516 THE CAT. [CHAP. xv. 



the specially marsupial character, u the pouch," is wanting. The 

 same is the case in the allied genus Phascogale, while in most of the 

 small American opossums (Didelphys), the pouch is not developed. 

 The character is still a very variable one in many forms of the order 

 as if it had not become even now a well-established character. In 

 the predatory opossum the so-called Tasmanian wolf (Thylacimis), 

 " the marsupial structure, if shown at all, is represented by a pair 

 of shallow, semi-lunar fossae, with their concave outlets opposite 

 each other as in Echidna." * 



It is the very highly specialized Australian kangaroos and pha- 

 langers forms which may be relatively modern developments 

 which have the pouch most completely formed and which may be 

 considered the typical representatives of marsupial life. 



Moreover the secondary fossils, Ampliitliorium, Amphilestes, Dro- 

 matherium and Phascolotherium, had, as we have seen,f but six 

 incisors in the mandible, while only in Phascolotherium was the 

 mandibular angle inflected. All then that the yet discovered meso- 

 zoic fossils can be held to demonstrate is, that there existed at the 

 time of their entombment, forms having both placental and didel- 

 phous aflinities and which may have been some of the as yet 

 undifferentiated ancestors whence those two now divergent sub- 

 classes of mammals have descended. 



And it is far from impossible that some existing marsupials may have 

 come from a different root from that which gave rise to others. Forms 

 may have grown alike from different origins, as few things are more 

 certain in the matter of development, than that similar structures 

 often arise independently, and causes which would induce marsupial 

 modifications in the descendants of one root-form might well also 

 induce them in those of another root-form. The singular difference 

 in the structure of the hind-paw in the more typical marsupials 

 the kangaroos and phalangers from that which is found in Didelphys, 

 Dasynrus, Phascogale and HyrmecoUm, seem to point to a twofold 

 origin of the modern order Marsupialia. The hypothesis then 

 which represents the most ancient mammals to have been allied to 

 the Insectivora, is one which appears to me best to accord with all 

 the facts yet known. Thus we may account for the low brain 

 structure and defective palate of Arctocyon; for the form of the 

 molars of Proviverra and its allies ; for the form of the astragalus 

 noted by Professor Cope ; for the number of the inferior incisors ; 

 for the rare inflection of the mandibular angle and for the existence 

 of a complete milk dentition. 



From this Insectivorous root then, the Marsupials, as we at 

 present know them, must have diverged as a relatively unimportant 

 branch, while the main stem of the mammalian tree was continued 

 on by the successively arising placental forms of life. It seems far 

 more likely that the allantois came to atrophy and the pouch to be 

 developed, and that so the modern marsupial structure was initiated ; 



* So says Professor Owen, Philoso- t See ante, p. 504. 

 phical Transactions, 1865, p. 676. | 



