MATTHEW, CLIMATE AND EVOLUTION 231 



and the alliance of the African to the South American genera and of 

 the New and Old World porcupines must be regarded as more remote 

 than it appears. Dispersal from South America by help of Antarctic 

 or transatlantic land-bridges will not solve the problems of their distri- 

 bution much better. The most specialized porcupines in most respects 

 are the hystricids of the Old World — late Tertiary in Europe, now chiefly 

 Oriental and African. The Nearctic porcupines (E retliizon) are more 

 advanced in several features than the Neotropical (Synetheres). Yet 

 the ancestors of the New World porcupines at least occur in the late 

 Tertiary of South America and are absent or unrecorded from the Ter- 

 tiary of North America. The distribution of the Oetodontidse in Africa 

 and South America would possibly admit of being interpreted by parallel 

 development from theridomyid ancestors; but the parallelism must have 

 been singularly close, and the absence or non-recognition of Therido- 

 myidas from the North American Tertiaries appears surprising. I have 

 been unable to frame any hypothesis which will fit all the facts of distri- 

 bution in this group,*^ except by assuming that the South American 

 Hystricomorpha, which as Scott has shown are all clearly derived from 

 a single stock, reached South America from Africa in the Oligocene by 

 over-sea raft transportation. This involves so long a voyage that I hesi- 

 tate to accept it as a reasonable probability, even though the winds and 

 currents would obviously favor transportation in this direction. 



The Hystricidse may fairly be assumed as of Old World origin, and 

 probably Palaearctic, since they are represented in the later Tertiary of 

 Europe and are unknown in the New World. The Erethizontidae must 

 apparently be derived from South America, since they are unknown in 

 the Old World, and unknown in the North American Tertiary, while 

 Steiromys of the Patagonian Miocene appears to be ancestral. 



The primary type of the simplicidentate rodents, as I have elsewhere 

 shown,*^ must be regarded as being represented by the Ischyromyidse of 

 the American and European Eocene, in particular by Paramys and Sciu- 

 ravus. All other rodents may be derived from this group by divergent, 

 parallel and in some respects convergent evolution. Modern rodents rep- 

 resent a great number of independent derivations from this primary 

 stock, their association into sections and families being to a considerable 

 extent artificial. 



■'■' The hypothesis of migration to or from South America across a land-bridge from 

 Africa to Brazil is equally unsatisfactory as an attempt to explain the relations of the 

 hystrlcomorph families and is entirely at variance with tlie evolution and distribution of 

 other mammalian orders, besides being highly improbable on isostatic grounds. The 

 supposed evidence in Its favor from lower vertebrates and invertebrates is due, so far as 

 I have been able to examine it, to a laclc of appreciation of the principles of dispersal 

 of races and of parallelism and of tlie imperfection of the geological record. 



■•^ "Osteology and Relationship of Paramys and Affinities of the Ischyromyidie," Bull. 

 Am. Mus. Nat. Hist., vol. xxviii, p. 43-71. 1010. 



