94 Dr. G. D. Hale (Jarpenter on 



Koine is a large island and, like Bugalla, has forest 

 along its shores and in patches on the high ground in its 

 centre. Ngamba and Kimmi are quite small islands both 

 covered almost entirely with forest. They are all so close 

 together that it would be absurd to account for differences 

 between the proportions of forms of Pseudacraea by climatic 

 conditions. At Kakindu is a large area of forest at very 

 little elevation above the lake level, extending from the 

 lake shore five-and-twenty miles inland. 



Now, on Kimmi the only Plmiema models in 1914 were 

 the species with black-and-white females of Combination 11, 

 and the male of one of them belonging to Combination I. 

 The only forms of Pseudacraea were Jiobleyi and tirikensis 

 mimicking these combinations. On Kome the orange tellus 

 of Combination IV predominated, and terra was the most 

 abundant form of Pseudacraea. 



On Ngamba, however, members of Combinations I and II 

 were the prevailing models and the mimics were in pro- 

 portions accordingly, to such an extent that just as Planema 

 tellus appeared to be absent so was terra its mimic ! But 

 at Kakindu tellus was very much the most abimdant, and 

 all the Pseudacraea save one were of the corresponding 

 form. 



But the most remarkable feature of the collections now 

 discussed is the difference between the proportions of 

 models and mimics on Kome island in 1914 and 1918-19. 

 On the former occasion 338 Planema models were taken, 

 and only 5G Pseudacraea, of which only 3 were transitional : 

 the proportion of model to mimic being 6 to 1. In 1918-19, 

 however, 25 Planema models were taken but 55 Pseudacraea, 

 of which 23 were transitional ; the proportion of model to 

 mimic being 1 to 2-2. 



This most interesting result seems to fall in almost too 

 well with the others ! It is supposed tjiat on an island , 

 such as Ngamba the insectivorous birds spare mimics in 

 Combination II but destroy the rest, while on Kome and 

 at Kakindu they find the model of Combination IV so 

 much more abundant that its mimics are more often spared. 

 This, of course, presupposes that the bird population of 

 an island stays there, and that birds from Ngamba do not 

 cross to Kome to hunt. Although this may seem an 

 extravagant theory yet a tour among many islands jn 

 1914 showed that the bird fauna of adjacent isles does 

 differ, and I pointed out in 1918, in reply to some remarks 



