( 25 ) 



In the north-western species the relative shape and propor- 

 tions of the chief markings on the upper surface are very 

 similar to each other and to the white markings of the Mada- 

 gascar saclava. In comorarum, however, the band of the hind- 

 wing is considerably narrower, and is not prolonged on to the 

 fore-wing as it is in all the other species of this group. This 

 extension on the fore- wing is largest in saclava and mayottensis, 

 smaller in dumetorum, and still" smaller but quite distinct in 

 frohenia. In the eastern species the band of the hind-wing 

 is not only still narrower, but possesses, in dumetorum, a 

 mai'kedly crenulate outer margin which gives it a very dis- 

 tinct appeai'ance. In some individuals of frohenia, slight 

 but obvious traces of this outline ai-e a further indication of 

 affinity between it and dumetorura. Each crenulation occu- 

 pies an internervular space, and is concave in form. In 

 saclava, on the contrary, the outer border of the band projects 

 as more or less of a convexity into each internervular space. 

 The difference may be expressed by saying that the outer 

 margin of the band is formed of concave crenulations in 

 dumetorum, of convex crenulations in saclava. 



As regards the markings of the under surface and of the 

 xxxix] 



upper (notwithstanding the one distinguishing feature described 

 above), comorarum is far nearer to saclava than is frohenia or 

 even dumetorum. In the north-western species the markings 

 on the under surface are paler than those of the upper, while 

 the minute markings are lighter than the larger. The under 

 side of dumetorum is remarkable in the fact that the chief 

 spot of the fore-wing is fulvous while the other markings and 

 those of the hind-wing are white. Indications of the same 

 contrast exist, although far less developed, in frohenia. 



These are the chief characteristics of colour and pattern 

 which distinguish the four island species from one other and 

 from the Madagascar saclava. It is of great interest to attempt 

 the difficult task of reconstructing some stages of the past 

 history of the group. "VVe may confidently assume, from the 

 greater affinity to saclava on the innermost island on each 

 side, that all the four forms were derived from Madagascar ;, 

 and, if amount of change be a measux'e of period of isola- 



