54 BIOLOGICAL LECTURES. 



analogy, we have here at least fragments, and sometimes very 

 extensive ones, of the various literatures which register the 

 changes of language, and in the original documents which bear 

 evidence of their dates and succession, and which, however 

 incomplete, have not been falsified by forgeries and late inter- 

 polations. In this way we may establish unequivocally some, 

 at least, of the animal pedigrees, which it is one of the great 

 objects of morphology to construct, and thus to correct the 

 results obtained by the other methods of inquiry. Palaeon- 

 tology further enables us accurately to discriminate between 

 resemblances which are due to genetic affinity and those which 

 result from parallelism or convergence. 



To illustrate : On grounds of comparative anatomy Flower 

 classified the land Carnivora into three sections : the Cynoidea, 

 or dogs ; the Arctoidea, containing the bears, raccoons, and 

 mustelines ; and the Aeluroidea, including the civets, hyenas, 

 and cats. This classification has found wide favor and very 

 general acceptance, but palaeontology shows it to be untenable. 

 The extinct phyla show that the dogs and bears are very closely 

 akin, as are the mustelines, civets, and hyenas, while the cats 

 occupy a very isolated position, and are not closely allied to 

 any of the other families. The anatomical characters which 

 suggested Flower's system are in part examples of conver- 

 gence and in part due to the retention of primitive characters 

 in some groups and their loss in others. 



Again, reasoning from embryological data, Rose and others 

 have propounded the theory that the complex, multicuspidate, 

 mammalian tooth has been formed by the coalescence of many 

 simple teeth. The phyletic series enable us to follow the 

 evolution of these teeth step by step, and demonstrate the 

 incorrectness of the " concrescence theory." In fact, the 

 great lesson which the study of the phyla continually brings 

 home to the observer is, that trustworthy results are to be 

 obtained only by the laborious and minute tracing of the 

 changes through every step of the way. Fragmentary series 

 are not to be depended upon, and the wider the gaps between 

 their members the more uncertain is their connection. 



(2) The reconstruction of pedigrees, the solving of homol- 



