THE OPOSSUM 53 



their mode of reproduction, are Jinown as "non-marsupial," 

 and " placental." But our short survey has shown us 

 that the marsupial beasts are very diverse in structure. 

 They are so much so that they contain groups which 

 run parallel with various orders of placental beasts, as 

 follows : 



Placentals. Maesupials. 



Eodents (marmot, &c. ). Wombat and its allies. 



Carnivora (cats, weasels, wolf, Native cats, weasels, and Tas- 



&c.). manian wolf. 



Insect-eaters, mole, &c. Phasoogales and bandicoots. 



Hoofed beasts (deer and an- Kangaroos. 



telopes, &c.). 



Edentates (sloths and ant- Native sloth and echidna, 



eaters, &o.). 



Now, according to the doctrine of evolution, all existing 

 species are the descendants of common ancestors, from the 

 structure of which they in various degrees diverge ; and 

 with regard to the origin of these two parallel series of 

 marsupial and placental beasts, two hypotheses are open 

 to us. One is that all beasts were at first of marsupial 

 nature, and that the rodent, carnivorous, insectivorous, 

 hoofed, and edentate placentals, are respectively the modi- 

 fied offspring and descendants of the rodent, carnivorous, 

 insectivorous, kangaroo-like, and edentate marsupials. 

 But on this hypothesis it is absolutely necessary that a 

 number of very similar structures must be affirmed to 

 have arisen independently. Such, in fact, must have 

 been the case with regard to all those structural and 

 functional characters by which the placental mammals 

 agree to differ from all marsupials, since these characters 

 must have similarly and separately arisen in each of 

 these several groups, if we suppose the various groups of 

 placentals to have severally descended independently from 

 antecedent separate sets of marsupial forms. 



