522 W. D. MATTHEW CORRELATION BY FOSSIL VERTEBRATES 



Mastodon is of Tertiary age and did not survive into the Pleistocene, as 

 we have erroneously supposed. 



-f . There are certain groups whose phyletic history is much clearer and 

 less complex than others, and whose evolution has involved greater struc- 

 tural change and more rapid progress toward an extreme of specialization. 

 These will naturally afford better evidence for correlation than more con- 

 servative groups. 



5. As we have generally to deal with fragmentary material, there are 

 some portions that show the structural changes better than others. The 

 cheek teeth of mammals are in general far more characteristic than any 

 other parts. In particular those phyla in which the evolutionary changes 

 in molar and premolar teeth are in the direction of a more complex struc- 

 ture are of most value, as the characteristic construction can be more 

 surely distinguished from parallel adaptations in other races of the same 

 stock or convergent adaptations from other stocks. 



6. Groups of wide geographic dispersal are more useful in correlation 

 than those of more restricted range. 



For all the above reasons the Equidas or fossil horses are the most use- 

 ful of all mammals in Tertiary correlation. Their evolution and dis- 

 persal is a comparatively simple record and has been more thoroughly 

 studied than in any other group. The cheek teeth have a characteristic 

 pattern, highly complex in all the later stages. They have a wide geo- 

 graphic range and a long geologic record and are very common and wide- 

 spread. I am disposed to regard the successive stages of North American 

 Equidaa as affording a standard of correlation for our Tertiary horizons, 

 in the same way that the Ammonites do for the marine Mesozoic horizons. 



Next to the Equidae come the rhinoceroses, proboscideans, and various 

 artiodactyl groups. These are less characteristic in tooth pattern and 

 their evolutionary record is more complex and less clearly understood. 

 Fossil Carnivora are less satisfactory because of their comparative scarcity 

 and because the simple construction of their cheek teeth renders it more 

 difficult to distinguish between parallelism and direct affinity on the evi- 

 dence of fragmentary material. In fossil rodents parallelism is very 

 prevalent and confusing, and there are many primitive survivals to com- 

 plicate the problem. These and other fossil mammals afford corrobora- 

 tive evidence of varying importance. 



Taking the Equidas as a standard, we find that there are none recorded 

 from the Paleocene or earliest Tertiary. None of the Paleocene ungu- 

 lates are directly ancestral to the Eocene horses. 



All Eocene horses are four-toed, with short-crowned teeth. All Oligo- 

 cene horses are three-toed ; with short-crowned teeth. All Miocene horses 



