110 Bulletin American Museum of Natural History. [Vol. XXVII, 



fail except in regard to minor details, but the cheek teeth correctly indicate 

 that Tiianotherium and Metamynodon belong to very distinct families of 

 Perissodactyls, — a conclusion which is supported by much additional 

 evidence. 



(/;) The fore arm and feet of Arsinoltherium (Andrews, 1906) from the 

 lower Oligocene of Egypt are remarkably similar to those of Proboscideans, 

 the hind limb and foot are full of resemblances to the heavy bodied Ambly- 

 poda; but the cheek teeth may be a development of the Hyracoid type. 

 In this instance it is proljable that we have a mingling of hereditary and 

 homoplastic resemblances, and that Arsinoltherium is actually related, 

 more or less distantly, to all three orders, Amblypoda, Proboscidea and 

 Hyracoidea. 



Both teeth and feet often inferior in interordincd value to the car pals, and 

 especialhj to the tarsals. — In brief it has been shown that in certain cases the 

 cheek teeth are misleading guides to the relationships between orders, and 

 in other cases the number of digits and the general characters of the manus 

 and pes are efjually misleading guides. But in perhaps the majority of 

 instances, and with the exceptions noted above (p. 109) tJie detailed charac- 

 ters of the carpals, and still more of the tarsals, especially the astragahts, yield 

 more reliable indications of ordinal relationships than the characters of the 

 cheek teeth. The fuller evidence for this view is given below, especially in 

 Chapter XI under the headings "Marsupialia," " Insectivora, " "Rodentia" 

 and "Ungulates." 



Each case to be judged on its oivn merits. — While the carpals and tarsals 

 are in the majority of instances reliable guides to ordinal affinities, it must be 

 understood that no single characters or set of characters are always reliable; 

 that, as Linnaeus discovered (p. 34), a character may be of great value to 

 the systematist in one order and of very little value in another; that in short 

 every case must be judged on its own merits. 



Why primitive characters survive in the carpals and tarsals. — Why is it 

 that the carpals and tarsals, in their detailed characters, correctly indicate 

 remote relationship (ordinal or interordinal) more often than do the cheek 

 teeth? Does this fact indicate: (a) that food-habits and tooth-structure 

 have in the cases cited changed more rapidly than locomotive habits and foot- 

 structure ? 



Does it indicate: (b) that changes in food-habits and tooth-structure have 

 a tvider range of possibilities than changes in locomotive habits and foot- 

 structure ? 



Does it indicate: (c) that the teeth are more subject to orthogenetically 

 divergent tendencies than are the carpals and tarsals ? 



Whatever may be the true explanation of these facts the following con- 

 siderations should not be lost sight of. 



