relates to 

 similarity of mode of formation. But when we compare the 

 development of an animal from an embryo cell with the pro- 

 gress of animals in time, though we have a curious analogy as 

 to the steps of the process, the conditions and causes at work 

 are known to be altogether dissimilar, and therefore we have no 



