324 FOUMER LAND COXXKCTIONS. 



We conclude that a mid-Tertiary African connection, which 

 provided a passage for swamp-dwellers, is a minimum estimate 

 of the requirements. Even this leaves unexplained such anomalies 

 as that of Rana mascareniensis which seem to imply a inore 

 recent connection, though only a very imperfect one in view of 

 the absence of characteristic African mammals and birds from 

 Madagascar. 



The greater portion of the paper is devoted to a proof that 

 nearly all the groups of mammals had their centres of dispersal 

 in the northern hemisiDhere. This has to be conceded, but the 

 extension of the principle to lower vertebrates and invertebrates 

 must not be assumed. Matthew attempts to prove it in the case 

 of certain toads — Cystignathidae, Discoglossidae and Pelobatidae, 

 but his distribution data are quite wrong and the map very mis- 

 leading (p. 296). According to the principle, the most primitive 

 families should be ■southern, and their more specialised successors 

 northern : but such is not the case, the southern Cystignathids 

 being less priinitive than the northern Pelobatids. 



(Gr.) The last note I have seen on the problem of the tiger 

 fish is one by Dr. C. R. Eastman," in which he concludes that 

 the ancestry of the modern Characinids may be traced back to 

 Onchosaurus of the Cretaceous. One of the fossil species is 

 found in the Upper Cretaceous of Egypt, and other representa- 

 tives occur in Europe and North America. The identification 

 is based solel.y on the characters of the teeth, which are very 

 distinctive. Mr. Boulenger seemed to favour Eastman's con- 

 clusion, remarking that: " Hydroci/on goliatJi from the Congo 

 has enormous shark- like teeth similar to fossil teeth occurring 

 in the Upper Cretaceous, apparently indicating the existence of 

 Characinidae in that epoch." 



However, this seems to have been a mere guess, for Mr. 

 Eegan has recently informed me that, as shown by Stromer, the 

 said teeth are actually rostral teeth of a sawfish ! 



Evidence for Eegarding South America, Africa and S.E. Asia 

 AS A Single Zoological Region. 



The occurrence of genetically related animals in Africa and 

 South America may arise from various causes, but some former 

 land-connection between the Old and New Worlds must be 

 postulated : thus, the explanations fall into two main groups, 

 some involving a trans-Paeific connection, apparently across 

 Behring Straits, and others a trans-Atlantic connection, either 

 north or south of the Equator. 



Now, considering each case on its merits, we have to admit 

 that various instances of Afro-American affinity seem explicable 

 more or less well by either of the alternatives, and such incon- 

 clusive cases should be reinoved from consideration. T refer to 

 very small and isolated groups of animals where fossil data is 

 completely lacking. These may conceivably be remnants or 

 refugees of formerlv widespread faunas. Thus we eliminate the 

 evidence of the Pipidae toads (Xeno]ms, Hymenochirus and 



