xix. 7 MESOZOIC MAMMALS 545 



there was also a quadrate-articular joint. This fossil has been appro- 

 priately named *Diarthrognathus. It shows that it is not possible to 

 find an absolutely rigid definition of a 'mammal', even by means of 

 the condition of the jaw. We shall classify this fossil as a mammal. 



Study of the synapsids, mainly from the Karroo system, shows us, 

 therefore, a series of types, of which the earliest were very like the 

 first reptiles and the latest very like true mammals. There can be no 

 doubt of the general tendency, but the series is not complete enough 

 to enable us to follow the details of the evolution of the populations. 

 These fossils are revealed by denudation and can be given only 

 approximate dates. It is certain that some of the mammal-like features 

 appeared independently in lines whose evolution proceeded separately 

 from a common ancestor. Thus the dicynodonts and later theriodonts 

 all had the mammalian phalangeal numbers, but each of these lines 

 has certainly evolved independently from pelycosaur-like ancestors 

 having a greater number of phalanges. 



The influences that produced the evolution of these populations 

 must have been quite complex, since they did not affect all parts of the 

 body at once. For instance, some early therapsids, in spite of their 

 mammalian phalangeal formula, still showed pelycosaur features in 

 the absence of a secondary palate, and presence of a single occipital 

 condyle and small dentary. If the presence of a squamo-dentary 

 articulation is taken as the criterion of 'a mammal' this condition was 

 almost certainly reached independently by several different lines (see 

 Simpson, 1959). We still know too little to be able to specify clearly 

 the conditions controlling such evolutionary changes, but it seems 

 possible that the gradual appearance of terrestrial life and of large 

 herbivores led various animals of a suitable structure and disposition 

 to a carnivorous life. For this purpose certain changes of the ancestral 

 structure would be suitable, leading to parallel evolution in related 

 stocks. However, at present we can hardly do more than pose questions 

 about such matters and resolve to be rigorous in interpretation of the 

 available evidence. 



7. Mammals from the Trias to the Cretaceous 



The types classified as synapsid reptiles, which we have been con- 

 sidering, are not found later than the early Jurassic, 170 million years 

 ago ; mammals of approximately the modern type appear in the late 

 Cretaceous. For the enormous time of more than 90 million years 

 between these dates the mammalian organization maintained itself 

 in the form mostly of small insectivorous animals, perhaps arboreal 



