INTEI!P-FT,ATIONS OF GTiXFRA. 



Ixi 



a little doTibtful. The fact is tliat, this genus has retained nian^- 

 characters of lioiiseUns, while in practically all the features in 

 ■which it differs from liousetttis it more or less closely approaches 

 to Cynopterus. Whether a genus exhibiting characters of this 

 description ought, in a linear arrangement, to be classed at or near 

 the end of the Rousettine section or as the "opening" genus of 

 the Cynopterine section, must necessarily remain a matter of 

 opinion. 



With regard to the principles for the arrangement of the genera, 

 only a few remarks are required. Putting for a moment J/yo- 

 nycteris and Nyctimene at one side, it would be easy to divide all 

 the remaining nine genera of this section into two groups accord- 

 ing to the number of incisors (|— | or ^— ^ ) ; but. as pointed out 

 elsewhere (p. 649, footnote), an arrangement based primarily on 



Splisenas 



Nyctimene 



[Rousenus 

 section ] 



Fig. IV.— Interrelations of the genera of the Cynopterus section. 



this character would be thoroughly artificial ; it would separate 

 Ptenochirns and Megrn'ops from their closest relative, Cynopterus; 

 Bdlionycteris from its Javan representative, Chironcuv ; and Pen- 

 thttor from its eastern representative, Thoopterns; there is no 

 doubt that the loss of i^ is a character developed independently in 

 difl'erent branches of the section. The diagram above (fig. IV.) is 

 based, therefore, on the following considerations : — .1///oh//W<'/-/<? 



